Amino acids are integral components of cancer cell metabolism. They serve as building blocks for peptide synthesis, as an energy source and as precursor molecules to secure an adequate supply of nucleotides, neurotransmitters and nitrogen. They also contribute to redox balance and serve as substrates for post-translational and epigenetic modifications [1
]. In recent years, there has been increasing interest in the role of essential and non-essential amino acids in the metabolic equilibrium that enables tumor cell proliferation. Despite the fact that glutamine is a non-essential amino acid, many cancer cells were shown to be addicted to glutamine, which fuels the tricarboxylic acid (TCA) cycle and contributes to the synthesis of lipids, nucleotides and non-essential amino acids [2
]. Published studies by ourselves and others also demonstrated that the growth and survival of various types of cancer cells, including sarcoma cells, depend on adequate availability of the non-essential amino acid asparagine (Asn) [3
Mammalian cells maintain intracellular asparagine availability through import from the extracellular space [5
] and de novo synthesis from glutamine (Gln) and aspartate (Asp) through a unidirectional ATP-dependent reaction catalyzed by asparagine synthetase (ASNS) (Figure 1
]. Physiological asparagine concentrations are 5.1 mg/L (38.8 µM [7
]) and 10.9 mg/L (82.4 µM [8
]) in mouse and human plasma, respectively (Figure 2
a,b). Asparagine uptake from the extracellular space and cell-intrinsic asparagine synthesis secure intracellular asparagine concentrations below 132 ng/L (1 nM) [5
ASNS expression in various types of solid tumor cells correlates with higher tumor grade, a propensity to metastasize and poor patient survival [9
]. If cells are deprived of nutrients, including asparagine, a conserved transcriptional program known as the integrated stress response is activated to restore homeostasis through upregulation of various nutrient transporters and enzymes, including ASNS [4
]. In other words, nutrient-deprived cells consume nitrogen and ATP to maintain intracellular availability of asparagine. Known functions of asparagine in tumor cells include the translation of new peptides [3
], activation of mechanistic Target of Rapamycin complex 1 (mTORC1) signaling [4
] and use as an amino acid exchange factor to regulate uptake of other amino acids from the extracellular space [5
Asparagine availability in cancer cells serves as a therapeutic target. Asparagine depletion through treatment with bacterially derived asparaginase has long been established as an important strategy in the treatment of leukemias expressing low levels of ASNS [6
]. Asparaginase was also shown to reduce the in vitro growth of sarcoma cells. Genetic silencing of ASNS combined with depletion of systemic asparagine via asparaginase decreased sarcoma growth in vivo [3
]. Yet, sarcoma cells express high levels of ASNS and therefore rely less on environmental asparagine supply. Indeed, asparaginase sensitivity of sarcoma cells is moderate to poor when compared to lymphoblasts [3
In this study, we interrogated changes in the sarcoma metabolome induced by asparagine depletion to better understand why cancer cells depend on adequate asparagine availability and to identify chemically actionable vulnerabilities that may be exploited to potentiate asparaginase effects. Our studies revealed relative excess of reducing equivalents in asparagine-starved sarcoma cells. We also report synergistic effects of asparaginase and complex 1 inhibitors, which block regeneration of nicotinamide adenine dinucleotide (NAD+) in the electron transport chain and enhance reductive stress in asparagine-starved sarcoma cells.
Rapidly proliferating cancer cells have to meet specific metabolic requirements to sustain homeostasis while supporting rapid expansion of biomass, which poses a profound metabolic challenge [19
]. In this study, we employed different mass spectrometry approaches to identify lower aspartate levels, higher aspartate/glutamine ratios and lower levels of TCA cycle metabolites in asparagine-deprived cells. These changes indicated a redirection of TCA cycle flux and were accompanied by reduced NAD+
/NADH ratios, consistent with a relative lack of electron acceptors. Reductive stress caused by deficiency of electron acceptors may be just as harmful as oxidative stress, as electron acceptors NAD+
and GSH are essential for maintaining cellular homeostasis. For example, NAD+
is a necessary cofactor for many enzymes, and a decrease in the NAD+
/NADH ratio causes these enzymes to decrease in activity [20
]. We discovered that asparagine depletion of sarcoma cells causes reductive stress and that exogenous supplementation with the electron acceptor pyruvate [13
] restored the changes in NAD+
/NADH ratios, proliferation and viability induced by asparagine deprivation. Treatment with chemicals disrupting the regeneration of NAD+
in the ETC further enhanced the anti-proliferative and pro-apoptotic effects of asparagine depletion.
Aspartate is known to serve as an important carbon donor for the synthesis of proteinogenic amino acids and nucleotides. Our metabolomic analyses demonstrated consistently reduced aspartate levels in asparagine-deprived cells, including shASNS cells grown in asparagine-free medium and asparaginase-treated cells. This was surprising at first as we had initially expected higher aspartate levels due to reduced aspartate consumption after ASNS knockdown or increased release of aspartate by asparaginase. Yet, addition of the electron acceptor pyruvate not only reversed reductive stress and growth arrest in asparagine-starved cells but also raised aspartate levels. Thus, we conclude that asparagine deprivation, through metabolic reprogramming, causes reductive stress, which, in turn, results in lower aspartate levels in asparagine-starved cells.
Growth of various types of tumor cells was recently shown to depend on adequate availability of the non-essential amino acid asparagine [3
]. Mammalian cells maintain intracellular asparagine levels through ATP-dependent conversion, catalyzed by the enzyme ASNS, from aspartate and glutamine into asparagine and glutamate (Figure 1
]. The growth-inhibitory effects of ASNS silencing are reversible through uptake of exogenously supplemented asparagine [3
]. Findings from our experiments confirm that asparagine deprivation stalls proliferation, induces autophagy and increases apoptosis in sarcoma cells grown in physiological and supraphysiological glucose and glutamine conditions. Consistent with the published literature [10
], asparagine did not reverse the growth inhibition observed in glutamine-starved sarcoma cells. Yet, exogenous asparagine partially reversed glutamine withdrawal-induced apoptosis in KRAS
-driven mouse sarcoma cells. Zhang et al. also reported complete reversal of glutamine withdrawal-induced apoptosis by supplemental asparagine in glioma and neuroblastoma cells [10
Thus, asparagine appears to play an important role in cancer cell metabolism [3
]. ASNS transcription in cancer cells is controlled by tumor-driving oncogenes [4
] and elevated in response to amino acid and glucose starvation through adaptive processes known as amino acid and endoplasmic reticulum stress responses [6
]. These cellular programs converge on increased translation of the transcription factor ATF4, which, in turn, stimulates expression of ASNS and various other enzymes and nutrient transporters [4
]. We speculate that solid tumors express high levels of ASNS to maintain intracellular asparagine supply as they outgrow nutrient supply via the existing vasculature and thereby promote tumor cell survival. This is further supported by correlative studies indicating that higher ASNS expression in tumor cells correlates with higher tumor grade, higher rates of metastases and worse patient outcomes [9
]. However, asparagine does not appear to serve as a major intermediary metabolite. Flux analyses demonstrated that 10% of its nitrogen and minimal amounts of its carbon were detected in purines and aspartate/malate, respectively [5
]. Nevertheless, asparagine is an important proteinogenic amino acid [3
], especially with respect to the production of asparagine-rich proteins involved in epithelial-to-mesenchymal transition [9
]. We demonstrate that transcripts involved in nascent peptide synthesis (e.g., KEGG categories ribosome and aminoacyl tRNA biosynthesis) are enriched among genes differentially regulated in asparagine-starved cells. Furthermore, intracellular asparagine serves as an amino acid exchange factor to secure intracellular levels of other amino acids, especially serine/threonine and non-polar amino acids [5
]. In fact, asparagine appears to coordinate protein and nucleotide synthesis in proliferating cells by regulating mTORC1 activity [4
] and uptake of serine [4
], which is crucial for protein and pyrimidine synthesis.
Asparagine metabolism in cancer cells represents a vulnerability with possible therapeutic value. Asparagine-depleting drugs have been long established as effective drugs in the treatment of lymphoblastic leukemia [6
]. Asparaginase was also shown to reduce sarcoma growth. However, when compared to lymphoblasts, asparaginase sensitivity of sarcoma cells is moderate to poor [3
]. This may be due to the fact that sarcoma cells and many other solid tumor cells express substantially higher levels of ASNS than lymphoblasts [9
], which are characterized by low ASNS levels and are much more dependent on environmental asparagine. Still, asparaginase is an attractive anti-cancer drug, as its side effect profile does not overlap with the toxicities of many other conventional chemotherapy drugs. It may be a valuable component in anti-sarcoma combination therapies. Synergistic/additive effects were previously reported for asparaginase and mTOR inhibitors [21
], autophagy inhibitors such as chloroquine [22
] and proteasome inhibitors such as bortezomib [23
]. Chemical inhibitors of the mitochondrial electron transport chain (ETC) were previously shown to decrease NAD+
/NADH ratios and halt growth in tumor cells [13
]. This led us to examine if complex 1 inhibitors deepened the reductive stress and growth arrest caused by asparagine depletion. Our experiments revealed a clear synergistic effect of asparaginase and complex 1 inhibitors on the growth of mouse and human sarcomas, including patient-derived rhabdomyosarcoma cultures. Complex 1 inhibitors may be applied in combination with asparaginase to further exploit the anti-cancer effects of asparagine depletion in sarcomas.
4. Materials and Methods
4.1. Cell Lines
-driven mouse sarcoma cell line used in this study was established from a genetically engineered Kras(G12v);p16p19null
mouse sarcoma with myogenic differentiation [3
]. The human embryonal rhabdomyosarcoma cell line RD was purchased from American Type Culture Collection (ATCC). Both cell lines were kept in DMEM with 10% fetal bovine serum (FBS, Sigma Aldrich, St. Louis, MO, USA) and 1% penicillin–streptomycin (PS). To interrogate asparagine effects, 4.15 g DMEM (D5030, Sigma Aldrich), 1.85 g NaHCO3 (Sigma Aldrich), 10% FBS and 1% PS were reconstituted in 500 mL dH2O together with defined amounts of glutamine (0 to 292 mg; Sigma), glucose (0.25 to 2.25 g; Sigma Aldrich) and asparagine (0 to 50 mg; A4159, Sigma Aldrich). Short tandem repeat analyses of mouse (Table S7
) and human (Table S8
) cell lines used in this study were performed by Eurofins.
RMSZH003_XC primary rhabdomyosarcoma cell cultures were established from a recurrent pelvic PAX3:FOXO1 fusion-positive rhabdomyosarcoma diagnosed in a 3-year-old female, IC-pPDX-29_XC from a 14-year-cold female with a relapsed, paravertebral PAX3:FOXO1 fusion-positive rhabdomyosarcoma, SJRHB011_YC from a 4-year-old boy with a head/neck fusion-negative rhabdomyosarcoma and SJRHB012_ZC from an 18-year-old male with a relapsed bladder/prostate fusion-negative rhabdomyosarcoma (Table S6
). Primary rhabdomyosarcoma cell cultures were maintained in neurobasal medium (Gibco, 21103049, Waltham, MA, USA) supplemented with 1% penicillin/streptomycin (Gibco, 15140-122), 1X Glutamax (Gibco, 35050) and 2X B27 (Life Technologies, 17504044, Carlsbad, CA, USA). For RMS-ZH003_XC and IC-pPDX-29_XC cells, 20 ng/ml basic fibroblast growth factor (bFGF) (Peprotech, AF-100-18B, Rocky Hill, NJ, USA) and 20 ng/mL epidermal growth factor (EGF) (Peprotech, AF-100-15) were added [24
4.2. ASNS Silencing
TRC clones TRCN0000324779, TRCN0000031703 and TRCN0000031702, delivered in PLKO vectors, were employed to knock down ASNS in mouse sarcoma cells as previously described [3
4.3. Proliferation Assays
Cells were seeded at 750 (mouse) or 1000 (human RD) cells per well in 96-well plates with 100 µL medium. After 24 h, cells were exposed to asparaginase (MBS142428, Biozol), phenformin (14997, Cayman Chemical, Ann Arbor, MI, USA), metformin (13118, Cayman Chemical) and imiquimod (14956, Cayman Chemical) at defined concentrations. Chemical treatments were repeated every 48 h for up to 6 days in total. To determine cell growth, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (M2003, Sigma-Aldrich) was solved in PBS to a final concentration of 12 mM. In brief, 10 µL of MTT solution was added per well and incubated for 4 h. Then, 100 µL dissolving solution (10% SDS (L3771, Sigma-Aldrich) with 0.01M HCL) was added to each well and incubated for 12–16 h. Optical densities at 570 nm were measured with a Tecan Sunrise absorbance microplate reader.
4.4. Cell Viability Assay
SJRHB012_ZC, SJRB011_YC, IC-pPDX-29_XC and RMS-ZH003_XC cells were plated at 6000 cells/well in 384-well plates coated with Matrigel. The medium was changed after one day, and cells were incubated with either asparaginase 1 u/mL (MBS142428, Biozol, Eching, Germany), imiquimod 50 uM (14956, Cayman Chemical) or both. DMSO-treated cells served as controls. After 72 h, cell viability was determined using the CellTiter-Glo 3D viability assay (Promega, G9681, Madison, WI, USA).
4.5. Annexin V Staining
Annexin V staining was performed according to the manufacturer’s instructions using Annexin V-APC (550474, BD Biosciences, San Jose, CA, USA) and 7AAD (559925, BD Biosciences) with a BD FACS Canto II flow cytometer.
4.6. Western Blots
Cells were detached using trypsin, washed twice in ice-cold PBS, lysed for 15 min on ice using lysis buffer (9803S, NEB, Ipswich, MA, USA) containing protease and phosphatase inhibitors (5872 S, Cell Signaling Tech, Denver, MA, USA) and centrifuged for 10 min at 13,000 g. Protein concentrations in supernatants were measured using the DC Protein Assay (Bio-Rad, # 5000111). Equal amounts of protein extracts were processed for Western blots. Specific antibodies and concentrations were used to detect expression of asparagine synthetase (1:250, Sigma Aldrich, HPA029385), LC3I/II (1:500, Cell Signalling Technology, #1241) and actin (1:10,000, Sigma Aldrich, A5441). Immune complexes were detected by chemiluminescence (ECL, RPN2235, GE Healthcare Life Sciences, Marlborough, MA, USA).
4.7. RNA Isolation, RNA-Seq and qRT-PCR
Total RNA was isolated using TRIzol Reagent (Ambion, 15596018, Waltham, MA, USA) and quantified using a 2.0 Qubit fluorometer (Invitrogen, Carlsbad, CA, USA). RNA integrity was confirmed using an Agilent 2100 Bioanalyzer.
RNA-Seq was performed in duplicate. Paired-end reads were aligned to the mouse genome (mm10) and quantified after adapter removal and bad quality trimming with Trimmomatic [25
] and STAR aligner [25
]. Identification of the differentially regulated genes between A-N0 and A-N5 cells versus A-N100 and C-N5 cells was done using the linear-based model limma voom [27
]. Gene set enrichment analysis was performed with GAGE [28
] using gene sets from MSigDB [29
]. Human Entrez IDs were converted to their mouse ortholog Entrez IDs. For all analyses, the significance thresholds were set to an adjusted p
-value of 0.05. Raw data are accessible on GEO (GSE153991; token: ozgdacsyhihnkf).
cDNA synthesis was performed using the Superscript III First Strand kit from Invitrogen. Reverse transcription was achieved using the Superscript III First-Strand Synthesis System, and qRT-PCR was conducted using an ABI7900 RT-PCR system (Applied Biosystem) with SYBR Green PCR reagents. Primer sequences are listed in Table S9
4.8. Metabolite Extraction
Metabolite extraction and quenching were performed on plate with 1.5 mL ice cold extraction medium (90% methanol, 10% water) containing 1 µg/mL ribitol (A5502, Sigma Aldrich), phenyl β-d-glucopyranoside (292710, Sigma Aldrich), isoguanosine (sc-207768, Santa Cruz, Dallas, TX, USA), d4-succinate (293075, Sigma Aldrich) and methyl-tyrosine (M8131, Sigma Aldrich) as the internal standard. Cells were detached on ice by using a cell scraper, transferred into screw-cap tubes prefilled with 300 mg glass beads (G4649, Sigma Aldrich) and immediately frozen in liquid nitrogen until homogenization. Cells were homogenized using a Precellys tissue homogenizer (P000669-PR240-A, Bertin instruments, Montigny-le-Bretonneux, France) at −10 °C. Three cycles of homogenization for 15 s at 6500 rpm were applied with 10-s breaks in between cycles. Samples were then centrifuged (20,000× g) for 10 min at 4 °C to remove cell debris and protein precipitates, and 500 µL of each supernatant was transferred into two new reaction tubes for GC-MS and LC-MS analyses. Finally, extracts were dried using a vacuum rotator (Eppendorf, Hamburg, Germany) and flushed with nitrogen. The DNA content in the extracts was measured using NanoDrop 1000 (Thermo Fisher Scientific).
4.9. Gas Chromatography–Mass Spectrometry (GC-MS)
To achieve methoxymethylation of keto- and aldehyde groups and trimethylsilylation of amines, hydroxyl groups and carboxylic groups, dried metabolite pellets were derivatized by adding 20 µL methoxyamine hydrochloride (226904, Sigma Aldrich) (20 mg/mL in pyridine) and incubated for 90 min at 28 °C at 1200 rpm and then incubated with 50 µL N
-(trimethylsilyl)-trifluoroacetamide (69479, Merck, Kenilworth, NJ, USA) for 30 min at 37 °C and 1200 rpm. GC-MS analysis was performed using an Agilent 7890 A/5975 C system with a Gerstel MPS2 XL autosampler. An HP5-MS column (5% diphenyl–95% dimethylpolysiloxane, 60 m × 0.25 mm × 0.25 μm) (AG19091S-436E, Agilent, Santa Clara, CA, USA) was used for GC separation (80 °C for 3 min, increase to 320 °C at 5 °C per minute, then hold for 14 min with a carrier gas flow rate of 1 mL/minute and septum purge flow rate of 3 mL/minute). Full scan spectra were acquired from 50 to 800 m/z at a scan rate of 1.99 per second. Equilibration time and post-run time were set to 1 min, inlet temperature to 230 °C, MS source temperature to 230 °C and quadrupole analyzer temperature to 150 °C. Pooled samples were injected at regular intervals to serve as internal controls. Raw data are accessible on https://www.ebi.ac.uk/metabolights/MTBLS2035
4.10. GC-MS Data Analysis
A system-independent Kovats retention index (RI) was generated by using one C10-C40 n-alkane standard sample within the GC-MS sample set. Peak identification, deconvolution and integration were performed by the automated mass spectral deconvolution and identification system (AMDIS) version 2.72 with the following parameters: component width: 12; omit TIC; adjacent peak subtraction: one; resolution: medium; sensitivity: medium; shape requirements: medium. Each sample was processed as a batch job and ELU files were saved as AMDIS outputs and further processed by the SpectConnect online service to generate a compound matrix [31
]. SpectConnect settings were set as follows: elution threshold, 1 min; support threshold, 50%; similarity threshold, 80%. Compound annotation was performed by matching the obtained spectra with the library spectra of FiehnLib [32
], Nist14 [33
], Golm DB [34
] and an in-house database. Compounds were accepted with a match factor threshold of more than 750 and a retention index deviation of less than 5%. Normalization was performed using the total peak area of the chromatogram with phenyl β-d
-glucopyranoside or ribitol as the internal standard. Division of the total peak areas of the chromatogram was previously shown to be an approximation for the total cell number [35
4.11. Liquid Chromatography–Mass Spectrometry (LC-MS)
LC-MS was used to evaluate the amino acid spectrum, NAD+/NADH ratios and TCA cycle metabolites.
For targeted amino acid analyses, cell extracts were directly injected onto a BEH Amide 1.8 µm, 150 × 2.1 mm column using two buffers: 0.1% formic acid (4724.1, Sigma Aldrich) was added to water (buffer A) and acetonitrile (HN40.2, Carl Roth, Karlsruhe, Germany; buffer B). Separation took place using the following program: 90% B to 85% B in 0.2 min, 85% B to 75% B in 0.8 min, 75% B to 40% B in 1 min, 40% B to 50% B in 0.1 min, 50% B for 5.9 min, 50% B to 90% B in 0.2 min and 90% B for 4.8 min. The flow rate was set to 600 µL per minute and column temperature to 50 °C. Methyl-tyrosine served as an internal standard. Pooled samples were injected at regular intervals as internal controls.
For NAD+/NADH-specific analyses, cell extracts were injected onto a Waters acquity HSS T3 1.8 µm, 100 × 2.1 mm column using two buffers: 0.1% formic acid was added to water (buffer A) and methanol (buffer B). Separation took place using the following program: 100% A for 1.5 min, 100% A to 95% A in 2.5 min, 95% A to 5% A in 4 min, 5% A for 8 min, 5% A to 100% A in 0.2 min and 100% A for 7.8 min. The flow rate was set to 200 µL per minute and the column temperature was set to 50 °C. Pooled samples were injected at regular intervals as internal controls.
4.12. LC-MS Data Analysis
LC-MS data analysis was performed using Quantitative Analysis software (Agilent) with specific MRM transitions, qualifier ratios and retention times for metabolite identification. Data were normalized using internal standards and the total peak area of the corresponding sample was obtained by GC-MS or DNA concentration. Division of total peak areas of a chromatogram was previously shown to be an approximation for the total cell number [35
4.13. Oxygen Consumption Rate
Oxygen consumption rate (OCR) was measured using a Seahorse XF96 analyzer (Agilent). Cells were seeded 48 h before the experiment. Cells were exposed to phenol red- and bicarbonate-free DMEM with 292.3 mg/L glutamine and 1.80 g/L glucose immediately prior to the experiment. Cells were incubated for at least one hour at 37 °C in a non-CO2 incubator. The mito stress test assay was performed at final concentrations of oligomycin at 3 µM, FCCP at 1 µM, antimycin A at 2 µM and rotenone at 2.5 µM.
MS data were analyzed using MetaboAnalyst 3.0 [37
]. Missing values were replaced by the half minimum positive value. All experiments were replicated two to three times.
For the GC-MS analyses on asparagine-deprived compared to control mouse sarcoma cells presented in Figure 3
, samples were run in three independent experiments with 4 replicates per condition. Each of the 3 experiments included wild-type control cells grown under physiological asparagine concentrations (W-5). For each experiment, MS data from W-5 runs were averaged and then used to normalize MS data obtained for the other samples. If ≥2 out of 4 W-5 replicates did not detect specific metabolites, these were excluded from the analysis. For all other conditions, missing values were replaced by the half minimum positive value. Only metabolites which were detected in all three experiments were included in the overall analysis.
For the GC-MS analyses on asparagine-deprived mouse sarcoma cells cultured with and without supplemental pyruvate, for the GC-MS analyses on asparaginase- and vehicle-treated mouse sarcoma cells and for the LC-MS analyses, samples were run as part of one single experiment each. Range scaling (mean-centered and divided by the range of each variable) was used for principal component analysis and heat map generation. Cluster and distance analyses were conducted following Pearson and Ward. Differentially regulated metabolites were determined using a one-way analysis of variance (ANOVA) with an adjusted p-value (FDR) cutoff of 0.05. Differences in mean values were tested for statistical significance using Tukey’s HSD post-hoc test.
Differences in cell growth, apoptosis and transcript expression were tested for statistical significance using t-tests or ANOVAs with an adjusted p-value (FDR) cut-off of 0.05. Differences in mean values were determined using Tukey’s HSD post-hoc tests.