Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

AGT is a DNA repair protein that acts on a variety of adducts at the O⁶-position of guanine including methyl-, ethyl-, benzyl-, 2-chloroethyl-, and pyridyloxobutyl- (Mitra and Kaina, 1993; Pegg et al., 1995a; Pegg, 2000). Repair is brought about in a direct single reaction in which the adduct is transferred to a cysteine acceptor site. The S-alkylcysteine is not regenerated, and the alkylated form of the protein is ubiquitinated and rapidly degraded. Thus, substrates of the AGT protein are also inactivators.

Repair of either O⁶-methylguanine (m⁶G) or O⁶-(2-chloroethyl)guanine by AGT protects cells from the cytotoxicity of methylating and chloroethylating agents, respectively (Erickson et al., 1980; Margison et al., 1996; Ludlum, 1997; Pegg et al., 2000). Therefore, attempts are being made to use inactivators of the AGT protein to increase the sensitivity of tumor cells to these agents. The most widely studied compound that is being used for this purpose is O⁶-benzylguanine (b⁶G) (Dolan et al., 1990; Dolan and Pegg, 1997; Kreklau et al., 1999; Pegg et al., 2000), although other agents that act similarly, such as O⁶-benzyl-2'-deoxyguanosine, O⁶-benzyl-N²-acetylguanosine, and O⁶-(5-bromothenyl)guanine, are also being pursued (Marathi et al., 1994; Kokkinakis et al., 2000; Middleton et al., 2000; Pegg et al., 2000). b⁶G is currently undergoing clinical trials, but its use may be limited by its poor solubility, relatively low potency, and inability to inactivate mutant forms of AGT that contain single point mutations rendering a high level of resistance to it (Pegg et al., 2000; Xu-Welliver and Pegg, 2000).

b⁶G is recognized as a substrate by AGT and is acted upon by its formation of free guanine and S-benzylcysteine at the acceptor site of the protein leading to irreversible inactivation of the AGT (Pegg et al., 1995a; Pegg, 2000). The reaction of AGT with b⁶G is much slower than the reaction with methylated DNA. This is likely to be due to the weak binding of b⁶G to the active site of the AGT protein since it lacks the ability to interact with the DNA binding domain of AGT. We have shown earlier that relatively short oligodeoxyribonucleotides that contain a b⁶G residue are much better substrates and effective inhibitors of both wild-type and the mutant human AGT proteins than the b⁶G free base (Goodtzova et al., 1997; Pegg et al., 1998). This is consistent with the concept that the interactions with the other components of the oligodeoxyribonucleotide lead to a more efficient transfer of the benzyl group to the AGT protein. Our previous studies showed that oligodeoxyribonucleotides as short as 7-mers containing a central b⁶G have excellent solubility in aqueous media and were at least 20 times more effective in inactivation of wild-type AGT than the free base and were 50- to 1000-fold more effective in inactivation of the b⁶G-resistant mutants P140A and G156A. Such oligodeoxyribonucleotides therefore provide an opportunity to produce more useful AGT inhibitors but only if the obvious problems relating to cellular uptake and metabolic stability can be solved. Nuclease-mediated degradation is well known to occur very readily with oligodeoxyribonucleotides (Crooke, 1998; Agrawal, 1999). Such degradation is greatly reduced by modification of the phosphodiester linkages (Miller et al., 1981, 2000). We have therefore investigated whether the replacement of the terminal linkages in 11-mer oligodeoxyribonucleotides containing b⁶G with methylphosphonates is compatible with the ability to serve as a substrate for AGT and allows the 11-mer to be used to sensitize HT29 cells to killing by the chloroethylating agent BCNU.

In our study, we used 11-mers containing either b⁶G or m⁶G with methylphosphonate linkages to both the 5' and the 3' terminal nucleosides. We compared the ability of these oligodeoxynucleotides to serve as substrates for AGT. It was found that 5'-d(TmpGTGAb⁶GCTGTmpG)-3' (11-mpBG) was an extremely potent inactivator of both wild-type and mutant b⁶G-resistant alkyltransferases and was able to inactive AGT in HT29 cells and render these cells susceptible to BCNU. Despite being able to inactivate AGT in vitro, the corresponding 11-mers lacking the methylphosphonates (11-BG) or containing m⁶G instead of b⁶G (11-mpMG) were ineffective in imparting sensitivity to BCNU in HT29 cells. These results suggest that derivatives of 11-mpBG may be clinically useful AGT inactivators if modifications enhancing uptake and stability can be made.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. Ampicillin, isopropyl -D-thiogalactopyranoside, and most of the other chemicals were purchased from Sigma (St. Louis, MO). Restriction enzymes were obtained from New England Biolabs (Beverly, MA). BCNU was obtained from the Drug Synthesis and Chemistry Branch, Division of Cancer Treatment, National Cancer Institute.

AGT Inhibitors. The synthesis of b⁶G (Dolan et al., 1990), 5'-d(AACAGCCATATb⁶GGCCC)-3' (Pauly et al., 1991), and 5'-d(TGTGAb⁶GCTGTG)-3' (11-BG) (Pauly et al., 1988; Pegg et al., 1998) were described previously. The 5'-d(TGTGAm⁶GCTGTG)-3' (11-MG) was synthesized in the Macromolecular Core Facility, Hershey Medical Center, by using a Milligen 7500 DNA synthesizer (Milligen/Biosearch, Burlington, MA). An impurity in this material, which did not react with AGT or affect the reaction with AGT, was not removed and can be seen on chromatography in Fig. 5 (peak X).

Oligodeoxyribonucleotide Synthesis and Purification. Previously undescribed oligodeoxyribonucleotides were synthesized on a 10-µmol scale using an Applied Biosystems, Inc. (Foster City, CA) model 394 DNA/RNA synthesizer. The O⁶-substituted guanine 2'-deoxyribonucleoside phosphoramidites were prepared as previously described (Pauly et al., 1988). All other DNA synthesis reagents were from Glenn Research (Sterling, VA). The standard Applied Biosystems, Inc. 10-µmol synthesis cycle was used except that methylphosphonates were allowed to couple for an additional 6 min, and O⁶-substituted guanine-containing phosphoramidites were allowed to couple for an additional 15 min. At the end of synthesis, the 4,4'-dimethoxytrityl (DMT) protecting group was not removed from the 5'-end of the oligodeoxyribonucleotides. Oligodeoxyribonucleotides were cleaved from the solid support by standard ammonium hydroxide treatment followed by the immediate removal of ammonium hydroxide under vacuum. Removal of the protecting groups was accomplished using a modification of a published method (Polushin et al., 1994). Briefly, oligodeoxyribonucleotides were exposed to 3.3 ml of a solution of hydrazine, ethanolamine, and methanol (1:5:5, v/v/v) for 1.5 h at room temperature. The solution was then treated with 1.18 ml of glacial acetic acid in 5 ml of water until the pH of the aqueous solution reached 7.5. At this point the solution became cloudy and it was stored at 20°C overnight. Centrifugation of the cold suspension produced a pellet. The supernatant was decanted, and the pellet was redissolved in 5 ml of 0.1 M triethylammonium acetate, pH 7, containing 10% acetonitrile by volume. The 5'-O-DMT-containing oligodeoxyribonucleotides were purified by HPLC on a 10 mm × 25 cm Luna column (Phenomenex, Torrance, CA). The solvents were 0.1 M triethylammonium acetate, pH 7 (A), and acetonitrile (B). Column elution was carried out using a linear gradient of 10 to 40% B over 60 min at a flow rate of 3 ml/min. UV absorbance was monitored at 270 nm. The m⁶G-containing oligodeoxyribonucleotide methylphosphonates were recovered in two peaks at 46 and 50 min that were designated DMT-11-mpMG-1 and DMT-11-mpMG-2, respectively. The O⁶-benzylguanine-containing oligodeoxyribonucleotide methylphosphonates chromatographed as two peaks at 48 and 52 min, and these were designated DMT-11-mpBG-1 and DMT-11-mpBG-2, respectively. The resolution of these oligodeoxyribonucleotides into two peaks was a consequence of the proximity of the bulky DMT group to the chiral methylphosphonate linkage at the 5'-end of each oligodeoxyribonucleotide. The solutions for these various samples were processed separately. Each was evaporated to dryness, and the resulting oligodeoxyribonucleotides were detritylated by treatment with acetic acid/water (8:2, v/v) for 15 min followed by coevaporation with excess ethanol under vacuum. The detritylated oligodeoxyribonucleotides were purified by HPLC using a gradient of 5 to 40% B over 60 min at 3 ml/min. The resulting 11-mpMG-1 and 11-mpMG-2 eluted at 22 min. Oligodeoxyribonucleotides 11-mpBG-1 and 11-mpBG-2 eluted at 25 min. After removal of the 5'-O-DMT group, the stereoisomeric methylphosphonate-containing oligodeoxyribonucleotides were not resolved under these preparative chromatographic conditions. However, many were resolved under the analytical chromatographic conditions described below.

Construction of Plasmid and Expression of b⁶G-Resistant Mutant Alkyltransferases. The wild-type human AGT and O⁶-benzylguanine-resistant mutants (P140A, P140K, G156A, Y158H, G160R, K165A) were made by inserting the relevant cDNA into the pQE30 vector from Qiagen (Chatsworth, CA) for the expression of the recombinant protein and purified to homogeneity by immobilized metal affinity chromatography as described earlier (Xu-Welliver et al., 1998, 1999, 2000).

Inactivation of AGT in Vitro. The purified wild-type or mutant AGT proteins were incubated with different concentrations of the potential inhibitors in 0.1 ml of reaction buffer (50 mM Tris-HCl, pH 7.6, 0.1 mM EDTA, 5.0 mM dithiothreitol) in the presence of 10 µg of hemocyanin for 30 min at 37°C. The remaining AGT activity was determined after incubation with [³H]methylated calf thymus DNA substrate for 30 min at 37°C as described previously (Xu-Welliver et al., 1998). The concentration of inhibitor which led to a 50% loss of AGT activity (ED₅₀) was calculated from graphs where percentage of remaining AGT activity was plotted against inhibitor concentration.

Separation of Oligodeoxyribonucleotides by HPLC. The 11-mer oligodeoxyribonucleotides containing O⁶-adducts were incubated with different amounts of wild-type AGT in 50 mM Tris-HCl, pH 7.6, 0.1 mM EDTA, and 0.5 mM dithiothreitol for 10 min at 37°C. The reaction was then stopped by the addition of 1% (final concentration) sodium dodecyl sulfate, and the mixture of oligodeoxyribonucleotides was separated on a reverse-phase C18 5-µm Ultrasphere ODS column (Beckman, Palo Alto, CA), and oligonucleotides were detected by UV at 254 nm as described earlier (Goodtzova et al., 1997; Pegg et al., 1998). Separation was carried out at 45°C using a linear gradient of 14 to 50% methanol in 50 mM sodium phosphate buffer (pH 6.3) over 60 min with a flow rate of 2 ml/min. In most cases, the oligodeoxyribonucleotides containing methylphosphonate linkages could be resolved into multiple peaks corresponding to the respective stereoisomers. The exact starting point of the gradient was varied slightly according to the experiment. This accounts for the slightly different retention times in Figs. 1 to 3 and 5, but in all cases all samples from a particular experiment were run under identical conditions.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Repair of 11-mpBG by wild-type AGT. After incubation with the amount of AGT protein shown for 10 min at 37°C, 11-mpBG and the products were separated by HPLC on a reverse-phase C18 Ultrasphere ODS column as detailed in the text. The left-hand panels show repair of 11-mpBG-1 (2.4 nmol), and the right-hand panels show repair of 11-mpBG-2 (2.4 nmol). The nonalkylated product peaks of 11-mpG were resolved by the chromatography into two isomers that are designated as 11-mpG-1a and -1b and 11-mpG-2a and -2b as shown.

Cell Culture, Inactivation of Cellular AGT, and Cytotoxicity Assays. HT29 cells were grown in Dulbecco's modified Eagle's medium containing 36 mM NaHCO₃ supplemented with 10% fetal bovine serum plus 3% glutamine and 50 µg/ml gentamycin (Pegg et al., 1995b). The loss of AGT activity after oligodeoxyribonucleotide addition was measured in HT29 cells that were plated at 1.5 × 10⁶ cells/100-mm dish in 10 ml of medium and grown at 37°C to 80% of confluence. The medium was replaced with 5 ml of fresh medium containing the oligodeoxyribonucleotide concentrations described in the figure legends, and cells were incubated for the times indicated. Cells were then harvested, and the AGT activity was assayed as previously described (Dolan et al., 1990).

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Ability of Wild-Type AGT to React with Oligodeoxyribonucleotides Modified with Terminal Methylphosphonate Linkages. Since introduction of a substituent on the phosphate linkages generates a new center of chirality, the 11-mers containing a m⁶G (11-mpMG) or a b⁶G (11-mpBG) each exist as four stereoisomers. Purification after synthesis of the 5'-O-DMT-containing oligodeoxyribonucleotides was able to separate both 11-mpBG and 11-mpMG into two mixtures of stereoisomers that are arbitrarily designated 1 and 2. This separation was a consequence of the proximity of the DMT group to the chiral methylphosphonate linkage at the 5'-end of each oligodeoxyribonucleotide.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Repair of 11-mpMG by wild-type AGT. After incubation with the amount of AGT protein shown for 10 min at 37°C, 11-mpMG and the products were separated as described in the legend to Fig. 1. The left-hand panels show repair of 11-mpMG-1 (1.4 nmol), and the right-hand panels show repair of 11-mpMG-2 (1.4 nmol). The HPLC separation was able to partially resolve both 11-mpMG-1 and 11-mpMG-2 into the two isomers, which are designated as 11-mpMG-1a and -1b and 11-mpMG-2a and -2b as shown. The product peaks are labeled as in Fig. 1.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Competition for repair by wild-type AGT of m6G and b6G present in 11-mer methylphosphonates. Various amounts of the AGT protein as indicated were incubated with an equimolar mixture (1.4 nmol) of 11-mpBG-1 and 11-mpMG-1 for 10 min at 37°C, and the formation of 11-mpG was measured by HPLC as in Fig. 1. As in Figs. 1 and 2, 11-mpMG-1 and 11-mpG could be resolved into two peaks, which are designated as a and b. The substrate and product peaks are labeled as in Figs. 1 and 2.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   The time course of repair of 11-mpMG-1 and 11-mpBG-1 by AGT. The AGT protein was incubated with either 11-mpMG-1 or 11-mpBG-1 as shown, and the extent of repair was determined by HPLC as in Figs. 1 and 2.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Competition for repair by wild-type AGT of m6G present in 11-mers with and without terminal methylphosphonate linkages. Various amounts of the AGT protein as indicated were incubated with an equimolar mixture (1.4 nmol) of 11-mpMG-1 and 11-MG (left panels) or with an equimolar mixture (1.4 nmol) of 11-mpMG-2 and 11-MG (right panels) for 10 min at 37°C, and the formation of 11-mpG was measured by HPLC as in Fig. 2. The peak marked X represents an impurity in the 11-MG preparation that was not affected by AGT. As in Fig. 2, 11-mpMG-1, 11-mpMG-2, 11-mpG-1, and 11-mpG-2 were partially separated by the HPLC into the two isomers, which are designated as a and b, respectively. The substrate and product peaks are labeled as in Figs. 1 and 2.

Inactivation of b⁶G-Resistant Mutant AGTs. A number of point mutations in the AGT sequence have been identified that lead to resistance to b⁶G (Xu-Welliver et al., 1998, 1999, 2000). A series of such mutant proteins were examined for their ability to be inactivated by either 11-mpBG-1 or the 16-mer 5'-d(AACAGCCATATb⁶GGCCC)-3' (Table 3). There was very little difference in the ED₅₀ values for these two oligodeoxyribonucleotides with only mutants Y158H and P140K showing a slight preference for inactivation by the 16-mer. More importantly, both of these oligodeoxyribonucleotides were much more potent inactivators of the mutant forms of AGT than was b⁶G itself by factors that ranged from >900 to >80,000. The effect was greatest with the most b⁶G-resistant mutants, Y158H and P140K, where the advantage of incorporation of b⁶G into 11-mpBG-1 was more than 10,000-fold (Table 3). Even with mutant P140K, which is totally impervious to free base b⁶G (Xu-Welliver et al., 1998), the ED₅₀ value for 11-mpBG-1 was only 0.15 µM.

Inactivation of AGT in HT29 Cells. The addition of 11-mpBG-1 to the culture medium of HT29 human colon carcinoma cells led to a loss of AGT activity in a dose- and time-dependent manner (Fig. 6). Approximately 7 µM 11-mpBG-1 was needed to achieve 50% inhibition in 4 h (Fig. 6A), but maximal inhibition was not reached until more than 12 h after exposure to 5 µM 11-mpBG-1 and complete inactivation was maintained for at least a further 60 h (Fig. 6B). These results suggest that uptake of the 11-mpBG-1 does occur but is relatively slow. Similar results (not shown) were obtained with 11-mpBG-2. No inactivation of AGT in HT29 cells was achieved with either the oligodeoxyribonucleotides 11-mpMG-1 (Fig. 6A) and 11-mpMG-2 (not shown), which contain m⁶G even at levels of 20 µM.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Inactivation of AGT in HT29 cells by 11-mpBG-1. A, the effect of addition of the various concentrations of 11-mpMG-1 or 11-mpBG-1 to the culture medium of HT29 cells for 4 h. The remaining AGT activity was then calculated. B, time course of the inactivation of AGT in HT29 cells following addition of 5 µM 11-mpBG-1 to the culture medium.

Ability of 11-mpBG to Sensitize HT29 Cells to Killing by BCNU. To test whether pretreatment with 11-mpBG-2 is able to enhance the cytotoxicity of BCNU, HT29 cells were preincubated with different concentrations of 11-mpBG-2, 11-mpMG-2, or 11-BG for 14 h. The cells were then treated with 40 µM BCNU for 2 h. This dose of BCNU alone did not alter survival of HT29 cells. As shown in Fig. 7, 11-mpBG-2 increased killing by BCNU with more than a 500-fold increase at 5 µM. In these experiments the medium that was added to the cells after the 2-h exposure to BCNU also contained the respective oligodeoxyribonucleotide, but this second addition was probably not necessary since a repetition of the study carried out at 5 µM 11-mpBG-2 without a second addition gave a similar increase in the cytotoxicity of BCNU (Fig. 7). Exposure to 11-mpBG-1 had a similar effect to that observed with 11-mpBG-2. Both 11-mpMG-2 and 11-BG had much less of an effect on BCNU toxicity with at most a 2-fold reduction in survival after 10 µM (results not shown).

View larger version (13K): [in this window] [in a new window]   Fig. 7.   Effect of BCNU on survival of HT29 cells treated with 11-mpBG-2. HT29 cells were treated with the concentrations of 11-mpBG-2 shown 14 h before exposure to BCNU as described under Experimental Procedures, and the cell survival was measured using a colony forming assay (open circles). The solid circle shows results for a second experiment in which the 11-mpBG-2 was not replaced in the medium after removing the BCNU.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Recent studies of the crystal structure of the protein and its benzylated form have provided plausible models for the binding of free base b⁶G and of alkylated DNA substrates (Daniels et al., 2000; Wibley et al., 2000). These show that b⁶G is held at the active site largely by hydrophobic interactions with the benzyl group and some interactions with the purine base, whereas the interaction of AGT with alkylated DNA also makes a substantial number of contacts with the surrounding DNA and flips the alkylated base into the binding pocket. It is also probable that there is a small change in the structure of the protein upon binding DNA that facilitates the alkyl transfer reaction. These factors produce a much higher affinity of AGT for the DNA substrate, improve the chances of its binding in a productive manner for reaction, and thus accelerate the reaction. This explains well in molecular terms why short oligodeoxyribonucleotides are much more potent inactivators of AGT than the free base.

At present, the knowledge of the binding of oligodeoxyribonucleotides to AGT is based only on modeling studies since cocrystals of the protein and DNA have not been achieved. However, these models are highly plausible since the AGT structure is very similar to that of helix-turn-helix-containing proteins such as Escherichia coli catabolite gene activator protein and Mu transposase for which definite structures of protein:DNA are available (Daniels et al., 2000; Wibley et al., 2000). The negatively charged DNA phosphodiester backbone matches a complementary positively charged surface on the AGT centered around Arg-128. It was therefore quite possible that the incorporation of methylphosphonate linkages into such short oligodeoxyribonucleotides was sufficient to block or distort binding to AGT. However, our results show clearly that this is not the case and that the ability of human AGT to interact with and repair O⁶-alkylguanine adducts is not affected adversely by the presence of methylphosphonate linkages to the 5' and 3' terminal residues of 11-mers as tested here. It is also apparent from the studies shown in Figs. 1, 2, and 5 that the presence of different stereoisomers in these linkages does not alter the ability to serve as AGT substrates since all four of the possible isomer products are formed to the same extent. Since the AGT reaction is not catalytic and a single repair event irreversibly inactivates the protein, 11-mers such as 11-mpBG are therefore useful potential inactivators of AGT. This fact is supported by the results in Table 2, which shows that 11-mpBG has a much lower ED₅₀ value for the inactivation of AGT than free b⁶G, which is currently undergoing clinical trials that have indicated it is able to reduce AGT levels in tumors (Dolan et al., 1998; Spiro et al., 1999; Friedman et al., 2000).

An additional potential advantage of using 11-mpBG over b⁶G itself is that the former was able to inactivate mutant forms of the AGT protein that are highly resistant to b⁶G. Detailed studies have indicated that there are many sites at which such resistance can be produced by point mutations (Xu-Welliver et al., 1998, 1999; Xu-Welliver and Pegg, 2000). Recent publications in which the crystal structure of AGT has been solved have provided plausible models of the manner by which free base b⁶G and DNA substrates are bound (Daniels et al., 2000; Wibley et al., 2000). These models show that the binding of b⁶G in the active site pocket occurs via a limited number of interactions with a hydrophobic binding surface. In contrast, O⁶-alkylguanine nucleosides in DNA are "flipped out" of the DNA helix by a concerted attack of residues interacting with the surrounding nucleotides and the residue Arg-128, which replaces the displaced base. These interactions are likely to provide a much stronger binding of an oligodeoxyribonucleotide such as 11-mpBG compared with b⁶G and to a large extent overcome the effect of point mutations such as P140K and Y158H, which strongly interfere with the binding of the free base. Thus, 11-mpBG-1 was able to inactivate all of the b⁶G-resistant mutants with ED₅₀ values of less that 0.2 µM. Although the amount of 11-mpBG that is needed to inactivate these mutants is greater than for wild-type AGT with an increase in the ED₅₀ values of 40- to 100-fold (Table 3), these differences are much less than for b⁶G itself where these mutations produce an increase in ED₅₀ of more than 1,500- to >30,000-fold.

Furthermore, as would be expected from the previous results showing that AGT has a significant preference for the repair of benzyl groups rather than methyl groups (Goodtzova et al., 1997; Pegg et al., 1998), which are confirmed by the competition experiments shown in Fig. 3, it was found that the presence of an O⁶-benzyl group is preferable to an O⁶-methyl group for the rapid inactivation of AGT. The lower ED₅₀ value for 11-mpBG compared with 11-mpMG (Table 2) is consistent with this. An even more compelling reason for the use of oligodeoxyribonucleotides containing b⁶G rather than m⁶G residues is that only 11-mpBG was able to inactivate AGT in HT29 cells (Fig. 6) and to abolish the resistance to BCNU provided by the AGT activity (Fig. 7). In addition to the greater potency for AGT inactivation, the presence of the benzyl group is likely to increase the uptake of the oligodeoxyribonucleotide into the cell. Based on many experiments using oligodeoxyribonucleotides as potential therapeutics, uptake is likely to be the limiting factor in the use of such compounds as drugs (Crooke, 1998; Stein, 1999; Lebedeva et al., 2000). When the results of addition of 11-mpBG to cells are compared with the results with b⁶G as a free base, there is a striking loss of potency of the former, and the time taken to achieve maximal inactivation of AGT is much longer. With b⁶G, which passes very well through mammalian cell membranes, inactivation of AGT in HT29 cells occurs within a few minutes, and the ED₅₀ value for loss of AGT activity at 4 h is similar to the ED₅₀ value for inactivation of AGT in vitro (Dolan et al., 1990, 1991). In contrast, with 11-mpBG the ED₅₀ value for reduction of AGT in HT29 cells is 1000-fold greater than with purified enzyme, and exposure of the cells for at least 12 h was needed to achieve a maximal effect. 11-mpBG-1 and 11-mpBG-2 were equally effective in reducing AGT activity and abolishing BCNU resistance in HT29 cells showing that the different isomers do not have altered uptake.

It is probable that the methylphosphonate modifications do provide significant stabilization of the oligodeoxyribonucleotides by preventing exonuclease digestion since the inactivation of AGT by 11-mpBG in the HT29 cell cultures was maintained over a 72-h period (Fig. 6B) and 11-BG was ineffective in producing sensitization to BCNU. The use of methylphosphonates to provide metabolically stable antisense oligodeoxyribonucleotides with potentially useful pharmacological activity is now well established (Miller et al., 2000), but the mechanisms by which such oligodeoxyribonucleotides are taken up by the cells, interact with plasma components, and accumulate within a cell are poorly understood. At low concentrations a receptor-like mechanism may predominate (Agarwal and Gewirtz, 1999), while at higher concentrations, some studies suggest that they are mainly taken up by a fluid phase process (Alama et al., 1997), while others suggest that methylphosphonates penetrate cell membrane by passive diffusion (Baertschi, 1994). It is probable that the presence of the benzyl group in 11-mpBG facilitates such uptake since 11-mpMG was ineffective in the cell cultures tested. We therefore conclude that such relatively stable 11-mers can be accumulated in tumor cells at levels sufficient to inactivate AGT activity and render them sensitive to BCNU. However, it is apparent that additional modifications or different protocols of administration will be needed to take better advantage of the potential of 11-mpBG and related compounds shown by the experiments with purified wild-type and mutant AGTs in vitro. If these problems can be solved using modifications that retain the excellent ability to serve as substrates and inactivate AGT and its b⁶G-resistant mutants, the resulting compounds would provide a major advance in the design of inactivators of this important cause of tumor cell resistance to therapy with alkylating agents. Future studies with cells expressing mutant forms of AGT and treatments with various alkylating agents in addition to BCNU that have been shown to be enhanced by the elimination of AGT-mediated repair will be helpful in demonstrating the potential of this approach.