Roles of cuproptosis-related gene DLAT in various cancers: a bioinformatic analysis and preliminary verification on pro-survival autophagy

Gene and protein expressions of DLAT

Two distinct analysis tools, which are the Xiantao tool (https://www.xiantao.love/) (Chu et al., 2022) and The University of Alabama at Birmingham Cancer data analysis Portal (UALCAN, http://ualcan.path.uab.edu/index.html) (Zhang et al., 2022b), were used to differentiate the DLAT expressions of tumors and the corresponding normal tissues in The Cancer Genome Atlas (TCGA) database. Because some tumors in TCGA lacked data on corresponding normal tissues, we referred to the Gene Expression Profiling Interactive Analysis 2 (GEPIA2) website. Then, we obtained these normal data with the log2FC cutoff value set as 0.25 in the analysis (http://gepia2.cancer-pku.cn/#analysis) (Li et al., 2021). With the UALCAN tool, we also compared the expression levels of DLAT in different tumor grades. Furthermore, we used the immunohistochemistry (IHC) staining images on the Human Protein Atlas (HPA) website to present the expression levels of DLAT in tumors and normal tissues (http://www.proteinatlas.org/) (Uhlen et al., 2015).

Cell cultures and DLAT siRNA breakdown

Two cell lines of LIHC, Huh7, and HepG2 were provided by the Cancer Research Institute of Central South University. The cells were cultured in a DMEM medium (HyClone, Logan, UT, USA), in which penicillin-streptomycin (1%; Gibco, Waltham, MA, USA) and fetal bovine serum (10%; Gibco, Waltham, MA, USA) were added to facilitate cell growth. The incubator conditions for culturing these two cell lines were set at 37 °C with 5% CO2. Two small interfering RNAs (siRNA) were specifically synthesized to target DLAT (siDLAT-1, 5′-GGCCAACCGAAGTAACAGA-3′, and siDLAT-2, 5′-CACTCTGTATCATTGTAGA-3′; RiboBio, Guangdong, China), with a scrambled siRNA as a negative control (RiboBio, Guangdong, China). Transfection for cell lines was carried out with the Lipofectamine L3000-015 reagent (Invitrogen, Waltham, MA, USA) following the protocol of Invitrogen.

Western blot

Using the RIPA buffer, we made lysates of two LIHC cell lines with phosphatase-protease inhibitors (Bimake, Houston, TX, USA). The BCA protein assay (Thermo Fisher, Waltham, MA, USA) was performed to calculate the protein concentrations. Protein lysates with the same volume were transferred into PVDF membranes of 0.22 µm and 0.45 µm (Millipore, USA) after exposure to SDS-PAGE electrophoresis. PVDF membranes were sealed with 5% skimmed dry milk for 1 h and then incubated with 1:1000 primary antibodies of DLAT (Signal Antibody, Greenbelt, MD, USA), LC3 I/II (CST, Danvers, MA, USA), Beclin-1 (CST, Danvers, MA, USA) and GAPDH (Proteintech, Wuhan, China) at 4 °C overnight. Immobilon Western chemiluminescent WBKLS0500 reagent (Millipore, USA) was used to present proteins on PVDF membranes.

Colony formation assay

The test protocol has been described in our earlier studies (Chen et al., 2021b) that cells were re-suspended and seeded in 6-well petri dishes with a density of about a thousand cells in each well. After at least 14 days of incubation, supernatants of the petri dish were discarded, and clones were fixed with pure methanol for 15 min and stained with crystal violet for 20 min.

Transmission electron microscopy (TEM)

The cells were digested with trypsin and then treated overnight with 2.5% glutaraldehyde reagent. Subsequently, the cells were transferred to the TEM Laboratory of Xiangya Hospital, China, and treated as follows. The cells were cleaned three times with an interval of 10 min in the Millonig phosphate buffer and then incubated with 1% osmium tetroxide for 1 h before being rewashed. Cells were dehydrated successively with 50%, 70%, and 90% acetone and then dehydrated twice with 100% acetone. The samples were soaked and embedded for 12 h in a mixture of acetone and resin (1:1) and then polymerized overnight with pure resin at 37 °C. After that, the samples were further incubated at 60 °C for 12 h and finally sliced into 50–100 nm ultra-thin cell sections with an ultramicrotome. The cell sections were observed under an HT-7700 transmission electron microscope (Hitachi, Japan) after being stained with uranyl acetate and lead nitrate.

DNA methylation and genetic alteration analysis

The UALCAN tool was applied to differentiate the DNA methylation of DLAT in tumors and normal tissues in the TCGA database. With the cbioportal tool, an online genetic alteration analysis of DLAT genes in the TCGA Pan-Cancer Atlas was carried out, with the frequency of genetic alteration provided by the “Cancer Types Summary” module and the “mutation” module on the genetic mutation site (https://www.cbioportal.org/) (Gao et al., 2013). Moreover, the survival rates of cancer patients in the DLAT-altered group and the DLAT-unaltered group in the TCGA database were compared in the “Comparison” module.

Survival prognosis analysis and function enrichment analysis

The survival maps for overall survival (OS) across all TCGA cancer types were analyzed with the GEPIA2 tool. Using the Kaplan–Meier Plotter website, we further validated the relationships between DLAT expression and multiple prognostic markers in five tumors, including breast, liver, lung, ovarian, and gastric cancers (https://kmplot.com/analysis/). In addition, with the Xiantao tool, the univariate and multivariate Cox regression analyses of DLAT mRNA levels were performed on OS endpoints of 33 different tumor types in the TCGA with all available clinical variables adjusted.

For LIHC, the “Survival analysis” module of the Biomarker Exploration of Solid Tumors (BEST) method (https://rookieutopia.com/) was applied to perform the Cox regression and Kaplan–Meier analyses on DLAT expressions. With the UALCAN tool, the Kaplan–Meier analysis on the TCGA data set revealed the impact of DLAT expression level on the LIHC patients’ survival rate. A prognosis analysis was also conducted with the Kaplan–Meier Plotter on sub-types of liver cancers with statistically significant clinical variables. In addition, the Gene Set Enrichment Analysis (GSEA) of DLAT-related genes was conducted with the Xiantao website to identify whether DLAT was involved in autophagy.

Expressions of DLAT and responses to chemotherapy

We applied the Receiver Operating Characteristic plotter (ROC plotter) website (https://www.rocplot.org/cells) (Tibor Fekete & Gyorffy, 2022) to verify the relationships between DLAT levels and responses to chemotherapy among LIHC patients. These patients were divided into sensitive and resistant groups according to their responses to chemotherapy drug treatment on the basis of lower and upper tertiles of AUDRC.

Immune infiltration analysis

With the “Immune” section on the Tumor Immune Estimation Resource 2.0 (TIMER2.0) website, the relationships between DLAT expressions in TCGA tumors and the infiltration of various immune cells, including B cells, DC cells, NK cells, CD8+ T cells, cancer-associated fibroblasts, macrophages, monocytes, and neutrophils can be revealed. Several algorithms were used to estimate immune infiltration, including TIMER, XCELL, QUANTISEQ, MCPCOUNTER, EPIC, TIDE, CIBERSORT, and CIBERSORT-ABS. Furthermore, an analysis of the correlations between the DLAT’s expression level and typical phenotype markers of Tregs was performed with the Xiantao tool and GEPIA2 tool. We used the “Immunotherapy” module of the BEST method to evaluate the correlations between different expressions of DLAT and the response and prognosis of immunotherapy.

Tissue microarrays and immunohistochemistry

We used tissue microarrays of 75 pairs of LIHCs and their adjacent normal tissues (Hliv-HCC150CS-02; Outdo Biotech, Shanghai, China) approved by the Ethics Committee of Shanghai OUTDO Company (YB M-05-02). The antibodies against DLAT and CD49d (19676-1-AP, Proteintech) were used for IHC. The concentrations of the antibodies against DLAT and CD49d were 1:100. Two professional pathologists scored the staining of DLAT and CD49d in tissue microarrays separately. According to the standard intensity scoring, a score of 0 indicates a negative intensity, a score of 1 indicates a weak intensity, a score of 2 indicates a moderate one, and a score of 3 indicates a strong one. The extent scores are divided into five grades, with 0, 1, 2, 3, and 4 indicating the extent of no more than 10%, 11%–25%, 26%–50%, 51%–75%, and more than 75%, respectively. Finally, the multiplying results of intensity scores and extent scores were viewed as the staining scores that were recorded.

Statistical analysis

All experiments were repeated three times independently. Also, the Student’s t-test method was used to compare the differences between these two groups’ sample data. The statistical analysis results on the online database were shown in the figures, with some analysis methods not clearly marked. Data processing was conducted with GraphPad Prism 8 and SPSS 23.0. All statistical calculation results were judged according to the following criteria: * P < 0.05; ** P < 0.01; *** P < 0.001.

DLAT pan-cancer expressions

In order to compare the mRNA expression levels of DLAT in tumors and normal tissues, the Xiantao tool, UALCAN tool, and GEPIA2 tool were used to perform differential analysis on the TCGA and GTEx databases. Figs. 1A–1C show the analysis results with different databases and tools that the mRNA levels of DLAT were all significantly upregulated in certain tumor tissues, including cholangiocarcinoma (CHOL), esophageal carcinoma (ESCA), LIHC, lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC) and stomach adenocarcinoma (STAD). However, the mRNA levels of DLAT were all significantly down-regulated in head and neck squamous cell carcinoma (HNSC) and kidney renal clear cell carcinoma (KIRC). The IHC staining images in the HPA database confirm that the above eight tumor types that show significantly different mRNA expressions of DLAT also present different protein expressions of DLAT. The tumor tissues of LIHC, STAD, and lung cancers have stronger DLAT staining than their corresponding normal tissues, consistent with the mRNA expression results. However, as for CHOL and ESCA, there lack the IHC results.

The normal kidney, head, and neck tissues show strong or moderate DLAT staining. In contrast, their corresponding tumor tissues show weak staining (Fig. 1D). The relationship between DLAT expression and tumor grade was investigated with the UALCAN, with the result that DLAT expressions varied with the tumor grades in several tumors, including KIRC, brain lower grade glioma (LGG), LIHC, ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD) and STAD (Figs. 1E–1J).

The expressions of DLAT are related to DNA methylation and genetic alteration

Epigenetic modifications are mainly realized by DNA methylation, which has been proven to be a key contributor to the process of tumorigenesis and tumor progression (Abante et al., 2021; Chen et al., 2021a; Fatima et al., 2021; Zhang et al., 2021). Through comparing the methylation roles of DLAT in tumors and normal tissues, we found that the levels of DNA methylation were significantly increased in bladder urothelial carcinoma (BLCA), colon adenocarcinoma (COAD), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), ESCA, HNSC, prostate adenocarcinoma (PRAD) and rectum adenocarcinoma (READ) (Figs. 2A–2G). On the contrary, we found relatively low DNA methylation levels of DLAT in KIRC, LIHC, LUAD, pheochromocytoma and paraganglioma (PCPG), and sarcoma (SARC) (Figs. 2H–2L). There were no significant differences in the DLAT methylation levels of tumors containing CHOL, glioblastoma multiforme (GBM), kidney renal papillary cell carcinoma (KIRP), LUSC, PAAD, STAD, thyroid carcinoma (THCA), thymoma (THYM) and uterine corpus endometrial carcinoma (UCEC) (Figs. S1A–S1I).

Using the cBioPortal tool, we analyzed the pan-cancer genetic alteration data of DLAT. In melanoma, the mutation frequency of DLAT genes is the highest among all cancers, with more than 4% higher than those frequencies of all other cancers. The “deep deletions” were the predominant type of mutation (Fig. S2A). In addition, Fig. S2 shows that the major variation type of DLAT is the missense mutation. The V242Wfs*5/E239G alteration, a truncating mutation in the Biotin_lipoyl domain, presented potential clinical significance in two cases of STAD, one case of UCEC, one case of invasive breast carcinoma (BRCA) and one case of COAD. Then, we investigated the effects of DLAT alterations on the survival rates of patients with different tumors. Compared with patients without DLAT alterations, the ESCA and LGG patients with DLAT alterations exhibited poor OS, progression-free survival (PFS), and disease-specific survival (DSS). In contrast, uterine carcinosarcoma (UCS) patients with DLAT alterations exhibited poor disease-free survival (DFS). Furthermore, the STAD patients with DLAT mutations showed favorable prognosis in PFS ( P = 0.0276), but not in OS, DSS, and DFS (Figs. S2K–S2N).

Significant pan-cancer prognostic potential of DLAT

The effects of DLAT levels on the clinical prognoses of different cancer patients were analyzed with the GEPIA2 website. High DLAT expression levels were observed among the BRCA (Logrank P = 0.038), LGG (Logrank P = 0.006), and LIHC (Logrank P = 0.007) patients and these levels were related to the poor OS of these patients. Meanwhile, those COAD (Logrank P = 0.037), KIRC (Logrank P = 0.000), and READ (Logrank P = 0.029) patients exhibited poorer OS than the patients mentioned above, which was related to their low levels of DLAT (Fig. 3A). In addition, with the Kaplan–Meier Plotter, we found that the patients of breast cancer, liver cancer, lung cancer, ovarian cancer, as well as gastric cancer had a relatively poor prognosis if they showed relatively high DLAT expression levels. Specifically, breast cancer patients with relatively high expressions of DLAT exhibited relatively poor OS, post-progression survival (PPS), relapse-free survival (RFS), and distant metastasis-free survival (DMFS) (Figs. 3B–3C, Figs. S3A–S3B). Meanwhile, liver cancer patients with relatively high expressions of DLAT showed relatively poor OS, PFS, RFS, and DSS (Figs. 3D–3E, Figs. S3D–S3E). Lung cancer patients with relatively high expressions of DLAT exhibited relatively poor OS, PPS, and first progression (FP) (Figs. 3F–3G, Fig. S3C). Furthermore, ovarian cancer patients with relatively high DLAT expressions presented relatively poor OS, PPS, and PFS (Figs. 3H–3I, Fig. S3F), while patients of gastric cancer with relatively high DLAT expressions showed relatively poor OS and PPS (Figs. 3J–3K).

In order to assess the prognostic significance of DLAT to the thirty-three tumor types in the TCGA, univariate and multivariate Cox regression analyses were conducted on the basis of the DLAT gene expressions and all available clinical variables, with OS being used as the dependent variable. The results of the univariate Cox regression analysis showed that DLAT expressions had an adverse effect on the OS of patients with adrenocortical carcinoma (ACC), BLCA, BRCA, HNSC, kidney chromophobe (KICH), LGG, LIHC and PAAD (Hazard ratio>1, P < 0.05). In contrast, DLAT expressions were related to a protective effect on the patients with COAD, KIRC, and READ (Hazard ratio<1, P < 0.05) (Fig. 3L). The variable criterion for the multivariate Cox regression analysis was that the P value of the univariate Cox regression should be less than 0.1. After all clinical variables were adjusted, the multivariate Cox regression analysis delivered the results which showed that the DLAT expressions had significantly adverse effects on patients with ACC (Hazard ratio =4.019; 95% CI [1.630–9.909]) and LIHC (Hazard ratio = 1.398; 95% CI [1.067–1.832]), and a significantly protective effect on patients with KIRC (Hazard ratio = 0.630; 95% CI [0.439–0.902]) (Fig. 3M).

Effects of DLAT expressions on survival and chemotherapeutic resistance of LIHC patients

After verifying the different databases described above repeatedly, we tried to single out the tumor types that were consistent in mRNA and protein expressions of DLAT and exhibited a significant connection between DLAT level and prognosis. Previously, we obtained the result that the gene and protein levels of DLAT were upregulated in LIHC (Figs. 1A–1D). We also figured out that the DLAT expressions had a significant adverse effect on LIHC and that high expressions of DLAT were related to the poor prognoses of LIHC (Figs. 2A, 2D–2E and 2L–2M). Then, other online tools, including the BEST and the UALCAN, were utilized to verify the effect of DLAT expressions on the survival of the LIHC patients on the basis of the TCGA dataset and different GEO datasets. The results showed that relatively high expressions of DLAT were related to the relatively poor prognoses of the LIHC patients in the TCGA (Figs. 4A–4B and 4D) and the GSE76427 dataset (Fig. 4C). Since the multivariate Cox regression analysis used such clinical variables of LIHC as T stage, M stage, pathologic stage, and Tumor status, a prognosis analysis was conducted for sub-types of LIHC accordingly. The LIHC patients at the AJCC T1 and stage1 with relatively high DLAT expressions exhibited relatively poor OS (Figs. 4E and 4I). Furthermore, for patients of LIHC at the AJCC T2 and stage2, relatively high DLAT expressions were related to their relatively poor PFS, RFS, and DSS (Figs. 4F–4H and 4J–4L). The ROC plotter tool was used to analyze the correlation between DLAT gene expressions and the sensitivity of all chemotherapy drugs for LIHC. We found that in the Depmap dataset, the DLAT levels of patients in the non-responding groups to 5-Fluorouracil and Lenvatinib were significantly higher than those in the responding groups (5-Fluorouracil, Fig. 4M; Lenvatinib, Fig. 4O; both P = 0.000). And the predicted AUC values of drug sensitivity towards 5-Fluorouracil and Lenvatinib were 1 (Figs. 4N and 4P).

The role of DLAT in regulating pro-survival autophagy in LIHC

Previous studies have found that DLAT colocalizes with autophagy markers LC3 and SQSTM1 among patients with primary biliary cirrhosis (PBC), and the autophagy induced in bile epithelial cells is related to the over-expression of DLAT (Sasaki et al., 2013; Zhang et al., 2020). Accordingly, we assessed the role of DLAT in mediating autophagy in LIHC cells. We conducted a single gene differential analysis to select the DLAT-related genes with which we performed the GSEA and verified that these DLAT-related genes might directly or indirectly affect autophagy (Fig. 5A). The DLAT broken-down LIHC cell lines (Huh7 and HepG2) were cultured with two siRNAs (Fig. 5B). A colony formation analysis revealed that the breaking down of DLAT significantly inhibited the colony formation of Huh7 and HepG2 cells (Fig. 5C). Furthermore, protein markers such as LC3 I/II and Beclin-1 involved in the autophagosome formation were determined in vitro. Inhibiting DLAT with siRNA other than with siNC in Huh7 and HepG2 cells could decrease the expressions of LC3 I/II and Beclin-1 (Fig. 5D). Consistently, the TEM analysis showed that the breaking down of DLAT in Huh7 and HepG2 cells reduced the number of autophagosomes significantly (P < 0.05; Fig. 5E). In general, all these analysis results showed that the breaking down of DLAT can inhibit the pro-survival autophagy in LIHC cells.

The effects of DLAT on Tregs infiltration in LIHC

Studies have revealed that DLAT plays an immunomodulatory role in some autoimmune diseases. Thus, some studies have analyzed the possibility of DLAT affecting the occurrences and development of tumors by regulating immune infiltration (Wang et al., 2016). Multiple algorithms from the TIMER2.0 website were applied to analyze the roles of DLAT in regulating the immune infiltration of all TCGA tumors. The analysis results showed that there were positive correlations between DLAT and B cells in KICH (spearman’s R: 0.252−0.548, P < 0.05, Fig. 6A) and between DLAT and DC cells in BRCA (spearman’s R: 0.025−0.209, P < 0.05) and PRAD (spearman’s R: 0.148−0.328, P < 0.05) (Fig. 6B). In addition, there was also a positive correlation between DLAT and Tregs in LIHC (spearman’s R: 0.114−0.267, Fig. 6C). However, DLAT was negatively correlated with the infiltration of CD8+ T cell in HPV+ HNSC (spearman’s R: −0.483-(−0.233), Fig. 6D). No significant correlation was observed between DLAT and other immune cells (Figs. S4A–S4E). Furthermore, when analyzing the relationships between immunotherapy responses and DLAT expressions of various tumors, we found that among all the patients in anti-PD-1 / PD-L1 treatment of GSE126044, the patients in the non-responding group showed relatively high DLAT expressions (Fig. 6L), which were related to the relatively poor PFS (Fig. 6N). From the ROC curve, it can be seen that the DLAT-based AUC value of immunotherapy response prediction is 0.873 (Fig. 6M).

Using the Xiantao tool and GEPIA2 tool, we conducted an analysis based on the LIHC dataset in TCGA to analyze the relationships between DLAT and such classical phenotypic markers of Tregs as CD25, CD31, TIM-3, CD83, IL-10, CD62L, CD44, CD28, CXCR4, NRP1, and CD49d (Doebbeler et al., 2018; Peterson, 2012; Sakaguchi et al., 2020; Scheinecker, Goschl & Bonelli, 2020) (Fig. 6E). The analysis results showed significant positive correlations between the Tregs markers of CD31, CD44, NRP1, and CD49d and DLAT (spearman’s R > 0.3, P < 0.05) in LIHC (Figs. 6F–6I). Furthermore, the IHC staining results of tissue micro-array specimens showed that the expression intensities of DLAT and CD49d in LIHC were significantly higher than those in normal specimens (Fig. 6J). In terms of staining intensity, there was a significant positive correlation between the typical Tregs marker of CD49d and DLAT (Pearson’s R = 0.6419, P = 0.000) (Fig. 6K).