Comparison of the Phosphorylation of 4′-Ethynyl 2′,3′-Dihydro-3′-Deoxythymidine with That of Other Anti-Human Immunodeficiency Virus Thymidine Analogs

Abstract

Thymidine analogs, including 3′-azido-3′-deoxythymidine (AZT) and 2′,3′-dideoxy-3′-deoxythymidine (D4T), are important antiretroviral agents. To exert antiretroviral activity, these analogs undergo a stepwise phosphorylation intracellularly to the active triphosphate metabolites. We previously reported that 4′-substituted D4T with an ethynyl group (i.e., 4′-ethynyl D4T) increased the anti-human immunodeficiency virus (HIV) activity and was active against multidrug-resistant HIV strains. 4′-Ethynyl D4T is a better substrate for phosphorylation by human thymidine kinase 1 than D4T is. In this report, we first studied the enzymes involved in the phosphorylation of 4′-ethynyl D4T from monophosphate to triphosphate metabolites. The 4′-ethynyl D4TMP is phosphorylated by recombinant human TMP kinase with a K_m of 19 ± 4 μM and a k_cat of 0.007 ± 0.001 s⁻¹; the relative efficiency is about 9 and 15% of those of D4TMP and AZTMP, respectively. Several enzymes from crude cellular extracts, including nucleoside diphosphate kinase, pyruvate kinase, creatine kinase, and 3-phosphoglycerate kinase, could phosphorylate 4′-ethynyl D4T-diphosphate. The relative phosphorylation efficiencies of 4′-ethynyl D4TDP were about 3 to 25% of those of D4TDP and were generally similar to those of AZTDP. In T-lymphoid cell lines, there was a preponderant accumulation of 4′-ethynyl D4TMP, suggesting that TMP kinase could be the rate-limiting enzyme in the metabolism of 4′-ethynyl D4T. Although the same enzymes are involved in the stepwise phosphorylation of thymidine analogs, their behaviors in phosphorylating metabolites of 4′-ethynyl D4T are different from those of D4T and AZT. Qualitatively, the metabolism of 4′-ethynyl D4T is more similar to that of AZT than to that of its progenitor, D4T.

Nucleoside analogs are one of the most important classes of compounds active against human immunodeficiency virus (HIV) infection (30). Pyrimidine 2′,3′-dideoxynucleoside analogs, including 3′-azido-3′-deoxythymidine (AZT or zidovudine), 2′,3′-dideoxy-3′-deoxythymidine (D4T or stavudine), and β-l-2′,3′-dideoxy-3′-thiacytidine (3TC or lamivudine), which are potent inhibitors of HIV replication, have been widely used as components of antiretroviral regimens for years (10, 22, 23, 30). The long-term use of antiviral nucleoside analogs is inevitably associated with delayed toxicity and/or development of drug-resistant virus (3, 5, 31, 31a). There is, therefore, an urgent need for the development of novel anti-HIV agents with better therapeutic indices, new mechanisms of action, and activity against HIV strains resistant to currently available drugs.

Recently, we reported that 4′ substitution of an ethynyl group to D4T resulted in an increased anti-HIV activity (6, 11, 12). In MT-2 cells infected with HIV IIIB virus, 4′-ethynyl D4T was shown to have a fivefold increase in antiviral activity compared to that of its parental compound, D4T (6). 4′-Ethynyl D4T was also found to have considerable inhibitory effect against the replication of multidrug-resistant HIV strains (25). Furthermore, compared to that of the parental D4T, it has fourfold better phosphorylation efficiency by human thymidine kinase 1 (TK1) (6). The 4′-ethynyl D4T-triphosphate (TP) has less inhibitory activity than D4TTP does against human DNA polymerases, including DNA polymerase γ (34). Taken together, these data suggest that 4′-ethynyl D4T has the potential to be developed as a clinically useful anti-HIV drug with a better clinical efficacy and toxicity profile than those of AZT and D4T.

To exert the antiviral effect, nucleoside analogs need to undergo stepwise phosphorylation to their respective triphosphate metabolites by intracellular enzymes (32, 33). The enzymes involved in the activation could be quite different qualitatively and quantitatively from one analog to another. For instance, those enzymes responsible for the activation of thymidine analogs, such as AZT and D4T, are different from those responsible for deoxycytidine analogs, such as lamivudine (3TC). The phosphorylation of naturally occurring pyrimidine deoxynucleosides or their analogs to their monophosphate (MP) metabolites in human cells is carried out by TK1 (the cytoplasmic TK), TK2 (the mitochondrial TK), or deoxycytidine kinase (dCK or CdRK) (33). The phosphorylation of pyrimidine nucleoside monophosphate metabolites to their diphosphate (DP) forms is catalyzed by TMP kinase (TMPK) or uridylate/cytidylate (UMP/CMP) kinase (19, 24, 32). Several enzymes have been shown to be able to phosphorylate nucleoside DPs. Enzymes of the nm23 family are members of the classic type of nucleoside diphosphate kinase (NDPK) using ATP as the phosphate donor. Among them, nm23-H1 and nm23-H2, the predominant isoforms residing in the cytoplasm (18), have been characterized in detail with regard to their kinetic properties and catalytic mechanisms (29). Other enzymes that are able to phosphorylate nucleoside diphosphates include creatine kinase (CK), pyruvate kinase (PK), and 3-phosphoglycerate kinase (PGK) (14-17). Our previous studies demonstrated that TK1 is the enzyme responsible for the phosphorylation of 4′-ethynyl D4T to 4′-ethynyl D4TMP (6). However, the enzymes responsible for the subsequent phosphorylation of 4′-ethynyl D4TMP to the DP and TP metabolites have not been characterized.

We hypothesized that the structural differences between 4′-ethynyl D4T and other thymidine analogs may affect the phosphorylation efficiency carried out by the metabolizing enzymes. Therefore, in this report, we evaluated the kinetics and activities of human pyrimidine nucleoside metabolizing enzymes that phosphorylate the MP and DP metabolites of 4′-ethynyl D4T in comparison with those of D4T and AZT in vitro and demonstrated the formation of phosphorylated metabolites of 4′-ethynyl D4T in cultured T-lymphoid cells. By comparing the behaviors of phosphorylated metabolites of 4′-ethynyl D4T and other anti-HIV thymidine analogs (i.e., D4T and AZT) toward the metabolizing enzymes, we intended to get a better insight about how the structural differences of the thymidine analogs are related to their phosphorylation efficiencies, which might be critical for their anti-HIV activities.

MATERIALS AND METHODS

Chemicals.

The 4′-ethynyl D4T was synthesized by the laboratory of Hiromichi Tanaka, School of Pharmaceutical Sciences, Showa University, Tokyo, Japan (11). Thymidine, D4T, AZT, dTMP (TMP), dTDP (TDP), and AZTMP were purchased from Sigma-Aldrich Corp. (St. Louis, MO). The MP and DP metabolites of 4′-ethynyl D4T were chemically synthesized and purified according to previously published procedures with minor modifications (26). AZTDP, D4TMP, and D4TDP were synthesized by the same method. Radiochemicals, including [5′-³H]-4′-ethynyl D4T (11 Ci/mmol), [methyl-³H]-D4T (11 Ci/mmol), and [methyl-³H]-AZT (12 Ci/mmol), were purchased from Moravek Biochemicals, Inc. (Brea, CA).

Cell lines.

CCRF-CEM (CEM) (ATCC CCL-119) is a human T-lymphoblastoma cell line. CEM cells were maintained in RPMI medium supplemented with 10% fetal bovine serum at 5% CO₂ and 37°C.

Preparation of crude extracts from cultured cells.

Cells in logarithmic growth were harvested. After being washed three times with ice-cold phosphate-buffered saline, the cells were incubated with lysis buffer, which was composed of 25 mM Tris-HCl (pH 7.5), 25 mM NaCl, 5 mM NaF, 2 mM dithiothreitol (DTT), 0.5 mM EDTA, 1% Nonidet-P 40 (NP-40), and protease inhibitor cocktail (Roche Diagnostics Corp., Indianapolis, IN). The cell extracts were incubated on ice for 1 h, and the supernatants were collected after ultracentrifugation at 13,500 rpm for 20 min. The nucleotides and small molecules, which might potentially interfere with the enzyme assays, were removed by G-50 Sephadex columns (Roche Diagnostics Corp., Indianapolis, IN) according to the manufacturer's instructions.

Recombinant human TMPK, nm23-H1, nm23-H2, and PGK.

Human TMPK was cloned from HeLa cells, a human cervical carcinoma cell line (13). Human nm23-H1 and nm23-H2 were cloned from KB cells, a human oropharyngeal carcinoma cell line (15). Human PGK was cloned from HepG2 cells, a human hepatoma cell line (16). These proteins, whose encoding genes were each inserted into the pET-28a expression vector (Novagen, Inc., Madison, WI), were expressed in Escherichia coli with an N-terminal His-tag configuration. The details of cloning, expression, and purification have been reported previously (13, 15, 16).

Phosphorylation of nucleoside monophosphates by human TMPK.

The assay of TMPK activity (13) was performed in a reaction mixture containing 60 mM Tris-HCl (pH 7.4), 2.5 mM DTT, 4 mM MgCl₂, 0.15% bovine serum albumin, and 4.5 mM ATP. The kinetic properties of TMP and thymidine analog MPs were examined under various concentrations of TMP (6.25 to 100 μM) or thymidine analog MPs (6.25 to 250 μM). The reactions were performed at 37°C and stopped on ice by adding trichloroacetic acid after various durations of reaction time. The samples were then neutralized by two extractions with a mixture of trioctylamine and 1,1,2-trichlorotrifluoroethane (Aldrich Chem. Co., Milwaukee, WI) in a ratio of 45:55. The phosphorylated metabolites of thymidine analog monophosphates and TMP were analyzed by high-pressure liquid chromatography (HPLC) (Shimadzu Class-VP HPLC; Shimadzu, Braintree, MA) with a binary gradient of water to 300 mM potassium phosphate buffer using an anion-exchange column (Partisil-SAX; Whatman, Inc., Clifton, NJ). Michaelis-Menten constants (K_m) and turnover number (k_cat) values were determined from Lineweaver-Burk plots.

Enzymatic phosphorylation of nucleoside analog diphosphates.

The assays of NDPK, CK, PK, and PGK (14-17) were performed with a reaction mixture containing 50 mM Tris-HCl (pH 7.5), 10 mM KCl, 5 mM MgCl₂, 1 mM NaF, 1 mM DTT, and 100 μM of TDP or nucleoside analog diphosphates in a total volume of 100 μl in which 2 mM ATP for the NDPK assay, 2 mM creatine phosphate for the CK assay, or 2 mM phosphoenolpyruvate for the PK assay as the phosphate donor was used, respectively. The 1,3-biphosphoglycerate, which is the phosphate donor for PGK, was generated by a coupled reaction using 2 mM sodium phosphate, 2 mM NAD, and 2 mM dl-glyceraldehyde-3-phosphate in the presence of 1 unit/ml of glyceraldehyde-3-phosphate dehydrogenase for 10 min before the PGK assay. Reactions were conducted, and the samples were processed for HPLC analysis by using the procedures described above (see “Phosphorylation of nucleoside monophosphates by human TMPK”). The efficiencies for nucleoside analog diphosphates to be phosphorylated were presented as nanomoles/minute/milligram of protein, which was further normalized to the units of enzyme activity contained in either crude cell extract or recombinant proteins. The unit of enzyme activity is defined as the conversion of 1 μmol of ADP to ATP in 1 min.

Intracellular metabolism of radiolabeled 4′-ethynyl D4T, D4T, and AZT in CEM cells.

The metabolism of the radiolabeled compounds was monitored according to previously established procedures (17). Briefly, CEM cells were seeded at 10⁶ cells/ml and were incubated with 2 μM (75 mCi/mmol) of various radiolabeled compounds. At 12 h, cells were harvested by centrifugation and washed in cold phosphate-buffered saline containing 20 μM dipyridamole (Sigma) and the cell pellets were extracted with 15% trichloroacetic acid for 15 min on ice. The supernatant containing the nucleoside and its phosphorylated forms was neutralized by two extraction methods with a 45/55 ratio of trioctylamine and 1,1,2-trichlorotrifluroethane. The nucleoside analog metabolites in the soluble fraction were analyzed by HPLC (Shimadzu, Braintree, MA) connected to a radiometric detector (Radiomatic 150TR flow scintillation analyzer; Packard) using a Partisil SAX column (Whatman, Clifton, NJ).

RESULTS

Phosphorylation of MPs of 4′-ethynyl D4T, AZT, and D4T by human TMPK.

Initially, we studied whether TMPK or UMP/CMP kinase, the two cellular enzymes responsible for the phosphorylation of pyrimidine MPs, could phosphorylate 4′-ethynyl D4TMP. The initial data revealed that TMPK, but not UMP/CMP kinase, could phosphorylate 4′-ethynyl D4TMP. We then evaluated the phosphorylation characteristics of anti-HIV thymidine analog monophosphates as well as the natural substrate TMP by recombinant human TMPK. As shown in Table 1, The K_m for 4′-ethynyl D4TMP was 19 μM, which was similar to that for D4TMP (22 μM) but higher than that for AZTMP (7.6 μM). The k_cat for 4′-ethynyl D4TMP was only 0.007 s⁻¹, which was much lower than that for D4TMP (0.093 s⁻¹) and slightly lower than that for AZTMP (0.018 s⁻¹). The characteristics of TMPK with a K_m of 8.3 μM and a k_cat of 1.8 s⁻¹ towards naturally occurring TMP were generally in agreement with previous reports using highly purified enzymes from human cells or recombinant human proteins (4, 20). Overall, the relative efficiency of 4′-ethynyl D4TMP phosphorylated by human TMPK was 8.8% of that of D4TMP (Table 1).

TABLE 1.

Kinetic characteristics of 4′-ethynyl D4TMP, AZTMP, D4TMP, and TMP as substrates for recombinant TMP kinase^a

Nucleoside analog	Km (μM)	kcat (s−1)	Relative efficiencyb (%)
4′-Ethynyl D4TMP	19 ± 4	0.007 ± 0.001	8.8
AZTMP	7.6 ± 2.3	0.018 ± 0.015	59
D4TMP	22 ± 4	0.093 ± 0.014	100
TMP	8.3 ± 4.0	1.8 ± 0.4	5.6 × 103

^aProcedures were as described in Materials and Methods. K_m and k_cat values were determined from Lineweaver-Burk plots. The data were expressed as means ± standard deviations from three independent experiments.

^bRelative efficiency was calculated from k_cat/K_m, with reference to D4TMP as 100%.

Phosphorylation of DPs of 4′-ethynyl D4T, AZT, and D4T by crude extracts from cultured cells.

Because the T lymphocyte is the primary host cell for HIV infection, we first used the crude extracts from CEM T-lymphoblastoma cells as the source of enzymes to study the phosphorylation of DPs of these thymidine analogs to their TPs. By supplying different phosphate donors, we were able to evaluate the activities of different enzymes, such as NDPK, CK, PK, and PGK. As shown in Table 2, the enzyme activities of phosphorylating 4′-ethynyl D4TDP were 0.032 ± 0.006, 0.091 ± 0.011, and 0.30 ± 0.02 nmol/min/mg for NDPK, PK, and PGK, respectively. These activities are approximately 30-fold, 4-fold, and 20-fold, respectively, less efficient than those of phosphorylating D4TDP. AZTDP could be phosphorylated by NDPK and PGK. The efficiencies of phosphorylating AZTDP by NDPK and PGK were similar to those of phosphorylating 4′-ethynyl D4TDP. However, neither AZTDP nor TDP could be phosphorylated by PK from CEM cells. It is also interesting to note that although we did detect CK activity in the phosphorylation of ADP to ATP (the activity was around 1.5 nmol/min/mg in CEM cell lysate) (Table 2), neither thymidine analog DPs nor TDP could be phosphorylated by CK from CEM cells.

TABLE 2.

Phosphorylation of 4′-ethynyl D4TDP, AZTDP, D4TDP, and TDP by crude extracts from CEM cells^a

Parameter for phosphate	Value for indicated enzyme
	Nucleoside diphosphate kinase	Creatine kinase	Pyruvate kinase	3-Phosphoglycerate kinase
Enzyme activity of crude extracts of CEM cells (nmol/min/mg)
4′-Ethynyl D4TDP	0.032 ± 0.006	NDc	0.091 ± 0.011	0.3 ± 0.02
AZTDP	0.018 ± 0.004	ND	ND	0.26 ± 0.02
D4TDP	1.1 ± 0.3	ND	0.35 ± 0.08	4.2 ± 0.8
TDP	(2.7 ± 0.3) × 103	ND	ND	21 ± 5
ADP	(2.5b ± 0.2) × 103	2.5 ± 0.3	3.6 ± 0.4	(10.6 ± 1.2) × 103
Conversion normalized to per unit activity (nmol/unit)
4′-Ethynyl D4TDP	0.013 ± 0.002	ND	25 ± 3	0.028 ± 0.002
AZTDP	0.007 ± 0.002	ND	ND	0.025 ± 0.001
D4TDP	0.42 ± 0.13	ND	99 ± 23	0.4 ± 0.08
TDP	(1.1 ± 0.1) × 103	ND	ND	2 ± 0.4
Relative phosphorylation efficiency (D4TDP as 100%)
4′-Ethynyl D4TDP	3 ± 0.6	ND	26 ± 3	7 ± 0.5
AZTDP	1.7 ± 0.4	ND	ND	6.2 ± 0.4
D4TDP	100	ND	100	100
TDP	(2.6 ± 0.3) × 105	ND	ND	490 ± 110

^aThe reaction mixtures contained nucleoside analog diphosphates as phosphate acceptors and cellular extracts from CEM cells as sources of enzymes. Experiment procedures were detailed in Materials and Methods. The CEM cell extracts were first characterized for the efficiencies in phosphorylating ADP by using different phosphate donors. One unit was defined as the amount of enzyme responsible for the conversion of 1 μmol ADP to ATP in 1 minute. The unit activity per mg of CEM cell extracts for different enzyme categories is shown in the last row of the “Enzyme activity of crude extracts of CEM cells” section. The enzyme activities of each enzyme category of CEM cell lysates in phosphorylating respective nucleoside analog diphosphates were expressed as nmol/min/mg protein as shown in the “Enzyme activity of crude extracts of CEM cells” section. The enzyme activities were further normalized to the unit activity of each enzyme category (nmol/unit) as shown in the “Conversion normalized to per unit activity” section. After being compared to the efficiency of D4TDP phosphorylation, the relative efficiency was shown in the “Relative phosphorylation efficiency” section. (D4TDP as 100%). Values are expressed as means ± standard deviations from three independent experiments.

^bTo determine the unit activity of NDPK, GTP was used as phosphate donor to test the efficiency of CEM cell lysates in conversion of ADP to ATP.

^cND, not detected.

In summary, the results from the in vitro enzyme assay using crude cell extracts indicate that while NDPK is the major enzyme responsible for the phosphorylation of naturally occurring TDP, other enzymes, such as PGK and PK, may play more important roles in phosphorylating anti-HIV thymidine analog DPs studied. PGK appears to play a predominant role in the phosphorylation of anti-HIV thymidine analog DPs.

Relative efficiencies in phosphorylating DPs of 4′-ethynyl D4T, AZT, and D4T by different enzymes from crude extracts of CEM cells.

In order to determine the relative efficiency of phosphorylating these thymidine analog DPs in comparison with that of phosphorylating ADP, we determined the activity of each enzyme in CEM cells and expressed the activity in units, which was defined as the activity of enzyme required to phosphorylate 1 μmol of ADP to ATP in 1 min using different donors. We normalized the conversion of TDP or thymidine analog DPs to the activity of each enzyme. As shown in Table 2, the most efficient enzyme in the phosphorylation of 4′-ethynyl D4TDP and D4TDP, in comparison with the phosphorylation of ADP, was PK. One unit of PK activity could phosphorylate 25 ± 3 and 99 ± 23 nmol of 4′-ethynyl D4TDP and D4TDP, respectively. The efficiencies of NDPK and PGK enzymes from CEM cells in the phosphorylation of 4′-ethynyl D4TDP or D4TDP, in comparison with the phosphorylation of ADP, were much less than those of PK. Furthermore, 4′-ethynyl-D4TDP, D4TDP, and AZTDP could be phosphorylated by NDPK and PGK with similar efficiencies, while naturally occurring TDP was preferentially phosphorylated by NDPK.

Taken together, our data indicate that PK may be a very efficient enzyme in phosphorylating 4′-ethynyl D4TDP and D4TDP in comparison with phosphorylating ADP. However, the contribution of PK in phosphorylating these anti-HIV thymidine analog DPs in CEM cells may be limited because of the relatively small activity of this enzyme in cells. The more predominant role of PGK in phosphorylating thymidine analog DPs (shown in Table 2) in CEM cells is likely due to the abundance of PGK enzymes.

Phosphorylation of DPs of 4′-ethynyl-D4T, AZT, and D4T by recombinant proteins of nm23-H1, nm23-H2, and PGK.

Our data for crude cell extracts showed that PGK and NDPK activities in cellular extract were capable of phosphorylating these anti-HIV thymidine analog DPs. The contribution of these enzymes was further verified by experiments using recombinant human enzymes. As shown in Table 3, we found that nm23-H1 and nm23-H2 behaved very similarly: both recombinant proteins had activities in phosphorylating naturally occurring TDP, and both could phosphorylate 4′-ethynyl D4TDP and AZTDP with similar efficiencies, which were about 15-fold less efficient than that of D4TDP. This pattern was basically in line with what was observed in the experiments using crude cell extracts (Table 2). However, the specific activities of recombinant PGK in terms of phosphorylating TDP and other thymidine analog DPs were considerably lower than that of nm23-H1 or nm23-H2. When normalized to per unit activity, PGK could phosphorylate TDP and these analog DPs only with activities between 0.001 and 0.050 nmol/unit and in the order of TDP > AZTDP ∼ D4TDP > 4′-ethynyl D4TDP (Table 3). These activities were not only less than those of nm23-H1 and nm23-H2 but also much less than those of PGK enzymes derived from CEM cell extracts (20- to about 100-fold difference). Furthermore, the relative efficiencies in phosphorylating TDP and thymidine analog DPs by recombinant PGK were different from the relative efficiencies derived from CEM cell extracts (comparison of Tables 2 and 3). For example, while PGK activity from cell extracts phosphorylated AZTDP much less efficiently than D4TDP, the recombinant PGK could phosphorylate AZTDP as efficiently as D4TDP.

TABLE 3.

Phosphorylation of 4′-ethynyl D4TDP, AZTDP, D4TDP, and TDP by recombinant proteins of nm23-H1, nm23-H2, and PGK^a

Parameter for phosphate	Value for indicated enzyme
	nm23-H1	nm23-H2	3-Phosphoglycerate kinase
Specific enzyme activity (nmol/min/mg)
4′-Ethynyl D4TDP	113 ± 18	149 ± 16	3.9 ± 0.2
AZTDP	107 ± 11	131 ± 15	14 ± 1
D4TDP	(1.5 ± 0.1) × 103	(1.7 ± 0.2) × 103	10 ± 1
TDP	(1.3 ± 0.2) × 106	(1.4 ± 0.2) × 106	137 ± 13
Conversion normalized to per unit activity (nmol/unit)
4′-Ethynyl D4TDP	0.09 ± 0.01	0.11 ± 0.01	0.0014 ± 0.0001
AZTDP	0.08 ± 0.01	0.1 ± 0.01	0.0050 ± 0.0003
D4TDP	1.2 ± 0.1	1.2 ± 0.2	0.0037 ± 0.0002
TDP	(1 ± 0.2) × 103	(1 ± 0.2) × 103	0.049 ± 0.004
Relative phosphorylation efficiency (D4TDP as 100%)
4′-Ethynyl D4TDP	7.7 ± 1.2	8.9 ± 1	37 ± 2
AZTDP	7.2 ± 0.8	7.6 ± 0.9	135 ± 9
D4TDP	100	100	100
TDP	(8.9 ± 1.6) × 104	(8.3 ± 1.4) × 104	(1.3 ± 0.1) × 103

^aProcedures were as described in Materials and Methods. The specific enzyme activities of each enzyme in phosphorylating respective nucleoside diphosphates were expressed as nmol/min/mg protein as shown in the “Specific enzyme activity” section. The specific enzyme activities were further normalized to the unit activity of each enzyme, defined as the amount of enzyme responsible for the conversion of 1 μmol ADP to ATP in 1 min, and were shown as nmol/unit in the “Conversion normalized to per unit activity” section. After being compared to the efficiency of D4TDP phosphorylation, the relative efficiency was shown in the “Relative phosphorylation efficiency” section. Values are expressed as means ± standard deviations from three independent experiments.

Phosphorylation of radiolabeled 4′-ethynyl D4T, D4T, and AZT in CEM cells (Fig. 1).

FIG. 1.

Intracellular metabolism of 4′-ethynyl D4T, D4T, and AZT. CEM cells were seeded at 10⁶ cells/ml and incubated with 2 μM of the radiolabeled compounds for 12 h. Cells were then harvested in cold phosphate-buffered saline and extracted with 15% trichloroacetic acid for 15 min on ice. The supernatant containing the nucleoside and its phosphorylated forms was extracted with a 45/55 ratio of trioctylamine and 1,1,2-trichlorotrifluroethane. The nucleoside analog metabolites in the soluble fraction were analyzed by high-pressure liquid chromatography. The detected radioactivity was then calculated into picomoles of each type of metabolite for each nucleoside analog/10⁶ cells. The results are derived from two independent experiments.

In order to ensure that enzymes in CEM cells could phosphorylate 4′-ethynyl D4T to its TP metabolites, cells were incubated with 2 μM radiolabeled 4′-ethynyl D4T for 12 h and the intracellular MP, DP, and TP metabolites of 4′-ethynyl D4T were found and determined to be 16.11 ± 0.22, 0.079 ± 0.036, and 0.42 ± 0.18 pmol/10⁶ cells, respectively. This predominance of MP metabolites was also found in cells exposed to AZT, but not in cells exposed to D4T. Furthermore, the ratio of DP metabolites to MP metabolites was highest in cells exposed to D4T (111 versus 0.49% in 4′-ethynyl D4T or 0.40% in AZT). The data corroborate the in vitro enzyme studies, which already demonstrated the inferior phosphorylation efficiency of 4′-ethynyl D4TMP and AZTMP by recombinant TMPK compared with that of D4T (Table 1). On the other hand, the intracellular ratio of TP metabolites to DP metabolites for 4′-ethynyl D4T was no less than those of AZT and D4T. The intracellular levels of TP metabolites, the active metabolites against HIV reverse transcriptase, of 4′-ethynyl D4T (0.42 ± 0.18 pmol/10⁶ cells) and of AZT (0.27 ± 0.15 pmol/10⁶ cells) were higher than those of D4T (0.15 ± 0.06 pmol/10⁶ cells) when cells were exposed to 2 μM of these thymidine analogs.

Overall, the intracellular metabolism study confirms the inferior phosphorylation efficiency from MPs to TPs for 4′-ethynyl D4T than D4T, as suggested by the in vitro enzyme assays. However, the higher level of TP metabolites of 4′-ethynyl D4T than that of D4T indicates that the overall intracellular metabolism is improved by 4′-ethynyl substitution of D4T.

DISCUSSION

To exert their anti-HIV activity, thymidine nucleoside analogs need to be phosphorylated in a stepwise fashion to their TP metabolites, which could be incorporated into HIV DNA and cause premature termination of viral DNA chain elongation (30). The interaction of these analogs and their phosphorylated metabolites with the metabolic enzymes would determine the relative ratios of their phosphorylated metabolites and the amount of TP-metabolites formed. Although 4′-ethynyl D4T is a derivative of D4T, the behavior of 4′-ethynyl D4T and its metabolites toward metabolizing enzymes is more similar to that of AZT than to that of D4T in the in vitro studies. This would predict that the 4′-ethynyl D4T would have an intracellular metabolism pattern that is similar to that of AZT and different from that of D4T. Indeed, this hypothesis was confirmed by a cell culture study using CEM cells and another T-lymphoid cell line, H9 (G. Dutschman and Y.-C. Cheng, data not shown). In CEM cells, the rate-limiting step of intracellular metabolism of AZT in T-lymphoid cells is TMP kinase (1, 2, 7-9) and the rate-limiting step for 4′-ethynyl D4T, as shown in the current report, could be TMP kinase as well. It is possible that other NMP kinases or related enzymes could be involved.

Although the in vitro enzyme assay showed that 4′-ethynyl D4T was phosphorylated from MPs towards TPs less efficiently than D4T was, the intracellular metabolism experiment indicated that 4′-ethynyl D4T formed more TP metabolites in CEM cells than D4T did. These two apparently contradictory observations could be explained in part by the fact that a much higher concentration of 4′-ethynyl D4TMP than D4TMP was formed in CEM cells (Fig. 1). In our previous work (6), 4′-ethynyl D4T was phosphorylated fourfold more efficiently than was D4T to MPs by TK1, with a lower K_m (52 μM versus 133 μM) and a better catalytic rate (180 versus 100%). In the presence of 2 μM of 4′-ethynyl D4T or D4T, a concentration which is much lower than the K_ms of the two nucleoside analogs towards TK1, the TK1 enzyme in CEM cells might preferentially phosphorylate 4′-ethynyl D4T over D4T to a degree of higher than a fourfold difference. The actual intracellular concentration of pyrimidine nucleosides is also affected by the catabolic enzymes. Thymidine phosphorylase is the major cellular enzyme responsible for the catabolism of thymidine and thymidine analogs. Using a partially purified preparation of human liver thymidine phosphorylase, we previously demonstrated that 4′-ethynyl D4T was much more resistant to thymidine phosphorylase than D4T was (6). Taken together, a nearly 100-fold-higher concentration of 4′-ethynyl D4TMP than of D4TMP was thus formed intracellularly and might therefore contribute to a higher concentration of 4′-ethynyl D4TTP than of D4TTP in CEM cells.

There was a discrepancy between our in vitro enzyme assays and the intracellular metabolism experiment, i.e., while the in vitro enzyme assays showed that the phosphorylation efficiencies of 4′-ethynyl D4TDP were much poorer than those of D4TDP, the intracellular metabolism in CEM cell showed that the TP/DP ratio of 4′-ethynyl D4T was sixfold higher than that of D4T (Fig. 1). The actual mechanism underlying this discrepancy is currently unknown. We speculate that this may be due to either a more efficient phosphorylation of 4′-ethynyl D4TDP or faster catabolism of D4TTP by undefined cellular mechanisms.

There are at least three classes of enzymes capable of phosphorylating 4′-ethynyl D4TDP to its TP metabolites. The in vitro enzyme assays using the crude extract from CEM cells indicated that PGK appeared to be the key enzyme. Intriguingly, our in vitro assay of PGK activity revealed that recombinant PGK and PGK from crude cellular extract behaved differently, with a less efficient phosphorylation of 4′-ethynyl D4T by the former. Likewise, the relative efficiency of phosphorylating D4TDP versus AZTDP by PGK was not comparable between recombinant protein and enzyme preparation from crude cellular extract. This could be due to the interaction of PGK with other cellular proteins that results in the change of enzyme behaviors or a posttranslational modification of PGK. Further studies are warranted.

The catalytic mechanisms and structure-activity relationship towards several antiviral thymidine analog DPs for nm23-H1 and nm23-H2 have been studied in detail before (27-29). It has been shown that NDPK can phosphorylate AZTDP and D4TDP with efficiencies at least 10⁴-fold and 10³-fold less than that for naturally occurring TDP, respectively (27-29). It is believed that the 3′-OH of the nucleotide sugar is involved in a network of hydrogen bonds with several residues in the active site and provides the oxygen bridging β and γ phosphates of the nucleotides, and is therefore crucial for the phosphorylation activity of NDPK enzymes, such as nm23-H1 or -H2 (27, 28). For AZT, the lack of 3′-OH and the presence of a 3′-azido group, which disturbs the interacting hydrogen bond network, significantly affects the phosphorylation efficiency. On the other hand, the double bond on the sugar moiety of D4T in some way counteracts the detrimental effect of lacking 3′-OH and results in a better phosphorylation efficiency for D4TDP than those for AZTDP and other dideoxynucleoside analog diphosphates (29). While our data about the phosphorylation of AZTDP and D4TDP by nm23-NDPK is generally comparable to previous reports, we show that the phosphorylation efficiency of 4′-ethynyl D4TDP is about 15-fold less than that of D4TDP and is similar to that of AZTDP. This indicates that the 4′-ethynyl substitute could impair the phosphorylation activity of NDPK enzymes by counteracting the beneficial effect from the double bond between 2′ and 3′ positions in D4T.

In summary, 4′-ethynyl D4T metabolites have unique behaviors toward their phosphorylation enzymes in cells. The profile of phosphorylated metabolites of 4′-ethynyl D4T in T-lymphoid cells is more similar to that of AZT than to that of the progenitor D4T. The rate-limiting step for the intracellular metabolism of 4′-ethynyl D4T could be at TMP kinase. Further studies of the metabolism of 4′-ethynyl D4T are needed.