Exploring Molecular Pathways in Metastasis of Breast Cancer Cells to the Brain: Insights from Gene Expression Analysis and Cellular Origins

Abstract

Introduction:

Breast cancer (BC) metastasis to the brain poses a significant challenge in improving outcomes for BC patients. This study aimed to identify the biological mechanisms and associated genes involved in BC cell metastasis to the brain.

Materials and Methods:

Genome-wide gene expression profiles (GSE42568 and GSE43837) generated on a microarray platform were obtained from the GEO database. GSE42568 included 104 BC and 17 normal breast biopsies, while GSE43837 comprised 19 human epidermal growth factor receptor 2 (HER2)-positive human BCs with brain metastases and 19 HER2-positive nonmetastatic primary BCs. Differential gene expression and protein interaction analysis were performed to identify significant gene clusters associated with BC metastasis to the brain.

Results:

Two significant clusters of genes were identified in BC. The first cluster included genes such as EEF1A1, RPL8, RPS3A, RPN2, PSMD2, RPL18, RPL7, and RPS3. The second cluster included genes such as ASPN, COL6A3, COL3A1, LUM, VCAM1, FASLG, GBP1, CXCL9, DAXX, H3F3A, H3F3B, and KMT2D. The Kyoto Encyclopedia of Genes and Genome pathway analysis revealed the significance of genes in the second cluster in pericentric heterochromatin assembly, telomeric heterochromatin assembly, and collagen type III trimer. The expression patterns of 15 hub genes in histopathology specimens and immunofluorescence microscopy images of BC tissue samples exhibited varying expression levels.

Conclusion:

The study identified specific genes and biological mechanisms associated with BC metastasis to the brain. The findings provide insights into the processes involved in cancer genesis and development. Understanding these mechanisms can aid in the development of targeted treatment strategies for BC patients.

INTRODUCTION

Breast cancer (BC) is the second-most common reason of cancer death among women.[1,2] BC initiates due to abnormal alterations of cell growth leading to tumor development. According to recent statistics from the World Health Organization, BC is the most commonly diagnosed cancer in women globally, with an estimated 2.3 million new cases diagnosed in 2020 alone. It is also the leading cause of cancer-related deaths among women, accounting for approximately 685,000 deaths in the same year.[3] In 2018, recent estimations have reported approximately 2.1 million cancer patients all over the world. BC makes up around 21.8% of all cancer occurrences in Saudi women. BC is considered the ninth leading cause of death in women in Saudi Arabia,[4] and has been reported as the second-most frequently occurring cancer.[5] The number of BC estimations will increase proportionally to community growth and age in Saudi Arabia over the coming years.[6] The current treatment options for BC include a combination of surgery, radiation therapy, chemotherapy, targeted therapy, and hormone therapy. The choice of treatment depends on various factors, such as the stage of cancer, hormone receptor status, human epidermal growth factor receptor 2 (HER2) status, and individual patient characteristics.[7]

Several studies have concentrated on inferring the molecular mechanisms of BC.[8] Multiple genetic alterations have been recognized as causative factors of functional defects in BC cells. Many receptors and the associated signaling pathways are considered prominent elements that impact BC development.[9] In addition, numerous oncogenes and anti-oncogenes are implicated.[10] Gene expression alteration can be effectively investigated by microarray data analysis, which is a highly sensitive and high throughput approach. Consequently, the expansion of microarray analysis will enhance the understanding of BC diagnosis and treatment.[11] Inherited mutations are responsible for around 5%–10% of all cases of BC. The most frequent causative genes are BRCA1 and BRCA2,[12] which increase the probability of hereditary BC.[13] The possibility of BC progression escalates up to 55%–65% in women carrying a mutated BRCA1. Moreover, the BC risk increases to 45% if BRCA2 is mutated.[14] Mutations in some genes such as STK11, TP53, ATM, PTEN, CDH1, CHEK2, and PALB2 are associated with the development of BC.[15] Nongenetic risk factors may initiate BC development, including family history, race/ethnicity, breast lesion, radiation exposure, hormones, overweight/obesity, and lack of physical activity.[16]

BM is one of most frequently occurring adult brain tumors,[17] affecting >25% of cancer patients. In the later stage of the disease, 10%–15% of patients with BC develop BM.[18] The estimation of BM incidence ranges from 7 to 14/100,000.[19] Studies have shown that BC metastasis to the brain involves complex interactions between cancer cells and the brain microenvironment. Factors such as angiogenesis, immune response, signaling pathways, and specific genes have been implicated in this process.[20,21] Oncogenes involved in BC to BM cascade can be classified into three types: initiation, progression, and virulence. Initiation genes include CDH2, TWIST, matrix metalloproteinases (MMPs), and vascular endothelial growth factor (VEGF). CDH2 and TWIST enhance detachment, whereas MMPs cause degeneration of the extracellular matrix (ECM). VEGF plays a role in angiogenesis, and PTGS2 and MMP-1 are involved in tumor progression. The PTGS2 and MMP-1 are implicated in the colonization of metastases by the extravasation of tumor cells. The virulence genes include interleukin 6 and tumor necrosis factor (TNF) α, which support cell survival during circulation and promote growth in the distant microenvironment. KAI-1, BRSM1, and NME1 are suppressor genes that inhibit tumor cell dissemination.[22] Patients with BM (brain metastases) have a bleak prognosis, as they have a median survival rate of approximately 7.8 months.[23]

Targeted therapy in BC involves the use of drugs that specifically target molecules involved in the growth and spread of cancer cells, like HER2-positive BCs, where drugs such as trastuzumab (Herceptin) and pertuzumab (Perjeta) block the HER2 protein. Identifying the genes associated with BC metastasis, particularly to the brain, is crucial as it can provide valuable insights into the underlying biological mechanisms driving cancer cell spread. This knowledge may potentially lead to the development of targeted therapies or interventions aimed at preventing or treating metastasis, ultimately improving patient outcomes and reducing the burden of metastatic BC.[24] Therefore, the aim of this study was to identify common genes in breast/metastatic brain cancer datasets involved in BC metastasis to the brain using microarray analysis. Enrichment analysis is used to determine the annotations for the common and exclusive genes of both datasets. The-protein-protein interactions (PPI) network analysis is applied to the common genes in both datasets and exclusive genes of metastatic brain cancer. Cluster identification is used to determine the common genes in both datasets and of metastatic brain cancer. Therefore, gene ontology (GO) analysis of the gene clusters is important.

MATERIALS AND METHODS

Data source

For the current research, two gene expression profiles (GSE42568 and GSE43837) were chosen and retrieved from the GEO database. The GSE42568 dataset utilized the GPL570 Affymetrix Human Genome U133 Plus 2.0 Array and included 104 BC samples and 17 normal breast biopsies. The platform for GSE43837 is the GPL1352 Affymetrix Human X3P Array. The expression profile was obtained by the analysis of 19 HER2-positive human BCs with brain metastases and 19 HER2-positive nonmetastatic primary human BCs.

Preprocessing of data and analysis of differentially expressed genes

The R affy package was utilized for analyzing the data, employing the Robust Multi-array average algorithm. The raw data underwent background correction, quantile normalization, and probe summarization to obtain expression values. To measure the significance of differentially expressed genes (DEGs) between GSE42568 and GSE43837 datasets, a paired t-test was performed with | log2 FC | 1.5 set as the cutoff value for DEG screening. The common DEGs between the two datasets were obtained through the Venn package in R.

Gene ontology enrichment analysis of differentially expressed genes

GO analysis and DEGs functional enrichment analysis were conducted. These analyses included enrichment for biological, cellular, and molecular GO, as well as the Kyoto Encyclopedia of Genes and Genome (KEGG) and Reactome pathways, and were performed using ClueGo on the Cytoscape software (version 3.7.2, Cytoscape Consortium, Seattle, Washington, United States). The analyses were conducted for the common genes between GSE42568 and GSE43837 datasets, unique BM and BC genes, and the modules of combined common and unique metastases genes.

Protein–protein interactions network construction and identification of gene clusters

The STRING database was used to analyze the PPI of DEGs identified as common genes. Unique BM genes were separately mapped to the databases like STRING for estimating the associations of PPI. The PPI network of combined common and unique metastases genes was also obtained. PPIs were extracted with an interaction score of > 0.4. The PPI network clusters of combined common and unique metastases genes were identified by conducting plug-in MCODE using the Cytoscape software (version 3.7.2).

Histopathological studies

Tissue samples from BC patients were obtained from the Protein Atlas (www.proteinatlas.org) database. The 15 genes of interest (EEF1A1, PSMD2, RPL7, ASPN, COL3A1, VCAM, H3F3B, GBP1, H3F3A, CXCL9, KMT2D, FASLG, DAXX, LUM, and COL6A3) were selected based on their relevance to various biological processes and diseases, including BC. The Protein Atlas provides high-resolution images of gene expression patterns in various tissues and cell types.

The immunofluorescence protein expression data for each query gene were analyzed using the Protein Atlas web-based tool. The tool provides information on the expression levels of each gene in different tissue types, including normal and cancerous tissues. A scoring system was used to measure expression levels, ranging from 0 (indicating no expression) to 5 (indicating high expression).

RESULTS

Identification of differentially expressed genes

We analyzed a total of 1262 DEGs from the GSE42568 database, consisting of 486 upregulated genes and 776 downregulated genes. In addition, we examined 77 DEGs from the GSE43837 database, comprising 26 upregulated genes and 51 downregulated genes. Overall, 15 common genes overlapped between the GSE42568 and GSE43837 databases [Figures 1 and 2].

Figure 1.

(a) Volcano plot of dataset 1 (normal breast/breast cancer [BC]), and (b) volcano plot of dataset 2 (BC versus BC with brain metastasis)

Figure 2.

Venn diagram showing 15 common differentially expressed genes (DEGs) between the two expression profile datasets (GSE42568 and GSE43837). Different colored areas display different datasets. The overlapping area represents the common DEGs. DEGs were identified using the t-test; statistically significant DEGs were defined with (log2FC) >1.5 as the cutoff criterion

Protein–protein interactions network analysis

After analyzing a total of 1339 genes from both the BC and BM datasets, we examined 15 common DEGs using the DEGs PPI network complex [Figure 2]. The analysis revealed a network of 15 nodes and eight edges [Tables 1-3 and Figure 3a]. The 62 unique BM genes examined through the DEGs PPI network complex showed 59 nodes and 64 edges [Figure 3b]. The 77 common genes were examined in a combined PPI network complex, which revealed 74 nodes and 104 edges [Figure 3c].

Table 1.

Top 10 of up- and downregulated genes of breat cancer dataset

Gene symbol	Gene name	LogFC	P	Adjusted P	Category
KRT19	Keratin 19	4.5998732	7.35E-17	4.20E-15	Upregulated gene
EPPK1	Epiplakin 1	4.370685	1.39E-19	1.20E-17	Upregulated gene
COL11A1	Collagen type XI alpha 1 chain	4.3041204	2.92E-14	1.09E-12	Upregulated gene
EPCAM	Epithelial cell adhesion molecule	4.2491246	2.35E-21	2.71E-19	Upregulated gene
ESRP1	Epithelial splicing regulatory protein 1	4.2442326	5.60E-27	1.64E-24	Upregulated gene
CDH1	Cadherin 1, fizzy and cell division cycle 20 related 1	4.1864842	2.83E-19	2.31E-17	Upregulated gene
EPPK1	Epiplakin 1, desmoplakin	4.1317431	1.30E-18	9.74E-17	Upregulated gene
DSP	Desmoplakin, dentin sialophosphoprotein	3.9872277	1.67E-25	3.68E-23	Upregulated gene
CD24	CD24 molecule	3.9286253	3.36E-15	1.48E-13	Upregulated gene
TFAP2A	Transcription factor AP-2 alpha	3.7988358	1.45E-21	1.74E-19	Upregulated gene
FABP4	Fatty acid-binding protein 4	−6.1588333	5.44E-28	1.98E-25	Downregulated gene
LEP	Leptin	−6.0406275	4.05E-25	8.45E-23	Downregulated gene
RBP4	Retinol-binding protein 4, RNA polymerase II subunit D	−5.9984488	3.95E-26	9.87E-24	Downregulated gene
ADH1B	Alcohol dehydrogenase 1B (class I), beta polypeptide	−5.7063392	3.15E-16	1.63E-14	Downregulated gene
ADH1B	Alcohol dehydrogenase 1B (class I), beta polypeptide	−5.6393477	2.65E-19	2.18E-17	Downregulated gene
CD36	CD36 molecule	−5.4911813	1.81E-17	1.13E-15	Downregulated gene
LPL	Lipoprotein lipase, lymphocyte cytosolic protein 1	−5.4491679	4.15E-19	3.32E-17	Downregulated gene
CD36	CD36 molecule	−5.3666825	1.77E-20	1.77E-18	Downregulated gene
FHL1	Four and a half LIM domains 1, complement factor H, hemophagocytic lymphohistiocytosis 1	−5.3010429	2.82E-27	8.90E-25	Downregulated gene
PCK1	Phosphoenolpyruvate carboxykinase 1	−5.1381722	1.58E-35	2.78E-32	Downregulated gene

Table 3.

Top 10 common genes according to degree of protein interactions

Gene symbol	Betweenness centrality	Closeness centrality	Degree	Eccentricity	Neighborhood connectivity
MMP1	0.6	0.75	4	2	2.25
COL1A2	0.43333333	0.75	4	2	2.5
POSTN	0.1	0.66666667	3	2	3.33333333
SULF1	0	0.5	2	3	3.5
TNFAIP8	0	0.46153846	1	3	4
CXCL10	0	0.46153846	1	3	4
COL10A1	0	0.46153846	1	3	4
PPM1H	0	0	0	0	0
HBB	0	0	0	0	0
ATP1A2	0	0	0	0	0

Figure 3.

Protein–protein interactions networks of (a) BC unique genes, (b) BC with Brain metastasis unique genes, and (c) Shared genes of BC and BC with brain metastasis

Table 2.

Top 10 of up- and downregulated genes of breast cancers with brain metastasis

Gene symbol	Gene name	LogFC	P	Adjusted P	Category
POSTN	Periostin	2.9196403	0.0001647	0.1600922	Upregulated gene
ASPN	Aspirin	2.6038626	0.0014545	0.173288	Upregulated gene
LUM	Lumican	2.3230211	0.0002723	0.1600922	Upregulated gene
COL6A3	Collagen type VI alpha 3 chain	2.2496997	0.0015194	0.1734185	Upregulated gene
COL3A1	Collagen type III alpha 1 chain	2.2488771	0.0007105	0.1603132	Upregulated gene
SFRP2	Secreted frizzled-related protein 2	2.207059	0.0019966	0.1734185	Upregulated gene
MUCL1	Mucin-like 1	2.1961584	0.0216071	0.2779443	Upregulated gene
SULF1	Sulfatase 1	2.1029217	0.0014886	0.1734185	Upregulated gene
COL10A1	Collagen type X alpha 1 chain	1.9544767	0.0007472	0.1603132	Upregulated gene
TNFAIP8	TNF alpha-induced protein 8	1.9319549	0.0002881	0.1600922	Upregulated gene
TUBB2B	Tubulin beta 2B class IIb	−2.6093223	0.0045415	0.1989037	Downregulated gene
RPS3	Ribosomal protein S3	−2.4931131	0.0003292	0.1600922	Downregulated gene
TUBB2B	Tubulin beta 2B class IIb	−2.3891195	0.0030103	0.1834509	Downregulated gene
RPL7	Ribosomal protein L7	−2.334293	0.0155453	0.2553604	Downregulated gene
PPM1H	Protein phosphatase, Mg2+/Mn2+dependent 1H	−2.2755322	3.09E-05	0.1505735	Downregulated gene
KDELR2	KDEL endoplasmic reticulum protein retention receptor 2	−2.2004147	0.0019326	0.1734185	Downregulated gene
BTBD3	BTB domain containing 3	−2.1423961	0.0009079	0.1625296	Downregulated gene
PPM1H	Protein phosphatase, Mg2+/Mn2+-dependent 1H	−2.1233854	6.87E-05	0.1505735	Downregulated gene
DAZAP1	DAZ-associated protein 1	−2.0532062	0.0013438	0.1702637	Downregulated gene
MAGI1	Membrane-associated guanylate kinase, WW and PDZ domain containing 1, CNKSR family member 3	−2.0096104	0.0029632	0.1834509	Downregulated gene

We used Cytoscape to generate the PPI network and performed module analysis using the MCODE plug-in. Two significant clusters, labeled clusters 1 and 2, were extracted from the PPI network [Figure 4a and b]. Cluster 1 contained eight genes, namely, EEF1A1, PSMD2, RPL7, RPL8, RPL18, RPS3, RPS3A, and RPN2, whereas Cluster 2 consisted of 12 genes, namely, ASPN, COL3A1, VCAM, H3F3B, GBP1, H3F3A, CXCL9, KMT2D, FASLG, DAXX, LUM, and COL6A3.

Figure 4.

Clusters of combined common and brain metastasis genes. (a) Cluster 1 and (b) cluster 2

Cluster analysis through gene ontology function plus Kyoto Encyclopedia of Genes and Genomes enrichment pathways

There are three classes of DEGs based on GO analysis, including biological, cellular components, and molecular pathways. For cluster 1, GO functions included GO: 1900022, GO: 1905051, and GO: 0032358. The GO: 1900022 regulates d-erythro-sphingosine kinase activity, whereas the GO: 1905051 maintains regulation of the cellular base excision repair process. The GO: 0032358 is involved in oxidized pyrimidine DNA binding. In contrast, cluster 2 GO functions included GO: 0031508, GO: 0031509, GO: 0005201, GO: 0030021, and GO: 0030021. The GO: 0031508 is responsible for pericentric heterochromatin assembly, and GO: 0031509 has a role in telomeric heterochromatin assembly. The GO term GO: 0005586 facilitates the formation of collagen type III trimer, whereas GO: 0005201 maintains the ECM structural integrity. In addition, GO: 0030021 regulates the composition of the ECM and confers resistance to compression [Tables 4-6].

Table 4.

Top 10 genes enriched for breast cancers with brain metastasis according to degree of protein interactions

Gene symbol	Betweenness centrality	Closeness centrality	Degree	Eccentricity	Neighborhood connectivity
ACTB	0.59975369	0.3372093	9	7	4.66666667
RPL8	0.09400657	0.29896907	8	8	7.5
RPL7	0.09400657	0.29896907	8	8	7.5
RPS3A	0.09400657	0.29896907	8	8	7.5
EEF1A1	0.04655172	0.29	8	8	7
PSMD2	0.01600985	0.24786325	7	9	6.85714286
RPS3	0.00640394	0.25438596	7	9	7.42857143
RPL18	0.00640394	0.25438596	7	9	7.42857143
RPN2	0.19211823	0.25217391	6	9	6.66666667
FASLG	0.27914614	0.24786325	4	9	3.5

Table 6.

Gene ontology analysis of genes unique to breast cancers with brain metastasis

GO ID	GO term	Genes
GO: 0015317	Phosphate: proton symporter activity	SLC25A3
GO: 0031508	Pericentric heterochromatin assembly	H3-3A, H3-3B
GO: 0031509	Telomeric heterochromatin assembly	H3-3A, H3-3B
GO: 0038044	Transforming growth factor-beta secretion	ITGB6
GO: 0061185	Negative regulation of dermatome development	SFRP2
GO: 1900022	Regulation of d-erythro-sphingosine kinase activity	EEF1A1
GO: 1902340	Negative regulation of chromosome condensation	H3-3A, H3-3B
GO: 1905051	Regulation of base excision repair	RPS3
GO: 1905053	Positive regulation of base excision repair	RPS3
GO: 2001272	Positive regulation of cysteine-type endopeptidase activity involved in the execution phase of apoptosis	RPS3
GO: 0005586	Collagen type III trimer	COL3A1
GO: 0034685	Integrin alphaV-beta6 complex	ITGAV, ITGB6
GO: 0150002	Distal dendrite	MAP2, METAP2
GO: 0150014	Apical distal dendrite	MAP2, METAP2
GO: 0015317	Phosphate: proton symporter activity	SLC25A3
GO: 0032358	Oxidized pyrimidine DNA binding	RPS3

Table 5.

Top 10 combined common and brain cancer metastasis unique genes according to degree of protein interactions

Gene symbol	Betweenness centrality	Closeness centrality	Degree	Eccentricity	Neighborhood connectivity
ACTB	0.59975369	0.3372093	9	7	4.66666667
EEF1A1	0.04655172	0.29	8	8	7
RPS3A	0.09400657	0.29896907	8	8	7.5
RPL7	0.09400657	0.29896907	8	8	7.5
RPL8	0.09400657	0.29896907	8	8	7.5
RPS3	0.00640394	0.25438596	7	9	7.42857143
RPL18	0.00640394	0.25438596	7	9	7.42857143
PSMD2	0.01600985	0.24786325	7	9	6.85714286
RPN2	0.19211823	0.25217391	6	9	6.66666667
COL1A2	0.43333333	0.75	4	2	2.5

According to the KEGG and Reactome enrichment pathways, cluster 1 genes were associated with nonsense-mediated decay independent of the exon junction complex. The cluster 1 genes control SLITs, ROBO expression, and ROBO signaling receptors. EEF1A1, PSMD, RPL8, RPL7, RPS3A, RPS3, RPL18, and RPN2 were involved in these signaling pathways [Tables 7-10]. In contrast to the enrichment pathways of cluster 2, these genes were related to the ECM proteoglycans in African trypanosomiasis. These genes, including ASPN, COL3A1, VCAM1, LUM, COL6A3, and FASLG, are also involved in interactions of cell surface integrin and ECM organization [Tables 11-13].

Table 7.

Pathway enrichment analysis of breast cancers with brain metastasis unique genes

Pathway ID	Pathway Name	Genes
KEGG: 03010	Ribosome	RPL18, RPL7, RPL8, RPS3, and RPS3A
R-HSA-2408557	Selenocysteine synthesis	RPL18, RPL7, RPL8, RPS3, and RPS3A
R-HSA-975957	NMD enhanced by the EJC	RPL18, RPL7, RPL8, RPS3, and RPS3A
R-HSA-9010553	Regulation of expression of SLITs and ROBOs	PSMD2, RPL18, RPL7, RPL8, RPS3, and RPS3A
R-HSA-376176	Signaling by ROBO receptors	PSMD2, RPL18, RPL7, RPL8, RPS3, and RPS3A
R-HSA-392499	Metabolism of proteins	LYZ, MUCL1, ACTB, DAXX, DDB1, EEF1A1, EPRS1, KDELR2, MSRB2, PSMD2, RPL18, RPL7, RPL8, RPN2, RPS3, TUBB2B, UBE2G1, ZRANB1, and RPS3A
R-HSA-216083	Integrin cell surface interactions	COL3A1, COL6A3, LUM, VCAM1, and ITGB6
R-HSA-3000178	ECM proteoglycans	ASPN, COL3A1, COL6A3, LUM, and ITGB6

NMD: Nonsense-mediated decay, EJC: Exon junction complex, ECM: Extracellular matrix

Table 10.

Gene ontology function of cluster 1

GO ID	GO term	Genes
GO: 1900022	Regulation of D-erythro-sphingosine kinase activity	EEF1A1
GO: 1905051	Regulation of base-excision repair cellular process	RPS3
GO: 0032358	Oxidized pyrimidine DNA binding	RPS3
GO: 0005201	ECM structural constituent	ASPN, COL3A1, LUM, and COL6A3
GO: 0030021	ECM structural constituent conferring compression resistance	ASPN and LUM

ECM: Extracellular matrix

Table 11.

Gene ontology function of cluster 2

GO ID	GO term	Genes
GO: 0031508	Pericentric heterochromatin assembly	H3-3A
GO: 0031509	Telomeric heterochromatin assembly	H3-3B
GO: 0005586	Collagen type III trimer	COL3A1

Table 13.

Signaling pathway enrichment analysis of cluster 2

Pathway ID	Pathway Name	Genes
KEGG: 05143	African trypanosomiasis	VCAM1 and FASLG
R-HSA-3000178	ECM proteoglycans	ASPN, COL3A1, LUM, and COL6A3
R-HSA-216083	integrin cell surface interactions	COL3A1, VCAM1, LUM, and COL6A3
R-HSA-1474244	ECM organization	ASPN, COL3A1, VCAM1, LUM, and COL6A3

ECM: Extracellular matrix

Table 8.

Gene ontology function of genes shared between breast cancers and breast cancers with brain metastasis

GO ID	GO term	Genes
GO: 0000415	Negative regulation of histone H3-K36 methylation	BCOR
GO: 0051941	Regulation of amino acid uptake involved in synaptic transmission	ATP1A2, PER2
GO: 0051946	Regulation of glutamate uptake involved in the transmission of nerve impulse	ATP1A2, PER2
GO: 0070208	Protein heterotrimerization	COL1A2
GO: 0071283	Cellular response to iron (III) ion	B2M, CCNB1
GO: 1903280	Negative regulation of calcium: sodium antiporter activity	ATP1A2
GO: 2000775	Histone H3-S10 phosphorylation involved in chromosome condensation	CCNB1
GO: 0005584	Collagen type I trimer	COL1A1, COL1A2
GO: 0140261	BCOR complex	BCOR
GO: 0008603	cAMP-dependent protein kinase regulator activity	CXCL10, PKIB

Table 9.

Signaling pathway enrichment analysis of common genes

Pathway ID	Pathway name	Genes
KEGG: 04974	Protein digestion and absorption	COL10A1, COL1A2, and ATP1A2
R-HSA-1442490	Collagen degradation	COL10A1, COL1A2, and MMP1
R-HSA-1474228	Degradation of the ECM	COL10A1, COL1A2, and MMP1

ECM: Extracellular matrix

Table 12.

Signaling pathway enrichment analysis of cluster 1

Pathway ID	Pathway Name	Genes
R-HSA-1643685	Disease	RPL8, RPL7, RPS3A, RPS3, PSMD2, and RPL18
R-HSA-975956	NMD independent of the EJC	RPL8, RPL7, RPS3A, RPS3, and RPL18
R-HSA-9010553	regulation of expression of SLITs and ROBOs	RPL8, RPL7, RPS3A, RPS3, PSMD2, and RPL18
R-HSA-376176	signaling by ROBO receptors	RPL8, RPL7, RPS3A, RPS3, PSMD2, and RPL18

NMD: Nonsense-mediated decay, EJC: Exon junction complex

Protein Atlas analysis

We analyzed the expression patterns of the 15 genes of interest in BC tissue samples using the Protein Atlas database [Figure 5]. The results showed that all 15 genes were expressed in BC tissue, with varying expression levels. Based on the BC tissue’s expression of various genes, they can be categorized into three groups: over-expressed, down-expressed, and moderately expressed genes. ASPN, GBP1, and CXCL9 are identified as over-expressed genes in BC tissue. EEF1A1, RPL7, and VCAM are classified as moderately expressed genes. On the other hand, PSMD2, COL3A1, H3F3B, KMT2D, FASLG, and LUM are identified as down-expressed genes. H3F3A is moderately expressed, and COL6A3 is also moderately expressed in BC tissue. This classification can help identify the molecular biological mechanisms involved in BC development and can be useful in developing targeted treatment strategies for BC patients. The results indicate that some of the 15 genes of interest may be involved in the development and progression of BC. Additional studies are warranted to support these results and to explore the potential therapeutic implications of targeting these genes for BC treatment [Figure 5].

Figure 5.

The Human Protein Atlas provided immunohistochemistry results, including representative images, for the 13 hub genes in both breast cancer (BC) tissues and normal breast tissues. Immunofluorescence image of the EEF1A1 gene showing the corresponding protein in MCF-7 BC cell lines

DISCUSSION

BC is one of the main causes of mortality in women, with brain metastasis (BM) representing a major risk factor for poor prognosis.[25,26,27] Despite significant research efforts, the underlying mechanisms of BC metastasis to the brain remain poorly understood.[25] Progression to BM may result from angiogenesis, immunity, signaling kinase, and genes such as COX2 and ST6GALNAC5. In recent studies, long noncoding RNA was also associated with metastatic BC to the brain.[28,29,30,31,32]

A previous study identified five DEGs, namely, MMP2, MMP11, VCAM1, CXCL12, and MME, implicated in primary brain cancer and BC metastasis to the brain. These genes play essential roles in metastases processes.[33] In this study, VCAM1, a gene that is expressed in cytokine-activated endothelium and encodes a cell surface sialoglycoprotein, was found in cluster 1. VCAM1 protein regulates adhesion between leukocytes and endothelial cells and contributes significantly to signal transduction. It has been associated with atherosclerosis and rheumatoid arthritis progression, and three alternative splice variants have been reported, which encode several isoforms. VCAM1 contributes to leukocyte circulation from blood to tissue during the inflammatory response and may be expressed abnormally in multiple types of cancers. It includes BC, melanomas, and gastric cancer.[34] Thus, VCAM1 may contribute to the progression of BC and BM.

A recent study has identified 12 differentially expressed collagen genes between metastatic brain cancer and primary BC. These genes are COL1A1, COL15A1, COL1A2, COL3A1, COL6A3, COL14A1, COL6A1, COL5A1, COL8A2, COL6A2, COL5A2, and COL10A1. Each of these genes was expressed considerably more in BC than in BM. Collagen shows for 35% of all the proteins produced by humans.[35] Collagen I, II, III, V, plus Collagen IX levels were increased at BC progression.[36] In our study, we also identified four collagen genes, namely, COL1A2, COL3A1, COL6A3, and COL10A1. Abnormal expression of collagen genes may have a contribution to metastatic brain cancer development in patients with BC.

Another study supports the concept that EEF1A1 may be pro-tumorigenic for particular cancers, including kidney cancer, liver cancer, gliomas, and glioblastomas. EEF1A1 expression is more likely to be a good prognostic indicator for several cancer types, including breast, brain, lung, and kidney cancers, and EEF1A1 has also been commonly described as a pro-apoptotic protein.[37] In this study, EEF1A1 was detected and may have been identified as either CCS3, EE1A1, EF1A, PTI1, or CCS-3. Other names such as LENG7, EEF1A, GRAF-1EF, EEF-1, EF-Tu, and eEF1A-1 may also be associated with EEF1A1. In addition, the alpha domain of the elongation factor-1 is encoded by EEF1A1. Alpha-1 isoform facilitates aminoacyl tRNA distribution to the ribosome. Its expression is widely demonstrated in the brain, placenta, lungs, liver, and kidney.

In this study, we also identified 12 genes, ASPN, FASLG, H3-3A, H3.3B, LUM, PSMD, RPL8, RPL7, RPS3A, RPS3 RPL18, and RPN2. PSMD2 is a type of proteasome that spreads among eukaryotic cells at high concentrations. They degrade peptides in a nonlysosomal pathway specifically through an adenosine triphosphate/ubiquitin-dependent manner. The modified proteasome has a particular role as an immunoproteasome. The immunoproteasome refers to the preparation process of class I major histocompatibility complex peptides. The immunoproteasome also contributes to the TNF signaling pathway. PSMD2 demonstrates ubiquitous expression in the brain (RPKM 40.8). RPL8, RPL7, RPS3A, RPS3 RPL18, and RPN2 are ribosomal proteins identified in our study, and they may have a role in brain metastases, with PSMD2, because they are involved in the regulation of SLIT and ROBO expressions.

ASPN encodes a cartilage extracellular protein, which belongs to the leucine-rich proteoglycan family, and contributes to chondrogenesis regulation. In cartilage, it acts as a suppressor for transforming growth factor-beta 1.[38] It connects calcium and collagen to promote collagen mineralization. ASPN polymorphism is associated with osteoarthritis susceptibility through the aspartic acid repeat domain and is also associated with intervertebral disc disease. Several transcript variants are initiated as a consequence of alternative splicing of ASPN.

The FASLG gene encodes a transmembrane protein that is essential for inducing apoptosis initiated by binding to FAS. The FAS/FASLG signaling pathway has a basic role in immune system regulation as it regulates activation-induced cell death which occurs in T-cells. It also initiates cytotoxic T-lymphocyte-induced cell death. In addition, FASLG has been implicated in the progression of several cancers and may be correlated with brain metastases from BC. H3-3A and H3-3B genes control the nucleosome composition of the chromosomal fiber. These genes comprise introns, and their mRNA is polyadenylated, unlike that of most histone genes. The encoded protein is a replication-independent element of the histone H3 group. As histone modification indicates epigenetic mechanisms, H3-3A and H3-3B genes might have a potential association with BC and the development of BM. LUM was upregulated in this study, which encodes a unit of the SLRP cascade. This involves keratocan, osteoglycin, biglycan, decorin, epiphycan, and fibromodulin. The protein attaches to collagen fibrils in these bifunctional molecules. They are maintaining interfibrillar spacing through increased charged hydrophilic glycosaminoglycans. Lumican is a primary keratan sulfate proteoglycan. However, it is found likewise in interstitial collagenous matrices. It also facilitates the regulation of collagen organization and growth.

The study has limitations to consider. First, the small sample size may limit generalizability. A larger sample would provide robust results and a better understanding of genes and mechanisms in BC brain metastasis. In addition, microarray data have biases like limited genome coverage and technical variation sensitivity. To enhance significance and guide future research, suggestions include validating identified genes and their roles using independent datasets and models. Functional studies can investigate gene-specific roles and prognostic potential through in vitro and in vivo experiments. Exploring gene interplay with angiogenesis, immune response, and signaling pathways would advance the understanding of brain metastasis mechanisms. Investigating cross talk between pathways and genes may unveil novel therapeutic strategies for inhibiting metastasis and improving outcomes.

CONCLUSION

In this study, we performed various bioinformatic analyses to determine the potential molecular contributions between BC and metastatic brain cancer. Twenty significant genes, namely, EEF1A1, RPL8, RPS3A, RPN2, PSMD2, RPL18, RPL7, RPS3, ASPN, COL6A3, COL3A1, LUM, VCAM1, FASLG, GBP1, CXCL9, DAXX, H3F3A, H3F3B, and KMT2D, in two PPI clusters were identified. However, further validations are necessary to support these findings. Further studies should involve both in vivo likewise in vitro analyses to validate the results obtained from this study. The genes identified are associated with the structural composition of the ECM, which confers compression resistance, which demonstrates the significance of these gene clusters.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.