Arginine Deiminase Resistance in Melanoma Cells Is Associated with Metabolic Reprogramming, Glucose Dependence, and Glutamine Addiction

Published OnlineFirst August 26, 2013; DOI: 10.1158/1535-7163.MCT-13-0302

Molecular Cancer Cancer Therapeutics Insights Therapeutics

Yan Long1, Wen-Bin Tsai1, Medhi Wangpaichitr2, Takashi Tsukamoto3, Niramol Savaraj2, Lynn G. Feun2, and Macus Tien Kuo1

Abstract Many malignant human tumors, including melanomas, are auxotrophic for arginine due to reduced expression of argininosuccinate synthetase-1 (ASS1), the rate-limiting enzyme for arginine biosynthesis. Pegylated arginine deiminase (ADI-PEG20), which degrades extracellular arginine, resulting in arginine deprivation, has shown favorable results in clinical trials for treating arginine-auxotrophic tumors. Drug resistance is the major obstacle for effective ADI-PEG20 usage. To elucidate mechanisms of resistance, we established several ADI-PEG20–resistant (ADIR) variants from A2058 and SK-Mel-2 melanoma cells. Compared with the parental lines, these ADIR variants showed the following characteristics: (i) all ADIR cell lines showed elevated ASS1 expression, resulting from the constitutive binding of the transcription factor c-Myc on the ASS1 promoter, suggesting that elevated ASS1 is the major mechanism of resistance; (ii) the ADIR cell lines exhibited enhanced AKT signaling and were preferentially sensitive to PI3K/AKT inhibitors, but reduced mTOR signaling, and were preferentially resistant to mTOR inhibitor; (iii) these variants showed enhanced expression of glucose transporter-1 and lactate dehydrogenase-A, reduced expression of pyruvate dehydrogenase, and elevated sensitivity to the glycolytic inhibitors 2- deoxy-glucose and 3-bromopyruvate, consistent with the enhanced glycolytic pathway (the Warburg effect); (iv) the resistant cells showed higher glutamine dehydrogenase and glutaminase expression and were preferentially vulnerable to glutamine inhibitors. We showed that c-Myc, not elevated ASS1 expression, is involved in upregulation of many of these enzymes because knockdown of c-Myc reduced their expression, whereas overexpressed ASS1 by transfection reduced their expression. This study identiﬁed multiple targets for overcoming ADI-PEG resistance in cancer chemotherapy using recombinant arginine-degrading enzymes. Mol Cancer Ther; 12(11); 2581–90. 2013 AACR.

Introduction reactions in the urea cycle: argininosuccinate synthetase-1 (ASS1), which catalyzes the reaction of L-citrulline The amino acid L-arginine has many physiologic functions because it is a precursor for the biosyntheses of and aspartic acid to argininosuccinate and argininosucci- proteins, nitric oxide, polyamines, proline, glutamate, nate lyase (ASL), which recycles argininosuccinate back to creatine, and agmatine. It also plays important role in arginine with the production of fumaric acid, an interme- immune response to antigen by impaired cytokine pro- diate in the tricarboxylic acid (TCA) cycle (Fig. 1). duction and an arrest in T-cell proliferation (1, 2). Arginine Many advanced human tumors, including melanomas, is primarily synthesized by two sequential enzymatic renal cell carcinomas, hepatocellular carcinomas, meso- theliomas, prostate cancers, and small-cell lung cancers do not express ASS1 (3). ASS1-negative tumor cells therefore Authors' Afﬁliations: 1Department of Translational Molecular Pathology, require arginine from the circulation for survival, whereas The University of Texas MD Anderson Cancer Center, Houston, Texas; normal cells that synthesize arginine endogenously will 2Sylvester Comprehensive Cancer Center, University of Miami, Miami, Florida; and 3Department of Neurology and Brain Science Institute, Johns survive under arginine deprivation conditions. Arginine Hopkins University School of Medicine, Baltimore, Maryland deiminase (ADI; refs. 4–7) and human arginase 1 (4, 8, 9) Y. Long and W.-B. Tsai contributed equally to this work. two arginine-depleting recombinant proteins (Fig. 1), have been under clinical development for treating argi- Corresponding Author: Macus Tien Kuo, The University of Texas MD Mycoplasma Anderson Cancer Center, 1515 Holcombe Blvd., Box 951, Houston, TX nine-auxotrophic tumors. ADI is produced by 77030. Phone: 713-834-6038; Fax: 713-834-6085; E-mail: spp. and humans do not produce this enzyme. Like most [email protected] recombinant bacterial proteins, it is immunogenic and doi: 10.1158/1535-7163.MCT-13-0302 has a short half-life in vivo. Pegylated ADI (ADI-PEG20) 2013 American Association for Cancer Research. has been formulated for clinical use. Clinical trials using

www.aacrjournals.org 2581

Long et al.

Figure 1. The metabolic interrelationships of glycolysis, the TCA cycle, the urea cycle, and glutaminolysis. Upward arrows in the boxes denote that the expression is increased in the ADIR cell lines or can be induced by ADI-PEG20; downward arrows in the boxes denote downregulation. Glu-6-P, glucose-6-phosphate; OTC, ornithine carbamoyltransferase; PHD, prolyl hydroxylation.

ADI-PEG20 for treating various cancers including mela- DG), 3-bromopyruvate (3-BP), 6-diazo-5-oxo-L-norleu- noma have produced favorable results (5, 6). cine (DON), azaserine, deferoxamine mesylate (DFX), Drug resistance remains an important contributor for and CoCl2, from Sigma-Aldrich; perifosine from Cay- the treatment failure, owing to reexpression of the once- man Chemical; temsirolimus (CCI-779) from Selleck- silenced ASS1. Using a cultured melanoma cell system, we chem. Bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl) previously showed that transcriptional regulation of ASS1 ethyl sulfide (BPTES) was prepared according to a by ADI-PEG20 involves a switch of promoter binding, c- previous report (12). Myc, which functions as a positive regulator replaces Antibodies were obtained from the following sources: hypoxia-inducible factor 1a (HIF-1a), which functions as mouse anti-human ASS1 monoclonal antibody from a negative regulator (10, 11). Polaris Pharmaceuticals; rabbit anti-ASL and rabbit The current investigation was initiated to elucidate anti-c-Myc (N262), rabbit anti-MCT1 and MCT4 anti- resistance mechanisms of ADI-PEG20 in melanoma bodies from Santa Cruz Biotechnology; rabbit antiglu- cells. We developed several stable ADI-resistant (ADIR) cose transporter-1 (Glut1) from FabGennix; antibodies cell lines and found that compared with their parental against lactate dehydrogenase-A (LDH-A), pyruvate counterparts, these ADIR variants exhibited elevated dehydrogenase (PDH), acetyl CoA carboxylase (ACC), glycolytic metabolic activities and were preferentially fatty acid synthase (FAS), mTOR complex 1 (mTORC1), sensitive to the killing by glycolytic inhibitors. More- p70S6 kinase, p-p70S6 kinase (Thr389), 4E-BP, eIF2a,p- over, these ADIR cells also exhibited elevated expression eIF2a (Ser51), ASCT2 (SLC15), AMPKa,andp-AMPKa of the kidney-type glutaminase (GLS1) and glutamine (Thr172) from Cell Signaling Technology; rabbit anti- dehydrogenase (GDH), key enzymes for glutaminolysis kidney type glutaminase (GLS1) and rabbit anti-GDH and were preferentially sensitive to glutamine inhibi- from Abcam; and rabbit anti-a-tubulin antibodies from tors. These results show that ADI resistance in arginine- Sigma-Aldrich. auxotrophic melanoma cells is associated with metabolic Recombinant short hairpin RNA (shRNA) for c-Myc reprogramming that provides potential targets for over- were purchased from Sigma-Aldrich. coming the resistance. Cell culture and establishment of ADIR cells Materials and Methods A2058, SK-Mel-2, and A375 melanoma cells were pur- Reagents, antibodies, and shRNA chased from American Type Culture Collection Center Reagents were obtained from the following sources: without further authentication. All cell cultures were ADI-PEG20 (specific activity, 5–10 IU/mg) from Polaris maintained in Dulbecco’s Modified Eagle Medium con- Pharmaceuticals, Inc.; LY294002, 2-deoxy-D-glucose (2- taining 10% FBS in 5% CO2 atmosphere.

2582 Mol Cancer Ther; 12(11) November 2013 Molecular Cancer Therapeutics

Metabolic Reprogramming in Arginine Deiminase-Resistant Cell

ADIR cell lines were established by selecting the Other procedures surviving population from A2058 and SK-Mel-2 cells Western blotting, chromatin immunoprecipitation that were continuously exposed to ADI-PEG20 with (ChIP), cytotoxicity assay using MTT were conducted as stepwise increases in concentrations, starting from 0.1 described previously (10, 11). mg/mL and 0.05 mg/mL for A2058 and SK-Mel-2, respectively. Independent ADIR cell lines were estab- Statistical analyses lished from single cells using 96-well plate. Increased The IC50 values were obtained from nonlinear regres- resistance in these clones were further developed by sion analysis of concentration–effect curves by the Graph- increased drug concentrations with 0.025 mg/mL incre- Pad Prism Software and represented by mean Æ SD of ments for over 6 months, at which time the resistant three independent experiments. For analysis with only cells were able to survive at 0.9 mg/mL of ADI-PEG20. two-groups (parental and ADIR cells), Student t test was m Cells were routinely maintained at 0.3 g/mL of ADI- used. The IC50 values were plotted in reference to that of PEG20. A2058, which was set at 1.0. Signiﬁcance was determined when P < 0.05. Glucose uptake Glucose transport was measured according to the pro- Results cedure described previously (13). In brief, the experiments Elevated ASS1 expression in ADIR cell lines were started by adding 1 mCi of 2-DG (5–10 Ci/mmol/L) We established ﬁve independent ADIR variants from into exponentially growing A2058 cells (5 Â 105 in 6- A2058 (A2058-R1 to A2058-R5) and 4 SK-Mel-2 (SK-Mel-2- well; Corning). Cultured cells were maintained at 37C R1 to SK-Mel-2-R4) lines. Cytotoxicity tests showed that for 20 minutes. Cells were harvested, washed three all these cell lines were more resistant to ADI-PEG20 than times with PBS, and solubilized in 0.1 N NaOH and were their respective parental cell lines. More than 95% of the radioactivity was measured in a liquid scintillation all ADIR-A2058 cells could survive at 0.9 mg/mL ADI- counter. PEG20 for 48 hours treatment, whereas less than 25% of the parental cells survived (Fig. 2A). Likewise, treating Establishment of stable ASS1-overexpressing cell SK-Mel-2 cells with 0.9 mg/mL ADI-PEG20 almost lines completely killed the cells, whereas more than 70% of The plasmid pCMV6-ASS1 carrying human full-length ADIR SK-Mel-2 cells were able to survive under the same ASS1 cDNA sequence (GenBank, NM_000050; Origene) treatment (Fig. 2B). and empty vector (pCMV6) were transfected into A2058 All the ADIR-A2058 cells exhibited increased ASS1 and A375 cells using lipofectamine and positive clones expression (about 15-fold by densitometry, not shown) were selected with G418. as compared with that in the drug-sensitive counterpart

Figure 2. Survivability test and protein expression of ADIR variants. A2058 and its ADIR variants (A2058-R1–A2058-R5; A) and SK-Mel-2 and its ADIR variants (SK-R1–SK-R4; B) were treated with different concentrations of ADI-PEG20 for 3 days. Viability of the treated cells was determined by MTT assay. C, Western blotting analyses of ASS1, c-Myc, HIF-1a, and a-tubulin expression in A2058 (denoted by A) and its ADIR variants (R1–R5). D, Western blotting of protein expression in SK-Mel-2 (SK) and its ADIR lines (R1–R4).

www.aacrjournals.org Mol Cancer Ther; 12(11) November 2013 2583

Long et al.

(Fig. 2C). Likewise, all the ADIR-SK-Mel-2 cells had higher sistent with our previous results (10), we found that HIF- ASS1 expression levels than did the parental cell lines, but 1a, but not c-Myc, binds to the ASS1 promoter in the the SK-Mel-2-R1 and SK-Mel-2-R2 cell lines had lower parental A2058 (Fig. 3C, lane 1) and SK-Mel-2 cells (Fig. ASS1 levels than did the SK-Mel-2-R3 and SK-Mel-2-R4 3D, lane 1). In contrast, in all the ADIR variants investi- lines (Fig. 2D). Reduced expression of ASS1 in the SK-Mel- gated, c-Myc, not HIF-1a, binds to the ASS1 promoter (Fig. 2-R1 and SK-Mel-2-R2 cell lines was correlated with less 3C and D, lanes 2 and 3). These results taken together, resistance to ADI-PEG20 (Fig. 2B). These observations, show that HIF-1a and c-Myc remain important in tran- together with our previous finding that modulation of scriptional regulation of ASS1 expression in ADIR ASS1 levels affects cell sensitivity to ADI-PEG20 (4, 10), variants. strongly support that ASS1 expression is the predominant mechanism for acquired resistance to ADI-PEG20 in cul- ADIR cell lines are sensitive to PI3K/AKT inhibitors tured melanoma cells. We previously observed that A2058 and SK-Mel-2 cells treated with ADI-PEG20 activated the Ras/ Roles of c-Myc and HIF-1a in the regulation of ASS1 PI3K/AKT signaling cascade, resulting in c-Myc stabi- expression in ADIR cell lines lization for ASS1 upregulation (11). Moreover, several We previously showed that induction of ASS1 expres- recent reports suggested that PI3K/AKT play important sion by ADI-PEG20 in melanoma cells was associated roles in regulating c-Myc expression (15–17). Activation with the upregulation of c-Myc, which functions as a of PI3K/AKT signal is elicited by phosphorylation of positive regulator and the downregulation of HIF-1a, AKT at threonine 308. We conducted Western blotting which functions as a negative regulator (10, 11). Figure analyses of 5 ADIR-A2058 cell lines using antipho- 2C and D shows that c-Myc expression levels were similar sphorylated AKT antibody and found that all these cells between all ADIR variants and their respective parental contained activated AKT signal (Fig. 4A). Activation of cell lines; whereas reduced HIF-1a expression levels were AKT in these ADIR cells was reflected by increased seen in A2058-R2, A2058-R4, A2058-R5, and all ADIR-SK- sensitivities to the killing by LY294002 and perifosine, Mel-2 lines, but not in A2058-R1 and A2058-R3 cells. The which inhibit phosphoinositide 3-kinase (PI3K) and observations that no elevated expression of c-Myc and no AKT1 activities, respectively (Fig. 4B). reduction of HIF-1a expression in all ADIR cells prompted us to investigate whether c-Myc and HIF-1a remain Downregulated mTOR signaling in ADIR cell lines important for regulating ASS1 expression in these ADIR Three well-characterized sensing mechanisms for cel- cell lines. lular amino acid and/or nutritional deprivation in We chose four representative ADIR cell lines (A2058-R1, eukaryotic cells include the mTORC1, general amino acid which expressed higher HIF-1a levels than the control, control nonderepressible 2 (GCN2), and AMP-dependent and A2058-R2, SK-Mel-2-R1, and SK-Mel-2-R3, which did protein kinase (AMPK; refs. 18, 19). mTORC1 phosphor- not express detectable levels of HIF-1a). To investigate ylates and activates protein kinase p70S6 kinase 1 (S6K1), whether accumulation of HIF-1a would suppress ASS1 which subsequently activates mRNA cap binding by expression in the ADIR cell lines, we treated these cells and phosphorylating eIF-4B for the formation of translational their drug-sensitive cells (as control) with hypoxic initiation complex. GCN2 is the only kinase for eukaryotic a a mimics, DFX, and CoCl2 (14). Treating A2058 or SK- initiation factor 2 (eIF2 ) that senses the absence of Mel-2 cells with DFX or CoCl2 suppressed ADI-PEG20– one or more amino acids by virtual of direct binding to induced ASS1 expression (Fig. 3A, left). However, no uncharged cognate tRNA (20). Phosphorylation of suppression of ASS1 expression was found in A2058- elF2a (p-eIF2a) by activated GCN2 prevents the assembly R1, A2058-R2, and SK-Mel-2-R3, except SK-Mel-2-R1 cells of eIF2-GTP-Met-tRNAMet translation initiation complex, (Fig. 3A, right). The suppression of ASS1 expression in SK- resulting in reduction of general protein synthesis (21, 22). Mel-2-R1 cells by elevated HIF-1a, but not in the other AMPK is a metabolic sensor of energy crisis resulting from ADIR cell lines may have been due to the lower expression nutritional deprivation (23). We investigated the expres- levels of ASS1 in SK-Mel-2-R1 cells as compared with sion of these signals in five ADIR-A2058 cells. Figure 4C those in the other cell lines (Fig. 2D). These results suggest shows that expression levels of mTOR and p-mTOR were that the extent of ASS1 suppression by overexpressed drastically reduced in all five ADIR cell lines. Likewise, the HIF-1a may depend upon the intrinsic ASS1 expression downstream signal for mTOR, p70S6K was also down- levels. regulated. Downregulation of these signals in these ADIR To investigate the role of c-Myc in regulating ASS1 cells is reflected by virtue of their resistance to clinically expression in ADIR variants, we used c-Myc shRNA approved mTOR inhibitor, temsirolimus (CCI-779; strategy. Knockdown of c-Myc expression by two inde- ref. 24; Fig. 4D). Expression levels of AMPKa are variable pendent shRNA in lentiviral vectors downregulated ASS1 among the five ADIR cell lines (Fig. 4C) and levels of expression (Fig. 3B), supporting the positive role of c-Myc phosphorylated AMPKa were too low to be detected in all in the regulation of ASS1 in the ADIR cells. cells (not shown). No apparent difference in eIF2a levels We also conducted ChIP assay to determine the occu- was observed between ADIR-A2058 cells and their paren- pancy of HIF-1a and c-Myc at the ASS1 promoter. Con- tal cells, but levels of p-eIF2a were low in all the cell lines

2584 Mol Cancer Ther; 12(11) November 2013 Molecular Cancer Therapeutics

Metabolic Reprogramming in Arginine Deiminase-Resistant Cell

Figure 3. Roles of c-Myc and HIF-1a in the expression of ASS1 in ADIR cell lines and their parental cell lines. A, effects of HIF-1a accumulation by hypoxia mimics,

DFX (50 mmol/L), or CoCl2, (100 mmol/L) on the expression of ASS1 in A2058, SK-Mel2 induced by ADI- PEG 20 (0.3 mg/mL), and their ADIR cell lines. Note that accumulation of HIF-1a suppresses ASS1 expression in SK-R1 but not in SK- R3. B, knockdown of c-Myc by shRNA (Sh1 and Sh2) suppressed ASS1 expression in the indicated ADIR cell lines. C, ChIP assay of HIF-1a and c-Myc interactions with the ASS1 promoter in A2058 (denoted by A, lane 1), A2058-R1 (lane 2), and A2058-R2 (lane 3). Top, the antibodies used for ChIP and control IgG. Input, genomic DNA before immunoprecipitation. D, similar experiment was carried out in SK-Mel-2 (SK) and its ADIR cell lines (R1 and R3). ASS1 promoter sequences from ChIP were quantiﬁed by PCR assay.

investigated, using cell lysate prepared from A2058 cells A2058 parental cells. Elevated Glut1 expression in these treated with ADI-PEG20 for 24 hours as a positive control the ADIR cell lines was correlated with the enhanced (Fig. 4C). These results showed that mTOR is the only transport of [3H]-2-DG, a nonmetabolizable form of glu- signal that is consistently downregulated in the ADIR cose (Fig. 5B). Compared with the parental cell lines, the cells. ADIR variants showed higher levels of LDH-A, which catalyzes the conversion of pyruvate into lactate (Fig. 1), ADIR cells are associated with enhanced expression but lower levels of PDH (Fig. 5A). PDH is the key enzyme of ASL and glycolytic activities (the Warburg effect) that catalyzes the conversion of pyruvate into acetyl CoA, We then investigated whether ADI-PEG20 resistance an important metabolic intermediate in the TCA cycle would alter the metabolic programs. Because ASL is (Fig. 1). directly downstream from ASS1 in the urea cycle (Fig. Monocarboxylate transporter family (MCT, SLC16) cat- 1), we conducted Western blotting of its expression in ﬁve alyzes the rapid transport across the plasma membrane of ADIR-A2058 lines. We found that ASL expression levels many monocarboxylates including lactate (25). Because were elevated in all 5 ADIR cell lines (Fig. 5A). these ADIR cells show elevated expression of LDH-A, we Next, we investigated the expression of several key examined the expression levels of MCT1 and MCT4 in players involved in the glycolytic pathway in the ADIR ADIR A2058 cells. Results showed that levels of MCT1 cell lines. We ﬁrst investigated Glut1, which is the major were increased in A2058-R1, A2058-R2, A2058-R4, and transporter for glucose and found that all 5 ADIR-A2058 A2058-R5, but reduced in A2058-R3; whereas levels of cells showed about 8-fold higher Glut1 levels than those in MCT4 were increased in A2058-R2, A2058-R3, and A2058-

www.aacrjournals.org Mol Cancer Ther; 12(11) November 2013 2585

Long et al.

Figure 4. Analyses of protein expression in A2058 and its ﬁve ADIR cells. A, activated phosphorylation AKT (p-AKT) signaling. B, enhanced sensitivity of ADIR cells to the PI3K inhibitor (LY294002) and AKT (perifosine) as compared with their parental line. C, Western blotting analyses of mTOR signaling, AMPKa, and eIF1a. D, sensitivity of A2058 and its ADIR cells to mTOR inhibitor CCI-779. Cells were incubated with CCI-779 at concentrations ranging from 1 to 100 mmol/L for 72 hours and their viabilities were determined by MTT assay.

R5, but reduced in A2058-R4 and no change in A2058-R1 ADIR-A2058 cell lines as compared with those in the (data not shown). Thus, the expression levels of MCT1 and parental control, indicating the increased glutaminolytic MCT4 are heterogeneous among the ﬁve ADIR-A2058 activities in these ADIR variants (Fig. 5D). These results cells. Their roles in ADI resistance are not certain. were supported by the demonstration that all the 5 ADIR cell lines were associated by preferential sensitivity to GLS ADIR cell lines are preferentially sensitive to inhibitors, DON, azaserine, and BPTES (Fig. 5E). Despite glycolytic inhibitors increased glutaminolytic enzymes in these ADIR cell lines, Because ADIR cells exhibit elevated expression of Glut1 we found only A2058-R4 exhibited-elevated ASCT2 glu- and enhanced glucose uptake, known as the Warburg tamine transporter among the 5 ADIR A2058 cells (data effect, we asked whether these cells would be preferen- not shown). tially killed by 2-DG, a commonly used inhibitor of glycolysis (26–28). We found that, indeed, all the 5 ADIR c-Myc, but not elevated ASS1, contributes to the -A2058 variants exhibited enhanced sensitivity to 2-DG enhanced glycolytic and glutaminolytic metabolisms (Fig. 5C). Moreover, we found that these ﬁve ADIR cells in ADIR cell lines were also vulnerable to another glucolytic inhibitor 3-BP, To investigate the kinetics of induction of these glyco- a lactate/pyruvate analogue (Fig. 5C; refs. 29, 30), show- lytic and glutaminolytic enzymes by ADI-PEG20 in the ing that these ADIR cells are preferentially sensitive to ADIR cells, we conducted an ADI-PEG20 time-course glycolytic inhibitors as compared with their parental cell treatment of A2058 cells from 8 to 72 hours. We found lines. that ADI-PEG20 induced not only ASS1 and c-Myc, but also Glut1, LDH-A, and GLS1 (but not GDH) within this ADIR cells exhibit elevated glutaminolytic time frame. These results indicate that induction of ASS1, metabolism and increased sensitivity to glutaminase Glut1, and LDH-A, occurred very early during the devel- inhibitors opment of ADI resistance and elevation of GDH levels Two key enzymes in glutaminolytic metabolism, glu- occurred subsequently (Fig. 6A). taminase (GLS) and GDH, catalyze the two-step sequen- As elevated expression of ASS1 is the principal mech- tial conversion of glutamine into a-ketoglutarate via glu- anism of ADI resistance, we investigated whether tamate intermediate (Fig. 1). Mammals express two GLS enhanced expression of these enzymes was the conse- isoforms, kidney-type (or GLS1) and liver-type (GLS2). quence of ASS1 upregulation in the ADIR cell lines. To this GLS1 is expressed in a broad spectrum of cell lines. We end, we used transfection to establish two independently showed elevated expression of GLS1 and GDH in all 5 overexpressed ASS1 cell lines in A2058 and A375 cells.

2586 Mol Cancer Ther; 12(11) November 2013 Molecular Cancer Therapeutics

Metabolic Reprogramming in Arginine Deiminase-Resistant Cell

Figure 5. ADIR cell lines exhibited elevated glycolytic and glutaminolytic metabolism and increased sensitivity to glucose and glutaminase inhibitors. A, Western blot analyses of glucolytic enzymes in A2058 and its ADIR cells. B, enhanced uptake of 2-DG in ADIR cell lines as compared with A2058 cells. C, enhanced sensitivity of ADIR cell lines to glucolysis inhibitors 2-DG and 3-BP as compared with A2058 cells. D, expression levels of GLS1 and GDH in A2058 and its ADIR cell lines. E, enhanced sensitivity of ADIR cells to glutamine inhibitors, DON, azerserine, and BPTES, as compared with A2058 cells.

While no apparent alteration of HIF-1a levels was dis- than Glut1 or GLS1. In any event, these results support cernible even using prolonged exposure of the Western that c-Myc plays a role in regulating metabolic rewiring blot analyses (not shown). We found that expression of during the development of ADI-PEG20 resistance. c-Myc, Glut1, LDH-A, GLS1, and GDH was reduced (Fig. 6B). This expression proﬁle is directly opposite from that Expression of enzymes involved in the lipid in the ADIR cell lines. These results show that elevated biosynthesis in the ADIR cell lines expression of these enzymes in ADIR cells was not likely We also determined the expression of enzymes due to the enhanced ASS1 expression. involved in fatty acid biosynthesis in ADIR cell lines. We next investigated whether c-Myc plays a role in Compared with their parental counterpart, all the ADIR regulating the glycolytic and glutaminolytic enzymes in A2058 cell lines, except A2058-R4, had lower expression these ADIR cells, in light that c-Myc has been recognized levels of ACC, which catalyzes the conversion of acetyl- as an important transcription regulator of 10% to 20% of CoA into malonyl-CoA (Fig. 1; data not shown) and all the total cellular genes through E-box interactions in mam- ADIR A2058 cell lines, except A2058-R2, had lower levels malian cells (31, 32). A2058-R1 and A2058-R2 cells were of FAS, which converts malonyl-CoA into palmitoyl-CoA treated with two independent c-Myc shRNA for 48 hours. (data not shown). These results suggest that fatty acid Although the shRNA we used could only partially reduce metabolism may also be reduced in the ADIR cell lines c-Myc levels, we found reduced expression of Glut1 and compared with their parental counterpart. GLS1 in these ADIR cells as compared with those in the scramble shRNA-treated controls, whereas levels of LDH- Discussion A and GDH were not affected. The inability of reducing The discovery that many human malignant tumors do LDH-A and GDH suggest that either other mechanisms not express ASS1 and are arginine-auxotrophic provides a are involved for the regulation of LDH-A and GDH, or mechanistic basis for using recombinant arginine-deplet- regulation of these enzymes by c-Myc is more stringent ing enzymes in targeted therapy of these diseases (4). In

www.aacrjournals.org Mol Cancer Ther; 12(11) November 2013 2587

Long et al.

Figure 6. Western blotting analyses of protein expression. A, time-course induction of various enzymes by ADI-PEG20 treatment (0.3 mg/mL) in A2058 cells. B, effects of ASS1 overexpression in A2058 and A375 cell lines by ASS1 transfection (two independent cell lines each) on the expression of several proteins as indicated. C, effects of c-Myc knockdown on the expression of proteins as indicated in two ADIR-A2058 cell lines.

this communication, we show that in a panel of nine We found that these ADIR cells have altered metabolic independently established ADIR cell lines all expressed program toward aerobic glycolysis by enhancing Glut1 elevated levels of ASS1, further strengthening the role of and LDH-A expression but reducing PDH expression, ASS1 in ADI resistance. These ADIR cell lines serve as a exhibiting the Warburg phenomenon. Since Warburg’s valuable resource for investigating the metabolic repro- work in 1929, it has been evidenced that cancer cells, even gramming associated with ADI resistance as summarized under aerobic conditions, frequently upregulate glucose in Fig. 1. Several important findings are discussed below. metabolism, leading to a high uptake and use of glucose, We found that these ADIR variants exhibited elevated but a moderate rate of mitochondrial respiration (39, 40). AKT signaling and are preferentially sensitive to PI3K/ This characteristic has led to the development of 18 AKT inhibitors. These findings support the important role 2[ F]fluoro-2-deoxy-D-glucose (FDG) in positron emis- of this pathway in ADI-PEG20 resistance (11). We also sion tomography (PET) for the imaging of patients with observed that expression of mTOR and its downstream primary and metastatic tumors(41). Our observations that signal p70S6K was downregulated in all ADIR cells. While ADIR cells exhibited elevated Warburg phenomenon sug- the canonical pathway of mTOR regulation is through the gest that it may be possible to use FDG-PET for monitoring PI3K/AKT pathway (33), however, downregulation of the development of ADI resistance during cancer chemo- mTOR independent of this pathway has also been therapy. In clinical setting, melanoma is known to be reported (34). The molecular mechanism for downregula- strongly PET-positive and usually become less FDG avid tion of mTOR signal in ADI resistance remains to be when tumor response to treatment. Once tumors become investigated. resistant to ADI-PEG20, melanoma will again become We further showed that ASL was elevated in these ADIR FDG avid. This was also observed clinically when patient lines, suggesting that the ADIR cells may require robust failed ADI-PEG20. ASS1/ASL activities in the urea cycle to overcome ADI- Metabolic reprogramming in ADIR cell lines also PEG20–induced autophagy and apoptosis in the other- include elevated expression of GLS1 and GDH, another wise arginine-auxotrophic cancer cells (11, 35). Although common phenomenon in cancer cells (42, 43). These ASS1 has often been considered as the rate-limiting enzymes are important for producing a-ketoglutarate, an enzyme for the biosynthesis of arginine (36), our results important metabolite in the TCA cycle. Tumor cells fuel suggest that ADIR cells may require additional enzymes to their metabolism with glucose and glutamine as impor- replenish arginine under prolong arginine deprivation tant bioenergetic sources for cell proliferation. The depen- conditions. Our results are consistent with recent findings dence of these ADIR cells to glucose and glutamine for that several enzymes in the urea cycle may be coregulated growth was shown by the findings that these ADIR cells (37), including ASS1, ASL, nitric oxide synthase (NOS), exhibited elevated sensitivity to the killing by glucose (2- and cationic amino acid transporter 1 and may form a DG and 3-BP) and glutamine inhibitors (DON, azaserine, supermolecular complex to conduct important physio- and BPTES). While DON and azaserine have long been logic function (38). known to be toxic in human clinical evaluations (44),

2588 Mol Cancer Ther; 12(11) November 2013 Molecular Cancer Therapeutics

Metabolic Reprogramming in Arginine Deiminase-Resistant Cell

presumably due to the presence of chemically reactive expression. These possible mechanisms would further diazo group. The increased sensitivity of these ADIR cells support the important roles of c-Myc in the regulation of to BPTES showed in this study may have promising glycolytic and glutaminolytic metabolism in melanoma therapeutic implication. New generation of GLS1 inhibi- cells. tors such as BPTES derivatives, which are devoid of diazo Finally, the identification of metabolic reprogramming group found in DON and azaserine (12), as well as associated with ADI resistance in the arginine-auxotro- compound 968 and its derivatives, another structurally phic melanoma model bears important clinical implica- distinct series of GLS1 inhibitor (45, 46) that are currently tions for targeted therapy of arginine-auxotrophic tumors. under clinical evaluations, may pave the way for future Many antitumor small molecules targeting PI3K/AKT, clinical applications. glycolytic, and glutaminolytic signalings have been We also found that c-Myc plays important role in the approved for clinical use. Combination of these agents metabolic rewiring of ADI resistance. Previous studies with arginine-degrading enzymes, such as ADI-PEG20 or have showed that c-Myc regulates genes involved in the pegylated human arginase 1 may improve the treatment glycolytic pathway by directly binding to the E-box locat- efficacy of many human cancers that require arginine for ed at the promoters of these genes (47). C-Myc can also survival. upregulates glutamine metabolism via transcriptional repression of miR-23a and miR-23b (48). Although levels Disclosure of Potential Conflicts of Interest of c-Myc were not elevated in these ADIR cells, however, No potential conflicts of interest were disclosed. our kinetic analysis showed that expression of c-Myc was induced by ADI-PEG20 within the same time frame when Authors' Contributions Conception and design: Y. Long, M.T. Kuo many of these enzymes were also induced. Elevated levels Development of methodology: W.-B. Tsai of glycolytic and glutaminolytic pathways, apparently, Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Long, W.-B. Tsai, M. Wangpaichitr, N. Savaraj, are needed for the maintenance of ADI resistance because L.G. Feun inhibitors to these pathways overcome the resistance. The Analysis and interpretation of data (e.g., statistical analysis, biostatis- observation that no elevated c-Myc levels in all the nine tics, computational analysis): Y. Long, W.-B. Tsai, M.T. Kuo R Writing, review, and/or revision of the manuscript: T. Tsukamoto, N. ADI cell lines, despite their early induction, suggests that Savaraj, L.G. Feun, M.T. Kuo persistently high levels of c-Myc may be disadvantageous Administrative, technical, or material support (i.e., reporting or orga- for cell growth during the development of ADI resistance. nizing data, constructing databases): Y. Long, T. Tsukamoto, M.T. Kuo We also discovered that elevated expression of ASS1 by transfection downregulates Glut1, LDH-A, GLS1, and Acknowledgments The authors thank Dr. Bor-Wen Wu (Polaris) for providing ADI-PEG20 GDH expression, showing that modulating ASS1 levels and anti-ASS1 antibody. can influence the expression of other enzymes in distance, i.e., glycolytic and glutaminolytic pathways. While the Grant Support precise mechanism(s) remain to be investigated, down- The work was supported in part by ROI-CA149260 from the National regulation of c-Myc may be responsible for the reduced Cancer Institute (to M.T. Kuo and L.G. Feun) and R21NS074151 from the ASS National Institute of Neurological Disorders and Stroke (to T. Tsukamoto). expression of these enzymes in these 1-transfected The costs of publication of this article were defrayed in part by the cells, although this remains to be elucidated. Moreover, payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate the findings of reduced levels c-Myc levels in these ASS1- this fact. transfected cells may explain why ADIR cells did not show elevated c-Myc, as this may be due to feed-back inhibition Received April 22, 2013; revised July 26, 2013; accepted August 15, 2013; mechanism of c-Myc regulation once ASS1 is overly published OnlineFirst August 26, 2013.

References 1. Bronte V, Zanovello P. Regulation of immune responses by L-arginine apy with pegylated arginine deiminase. Br J Cancer 2012;106: metabolism. Nat Rev Immunol 2005;5:641–54. 1481–5. 2. Rodriguez PC, Ochoa AC. Arginine regulation by myeloid derived 6. Ott PA, Carvajal RD, Pandit-Taskar N, Jungbluth AA, Hoffman EW, Wu suppressor cells and tolerance in cancer: mechanisms and therapeutic BW, et al. Phase I/II study of pegylated arginine deiminase (ADI-PEG perspectives. Immunol Rev 2008;222:180–91. 20) in patients with advanced melanoma. Invest New Drugs 2013;31: 3. Dillon BJ, Prieto VG, Curley SA, Ensor CM, Holtsberg FW, Bomalaski 425–34. JS, et al. Incidence and distribution of argininosuccinate synthetase 7. Yang TS, Lu SN, Chao Y, Sheen IS, Lin CC, Wang TE, et al. A deﬁciency in human cancers: a method for identifying cancers sen- randomised phase II study of pegylated arginine deiminase (ADI-PEG sitive to arginine deprivation. Cancer 2004;100:826–33. 20) in Asian advanced hepatocellular carcinoma patients. Br J Cancer 4. Kuo MT, Savaraj N, Feun LG. Targeted cellular metabolism for cancer 2010;103:954–60. chemotherapy with recombinant arginine-degrading enzymes. Onco- 8. Mauldin JP, Zeinali I, Kleypas K, Woo JH, Blackwood RS, Jo CH, et al. target 2010;1:246–51. Recombinant human arginase toxicity in mice is reduced by citrulline 5. Feun LG, Marini A, Walker G, Elgart G, Moffat F, Rodgers SE, et al. supplementation. Transl Oncol 2012;5:26–31. Negative argininosuccinate synthetase expression in melanoma 9. Yau T, Cheng PN, Chan P, Chan W, Chen L, Yuen J, et al. A phase 1 tumours may predict clinical beneﬁt from arginine-depleting ther- dose-escalating study of pegylated recombinant human arginase 1

www.aacrjournals.org Mol Cancer Ther; 12(11) November 2013 2589

Long et al.

(Peg-rhArg1) in patients with advanced hepatocellular carcinoma. 29. Ko YH, Smith BL, Wang Y, Pomper MG, Rini DA, Torbenson MS, et al. Invest New Drugs 2013;31:99–107. Advanced cancers: eradication in all cases using 3-bromopyruvate 10. Tsai WB, Aiba I, Lee SY, Feun L, Savaraj N, Kuo MT. Resistance to therapy to deplete ATP. Biochem Biophys Res Commun 2004;324: arginine deiminase treatment in melanoma cells is associated with 269–75. induced argininosuccinate synthetase expression involving c-Myc/ 30. Yun J, Rago C, Cheong I, Pagliarini R, Angenendt P, Rajagopalan H, HIF-1alpha/Sp4. Mol Cancer Ther 2009;8:3223–33. et al. Glucose deprivation contributes to the development of KRAS 11. Tsai WB, Aiba I, Long Y, Lin HK, Feun L, Savaraj N, et al. Activation of pathway mutations in tumor cells. Science 2009;325:1555–9. Ras/PI3K/ERK pathway induces c-Myc stabilization to upregulate 31. Fernandez PC, Frank SR, Wang L, Schroeder M, Liu S, Greene J, et al. argininosuccinate synthetase, leading to arginine deiminase resis- Genomic targets of the human c-Myc protein. Genes Dev 2003;17: tance in melanoma cells. Cancer Res 2012;72:2622–33. 1115–29. 12. Shukla K, Ferraris DV, Thomas AG, Stathis M, Duvall B, Delahanty G, 32. Li Z, Van Calcar S, Qu C, Cavenee WK, Zhang MQ, Ren B. A global et al. Design, synthesis, and pharmacological evaluation of bis-2-(5- transcriptional regulatory role for c-Myc in Burkitt's lymphoma cells. phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulﬁde 3 (BPTES) ana- Proc Natl Acad Sci U S A 2003;100:8164–9. logs as glutaminase inhibitors. J Med Chem 2012;55:10551–63. 33. Zoncu R, Efeyan A, Sabatini DM. mTOR: from growth signal integration 13. Fan Y, Dickman KG, Zong WX. Akt and c-Myc differentially activate to cancer, diabetes and ageing. Nat Rev Mol Cell Biol 2011;12:21–35. cellular metabolic programs and prime cells to bioenergetic inhibition. 34. Lee YK, Park SY, Kim YM, Kim DC, Lee WS, Surh YJ, et al. Suppression J Biol Chem 2010;285:7324–33. of mTOR via Akt-dependent and -independent mechanisms in sele- 14. Kaelin WG Jr. Cancer and altered metabolism: potential importance of nium-treated colon cancer cells: involvement of AMPKalpha1. Carci- hypoxia-inducible factor and 2-oxoglutarate-dependent dioxy- nogenesis 2010;31:1092–9. genases. Cold Spring Harb Symp Quant Biol 2011;76:335–45. 35. Kim RH, Coates JM, Bowles TL, McNerney GP, Sutcliffe J, Jung JU, 15. Ilic N, Utermark T, Widlund HR, Roberts TM. PI3K-targeted therapy can et al. Arginine deiminase as a novel therapy for prostate cancer induces be evaded by gene ampliﬁcation along the MYC-eukaryotic translation autophagy and caspase-independent apoptosis. Cancer Res 2009;69: initiation factor 4E (eIF4E) axis. Proc Natl Acad Sci U S A 2011;108: 700–8. E699–708. 36. Delage B, Fennell DA, Nicholson L, McNeish I, Lemoine NR, Crook 16. Muellner MK, Uras IZ, Gapp BV, Kerzendorfer C, Smida M, Lechter- T, et al. Arginine deprivation and argininosuccinate synthetase mann H, et al. A chemical-genetic screen reveals a mechanism of expression in the treatment of cancer. Int J Cancer 2010;126: resistance to PI3K inhibitors in cancer. Nat Chem Biol 2011;7:787–93. 2762–72. 17. Liu P, Cheng H, Santiago S, Raeder M, Zhang F, Isabella A, et al. 37. Oyadomari S, Gotoh T, Aoyagi K, Araki E, Shichiri M, Mori M. Coin- Oncogenic PIK3CA-driven mammary tumors frequently recur via PI3K duction of endothelial nitric oxide synthase and arginine recycling pathway-dependent and PI3K pathway-independent mechanisms. enzymes in aorta of diabetic rats. Nitric Oxide 2001;5:252–60. Nat Med 2011;17:1116–20. 38. Erez A, Nagamani SC, Shchelochkov OA, Premkumar MH, Campeau 18. Oliveira SM, Zhang YH, Solis RS, Isackson H, Bellahcene M, Yavari A, PM, Chen Y, et al. Requirement of argininosuccinate lyase for systemic et al. AMP-activated protein kinase phosphorylates cardiac troponin I nitric oxide production. Nat Med 2011;17:1619–26. and alters contractility of murine ventricular myocytes. Circ Res 39. Dang CV, Hamaker M, Sun P, Le A, Gao P. Therapeutic targeting of 2012;110:1192–201. cancer cell metabolism. J Mol Med 2011;89:205–12. 19. Yuan HX, Xiong Y, Guan KL. Nutrient sensing, metabolism, and cell 40. DeBerardinis RJ, Lum JJ, Hatzivassiliou G, Thompson CB. The biology growth control. Mol Cell 2013;49:379–87. of cancer: metabolic reprogramming fuels cell growth and prolifera- 20. Wek RC, Jiang HY, Anthony TG. Coping with stress: eIF2 kinases and tion. Cell Metab 2008;7:11–20. translational control. Biochem Soc Trans 2006;34:7–11. 41. Gambhir SS. Molecular imaging of cancer with positron emission 21. Kimball SR, Jefferson LS. Control of translation initiation through tomography. Nat Rev Cancer 2002;2:683–93. integration of signals generated by hormones, nutrients, and exercise. 42. Dang CV. Glutaminolysis: supplying carbon or nitrogen or both for J Biol Chem 2010;285:29027–32. cancer cells? Cell Cycle 2010;9:3884–6. 22. Jackson RJ, Hellen CU, Pestova TV. The mechanism of eukaryotic 43. DeBerardinis RJ, Mancuso A, Daikhin E, Nissim I, Yudkoff M, translation initiation and principles of its regulation. Nat Rev Mol Cell Wehrli S, et al. Beyond aerobic glycolysis: transformed cells can Biol 2010;11:113–27. engage in glutamine metabolism that exceeds the requirement for 23. Carling D, Thornton C, Woods A, Sanders MJ. AMP-activated protein protein and nucleotide synthesis. Proc Natl Acad Sci U S A 2007; kinase: new regulation, new roles? Biochem J 2012;445:11–27. 104:19345–50. 24. Stock C, Zaccagnini M, Schulze M, Teber D, Rassweiler JJ. Temsir- 44. Magill GB, Myers WP, Reilly HC, Putnam RC, Magill JW, Sykes MP, olimus. Recent Results Cancer Res 2010;184:189–97. et al. Pharmacological and initial therapeutic observations on 6-diazo- 25. Pinheiro C, Longatto-Filho A, Azevedo-Silva J, Casal M, Schmitt FC, 5-oxo-1-norleucine (DON) in human neoplastic disease. Cancer Baltazar F. Role of monocarboxylate transporters in human cancers: 1957;10:1138–50. state of the art. J Bioenerg Biomembr 2012;44:127–39. 45. Wang JB, Erickson JW, Fuji R, Ramachandran S, Gao P, Dinavahi R, 26. Kurtoglu M, Gao N, Shang J, Maher JC, Lehrman MA, Wangpaichitr M, et al. Targeting mitochondrial glutaminase activity inhibits oncogenic et al. Under normoxia, 2-deoxy-D-glucose elicits cell death in select transformation. Cancer Cell 2010;18:207–19. tumor types not by inhibition of glycolysis but by interfering with N- 46. Katt WP, Ramachandran S, Erickson JW, Cerione RA. Dibenzophe- linked glycosylation. Mol Cancer Ther 2007;6:3049–58. nanthridines as inhibitors of glutaminase C and cancer cell prolifera- 27. Pradelli LA, Beneteau M, Chauvin C, Jacquin MA, Marchetti S, Munoz- tion. Mol Cancer Ther 2012;11:1269–78. Pinedo C, et al. Glycolysis inhibition sensitizes tumor cells to death 47. Dang CV, O'Donnell KA, Zeller KI, Nguyen T, Osthus RC, Li F. The c- receptors-induced apoptosis by AMP kinase activation leading to Mcl- Myc target gene network. Semin Cancer Biol 2006;16:253–64. 1 block in translation. Oncogene 2010;29:1641–52. 48. Gao P, Tchernyshyov I, Chang TC, Lee YS, Kita K, Ochi T, et al. c- 28. Aft RL, Zhang FW, Gius D. Evaluation of 2-deoxy-D-glucose as a Myc suppression of miR-23a/b enhances mitochondrial glutamin- chemotherapeutic agent: mechanism of cell death. Br J Cancer ase expression and glutamine metabolism. Nature 2009;458: 2002;87:805–12. 762–5.

2590 Mol Cancer Ther; 12(11) November 2013 Molecular Cancer Therapeutics

Arginine Deiminase Resistance in Melanoma Cells Is Associated with Metabolic Reprogramming, Glucose Dependence, and Glutamine Addiction

Yan Long, Wen-Bin Tsai, Medhi Wangpaichitr, et al.

Mol Cancer Ther 2013;12:2581-2590. Published OnlineFirst August 26, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-13-0302

Cited articles This article cites 48 articles, 15 of which you can access for free at: http://mct.aacrjournals.org/content/12/11/2581.full#ref-list-1

Citing articles This article has been cited by 10 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/12/11/2581.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/12/11/2581. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.