Introduction

Top Summary Introduction Procedures Results Discussion References

Antisense ODNs have been shown to interdict gene expression at various levels (Stein and Cheng, 1993). Because their inhibitory activity is highly specific and their action rarely accompanied by toxic side effects, ODNs form promising clinical alternatives for conventional drug therapy. For effective therapeutic application, the pharmacokinetics of ODNs must be mapped thoroughly. The in vivo fate of ODNs and DNA in animal models has been addressed in a number of studies (for a review, see Crooke and Bennett, 1996). ODNs, and phosphodiester ODNs in particular, were reported to be cleared rapidly from the bloodstream after injection into rats due to extensive liver and kidney uptake (Emlen et al., 1988; De Smidt et al., 1991; Zendegui et al., 1992; Inagaki et al., 1994; Sands et al., 1994; Rifai et al., 1996; Sawai et al., 1996). The qualitative studies of Emlen et al. (1988) and Rifai et al. (1996) suggested that within the liver, nonparenchymal liver cells were responsible for hepatic uptake of double-strand DNA and phosphorothioate ODNs, respectively. Although the renal disposition of ODNs was claimed to be mediated by scavenger receptor A-I/II (Sawai et al., 1996), the hepatic recognition sites responsible for the ODN uptake remain to be identified.

On the basis of in vitro studies, a number of cellular uptake pathways for ODNs and DNA have been suggested (Bennet et al., 1985; Loke et al., 1989; Yabukov et al., 1989; Pearson et al., 1993; Stein et al., 1993; Zhao et al., 1993; Benimetskaya et al., 1997). Bennett et al. (1985) showed that leukocytes bind and take up -phage DNA via intracytoplasmic vacuoles through a 30-kDa receptor. Yabukov et al. (1989) and Stein et al. (1993) characterized the mechanism of binding and uptake of ³²P-ODN by fibroblasts and HL-60 cells. The uptake of ODN seemed to be mediated in part by pinocytosis (Stein et al., 1993) and in part by MAC-1 (CD11b/CD18; Benimetskaya et al., 1997). In addition, Pearson et al. (1993) recently demonstrated that ³²P-ODN uptake was increased in scavenger receptor-transfected CHO cells. Together with the finding that scavenger receptor-expressing nonparenchymal liver cells contribute to hepatic uptake (Emlen et al., 1988), this is suggestive of an important role for scavenger receptors in ODN clearance in vivo.

Despite the fact that the above studies indicate that different and cell-specific pathways may be implicated in uptake of ODNs, the actual nature of the elimination sites of the major elimination site, the liver, is still under investigation. We therefore have mapped the pathway or pathways for hepatic uptake of ODNs in greater detail so ODN delivery protocols can be designed that either use or avoid these recognition systems.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. Na¹²⁵I in 0.1 M NaOH (13.5 mCi/µg) and [-³²P]ATP was purchased from Amersham (Buckinghamshire, England). Collagenase (type IV), trypan blue, BSA (fraction V), levamisole·HCl, chloroquine diphosphate, polyadenosinic acid (potassium salt), polyguanosinic acid (potassium salt), polycytidinic acid (potassium salt), polyinosinic acid (potassium salt), heparin (from porcine intestinal mucosa, grade I), and fucoidin (from Fucus vesiculosus) were purchased from Sigma Chemical (St. Louis, MO). Salmon sperm DNA (native; molecular mass, 3.8 × 10⁶ Da) and T4-polynucleotide kinase were purchased from Pharmacia (Woerden, The Netherlands). Monensin (sodium salt) was obtained from Calbiochem-Behring (La Jolla, CA). ODN were synthesized at the Department of Organic Chemistry (Dr. J. H. Van Boom) (5-GAC.TTT.AGT.CGT.CGT.GGA-cap, ODN1; 5-TCC.ACG.ACG.ACT.AAA .GTC.TTT-cap, ODN2) or purchased from Eurogentec (Seraign, Belgium) (GTG.CCG.GGG.TCT.TCG.GGC-cap, ODN3; TTT.GTT.TGT. TTG.TTT.T, dG₃T₁₃; TGG.TGG.TGG.TGG.TGG.T, dG₁₀T₆; GGG.TGG.GTG.GGT. GGG.G, dG₁₃T₃; GGG.GGG.GGG.GGG.GGG.G, dG₁₆).

In vivo serum clearance and liver association. Male Wistar rats, weighing ~250-300 g, were anesthetized by intraperitoneal injection of 15-20 mg of sodium pentobarbital. The abdomen was opened, and ³²P-ODN [10 µg in 500 µl of PBS (10 mM NaP_i, 150 mM NaCl, pH 7.4)] was injected into the inferior vena cava. At the indicated times, blood samples of 0.2-0.3 ml were taken from the inferior vena cava. The samples were centrifuged for 2 min at 16,000 × g, and the serum was counted for radioactivity. The total amount of radioactivity in serum was calculated using the equation: serum volume (ml) = [0.0219 × body weight (g)] + 2.66 (Bijsterbosch et al., 1989). At the indicated times, liver lobules were tied off, excised, weighed, and counted for radioactivity. The total excised liver tissue amounted to <15% of the total liver mass. The liver uptake of the injected compound was corrected for the radioactivity in serum assumed to be entrapped in the tissue at the time of sampling (85 µl/g wet weight) (Caster et al., 1955).

Isolation of liver cells. Rats were anesthetized and injected with ³²P-ODN (10 µg in 500 µl of PBS) as described. At 10 min after the injection, the vena porta was cannulated, and the liver was perfused with Ca²⁺-free Hanks' balanced salt solution plus 10 mM HEPES, pH 7.4 (8°), at a flow rate of 14 ml/min. After 8 min, a lobule was tied off for determination of the total liver uptake. Subsequently, the liver was perfused with 0.01% (w/v) collagenase at 8° in Hanks' balanced salt solution containing 10 mM HEPES, pH 7.4, and parenchymal and nonparenchymal cells were isolated as described previously (Van Berkel et al., 1985). The nonparenchymal cell preparation was fractionated further into EC and KC by centrifugal elutriation as described in detail previously (Kuiper et al., 1994). The contributions of the various cell types to the total liver uptake were calculated from the radioactivity recovered in the respective liver cell fractions after correction for the protein content, with the assumption that 92.5%, 2.5%, and 3.3% of the total liver protein content can be ascribed to the parenchymal cells, KC, and EC, respectively (Kuiper et al., 1994). As found previously with other substrates (Van Berkel et al., 1985; Kuiper et al., 1994), no significant loss of radioactivity from the cells during the isolation procedure was observed. This was checked in each experiment by comparing the calculated liver uptake (i.e., the summation of the contributions of the various cell types) with the value actually measured in the liver lobule, with the average value being 98 ± 5%. The purity of EC and KC preparations as monitored by peroxidase staining was 95% and 90%, respectively. The viability of the cells used for in vitro experiment was 90% as judged by 0.2% Trypan blue exclusion.

Isolation, modification, and radioiodination of human LDL. Human LDL (1.024 < d < 1.055) was isolated from human serum by two repetitive centrifugations according to Redgrave et al. (1975). After isolation, the purity and integrity of LDL were checked using agarose gel electrophoresis (0.8%) and PAGE. LDL was acetylated as described by Van Berkel et al. (1991). After modification, AcLDL was stored in PBS containing 2 mM EDTA at 4° under nitrogen and used within 4 weeks. The electrophoretic mobility of AcLDL on native agarose gel electrophoresis was 0.54 compared with 0.21 for native LDL. AcLDL was radioiodinated with Na¹²⁵I to a specific activity of 110-230 cpm/ng as described previously (Van Berkel et al., 1991).

Radiophosphorylation of ODNs. The ODNs were 5-end labeled with [-³²P]ATP using T4-polynucleotide kinase and monitored on purity and by gel electrophoresis on 19% polyacrylamide gel (80 mM Tris, 2 mM EDTA, 90 mM boric acid, 7 M urea, pH 8.8) (Sambrook et al., 1989). Typically, the specific activity of ³²P-ODN was 100-250 cpm/ng.

In vitro binding of ³²P-ODN1 and ¹²⁵I-AcLDL to EC and KC. Binding studies of ³²P-ODN1 to EC and KC were performed as described previously (Biessen et al., 1994). KC (1.5 × 10⁶; 150 µg of cell protein) or EC (2 × 10⁶; 150 µg of cell protein) in DMEM (0.5 ml) containing 2% (w/v) BSA were incubated for 2 hr at 4° with a concentration of radiolabeled ligand of 5-250 nM. After incubation, cells were washed twice with DMEM plus 0.2% BSA and once with DMEM, and the cell-associated radioactivity was counted. Nonspecific binding was defined as the binding of ³²P-ODN1 in the presence of 100 µg/ml polyinosinic acid.

Kinetics of initial uptake of ³²P-ODN1 by EC and KC. Uptake of ³²P-ODN1 by EC and KC was studied as follows. KC (1.5 × 10⁶; 150 µg of cell protein) or EC (2 × 10⁶;150 µg of cell protein) in DMEM (0.5 ml) containing 2% (w/v) BSA were incubated for 10 min at 37° with radiolabeled ligand in the absence or presence of polyinosinic acid (100 µg/ml) or GMP (10 mM), 4-nitrophenylphosphate (10 mM), L-(+)-tartaric acid (1 mM), and levamisole (10 µM). After incubation, cells were placed on ice, washed with ice-cold DMEM plus 2.0% (w/v) BSA, and incubated for 10 min at 4° with DMEM plus 2.0% BSA plus polyinosinic acid [(w/v) 100 µg/ml] to remove membrane-associated ³²P-ODN1. In this time span, membrane-bound ³²P-ODN1 was almost quantitatively released from EC and KC (>90%). Subsequently, cells were processed as described before, and the cell-associated radioactivity was counted. Nonspecific uptake was defined as the binding of ³²P-ODN1 in the presence of 100 µg/ml polyinosinic acid. To test the effect of various agents that affect lysosomal uptake, cells were incubated for 30 min at 37° with DMEM plus 2.0% BSA in the absence or presence of NaN₃ (10 mM), chloroquin (1 mM), monensin (25 µM), or sucrose (250 mM). Subsequently, ³²P-ODN1 was added, and the cells were incubated and processed as described above.

Determination of protein, cholesterol, and carbohydrate content. Protein concentrations in cell suspensions were determined according to the method of Lowry et al. (1951), with BSA as the standard.

Data processing. Saturation binding data and substrate curves of initial uptake were analyzed according to a single-site binding model using nonlinear regression (GraphPAD, ISIS Software). Displacement studies were analyzed according to a single-site competitive displacement model using nonlinear regression (Biessen et al., 1988). Statistical significance of the differences was quantified by Student's t test.

	Results

Top Summary Introduction Procedures Results Discussion References

The in vivo behavior of three phosphodiester ODNs (miscellaneous sequences with G contents of 16-50%) has been addressed in rats. As a measure for the stability of ODNs in the circulation, we first determined the rate of degradation of ³²P-ODN1 in the presence of serum (50% v/v) at 37°. ODN1 is degraded slowly in the presence of serum at an apparent half-life of 19 ± 6 min (data not shown). On intravenous injection into rats, the miscellaneous ODN sequences (ODN1, ODN2, and ODN3) were cleared rapidly from the bloodstream with similar elimination half-lives of ~0.65 min (Fig. 1). Within 2 min after injection, only 11.2-12.9% of the injected dose resided in the serum. At this time, <10% of ³²P-ODN1 is degraded as determined by PAGE. The liver seemed to be the main site of clearance: ³²P-ODN1 (G content, 33%) displayed the highest uptake (36 ± 4% of the injected dose), whereas liver uptake of ³²P-ODN2 (16% G content) and ³²P-ODN3 (an antisense sequence for murine c-myb; 50% G content) amounted to 22 ± 2% and 23 ± 2%, respectively. Other major sites of ODN recovery were skin (recovery ranging from 9.0-13.7% of the injected dose for the various ODNs), muscles (8.8-13.8%), small intestine (5.0-7.5%), kidneys (1.3-3.5%), and bone marrow (7.0-13.4%).

View larger version (36K): [in this window] [in a new window]   Fig. 1.   Liver uptake (top) and serum half-life (bottom) of various 32P-ODNs in the rat. 32P-ODNs (ODN1, ODN2, ODN3, dG3T10, dG10T6, and dG16; 4 µg) were injected intravenously into rats. At 1, 2, 5, 10, 20, 30, and 40 min after injection, radioactivities in serum was determined, and the serum half-life was calculated using a computerized single-phase exponential decay algorithm. Liver-associated radioactivities was determined at 40 min after injection. Values are mean ± standard deviation of a determination performed in triplicate (ODN1 and ODN2) or duplicate (ODN3, dG3T10, dG10T6, and dG16).

Because liver uptake varied considerably from one sequence to another and because it has been suggested that receptor recognition may be influenced by the G content (Pearson et al., 1993), we assessed the effect of G content on the in vivo fate of GT-containing model ODNs. These GT-containing ODNs are telomere-like oligodeoxyribonucleotides that have the tendency to form quadruplex structures at high G contents. These quadruplexes are very good substrates for SR-AI/II (Pearson et al., 1993). The serum decay of these GT-ODNs was essentially similar to that of the miscellaneous sequences. Liver uptake was maximal for dG₁₀T₆ (G content, 62%; 31 ± 2% liver uptake). At higher G contents [i.e., dG₁₃T₃ (81%) and dG₁₆ (100%)], liver uptake was strongly impaired due to a significantly increased bone marrow uptake (up to 45% of the injected dose for dG₁₆). Hepatic uptake of ³²P-ODN1 was significantly reduced after heat-induced denaturation (22 ± 2% versus 36 ± 4% for control ODN1; p < 0.01). Isolation of the various liver cell types at 10 min after injection revealed that hepatic ODN uptake could be mainly ascribed to KC (39% average) and EC (43%) (Fig. 2). Parenchymal liver cells, although constituting 92.5% of the total liver mass, were responsible for only 13-22% of the liver uptake. There was no direct correlation between the ratio of EC to KC uptake and the G content or sequence of the ODN.

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Contribution of various cell types to the liver association of 32P-ODN1. 32P-labeled ODN (ODN1, dG3T13, dG10T6, dG13T3, and dG16; 10 µg in 500 µl of PBS) was injected into rats. At 10 min after injection, parenchymal cells (PC), EC, and KC were isolated from the liver, and cellular radioactivity was counted. Values are mean ± standard deviation of three (ODN1) or two (other ODNs) experiments and are expressed as percentage of the total injected dose that is recovered in the various liver cell types, assuming the parenchymal liver cell, EC, and KC contribute 92.5%, 3.3%, and 2.5% to the liver protein content, respectively.

To identify the recognition site or sites responsible for liver uptake, we investigated the interaction of the ODN with the highest liver uptake, ODN1, with isolated EC and KC. To establish conditions for equilibrium binding, we first determined the kinetics of the association of ³²P-ODN1 to EC and KC showing that at 4° and 1 nM ³²P-ODN1, equilibrium binding was achieved within 2 hr of incubation. ³²P-ODN1 binding to EC and KC was monophasic (Hill coefficient close to unity), saturable (B_max = 112.5 ± 4.9 and 51.9 ± 5.2 ng/mg, respectively), and of moderate affinity (K_d = 109 ± 22 and 102 ± 9 nM, respectively) (Fig. 3). The protein nature of the ³²P-ODN1 binding site or sites on both cell types was established by determining the effect of pretreatment of the cells with trypsin (37° for 15 min). Because binding to EC and KC was reduced by 60-80% (Fig. 4), it can be concluded that ³²P-ODN1 binding is largely mediated by a membrane-bound protein. Incubation of the cells with glycine buffer (pH 2.8) also impaired binding by 70%, confirming that a protein receptor may be involved in ³²P-ODN1 binding to both cell types. ³²P-ODN1 binding to nonparenchymal liver cells was not influenced by incubation with EDTA (10 mM), indicating that ³²P-ODN1 binding is a Ca²⁺-independent process.

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Left, 32P-ODN1 binding to EC () and KC (). Effect of trypsin digestion. EC or KC (2-3 × 106 cells) were incubated for 15 min at 37° in 1 ml of DMEM or trypsin (500 µg/ml) and subsequently washed thoroughly with DMEM supplemented with 2% BSA. Then, cells were incubated for 2 hr at 4° with 32P-ODN1 (20 nM). After incubation, cells were immediately washed as described in Experimental Procedures. Cell-bound radioactivity was determined and corrected for protein content (100% binding corresponds with 10.0 ± 0.9 ng/mg for EC and 4.5 ± 0.4 ng/mg for KC). Right, Effect of pH 2.8 shock or EDTA treatment. EC or KC (2-3 × 106 cells/ml) in 1 ml of DMEM were incubated for 2 hr at 4° with 20 nM 32P-ODN1 in the absence or presence of EDTA (10 mM). After incubation, cells were either washed as described above or incubated with glycine buffer (200 mM, pH 2.8) for 10 min at 4° and subsequently washed. Cell-bound radioactivity was determined and corrected for protein content.

View larger version (27K): [in this window] [in a new window]   Fig. 4.   Saturation curve of 32P-ODN1 binding to EC and KC at 4°. EC (, ) or KC (, ) (2-3 × 106 cells/ml) were incubated for 2 hr at 4° with 0-250 nM 32P-ODN1 in the absence (, ) or presence (, ) of 100 µg of polyinosinic acid. After incubation, the cells were washed thoroughly, and the cell-associated radioactivity was determined. Inset, Scatchard plots of the two binding curves are given (, EC; , KC). Lines, nonlinear regression analysis of the binding curves.

Subsequently, we investigated the effect of various anionic competitors on the binding of ³²P-ODN1 to EC and KC (Fig. 5). ³²P-ODN1 binding was inhibited for 85-90% by polyinosinic acid (100 µg/ml) and polyguanosinic acid (100 µg/ml), whereas polyadenosinic acid (100 µg/ml) and polycytidinic acid (100 µg/ml) were not effective. In addition, it was demonstrated that the ability of mononucleotides to inhibit ³²P-ODN1 binding decreased from ATP and GTP (data not shown) via ADP to AMP, which was only a poor inhibitor at 50 mM. Similarly, the potency of phosphodiester 16-mers to interfere with ³²P-ODN1 binding increased with increasing G content from an insignificant 5% inhibition for dG₃T₁₃ (at 1 µM) to 40% for dG₁₀T₆ and even 90% inhibition for dG₁₆. Of the other polyanions, only fucoidin gave full inhibition of ³²P-ODN1 binding. Lipopolysaccharide (150 µg/ml) and heparin (100 units/ml) were moderately potent, whereas salmon sperm DNA (500 µg/ml) was completely ineffective as an inhibitor. The inhibition profiles of ³²P-ODN1 binding to EC and KC not only were mutually identical but also closely matched that of ¹²⁵I-AcLDL binding to both cell types, suggesting that scavenger receptor type binding sites are involved in ³²P-ODN1 binding to nonparenchymal liver cells. Therefore, we monitored the effect of AcLDL and OxLDL, both established substrates for the most relevant scavenger receptors [scavenger receptor class A-I/II (Horiuchi et al., 1985), CD36 (Endeman et al., 1993), and Fc/RII (Stanton et al., 1992)] on ³²P-ODN1 binding (Fig. 6). AcLDL seemed to reduce ³²P-ODN1 binding to EC and KC by ~40% (six experiments), whereas OxLDL led to a small but significant 15-20% inhibition; however, there was a considerable interexperimental variation in the effect of both modified lipoproteins on ³²P-ODN binding.

View larger version (38K): [in this window] [in a new window]   Fig. 5.   Effect of various (poly)anions on the binding of 32P-ODN1 (closed bars) or 125I-AcLDL (open bars) to EC (top) and KC (bottom). EC or KC (2-3 × 106 cells/ml) were incubated in DMEM (supplemented with 2% BSA), with 32P-ODN1 (20 nM) or 125I-AcLDL (9.6 nM), in the absence (control) or presence of polyinosinic acid (polyI, 100 µg/ml), polyadenosinic acid (polyA, 100 µg/ml), polyguadinilic acid (polyG, 100 µg/ml), polycytidinic acid (polyC, 100 µg/ml), AMP (50 mM), ADP (50 mM), ATP (50 mM), dG16 (1 µM), dG10T6 (1 µM), dG3T13 (1 µM), lipopolysaccharide (LPS) (150 µg/ml), heparin (100 units/ml), salmon sperm DNA (ssDNA) (300 µg/ml), or fucoidin (100 µg/ml) for 2 hr at 4°. After incubation, cells were washed thoroughly, and the cell-bound radioactivity was determined. Radioligand binding is plotted as percentage of total binding (in the absence of displacer) and is the mean of a duplicate determination (100% binding corresponds with 14.2 ± 1.5 ng/mg for EC and 7.1 ± 1.0 ng/mg for KC).

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Effect of AcLDL () and OxLDL () on 32P-ODN1 binding to EC (top) and KC (bottom). EC or KC (2-3 × 106 cells/ml in DMEM plus 2% BSA) were incubated for 2 hr at 4° with 20 nM 32P-ODN1 and 0-150 µg/ml AcLDL or OxLDL. After incubation, the cells were washed thoroughly, and the cell-bound radioactivity was determined. 32P-ODN1 binding is plotted as percentage of total binding (in the absence of displacer) and is the mean ± standard deviation of six experiments.

The efficacy of two ODNs (ODN1 and ODN3), the polyanion polyinosinic acid, and the mononucleotide ADP to displace ³²P-ODN1 binding to EC and KC was studied in more detail using competition binding studies (Fig. 7). All compounds fully inhibited specific ³²P-ODN1 binding in a competitive fashion with inhibition constants of 74 ± 23 nM (unlabeled ODN1), 10 ± 3 nM (ODN3), 0.052 ± 0.008 nM (polyinosinic acid), and 31 ± 7 mM (ADP) for EC and 68 ± 16 nM (unlabeled ODN1), 38 ± 6 nM (ODN3), 0.036 ± 0.004 nM (polyinosinic acid), and 27.1 ± 4.4 mM (ADP) for KC.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Competition curves of 32P-ODN1 binding to EC (closed symbols) and KC (open symbols). Liver EC and KC were incubated for 2 hr at 4° with 20 nM 32P-ODN1 and polyinosinic acid (,  ), unlabeled ODN1 (, ), ODN3 (, ), or ADP (, ). The 100% specific binding is defined as the difference of 32P-ODN1 binding in the presence and absence of 100 µg/ml polyinosinic acid and is determined in duplicate.

To verify whether the ³²P-ODN1 binding site or sites also mediate ligand internalization, we determined uptake at 37° (i.e., total cell-associated radioactivity after removal of the membrane-bound ³²P-ODN1 by treatment with 100 µg/ml polyinosinic acid). Uptake proceeded linearly in time for 10-15 min and leveled off after 60 min of incubation (data not shown). Hence, a 10-min incubation time was selected for uptake studies (unless otherwise stated). The stability of ³²P-ODN1 in the presence of EC and KC at 37° was sufficient to allow a 10-min incubation with both cell types (half-lives of EC and KC, 48 ± 7 and 55 ± 7 min, respectively). Uptake of ³²P-ODN1 followed monophasic Michaelis-Menten kinetics with a K_m value of 270 ± 21 and 106 ± 64 nM for EC and KC, respectively, and was inhibited for 80-90% in the presence of 100 µg/ml polyinosinic acid. Apparently, ³²P-ODN1 is internalized efficiently after initial binding. To characterize the pathway of ³²P-ODN1 uptake, we studied the effect of various agents that interfere with lysosomal uptake on uptake of ³²P-ODN1 (Fig. 8A). Sodium azide (84 ± 4% and 77 ± 4%, respectively), sucrose (74 ± 2% and 63 ± 3%), monensin (70 ± 7% and 63 ± 4%), and chloroquine (86 ± 4% and 71 ± 8%) significantly reduced uptake of ³²P-ODN1 by EC and KC. Although these results demonstrate that uptake of ³²P-ODN1 proceeds via an energy-dependent uptake mechanism, they leave unresolved whether ³²P-ODN1 enters the cell intact or as an metabolite formed after extracellular degradation of membrane-bound ³²P-ODN1. The effects of phosphomonoesterase, phosphodiesterase, and phosphatase inhibitors on the rate of uptake were determined to discriminate between these two pathways (Fig. 8B). GMP, an analogue of the 5-terminal nucleoside (thus inhibiting receptor-mediated uptake of exonuclease-digested ³²P-ODN1 metabolites); 4-nitrophenyl phosphate, a phosphomonoesterase inhibitor; L-(+)-tartaric acid, an acid phosphatase inhibitor; levamisole, a alkaline phosphatase inhibitor; and a mixture of these agents had no affect on ³²P-ODN1 uptake. Only 4-nitrophenyl phosphate tended to enhance ³²P-ODN1 uptake by EC, probably by increasing the stability of ³²P-ODN1 in the incubation buffer.

View larger version (41K): [in this window] [in a new window]   Fig. 8.   Top, effect of classic uptake inhibitors on the rate of initial 32P-ODN1 uptake by EC () and KC (). EC or KC (2-3 × 106 cells/ml) were incubated for 30 min at 37° with sodium azide (NaN3, 10 mM), sucrose (250 mM), monensin (10 µM), colchicine (100 µM), or chloroquine (1 mM). Subsequently, the cells were incubated for 10 min at 37° with 20 nM 32P-ODN1 in DMEM plus 2% BSA. After incubation, the cells were put on ice, washed once with DMEM plus 2% BSA, and incubated with polyinosinic acid (100 µg/ml in DMEM plus 2% BSA) to remove membrane-associated radioactivity. Finally, the cells were washed thoroughly, and the cell-associated radioactivity was determined. Uptake is given as percentage of initial uptake in the absence of displacer. Bottom, effect of enzyme inhibitors on the rate of initial 32P-ODN1 uptake by EC () and KC (). EC or KC (2-3 × 106 cells/ml) were incubated in DMEM (supplemented with 2% BSA) for 10 min at 37° with 20 nM 32P-ODN1 in the absence or presence of GMP (10 mM), 4 nitrophenol phosphate (4-NPP, 10 mM), tartaric acid (1 mM), levamisol (10 µM), or a mixture of GMP, 4-NPP, tartaric acid, and levamisol (mix). After incubation, the cells were put on ice, washed once with DMEM plus 2% BSA, and incubated with polyinosinic acid (100 µg/ml in DMEM plus 2% BSA) to remove membrane-associated radioactivity. Finally, the cells were washed thoroughly, and the cell-associated radioactivity was determined. Uptake is given as percentage of initial uptake in the absence of displacer.

Finally, we analyzed the cell lysates of EC and KC, which had have been incubated for 0-30 min at 37° with ³²P-ODN1, on PAGE for the presence of intact ³²P-ODN1 (Fig. 9). In both cell types, intact ³²P-ODN1 could be detected in the lysosomal compartment, and the intracellular amount of intact ³²P-ODN1 in both cell types increased in time after a 2-min lag phase, reaching a maximum at 20 min of incubation at 37°. Degradation products of ³²P-ODN1 were not observed during this time course.

View larger version (90K): [in this window] [in a new window]   Fig. 9.   PAGE analysis of the kinetics of 32P-ODN1 internalization by EC and KC. EC or KC (2-3 × 106 cells/ml) were incubated in DMEM (supplemented with 2% BSA) with 20 nM 32P-ODN1 for 0, 1, 2, 5 10, 20, or 30 min at 37°. After incubation, the cells were put on ice, washed once with ice-cold DMEM (1 ml; supplemented with 2% BSA), and incubated for 5 min at 4° with polyinosinic acid (100 µg/ml in DMEM plus 2% BSA) to remove membrane-bound ODN1. Subsequently, the cells were washed thoroughly, heated for 5 min at 96°, and sonicated, after which 32P-ODN1 was isolated from the cell lysate by phenol/chloroform extraction. The isolated 32P-ODN1 was subjected to gel electrophoresis on 19% polyacrylamide gel under denatured conditions (80 mM Tris/90 mM boric acid/2 mM EDTA/7 M ureum), and the gel was autoradiographed using the PhosphorImager.

	Discussion

Top Summary Introduction Procedures Results Discussion References

A number of studies have shown that phosphodiester ODNs injected intravenously into rats are eliminated from the circulation partly as a result of rapid redistribution over the body fluid and partly as a result of renal excretion and extensive liver uptake (Emlen et al., 1988; De Smidt et al., 1991 ; Zendegui et al., 1992; Inagaki et al., 1994; Sands et al., 1994; Rifai et al., 1996; Sawai et al., 1996). A quantitative analysis of the cell types responsible for this liver uptake and a study of the actual uptake mechanism are still lacking. This prompted us to analyze the intrahepatic cellular distribution of ³²P-ODNs and characterize the recognition site or sites responsible for this uptake. In agreement with previous studies (De Smidt et al., 1991; Zendegui et al., 1992; Inagaki et al., 1994; Sands et al., 1994; Rifai et al., 1996), all of the tested phosphodiester ODNs were cleared rapidly from the bloodstream. Liver uptake of the miscellaneous ODNs ranged from 22% to 36%, whereas for GT-containing model ODNs, uptake ranged from 15% to 31%. Liver uptake was maximal for ODNs with an intermediate G content and markedly reduced for G-rich ODNs (>50-60%). The reduced uptake was compensated for by an enhanced bone marrow disposition, possibly by the dextran-sensitive binding site for ODNs on pro-B/pre-B cells (Zhao et al., 1994).

Tissue distributions of the ³²P-ODNs were similar and indicative of redistribution of the ODNs and/or their metabolites over the total body fluid. Apart from the nonparenchymal liver cells, specific disposition of radioactive ODNs was observed only in bone marrow (for G-rich GT-containing ODNs) and in the intestine, the latter probably reflecting hepatic degradation of ³²P-ODNs and subsequent biliary secretion. Hepatic uptake of ODNs could be ascribed mainly to nonparenchymal liver cells, which concurs with the qualitative findings of Emlen et al. (1988), Rifai et al. (1996), and Inagaki et al. (1994) that, in rats, both DNA and phosphorothioates are taken up rapidly by adherent cells in the liver. No correlation was found between ODN sequence and the ratio of KC to EC uptake.

These results are surprising in view of the fact that G-rich ODNs display a higher affinity for scavenger receptor class A-I/II (Pearson et al., 1993), which is localized mainly on liver EC (Van Berkel et al., 1991: Nagelkerke et al., 1983). It therefore was anticipated that G-rich ODNs would be eliminated more efficiently from the bloodstream by hepatic EC. Additional receptor systems (Suzuki et al., 1997) on EC, KC, and bone marrow cells, with similar substrate recognition profiles, also probably contribute to ODN clearance.

The mechanism of the rapid and substantial hepatic uptake of ODNs in vivo was addressed further in vitro by studying ³²P-ODN1 binding to isolated EC and KC. ³²P-ODN1 binding to both cell types seemed to be saturable and of moderately high affinity, involves a protein receptor, and is Ca²⁺ independent. The latter makes it unlikely that the ³²P-ODN1 binding site is identical to the DNA receptor on leukocytes, which was reported to be Ca²⁺ dependent (Bennett et al., 1985). Likewise, the recently identified ODN receptor MAC-1 (CD11b/CD18; Benimetskaya et al., 1997) differs from the hepatic ³²P-ODN binding site on EC and KC in that ODN binding to MAC-1 is Ca²⁺ dependent.

To characterize the ³²P-ODN1 recognition site or sites on nonparenchymal liver cells in more detail, the binding profile was monitored using displacement studies. Of the tested polyanions, only unlabeled ODN1, polyinosinic acid, polyguanosinic acid, and fucoidin inhibited ³²P-ODN1 binding, whereas structural analogues like polyadenosinic and polycytidinic acid were not capable of displacing ³²P-ODN1 binding. In concert with the data of Pearson et al. (1993), the ability of 16-mers to displace ³²P-ODN binding increased with increasing G content. Of the other anions, only nucleotide triphosphates, lipopolysaccharide (Hampton et al., 1991), and fucoidin inhibited ODN binding. This illustrates that recognition of polyanions is a highly specific process.

Three main conclusions can be drawn from the competition studies. First, the binding profiles of EC and KC are essentially equal, making it likely that the ³²P-ODN1 recognition sites on both cell types are identical. Second, the binding characteristics of ³²P-ODN1 binding to EC and KC do not concur with that of p80, which is involved in ODN binding to HL-60 cells, or with the ODN receptor described by Yabukov et al. (1989). Binding of ³²P-ODN to HL-60 cells seemed to be inhibitable by nucleotide monophosphates and tRNA (Loke et al., 1989), whereas we did not observe any effect of these inhibitors. Moreover, double-stranded DNA, which is an efficient inhibitor of ODN binding to the receptor of Yabukov et al. on fibroblasts, is unable to displace ³²P-ODN1 binding to nonparenchymal liver cells. Third, the characteristics of ³²P-ODN1 binding to EC and KC closely parallel that of ¹²⁵I-AcLDL binding to these cell types. Recent studies have shown that a number of scavenger receptors are able to take up ¹²⁵I-AcLDL, including scavenger receptor class A-I/II (Horiuchi et al., 1985; Ashkenas et al., 1993; Doi et al., 1993; Pearson et al., 1993; De Rijke et al., 1994), CD36 (Endeman et al., 1993), the Fc/RII receptor (Stanton et al., 1992), and other still-unidentified anion receptors (Suzuki et al., 1997). In fact, Suzuki et al. (1997) demonstrated that the in vivo fate of AcLDL in scavenger receptor AI/II-deficient mice was equal to that in control mice. This suggests that in addition to scavenger receptor class AI/II, alternative receptors exist in the liver. Of the potential candidate scavenger receptors for mediating ODN uptake by nonparenchymal liver cells, CD36 can be excluded because ligand binding to CD36 is not inhibitable by polyinosinic acid (Endeman et al., 1993). The in vitro data presented in this study confirm the in vivo finding that multiple receptors may contribute in ODN clearance. Excess AcLDL and OxLDL (being good substrates for scavenger receptor AI/II and CD36) only partly inhibit ³²P-ODN1 binding to EC and KC, whereas polyinosinic acid completely inhibits binding. In addition, the ODN binding capacities of EC and KC (i.e., 18.8 and 9.0 pmol/mg of cell protein, respectively) greatly exceed that of ¹²⁵I-AcLDL binding to both cell types (271 ± 33 and 58 ± 19 fmol/mg of cell protein, respectively; three experiments; data not shown).

Uptake studies at 37° demonstrated that ³²P-ODN1 binding to EC and KC is followed by internalization. At least 54% (KC; turnover rate, 18 min) and 89% (EC; turnover rate, 11 min) of the membrane-bound ODN was internalized within 10 min at 37°, which is slightly slower than that of ¹²⁵I-AcLDL by both cell types (Van Berkel et al., 1981; Nagelkerke et al., 1983). Internalization (which includes both uptake and initial degradation) was markedly impaired after energy depletion of EC and KC (sodium azide), inhibition of membrane clustering (sucrose), or inhibition of lysosomal acidification (chloroquine and monensin). This is in agreement with previous studies of Van Berkel et al. (1981) and Nagelkerke et al. (1983) showing that AcLDL degradation (±90%) and, to a lesser extent, uptake (±40%) is inhibited in the presence of lysosomotropic agents such as chloroquine. Apparently, the segregation of scavenger receptors and associated substrates is slightly affected in the presence of these agents. Neither phosphodiesterase, phosphomonoesterase, and phosphatase inhibitors nor a mixture of these agents significantly affected ODN uptake, suggesting that intact ³²P-ODN1 is internalized. This is confirmed by autoradiographic analysis of EC and KC cell lysates, demonstrating, inconceivably, the presence of intact ³²P-ODN1 at 2 min after onset of cell uptake. The amount of internalized ³²P-ODN1 increased, reaching a maximum at 10-20 min of uptake.

In conclusion, liver KC and EC possess specific binding sites for ODNs. The inhibition profiles of the ³²P-ODN1 binding site on both cell types are essentially equal and correspond closely to that of ¹²⁵I-AcLDL binding receptors. Multiple scavenger receptors on nonparenchymal liver cells and bone marrow cells may participate in the elimination of phosphodiester ODNs from the bloodstream. These receptors are involved in the clearance of small DNA fragments from the circulation and supplement the DNA receptor described by Emlen et al. (1988), which is implicated in the removal and degradation of large and double-stranded DNA fragments. Minor changes in ODN sequence markedly affect ODN recognition by these scavenger receptors and thus its in vivo fate; this further emphasizes that not only the intrinsic in vitro activity of an antisense ODN but also the biodistribution profile are crucial denominators of in vivo activity of a specific ODN. Through careful design of the antisense sequence, both nonparenchymal liver cell and bone marrow uptake can be modulated.