Cysteine dioxygenase 1 attenuates the proliferation via inducing oxidative stress and integrated stress response in gastric cancer cells

Abstract

Whereas cysteine dioxygenase 1 (CDO1) expression is lost due to its hypermethylated promoter across a range of cancer types including gastric cancer (GC), its functions and molecular underpinnings remain largely unknown. Here we demonstrate that reduced CDO1 expression is indicative of unfavorable prognosis in patients with GC. CDO1 overexpression in GC cells markedly inhibits cellular proliferation in vitro and in vivo. Mechanistically, CDO1 exerts this cytostatic effect via increasing oxidative stress and thus activating integrated stress response (ISR) in GC cells. High throughput screening (HTS) of antioxidants library identifies that Engeletin, a flavanonol glycoside, blunts oxidative stress and the ISR to relieve the inhibitory effect of CDO1 on the proliferation in GC cells. Additionally, genetic disruption or pharmaceutical inhibition of the ISR boosts the growth in the GC cells with CDO1 expression. Our data uncover the molecular mechanisms underlying the cytostatic function of CDO1 in the proliferation of GC cells.

Fig. 1. Reduced expression of CDO1 in GC tissue is indicative of poor prognosis.

A CDO1 mRNA level was examined in 30 paired GC and adjacent normal tissues, showing that it decreased in GC specimens relative to normal ones. B Analysis of CDO1 mRNA expression in one independent GC patient cohort (GSE29272) demonstrated that it was lower in GC samples than in normal counterparts. C IHC showed that CDO1 staining was stronger in normal gastric mucosa than in GC tissues. The representative images of one pair of tissues are presented here. Scale bars, 400 μm and 100 μm respectively. D Statistical analysis showed 50% of normal samples with high CDO1 protein expression while around 20% of GC tissues with high CDO1 protein level (N = 130). E Kaplan–Meier analysis of overall survival in GC patients with high or low CDO1 protein level showed that patients with higher CDO1 expression had better survival rate. The high or low expression of CDO1 in Kaplan–Meier analysis was stratified by median CDO1 protein level from IHC staining quantifications. **p < 0.01, ***p < 0.001.

Fig. 2. The inhibitory effect of CDO1 on the proliferation in GC cells.

A The re-expression efficiency of CDO1 in MKN45 (left panel) and NCI-N87 (right panel) cells was verified by immunoblots. B Cellular growth of MKN45 (left panel) or NCI-N87 (right panel) cells from N.C. or CDO1-restored group was assayed using CCK-8 at the indicated time points respectively (N = 6, mean ± SD), showing that CDO1 remarkably suppressed the proliferation in these two GC cell lines. C MKN45 (left panel) or NCI-N87 (right panel) cells from N.C. or CDO1-restored group were stained with EdU and analyzed flow cytometry, showing that CDO1 markedly blocked the proliferation in GC cells. One representative result (N = 3, mean ± SD) are graphically presented here. D, E Tumor xenografts model based on female Balb/c nude mice demonstrated that the proliferation of MKN45 (D) and NCI-N87 (E) cells was substantially inhibited by CDO1 in vivo, as evidenced by the delayed growth at the indicated time intervals and the decreased weights of the tumor masses from MKN45 or NCI-N87 cells. **p < 0.01, ***p < 0.001.

Fig. 3. The anti-proliferation function of CDO1 is in an enzymatic activity-dependent manner.

A The expression efficiency of CDO1^WT and CDO1^Y157F was confirmed via immunoblots in MKN45 (left panel) and NCI-N87 (right panel) cells. B CDO1^Y157F, unlike CDO1^WT, did not attenuate the growth of MKN45 (left panel) or NCI-N87 (right panel) cells in vitro, as indicated by CCK-8 assay at the indicated time points (N = 6, mean ± SD). C Tumor xenografts model showed that CDO1^Y157F could not inhibit the growth of MKN45 cells in vivo. The tumor masses from the indicated three groups were photographed and weighed (N = 10 mice per group). ***p < 0.001, n.s. means no significance.

Fig. 4. CDO1 induces oxidative stress in GC cells.

A, B CDO1^WT markedly increased ROS level while CDO1^Y157F did not alter ROS production in MKN45 (A) and NCI-N87 (B) cells. C, D CDO1^WT rather than CDO1^Y157F decreased GSH/GSSG ratio in MKN45 (C) and NCI-N87 (D) cells. E Heatmap of the fold changes in all antioxidants subjected to HTS relative to MKN45 cells treated with DMSO (Vehicle). Intracellular ATP level, measured using CellTiter-Glo, was used as the surrogate of viable cells. Luminescent reads of engeletin-treated MKN45 cells approximately increased by two times. The fold changes were measured as: Fold change = (Reads _{(antioxidant)}-Reads _(Vehicle)) / Reads _(Vehicle). F Engeletin-treated MKN45 (left panel) or NCI-N87 (right panel) cells with CDO1^WT restoration displayed reduced ROS production compared to these cells exposed to vehicle. G Treatment of Engeletin in MKN45 (left panel) or NCI-N87 (right panel) cells with restored CDO1^WT increased ATP generation relative to vehicle-treated cells. H CCK-8 assays demonstrated that engeletin treatment relieved the inhibition of CDO1^WT on the proliferation in MKN45 (upper panel) or NCI-N87 (bottom panel) cells in vitro. *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 5. CDO1 restoration impairs mitochondrial function.

A JC-1 staining was performed as described in Materials and Methods. The representative images of JC-1 staining in the control or the CDO1-restored MKN45 or NCI-N87 cells are presented here (left panel). The fluorescence intensity ratio of JC-1 aggregates (red) against JC-1 monomers (green) markedly decreased in MKN45 and NCI-N87 cells, indicating that mitochondrial depolarization occurred in GC cells with CDO1^WT re-expression. Scale bars, 100 μm. B ATP generation, which is the major function of mitochondria, was substantially reduced as the result of CDO1^WT restoration. C Test of OCR in MKN45 (left panel) and NCI-N87 (right panel) cells from the N.C. and CDO1^WT group, showing that respiration activity of mitochondria was markedly blocked by CDO1^WT. Bottom, quantification of basal respiration and maximal respiration as measured by OCR (mean ± SD; N = 3). Data shown is a representative experiment of two independent ones. D Metabolomics study targeting key metabolites implicated in cellular ATP generation process showed that succinate, AMP, cis-aconitate, GMP, NADH, and NADPH were all significantly decreased in CDO1^WT-overexpressing MKN45 cells compared to N.C. ones. *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 6. CDO1-induced oxidative stress triggers ISR.

A The expression of several key downstream genes of ISR, such as ATF3, ATF4, TRIB3 and GADD34, was examined by qPCR, demonstrating that CDO1^WT, rather than CDO1^Y157F, upregulated the mRNA expression of these genes in MKN45 (upper panel) and NCI-N87 (bottom panel) cells. B Immunoblots showed that ATF4 accumulated in the nucleus and phosphorylated eIF2α (Ser51) level in the cytoplasm increased in MKN45 (left panel) and NCI-N87 (right panel) cells with CDO1^WT re-expression. However, CDO1^Y157F did not activate ISR. C Engeletin reduced the nuclear accumulation of ATF4 and cytoplasmic increase in phosphorylated eIF2α (Ser51), which were induced by CDO1^WT, in MKN45 (left panel) and NCI-N87 (right panel) cells. D HRI knockdown relieved the CDO1^WT-induced suppression on the proliferation in MKN45 and NCI-N87 cells in vitro. E Immunoblots showed that HRI knockdown decreased nuclear expression of ATF4 and cytoplasmic level of phosphorylated eIF2α (Ser51) in CDO1^WT-restored MKN45 (left panel) and NCI-N87 (right panel) cells. Lamin-B1 was used as the loading control for nuclear proteins and Vinculin for cytoplasmic proteins, respectively. *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 7. The inhibitory effect of CDO1 on the proliferation in GC cells relies on ISR.

A, B ISRIB, a small inhibitor of ISR, was added in N.C. and CDO1^WT GC cells, showing that it attenuated ISR activation by reducing the nuclear level of ATF4 and cytoplasmic level of phosphorylated eIF2α (Ser51) (A). ATP production was increased in response to ISRIB treatment in CDO1^WT-restored MKN45 (left panel) and NCI-N87 (right panel) cells (B). C, D The Ser51 was mutated to Ala51 to generate the activity-dead eIF2α (eIF2Α^S51A). Immunoblots showed that eIF2Α^S51A, compared to eIF2Α^WT, remarkably blocked the activation of ISR (C) and, as uncovered by luminescent assay, increased ATP generation to some extend (D) induced by CDO1^WT in GC cells. Left panel, MKN45 cells; right panel, NCI-N87 cells. E Subcutaneous tumor xenograft model showed that eIF2Α^S51A markedly relieved the inhibition of CDO1^WT on the growth of MKN45 cells in Balb/c nude mice (N = 5 per the indicated group). F, G GADD34 was expressed in MKN45 (left panel) and NCI-N87 (right panel) cells with overexpressed CDO1, showing that GADD34 partially suppressed ISR activation (F) and restored ATP production (G). H GADD34 markedly restored the growth of MKN45 cells in vivo (N = 6 per the indicated group). Lamin-B1 was used as the loading control for nuclear proteins and Vinculin for cytoplasmic proteins, respectively. **p < 0.01, ***p < 0.001.