Hypoxia-induced cysteine metabolism reprogramming is crucial for the tumorigenesis of colorectal cancer

Abstract

Metabolic reprogramming is a hallmark of human cancer, and cancer-specific metabolism provides opportunities for cancer diagnosis, prognosis, and treatment. However, the underlying mechanisms by which metabolic pathways affect the initiation and progression of colorectal cancer (CRC) remain largely unknown. Here, we demonstrate that cysteine is highly enriched in colorectal tumors compared to adjacent non-tumor tissues, thereby promoting tumorigenesis of CRC. Synchronously importing both cysteine and cystine in colorectal cancer cells is necessary to maintain intracellular cysteine levels. Hypoxia-induced reactive oxygen species (ROS) and ER stress regulate the co-upregulation of genes encoding cystine transporters (SLC7A11, SLC3A2) and genes encoding cysteine transporters (SLC1A4, SLC1A5) through the transcription factor ATF4. Furthermore, the metabolic flux from cysteine to reduced glutathione (GSH), which is critical to support CRC growth, is increased due to overexpression of glutathione synthetase GSS in CRC. Depletion of cystine/cysteine by recombinant cyst(e)inase effectively inhibits the growth of colorectal tumors by inducing autophagy in colorectal cancer cells through mTOR-ULK signaling axis. This study demonstrates the underlying mechanisms of cysteine metabolism in tumorigenesis of CRC, and evaluates the potential of cysteine metabolism as a biomarker or a therapeutic target for CRC.

Keywords: ATF4; Colorectal cancer; Cysteine/cystine; Hypoxia; ROS homeostasis; Transporter genes.

Graphical abstract

Fig. 1

Cysteine is enriched in colorectal tumors. (A) OPLS-DA models for metabolites profiling of tumors (Tumor) and adjacent non-tumor samples (NT) from CRC patients by GC-TOF-MS (cohort 1, n = 20; cohort 2, n = 40). (B) Heatmaps of 41 and 67 differentially expressed metabolites (DEMs) identified from cohort 1 and cohort 2 respectively. (C) Venn diagram of DEMs in different cohorts between colorectal tumors and corresponding adjacent non-tumor tissues. Cohort 1 and cohort 2 were as described in (A). DEMs by Qiu et al. indicated the DEMs which were identified with 376 surgical specimens of CRC from four different hospitals with similar analysis platform [16]. (D) Volcano plots of DEMs from cohort 1 and cohort 2. Y axis represents -log₁₀ (p value) between tumors and non-tumor tissues. X axis represents log₂ (fold change) of tumors versus non-tumor tissues. The bubble size represents variable importance in the projection (VIP) value which reflects the contribution of the metabolites to differentiate tumors and non-tumor tissues. (E) KEGG pathway analysis of DEMs from cohort 1 and cohort 2. Bubble chart shows the enrichment of DEMs in metabolic pathways. Y axis represents -ln (p value) of pathway analysis. X axis represents impact factor (amount of DEMs enriched in the pathway/amount of all metabolites in the pathway). (F) l-cysteine was consistently increased in over 95 % tumors compared to their paired non-tumor tissues in both cohort 1 and cohort 2. (G) The H&E and Ki67 staining images of spontaneous colon tumors which were developed at two months post tamoxifen administration in a genetically engineered mouse model CDX2P-CreER^T2;Apc^15lox/+;Kras^LSL-G12D;R26-Pik3ca^H1047R (CAKP mice). (H) ¹³C₃-l-cysteine was infused in CAKP mice harboring spontaneous colon tumors. (I) Cysteine was enriched in CRC (Tumor) compared with adjacent non-tumor tissues (NT) in mice. Both ¹³C-labeled (red) and unlabeled (black) cysteine were measured, and relative abundance (peak area normalized to internal standard) were present (n = 8). The stacked column suggests total amount of cysteine, and ¹³C-labeled (red) column suggests the exogenous cysteine uptake specifically. Student's t-test was applied. Data are presented as mean ± SD. **p < 0.01; ***p < 0.001; ****p < 0.0001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2

Cystine/Cysteine transporters are upregulated in CRC cells to promote tumor growth. (A) Schematic diagram of exogenous ¹³C₂-cystine or ¹³C₃-cysteine uptake through various transporters. (B) Colorectal cancer cells import both cystine and cysteine. HCT116 or DLD1 cells was treated by either ¹³C₂-cystine (100 μM) or ¹³C₃-cysteine (100 μM). Intracellular cysteine was labeled as ¹³C₁-cysteine (M1) by ¹³C₂-cystine treatment, or as ¹³C₃-cysteine (M3) by ¹³C₃-cysteine treatment. Both unlabeled and ¹³C-labeled cysteine were presented (n = 6). (C) The transporter genes of both cystine and cysteine were significantly upregulated in CRC compared with paired non-tumor colon tissues in TCGA database (n = 50). SLC7A11, the light chain of cystine/glutamate antiporter system; SLC3A2, the heavy chain of cystine/glutamate antiporter system; SLC1A4, alanine/serine/cysteine/threonine transporter 1 (ASCT1); SLC1A5, alanine/serine/cysteine/threonine transporter 2 (ASCT2). mRNA levels represent FPKM value downloaded from TCGA database. (D and E) Knockdown of SLC1A4, SLC1A5, SLC7A11 or SLC3A2 reduced intracellular cysteine abundance in CRC cells. SLC1A4, SLC1A5, SLC7A11 or SLC3A2 was knocked down by siRNAs mixture in HCT116 or DLD1 cells. The knockdown efficiency was validated by qRT-PCR (n = 3) (D), and intracellular cysteine abundance was measured by UHPLC-QTRAP/MS (n = 6 for DLD1 SLC1A4 knockdown cells; n = 3 for others) (E). (F–H) Knockdown of SLC7A11 or SLC1A5 reduced the proliferation of CRC cells. SLC1A5 or SLC7A11 was knocked down by shRNA in HCT116 or DLD1 cells. The knockdown efficiency of SLC1A5 (F) and SLC7A11 (G) was validated by qRT-PCR (n = 3) and Western blots. The proliferation rates were measured by CCK-8 (n = 3) (H). (I–L) Knockdown of SLC1A5 on top of SLC7A11 knockdown further reduced intracellular cysteine content and significantly impaired CRC tumor growth. SLC1A5 was knockdown by siRNAs in DLD1 cells or DLD1 shRNA-SLC7A11 knockdown cells, and the cells were assayed for abundance of intracellular cysteine by UHPLC-QTRAP/MS (n = 3) (I); cell proliferation by CCK-8 (n = 3) (J); xenograft tumor (n = 10) growth rate (K) and weight (L). Two-way ANOVA was used for statistical analyses of H, J, K. Student t-test was used for others. Data of K is presented as mean ± SEM, and data of others are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001; n.s., not significant.

Fig. 3

Hypoxia induces the transcription of cystine/cysteine transporters through ATF4. (A) SLC1A4, SLC1A5, SLC7A11 and SLC3A2 were upregulated concurrently in CRC tumors. The RNA-seq data of SLC1A4, SLC1A5, SLC7A11 and SLC3A2 in CRC and paired non-tumor tissues from TCGA (n = 50) and GSE223120 (n = 20) were analyzed. The fold change (tumor/non-tumor) >1.1 was considered as upregulation. The percentage of tumors which upregulated 4 genes, 3 genes, 2 genes, or 1 gene were presented. (B and C) Hypoxia induced the expression of SLC1A4, SLC1A5, SLC7A11 and SLC3A2. HCT116 and DLD1 cells were incubated in hypoxia chamber (1.0 % O₂, 5.0 % CO₂, 94 % N₂ 37 °C) or normoxia chamber overnight respectively. qRT-PCR was performed to examine the expression levels of SLC1A4, SLC1A5, SLC7A11 and SLC3A2 (n = 3) (B). Two independent sets of HCT116 cells were transfected with indicated pGL3-promoter constructs and internal control Renilla luciferase construct. 2 days post-transfection, two sets of cells were incubated under hypoxia or normoxia conditions respectively overnight. Luciferase activities were then assayed by dual-luciferase reporter assay system (n = 3) (C). (D) The flowchart to screen transcription factors which are responsible for co-expression of SLC1A4, SLC1A5, SLC7A11 and SLC3A2. ChEA3, ChIP-X Enrichment Analysis Version 3; GTRD, Gene Transcription Regulation Database; Hypoxia, the transcription factors which were reported as hypoxia inducible genes; STRING, functional protein association networks. (E and F) Hypoxia-related ATF4 regulates the expression of SLC1A4, SLC1A5, SLC7A11 and SLC3A2. ATF4 was knocked down by shRNAs in HCT116 cells. Cells were incubated under hypoxia condition overnight and the expression levels of indicated genes were examined by qRT-PCR (n = 3) (E). pCMV-ATF4 or empty vector was transfected into HCT116 cells. 48 h post-transfection, the expression levels of indicated genes were examined by qRT-PCR (n = 3) (F). (G) ROS induced the expression of SLC1A4, SLC1A5, SLC7A11 and SLC3A2. Cells were treated with H₂O₂, and the expression levels of indicated genes were examined by qRT-PCR (n = 3). (H) ER stress regulates the expression of SLC1A4, SLC1A5, SLC7A11 and SLC3A2 via ATF4. HCT116 cells were treated with Tg (thapsigargin), and the expression levels of indicated genes were examined by Western blots. (I) Fold enrichment of DNA fragments surrounding putative ATF4 binding sites of indicated gene promoters by ChIP-qPCR with or without H₂O₂ (250 μM) treatment (n = 3). (J) Knockdown of ATF4 significantly inhibited the expression of cystine and cysteine transporter genes induced by hypoxia (overnight), H₂O₂ (250 μM, 6 h), or Tg (250 nM, 6 h). (K and L) Hypoxia induced ATF4 expression through PERK. ATF4 protein levels in PERK knockdown cells (K) or ISRIB-treated cells (p-eIF2α inhibition) (L) were detected. Student t-test was used for statistical analyses. Data are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001; n.s., not significant.

Fig. 4

The flux from cysteine to GSH increases in CRC. (A) Volcano plots of differential expressed metabolites of HCT116 cells cultured with or without cystine/cysteine. Metabolites in cysteine-GSH pathway are highlighted as red. (B) KEGG pathway analysis of DEMs from HCT116 cells with or without cystine/cysteine. Y axis represents -log₁₀ (p value) of pathway analysis. (C) GSH rescued the growth of CRC cells in cystine/cysteine depletion condition. 200 μM of cystine, cysteine, or NAC (N-acetyl cysteine), 400 μM GSH (reduced glutathione) or taurine, nucleotides (1 × EmbyoMax Nucleosides), NEAA (1 × non-essential amino acids), 2 mM glutamate or α-KG (dimethyl α-ketoglutarate), 1 mM sodium pyruvate, 200 μM cystathionine or homocysteine were individually supplemented into cystine/cysteine deficient medium. 48 h later, cell numbers were counted with hemocytometer by Trypan-Blue exclusive assay (n = 3). (D–F) The flux from cysteine to GSH is elevated in CRC cells. Cells were supplied with both 100 μM ¹³C₂-cystine (GSH labeled as M1) and 100 μM ¹³C₃-cysteine (GSH labeled as M3) for indicated time points, fraction of ¹³C-labeled GSH was measured (n = 6) (D). Schematic diagram of ¹³C₃-cysteine flux and related enzymes is shown in (E). Cells were cultured in the presence of 100 μM ¹³C₃-cysteine for 2 h, and percentages of labeled metabolites in total pool were plotted (n = 6) (F). (G) The flux from cysteine to GSH is elevated by hypoxia induced ER stress. DLD1 cells were treated with Tg (250 nM), H₂O₂ (250 μM), or hypoxia overnight, and then incubated with 100 μM ¹³C₃-cysteine for 30 min. Percentages of labeled GSH in total pool were plotted (n = 8). (H) Spontaneous colon tumors were generated with CAKP mice. After infusion with ¹³C₃-cysteine for 4 h, the serum, colorectal tumors and adjacent non-tumor colon tissues were collected, and percentages of labeled metabolites were plotted (n = 8). Student t-test was used for statistical analyses. Data are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001; n.s., not significant. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5

Overexpression of GSS plays important role in CRC growth. (A and B) Glutathione synthetase (GSS) was significantly upregulated in CRC samples compared with normal colon tissues. The mRNA levels of indicated genes of paired CRC and normal tissues in TCGA (n = 50, A) and GSE223120 (n = 20, B) datasets were plotted. (C–H) Knockdown of GSS reduced the growth of colorectal cancer cells. HCT116 or DLD1 cells were transfected with siRNAs against GSS or control siRNA. 48 h post-transfection, the cells were assayed for knockdown efficiency by qRT-PCR (n = 3) (C) or Western blots (D); relative abundance of γ-glutamylcysteine (E) or GSH (F) by LC-MS (n = 6); cell proliferation rate by CCK-8 (n = 3) (G); subcutaneous xenograft tumor growth (n = 10) (H). Two-way ANOVA was used for statistical analyses of G, H. Student t-test was used for others. Data of H is presented as mean ± SEM, and data of others are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001; n.s., not significant.

Fig. 6

Scavenging exogenous cystine/cysteine represses CRC growth. (A) Relative survival of colorectal cancer cells and normal colon epithelial cells under cystine/cysteine depletion condition (n = 3). (B and C) Cyst(e)inase efficiently depletes cystine and cysteine. HCT116 cells were treated with 1 mg/ml cyst(e)inase for 24 h. Intracellular (cells) and exogenous (medium) cystine or cysteine were measured by UHPLC-QTRAP/MS (n = 6) (B). C57BL/6 mice (n = 6) were intraperitoneally injected with cyst(e)inase (50 mg/kg). 24 h later, mice serum was collected to measure cystine or cysteine levels by UHPLC-QTRAP/MS (C). (D) The fold changes of cysteine-related metabolites abundance after HCT116 cells were treated with cystine/cysteine depletion condition or cyst(e)inase for 24 h. (E–G) Cyst(e)inase inhibited the CRC tumor growth. Nude mice were subcutaneous injected with either HCT116 cells (E) or small pieces of patient-derived xenograft tumors (F). Once tumors reached 100–150 mm³, mice were divided into two groups to intraperitoneally inject with PBS vehicle or cyst(e)inase (50 mg/kg) thrice a week. CAKP mice were administrated with tamoxifen. 40 days post injection, mice were divided into two groups to intraperitoneally inject with PBS vehicle or cyst(e)inase (50 mg/kg) thrice a week. Kaplan-Meier analyses was used to show the survival of mice with or without cyst(e)inase treatment (G). Student t-test was used for statistical analyses. Data are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001; n.s., not significant.

Fig. 7

Depletion of cystine/cysteine induces autophagy of CRC cells through mTOR-ULK pathway. (A) Depletion of cystine/cysteine affects cellular autophagy, apoptosis or ferroptosis. Cells were cultured under cystine/cysteine depletion condition (-cyst(e)ine) or treated with cyst(e)inase (+cyst(e)inase) for 48 h. Cell lysates were collected and cell death markers were detected by Western blot assays. (B) Depletion of cystine/cysteine, but not glutamine or methionine, significantly induced autophagy in CRC cells but not in normal epithelial cells. (C–E) Cystine/cysteine depletion significantly induces autophagy in CRC cells. mCherry-GFP-LC3-expressing HCT116 and DLD1 cells were cultured under cystine/cysteine depletion condition or treated with cyst(e)inase for 24 h. Representative images of fluorescent LC3 puncta were show in (C). Mean number of GFP and mcherry dots per cell (n = 3) (D); and mean number of autophagosomes (yellow dots in merged images) and autolysosomes (red dots in merged images) per cell were counted (n = 3) (E). (F and G) Cystine/cysteine depletion induces autophagy of CRC cells through mTOR-ULK1 pathway. The dose-dependence (F) and time-dependence (G) of autophagy (the expression of LC3A and LC3B), mTOR activity (p-p70S6K) and ULK activity (p-ULK) upon cystine/cysteine depletion were assayed by Western blots. Student t-test was used for statistical analyses. Data are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001; n.s., not significant. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)