Epigenetic status of argininosuccinate synthetase and argininosuccinate lyase modulates autophagy and cell death in glioblastoma

Abstract

Arginine deprivation, either by nutritional starvation or exposure to ADI-PEG20, induces adaptive transcriptional upregulation of ASS1 and ASL in glioblastoma multiforme ex vivo cultures and cell lines. This adaptive transcriptional upregulation is blocked by neoplasia-specific CpG island methylation in either gene, causing arginine auxotrophy and cell death. In cells with methylated ASS1 or ASL CpG islands, ADI-PEG20 initially induces a protective autophagic response, but abrogation of this by chloroquine accelerates and potentiates cytotoxicity. Concomitant methylation in the CpG islands of both ASS1 and ASL, observed in a subset of cases, confers hypersensitivity to ADI-PEG20. Cancer stem cells positive for CD133 and methylation in the ASL CpG island retain sensitivity to ADI-PEG20. Our results show for the first time that epigenetic changes occur in both of the two key genes of arginine biosynthesis in human cancer and confer sensitivity to therapeutic arginine deprivation. We demonstrate that methylation status of the CpG islands, rather than expression levels per se of the genes, predicts sensitivity to arginine deprivation. Our results suggest a novel therapeutic strategy for this invariably fatal central nervous system neoplasm for which we have identified robust biomarkers and which overcomes the limitations to conventional chemotherapy imposed by the blood/brain barrier.

Figure 1

Methylation-dependent transcriptional silencing of ASS1 in GBM. (a) Expression of novel candidate genes is upregulated by demethylation. The figure shows qPCR analysis of the indicated genes in GAMG cells treated (black) or untreated (clear) with 5′AZA. Experiments were performed in triplicate and data shown are mean fold increases (+/−1 S.D.) in 5′AZA-treated cells relative to control cells treated with dimethylsulphoxide. (b) 5′AZA upregulates ASS1 in GAMG but not 42MG cells. The top panel shows qPCR, the middle panel MSP and the bottom panel western blot analysis of ASS in GAMG and 42MG cells with and without exposure to 5′AZA as indicated. Actin is used as a loading control for the western blot. (c) qPCR and western blot analysis of ASS1 in primary GBM explants. qPCR was performed in triplicate and data shown are expression relative to GBM 7 (+/−1 S.D.). (d) MSP analysis of ASS1 CpG island in primary GBM explants. The figure shows unmethylated (U) and methylated (M) reactions for each case. Also shown are control U and M DNA samples modified in parallel with the experimental DNA samples. (e) Pyrosequencing analysis of ASS1 CpG island in primary GBM explants. The level of methylation in individual CpG dinucleotides is indicated by the intensity of shading as shown

Figure 2

Methylation-dependent transcriptional silencing of ASL. (a) qPCR analysis of ASL in primary GBM explants. qPCR was performed in triplicate and data shown are expression relative to GBM 6 (+/−1 SD). (b) MSP analysis of ASL CpG island in primary GBM explants. The figure shows unmethylated (U) and methylated (M) reactions for each case. Also shown are control U and M DNA samples modified in parallel with the experimental DNA samples. (c) Pyrosequencing analysis of ASL CpG island in primary GBM explants. The level of methylation in individual CpG dinucleotides is indicated by the intensity of shading as shown. (d) 5′AZA upregulates ASL in primary GBM 59 but not in primary GBM 6 cells. The top panel shows qPCR and the bottom panel MSP analysis of ASL in GBM 59 and GBM 6 cells with and without exposure to 5′AZA as indicated

Figure 3

Methylation in the ASS1 and ASL CpG islands blocks transcriptional upregulation upon arginine deprivation and confers arginine auxotrophy and sensitivity to arginine deiminase (ADI-PEG20) in primary GBM explants. (a) Arginine deprivation induces ASS1 and ASL mRNA in primary GBM cells, but this is abrogated by CpG island methylation. The indicated GBM explants were grown in the presence of ADI-PEG20 (1 μg/ml) or 5′AZA (1 μM) as shown. RNA was harvested after 48 h and subjected to qPCR analysis of ASS1 and ASL as indicated. Each experiment was performed at least twice and the values shown are means (+/−1 SD) relative to untreated control cells. (b) Western blot analysis of ASS in primary GBM cells. The indicated primary explants were grown in the presence of ADI-PEG20 (1 μg/ml) or 5′AZA (1 μM) as shown. ASS levels were analysed by western blotting after 48 h. Actin is used as a control protein. (c) Dose response curves for ADI-PEG20 in primary GBM of varying CpG island methylation status. Logarithmic phase primary GBM cells were exposed to the indicated concentrations of ADI-PEG20 and proliferation assessed by measurement of SRB as described in Methods. The most sensitive tumour is GBM 59 in which both ASS1 and ASL CpG islands are methylated and which is fully inhibited by 0.06 μg/ml ADI-PEG20. By contrast, GBM16 (ASS1 and ASL unmethylated) is unaffected by ADI-PEG20. GBM 27 and GBM 31 (either ASS1 or ASL CpG islands methylated) show intermediate sensitivity to ADI-PEG20

Figure 4

ADI-PEG20 kills ASS1 knockdown cells, ASL negative CSCs and induces autophagy. (a) ASS1 knockdown cells are sensitive to ADI-PEG20. Stable ASS1 knockdown cells (T98G-A1) were exposed to ADI-PEG20 and analysed for proliferation as described in Methods. Cells containing empty vector were treated similarly. Upper panel confirms knock down of ASS1 by qPCR and western blotting. T98G-A1 cells are more sensitive to killing than mock transfected cells, lower panel. (b) ADI-PEG20 kills CD133-positive GBM cells. DBTRG cells (ASS1 unmethylated, ASL methylated) were sorted for CD133-positive cells (left histogram, IgG isotype control PE antibody; right histogram, anti-CD133 showing 10.27% CD133-positive cells). CD133-positive and -negative fractions were then exposed to ADI-PEG20 (1 μg/ml) and proliferation assessed by SRB staining. CD133-positive and -negative fractions are equally sensitive to inhibition by ADI-PEG20, lower graph. (c and d) ADI-PEG20 induces autophagic markers. LN229 and T98G cells were exposed to ADI-PEG20 (1 μg/ml) and analysed for autophagic markers by western blotting (upregulation of Beclin 1 and Atg proteins, degradation of P62 and conversion of LC3-1 to LC3-11) (c) and by acridine orange staining (formation of acidic vacuoles) in the presence or absence of bafilomycin 48 h after treatment (d). Autophagic changes are observed in the ASS1 methylated cell line, LN229, only. Bafilomycin abrogates the formation of acidic vacuoles, which stain red in the absence of bafilomycin

Figure 5

Caspase-independent cell death is induced by ADI-PEG20 and is accelerated by chloroquine. (a) CQ inhibits autophagic markers. LN229 cells were treated with 1 μg/ml ADI-PEG20 with and without 10 μg/ml CQ and analysed for conversion of LC3-1 to LC3-11 and degradation of P62 by western blotting 48 h post treatment. CQ inhibits degradation of P62 and prevents autophagic flux as evidenced by accumulation of the LC3-11 band. (b) LN229 cells stably expressing GFP-LC3 or empty GFP vector were treated as above and analysed for punctate, autophagosome-associated LC3-11 GFP by fluorescence microscopy. (c) Inhibition of autophagy accelerates and enhances ADI-PEG20 induced cell death. Primary and GBM cell lines were treated with 10 μM CQ, 1 μg/ml ADI-PEG20 or in combination and analysed for sub G1 content by flow cytometry at various time points. (Primary lines – GBM 31 ASS negative, GBM 27 ASL negative, GBM 16 ASS and ASL positive; GBM cell lines – LN229 ASS negative, DBTRG ASL negative). CQ in combination with ADI-PEG20 accelerates and enhances cell death in ASS1/ASL negative lines only as evidenced by an increase in the sub-G1 compartment. The results shown are mean values from three independent experiments. (d) Caspase inhibitors do not prevent ADI-PEG20-induced cell death. LN229 and T98G cells were exposed to 1 μg/ml ADI-PEG20 and analysed for sub-G1 content as above. Where indicated, the caspase inhibitor Z-VAD-fmk was added to cells 24 h before addition of ADI-PEG20. Results are expressed as the mean percentage of live cells remaining from three independent experiments. ADI-PEG20 has no effect on cell death in the unmethylated cell line, T98G but induces marked cell death (78%) in the methylated line, LN229. The percentage of cell death induced by ADI-PEG20 is only partially blocked by Z-VAD-fmk (51%)

Figure 6

ASS and ASL in clinical cases of GBM. (a) Primary lines generated from tumour explants were stained for GFAP and hematoxylin and eosin. Representative examples are shown. (b) Immunohistochemistry showing ASS and ASL in clinical cases of GBM. (c) Immunohistochemical analysis for the microglial marker, Iba-1, in ASS-negative GBM. (d) MSP for ASS1 and ASL CpG islands in clinical cases of GBM and primary lines generated from them. The figure shows unmethylated (U) and methylated (M) reactions for each case. Also shown are control U and M DNA samples modified in parallel with the experimental DNA samples

Figure 7

Methylation of ASS1 and ASL CpG island is a predictive biomarker in GBM. To determine whether methylation of ASS1 and/or ASL affects outcomes of patients, Kaplan–Meier curves were used to estimate the probabilities of survival and the log-rank test to assess the statistical significance of differences in event rates using Prism 5. (a) Median overall survival (OS) for GBM patients with unmethylated ASS1 CpG island was 496 days versus 309 days for patients with methylated ASS1 CpG island (P=0.11). (b) Median OS for patients with unmethylated ASL CpG island was 443 days compared with 299 days for patients with methylated ASL CpG island (P=0.16). (c) Median OS for patients whose GBM showed methylation in the CpG islands of both ASS1 and ASL was significantly shorter than those in whom neither CpG island was methylated or only one of the two CpG islands was methylated (P=0.0468). (d) Median OS in cases with either CpG island or none methylated=436 days versus 299 days (P=0.18)