Intracellular Metabolism of β-l-2′,3′-Dideoxyadenosine: Relevance to Its Limited Antiviral Activity

Abstract

The intracellular metabolism of the β-l- enantiomer of 2′,3′-dideoxyadenosine (β-l-ddA) was investigated in HepG2 cells, human peripheral blood mononuclear cells (PBMC), and primary cultured human hepatocytes in an effort to understand the metabolic basis of its limited activity on the replication of human immunodeficiency virus and hepatitis B virus. Incubation of cells with 10 μM [2′,3′,8-³H]-β-l-ddA resulted in an increased intracellular concentration of β-l-ddA with time, demonstrating that these cells were able to transport β-l-ddA. However, it did not result in the phosphorylation of β-l-ddA to its pharmacologically active 5′-triphosphate (β-l-ddATP). Five other intracellular metabolites were detected and identified as β-l-2′,3′-dideoxyribonolactone, hypoxanthine, inosine, ADP, and ATP, with the last being the predominant metabolite, reaching levels as high as 5.14 ± 0.95, 8.15 ± 2.64, and 15.60 ± 1.74 pmol/10⁶ cells at 8, 4, and 2 h in HepG2 cells, PBMC, and hepatocytes, respectively. In addition, a β-glucuronic derivative of β-l-ddA was detected in cultured hepatocytes, accounting for 12.5% of the total metabolite pool. Coincubation of hepatocytes in primary culture with β-l-ddA in the presence of increasing concentrations of 5′-methylthioadenosine resulted in decreased phosphorolysis of β-l-ddA and formation of associated metabolites. These results indicate that the limited antiviral activity of β-l-ddA is the result of its inadequate phosphorylation to the nucleotide level due to phosphorolysis and catabolism of β-l-ddA by methylthioadenosine phosphorylase (EC 2.4.2.28).

Recent findings have indicated that several β-l-nucleoside analogs exhibit high and potent antiviral activity against both human immunodeficiency virus (HIV) (12) and hepatitis B virus (HBV) (2, 3, 5, 9, 13) replication accompanied by low host cellular toxicity when compared to their respective natural β-d- counterparts (15). These findings prompted a search for novel β-l-nucleoside derivatives which could have synergistic activities and/or do not exhibit cross-resistance with currently available chemotherapeutic nucleoside analogs so that they could be used in combination therapy regimens. We have found that β-l-dideoxyadenosine (ddA)-5′-triphosphate (β-l-ddATP) is a potent inhibitor of both HIV type 1 reverse transcriptase and woodchuck hepatitis virus DNA polymerase (A. Faraj, L. Placidi, C. Perigaud, E. Cretton-Scott, G. Gosselin, L. T. Martin, C. Pierra, R. F. Schinazi, J. L. Imbach, and J. P. Sommadossi, Prog. Abstr. 13th Int. Round Table Nucleosides Nucleotides Biol. Appl., abstr. 135, 1998). However, β-l-ddA per se has a limited anti-HBV activity (50% effective concentration [EC₅₀], 5 to 6 μM) in HBV DNA-transfected human hepatoblastoma-derived HepG2 cells (2.2.15 cells) and no anti-HIV activity (EC₅₀, >100 μM) in peripheral blood mononuclear cells (PBMC) (1, 4, 8, 10). Previous studies showed that β-l-ddA is phosphorylated by 2′-deoxycytidine kinase (EC 2.7.1.74) with a high K_m of 220 μM (6). In addition, β-l-ddA is not a substrate for the catabolic enzyme purine nucleoside phosphorylase (EC 2.4.2.1) (11). The purpose of the present study was to investigate the intracellular metabolism of β-l-ddA in HepG2 cells, PBMC, and primary cultured hepatocytes in order to understand its limited in vitro antiviral activity.

MATERIALS AND METHODS

β-l-ddA was synthesized as previously described (10). [2′,3′,8-³H]β-l-ddA (18.2 mCi/mmol) was obtained by tritium reduction of β-l-2′,3′-didehydro-2′,3′-dideoxyadenosine (Moravek Biochemical). [2′,3′,8-³H]β-l-ddA was prepared by heterogeneous catalytic exchange with tritium gas in the presence of palladium catalyst and was >96% pure as ascertained by the high-performance liquid chromatography (HPLC) method described below. The presence of the tritium in both the base and l-dideoxyribose allowed us to follow the metabolism of this molecule.

Cell culture conditions and preparation of samples.

HepG2 cells were grown in 225-cm² tissue culture flasks in minimal essential medium with nonessential amino acids supplemented with 10% heat-inactivated dialyzed fetal bovine serum (FBS), 1% sodium pyruvate, and 1% penicillin-streptomycin. The medium was changed every 3 days, and the cells were subcultured once a week. After detachment of the adherent monolayer with a 10-min exposure to 30 ml of trypsin-EDTA and three consecutive washes with medium, confluent HepG2 cells (2 × 10⁶/ml) were resuspended in a final volume of 10 ml of medium per time period and exposed to 10 μM [³H]β-l-ddA (1,000 dpm/pmol). The cells were maintained at 37°C under a 5% CO₂ atmosphere for specified time periods. At the selected time points, the cells were centrifuged at 1,200 rpm for 10 min in a GPR centrifuge (Beckman Instruments, Palo Alto, Calif.), washed three times with 10 ml of cold phosphate-buffered saline (PBS), and counted on a hemacytometer. Intracellular β-l-ddA and metabolites were extracted by incubating the cell pellet overnight at −20°C with 1 ml of 60% methanol and were then extracted with an additional 500 μl of cold methanol for 1 h in an ice bath. The extracts were then dried under a gentle filtered air flow and stored at −20°C until they were analyzed by HPLC. The residues were resuspended in 250 μl of water, and 200 μl was injected onto the HPLC system described below.

Human PBMC were obtained from healthy HIV- and HBV-seronegative donors. The cells were then separated by single-step Ficoll-Hypaque discontinuous-gradient centrifugation. The mononuclear-cell layer was collected and washed with cold PBS, and the pellet was resuspended in RPMI 1640 medium supplemented with 10% FBS, 1% sodium pyruvate, 1% nonessential amino acids, and 1% penicillin-streptomycin. The cells were then stimulated with phytohemagglutinin at a final concentration of 5 μg/ml and incubated for 48 h at 37°C under a 5% CO₂ atmosphere. After stimulation, the cells were washed, counted, and resuspended in medium at a density of 20 × 10⁶/time point in 25-cm² tissue culture flasks and exposed to 10 μM [³H]β-l-ddA (specific activity, 1,000 dpm/pmol) for specified time periods. The extraction procedure of β-l-ddA and its intracellular metabolites was similar to that described for the HepG2 cells, and samples were further analyzed by HPLC.

Isolation of human hepatocytes.

Human livers were obtained through the University of Alabama at Birmingham Liver Center. All livers had normal histology and had tested negative for HIV and HBV. In addition, no specific drug history that might have potentially affected the content or function of the enzymes studied was reported for any of the donors. The livers were washed in situ with Eurocollins buffer at 4°C supplemented with heparin to remove blood from the vessels. The liver samples were then perfused with previously oxygenated calcium-free HEPES buffer (2.4 g/liter), pH 7.4, followed by treatment with a 0.05% (wt/vol) collagenase solution containing calcium under recirculation and continuous oxygenation. After 15 to 20 min of perfusion necessary for the disruption of the Glisson's capsule, hepatocytes were suspended in Leibovitz medium (L15) containing 5% fetal calf serum. The freshly isolated cells were then washed three times and centrifuged at 40 g at 4°C for 10 min in L15 supplemented with 10% fetal calf serum to remove debris and damaged cells. After the final wash, the cells were immediately cryopreserved as described below. The number of cells was determined by an erythrosin B exclusion test, and viability was higher than 80%.

Cryopreservation and thawing of human hepatocytes.

Freshly isolated cells were immediately cryopreserved in L15 containing 25 g of bovine serum albumin/liter, 20 g of polyvinylpyrrolidone/liter, 10% dimethyl sulfoxide, and 20% FBS in sterile polypropylene vials. Cell freezing was performed with a Cryomed model 1010 apparatus (Forma Scientific, Marietta, Ohio), which was previously programmed to optimize the temperature drop. The hepatocytes were subsequently stored in liquid nitrogen until use. The cells were thawed by immersing the vials in a 37°C water bath and then purified by Percoll density gradient. The viable hepatocytes were resuspended in William's medium containing 2 mM glutamine and antibiotics. The hepatocytes were then seeded at a density of 4 × 10⁵/ml in 6-well plates previously coated with rat tail collagen and were incubated in a humidified 5% CO₂ atmosphere at 37°C. After 4 h, the medium was replaced by the same medium without FBS and containing 10 μM hydrocortisone hemisuccinate, 10 mM sodium pyruvate, 10 ng of selenium/ml, 4 μg of glucagon/ml, 6.8 μM ethanolamine, and 10 μg of human transferrin/ml. After 14 to 16 h, the medium was renewed and drug metabolic assays were initiated. The hepatocytes were incubated with 10 μM [³H]β-l-ddA (1,000 dpm/pmol) for specified time periods. At the selected time, the medium was removed and the cell layer was washed with cold PBS; then, after cell scraping, β-l-ddA and metabolites were extracted with 60% methanol by incubation overnight at −20°C and then with an additional 500 μl of cold methanol for 1 h in an ice bath. The extracts were dried under a gentle filtered air flow and stored at −20°C until they were analyzed by HPLC. The dried samples were resuspended in 250 μl of water, and 200-μl fractions were injected onto the HPLC system described below.

Coincubation of [³H]β-l-ddA and MTA in hepatocytes in culture.

Hepatocytes were incubated with 10 μM [³H]β-l-ddA (1,000 dpm/pmol) alone (control) or in the presence of 10 or 50 μM 5′-methylthioadenosine (MTA) for 24 h. At the end of the incubation period, the medium was removed and the cell layer was washed with cold PBS; then, after cell scraping, β-l-ddA and metabolites were extracted by 60% ice-cold methanol at −20°C overnight. The extracts were dried and analyzed by HPLC.

HPLC analysis of β-l-ddA and its intracellular metabolites.

Samples were analyzed by reverse-phase HPLC performed with a Hypersil ODS 5-μm column using a model 1090 with automatic injection and a fixed-wavelength spectrophotometer (Hewlett-Packard, Palo Alto, Calif.). The mobile phase consisted of two buffers: buffer A (100 mM triethylamine, pH 7.4) and buffer B (acetonitrile). Elution was performed at a constant flow rate of 1 ml/min using a multistage linear gradient of buffer B from 2 to 3% during the initial 10 min, then increasing from 3 to 6% from 20 to 30 min, to 25% at 40 min, and to 80% at 55 min. Radioactivity was analyzed by use of a 500TR radiomatic FLO-ONE radiochromatography analyzer (Packard Instrument Company, Inc., Meriden, Conn.). Under these conditions, the retention times for pure standards of β-l-ddA, β-l-ddATP, β-l-ddADP, and β-l-ddAMP were about 35, 33, 31, and 27 min, respectively. In addition to these derivatives, five other metabolites labeled A, B, C, D, and E were detected and eluted at 3, 5, 10, 15, and 20 min, respectively. Furthermore a glucuronic derivative of β-l-ddA (β-l-ddA-Glu) was observed only in the primary cultured human hepatocytes and eluted at 33 min.

All metabolites, including β-l-ddA-Glu, were identified by mass spectrometry. The metabolites were isolated from the intracellular medium of HepG2 cells and human hepatocytes by HPLC as described above; fractions were then pooled and lyophilized. The dry residues were then dissolved in 1 ml of water passed through a 0.45-μm-pore-size Acro L13 filter (Gelman Sciences, Ann Arbor, Mich.), and applied to a C₁₈ Sep-Pak cartridge (Waters, Milford, Mass.) that had been preconditioned with 1 ml of acetonitrile and 1 ml of water. After sample loading, the cartridge was washed with 1 ml of water and increasing percentages of acetonitrile from 2 to 100%. After elution, 50-μl samples were chromatographed to ensure purity and the remaining eluents were lyophilized and analyzed by mass spectrometry. In addition, the identification of purified β-l-ddA-Glu was verified by treatment with 20,000 U of β-glucuronidase in 0.1 M Tris HCl and further HPLC analysis.

Mass spectrometry analysis.

Samples were analyzed by electrospray ionization mass spectrometry on a PE Sciex API III mass spectrometer (Thornhill, Ontario, Canada). Samples in aqueous solution were injected through a Harvard Apparatus model 22 syringe pump into the system at a flow rate of 10 μl/min of 50% acetonitrile buffer containing 0.1% formic acid.

RESULTS

HPLC analysis of [³H]β-l-ddA and its intracellular metabolites.

Figure 1 illustrates the HPLC radiochromatogram of intracellular extracts from phytohemagglutinin-stimulated PBMC following exposure of the cells to 10 μM [³H]β-l-ddA for 48 h. In addition to unchanged β-l-ddA, five other metabolites labeled A, B, C, D, and E with retention times of 3, 5, 10, 15, and 20 min, respectively, were detected. Similar HPLC profiles were obtained in HepG2 cells and hepatocytes.

FIG. 1.

HPLC radiochromatogram of intracellular extracts from PBMC following exposure to 10 μM [³H]β-l-ddA for 48 h.

Identification of β-l-ddA metabolites.

Mass spectrometry analysis of metabolites A and B showed single positive molecular ions (M + H)⁺ at m/z's of 117 and 137, respectively. These molecular weights (116 and 136) as well as their breakdown patterns were consistent with those of authentic standards of β-ribonolactone and hypoxanthine analyzed under similar conditions. Moreover, metabolites C, D, and E showed single negative molecular ions (M − H)⁻ at m/z's of 267, 426, and 506, respectively, by mass spectrometry. The molecular weights (268, 427, and 507) of these metabolites as well as their breakdown patterns were consistent with those of authentic standards of inosine, ADP, and ATP, respectively, analyzed under similar conditions. Similarly, the identity of the chromatographic peak corresponding to β-l-ddA-glucuronide was confirmed by mass spectrometry analysis. The mass spectrometry spectrum of the glucuronic derivative of β-l-ddA demonstrated a single positive molecular ion, (M + H)⁺, at an m/z of 412 and a single negative molecular ion (M − H)⁻ at an m/z of 410.

Analysis of the time course of accumulation of β-l-ddA and its metabolites (Fig. 2).

FIG. 2.

Proposed catabolic pathway of β-l-ddA.

Table 1 shows the intracellular concentrations and percentages of radioactivity of β-l-ddA and its metabolites in HepG2 cells after a 0- to 24-h exposure to 10 μM [³H]β-l-ddA. β-l-ddA levels gradually increased with time, attaining a maximum concentration of 5.08 ± 3.30 pmol/10⁶ cells at 8 h, and then decreased to 2.85 ± 0.09 pmol/10⁶ cells after 24 h. β-l-ddA was modestly phosphorylated in HepG2 cells, as the concentration of β-l-ddAMP was only 0.12 ± 0.14 pmol/10⁶ cells (0.5% of total radioactivity) after 24 h. Neither β-l-ddADP nor β-l-ddATP was detected under these conditions. The predominant metabolite, ATP, reached a maximum concentration of 5.14 ± 0.95 pmol/10⁶ cells (43% of total radioactivity) at 8 h and subsequently declined to 2.26 ± 0.65 pmol/10⁶ cells at 24 h. Inosine and hypoxanthine accounted for 0.87 ± 0.27 and 0.65 ± 0.42 pmol/10⁶ cells at 24 and 8 h of incubation, respectively. β-l-2′,3′-dideoxyribonolactone and ADP reached steady-state levels of 0.18 ± 0.01 and 0.34 ± 0.04 pmol/10⁶ cells, respectively, within 2 h and remained unchanged for the rest of the experiment.

TABLE 1.

Intracellular concentrations of β-l-ddA and metabolites in HepG2 cells and percent total radioactivity in each metabolite after incubation with 10 μM [2′,3′,8-³H]β-l-ddA for specific time periods

Compound	Concna
	2 h	4 h	8 h	24 h
β-l-ddA	2.50 ± 1.28 (41.3)	3.40 ± 1.54 (39.9)	5.08 ± 3.30 (42.5)	2.85 ± 0.09 (40.3)
β-l-ddAMP	0.10 (1.65)	0.05 ± 0.04 (0.6)	0.06 ± 0.07 (0.5)	0.12 ± 0.14 (1.7)
β-l-ddADP	ND	ND	ND	ND
β-l-ddATP	ND	ND	ND	ND
Ribonolactone	0.18 ± 0.01 (3.0)	0.14 ± 0.06 (1.6)	0.17 ± 0.06 (1.4)	0.17 ± 0.06 (2.4)
Hypoxanthine	0.23 ± 0.04 (3.8)	0.50 ± 0.17 (5.9)	0.65 ± 0.42 (5.4)	0.49 ± 0.37 (6.9)
Inosine	0.25 ± 0.04 (4.1)	0.37 ± 0.14 (4.3)	0.59 ± 0.12 (4.9)	0.87 ± 0.27 (12.3)
ADP	0.34 ± 0.04 (5.6)	0.28 ± 0.09 (3.3)	0.27 ± 0.20 (2.3)	0.32 ± 0.06 (4.5)
ATP	2.46 ± 0.05 (40.6)	3.79 ± 0.76 (44.4)	5.14 ± 0.95 (43.0)	2.26 ± 0.65 (31.9)

^aPicomoles/10⁶ cells ± standard deviation of three replications; percent total radioactivity is shown in parentheses. ND, not detectable. 

Table 2 shows the profile of intracellular metabolite formed after a 0- to 24-h exposure of human PBMC to a 10 μM concentration of [³H]β-l-ddA. Unchanged β-l-ddA levels gradually increased with time, attaining a maximum value of 1.24 ± 0.21 pmol/10⁶ cells at 8 h, and then decreased to 0.26 ± 0.07 pmol/10⁶ cells at 24 h. As was the case with HepG2 cells, the anabolism of β-l-ddA was very limited. The concentration of β-l-ddAMP was only 0.19 pmol/10⁶ cells at 24 h, and similarly, no β-l-ddADP or β-l-ddATP was detected. ATP was the predominant metabolite detected, reaching a maximum concentration of 8.15 ± 2.64 pmol/10⁶ cells (62% of total radioactivity) at 4 h and then decreasing to 5.51 ± 1.12 pmol/10⁶ cells at 24 h of incubation. ADP and inosine accounted for 2.24 ± 1.53 and 1.53 ± 0.40 pmol/10⁶ cells, respectively, at 4 h. Hypoxanthine and β-l-2′,3′-dideoxyribonolactone reached maximum concentrations of 0.50 ± 0.08 and 0.46 ± 0.01 pmol/10⁶ cells at 2 and 24 h, respectively.

TABLE 2.

Intracellular concentrations of β-l-ddA and metabolites in primary human PBMC and percent total radioactivity in each metabolite after incubation with 10 μM [2′,3′,8-³H]β-l-ddA for specific time periods

Compound	Concna
	2 h	4 h	8 h	24 h
β-l-ddA	0.32 ± 0.18 (2.7)	0.41 ± 0.27 (3.1)	1.24 ± 0.21 (11.2)	0.26 ± 0.07 (2.8)
β-l-ddAMP	0.11 (0.9)	0.16 (1.2)	0.11 (1.0)	0.19 (2.0)
β-l-ddADP	ND	ND	ND	ND
β-l-ddATP	ND	ND	ND	ND
Ribonolactone	0.14 ± 0.03 (1.2)	0.18 ± 0.12 (1.4)	0.21 ± 0.01 (1.9)	0.46 ± 0.01 (4.9)
Hypoxanthine	0.50 ± 0.08 (4.2)	0.49 ± 0.36 (3.7)	0.46 ± 0.13 (4.2)	0.36 ± 0.16 (3.9)
Inosine	1.36 ± 0.25 (11.3)	1.53 ± 0.40 (11.6)	1.27 ± 0.24 (11.5)	1.12 ± 0.37 (12.0)
ADP	1.87 ± 1.07 (15.6)	2.24 ± 1.53 (17.0)	1.33 ± 0.71 (12.0)	1.40 ± 0.65 (15.1)
ATP	7.70 ± 0.81 (64.2)	8.15 ± 2.64 (61.9)	6.43 ± 1.79 (58.2)	5.51 ± 1.12 (59.2)

^aPicomoles/10⁶ cells ± standard deviation of three replications; percent total radioactivity is shown in parentheses. ND, not detectable. 

Table 3 shows the intracellular metabolite formation profile obtained after a 0- to 24-h exposure of hepatocytes to a 10 μM concentration of [³H]β-l-ddA. In contrast to HepG2 cells and PBMC, there was a much larger accumulation of β-l-ddA in hepatocytes. Furthermore, although the formation of β-l-ddAMP was not detected at 2 h, there was a larger formation of β-l-ddAMP at 4 h than in HepG2 cells or PBMC. This larger formation of β-l-ddAMP was also accompanied by β-l-ddADP formation at 8 and 24 h. Maximal concentrations of β-l-ddAMP and β-l-ddADP were 1.12 ± 0.13 and 0.92 ± 0.42 pmol/10⁶ cells by 24 and 8 h, respectively, and β-l-ddATP was not detected at any time. Nevertheless, as with HepG2 cells and PBMC, ATP was the predominant metabolite, reaching values of 15.60 ± 1.74 pmol/10⁶ cells (32% of total radioactivity) by 2 h and then decreasing to 9.38 ± 1.19 pmol/10⁶ cells at 24 h. Inosine, ADP, hypoxanthine, and β-l-2′,3′-dideoxyribonolactone accounted for 4.60 ± 2.57, 2.19 ± 0.61, 0.69 ± 0.31, and 0.29 ± 0.12 pmol/10⁶ cells at 24, 8, 4, and 4 h, respectively. In addition to these metabolites, a 5′-glucuronidated derivative of β-l-ddA was observed, gradually increasing to 2.60 ± 0.36 pmol/10⁶ cells at 24 h.

TABLE 3.

Intracellular concentrations of β-l-ddA and metabolites in human primary cultured hepatocytes and percent total radioactivity in each metabolite after incubation with 10 μM [2′,3′,8-³H]β-l-ddA for specific time periods

Compound	Concna
	2 h	4 h	8 h	24 h
β-l-ddA	29.05 ± 7.34 (58.6)	25.36 ± 3.34 (54.4)	22.98 ± 2.65 (52.8)	24.07 ± 1.55 (53.5)
β-l-ddAMP	ND	0.41 ± 0.07 (0.9)	0.75 ± 0.12 (1.7)	1.12 ± 0.13 (1.8)
β-l-ddADP	ND	ND	0.92 ± 0.42 (2.1)	0.81 ± 0.41 (1.3)
β-l-ddATP	ND	ND	ND	ND
Ribonolactone	0.24 ± 0.10 (0.5)	0.29 ± 0.12 (0.6)	ND	ND
Hypoxanthine	0.43 ± 0.20 (0.9)	0.69 ± 0.31 (1.5)	0.52 ± 0.16 (1.2)	0.27 ± 0.07 (0.6)
Inosine	1.90 ± 0.81 (3.8)	2.68 ± 0.59 (5.7)	3.90 ± 1.97 (8.9)	4.60 ± 2.57 (10.2)
ADP	1.73 ± 0.45 (3.5)	1.98 ± 0.32 (4.2)	2.19 ± 0.61 (5.0)	2.16 ± 0.53 (4.8)
ATP	15.60 ± 1.74 (31.5)	13.99 ± 1.43 (30.0)	10.99 ± 1.30 (25.2)	9.38 ± 1.19 (20.8)
β-l-ddA-Gluc	0.61 ± 0.34 (1.2)	1.25 ± 0.47 (2.7)	1.31 ± 0.29 (3.0)	2.60 ± 0.36 (5.8)

^aPicomoles/10⁶ cells ± standard deviation of three replications; percent total radioactivity is shown in parentheses. ND, not detectable. 

It should also be noted that approximately 40 to 50% of the total radioactivity in HepG2 cells and hepatocytes was associated with 2′,3′-dideoxyribonolactone, hypoxanthine, inosine, ADP, and ATP. In contrast, these metabolites accounted for over 90% of the total radioactivity in PBMC. These metabolites resulted from the phosphorolysis and catabolism of β-l-ddA as discussed below.

When radiolabeled β-l-ddA was incubated for 24 h in the presence of increasing concentrations of MTA in primary cultured hepatocytes, a statistically significant reduction of β-l-ddA catabolism was observed compared to that in control cells. Inosine levels were reduced threefold, from 4.91 ± 0.90 to 1.60 ± 0.53 μM; ADP levels were decreased more than twofold, from 1.86 ± 0.31 to 0.72 ± 0.20 μM; and ATP concentrations were decreased threefold, from 10.49 ± 4.27 to 3.32 ± 1.79 μM, with the addition of 10 μM MTA to cells. Moreover, unchanged β-l-ddA concentrations were increased from 13.50 ± 4.82 to 25.49 ± 5.62 μM by incubation of cells with MTA. These results strongly demonstrate that MTA phosphorylase is responsible for β-l-ddA degradation (Fig. 3).

FIG. 3.

Intracellular concentrations of β-l-ddA and metabolites, inosine, ADP, and ATP after a 24-h incubation of cells to 10 μM [³H]β-l-ddA in the absence or presence of 10 and 50 μM MTA. The results represent the mean of three experiments with triplicate incubations. The P values represent the statistical significance of the differences in the observed concentrations of β-l-ddA and metabolites determined in the presence of MTA compared to that of the control.

DISCUSSION

Preliminary studies demonstrated that β-l-ddATP potently and selectively inhibited HIV reverse transcriptase with a K_i of 2.0 μM and inhibited HBV DNA polymerase with an EC₅₀ of 2.1 μM while it did not have any effect on human DNA polymerase α, β, and γ when tested at up to 100 μM (Faraj et al., Prog. Abstr. 13th Int. Round Table Nucleosides Nucleotides Biol. Appl.). On the other hand, β-l-ddA per se exhibited limited activity against HBV replication in human hepatoblastoma-derived HepG2 cells, with an EC₅₀ of 5 to 6 μM (4, 8) and had no anti-HIV activity in PBMC, with an EC₅₀ of >100 μM (1, 10). In order to understand the reason for the weak antiviral activity of β-l-ddA, we studied its intracellular metabolism. The accumulation of unchanged β-l-ddA with time suggests that the uptake of β-l-ddA is not impaired in these cell cultures and that other metabolic factors are involved in the reduced antiviral activity of β-l-ddA.

Previous studies showed that β-l-ddA is not a substrate or, at best, is a very poor substrate for adenosine kinase (EC 2.7.1.20) (1) and adenosine deaminase (EC 3.5.4.4) (6, 12) but is phosphorylated by 2′-deoxycytidine kinase (EC 2.7.1.74) with a high K_m of 220 μM (6). As illustrated in Tables 1, 2, and 3, the levels of 5′-phosphorylated derivatives of β-l-ddA in all cell types studied were extremely low after incubation with radiolabeled β-l-ddA. In fact, β-l-ddATP was below the limit of detection at all time points. Furthermore, β-l-ddAMP reached concentrations of only 0.18 ± 0.13, 0.32 ± 0.12, and 1.12 ± 0.13 pmol/10⁶ cells at 24, 4, and 24 h in HepG2, hepatocytes, and PBMC, respectively. These results suggest that there was minor anabolism of β-l-ddA in these cells. On the other hand, there was significant catabolism.

Five other metabolites of β-l-ddA were detected and identified as β-l-2′,3′-dideoxyribonolactone, hypoxanthine, inosine, AMP, and ATP. ATP was the most prominent metabolite observed in all three cell culture systems. These metabolites could only have been formed via the catabolism of β-l-ddA through phosphorolysis and cleavage of the glycosidic bond of the nucleoside to yield ribonolactone and purine base. Hypoxanthine could then be salvaged by hypoxanthine-guanine phosphoribosyltransferase (EC 2.4.2.8), leading to the formation of inosine 5′-monophosphate, and metabolized further to the 5′-phosphorylated derivatives of adenosine as illustrated in Fig. 2. Since β-l-ddA is not a substrate for purine nucleoside phosphorylase (EC 2.4.2.1) (10), the cleavage of the glycosidic bond must be the result of another purine nucleoside phosphorylase. MTA phosphorylase (EC 2.4.2.28) is known to catalyze the degradation of MTA to yield adenine and 5-methylthioribose-1-phosphate. Adenine is then recycled via purine salvage pathways, and 5-methylthioribose-1-phosphate is converted via a multistep pathway to methionine (14). Such degradation of MTA is critical in preventing its accumulation and its inhibitory effect on cell growth (11).

This degradation of β-l-ddA would lower its intracellular concentration below its K_m with 2′-deoxycytidine kinase. This in turn would lower the efficiency of 2′-deoxycytidine kinase in phosphorylating β-l-ddA to β-l-ddAMP.

The metabolic profile observed for β-l-ddA contrasts with the metabolic pathway observed with its β-d- enantiomer. As described by Johnson and Fridland (7), β-l-ddA is rapidly deaminated by adenosine deaminase to 2′,3′-dideoxyinosine, which can be further phosphorylated by 5′-nucleotidase (EC 3.5.3.5). Its 5′-monophosphate derivative is then converted by adenylate synthase or lyase to ddAMP, which is further phosphorylated by adenylate kinase and nucleoside diphosphate kinase to the pharmacologically active ddATP. In addition, while β-l-ddA is extensively degraded by MTA phosphorylase, β-d-ddI is a poor substrate for purine nucleoside phosphorylase, minimizing the catabolism of ddI to hypoxanthine and dideoxyribose-1-phosphate. Therefore, β-d-ddI is accumulated in the cell and then activated by phosphorylation, resulting in high antiviral activity.

Finally, the rate of β-l-ddA catabolism can explain its differential antiviral activity against HIV (EC₅₀, >100 μM) grown in PBMC (1) and HBV (EC₅₀, 5 to 6 μM) grown in HepG2 cells (4, 8). In PBMC, the metabolites resulting from [³H]β-l-ddA catabolism accounted for over 90% of the intracellular radioactivity (Table 2), while in HepG2 cells these metabolites accounted for only 40% of the total radioactivity (Table 1). Therefore, there is more β-l-ddA available for phosphorylation to the nucleotide level in HepG2 cells (Table 1) than in PBMC (Table 2), consistent with the greater impact of viral replication inhibition in HepG2 cells than in PBMC.

In conclusion, the present study clearly demonstrates that the limited antiviral activity of β-l-ddA is mainly due to its rate of catabolism via the breaking of the glycosidic bond by MTA phosphorylase, resulting in low intracellular concentration of β-l-ddA, far below its K_m for the activating enzyme, 2′-deoxycytidine kinase; hence the low observed phosphorylation of β-l-ddA. The present study emphasizes that the enantiomeric selectivity of the catabolic enzymes as well as the anabolic enzymes in host cells is of particular importance in the possible role of β-l-ddA as an antiviral agent.