Gene expression profiling after LINC00472 overexpression in an NSCLC cell line 1

Abstract

Lung cancer accounts for a large proportion of cancer-related deaths worldwide. Personalized therapeutic medicine based on the genetic characteristics of non-small cell lung cancer (NSCLC) is a promising field, and discovering clinically applicable biomarkers of NSCLC is required. LINC00472 is a long non-coding RNA and has been recently suggested to be a biomarker of NSCLC, but little is known of its mechanism in NSCLC. Thus, the current study was performed to document changes in gene expression after LINC00472 overexpression in NSCLC cells. As a result of cell viability and migration assay, LINC00472 downregulated cell survival, proliferation, and motility. Transcriptome sequencing analysis showed 3,782 genes expression were changed in LINC00472 overexpressing cells. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis showed most genes were associated with intracellular metabolism. The PPP1R12B, RGS5, RBM5, RBL2, LDLR and PTPRM genes were upregulated by LINC00472 overexpression and these genes functioned as tumor suppressors in several cancers. In contrast, SPSB1, PCNA, CD24, CDK5, CDC25A, and EIF4EBP1 were downregulated by LINC00472, and they functioned as oncogenes in various cancers. Consequently, the function of LINC00472 in tumorigenesis might be related to changes in the expressions of other oncogenes and tumor suppressors.

1. Introduction

Lung cancer is a common cause of cancer death because during the early stage the disease is asymptomatic, and thus, it is usually detected at an advanced stage [1]. Lung cancer is traditionally classified as non-small cell lung cancer (NSCLC) or small cell lung cancer (SCLC), and NSCLC accounts for about 80% of patients [2]. Although most cases of SCLC are treated non-surgically, NSCLC is treated using surgery and adjuvant therapy [2]. The most common cause of lung cancer is smoking [3], and thus, preventing tobacco use is the most important factor in terms of overcoming NSCLC. Nevertheless, more aggressive, sophisticated treatment regimens are needed to completely eradicate the disease. The treatment of NSCLC has undergone many changes, including advances in targeted therapies and immunotherapy. Furthermore, personalized therapeutic medicine shows great promise for the treatment of NSCLC [4]. Therefore, it is necessary to discover the applicable biomarkers and to profile gene expression changes in NSCLC.

LncRNAs are generally non-coding RNAs longer than 200 nucleotides [5]. In general, although the intracellular expressions of lncRNAs are low [6], lncRNA is importantly involved in various cellular processes including transcription, splicing, translation, cell cycle, and heat shock response [7, 8]. In addition, lncRNAs are known to be involved in cancer progression [9, 10] and in the pathogeneses of several other diseases [11]. Abnormal lncRNA expression is associated with tumor formation and can be used as a biomarker to predict cancer prognosis [12]. Interestingly, some lncRNAs function as oncogenes, whereas others function as tumor suppressors [13]. For example, the oncogenic lncRNA, lncRNA PTAR directly targets mir-101 to promote NSCLC cell growth [14], but lncRNA FENDRR acts as a tumor suppressor by binding to oncogenic mir-761 and inhibiting NSCLC aggressiveness [15].

One type of lncRNA, LINC00472, is a long intergenic non-protein coding RNA 472 located on human chromosome 6q13 and has been recently reported to act as a tumor suppressor in various cancers. Meta-analyses showed LINC00472 lowered the risk of tumor recurrence in breast cancer patients with a grade 2 tumor [16] and negatively controlled the cell cycle, metastasis, and proliferation in colorectal cancer [17]. The tumor inhibitory effect of LINC00472 has also been reported in NSCLC. LINC00472 inhibited cell invasion and migration of NSCLC cells and reduced tumor growth in nude mice [18]. However, despite evidence that LINC00472 acts as a tumor suppressor, few studies have investigated LINC00472-related genes and signaling pathways in NSCLC.

In this study, after overexpressing LINC00472 in the NCI-H838 cells (an NSCLC cell line), Transcriptome Sequencing analysis (total RNA-seq) was used to profile LINC00472-induced gene expressional changes to provide important data for future investigations on the LINC00472 signaling pathway in NSCLC.

2. Materials and methods

2.1 Cell culture and transfection

Cell culture and transfection were performed as previously described [19]. NCI-H838 cells (a human NSCLC cell line) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA), authenticated, and then maintained in early passages for no more than 6 months after receipt. Cells were grown in RPMI-1640 (HyClone, Marlborough, MA, USA) supplemented with 10% FBS (Hyclone), 100 U/ml penicillin and 100 $\mu$ g/ml streptomycin (Gibco, Waltham, MA, USA) at 37 ${}^{\circ }$ C in a 5% CO ${}_2$ /95% air atmosphere.

Cells were transiently transfected with a plasmid containing a full-length LINC00472 construct (pCDH-LINC00472) and a control plasmid (pCDH empty vector) using Lipofectamine 2000 (Invitrogen, Waltham, MA, USA), incubated for 24 h, and subjected to cell survival and metastasis assays.

2.2 Total RNA preparations

After transient transfection, total RNA was isolated from cells using TRIzol (Invitrogen) [20]. RNA qualities and concentrations were assessed using OD 260/280 ratios using a Nanodrop 2000 (Thermo Scientific, Waltham, MA, USA). Extracted RNAs were used for reverse transcription PCR and forwarded to Macrogen (Seoul, Republic of Korea) for transcriptome sequencing.

2.3 Reverse transcription PCR

A Total RNA isolated from cells was synthesized as cDNA using the MMLV Reverse Transcriptase system (Bioneer, Daejeon, Republic of Korea). The temperature conditions for reverse transcription (RT) were 60 min at 37 ${}^{\circ }$ C and 5 min at 99 ${}^{\circ }$ C, and the RT product was stored at 4 ${}^{\circ }$ C until next use. Each RT product was used for a polymerase chain reaction (PCR) with a specific primer, Taq DNA polymerase and dNTP mix (BeamsBio, Seongnam, Republic of Korea). Primers used in PCR were GAPDH (F: ATGACATCAAGAAGGTGGTG, R: CATACCAGGAAATGAGCTTG), LINC00472 (F: GATGGCAGCTGTCTCTCTCC, R: GGGCCTCTCTGACCGTATCT). The temperature conditions for PCR to confirm the expression of GAPDH and LINC00472 are as follows. GAPDH: After initial denaturation at 95 ${}^{\circ }$ C for 5 min, 30 cycles of 94 ${}^{\circ }$ C for 1 min, 50 ${}^{\circ }$ C for 1 min, 72 ${}^{\circ }$ C for 1 min were performed, and late denaturation at 72 ${}^{\circ }$ C for 10 min. LINC00472: After initial denaturation at 94 ${}^{\circ }$ C for 5 min, 30 cycles of 94 ${}^{\circ }$ C for 30 s, 52 ${}^{\circ }$ C for 30 s, 72 ${}^{\circ }$ C for 45 s were performed, and late denaturation at 72 ${}^{\circ }$ C for 10 min. The PCR product were then subjected to electrophoresis, and the relative intensity of the band photographed in UV was calculated through Image J (National Institutes of Health, Bethesda, Maryland, USA).

2.4 Cell viability assay

A Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) was used to measure cell survival and proliferation. A cell suspension of NCI-H838 cells (5,000 cells/100 $\mu$ L) was inoculated into a 96-well plate and cultured in an incubator for 24 h. After transient transfection, cells were incubated for 0, 24, 48, or 72 h, and 10 $\mu$ L of CCK-8 solution was added to each well. After incubation for 1–4 h, absorbances were measured at 450 nm.

In addition, trypan blue solution (Gibco) was used to measure cell viability [20]. Briefly, cells (1 $\times$ 10 ${}^5$ ) were transfected with pCDH-LINC00472 (plasmid containing, full-length LINC00472) or control plasmid (pCDH empty vector) and cultured in 60-mm dishes for 24 h. Cells were detached with 10% trypsin-EDTA (Gibco), washed with PBS (WELGENE Inc., Gyeongsan-si, Republic of Korea) resuspended in PBS, and diluted 1:1 with trypan blue solution. Cell viabilities were calculated by expressing the number of viable cells (not stained with trypan blue).

2.5 Colony-forming assay

For colony-forming assays [21], NCI-H838 cells were plated at a density of 600 cells per 35 mm dish. After 24 h, cells were transfected with pCDH-LINC00472 or pCDH empty vector, incubated at 37 ${}^{\circ }$ C in a 5% CO ${}_2$ /95% air atmosphere for 12 days, fixed with 10% methanol/10% acetic acid, and stained with 1% crystal violet (Sigma, St. Louis, MO, USA). Colonies containing more than 50 cells were identified using ImageJ software and scored as survivors.

2.6 Wound healing assay

For wound healing assays [22], cells were cultured until 70% confluent in growth medium. Cell monolayers were then scratched with a 200 $\mu$ L pipette tip and cells were further incubated with fresh culture medium in the presence of pCDH-LINC00472 or pCDH empty vector for 24 h. Photomicrographs were taken at $\times$ 100 using an Olympus CKX53 inverted microscope (Olympus Optical, Tokyo).

2.7 Cell migration assay

The Radius™ 96-well Cell Migration Assay (Cell Biolabs, Inc., San Diego, CA, USA) was used to measure cell migration ability. After wells had been pretreated with Radius™ Pretreatment Solution and Radius™ Wash Solution, 3.0 $\times {10}^4$ cells/100 $\mu$ L were seeded into each well and incubated for 24 h. After removing gel in the center of each well using 1 $\times$ Radius™ Gel Removal Solution, transient gene transfection was performed. Cells were then incubated for 24 h and areas covered by cells that migrated toward well centers were calculated.

2.8 Transcriptome sequencing

Total RNA concentrations were calculated using Quant-IT RiboGreen (Invitrogen). Samples were run on the TapeStation RNA screentape (Agilent, Santa Clara, CA, USA) to assess total RNA integrities. Only high-quality RNA preparations with RIN values of $>$ 7.0 were used for RNA library construction. A library was independently prepared for each sample using 0.5ug of total RNA using Illumina TruSeq Stranded Total RNA Library Prep Gold Kit (Illumina, Inc., San Diego, CA). After removing rRNA from total RNA using the Ribo-Zero rRNA Removal Kit (Human/Mouse/Rat Gold) (Illumina), remaining mRNA was fragmented using divalent cations under elevated temperature. RNA fragments were then copied into first-strand cDNA using SuperScript II reverse transcriptase (Invitrogen) and random primers. This was followed by second-strand cDNA synthesis using DNA Polymerase I, RNase H, and dUTP. These cDNA fragments then go through an end repair process, the addition of a single ‘A’ base, and then ligation of the adapters. The products were then purified and enriched by PCR to create final cDNA libraries, which were quantified using KAPA Library Quantification Kits for Illumina Sequencing platforms according to the qPCR Quantification Protocol Guide (Kapa Biosystems Inc., Wilmington, MA, USA) and then qualified using the TapeStation D1000 ScreenTape (Agilent). Indexed libraries were then sequenced using an Illumina NovaSeq sequencing system (Illumina, Inc.). Paired-end (2 $\times$ 100 bp), sequencing was performed by the Macrogen.

Table 1

LINC00472 acts as a tumor suppressor in various cancers

Cancer type	Downstream gene	Effects on cancer	Ref.
Bladder carcinoma	PDCD4 (upregulated)	Induction of apoptosis, Inhibition of invasion and tumor growth	[26]
Breast cancer	NF- $\kappa$ B (inhibition of phosphorylation)	Inhibition of proliferation, migration, invasion, survival and cell cycle	[30]
Colorectal cancer	PDCD4 upregulated by downregulated mir-196a	Induction of apoptosis, Inhibition of proliferation and tumor growth	[27]
Hepatocellular carcinoma	PDCD4 upregulated by downregulated mir-93-5p	Induction of apoptosis, Inhibition of proliferation, migration and invasion	[28]
Lung adenocarcinoma	DEDD upregulated by downregulated mir-24-3p	Induction of apoptosis, Inhibition of cell proliferation	[31]
Lung cancer	Mir-196b-5p (downregulated)	Induction of apoptosis, Inhibition of viability, migration, invasion and EMT	[32]
NSCLC	KLLN upregulated by downregulated mir-149-3p and mir-4270	Inhibition of cell migration, invasion, EMT and tumor growth	[18]
Osteosarcoma	FOXO1 upregulated by downregulated mir-300	Inhibition of cell proliferation, colony formation abilities and tumor growth	[29]

2.9 RNAseq data analysis

Raw reads were preprocessed from the sequencer to remove low quality and adapter sequences before analysis and processed reads were aligned to Homo sapiens (hg19) using HISAT v2.1.0 [23]. HISAT utilizes two index types for alignment, that is, a global, whole-genome index and tens of thousands of small local indexes. These two index types’ are constructed using the same BWT (Burrows-Wheeler transform)/a graph FM index (GFM) as Bowtie2. Because of the use of these efficient data structures and algorithms, HISAT generates spliced alignments several times faster than the widely used Bowtie and BWA. The reference genome sequence of Homo sapiens (hg19) and annotation data were downloaded from the NCBI, and transcript assembly of known transcripts, novel transcripts, and alternative splicing transcripts was processed using StringTie v1.3.4d [24, 25]. Base on the results obtained, expressional abundances of transcripts and genes were calculated as read counts or FPKM values (Fragments Per Kilobase of exon per Million fragments mapped) per sample. Expression profiles were for additional analysis, for example, to Differentially Expressed Gene (DEG). In samples obtained using different conditions, differentially expressed genes or transcripts were filtered by statistical hypothesis testing.

Statistical analysis was performed to identify differentially expressed genes using estimates of gene abundances in samples. Relative gene abundances were measured in FPKM using StringTie. Genes with one more that zeroed FPKM values in samples were excluded. To facilitate log2 transformation, 1 was added to the FPKM values of filtered genes. Filtered data were log2-transformed and subjected to quantile normalization. The statistical significances of differential expression data were determined using fold changes. The sequencing results have been deposited in the Gene Expression Omnibus database (GEO Series accession number GSE164230). For DEG sets, hierarchical clustering analysis was performed using complete linkage and Euclidean distance as measures of similarity. Gene-enrichment analysis and KEGG pathway analysis of DEGs were performed using Gene Ontology (http://geneontology.org/) and the KEGG pathway (https://www.genome.jp/kegg/) database, respectively.

2.10 Statistical analysis

Results are presented as the mean $\pm$ standard deviation (SD) at least three independent experiments were performed, and sample sizes were calculated to allow significances to be determined. Experimental results were analyzed by one-way ANOVA for ranked data followed by Tukey’s honestly significant difference test. Prism 5 software (GraphPad Software, San Diego, CA, USA) was used for the statistical analysis, and statistical significance was accepted for $p$ values of $<$ 0.05.

Figure 1.

Effects of LINC00472 overexpression on NSCLC cell viability and metastasis. (A) Expression of LINC00472 in NCI-H838 after transient transfection. Relative intensity $=$ (Control Vector or LINC00472 oe intensity)/(Negative Control intensity), Intensity $=$ (LINC00472 band intensity)/(GAPDH band intensity). ${}^{***}p<$ 0.001 (B, C) Short-term effects of LINC0472 overexpression on NCI-H838 proliferation as determined by CCK-8 and trypan blue assays as compared with treatment negative control cells or empty vector transfected controls. ${}^{*}p<$ 0.05, ${}^{**}p<$ 0.01, ${}^{***}p<$ 0.001 (D) Long-term effects of LINC0472 overexpression on cell proliferation were assessed using colony-forming assays. The number of colonies (%) $=$ (Number of colonies in Control Vector or LINC00472 oe/Number of colonies in Negative Control) $\times$ 100. ${}^{***}p<$ 0.001 (E, F) Effects of LINC00472 overexpression on cell invasion and migration were determined using wound healing and cell migration assays. Covered area by cells (%) and Migration ability (%) $=$ (Area covered by cells after 24 h/Area of wound at 0 h) $\times$ 100. ${}^{***}p<$ 0.001.

3. Results

3.1 Effects of LINC00472 on the survival and invasiveness of NSCLC cells

LINC00472 has been demonstrated to act as a tumor suppressor in several cancers including breast cancer, colorectal cancer, and lung cancer (Table 1). In bladder carcinoma, LINC00472 functioned as a tumor suppressor by upregulating the expression of programmed cell death protein 4 (PDCD4) [26], and in colorectal cancer cells, LINC00472 directly targeted mir-196a to increase the expression of PDCD4, and thereby, inhibited proliferation and tumor growth [27]. In hepatocellular carcinoma, LINC00472 inhibited cell proliferation, migration, and invasion by upregulating PDCD4 expression by sponging mir93-5p [28]. In a study of osteosarcoma tumorigenesis, LINC00472 inhibited cancer cell proliferation by upregulating FOXO1 via the downregulation of mir-300 [29]. In breast cancer, LINC00472 was highly expressed in estrogen receptor (ER)-positive cells, and the presence of these cells was associated positively with survival and prognosis as compared with ER-negative cells, and LINC00472 was found to act as a tumor suppressor by inhibiting phosphorylation of nuclear factor $\kappa$ B (NF- $\kappa$ B) [30]. In lung adenocarcinoma, LINC00472 functioned as a ceRNA of mir24-3p and indirectly upregulated the expression of death effector domain-containing DNA binding protein (DEDD) [31]. In addition, LINC00472 has been reported to inhibit metastasis by inhibiting the expression of mir-196b-5p in lung cancer [32], and in an NSCLC study, direct interactions between LINC00472 and mir-149-3p or mir-4270, inhibited metastasis and tumor growth by increasing the expression of killin (KLLN) [18].

Table 2

Sequence data quality check results

Sample name	Total read bases	Total reads	GC (%)	Q20 (%)	Q30 (%)
Control vector	13,306,131,842	132,325,002	40.15	98.81	95.46
LINC00472-oe	11,831,282,397	117,742,244	45.30	98.80	95.55

Total read bases: Total number of sequenced bases after trimming Total reads: Total number of reads after trimming Q20 (%): The proportion of bases with a quality of Phred quality score 20 or higher Q30 (%): The proportion of bases with a quality of Phred quality score 30 or higher.

To confirm that LINC00472 acts as a tumor suppressor in NSCLC cells, CCK-8, trypan blue, cell migration, wound healing, and colony-forming assays were performed on LINC00472-overexpressing NSCLC cells before proceeding with total RNA-seq. Successful transfection of LINC00472 in NCI-H838 was confirmed by PCR (Fig. 1A). CCK-8 assays (Fig. 1B) showed the OD of cells overexpressing LINC00472 (LINC00472-oe) was lower than that of the negative or vector controls ( $p<$ 0.05). Trypan blue assays (Fig. 1C) showed the number of control cells significantly increased with time, while the number of LINC00472-oe cells did not significantly increase as compared with control cells ( $p<$ 0.05), that is, results of the CCK-8 and trypan blue assays showed that cell survival and proliferation were inhibited in the LINC00472-oe cells. After LINC00472 transfection, a colony-forming assay was performed over 12 days (Fig. 1D). We found colony formation ability was significantly poorer for LINC00472-oe cells than for controls ( $p<$ 0.05), which showed LINC00472 inhibits cell proliferation ability and colony formation ability in the long-term. To evaluate cell invasion and migration abilities, wound recovery ability was measured. A 24 h wound healing assay (Fig. 1E) showed that wounds healed more for controls than for LINC00472-oe cells ( $p<$ 0.05). In addition, the cell migration assay (Fig. 1F) showed migration ability was significantly lower for LINC00472-oe cells than control cells ( $p<$ 0.05). Altogether, these results demonstrate that LINC00472 inhibits NSCLC proliferation and invasion and migration abilities.

3.2 Expression profiles of mRNAs in LINC00472-overexpressing NSCLC cells

After transfecting NCI-H838 cells with pCDH-LINC00472 or pCDH empty vector, total RNA-seq was performed to investigate mRNA profile differences. The results of the quality check of sequencing data are presented in Table 2. Cells transfected with control vector had 132,325,002 total reads whereas LINC00472-oe cells had 117,742,244 total reads. Since Phred quality scores (Q20 and Q30), which represent the accuracy of nucleotide sequence analysis, were over 90%, the sequencing data was deemed reliable and used for further analysis. The sequencing data was deposited in the Gene Expression Omnibus database (GEO Series accession number GSE164230).

After performing total RNA-seq, only genes whose expressional values differed by more than 2-fold between LINC00472-oe cells and controls ( $|\log 2$ fold change (FC) $|⩾$ 2) were used to analyze DEG, GO and KEGG pathways. In LINC00472-oe cells, 1,922 genes were upregulated and 1,860 genes were downregulated (Fig. 2A). Gene expression levels are shown as scatter plots in Fig. 2B. Red dots in scatter plots indicate genes with $|\text{FC}|$ of $⩾$ 2.

Figure 2.

Differentially expressed genes as determined by transcriptome sequencing and the results of GO and KEGG enrichment analyses. The sequencing results were registered in the GEO (GSE164230). (A) 1,922 genes were upregulated by $⩾$ 2 fold in LINC00472 overexpressing cells versus cells treated with control vector. 1,860 genes were downregulated by $⩾$ 2 fold in LINC00472 overexpressing cells. Only genes with $p$ value of $<$ 0.05 were counted. (B) The expression levels of DEGs are shown as scatter plots. The red dots correspond to genes with expressions changed by $⩾$ 2 fold ( $|\text{FC}|⩾$ 2, all $p<$ 0.05). (C) The 10 most abundant terms are shown for each of the three categories of GO terms, that is, Biological Process, Molecular Function, and Cellular Component (all $p<$ 0.05). (D) The 32 most enriched pathways are also shown for each of the 5 categories of KEGG pathways: Metabolism, Genetic Information Processing, Environmental Information Processing, Cellular Processes, Organismal Systems, Human Diseases (all $p<$ 0.05).

GO terms and KEGG pathways were analyzed to evaluate DEG functions and associated signaling pathways. GO and KEGG enrichment analyses were used to identify the top 10 terms of GO category and the top 32 KEGG pathways. In the Biological Process GO category, enriched DEGs were associated with the following processes; “cellular process” (2,308 genes, $p=$ 0), “metabolic process” (1,989 genes, $p=$ 0), “cellular metabolic process” (1,747 genes, $p=$ 2.7528E-287), “organic substance metabolic process” (1,749 genes, $p=$ 3.5884E-274), “primary metabolic process” (1,693 genes, $p=$ 7.2305E-263), “single-organism process” (1,990 genes, $p=$ 2.122E-259), “macromolecule metabolic process” (1,511 genes, $p=$ 5.506E-229), “single-organism cellular process” (1,803 genes, $p=$ 3.7305E-223), “cellular macromolecule metabolic process” (1,427 genes, $p=$ 2.6617E-221), and “nitrogen compound metabolic process” (1,218 genes, $p=$ 3.4898E-201) (Fig. 2C). In the Molecular Function GO category, DEGs were associated with; “binding” (2,273 genes, $p=$ 0), “protein binding” (1,793 genes, $p=$ 6.9885E-292), “heterocyclic compound binding” (1,041 genes, $p=$ 1.8045E-156), “organic cyclic compound binding” (1,046 genes, $p=$ 3.7335E-155), “nucleic acid binding” (768 genes, $p=$ 1.7049E-127), “catalytic activity” (937 genes, $p=$ 2.2919E-123), “ion binding” (939 genes, $p=$ 4.2101E-106), “RNA binding” (408 genes, $p=$ 4.1083E-95), “poly (A) RNA binding” (329 genes, $p=$ 3.0806E-83), and “metal ion binding” (626 genes, $p=$ 1.9011E-66). In the Cellular Component GO category, DEGs were associated with; “cell” (2,493 genes, $p=$ 0), “cell part” (2,487 genes, $p=$ 0), “intracellular” (2,363 genes, $p=$ 0), “intracellular part” (2,336 genes, $p=$ 0), “organelle” (2,223 genes, $p=$ 0), “membrane-bounded organelle” (2,117 genes, $p=$ 0), “intracellular organelle” (2,076 genes, $p=$ 0), “intracellular membrane-bounded organelle” (1,936 genes, $p=$ 0), “intracellular organelle part” (1,559 genes, $p=$ 0), and “organelle part” (1,580 genes, $p=$ 0).

KEGG pathway analysis revealed significant gene enrichment in the 32 top pathways (Fig. 2D). Of the 32 top-ranked pathways, “metabolic pathways” (234 genes, $p=$ 6.8992E-98), “pathways in cancer” (81 genes, $p=$ 3.0798E-28), “thermogenesis” (70 genes, $p=$ 2.2092E-39), “human papillomavirus infection” (70 genes, $p=$ 2.1992E-31), “Huntington disease” (63 genes, $p=$ 8.5906E-37), and “PI3K-Akt signaling pathways” (62 genes, $p=$ 4.1818E-24) were the six most enriched. In particular, most genes were associated with “Metabolic pathways”, indicating these pathways were substantially more enriched than other pathways.

Table 3

Selected genes among DEG when LINC00472 is overexpressed

Gene ID	Gene symbol	Description	$\log 2$ fold change	Functions	Ref.
4660	PPP1R12B	Protein phosphatase 1 regulatory subunit 12B	3.308246	Interacting with Bcl-2 to regulate apoptosis	[34]
8490	RGS5	Regulator of G protein signaling 5	3.191714	Major regulators of vascular remodeling and pericyte maturation in tumors	[35]
10181	RBM5	RNA binding motif protein 5	3.170083	Positively regulating of cell death	[36]
5934	RBL2	RB transcriptional corepressor like 2	3.029581	Inducing cellular senescence through DNMT regulation	[37]
3949	LDLR	Low density lipoprotein receptor	2.791828	Associated with many cancers, including liver, lung and breast cancer	[39]
5797	PTPRM	Protein tyrosine phosphatase, receptor type M	2.630373	Regulating a variety of cellular processes including cell growth, differentiation, cell cycle and oncogenic transformation	[40]
80176	SPSB1	splA/ryanodine receptor domain and SOCS box containing 1	$-$ 2.014572	Enhancing cellular immune response by interacting with iNOS	[41]
5111	PCNA	Proliferating cell nuclear antigen	$-$ 2.212547	Components of DNA replication and repair machinery	[42]
100133941	CD24	CD24 molecule	$-$ 2.360281	Affecting immune response, inflammation, autoimmune diseases and cancer	[43]
1020	CDK5	Cyclin dependent kinase 5	$-$ 2.474451	Inducing anti-apoptotic function	[44]
993	CDC25A	Cell division cycle 25A	$-$ 3.189397	One of the cell cycle regulators that remove phosphorylation from CDK	[46]
1978	EIF4EBP1	Eukaryotic translation initiation factor 4E binding protein 1	$-$ 4.266242	A member of the translation inhibitory protein family, mechanical target of mTOR	[47]

3.3 Expression and function analysis of selected DEGs in LINC00472 overexpressing cells

To identify genes that interact with LINC00472, 12 genes were selected from among DEGs identified by total RNA-seq (each $p<$ 0.05, Table 3). Genes were selected with a focus on cellular processes. The keywords used for screening were “Cell proliferation”, “Cell cycle”, “Cell death” and “Cell immunity”, which are associated with cancer development. Protein phosphatase 1 regulatory subunit 12B (PPP1R12B) is one of the regulatory subunits of protein phosphatase 1 (PP1) and plays a role in insulin signaling [33], and is also involved in the regulation of apoptosis by interacting with Bcl-2 [34]. Regulator of G protein signaling 5 (RGS5) is one of the RGS family members and a major regulator of revascularization and pericyte maturation in tumors [35]. RNA binding motif protein 5 (RBM5) is a member of a structurally related group of genes that include RBM6 and RBM10, and functions as a tumor suppressor in lung cancer by regulating cell death [36]. RB transcriptional corepressor like 2 (RBL2) is a member of the rhamnose-binding lectin (RBL) family, exhibits high activity in G0-arrested cells. RBL2 induces cell aging through the regulation of DNA methyltransferase (DNMT), which can promote telomere shortening [37]. Low density lipoprotein receptor (LDLR) is a cell membrane glycoprotein involved in the binding and internalization of circulating cholesterol-containing lipoproteins [38] and is associated with many cancers, including liver cancer, lung cancer, and breast cancer [39]. Protein tyrosine phosphatase, receptor type M (PTPRM) is a member of the protein tyrosine phosphatase family and acts as a signaling molecule that regulates various cellular processes including cell proliferation, anti-apoptosis, differentiation, cell cycle and carcinogenic transformation [40]. SplA/ryanodine receptor domain and SOCS box containing 1 (SPSB1) is a major physiological regulator of inducible nitric oxide synthase (iNOS), and inhibitors of the SPSB-iNOS interaction enhance cellular immune response [41]. Proliferating cell nuclear antigen (PCNA) is a component of the DNA replication and repair machinery and participates in DNA synthesis acceleration by linking the polymerase catalytic unit to DNA [42]. CD24 has carcinogenic properties and affects immune response, inflammation, autoimmune diseases, cancer and is overexpressed in many cancers [43]. Cyclin dependent kinase 5 (CDK5) is activated by cyclin I and increases the expressions of anti-apoptotic Bcl-2 family proteins [44]. In addition, one study showed that CDK5 is associated with the incidence and progression of various cancers, including breast cancer, lung cancer, colon cancer, and pancreatic cancer [45]. Cell division cycle 25A (CDC25A) is a cell cycle regulator that dephosphorylates CDK proteins and positively regulates its activity, thus promoting cell cycle progression [46]. Eukaryotic translation initiation factor 4E binding protein 1 (EIF4EBP1) is a member of the translation inhibitory protein family and upregulation of EIF4EBP1 has been shown in several cancers [47].

Total-RNA seq showed PPP1R12B, RGS5, RBM5, RBL2, LDLR, and PTPRM were upregulated in LINC00472-oe cells, and that SPSB1, PCNA, CD24, CDK5, CDC25A, EIF4EBP1 were downregulated (each $p<$ 0.05, Fig. 3). Twelve DEGs were further investigated using the bioinformatic database Oncomine (http://www.oncomine.org) to analyze their expression patterns in NSCLC tissues. According to database information, PPP1R12B is downregulated $-$ 1.519 fold in lung adenocarcinoma as compared with normal lung tissue [48], RGS5 is downregulated $-$ 2.132 fold in lung adenocarcinoma [49], RBM5 is downregulated $-$ 2.470 fold in lung adenocarcinoma [50], RBL2 is downregulated $-$ 1.405 fold in lung adenocarcinoma [51], LDLR is downregulated $-$ 3.963 fold in lung adenocarcinoma [51], and PTPRM is downregulated $-$ 6.322 fold in lung adenocarcinoma (each $p<$ 0.05, Figs 4A–F) [50]. In contrast, SPSB1 is upregulated 2.149 fold in lung adenocarcinoma [49], PCNA is upregulated 2.322 fold in lung adenocarcinoma [52], CD24 is upregulated 3.209 fold in lung adenocarcinoma [51], CDK5 is upregulated 2.203 fold in lung adenocarcinoma [53], CDC25A was upregulated 1.775 fold in lung adenocarcinoma [51], and EIF4EBP1 is upregulated 2.044 fold in lung adenocarcinoma (each $p<$ 0.05, Figs 4G–L) [51]. PPP1R12B, RGS5, RBM5, RBL2, LDLR, PTPRM, which were upregulated in response to LINC00472 overexpression, were found to be downregulated in NSCLC tissues, and SPSB1, PCNA, CD24, CDK5, CDC25A, EIF4EBP1, which were downregulated in response to LINC00472 overexpression are upregulated in NSCLC tissues. These results suggest that the selected DEGs might interact with LINC00472 and function as potential prognostic markers for NSCLC.

Figure 3.

The expression levels of 12 selected DEGs in the LINC00472 overexpressing NCI-H838 cells as compared with those in cells treated with control vector. Six genes (PPP1R12B, RGS5, RBM5, RBL2, LDLR, and PTPRM) were upregulated, and the other six genes (SPSB1, PCNA, CD24, CDC5, CDC25A, EIF4EBP1) were downregulated (all $p<$ 0.05).

Figure 4.

The expression patterns of 12 selected DEGs in NSCLC tissues based on the Oncomine database. (A) The Landi dataset showed PPP1R12B is downregulated in lung adenocarcinoma (fold change $=$ $-$ 1.519, $p=$ 6.96E-12). (B) The Selamat dataset showed RGS5 is downregulated in lung adenocarcinoma (fold change $=$ $-$ 2.132, $p=$ 1.42E-13). (C) The Bhattacharjee dataset showed RBM5 is downregulated in lung adenocarcinoma (fold change $=$ $-$ 2.470, $p=$ 0.002). (D) The Hou dataset showed RBL2 is downregulated in lung adenocarcinoma (fold change $=$ $-$ 1.405, $p=$ 5.75E-11). (E) The Hou dataset showed LDLR is downregulated in lung adenocarcinoma (fold change $=$ $-$ 3.963, $p=$ 9.582E-18). (F) The Bhattacharjee dataset showed PTPRM is downregulated in lung adenocarcinoma (fold change $=$ $-$ 6.322, $p=$ 8.48E-7). (G) The Selamat dataset showed SPSB1 is upregulated in lung adenocarcinoma (fold change $=$ 2.149, $p=$ 6.61E-17). (H) The Stearman dataset showed PCNA is upregulated in lung adenocarcinoma (fold change $=$ 2.322, $p=$ 1.39E-6). (I) The Hou dataset showed CD24 is upregulated in lung adenocarcinoma (fold change $=$ 3.209, $p=$ 4.73E-10). (J) The Su dataset showed CDK5 is upregulated in lung adenocarcinoma (fold change $=$ 2.203, $p=$ 9.26E-6). (K) The Hou dataset showed CDC25A is upregulated in lung adenocarcinoma (fold change $=$ 1.775, $p=$ 5.61E-10). (L) The Hou dataset showed EIF4EBP1 is upregulated in lung adenocarcinoma (fold change $=$ 2.044, $p=$ 1.72E-11).

4. Discussion

According to recent cancer studies, LINC00472 acts as a tumor suppressor. However, despite strong evidence supporting this role, few studies have investigated LINC00472-related mechanisms in NSCLC. In this study, to identify genes that interact with LINC00472 in NSCLC, mRNA expressions were profiled by transcriptome sequencing analysis of a LINC00472 overexpressing NSCLC cell line.

We prepared an NSCLC cell line overexpressing LINC00472 by transient transfection. As a result of measuring the cell viability, proliferative capacity, and migration capacity, it was confirmed that LINC00472 has a tumor-suppressive function in NSCLC cells. Transcriptome sequencing of empty vector controls and LINC00472-oe cells resulted in the screening of 3,782 genes as DEGs. Many genes were found to be related to metabolism by GO and KEGG pathway analyses. The gene regulations of intracellular metabolism have been linked to many cancers. For example, spermidine/ spermine N1-acetyltransferase 1 (SAT1), which was not included in the list of 12 selected DEGs but was upregulated by 3.64 fold in LINC00472-oe cells, functions as a rate-limiting enzyme in the catabolic pathway of polyamine metabolism and acetylates spermidine and spermine [54], and disruption of polyamine metabolism by abnormal SAT1 expression is associated with cancer [55]. In a study of the mammalian cell line 293T, it was demonstrated that SAT1 overexpression rapidly depleted intracellular spermidine and spermine, and thus, inhibited cell growth and the induction of mitochondrial apoptosis [56]. In addition, it has been reported that p53 activates the expression of SAT1, and that activated SAT1 induces lipid peroxidation resulting in ferroptosis caused by ROS-induced oxidative stress [57]. SAT1 was included in KEGG “metabolic pathways” and in GO “metabolic process”. Adenylosuccinate lyase (ADSL) provides another example. ADSL is involved in the synthesis of de novo purine, and ADSL deficiency leads to defects in the metabolism of purines [58], which are the basic building blocks of nucleotides. In actively dividing cancer cells, the de novo purine biosynthetic pathway is required to supplement purines [59], and thus, the expression of ADSL may be related to tumor growth. In fact, ADSL has been shown to be upregulated in several cancers [60, 61, 62]. In this study, ADSL was included in KEGG “metabolic pathways” and GO “metabolic process”. Also, ADSL was downregulated $-$ 3.07 fold in LINC0472-oe cells, although it was not included among the 12 selected DEGs. Summarizing, our findings suggest that the anticancer effect of LINC00472 is strongly related to the expressional regulations of genes involved in various metabolic pathways that contribute to cancer development.

Twelve DEGs (PPP1R12B, RGS5, RBM5, RBL2, LDLR, PTPRM, SPSB1, PCNA, CD24, CDK5, CDC25A, EIF4EBP1) selected based on cell cycle, survival, and immunity were not included in KEGG “metabolic pathways”, but were included in GO “metabolic process”. In LINC00472-oe cells, genes PPP1R12B, RGS5, RBM5, RBL2, LDLR and PTPRM were upregulated, but are downregulated in NSCLC tissues. These genes are expected to function as biomarkers as tumor suppressors in lung cancer. In fact, it has been reported that tumor cells exhibiting high RGS expression in NSCLC have a low risk of lymph node metastasis and vascular invasion [63]. RBM5 is a tumor suppressor gene in lung adenocarcinoma cells and induces apoptosis and autophagy, inhibits tumor growth, and improves chemosensitivity [64]. On the other hand, reduced RBM5 expression in an in vivo study on RBM5 loss-of-function mice was associated with more aggressive lung cancer in terms of cell numbers and tumor size [65]. RBL2 has also been reported to suppress tumor cell proliferation in vitro [66]. A study on lung cancer and mesothelioma suggested that RBL2/p130 is a direct substrate of AKT kinase and that it is required for AKT inhibition-induced apoptosis [67]. In lung cancer, PTPRM dephosphorylated STAT3 and its overexpression inhibited cell growth [68]. Although PPP1R12B and LDLR have not been sufficiently studied in lung cancer, there is evidence that they function as tumor suppressors in other cancers. PPP1R12B inhibited tumor growth and metastatic capacity by deregulating Grb2/PI3K/Akt signaling in colorectal cancer [69], and LDLR overexpression reduced cell proliferation in cervical cancer [70]. Consequently, it appears PPP1R12B, RGS5, RBM5, RBL2, LDLR, and PTPRM act as tumor suppressors in several cancers, and that LINC00472 might act as a tumor suppressor by upregulating the expressions of these genes.

In contrast, SPSB1, PCNA, CD24, CDK5, CDC25A, and EIF4EBP1 were downregulated in LINC00472-oe cells but upregulated in NSCLC tissues. These genes are expected to function as oncogenic biomarkers in lung cancer. For example, in NSCLC, PCNA promoted cell proliferation and tumorigenesis and inhibited cell death by phosphorylating STAT3 [71], which contrasts to the activities of PTPRM, which dephosphorylates STAT3 in lung cancer cells. As regards CD24, its downregulation in lung cancer inhibited metastasis of lung cancer cells to bone in vivo [72]. High CDK5 expression in lung cancer was reported to be responsible for poorer prognosis, and CDK5 was also found to promote proliferation and metastasis of lung cancer cells [73], and to upregulate a transcriptional coactivator with a PDZ-binding motif (TAZ) to activate the Hippo signaling pathway, and thus, increasing radiation resistance of lung cancer cells [74]. On the other hand, reduced expression of CDC25A targeted by mir-365 improved the radiation sensitivity of NSCLC cells [75]. Finally, the expression of phosphorylated EIF4EBP1 was reported to be significantly correlated with poor prognosis in SCLC [76]. In contrast to EIF4EBP1 in the nucleus responsible for translation inhibition, EIF4EBP1 overexpressed in the cytoplasm has been reported to promote tumor growth by activating a hypoxic response to convert cap-dependent translation into cap-independent translation [47]. Studies on SPSB1 in lung cancer are lacking, but it has been reported to function as an oncogene in several cancers. SPSB1 has been associated with breast cancer recurrence [77]. and the survival of cancer cells in ovarian cancer [78]. Taken together, SPSB1, PCNA, CD24, CDK5, CDC25A, and EIF4EBP1 act as oncogenes in various cancers, and LINC00472 is expected to downregulate the expressions of these genes, and thereby, inhibit tumor growth.

Our mRNA profiling data indicate that LINC00472 regulates the expressions of genes that are important in various signaling pathways, such as metabolic pathways, involved in metabolism, cell cycle, and tumorigenesis. Analysis of selected DEGs confirmed that several genes expressionally regulated by LINC00472 are involved in various cancer-related signaling pathways. Consequently, our findings indicate LINC00472 acts as a tumor suppressor by increasing the expressions of tumor suppressor genes and decreasing the expressions of oncogenes.

The selected DEGs regulated by LINC00472 are expected to interact closely with LINC00472, but, the molecular mechanisms of these interactions, including intracellular interactions, need to be demonstrated experimentally in vitro and in vivo. Nevertheless, the profiling of mRNA expressions in LINC00472 overexpressing NSCLC cells provides a blueprint for the function and role of LINC00472 in NSCLC cells. We believe this will considerably aid understanding of the function of LINC00472 in NSCLC cells and provide guidance for biochemical studies aimed at elucidating signaling pathways in cancer involving LINC00472.