NDUFA4L2 is associated with clear cell renal cell carcinoma malignancy and is regulated by ELK1

Gene expression and clinical data in GEO and TCGA database

The microarray data from Gene Expression Omnibus (accession number GSE6344) and mRNA data (RNA Seq v2) from The Cancer Genome Atlas (TCGA) were used. TCGA_KIRC IlluminaHiSeq_RNASeqV2 (Synapse ID: syn2320105) data were obtained from https://www.synapse.org/#!Synapse:syn2320105. The NDUFA4L2, nuclear factor kappa-B1 (NF-κB1), REL proto-oncogene nuclear factor-κB p65 subunit (REL), ELK1, ever shorter telomeres 1 (EST1) and hypoxia inducible factor-1α (HIF1α) mRNA levels were used in this study. Clinical Kidney Renal Clear Cell Carcinoma (TCGA, Nature 2013) data were downloaded from cBioPortal database (http://www.cbioportal.org/study?id=kirc_tcga_pub).

Gene set enrichment analysis

The association between expression level of NDUFA4L2 and biological processes was analyzed using Gene Set Enrichment Analysis (GSEA v2.2, http://www.broad.mit.edu/gsea/). GSEA calculates a gene set Enrichment Score (ES) that estimates whether genes from pre-defined gene set (obtained from the Molecular Signatures Database, MSigDB) are enriched in the NDUFA4L2 high/low expression group or distributed randomly. Thresholds for significance were determined by permutation analysis (1,000 permutations). False Discovery Rate (FDR) was calculated. A gene set is considered significantly enriched when the FDR score is < 0.25 (Subramanian et al., 2005).

Protein-protein interaction (PPI) network construction

Search Tool for the Retrieval of Interacting Genes/Proteins (IntAct; https://www.ebi.ac.uk/intact/?conversationContext=2) is a database of known and predicted protein interactions that may aid the comprehensive description of cellular mechanisms and functions. The PPI network of NDUFA4L2 was constructed using the IntAct database. Experimental result evidences about the interaction among proteins were downloaded from the BioGRID database (https://thebiogrid.org/).

GO and KEGG pathway analyses

To explore the functional annotation enrichment of NDUFA4L2 and its binding partners in ccRCC, Gene Ontology (GO) and unifying functional Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were analyzed in WebGestalt (http://www.webgestalt.org/).

Prediction of TFs associated with binding motifs of NDUFA4L2

Transcription Factors (TFs) bound specifically to their corresponding binding motif (BMo) (Srivastava et al., 2014) and regulated the expression of their target genes. Potential TFs of NDUFA4L2 gene in ccRCC were predicted by Gene Regulation program (P Match 1.0 and Match 1.0; http://gene-regulation.com/pub/programs.html). The JASPAR public database was used to verify the binding sites of these TFs (Mathelier et al., 2016).

Cell culture, ELK1 knockdown and Western blotting

Human ccRCC 786-O (Williams et al., 1978) and ACHN (Kochevar, 1990) cell lines (American Type Culture Collection, ATCC, Manassas, VA, USA) were grown in RPMI 1640 medium and Dulbecco’s modified Eagle’s medium (DMEM) (both from Gibco, Waltham, MA, USA), respectively. These media contain 10% fetal bovine serum (FBS, Hyclone, Logan, UT, USA) and 1% antibiotic-antimycotic agent (Life Technologies, Inc., Grand Island, NY, USA). Cells were grown at 37 °C and 5% CO₂.

Small interfering RNA (siRNA) duplexes pools directed against ELK1 (sc-35290) and control scrambled RNAi (sc-37007) were bought from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Cells were grown to 80% confluency in 35-mm dishes, transfected with 2 µl Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA), and mixed with 30 pmol of the synthetic ELK1 siRNA pool. The cells were then harvested and analyzed after 48 h of transfection.

The primary antibodies specific for ELK1 was bought from Santa Cruz, NDUFA4L2 and GAPDH antibodies were purchased from Abcam (Cambridge, MA, USA).

Statistical analysis

Paired and unpaired samples were analyzed by paired and independent sample t-test, respectively. The relationship between the NDUFA4L2 expression level and clinical stages was analyzed by one-way analysis of variance (ANOVA). Receiver operator characteristic (ROC) curve analysis was applied to detect the optimal cutoff point discriminating ccRCC and normal tissues. The Log rank test for the generated Kaplan–Meier (KM) curve was conducted to evaluate the association between the expression level of NDUFA4L2 and survival rate. The correlations of NDUFA4L2 and transcription factor expression levels were analyzed by Pearson correlation. Statistical analyses were performed using the SPSS 19.0 (SPSS Inc, Chicago, IL, USA) and Graphpad Prism 5 (Graphpad software Inc, San Diego, CA, USA). p < 0.05 was considered to be statistically significant.

NDUFA4L2 is overexpressed and correlates with poor prognosis in ccRCC

To identify differentially expressed genes related with ccRCC occurrence and development, 10 pairs of ccRCC and matched normal samples in GEO database (GSE6344) were analyzed. 1,549 differentially expressed genes were identified (p < 0.05). Expression heatmap of top 20 genes was constructed by R 3.2.0 (R Core Team, 2015) and NDUFA4L2 was the most prominently upregulated gene (Figs. 1A and 1B). Analysis results in TCGA dataset further validated that the expression level of NDUFA4L2 was abnormally upregulated in ccRCC tissues compared with normal tissues and matched adjacent nontumor tissues (Fig. 1C). The expression level of NDUFA4L2 was increased accordingly as the American Joint Committee on Cancer (AJCC) stage progressed (Fig. 1D). The NDUFA4L2 expression level could discriminate between ccRCC and normal tissues with an area under the ROC curve of 0.969 (Fig. 1E) and was correlated with prognosis in ccRCC (Fig. 1F). The KM curve revealed that patients with a high NDUFA4L2 expression level presented lower overall survival probability than those with a low NDUFA4L2 expression level. These results indicated that upregulated NDUFA4L2 level in ccRCC tissues was correlated with the tumorigenesis, development and poor prognosis of ccRCC.

NDUFA4L2 is positively correlated with cell proliferation and anti-apoptosis in ccRCC

To further verify that NDUFA4L2 was related with ccRCC tumorigenesis, the correlation of NDUFA4L2 expression level with cell proliferation and apoptosis were analyzed by GSEA. The ccRCC patients in TCGA dataset were divided into high and low NDUFA4L2 expression groups according to the median value of the NDUFA4L2 level. Gene sets of cell proliferation and negative regulation of apoptotic signaling pathway were highly enriched in the high NDUFA4L2 expression group (Figs. 2A and 2B, FDR < 0.25). These data suggested that a high NDUFA4L2 expression level correlated with enhanced cell proliferation and decreased apoptosis.

NDUFA4L2 is involved in multiple types of cancer-related biological processes and pathways

Proteins exert their functions via interacting proteins and involved pathways. To gain insight into the underlying mechanism by which NDUFA4L2 is involved in ccRCC development, we performed Protein-Protein Interaction (PPI), GO and KEGG pathway analyses to find out the interacting proteins and involved pathways of NDUFA4L2. PPI analysis showed 14 proteins could interact with NDUFA4L2 (Fig. 3A) and these interactions were experimentally verified by affinity capture-mass spectrometry (MS) and two-hybrid assays (https://thebiogrid.org/121231/summary/homo-sapiens/ndufa4l2.html, Fig. 3B). These proteins participated in 13 significant biological processes, of which one was related to death and two were involved in growth, indicating a potential correlation of NDUFA4L2 with ccRCC tumorigenesis and development (Fig. 3C). NDUFA4L2 and its binding partners were involved in multiple types of ccRCC-related signaling pathways, such as the insulin-like growth factor 1 (IGF-1) signaling pathway, the mammalian target of Rapamycin (mTOR) signaling pathway, and the phosphoinositide 3 kinase serine/threonine protein kinase (PI3K/AKT) signaling pathway (Fig. 3D). It was known that IGF-1 pathway played an important role in cell proliferation and apoptosis resistance in RCC (Tracz et al., 2016). Hence, NDUFA4L2 might exert its functions via these binding partners and associated pathways.

The expression levels of NDUFA4L2 and ELK1 are positively correlated in ccRCC tissues and ELK1 regulates the expression of NDUFA4L2 in ccRCC cells

Since the upregulation of NDUFA4L2 expression level played the important role in ccRCC occurrence and development, it is necessary to explore the reasons why the expression level of NDUFA4L2 was increased in ccRCC. In genomic level, gene copy number amplification and gene mutation are one of the reasons for the upregulation of gene expression level. However, bioinformatics analysis results revealed that copy number variation and mutation for NDUFA4L2 gene were not the reasons for NDUFA4L2 level upregulation in ccRCC (Fig. S1). Transcriptional regulation often affects gene expression at transcriptional level and the transcription factors (TFs) play important roles in this process. Hence, we paid attention to the TFs which could regulate the expression of NDUFA4L2. Unexpectedly, the level of reported TF HIF1α was not positively correlated with the level of NDUFA4L2 in the TCGA dataset (Fig. 4A), implying that there were unidentified TFs correlating with the level of NDUFA4L2. NF-κB1, ELK1, ETS1 and REL were predicted TFs of NDUFA4L2 by gene regulation program and were expressed in ccRCC patients. Of them, NF-κB1, ELK1 and ETS1, but not REL expression levels were significantly upregulated in ccRCC (all p < 0.01) compared with normal renal tissues (Figs. 4B–4E). JASPAR database analysis results also revealed these three TFs could bind to the promoter region of NDUFA4L2. To further verify the correlation of the NDUFA4L2 expression level with the levels of these three TFs from a clinical aspect, the Pearson correlation analysis between expression levels of NF-κB/ELK1/ETS1 and NDUFA4L2 was done in ccRCC tissues. Results showed that the ELK1 was the only transcription factor positively correlated with the expression level of NDUFA4L2 (Figs. 5A–5C).

Since the correlation between the expression of NDUFA4L2 and ELK1 is very weak, we further performed the knockdown of ELK1 and detected the levels of NDUFA4L2 in ccRCC cells 786-O and ACHN to elucidate the direct regulatory role of ELK1 on NDUFA4L2 expression. Results showed that with the knockdown of ELK1, the protein level of NDUFA4L2 was downregulated, verifying that the levels of NDUFA4L2 could be regulated by ELK1 expression (Figs. 5D–5E).