Exploration of a prognostic signature for mitochondria-related genes and the therapeutic prospects of vorinostat in clear cell renal cell carcinoma

Introduction

Renal cell carcinoma (RCC) constitutes a small proportion (2–3%) of adult malignant tumors worldwide (1,2). The predominant subtype, clear cell RCC (ccRCC), accounts for a substantial majority (70–80%) of cases (3). Although localized ccRCC generally carries a favorable prognosis, the overall survival rate for metastatic ccRCC (mccRCC) remains dismally low (less than 10%) (4). Despite ongoing research, the precise interplay between environmental factors, genetics, and epigenetic alterations in ccRCC development remains elusive. Furthermore, the current therapeutic options for ccRCC remain limited (5,6).

Historically, the significance of mitochondrial metabolism in meeting the metabolic demands of rapidly dividing cancer cells has been disregarded (7,8). Otto Warburg’s findings in the 1920s revealed that tumor slices exhibited glucose consumption and excessive lactate production independent of oxygen availability (9). Consequently, Warburg postulated that glycolysis served as the principal metabolic pathway essential for tumor proliferation, with impaired respiration being a prerequisite for malignant transformation (7). Multiple clinical trials have been performed to estimate the effectiveness of inhibiting mitochondrial metabolism as an innovative approach to cancer treatment, thereby questioning the belief that metabolism is not essential for tumor proliferation (10). Chen et al. have highlighted the clinical relevance of inner mitochondrial membrane proteins (IMMT) in predicting prognosis, understanding the immune microenvironment of tumors, and advancing precision medicine for renal cancers (11). Additionally, Bai et al. have explored the involvement of cuproptosis-correlated long non-coding RNA (lncRNA) in the ccRCC prognosis and immunotherapy (12). Thus, there is a close relationship between changes in mitochondrial function and morphology and the onset and advancement of tumors.

Through the utilization of The Cancer Genome Atlas (TCGA) ccRCC database, we conducted an analysis to ascertain the connection between mitochondria-related genes (MRGs) and pertinent clinical data. Employing machine learning techniques and bioinformatics tools, we successfully established a prognostic signature for ccRCC comprising 16 genes. By evaluating the risk value of this prognostic signature, we were able to predict the probability of high or low patient survival rates. Furthermore, we subsequently identified vorinostat, a drug with potential prognostic implications for high-risk patients, by leveraging the Connectivity Map (CMap) database and L1000CDS² dataset. By studying MRGs, we learned about their potential prognostic value, immune therapy strategies, and ccRCC outcomes. Simultaneously, we present this article in accordance with the TRIPOD reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-24-565/rc).

Methods

Materials

Vorinostat (GC17390) and Cell Counting Kit-8 (CCK-8) (GK10001) were purchased from GLPBIO (Shanghai, China). The lactate dehydrogenase (LDH) release detection kit (C0016) was sourced from Beyotime Biotechnology (Shanghai, China). MitoTracker™ Green (M7514), DAPI (62248), and Live/Dead Cell Viability Assays Kit (L3224) were procured from Thermo Fisher (Shanghai, China). Transwell chambers (3401) and Matrigel Matrix (354248) were obtained from Corning (Shanghai, China). The 786-O and Caki-1 cell lines, along with related cell culture reagents, were acquired from Shanghai Zhong Qiao Xin Zhou Biotechnology Co., Ltd (Shanghai, China).

Data acquisition and preprocessing

Data on RNA-HTSeq (RNA High-throughput sequence) and clinical information about kidney renal clear cell carcinoma (KIRC) patients were obtained from The Cancer Genome Atlas (TCGA) (data cutoff date was August 22, 2021). Following the exclusion of samples with incomplete data, a total of 539 ccRCC tumor samples and 72 normal kidney samples were retained for subsequent analysis. R Studio was used to process data to ensure the protection of human subjects and access to data.

Development of the prognostic gene signature

Genes significantly correlated with overall survival were identified by subjecting each MRG to univariate Cox proportional hazards regression analysis as part of the TCGA training dataset (13-15). Least absolute shrinkage and selection operator (LASSO) Cox regression was then applied to these genes (15,16). This led to the creation of a signature that incorporates genes from mitochondria. In order to determine risk scores, genes were chosen based on their coefficients. A prognostic risk score was computed for each patient employing the following formula: risk score = expression level of gene 1 × j1 + expression level of gene 2 × j2 + expression level of gene x × jx, where j represents the Coef (i). On the basis of the median risk score, the cut-off point for stratifying TCGA ccRCC patients into high-risk group (HRG) and low-risk group (LRG) was chosen. The specimens were separated into two datasets, namely the risk train and risk test datasets in order to estimate the prognostic signature of the MRGs. Subsequently, Cox proportional hazards regression analyses, univariate and multivariate, were conducted in all datasets. Prognostic performance was estimated employing the receiver operating characteristics (ROC) curve analysis, which evaluated both sensitivity and specificity. Discrimination was determined by area under the curve (AUC) values. The specimens were separated into two datasets, namely the risk train and risk test datasets in order to estimate the prognostic signature of the MRGs. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Determination of differentially expressed genes (DEGs), functional analysis, and tumor immune cell infiltration

The “edgeR” R package was employed to differentiate high- and low-risk DEGs (17). To comprehend the biological implications of molecular discoveries acquired through high-throughput methods, the examination of gene functionality is of utmost importance. DEG functional profiles, encompassing Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses, were scrutinized and depicted deploying the cluster profiler package in R (18-20). The “estimate” R package was utilized to evaluate the Immune Score and Stromal Score. Tumor-infiltrating immune cells were analyzed through the CIBERSORT database (21).

Potential drug exploration

Using the “limma” R package, we identified the up-regulated and down-regulated messenger RNA (mRNAs) between the HRG and LRG. These up-regulated and down-regulated mRNAs were then submitted to the CMap (https://clue.io/) (22) and the library of integrated network-based cellular signatures (LINCS) L1000 dataset (https://maayanlab.cloud/L1000CDS2/#/index) (23). Subsequently, we obtained potential intervention drugs that could reverse the expression of these mRNAs.

CCK-8 assay

The CCK-8 assay was employed to estimate cell viability and proliferation. Cells were cultured and seeded in a 96-well plate. Following cell adhesion, cells were co-cultured with various concentrations of vorinostat. After 24 and 48 h of incubation at 37 ℃ with CCK-8 solution, absorbance was measured at 490 nm. Data were normalized to controls and statistically analyzed.

LDH release assay

The LDH release assay evaluates the extent of damage to cell membrane integrity. Firstly, cells are cultured in a 96-well plate and subsequently subjected to predetermined treatments. Post-treatment, the culture medium is collected for LDH release assessment. The LDH assay is performed using a commercially available LDH detection kit following the manufacturer’s instructions, and absorbance values are measured at 490 nm.

Wound healing assay

The wound healing assay is a technique widely used to study cell migration and tissue repair. It involves creating a controlled gap in a confluent cell monolayer and then monitoring the migration of cells to close the wound. After seeding cells and inducing the wound with pipette tips (200 µL), time-lapse images are captured at specific intervals of 0 and 24 h for analysis. The width of the wound is measured using Image-Pro Plus, and the percentage of closure is calculated.

Transwell assay

The transwell chamber was coated with Matrigel Matrix twenty-four h before. After that, 100 µL of a single-cell suspension containing 3×10⁴ cells was introduced to the chamber with serum-free media, and 600 µL of medium encompassing 10% fetal bovine serum was placed in the bottom compartment. Remove the transwell chamber after 48 h of incubation. Fix with methanol for 15 min. Stain with 0.1% crystal violet for 20 min. Gently wipe to remove the top layer of cells on the microporous membrane. Phosphate buffer saline (PBS) was used twice for rinsing. We used Image-Pro Plus to tally the cells that had translocated through the microporous membrane after photographing the bottom layer of cells on the membrane under a microscope.

Statistical analysis

We conducted all statistical analyses and generated the graphics using R software 4.0.2 and GraphPad Prism 8. When comparing two groups statistically, T-tests were employed. One-way analysis of variance (one-way ANOVA) was employed when comparing more than two groups.

Results

Identification of survival-associated MRGs

The analysis workflow of this investigation is visually manifested in Figure 1. The ccRCC RNA-HTSeq dataset comprises 539 tumors and 72 normal kidney specimens. Clinical data and expression matrix files encompassing mRNA and lncRNA expression profiles were sourced from the TCGA database. A total of 1,136 gene names associated with human MRGs were acquired from MitoCarta3.0 (https://personal.broadinstitute.org/scalvo/MitoCarta3.0/human.mitocarta3.0.html). Subsequently, the expression data of genes associated with mitochondria were extracted from each clinical sample and integrated with the corresponding clinical information (futime, fustat, age, gender, stage, T, M, N) for the purpose of conducting weighted gene co-expression network analysis (WGCNA). Figure 2A,2B illustrate the application of WGCNA to analyze the data of tumor specimen’s gene expression, resulting in the identification of five co-expression modules: blue, green, turquoise, yellow, and grey modules, comprising 411, 97, 458, 136, and 31 genes, respectively. Among these, blue, green, and yellow modules exhibited a significant connection with patients’ futime and fustat (P<0.05). Consequently, we proceeded to analyze a total of 644 genes from these three modules further.

Figure 1 The flowchart delineates the procedure employed in generating MRGs prognostic signatures for ccRCC. ccRCC, clear cell renal cell carcinoma; KIRC, kidney renal clear cell carcinoma; MRGs, mitochondria-related genes; OS, overall survival; WGCNA, weighted gene co-expression network analysis.

Figure 2 Screening of KIRC data and development of a clinical prognostic signatures based on MRGs. (A) Visualization of gene dendrogram and module colors. (B) Visualization of the correlation between the module eigenvectors of KIRC and clinical features using WGCNA analysis. (C) Utilizing Cox’s proportional hazard signatures, a prognostic classifier was created through lasso regression analysis, incorporating the gene expression levels of all 16 genes in KIRC tissues. Expression of these genes in (D) the overall risk database, (E) the risk train database, and (F) the risk test database. KIRC, kidney renal clear cell carcinoma; LncRNA, long non-coding RNA; MRGs, mitochondria-related genes; ME, module eigengene; mRNA, messenger RNA; WGCNA, weighted gene co-expression network analysis.

By establishing a prognostic correlation threshold of P<0.001, we have successfully estimated a comprehensive set of 256 genes that exhibit a strong association with the prognosis of ccRCC patients within the three aforementioned modules (Table S1). Furthermore, employing a correlation coefficient of 0.8 in conjunction with the aforementioned 256 genes and adopting a prognostic correlation threshold of P=0.001, we have successfully identified eight lncRNAs that demonstrate a significant correlation with MRGs, thereby exerting a substantial influence on the prognosis of ccRCC (Table S2).

Creation of the MRGs prognostic signature

Subsequently, we employed LASSO regression analysis to ascertain the ccRCC MRGs prognostic signature, which consisted of 16 MRGs, comprising 1 lncRNA (AC008870.2) and 15 mRNAs (LYRM7, THNSL1, FKBP10, PDK4, ACOT13, SLC25A23, ALDH3A2, GLYAT, CRYZ, FAHD2A, SARS2, PISD, FAM136A, RECQL4, and ETFBKMT) (Figure 2C). The risk score for each specimen was computed employing the following formula: risk score = AC008870.2 expression * 0.10089292 + LYRM7 expression * (−0.039037334) + THNSL1 expression * (−0.037233383) + FKBP10 expression * 0.089483579 + PDK4 expression * (−0.061487668) + ACOT13 expression * (−0.042435961) + SLC25A23 expression * (−0.030000909) + ALDH3A2 expression * (−0.000327202) + GLYAT expression * (−0.021485389) + CRYZ expression * (−0.014767018) + FAHD2A expression * 0.038904691 + SARS2 expression * 0.127702113 + PISD expression * 0.043581241+ FAM136A expression * 0.149183099 + RECQL4 expression * 0.229143844 + ETFBKMT expression * (−0.035058912). In the future, a small size of tumor tissue samples can be extracted to detect these genes, and the risk value can be calculated to predict the high and low risk of patients, so as to intervene early. Furthermore, patient samples were categorized according to their risk scores in descending order, with the top 50% being categorized as high-risk and the bottom 50% as low-risk. In order to estimate the prognostic signature of the MRGs, the specimens were separated into two datasets, namely the risk train and risk test datasets, with an equal ratio of 1:1. Initially, we examined the expression differences of these 16 MRGs between the HRG and LRG. Figure 2D-2F reveal statistically significant changes in the 16 MRG expression levels between the HRG and LRG. Specifically, the HRG exhibited upregulation in AC008870.2, FKBP10, FAHD2A, SARS2, PISD, FAM136A, and RECQL4, whereas LYRM7, THNSL1, PDK4, ACOT13, SLC25A23, ALDH3A2, GLYAT, CRYZ, and ETFBKMT were downregulated compared to the LRG.

Moreover, we conducted a comparative analysis of the clinical data across different datasets. According to the risk-all dataset, the HRG had a significantly greater likelihood of presenting with grades 3–4, stage III–IV, T3–4, and M1 stages than the LRG (Table S3). In addition, high-risk participants also displayed higher grades, stages, and T grades than low-risk participants in both the risk train and test datasets (Tables S4,S5). In other words, patients in the HRG tend to have a more advanced TNM staging.

In the subsequent phase, we proceeded to validate the disparities in survival rates observed between the HRG and LRG. Concurrently, across the risk all, train, and test datasets, it was evident that patients belonging to the HRG exhibited significantly lower survival rates than those in the LRG (Figure 3A-3C). Furthermore, the AUC values for 1-year survival were calculated to be 0.768, 0.801, and 0.747 in all three datasets (Figure 3D-3F). Additionally, the AUC values for 5-year and 10-year survival surpassed the threshold of 0.75 (Figure S1). The scatter plots of survival time and status corresponding to different sample risk scores were also observed. Figure 3G shows that high-risk patients had higher risk scores, but their survival rates were lower. Risk training and test data also showed this trend (Figure 3H,3I).

Figure 3 Validation of the prognosis signature and survival outcomes of various risk groups. Kaplan-Meier survival curves for various risk groups in the risk all database (A), the risk train database (B), and the risk test database (C). ROC curve analysis of the MRGs prognostic signature in the risk all database (D), risk train database (E), and risk test database (F) for 1 year. The distributions of risk scores, survival time, and patient statuses are depicted in the risk all database (G), risk train database (H), and risk test database (I), the dotted lines demarcate the optimal threshold distinguishing between low- and high-risk groups. (J-M) Decision curve analysis curves for the MRGs prognostic signature at 1, 3, 5, and 10 years. AUC, area under curve; MRGs, mitochondria-related genes; ROC, receiver operating characteristic.

Furthermore, we assessed the prognostic signature using decision curve analysis (DCA). All signature curves for 1, 3, 5, and 10 years were above the “None” and “All” reference lines, with the 10-year DCA curve being particularly significant (Figure 3J-3M).

The findings from the Cox analysis revealed that within the risk all dataset, both univariate and multivariate Cox analyses establish a significant correlation between age, grade, and stage with patient prognosis. Furthermore, the patient’s risk score exhibits an independent ability to anticipate the prognosis of ccRCC patients, surpassing the implications of age, grade, and stage. This pattern is similarly observed in the risk train and test datasets (Figure 4).

Figure 4 Conducting univariate Cox analysis on clinical parameters and risk scores in risk all database (A), risk train database (C), and risk test database (E). Executing multivariate Cox analysis on clinical parameters and risk scores in risk all database (B), risk train database (D), and risk test database (F).

The impact of MRGs prognostic signature on immune infiltration

The immune microenvironment is strongly associated with the development and prognosis of tumors (24,25). The stromal cell scores did not exhibit a significant disparity between the HRG and LRG (Figure 5A). However, the immune cell scores of the HRG were notably greater compared to those of the LRG (Figure 5B). The ESTIMATE score, which encompasses both stromal and immune cell scores, was also greater in the HRG than in the LRG, as depicted in Figure 5C. In the HRG, immune cells were found to be more abundant than in the LRG, while stromal cell components were similar. For patients with ccRCC, stromal cell and immune cell scores did not show a significant difference in survival time (Figure 5D-5F).

Figure 5 Investigating the variance in immune infiltration within the tumor microenvironment between HRG and LRG. Evaluating the disparity in the stromal score (A), immune score (B) and ESTIMATE score (C) between HRG and LRG. Analyzing the influence of high and low stromal scores (D), immune scores (E), and ESTIMATE score (F) on the survival likelihood of ccRCC patients. (G) Illustrating the expression levels of immune cells through Vioplot between HRG and LRG. Assessing the influence of high and low levels of resting dendritic cells (H), resting mast cells (I), T cells follicular helper (J), and T cells regulatory (Tregs) (K) on the survival probability of ccRCC patients. ccRCC, clear cell renal cell carcinoma; ESTIMATE, Estimation of Stromal and Immune cells in MAlignant Tumour tissues using Expression data; HRG, high-risk group; LRG, low-risk group.

Moreover, we employed CIBERSORT to ascertain the change in immune cell levels between LRG and HRG. According to Figure 5G, the HRG exhibited decreased levels of B cell memory, T cells CD4 memory resting, T cells gamma delta, monocytes, macrophages M1, macrophages M2, and activated dendritic cells. Conversely, the HRG demonstrated significantly elevated levels of Plasma cells, T cells follicular helper, T cells regulatory (Tregs), and Macrophages M0.

A study was conducted to examine the survival time of ccRCC patients by evaluating the levels of various immune cells. The findings revealed that patients with elevated levels of activated Dendritic cells and resting Mast cells in their ccRCC tissue samples had notably higher survival rates compared to those with lower levels. Conversely, patients with lower levels of T cells follicular helper and Tregs exhibited significantly higher survival rates than those with higher levels (Figure 5H-5K).

Differential gene enrichment and hub genes between HRG and LRG in the MRGs prognostic signature

On the basis of the aforementioned findings, it is apparent that there are notable disparities in patient survival and immune infiltration between the HRG and LRG. Consequently, we proceeded to examine the distinct genes between these groups. By employing a cut-off of log fold change (FC) ≥2 and P<0.05, typically, 563 differential genes were identified. Then, we conducted GO and KEGG enrichment analyses on these genes.

The differential genes were manifested to mainly contribute to immune system processes, according to the GO enrichment analysis. These processes include the following: the classical pathway, immunoglobulin-mediated immune response, lymphocyte-mediated immunity, complement activation, adaptive immune response according to somatic recombination of immune receptors constructed from immunoglobulin superfamily domains, humoral immune response mediated by circulating immunoglobulin, as well as humoral immune response mediated by lymphocytes (Figure 6A). The KEGG enrichment analysis manifested that the differential genes were mainly linked to pathways like complement activation, lymphocyte-mediated immunity, the adaptive immune response through somatic recombination of immune receptors built from immunoglobulin superfamily domains, and the humoral immune response (Figure 6B). It concludes that changes in immune function are significantly associated with mitochondrial function changes in ccRCC tissue samples.

Figure 6 Analysis of gene enrichment pathways and regulatory networks associated with DEGs between HRG and LRG. (A) The ten most highly enriched pathways in GO across biological processes, cellular components, and molecular functions. (B) The thirty most enriched pathways in the KEGG. (C) Network diagram illustrating the regulatory relationships among differentially expressed genes between high- and low-risk groups. (D) The top ten hub genes were identified through Cytoscape analysis. BP, biological processes; CC, cellular components; DEGs, differential expressed genes; GO, Gene Ontology; HRG, high-risk group; KEGG, Kyoto Encyclopedia of Genes and Genomes; LRG, low-risk group; MF, molecular functions.

Subsequently, a gene interaction network was created employing STRING to analyze the differential genes (Figure 6C). Subsequently, hub genes were identified through Cytoscape, with the top 10 genes being CDC20, AURKB, UBE2C, CDCA8, CCNB2, BIRC5, KIF2C, KIF20A, DLGAP5, and RRM2 (Figure 6D). It was observed that nine of these genes significantly impacted the survival time of ccRCC patients. Specifically, patients with great expression levels of CDC20, AURKB, UBE2C, CDCA8, CCNB2, BIRC5, KIF2C, KIF20A, and RRM2 exhibited lower survival rates within the same time intervals (Figure S2).

Exploration of potential drugs for treating high-risk patients

The study utilizes the CMap to exploit cellular responses to perturbations, aiming to uncover connections between diseases, genes, and treatments. CMap was employed to screen drugs targeting the reversal of up-regulated and downregulated mRNAs in both HRG and LRG (15,26,27). Initially, the differential genes were ranked in ascending order based on their corrected P values (FDR, false discovery rate). Subsequently, the top 100, 200, and 300 differential genes were individually analyzed to identify potential drugs capable of reversing these mRNAs. From this analysis, the top 20 drugs were extracted (Figure 7A and Figure S3).

Figure 7 Conducting a selection of potential reversing drugs targeting DEGs between HRG and LRG. (A) The outcome of possible drugs targeting the reversal of the top 200 DEGs between HRG and LRG utilizing Connectivity Map. (B) The outcome of potential drugs targeting the reversal of all DEGs between HRG and LRG using. (C) Venn diagram displaying drugs identified by both Connectivity Map and L1000CDS² databases. (D,E) Depicting the 2D and 3D structures of vorinostat, a potential reversing drug. 2D, two-dimensional; 3D, three-dimensional; DEGs, differential expressed genes; HRG, high-risk group; LRG, low-risk group.

The existing collection of LINCS L1000 dataset comprises over one million gene expression profiles of chemically perturbed human cells (28). Through the utilization of the L1000CDS² database, we have successfully identified 50 potential drugs capable of reversing these differential genes, as depicted in Figure 7B. By intersecting the results obtained from two distinct databases, we have determined vorinostat as a potential therapeutic drug, as illustrated in Figure 7C. The chemical structures of vorinostat, both in 2D and 3D representations, are presented in Figure 7D and Figure 7E, respectively.

Effects of vorinostat on ccRCC cell death, proliferation, migration, and invasion

Two human renal clear cell carcinoma cell lines were chosen for validation in order to examine the implications of vorinostat on ccRCC. Initially, the renal clear cell carcinoma cells, 786-O and Caki-1, were cultured and subjected to vorinostat treatment for 48 h, after which mitochondrial morphology was observed. The findings from Figure 8A,8B demonstrated that in the control group, the mitochondria within the cells were predominantly punctate or elongated. However, upon exposure to vorinostat, the mitochondria in 786-O cells underwent a transformation into continuous filamentous structures.

Figure 8 Effect of vorinostat on 786-O and Caki-1 Cells. (A,B) The impact on mitochondrial morphology in 786-O cells and Caki-1 was assessed after 48 h of treatment with 5 µM vorinostat. The image labeled “Amplified” in the third column presents an enlarged view of the area highlighted in yellow within the second column figure. (C,D) Assessment of the impact of vorinostat at different concentrations on the viability of 786-O cells after 24 and 48 h using the CCK-8 assay. P vs. 0 µM. (E,F) Analysis of the influence of vorinostat at different concentrations on the viability of Caki-1 cells after 24 and 48 h using the CCK-8 assay. P vs. 0 µM. (G,H) Evaluation of the effect of vorinostat at different concentrations on LDH release in 786-O cells after 24 and 48 h. P vs. 0 µM. (I,J) Examination of the effect of vorinostat at different concentrations on LDH release in Caki-1 cells after 24 and 48 h by LDH release assay. P vs. 0 µM. (K) Impact of vorinostat at different concentrations on the induction of cell death in 786-O after 24 and 48 h, along with corresponding statistical data. Cell counts were obtained from eight different regions. (L,M) Influence of 0.5 and 1 µM vorinostat on the migratory capacity of 786-O cells after 24 and 48 h by wound healing assay, along with corresponding statistical data. Cell counts were obtained from eight different regions. (N,O) Impact of 0.5 and 1 µM vorinostat on the invasive capacity of 786-O cells after 48 h by transwell assay, along with corresponding statistical data (The quantity of cells stained with crystal violet within a random region was enumerated). Cell counts were obtained from eight different regions. Calcein AM, calcein acetoxymethyl ester; CCK-8, cell counting kit-8; EthD-1, ethidium homodimer-1; LDH, lactate dehydrogenase.

In addition, we conducted an assessment to determine the impact of various doses of vorinostat on the viability of ccRCC cells, specifically 786-O and Caki-1. The results obtained from the CCK-8 assay indicated that, following a 24-hour treatment period, concentrations of 1, 2, 5, 10, 20, and 50 µM of vorinostat significantly impeded the viability of 786-O cells (Figure 8C). Similarly, after a 48-hour treatment period, concentrations of 2, 5, 10, 20, and 50 µM of vorinostat exhibited notable inhibitory effects on the viability of 786-O cells (Figure 8D). This observed trend was also consistent in the experiments involving Caki-1 cells (Figure 8E,8F). However, the results of LDH release indicated that vorinostat, at concentrations ranging from 0 to 20 µM, did not have an impact on the release of LDH from either 786-O or Caki-1 cells within a 24-hour period, nevertheless, it was observed that vorinostat, at a concentration of 2 µM or higher, could induce an increased release of LDH in 786-O and Caki-1 cells following 48 h (Figure 8G-8J). These findings suggest that vorinostat primarily inhibits renal clear cell carcinoma cell proliferation for a short duration and induces cell death over a longer duration.

These findings were further confirmed through live/dead cell viability assays. Given the similar impact of vorinostat on both cell lines, subsequent experiments were conducted using 786-O cells. As depicted in Figure 8K, the exposure of 786-O cells to vorinostat at concentrations ranging from 0-10 µM for a duration of 24 h did not result in an escalation of dead cells labeled with EthD-1 dye. Conversely, the number of viable cells labeled with Calcein AM dye exhibited a gradual decline with increasing vorinostat dosage. Following a 48-hour treatment of 786-O cells with vorinostat concentrations ranging from 0 to 10 µM, there was a gradual decrease in the population of viable cells labeled with Calcein AM dye, accompanied by an increase in the population of deceased cells labeled with EthD-1 dye, both of which were dose-dependent.

In addition, we performed Wound Healing and Transwell assays to assess the impact of vorinostat on the migratory and invasive properties of 786-O cells. The findings depicted in Figure 8L,8M indicate a significant reduction in the migratory capacity of 786-O cells following treatment with 0.5 and 1 µM of vorinostat for 24 h and 48 h. Moreover, Figure 8N,8O demonstrate that a concentration of 1 µM of vorinostat effectively inhibits the invasion of 786-O cells over a 48-hour period.

To summarize, the administration of vorinostat has been found to impede the proliferation, migration, and invasion of ccRCC cells while also promoting cell death when administered at high dosages or for extended periods. However, it is noteworthy that within a specific dosage range (up to 20 µM), vorinostat does not elicit apoptosis or cell death in ccRCC cells. It is important to mention that this study did not explore the effects of higher dosages or longer treatment durations of vorinostat.

Discussion

RCC is essentially a metabolic disease characterized by a reprogramming of energetic metabolism (29-32). In particular, the metabolic flux through glycolysis is partitioned (33-35), and mitochondrial bioenergetics and OxPhox are impaired, as well as lipid metabolism (33,36-39). We propose to ascertain the substantial contribution of mitochondria in the advancement of ccRCC in this study. Through an extensive bioinformatics analysis, a prognostic signature associated with mitochondrial function was identified, along with potential therapeutic strategies, including drug predictions. Furthermore, the investigation encompassed data pertaining to immune cell infiltration and the immune microenvironment. Ultimately, experimental validation substantiated the inhibitory impacts of vorinostat, a promising therapeutic agent, on the proliferation, migration, and invasion of ccRCC cells (Figure 9).

Figure 9 Summary roadmap of the study. ccRCC, clear cell renal cell carcinoma; MRGs, mitochondria-related genes.

The potential link between the Warburg effect and changes in mitochondrial function and morphology in tumor tissues has been observed (9). In cancer cells, the tricarboxylic acid (TCA) cycle and mitochondrial oxidative phosphorylation become decoupled, leading to the redirection of glycolysis-derived pyruvate towards lactate fermentation rather than mitochondrial oxidative metabolism. This metabolic phenomenon, known as the Warburg effect, is believed to have an involvement in the alterations observed in mitochondrial form and function within tumors (40). Additionally, scholarly investigations indicate that the inhibition of the Warburg effect in ccRCC can effectively impede tumor formation. For example, the deacetylation of PDHA1 at K351 by SIRT5 enhances the activity of pyruvate dehydrogenase complex (PDC), consequently modifying the metabolic interplay with the TCA cycle, suppressing the Warburg effect and ultimately restraining tumor proliferation (41). Thus, it is plausible that alterations in genes associated with mitochondria may have an essential involvement in initiating and developing kidney cancer. These modifications in mitochondria-associated genes hold significant implications for the prognosis and therapeutic approaches for ccRCC patients.

Multiple studies have documented the development of prognostic signatures for ccRCC through the utilization of machine learning and bioinformatics methodologies, underscoring their significant implications for diagnosis, prognosis, and treatment. Notably, Zuo et al. successfully established a novel prognostic biomarker that is applicable to all three subtypes of RCC. This biomarker comprises a 6-lncRNA signature, namely AC003092.1, AC079160.1, COL18A1-AS1, LINC00520, LINC02154, and SLC7A11-AS1. The integration of this signature into prognostic models has been shown to enhance the accuracy of predicting overall survival (42). Zhiyong Cai and colleagues have developed cuproptosis-associated alteration patterns, which serve a dual purpose of predicting immune cell infiltration in the tumor microenvironment (TME) and assessing an individual’s responsiveness to immune checkpoint blockers (ICBs). These patterns offer valuable guidance for precise treatment and accurate prognostic prediction, specifically for KIRC patients (43). Yang et al. performed bioinformatics analyses utilizing online databases to assess the disparities in expression, survival implications, immune infiltration, and prognostic importance of iron-sulfur protein genes linked to multiple mitochondrial dysfunction syndrome (MMDS) in KIRC patients. The authors suggest that NFU1, ISCA1, ISCA2, and C1ORF69 exhibit promising potential as therapeutic targets for KIRC, thereby potentially enhancing survival rates and prognostic precision through the utilization of these prognostic markers (44).

This study employed machine learning and bioinformatics techniques to develop a prognostic signature for ccRCC patients. The signature consisted of 16 MRGs, including one lncRNA (AC008870.2) and 15 mRNAs (LYRM7, THNSL1, FKBP10, PDK4, ACOT13, SLC25A23, ALDH3A2, GLYAT, CRYZ, FAHD2A, SARS2, PISD, FAM136A, RECQL4, and ETFBKMT). Furthermore, alterations in these genes were found to impact the survival duration of ccRCC patients. The prognostic signature facilitates the computation of a risk score for patients with ccRCC, thereby facilitating their categorization into HRG and LRG for the purpose of predicting survival. Clinical data analysis from the TCGA database revealed that patients classified as high-risk manifested lower rates of survival contrasted with those classified as low-risk. Consequently, it is worth considering the possibility of conducting biopsies to obtain a small tumor sample from ccRCC patients in order to analyze the MRG expression within this signature and subsequently compute the risk score for the purpose of evaluating the patient’s prognosis in the foreseeable future. This has the potential to facilitate enhanced individualized and rationalized pharmacotherapy, thereby exemplifying a prospective utilization of the prognostic signature within the realm of clinical practice.

Damaged mitochondria were selectively enclosed within autophagosomes and subsequently merged with lysosomes, thereby accomplishing the degradation of mitochondria and preserving cellular homeostasis. The phenomenon known as mitochondrial autophagy is distinguished by tumors with impaired autophagy, which exhibit abnormal mitochondrial structures akin to benign tumors (45). Otto Warburg’s research revealed that the deficiency in mitochondrial respiration under aerobic conditions may serve as the primary catalyst for cancer (46). It is now recognized that genetic events responsible for abnormal proliferation in cancer cells can also impact biochemical metabolism, such as promoting aerobic glycolysis (47), although mitochondrial function is typically unaffected by these alterations (48-50). The mitochondria have an essential involvement in different cellular processes, including energy production, cell proliferation, redox homeostasis, signal transduction, immune response regulation, and programmed cell death. Notably, the biogenesis and maintenance of mitochondria are frequently enhanced in cancer (51). Certain types of cancer exhibit mutations in nuclear DNA-encoded enzymes involved in the TCA cycle within the mitochondria, involving in the generation of metabolites with carcinogenic properties. Conversely, pathogenic mutations in the mitochondrial genome are typically eliminated through negative selection mechanisms (52,53). The exclusion of mitochondrial DNA (mtDNA) has been found to restrict the development of tumors (54), and the most of human tumors harboring mutated mitochondrial genomes tend to be benign (55). Consequently, mitochondria is essential and multifaceted involvement in progressing cancerous growth.

In this study, an analysis of the differential genes between HRG and LRG led to the identification of vorinostat as a potential drug capable of reversing the differential gene expression. Interestingly, the experimental findings revealed that vorinostat induced elongation of the mitochondria in renal clear cell carcinoma cells. Furthermore, it was observed that vorinostat effectively inhibited the proliferation, migration, and invasion of 786-O and Caki-1 cells, thereby demonstrating promising anti-ccRCC activity. The histone deacetylase (HDAC) inhibitor vorinostat effectively hinders the elimination of acetyl groups from histones, resulting in histone acetylation, which subsequently induces apoptosis and halts the cell cycle. The inhibitory impact of HDAC on malignant cells, particularly malignant T-cells, is notably significant (56). In a study involving refractory cutaneous T-cell lymphoma (CTCL) patients, vorinostat monotherapy demonstrated an overall response rate of 30%, with certain responders achieving prolonged remission and qualifying for stem cell transplantation (57). Vorinostat has also been assessed in B-cell lymphomas and diverse solid tumors, although the outcomes have been less remarkable (58-60). In 2006, vorinostat became the first HDAC inhibitor to receive approval as an anticancer agent in the United States, specifically as a monotherapy for refractory or relapsed CTCL. In the context of renal physiology, research indicates that vorinostat has the potential to modulate the expression levels of epidermal growth factor (EGF), thereby enhancing kidney enlargement in instances related to diabetes (61). Furthermore, prolonged administration of the HDAC inhibitor vorinostat has been manifested to alleviate renal damage in experimental diabetic subjects through a mechanism reliant on endothelial nitric oxide synthase (62). Nevertheless, there is a dearth of literature specifically investigating the effects of vorinostat on RCC (63), which corroborates our experimental findings. Therefore, vorinostat holds significant potential in treating high-risk renal clear cell carcinoma patients.

In relation to the tumor immune microenvironment, high-risk ccRCC tumors exhibit a notably elevated level of immune cell infiltration compared to low-risk tumors. The HRG demonstrates an increased expression of T cells follicular helper and Tregs, which is linked to diminished survival rates. This finding is consistent with one of the underlying rationales for the potential efficacy of vorinostat in treating CTCL patients. Moreover, within the HRG, the expression levels of activated Dendritic cells and resting Mast cells are comparatively lower than those observed in the LRG, suggesting a less favorable prognosis. This observation aligns with previous studies (64,65). Additionally, it is plausible that variations in the expression of other immune cells exist; however, their influence on the survival outcomes of ccRCC patients appears to be relatively insignificant.

This experiment undoubtedly possesses certain limitations. The validation of the impact of vorinostat on animal signatures was not pursued further. Furthermore, the prognostic signature encompasses a set of 16 genes, which may be deemed relatively extensive in terms of quantity. Nevertheless, in its entirety, we have successfully developed a prognostic signature utilizing bioinformatics methodologies, focusing on genes associated with mitochondria. The signature possesses the capability to function as an autonomous prognostic indicator for patient survival risk. Additionally, we have successfully discerned a prospective therapeutic agent, vorinostat, by means of differential gene analysis conducted on HRG and LRG. Furthermore, we have substantiated its efficacy at the cellular level. Consequently, we have not only established a resilient prognostic signature but also acquired a potential remedy for high-risk ccRCC. This investigation carries substantial implications for the prognosis and treatment of clinical ccRCC instances.

Conclusions

Utilizing a combination of integrated bioinformatics and machine learning methodologies, we have successfully devised a prognostic signature consisting of 16 genes associated with mitochondria. This signature effectively stratifies patients with ccRCC into HRG and LRG, thereby facilitating accurate prognostication of survival outcomes. Notably, individuals belonging to the HRG exhibit diminished rates of survival, thereby emphasizing the substantial prognostic value of this particular factor. Moreover, our investigation has shed light on potential therapeutic avenues for high-risk ccRCC patients, with vorinostat emerging as a promising treatment agent that demonstrates noteworthy inhibitory effects. This research provides crucial insights for understanding ccRCC pathogenesis and devising personalized treatment approaches.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tau.amegroups.com/article/view/10.21037/tau-24-565/rc

Data Sharing Statement: Available at https://tau.amegroups.com/article/view/10.21037/tau-24-565/dss

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-24-565/prf

Funding: This study was supported by the Jiaxing Public Welfare Science and Technology Program Fund (No. 2023AY40023), Zhejiang Provincial Natural Science Foundation of China (Grant No. LTGD24H050001), National Natural Science Foundation of China (Grant No. 82000230 to J.G. and No. 82104197 to J.J.), and the Qin Shen Scholar Program of Jiaxing University (No. CD70623040).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-24-565/coif). J.J. and J.G. report that this study was supported by the National Natural Science Foundation of China (Grant No. 82104197 to J.J. and No. 82000230 to J.G.). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Ljungberg B, Albiges L, Abu-Ghanem Y, et al. European Association of Urology Guidelines on Renal Cell Carcinoma: The 2019 Update. Eur Urol 2019;75:799-810. [Crossref] [PubMed]
2. Young M, Jackson-Spence F, Beltran L, et al. Renal cell carcinoma. Lancet 2024;404:476-91. [Crossref] [PubMed]
3. Nerich V, Hugues M, Paillard MJ, et al. Clinical impact of targeted therapies in patients with metastatic clear-cell renal cell carcinoma. Onco Targets Ther 2014;7:365-74. [Crossref] [PubMed]
4. Simonaggio A, Epaillard N, Pobel C, et al. Tumor Microenvironment Features as Predictive Biomarkers of Response to Immune Checkpoint Inhibitors (ICI) in Metastatic Clear Cell Renal Cell Carcinoma (mccRCC). Cancers (Basel) 2021;13:231. [Crossref] [PubMed]
5. Schmidt LS, Linehan WM. Genetic predisposition to kidney cancer. Semin Oncol 2016;43:566-74. [Crossref] [PubMed]
6. Liao L, Testa JR, Yang H. The roles of chromatin-remodelers and epigenetic modifiers in kidney cancer. Cancer Genet 2015;208:206-14. [Crossref] [PubMed]
7. Warburg O. On the origin of cancer cells. Science 1956;123:309-14. [Crossref] [PubMed]
8. Kenny TC, Birsoy K. Mitochondria and Cancer. Cold Spring Harb Perspect Med 2024;14:a041534. [Crossref] [PubMed]
9. Koppenol WH, Bounds PL, Dang CV. Otto Warburg’s contributions to current concepts of cancer metabolism. Nat Rev Cancer 2011;11:325-37. [Crossref] [PubMed]
10. Vasan K, Werner M, Chandel NS. Mitochondrial Metabolism as a Target for Cancer Therapy. Cell Metab 2020;32:341-52. [Crossref] [PubMed]
11. Chen CC, Chu PY, Lin HY. Supervised Learning and Multi-Omics Integration Reveals Clinical Significance of Inner Membrane Mitochondrial Protein (IMMT) in Prognostic Prediction, Tumor Immune Microenvironment and Precision Medicine for Kidney Renal Clear Cell Carcinoma. Int J Mol Sci 2023;24:8807. [Crossref] [PubMed]
12. Bai Z, Lu J, Chen A, et al. Identification and Validation of Cuproptosis-Related LncRNA Signatures in the Prognosis and Immunotherapy of Clear Cell Renal Cell Carcinoma Using Machine Learning. Biomolecules 2022;12:1890. [Crossref] [PubMed]
13. Nagashima K, Sato Y. Information criteria for Firth’s penalized partial likelihood approach in Cox regression models. Stat Med 2017;36:3422-36. [Crossref] [PubMed]
14. Linden A, Yarnold PR. Modeling time-to-event (survival) data using classification tree analysis. J Eval Clin Pract 2017;23:1299-308. [Crossref] [PubMed]
15. Xiaoqin Z, Zhouqi L, Huan P, et al. Development of a prognostic signature for immune-associated genes in bladder cancer and exploring potential drug findings. Int Urol Nephrol 2024;56:483-97. [Crossref] [PubMed]
16. Zheng H, Wu R, Zhang G, et al. Nomograms for prognosis prediction in esophageal adenocarcinoma: realities and challenges. Clin Transl Oncol 2025;27:449-57. [Crossref] [PubMed]
17. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43:e47. [Crossref] [PubMed]
18. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000;25:25-9. [Crossref] [PubMed]
19. Kanehisa M, Furumichi M, Tanabe M, et al. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res 2017;45:D353-61. [Crossref] [PubMed]
20. Yu G, Wang LG, Han Y, et al. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 2012;16:284-7. [Crossref] [PubMed]
21. Chen B, Khodadoust MS, Liu CL, et al. Profiling Tumor Infiltrating Immune Cells with CIBERSORT. Methods Mol Biol 2018;1711:243-59. [Crossref] [PubMed]
22. Subramanian A, Narayan R, Corsello SM, et al. A Next Generation Connectivity Map: L1000 Platform and the First 1,000,000 Profiles. Cell 2017;171:1437-1452.e17. [Crossref] [PubMed]
23. Wang Z, Clark NR, Ma’ayan A. Drug-induced adverse events prediction with the LINCS L1000 data. Bioinformatics 2016;32:2338-45. [Crossref] [PubMed]
24. Gajewski TF, Schreiber H, Fu YX. Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol 2013;14:1014-22. [Crossref] [PubMed]
25. Baharom F, Ramirez-Valdez RA, Khalilnezhad A, et al. Systemic vaccination induces CD8(+) T cells and remodels the tumor microenvironment. Cell 2022;185:4317-4332.e15. [Crossref] [PubMed]
26. Blasco A, Natoli T, Endres MG, et al. Improving deconvolution methods in biology through open innovation competitions: an application to the connectivity map. Bioinformatics 2021;37:2889-95. [Crossref] [PubMed]
27. Fan LY, Yang J, Liu RY, et al. Integrating single-nucleus sequence profiling to reveal the transcriptional dynamics of Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis. J Transl Med 2023;21:649. [Crossref] [PubMed]
28. Duan Q, Reid SP, Clark NR, et al. L1000CDS(2): LINCS L1000 characteristic direction signatures search engine. NPJ Syst Biol Appl 2016;2:16015. [Crossref] [PubMed]
29. di Meo NA, Lasorsa F, Rutigliano M, et al. The dark side of lipid metabolism in prostate and renal carcinoma: novel insights into molecular diagnostic and biomarker discovery. Expert Rev Mol Diagn 2023;23:297-313. [Crossref] [PubMed]
30. Lucarelli G, Loizzo D, Franzin R, et al. Metabolomic insights into pathophysiological mechanisms and biomarker discovery in clear cell renal cell carcinoma. Expert Rev Mol Diagn 2019;19:397-407. [Crossref] [PubMed]
31. di Meo NA, Lasorsa F, Rutigliano M, et al. Renal Cell Carcinoma as a Metabolic Disease: An Update on Main Pathways, Potential Biomarkers, and Therapeutic Targets. Int J Mol Sci 2022;23:14360. [Crossref] [PubMed]
32. De Marco S, Torsello B, Minutiello E, et al. The cross-talk between Abl2 tyrosine kinase and TGFβ1 signalling modulates the invasion of clear cell Renal Cell Carcinoma cells. FEBS Lett 2023;597:1098-113. [Crossref] [PubMed]
33. Bianchi C, Meregalli C, Bombelli S, et al. The glucose and lipid metabolism reprogramming is grade-dependent in clear cell renal cell carcinoma primary cultures and is targetable to modulate cell viability and proliferation. Oncotarget 2017;8:113502-15. [Crossref] [PubMed]
34. Ragone R, Sallustio F, Piccinonna S, et al. Renal Cell Carcinoma: A Study through NMR-Based Metabolomics Combined with Transcriptomics. Diseases 2016;4:7. [Crossref] [PubMed]
35. Lucarelli G, Galleggiante V, Rutigliano M, et al. Metabolomic profile of glycolysis and the pentose phosphate pathway identifies the central role of glucose-6-phosphate dehydrogenase in clear cell-renal cell carcinoma. Oncotarget 2015;6:13371-86. [Crossref] [PubMed]
36. Lucarelli G, Rutigliano M, Sallustio F, et al. Integrated multi-omics characterization reveals a distinctive metabolic signature and the role of NDUFA4L2 in promoting angiogenesis, chemoresistance, and mitochondrial dysfunction in clear cell renal cell carcinoma. Aging (Albany NY) 2018;10:3957-85. [Crossref] [PubMed]
37. Bombelli S, Torsello B, De Marco S, et al. 36-kDa Annexin A3 Isoform Negatively Modulates Lipid Storage in Clear Cell Renal Cell Carcinoma Cells. Am J Pathol 2020;190:2317-26. [Crossref] [PubMed]
38. Lucarelli G, Rutigliano M, Loizzo D, et al. MUC1 Tissue Expression and Its Soluble Form CA15-3 Identify a Clear Cell Renal Cell Carcinoma with Distinct Metabolic Profile and Poor Clinical Outcome. Int J Mol Sci 2022;23:13968. [Crossref] [PubMed]
39. Milella M, Rutigliano M, Lasorsa F, et al. The Role of MUC1 in Renal Cell Carcinoma. Biomolecules 2024;14:315. [Crossref] [PubMed]
40. Lu J, Tan M, Cai Q. The Warburg effect in tumor progression: mitochondrial oxidative metabolism as an anti-metastasis mechanism. Cancer Lett 2015;356:156-64. [Crossref] [PubMed]
41. Yihan L, Xiaojing W, Ao L, et al. SIRT5 functions as a tumor suppressor in renal cell carcinoma by reversing the Warburg effect. J Transl Med 2021;19:521. [Crossref] [PubMed]
42. Zuo S, Wang L, Wen Y, et al. Identification of a universal 6-lncRNA prognostic signature for three pathologic subtypes of renal cell carcinoma. J Cell Biochem 2019;120:7375-85. [Crossref] [PubMed]
43. Cai Z, He Y, Yu Z, et al. Cuproptosis-related modification patterns depict the tumor microenvironment, precision immunotherapy, and prognosis of kidney renal clear cell carcinoma. Front Immunol 2022;13:933241. [Crossref] [PubMed]
44. Yang L, Chen YX, Li YY, et al. Systematic analysis of expression profiles and prognostic significance for MMDS-related iron-sulfur proteins in renal clear cell carcinoma. Sci Rep 2022;12:19637. [Crossref] [PubMed]
45. Ferro F, Servais S, Besson P, et al. Autophagy and mitophagy in cancer metabolic remodelling. Semin Cell Dev Biol 2020;98:129-38. [Crossref] [PubMed]
46. Warburg O. On respiratory impairment in cancer cells. Science 1956;124:269-70.
47. Vander Heiden MG, Cantley LC, Thompson CB. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 2009;324:1029-33. [Crossref] [PubMed]
48. Zu XL, Guppy M. Cancer metabolism: facts, fantasy, and fiction. Biochem Biophys Res Commun 2004;313:459-65. [Crossref] [PubMed]
49. Fan J, Kamphorst JJ, Mathew R, et al. Glutamine-driven oxidative phosphorylation is a major ATP source in transformed mammalian cells in both normoxia and hypoxia. Mol Syst Biol 2013;9:712. [Crossref] [PubMed]
50. Gaglio D, Metallo CM, Gameiro PA, et al. Oncogenic K-Ras decouples glucose and glutamine metabolism to support cancer cell growth. Mol Syst Biol 2011;7:523. [Crossref] [PubMed]
51. Hasumi H, Baba M, Hasumi Y, et al. Regulation of mitochondrial oxidative metabolism by tumor suppressor FLCN. J Natl Cancer Inst 2012;104:1750-64. [Crossref] [PubMed]
52. Adam J, Hatipoglu E, O’Flaherty L, et al. Renal cyst formation in Fh1-deficient mice is independent of the Hif/Phd pathway: roles for fumarate in KEAP1 succination and Nrf2 signaling. Cancer Cell 2011;20:524-37. [Crossref] [PubMed]
53. Sullivan LB, Martinez-Garcia E, Nguyen H, et al. The proto-oncometabolite fumarate binds glutathione to amplify ROS-dependent signaling. Mol Cell 2013;51:236-48. [Crossref] [PubMed]
54. Tan AS, Baty JW, Dong LF, et al. Mitochondrial genome acquisition restores respiratory function and tumorigenic potential of cancer cells without mitochondrial DNA. Cell Metab 2015;21:81-94. [Crossref] [PubMed]
55. Lang M, Vocke CD, Merino MJ, et al. Mitochondrial DNA mutations distinguish bilateral multifocal renal oncocytomas from familial Birt-Hogg-Dubé tumors. Mod Pathol 2015;28:1458-69. [Crossref] [PubMed]
56. Olsen EA, Kim YH, Kuzel TM, et al. Phase IIb multicenter trial of vorinostat in patients with persistent, progressive, or treatment refractory cutaneous T-cell lymphoma. J Clin Oncol 2007;25:3109-15. [Crossref] [PubMed]
57. Duvic M, Talpur R, Ni X, et al. Phase 2 trial of oral vorinostat (suberoylanilide hydroxamic acid, SAHA) for refractory cutaneous T-cell lymphoma (CTCL). Blood 2007;109:31-9. [Crossref] [PubMed]
58. Ogura M, Ando K, Suzuki T, et al. A multicentre phase II study of vorinostat in patients with relapsed or refractory indolent B-cell non-Hodgkin lymphoma and mantle cell lymphoma. Br J Haematol 2014;165:768-76. [Crossref] [PubMed]
59. Chalasani N, Bonkovsky HL, Fontana R, et al. Features and Outcomes of 899 Patients With Drug-Induced Liver Injury: The DILIN Prospective Study. Gastroenterology 2015;148:1340-52.e7. [Crossref] [PubMed]
60. Krug LM, Kindler HL, Calvert H, et al. Vorinostat in patients with advanced malignant pleural mesothelioma who have progressed on previous chemotherapy (VANTAGE-014): a phase 3, double-blind, randomised, placebo-controlled trial. Lancet Oncol 2015;16:447-56. [Crossref] [PubMed]
61. Gilbert RE, Huang Q, Thai K, et al. Histone deacetylase inhibition attenuates diabetes-associated kidney growth: potential role for epigenetic modification of the epidermal growth factor receptor. Kidney Int 2011;79:1312-21. [Crossref] [PubMed]
62. Advani A, Huang Q, Thai K, et al. Long-term administration of the histone deacetylase inhibitor vorinostat attenuates renal injury in experimental diabetes through an endothelial nitric oxide synthase-dependent mechanism. Am J Pathol 2011;178:2205-14. [Crossref] [PubMed]
63. Lee JS, Yoon YG, Yoo SH, et al. Histone deacetylase inhibitors induce mitochondrial elongation. J Cell Physiol 2012;227:2856-69. [Crossref] [PubMed]
64. Wculek SK, Cueto FJ, Mujal AM, et al. Dendritic cells in cancer immunology and immunotherapy. Nat Rev Immunol 2020;20:7-24. [Crossref] [PubMed]
65. Tamminga M, Hiltermann TJN, Schuuring E, et al. Immune microenvironment composition in non-small cell lung cancer and its association with survival. Clin Transl Immunology 2020;9:e1142. [Crossref] [PubMed]