Identification of lncRNAs via microarray analysis for predicting HER2-negative breast cancer response to neoadjuvant chemotherapy

Abstract

Mortality is high in patients with locally advanced HER2-negative breast cancer, especially those with residual tumor after neoadjuvant chemotherapy (NAC). Tissue-specific long non-coding RNAs (lncRNAs) are responsible for specific breast cancer subtypes. To identify the lncRNAs involved in residual cancer tissues (RCTs) and to evaluate their potential for predicting HER2-negative breast cancer response to NAC, we used three paired tissues to compare differences in gene expression between RCTs and remittent tissues (RTs) after NAC in human HER2-negative breast cancer. Subsequently, we detected expression of the top ten up-regulated and down-regulated lncRNAs in 11 paired tissues via quantitative RT-PCR analysis. Finally, we explored the potential function of these dysregulated lncRNAs through bioinformatics analysis. Our results indicate that 1348 mRNAs and 183 lncRNAs were differentially expressed in RCTs compared with RTs, and the expression levels of four novel lncRNAs (DSCAM-AS1, LINC01508, lnc-MGST1-2 and lnc-BTG2-2) were in agreement with the microarray analysis results. Furthermore, we found that the expression level of LINC01508 was significantly related to poor prognosis, suggesting that LINC01508 is a potential biomarker for predicting breast cancer response to NAC, which might be helpful in exploring potential diagnostic factors and therapeutic targets for chemo-resistant HER2-negative breast cancer.

Introduction

Breast cancer (BC) is the second leading cause of cancer mortality among women worldwide [1]. Based on global transcriptomic analysis, BC is subdivided into different molecular subtypes, including luminal A, luminal B, HER2-enriched, claudin-low, basal-like, and normal breast-like subtypes [2,3]. Moreover, luminal B BC can be further classified as either HER2 positive or HER2 negative [4]. For HER2-negative BC, the therapeutic strategy is restricted to chemotherapy and endocrine therapy. However, some patients acquire chemo-resistance during chemotherapy, which leads to treatment failure and high mortality. Therefore, it is important to clarify the molecular mechanism of chemotherapy resistance in HER2-negative BC patients.

Neoadjuvant chemotherapy (NAC) is associated with rates of disease-free survival (DFS) and overall survival (OS) comparable to those for adjuvant (post-surgical) chemotherapy [5]. NAC is standard in locally advanced and operable BC because it is intended to shrink the tumor and improve the chance for breast-conserving surgery [6]. BC represents a highly heterogeneous disease due to its diverse clinical outcomes and biological behavior [7]. Pathological complete response (pCR) is the best predictor of patient outcome after NAC [6,8,9]. Clinical trials have found that different BC subtypes have different rates of pCR and that patients who show pCR have a different prognosis depending on subtype [10]. Many studies and clinical observations have demonstrated that the prognosis of the luminal subtype is better than that of HER2-positive or triple-negative BC (TNBC). However, Niikura et al. [11] have suggested that patients with luminal tumors have lower pCR rates than those with HER2-positive or triple-negative tumors. The differential response in a single BC tumor tissue may be caused by clinical heterogeneity, which is an inherent feature of breast tumors. In residual cancer tissues (RCTs), the shape and structure of some cancer cells are changed [12], and cells can develop drug resistance. Thus, we speculate that the heterogeneity of BC tumor tissues after NAC may lead to poor outcomes, and the residual tumor is a critical indicator of treatment failure.

Long non-coding RNAs (lncRNAs) are defined as RNA genes longer than 200 bp with no coding potential [13], and recent studies have shown that lncRNA gene silencing is involved in chromatin modification, transcriptional activation, and many other important biological process [14]. In studies of lncRNAs in BC, several well-known lncRNAs have been identified to regulate BC endocrinotherapy [15,16], such as HOTAIR [15,16], lncRNA-ARA [17] and lncRNA-PANDA [18]. However, there is little information reported about the link between systematic expression of lncRNAs and HER2-negative BC.

In this study, we performed microarray analysis to establish an mRNA and lncRNA gene expression profile to identify genes that are differentially expressed between RCTs and remittent tissues (RTs) of HER2-negative BC after NAC. We also performed lab experiments and bioinformatics analysis to analyze the biological functions of the genes to identify mRNAs or lncRNAs that play important roles in chemo-resistance, with the goal of providing new targets for chemo-resistance studies of HER2-negative BC.

Materials and methods

Tissue collection

In this study, we recruited 48 BC patients who had undergone NAC, from which 34 patients were excluded according to our inclusion criteria, which is presented in Figure 1. BC tissue samples consisting of RCTs and RTs from patients who had invasive BC diagnosed histopathologically and had been treated with NAC were collected consecutively between September 2015 and May 2016. None of these patients accepted radiotherapy or endocrine therapy prior to breast tumor resection. The RTs were resected 2 mm from the edge of the tumor (according to the standard of safe cutting for breast-conserving surgery after NAC [19]). This study and the necessary sample collection were approved by the Ethical Review Committee of the Second Xiangya Hospital of Central South University.

Figure 1.

Flowchart of inclusion criteria for patient enrollment.

RNA extraction and quality control

To isolate total RNA from each tissue, frozen tissues were resuspended in TRIzol reagent (Life Technologies) and were finally eluted into 100 μL of Elution Solution according to the manufacturer’s instructions. Quantification and quality checks were performed with a NanoDrop ND-2000 spectrophotometer (Thermo Scientific). All the RNA samples were stored at -80°C until further analysis.

LncRNA and mRNA microarray expression profiling

An Agilent Human Gene Expression (8*60 K, Design ID: 039494) microarray was used in this study. The lncRNAs were obtained from authoritative databases (i.e., RefSeq, Ensembl, UCSC Known genes, and LNCipedia) and related literature. The mRNAs were collected from RefSeq and GENCODE. Each transcript was represented by a specific exon or splice junction probe, which could identify individual transcripts accurately. Sample labeling, microarray hybridization, and washing were performed based on the Agilent One-Color Microarray-Based Gene Expression Analysis protocol (Agilent Technology). Briefly, total RNA was transcribed to double-stranded cDNA, synthesized into cRNA and labeled with cyanine-3-CTP. Labeled cRNAs were hybridized onto the microarray. After washing, the arrays were scanned with an Agilent Scanner G2505C (Agilent Technologies) and microarray profiling was conducted in the laboratory of the OE Biotechnology Company in Shanghai, People’s Republic of China.

Microarray results analysis and prediction of the functions of lncRNAs

Differential mRNA and lncRNA expression level based on microarray analysis

Feature Extraction software (version 10.7.1.1, Agilent Technologies) was used to analyze the array images to obtain the raw data. GeneSpring (version 13.1, Agilent Technologies) was employed to complete the basic analysis of the raw data. Differentially expressed genes were then identified through the observed fold change (FC), and P values were calculated with a t-test. The threshold set for up- and down-regulated genes was an FC≥2.0 and a P value ≤0.05.

Co-expression of lncRNAs with mRNAs and function prediction

The gene ontology (GO) project provides a controlled vocabulary to describe gene and gene product attributes (http://www.geneontology.org) and has been widely used in large-scale genomic and transcriptional data functional studies. KEGG pathway analysis offered us a reliable method to elucidate the candidate biological pathways in which the lncRNAs interacted with the mRNAs. In this study, a bioinformatics analysis was performed to annotate the functional roles of the mRNAs, which were significantly correlated with the aforementioned lncRNAs.

qRT-PCR validation of 3 differentially expressed lncRNAs

Quantitative real-time PCR was performed to validate the microarray experiments using an independent cohort of 11 paired BC RCT and RT samples. Total RNA was extracted following the manufacturer’s protocols for each kit. RNA quality was confirmed using a NanoDrop 1000 spectrophotometer. An OD_260/280 of approximately 1.8 was set as the criterion of acceptable purity. Reverse transcription was performed by following the manufacturer’s protocols for a First Strand cDNA Synthesis Kit. The relative levels of the top ten up-regulated and down-regulated lncRNAs were determined via quantitative real-time PCR, which was performed using ABI Power SYBR1Green PCR Master Mix (Applied Biosystems, USA). Relative lncRNA expression levels were calculated using the 2-ΔΔCt method and were normalized to β-actin expression.

Kaplan-Meier analysis

The prognostic value of lncRNA expression was evaluated using the online database Kaplan-Meier Plotter (www.kmplot.com), which contained lncRNA expression data and survival information for 761 clinical BC patients. To analyze the relapse-free survival (RFS) of patients with BC, patient samples were split into two groups by median expression (high vs. low expression) and assessed using a Kaplan-Meier survival curve, with a hazard ratio (HR) with 95% confidence intervals (CI) and a log-rank P value.

Results

General expression profiles of differentially expressed lncRNAs and mRNAs

Feature Extraction software was used to obtain standardized data (Table 1). In unsupervised hierarchical clustering analysis, the differentially expressed lncRNAs were used to generate a heat map (Figure 2A). In total, 1801 genes were significantly altered (FC>2; P<0.05) in 3 pairs of samples, including 1348 mRNAs, 183 lncRNAs and 270 other genes (Figure 2B). The chromosome location data showed the number of up- or down-regulated lncRNAs located on specific human chromosomes (Figure 2C).

Table 1.

The top 20 dysregulated lncRNAs (RCT vs. RT)

LncRNAs	P-value	FC	Regulation
TRHDE-AS1	0.012647376	124.45968	Down
DSCAM-AS1	0.027974129	115.70292	Up
Lnc-MGST1-2	0.04264511	82.2943	Down
LINC01070	0.044032626	65.88673	Down
XLOC_l2_008043	0.023433592	46.68634	Down
lnc-BTG2-2	0.018294355	36.8005	Down
PP14571	0.04138621	35.61804	Up
PGM5P3-AS1	0.002629558	35.59262	Down
LINC01497	0.010669898	35.54846	Down
LOC284825	0.008203023	27.27142	Down
LINCR-0003	0.030180234	27.17261	Down
MEOX2-AS1	0.034956995	25.0278	Down
lnc-CCDC74A-1	0.04868132	17.09113	Up
LINC01136	0.018502263	16.6816	Up
HOTAIR	0.029617796	12.68415	Up
lnc-C10orf96-1	0.03701051	8.49516	Up
XLOC_l2_011873	0.03658473	8.17016	Up
PVT1	0.026796862	7.97259	Up
LINC01508	0.028164737	7.10403	Up
lnc-CNTNAP5-2	0.020194884	6.84649	Up

Figure 2.

LncRNA microarray data of three paired BC tissues (between RCT and RT). The lncRNA expression patterns in samples is shown as a heatmap based on hierarchical clustering (A); Using second-generation RNA microarray analysis, 1801 RNAs (log fold-changes >2) were detected. Pie chart showing the proportion of components (B); Chromosome distribution showing the number of up- or down-regulated lncRNAs located in different chromosomes (C).

LncRNA function prediction

Hundreds of lncRNAs were co-expressed with thousands of mRNAs. The lncRNAs were clustered into hundreds of GO and KEGG pathway annotations, and the top ten GO biological processes and KEGG pathways enriched for the up-regulated or down-regulated lncRNAs are presented in Figure 3. In our results, “positive regulation of cell proliferation” was enriched in GO biological process for down-regulated lncRNAs, while “drug metabolism-cytochrome P450” was enriched in the KEGG pathway for down-regulated lncRNAs, which indicates the possibility that the down-regulated lncRNAs participate in the chemo-resistance process.

Figure 3.

GO and KEGG pathway analysis. The top 10 GO terms associated with the coding gene functions of up-regulated lncRNAs (left upper panel) and down-regulated lncRNAs (right upper panel) are listed. The top 10 pathways associated with the coding genes of up-regulated lncRNAs (left lower panel) and down-regulated lncRNAs (right lower panel) are listed.

RT-PCR confirmation and prognostic value

To validate the microarray data, the top ten up-regulated and down-regulated lncRNAs were examined by RT-PCR in 11 pairs of BC tissues (RCT and paired RT). In our results, we found that the expression of two up-regulated lncRNAs (LINC01508 and DSCAM-AS1) and two down-regulated lncRNAs (lnc-MGST1-2 and lnc-BTG2-2) was consistent with the microarray analysis and showed a significant difference between RCT and RT (Figure 4A).

Figure 4.

Validation of dysregulated lncRNAs and prediction of their prognostic value. We selected 20 lncRNAs (the top ten up-regulated and down-regulated lncRNAs) for real-time PCR quantification in 11 paired BC tissues. The data show that lnc-MGST1-2 and lnc-BTG2-2 are down-regulated in the residual cancer tissues, while LINC01508 and DSCAM-AS1 are significantly up-regulated in the residual cancer tissues compared with their levels in the remittent tissues (A). Prognostic value of the LINC01508 expression level in BC patients (B).

We also predicted the prognostic value of lncRNA expression using Kaplan-Meier Plotter, and the clinical data co-related to 11 lncRNAs (6 up-regulated lncRNAs and 5 down-regulated lncRNAs) was acquired from TCGA database (The Cancer Genome Atlas). Interestingly, we found that the expression level of the novel lncRNA LINC01508 was significantly related to the overall survival of BC patients (P=0.014), and a high LINC01508 expression level indicated a poor BC patient prognosis (Figure 4B).

Discussion

BC is the most frequently diagnosed cancer in females, and a large number of patients, especially those with HER2-negative BC, experience treatment failure due to recurrence and metastasis. Various drugs have been approved for BC, but the acquisition of resistance remains a substantial obstacle for clinical management of the disease. After chemotherapy, the morphology and structure of some cancer cells are altered in RCTs. These tissues are characterized by degeneration of cancer cells, proliferation of fibrous tissue cells and infiltration of lymphocytes [12]. Additionally, epithelium- and cancer stem cell-related markers have been found in RCTs [20]. At the same time, drug resistance is enhanced in residual cancer cells. The resistant and normal tumor stem cells exhibit the same pattern of accumulated mutations, which leads to newly acquired drug resistance via gene activation, point mutations and gene amplification [21]. However, the specific mechanism is unclear. Therefore, elucidation of these molecular mechanisms, particularly the mechanism associated with chemotherapy resistance, is crucial for better prediction of BC patient outcome and response to therapy.

Although study of lncRNAs has increased rapidly, only a few lncRNAs are well understood. Among the 183 differentially expressed lncRNAs we identified from the microarray results, only four were verified to be dysregulated in chemo-resistant tissues. Expression of lncRNAs exhibits temporal-spatial specificity. For instance, different lncRNAs derived from different types of tumor tissues and specific tissue-based lncRNAs are responsible for specific subtypes of BC. Therefore, determining how to design different groups for analysis is critical for identifying the subset-specific lncRNAs. This study is the first to perform a microarray analysis comparing RCTs and RTs, and the majority of the differentially expressed lncRNAs we found have seldom been reported previously.

The new lncRNAs are generally less evolutionarily conserved than known lncRNAs, with a large fraction unique to humans or primates. Our predicted function results showed enrichment in hundreds of biological processes and pathways, among which we found some cancer behaviors that were specifically enriched in the down-regulated lncRNA group, indicating that chemo-resistance is related to silencing of tumor suppressor genes. Through analysis of RNA-seq samples in a cohort of 947 BC patients, Yashar et al. [22] found that DSCAM-AS1 was down-regulated in BC tissues and participated in tumor progression and tamoxifen resistance.

Although the present results provided obvious evidence of some novel lncRNAs that may serve as BC biomarkers of poor survival, such as LINC01508, whose overexpression in BC tissues was related to poor clinical outcomes based on comparison of the Kaplan-Meier plots, further research should be performed in the future to verify our results. Particularly, it is difficult to obtain information regarding the location of the novel lncRNAs and the specific mechanisms underlying the process of BC cell proliferation, invasion, metastasis. In addition, we plan to use in vivo and in vitro experiments to further study the specific role of the differentially expressed genes in HER2-negative BC in chemo-resistance mechanisms.

Disclosure of conflict of interest

None.