Modulation of Gene Expression in Human Central Nervous System Tumors under Methionine Deprivation-induced Stress Free

Most human tumor cells are dependent on exogenous methionine, whereas normal tissues can rely on homocysteine to meet their methionine needs (1). This is known as the methionine defect, and although it has been attributed to failure of methionine synthase to recycle homocysteine in rapidly dividing cells (2), its cause is far more complex and may involve altered methylation patterns in tumor DNA that affect gene expression. Because of the property of methionine stress to target only cancer and not normal cells, its potential for clinical use against many malignancies has outweighed the technical problems encountered in the past (3, 4) that have kept this approach in obscurity. Theoretically, methionine-dependent tumors could be selectively “starved” under methionine-stress conditions, whereas normal tissues are rescued with homocysteine. This theory was validated in central nervous system (CNS) tumor xenografts implanted in athymic mice with both dietary and pharmacological suppression of methionine (3, 4). Methionine-depleting regimens caused three distinct physiologic responses: (a) mitotic and cell cycle blocks, which lead to cell death and a marked and sustained regression during methionine-depleting regimens (5); (b) selective sensitization of tumors to DNA alkylating damaging agents because of down-regulation of DNA repair protein O⁶-methylguanine-DNA methyltransferase (3, 4, 5, 6); and (c) a marked increase in the immune response of the host against tumors indicated by the macrophage infiltration (4).

Given the similar response of CNS tumors to methionine stress, it is possible that the restriction of methionine mobilizes cellular elements for growth control and death that are common in these tumors. The effect of methionine stress may be exerted via the induction of a single gene with multiple regulatory functions leading to proapoptotic and/or antisurvival events such as p53. The alternative to a single gene involvement is the mobilization of several proapoptotic or antiproliferative elements acting via a variety of mechanisms that are either common or depend on the mutational status of the tumor cell. A marked up-regulation of more than a single tumor suppressor gene or of a single signal transduction pathway by methionine stress could result in a synergy that would be more toxic to tumors than the targeting of a single gene or gene product. Whether the toxicity of methionine stress is regulated by the induction of a single or by multiple tumor suppressor genes, such toxicity should also depend on how the methionine stress influences the induction of proapoptotic and/or the abrogation of prosurvival (proliferative) pathways by altering signal transduction patterns through direct regulation of growth factors and epigenetic events. Because of the complexity of testing this hypothesis, understanding methionine stress requires the use of holistic approaches to identify the multiple changes in gene expression and protein synthesis between this stress and the normal condition. To determine the effect of methionine stress on gene expression in CNS tumors, and to subsequently link key elements of gene expression to physiologic effects of the stress, such as cell cycle arrest and apoptosis, we compared gene expression and key protein synthesis profiles between pairs of CNS tumor cells cultured either in methionine or homocysteine in the following set of lines: (a) the glioblastoma multiforme SWB77 (mut p53), which is resistant to methionine stress in xenograft experiments (4); (b) the glioblastoma multiforme SWB39 (mut p53); (c) the meduloblastoma DAOY (mut p53), which is sensitive to methionine stress both in culture and in xenograft experiments (3, 4, 5); (d) the anaplastic oligodendroglioma SWB61 (wt p53); and (e) the anaplastic astrocytoma SWB40 (wt p53). In this article we show that methionine stress in CNS tumors induces cell cycle inhibition and proapoptotic events regardless of p53 mutational status and inhibits prosurvival pathways by partially overlapping mechanisms.

The CNS tumor cell lines SWB40 (anaplastic astrocytoma), SWB61 (anaplastic oligodendroglioma), SWB77 (glioblastoma multiforme), and SWB39 (glioblastoma multiforme), which have been isolated and characterized in this laboratory (7), as well as the medulloblastoma DAOY and the modified fibroblast 3T3 (American Type Culture Collection, Manassas, VA) were used in this study. Additional cell lines, namely SWB85, SWB83, SWB95, and SWB33, also isolated in this laboratory from glioblastoma multiformes were used in some experiments. All of the cell lines were cultured in DMEM (BioWhittaker, Walkersville, MD) with 5% fetal bovine serum (Media Tech, Inc., Herndon, VA) in 6.5% CO₂ atmosphere at 37°C until confluent. Cultures were subsequently adapted to DMEM A (Media Tech, Inc.) supplemented with an additional 584 mg/L l-glutamine (Sigma, St. Louis, MO) to a final concentration of 8 mmol/L, 10 ml/L of Pen/Strep (Sigma), and 50 ml/L dialyzed fetal bovine serum (Life Technologies, Inc., Grand Island, NY). Once the cell lines were passed three times in medium A, they were split into 40 T-75 flasks and maintained in medium A until they reached 60 to 70% confluence. At that time, 10 plates were maintained in medium A, whereas the rest were transferred in methionine-free DMEM B (Media Tech, Inc.) supplemented with 27 mg/L homocysteine thiolactone hydrochloride (Sigma), 62.5 mg/L l-cystine dihydrochloride (Sigma), and l-glutamine to a final concentration of 8 mmol/L, Pen/Strep, and dialyzed fetal bovine serum. Media was replaced in all of the cultures every 48 hours, and detached cells were collected, counted, and processed for determining apoptosis by the terminal deoxynucleotidyl transferase-mediated nick end labeling assay. Growth patterns were observed on a daily basis, and plates were sacrificed approximately every two days to determine cell number progression in each culture and cell cycle measurement by flow cytometry. Additional plates were used to harvest cells at various intervals as indicated one day after medium replacement and frozen at −80°C until used for c-DNA oligoarray and protein analysis.

Cells were trypsinized, washed with PBS twice, and centrifuged (300 rpm) at 4°C. The cell count was adjusted to 2 to 3 × 10⁵/ml cells in PBS per sample. From this cell suspension, 1 ml was taken in duplicate centrifuged at 300 rpm for 10 minutes at 4°C, reduced to 100 μL, and fixed in 500 μL of ice-cold 70% ethanol overnight at 4°C. Fixed cells were centrifuged at 500 rpm at 4°C for 10 minutes and suspended in 500 μL of PBS containing 50 μL of 50 μg/ml propidium iodide (Sigma) and 10 μL of 5 μg/ml RNase for 45 to 60 minutes at 4°C. Flow cytometric analysis was done within 3 hours of propidium iodide staining on a Beckman-Coulter Epics XL. EXPO32 software was used for optimum analysis of cell cycle (G₁, G₂, and S phase).

Total RNA was isolated from control and methionine-stressed lines (at 24 hours before massive apoptosis) in triplicate with a two-step isolation procedure as follows: (1) RNA from the cell samples was extracted with TRIzol reagent; and (2) small fragments of RNA and genomic DNA contamination were eliminated by additionally purifying the total RNA with RNEASY columns. The quality of the RNA samples was examined by quantifying the 260:280 ratio (the minimal acceptable ratio is 1.8) and by visualizing the 28S to 18S rRNA bands in agarose gel (the minimal acceptable ratio is 1.5). The first-strand cDNA synthesis was carried out with 5 μg total RNA with a T7-d(T)24 primer GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)24 (Superscript II, Life Technologies, Inc., Rockville, MD). This was followed by the second-strand cDNA synthesis at 16°C by adding Escherichia coli (E. coli) DNA ligase, E. coli DNA polymerase I, and RNase H into the reaction. The cDNA was blunt-ended by the addition of T4 DNA polymerase and purified with phenol/chloroform extractions and ethanol precipitation. The cDNA was then incubated at 37°C for 4 hours in an in vitro transcription reaction to produce biotin-labeled cRNA with the MEGAscript system (Ambion, Inc., Austin, TX).

The mRNA expression of over 22,000 genes and expression sequence tags was done for pairs of tumor cells grown in methionine or homocysteine media with the use of U133A (22,000 genes) and U95 (12,600 genes) chips from Affymetrix in duplicate. Briefly, 15 to 20 μg of biotin-labeled cRNA was fragmented by incubating in a buffer containing 200 mmol/L Tris acetate (pH8.1), 500 mmol/L KOAc, and 150 mmol/L MgOAc at 95°C for 35 minutes. The fragmented cDNA was hybridized with a pre-equilibrated Affymetrix chip at 45°C for 14 to 16 hours. The hybridizations were washed in a fluidic station with nonstringent buffer (6 × SSPE, 0.01% Tween 20, and 0.005% antifoam) for 10 cycles and stringent buffer (100 mmol/L 2N-morpholino-ethanesulfonic acid, 0.1 m NaCl, and 0.01% Tween 20) for 4 cycles and stained with strepto-avidin phycoerythrin. This was followed by incubation with biotinylated mouse antiavidin antibody and restained with strepto-avidin phycoerythrin. The chips were scanned in an Agilent ChipScanner (Affymetrix Inc., Santa Clara, CA) to detect hybridization signals. Baseline analyses were done with Oligoarray suite 5.0 to identify statistically significant gene expression alterations between samples derived from cells grown in methionine and those maintained in homocysteine containing medium. Because samples were analyzed in triplicates, these results were additionally screened for consistent P by the student t tests (P < 0.05) to eliminate random sampling errors.

Cyber green light cycler PCR was used to estimate relative mRNA expression with high confidence. An aliquot (200 ng total RNA) of the first-strand cDNA was added to a mastermix of forward and reverse primer mix (30 pmol), SYBR green, two OmniMix HS beads (TaKaRa Bio Inc., Otsu, Shiga, Japan) containing 3 units TaKaRa hot start Taq polymerase, 200 μmol/L deoxynucleoside triphosphate, and 4 mmol/L MgCl₂ in 25 mmol/L HEPES buffer (pH 8.0). PCR was done with the following custom primers: inhibitory κB (IκB)e, 5′-GCT GGA AGC ACT CAC TTA CAT CT-3′ (forward) and 5′-GGA GTG CTG TCT GGT AAA GGT TAT-3′ (reverse); NF-κB65, 5′-AGG ACA TAT GAG ACC TTC AAG AGC-3′ and 5′-CTC ATC ATA GTT GAT GGT GCT CAG-3′; RkB, 5′-CTT AAG GGA AGT GAA AGA GGA GTG-3′ and 5′-CAC TGT CCC CCA GTT CTA CTT TAT-3′; NF-κB, 5′-TGG ACA GTA CTA CCT ACG ATG GAA-3′ and 5′-AGA GTC ATC CAG GTC ATA GAG AGG-3′; NF-κB p50, 5′-AGA TCT GTA ACT ACG AGG GAC CAG-3′ and 5′-TAG TCC CCA TCA TGT TCT CCT TAG-3′; BAX, 5′-GAC CCA GAG ATG GTC ACC TTA C-3′ and 5′-GAA GTA CTC ATA GGC ATT CTC TGC-3′; BAX β, 5′-CAG CTC TGA GCA GAT CAT GAA GAC-3′ and 5′-GCC CAT CTT CTT CCA GAT GGT GAG-3′; BAX δ, 5′-AGC TCT GAG CAG ATC ATG AAG AC-3′ and 5′-CAA AGT AGA AAA GGG CGA CAA C-3′; GADD34, 5′-GAT GAG GAA GAA GAT GAG GAC AGT-3′ and 5′-GAC TTC AAG AAG ACA CCT GTA GCA-3′; GADD45, 5′AGC AGT TAC TCC CTA CAC TGA TGC-3′ and 5′-CTG CAA AGT CAT CTA TCT CCG GGC-3′; TRAIL, 5′-CAG TCT CTC TGT GTG GCT GTA ACT-3′ and 5′-TAT GTG AGC TGC TAC TCT CTG AGG-3′; DR6, 5′-ACC TTC TCA GAT GTG CCT TCT AGT-3′ and 5′-CTA ACA GAG GCA GAA GAG TTG GAT-3′; TRAIL R3, 5′-TCA GAA CAT ACT GGA GCC TGT AAC-3′ and 5′-CAG GAC GTA CAA TTA CTG ACT TGG-3′; MDA7, 5′-TAA GCA GCG CCM GMT GAC AGT CAG-3′ and 5′-CAC ACT GTG GCA AGG ATT GTG AGG-3′; BCL-2, 5′-CTG TGG ATG ACT GAG TAC CTG AAC-3′ and 5′-AAC TGA GCA GAG TCT TCA GAG ACA-3′; BCL-2, 5′-CCT TCT TTG AGT TCG GTG GG-3′ and 5′-CAG CCA GGA GAA ATC AAA CAG AGG-3′; BCL-XL, 5′-TTG GAC AAT GGA CTG GTT GA-3′ and 5′-GTA GAG TGG ATG GTC AGT G-3′; CASP-1, 5′-AAC TTA ATA TGC AAG ACT CTC AAG-3′ and 5′-GCG GCT TGA CTT GTC CAT TAT TGG-3′; CASP-3, 5′-ATG GGA ACA CTG AAA ACT CAG-3′ and 5′-GTC ATC ATC AAC ACC TCA GTC T-3′; CASP-2, 5′-GTT ACC TGC ACA CCG AGT CAC G-3′ and 5′-GCG TGG TTC TTT CCA TCT TGT TGG TCA-3′; IRF-1, 5′-ATT CAC ACA GGC CGA TAC AA-3′ and 5′-ACT ATG GAA GAG ACT ACC TGA-3′; IRF-2, 5′-ATC CAT ACA GGA AAG CAT CA-3′ and 5′-AAT CAT TAC CTC ATT CAC TA-3′; CASP-3, 5′-TGT TTC ATC CAG TCG CTT TG-3′ and 5′-ATT CTG TTG CCA CCT TTC GG-3′; CASP-7-RT, 5′-GGA GAA AGC TCA TGG CTG TGT-3′ and 5′-TCC5′-AAG CAA ACC TCG GGG ATA CT-3′; CASP-8-RT, 5′-AAG CAA ACC TCG GGG ATA CT-3′ and 5′-GGG GCT TGA TCT CAA AAT GA-3′; CASP-8-VC, 5′-ATT CAG CAA AGG GGA GGA GT-3′ and 5′-TGC ATC CAA GTG TGT TCC AT-3′; CASP-8-VD, 5′-GGA CTC TCC AAG AGA ACA GGA TAG-3′ and 5′-TCT ACT GTA CCT GCA TCC AAG TGT-3′; and CASP-8-VE, 5′-GAC TGG ATT TGC TGA TTA CCT ACC-3′ and 5′-ATG ACC CTC TTC TCC ATC TCT ATG-3′. Amplification was carried for 35 cycles (hot start 2 minutes at 95°C, 15 seconds at 95°C, 25 seconds at 54°C, and 25 seconds at 72°C). To determine the amplification efficiency of the target, serial dilutions of the samples were amplified, and the run was analyzed with light cycle software under conditions of a low error rate (<0.2). The formula (efficiency = 10−^1/slope) was used, where slope was determined from standard curves. The specificity of the primers was determined by agarose electrophoresis of the PCR product. β-actin was used as the internal control, and leukemic K562 cells containing 1,544 copies of β-actin mRNA per cell (8) were used as standard to determine the approximate number of copies in the target gene.

Cultures used in this experiment were harvested at the same time intervals used for harvesting RNA. Cells were washed (2×) with cold PBS and lysed (while still attached) for 30 minutes in 100 μL of ice-cold radioimmunoprecipitation assay buffer [50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl containing 0.1% SDS, 1% NP40, 0.5% sodium deoxycholate, 0.1 mg/ml phenylmethylsulfonyl fluoride, 1 mmol/L sodium orthovanadate, and 1 mg/ml aprotinin]. Cell debris was removed by centrifugation at 14,000 × g for 10 minutes at 4°C. Protein was determined with the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA). Aliquots of cell extracts containing 50 μg of total protein were resolved by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Perkin-Elmer Life Sciences, Inc., Boston, MA). Filters were blocked for 1 hour at room temperature in Blotto A [5% nonfat dry milk powder in 1 m Tris (pH 7.4), 5 m NaCl, and 0.05% Tween 20] and then in Blotto A containing a 1: 500 to 1:100 dilution of the following: (a) the rabbit polyclonal GADD45, GADD34, GADD153, and p-38 antibodies or the mouse monoclonal P-p38, p21, CDC2, and SMAD4 antibodies from Santa Cruz Biotechnology (Santa Cruz, CA); and (b) the rabbit polyclonal P-AKT(Ser473) and P-AKT(Thr308), P-Rb(Ser780), P-Rb(Ser795), P-Rb (Ser807/811) or mouse monoclonal Rb antibodies from Cell Signaling Technology (Beverly, MA). After washing in 1 m Tris (pH 7.4), 5 m NaCl, and 0.05% Tween 20 buffer (4 × 10 minutes, reverse transcription), filters were incubated for 1 hour at room temperature in Blotto A containing 1:4,000 dilution of peroxidase-conjugated antirabbit secondary antibody (Amersham Biosciences Corp., Piscataway, NJ) or Blotto A containing 1:4,000 dilution of peroxidase-conjugated antimouse secondary antibody (Amersham Biosciences Corp.). After washing in 1 m Tris (pH 7.4), 5 m NaCl, and 0.05% Tween 20, enhanced chemiluminescence was done according to the manufacturer’s recommendation. Protein expression was quantitatively analyzed via laser-scanning densitometry with NIH Image Version 1.61 software. All results were normalized to β-actin.

Cells were cultured on Lab-Tek chamber slides (Nunc Inc., Naperville, IL) under normal and methionine-stress conditions. Slides were blocked with 3% BSA in PBS (pH 7.4) with 0.25% Tween 20 for 30 minutes. Slides were then incubated overnight at 4°C in primary antibodies, mouse anti-p65 and rabbit p50 (Santa Cruz Biotechnology), diluted 1:100 with the blocking buffer. After washes in PBS (pH 7.4) with 0.25% Tween 20 (3 × 15 minutes), the cells were overlaid with secondary antibodies, Cy2-conjugated antimouse IgM and Cy3-conjugated antirabbit IgG, diluted 1:1000 in blocking buffer. After three washes in PBS (pH 7.4) with 0.25% Tween 20 (3 × 15 minutes) and one wash in PBS (1 × 15 minutes), the slides were mounted with aqueous mounting media with antifade and 4′,6-diamidino-2-phenylindole (VectaShield, Vector, Burlingame, CA) and visualized with a Zeis epifluorescence microscope.

Fig. 1 illustrates the response of the tumor cell lines DAOY, SWB77, SWB40, SWB39, and SWB61 to methionine-stress culture conditions. Replacement of methionine with homocysteine resulted in marked decrease of mitotic index in all of the cultures. Cell death was most prominent in SWB61 followed by DAOY, SWB40, SWB39, and SWB77, in which cellular death was protracted and sporadic. In contrast, the control 3T3 continued to proliferate in homocysteine (data not shown). Massive cell death was evident in SWB61 and DAOY after 2 days of methionine stress, whereas longer periods were required for SWB40 and SWB39 to respond. SWB77 overcame an initial mitotic arrest and, like the control 3T3 (not shown), it adapted to methionine stress assuming a different morphology. In addition to cell division, methionine stress inhibited cell migration and colony formation. Cell migration and colony formation were restored by replenishing methionine, and mitosis was evident in or near colonies but not in migrating single cells (Fig. 2). The effect of methionine stress on the cell cycle is shown in Table 1. SWB61 and DAOY displayed a marked G₁ arrest and a repression of the S phase after methionine withdrawal, respectively. SWB39 showed a G₂ arrest followed by a G₁ arrest after methionine withdrawal. SWB40 showed a S-phase increase concomitant with G₁ depletion after several days in methionine-void medium. The peak of cell cycle arrest coincided with increased apoptosis in all of the methionine-dependent lines tested, and such correlation was more pronounced in SWB61 and DAOY. The decline in tumor cell proliferation under methionine stress cannot be attributed to a single cell cycle block and may involve additional factors, such as failure to complete mitosis. The low mitotic index in methionine-stressed tumor lines combined with the decline of G₂ cells in DAOY (5 days), SWB61 (2 days), and SWB40 (4 days) indicate that these cells may die during an attempt to divide.

The relative gene expression in cell lines grown under control conditions or maintained under methionine stress for a period of time required to induce a peak of cell cycle arrest 24 hours before a sharp increase of the apoptotic index (Table 1) was determined by c-DNA array analysis. Genes, the expression of which was significantly changed (P > 0.05; see supplement), were classified based on their function, such as cell cycle related, p53 related, related to apoptosis/antiapoptosis, growth factor/angiogenesis related, cytokine dependent, and so forth.

The most notable changes in gene expression induced by methionine stress in DAOY are described below.

Down-regulated: cyclins A1 (×2.6), A2 (×2.2), B1 (×2.1), B2 (×2.9), D2 (×1.8), CDC2 (×2.8), CDC25A (×1.9), and B (×2.0), proliferating cell nuclear antigen (PCNA; ×1.5), E2F-4 transcription factor (×1.4), and CDC2-related protein kinase (×1.4); and up-regulated: Cyclins D1 (×3.9) and D3 (×2.1) and cyclin-dependent kinase (CDK)6 (×6.6).

Up-regulated: MAD1 (×2.7), CDC5L (×1.4), and CDC14A2 phosphatase (×5.9); and down-regulated: BUB1 (×2.0) and 3 (×1.3), MAD2 (×2.0), CKS1 (×2.0), CDC20 (×2.6), and NET1 (×2.2).

Up-regulated: p21 (×2.9), SKP1 (1.6), GADD34 (×3.4), GADD45α (×3.1), BRCA-1 (×1.9), 14-3-3σ (×1.6), and GADD45γ (×53).

Up-regulated: BCL1 (×4), BCL-2 interacting killer (BIK; ×2), TRAIL (×32), TRAIL-R2 (×1.9), and DR6 (×43).

Up-regulated: TNF-α (×1.7), TRADD (×8.5), TRAF-interacting protein (I-TRAF; ×3.1), and TNFSF10 (×36); and down-regulated: inhibitor of apoptosis protein (IAP)-β.

Induced or up-regulated: IL-24 (MDA7 or ST16) (×89), SCYA20 (×322), SCYA7 (×11), interleukin (IL)-1B (×24), IL-1A (×25), IL-6 (×24), IL-8 (×80), SGRF (×5.1), IL-11 (×7.4), IL-1R type I (×7.5), and type II (×7.2).

Down-regulated: IL-1 signaling factors Tollip-like receptor (TLR)3 (×5.3), interleukin receptor-associated kinase (IRAK; ×1.7), TAB1 (×2.5), and myeloid differentiation primary response protein (MyD88; ×1.9); and up-regulated: IL-1B converting enzyme γ (×2.3), δ (×3.0), ε (×27.4), ICE (×26), and MAP kinase kinase (MKK)3 (×2.1).

Induced or up-regulated: IFN-α (×4.1), IFN-β (×73), IRF1 (×4.8), IRF3 (×2.2), IRF7 (×1.7), IFN-inducible protein kinase (PKR; ×2.1), IFN-γ receptor α chain (×2.1), IFN-γ receptor I (×1.7) and II (×2.1), IFN-γ-inducible indoleamine-dioxygenase (IDO; ×3.6), IFN-γ inducible protein (IP-30; ×1.9), IFN-induced transmembrane protein II (×2.1), IFN-γ–inducible mRNA (×3.9), and ISG15 (×6.9).

Down-regulated: IGF2 (×2.3) and IGFBP5 (×8.3); and up-regulated: IGFBP4 (×1.9) and IGFBP7/mac25 (×54).

Up-regulated: TGF-β (×3.8) and the inducible early growth response protein α (TIEG; ×3.2); and down-regulated: TAK1 (×16.7).

Up-regulated: NF-κB p50 (×2.8), JUN-B (×4), c-JUN (×3.9), plasminogen activator inhibitor (PAI)2 (6.6), and c/EBP-β (×1.9); and down-regulated: c-FOS (×2).

Down-regulated: topoisomerases II (×2) and III (×1.5), thymidine kinase (×1.6), MAP4 (×2), stathmin (×1.8); and up-regulated: integrin (×2.0), inositol 1, 4, 5-triphosphate receptor type I (×4.4), and adrenomedullin (×5.1).

Up-regulated: p53 (×4.7), p21/Waf-1 (×2.6), GADD45α (×3.7), BRCA1 (×3.3), GADD34 (8.3), 14-3-3 η (×1.4), ε (x7.3), σ (x1.6), cyclin D1 (×2.5) and D3 (×2.9), CDK4 (×6.1), and SKP1 (×1.7); down-regulated: CDC25B (×1.8), cyclins A1 (×4), A2 (×2.2), B1 (×7.1), and B2 (×5.3), CDC2 (×2.3), and its kinase (×1.7), CDC25B (×1.8), ASK (2.6), and DNA polymerase α (×7.1).

Up-regulated: CDC5L (×1.3), MAD1 (×1.7), tumor endothelial marker precursor (TEM)1 (2.4), and CDC5L (1.3); and down-regulated: MAD2 (×1.5), CDC14B (×4), and BUB1 (2.6).

Up-regulated: BCL-1 (×15), BAX δ (×21), NIP3 (×2.7), TRAIL (×5.5), TRAIL-R 2 (×2.9), and DR6 (×8.6); and down-regulated: BCL-2 (×5).

Up-regulated or induced: TGF-β (×13), β2 (×3), TIEG (×40), TGF-β receptor associated protein 1 (×4.5), SMAD5 (×2.6), SMAD2 (×10), JUN-B (×20), and tissue inhibitor of metalloproteinases (TIMP; ×31); and down-regulated: activin receptor-like kinase 4 (ALK4; ×4), TAK1 (×2.8), SMAD3 (×2), TIAF (×2.3), c-FOS (×3.0), MMP1 (×6.7), and phospholipase c (×7.7).

Down-regulated: IGF-2 (×5.9); and up-regulated: IGFBP6 (×8.8) and IGF-2R (×2.3).

Down-regulated: TNF-α (×10) and TRADD (×1.8); and up-regulated: TACE (×3.5), TRAF-2 (×6.3), TRAF-6 (×6.1), I-TRAF (×2.4), and TNF-R2 (×1.7).

Down-regulated: IL-1B (×4.4), IL-11 (×2.3), IL-10 (×10), IL-13 (×12), IL-4 (×4.4), IL-8 (×1.6), the receptors IL-1R2 (×3.5), IL-8RB (×3.9), IL-5RA (×9.1), IL-10R (×20), IL-11RA (×16.7), IL-9R (×6.7), TAB (×3.3), and Tollip (×100); and up-regulated: IL enhancer binding factor 3 (×2.8), TACE (×6.1), IRAK (×2.1), NIK (×9.4), MyD88 (×5.8), Iκ kinase (IKK)-α (×1.4), and IKB-ε (×2).

Up-regulated: IRF1 (×5.1), IRF3 (×3), IRF5 (3.1), and IRF7 (1.7), the IFN-γ–receptor α chain (×2.9), IP-3 (×2.1), AF-1 (×3.1), FRD1 (×3.9), growth-arrest-specific protein (GAS; ×4.1), and ISG-15 (×2.8); and down-regulated: IFNα (×25), ISG3 (×1.3), IRF2 (×1.4), IRF6 (×6.7), and PKR (×2.7).

Down-regulated: extracellular signal-regulated kinase (ERK1; ×1.6), ELK1 (×2), R-kB (×3.5), cAMP responsive element-binding protein (CREB; ×2.5), ELK1 (×1.7), PAI2 (×5), and c-FOS (×3.0); up-regulated: NF-κB p50 (×15), p65 (×3.8), JUN-D (×1.6), JUN-B (19.7), c-JUN (×6.9), v-Fos transformation effector (FTE)1 (×2.9), B-cell translocation gene (BTG)1 (×3.1), PC4 (×3.9), ERK3 (×2.1), ELK3 (10.3), PAI1 (3.6), c/EBP-β (×3.6), c-EBP-γ (×3.2).

Down-regulated: phospholipase c (7.7), MAP4 (×1.9), separatin (×3.6), topoisomerase I (×3.0), and II (×3.6); and up-regulated: integrin (×1.4) and adrenomedullin (×3.2).

Down-regulated: cyclins A1 (×3.6), A2 (×2.9), B1 (×8.3), B2 (×5.3), D3 (×1.5), CDC2 (×2.4), CDK5 (×1.6), CDK6 (×2.6), CDK4 (×4.3), SKP1 (×1.6), CDC25A (×1.5), and B (×4.0), and DNA polymerase α (×2.5); and up-regulated: cyclins D1 (×5.5).

Up-regulated: MAD1 (×1.6) and CDC5L (×1.2); and down-regulated: BUB1 (×12.5), BUB3 (×1.8), MAD2 (×2.0), CKS1 (×1.5) and 2 (×2.1), CDC20 (×4.6), CDC14A2 phosphatase (×3.0), and NET1 (×2.0).

Up-regulated: p21 (×2.7), GADD34 (×3.2), GADD45α (×1.4), BRCA1 (×1.4), 14-3-3σ (×1.6), and GADD45γ (×2.1).

Up-regulated: BCL-1 (×2.8), BAX δ (2.1) and β (×2), TRAIL-R2 (×3.6), and DR6 (×2.6); and down-regulated: BCL-2 (×2.0) and IAP-β (×2).

Down-regulated: TNF-R2 (×1.4); and up-regulated: TACE (×1.8) and I-TRAF (×2.4).

Up-regulated: IL-8 (×15) and the interleukin 13 receptor IL-13R (×2.4).

Down-regulated: IRAK1 (×1.3), TAK-1 binding protein (TAB-1; ×1.9), and MyD88 (×2); and up-regulated: IL-1B converting enzyme γ (×2.1), IKB-ε (×2), and NIK (×1.8).

Induced or up-regulated: IFN regulatory factors 1 (IRF-1; ×2.4), IRF-5 (×1.7), and IRF-7 (×69), IFN-inducible protein kinase (PKR; ×1.9), IFN-γ as reflected by up-regulation of IFN-γ–receptor α chain (×1.6), FRD1 (×3.2), and IFN-γ–inducible protein (IP-30; ×1.4).

Down-regulated: IGFBP4 (×4.0), IGFBP6 (×2.4), and IGFBP5 (×2.5); and up-regulated: IGF-2R (×1.8).

Up-regulated: TGF-β (×11), TIEG (×7), and SMAD5 (×2); and down-regulated: TAK1 (×1.4) and SMAD6 (×4.0).

Up-regulated: JUN-D (1.7), FTE1 (1.6), BTG1 (×2.3), PC4 (×3.2), CREB (×2.0), c/EBP β (×2.2) and γ (×5.6), and ERK3 (×1.5); and down-regulated: c-JUN (×1.3), JUN-B (×4), ERK1 (×1.7), and c-FOS (×5).

Down-regulated: topoisomerase II (×2), dehydrofolate reductase (×6.7), thymidine kinase (×3.9), adrenomedullin (1.6), MAP4 (×1.5), and stathmin (×2.3); and up-regulated: inositol 1, 4, 5-triphosphate receptor type I (×23) and integrin (×1.5).

Down-regulated: cyclins A2 (×5.6), B1 (×2.7), B2 (×4.6), and D3 (×1.7), CDC2 (×2.3), CDK6 (×1.9), CDC25A (×2) and B (×1.8), and CDC2-related protein kinase (×1.5); and up-regulated: cyclins D1 (×1.2).

Up-regulated: MAD1 (×2.4) and CDC5L (×1.2); and down-regulated: BUB1 (×3.5) and 3 (×1.9), MAD2 (×2.9), CDC28 kinase (CKS2; ×1.6), and CDC20 (×2.1).

Up-regulated: p21 (×2.1), GADD34 (×2.3), GADD45α (×2.4), BRCA1 (×2.4), BRCA2 (×3), 14-3-3 ε (×2.3), and GADD45γ (×6.1); and down-regulated: p53 binding protein (×2.9) and SKP1 (×2.3).

Up-regulated: BCL1 (×1.4), NIP3 (5.5), TRAIL (×7), TRAIL-R2 (×2), and DR6 (×3.7); and down-regulated: IAP-β (×2).

Up-regulated: TRADD (×11), I-TRAF (×3.6), and TNFSF10 (×1.8); and down-regulated: TNF-R2 (×1.5).

Induced or up-regulated: IL-24 (MDA7 and ST16) (×21), SCYA20 (×320), IL-1A (×3.3), IL-6 (7.3), IL-8 terminal variant (×6.4), and the interleukin receptors IL-1R type I (×4.0).

Down-regulated: IRAK (×2.1); and up-regulated: IL-1B converting enzyme γ (×2.3), δ (×3.3), ε (×2.9), and IKB ε (×1.6).

Induced or up-regulated: IRF-1 (×1.8), IRF-3 (×2.1), IRF-5 (1.7), IRF-7 (2.5), GAS (×2), IFN-γ receptor I (×1.6), and ISG15 (×4.5); and down-regulated: IRF 6 (×1.9), IFN-inducible peptide (×16.7), and IFN-inducible protein kinase (PKR; ×2.6).

Down-regulated: IGFBP-6 (×1.5); and up-regulated: IGF-2R (×1.5) and IGFBP4 (×1.8).

Up-regulated: TGF-α (×5.5), TGF-β (×6.6) and of the inducible early growth response protein α (TIEG; ×2.7), and TMP (2.2); and down-regulated: TGF-β2 (×2.2) and TAK1 (×2.0).

Up-regulated: NF-κB p50 (×1.6), p65 (×1.5), BTG1 (×2), PC4 (×1.3), CREB (×2) and c/EBP-β (×2.4); and down-regulated: JUN-B (×3.1) and ELK1 (×1.5).

Down-regulated: topoisomerase II (×2.7) and III (×1.5), stathmin (×1.8); and up-regulated: inositol 1, 4, 5-triphosphate receptor type I (×2.4), BCL-6 (×2), and adrenomedullin (×2.3).

Down-regulated: cyclins A2 (×1.9), D1 (×1.5), CDC2 (×1.3), CDC25A (×3.3), and CDC25B (×1.5); and up-regulated: cyclins D2 (×2.3) and D3 (×2.4).

Down-regulated: MAD1 (×3.5) and CDC20 (×1.8).

Up-regulated: GADD45α (×1.4) and BRCA-1 (×2.1); and down-regulated: p21 (×2.3), E2F-4 (×1.4), and 14-3-3 σ (×1.4).

Down-regulated: TRAIL-R2 (×1.6).

Up-regulated: TRADD (×1.7).

Down-regulated: IL-11 (×5.9).

Induced or up-regulated: IFN-γ–receptor α chain (×2.6), IP-30 (×1.7), and FRD1 (×1.8).

Up-regulated: TGF-β (×1.4) and TGF-β–activated kinase 1a (×1.6); and down-regulated: TIEG (×3.1).

Up-regulated: JUN-D (×1.4), FTE1 (×1.4), and PC4 (×1.8).

Up-regulated: inositol 1, 4, 5-triphosphate receptor type I (×16) and adrenomedullin (×2.9); and down-regulated: topoisomerase II (×10) and III (×1.3).

Validation of Affymetrix chip hybridization was done with real-time PCR for several genes of interest, and expression was determined in comparison to β-actin (Table 2). There was a 94% agreement with the cDNA microarray analysis. According to reverse transcription-PCR data, TRAIL was up-regulated only in SWB39 and DAOY, whereas TRAIL-R3 was up-regulated in SWB40. BCL-XL was up-regulated in DAOY. Expression of p50-NF–κB and NF-κB–R were markedly up-regulated in DAOY, SWB61, and SWB39 by methionine stress. Examination of levels of expression of caspases (not provided by Affymetrix hybridization chip) indicated up-regulation of caspase 1 in DAOY and SWB61 and of caspase 2 in SWB40 by methionine stress, whereas caspase 3 was expressed marginally in all of the tumors and was modestly up-regulated only in DAOY under methionine stress. Similarly, robust up-regulation of caspase 8 was observed in DAOY and SWB40, whereas its expression was down-regulated in SWB61 and SWB39 under methionine stress. One of the genes that were strongly affected by methionine stress was the tumor suppressor/cytokine MDA7 (also known as IL-24). In accordance with c-DNA oligoarray analysis, low levels of MDA7 expression were present in all of the CNS tumors tested; however, substantial up-regulation by methionine stress occurred in DAOY and SWB39 and to a lesser extent in SWB40 and SWB61 (Fig. 3). No MDA7 signal was detected in the methionine-resistant SWB77 (Fig. 3). MDA7 mRNA levels increased progressively to reach maximum expression at days 6 to 8 in DAOY. Replenishing methionine resulted in down-regulation and loss of signal (at day 2). Relative ratios of MDA7 to actin were <0.005, 0.08, 0.21, and 0.30 at days 0, 2, 4, and 6, respectively, illustrating the rapid and substantial (>60-fold) increase.

Induction of MDA7 mRNA expression by methionine stress resulted in the synthesis of MDA7 protein in DAOY and SWB39 with concomitant activation of p38 (Fig. 4; Table 3). Although low levels of MDA7 mRNA were observed in SWB61 and SWB40, protein was not detectable by Western blot analysis. Induction of MDA7 protein in DAOY and SWB39 was accompanied by phosphorylation of p38, with no increase in total p38 expression. In contrast, p38 activation was inhibited in methionine-resistant SWB77. An increase (P < 0.05) in GADD45 protein was observed in DAOY under methionine stress. GADD45 protein was more pronounced in p53-efficient tumors SWB61 and 40 and to a lesser extent in mut p53 SWB39, indicating that GADD45 induction is p53 dependent. GADD34 was markedly up-regulated in SWB61 and to a lesser extent in SWB40, DAOY, and SWB39. Similarly, methionine stress resulted in the increase of p21 protein levels in all of the lines except in SWB77. The p21 synthesis was induced by methionine stress regardless of p53 status and was pronounced in SWB61 (wt p53), DAOY, and SWB39 (mut p53). In contrast to MDA7 and the GADDs, the Rb protein and its phosphorylated forms were substantially suppressed (Table 3; Fig. 4). In methionine medium, levels of unphosphorylated Rb were high in SWB61 and 40 and less abundant in the glioblastoma multiformes SWB77 and 39, whereas such levels were markedly lower in DAOY. Under methionine stress, unphosphorylated Rb accumulated in all of the cell lines compared with its hyperphosphorylated form in the following order: DAOY = SWB61>SWB39>SWB40>SWB77. Methionine stress reduced phosphatidylinositol 3′-kinase (PI3k)/AKT phosphorylation at ser 473 and 308 in DAOY, SWB40, and SWB77; however, phosphorylation of AKT under methionine-stress conditions was more prominent in SWB77 than in all of the other tumors examined, regardless of culture conditions. Examination of the Affymetrix array data and the protein levels determined in this study suggests that RNA levels reflect protein synthesis in some cell lines but not in all of them. A reason for such discrepancy is not fully understood, but in this case it may be because of measuring gene expression and protein levels at the same time point. Therefore, it is possible that protein synthesis in response to RNA changes was occurring but had not reached its zenith at the time the measurement was made. In this regard small changes in protein levels between controls and methionine-stressed lines should not be disregarded, as they may increase in magnitude with time.

Active NF-κB, as determined by colocalization of NF-κB subunits p50 and p65 in the nucleus, was observed in all of the cell lines tested and in SWB77 in particular (Fig. 5). Methionine withdrawal resulted in the failure of NF-κB subunits to enter the nucleus in SWB61 and DAOY. Methionine stress increased accumulation of NF-κB subunits in the cytosol. These findings strongly suggest that methionine stress causes substantial inhibition of NF-κB activity.

Methionine stress induces an arrest in cell proliferation with concomitant cell cycle blocks and induction of cell death, loss of motility, and failure to form colonies. These antitumor phenotypes are related to a range of altered signaling mechanisms. Although functional studies are required to pinpoint the exact pathways mediating the responses to methionine stress, the work presented here provides a comprehensive overview of the patterns of gene regulation in response to methionine stress. The major signal transduction pathways in CNS tumors that are affected by methionine stress and differences among CNS tumors in response to such stress are classified below.

Up-regulation of TGF-β by methionine stress in all of the methionine-dependent tumors tested coupled with the induction of p53 and GADD45 in p53-competent tumors inhibits early progression through G₁. Both the G₁ to S and the G₂ to M transitions are blocked by the up-regulation of p21 (9, 10). Up-regulation of BRCA1 and 14-3-3σ is also expected to exert an inhibitory effect on G₂-M progression of cell cycle, regardless of p53 status, by binding to phosphorylated CDC25. Down-regulation of CDC2 expression by methionine stress is indicative of the inhibitory effect of such stress on S entry in all of the methionine-dependent tumors and to some extent in SWB77. The most notable events in consequence to methionine stress in methionine-dependent tumors tested are the down-regulation of cyclin A1 and 2, cyclin B2, CDC2 and CDC25 (A and B), and BUB1, which is in agreement with the observed inhibition of mitosis. Down-regulation of CDC20 and that of CDC28 kinases is expected to contribute to the induction of mitotic checkpoints. Of particular interest is the marked reduction of P-Rb species, as compared with free Rb, which is expected to prevent entry into the S phase under methionine stress.

The transcriptional families that are affected during methionine stress are identified as follows: (a) AP-1, which regulates transcriptional control of c-JUN, JUN-B, c-FOS, and so forth; (b) c/EBP-β or γ; (c) the NF-κB subunits, p50 and p65; and (d) the early response gene BTG-1, c-MYC, and KROX-20. The c-JUN and JUN-B are up-regulated in DAOY and SWB61 by methionine stress and are also expressed in modest levels in all of the other tumors. A highly expressed in CNS tumors c/EBP-β (NF-IL6) is markedly up-regulated by methionine stress in all of the tumors, whereas c/EBP-γ is elevated in SWB61 and SWB40 but not in DAOY or SWB39. Activation of JNK pathway by methionine stress is in agreement with the overexpression of c-JUN (DAOY and SWB61) and c-MYC (SWB61 and 40), and it may be linked to the up-regulation of MKK4. The down-regulation of expression of ELK1 in SWB61 and 39 and of c-FOS in all of the methionine-dependent cell lines may be related to a modest down-regulation of PI3K by methionine stress. Transcriptional activation of NF-κB subunits p50 and p65 by methionine stress, in DAOY, SWB39, and SWB61 may be attributed to some extent in the persistent PI3K function and AKT phosphorylation in these three lines but not in SWB40. Methionine stress enhances the expression of NF-κB subunits by a control that might also be related to interleukin production, but it fails to maintain its activation, which requires not only a functional PI3K but also the participation of H-Ras/MKK-3 activity, which is diminished under methionine stress (Fig. 6 A). Failure to maintain active NF-κB is possibly linked to the suppression of E2F-1 response, which is reflected by the down-regulation of cyclins A/E and CDC2 in all of the methionine-dependent cell lines and by the diminished expression of polymerase α, dehydrofolate reductase, and thymidine kinase (11). Suppression of E2F-1 response is supported by the increase of Rb to P-Rb ratios, and it may be related to high levels of expression of p21 under methionine-stress conditions that prevent the dissociation of E2F-1 from Rb, despite declining levels of the latter. Up-regulation of p21 function is in accordance with the suppression of topoisomerase II, cyclin B, MMP1, MAP4, and MAD1 in the p53 competent lines SWB61 and 40 and to a lesser extent in DAOY and SWB39.

Interleukins and their receptors are generally up-regulated in DAOY, SWB40, and SWB39 by methionine stress. Most remarkable is the induction of IL-8 in DAOY and SWB40. It is likely that IL-8 induces an AP-response in DAOY and SWB40 that is consistent with inhibition of NF-κB subunit transcription and subsequent activation (12, 13, 14). Methionine stress also results in up-regulation of the already highly expressed NF-IL6 and subsequent increase of the adrenomedullin signal (15) in all of the methionine-dependent cell lines. High expression of NF-IL6 and adrenomedullin is in agreement with the active involvement of phospholipase C and protein kinase C in SWB77, regardless of culture conditions. An impressive induction/up-regulation of the cytokines SCYA7, SCYA 20, and ST16 (MDA7 or IL24) is observed only in DAOY and in SWB39. SCYA20 induction correlates with up-regulation of TNF-α and related signal transduction pathways in these two lines (16). The induction of MDA7 in CNS tumors by methionine stress may be responsible for the several cytotoxic responses of methionine stress in DAOY. MDA7, also known as IL-24, is a novel cancer suppressing and apoptosis-inducing gene of the IL-10 family (17) with broad antitumor properties (18). In contrast to its antitumor effect, its expression, like methionine stress, does not induce toxicity in normal cells (17, 18, 19, 20, 21, 22, 23). MDA7 induces toxicity and apoptosis in many tumor cells containing various genetic defects such as p53, p16/INK4a, and/or Rb (17, 18, 22, 23, 24). The mechanism of action of the tumor suppressor MDA7 in melanoma involves up-regulation of the GADD family of genes including GADD 45-α, GADD 34, and GADD 153 and to a lesser extent GADD 45-γ with subsequent activation of p38 mitogen-activated protein kinase (MAPK) and the increase of the ratio of BAX to BCL-2 leading to apoptosis (21). Up-regulation of expression of c-JUN in DAOY by methionine stress and the absence of an effect in SWB77 are in agreement with patterns of MDA7 levels in these lines. Previous studies have shown induction of c-JUN by adenoviral delivery of MDA7 treatment of lung cancer cells (23). Up-regulation of IFN-γ and -β, TNF-α, IL-1B but not IL-12 or GM-CSF in DAOY, additionally support a functional MDA7 pathway under methionine stress. Furthermore, previous studies have shown that MDA7 treatment of human peripheral blood mononuclear cell results in induction of TH1 secondary cytokines, including IL-6, TNF-α, IFN-γ, and so forth (25). This study is the first linking endogenous MDA7 expression to growth control in CNS tumors, and correlates with previous studies in melanoma (17, 20). It is interesting that all of the methionine-dependent cell lines tested expressed low levels of MDA7 mRNA under normal growth conditions; however, when induced by methionine stress, both MDA7 mRNA and protein expression were rapidly and coordinately activated.

The inactivation of NF-κB by methionine stress despite the rise of IL-1 in DAOY is of interest. IL-1, like TNF-α, is a master inflammatory switch that initiates synthesis of a variety of factors including IL-6 and IL-8 (26), both of which alert the immune system. The action of IL-1 is mediated through binding to the interleukin Toll/IL-1 receptor superfamily, a member of receptors responding to injury and infection (27), and also by the activation of the IRAK-TRAF6 signaling module, which ultimately causes the release and nuclear translocation of NF-κB (26, 27, 28, 29). The IL-1R–mediated signaling operates with the recruitment of Myd88 and Tollip, which activate IRAK-4, a kinase that phosphorylates and activates IRAK-1 (30). The module Myd88-IRAK–TRAF6 also controls signaling of p38 mitogen-activated protein kinase (MAPK), which, however, can be modulated by an IRAK-independent fashion involving Ras, TAB1-TAK1, and MKK3 with a direct effect on the activation of the transcription factors AFT2 and ELK1 (31). Activated TRAF and IRAK also interact with TGF-β–activated kinase TAK1 and with two TAK1 binding proteins, TAB1 and TAB2. Interaction of TAK1 and TAB1 with TRAF6 and IRAK induces TAK1 kinase, a mitogen activated protein kinase kinase kinase initially identified in the TGF signaling path. In turn, the activation of TAK1 results in phosphorylation of IKK and MAPK kinase (MKK3, 4, and 6) and inhibition of I-TRAF, leading to activation of NF-κB. An IRAK-independent pathway in NF-κB activation that uses TLR3, TRAF6, TAK1, TAB2, and PKR has been proposed, and it perhaps involve IKK, an inhibitor of I-TRAF (31, 32). I-TRAF is phosphorylated by inducible IKK-I and releases TRAF-2 from the I-TRAF.TRAF2 complex, which is then involved in activation of NF-κB (33). According to the above, the down-regulation of Myd88, IRAK-1, TAB1, and TAB2, when combined with the lack of IKK activation and the up-regulation of I-TRAF, are in agreement with inactivation of NF-κB under methionine stress in DAOY and other methionine-dependent CNS tumor cell lines.

Methionine-stress results in a modest increase of TNF-α expression in DAOY. In the rest of the cells tested, TNF-α was expressed weakly. However, up-regulation of expression of TNF receptors in DAOY, SWB39, and especially SWB61 is likely to enhance TRAF2 degradation, an event that could determine the dominant cellular response phenotype of methionine stress via the TNF pathway (34, 35). That phenotype is proapoptotic, as contrasted with the proliferative NF-κB pathway supported by TRAF2. The most important event associated with TNF-α is the up-regulation of TRAIL in DAOY and SWB39 and that of TRAIL receptors in SWB61, 40, and 39 but not in SWB77. Therefore, the apoptotic response to methionine stress in methionine-dependent tumors is related to up-regulation of TRAIL (Fig. 6 B), which in conjunction with DR6 and DR5 and the subsequent recruitment of FADD should result in caspase-8 activation (36). Of particular significance is that regulation of genes associated with the above pathways corroborates the observation that nuclear translocation of NF-κB subunits is abolished under methionine stress in DAOY despite marked up-regulation of expression of inflammatory cytokines. In addition, activation of NF-κB has been shown to inhibit the TNF-mediated apoptosis by triggering the expression of BCL-2 family of proteins and of the small death effectors domain containing the authentic protein NDED, also known as GG2–1 or SCC-S2 (37, 38). Sustained inhibition of NF-κB by I-kB suppresses the TNF-induced expression of NDED and IAP and facilitates the TNF-mediated activation of caspase 8 (37, 38). The TRAIL pathway is particularly favored in SWB61, in which high levels of expression of BAX (δ) and down-regulation of NDED and BCL-2 are in agreement with the activation of TNF receptor to suppress the NF-κB function.

Up-regulation of IRF-1 by methionine stress is expected to promote apoptosis by regulating the expression of caspases, Fas/CD95, Fas ligand, and BCL-2 (39, 40, 41, 42, 43, 44, 45). The robust inductive effect of methionine stress on TRAIL and DR6, and a correlation between IRF-1, TRAIL and DR6 up-regulation under methionine stress in DAOY, suggests that IRF-1 may precede the induction of TRAIL (Fig. 6 C) and that a pathway involving both IRF-1 and TRAIL could be a major route leading to apoptosis (46, 47). This response is not fully operational in SWB61 in which up-regulation of IRF-1 does not enhance expression of IFN-β and appears to proceed exclusively via the IFN-γ–mediated pathway. The antiproliferative and proapoptotic activity of IRF-1 is also mediated via the activation of p21, which induces growth arrest in G₀/G₁ by inhibiting cyclin A/CDK2 kinase (48, 49). Interferon-mediated growth arrest via the induction of p21 is most likely to occur in DAOY, in which p21 protein levels rise by at least 20-fold. Induction of PC4 in SWB61 and SWB40 by IRF-1 or IFN-α/β additionally limits promoter-driven transcription as well as nonspecific transcription, which is additionally enhanced by the nonexpression of TFIIH and the suppression of ERCC3 expression (50). Another pathway involving IRF-1 in DAOY could be mediated by the induction of IFN α and β, which up-regulate synthesis of ISG15 and suppress proliferation (49, 50).

TGF-β is a potent growth inhibitor exerting its effect by binding to specific type I and II receptors, which in turn initiate intracellular signaling via the SMAD proteins. Phosphorylated SMADS translocate to the nucleus and modulate target gene transcription resulting in down-regulation of c-JUN, c-FOS, c- MYC, CDC25, up-regulation of p21, p15, and the plasminogen activator inhibitors 1 and 2 (51, 52, 53, 54, 55, 56). In addition, TGF-β initiates signal transduction pathways, including AKT (AKT phosphorylation at Ser-473), stress-activated protein kinase/JUN-N terminal kinase (SAPK/JNK), AP-1, p38/MAPK, ERK 1 and 2, and MAPK (57, 58). TGF-β signal is up-regulated by methionine stress predominantly in SWB61 and 40 (Fig. 6 D), indicating that regulation of growth in these lines under methionine stress is controlled by the TGF-β pathway. The R-SMADs, SMAD 2 and 5 are induced in SWB61 and 40, whereas the I-SMADS, SMAD 6 and 7 are transcriptionally down-regulated or unaffected under methionine stress in DAOY and SWB40 in particular. This pattern of SMAD activation could be a response to the up-regulation of the TGF-β–inducible early gene (TIEG), a transcriptional regulator of thyroid hormone receptor as well as of the TGF-β receptor itself (51, 59, 60, 61). TIEG amplifies the effect of R-SMADS, triggers expression of p21, enhances TGF-mediated expression of SMAD2, and represses expression of SMAD6 (61, 62, 63, 64). In addition, SMAD-dependent mechanisms are responsible for the activation of both the c-JUN and JUN-B promoters by TGF-β (63), which could also apply in the case of the response of glial tumors to methionine-stress. TGF-β has been shown to repress MMP1 by altering ratios of B-JUN, JUN-D, or c-JUN containing AP-1 complexes. The modulation of expression of these gene products under methionine stress is in accordance with the control of TIMP1 and MMP1 expression via the TGF-β pathway in SWB61 and SWB40 (63).

Given the many potential binding sites for p53 (>100), it is likely that tumor growth suppression by p53 is achieved by regulating the physiologic function of a multitude of genes (65). IRF-5 has been shown to be a direct target for p53, and as such, it may respond to stress that affects p53 expression (66). Methionine stress up-regulates p53-regulated genes GADD45, BRCA-1, and p21 in mut p53 DAOY and SWB39. This indicates that these genes are not always dependent on wt-p53 function and can be induced regardless of the mutational status of p53. The only exception is the resistant line SWB77, which did not respond to methionine stress and expressed the same or lower levels of GADDs under methionine stress. The ability of methionine stress to induce p21 and GADDs independently of wt-p53 is expected to steer p53-mutant cells toward death by acting in lieu of wt p53, thus causing a switch from a proliferative state to an apoptotic one. Nevertheless, the role of wt-p53 in enhancing the effect of methionine stress is evident, considering differences in the resistance between SWB61, in which p53 is up-regulated, and SWB40, in which methionine stress has no effect on its expression.

In summary, methionine stress results in partial suppression of PI3K/AKT and RAF/RAS proliferative/prosurvival pathways and inactivation of NF-κB in all of the methionine-dependent tumors. MDA7 protein levels increase by >60- to 100-fold in DAOY and SWB39, and p21 is up-regulated in these cells in a p53-independent manner. MDA7 induction and TNF-α activation of TRAIL, its receptor, and of DR6 are most likely responsible for apoptotic events in SWB39 and DAOY. Of particular interest is the marked induction of the cytokines IL-1, IL-8, IL-6, SCYA20, SCYA7, and IL-24 primarily in DAOY and to some extent in SWB39 by methionine stress. The up-regulation of IL-6 in DAOY and SWB61 by methionine stress could be associated to induction of IRF-1 and subsequently to up-regulation of p21 as well as to the activation of TRAIL and the TRAIL receptor 3. These events could lead to a p53-independent cell cycle arrest and death. The induction of PC4 and IFRD-1 in SWB61 and 40 could be linked to inhibition of transcription, whereas induction of ISG15 by IFN-β could lead to an inflammatory response under methionine stress. Activation of the IFN regulated pathways is fully supported by the induction of MIP1, MCP1, IL-8, and IP-10 genes in DAOY and to a lesser extent in SWB39 and 40. The wt-p53 in SWB61 may contribute directly to cell death via the induction of BAX and down-regulation of BCL-2. Whereas the MDA7, TNF, and IFN-regulated pathways are likely to be involved in induction of proapoptotic and antiproliferative responses in DAOY, the TGF-β pathway appears to be primarily involved in the toxicity of methionine stress in SWB61. The postulated mechanism for the TGF-β-mediated effect on SWB61 and to a lesser extent on SWB40, SWB39, and DAOY is mediated via the induction of the R-SMADS 2 and 5 and the down-regulation of I-SMAD 6. This mechanism is also supported by the up-regulation of expression of TIEG and by the marked induction of TIMP in SWB61. Cell cycle control in response to TGF-β is induced by the up-regulation of ERK2 and induction of p21 and PAI1.