Microarray and pattern miner analysis of AXL and VIM gene networks in MDA‑MB‑231 cells

  • Authors:
    • Sudhakar Natarajan
    • Venil N. Sumantran
    • Mohan Ranganathan
    • Suresh Madheswaran
  • View Affiliations

  • Published online on: August 20, 2018     https://doi.org/10.3892/mmr.2018.9404
  • Pages: 4147-4155
Metrics: Total Views: 0 (Spandidos Publications: | PMC Statistics: )
Total PDF Downloads: 0 (Spandidos Publications: | PMC Statistics: )


Abstract

MDA‑MB‑231 cells represent malignant triple‑negative breast cancer, which overexpress epidermal growth factor receptor (EGFR) and two genes (AXL and VIM) associated with poor prognosis. The present study aimed to identify novel therapeutic targets and elucidate the functional networks for the AXL and VIM genes in MDA‑MB‑231 cells. We identified 71 genes upregulated in MDA‑MB‑231 vs. MCF7 cells using BRB‑Array tool to re‑analyse microarray data from six GEO datasets. Gene ontology and STRING analysis showed that 43/71 genes upregulated in MDA‑MB‑231 compared with MCF7 cells, regulate cell survival and migration. Another 19 novel genes regulate migration, metastases, senescence, autophagy and chemoresistance. The Pattern Miner systems biology tool uses specific genes as inputs or ‘baits’ to identify outputs from the NCI‑60 database. Using five genes regulating cancer cell migration (AXL, VIM, EGFR, CAPN2, and COL4A1) as input ‘baits’, we used pattern miner to identify statistically significant, co‑expressed genes from the list of 71 genes upregulated in MDA‑MB‑231 compared with MCF7 cells. Outputs were subsets of the 71 genes, which showed significant co‑expression with one or more of the five input genes. These outputs were used to develop functional networks for AXL and VIM. Analysis of these networks verified known properties of AXL and VIM, and suggested novel functions for these two genes. Thus, genes in the AXL network promote migration, metastasis and chemoresistance, whereas the VIM gene network regulates novel tumorigenic processes, such as lipogenesis, senescence and autophagy. Notably, these two networks contain 12 genes not reported for TNBC.

Introduction

Triple negative breast cancers (TNBC) lack expression of three important receptors (ER, PR, and HER2). These cancers account for 10–15% of breast cancers, and are characterized by overexpression of epidermal growth factor receptor (EGFR), high proliferative rate, and mutations in the p53 and BRCA1 tumour suppressor genes (1). TNBC are aggressive, metastatic cancers associated with early age of onset, high relapse rates, and poor clinical outcome (1). The MDA-MB-231 cell line is a model for TNBC, and is classified as a mesenchymal, stem-like, subtype of TNBC (2). TNBC is managed with standard chemotherapy agents since there are no suitable therapeutic targets (1,3). Therefore, it is important to identify new genetic markers and therapeutic targets for TNBC. Accordingly, we re-analysed 6 microarray GEO datasets and identified a common list of 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. Notably, this list contains known and novel genes that regulate the invasive phenotype of MDA-MB-231 cells. It also contain known and novel therapeutic target genes for TNBC.

The AXL and VIM genes are linked with very poor prognosis in TNBC (4). Furthermore, mechanisms of action of the AXL and VIM proteins are unclear. Therefore, elucidation of functional gene networks for AXL and VIM, could provide insights on therapeutic targets for TNBC. As reported earlier (5), our data showed significant upregulation of the AXL and VIM genes in MDA-MB-231 vs. MCF7 cells. Interestingly, pattern miner analysis showed that different subsets of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, were significantly and specifically co-expressed with either AXL or VIM. These co-expressed gene subsets were used to model distinct functional networks for AXL and VIM. Notably, these networks contain several novel genes and potential therapeutic targets in TNBC.

Materials and methods

Microarray analysis

BRB-Array Tools is an integrated software package for visualization and statistical analysis of raw microarray data (6). We used the BRB-Array (v.4.3.2) class comparison tool to identify differentially expressed genes (DEGs) between MDA-MB-231 vs. MCF7 cell lines from 6 published Gene Expression Omnibus (GEO) datasets. These 6 datasets were deposited into the GEO database. Four of the 6 datasets have not been published yet, and include GSE 26370 (Kung et al, unpublished data), GSE 34987 (Lin et al, unpublished data), GSE 41445 (Groth et al, unpublished data), and GSE 54326 (Braunstein et al, unpublished data). The remaining 2 datasets have published data, and include GSE 29682 (7) and GSE 32474 (8). Class comparison data from each dataset was analysed with the multivariate permutation test computed with 1,000 random permutations and a false discovery rate of 1%. For each dataset, we obtained a list of genes with >2-fold change in expression between the 2 cell lines with high statistical significance (P=10−4 to 10−8). The VENNY program is a web-server that compares gene lists. Accordingly, VENNY analysis identified 71 genes as significantly upregulated in MDA-MB-231 vs. MCF7 cells in at least 3 of the 6 datasets.

Gene ontology analysis

Gene ontology (GO) analysis of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells was done. Accordingly, Table I shows functional categories of genes regulating the 4 most statistically significant biological processes. The Gene Card database (http://www.genecards.org/) also provided additional information on these 71 genes.

Table I.

Gene Ontology of 71 genes upregulated in MDA-MB-231 vs. MCF7 cells.

Table I.

Gene Ontology of 71 genes upregulated in MDA-MB-231 vs. MCF7 cells.

GO termGene symbols (Gene count)FDR adjusted P-value
GO:0006928 Cell movementIL8a, EXT1, F2RL1a, MAP1B, RAC2a, ARHGD1ba, CD97, GPX1, MSNa.c, VIMc, CD44a,c, CTGFa, HBEGFa, CALD1c, PLAUa,b, CYR61b, SLC16A3, AXLb, PRNPc, COL4A1a, VEGFCa.c (21) 6.11×10−06
GO:0009611 Response to woundingF2RL1a, ANXA1a, AXLb, ADM, RAC2, CTGFa,b, SRGN, OPTNb, GPX1, SERPINE1, VEGFCa,c, AOX1b, HBEGFa, LOXb, SLC16A3 (15) 5.20×10−05
GO:0005925 Focal adhesionANXA1a, CAPN2, CNN3, CD44a,c, CD97, EGFRa, MSNa,c, RAC2a, TGM2b, PLAUa,b, VIMc (11) 2.55×10−05
GO:0043066 Negative regulation of apoptosisANXA1a, BIRC3a, CD44a,c, CYR61a,b, EGFRa, GPX1, GSTP1a, OPTNb, PLAC8, PRNPc, SERPINE1a, SPRY2a (12) 3.59×10−03

{ label (or @symbol) needed for fn[@id='tfn1-mmr-18-04-4147'] } The 71 genes upregulated in MDA-MB-231 vs MCF7 cells have 59 genes in 4 statistically significant GO terms.

a Genes encoding interacting proteins in Figure 1 are indicated.

b Genes in the AXL network are indicated.

c Genes in the VIM network are indicated. P-values for each GO term are adjusted for FDR. FDR, false discovery rate; GO, gene ontology.

Protein-protein interactions

Protein-protein interactions (PPIs) were analysed by STRING [https://string-db.org/, v.10.5, with a setting of high confidence (0.70)]. Combined STRING scores for each PPI are based on three types of evidence (experimental data, databases, and published texts). PPIs with combined STRING scores ranging from 0.700–0.990 were considered for the present study.

Pattern miner systems biology tool

The Pattern miner tool provides Pearson's correlation coefficients (R-values) between input and output parameters such as gene expression and a cell line (9). All significant correlations are based on data from a minimum of 35 of 60 cell lines in the NCI-60 cancer cell line database. R values > +0.340, indicate that input and output values increase in a statistically significant manner (P<0.05). For example, a ligand and its significantly co-expressed receptor, would share an R-value > +0.340. For this study, we used higher Pearson's correlation coefficients (R>0.450, P<0.001).

Statistical analysis

P-values adjusted for false positive rates for the 71 individual DEG were obtained from microarray analysis. The P-values for 19 of these 71 DEGs are in Table II. In all Tables, P-values are expressed as negative exponents. For example, a P-value of 1E-06 indicates P=0.000001.

Table II.

Novel genes upregulated in MDA-MB-231 vs. MCF7 cells.

Table II.

Novel genes upregulated in MDA-MB-231 vs. MCF7 cells.

GeneMean ± SD fold changeP-valueGene name and function
Cell migration
  ADORA2B23.80±5.80 5.00×10−08Adenosine Receptor A2B
  BCAT125.15±6.00 8.30×10−08Branched Chain Amino Acid Transaminase1
  COL13A1a105.90 ±1.30 2.80×10−05COL13A1 Isoform of Collagen
  EXT118.10±2.60 3.30×10−06EXT1 Glycosyltransferase
  LIMCH1a48.80±6.10 6.00×10−06LIM and Calponin-homology domain 1
Metastasis
  AKR1B1a41.60±6.30 1.20×10−07Aldo-Keto Reductase 1
  CNN3a212.00±27.81 1.00×10−06Calponin 3
  CYBRD124.20±1.70 5.00×10−08Cytochrome B Reductase1
  FXYD527.70±6.60 1.50×10−07FXYD Domain containing Ion Transport Regulator 5
  SRGNa123.60±16.80 5.00×10−08Serglycin is a Proteoglycan
Senescence
  IFI16a69.80±8.70 1.10×10−04Interferon Gamma Inducible Protein 16
  IGFBP7a67.20±22.40 5.00×10−08Insulin Like Growth Factor Binding Protein 7
  TMEM15828.90±7.30 1.00×10−06TMEM158 encodes a Ras induced Senescence protein (RIS1)
  GNG11a67.80±28.70 5.00×10−08G Protein Subunit Gamma 11 protein
Autophagy
  PLAC8a85.60±15.00 3.00×10−06PLAC8 (Placenta Specific 8)
  SRPX
33.40±2.30 3.00×10−07SRPX (Sushi Repeat Containing Protein-X-Linked). SRPX is also named DRS
Multi-drug resistance
  AGPS10.40±1.80 5.00×10−08Alkylglycerone Phosphate Synthase
  PTRF22.10±5.40 2.80×10−07Polymerase 1 and Transcript Release factor

{ label (or @symbol) needed for fn[@id='tfn5-mmr-18-04-4147'] } Names and functions of 19 novel genes upregulated in MDA-MB-231 vs MCF7 cells are shown. Using class comparison data from 6 GEO datasets, mean Fold change in upregulation of each gene and its Standard deviation, were calculated. Genes showing >40 fold change in upregulation in MDA-MB-231 vs MCF7 cells are indicated

a GEO, Gene Expression Omnibus.

Results

Functional analysis of genes upregulated in MDA-MB-231 vs. MCF7 cells

Our microarray analysis identified 71 genes significantly upregulated in MDA-MB-231 vs. MCF7 cells from 6 GEO datasets (P=5E-08 to 1E-04). GO analysis of these 71 genes put 59 genes into four statistically significant GO terms (Table I). Since 16 genes were in more than 1 GO term, a total of 43 (59–16) of the 71 genes regulate cell movement, response to wounding, adhesion, and survival (Table I). Interestingly, another 19 of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, are not reported for TNBC (Table II). The gene card database showed that 5 of these 19 novel genes promote migration (ADORA2B, BCAT1, COL13A1, EXT1, and LIMCH1), whereas another 5 novel genes promote metastasis (AKR1B1, CNN3, CYBRD1, FXYD5, and SRGN).

We also identified genes in two pathways which are poorly understood in TNBC. The first pathway is senescence, which is promoted by 6 genes with a 28–82 fold upregulation in MDA-MB-231 vs. MCF7 cells. These 6 genes include IFI16, IGFBP7, TMEM158, GNG11, PLAU and SERPINE1 (Tables I and II). The second pathway is autophagy, which is regulated by 4 novel genes (PLAC8, SRPX, CTGF, and OPTN) that were 12–85 fold upregulated in MDA-MB-231 vs. MCF7 cells. These data strongly suggest that senescence and autophagy play important roles in promoting survival and chemo-resistance in MDA-MB-231 cells (3). The literature also shows that senescence plays a key role in breast cancer progression, and that autophagy regulates maintenance of ‘stem cell-like’ properties of MDA-MB-231 cells (3,10).

In summary, GO analysis of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells shows 43 known genes regulating cell adhesion, migration, and survival (Table I). As Table II shows 10 novel upregulated genes promoting known processes (migration and metastases), and 9 novel genes regulating senescence, autophagy, and drug resistance. Therefore, 62 (43+10+9) of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, account for the invasive and malignant phenotype of MDA-MB-231 cells.

Interactions of 71 proteins overexpressed in MDA-MB-231 cells

STRING analysis showed that the 71 gene products overexpressed in MDA-MB-231 vs. MCF7 cells showed 42 protein-protein interactions (PPIs) with significant enrichment (P=4.14×10−11). STRING data also provided robust evidence for interactions between EGFR and 5 proteins which maintain cell survival (STRING scores of 0.730–0.950). These 5 proteins are ANXA1, BIRC3, CD44, GSTP1, and SPRY2. Notably, genes encoding these 5 survival proteins are in the GO term for ‘negative regulation of apoptosis’, and were 15–40 fold upregulated in MDA-MB-231 vs. MCF7 cells (Table I). EGFR also binds proteins that induce angiogenesis (CTGF and HBEGF), drug resistance (PTRF), matrix remodelling (CTGF), and epithelial-mesenchymal transition (ANXA1, CTGF). Combined STRING scores for these EGFR interactions ranged from 0.715–0.998. Interestingly, Fig. 1 also shows that interactions between EGFR, HBEGF, and FOSL1, modulate expression of the pro-inflammatory cytokine, IL8. This is important since IL8 is linked with poor prognosis in node-negative breast cancers, and resistance to EGFR inhibitors (11).

In summary, STRING data in Fig. 1 show that specific EGFR interactions promote cell survival, epithelial-mesenchymal transition (EMT), angiogenesis, drug resistance, and inflammation. This is consistent with reports showing that EGFR signalling drives migration and metastasis of MDA-MB-231 breast cancer cells (12).

Importance of the AXL and VIM genes in TNBC

Although EGFR is a promising therapeutic target in TNBC, anti-EGFR monoclonal antibodies have poor cytotoxicity in TNBC derived cells. This can occur because the AXL-receptor tyrosine kinase dimerizes with EGFR, and initiates signalling that blocks activity of anti-EGFR therapies (5). AXL is upregulated by VIM, which encodes a key marker of EMT (13). Furthermore, overexpression of VIM correlated with increased migration and invasion of breast cancer cells (13). Clinical data show that TNBC patients overexpressing AXL and VIM, have poorer prognosis and survival than TNBC patients expressing high levels of either gene (4,5).

Our data showed that AXL and VIM were 80 and 100 fold upregulated in MDA-MB-231 vs. MCF7 cells, respectively. Although AXL and VIM regulate cell migration and invasion, these proteins are not involved in PPIs (Fig. 1), and their signaling mechanisms are unclear. Therefore, identifying functional networks for AXL and VIM requires new approaches. Accordingly, we used the pattern miner tool to identify members of the 71 genes which show statistically significant co-expression with AXL or VIM. This approach is supported by other studies on molecular networks in breast cancer. For example, a co-expression approach analysed interactions between mRNAs and long noncoding RNAs in TNBC (14). Wang et al (15), used gene co-expression networks built by Pearson's correlation coefficients, to compare metastatic and non-metastatic breast cancers. The pattern miner tool also uses Pearson's correlation coefficients to identify significantly co-expressed genes in the NCI-60 cancer cell database. Therefore, we tested and used this tool to develop functional networks for the AXL and VIM genes in MDA-MB-231 cells.

Testing pattern miner tool on genes upregulated in MDA-MB-231 vs. MCF7 cells

Genes which are significantly co-expressed with each other have a high probability of sharing similar functions. Interestingly, Kohn et al (9), reported 5 co-expressed genes which regulate cancer cell migration by different mechanisms. Notably, all 5 genes from Kohn's study (EGFR, AXL, VIM, CAPN2, and COL4A1), were significantly upregulated in MDA-MB-231 vs. MCF7 cells. The functions of EGFR, AXL, and VIM, are explained above. CAPN2 encodes a calpain 2 protease which triggers invasion of breast cancer cells (16), whereas COL4A1 encodes an isoform of collagen which controls cell invasion (17).

Since the pattern miner tool identifies co-expressed genes, we tested this tool with our data. Accordingly, we determined whether the 5 ‘migration-regulating genes’ identified by Kohn et al (9), are significantly co-expressed with any of the other 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. Our results showed that all 5 ‘migration-regulating genes’ (EGFR, AXL, VIM, COL4A1, CAPN2) are significantly correlated with 7 of the 71 genes upregulated in MDA-MB-231 cells (R= +0.450 to +0.699, P<0.001). These seven genes are EXT1, MAP1b, VEGFC, SERPINE1, CNN3, PTRF, and MXRA7. Five of these 7 genes regulate cell migration (EXT1, MAP1b, VEGFC, SERPINE1, and CNN3) (Table I and Fig. 1). The sixth gene (PTRF), regulates multidrug resistance (18), and the function of the seventh gene (MXRA7), is unknown. Since PTRF and MXRA7 were significantly co-expressed with the 5 genes controlling migration (EGFR, AXL, VIM, CAPN2, and COL4A1), pattern miner predicts that both PTRF and MXRA7 can also regulate cancer cell migration. This pilot test showed that pattern miner is a valid and reliable tool to identify novel genes which regulate cell migration, and show co-expression with AXL or VIM.

Strategy to identify AXL and VIM gene networks with pattern miner

The previous section showed that only 7 of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, were co-expressed with all 5 ‘migration-regulating genes’ (EGFR, AXL, VIM, CAPN2, and COL4A1) (9). However, Tables I and II clearly showed that additional members of these 71 genes regulate cell migration. In order to find these additional members, we used different combinations of these 5 ‘migration-regulating genes’ as input ‘baits’ for pattern miner analysis. Thus, 7 gene pairs (AXL-EGFR, AXL-VIM, AXL-COL4A1, AXL-CAPN2, VIM-EGFR, VIM-COL4A1, and VIM-CAPN2), 2 gene triplets (AXL-COL4A1-VIM and AXL-EGFR-CAPN2), and 2 gene quadruplets (AXL-COL4A1-EGFR-CAPN2 and AXL-VIM-EGFR-CAPN2) were used as input ‘baits’. Next, we used the pattern miner tool to determine whether each of these ‘baits’ were specifically and significantly co-expressed with any of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. This approach should identify subsets of the 71 upregulated genes which co-express with AXL or VIM, and form functional networks that regulate migration of MDA-MB-231 cells.

Identifying and modelling the AXL gene network

We used pattern miner to determine whether any of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, were co-expressed with the different input ‘baits’ listed above. Table III shows that 4 input baits containing AXL, were specifically and significantly co-expressed with 4 different subsets of these 71 genes. This data was used to develop a model for the AXL gene network (Fig. 2). Thus, Fig. 2 shows that the first ‘bait’ (AXL-COL4A1) was significantly co-expressed with CTGF, which regulates EMT (19). The second ‘bait’ (AXL-COL4A1-VIM) was co-expressed with the HEG1 and LOX genes which control vasculogenesis and metastasis (20). The third ‘bait’ (AXL-COL4A1-EGFR-CAPN2), was significantly co-expressed with 3 genes regulating wound healing, angiogenesis, and cell survival, (PDGFC, CYR61, and OPTN; Table I). This bait was also co-expressed with a gene that regulates tumorigenicity (NNMT) (21). The fourth ‘bait’ (AXL-EGFR-CAPN2) was co-expressed with 2 genes regulating adhesion and drug resistance (TGM2, PLAU), and 3 genes controlling stress response (MT1E, NT5E, and AOX1) (Tables I and III) (22).

Table III.

Pattern Miner data for genes co-expressed with AXL.

Table III.

Pattern Miner data for genes co-expressed with AXL.

Input gene Colum 1 inputGenes co-expressed with each input Column 2 outputR-value with AXL Column 3R-value with COL4A1 Column 4R-value with VIM Column 5R-value with EGFR Column 6R-value with CAPN2 Column 7
AXL-COL4A1
CTGF0.5810.5750.4170.4030.357
AXL-COL4A1-VIM
HEG10.5550.4900.632X0.416
LOX0.6050.7110.457X0.379
AXL-COL4A1-EGFR-CAPN2
CYR610.7640.6450.3900.6640.554
OPTN0.5320.4920.4270.5520.626
PDGFC0.5680.533X0.5870.450
NNMT0.7300.7400.4400.7000.500
AXL-EGFR-CAPN2
TGM20.564XX0.6480.549
PLAU0.7260.410X0.6750.574
MT1E0.6370.387X0.5410.523
NT5E0.594X0.4030.4800.604
AOX10.708X0.4130.5240.470

[i] Four gene combinations are inputs for Pattern miner (column 1). Genes significantly correlated with each input, are outputs in column 2. Pearson's correlation coefficients (R) measure correlation between inputs and outputs (columns 3–7). Strong correlation and co-expression of an input with its output, is indicated by significant R-values (R >0.450, P<0.001). For example, the AXL-COL4A1 input is significantly co-expressed with the output, CTGF. Lack of significant correlation is indicated by ‘X’.

In summary, the AXL network has 4 distinct sets of co-expressed genes with different functions. These 4 sets contain 12 genes which were 12–85 fold upregulated in MDA-MB-231 vs. MCF7 cells, and account for key functions of AXL (Tables I, III, and Fig. 2). Notably, 6 of these 12 genes (HEG1, OPTN, MT1E, NNMT, AOX1, and TGM2) have not been reported for TNBC.

Identifying and modelling the VIM gene network

Since pattern miner analysis provided insights into the AXL gene network, we undertook a similar study for the VIM gene. Table IV shows that VIM, and 3 input baits containing VIM, were specifically co-expressed with 4 different subsets of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. This data was used to develop a model for the VIM gene network (Fig. 3). Thus, VIM alone was co-expressed with 5 genes (MSN, PRNP, ACSL4, IFI16, and SRPX) with varying functions. MSN regulates invasiveness, PRNP protects against endoplasmic reticular stress (Table I), and ACSL4 promotes lipogenesis (23). The IFI16 and SRPX genes regulate senescence and autophagy, respectively (24,25) (Tables II and IV). The first ‘bait’ (VIM-COL4A1), was co-expressed with three genes (CALD1, AKR1B1, and IGFBP7) that regulate motility, detoxification, and senescence (26,27) (Tables II and IV). The second ‘bait’ (VIM-AXL), was exclusively co-expressed with an oncogene correlated with lymph node metastases (EMP3) (28). The third bait (VIM-AXL-EGFR-CAPN2), was co-expressed with the FOSL1 and CD44 genes. Notably, the FOSL1/Fra-1 transcription factor controls the invasive phenotype of MDA-MB-231 cells (29), and can regulate expression of 3 other genes upregulated in these cells (CD44, VEGFC, and ADORA2B). CD44 regulates extracellular interactions, and is a marker of tumour progression (30), whereas VEGFC and ADORA2B regulate migration (31).

Table IV.

Pattern Miner Data for Genes Co-expressed with VIM.

Table IV.

Pattern Miner Data for Genes Co-expressed with VIM.

Input gene Colum 1 inputGenes co-expressed with each input Column 2 outputR value with VIM Column 3R value with COL4A1 Column 4R value with AXL Column 5R value with EGFR Column 6R value with CAPN2 Column 7
VIM
MSN0.8020.3510.355XX
PRNP0.624XXX0.445
ACSL40.565X0.436X0.390
IFI160.555XXXX
SRPX0.533XXXX
VIM-COL4A1
AKR1B10.6140.5840.4230.379X
IGFBP70.4930.5590.363XX
CALD10.4650.651XXX
VIM-AXL
EMP30.754X0.466X0.363
VIM-AXL-EGFR-CAPN2
CD440.5940.3970.5980.4920.650
FOSL10.589X0.7500.5790.722

[i] Four gene combinations are inputs for Pattern miner (column 1). Genes significantly correlated with each input, are outputs in column 2. Pearson's correlation coefficients (R) measure correlation between inputs and outputs (columns 3–7). Strong correlation and co-expression of an input with its output, is indicated by significant R-values (R >0.450, P<0.001). For example, the VIM-COL4A1 input is significantly co-expressed with 3 outputs (AKR1B1, IGFBP7, and CALD1). Lack of significant correlation is indicated by ‘X’.

In summary, the 4 components of the VIM network contain 11 genes which were 12–105 fold upregulated in MDA-MB-231 vs. MCF7 cells, and account for key functions of VIM (Tables I, II and IV and Fig. 3). Notably, 6 of these 11 genes (PRNP, AKR1B1, CALD1, EMP3, SRPX, and IFI16) have not been reported for TNBC.

Comparison of the AXL and VIM gene networks

We carefully compared the AXL and VIM gene networks shown in Figs. 2 and 3. Notably, these networks contain very different subsets of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. The AXL network has 12 genes regulating EMT, invasion, adhesion, and chemo-resistance, whereas the VIM network has 11 genes regulating lipogenesis, senescence, autophagy, and metastases. These results strongly support our hypothesis that AXL and VIM participate in 2 separate networks of co-expressed genes that are highly upregulated in MDA-MB-231 vs. MCF7 cells.

Identifying therapeutic targets in MDA-MB-231 cells

One of our objectives was to identify novel therapeutic targets in MDA-MB-231 cells. The gene card database showed that 15 of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, are therapeutic targets in other cancers. These 15 therapeutic targets regulate EGFR signalling, invasion, inflammation, angiogenesis, metastases, senescence, and autophagy (CD44, CTGF, CYR61/CCN1, EMP3, HEG1, IL8, ADORA2B, AGPS, AKR1B1, EXT1, IGFBP7, PLAC8, PTRF, SRGN, and TMEM158/RIS1). Six of these 15 target genes (CD44, CTGF, CYR61/CCN1, EMP3, HEG1, and IL8) are reported for TNBC. Notably, three of these 6 known therapeutic targets are in the AXL network in Fig. 2 (CTGF, CYR61/CCN1, and HEG1), and another 2 are in the VIM network in Fig. 3 (CD44 and EMP3). However, 9 of these 15 therapeutic targets are novel for TNBC, and are in Table II (ADORA2B, AGPS, AKR1B1, EXT1, IGFBP7, PLAC8, PTRF, SRGN, and TMEM158/RIS1). Interestingly, 2 of these 9 novel genes are in the VIM network. (AKR1B1 and IGFBP7) (Fig. 3). We also note 2 other novel, potential therapeutic target genes which regulate iron uptake (CYBRD1 and FXYD5, Table II).

To summarize, our re-analysis of the 6 GEO datasets identified many therapeutic targets from the list of 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. Thus, we identified 6 known therapeutic gene targets reported in other cancers, 9 novel therapeutic targets for TNBC, and 2 potential targets which regulate iron uptake (Table II).

Discussion

This study is unique for five reasons. First, we re-analysed raw data from 6 microarray GEO datasets and identified a common list of 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. Second, we identified novel upregulated genes regulating migration and metastasis in MDA-MB-231 vs. MCF7 cells. This is important, since migration and metastasis are phenotypic hallmarks of TNBC. Third, we found 9 novel upregulated genes that promote senescence, autophagy, chemo-resistance, and stem-cell like properties of MDA-MB-231 cells. Fourth, we used pattern miner to determine whether any of the 71 genes upregulated in MDA-MB-231 vs. MCF7 cells, were significantly co-expressed with different combinations of the 5 ‘migration-regulating genes’ (EGFR, AXL, VIM, CAPN2, and COL4A1). Interestingly, our results identified 2 distinct, non-overlapping networks of upregulated genes significantly co-expressed with either AXL or VIM. Notably, these networks can drive malignancy by regulating very different processes in MDA-MB-231 cells. Furthermore, these networks contain known therapeutic target genes, and other novel genes that can expand the signalling network and functionality of the AXL and VIM proteins in MDA-MB-231 cells. Fifth, we identified 6 known therapeutic targets, 9 novel therapeutic targets, and 2 potential therapeutic target genes; from the list of 71 genes upregulated in MDA-MB-231 vs. MCF7 cells. All novel therapeutic targets for TNBC require experimental verification.

Acknowledgements

The authors sincerely thank the Management of Dr. M.G.R. Educational and Research Institute (Chennai, India) for providing laboratory and computational facilities. The authors thank Dr. Rama Vaidyanathan, Director R&D (Dr. M.G.R. Educational and Research Institute) for valuable support.

Funding

The present study was supported by the Science and Engineering Research Board, Government of India under the Fast track Scheme for Young Scientists (grant no. SR/FT/LS-140/2010).

Availability of data and materials

The original microarry data are freely available from the GEO datasets of NCBI (https://www.ncbi.nlm.nih.gov/gds). Analysis of these microarray data for this study will be made available from the corresponding author on reasonable request.

Authors' contributions

SM and MR contributed to the acquisition and analysis of data, and preparation of figures. VNS and SN performed the interpretation of data, manuscript writing and critical revisions for important intellectual content. All authors approved the final version to be published.

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.

Glossary

Abbreviations

Abbreviations:

DEGs

differentially expressed genes

EGFR

epidermal growth factor receptor

EMT

epithelial-mesenchymal transition

GEO

Gene Expression Omnibus

GO

Gene Ontology

ER

estrogen receptor

PPIs

protein-protein interactions

PR

progesterone receptor

TNBC

triple negative breast cancers

References

1 

Cleator S, Heller W and Coombes RC: Triple-negative breast cancer: Therapeutic options. Lancet Oncol. 8:235–244. 2007. View Article : Google Scholar : PubMed/NCBI

2 

Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Y and Pietenpol JA: Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest. 121:2750–2767. 2011. View Article : Google Scholar : PubMed/NCBI

3 

O'Reilly EA, Gubbins L, Sharma S, Tully R, Guang MH, Weiner-Gorzel K, McCaffrey J, Harrison M, Furlong F, Kell M and McCann A: The fate of chemoresistance in triple negative breast cancer (TNBC). BBA Clin. 3:257–275. 2015. View Article : Google Scholar : PubMed/NCBI

4 

Tanaka K, Tokunaga E, Inoue Y, Yamashita N, Saeki H, Okano S, Kitao H, Oki E, Oda Y and Maehara Y: Impact of expression of vimentin and Axl in breast cancer. Clin Breast Cancer. 16:520–526.e2. 2016. View Article : Google Scholar : PubMed/NCBI

5 

Meyer AS, Miller MA, Gertler FB and Lauffenburger DA: The receptor AXL diversifies EGFR signaling and limits the response to EGFR-targeted inhibitors in triple-negative breast cancer cells. Sci Signal. 6:ra662013. View Article : Google Scholar : PubMed/NCBI

6 

Simon R, Lam A, Li MC, Ngan M, Menenzes S and Zhao Y: Analysis of gene expression data using BRB-ArrayTools. Cancer Inform. 3:11–17. 2007. View Article : Google Scholar : PubMed/NCBI

7 

Reinhold WC, Mergny JL, Liu H, Ryan M, Pfister TD, Kinders R, Parchment R, Doroshow J, Weinstein JN and Pommier Y: Exon array analyses across the NCI-60 reveal potential regulation of TOP1 by transcription pausing at guanosine quartets in the first intron. Cancer Res. 70:2191–2203. 2010. View Article : Google Scholar : PubMed/NCBI

8 

Pfister TD, Reinhold WC, Agama K, Gupta S, Khin SA, Kinders RJ, Parchment RE, Tomaszewski JE, Doroshow JH and Pommier Y: Topoisomerase I levels in the NCI-60 cancer cell line panel determined by validated ELISA and microarray analysis and correlation with indenoisoquinoline sensitivity. Mol Cancer Ther. 8:1878–1884. 2009. View Article : Google Scholar : PubMed/NCBI

9 

Kohn KW, Zeeberg BR, Reinhold WC, Sunshine M, Luna A and Pommier Y: Gene expression profiles of the NCI-60 human tumor cell lines define molecular interaction networks governing cell migration processes. PLoS One. 7:e357162012. View Article : Google Scholar : PubMed/NCBI

10 

Cufí S, Vazquez-Martin A, Oliveras-Ferraros C, Martin-Castillo B, Vellon L and Menendez JA: Autophagy positively regulates the CD44(+) CD24(−/low) breast cancer stem-like phenotype. Cell Cycle. 10:3871–3885. 2011. View Article : Google Scholar : PubMed/NCBI

11 

Liu YN, Chang TH, Tsai MF, Wu SG, Tsai TH, Chen HY, Yu SL, Yang JC and Shih JY: IL-8 confers resistance to EGFR inhibitors by inducing stem cell properties in lung cancer. Oncotarget. 6:10415–10431. 2015.PubMed/NCBI

12 

Price JT, Tiganis T, Agarwal A, Djakiew D and Thompson EW: Epidermal growth factor promotes MDA-MB-231 breast cancer cell migration through a phosphatidylinositol 3′-kinase and phospholipase C-dependent mechanism. Cancer Res. 59:5475–5478. 1999.PubMed/NCBI

13 

Satelli A and Li S: Vimentin in cancer and its potential as a molecular target for cancer therapy. Cell Mol Life Sci. 68:3033–3046. 2011. View Article : Google Scholar : PubMed/NCBI

14 

Liu YR, Jiang YZ, Xu XE, Yu KD, Jin X, Hu X, Zuo WJ, Hao S, Wu J, Liu GY, et al: Comprehensive transcriptome analysis identifies novel molecular subtypes and subtype-specific RNAs of triple-negative breast cancer. Breast Cancer Res. 18:332016. View Article : Google Scholar : PubMed/NCBI

15 

Wang X, Qian H and Zhang S: Discovery of significant pathways in breast cancer metastasis via module extraction and comparison. IET Syst Biol. 8:47–55. 2014. View Article : Google Scholar : PubMed/NCBI

16 

Cortesio CL, Chan KT, Perrin BJ, Burton NO, Zhang S, Zhang ZY and Huttenlocher A: Calpain 2 and PTP1B function in a novel pathway with Src to regulate invadopodia dynamics and breast cancer cell invasion. J Cell Biol. 180:957–971. 2008. View Article : Google Scholar : PubMed/NCBI

17 

Miyake M, Hori S, Morizawa Y, Tatsumi Y, Toritsuka M, Ohnishi S, Shimada K, Furuya H, Khadka VS, Deng Y, et al: Collagen type IV alpha 1 (COL4A1) and collagen type XIII alpha 1 (COL13A1) produced in cancer cells promote tumor budding at the invasion front in human urothelial carcinoma of the bladder. Oncotarget. 8:36099–36114. 2017. View Article : Google Scholar : PubMed/NCBI

18 

Yi JS, Mun DG, Lee H, Park JS, Lee JW, Lee JS, Kim SJ, Cho BR, Lee SW and Ko YG: PTRF/cavin-1 is essential for multidrug resistance in cancer cells. J Proteome Res. 12:605–614. 2013. View Article : Google Scholar : PubMed/NCBI

19 

Zhu X, Zhong J, Zhao Z, Sheng J, Wang J, Liu J, Cui K, Chang J, Zhao H and Wong S: Epithelial derived CTGF promotes breast tumor progression via inducing EMT and collagen I fibers deposition. Oncotarget. 6:25320–25338. 2015. View Article : Google Scholar : PubMed/NCBI

20 

Liu JL, Wei W, Tang W, Jiang Y, Yang HW, Li JT and Zhou X: Silencing of lysyl oxidase gene expression by RNA interference suppresses metastasis of breast cancer. Asian Pac J Cancer Prev. 13:3507–3511. 2012. View Article : Google Scholar : PubMed/NCBI

21 

Zhang J, Wang Y, Li G, Yu H and Xie X: Down-regulation of nicotinamide N-methyltransferase induces apoptosis in human breast cancer cells via the mitochondria-mediated pathway. PLoS One. 9:e892022014. View Article : Google Scholar : PubMed/NCBI

22 

Loi S, Pommey S, Haibe-Kains B, Beavis PA, Darcy PK, Smyth MJ and Stagg J: CD73 promotes anthracycline resistance and poor prognosis in triple negative breast cancer. Proc Natl Acad Sci USA. 110:pp. 11091–11096. 2013; View Article : Google Scholar : PubMed/NCBI

23 

Wu X, Li Y, Wang J, Wen X, Marcus MT, Daniels G, Zhang DY, Ye F, Wang LH, Du X, et al: Long chain fatty Acyl-CoA synthetase 4 is a biomarker for and mediator of hormone resistance in human breast cancer. PLoS One. 8:e770602013. View Article : Google Scholar : PubMed/NCBI

24 

Clarke CJ, Hii LL, Bolden JE and Johnstone RW: Inducible activation of IFI 16 results in suppression of telomerase activity, growth suppression and induction of cellular senescence. J Cell Biochem. 109:103–112. 2010.PubMed/NCBI

25 

Tambe Y, Yamamoto A, Isono T, Chano T, Fukuda M and Inoue H: The drs tumor suppressor is involved in the maturation process of autophagy induced by low serum. Cancer Lett. 283:74–83. 2009. View Article : Google Scholar : PubMed/NCBI

26 

Wu X, Li X, Fu Q, Cao Q, Chen X, Wang M, Yu J, Long J, Yao J, Liu H, et al: AKR1B1 promotes basal-like breast cancer progression by a positive feedback loop that activates the EMT program. J Exp Med. 214:1065–1079. 2017. View Article : Google Scholar : PubMed/NCBI

27 

Benatar T, Yang W, Amemiya Y, Evdokimova V, Kahn H, Holloway C and Seth A: IGFBP7 reduces breast tumor growth by induction of senescence and apoptosis pathways. Breast Cancer Res Treat. 133:563–573. 2012. View Article : Google Scholar : PubMed/NCBI

28 

Wang YW, Cheng HL, Ding YR, Chou LH and Chow NH: EMP1, EMP 2, and EMP3 as novel therapeutic targets in human cancer. Biochim Biophys Acta. 1868:199–211. 2017.PubMed/NCBI

29 

Young MR and Colburn NH: Fra-1 a target for cancer prevention or intervention. Gene. 379:1–11. 2006. View Article : Google Scholar : PubMed/NCBI

30 

Bourguignon LY: CD44 mediated oncogenic signaling and cytoskeleton activation during mammary tumor progression. J Mammary Gland Neoplasia. 6:287–297. 2001. View Article : Google Scholar

31 

Desmet CJ, Gallenne T, Prieur A, Reyal F, Visser NL, Wittner BS, Smit MA, Geiger TR, Laoukili J, Iskit S, et al: Identification of a pharmacologically tractable Fra-1/ADORA2B axis promoting breast cancer metastasis. Proc Natl Acad Sci USA. 110:pp. 5139–5144. 2013; View Article : Google Scholar : PubMed/NCBI

Related Articles

Journal Cover

October-2018
Volume 18 Issue 4

Print ISSN: 1791-2997
Online ISSN:1791-3004

Sign up for eToc alerts

Recommend to Library

Copy and paste a formatted citation
x
Spandidos Publications style
Natarajan S, Sumantran VN, Ranganathan M and Madheswaran S: Microarray and pattern miner analysis of AXL and VIM gene networks in MDA‑MB‑231 cells. Mol Med Rep 18: 4147-4155, 2018.
APA
Natarajan, S., Sumantran, V.N., Ranganathan, M., & Madheswaran, S. (2018). Microarray and pattern miner analysis of AXL and VIM gene networks in MDA‑MB‑231 cells. Molecular Medicine Reports, 18, 4147-4155. https://doi.org/10.3892/mmr.2018.9404
MLA
Natarajan, S., Sumantran, V. N., Ranganathan, M., Madheswaran, S."Microarray and pattern miner analysis of AXL and VIM gene networks in MDA‑MB‑231 cells". Molecular Medicine Reports 18.4 (2018): 4147-4155.
Chicago
Natarajan, S., Sumantran, V. N., Ranganathan, M., Madheswaran, S."Microarray and pattern miner analysis of AXL and VIM gene networks in MDA‑MB‑231 cells". Molecular Medicine Reports 18, no. 4 (2018): 4147-4155. https://doi.org/10.3892/mmr.2018.9404