A Transition State Analogue of 5′-Methylthioadenosine Phosphorylase Induces Apoptosis in Head and Neck Cancers*

Methylthio-DADMe-immucillin-A (MT-DADMe-ImmA) is an 86-pm inhibitor of human 5′-methylthioadenosine phosphorylase (MTAP). The sole function of MTAP is to recycle 5′-methylthioadenosine (MTA) to S-adenosylmethionine. Treatment of cultured cells with MT-DADMe-ImmA and MTA inhibited MTAP, increased cellular MTA concentrations, decreased polyamines, and induced apoptosis in FaDu and Cal27, two head and neck squamous cell carcinoma cell lines. The same treatment did not induce apoptosis in normal human fibroblast cell lines (CRL2522 and GM02037) or in MCF7, a breast cancer cell line with an MTAP gene deletion. MT-DADMe-ImmA alone did not induce apoptosis in any cell line, implicating MTA as the active agent. Treatment of sensitive cells caused loss of mitochondrial inner membrane potential, G2/M arrest, activation of mitochondria-dependent caspases, and apoptosis. Changes in cellular polyamines and MTA levels occurred in both responsive and nonresponsive cells, suggesting cell-specific epigenetic effects. A survey of aberrant DNA methylation in genomic DNA using a microarray of 12,288 CpG island clones revealed decreased CpG island methylation in treated FaDu cells compared with untreated cells. FaDu tumors in a mouse xenograft model were treated with MT-DADMe-ImmA, resulting in tumor remission. The selective action of MT-DADMe-ImmA on head and neck squamous cell carcinoma cells suggests potential as an agent for treatment of cancers sensitive to reduced CpG island methylation.

Cellular proliferation is associated with increased levels of polyamine biosynthesis and polyamine pools (1). Target validation for the polyamine pathway as an anticancer approach has come from ␣-difluoromethylornithine (DFMO), 3 a suicide inhibitor of ornithine decarboxylase (ODC) and the committed step of polyamine biosynthesis (2,3). ODC is a difficult cancer target because of its rapid turnover and the dose-limiting toxicity of anti-ODC agents (4). Because of these difficulties, DFMO has not gained wide use. But the polyamine pathway, through its close interaction with S-adenosylmethionine (AdoMet) recycling, remains a target for cancer therapy. We investigated the possibility that feedback inhibition by 5Јmethylthioadenosine (MTA), induced by a transition state analogue inhibitor of 5Ј-methylthioadenosine phosphorylase (MTAP), could be used to block this pathway and initiate anticancer effects. The results indicate that blocking MTA recycling with transition state analogues of MTAP induces apoptosis through specific epigenetic changes in specific cultured cancer cell lines. Inhibition of MTAP is effective in treating a xenograft model of head and neck cancer in mice.
MTA is a product of both spermidine and spermine synthases and provides product inhibition at two sequential sites in the polyamine pathway (Fig. 1A). In humans, MTA is degraded exclusively by MTAP (EC 2.4.2.28), expressed from a single gene locus at 9p21. MTAP produces adenine and 5-methylthio-␣-D-ribose-1-phosphate (Fig. 1B), and these products are recycled to AdoMet. Inhibitors of MTAP are therefore expected to increase intracellular MTA, cause feedback inhibition of polyamine biosynthesis, prevent AdoMet recycling, and disrupt AdoMet-dependent methylation activity. One or more of these activities is expected to be associated with antiproliferative activity (5)(6)(7)(8).
The transition state structure of human MTAP has been established by kinetic isotope effects and quantum chemical calculations. It is characterized by a late transition state with weak participation of the phosphate nucleophile, similar to that of human purine nucleoside phosphorylase but slightly more advanced (Fig. 1B) (9 -15). Analogues of the human MTAP transition state have been synthesized and are powerful and specific inhibitors (16 -18). Methylthio-DADMe-Immucillin-A (MT-DADMe-ImmA) is a chemically stable transition state analogue of human MTAP and is a slow onset tightly binding inhibitor with a dissociation constant of 86 pM (18).
Here we report the selective toxicity of MT-DADMe-ImmA in FaDu and Cal27, MTAP-positive human head and neck squamous cell carcinoma cell lines. An increased cellular concentration of MTA is required for this activity. The specific action against some cancer cell lines but not others is proposed to be the result of changes in DNA methylation events to cause different responses to cells with diverse gene expression patterns.
Homozygous deletions of the MTAP gene together with the tumor suppressor p16 are common in non-Hodgkin lymphoma and acute lymphoblastic leukemia and have also been described in lung, bladder, pancreatic, endometrial, breast, ovarian, mantle cell lymphoma, conventional chondrosarcomas, and biliary tract cancers (19 -24). Thus, progression of certain tumors can occur with a complete lack of MTAP activity. The deletion of p16 now appears to be more closely associated with tumor development. MTAP deletions in specific tumor cells in vivo have a metabolic effect different from whole organism inhibition of MTAP. Feedback inhibition effects associated with MTA accumulation will not occur in a MTAP-deficient tumor in the background of an intact mammal, since surrounding tissues are MTAP-normal and facilitate MTA removal. In whole animal MTAP inhibition, MTA accumulation is expected to permeate all tissues and may modulate cancer proliferation and metastases by metabolite feedback inhibition pathways described above.
Metabolite accumulation effects are well known in human genetic diseases and are caused by specific enzyme defects. For example, homozygous null mutation of human adenosine deaminase causes accumulation of 2Ј-deoxyadenosine and severe combined immunodeficiency disease, specifically killing dividing B-and T-lymphocytes. Likewise, mutation or deletion of purine nucleoside phosphorylase causes accumulation of 2Јdeoxyguanosine and a specific T-cell immune deficiency disease (25,26). Inhibitors against either of these enzymes have been shown to have anticancer effects against the cell types influenced by the genetic deficiency (10,26). Although no genetic deficiency model is known for MTAP, we explored the effects of cellular and whole animal MTAP ablation with MT-DADMe-ImmA, a slow onset tightly binding transition state analogue of the enzyme. . FaDu, MCF7 (wild type), U87, and CRL2522 cells were maintained in Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, 0.1 mM nonessential amino acids, and 1 mM sodium pyruvate. Cal27 and GM02037 cells were maintained in Dulbecco's mod-ified Eagle's medium with high glucose and 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. A549 cells were maintained in Dulbecco's modified Eagle's medium/F-12 medium (50:50) with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were treated with either 20 M MTA or 1 M MT-DADMe-ImmA alone or in combination and compared with untreated controls. MT-DADMe-ImmA was synthesized by Industrial Research Ltd. (17). Cells were photographed by phase-contrast microscopy, and cell numbers were determined by standard procedures.

EXPERIMENTAL PROCEDURES
Cytotoxicity Assay-Cell viability was evaluated using the Alamar Blue assay. Cells were seeded onto 96-well plates at a density of 10 4 cells/well and incubated with increasing concentrations of MT-DADMe-ImmA (100 pM to 100 M) for 4 days at fixed MTA concentrations (0, 5, 10, and 20 M). IC 50 was determined following the manufacturer's instructions (Biotium, Inc., Hayward, CA).
Apoptotic Assay-Apoptosis was measured by fluorescenceactivated cell sorter analysis. Cells (both free and attached) were harvested, and cell cycle analyses were done with a FACScan flow cytometer (BD Biosciences) (27). Cell cycle distributions were calculated using ModFit LT software (Verify Software House, Inc., Topsham, ME). Mitochondrial membrane potential was evaluated by cell incubation for 30 min in 40 nM DiOC 6 (3,3Ј-dihexyloxacarbocyanine iodide; Molecular Probes, Inc.) before harvest (27).
Immunoblot Analysis-Cells were harvested, washed with phosphate-buffered saline, and lysed using radioimmunoprecipitation assay buffer with complete protease inhibitor mixture (Sigma). Cell lysates were subjected to SDS-PAGE, transferred to polyvinylidene difluoride membrane, and immunoblotted with primary antibodies and horseradish peroxidase-conjugated secondary antibody. The blots were developed using the ECL kit (Amersham Biosciences). Anti-MTAP antibody was produced from the custom antibody production services of Harlan Bioproducts for Science (Indianapolis, IN) using purified recombinant human MTAP expressed in Escherichia coli as antigen (18). Anti-MTAP antibody was purified using a Melon gel IgG spin purification kit (Pierce) and used as described (28). Primary antibodies to caspase-9, caspase-3, caspase-7, poly(ADP-ribose) polymerase, secondary anti-rabbit, and anti-mouse antibody were from Cell Signaling, Inc. (Beverly, MA). Antibody for actin was purchased from Sigma.
MTAP Activity Assay-Cells were harvested, washed three times with cold phosphate-buffered saline, and lysed in 0.6% Triton X-100 in phosphate-buffered saline buffer. Assays for cellular MTAP activity contained 100 g of protein from cell lysates in 20-l reaction mixtures containing 200 mM Na 2 HPO 4 , pH 7.56, 10 mM KCl, and 50 M MTA containing 18,000 cpm [8-14 C]MTA. Labeled MTA was synthesized from [8-14 C]S-adenosylmethionine by adjusting the pH to 3.0 followed by 3.5 h at 75°C and purified by HPLC (18). Products of the MTAP reaction were resolved on thin layer silica with 1 M ammonium acetate, pH 7.55, containing 10% isopropyl alcohol. The adenine spots were excised and counted. MTAP activity in mouse blood was assayed in 10 mM sodium phosphate, pH 7.4, 10 mM KCl, 50 M MTA containing 15,000 cpm of [2, H]MTA and 7 l of a 1:1 mixture of whole mouse blood and 0.6% Triton X-100 in phosphate-buffered saline for a total volume of 10 l. Samples were incubated at room temperature for 4 and 8 min, quenched with 1 l of 70% perchloric acid, neutralized with KOH, centrifuged, and resolved on cellulose thin layer sheets with 1 M ammonium acetate, pH 7.5, in 10% (v/v) isopropyl alcohol.
Determination of Polyamines-Polyamines from 0.6 M perchloric acid extracts were purified via cation exchange chromatography and dansyl-derivatized with minor changes from a previous method (29). Samples were eluted from 10-ml Bio-Rad columns by centrifugation at 4,000 rpm for 3 min. The pH of the sodium carbonate used for derivatization was adjusted to 9.3, and the concentration of dansyl-chloride added to samples was adjusted to 100 mM. Dansyl-polyamines were quantitated by HPLC/fluorescence on a Waters Millennium system. Elution from a Phenomenex Luna 5 C18 (2) column used a mobile phase of 30% acetonitrile in a 50 mM ammonium acetate buffer at pH 6.8 (eluent A) and 100% acetonitrile (eluent B) with a gradient of 80% eluent A to 95% eluent B from 2 to 20 min. Fluorescence detection was by excitation at 338 nm and emission at 500 nm.
siRNA-mediated Knockdown of MTAP Expression-FaDu cells were plated in 6-well plates and transfected with Oligofectamine (Invitrogen) according to the manufacturer's protocol. siRNA against MTAP was purchased from the Dharmacon SMARTpool selection. A negative control was included in all siRNA experiments. Forty-eight h after transfection, cells were split into fresh medium for another 36 h before analysis.
Microarray Profiling of CpG Island DNA Methylation-Global profiling of aberrant DNA methylation of CpG islands in FaDu genomic DNA was carried out using the methylationspecific restriction enzyme microarray assay described previously (32). The methylation-specific restriction enzyme microarray assay compares a single DNA sample's response to a methylation-sensitive restriction enzyme (HpaII) (Cy5-red channel) and its corresponding methylation-insensitive isoschizomer (MspI) (Cy3-green channel). Cy5/Cy3 ratios were therefore a qualitative measure of DNA methylation corresponding to each CpG island on the microarray. Red (Cy5) and green (Cy3) signal intensities for each element on the array were calculated using the GenePix Pro 3.0 software package. Data designated to be of poor quality or that did not achieve a signal/noise ratio of at least 2-fold were discarded from subsequent analysis.
FaDu Xenografts-Male NOD.SCID mice (6 -8 weeks old) were obtained from Jackson Laboratory (Bar Harbor, ME). Animal experiments were conducted in accordance with the approved protocol guidelines of the Animal Committee of the Albert Einstein College of Medicine. FaDu cells (10 6 ) were inoculated into the dorsum of the hind foot. Tumors were established for 5 days and were followed by treatment with MT-DADMe-ImmA at 9 or 21 mg/kg body weight in drinking water or by daily intraperitoneal injections of 5 mg/kg body weight. After tumors were established, mice were randomly assigned to treatment or control groups of five animals each. Tumor volume (V) was determined from the equation, V ϭ (4/3)ϫ(22/ 7)ϫ(1/8)ϫ(length ϫ width ϫ height). Differences between treatment cohorts were determined using Student's t test. Mice were weighed every 4 -5 days and monitored for hair loss, loss of appetite, vomiting, and diarrhea. Total and differential blood and bone marrow analyses were performed after MT-DADMe-ImmA treatment.

RESULTS
MT-DADMe-ImmA Treatment and Cellular MTAP Activity-Treatment of FaDu cells with 1 M MT-DADMe-ImmA in combination with 20 M MTA for 24 h caused a 96% reduction in MTAP activity (control ϭ 0.535 Ϯ 0.01 and treated ϭ 0.020 Ϯ 0.004 nmol of adenine formed/min/mg of protein). In this assay, cells are first washed free of unbound inhibitor followed by lysis in detergent. In this protocol, less than 5% of the tightly bound inhibitor dissociates during the assay procedure. Thus, in cultured cells, at least 96% of the MTAP activity is inhibited. The result also establishes that the inhibitor is permeable to FaDu cells. The MTAP activity of cultured human fibroblasts (CRL2522) showed similar inhibition (94% reduction) in this protocol (control ϭ 0.245 Ϯ 0.011 and treated ϭ 0.015 Ϯ 0.003 nmol of adenine formed/min/mg of protein).
Effect of MT-DADMe-ImmA on Polyamine Levels-FaDu cells treated with MT-DADMe-ImmA and MTA (to prevent cellular MTA loss to medium) for 1-3 days showed increased putrescine and decreased spermine levels (Fig. 1C). In addition to the changes in cellular polyamines, putrescine in the medium becomes the dominant polyamine (greater than all cellular polyamines) in day 5 cultures treated with MTA ϩ MT-DADMe-ImmA but not in control FaDu cells or those treated with MTA or MT-DADMe-ImmA alone (not shown). Thus, the combination of MTA and MT-DADMe-ImmA provides the most effective block of polyamine synthesis. Spent medium from 5-day FaDu cultures treated with MT-DADMe-ImmA and MTA revealed a 3.7-fold increase in putrescine relative to untreated culture medium. Supplementation of culture medium with putrescine, spermidine, and spermine alone or in combination did not prevent the apoptosis of FaDu cells caused by MT-DADMe-ImmA and MTA.
Effects of DFMO on Polyamines-To test the hypothesis that decreased polyamine levels are responsible for FaDu cell death, cultures were treated with 5 mM DFMO, a known inhibitor of ornithine decarboxylase. After 24 h of DFMO treatment (roughly one cell division time) the total intracellular polyamines were decreased from 12 Ϯ 3 to 5 Ϯ 1 nmol/mg protein.
Extended culture of FaDu cells for 5 days in the presence of DFMO continued the depletion of polyamines to less than 1 nmol/mg protein. FaDu cells showed no decrease in growth rate from this treatment and did not undergo apoptotic changes. Thus, depletion of polyamines is not responsible for the onset of apoptosis in FaDu cells.
Effect of MT-DADMe-ImmA on Cellular MTA-Complete inhibition of MTAP would be expected to cause intracellular accumulation of MTA. FaDu cells treated with MT-DADMe-ImmA alone or in combination with MTA for 3 days showed a 3-4-fold increase in MTA levels in washed cell extracts. MTA levels increased 6 -7-fold at day 5 of treatment relative to untreated control FaDu cells (Fig. 1D).
MTAP Expression in Tumor Cell Lines-Human cancer cell lines of different origins (breast MCF7, lung A549, head and neck FaDu or Cal27, glioblastoma U87, and fibroblast CRL2522) were screened for MTAP protein expression by Western blot using the anti-MTAP antibody ( Fig. 2A). MTAP protein was readily detected in FaDu, Cal27, and CRL 2522 but not in MCF7, A549, or U87. Transfection of MCF7 with an expression vector for MTAP converted MCF7 to a strong positive response for MTAP.
MT-DADMe-ImmA ϩ MTA Inhibits FaDu Cell Growth-FaDu cells cultured in the presence of 1 M MT-DADMe-ImmA alone or 20 M MTA alone showed little growth inhibi-tion and no apoptosis from these agents. MT-DADMe-ImmA and MTA used in combination gave powerful inhibition of cell growth (Fig. 3, A and B). In contrast to FaDu cells, a human fibroblast cell line was not affected by the same treatment (Fig.  3B). Inhibition of FaDu cell growth by MT-DADMe-ImmA is dependent on MTA concentration. The experimental IC 50 value for MT-DADMe-ImmA decreased from 500 to 50 nM as MTA was increased from 5 to 20 M (Fig. 3C). MT-DADMe-ImmA alone caused little growth effect on FaDu cells even at  MT-DADMe-ImmA and MTA Cause G 2 /M Arrest and Apoptosis in FaDu and Cal27 Cells-FaDu cells were exposed to 1 M MT-DADMe-ImmA or 20 M MTA alone or in combination for up to 6 days and subjected to flow cytometry analysis. Combined treatment arrested cells in G 2 /M starting at day 2, followed by a significant increase in sub-G 1 cell fractions by day 4 and onward, indicating apoptotic activity (Fig. 4A).
Another head and neck cancer cell line, Cal27, was also found to be sensitive to MT-DADMe-ImmA and MTA. After 8 days of treatment, the number of viable Cal27 cells decreased (supplemental Fig. A) as a result of G 2 /M arrest and apoptosis when compared with controls (Fig. 4B). Cell selectivity was tested by exposing normal human fibroblast cells (CRL2522) to MT-DADMe-ImmA and/or MTA for 21 days followed by fluorescent cell sorting analysis. No cytotoxicity was observed in CRL2522 cells after prolonged treatment (supplemental Fig. B).
Mitochondrial Integrity-Initiation of apoptosis is a common response to cytotoxic agents used in cancer chemotherapy. However, neither MT-DADMe-ImmA nor MTA are considered to be cytotoxic agents. It was of interest to determine if the mechanism of apoptosis seen in FaDu cells involved mitochondrial damage. Apoptotic initiation can be mitochondria-  or -independent; therefore, mitochondrial membrane integrity was measured by DiOC 6 staining. Combined treatment of FaDu cells led to the loss of mitochondrial retention of DiOC 6 (Fig. 4C). This is not a consequence of the polyamine depletion, since treatment with 5 mM DFMO did not alter DiOC 6 staining (data not shown). In addition to mitochondrial damage, FaDu cells containing apoptotic bodies also showed clumped or fragmented chromatin by Hoechst staining (supplemental Fig. C). We conclude from these results that treatment with MT-DADMe-ImmA ϩ MTA causes G 2 /M cell cycle arrest and mitochondrial damage in FaDu cells to induce apoptosis. In contrast, no effect is seen in CRL2522 normal human fibroblasts treated in the same way (supplemental Fig. B).
Activation of Caspase Pathways-The pathway of apoptosis induced by MT-DADMe-ImmA ϩ MTA was investigated in FaDu cells by analyzing the activation status of initiator and effector caspases and cleavage of the caspase substrate poly-(ADP-ribose) polymerase. Treatment of FaDu cells with MT-DADMe-ImmA and MTA for 4 days caused significant cleavage of caspase-9, -3, and -7 and poly(ADP-ribose) polymerase (Fig. 5A). The Bcl 2 family members Bcl 2 and Bax are markers of mitochondrial integrity, and the Bcl 2 and Bax protein levels did not change detectably after treatment (data not shown). Thus, MT-DADMe-ImmA ϩ MTA treatment causes activation of a mitochondria-dependent apoptosis cascade.   (Fig. 6A). FaDu cells with MTAP knockdown showed substantial mitochondrial damage after a 1.5-day exposure to 20 M MTA without added MT-DADMe-ImmA. In contrast, the same MTA treatment had no effect on the mitochondrial integrity of control or scrambled siRNA transfectants (Fig. 6B). Therefore, the role of MT-DADMe-ImmA in FaDu sensitivity to MTA is to inhibit MTAP and prevent intracellular degradation of the MTA. MTA alone is sufficient for induction of apoptosis in MTAP Ϫ FaDu cells.
Cell Line Specificity of MTA Effects-MTA accumulation causes metabolic perturbation of polyamine/methionine/adenine/AdoMet pathways and could thereby perturb epigenetic control by altering AdoMet methyltransferase potential or specificity on DNA. Metabolite effects would be expected to have similar effects on multiple cell lines, whereas epigenetic effects would be cell line-specific related to differential gene expression patterns in cancer cell lines of different origins. MTAP-negative cell lines A549, U87, and MCF7 were tested for MTA-induced cytotoxicity both alone and in the presence of MT-DADMe-ImmA. No adverse effects were seen after 28 days of treatment (IC 50 values Ͼ100 M in the presence or absence of 20 M MTA). These cell lines respond differently from FaDu, implicating epigenetic effects.

Microarray Profiling of CpG Island DNA Methylation in
FaDu Cells-In order to evaluate any epigenetic effects of MTA and MT-DADMe-ImmA treatment, we examined the global pattern of DNA methylation of CpG islands in FaDu cells using the methylation-specific restriction enzyme assay with a microarray containing 12,288 CpG island clones. For each genomic DNA sample from FaDu cells, we identified hypermethylated CpG island clones based on a Cy5/Cy3 ratio of 3.0 or greater and a signal/noise ratio of at least 2.0. These cut-offs were optimized to maximize the number of signals and dynamic range used for profiling of head and neck squamous cell carcinoma samples while maintaining a high intraclass correlation coefficient between replicate measurements (32). For untreated FaDu cells, the median number of hypermethylated CpG island clones identified was 613 (range 597-628). In contrast, treated FaDu cells contained a median of 211 hypermethylated CpG island clones (range 114 -308). An overlap of the replicate microarray data sets identified 118 CpG island clones in which DNA methylation was consistently reduced in response to MTA and MT-DADMe-ImmA treatment of FaDu cells. Association of this subset of CpG island clones with the promoter elements and first exons of corresponding genes, as well as the effect of treatment on the relative expression of these genes, is ongoing.
Oral Availability of MT-DADMe-ImmA in Mice-Systemic blocking of MTAP activity in mice was accomplished by oral or intraperitoneal administration of MT-DADMe-ImmA. Administration of a single oral dose of 100 g (4 mg/kg) MT-DADMe-ImmA gave complete inhibition of blood MTAP activity. The t1 ⁄ 2 for onset of inhibition was 50 min with complete inhibition by 250 min. MTAP activity slowly returned, giving a biological half-life for the action of oral MT-DADMe-ImmA of 9050 min (6.3 days). In separate experiments, a single oral dose of 200 g of MT-DADMe-ImmA (8 mg/kg) showed Ͼ95% inhibition of MTAP activity in liver extracts at both 20 and 240 min following a single oral dose (not shown). Based on these studies, immunodeficient mice bearing human FaDu tumors were given daily treatment of MT-DADMe-ImmA to inhibit whole body MTAP. Oral administration was 9 and 21 mg/kg/day, and intraperitoneal injections were 5 mg/kg/day.

MT-DADMe-ImmA Inhibits the Growth of FaDu Xenografts-
The time-dependent growth of FaDu tumors in immunodeficient mice is suppressed by oral or intraperitoneal treatment with MT-DADMe-ImmA (Fig. 7A). Tumors were established in mice for 5 days prior to oral or intraperitoneal treatments with MT-DADMe-ImmA. Tumor growth in animals treated with MT-DADMe-ImmA was dose-responsive and was significantly slower than in controls (p Ͻ 0.06). Representative tumors from the treatment cohorts are shown at 28 days after therapy began (Fig. 7B). No significant differences in animal weight or in total and differential blood counts were seen between treatment and control groups after this treatment. Thus, MT-DADMe-ImmA administration suppresses FaDu growth in vivo with low cytotoxicity to mice.
MT-DADMe-ImmA Causes Remission of FaDu Tumors-Subsequent to the 28-day MT-DADMe-ImmA therapy, treatment was removed for a subsequent period of 28 days. There was no regrowth of tumor in those mice receiving an

DISCUSSION
The Polyamine Pathway in Cancer-Rapidly dividing cells synthesize large amounts of polyamines as counter ions for nucleic acids. Enzymes in these pathways are overexpressed in many cancer cells (33). Targeting the de novo polyamine biosynthetic pathway in the treatment of human cancer has a long history in DFMO treatment for inhibition or ornithine decarboxylase, the committed step of polyamine synthesis. But DFMO has found limited application in cancers because of the rapid turnover of the target enzyme. Targeting other synthetic steps has more recently resulted in S-adenosylmethionine decarboxylase inhibitors (MGBG and SAM486A), other ODC inhibitors (MDL72,175 and MAP), and polyamine analogues (DENSPM and DEHSPM). Their use in cancer therapy has been limited because of drug-induced toxicity to bone marrow and intestinal tract and also because of limited cancer therapeutic efficacy (21).
MTAP in the Polyamine Pathway-The major polyamine in humans is spermine, and its biosynthesis from ornithine and AdoMet produces 2 mol of MTA for each mol of spermine. We selected MTAP as a potential antiproliferative target because of its central function in removal of MTA, a 0.3 M product inhibitor for bovine spermine synthase and a 50 M product inhibitor for rat spermidine synthase (34 -37). MTAP also catalyzes an essential step in recycling both the methylthiol group and adenine of MTA to reform AdoMet in the polyamine pathway. The MTAP reaction is the only enzyme that forms adenine in humans, and the adenine salvage pathway is known to decrease cellular dependence on de novo purine synthesis (28). Thus, MTAP Ϫ cells are more sensitive to the inhibitory effects of antifolates and L-alanosine (37). Deletion of MTAP, together with p16, a tumor suppressor gene, and other genes near 9p21 is a frequent genotype in human tumor-derived cell lines and primary tumors (24). It has therefore been suggested that MTAP inhibition may be a poor anticancer target, since it is deleted in many tumor types. Our work tests this hypothesis by providing inhibition of MTAP catalytic activity without gene deletion in the 9p21 region.
Deletion of MTAP in Tumors-Chromosomal deletion of MTAP in tumors differs from drug-based whole animal inhibition of the enzyme. When only tumor cells are MTAP-deleted, MTA produced in MTAP Ϫ cancer cells will be removed by MTAP activity of neighboring tissues, and MTA will not accumulate.
MT-DADMe-ImmA Administration to Mice-MT-DADMe-ImmA given orally or intraperitoneally inhibits blood and liver MTAP and probably that of other tissues, since MTA is found in the urine of animals treated with MT-DADMe-ImmA. 4 This inhibitor provides an opportunity to examine the effect of whole animal inhibition of MTAP activity. Inhibition of MTAP increases cellular and mouse MTA concentrations and decreases polyamine levels. However, these metabolic changes are not generally inhibitory to growth of tumor cell lines. Rather, these changes led to tumor-specific effects such that the FaDu and Cal27 cell lines are highly susceptible in vitro, and FaDu can be eliminated in the in vivo immunodeficient mouse model. This study is the first to report the effects of MTAP inhibition on polyamine levels, MTA accumulation, and the growth of cancer cell lines in culture and in mice.
Cell Line Sensitivity to MTA and MTAP Inhibition-MTAP ϩ and MTAP Ϫ human cancer cell lines were evaluated for sensitivity to MTA or MT-DADMe-ImmA alone or in combinations. MTA alone had no effect on MTAP ϩ cancer-derived cells or in normal fibroblasts. MTAP ϩ cell lines degrade MTA. However, MTAP Ϫ lines may or may not be sensitive to added MTA, depending on their epigenetic response to MTA perturbation of DNA methylation patterns.
Cell Line Sensitivity to MT-DADMe-ImmA-No cell lines were found to be sensitive to MT-DADMe-ImmA alone, even at concentrations of 100 M, a concentration 11,600 times greater than the K d for the inhibitor-MTAP complex (18). However, FaDu and Cal27 are sensitive to the combination of MTA and inhibitor. This finding is significant for cancer, since head and neck cancers represent the sixth most frequent cancer, and 90% of these are squamous cell carcinomas (39). Treatment with MT-DADMe-ImmA and MTA arrests FaDu cells in G 2 /M, induces apoptosis in culture, and may also induce apoptosis to clear the tumor in the mouse model (see below).
Polyamine and MTA Levels Related to Cytotoxicity-Polyamine levels in treated FaDu cells were decreased prior to the 4 G. Cordovano and V. L. Schramm, unpublished observations. onset of apoptosis, but the addition of spermine, spermidine, and putrescine in growth medium did not reverse the effect. Treatment of FaDu cells with DFMO also inhibited polyamine synthesis, but this did not induce apoptosis in FaDu cells. MTA accumulated in cells treated with MT-DADMe-ImmA alone and to a greater extent by treatment with MT-DADMe-ImmA and MTA. Polyamine and MTA levels reach similar levels in sensitive and nonsensitive cell lines. Although the inhibitor influences both MTA and the polyamine synthetic pathway, these metabolite changes are not sufficient to explain the onset of apoptosis in FaDu and Cal27 cells without toxicity to other cell lines.
Pathway of Apoptotic Induction-Mitochondrial damage is linked to apoptosis through activation of caspase-9, -3, and -7 and poly(ADP-ribose) polymerase cleavage. Cleaved caspases and poly(ADP-ribose) polymerase are highest in FaDu cells treated with MT-DADMe-ImmA and MTA, and these cells also demonstrate the highest levels of MTA. Apoptosis induced by DNA damage signals through p53; however, apoptosis in FaDu involves a p53-independent pathway, since p53 of FaDu contains an R248L point mutation. This mutant p53 is unable to activate the target gene, p21 Cip1 (40). In response to MT-DADMe-ImmA plus MTA treatment, there is no change in p53 protein or phosphorylation levels compared with untreated control (results not shown). Induction of apoptosis is proposed to initiate through mitochondrial-mediated damage.
DNA Methylation in FaDu Cells-Responses of specific cell lines to agents that perturb the AdoMet pathway suggest action at the level of gene expression. AdoMet is well known as the methyl donor for cytosine DNA methyltransferases that methylate CpG sites. It is therefore not surprising that DNA methylation patterns are altered in FaDu DNA by treatment of cells with MT-DADMe-ImmA and MTA. Specifically, we observe an overall reduction in the number of hypermethylated CpG islands in FaDu cells in response to MT-DADMe-ImmA and MTA treatment. Of particular interest are the subset of 118 CpG island clones (1% of the 12,288 CpG screen) that show a consistent response to treatment. One or more of these sites are likely to alter expression of associated genes in a manner leading to apoptosis. Identification of these pathways is the next goal in characterizing this potentially useful anticancer mechanism.
Tumor Suppression in Mice-Administration of MT-DADMe-ImmA to mice resulted in an inhibition of MTAP activity, as indicated by the appearance of MTA in the plasma and urine. 4 Treatment with MT-DADMe-ImmA caused suppression of tumor growth in the FaDu xenograft mice. Significantly, the treated animals showed no significant weight loss, hair loss, loss of appetite, diarrhea, or blood changes after treatment for 28 days. With optimal doses of MT-DADMe-ImmA, 28 days of treatment caused remission of tumor growth, which did not recur following 28 days post-treatment. Thus, therapy in the mouse model may reflect the apoptotic reaction seen in cell culture. MTA accumulation is proposed to act through AdoMet-induced changes in DNA methylation. Each cancer cell line has a different gene expression pattern; thus, MTA perturbation of DNA methylation is expected to have different effects on each cell type. Gene expression patterns for cancer cell lines differ considerably in culture and in vivo, and it is possible that different effects of MT-DADMe-ImmA will be seen in mammalian tumor models than in cultured cells. Here, the FaDu head and neck line is shown to be driven into remission in SCID mice by oral administration of the apparently nontoxic MTAP inhibitor.