This article is mentioned in:

Introduction

Lung cancer is the most common cause of cancer-associated death (1). Every year, ~234,030 patients are newly diagnosed with cancer in the United States, while nearly ~154,050 patients succumb to lung cancer, with the majority of mortalities resulting from metastasis (1,2).

The brain is one of the most common sites of distant metastasis of lung cancer, particularly in lung cancer cases without lymph node involvement (3,4). The incidence of brain metastasis from lung cancer (BMLC) has been reported as 23–65% (5). The prognosis of patients with BMLC is poor, with a median survival of only 4–5 months (6). Over the past decades, several genetic alterations associated with the occurrence of BMLC were identified. For instance, vascular endothelial growth factor (VEGF)-C, a member of the VEGF family (7), was associated with BMLC. Chen et al (8) reported that high expression of VEGF-C is positively associated with brain metastasis in patients with lung cancer. Furthermore, the incidence of BMLC in a VEGF-C-positive expression group was higher compared with that in a negative expression group. In addition, inflammatory chemokines have been identified to be associated with brain metastases. The expression of C-X-C motif chemokine receptor (CXCR)4, receptor of the CXC chemokine ligand 12 (CXCL12), in the primary tumor tissues and distant metastatic lung tumors in the brain was reported to be higher than that in lung cancer patients without distant metastasis (9). In addition, epidermal growth factor receptor (EGFR) mutations and anaplastic lymphoma kinase (ALK) rearrangement are also thought to be associated with the BMLC (10).

Previous studies that focused on the role of genes in the occurrence of BMLC provide information on the molecular mechanisms of BMLC; however, they have remained to be fully elucidated. Recently developed gene expression profiling arrays may be used to assess the expression of thousands of genes simultaneously, providing a tool to comprehensively elucidate the mechanisms of BMLC (11).

Traditionally, lymph node metastasis was considered to be closely associated with spread of tumor cells as the origin of distant metastasis (12). Recently, a study on human colorectal cancer suggested that for most patients, metastasis to lymph nodes and distant sites were of independent origin, which suggested two different lineage associations between lymphatic and distant metastases in colorectal cancer (13). Consistent with this result, the brain was the most frequent site of distant metastasis in patients with lung cancer without lymph node metastasis (4).

In the present study, based on the data of a gene expression chip, the mRNA expression levels were compared between lung cancer with brain metastasis and lung cancer with lymph node metastasis. In addition, pathways and functional annotation was used to determine associations among the differentially expressed genes (DEGs). Furthermore, protein-protein interaction analysis was used to determine modules and hub genes. The results of the present study may enhance the current understanding of the mechanisms of BMLC.

Materials and methods

Expression profile microarray

The dataset of GSE18549, downloaded from Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/geo), is based on the GPL570 platform (Affymetrix Human Genome U133 Plus 2.0 Array). GSE18549 is a dataset containing three lymph node metastases and six brain metastases from lung cancer.

Identification of DEGs

The GSE18549 dataset was divided into two groups, namely the lymph node metastasis group and brain metastasis group. R (version 3.4.4) was used to identify the DEGs. First, background correction and normalization of the microarray data was performed in R and the Limma package (14) was then used to identify the DEGs. Multiple t-tests were used to detect the DEGs with the cut-off criteria of log2|fold change|>1 and adjusted P-value <0.05.

Gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis

GO analysis is a common method to annotate genes and contains three categories: Cellular component (CC), molecular function (MF) and biological process (BP) (15). KEGG analysis may be used to determine the pathways of DEGs between two groups (16). GO and KEGG analyses were performed for the identified DEGs using the Database for Annotation, Visualization and Integrated Discovery (DAVID, version 6.7; http://david.abcc.ncifcrf.gov) (17). P<0.05 and the number of involved genes of ≥2 were selected as cut-off criteria.

Construction of protein-protein interaction (PPI) network and module analysis

The Search Tool for the Retrieval of Interacting Genes (STRING; version 10.5) (18) is an online tool for determining interactions among DEGs. The screened DEGs were mapped using STRING, and a combined score of >0.4 was considered to indicate significance. Cytoscape software (version 3.6.1) was then used to visualize the PPI network (19), and the Network Analyzer (version 2.7), a Cytoscape plugin, was used to compute the basic properties of the PPI network, including average clustering coefficient distribution, closeness centrality, node degree distribution and shortest path length distribution. In addition, module analysis was performed by the plug-in Molecular Complex Detection (MCODE; version 1.5.1) with the following settings: Degree cutoff, 2; node score cutoff, 0.2; k-core, 2; and maximum depth, 100, and the following criteria: MCODE score >4; number of nodes >5. Finally, the hub genes in the PPI network were determined, defined as those with a degree of connectivity of >10.

Results

Identification of DEGs

First, the raw data of GSE18549 were normalized, as presented in Fig. 1A and B. Subsequently, the significant DEGs in BMLC compared with lung cancer with lymph node metastasis were identified. A total of 190 DEGs were identified, which consisted of 129 downregulated genes and 61 upregulated genes. A volcano plot of the differential expression analysis is presented in Fig. 1C. In addition, the top 40 DEGs were displayed in a heat map in Fig. 1D.

Figure 1. Normalization of gene expression data, volcano plot of DEGs and heatmap of Top 40 DEGs in mRNA expression profiling datasets. (A and B) Gene expression profiles (A) prior to normalization and (B) after normalization. (C) Volcano plot of DEGs between lymph node metastasis group and brain metastasis group. DEGs were selected with P<0.05 and log2|FC|>1. The x-axis presents the log2|FC| in gene expression between the lymph node metastasis group and brain metastasis group, and the y-axis displays the statistical significance of the differences. Red dots indicate genes without significantly different expression and blue dots indicate significant DEGs. (D) Representative heat map of the top 40 DEGs in the dataset GSE18549. Each column represents one of 40 genes and each row represents one sample. Green and red color represents high and low expression level, respectively. The color scale on the right represents the raw Z-score ranging from red to green (low to high expression level). FC, fold change; LN Met, lymph node metastasis; Br Met, brain metastasis; DEG, differentially expressed gene.

GO terms and KEGG pathway enrichment analysis

To gain insight into the GO categories of DEGs between the lymph node metastasis group and brain metastasis group, all DEGs were uploaded to the DAVID database. The results suggested that downregulated DEGs were mainly enriched in the category BP, including ‘immune response’, ‘cell activation’ and ‘leukocyte activation’, while upregulated DEGs were significantly enriched in ‘cell division’, ‘DNA repair’ and ‘viral process’. In the category CC, the downregulated DEGs were significantly enriched in ‘plasma membrane’, ‘integral to plasma membrane’ and ‘intrinsic to plasma membrane’, whereas the upregulated DEGs were significantly enriched in ‘nucleoplasm’, ‘cytoplasm’ and ‘condensed nuclear chromosome’. Furthermore, in the category MF, the downregulated genes were mainly enriched in ‘cytokine binding’, ‘nucleotide receptor activity’ and ‘purinergic nucleotide receptor activity’, and conversely, upregulated genes were significantly enriched in ‘poly-(a)-RNA binding’, ‘chromatin binding’ and ‘ATP-dependent helicase activity’. Further details on the results of the GO enrichment analysis are provided in Table I.

Table I. GO analysis of differentially expressed genes in the GSE18549 dataset.

Table I.

GO analysis of differentially expressed genes in the GSE18549 dataset.

A, Downregulated genes
Category/GO term	P-value	Genes
Biological Process
Immune response	2.52×10−20	ITGAL, TNF, CCL2, TRGC2, FASLG, CXCL6, TLR6, CLEC10A, CHIT1, FCRL4, CD96, SH2D1A, LILRA4, POU2F2, MS4A1, LTB, CD27, CD28, CR1, CRTAM, SIT1, GBP5, CR2, PTGER4, CMKLR1, CCL19, TNFRSF17, CD180, IGSF6, CYBB, CCR6, IL18BP, KCNJ8, CCR2, LCP2
Cell activation	1.38×10−12	ITGAL, CRTAM, TNF, CD3G, IKZF1, PLEK, IL21R, TLR6, SLAMF1, GIMAP1, CD48, P2RY12, VCAM1, P2RX7, MS4A1, IRF4, LTB, CD28, LCP2
Leukocyte activation	1.54×10−9	ITGAL, CRTAM, CD3G, IKZF1, IL21R, TLR6, SLAMF1, GIMAP1, CD48, VCAM1, P2RX7, MS4A1, IRF4, LCP2, CD28
Defense response	8.83×10−9	ITGAL, CR1, TNF, CCL2, CR2, CYSLTR1, CCL19, CXCL6, TLR6, CD180, CD48, CD84, SH2D1A, CYBB, P2RX7, CCR6, TRAC, KCNJ8, CCR2, PLA2G2D, AOC3
Lymphocyte activation	1.56×10−8	CD48, VCAM1, ITGAL, P2RX7, CRTAM, CD3G, IKZF1, IL21R, MS4A1, IRF4, SLAMF1, CD28, GIMAP1
Cellular Component
Plasma membrane	4.92×10−11	TRGC2, FASLG, TLR6, DDR2, FCRL4, FCRL3, CD48, ART4, CD96, TRAC, SLC2A3, CXCR6, MS4A1, CSF2RB, RECK, CRTAM, CD3G, SIT1, GBP5, PTGER4, GPR18, CMKLR1, TNFRSF17, THY1, PRKCB, IGSF6, CD84, ARHGAP31, CCR6, CD34, CCR2, PTGDR, ADAM12, AOC3, ITGAL, CD244, TNF, CYSLTR1, CD247, BRSK1, CLEC10A, CSMD2, VCAM1, CD27, CD28, IL2RB, CR1, CR2, PLEK, SLAMF1, CD180, P2RY12, P2RY13, P2RY10, P2RX7, CYBB, LYVE1, KCNJ8, GFRA1
Integral to plasma membrane	1.14×10−8	ITGAL, TNF, CYSLTR1, TRGC2, FASLG, TLR6, DDR2, CD48, CD96, TRAC, CXCR6, MS4A1, CSF2RB, CD27, CD28, IL2RB, CR1, SIT1, CD3G, CMKLR1, THY1, IGSF6, CD84, CYBB, P2RX7, LYVE1, CCR6, KCNJ8, CCR2
Intrinsic to plasma membrane	1.86×10−8	ITGAL, TNF, CYSLTR1, TRGC2, FASLG, TLR6, DDR2, CD48, CD96, TRAC, CXCR6, MS4A1, CSF2RB, CD27, CD28, IL2RB, CR1, SIT1, CD3G, CMKLR1, THY1, IGSF6, CD84, CYBB, P2RX7, LYVE1, CCR6, KCNJ8, CCR2
Plasma membrane part	2.11×10−7	ITGAL, CD244, TNF, CYSLTR1, CD247, TRGC2, FASLG, BRSK1, TLR6, DDR2, CD48, VCAM1, CD96, TRAC, CXCR6, MS4A1, CSF2RB, CD27, CD28, IL2RB, CR1, SIT1, CD3G, GBP5, PLEK, CMKLR1, SLAMF1, THY1, CD84, IGSF6, ARHGAP31, CYBB, P2RX7, LYVE1, CCR6, CD34, KCNJ8, CCR2
Intrinsic to membrane	3.72×10−7	IL21R, TRGC2, FASLG, TLR6, DDR2, FCRL4, FCRL3, CD48, ART4, CD96, TRAC, SLC2A3, LILRA4, CXCR6, MS4A1, CSF2RB, MCOLN2, LTB, RECK, CRTAM, CD3G, SIT1, PTGER4, GPR18, CMKLR1, TNFRSF17, PKHD1L1, THY1, MCTP1, IGSF6, CD84, CCR6, CD34, PTGDR, CCR2, ADAM12, AOC3, ITGAL, CD244, TNF, CYSLTR1, CD247, KMO, CLEC10A, CSMD2, VCAM1, FMO2, CD27, CD28, IL2RB, CR1, CR2, SLAMF1, CD180, GIMAP1, P2RY12, P2RY13, CYBB, P2RY10, P2RX7, LYVE1, RNF150, KCNJ8, ST8SIA4, GFRA1
Molecular Function
Cytokine binding	7.66×10−7	IL2RB, CCR6, IL18BP, CMKLR1, CCR2, IL21R, CXCR6, CSF2RB, GFRA1
Nucleotide receptor activity	1.03×10−4	P2RY12, P2RY13, P2RX7, P2RY10, GPR18
Purinergic nucleotide receptor activity	1.03×10−4	P2RY12, P2RY13, P2RX7, P2RY10, GPR18
Cytokine activity	3.77×10−4	TNF, CCL2, CCL19, FASLG, CXCL6, IL33, GREM1, LTB
Chemokine receptor activity	6.42×10−4	CCR6, CMKLR1, CCR2, CXCR6
B, Upregulated genes
Category/GO term	P-value	Genes
Biological process
Viral process	1.98×10−4	DDX11, BTRC, RBM15B, TSC2, BRD4, TPR, RCC1
RNA processing	2.68×10−3	DHX9, DDX54, HNRNPDL, DHX30
Cell division	1.76×10−2	ANAPC1, BRCC3, TPR, RCC1, SMC1A
Positive regulation of chromatin binding	2.54×10−2	KDM1A, DDX11
DNA repair	2.95×10−2	BRCC3, DDX11, TDP1, SMC1A
Cellular Component
Nucleoplasm	2.85×10−5	ANAPC1, DHX9, BRCC3, ZMYM3, TONSL, BTRC, RBM15B, SNIP1, HNRNPDL, RCC1, BMS1, GTSE1, GPS2, KDM1A, DDX11, GTF2IRD1, BRD4, TPR, PPP4C, SMC1A, HDAC8
Cytoplasm	7.46×10−5	SHROOM3, BTRC, SNIP1, RCC1, TIPRL, ZIC2, DDX11, DLG3, BRD4, TPR, DHX30, PPP4C, SAMD4B, DHX9, BRCC3, ZMYM3, PIK3C2A, TONSL, AMBRA1, HNRNPDL, MID1, GCN1, GTF2IRD1, TDP1, TSC2, USP47, SPTBN1, SMC1A, HDAC8
Condensed nuclear chromosome	2.77×10−3	BRD4, RCC1, SMC1A
Nucleus	2.86×10−3	BTRC, SNIP1, RCC1, BMS1, ZIC2, KDM1A, DDX11, BRD4, DHX30, PPP4C, TPR, SAMD4B, DHX9, PGAP2, BRCC3, PIK3C2A, CIZ1, HNRNPDL, GTF2IRD1, TDP1, TSC2, USP47, SPTBN1, DDX54, SMC1A, HDAC8
Cytoplasmic microtubule	8.97×10−3	SPACA9, MID1, GTSE1
Molecular Function
Poly(A) RNA binding	3.79×10−4	DHX9, RBM15B, SNIP1, SPTBN1, DDX54, HNRNPDL, TPR, DHX30, SMC1A, BMS1, GCN1, SAMD4B
Chromatin binding	9.98×10−4	KDM1A, DDX11, BRD4, TPR, RCC1, DHX30, SMC1A
ATP-dependent helicase activity	3.32×10−3	DHX9, DDX11, DHX30
Protein binding	7.47×10−3	REPS1, BTRC, RBM15B, SNIP1, RCC1, TIPRL, GTSE1, KDM1A, DDX11, ILVBL, DLG3, PEX13, BRD4, TPR, DHX30, PPP4C, DHX9, PGAP2, CHDH, BRCC3, SPACA9, TONSL, CIZ1, AMBRA1, HNRNPDL, MID1, GPS2, ARHGAP32, TDP1, TSC2, USP47, SPTBN1, SMC1A, HDAC8, GATC, IQCE
Nucleic acid binding	7.89×10−3	DHX9, DDX11, CIZ1, RBM15B, GPATCH1, DDX54, HNRNPDL, DHX30, ZIC2

[i] GO, Gene Ontology.

As presented in Table II, the KEGG pathway enrichment analysis of DEGs indicated that the downregulated DEGs were mainly enriched in ‘cytokine-cytokine receptor interaction’, ‘natural killer cell-mediated cytotoxicity’, ‘hematopoietic cell lineage’, ‘chemokine signaling pathway’ and ‘t-cell receptor signaling pathway’, and upregulated DEGs were mainly enriched in ‘oocyte meiosis’.

Table II. KEGG pathway analysis of differentially expressed genes in the GSE18549 dataset.

Table II.

KEGG pathway analysis of differentially expressed genes in the GSE18549 dataset.

KEGG pathway	P-value	Genes
Downregulated genes
Cytokine-cytokine receptor interaction	3.90×10−7	IL2RB, CCL2, TNF, IL21R, FASLG, TNFRSF17, CCL19, CXCL6, CCR6, CXCR6, CCR2, CSF2RB, LTB, CD27
Natural killer cell-mediated cytotoxicity	2.36×10−5	CD48, ITGAL, CD244, SH2D1A, TNF, CD247, FASLG, PRKCB, LCP2
Hematopoietic cell lineage	1.08×10−3	CR1, TNF, CR2, CD3G, CD34, MS4A1
Chemokine signaling pathway	6.98×10−3	CCR6, CCL2, CCR2, CXCR6, CCL19, CXCL6, PRKCB
T-cell receptor signaling pathway	1.70×10−2	TNF, CD3G, CD247, CD28, LCP2
Upregulated genes
Oocyte meiosis	3.89×10−2	ANAPC1, BTRC, SMC1A

[i] KEGG, Kyoto Encyclopedia of Genes and Genomes.

Construction of PPI network and module analysis

The STRING database was used to predict the interaction between the 190 DEGs, and the PPI was visualized using Cytoscape software. Initially, basic properties of the network were computed. The network contained 90 nodes and 205 interaction edges, where the average degree of connectivity (i.e., average number of neighbors) was 4.56 (Fig. 2A). Subsequently, the Network Analyzer tool was used to compute the basic properties of the PPI network, including the degree distribution, clustering coefficient, average shortest path and the closeness centrality of the PPI network (Fig. 3). The analysis indicated that the degree distribution of PPI network nodes followed the power-law of network, and had small-world network characteristics, including a short average shortest path (20,21). In addition, one significant module consisting of 8 nodes and 28 edges was obtained from the PPI network of DEGs using MCODE (Fig. 2B). As provided in Table III, enrichment analysis suggested that the genes in this significant module were mainly associated with functional terms in the category BP, including ‘G-protein-coupled receptor protein signaling pathway’, ‘cell surface receptor-linked signal transduction’, ‘taxis’, ‘chemotaxis’ and ‘locomotory behavior’. In the category CC, the genes in this significant module were significantly enriched in ‘plasma membrane’, and in the category MF, the genes were mainly enriched in ‘C-C chemokine receptor activity’, ‘C-C chemokine binding’, ‘chemokine receptor activity’, ‘chemokine binding’ and ‘nucleotide receptor activity, G-protein coupled’. Furthermore, results from KEGG analysis demonstrated that the genes in this significant module were associated with ‘chemokine signaling pathway’ and ‘cytokine-cytokine receptor interaction’ (Table IV). Finally, the hub genes with a degree of connectivity of >10 were identified, including tumor necrosis factor (TNF), C-C motif chemokine ligand 2 (CCL2), CD34, vascular cell adhesion molecule 1 (VCAM1), CD48, CD27, C-C motif chemokine ligand 19 (CCL19), C-X-C motif chemokine receptor 6 (CXCR6) and C-C motif chemokine receptor 2 (CCR2).

Figure 2. PPI network and a significant module constructed from the DEGs in dataset GSE18549. (A) Using the STRING online database, a total of 190 DEGs were filtered into the PPI network complex. (B) A significant module selected from the PPI network. The nodes represent proteins, the edges represent the interaction of proteins, the green circles represent downregulated genes and the white circles represent the upregulated genes. PPI, protein-protein interaction; DEG, differentially expressed gene.

Figure 3. Topological properties of the protein-protein interaction network. (A) Node-degree distribution; (B) average clustering coefficient distribution; (C) shortest path length distribution; (D) closeness centrality.

Table III. GO analysis of genes in the significant module.

Table III.

GO analysis of genes in the significant module.

Category/GO term	P-value	Genes
Biological Process
G-protein coupled receptor protein signaling pathway	2.67×10−8	P2RY12, P2RY13, CCR6, GPR18, CCR2, CXCR6, CCL19, CXCL6
Cell surface receptor linked signal transduction	9.06×10−7	P2RY12, P2RY13, CCR6, GPR18, CCR2, CXCR6, CCL19, CXCL6
Taxis	5.49×10−5	CCR6, CCR2, CCL19, CXCL6
Chemotaxis	5.49×10−5	CCR6, CCR2, CCL19, CXCL6
Locomotory behavior	2.71×10−4	CCR6, CCR2, CCL19, CXCL6
Cellular Component
Plasma membrane	2.69×10−2	P2RY12, P2RY13, CCR6, GPR18, CCR2, CXCR6
Molecular Function
C-C chemokine receptor activity	2.98×10−5	CCR6, CCR2, CXCR6
C-C chemokine binding	2.98×10−5	CCR6, CCR2, CXCR6
Chemokine receptor activity	7.43×10−5	CCR6, CCR2, CXCR6
Chemokine binding	8.69×10−5	CCR6, CCR2, CXCR6
Nucleotide receptor activity, G-protein coupled	1.00×10−4	P2RY12, P2RY13, GPR18

[i] GO, Gene Ontology.

Table IV. KEGG pathway analysis of genes in the significant module.

Table IV.

KEGG pathway analysis of genes in the significant module.

KEGG pathway	Count	P-Value	Genes
hsa04062:Chemokine signaling pathway	5	8.61×10−6	CCR6, CCR2, CXCR6, CCL19, CXCL6
hsa04060:Cytokine-cytokine receptor interaction	5	3.31×10−5	CCR6, CCR2, CXCR6, CCL19, CXCL6

[i] Hsa, Homo sapiens; KEGG, Kyoto Encyclopedia of Genes and Genomes; CXCL6, chemokine (C-X-C) ligand 6; CCL, C-C motif chemokine ligand; CCR, C-C chemokine receptor.

Discussion

Brain metastasis is a frequent complication in patients suffering from advanced lung cancer (3). It has been estimated that 50% of the patients diagnosed with lung cancer will develop metastatic brain lesions, which results in a dismal prognosis (22). Recently, a series of biomarkers have been identified to be associated with the development of brain metastasis, such as integrins, cell adhesion molecules, cadherins, VEGF, chemokines, matrix metalloproteinase, EGFR mutations and ALK translocations (23). However, the biology of brain metastasis is still poorly understood, as a result, the specific and effective strategies used to control or treat for brain metastasis are currently unavailable.

In the present study, a microarray dataset was analyzed to screen the DEGs between lung cancers with lymph node metastasis and brain metastasis. A total of 190 DEGs were obtained, which included 129 downregulated genes and 61 upregulated genes. To further investigate the functions of the DEGs, GO functional annotation and KEGG pathway enrichment analysis were used based on the DAVID database. The GO analysis demonstrated that downregulated DEGs were mainly associated with ‘immune response’, ‘cell activation’ and ‘leukocyte activation’, which was in accordance with previous studies. For instance, secondary brain cancers frequently exhibit high expression of programmed cell death 1 (PD-1) ligand 1, which may be inhibited by novel treatments that activate the immune system (24,25). Furthermore, degraded white-matter tract integrity in the areas with high T-cell densities are thought to provide active microenvironments for brain metastases (26); in addition, immune checkpoint inhibitors targeting PD-1 and cytotoxic T lymphocyte-associated protein 4 are becoming a frontline therapy in melanoma, which also suggests that the immune response is involved in the development of brain metastases (27). On the contrary, the upregulated DEGs were involved in ‘DNA repair’ and ‘viral process’, which were also consistent with previous data. Overexpression of DNA repair genes BRCA1-associated RING domain 1 and RAD51 recombinase frequently occur in brain metastases from breast cancer (28). The DNA-damage response pathway was also determined to be involved in leptomeningeal metastasis of non-small cell lung cancer (29). Furthermore, high rates of DNA repair mutations were identified in brain metastases from prostate cancer (30). In addition, the prevalence of human cytomegalovirus proteins and nucleic acids is high in primary and metastatic tumors, indicating that this virus may drive the development of metastatic brain tumors (31).

The results of the KEGG analysis indicated that the downregulated DEGs were mainly enriched in ‘chemokine signaling pathway’. Previous studies demonstrated that inflammatory chemokines and their receptors regulate tumor cell migration and participate in tumor growth, metastasis, angiogenesis and invasion through the interaction between mesenchymal cells and neoplastic cells (32,33). Of note, the upregulated DEGs were mainly associated with ‘Oocyte meiosis’. However, there is no evidence to support that this pathway was associated with BMLC or brain metastasis from other types of tumor.

Furthermore, the protein interactions among the screened DEGs were predicted. In the PPI network, 9 hub genes with the highest degree of connectivity were selected, which included TNF, CCL2, CD34, VCAM1, CD48, CD27, CCL19, CXCR6 and CCR2. TNF encodes a multifunctional proinflammatory cytokine that belongs to the TNF superfamily. TNF-α has an important role in the adhesion of non-small cell lung cancer cells to brain endothelium mediated by CD62E (34). A previous study suggested that microRNA (miR)-509 has a critical role in brain metastasis of breast cancer by modulating the Ras homolog family member C/TNF-α network (35). Among the above genes, half of the hub genes, including CCL2, CCL19, CCR2 and CXCR6, were involved in the chemokine signaling pathway. CCL2, a small cytokine that belongs to the CC chemokine family, is anchored in the plasma membrane of endothelial cells by glycosaminoglycan side chains of proteoglycans; CCL2 exhibits a chemotactic activity for monocytes and basophils. miR-19a contained in astrocyte-derived exosomes reversibly downregulates phosphatase and tension homolog in tumor cells, resulting in secretion of CCL2, which is able to recruit brain metastasis-promoting myeloid cells (36). Conceivably, CCR2, as one of the receptors for CCL2, mediated the roles of CCL2 in breast cancer metastasis (37). Similar to CCL2, CCL19 is another member of the CC motif chemokine superfamily and functions as a tumor suppressor in lung cancer; CCL19-expressing fibroblastic stromal cells were indicated to inhibit lung carcinoma growth by promoting local anti-tumor T-cell responses (38). CCL19 exhibited anti-tumor effects by promoting interferon-γ-dependent responses in a lung cancer model (39); however, the roles of CCL19 in BMLC have remained to be determined. CXCR6, belonging to family A of the G protein-coupled receptor superfamily, was identified to be associated with B-lineage maturation and antigen-driven B-cell differentiation. As a binding partner, CXCR6 mediates CXCL16-associated signaling. Recently, several studies indicated that CXCL16/CXCR6 signaling drives metastasis of different cancer types by promoting a protumor inflammatory environment or attracting cancer cells (40,41).

Another group of hub genes, CD, are a defined subset of cellular surface receptors, which include CD48, CD27 and CD34. CD48, also known as B-lymphocyte activation marker or signaling lymphocytic activation molecule 2, encodes a member of the CD2 subfamily of immunoglobulin-like receptors. Activation of CD48 by injecting anti-CD48 monoclonal antibodies resulted in the inhibition of tumor metastasis of melanoma (42). CD27, a member of the TNF-receptor superfamily, is required for generation and long-term maintenance of T-cell immunity; Pagès et al (43) discovered that the expression of CD27 was associated with early metastasis in colorectal cancer. CD34, a single-pass membrane protein, is highly glycosylated and phosphorylated by protein kinase C and expressed on human hematopoietic progenitor cells and the small-vessel endothelium of a variety of tissue types (44); Upregulation of CD34-positive lymphatic/vascular endothelial progenitor cells are associated with metastasis of ovarian cancer (45).

VCAM1 is an Ig-like cell adhesion molecule expressed by cytokine-activated endothelium (46). One study based on RNA sequencing analysis confirmed that the expression of VCAM1 was upregulated in brain tissue harboring metastases, which suggested that this gene may contribute to the establishment of brain metastases from breast cancer or melanoma (47); however, the functions VCAM1 in BMLC have so far remained elusive.

In conclusion, although the present study had certain limitations, including the small number of cases and the lack of validation in clinical samples, the present analysis identified distinct key genes and pathways closely associated with BMLC, which may contribute to the current knowledge of the complex mechanisms of BMLC. Of note, the present results warrant confirmation by further investigations.

Acknowledgements

Not applicable.

Funding

This study was supported by The National Natural Science Foundation of China (grant no. 81501959), The Natural Science Foundation of Liaoning Province (grant no. 20180530080), The Technological Project of Liaoning Province (grant no. 20170540392), The Innovation Foundation for the University Students (grant no. 201510160000013) and The Biological Anthropology Innovation Team Project of JZMU (grant no. JYLJ201702).

Availability of data and materials

The datasets used and/or analyzed in the present study are available from the corresponding authors on reasonable request.

Authors' contributions

FR and PW designed the experiments, and XZ and NW collected and analyzed the data. TC and HG wrote the manuscript. YL and HC downloaded the gene expression profile from the GEO. All authors critically reviewed the content and approved the final version for publication.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References