Abstract
Phosphorothioate antisense oligodeoxynucleotides are novel therapeutic agents designed to selectively and specifically inhibit production of various disease-related gene products. In vivo pharmacokinetic experiments indicate that these molecules are widely distributed in many species, with the majority of oligomers accumulating within liver and kidney. To better understand the metabolism of these agents, we studied the stability of several phosphorothioate oligodeoxynucleotides, their congeners, and second generation oligomer chemistries in rat liver homogenates. To examine metabolism, background nuclease activity was characterized in whole liver homogenates by using ISIS 1049, a 21-mer phosphodiester oligodeoxynucleotide. Nuclease activity could readily be detected in liver homogenates. Under optimized conditions, the predominant enzymatic activity was 3′-exonucleolytic and could be influenced by pH and ionic conditions. However, in addition to 3′ exonucleases, 5′ exo- and endonuclease activities were also observed. Our data indicate that metabolism of phosphorothioate oligodeoxynucleotides was more complex than that of phosphodiesters for many reasons, including phosphorothioate oligodeoxynucleotide inhibition of nucleases and the presence ofRp and Spstereoisomers. The rate of phosphorothioate metabolism also appeared to be influenced by sequence, with pyrimidine-rich compounds being metabolized to a greater extent than purine-rich oligomers. Other factors affecting stability included oligomer chemistry and length. Concomitant experiments performed in rats dosed systemically with the same compounds mimic the activities seen in vitro and suggest that this liver homogenate system is a valuable model with which to study the mechanism of metabolism of antisense oligonucleotides.
Antisense oligonucleotides are new therapeutic agents designed to specifically and selectively inhibit production of disease-related proteins (Crooke ST, 1996, 1998; Crooke and Bennett, 1996). Phosphorothioate oligodeoxynucleotides have proved to be the most promising of the first generation compounds, with efficacy being demonstrated at reasonable concentrations with little or no toxicity (Bennett, 1993; Crooke ST, 1997). The pharmacokinetic properties of these compounds have been well characterized. Data from a variety of studies in mouse, rat, and monkey have shown that phosphorothioate oligodeoxynucleotides are widely distributed in animals after systemic administration, with spleen, lymph nodes, bone marrow, kidney, and liver representing the principal sites of distribution (Geary et al., 1997b; Nicklin et al., 1998). More recent studies have begun to address their suborgan distribution in kidney and liver by using a variety of localization techniques, including fluorescent labeling, autoradiography, and immunostaining (Plenat et al., 1995; Rifai et al., 1996; Butler et al., 1997). Our laboratory, using subcellular fractionation and capillary gel electrophoresis (CGE) analytical techniques, recently reported the cellular and subcellular distribution of a 21-mer phosphorothioate oligodeoxynucleotide, ISIS 1082, in rat liver (Graham et al., 1998).
Information concerning the stability of phosphorothioate oligodeoxynucleotides in vivo has been provided by some of the pharmacokinetic experiments described above (Graham et al., 1998;Nicklin et al., 1998). Although detailed analytical profiles indicated that these compounds were metabolized primarily by exonucleases present in plasma and many tissues, mechanistic studies in liver and kidney and within suborgan and subcellular compartments are obviously problematic in vivo. As a result, the kinetics of phosphorothioate degradation and comparative studies examining metabolism of new oligonucleotide chemistries have been studied traditionally in biological fluids in vitro, including tissue culture medium, serum, urine, cerebrospinal fluid (Crooke RM, 1998), human and animal plasma (Gilar et al., 1997;Koziolkiewicz et al., 1997), solutions containing purified bacterial enzymes (Koziolkiewicz and Stec, 1992), nuclease S1 and snake venom phosphodiesterase (Crooke RM, 1998), and, finally, cellular extracts derived from a variety of tissue culture cells (Crooke RM et al., 1995). Although those experimental systems provided data showing that phosphorothioates were more resistant to nucleolytic degradation than their phosphodiester congeners, the pharmacokinetic behavior of these same compounds in plasma and organs, and on the suborgan level could not be predicted with accuracy.
To better understand the metabolism of phosphorothioate antisense oligodeoxynucleotides in vivo, we developed a system consisting of homogenates derived from whole rat liver and purified parenchymal and nonparenchymal cells that allowed us to characterize the kinetics of oligonucleotide metabolism. More importantly, we were able to investigate potential processes that might influence phosphorothioate degradation. This model had several obvious advantages over the systems described above, including the fact that metabolism was studied by using endogenous nucleases actually present in rat liver rather than unrelated bacterial, fungal, or tissue culture-derived enzymes. Concomitant experiments performed in rats treated systemically with the same compounds (Graham et al., 1998) suggest that this liver homogenate system is a valuable model with which to study the mechanism of metabolism of antisense oligonucleotides.
Materials and Methods
Oligonucleotide Synthesis.
Three phosphodiesters, ISIS 1049 (5′-GCCGAGGTCCATGTCGTACGC-3′), ISIS 8651 (5′-TCGATCCCCCAGGCCACCAT-3′), and ISIS 20548, a 35-mer phosphodiester (5′-GCCGAGGTCCATGTGCCGAGGTCCATGTCGTACGC-3′); several phosphorothioates; ISIS 1082 and shortmers, ISIS 1939, ISIS 3067, ISIS 3082, ISIS 3521, ISIS 5132, and ISIS 4189 (see Table1) were synthesized at Isis Pharmaceuticals (Carlsbad, CA) on a Milligen 8800 DNA synthesizer (Millipore Corp., Bedford, MA) by the phosphoramidite method. The phosphorothioate oligodeoxynucleotide thiation reagent was synthesized as previously described (Graham et al., 1998). Two gapmer chimeric oligonucleotides were also used in our studies, ISIS 13543 (5′-C*A*G*C*C*A*TGGTTGGG C*C*C*A*A*C-3′) and ISIS 13771 (5′-CAGCCATGGTTCCCCC*C*C*A* A*C*-3′), where the asterisk (*) indicates 2′-O-propyl modifications and the underlined bases represent phosphodiesters. The parent compound for the chimeric oligonucleotides was ISIS 4189, a deoxyphosphorothioate (5′-CAGCCATGGTTCCCCCCAAC-3′). All oligomers were reversed phase HPLC-purified and were shown to be greater than 85 to 90% full-length by reversed phase HPLC and CGE.
The poly(T) oligonucleotides, ISIS 17242 (5′-TTTTTTTTTTTTTTTTTTTTSp T-3′) and ISIS 17243 (5′-TTTTTTTTTTTTTTTTTTTTRpT-3′) carrying 3′ chirally pure phosphorothioate linkages, as well as ISIS 17239, the parent poly(T) phosphorothioate (5′-TTTTTTTTTTTTTTTTTTTTSp/Rp T-3′) with the thymidine containing the Sp andRp diastereoisomer mix, were also synthesized at Isis Pharmaceuticals. The nucleotide dimers were synthesized by using phosphite triester chemistry and 2-(diphenylmethylsilyl)ethyl (DEPSE) phosphate-protecting groups in solution (Krotz et al., 1996). DEPSE-protected phosphoroamidites were coupled with 3′-O-methoxyacetylthymidine in the presence of 1H-tetrazole in anhydrous acetonitrile, followed by oxidation with a 5% solution of elemental sulfur in CS2/pyridine/Et3N (14:14:1, v/v/v), affording products with 90% yield. The 3′-OH protecting groups were selectively removed from the fully protected dimers by using a 30-min treatment with a solution of NH4OH in EtOH (1:1, v/v). TheSp andRp diastereoisomers were separated and purified by using silica gel flush chromatography with a stepwise gradient of ethyl acetate in hexane (60–100%). Chirally pure dimers were loaded on long-chain alkyl amino-controlled pore glass solid support according to the protocol of Alul et al. (1991), yielding controlled pore glass with typical loads ranging from 30 to 40 μmol/g.
Phosphorothioate oligonucleotides containing theSp andRp diastereoisomers were synthesized on a 1-μmol scale on a model 398B ABI synthesizer (Applied Biosystems, Foster City, CA) by using the standard phosphoramidite method. Phosphorothioate linkages were introduced by sulfurization with a 0.2 M solution of 3H-1,2-benzodithiol-3-one 1,1-dioxide in CH3CN for 120 s. Oligonucleotide products were cleaved from the support and deprotected by treatment with concentrated NH4OH at 55°C for 16 h and evaporated to dryness. The products were resuspended in water and purified by polyacrylamide gel electrophoresis (20% gel, 8 M urea) and desalted by gel filtration using 30 × 1 cm Sephadex G-25 columns. The structural identity and purity of the oligomers was confirmed by CGE, reversed phase HPLC, and electrospray mass spectrometry (ES/MS).
Chemicals.
All chemicals were reagent grade and were obtained from various sources as described below in greater detail. UltraPure Tris-HCl (catalog no. 15506-017), and 1× PBS without calcium or magnesium (catalog no. 14190-144) were obtained from Life Technologies, Inc. (Gaithersburg, MD). Magnesium acetate tetrahydrate (catalog no. M9147) was obtained from Sigma Chemical Co. (St. Louis, MO). Bio-Rad protein assay reagent (catalog no. 500-0006) was obtained from Bio-Rad Laboratories (Hercules, CA).
Animals.
Male Sprague-Dawley rats (200–300 g) were obtained from Harlan Sprague-Dawley (Madison, WI). The animals were housed in polycarbonate cages and had access to rat chow and water ad libitum in compliance with Institutional Animal Care and Use Committee guidelines.
Preparation of Whole Rat Liver.
Whole rat livers were perfused by using modifications of a procedure described previously (Green et al., 1983). Briefly, rats were anesthetized with a 50-mg/kg i.p. injection of sodium pentobarbital. Whole livers were then perfused with 500 ml of ice-cold 1× PBS without calcium or magnesium via the hepatic portal vein with a flow rate of 35 ml/min. After the blood was flushed, the liver was excised and placed in a 50-ml polycarbonate centrifuge tube containing an ice-cold buffer consisting of 100 mM Tris-HCl and 1 mM magnesium acetate, pH 8.0 (nuclease buffer). The liver was then transferred to a large plastic weigh boat on ice with 5 ml of fresh nuclease buffer and minced into at least thirty 1- to 2-mm pieces. These smaller pieces were transferred into 2-ml Fastprep tubes (Bio 101, Inc., Vista, CA) containing 1 ml of cold nuclease buffer and 100 μl of Matrix Green lysing beads (Bio 101, Inc.) and homogenized with a Bio 101 Fastprep Savant Tissue/Cell disruptor for 8 s at an energy setting of 4.5. After homogenization, the tubes were placed on ice, the homogenates were pooled, and the protein concentration was determined by using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA) based on the method of Bradford (1976). The homogenates were diluted to varying protein concentrations and distributed to 2.0-ml microfuge tubes to perform nuclease/stability assays as described below.
Isolation of Parenchymal and Nonparenchymal Cells: Liver Perfusion.
To isolate purified parenchymal and nonparenchymal cell types, the liver was perfused with collagenase as described previously (Graham et al., 1998). After collagenase treatment, the liver was removed and placed in 100 ml of ice-cold 1× PBS. After gentle mincing, the suspension was poured through sterile 150-μm nylon mesh (Tetko, Buffalo, NY).
Purification of Parenchymal and Nonparenchymal Cells.
Hepatocytes, Kupffer cells, and endothelial cell types were isolated from whole liver as described previously (Graham et al., 1998). Hepatocytes were isolated from the liver perfusate described above by centrifugation at 50g for 5 min in a Beckman tabletop centrifuge (Fullerton, CA). The supernatant containing nonparenchymal cells was removed and placed on ice for additional purification steps. Centrifugation was repeated three additional times to remove contaminating cells. After the final centrifugation, hepatocytes were resuspended in 1× PBS and counted in the presence of trypan blue. Viability of hepatocytes was generally greater than 90% after this procedure.
The initial supernatant, containing nonparenchymal cells, was separated into endothelial and Kupffer cell populations by brief adherence to plastic at 37°C followed by trypsinization and cell scraping to take advantage of the differential adherence characteristics of the two cell types (Garvey and Caperna, 1981). By isolating endothelial cells with trypsin, washing, and then removing Kupffer cells by scraping, a rapid and effective nonparenchymal cell type enrichment was accomplished.
The identity and purity of liver cell types was validated by flow cytometry and double-label immunohistochemistry as described previously (Graham et al., 1998). Briefly, endothelial cells were identified with monoclonal antibodies to rat platelet endothelial cell adhesion molecule-1 and detected with horseradish peroxidase-conjugated donkey anti-mouse IgG F(ab′)2. Hepatocyte preparations were stained with rabbit antibodies to mouse albumin and detected with alkaline phosphatase-conjugated donkey anti-rabbit IgG F(ab′)2. Kupffer cells were stained with mouse anti-rat mononuclear phagocyte antibodies and were detected with either horseradish peroxidase- or alkaline phosphatase-conjugated donkey anti-mouse IgG F(ab′)2.
Once isolated, the cells were resuspended in varying amounts of ice-cold nuclease buffer and transferred to 2-ml Fastprep tubes; homogenates were prepared as described above with the Bio 101 Fastprep Savant Tissue homogenizer. Protein concentration was determined by using the Bio-Rad assay.
Metabolism of Oligonucleotides in Liver and Purified Liver Cell Homogenates: Nuclease Assays.
The metabolism of phosphodiester and phosphorothioate oligodeoxynucleotides was studied in whole liver or purified liver cell populations by incubating varying amounts of liver protein with 0.1 to 50 μM oligonucleotide in nuclease buffer. Briefly, 450 μl of liver homogenate was added to sterile 2.0-ml microfuge tubes and placed on ice. Oligonucleotides, previously prepared as 10× stocks in nuclease buffer, were added to tubes and reactions were initiated by placing tubes in a 37°C New Brunswick gyrotory water bath shaker (New Brunswick Scientific Company, Inc., Edison, NJ). At the end of various incubation times, reactions were terminated by placing tubes on ice and adding 62.5 μl proteinase K for a final concentration of 100 μg/ml (Boehringer Mannheim, Mannheim, Germany) and 100 μl of a 5× stop buffer solution containing 5% Nonidet P-40 (Calbiochem-Novabiochem Corp., La Jolla, CA), 1.0 M NaCl, 200 mM EDTA, and 200 mM Tris, pH 8.0.
Cellular Digestion and Organic Extraction for CGE Analysis.
Whole cells were digested with proteinase K extraction solution as described previously (Graham et al., 1998). To each homogenate containing proteinase K and stop buffer, 30 pmol of an internal standard (homopolymer T 27-mer phosphorothioate oligodeoxynucleotide) was added before enzymatic digestion to permit accurate quantitation of phosphodiester and phosphorothioate oligodeoxynucleotides. Samples were then incubated for 2 h at 55°C to digest proteins. After digestion, 200 μl of 30% ammonium hydroxide was added to each sample before organic extraction with 1 ml of phenol/isoamyl alcohol/chloroform (24:1:24), as described previously (Cossum et al., 1993).
Solid Phase Extraction (SPE).
To purify samples sufficiently for CGE, two SPE columns were required. Removal of residual contaminants was accomplished by using a strong anion exchange SPE column (J & W Scientific, Folsom, CA) followed by desalting with a reversed phase SPE column [Isolute C18(EC), International Sorbent Technology, Mid Glamorgan, UK]. As a final step before CGE analysis, samples were placed on 0.025-μm dialysis membranes (Millipore, Bedford, MA) and floated over 60-mm culture dishes containing 10 ml of 18.3 MΩ/cm dH2O for 30 min before analysis to further reduce the amount of competitive anions that would be loaded during electrokinetic injections.
CGE Sample Analysis.
Samples were placed into microvials and analyzed with a Beckman PA/CE System Gold 5010 capillary electrophoresis system with UV detection at A260nm. Samples were resolved by using a 100-μM i.d. capillary column (Polymicro Technologies Inc., Phoenix, AZ) filled with 11% polymerized acrylamide (Fluka, Neu-Ulm, Switzerland). The electrophoretic buffer used in the capillary and running buffer contained 200 mM bis-(2-hydroxyethyl)imino-tris-(hydroxymethyl)methane (Sigma Chemical Co.), 200 mM boric acid (Fluka), and 8.3 M urea (Boehringer Mannheim). Samples were electrokinetically applied by using 5 to 10 kV for 5 to 10 s; separations were achieved by operating at 20 kV constant voltage for approximately 5 min at 50°C. Samples were injected and quantified within the linear range of the detector, which spanned approximately 0.01 to 0.001 measured absorbance units. Oligonucleotide was quantitated by comparing sample peak areas relative to the T 27-mer internal standard as described previously (Graham et al., 1998).
Measurement of Oligonucleotide Oxidation.
Varying concentrations of ISIS 1082 were incubated with whole liver homogenates as described above. To correlate oligonucleotide metabolism with rates of oxidation, samples were prepared for CGE analysis as described above and additionally desalted before mass spectrometric analysis by reversed phase HPLC with a Poros R2/H 2.1 × 30 mm column (Perseptive Biosystems, Framingham, MA). Buffer A consisted of 5 mM tripropylamine (Aldrich, Milwaukee, WI) in H2O, and buffer B consisted of 20 mM tripropylamine in acetonitrile. The tripropylamine was redistilled before use. A linear gradient of 10 to 40% B was maintained by using two Micro-Tech Scientific Ultra Plus micropump modules (Micro-Tech Scientific, Sunnyvale, CA) at a flow rate of 50 μl/min. Mass spectra were acquired with a Finnegan LCQ quadrupole ion trap mass spectrometer (ThermoQuest Corporation, San Jose, CA) equipped with an ES ionization source. The spray needle voltage was set at −2.8 kV, with the sheath gas and auxiliary gas flows set to 50 and 60 psi, respectively. The automatic gain control was used to fill the trap. A mass range ofm/z 600 to 1200 was scanned, and the entire analyte peak was averaged for phosphorothioate and phosphodiester analysis, which was carried out gravimetrically. The percentage of phosphodiester reported was the average value of two HPLC-ES/MS analyses for the 1.0- and 10.0-μM studies and a single value of one HPLC-ES/MS analysis for the 0.1-μM study. The percentage of phosphodiester was determined by calculating the average percentage of phosphodiester for the −6 to −9 charge states.
Results
Characterization of Nuclease Activity in Liver Homogenate
The nuclease activity in whole rat liver homogenate was characterized by studying the degradation of a model 21-mer phosphodiester antisense oligodeoxynucleotide, ISIS 1049. That oligomer chemistry was chosen because it resembles endogenous nucleic acids, and as such, would serve as a natural substrate for nucleases found within the liver. As with any other purified enzymatic system, we were able to show that various factors, including the amount of liver protein (enzyme), oligonucleotide concentration (substrate), pH, and ionic milieu, influenced the kinetics of nuclease activity in whole rat liver homogenate (data not shown). After optimizing the system, final assay conditions were chosen to be 1 μM oligonucleotide, which is within the concentration range of drug previously calculated to be present in rodent and monkey liver from various in vivo experiments (Cossum et al., 1993; Leeds et al., 1997), a simple Tris-HCl buffer containing 1 mM magnesium acetate, pH 8.0, and 25 to 50 μg liver protein per reaction tube.
As stated above, we wished to develop a model system that used whole liver homogenates to determine the mechanisms of oligonucleotide metabolism in that organ. However, the liver is composed of connective tissue and three major cell types. Hepatocytes, or parenchymal cells, constitute approximately 80% of all liver cells. Nonparenchymal cells, which consist of endothelial and Kupffer cells, constitute the remaining 20% of liver cells. Figure1 shows that the degradation of ISIS 1049 in whole liver homogenates was similar to that seen with purified hepatocyte homogenates, the time to degrade compound by 50% (t50%) for whole liver and hepatocyte homogenates being 15 min versus 13 min, respectively. This minor difference in activity can be attributed to the presence of noncellular-associated proteins (connective tissue) in the whole liver homogenate. Although hepatocyte and Kupffer cell nuclease activities were very similar, endothelial cell nuclease activity was slightly greater under identical assay conditions.
The durability of the enzymatic activity in liver homogenate was investigated by comparing nuclease activity of freshly prepared homogenate with that of homogenate incubated in a 37°C water bath for 8 h. The degradation of ISIS 1049 under both conditions was essentially identical, indicating no loss of enzymatic activity (data not shown). Additional experiments showing activity of a carboxyesterase that removes S-acyl-2-thioethyl groups from poly(T) oligomers also suggested that the homogenate remained enzymatically active over an 8-h period (data not shown).
Metabolism of ISIS Oligonucleotides in Liver Homogenate
Phosphodiesters.
By using optimized assay conditions, the metabolism of ISIS 1049 in whole liver homogenates was studied in greater detail. A typical electropherogram generated from these assays is shown in Fig. 2. The first peak represents full-length ISIS 1049, whereas the metabolites (processiven-1 shortmers) are represented by the peaks that migrate earlier. This pattern of degradation, i.e., the processive laddering of smaller-length products, indicated that the principal enzymatic activity under these conditions was exonucleolytic.
The graphical representation of the kinetics of degradation of ISIS 1049 is shown in Fig. 3. The compound was metabolized rapidly by liver nucleases, with full-length product being degraded by 50% in approximately 12.5 min. By 60 min, only 20% of the full-length product remained. Concomitant with the loss of full-length material, shortmers were processively generated and subjected to additional degradation. By 30 min, a greater percentage of then-1 metabolite existed than the parent compound. No meaningful differences were observed in rate or pattern of cleavage when phosphodiester oligonucleotides of similar length but consisting of different sequences (ISIS 1049 versus ISIS 8651) were compared (data not shown).
Phosphorothioate Oligodeoxynucleotides.
It has been well documented that phosphorothioate oligodeoxynucleotides are more resistant to nucleolytic degradation than their phosphodiester congeners under a variety of in vitro and in vivo experimental conditions (Crooke RM, 1998). This can be seen in Fig.4, when the stability of ISIS 1082 was measured over an 8-h period. After an initial phase of relatively rapid degradation (100 to ∼68% full-length in 60 min), metabolism of the compound slowed considerably, plateauing between 2 and 8 h. At the end of the incubation period, thet50% for the phosphorothioate was barely attained. Degradation of ISIS 1082, like that of ISIS 1049, appeared to be the result of exonucleolytic activity producing processive n-1 shortmers (electropherogram not shown). However, unlike ISIS 1049, the production of the ISIS 1082 metabolites appeared to plateau after 1 h.
Inhibition of Liver Nucleases by Phosphorothioates
The influence of phosphorothioates on nucleolytic activity in liver was examined by pretreating 50 μg/ml homogenate with ISIS 1082 and then incubating the same samples with ISIS 20548, a 35-mer phosphodiester whose metabolic profile could easily be separated from that of ISIS 1082 by CGE analysis. As shown in Fig.5, 1 μM ISIS 20548 in control homogenates was metabolized rapidly over 60 min, with at50% of approximately 18.5 min. However, preincubation of the liver homogenate with 1 μM ISIS 1082 for 1, 2, or 4 h significantly inhibited degradation of the phosphodiester. This inhibition of nucleolytic activity was not observed when the homogenate was preincubated with 1 μM ISIS 1049 and then challenged with an equimolar amount of the same phosphodiester (data not shown).
Effect of Sp andRp Diastereoisomerism on Nucleases
Routine synthesis of phosphorothioate oligodeoxynucleotides results in a mixture of Sp andRp isomers that display varying sensitivities to a variety of purified nucleases (Spitzer and Eckstein, 1988; Koziolkiewicz et al., 1997). The effect of phosphorothioate chirality was studied in liver homogenate by incubating 21-mer T homopolymers that were modified at the 3′-end with a chirally pureSp (ISIS 17242) orRp (ISIS 17243) nucleotide, or with the racemicSp/RpT (ISIS 17239). The metabolism of ISIS 17239, the racemic phosphorothioate compound, proceeded similarly to that seen with ISIS 1082 (Fig. 6A), with a more rapid initial degradation, gradual slowing over 4 h, and a plateau of then-1 metabolite between 1 and 2 h. The phosphorothioate capped with the pure Sp nucleotide, ISIS 17242, was barely degraded over the experimental time course (Fig.6B), whereas the metabolism of ISIS 17243, the phosphorothioate with the Rp nucleotide on the 3′-end, was much more rapid than the two congeners (Fig. 6C). In fact, the pattern of degradation very much resembled that of a phosphodiester, with at50% being reached at 1.5 h and a greater amount of the n-1 metabolite existing than full-length material by the end of the 4-h incubation period.
Effect of Length on Phosphorothioate Oligodeoxynucleotide Metabolism
As shown in Fig. 7, varying the length of the oligomers (Table 1) altered the rate of degradation by liver nucleases, with ISIS 20425, the 6-mer, displaying the least amount of metabolism. Increasing the size of the oligomer from 6 to 10 nucleotides resulted in a proportional increase in the rate of metabolism. However, for compounds longer than 11 nucleotides, the rate of cleavage did not vary directly with length. The rank order for cleavage was: 12-mer > 18-mer > 15-mer > 10-mer = 21-mer > 9-mer > 8-mer > 7-mer > 6-mer.
Sequence-Dependent Metabolism of Phosphorothioate Oligodeoxynucleotides
Figure 8 compares the metabolism of 1 μM ISIS 1082 with five similar-length phosphorothioate oligodeoxynucleotides (see Table 1). ISIS 1082, ISIS 2302, ISIS 3067, and ISIS 3082 displayed similar rates of metabolism (t50% > 8 h). In contrast, ISIS 3521 and ISIS 1939 were more rapidly degraded (t50%: 4 h and 2 h, respectively). These differences in stability may be explained, in part, by the base composition of the oligonucleotides. ISIS 1939, which was metabolized to the greatest extent, is pyrimidine-rich, its AG/CT ratio being 1:9. However, other factors are obviously involved in metabolism because ISIS 3521, consisting of 60% pyrimidines, was metabolized faster than ISIS 3082, which is even more pyrimidine-rich (65%).
Effect of Chemical Modifications on Oligonucleotide Stability
Phosphodiester and phosphorothioate oligonucleotides have been chemically modified to alter various biochemical and biophysical parameters, including nuclease resistance, lipophilicity, and binding affinity (Cook, 1998). Altering the 2′-position on the B-d-ribofuranosyl moiety of oligomers with various alkyl groups, including 2′-O-propyls, especially in a chimera or “gapmer motif” to support an RNase H mechanism, is a widely used strategy to enhance nuclease resistance both in vitro and in vivo (Cook, 1998).
The metabolism of two alkyl-modified oligonucleotides, ISIS 13543, a gapmer with 2′-O-propyl wings on a phosphorothioate backbone with a 10-base phosphodiester gap, and ISIS 13771, a phosphorothioate hemimer with a 6-base 2′-O-propyl wing on the 3′-end, was compared with their parent phosphorothioate oligodeoxynucleotides, ISIS 4189 and ISIS 1082. The latter two compounds were degraded in the liver homogenate in a manner consistent with that seen by other phosphorothioates (Fig. 9A), and ISIS 4189, which is pyrimidine-rich, was metabolized faster and to a greater extent than ISIS 1082. The degradation profile for ISIS 13543, the chimera with a phosphodiester in the gap and two 2′-O-propyl wings was similar to that observed for ISIS 4189. However, a single 2′-O-propyl wing on the 3′-end of a molecule (ISIS 13771) provided the greatest stability of the congeners tested over the 4-h incubation period.
Examination of the electropherograms generated from this experiment (Fig. 9B) suggested that altering oligonucleotide chemistry can also change the types of nucleolytic activity involved in metabolism. ISIS 4189 (Panel A) illustrates the typical processive pattern of degradation observed with phosphodiesters and phosphorothioate oligodeoxynucleotides. However, by blocking both 5′- and 3′-ends of a molecule with nuclease-resistant 2′-O-propyl modifications, as with ISIS 13543, endonuclease activity can readily be detected as shown by the discontinuous pattern of the metabolites (panel B).
Phosphorothioate Oxidation
HPLC-ES/MS was used to determine whether oxidation at the phosphorothioate linkages might contribute to the metabolism of ISIS 1082 and, potentially, other phosphorothioate oligodeoxynucleotides. As shown in Fig. 10, incubation of 0.1, 1, and 10 μM ISIS 1082 in whole liver homogenate over an 8-h period did not result in a significant change in the phosphodiester content of the compound at any of the oligonucleotide concentrations studied. Analysis of ISIS 1082 extracted from the homogenate at t = 0 indicated that the phosphodiester content ranged from 7.0 to 11.0%. The largest percentage of phosphodiester difference (4.3%) was observed at t = 4 h at 1.0 μM oligomer and att = 2 and 4 h (2.0%) at 0.1 μM compound. Because the estimated error for phosphodiester detection in the hepatic homogenates ranged from 1 to 5%, these variations do not represent a significant change from the t = 0 value. These data suggest that oxidation of phosphorothioate linkages does not contribute to the metabolism of ISIS 1082 in rat liver under these experimental conditions.
Discussion
To characterize metabolism of oligonucleotides in liver, a principal site of phosphorothioate oligodeoxynucleotide accumulation (Geary et al., 1997a,b; Graham et al., 1998), perfused liver was gently homogenized and nuclease assays were performed in the presence of a variety of buffer fluids, including serum-free medium, Williams E medium, a commonly used liver tissue culture medium, and Tris-HCl supplemented with various ions and sucrose (data not shown). Ultimately, Tris-HCl, pH 8.0, with 1 mM magnesium was chosen as our buffer because it had been used previously to examine rat liver nucleolytic enzymes isolated from rat liver nuclear, cytosolic, and microsomal fractions (Kouidou et al., 1987; Malicka-Blaszkiewica, 1990;Vavatsi et al., 1991). By using ISIS 1049, an unmodified oligodeoxynucleotide, which, like natural DNA and RNA, contains phosphodiester linkages, we demonstrated that our liver homogenate system was time, enzyme, and substrate dependent. Our data also suggested that liver enzymatic activities, i.e., the aggregate nucleases and an unrelated carboxyesterase, could be maintained over an 8-h period with these standardized conditions. Our system has also proved to be extremely consistent as there was little variation in homogenates prepared from livers of multiple animals over a 2-year period.
By using optimized conditions consisting of Tris-HCl with 1 mM magnesium, 50 μg of liver protein, and 1 μM drug, which was within the concentration range of oligonucleotide previously detected in vivo (Cossum et al., 1993; Leeds et al., 1997), we examined the stability of ISIS 1082, the phosphorothioate congener of ISIS 1049. As predicted from in vitro systems and the chemical nature of the molecule, ISIS 1082 was significantly more stable (38-fold) than the parent phosphodiester. Nonetheless, the compound still was metabolized by liver nucleases in a time-dependent fashion. In addition to obvious differences in rates of metabolism, the pattern and extent of degradation of ISIS 1082 differed from that of ISIS 1049 in that the loss of full-length ISIS 1082 and production of shortmers was very rapid within the first hour of incubation and then appeared to plateau.
Unlike in vivo pharmacokinetic experiments, the liver homogenate allowed us to explore in greater detail the mechanistic differences in the patterns and extent of degradation between ISIS 1082 and ISIS 1049. Some of the factors that could affect, independently or in aggregate, the metabolism of phosphorothioates include 1) inhibition of liver nucleases by phosphorothioates and their metabolites over time, 2) selectivity of nucleolytic enzymes toward specific phosphorothioate diastereoisomers, 3) competition for nucleolytic degradation of slower-degrading shortmer metabolites with full-length oligomer, and/or 4) sequence specificity of liver nucleases.
Phosphorothioates, because of the negative charge delocalization at the internucleotide thioate bond, are “sticky” and bind nonspecifically to many proteins (Eckstein, 1985; Crooke RM et al., 1995). Over the past 10 years, these compounds have been shown by a number of groups to inhibit a variety of enzymes in vitro, including specific nucleases, HIV reverse transcriptase, human DNA polymerases α and γ, topoisomerase, and human RNases H1 and H2 (Spitzer and Eckstein, 1988;Gao et al., 1992; Crooke RM, 1998). There is additional evidence to suggest that phosphorothioates and their metabolites also inhibit nucleases derived from cells cultured in vitro (Crooke RM et al., 1995) and 3′ exonucleases derived from human plasma (Koziolkiewicz et al., 1997). We have demonstrated (Fig. 5) that the ex vivo preincubation of the rat liver homogenate with ISIS 1082 inhibited metabolism of a phosphodiester oligodeoxynucleotide. The fact that preincubation with ISIS 1049, the phosphodiester congener, did not affect degradation in a similar fashion and that decomposition of liver enzymes did not occur is consistent with enzymatic inhibition by phosphorothioates.
Another factor potentially influencing phosphorothioate metabolism is the chirality of the molecules themselves. Under normal conditions, phosphorothioates are synthesized as a 1:1 mixture ofRp andSp diastereoisomers (Eckstein, 1985;Koziolkiewicz et al., 1997). For years, it has been known that various purified enzymes under in vitro incubation conditions are stereoselective (Potter et al., 1983; Spitzer and Eckstein, 1988). Because purified phosphorothioate-specific nucleases have not been isolated from rat liver, the determination of stereoselectivity of those enzymes is obviously problematic. By using the liver homogenate, however, we could determine the general type or types of nucleases involved in metabolism, whether these activities differed between phosphodiesters and phosphorothioates, and finally, the general effect of chirality on the rates of degradation.
Examination of the metabolic profiles of ISIS 1082 and ISIS 1049 seen on capillary gel electropherograms indicated that both compounds were degraded by exonucleases, resulting in the typical pattern of chain-shortened metabolites or shortmers. A more detailed analysis of the metabolites by MS indicated that degradation occurred primarily through the action of 3′ exonucleases, although a minor 5′-exonucleolytic activity was detected (H. Gaus, personal communication). The question of whether metabolism of both chemistries resulted from identical sets of enzymes is still unknown. Nonetheless, the predominant 3′ exonuclease pattern of degradation we observed was consistent with in vivo data obtained from mouse, rat, monkey, and pig with a variety of phosphorothioate oligodeoxynucleotides (Cummins et al., 1997; Graham et al., 1998; Nicklin et al., 1998).
In addition to 5′ and 3′ exonucleases, endonucleases were also detected in the homogenate as shown by the change in the pattern of metabolism derived from chimeric oligonucleotides modified in the 2′-position of the B-d-ribofuranosyl moiety (Cook, 1998). In these experiments (Fig. 9B), where the 3′-ends or both 3′- and 5′-ends of the oligomers were blocked by nuclease-resistant chemistries, a discontinuous pattern of metabolite production, representative of endonucleolytic activity, was generated. These results are also consistent with data from several groups demonstrating the isolation of single-strand-specific endonucleases from rat liver microsomal and cytosolic fractions (Kouidou et al., 1987; Vavatsi et al., 1991).
Experiments using phosphorothioate T homopolymers 21 nucleotides in length modified at the 3′-ends with pureRp andSp nucleotides highlight the effect of phosphorothioate diastereoisomerism on nucleolytic degradation (Fig.6). We demonstrated that rat liver nucleases preferentially digested phosphorothioates in the Rp andRp/Spracemic configurations, with the Rpisomer being metabolized at a much faster rate than the mixture, whereas the compound modified by the chirally pureSp nucleotide was barely metabolized. The data also emphasize the dominance of the 3′-exonucleolytic activities in rat liver homogenate and suggest that the proper chemical modifications on the 3′-end of oligonucleotides can potentially stabilize compounds in vivo.
Our experiments also demonstrated that the rate and extent of phosphorothioate degradation was affected by the length of oligonucleotide (Fig. 7). The influence of substrate length on nucleolytic degradation was reported over 30 years ago by two groups examining the mechanism of action of deoxyribonuclease I (Potter et al., 1958; Ralph et al., 1962). Both laboratories reported that a minimum requirement for enzymatic activity was a stretch of three internucleotide bonds. Additionally, Ralph et al. (1962) noted that rates of degradation increased with chain length. Although the experiments were performed with a purified endonuclease, those studies and our data strongly suggest that competition of shorter metabolic products with full-length material for nucleases can affect the overall rate of degradation of phosphorothioate oligodeoxynucleotides.
We previously reported minor sequence-specific uptake and stability differences when examining the pharmacokinetics of phosphorothioate oligodeoxynucleotides in tissue culture cells (Crooke RM et al., 1995). Data presented here comparing the stability of six phosphorothioate oligodeoxynucleotides also demonstrate differences in the rate and extent of metabolism as a function of sequence. In general, although the majority of compounds were metabolized to comparable extents at similar rates, ISIS 1939, a pyrimidine-rich oligomer, was clearly more labile than the other compounds, with at50% of 1.5 h. These data are consistent with other studies showing differences in degradation with purified nucleases depending on the purine and pyrimidine content of substrates (Ralph et al., 1962; Koziolkiewicz et al., 1997). Some in vivo metabolism data from our laboratories also corroborate minor sequence-dependent sensitivity to nucleases (J. Leeds, personal communication).
There has been some speculation that phosphorothioates undergo metabolic oxidation, i.e., an exchange of sulfur for oxygen at the phosphorothioate linkages, in vivo in liver and kidney via the cytochrome P-450 system or flavin-containing monooxygenases, which are known to oxidize various organic sulfur and phosphorus compounds (Cohen et al., 1997). Although the bulk of the in vivo data suggest that the principal metabolic pathway for oligonucleotides is via nucleolytic degradation and not a flavin or cytochrome monooxygenase-driven metabolic oxidation process (Geary et al., 1997b; Crooke ST, 1998;Nicklin et al., 1998), some ES/MS data derived from rodent plasma and liver samples have detected minor products consistent with sulfur for oxygen metabolic oxidation (Nicklin et al., 1998). ES/MS analysis of three concentrations of ISIS 1082 in whole liver homogenates over an 8-h period indicated that oxidation did not contribute to metabolism. These data are consistent with additional in vivo pharmacokinetic experiments performed in our laboratories using rats, mice, monkeys, and humans (Geary et al., 1997a,b; Leeds and Geary, 1998). Inconsistencies between our ex vivo and in vivo data and that of other laboratories could be explained by potential differences in processing and handling (e.g., extracting or desalting) of tissue and plasma samples.
In conclusion, we believe that our ex vivo liver homogenate is a reliable system in which to assess and compare the stability of first generation and newer chemically modified antisense compounds. Obviously, such an experimental approach does not replace in vivo pharmacokinetic experiments and can be used only for the extrapolations that have already been validated. Nonetheless, in vivo data generated by multiple groups, showing that the patterns of degradation and the types of nucleolytic activities were similar to those observed in liver homogenates, confirmed the validity of the model. More importantly, the data are more relevant and predictive of stability in animals, unlike those obtained from in vitro experimental systems using enzymes derived from bacteria, fungi, and tissue culture cells (Crooke RM, 1998; Crooke ST, 1998). The reliability and relative simplicity of homogenate preparation highlights the broad applicability of this approach to analyses in multiple organs and across species. Analysis of species- and organ-specific variations in metabolism and the potential mechanisms of these differences will provide valuable information toward the clinical development of more efficacious antisense therapeutics.
Acknowledgments
We thank Drs. Stanley T. Crooke, Janet Leeds, Frank Bennett, and Laurel Bernstein for critical evaluation of the manuscript.
Footnotes
-
Send reprint requests to: Dr. Rosanne M. Crooke, Isis Pharmaceuticals, Carlsbad Research Center, 2292 Faraday Ave., Carlsbad, CA 92008. E-mail:rcrooke{at}isisph.com
- Abbreviations:
- CGE
- capillary gel electrophoresis
- DEPSE
- 2-(diphenylmethylsilyl)ethyl
- ES/MS
- electrospray mass spectrometry
- SPE
- solid phase extraction
- t50%
- time to degrade compound by 50%
- Received May 18, 1999.
- Accepted September 14, 1999.
- The American Society for Pharmacology and Experimental Therapeutics