Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid DNA (pDNA) encoding therapeutic or antigenic protein is an important macromolecular agent as a nonviral vector for gene therapy or DNA vaccination. While relatively low efficiency in vivo is the main limiting factor of nonviral gene transfer methods, pDNA has advantages in its safety compared with viral vectors, which are liable to undergo mutation thereby reacquiring the ability to produce infection. Thus, pDNA should be used particularly for long-term and repeated gene therapy requiring improved gene expression efficiency. In addition, it has been proved that bacterial DNA and oligodeoxynucleotides containing immunostimulatory sequences, i.e., the CpG motif, in their backbone have an ability to activate innate immune responses and consequently induce inflammatory cytokines (Krieg et al., 1995, 1996). It has been suggested that oligonucleotides containing the CpG motif represent a powerful adjuvant for both humoral and cellular immune responses (Lipford et al., 1997). Since pDNA amplified in Escherichia coli is a bacterial DNA, pDNA containing the CpG motif works as an adjuvant in DNA vaccines (Davis, 1997).

As regards the in vivo disposition of pDNA, we have previously reported that pDNA is degraded quickly by nucleases in the blood and other compartments and is rapidly removed from the circulation and taken up by the liver, predominantly by the liver nonparenchymal cells (NPCs), after intravenous administration into mice (Kawabata et al., 1995). We have also shown that this rapid hepatic uptake occurs in perfused rat liver (Yoshida et al., 1996). Moreover, we have demonstrated that pDNA is taken up by macrophages via a mechanism mediated by receptors resembling scavenger receptors (Takagi et al., 1998). In addition, class A scavenger receptors that recognize a wide variety of anionic macromolecules are unlikely to be responsible for pDNA uptake as shown by in vivo and in vitro experiments using class A scavenger receptor-knockout mice and peritoneal macrophages from these animals (Takakura et al., 1999).

On the other hand, it has recently been reported that a high level of gene expression could be obtained in mouse liver by direct injection of pDNA by itself (i.e., naked pDNA) in hyperosmotic solution into the portal vein with transient occlusion of the outflow (Budker et al., 1996). More recently, it has been found that a high level of gene expression can be easily obtained by simple injection of naked pDNA into the tail vein with a large volume of saline at a high velocity (Liu et al., 1999; Zhang et al., 1999). This is the so-called hydrodynamics-based transfection procedure (Liu et al., 1999). The hydrodynamics-based procedure is very attractive since a selective and significant gene expression level in the liver can be obtained without any DNA carriers. This procedure has been applied as a simple and efficient in vivo transfection method for screening genes in mice (He et al., 2000; Zhang et al., 2000).

To develop a strategy for optimizing pDNA delivery in gene therapy and DNA vaccination, it is important to elucidate the mechanism of hepatic recognition of naked pDNA because this would be related to pDNA-derived gene expression and pDNA-induced immune responses. However, the detailed mechanism underlying the hepatic uptake of naked pDNA is still unclear. Also, the biodistribution as well as the hepatic uptake mechanism of pDNA injected by the hydrodynamics-based procedure is not yet understood. In the present study, therefore, we studied the in vivo characteristics of the hepatic uptake of pDNA injected by the normal intravenous administration procedure in more detail and also examined the hepatic uptake and gene expression mechanisms after intravenous injection of pDNA by the hydrodynamics-based procedure.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. [-³²P]dCTP (3000 Ci/mmol), [¹⁴C]polyethylene glycol 4000 (PEG₄₀₀₀, mol. wt. = 4,000) and dextran (mol. wt. = 70,000) were obtained from Amersham Pharmacia Biotech (Buckinghamshire, England). ¹¹¹Indium chloride was supplied by Nihon Medi-Physics Co. (Takarazuka, Japan). Polyinosinic acid (poly I, mol. wt. = 103,000) and polycytidylic acid (poly C, mol. wt. = 99,500) were purchased from Amersham Pharmacia Biotech (Uppsala, Sweden). Dextran sulfate (mol. wt. = 150,000) and heparin sodium salt was purchased from Nacalai Tesque (Kyoto, Japan). Gadolinium chloride (GdCl₃), bovine serum albumin (BSA), and IgG were purchased from Sigma (St. Louis, MO). pGL3-control vector was purchased from Promega (Madison, WI). pcDNA3 vector was purchased from Invitrogen (Carlsbad, CA). All other chemicals used were of the highest purity available.

Plasmid DNA. pCMV-Luc encoding firefly luciferase was used as a model pDNA throughout the present study. The pDNA was constructed by subcloning the HindIII/XbaI firefly luciferase cDNA fragment from pGL3-control vector into the polylinker of the pcDNA3 vector (Nomura et al., 1999). The pDNA amplified in the DH5a strain of E. coli was extracted and purified by a QIAGEN Endofree Plasmid Giga kit (QIAGEN GmbH, Hilden, Germany). The purity was checked by 1% agarose gel electrophoresis followed by ethidium bromide staining. The pDNA concentration was measured by UV absorption at 260 nm. For biodistribution and inhibition studies, pDNA was radiolabeled with [-³²P]dCTP by nick translation (Sambrook et al., 1989) and, for confocal microscopic observation, pDNA was labeled with fluorescein using a FastTag FL labeling kit (Vector Laboratories, Burlingame, CA).

Biodistribution Experiments after Intravenous Administration of pDNA by Normal and Hydrodynamics-Based Procedures. Female ICR mice (18-20 g) received the described dose of [³²P]pDNA diluted with unlabeled pDNA in sterilized saline by tail vein injection. For the normal procedure, a normal volume of 10 µl/1 g of body weight, i.e., 200 µl/20-g mouse, was injected over about 5 s. For the hydrodynamics-based procedure, mice were rapidly injected with pDNA in a large volume of saline (1.6 ml) within 5 s. The tail vein injection was performed by using a 26-gauge needle for both the normal and hydrodynamics-based procedure. In the inhibition experiments, 20- or 200-fold higher dose of each polyanion was either injected 1 min before the [³²P]pDNA injection or coadministered. Control mice were injected with saline without any inhibitors. Blood was withdrawn from the vena cava and then centrifuged to obtain plasma and urine was collected from the urinary bladder. The mice were killed after 0.5, 1, 2, 5, 10, or 30 min for the distribution experiments and after 2 or 10 min for the inhibition experiments following [³²P]pDNA injection and the kidney, liver, spleen, lung, and heart were excised at each sampling point, rinsed with saline, and weighed. The samples of plasma, urine, and small pieces of tissue were dissolved in 0.7 ml of Solene-350 at 45°C, and 0.2 ml of isopropanol, 0.2 ml of H₂O₂, 0.1 ml of 5 N HCl, and 5 ml of Clearsol I (scintillation medium) were added to each sample. The radioactivity was measured using a liquid scintillation counter (LSA-500; Beckman, Tokyo, Japan). The radioactivity measured was the total of intact pDNA and its fragments. We examined the pDNA disposition only up to 30 min after injection since the disposition of pDNA after longer periods would not represent that of intact pDNA due to rapid degradation (Kawabata et al., 1995). Only the part of pDNA taken up by the organs in an intact form should be related to gene expression. The plasma volume of each organ was determined from the distribution data of ¹¹¹In-labeled BSA in a separate set of mice at 10 min after intravenous injection and used for the correction of tissue concentrations, assuming that ¹¹¹In-BSA was not taken up by tissues during the 10-min experimental period (Nishikawa et al., 1995).

Data Analysis. The tissue distribution data of various doses of pDNA were evaluated by pharmacokinetic analysis involving the clearance (Takakura et al., 1990). Plasma radioactivity concentrations were normalized to the percentage of dose per milliliter and analyzed by an exponential function using the nonlinear least-square program MULTI (Yamaoka et al., 1981). The total body clearance was calculated by dividing the injected dose by the area under the plasma concentration-time curve (AUC) extrapolated to infinite time. The organ uptake clearance and urinary excretion clearance were calculated as the slope of the integration plot, (X_t/C_t) versus (AUC_0-t/C_t), where X_t is the amount of radioactivity in each organ or urine at an appropriate time interval t and C_t is the plasma radioactivity concentration at the same time t. The data up to 2 min were used to minimize the influence of [³²P]pDNA degradation products.

Confocal Microscopic Study. Mice received fluorescein-labeled pDNA (1 mg/kg) by the normal or hydrodynamics-based procedure. Control mice were injected with 1.6 ml of saline by the hydrodynamics-based procedure. At 10 min after injection, mice were killed by cutting the vena cava and then the liver was gently infused with 5 ml of saline through the portal vein to remove remaining blood. Then, the liver was excised and fixed with 10% neutral formalin. A paraffin-embedded section of the liver was incubated with xylene followed by 100% ethanol to remove the paraffin. Then, the section was incubated with 15 µg/ml RNase (type 1-A; Sigma) at 37°C for 20 min, stained with 0.5 mg/ml propidium iodide (Sigma) at room temperature for 20 min, and finally subjected to confocal microscopy (MRC-1024; Bio-Rad, Hercules, CA).

In Vivo Gene Expression Experiments. Mice received 10 µg of pCMV-Luc (corresponding to 0.5 mg/kg) in 200 µl (normal) or 1.6 ml (hydrodynamics-based) of saline. In some experiments, polyanions (100 or 10 mg/kg) in 100 µl of saline were injected 1 min before pDNA injection. At 6 h after injection, the organs were excised, homogenized in 5 ml/g (liver) or 4 ml/g (other tissues) lysis buffer (0.1 M Tris, 0.05% Triton X-100, 2 mM EDTA, pH 7.8), and subjected to three cycles of freezing (190°C) and thawing (37°C). The homogenates were centrifuged at 10,000g for 10 min at 4°C. Then, 10 µl of appropriately diluted supernatant was mixed with 100 µl of luciferase assay buffer (Picagene; Toyo Ink, Tokyo, Japan) and the chemiluminescence produced was measured in a luminometer (Lumat LB 9507; EG & G Berthold, Bad Wildbad, Germany).

GdCl₃-Induced Kupffer Cell Blockade. Distearoylphosphatidylcholine/cholesterol/Man-C4-cholesterol (60:35:5) liposomes labeled with [³H]cholesteryl hexadecyl ether (Man-Liposome) were prepared as previously described (Kawakami et al., 2000). Mannosylated bovine serum albumin (Man-BSA) was synthesized by reacting BSA with 2-imino-2-methoxyethyl 1-thiomannoside as previously described (Fujita et al., 1992) and radiolabeled with ¹¹¹In using the bifunctional chelating agent diethylenetriaminepentaacetic acid anhydride, according to the method reported previously (Hnatowich et al., 1982). Both Man-Liposome and Man-BSA were used as a positive control because these mannosylated derivatives were taken up by the liver NPCs, predominantly Kupffer cells, via the mannose receptors expressed on the cell membrane (Nishikawa et al., 1992; Kawakami et al., 2000). Mice received GdCl₃ (30 mg/kg) intravenously 24 h before injection of [³²P]pDNA (0.5 mg/kg), Man-Liposome (25 mg/kg), or Man-BSA (1 mg/kg). Hepatic accumulation of each form of radioactivity was determined at 5 min (Man-BSA, Man-Liposome) or 10 min (pDNA) after injection. The effect of GdCl₃-induced Kupffer cell blockade on the gene expression was also examined.

Intravenous Administration of Other Macromolecules by the Hydrodynamics-Based Procedure. To examine the generality of enhanced hepatic uptake by the hydrodynamics-based procedure, three types of model macromolecules with different molecular weights, ¹¹¹In-labeled IgG (mol. wt. = 150,000), ¹¹¹In-labeled BSA (mol. wt. = 67,000), and [¹⁴C]PEG₄₀₀₀ were used for the biodistribution experiments. These macromolecules are not taken up by the liver and any other organs after normal intravenous injection. In the case of PEG₄₀₀₀, mice were anesthetized with a peritoneal injection of pentobarbital sodium (50 mg/kg) and some of them were subjected to renal artery ligation to exclude the effect of glomerular filtration.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

In Vivo Disposition of pDNA after Normal Intravenous Injection. Figure 1 shows the time course of the radioactivity concentration in the plasma and the amounts of radioactivity in the liver, kidney, spleen, lung, heart, and urine after normal intravenous injection of [³²P]pDNA at a dose of 0.5 mg/kg. pDNA was rapidly removed from the circulation with the half-life of about 2 min and mainly taken up by the liver. Liver accumulation of radioactivity reached about 60% of the dose at 1 min and then gradually decreased probably due to the release of degradation products into the circulation. Consistent with this observation, both renal accumulation and urinary excretion of radioactivity increased with time, suggesting that degradation products of [³²P]pDNA were excreted via the kidney.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Plasma concentration and tissue accumulation of radioactivity after normal intravenous injection of [32P]pDNA at a dose of 0.5 mg/kg into mice. Mice were injected with [32P]pDNA in a normal volume of saline (10 µl/1 g of body weight). , plasma; , liver; , spleen; , lung; , heart; , kidney; , urine. The results are expressed as mean ± S.D. of three mice.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Dose dependence of plasma concentration and liver accumulation of [32P]pDNA after normal intravenous injection into mice. [32P]pDNA in a normal volume of saline was injected into mice at a dose of 0.1 (), 0.5 (), or 1 () mg/kg. The results are expressed as mean ± S.D. of three mice.

In Vivo Disposition of pDNA Injected by the Hydrodynamics-Based Procedure. Figure 3 shows the time course of the plasma concentration and tissue accumulation of radioactivity of [³²P]pDNA after intravenous injection by the hydrodynamics-based procedure at a dose of 0.5 mg/kg into mice. The apparent amount of hepatic uptake of [³²P]pDNA by the hydrodynamics-based procedure was similar to that obtained by the normal procedure at the same dose (Fig. 1). However, the plasma concentration profile was significantly different in that the plasma concentration of radioactivity increased with time until 10 min after injection. Moreover, approximately 20% of the radioactivity remained in the plasma pool at 30 min in the hydrodynamics-based procedure while almost the entire radioactivity was eliminated from the circulation at the same time point in the normal procedure. It was also found that approximately 10% of the radioactivity remained in the plasma pool under these conditions at 60 min after [³²P]pDNA injection (data not shown). As shown in the normal procedure, both renal accumulation and urinary excretion of radioactivity increased with time, suggesting that degradation products of [³²P]pDNA were also excreted via the kidney under these conditions. However, the rate of renal accumulation of radioactivity was slightly slower than that obtained by the normal procedure. Pharmacokinetic analysis of the data of pDNA injected by the hydrodynamics-based procedure was not carried out due to possible influence in organ flow rates and the unusual plasma concentration-time profile.

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Plasma concentration and tissue accumulation of radioactivity after intravenous administration of [32P]pDNA by the hydrodynamics-based procedure at a dose of 0.5 mg/kg into mice. , plasma; , liver; , spleen; , lung; , heart; , kidney; , urine. The results are expressed as mean ± S.D. of three mice.

Intrahepatic Distribution of Fluorescein-Labeled pDNA after Intravenous Administration by the Normal and Hydrodynamics-Based Procedures. Figure 4 shows the confocal microscopic images of fluorescein-labeled pDNA in the liver after intravenous injection. In the case of the normal intravenous injection, fluorescein-labeled pDNA was detected selectively in NPCs along the sinusoid (Fig. 4A), being consistent with a previous study involving collagenase perfusion experiments (Kawabata et al., 1995). On the other hand, no concentrated fluorescein signal was observed in any cells after the hydrodynamics-based procedure (Fig. 4B). This image was almost identical to that obtained from control mice that received saline alone (data not shown).

View larger version (45K): [in this window] [in a new window]   Fig. 4.   Confocal microscopic image of fluorescein-labeled pDNA in the liver after intravenous injection by normal (A) and hydrodynamics-based (B) procedures. Mice received fluorescein-labeled pDNA (1 mg/kg) by tail vein injection. The images shown are typical of those observed in several visual fields.

Effects of GdCl₃-Induced Kupffer Cell Blockade on Hepatic Uptake of pDNA. Figure 5 shows the effects of Kupffer cell blockade induced by GdCl₃ on the hepatic uptake of pDNA. Man-Liposome and Man-BSA were used as positive controls that were supposed to be affected by GdCl₃ because these mannosylated derivatives were mainly taken up by the liver NPCs, predominantly Kupffer cells, via the mannose receptors expressed on these cells (Nishikawa et al., 1992; Kawakami et al., 2000). The amount of hepatic uptake of both Man-Liposome and Man-BSA was significantly less in GdCl₃-treated mice compared with control mice. In contrast, the amount of hepatic uptake of pDNA after intravenous administration by the normal and hydrodynamics-based procedures was not significantly affected by GdCl₃ treatment, suggesting that the contribution of Kupffer cells to the overall hepatic uptake was small in both cases.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Effects of Kupffer cell blockade by GdCl3 on hepatic accumulation of [32P]pDNA, 111In-Man-BSA, or [3H]Man-Liposome. Mice received GdCl3 (30 mg/kg) by tail vein injection 24 h before [3H]Man-Liposome (25 mg/kg), 111In-Man-BSA (0.1 mg/kg), or [32P]pDNA (0.5 mg/kg) injection. [32P]pDNA was injected by both normal and hydrodynamics-based procedures. Control mice received saline instead of GdCl3. At 5 min (Man-BSA, Man-Liposome) or 10 min (pDNA) after injection, hepatic accumulation was determined. The results are expressed as mean ± S.D. of three mice. Significantly different (*p < 0.01) from the value for control mice. , control mice; , GdCl3-treated mice.

Inhibition of Hepatic Uptake of pDNA by Various Polyanions. Figure 6 shows the effects of various polyanions on the plasma concentration and hepatic uptake of [³²P]pDNA after normal intravenous injection. We used various polyanions such as poly I, which is a ligand of scavenger receptors; poly C, which is not a ligand of scavenger receptors; synthetic and natural DNA; sulfated and nonsulfated glycosaminoglycans; dextran; and its sulfate ester. Mice received 20-fold higher doses of polyanions either beforehand or coadministered with [³²P]pDNA. The hepatic uptake of [³²P]pDNA (0.1 mg/kg) was inhibited by a simultaneously administered excess dose of unlabeled pDNA (1 or 10 mg/kg), indicating that the hepatic uptake of pDNA was saturable. Heparin, calf thymus DNA, and polydeoxyinosinic-polydeoxycytidylic acid, as well as dextran sulfate and poly I, proved to be potent inhibitors of the hepatic uptake of pDNA as far as the various polyanions tested were concerned. In addition, heparan sulfate and polydeoxycytidylic acid slightly inhibited the uptake.

View larger version (26K): [in this window] [in a new window]   Fig. 6.   Inhibition of hepatic uptake of [32P]pDNA by various polysaccharides or polynucleotides after normal intravenous injection. A, [32P]pDNA (1 mg/kg) was injected at 1 min after administration of polyanion (20 mg/kg). B, [32P]pDNA (0.1 mg/kg) was coadministered with unlabeled pDNA (1 or 10 mg/kg) or polyanion (2 mg/kg). The plasma concentration and hepatic accumulation were determined at 10 (A) or 2 min (B) after [32P]pDNA injection. The results are expressed as mean ± S.D. of three mice. Significantly different (*p < 0.05, **p < 0.01) from the control values. , plasma concentration; , hepatic accumulation.

View larger version (19K): [in this window] [in a new window]   Fig. 7.   A, inhibition of hepatic uptake of [32P]pDNA by various polysaccharides or polynucleotides after intravenous administration by the hydrodynamics-based procedure. [32P]pDNA (0.5 mg/kg) was administered at 1 min after dextran (100 mg/kg), dextran sulfate (10 mg/kg), heparin (100 mg/kg), or poly I (100 mg/kg) injection. The plasma concentration and hepatic accumulation were determined at 10 min after [32P]pDNA injection. The results are expressed as mean ± S.D. of three mice. Significantly different (*p < 0.05) from the control values. , plasma concentration; , hepatic accumulation. B, effects of polyanions on gene expression produced by pDNA injected by the hydrodynamics-based procedure. pDNA (10 µg in 1.6 ml saline) was administered at 1 min after dextran (100 mg/kg), dextran sulfate (10 mg/kg), heparin (10 mg/kg), poly C (10 mg/kg), or poly I (10 mg/kg) injection. At 6 h after injection, luciferase activity in each tissue was measured. The results are expressed as mean ± S.D. of three mice.

Effects of Polyanions and GdCl₃-Induced Kupffer Cell Blockade on Gene Expression after the Hydrodynamics-Based Procedure. Figure 7B illustrates the effects of other macromolecules on gene expression after the hydrodynamics-based procedure. The high level of gene expression obtained by this procedure was not affected by polyanions such as poly I, heparin, and dextran sulfate. GdCl₃ treatment also did not reduce the gene expression of pDNA injected by the hydrodynamics-based procedure (data not shown).

Hepatic Uptake of Macromolecules after the Hydrodynamics-Based Procedure. To examine whether the hydrodynamics-based procedure delivers not only pDNA but also any other macromolecules to the liver, ¹¹¹In-IgG, ¹¹¹In-BSA, and [¹⁴C]PEG₄₀₀₀ were used as model macromolecules unlikely to be taken up by the liver after intravenous injection. Figure 8, A and B, shows the plasma concentration and tissue accumulation of ¹¹¹In-IgG and ¹¹¹In-BSA after intravenous injection by the normal or hydrodynamics-based procedures. Both ¹¹¹In-IgG and ¹¹¹In-BSA injected by the normal procedure were not distributed to any organs, while a significant amount of their associated radioactivity was detected in the liver by the hydrodynamics-based procedure. Figure 8, C and D, shows the plasma concentration and tissue accumulation of [¹⁴C]PEG₄₀₀₀ after intravenous injection by the normal or hydrodynamics-based procedures with or without renal artery ligation. Since [¹⁴C]PEG₄₀₀₀ is small enough to undergo glomerular filtration, a high degree of urinary excretion of [¹⁴C]PEG₄₀₀₀ injected by both the normal and hydrodynamics-based procedure was detected in the mice that did not undergo renal artery ligation. The hepatic accumulation of [¹⁴C]PEG₄₀₀₀ injected by the hydrodynamics-based procedure was not significantly higher than that obtained by the normal procedure. Similarly, no significant increase in the hepatic accumulation of [¹⁴C]PEG₄₀₀₀ injected by the hydrodynamics-based procedure was observed in the mice undergoing renal artery ligation, even though the glomerular filtration of [¹⁴C]PEG₄₀₀₀ was blocked.

View larger version (21K): [in this window] [in a new window]   Fig. 8.   Plasma concentration and tissue accumulation of radioactivity of 111In-IgG (A), 111In-BSA (B), and [14C]PEG4000 (C and D) after intravenous injection by the normal or hydrodynamics-based procedure. 111In-IgG (0.5 mg/kg), 111In-BSA (0.5 mg/kg), or [14C]PEG4000 (0.5 mg/kg) was injected into mice either by the normal or hydrodynamics-based procedure. [14C]PEG4000 was injected into mice under normal conditions (C) or after renal artery ligation (D). The plasma concentration and tissue accumulation were determined at 10 min after injection. The results are expressed as mean ± S.D. of three mice. Significant difference (*p < 0.01) between procedures. P, plasma; Li, liver; Sp, spleen; Lu, lung; H, heart; K, kidney; U, urine. , normal procedure; , hydrodynamics-based procedure.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The biodistribution after intravenous injection of various naked DNAs and their complexes, such as single-stranded DNA, double-stranded DNA, oligonucleotide, DNA anti-DNA immune complex, mononucleosome, or chromatin, has been examined. Although these studies have shown that the liver is the main organ responsible for the rapid clearance of these DNAs from the circulation (Emlen and Mannik, 1978, 1984; Emlen and Burdick, 1988; Gauthier et al., 1996; Du Clos et al., 1999), the uptake mechanism and the cell type(s) contributing to this hepatic uptake remain to be elucidated.

In the present study, we have confirmed that pDNA is rapidly eliminated from the circulation and taken up by the liver after conventional intravenous injection (Fig. 1). The hepatic uptake was dose-dependent and saturable (Fig. 2; Table 1). We also found that fluorescein-labeled pDNA injected by the normal procedure was selectively observed in the hepatic cells along the sinusoid (Fig. 4A), consistent with our previous observation that pDNA is preferentially taken up by the liver NPCs (Kawabata et al., 1995; Yoshida et al., 1996). Furthermore, our present results suggest that, among the NPCs, liver endothelial cells rather than Kupffer cells make the major contribution to the overall uptake, at least in the early phase, since GdCl₃-induced Kupffer cell blockade did not affect the degree of hepatic uptake of pDNA (Fig. 5). GdCl₃ blocks Kupffer cell function and eliminates the population of Kupffer cells by causing macrophage apoptosis (Hardonk et al., 1992). The dose of GdCl₃ used in the present study was relatively higher than that used in other studies for the same purpose and appeared to be enough to impair Kupffer cell function because the hepatic uptake of both Man-Liposome and Man-BSA was reduced in GdCl₃-treated mice. Emlen et al. (1988) have shown that single-stranded linear DNA binds to liver sinusoidal lining cells, primarily Kupffer cells, via a receptor-mediated process using a mouse perfused liver system. Differences in DNA type (single and linear versus double and circular) and experimental systems (in situ versus in vivo) may account for this discrepancy.

Moreover, the present study has provided a novel insight into the specificity of the mechanism underlying the in vivo pDNA uptake by the liver after normal intravenous injection. In the inhibition experiments using various polyanions, the hepatic uptake of pDNA was inhibited by heparin, heparan sulfate, and polydeoxyinosinic-polydeoxycytidylic acid, as well as typical ligands for scavenger receptors such as dextran sulfate and poly I, but not by poly C, hyaluronic acid, chondroitin sulfate, or dextran (Fig. 6). The inhibition profile was very similar to that observed in our in vitro experiments using mice peritoneal macrophages (Takakura et al., 1999). Poly G, poly I, and other polynucleotides are known to form a base-quartet-stabilized four-strand helix (quadruplex), which is presumably highly polyanionic structure (Pearson et al., 1993; Krieger and Herz, 1994). The density of negative charges might be important in the hepatic uptake because the most potent inhibitors seem to be more densely sulfated than the weaker inhibitors on a basis of a disaccharide unit. Taken together, these results suggest that pDNA is taken up by the liver NPCs, predominantly by the endothelial cells, after conventional intravenous injection via a mechanism that recognizes pDNA as a polyanion, presumably based on its three-dimensional structure. Further studies are required to clarify the details of this mechanism.

It seems that there is a concern of toxicity of injected pDNA itself or the injection technique of the hydrodynamics-based procedure. Although pDNA is known to be capable of eliciting an acute inflammatory response (Krieg et al., 1995, 1996), a high dose of pDNA (300 µg/mouse) did not cause liver damage in our preliminary study (N. Kobayashi, K. Yamaoka, and Y. Takakura, unpublished data). Hydrodynamics-based procedure appears to be an invasive method. However, Liu et al. (1999) found that there was no indication of serious liver damage assessed by animal growth and clinical biochemistry tests, which were all in normal range with the exception of transient increase of serum concentration of alanine aminotransferase. This minor liver damage was not due to pDNA but the large volume of saline injected.

Although marked gene expression could be obtained in the liver by the hydrodynamics-based procedure but not the normal one, the time course profile and the amount of hepatic accumulation of pDNA injected by the hydrodynamics-based and normal procedures were similar (Figs. 1 and 3). The large volume of injected saline was almost equal to the blood volume and may dilute the radioactivity in the plasma pool, resulting in the distinct plasma-concentration profile in the hydrodynamics-based administration experiment. Delayed elimination of pDNA from the plasma pool may be ascribed to reduced pDNA degradation by nuclease. While the hepatic accumulation profiles of [³²P]pDNA were almost identical, the results of the confocal microscopic studies and inhibition experiments were different. Neither poly I, heparin, nor dextran sulfate, which are potent competitors of pDNA injected by the normal procedure, affected the hepatic uptake of pDNA injected by the hydrodynamics-based procedure (Fig. 7A). These results suggest that the hepatic uptake of pDNA injected by the hydrodynamics-based procedure is a nonspecific process and that fluorescein-labeled pDNA spreads out into the whole liver so that a strong signal could not be detected (Fig. 4B). Similar to the hepatic uptake of pDNA, these tested polyanions did not affect the high level of gene expression in the liver (Fig. 7B).

Due to the large amount of nuclease in the blood and in other compartment such as on the surface of tissues (Emlen et al., 1988), pDNA injected by the normal procedure is likely to be rapidly degraded in circulation and subsequently in liver cells after it is taken up via specific receptors. On the other hand, a part of pDNA injected by the hydrodynamics-based procedure would be directly exposed to the liver cells and nonspecifically taken up by the cells through the cellular membrane as intact molecules before mixed with blood. This accounts for the finding that a high level of gene expression can be obtained by the hydrodynamics-based procedure, and not by the normal one.

Very recently, Budker et al. (2000) hypothesized that naked pDNA is taken up by parenchymal cells in vivo by a receptor-mediated process based on the finding that coinjection with excess polyanions, including nonexpressing pDNA, sonicated salmon sperm DNA, polyglutamic acid, poly C, poly I, and high-density lipoprotein, inhibited the gene expression of pDNA injected under high-pressure conditions. In this report, pDNA and polyanions were coinjected under high-pressure conditions, whereas we preinjected polyanions by the normal procedure, followed by pDNA injection by the hydrodynamics-based procedure. Our preliminary study showed that coinjection of 20-fold higher dose of dextran sulfate did not inhibit the hepatic uptake of [³²P]pDNA (0.5 mg/kg) injected by the hydrodynamics-based procedure (data not shown), suggesting that the uptake of pDNA is not affected by an excess of coadministered polyanion, while the subsequent gene expression process may be inhibited by polyanions.

We further characterized the hydrodynamics-based procedure by using other macromolecules to examine whether the hydrodynamics-based procedure delivered not only pDNA but also other macromolecules to the liver. As shown in Fig. 8, A and B, a significant amount (approximately 20%) of hepatic accumulation of both ¹¹¹In-IgG and ¹¹¹In-BSA was observed in the hydrodynamics-based procedure, indicating that this procedure is also applicable to the hepatic delivery of other macromolecules via a nonspecific mechanism. However, in the case of PEG₄₀₀₀, much lower than IgG (mol. wt. = 150,000) and BSA (mol. wt. = 67,000), a significant amount of hepatic accumulation was not seen in the hydrodynamics-based procedure both in mice under normal conditions and with renal artery ligation for the purpose of excluding glomerular filtration (Fig. 8, C and D). Consistently, TOTO-1, a molecular probe with a molecular weight less than 1000, was not observed when it was injected into the tail vein under high-pressure and high-volume injection conditions (Budker et al., 2000). These results suggest that hepatic delivery by the hydrodynamics-based procedure depends on the molecular weight of the injected compound and requires a relatively high molecular weight. Although this molecular weight-dependent phenomenon requires further investigation, we speculate that transient enhanced membrane permeability, such as the formation of small pores, may be involved. Immediately after injection, any macromolecules, including pDNA, might be taken up by the liver via transient small pores on the cell membrane on the flow caused by high blood pressure, which would be independent of the molecular weight. However, smaller macromolecules could be released from the cells more rapidly via these pores by diffusion, a molecular weight-dependent process. Thus, it is possible that apparent amount of hepatic uptake depends on the physicochemical characteristics of the macromolecules involved.

In conclusion, we have identified some characteristics in the process of in vivo hepatic uptake of pDNA following intravenous injection via normal and hydrodynamics-based procedures. Under normal conditions, the contribution of endothelial cells to the rapid hepatic uptake of pDNA is greater than that of Kupffer cells. The hepatic uptake mechanism of pDNA is still unclear, but any such mechanism is likely to recognize some specific structure of pDNA and other polyanions. In the hydrodynamics-based procedure, pDNA was taken up by the liver via a nonspecific mechanism. This method will be applicable to the delivery of macromolecules with a relatively high molecular weight, as well as pDNA, to the liver. These findings provide a useful basis for pDNA delivery in gene therapy and DNA vaccination procedures.