ncRNAs-mediated high expression of TIMM8A correlates with poor prognosis and act as an oncogene in breast cancer

The breast cancer clinical samples involved in this experiment were obtained from the Second People’s Hospital of Lianyungang in 2021. This study was approved by the hospital ethics committee, and samples were obtained upon informed consent from the patients.

The TIMM8A mRNA expression data of breast cancer was downloaded from The Cancer Genome Atlas (TCGA) database (https://​tcga-data.​nci.​nih.​gov/​tcga/​). These data were normalized and then analyzed for differential expression using limma R package.

Oncomine database (https://​www.​oncomine.​org/​) is a public online cancer database for RNA and DNA sequences. In our study, firstly, it was used to analyzed the transcriptional levels of TIMM8A in 20 types of human cancer. Then, we compared the mRNA expression of TIMM8A in breast cancer samples with those in normal samples using Students t-test. Statistically significant values and fold change were defined as p-value ≤ 1E-4, and 2, respectively.

GEPIA database (http://​gepia.​cancer-pku.​cn/​) is a web database for differential expression analysis based on TCGA and The Genotype-Tissue Expression (GTEx) data. GEPIA was utilized to discover TIMM8A expression in breast cancer. GEPIA was employed to perform survival analysis for TIMM8A in breast cancer, including overall survival (OS) and relapse-free survival (RFS). p value < 0.05 was considered as statistically significant.

The Kaplan Meier Plotter (http://​kmplot.​com/​analysis/​) provides survival information for patients with breast cancer. We used it to reveal the clinical relationships between miRNAs expression and survival information including RFS and OS. The prognostic value of upstream binding miRNAs of TIMM8A, such as p-values, hazard ratios (HR), and 95% confidence intervals could be automatically calculated on the basis of the RNA expression (high vs. low) of miRNAs.

Upstream binding miRNAs and circRNAs of TIMM8A were predicted by several target gene prediction databases. miRNAs targeting TIMM8A were predicted by starBase (http://​starbase.​sysu.​edu.​cn/​) and TargetScan (http://​www.​targetscan.​org/​vert_​72/​). The Cancer-specific circRNAs database (CSCD) [17] and circbank (http://​www.​circbank.​cn/​index.​html) were utilized to predict target circRNAs. Overlapped circRNAs in the two databases were considered as potential target ciriRNAs of miRNAs.

KEGG pathway enrichment and GO functional enrichment were respectively implemented by the KEGG database (http://​www.​genome.​jp/​kegg/​) and GO database (http://​geneontology.​org). The statistical significance was examined via the hypergeometric test, and p < 0.05 was considered the threshold to identify significantly enriched KEGG pathways and GO terms as reported [18]. GSEA tools (http://​www.​broadinstitute.​org/​gsea) were used to analyze the potential biological processes and pathways regarding to TIMM8A [19]. And the STRING database (http://​string-db.​org, access date 30 July, 2021) was employed to predict protein–protein interactions (PPI) and constructing the PPI network.

Human breast cancer cell lines MCF-7 and MDA-MB-231 were purchased from American Type Culture Collection (ATCC). HEK-293T cell line was preserved by our lab. Cells were cultured in Dulbecco’s modified eagle medium (DMEM) (Gibco, USA) containing 10% fetal bovine serum (Wisent, Canada), 100 U/ml penicillin and 100 mg/ml streptomycin. Cells were cultured and maintained in an incubator containing 5% CO₂ at 37 °C. siRNA (GenePharma, China) was used to knockdown TIMM8A. Cells were grown to 80–90% confluency at the time of transfection. Cells were transfected with siRNA by Lipofectamine^® 3000 reagent (ThermoFisher Scientific, USA) and 72 h after transfection, cells were harvested to extract total RNA and protein. qRT-PCR and western blot were used to detect the transfection efficiency.

Total RNA was extracted from the cells or tissues by using Trizol Reagent (TaKaRa, Japan). cDNA synthesis was performed using the HiScript qRT SuperMix (Vazyme, China). AceQ qPCR SYBR Green Master Mix (Vazyme, China) was used for qRT-PCR in a real-time PCR instrument (Roche, USA) by manufacturer’s instructions. β-actin was used as the endogenous control. The sequences of primers used are as follows: β-actin: forward 5′-GCTGTGCTATCCCTGTACGC-3′ and reverse 5′-TGCCTCAGGGCAGCGGAACC-3′; TIMM8A: forward 5′-CAGCATTTCATCGAGGTAGAGAC-3′ and reverse 5′- AGCCCGACTGTCCAACTTTG-3′; Ki67: forward 5′-GACCTGTTCTTTGAGGCTGAC-3′ and reverse 5′-TCCATCTTCTTCTTTGGGTATTGTT-3′.

The total protein was extracted from cells with RIPA buffer (Beyotime, China) containing phenylmethylsulphonyl fluoride (PMSF), protease inhibitors and phosphatase inhibitors (Beyotime, China). Total protein was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) by a 10% SDS-PAGE gel, and then transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore, USA). The membranes were blocked with 5% skimmed milk powder in Tris buffered saline containing 0.1% Tween 20 (TBST) for 2 h at room temperature and probed with primary antibody against TIMM8A, Bax, Bcl-2 and GAPDH (CST, USA). The membranes were then washed three times with TBST, and incubated with the appropriated secondary antibodies for 2 h. Finally, the bands were examined through Immobilob™ Western Chemiluminescent HRP Substrate (Millipore, USA) to detect the expression levels of target proteins.

Cell proliferation was determined using CCK-8 kit (Dojindo, Japan) according to the instructions of the manufacturer. 3000 cells were suspended in 200 μl medium and seeded in triplicate in a 96-well plate, grown in an incubator containing 5% CO₂ at 37 °C overnight. The original medium was replaced by medium containing 10% CCK8 and then incubated at 37 °C for 2 h. Next, a microplate reader was determined the absorbance at 450 nm.

EdU assay was performed using an EdU assay kit (RiboBio, China). In short, cells were seeded in 96-well plates (1 × 10⁴ cells/well) overnight and transfected with control siRNA and TIMM8A siRNA for 72 h. After incubated with EdU (50 µM) for 2 h, the samples were fixed in 4% formaldehyde for 15 min. Subsequently, cells were permeabilized with 0.3% TritonX-100 (Beyotime, China) for 10 min and then reacted with Apollo reaction mixture for 30 min. Nuclei were stained with DAPI (Beyotime, China) for 15 min and images were taken by a fluorescent microscope (Nikon, Japan).

JC-1 assay was performed to measure mitochondrial membrane potential (ΔΨm) using the JC-1 mitochondrial membrane potential assay kit (KeyGen, China). In short, cells were seeded in 6-well plate and 72 h after siRNA transfection, the detection of ΔΨm was carried out according the guidelines of JC-1 kit. Images were taken by a fluorescent microscope (Nikon, Japan).

Dual-luciferase reporter assay was used to illustrate the targeting binding relationship between miRNA and the mRNA of TIMM8A in HEK-293T cells. In short, wild-type or mutant TIMM8A 3′UTR dual-luciferase reporter plasmid (Gene Pharma, China) and hsa-miR-34c-5p mimics or mimics control (RiboBio, China) were co-transfected into HEK-293T cells using the Lipofectamine 3000 reagent (Invitrogen, USA). After 48 h of transfection, the activities of Firefly and Renilla luciferase were determined according to the manufacturer’s instructions.

The patient-derived tumor fragment platform was constructed as previously described [20]. Briefly, fresh tumor tissue from surgical resections was dissected into fragments of approximately 1 mm³. The fragments were embedded into an artificial extracellular Matrigel matrix (BD Biosciences, USA) and then seeded in 24-well plates. The ex vivo tumor fragment platform was cultured using organoid growth medium (TEMCELL™ Technologies, Canada) and maintained in an incubator containing 5% CO₂ at 37 °C.

The patient-derived xenograft (PDX) model was carried out as previously described [21]. Briefly, we implanted breast cancer samples of 75 mm³ into the fourth mammary fat pad of severely immunocompromised NOD/SCID mice (aged 4 weeks, 18–22 g) and estradiol were added to the mice using patches. Tumor volume was measured every 3 days using a caliper calculated as (length width height)/2. One week after implantation, PBS or TIMM8A inhibitor was intra-tumor injection every 3 days until mice were sacrificed. Euthanasia was performed by cervical dislocation after induction of terminal anesthesia using 5% isoflurane which was in accordance with the American Veterinary Medical Association (AVMA) Guidelines: for the Euthanasia of Animals (2020 Edition). All animal experiments were approved by the Animal Use Committee of the Nanjing Medical University.

The data were presented as mean ± standard error of the mean (SEM) and analyzed using the GraphPad Prism (Version 8.0) and SPSS 20.0 software. The relationship between TIMM8A expression and clinicopathologic features was analyzed by Pearson chi-square test or Fisher exact test. The univariate and multivariate analysis of the effect of each variable on survival was performed using cox proportional hazards regression model. All experiments were least repeated in triplicate, unless otherwise specified and p < 0.05 was considered as a threshold to indicate a statistical significance.

We first analyzed the transcriptional levels of TIMM8A in 20 types of human cancer with those in normal samples using Oncomine databases (Fig. 1A). The mRNA expression levels of TIMM8A were significantly upregulated in patients with breast cancer based on TCGA and GTEx database (Fig. 1B). Richardson’s study found that TIMM8A is highly expressed in breast cancer compared with normal tissues (Fig. 1C) [22]. Moreover, the lobular epithelial (N-Lobular) tissues exhibited significantly lower expression of TIMM8A relative to ductal carcinoma (Fig. 1D) [23]. Furtherly, immunohistochemistry (IHC) from the Human Protein Atlas database confirmed the higher expression of TIMM8A in breast cancer ductal carcinoma cells in protein level (Fig. 1E). As presented in Fig. 1F, we further demonstrated that TIMM8A was expressed higher in breast cancer tissues than normal tissues of patients via detecting TIMM8A mRNA from total of 40 samples. In addition, the gene expression profile of TIMM8A between tumor and normal breast tissues was obtained using the GEPIA database (Fig. 1G). An oncoprint plot using the cBioPortal was generated from the data extracted from the TCGA database to analyze and visualize the somatic mutation of TIMM8A. DNA copy number amplifications, mutations, and deep deletion were the main genetic mutations of breast cancer (Fig. 1H). The percentage of TIMM8A genetic alteration among breast cancer patients is 0.68% (74/10950) (Fig. 1I).

The OS and RFS analysis for TIMM8A in breast cancer determined by the GEPIA database were shown in Fig. 2A, indicating that the higher TIMM8A mRNA expression group has a poorer prognosis than the lower TIMM8A group. By the combination of OS and RFS analysis, TIMM8A may be used as an unfavorable prognostic biomarker in patients with breast cancer. Using logistic regression model, we illustrated the association between TIMM8A and clinicopathological factors in patients with breast cancer from TCGA. As shown in Fig. 2B, TIMM8A expression as a categorical dependent variable (based on a median expression of 25.01) was remarkably associated with stage (I vs II; OR = 1.66, p < 0.01), (I vs III; OR = 1.67, p < 0.01), (I vs IV; OR = 3.62, p < 0.05), T classification (I vs II; OR = 1.59, p < 0.05), (I vs IV; OR = 2.34, p < 0.05) and metastasis (negative vs positive; OR = 2.73, p < 0.05). To identify the prognostic value of TIMM8A in clinical application, we performed the univariate and multivariate cox regression analyses. For TCGA-BRCA data, the TIMM8A in univariate analysis was correlated with OS (OR = 1.15, 95% CI 1.05–1.26, p < 0.01) (Fig. 2C). Multivariate analysis showed that the TIMM8A was an independent prognostic indicator (OR = 1.16, 95% CI 1.06 1.27, p < 0.01) (Fig. 2D).

We performed GSEA bioinformatics to analyze the associated biological processes and signaling pathways based on the data acquired from TCGA dataset. Top five upregulated and downregulated pathways were shown in Fig. 3A: including CELL_CYCLE; CYSTEINE_AND_METHIONINE_METABOLISM; DILATED_CARDIOMYOPATHY; ECM_RECEPTOR_INTERACTION; FOCAL_ADHESION; HOMOLOGOUS_RECOMBINATION; OTHER_GLYCAN_DEGRADATION; PYRIMIDINE_METABOLISM; RNA_DEGRADATION; VASOPRESSIN_REGULATED_WATER_REABSORPTION. Additional file 1: Tables S1, S2 show all significantly up- or down-regulated pathways. Thirty outstanding proteins related to TIMM8A were identified by PPI analysis of STRING software (Fig. 3B), such as SLC25A12, TIMM13, COX17, TOMM22, CHCHD4. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (Fig. 3C) and Gene Ontology (GO) annotation (Fig. 3D–F) were performed to predict the functions of TIMM8A and the genes dramatically associated with TIMM8A alterations. Metabolic pathways, Oxidative phosphorylation, Calcium signaling pathway and cGMP-PKG signaling pathway are enriched with the most TIMM8 related genes. The complete results of GO and KEGG analyses are shown in Additional file 1: Table S3.

A total of 96 likely target miRNAs of TIMM8A were predicted by starBase (Additional file 1: Table S4). According to the negative correlation principle of miRNA and mRNA of TIMM8A, two candidate miRNAs, hsa-miR-34c-5p and hsa-miR-449a, were selected (Fig. 4A). The correlation between low expression of hsa-miR-34c-5p and hsa-miR-449a and poor prognosis in breast cancer patients was shown in the Kaplan–Meier Plotter database (Fig. 4B). The miRNA target prediction software Targetscan revealed that there was a same binding site (marked with a black line) between hsa-miR-34c-5p and hsa-miR-449a and TIMM8A (Fig. 4C). Furthermore, dual-luciferase reporter gene assay was used to ascertain whether TIMM8A was the direct target gene of hsa-miR-34c-5p and hsa-miR-449a. The results revealed that the relative luciferase activity of TIMM8A-WT was obviously decreased by hsa-miR-34c-5p WT, whereas no similar reduction was observed in the luciferase activity of hsa-miR-34c-5p Mut (Fig. 4D). Moreover, the present study screened the circRNAs that may interact with miRNAs using circBank. 192 candidate circRNAs for hsa-miR-34c-5p and 182 candidate circRNAs for hsa-miR-449a were shown in Additional file 1: Tables S5, S6. Based on conserved sequence, length and binding sites, 24 and 22 potential circRNAs that may interact with hsa-miR-34c-5p and hsa-miR-449a were screened, respectively. Of these, hsa-circ-0107314/hsa-circ-0021867/hsa-circ-0122013 were selected by analyzing their cancer specific property and cytosolic location through Cancer Specific CircRNA Database (Fig. 4E). Taking together, hsa-circ-0107314/hsa-circ-0021867/hsa-circ-0122013 might be the three most potential upstream circRNAs of hsa-miR-34c-5p/hsa-miR-449a-TIMM8A axis in breast cancer (Fig. 4F).

We selected different types of breast cancer cell lines (MDA-MB-231 and MCF7) as representatives to explore the function of TIMM8A. By CCK8 assay, we found that inhibition of TIMM8A inhibited proliferation of both MDA-MB-231 and MCF7 cells (Fig. 5A). We further verified the effect of TIMM8A on the proliferation of breast cancer cells using EdU assay. The results showed that in both MDA-MB-231 and MCF7 cell lines, the proliferating cells was reduced after the expression of TIMM8A was inhibited, indicating that TIMM8A has a promoting effect on cell proliferation. (Fig. 5B) Next, we used JC-1 reagent to explore the effect of TIMM8A on mitochondrial membrane potential. We found that after the expression of TIMM8A was inhibited, the JC-1 dimers were reduced, indicating that TIMM promotes proliferation by affecting mitochondrial function (Fig. 5C, D). In order to evaluate the effect of TIMM8A on apoptosis of breast cancer cells, we detected the levels of apoptosis-related proteins in different treatment groups by western blot, and the results showed that after the expression of TIMM8A was inhibited, the pro-apoptotic protein Bax was significantly increased, while the anti-apoptotic protein Bcl-2 was decreased, indicating TIMM8A’s ability to inhibit apoptosis (Fig. 5E). Using the patient-derived tumor fragment platform, we assessed the expression of Ki67 and found that Ki67 expression in human breast cancer fragments decreased after incubation with TIMM8A inhibitors for one week (Fig. 5F). These results indicate that silencing TIMM8A can inhibit the proliferation of breast cancer in vitro. Similarly, silencing TIMM8A inhibits tumor growth in the breast cancer PDX mouse model. The tumor growth rate and tumor weight in TIMM8A inhibited group were significantly lower than those in control group (Fig. 6A, B). We further verified the expression levels of TIMM8A mRNA in tumors (Fig. 6C), and detected apoptosis and proliferation related indicators, such as Caspase 3, proliferating cell nuclear antigen (PCNA), and found that the results were consistent with in vitro cell experiments (Fig. 6D). In addition, IHC results showed that Ki67 levels in tumors decreased and TUNEL levels in tumors increased after TIMM8A inhibitors were administered (Fig. 6E), which further indicates that TIMM8A inhibition can suppress the proliferation and growth of breast cancer.