Lower expression of LINC00092 in lung adenocarcinoma might mean poorer prognosis: A study based on data mining and bioinformatics.

The mechanisms that underlie long non-coding RNA 00092 (LINC00092) in lung adenocarcinoma (LUAD) remain unclear. In this study, by mining the Cancer Genome Atlas and Gene Expression Omnibus databases and using bioinformatics tools, we try to elucidate the function of LINC00092 in LUAD.The the Cancer Genome Atlas and gene expression Omnibus microarray datasets were used to analyze and evaluate the expression of LINC00092 in LUAD and its clinical significance. Clinical samples were collected and the relative expression level of LINC00092 were identified by quantitative real time polymerase chain reaction. The LINC00092 related genes were identified by Multi Experiment Matrix, The Atlas of ncRNA in Cancer and the database of RNA-Binding Protein specificities. The predicted genes were then sent to the Gene Ontology enrichment and the Kyoto Encyclopedia of Genes and Genomes pathway analysis.The expression of LINC00092 was significantly decreased in LUAD tissues compared to non-tumor tissues (standard mean difference =-1.10, 95% confidence interval: -1.87 to -0.32, P < .001, random). Low expression of LINC00092 was associated with the poor overall survival (hazard ratio = 1.32, 95% confidence interval: 1.08-1.62, P < .05, fixed) and high pathological stage (P < .05). The relative expression level of LINC00092 in clinical samples were significantly lower in LUAD tissues compared with adjacent normal tissues. (P < 0.05) 61 LINC00092 related genes were identified; the Kyoto Encyclopedia of Genes and Genomes analysis showed that the most significant signaling pathways were: NF-κB, HIF-1 and ErbB signaling pathways.In this study, we found that the decrease of LINC00092 expression was involved in LUAD tumorigenesis and metastasis, and the depletion of LINC00092 was associated with a poor prognosis in patients with LUAD. The mechanisms that underlie LINC00092 in LUAD might be related to the NF-κB, HIF-1 and ErbB signaling pathways.

Introduction

Lung cancer is the most frequent cause of cancer-related deaths worldwide. About 1.8 million people were diagnosed with lung cancer, and 1.6 million people died as a result of the disease every year. Lung adenocarcinoma (LUAD), as the most common histologic type of lung cancer, accounts for almost half of the lung cancers.[ Despite the development of various detection measures and clinical treatments, LUAD remains a major public health problem. Lots of patients with LUAD are diagnosed with advanced stage or distant metastasis when they initially seek medical treatment. The overall cure rate and 5-year survival rate of LUAD is still slim.[

Long non-coding RNAs (lncRNAs) are a class of non-coding RNAs that have no protein-coding capacity and are more than 200 nucleotides in length. lncRNAs play an important role in regulating gene expression at various levels, such as the regulation of the targeted gene transcription, post-transcriptional function as well as epigenetic regulation of genes.[ Recent studies have reported that lncRNAs have been involved in the process of occurrence, progression and treatment of various cancers, such as breast cancer, leukemia, colorectal cancer, gastric cancer, and liver cancer.[ Therefore, the discovery and identification of novel functional lncRNAs as candidates for diagnosis and intervention of LUAD is of great significance.

Public databases such as the Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) have been proved valuable tools because they allow us to access large amounts of high-throughput data and related clinical data. Those data are informative and far from being fully applied. By analyzing these data, we can identify numerous prognostic biomarkers for various malignancies. In this study, we tried to use TCGA and GEO databases as source of relevant data and clinical information, and followed by using multiple public bioinformatics tools to explore the potential role of long non-coding RNA 00092 (LINC00092) in LUAD tumorigenesis and prognosis.

Materials and methods

Data acquired from TCGA

The relevant LINC00092 data in LUAD tumor tissues and normal tissues were obtained from the TCGA database (https://portal.gdc.cancer.gov/) up to December 2019. Primary Site: bronchus and lung; Program: TCGA; Project: TCGA-LUAD; Disease Type Key words adenomas and adenocarcinomas, acinar cell neoplasms, cystic, mucinous and serous neoplasms. The RNA expression data were classified as Level 3. All RNA-Seq and clinical data were chosen and archive-building function was performed. All quantification files with “.txt” were selected, and all files in the data and clinical sections were examined.

Data acquired from GEO

GEO databases, including GEO datasets (http://www.ncbi.nlm.nih.gov/gds/) and GEO profiles (http://www.ncbi.nlm.nih.gov/geoprofiles/), were also used to gain the relevant LINC00092 data. The Key words “lung cancer”, “non-small cell lung cancer”, “NSCLC”, “lung adenocarcinoma(s)”, “lncRNA”, “long non-coding RNA” and “LINC00092” were used as search terms. The following criteria were strictly observed:

study type: Expression profiling by array;

datasets were obtained from Homo sapiens;

datasets included LUAD tissues and non-tumor tissues;

for analyzing of expression, the datasets included more than 5 samples in each group; for analyzing of hazard ratio (HR), the datasets included no less than 20 samples;

datasets provided overall survival (OS) information;

the expression data of LINC00092 from the cases and the controls were provided or could be calculated;

the non-small cell lung cancer (NSCLC) datasets should contain a clearly defined subset of adenocarcinomas.

Potentially relevant datasets were further screened by reviewing titles, summary and overall design. Relevant full articles and samples information were evaluated in order to identify studies that met the eligible criteria. Data for eligible datasets and series were up to December 2019.

Confirmation by quantitative real time polymerase chain reaction

According to the results of TCGA and GEO database analysis, clinical tissue samples of LUAD (n = 20) and their normal adjacent (n = 20) were obtained. All samples collection and analyzing have been approved by the ethics institute of the People's Hospital of Deyang City and all the informed consents of the participants have been obtained. Total RNA was extracted by RNeasy FFPE Kit (Cat. No. 73504, QIAGEN, Shanghai, China). Complementary DNAs were generated using the QuantiNova Reverse Transcription Kit (Cat. No. 205411, QIAGEN, Shanghai, China). Quantitative real-time PCR were conducted by using the QuantiNova SYBR Green PCR Kit (Cat. No. 208054, QIAGEN, Shanghai, China) on GENTIER 96E PCR system (Tianlong, Xian, China). Relative expression level of LINC00092 were analyzed by using the 2-ΔΔCt method. The PCR primers were designed and synthesized by Takara (TAKARA, Dalian, China). The human GAPDH was used as endogenous controls for lncRNA expression in this analysis. All PCR primers are listed in Table 1.

Table 1

Primer sequences used for quantitative real time polymerase chain reaction.

Statistical analysis

LINC00092 expression data were presented as mean ± standard deviation, and SPSS v25 software (IBM Corporation, Armonk, NY, USA) was used for data analysis. Independent-samples t test was employed for the analysis of LINC00092 expression between 2 groups. To evaluate the relationship between LINC00092 expression and OS in LUAD patients, the Kaplan–Meier method and log-rank P-value were calculated with the analysis tool SPSS Statistics 25. The cutoff value of LINC00092 expression was determined by its median value. Survival curves were plotted with GraphPad Prism v8.0 (GraphPad Software, Inc., La Jolla, CA, USA) and also harvest from the analysis tools, which can be accessed online at OncoLnc (http://www.oncolnc.org).[ GraphPad Prism 8 was also used to complete the box plot and symbols & lines plot of clinical samples.

Meta-Analyses were accomplished by using STATA software package v15.0 (Stata Corporation, College Station, TX, USA). All data obtained from TCGA and GEO datasets were estimated as standard mean difference with a 95% Confidence Interval (95% CI). HR and 95% CI were calculated to represent the prognosis of LUAD with expression of LINC00092. Effect sizes were pooled with a random or fixed effects model. HR>1 represented a poor prognosis for the group with low LINC00092 expression and would be considered to be statistically significant if the 95% CI did not overlap 1. P < .05 was considered to a significant difference.

Genes that related to LINC00092 in LAUD

Multi Experiment Matrix (MEM) (http://biit.cs.ut.ee/mem/) is a web bioinformatic tool, it can identify lncRNAs related genes by gathering publicly available gene expression datasets from the ArrayExpress (http://www. ebi.ac.uk/arrayexpress/) database.[ The Atlas of ncRNA in Cancer (TANRIC) (https://ibl.mdanderson.org/tanric/_design/basic/index.html) is an open-access web resource for interactive exploration of lncRNAs in cancer.[ These 2 bioinformatic tools along with the database of RNA-Binding Protein specificities (RBPDB) (http://rbpdb.ccbr.utoronto.ca/),[ a collection of experimental observations of RNAbinding sites, were used to determine genes that related to LINC00092 in LUAD. Repeat genes from any 2 of those 3 sites were included as LINC00092 related genes in LUAD.

The comprehensive analysis of LINC00092 related genes

Gene ontology (GO) analysis, including 3 major groups: biological processes, cellular component (CC) and molecular functions, were performed by using DAVID website (https://david.ncifcrf.gov/, version 6.8).[ Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was implemented in KOBAS website (http://kobas.cbi.pku.edu.cn/, version 3.0).[ The protein-to-protein interaction (PPI) network analysis was conducted by STRING website (https://string-db.org/cgi/input.pl).[

Results

Basic information, expression and HRs of LINC00092 in LUAD

According to the inclusion criteria, 9 GEO series (GSE) in the GEO database were obtained for this study. The evaluation process of the datasets was shown in Figure 1. With the combination of TCGA and GEO data, a total number of 1518 LUAD samples were enrolled. Sample sizes ranged from 12 to 592 cases. Detailed information was shown in Tables 2–4 and Figure 2.

Figure 1

The process of data identification and acquisition. GEO = gene expression Omnibus.

Table 2

Basic information about enrolled datasets.

Table 4

Analysis of overall survival in datasets obtained from gene expression Omnibus and the cancer genome atlas.

Figure 2

Expression levels of long non-coding RNA 00092 in lung adenocarcinoma tumor tissues and non-tumor tissues were measured in a total of 7 datasets. ∗∗∗∗P < .001. TCGA = the cancer genome atlas.

Combined with TCGA and GEO data, Meta-analysis revealed that the expression of LINC00092 was significantly downregulated in LUAD tumor tissues compared with non-tumor tissues (standard mean difference = –1.10, 95% CI: –1.87 to –0.32, P < .001, random) (Fig. 3A). The low expression of LINC00092 was associated with poorer OS in LUAD patients (HR = 1.32, 95% CI: 1.08–1.62, P < .05, fixed) (Fig. 3B).

Figure 3

Meta-analysis based on the cancer genome atlas and gene expression omnibus datasets. Notes: (A) forest plots to assess the expression of long non-coding RNA 00092 in tumor tissues and non-tumor tissues. (B) Forest plots to assess the effect of downregulated long non-coding RNA 00092 expression on overall survival in patients with lung adenocarcinoma. Weights are from random effects analysis. GEO = gene expression Omnibus, LUAD = Lung adenocarcinoma, OS = overall survival, SMD = standard mean difference, TCGA = the cancer genome atlas.

Analysis of clinicopathological parameters based on TCGA

The analysis of LINC00092 expression in TCGA between 2 groups showed that the expression level of LINC00092 was lower in tumor tissues (6566.9673 ± 7211.8636), compared with normal tissues (22631.9661 ± 8569.0259; P < .001). Additionally, in the pathologic T group, LINC00092 expression was significantly downregulated in T2-T4 group (5909.8924 ± 6891.3360), which represented for bigger tumor size, compared with the group T1 (8004.4016 ± 7789.4820; P = .003). Furthermore, the expression of LINC00092 was significantly downregulated in pathologic stage III-IV group (5365.8887 ± 6987.8984), compared with the group I-II (6949.9821 ± 7346.8671; P = .044). However, LINC00092 was not associated with age, gender and “pathologic N”. Clinicopathological characteristics are presented in Table 5. Survival curves found a marked difference between high and low LIINC00092 expression group (P < .05). (Fig. 4).

Table 5

Expression of long non-coding RNA 00092 and clinicopathological parameters in lung adenocarcinoma obtained from the cancer genome atlas.

Figure 4

Kaplan–Meier survival curves for patients with high and low long non-coding RNA 00092 expression. Notes: (A) survival curve based on the cancer genome atlas data and plotted by GraphPad Prism 8. (B) Survival curve came from the OncoLnc website (http://www.oncolnc.org/).

Relative expression level of LINC00092 in clinical samples

In the present study, we compared the expression of LINC00092 in 20 cases of LUAD tissues and their matched adjacent normal tissues. Total RNA was isolated from the tissues, and quantitative real-time PCR was performed. Relative expression level of LINC00092 were analyzed by using the 2-ΔΔCt method. Our data showed that the relative expression level of LINC00092 were markedly decreased in LUAD tissues compared to their adjacent normal tissues (Fig. 5A). A line connects 2 points from LUAD to normal with an increased trend, suggesting that the expression level of LINC00092 were significantly lower in LUAD tissues than in the matched normal tissues (Fig. 5B).

Figure 5

Relative expression level of long non-coding RNA 00092 in clinical samples. Notes: (A) box plot to uncover the relative expression level of long non-coding RNA 00092 between lung adenocarcinoma and normal tissues. (B) Symbols & lines plot suggesting that the relative expression level of long non-coding RNA 00092 was lower in lung adenocarcinoma tissues than in the matched normal tissues. LUAD = Lung adenocarcinoma.

Identification of LINC00092 related genes in LUAD

Three public bioinformatic tools were used to identify genes related to LINC00092: MEM, TANRIC, and RBPDB. The number of genes identified from MEM was 7160, when the score was set at 10-4. A total number of 78 genes related to LINC00092 were identified from TANRIC and another 39 genes were recognized by PBPDB. After calculating out genes repeated from any 2 of those 3 sites, 61 genes were eventually used as genes associated with LINC00092 in LUAD.

GO enrichment analysis

DAVID was enrolled for GO analysis. There were 14 significant annotations of the biological process (DIRECT) (P < .05), including mRNA processing, positive regulation of phosphatidylinositol 3-kinase activity, B cell receptor signaling pathway, negative regulation of translation, MAPK cascade, and so on. When comes to the GO analysis about CC (DIRECT), 4 significant terms were observed (P < .05). They are cytoplasmic stress granule, extracellular exosome, growth cone, and cell cortex. As to the molecular function (DIRECT) of GO analysis, 7 significant terms were procured (P < .05). They were contributed to cytokine activity, RNA and poly (A) RNA binding, receptor signaling protein tyrosine kinase activity, Ras guanyl-nucleotide exchange factor activity, etc. The outcomes showed in Table 6 and Figure 6. The top 5 results were listed according to P-value.

Table 6

The gene ontology analysis of predicted target genes of long non-coding RNA 00092.

Figure 6

GO enrichment analysis of long non-coding RNA 00092 related genes in lung adenocarcinoma. Note: long non-coding RNA 00092 related genes were divided into 3 functional groups: biological process, cellular component, and molecular function. GO = gene ontology, LUAD = lung adenocarcinoma.

KEGG pathway analysis

KOBAS was responsible for the KEGG pathway analysis, there were 30 signaling pathways that were statistically significant (P < .05). The top 20 items were listed in Figure 7 according to corrected P-value. The signaling pathways were mainly enriched in the NF-kappa B signaling pathway, HIF-1 signaling pathway and ErbB signaling pathway, etc.

Figure 7

Pathway analysis of the predicted target genes of long non-coding RNA 00092. Note: Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis. Top 20 pathways were listed in this graph (P < .05).

PPI interaction of the 61 genes

The protein-to-protein network analysis was conducted by STRING website (Fig. 8). In the networks, the nodes represent for proteins, and each color corresponds to a cluster. The line thickness is an indicator for the strength of the evidence. The edges indicate the predicted functional associations, which are colored by the types of predicted associations using different evidence. The red lines, green lines, blue lines, purple lines, yellow lines, light blue lines, and black lines indicate fusion evidence, neighborhood evidence, co-occurrence evidence, experimental evidence, text mining evidence, database evidence, and co-expression evidence, respectively. As showed in this PPI network, several genes demonstrated higher connectivity, including SYK, ERBB4, NCL, IGF2BP1, PRKCB, WDR12, and CD40LG, etc.

Figure 8

Protein–protein interactions of the long non-coding RNA 00092 related genes.

Discussion

The development of LUAD involves molecular mechanisms, and many studies have shown that lncRNA plays a key role in this process.[ Therefore, exploration of tumor-associated lncRNAs and its related signaling pathways is of great significance, which can provide a sufficient theoretical basis for elucidating the potential mechanisms of LUAD patients and finding effective therapeutic targets. The TCGA and GEO databases provided a large amount of open raw data from multiple experiments on multiple platforms, making big data analysis possible.

In this paper, we attempted to investigate the potential roles of LINC00092 in LUAD by analyzing the data excavated from TCGA and GEO databases, and we conformed the expression level of LINC00092 in in-house real tumor samples. Our results indicated that the expression of LINC00092 was significantly decreased in LUAD tissues compared with non-tumor tissues. Then the HR of meta-analysis established that low LINC00092 expression was a high HR indicator. Thereafter, analysis of TCGA data showed that the decline in LINC00092 expression was not related to gender and age, but was associated with pathologic T and pathological stage. As to the pathologic M, an indicator for remote metastasis, statistical analysis showed that there was no difference between 2 groups. The reason for this phenomenon might contributed to “MX”. “MX” means that the remote metastasis status cannot be assessed. Among the 511 eligible “pathologic M” data, there were 140 cases of “MX”, accounting for more than 27.3%. Simply categorizing “MX” as “M0” or “M1” might result in false positive or false negative results. Therefore, it has not been clarified whether the expression of LINC00092 is related to remote metastasis. We also found that patients with lower expression of LINC00092 exhibited poorer OS compared to patients with higher expression of LINC00092. Taken together, our findings suggested that the absence of LINC00092 played a crucial role in tumorigenesis and prognosis in patients with LUAD.

At present, there is still no related research on the molecular mechanism of LINC0092 in LUAD patients. So, the second step of this study is to explore the potential mechanism about LINC00092 in LUAD. Here, we enrolled 3 biological databases: MEM, TANRIC, and RBPDB. Those 3 biological tools belong to 3 different kinds of category. MEM recognizes lncRNA-related genes based on publicly available gene expression data sets from ArrayExpress database. TANRIC characterizes the expression profiles of lncRNAs in large patient cohorts of 20 cancer types, while the RBPDB database predicts genes associated with lncRNA based on the binding properties of the protein. We used these tools separately and conjointly to analyze the outcomes. Hence, 61 LINC00092 related genes were harvested. The next, we performed GO and KEGG pathway analysis to identify biological functions enriched among those LINC00092-related genes. The results of GO analysis demonstrate that LINC00092 is closely related to cellular RNA related processing and phosphorylating activities. In addition, the CC of LINC00092 related genes is mainly associated with cytoplasmic stress granule and exosome.

KEGG pathway analysis found that multiple signaling pathways may be regulated by LINC00092. Among these signaling pathway, NF-κB, HIF-1 and ErbB signaling pathways were significant. These signaling pathways have been proved to be clearly involved in the process of tumorigenesis. NF-κB signaling pathway is activated in over 60% of lung cancers. Aberrant activation of NF-κB signaling in LUAD cell lines and tumor tissues could be involved in tumorigenesis and metastasis.[ In addition to regulating cell proliferation, angiogenesis, and apoptosis, NF-kappaB is also involved in mechanism of epithelial-mesenchymal transition,[ which is critical for cancer cell invasion and metastasis. Activation of HIF-1 signaling is considered as tumor-initiating factor. It may involve in tumor angiogenesis, energy metabolism, cell proliferation, invasion, and metastasis.[ HIF-1alpha antagonist can inhibit progression and spread of LUAD.[ The ErbB receptor tyrosine kinase family contains 4 cell surface receptors: ErbB1/EGFR/HER1, ErbB2/HER2, ErbB3/HER3, and ErbB4/HER4.[ Besides its involvement in tumorigenesis and metastasis through Akt, MAPK and many other pathways, ErbB also appears as a kinase and transcriptional regulatory molecule in the nucleus. In many forms of malignancy, members of the ErbB family and some of their ligands are often overexpressed, amplified, or mutated.[ Therefore, we suppose that LINC092 inhibits tumorigenesis and progression by modulating these pathways in LUAD.

Finally, the following limitations should be considered in our analysis:

the inclusive data and clinical samples were limited. We believe that as the data grows and becomes more sophisticated, there may be more useful results;

This study was based on bioinformatics analysis, and the genes involved were predicted. Therefore, we still need further experiments to evaluate and verify the contribution of LINC00092 in LUAD.

Conclusion

In conclusion, based on the data we obtained from TCGA and GEO datasets and the information we harvested from real tumor samples, we conclude that the expression of LINC00092 might play a pivotal role in the initiation and progression of LUAD. LINC00092 was detected at low expression levels in LUAD tissues and was linked to high pathologic stage. Moreover, the lower expressions of LINC00092 in LUAD indicated poorer OS. The possible signaling pathways for LINC00092 contributing in LUAD were NF-κB, HIF-1 and ErbB signaling pathways. Overall, this study might provide a potential approach to molecular diagnosis, evaluation of prognosis, and targeted therapy of LUAD in the future.

Author contributions

Xin Wang designed the study and revised the article. Xianwei Wang collected the clinical samples and completed the experiments of clinical samples. Xin Wang reviewed the literature and wrote the article. Guichuan Huang reviewed the literature and revised the article. All authors read and approved the final manuscript.

Conceptualization: Xin Wang.

Data curation: Xin Wang.

Formal analysis: Xin Wang.

Investigation: Xianwei Wang.

Methodology: Xianwei Wang, Guichuan Huang.

Software: Yi Chen.

Writing – original draft: Xin Wang.

Table 3

Characteristics of long non-coding RNA 00092 expression profiling datasets obtained from gene expression Omnibus and the cancer genome atlas.