Aspartate is a limiting metabolite for cancer cell proliferation under hypoxia and in tumours

Abstract

As oxygen is essential for many metabolic pathways, tumour hypoxia may impair cancer cell proliferation^1-4. However, the limiting metabolites for proliferation under hypoxia and in tumours are unknown. Here, we assessed proliferation of a collection of cancer cells following inhibition of the mitochondrial electron transport chain (ETC), a major metabolic pathway requiring molecular oxygen⁵. Sensitivity to ETC inhibition varied across cell lines, and subsequent metabolomic analysis uncovered aspartate availability as a major determinant of sensitivity. Cell lines least sensitive to ETC inhibition maintain aspartate levels by importing it through an aspartate/glutamate transporter, SLC1A3. Genetic or pharmacologic modulation of SLC1A3 activity markedly altered cancer cell sensitivity to ETC inhibitors. Interestingly, aspartate levels also decrease under low oxygen, and increasing aspartate import by SLC1A3 provides a competitive advantage to cancer cells at low oxygen levels and in tumour xenografts. Finally, aspartate levels in primary human tumours negatively correlate with the expression of hypoxia markers, suggesting that tumour hypoxia is sufficient to inhibit ETC and, consequently, aspartate synthesis in vivo. Therefore, aspartate may be a limiting metabolite for tumour growth, and aspartate availability could be targeted for cancer therapy.

Figure 1. Diversity of cancer metabolic responses to ETC inhibition

A) Proliferation of 28 cancer cell lines treated with electron transport chain inhibitors. Graphical scheme depicting the targets of complex I (100 uM phenformin, 10 nM piericidin), complex III (30 nM antimycin A) and complex V (100 nM oligomycin) (top). Heat map indicating the changes in cell numbers upon treatment with ETC inhibitors as calculated z-scores (bottom). B) Correlation of the sensitivities of 28 cancer cell lines to different ETC inhibitors. n=3 biologically independent samples per cell line. C) Metabolites significantly altered between ETC inhibition resistant and sensitive cell lines upon 8 h piericidin (10 nM) treatment, ranked by p-value(-log₂ transformed). D) The change (log₂) in amino acid abundance of ETC inhibition resistant and sensitive cell lines upon piericidin (10 nM) treatment (The boxes represent the median, and the first and third quartiles, and the whiskers represent the minimum and maximum of all data points. P =0.002). For each experiment, 6 ETC inhibition resistant and 6 sensitive cell lines were used. n=3 biologically independent samples per cell line. Statistics: two-tailed unpaired t-test. For individual P values, see Supplementary Table 1. Statistics source data are provided in Supplementary Table 1.

Figure 2. Aspartate import underlies the resistance of cancer cells to ETC inhibition

A) ¹⁴C-Aspartate uptake in ETC inhibition resistant and sensitive cancer cell lines. (mean ± S.E.M., n=3 biologically independent samples). B) Aspartate depletion sensitizes aspartate importing cancer cell lines to ETC inhibitors. L-Aspartate (150 uM), antimycin (30 nM) and piericidin (10 nM) were added where indicated. (mean ± S.E.M., 5 cell lines in each group, n=3 biologically independent samples.) C) Correlation of the sensitivities to ETC inhibitors with transcriptome-wide mRNA expression data from the Cancer Cell Line Encyclopedia (CCLE). The resulting Pearson correlation coefficients were sorted and plotted. The red dot indicates SLC1A3. (n=28 cell lines from Fig. 1A). D) Immunoblot analysis of SLC1A3 in ETC inhibition resistant and sensitive cell lines. Actin, loading control. E) Immunoblot analysis of SLC1A3 in wild-type, SLC1A3-null, and rescued null SNU-1 cells. Actin, loading control. F) ¹⁴C-Aspartate uptake in wild type, SLC1A3 null and SNU-1 cells treated with an SLC1A3 inhibitor (TFB-TBOA, 20 uM). (mean ± S.E.M., n = 3 biologically independent samples). G) Loss of SLC1A3 sensitizes cancer cells to antimycin and piericidin in RPMI. Relative cell number of wild-type (black), SLC1A3-null (blue) and rescued SLC1A3-null (gray) SNU-1 cells in the absence and presence of pyruvate (1 mM) or TFB-TBOA (20 uM) after treatment with piericidin (10 nM) and antimycin (30 nM) for 5 days. (mean ± S.E.M., n=3 biologically independent samples). H) Pharmacologic inhibition of SLC1A3 sensitizes aspartate importing cancer cell lines to antimycin (30 nM) and piericidin (10 nM). Cells were grown for 5 days in the absence or presence of TFB-TBOA (20 uM). (mean ± S.D., 5 cell lines in each group, n=3 biologically independent samples). I) Expression of SLC1A3 rescues A549 from the anti-proliferative effects of antimycin and piericidin in standard RPMI media, which contains 150 uM aspartate. Relative cell number of control (black), and SLC1A3 overexpressing (gray) A549 cells in the absence and presence of pyruvate (1 mM) or TFB-TBOA (20 uM) after treatment with piericidin (10 nM) and antimycin (30 nM) for 5 days. (mean ± S.E.M, n=3 biologically independent samples). Statistics: two-tailed unpaired t-test. For individual P values, see Supplementary Table 1. Statistics source data are provided in Supplementary Table 1.

Figure 3. Aspartate is a limiting metabolite for cancer cell proliferation under hypoxia and in tumors

A) Differential intracellular amino acid abundances (log₂) of A549 and PANC-1 cells upon piericidin (10 nM) treatment or under 0.5% oxygen, relative to untreated control cells under 21% oxygen. The cells were cultured in RPMI media without glutamate and asparagine. B) Aspartate abundance in control (Vector, black) and SLC1A3 expressing (SLC1A3, gray) cell lines under hypoxia, relative to controls under 21% oxygen. (mean ± S.E.M, n=3 biologically independent samples). C) Relative cell number of control (Vector) or SLC1A3 overexpressing (SLC1A3) cell lines under 21% and 0.5% oxygen with or without supplementation of aspartate (150 uM) and TFB-TBOA (20 uM). The cells were cultured in RPMI media without aspartate, glutamate and asparagine. KP lung and pancreas indicates Kras^G12D/p53^−/− mouse cancer cell lines. (mean ± S.E.M., n=4 biologically independent samples for PANC-1 and A549, 3 biologically independent samples for KP lines). D) General scheme of the cancer cell competition experiment (left). Relative abundance of A549 and Kras/p53 lung cancer cells transduced with a control vector or with SLC1A3 cDNA grown in vivo as xenografts or in vitro under antimycin (30 nM) treatment or 0.5% oxygen (right). Results are plotted relative to untreated cells under 21% oxygen. (mean ± S.E.M, n=6 biologically independent samples for in vitro conditions, n=8 biologically independent samples for A549 tumors, n=14 biologically independent samples for KP Lung tumors). E) Schematic depicting the metabolic routes of aspartate in pyrimidine and purine synthesis. Filled circles represent ¹³C or ¹⁵N atoms derived from [U-¹³C]-L-aspartate or [¹⁵N]-L-Aspartate (left). Fraction of labeled nucleotide precursors derived from labeled aspartate in control and SLC1A3 overexpressing A549 cells cultured for 24 hr with [U-¹³C]-L-Aspartate (150 uM) or [¹⁵N]-L-Aspartate (150 uM) upon piericidin treatment (10 nM) or under 0.5% oxygen. Colors indicate mass isotopomers (mean ± S.D., n=3 biologically independent samples) (right). F) Relative cell number of control (Vector) or SLC1A3 overexpressing (SLC1A3) cell lines under 0.5% oxygen with or without supplementation of aspartate (150 uM) or nucleosides (thymidine, uridine, adenosine, cytidine and inosine; 100 uM). The cells were cultured in RPMI media without aspartate, glutamate and asparagine. KP lung and pancreas indicates Kras^G12D/p53^−/− mouse cancer cell lines. Results are plotted relative to cells cultured under 0.5% oxygen without aspartate or nucleoside supplementation. (mean ± S.E.M., n=3 biologically independent samples, *p < 0.01, **p < 0.001, ***p < 0.0001). Statistics: two-tailed unpaired t-test. For individual exact P values, see Supplementary Table 1. Statistics source data are provided in Supplementary Table 1.

Figure 4. Aspartate levels inversely correlate with the expression of hypoxia-induced genes in primary human tumors

A) Correlation of 128 metabolites and hypoxia induced mRNA markers (CA9, VEGF and HK2) in 24 human glioblastoma tumors. B) Relative aspartate, lactate and acyl-carnitine levels in VEGF high (n=12 biologically independent samples) and low (n=12 biologically independent samples) tumors to population minimum. The boxes represent the median, and the first and third quartiles, and the whiskers represent the minimum and maximum of all data points. Statistics: two-tailed unpaired t-test. C) Proposed mechanism of the limiting role of aspartate in cell proliferation under tumor hypoxia.