Methotrexate enhances the antianabolic and antiproliferative effects of 5-aminoimidazole-4-carboxamide riboside

Malignant cells typically show a high rate of anabolic processes. They frequently overexpress lipogenic enzymes, such as fatty acid synthase, show enhanced protein biosynthesis, and are more active in synthesizing DNA. All these processes require energy; hence, interference with the cellular mechanisms governing energy homeostasis was proposed recently as a novel strategy in the battle against cancer (1–3). Key target in this strategy is the AMP-activated protein kinase (AMPK), a cellular fuel gauge that is exquisitely sensitive to changes in the levels of the low energy indicator AMP (4–6). On binding of AMP, AMPK phosphorylates several downstream targets, including key anabolic enzymes, inhibiting their activity and this way switching off major energy consuming pathways. In a recent attempt to explore the feasibility of targeting the energy sensing system as a strategy in the treatment of cancer, cancer cells were treated with 5-aminoimidazole-4-carboxamide (AICA) riboside, a cell-permeable nucleoside (1–3). On cell entry, AICA riboside is phosphorylated by adenosine kinase to AICA ribotide, also known as ZMP, which mimics the effects of AMP in activating AMPK (6, 7). In cultures of cancer cells, the AICA riboside–induced increase of ZMP levels blocks tumor-associated anabolism, attenuates tumor cell proliferation, impedes cell invasiveness, and inhibits the ability of cancer cells to form colonies (1–3). One important shortcoming of AICA riboside is the large amount of drug that is needed to exert effects. This problem is related in part to the rapid metabolism of ZMP entering into the de novo purine synthesis pathway (8) and to the inhibitory influence of ATP on the activation of AMPK (9). Here, reasoning that these limitations could be overcome by inhibition of the de novo purine synthesis, we cotreated cancer cells with AICA riboside and methotrexate (4-amino-10-methylpteroylglutamic acid). Methotrexate is an antimetabolite that already for several decades is being used clinically in the treatment of a variety of malignancies (10). It exerts cytotoxic effects through direct inhibition of dihydrofolate reductase and inactivation of folate-dependent enzymes, including glycinamide ribonucleotide transformylase (GART) and AICA ribotide transformylase/IMP cyclohydrolase (ATIC), two key enzymes of the de novo purine biosynthesis pathway, responsible for the conversion of glycinamide ribonucleotide to formyl-glycinamide ribonucleotide and of ZMP to IMP, respectively, in the final steps of the de novo purine synthesis (11, 12). The major inhibitory effect of methotrexate is due to depletion of 10-formyltetrahydrofolate, the one-carbon donor for both of these reactions. In addition, methotrexate polyglutamate derivatives that accumulate in cells are the direct inhibitors of both ATIC and GART, but they most potently inhibit ATIC (13–15). The net result is an inhibition of the de novo purine biosynthesis leading to a depletion of ATP and to a lesser extent of GTP (15, 16). In contrast to normal cells, which prefer the purine salvage pathway, rapidly dividing tumor cells rely on the de novo pathways explaining part of the chemotherapeutic action of methotrexate. Here, we show that methotrexate, by blocking the de novo purine biosynthesis pathway, enhances the ability of exogenous AICA riboside to increase the levels of ZMP and leads to a decrease in the levels of ATP. Consequently, methotrexate sensitizes AMPK for activation by AICA riboside and enhances the AICA riboside–induced inhibition of tumor-associated anabolism and colony formation.

MDA-MB-231 and A-431 cells were obtained from the American Type Culture Collection (Manassas, VA) and maintained at 37° C in a humidified incubator with a 5% CO₂/95% air atmosphere (MDA-MB-231) or a 10% CO₂/90% air atmosphere (A-431) in RPMI 1640 (MDA-MB-231) or DMEM (A-431) supplemented with 10% FCS (Invitrogen, Carlsbad, CA). PC-3M-luc-C6 cells were obtained from Xenogen Corp. (Alameda, CA). These cells were maintained at 37°C in a 5% CO₂/95% air atmosphere in MEM supplemented with 10% FCS, 1% sodium pyruvate (Invitrogen), 1% MEM vitamin solution (Invitrogen), and 1% MEM nonessential amino acids (Invitrogen).

Methotrexate and albumin were purchased from Sigma (St. Louis, MO). Methotrexate-albumin was prepared as described by Stehle et al. (17).

Cells were plated on day 0. On day 1, they were treated with methotrexate, methotrexate-albumin, or vehicle. One day later, cultures received different concentrations of AICA riboside. After 2 hours of incubation, cultures were washed twice with PBS, harvested, and centrifuged. ZMP was measured by quantification of the m/z 127 product ion by tandem mass spectrometry on a Finnigan LTQ Linear Ion Trap mass spectrometer (Thermo Electron Corporations, San Jose, CA). ATP levels were determined using the luciferase/luciferin-based ENLITEN ATP assay system (Promega, Madison, WI).

Cells treated with different concentrations of methotrexate, methotrexate-albumin, and/or AICA riboside were washed with PBS and lysed in a reducing NuPage sample loading buffer (Invitrogen). Equal amounts of protein were separated on NuPage gels (Invitrogen) and processed for immunoblot analysis for AMPKα, phosphorylated Thr¹⁷² AMPKα, phosphorylated Ser⁷⁹ acetyl-CoA carboxylase, and cytokeratin-18 as described (1).

Cells were treated with methotrexate, methotrexate-albumin, and/or AICA riboside as indicated. One hour after addition of AICA riboside (or vehicle), 2-[¹⁴C]-labeled acetate (2 μCi/6-cm dish; 57 mCi/mmol; Amersham Biosciences, Piscataway, NJ) was added to the cultures. Four hours later, cells were washed with PBS, scraped, and resuspended in 0.9 mL PBS. Lipids were extracted using the Bligh-Dyer method as described previously (18). Incorporation of ¹⁴C into lipids was measured by scintillation counting.

Cells were treated with methotrexate, methotrexate-albumin, and/or AICA riboside as described. DNA synthesis was measured using the bromodeoxyuridine (BrdUrd) labeling and detection kit III (Roche Applied Science, Hague Road, IN).

The combination index (CI) method of Chou and Talalay (19) was used to analyze the nature of the pharmacologic interaction between methotrexate and AICA riboside. Synergism is defined as more than the expected additive effect with CI < 1, whereas CI > 1 indicates antagonism (19, 20).

Cells were seeded at a density of 5 × 10³ per 6-cm dish and treated as indicated. Eight days after initial treatment, cultures were fixed with 4% formaldehyde in PBS and stained with a 0.5% crystal violet solution in 25% methanol. Plates were photographed with a Nikon Coolpix digital camera (Nikon Corp., Tokyo, Japan).

The day before transfection, cells were plated at a density of 5 × 10⁵ per 6-cm dish. Transfection of cells with Stealth small interfering RNA (siRNA) using LipofectAMINE 2000 (Invitrogen) was carried out according to the manufacturer's protocol. Stealth siRNA molecules targeting ATIC (5′-CCUGCAAUCUCUAUCCCUUUGUAAA-3′) and GART (5′-GGCAGCUGGAGAAACAAUUGUCAUU-3′) and a nonspecific control double-stranded RNA, designed by scrambling the nucleotide sequence of the ATIC siRNA (5′-CCUAACUCUAUCUCCUUGUUGGAAA-3′), were purchased from Invitrogen. Twenty-four hours after transfection, cells were trypsinized and reseeded at a density of 5 × 10³ per dish for colony formation assay.

A 556-bp cDNA probe for ATIC and a 620-bp cDNA probe for GART were synthesized by PCR on human cDNA (generated by reverse transcription as described previously; ref. 15). For the ATIC probe, the primers were 5′-CCAGCTGTACACACTGCAGCC-3′ (forward) and 5′-TTATGACTTGTCGACGACACC-3′ (reverse). The GART probe was synthesized using 5′-AGGCTCAAGCTCTAGGACTGG-3′ and 5′-TCCAAGGAGAGCACATCCAGC-3′ as forward and reverse primer, respectively. Probes for 18S rRNA, RNA preparation, and Northern blot procedures have been described previously (21).

Statistical analyses were done using one-way ANOVA with Tukey's multiple comparison test.

To examine the ability of methotrexate to enhance the effects of AICA riboside, three human cancer cell lines were used, MDA-MB-231 breast cancer cells, PC-3M-luc-C6 prostate cancer cells, and A-431 epidermoid carcinoma cells. MDA-MB-231 cells are the most thoroughly characterized in terms of their response to AICA riboside (1). Like many clinical cancers, they are resistant to native methotrexate due to transport deficiencies caused by an epigenetic alteration that down-regulates the expression of the reduced folate carrier (22, 23). Hence, in this study, MDA-MB-231 cells were treated with methotrexate bound to albumin (17, 24, 25), which enters the cells by albumin-mediated endocytosis (25) and circumvents the methotrexate transport deficiency of MDA-MB-231 cells. The other cell lines were treated with free methotrexate. Sixteen hours after treatment with methotrexate or methotrexate-albumin, cells were treated with AICA riboside and, 2 hours later, intracellular ZMP accumulation was measured. After treatment with AICA riboside alone, a measurable increase in intracellular ZMP was found at AICA riboside concentrations starting from 0.250 mmol/L (P < 0.05 compared with control; Fig. 1A). On pretreatment with methotrexate-albumin, increases in the levels of ZMP were noticeable at AICA riboside concentrations as low as 0.015 mmol/L. Similar effects were observed in PC-3M-luc-C6 and A-431 cells where, at an AICA riboside concentration of 0.125 mmol/L, the addition of methotrexate induced a 5-fold increase in the ZMP levels.

As methotrexate-induced inhibition of the de novo purine synthesis is expected to decrease the cellular ATP pool (15, 16) and ATP is an intrasteric inhibitor of AMPK (9), we also examined the effect of methotrexate on the levels of ATP. As shown in Fig. 1A, methotrexate caused an overall 2-fold decrease of the levels of ATP in MDA-MB-231 cells, with somewhat less pronounced effects at high concentrations of AICA riboside. In PC-3M-luc-C6 and A-431 cells, methotrexate caused a 5-fold decrease of the ATP levels. Consistent with the increased ZMP levels at low concentrations of AICA riboside and with the fact that AMPK is more sensitive to activation at lower ATP levels (5, 9), methotrexate-albumin potently increased the ability of AICA riboside to activate AMPK in MDA-MB-231 cells, as shown by Western blot analysis with an antiserum recognizing the Thr¹⁷²-phosphorylated (and thus active) form of AMPK (Fig. 1B). In PC-3M-luc-C6 and A-431 cells, similar results were obtained using free methotrexate as shown in Fig. 1B. In agreement with the transport deficiencies of methotrexate in MDA-MB-231 cells, free methotrexate did not enhance the effects of AICA riboside on the phosphorylation of MDA-MB-231 cells (data not shown).

One of the main targets of AMPK is acetyl-CoA carboxylase, one of the key enzymes involved in fatty acid synthesis and in the regulation of fatty acid oxidation (4, 26). In previous studies, AICA riboside at high concentrations was shown to affect the phosphorylation state of acetyl-CoA carboxylase in MDA-MB-231 cells (1). To examine whether methotrexate enhances these effects, MDA-MB-231 cells were incubated with methotrexate-albumin, with AICA riboside or with a combination of both drugs. As shown in Fig. 2A, phosphorylation of acetyl-CoA carboxylase was observed at lower concentrations of AICA riboside when cells were pretreated with methotrexate-albumin. These effects were paralleled by the inhibition of the incorporation of ¹⁴C of ¹⁴C-labeled acetate into extractable lipids (Fig. 2B). Treatment of MDA-MB-231 cells with AICA riboside alone resulted in a dose-dependent decrease of lipid production with half-maximal effects at 0.260 mmol/L. In the presence of methotrexate-albumin, effects of AICA riboside on lipid production were half-maximal at 0.025 mmol/L. At 0.125 mmol/L, lipid production was almost completely shut down, reaching values obtained at 1 mmol/L AICA riboside alone. Similar enhancing effects of methotrexate were observed in PC-3M-luc-C6 and A-431 cancer cells (Fig. 2A and B).

Methotrexate is known to potently inhibit DNA synthesis. AICA riboside has also been shown to attenuate DNA synthesis (1). Here, we examined whether cotreatment of cancer cells with both compounds results in stronger effects on DNA synthesis than with each compound individually. To assess the effects on DNA synthesis, the incorporation of the thymidine analogue BrdUrd was measured using an ELISA technique. Treatment of cancer cells with AICA riboside alone caused a potent and dose-dependent reduction of BrdUrd incorporation in all three tested cell lines (Fig. 3). Methotrexate or methotrexate-albumin alone also caused a dose-dependent decrease in BrdUrd incorporation and, importantly, caused a major shift in the dose-response curve of AICA riboside.

To determine whether the inhibitory effects of methotrexate and AICA riboside were synergistic, CI analysis (19) of the effects on DNA synthesis was done on PC-3M-luc-C6 cells. Cells were treated with different concentrations of methotrexate or AICA riboside alone or with a fixed ratio (methotrexate/AICA riboside: 1:25, 1:50, or 1:125) of both drugs simultaneously as shown in Fig. 4. Synergistic effects (CI < 1) were observed for all three tested ratios of methotrexate/AICA riboside. For the combination of 5 μmol/L methotrexate with 0.125 and 0.250 mmol/L AICA riboside that we also used in other experiments, CI indicating synergism (0.3 < CI < 0.7) and strong synergism (0.1 < CI < 0.3), respectively, were observed (20).

To assess the effect of drug treatment on the growth of cancer cells, we treated MDA-MB-231, PC-3M-luc-C6, and A-431 cells with AICA riboside in the presence or absence of methotrexate or methotrexate-albumin and assessed the ability of individual cancer cells to form colonies. As shown in Fig. 5, a remarkable concerted effect of both compounds was observed. Whereas 5 μmol/L methotrexate or methotrexate-albumin alone and 0.125 mmol/L AICA riboside alone had little or no effect on colony formation, the combination of both compounds led to a severe reduction both in the size and in the number of colonies.

To confirm that methotrexate sensitizes cancer cells for the effect of AICA riboside by inhibition of enzymes of the de novo purine synthesis pathway (ATIC and/or GART), we specifically inhibited these enzymes by the RNA interference technique. MDA-MB-231 and PC-3M-luc-C6 cells were transiently transfected with Stealth siRNA oligonucleotides targeting ATIC and/or GART or with a scrambled oligonucleotide as a negative control. Northern blot analysis showed a marked suppression of ATIC and GART after transfection of PC-3M-luc-C6 cells with respective siRNAs (Fig. 6A–B). As shown in Fig. 6C, RNA interference–mediated silencing of ATIC or GART in the absence of AICA riboside had little or no effects on colony formation. In the presence of exogenous AICA riboside, however, knockdown of ATIC enhanced ZMP accumulation and potently sensitized PC-3M-luc-C6 cancer cells for the inhibiting effect of AICA riboside on colony formation. Despite the fact that exogenous AICA riboside enters the de novo purine synthesis pathway downstream of GART, we also observed that inhibition of GART enhanced AICA riboside–induced ZMP accumulation and sensitized cancer cells for the effect of AICA riboside (data not shown). These findings suggest that silencing of GART may have a more complex effect on the de novo purine biosynthesis pathway than expected and indicate that methotrexate may sensitize cancer cells for AICA riboside through inhibition of both ATIC and GART. Similar effects were observed in MDA-MB-231 cells (data not shown).

The clinical use of antineoplastic agents is often limited by the high doses of compounds that are required to exert effects and by the accompanying dose-related side effects. Cotreatment with multiple chemical compounds that have complementary mechanisms of action or that elicit effects through different antineoplastic pathways often provides interesting avenues toward more effective cancer interventions. Here, we combined two antimetabolic agents that, both in terms of mechanisms of action and in terms of the affected cellular pathways, are complementary. AICA riboside evokes an increase of the intracellular levels of ZMP and through subsequent activation of AMPK leads to inhibition of major energy-demanding cancer-associated processes, including lipid production, DNA synthesis, and matrix invasion. Methotrexate exerts its antineoplastic effects through inhibition of folate-dependent enzymes, including ATIC, the main enzyme metabolizing ZMP, in the de novo purine synthesis pathway. Here, we took advantage of this ability of methotrexate to block the metabolism of ZMP to augment the ZMP levels generated after treatment with exogenous AICA riboside and to decrease simultaneously the levels of ATP, which functions as an inhibitor of AICA riboside–induced activation of AMPK. The net result is a marked sensitization of AMPK for activation by AICA riboside and a major sensitization of the cells toward AICA riboside–induced inhibition of key tumor-associated processes, including lipid production, DNA synthesis, and colony formation. Importantly, this sensitization was already observed at concentrations of compounds that were only marginally effective when used alone and led to synergistic effects. The end result of this cooperative action is that much lower concentrations of each compound are needed to exert antineoplastic effects compared with monotreatment, alleviating several of the limitations related to the high doses required of each compound when given alone. This way, our findings provide new opportunities for both the application of classic antimetabolic chemotherapeutics, such as methotrexate and its derivates, and the exploitation of the energy-sensing machinery as a target for cancer intervention. The recent development of more selective inhibitors of ATIC and GART (27) may improve the efficacy and selectivity of the chemotherapeutic strategy described in this report.