Dysregulated lncRNAs were found in triple-negative breast cancer
Two microarray datasets (GSE60689 and GSE112848) were analyzed according to the flow chart as shown in Figure 1. 3496 upregulated and 2677 downregulated lncRNAs were detected in breast cancer tissues comparing with normal breast tissues (BC groups). 723 upregulated and 753 downregulated lncRNAs were found in triple-negative breast cancer tissues comparing with normal breast tissues (TNBC groups). The results were shown as volcano plots according to the criteria of |log fold change (FC)| ≥ 1.0 and p value ≤ 0.05 (Figure 2A). In addition, 162 downregulated lncRNAs and 170 upregulated lncRNAs were investigated to be overlapped lncRNAs in both BC groups and TNBC groups. The overlapped lncRNAs were shown as the Venn diagram (Figure 1 and 2B). Among them, the top differentially expressed lncRNAs including 4 downregulated lncRNAs and 17 upregulated lncRNAs were screened out and demonstrated as heatmaps with the criteria of |logFC| ≥ 3.0 and p value ≤ 0.05 (Figure 1 and 2C). The 21 lncRNAs were applied for further analysis. As shown in Figure 1, after the annotation and verification of candidate 21 lncRNAs, three upregulated lncRNAs (LINC02303, LINC01351 and LINC01511) and two downregulated lncRNAs (CHL1-AS2 and LINC01612) were obtained from the strict verification via NONCODE and BLAST online software. The preliminary data of survival analysis by TANRIC database showed that only the long intergenic non-protein coding RNA 1351 (LINC01351) was related with the clinical feature of patients with breast cancer. Consequently, LINC01351 was further detected in the following analysis about the potential roles on breast carcinogenesis.
LINC01351 was an oncogenic indicator in breast cancer
To date, a large number of lncRNAs have been discovered through functional genomic studies, however, the function and mechanism of LINC01351 has not been well investigated yet. So, the function and regulation of LINC01351 were explored in the study. The expression of LINC01351 was tested by using the data from GEO datasets (GSE60689 and GSE112848). All the data was processed to be two groups including BC groups (BC vs. normal breast) and TNBC groups (TNBC vs. normal breast). As shown in Figure 3A-a/b, LINC01351 was highly expressed in both BC tissues or TNBC tissues compared with normal breast tissues (BC vs. normal, p=0.007; TNBC vs. normal, p=0.034). To further test the regulation of expression, the data from GDC TCGA breast cancer (BRCA) database was induced to be analyzed. 49 cases of normal breast samples, 595 cases of breast cancer samples, and 8 cases of TNBC samples were included in the analysis. Similarly, LINC01351 was found to be overexpressed in TNBC and breast cancer tissues comparing to normal breast tissues (Figure 3A-c, p<0.001).
It is well known that there are four main molecular subtypes of breast cancer that are based on the gene expression including estrogen-receptor (ER), progesterone-receptor (PR), human epidermal growth factor receptor 2 (Her2) and Ki-67. Then breast cancer is classified as luminal A, luminal B, triple-negative breast cancer and Her2-enriched breast cancer. Hormone-receptor negative (ER and PR negative) and Her2 positive are considered to be poor prognostic markers. So, the correlation between the expression of LINC01351 and molecular pathological features of breast cancer were further detected via TCGA and TANRIC database. As shown in Figure 3B, increased expression of LINC01351 was discovered in ER-negative, PR-negative and Her2-positive breast cancer tissues (ER-negative vs. ER-positive, p=3.21e-4; PR-negative vs. PR-positive, p=9.30e-8; Her2-positive vs. Her2-negative, p=3.83e-8), which suggested that LINC01351 contributed to the malignant phenotypes of breast cancer. In the GDC TCGA BRCA data, 18 cases of luminal A, 24 cases of luminal B, 10 cases of Her2-enriched and 15 cases of TNBC samples were induced into the analysis. Similarly, the expression of LINC01351 was significantly higher in Her2-enriched or triple-negative breast cancer tissues than in luminal subtypes of breast cancer tissues (TNBC vs. luminal A, p=0.012; TNBC vs. luminal B, p=0.045; Figure 3C). There was no significant difference in the expression of LINC01351 between Her2-enriched and triple-negative breast cancer (TNBC vs. Her2-positive, p=0.073).
Moreover, the expression of LINC01351 was progressively increasing with the TNM stages of breast cancer (p=0.007, Figure 3D). Generally, TNBC is an aggressive subtype of breast cancer with early metastasis. Then the expression of LINC01351 was further detected in both metastatic tumor samples and primary tumor samples by exploiting the GDC TCGA BRCA data. The higher expressed LINC01351 was found in the metastatic breast cancer tissues than in the primary breast cancer tissues (Figure 3E, p=0.003).
All these results indicated that highly expressed LINC01351 contributed to the aggressive subtypes, advanced TNM stages, and metastasis of breast cancer. It was the first time to report that LINC01351 played as an oncogenic and poor molecular pathological indicator of breast cancer.
LINC01351 was a poor prognostic marker in triple-negative breast cancer
To understand the potential role of LINC01351 on the survival of patients with breast cancer, the survival data was further analyzed via the TCGA and TANRIC database. The primary breast cancer tissues were divided equally into two groups based on the expression of LINC01351 (low-expression and high-expression, n=372, separately). In the study, all the duration of following-up was 10 years. As demonstrated in Figure 4A, the result of survival analysis showed that LINC01351 played no effect on the overall survival (OS) of breast cancer (p=0.3265). For the relapse-free survival (RFS), 588 cases of breast cancer with survival data were introduced into the analysis. Low-expressed group (n=294) and high-expressed group (n=294) were also divided by the expression of LINC01351. LINC01351 showed no effect on the RFS of breast cancer (p=0.06, Figure 4B).
Then for the analysis about TNBC, the higher expression of LINC01351 was found to be significantly related with poorer prognosis of patients. 83 cases of TNBC were divided into low-expressed group (n=42) and high-expressed group (n=41). Higher expressed LINC01351 was associated with poor OS of patients with TNBC (p=0.047, Figure 4C). Similarly, LINC01351 was significantly related to poor RFS of patients with TNBC from the analysis about 30 cases of low-expressed TNBC and 29 cases of high-expressed TNBC (p=0.029, Figure 4D).
All the above data indicated that highly expressed LINC01351 was associated with poor prognosis of patients with TNBC. LINC01351 was an unfavorable prognostic marker in TNBC.
Integrative analysis of lncRNA, mRNA and microRNA signature
Long non-coding RNA presents interaction domains for DNA, mRNA, microRNA (miRNA, miR), and protein, depending on both sequence and secondary structure [7]. Emerging evidence indicates that lncRNA could be ceRNA for miRNAs in cancer [12]. Thus, we explored the potential regulation of LINC01351 to mRNAs and miRNAs. The network of regulation was inquired and further visualized by Cytoscape software. Among them, 6 miRNAs (miR-504, miR-548n, miR-567, miR-618, miR-718, and miR-1280) and 4 mRNAs (ELK4, ALPI, SCEMOL and RAD23B) were identified to be targets of LINC01351. Based on the analysis from TANRIC database, LINC01351 downregulated ELK4, SCEMOL and RAD23B and upregulated ALPI (all p<0.001). For the miRNAs, miR-504 was downregulated by LINC01351, while miR-548n, miR-567, miR-618, miR-718, and miR-1280 were upregulated by LINC01351 (all p<0.01). The positive and negative regulation between LINC01351 and target mRNAs or miRNAs were all described in Figure 5.
It is well known that miRNAs are small non-coding RNAs that usually inhibit gene expression through partially complementary elements in the 3′ UTR of their target mRNAs [13]. Furthermore, the downstream target genes of six miRNAs (miR-504, miR-548n, miR-567, miR-618, miR-718, and miR-1280) were obtained from TargetScan and miRTarBase database. All the target mRNAs were predicted to be downregulated by their miRNAs. Then the overlapped target mRNAs of both TargetScan and miRTarBase databases were also shown by Cytoscape software in Figure 5. The above result demonstrated that LINC01351 was involved in the complex regulation network of mRNAs and miRNAs.
ELK4 was down-regulated by LINC01351 and associated with transcriptional regulation in cancer
To further understand the effect of LINC01351 during carcinogenesis, functional enrichment and GO annotations of both overlapped genes and target genes of LINC01351 were analyzed by Metascape database. It was found that gene signature was enriched in the transcriptional misregulation in cancer (p<0.001, Figure 6A). In the enriched functional group ETS transcription factor ELK4, also known as SAP1, was chosen for the following analysis, because it was identified as the direct target negatively regulated by LINC01351 (p=0.00034). ELK4 is a member of the ETS family of transcription factors and of the ternary complex factor (TCF) subfamily.
To clarify the effect of ELK4 and the inhibition role regulated by LINC01351, the expression of ELK4 was tested via immunohistochemistry (IHC) in breast cancer tissues compared with normal breast tissues. As shown in Figure 6B, the results of IHC showed that ELK4 was high-expressed in normal breast tissue and weak-expressed in breast cancer tissue, which was exactly opposite to the expression of LINC01351 in breast cancer compared with normal breast tissue. Moreover, the correlation between the expression of ELK4 and LINC01351 was investigated by analyzing the TCGA and TANRIC database. As shown in Figure 6C, the expression of ELK4 showed negative correlation with the expression of LINC01351 (p=0.019). All the results demonstrated that LINC01351 down-regulated the expression of ELK4 in breast cancer.
LINC01351/ELK4 axis predicted prognosis and sensitivity of therapy in triple-negative breast cancer
In the study, to explore the regulation of LINC01351/ELK4 axis on TNBC, the survival analysis was performed by using Kaplan–Meier Plotter database. The high expression of ELK4 was associated with the favorable prognosis of patients with TNBC (p=0.0035, Figure 6D). TNBC is defined by lack of expression of ER, PR and Her2 and is still a big challenge for anticancer therapy. To identify the best treatment strategy, 6-subtype-classification of TNBC was reported by Lehmann in 2011 [14]. For each subtype, available therapy was predicted and recommended in the study. Based on the classification, TNBC was classified as basal-like 1 (BL1), basal-like 2 (BL2), immunomodulatory (IM), mesenchymal (M), mesenchymal stem-like (MSL), and luminal androgen receptor (LAR) subtype. Furthermore, the relationship between the expression of ELK4 and different subtypes of TNBC (BL1, BL2, IM, M, MSL and LAR) was detected in the following analysis. The survival results demonstrated that higher expression of ELK4 was significantly related to better overall survival in basal-like 1 and LAR subtypes of TNBC after 10-year-following up (all p<0.05, Figure 6D). The survival data was exactly opposite to the survival data of LINC01351 in TNBC, which suggested that LINC01351 negatively regulated the expression of ELK4 and affected its roles on the overall survival of patients with BL1 and LAR subtypes of TNBC. Besides, LINC01351/ELK4 axis could be a potential prognostic marker of TNBC.
Based on the subtypes of Lehmann classification, BL1 and BL2 subtypes preferentially respond to the therapy regulating the cell cycle and DNA damage. IM subtype is associated with immunotherapy. M and MSL subtypes are related to epithelial-mesenchymal transition, growth factor pathways and PI3K/mTOR pathway. The LAR subtype is characterized by androgen receptor (AR) signaling and sensitive to AR antagonist [14]. Then, the sensitivity to anticancer therapy could be easily predicted by analyzing the expression of LINC01351/ELK4 axis. The TNBC tumor with low-expressed LINC01351 and high-expressed ELK4 was more sensitive to the treatment of chemotherapy and AR antagonist. The patient could receive more benefits from chemotherapy (e.g., Cisplatin) and AR antagonist drugs (e.g., bicalutamide).