The expression matrix and clinical data of patients (TCGA-STAD) with gastric cancer were obtained The Cancer Genome Atlas (TCGA; http://portal.gdc.cancer.gov/) database. To enhance the reliability of the analysis, patients whose survival time was <30 days were excluded from the study. A final total of 348 clinical cases were randomly divided into two groups, each comprising 174 patients, with one used as a training set and the other as a testing set. A review of relevant literature was conducted using the PubMed database (https://pubmed.ncbi.nlm.nih.gov/) to identify genes associated with cuproptosis. The search strategy included the use of keywords such as ‘cuproptosis’, ‘copper-induced cell death’, and ‘copper toxicity’ in combination with terms like ‘genes’ and ‘molecular mechanisms’. Articles published in peer-reviewed journals were considered during the search, and the inclusion criteria required studies to explicitly discuss the genetic or molecular mechanisms underlying cuproptosis. Exclusion criteria included review papers, studies focusing solely on non-genetic aspects, and those unrelated to copper-mediated cell death. The search covered the date range from March 2022 to May 2024. A total of 96 genes were identified from the selected studies. Using the R package ‘limma’ (version 3.58.1, bioconductor.org/packages/3.17/bioc/html/limma.html) for differential expression analysis, 41 CRGs were identified (15).

Single-cell sequencing data were acquired from the Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) database, specifically from dataset GSE163558. This dataset included three primary gastric cancer samples (GSM5004180, GSM5004181 and GSM5004182) and seven metastatic gastric cancer samples (GSM5004183, GSM5004184, GSM5004185, GSM5004186, GSM5004187, GSM5004188 and GSM5004189). These data were subsequently integrated and analyzed for further investigation.

The survival analysis data for CIAO1 were obtained from the Kaplan-Meier Plotter (http://kmplot.com/analysis/) to validate the impact of CIAO1 expression on the overall survival of patients with gastric cancer.

After obtaining the 41 CRGs, 18 prognostic CRGs were subsequently identified using multivariate Cox analysis. Least absolute shrinkage and selection operator (LASSO) regression analysis was then performed, resulting in a risk model that included 15 of these prognostic CRGs. The expression levels of individual genes were multiplied by their specific coefficients to calculate a cumulative gene score for each patient, and the median cumulative gene score was used as a risk score to differentiate the patients into high- or low-risk groups. Receiver operating characteristic (ROC) curves were generated using the ‘pROC’ package (version 1.18.5, cran.r-project.org/package=pROC) (16) to predict survival rates at 1, 3 and 5 years. Additionally, nomograms were created using the ‘survival’ (version: 3.5.8, http://cran.r-project.org/package=survival) (17), ‘regplot’ (version: 1.1, http://cran.r-project.org/package=regplot) and ‘rms’ (version: 6.8.0, cran.r-project.org/package=rms) (18) packages to assess the accuracy and efficiency of the prediction model. These tools were essential for confirming the robustness of the model in terms of its predicting the survival outcomes of patients with gastric cancer.

To investigate the interactions between model genes, a PPI network was constructed using the STRING database (https://string-db.org). The analysis was performed with the interaction score set to a confidence threshold of 0.4 (medium confidence), and organism-specific settings were configured for Homo sapiens. Only experimentally validated interactions and those sourced from curated databases were included to ensure reliability.

The CIAO1 expression data of TCGA-STAD samples were derived from the RNA-seq transcriptome dataset from TCGA). A t-test was used to assess the differences in expression between the tumor and normal samples. CIAO1 expression was then compared across subgroups defined by several clinical characteristics, including treatment outcome, histological type, tumor location, sex, pathological tumor (T)/lymph node (N) stage or age, using a Wilcoxon rank-sum test. Differentially expressed genes (DEGs) between these subgroups were identified using the R package ‘limma’ (version 3.58.1, bioconductor.org/packages/release/bioc/html/limma.html). To elucidate the biological processes and pathways associated with CIAO1, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis and Gene Set Enrichment Analysis (GSEA) were performed on the DEGs using the R packages ‘clusterProfiler’ (version 4.16.0, bioconductor.org/packages/release/bioc/html/clusterProfiler.html) and ‘org.Hs.eg.db’ (version 3.18.0, bioconductor.org/packages/release/bioc/html/org.Hs.eg.db.html) (19).

The distribution of the CIAO1 gene in patients with gastric cancer was assessed using the singlecell RNA-sequencing dataset, GSE163558. Dimensionality reduction was performed using t-distributed stochastic neighbor embedding. Principal cell types within this dataset were identified and annotated using the Leiden clustering method. The ‘tidyverse’ package (version 2.0.0, tidyverse.tidyverse.org/) was subsequently utilized to analyze the expression data of immune checkpoints in different CIAO1 expression groups (20). Furthermore, the tumor mutational burden (TMB) was compared between these two groups using the ‘maftools’ package (version 2.18.0, bioconductor.org/packages/release/bioc/html/maftools.html) (21). Finally, the ‘estimate’ package (version 1.0.13, bioinformatics.mdanderson.org/estimate/) was utilized to calculate and compare immune scores between high- and low-CIAO1 expression groups.

The CellMiner database (https://discover.nci.nih.gov/cellminer/home.do) was used to perform an analysis of drug sensitivity, integrating RNA expression data with drug activity information (22). The analysis was focused on drugs that had been approved through clinical trials by the United States Food and Drug Administration (FDA). Pearson correlation analysis was used to evaluate the correlation between CIAO1 expression and drug sensitivity. Only results with significant significance (P<0.05) were retained. To ensure robust data processing and statistical evaluation, the ‘impute’ (version 1.76.0, bioconductor.org/packages/release/bioc/html/impute.html) and ‘limma’ (version 3.58.1, bioconductor.org/packages/release/bioc/html/limma.html) packages were utilized. Finally, the results were visualized using the ‘ggplot2’ (version 3.5.1, cran.r-project.org/package=ggplot2) and ‘ggpubr’ (version 0.6.0, cran.r-project.org/package=ggpubr) packages (23).

Human gastric mucosa GES-1 (cat. no. HTX1964; Otwo Biotech, Inc.) cells, gastric cancer AGS (cat. no. CL-0022; Pricella Biotechnology), SNU-1 (cat. no. CH1029; Newgainbio, Inc.), MKN-45 (cat. no. CL-0292; Wuhan Pricella Biotechnology Co., Ltd.), HGC-27 (cat. no. CL-0107; Wuhan Pricella Biotechnology Co., Ltd.) cells, 293 (cat. no. CL-0001; Wuhan Pricella Biotechnology Co., Ltd.) cells and cervical cancer HeLa (cat. no. CL-0101; Wuhan Pricella Biotechnology Co., Ltd.) cells were authenticated by short tandem repeat profiling, and used within 6 months after resuscitation of frozen aliquots. Mycoplasma contamination was regularly assessed using the LookOut^® Mycoplasma PCR Detection Kit (cat. no. MP0035; Sigma-Aldrich; Merck KGaA). The gastric-related cell lines GES-1, SNU-1 and MKN-45 were cultured in RPMI-1640 (cat. no. PM150110; Wuhan Pricella Biotechnology Co., Ltd.) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.); the AGS cells were maintained in Ham's F-12K (cat. no. PM150910; Pricella Biotechnology) with 10% FBS; and HGC-27 was cultured in DMEM (cat. no. PM150210; Pricella Biotechnology Co., Ltd.) with 10% FBS. For non-gastric lines, 293 cells were maintained in DMEM (cat. no. PM150210; Pricella Biotechnology Co., Ltd.) with 10% FBS for adherent culture, and HeLa cells were cultured in MEM (cat. no. PM150411; Pricella Biotechnology Co., Ltd.) with 10% FBS. All cell lines were incubated under standard conditions at 37°C with 5% CO₂.

Total RNA was extracted from human gastric cancer cell lines AGS and MKN-45 using the RNeasy^® Mini kit (Qiagen GmbH) according to the manufacturer's instructions. Reverse transcription was carried out using the Transcriptor First Strand cDNA Synthesis Kit (Roche Diagnostics) with the following thermal protocol: 10 min at 65°C for primer annealing, 30 min at 55°C for reverse transcription, and 5 min at 85°C for enzyme inactivation. qPCR was performed using PowerUp SYBR™ Green Master Mix (Thermo Fisher Scientific, Inc.) on the StepOne Real-Time PCR System (Thermo Fisher Scientific, Inc.). The thermocycling conditions were as follows: initial activation at 50°C for 2 min and 95°C for 2 min, followed by 40 cycles of denaturation at 95°C for 15 sec and annealing/extension at 60°C for 1 min. Quantification cycle (Cq) values were determined for each sample, and relative transcript levels were calculated using the 2^−ΔΔCq method, with β-actin serving as the internal reference gene (24). Primer sequences used in this study are listed in Table SI.

Protein extracts were obtained from human gastric cancer cell lines AGS and MKN-45 using RIPA lysis buffer (Promega Corporation). Protein concentrations were determined using the Pierce BCA Protein Assay Kit (cat. A55865; Invitrogen^™; Thermo Fisher Scientific, Inc.). Subsequently, 30 µg aliquots of protein from each sample were subjected to SDS-PAGE on 12% Bis-Tris polyacrylamide gels, and the separated protein products were transferred onto 0.45-µm polyvinylidene fluoride membranes. The membranes were blocked with 5% skimmed milk (Beyotime Institute of Biotechnology) at room temperature for 1 h and incubated with primary antibodies for 4 h at 4°C, followed by incubation with secondary antibodies for 1 h at room temperature. Protein levels were normalized against those of β-actin. The primary antibodies used in the present study included the following: CIAO1 (1:1,000; cat. no. 10295-1-AP; Proteintech Group, Inc.), FDX1 (1:1,000; cat. no. 12592-1-AP; Proteintech Group, Inc.), LIAS (1:1,000; cat. no. 11577-1-AP; Proteintech Group, Inc.) and β-actin (1:20,000; cat. no. 66009-1-Ig; Proteintech Group, Inc.). The secondary antibodies were HRP-conjugated goat anti-rabbit IgG (1:5,000; cat. no. GB23303) and HRP-conjugated goat anti-mouse IgG (1:5,000; cat. no. GB23301; both Wuhan Servicebio Technology Co., Ltd.). Detection of chemiluminescence was performed using an ECL chemiluminescence substrate (Jiangsu Keygen Biotech Co., Ltd.), and protein bands were visualized using an imaging system (cat. no. A44116; Invitrogen; Thermo Fisher Scientific, Inc.) (25).

Oligonucleotides encoding short hairpin (sh)RNAs targeting CIAO1 were sourced from Merck KGaA. The shRNA sequences were subcloned into the GV298 vector (cat. no. GCD0316554; Shanghai Genechem Co., Ltd.), with 2 µg nucleic acid. Transfections were performed using the CHO Transfection Kit (cat. no. A29129; Invitrogen™; Thermo Fisher Scientific, Inc.), conducted at 37°C for 24 h. Following transfection, AGS and MKN-45 cells were cultured for 24 h. The sequences of the oligonucleotides are listed in Table SII. Stable human gastric cancer cell lines AGS and MKN-45 expressing the shRNAs were subsequently selected through antibiotic resistance, employing puromycin (cat. no. A1113803; Invitrogen^™; Thermo Fisher Scientific, Inc.) at cell line-specific concentrations: 2.0 µg/ml for AGS and 1.5 µg/ml for MKN-45 during initial selection (7-day treatment), followed by maintenance at 0.5 µg/ml for both lines (26).

Human gastric cancer AGS and MKN-45 cells were seeded into 96-well plates at a density of 5,000 cells per well (with five replicates per group), before being stably transfected with sh-scramble (scb), sh-CIAO1 #1 or sh-CIAO1 #2 vectors. After a 48-h incubation period at 37°C, 20 µl MTT solution was added to each well, and the plates were incubated at 37°C for a further 4 h. Subsequently, the medium (Ham's F12K) was removed, and DMSO was added to solubilize the formed formazan crystals. The absorbance was then immediately measured at 490 nm using an microplate reader (cat. VLBL00GD2; Invitrogen^™; Thermo Fisher Scientific, Inc.), and the average absorbance for each group was subsequently calculated (27).

A total of 6×10³ cells (AGS and MKN-45) were suspended in 0.3% Noble Agar (Thermo Fisher Scientific, Inc.) and layered on to a base of 0.6% solidified agar in Ham's F12K medium supplemented with 10% FBS in 6-well plates. Following a 3-week incubation period (37°C), the cells were fixed with 100% methanol for 20 min at room temperature and stained with 0.5% crystal violet dye for 30 min at room temperature. After staining, the cells were visualized using a light microscope (28).

A 24-well plate was prepared with 500 µl DMEM per well, and Transwell inserts were precoated with 50 µl Matrigel (cat. no. G4130-5ML; Wuhan Servicebio Technology Co., Ltd.) at 4°C for 30 min before being placed into the wells containing Ham's F12K. Following trypsinization, cells (human gastric cancer cell lines AGS and MKN-45) were centrifuged at 1,200 × g) for 5 min at room temperature, then resuspended and counted. A serum-free cell suspension at an appropriate concentration (10,000 cells per well) was prepared and added to each upper Transwell insert. A total of three replicates were set up for each group. The incubation was performed at 37°C for 24 h. Medium was discarded and the cells were fixed with 100% methanol for 20 min at room temperature before being stained with crystal violet for 30 min at room temperature. Non-invading cells in the upper chambers were carefully removed using cotton swabs. Images of invaded cells on the lower membrane surface were captured using a fluorescent microscope, and the number of cells was counted. For quantification, images of nine random fields per insert were captured at identical magnification and analyzed using the latest version (1.53k) of ImageJ software (National Institutes of Health) (29).

Human gastric cancer cell lines AGS and MKN-45 were cultured on coverslips, fixed with 4% paraformaldehyde for 20 min at room temperature, blocked with 5% milk for 1 h at room temperature and incubated overnight at 4°C with antibodies specific for CIAO1 (1:200, cat. no. 10295-1-AP; Proteintech, Group, Inc.). Subsequently, coverslips were treated with Alexa Fluor 594-conjugated goat anti-rabbit IgG (1:1,000; cat. no. ab150160; Abcam) for 1 h at room temperature, followed by staining with 300 nmol/l DAPI (cat. no. D9542; Sigma-Aldrich; Merck KGaA) for 15 min at room temperature. Finally, images were captured using a Nikon A1Si laser-scanning confocal microscope (Nikon Corporation) (30).

The bioinformatics data were analyzed using R software (version 4.2.1, The R Foundation), SangerBox (version 3.0) (31) and Microinformatics (bioinformatics.com.cn/). For the experimental data obtained from basic research, statistical analysis was conducted using GraphPad Prism 9.0 (Dotmatics). Data are presented as the mean ± SD (n=3). For parametric comparisons between two groups, either unpaired or paired two-tailed Students' t-tests were performed, depending on the experimental design. For non-parametric comparisons between two groups, the Wilcoxon rank-sum test was applied. Moreover, the Pearson correlation coefficient test was used for assessing linear relationships between variables. For comparisons involving ≥3 groups meeting non-parametric assumptions, the Kruskal-Wallis test was employed, followed by Dunn's post hoc test. Clinicopathological characteristics between the training and testing datasets were analyzed using the χ² test or Fisher's exact test, as appropriate. P<0.05 was considered to indicate a statistically significant difference.

A cohort of 348 patients with STAD were selected from TCGA database for the present study, as in Fig. 1. Transcriptome data identified a total of 22,628 differentially expressed genes (DEGs) through a comparison of gene expression levels between gastric cancer tissues and adjacent normal tissues. These were subsequently visualized using a volcano plot (P<0.05 and |log₂FoldChange (FC)|>0.585; Fig. 2A). Through intersecting these DEGs with genes associated with cuproptosis, a total of 41 differentially expressed CRGs (DE-CRGs) were identified (Fig. 2B). The 348 patients with STAD were randomly assigned to either a training or a testing cohort. No significant differences in clinical characteristics were observed between these two groups (Table I). Multivariate Cox regression analysis pinpointed 18 prognosis-associated CRGs. Furthermore, LASSO regression analysis was utilized to construct a risk model comprising a 15-gene signature (Fig. 2C and D). Of these genes, six were demonstrated to perform roles in the assembly of Fe/S proteins, whereas the remaining five were revealed to be associated with metallothionein-binding metals. The risk model score was determined as follows:=(ABCB4 × 0.387)+ (AOC3 × 0.004) + CDKN3 × 0.014)+(CIAO1 × 0.005)+(COA6 × 0.02)-(FDXR × 0.016)+(GLS × 0.02)-(HEPH × 0.009)+ [heat-shock protein (HSP)A9 × 0.006]+(LIPT2 × 0.034)-(MT1A × 0.043)+(MT2A × 0.003)+(MT3 × 0.395)-(NDOR1 × 0.053)-[nitrogen fixation 1 (NFS1) × 0.082]. Kaplan-Meier survival curves demonstrated that patients with STAD who had lower risk scores exhibited significantly improved overall survival (OS) rates compared with those with higher risk scores (Figs. 2E and S1). Additionally, the survival rates at 1, 3 and 5 years, determined by the risk scores of the DE-CRGs, were reflected by areas under the ROC curve (AUC) values of 0.822, 0.811 and 0.922, respectively, indicating high sensitivity and specificity (Fig. 2F). ROC curves were subsequently generated to evaluate clinically relevant factors, and this analysis revealed that the risk score was the most valuable prognostic predictor when compared with other clinical features, such as age, sex, grade and tumor stage (Fig. 2F). To estimate the 1-, 3- and 5-year OS rates of patients with STAD, a nomogram was constructed that included the clinical characteristics of grade, sex, age, stage and risk score (Fig. 2G). Collectively, these analyses demonstrated that, based on the CRGs, the developed risk model exhibited independent and reliable values in terms of predicting the prognosis of patients with STAD.

To further assess the potential impact of risk scores on patients with STAD, the DEGs between these two groups with a significance threshold of P<0.05 and |log2 FC|>1 were identified (Fig. S2A). Gene set enrichment analysis (GSEA) revealed that these DEGs were primarily involved in base-excision repair, one-carbon pool by folate or pantothenate and CoA biosynthesis (Fig. S2B). GO enrichment analysis indicated that these DEGs were primarily associated with digestion, brush border membrane and receptor ligand activity in the high- vs. low-risk groups (Fig. S2C). KEGG pathway enrichment analysis demonstrated that these genes were significantly associated with ‘cAMP signaling pathway’, ‘pancreatic secretion’ and bile secretion (Fig. S2D).

Furthermore, a strong association between risk score and several of the clinical characteristics of patients with STAD was demonstrated. Notably, the risk scores of deceased patients were revealed to be significantly higher compared with those of survivors (P<0.05; Fig. 3A). Additionally, in terms of the primary tumor site, the risk scores for cancers located at the gastroesophageal junction or antrum/distal part of stomach were revealed to be significantly higher compared with those of the fundus/body (P<0.05; Fig. 3H). However, the analysis did not demonstrated any significant differences in risk scores for different subtypes based on parameters such as age, sex, grade, T stage, N stage or M stage (Fig. 3B-G). These results suggest that patients with worse outcomes have higher risk scores in the CRG model.

A correlation heatmap was generated to illustrate the correlations among these risk model signature genes (Fig. 4A). The protein-protein interaction (PPI) networks were obtained from the STRING database (Fig. 4B). To assess the importance of individual genes in 32 gastric cancer cell lines, public gene knockout datasets generated using CRISPR-Cas9 technology were obtained from the iCSDB (icsdb.lk/) database, which indicated that HSPA9, NFS1 and CIAO1 were the genes essential for the viability of gastric cancer cells (Fig. 4C). Based on its known functions, described in the existing literature, CIAO1, an essential gene involved in the assembly of Fe/S proteins and strongly associated with other genes (32) (Fig. 4D), was chosen for further investigation. Kaplan-Meier survival curve analysis revealed that patients with gastric cancer with elevated levels of CIAO1 had a worse prognosis than those with a low expression of CIAO1 (Fig. 4E). Furthermore, differential analysis demonstrated that the level of CIAO1 expression was significantly higher in STAD tissues compared with that in normal tissues (Fig. 4F). Moreover, through mining the Human Protein Atlas (HPA) database (https://www.proteinatlas.org/), immunohistochemistry staining data for the CIAO1 protein indicated that its primary localization was within the cytoplasm, with a markedly stronger staining intensity in gastric cancer tissues compared with that in noncancerous tissues (Fig. 4G). Taken together, these data indicate that an increased level of CIAO1 is crucial for the assembly of Fe/S clusters in STAD.

To further evaluate the association between CIAO1 expression and the prognosis of patients with STAD, expression data and clinical information from TCGA database were analyzed. From a treatment efficacy perspective, CIAO1 expression was revealed to be significantly higher in patients who experienced progressive disease after the first course of treatment, compared with those who achieved complete remission/response and stable disease (Fig. 5A). From a histological standpoint, CIAO1 expression was significantly elevated in the intestinal type of STAD compared with the diffuse type (Fig. 5B). In terms of age stratification, the expression level of CIAO1 in older patients was significantly higher than that in younger patients (Fig. 5E). However, there was no significant difference in CIAO1 expression among subgroups stratified by sex, primary tumor site or T stage (Fig. 5C, D and F). Considered altogether, the aforementioned results suggest that CIAO1 expression is associated with clinicopathological features of STAD.

The distribution of CIAO1 in patients with STAD was subsequently assessed using the single-cell RNA sequencing dataset, GSE163558. Annotation of principal cell types within these datasets revealed that CIAO1 expression was markedly higher in tumor cells compared with other cell types (Figs. 6A and B, and S3). Additionally, analyses were performed to detect differences in other significant characteristics among the several groups. The expression levels of immune checkpoints were specifically assessed, revealing the presence of significantly higher levels of prephenate dehydratase 1, programmed cell death 1 ligand 2, cytotoxic T-lymphocyte associated protein 4, CD80, CD86, lymphocyte activating 3, hepatitis A virus cellular receptor 2, CEA cell adhesion molecule 1 and signal regulatory protein α in the high CIAO1 expression group compared with the low expression group (Fig. 6C). Furthermore, the group with elevated CIAO1 expression exhibited a TMB, indicating a stronger immune response compared with the lower expression group (Fig. 6D). To evaluate tumor microenvironment differences, ESTIMATE scores were calculated for the samples. The results obtained demonstrated that the high CIAO1 expression group had significantly lower ESTIMATE scores compared with the low expression group (Fig. 6E). Additionally, this high expression group exhibited significantly lower immune and stromal scores, whereas tumor purity was significantly higher, compared with the low expression group (Fig. S4). Taken together, these findings indicate that CIAO1 expression is associated with treatment outcomes, histological type, immune checkpoint expression, TMB and the tumor microenvironment in patients with gastric cancer.

To assess the cellular functions influenced by CIAO1, a volcano plot was used to visualize 368 DEGs between the high- and low-CIAO1 expression groups within TCGA database (P<0.05; |log2FC|>1; Fig. 7A). GO enrichment analysis revealed that these DEGs were mainly associated with chromosome segregation, spindle function and microtubule motor activity (Fig. 7B). Furthermore, GSEA revealed that these DEGs were predominantly associated with the cell cycle, pyruvate metabolism and metabolism of alanine, aspartate and glutamate (Fig. 7C). The STRING database was subsequently used to evaluate the PPIs involving CIAO1, resulting in the construction of a network diagram to illustrate these interactions. This analysis revealed that CIAO1 predominantly interacts with proteins such as NUBP Fe/S cluster assembly factor 2, cytosolic, DNA polymerase delta 1 catalytic subunit, ERCC excision repair 2 TFIIH core complex helicase subunit, MMS19, solute carrier family 25 members 5 and 6, CIAO2A and CIAO3 (Fig. 7D). Moreover, KEGG pathway enrichment analysis revealed that these DEGs were significantly associated with the cell cycle, motor proteins and the Fanconi anemia pathway (Fig. 7E). The threedimensional structures of CIAO1 and its interacting proteins were obtained from the UniProt database. Utilizing the HDOCK program (http://hdock.phys.hust.edu.cn/), molecular docking studies were then performed on these proteins and the optimal docking model was selected for further analysis. PyMOL was used to visualize these interactions (Figs. 7F-H and S5). These comprehensive analyses indicate potential molecular mechanisms through which CIAO1 may influence cellular processes in STAD.

To assess potential therapeutic effects of different drugs on CIAO1 expression, the IC₅₀ values of drugs were calculated using the R package ‘pRRophetic’. An initial screening identified 563 drugs, of which 318 were in clinical trials and received FDA approval. The results demonstrated that treatment with the drugs nelarabine, cladribine, nitazoxanide, fludarabine, masoprocol, econazole nitrate, methylprednisolone, ZM336372, idebenone or vorinostat was positively associated with the expression of CIAO1. By contrast, the drugs that were negatively correlated with CIAO1 were dasatinib and AT-9283 (Fig. 8). These data suggest that these two drugs may potentially be suitable as transcriptional inhibitors of CIAO1 expression.

The protein levels of CIAO1 were detected in several human cell lines, including GES-1, AGS, SNU-1, MKN-45, HGC-27, 293 and HeLa cell lines (Fig. 9A). Among the selected cell lines, AGS and MKN-45, which exhibit higher CIAO1 expression in gastric cancer cells, were chosen as model systems for further investigation. Additionally, to evaluate the functional role of CIAO1 in other cancer types, complementary experiments were performed using the HeLa cell line. Gastric cancer (AGS and MKN-45) and HeLa cells with CIAO1 knocked-down were established, and the knockdown of CIAO1 was assessed using RT-qPCR (Figs. 9B and S6A). The RT-qPCR results revealed that the knockdown efficiencies of the two sequences were similar, both effectively reducing CIAO1 expression to a comparable extent. Therefore, both knockdown approaches were utilized in subsequent experiments to ensure consistent validation. As an excessive amount of copper ions directly interact with lipoylated proteins in the tricarboxylic acid cycle, causing their aggregation and the degradation of Fe/S cluster proteins (6), and also in view of the fact that there is evidence to suggest that HSP70 is a heat-shock protein that triggers the proteotoxic stress response during copper death (or cuproptosis) (33,34), FDX1 or LIAS (as Fe/S cluster proteins) and HSP70 were selected as biomarkers of copper-induced cytotoxicity. Notably, the western blot analysis experiments revealed a marked decrease in the protein levels of FDX1 and LIAS, but an increase in the expression levels of HSP70 following CIAO1 knockdown in AGS, MKN-45 and HeLa cell lines (Figs. 9C and S6B). Moreover, immunofluorescence demonstrated that localization of CIAO1 was observed in the nucleus and cytosol of the AGS, MKN-45 and HeLa cells (Figs. 9D and S6C). CIAO1 knockdown was also associated with a notable reduction in the fluorescence intensity of CIAO1. Furthermore, treatment with the CIAO1 inhibitors dasatinib and AT-9283 markedly reduced the expression of CIAO1 and the Fe/S cluster proteins (FDX1 and LIAS), but led to an increase in the levels of HSP70 (Figs. 9E and S6D). MTT colorimetric, soft agar and Matrigel invasion assays also revealed significant reductions in the viability, proliferation and invasion rates of AGS, MKN-45 and HeLa cells following stable CIAO1 knockdown (Figs. 9F, H and I and S6E, G and H). Moreover, MTT colorimetric assays demonstrated significant reductions in the viability of AGS, MKN-45 and HeLa cells following treatment with Dasatinib or AT-9283 (Figs. 9G and S6F). Taken together, these results suggest that pharmacological inhibition of CIAO1 both induces cuproptosis and inhibits the aggressiveness of cancer cells.