Cytogenetic and cDNA Microarray Expression Analysis of MCF10 Human Breast Cancer Progression Cell Lines

Breast cancer is the most common cancer and accounts for the second highest mortality rate worldwide among cancer-related deaths in women (1). Numerous studies have indicated that initiation and progression toward a breast cancer phenotype is a multistep process involving accumulation of genomic aberrations (2). In particular, amplifications, deletions, and rearrangements have been observed in breast cancer of critical genes involved in cell growth, differentiation, and cell death (3). Breast cancers can be either hereditary or sporadic. Germ-line mutations in BRCA1 and BRCA2 (4) genes have been linked to the development of familial breast cancer.

Although the genomic and molecular changes in breast cancer have been studied in detail, the initial events leading to tumor formation remain to be elucidated. Cancer progression models have become an invaluable tool in studying the precise genetic aberrations that correlate with a shift from a normal to a disease phenotype. In this investigation, we used the MCF10 human breast cancer progression model cell lines to study the genetic changes that occur during the transformation of breast epithelial cells into breast cancer. The MCF10 progression model consists of three directly derived cell lines: (a) the spontaneously immortalized cell line MCF10A (5), which do not show any characteristics of invasiveness or tumor formation and hence are considered to be a normal-like breast epithelial cell line (6); (b) MCF10A cells were transformed with c-Ha-Ras to yield the premalignant MCF10AT1 cell line (7, 8); and (c) MCF10AT1 cells were xenografted onto mice and a third cell line, MCF10CA1a, was selected, which shows all the characteristics of a fully malignant breast cancer cell type (9). Karyotypic analysis of these three cell lines showed a characteristic (3;9) translocation that confirmed a common lineage (9).

Spectral karyotyping (10) has been employed previously to study chromosomal aberrations in breast cancer cell lines and patient tumors and led to the identification of numerous chromosomal translocations, deletions, and rearrangements (11). Array comparative genomic hybridization (aCGH) is useful in the detection of copy number variations that arise due to the genomic instability in cancer cells or tissues (12) and has led to the detection of gains within chromosome arms 1q, 3p, 4q, 8q, 11q, 17q, and 20q and losses in 6q,11q, 8p, 9p, 13q,16q, and 17p (13) in breast cancer. In most of these studies, the altered regions have been shown to harbor essential gene(s), the deregulation of which leads to the establishment of tumorigenesis (3). cDNA microarray analysis makes it feasible to study the expression patterns of thousands of genes in cancer (14). This technique has resulted in the identification of numerous changes in gene expression associated with breast cancer (15).

In the present study, we attempted to correlate the changes observed in the immortalized normal human breast epithelial cell line MCF10A and its malignant counterpart, MCF10CA1a, by combined spectral karyotyping, aCGH and cDNA microarray analysis. We observed numerous changes in gene expression patterns that were indicative of tumor progression. Thus, combining these three approaches has provided insights into how DNA copy number variations affect the gene expression patterns and for the identification of new candidate genes that might be associated with breast cancer and its progression.

Cell culture. The breast cancer progression model cell lines (MCF10A, MCF10AT1, and MCF10CA1a) used in these studies were obtained from the Barbara Ann Karmanos Center. The MCF10A cell line was cultured in DMEM supplemented with horse serum (5%), insulin (2.5 mg/mL), epidermal growth factor (50 μg/mL), cholera enterotoxin (150 μg/mL), hydrocortisone (2.5 mg/mL), and HEPES (5 mmol/L), whereas MCF10AT1 and MCF10CA1a cell lines were cultured in DMEM supplemented with only horse serum (5%). All three cell lines were grown at 37°C in a 5% CO₂ substituted incubator. Cells were collected at the same passage and used for each of the analyses (spectral karyotyping, CGH, and gene expression microarray) done in this study.

Spectral karyotyping. Preparations of metaphase chromosome spreads were subjected to the spectral karyotyping procedure recommended by Applied Spectral Imaging. The images were captured using a combination of rhodamine, Texas red, Cy5, FITC, and Cy5.5 filter sets mounted on a Nikon fluorescence microscope equipped with a spectral cube and an interferometer module. Karyotypic analysis of the images was carried out using the Spectral Karyotyping View software.

aCGH array analysis. DNA printing solutions were prepared from sequence connected RPCI-11 BAC by ligation-mediated PCR as described previously (16). The array contains ∼19,000 BAC clones that were chosen by virtue of their STS content, end sequence, and association with heritable disorders and cancer (17). Each clone is printed in duplicate on amino-silanated glass slides (Schott Nexterion type A+) using a MicroGrid ll TAS arrayer (Genomic Solutions). The BAC DNA products have ∼80 μm diameter spots with 150 μm center to center spacing creating an array of ∼39,000 elements. The printed slides dry overnight and are UV crosslinked (350 mJ) in a Stratalinker 2400 (Stratagene) immediately before hybridization.

Reference and test sample genomic DNA (1 μg each) were individually fluorescently labeled using the BioPrime DNA labeling kit (Invitrogen) for 18h at 37°C with the appropriate Cy dye (Cy3 or Cy5). After ethanol precipitation, the probes are resuspended in H₂O, combined, and purified of unincorporated dye by passage over a Qiagen spin column. Before hybridization, the test and reference probes were resuspended in 110 μL SlideHyb Buffer 3 (Ambion) containing 5 μL of 20 μg/μL Cot-1 and 5 μL of 100 μg/μL yeast tRNA (Invitrogen), heated to 95°C for 5 min, and placed on ice. Hybridization to the 6K BAC arrays was done for 16 h at 55°C using a GeneTAC hybridization station (Genomic Solutions) as described (18). The hybridized aCGH slides are scanned using a GenePix 4200A scanner (Molecular Devices) to generate high-resolution (5 μm) images for both Cy3 (test) and Cy5 (control) channels. Image analysis was done using the ImaGene version 6.1.0 (Bio Discovery). The log₂ test/control ratios were normalized using a subgrid Loess correction. Mapping information was added to the resulting log₂ test/control values.

cDNA microarray analysis. cDNA microarray analysis was done on an Agilent 44K whole human genome oligo microarray. Total RNA from the cell cultures (MCF10A and MCF10CA1a) were prepared using the RNeasy mini kits (Qiagen) following the manufacturer's instructions. After elution, RNA samples were concentrated by ethanol precipitation at −20°C overnight and resuspended in nuclease-free water. Before labeling, RNA samples were quantified using a Genequant spectrophotometer (GE Healthcare) and evaluated for degradation using a 2100 Bioanalyzer (Agilent Technologies).

To screen the samples for gene expression, cRNA was synthesized by invitro transcription and directly labeled with Cy3-CTP or Cy5-CTP using the low RNA input linear amplification kit as per the manufacturer's instructions (Agilent Technologies). Initially, first-strand cDNA was synthesized by incubating 0.2 to 2 μg total RNA and spiking controls and T7 oligo(dT) for 10 min at 65°C. Following the addition of first-strand reaction components, the reaction continued for 2 h at 40°C. cRNA was synthesized by incubating the resuspended double-stranded cDNA with the transcription master mix containing cyanine-CTP, ribonucleotides, and T7 RNA polymerase for 2 h at 40°C. The cyanine-labeled cRNA was recovered by column purification and eluted in 30 μL RNase-free water. The cRNA quality and concentration were assessed using a NanoDrop ND-1000 UV-visible spectrophotometer (Thermo Scientific). cRNA samples were required to have a yield >750 ng and a specific activity >8.0 pmol Cy3 or Cy5/μg cRNA to proceed to the hybridization step.

Before hybridization, 0.75 μg Cy5-labeled, linearly amplified cRNA (test) and 0.75 μg Cy3-labeled, linearly amplified cRNA (control) were combined and fragmented by incubating at 60°C for 30 min in fragmentation buffer. The hybridization reaction mixture was prepared by combining 2× hybridization buffer to the fragmented cRNA to a final volume of 210 μL. The hybridization reaction mixture was placed on ice and then loaded slowly onto the 44K array, allowing even flow and distribution of hybridization cocktail across the surface. The arrays were secured in a SureHyb chamber cover and then placed in a rotisserie hybridization oven at 65°C for 17 h. After hybridization, the slides were removed from the hybridization oven and washed with Gene Expression Wash Buffers 1 and 2. Following the wash steps, the array images were captured by scanning at 532 and 635 nm with an Agilent scanner and analyzed with Agilent feature extraction software version 8.5.

Data analysis for aCGH and cDNA microarrays. The 19K aCGH array was analyzed using DNAcopy R package, which uses the circular binary segmentation algorithm (19). A segment was identified as gain or loss if the estimated copy number ratio is more than 0.2 or less than −0.2 (corresponding to a genomic fold increase of 1.15 and decrease of 1.15), which is roughly eight times the SE log ratio on each chip. For Agilent gene expression arrays using the dye-swap design, significantly changed genes are identified using limma R package (20), which uses a modified t test with multiple tests in consideration. The gene ontology analysis for the significant changed genes was done by GOstat software (21). To better plot the aCGH results, an in-house ideogram and aCGH data plotting program was developed and used.

Expression analysis of normal and malignant breast cell lines. Microarray expression analysis of immortalized normal MCF10A and malignant MCF10CA1a cell lines was done to determine the changes in gene expression patterns. A total of 42,000 genes yielded hybridization signals for both samples (Supplementary Table S1). Analysis of the data yielded a total of ∼7,000 genes that showed at least 2-fold expression level changes (∼3,044 genes were up-regulated and ∼3,829 genes were down-regulated) in the MCF10CA1a cell line (Supplementary Table S2).

A group of genes that were either highly up-regulated or greatly repressed in the MCF10CA1a cell line is listed in Table 1. All these genes were chosen based on their functional involvement in breast cancer. Up-regulated genes included SEPP1 and DCN, which have an antimetastatic role in breast cancer (22–24); FBN1, AOX1, and PTERG2, which are known to promote tumor growth (25–27); and, surprisingly, ANGPT1, which when overexpressed has antitumor properties (28).

Our results showed a down-regulation of CDH1. Deregulation of this gene was observed previously in breast and other types of cancer (29). The IL1B and S100A14 genes were both greatly repressed, which is contrary to the reported high levels of protein expression of these two proteins in breast and other cancers (30). Moreover, the BDKRB2 gene is known to induce mitosis in breast cells and enhance progression of cancer (31). Yet, our study shows that this gene was significantly repressed.

Further analyses of the microarray data were done to establish the relationships between expression levels and gene function (Tables 2 and 3). Several keratin, connexin, and claudin genes involved in the formation of the extracellular matrix and cell-cell communication were down-regulated. Genes involved in signal transduction pathways such as ERBB2, HRAS, BRAF, VEGF, and EGFR, were also down-regulated in the cancer cell line. Oncogenes and tumor suppressors, including MYC, PTEN, and BRCA2, were down-regulated, whereas RB1, CDKNB1, and CCND3 were overexpressed.

Several interleukin genes such as IL6, IL8, IL1A, and IL1B were repressed in MCF10CA1a cell line, whereas IL7 and IL18 genes were expressed at higher levels. Finally, many genes that are involved in the formation of the extracellular matrix such as COL6A1, KRT7, FBN1, MMP2, MUC1, and FN1 showed higher levels of expression in MCF10CA1a (Table 3).

Spectral karyotyping analysis of MCF10 breast cancer progression model cell lines. Spectral karyotyping identified several chromosomal rearrangements in each of the three breast cancer cell lines (Fig. 1). A karyotype of 47 chromosomes was found for normal MCF10A and the premalignant MCF10AT1 cell lines (Fig. 1 A and B; Supplementary Table S3), whereas the malignant MCF10CA1a cell line had 50 chromosomes (Fig. 1C; Supplementary Table S3). Four marker chromosomes were identified in MCF10A and MCF10AT1 and nine in the malignant MCF10CA1a cell line (Supplementary Table S3). MCF10A showed the characteristic (3;9) reciprocal translocation that has been described as the single most important event in the immortalization and transformation of the MCF10 breast epithelial cell line (32). Other karyotypic changes including a gain of chromosome 8 (+8) and translocations involving chromosomes 5, 3, and 9 [t(5;3;9)] and chromosomes 6 and 19 [t(6;19); Supplementary Table S3]. Karyotypic changes in the MCF10AT1 included a reciprocal translocation involving chromosomes 3 and 17 [rept(3;17)] in addition to those found in MCF10A (Supplementary Table S3). Karyotypic alterations in MCF10CA1a also included a gain of chromosome 20 (+20) and translocation between chromosomes 5 and 9 [t(5;9)] and chromosomes 2 and 10 [t(2;10)] in addition to the changes that are already present in MCF10A (Supplementary Table S3).

aCGH analyses of chromosomal gains and losses in the breast cancer progression model cell lines. Analysis of aCGH data of the three cell lines revealed chromosomal breakpoints that were reported previously (33) as well as several new chromosomal aberrations. Three independent aCGH experiments were done in the following manner: MCF10A versus a normal diploid cell line, MCF10A versus MCF10AT1, and MCF10A versus MCF10CA1a. Ideograms of the complete analysis are presented in Supplementary Fig. S1.

The MCF10A cell line showed gains at 5q23.1-35.3, 19q13.11q13.43, and 13q32.1-p32.2 (Table 4; Supplementary Fig. S1A). A gain of 8p23.3-q24.3 was seen in MCF10A as evidenced by the presence of an extra copy of chromosome 8 in this cell line compared with the normal diploid karyotype. Losses were observed at 9p21.3, 3p26.3, 16p11.2, 21p11-q11.2, and 22q11.1 (Table 4; Supplementary Fig. S1A). The premalignant MCF10AT1 cell line showed gains at 3p14.3, 3q13.3, 19q12-q34.3, 10q22.1-q22.2, 16q23.3, and 17p11.2 and losses of 5q12.1, 5q14.3-15, and 15q21.1 (Table 4; Supplementary Fig. S1B). The additional chromosome 8 copy found in the normal MCF10A cell line is deleted in this cell line. The malignant MCF10CA1a cell line showed several more genomic aberrations compared with the normal and premalignant MCF10 cell lines. Gains include 2p25.3-q21.2, 3p14.1-q29, 9p24.3-p11.2, 9p34.13-p34.3, 10q11.1-q26.3, 17p11.2, and 20p13-q13.3 (Table 4; Supplementary Fig. S1C). Moreover, loss of genomic regions 2q21.2-q22.1, 5q12.1, 5q14.3-q15, −8p23.3-q24.3, and 16q23.1 were observed (Table 4; Supplementary Fig. S1C). These cytogenetic gains and losses were concordant with the loss and gain of genes that have been reported previously (33) in these cell lines.

Correlation of chromosomal gains and losses with gene expression. Gene expression changes for 40 genes found in 36 different genomic regions showing gains or losses in MCF10CA1a are shown in Supplementary Table S4. The ratio of genomic gains range from 1.18 to 1.49, whereas losses range from 0.71 to 0.77. We also identified 7 genes that are important in breast cancer within the regions of genomic loss/gain in MCF10CA1a (Supplementary TableS4, double asterisks). These genes include CA8, SULF1, C8orf13, C2orf32, TACSTD1, COL9A3, and tissue PLAT. These genes are located within the chromosome bands 8q12.1, 8q13.2, 8p23.1, 2p14, 2p21, 20q13.3, and 8p11.21, respectively. Among the genes found within genomic gain regions, 13 were up-regulated and 18 were down-regulated, whereas, in the loss regions, 3 genes were down-regulated and 6 genes were up-regulated (Supplementary Table S4). This indicated that in many cases there was no direct relationship between gains or losses of chromosomal regions and gene expression in those regions.

We further investigated this question at the global level of gene expression. About 1,000 genes were identified in MCF10CA1a that showed both changes in genomic gains or losses and at least a 2-fold change in their expression levels (Supplementary Table S5). To our surprise, we found that all 6 regions showing chromosomal gains had a much higher number of down-regulated genes and 3 of 5 regions with genomic losses showed higher numbers of up-regulated genes (Table 5). Of the 701 genes identified in the genomic gains regions, 76% were down-regulated, whereas 56% of the 301 genes identified in the loss regions were up-regulated (Table 5; Supplementary Table S5).

Cancer progression models provide an approach to elucidate the intermediate molecular and genetic changes that ultimately lead to the transformation of a normal cell into a metastatic cell type. In this present study, we used a combination of spectral karyotyping and aCGH methods to identify regions of chromosomal aberrations in the MCF10 human breast cancer progression model cell lines (9). Gene expression analysis was also done in the immortalized normal MCF10A and malignant MCF10CA1a cell lines as a step toward identifying novel candidate protein factors that potentially correlate with cancer metastasis. In addition, because cancer is characterized by numerous alterations in the genome compared with normal cells, we have attempted to correlate changes in gene expression with genomic gains or losses of the individual genes in the malignant cellline.

About 7,000 genes were identified in MCF10CA1a that showed a 2-fold difference in expression level (Supplementary Table S2). We further classified genes based on their relevance to breast cancer (Table 1). Consistent with previous reports in breast cancer, we detected up-regulation in MCF10CA1a of FN1 and MMP2 (34, 35). In contrast, although overexpression of SEPP1 and DCN has been shown to have an antimetastatic role in breast cancer (23, 24), these genes are highly expressed in the malignant MCF10CA1a line compared with MCF10A (Table 1). Whereas numerous studies showed that overexpression or amplification of the genes ERBB2 and EGFR promotes breast cancer tumorigenesis (36), we find that they are significantly down-regulated in the MCF10CA1a cells (Tables 2 and3). Repression of the HRAS gene in MCF10CA1a is consistent with the overexpression of a mutant form of this protein engineered into this cell line (7).

Genes such as MYC, PTEN, and BRCA2 (Table 2), which are involved in cell cycle control and DNA repair, are down-regulated. Germ-line mutations in BRCA2 predisposes humans toward development of inherited breast cancer (4) and studies on PTEN showed that deregulation of this gene is associated with increased cell proliferation and tumorigenicity (37). Cytokines including IL1A and IL1B are highly down-regulated in MCF10CA1a (Table 2) despite their reported contribution to the invasiveness and aggressive phenotype of breast cancer (38). In contrast, IL7, which has also been implicated in cancer progression (39), is significantly up-regulated in MCF10CA1a (Table 3).

Previous spectral karyotyping showed a variety of chromosomal rearrangements in breast cancer (11). In particular, the reciprocal translocation involving chromosomes 3 and 9 [t(3;9)(p14;p21)] identified in our analysis has been described as the single most prominent event in the transformation of this cell type (32, 33). We also observed several other translocations involving chromosomes 2, 3, 5, 10, and 17. Aberrations in chromosomes 3 and 17 were reported previously in breast cancer (40). Chromosome 17 has several putative breakpoints that might be associated with higher frequency of translocations of this chromosome. A gain of chromosome 8 was observed in MCF10A cells, which was lost in the subsequent premalignant (MCF10AT1) and malignant (MCF10CA1a) progression cell lines. We also identified a gain of an entire chromosome 20 in the malignant breast cell line. Consistent with these findings, chromosomal rearrangements and amplification of chromosomes 8 (41) and 20 (42) have been observed frequently in breast cancer.

Using aCGH, the parental MCF10A breast cell line showed gains in 5q23.1-35.3, 19q13.11q13.43, and 13q32.1-p32.2 regions and loss of 9p21.3, 3p26.3, 16p11.2, 21p11-q11.2, and 22q11.1 regions in comparison with a normal diploid cell line (Supplementary Fig. S1A; Table 4). The loss of the 9p21.3 region is consistent with the findings that a deletion in this region was associated with the t(3;9) translocation event that leads to immortalization of this cell line (32). A gain was also detected in the 8p23.3-q24.3 region, which is consistent with the gain of an additional copy of chromosome 8 in the spectral karyotyping analysis. Gain of 3p14.3, 3q13.3, 19q12-q34.3, 10q22.1-q22.2, 16q23.3, and 17p11.2 and loss of 5q12.1, 5q14.3-15, and 15q21.1 were observed in the premalignant MCF10AT1 cell line (Table 4; Supplementary Fig. S1B). Cytogenetic changes in chromosomes 17 (43) and 5 (44) have been reported in other studies. In particular, losses in the 5q13-q23.3 region have been associated with breast cancers showing a mutation for the BRCA1 gene (44). This region has also been shown to harbor several putative tumor suppressor genes.

Additional changes in MCF10CA1a include gains of 2p25.3-q21.2, 3p14.1-q29, 9p24.3-p11.2, 9p34.13-p34.3, 10q11.1-q26.3, and 20p13-q13.3 and losses of genomic regions 2q21.2-q22.1, 5q12.1, 8p23.3-q24.21, and 16q23.1 (Table 4; Supplementary Fig. S1C). The additional chromosome 8 seen in MCF10A was deleted in both premalignant and malignant cell lines. Deletions of certain genomic regions or the entire chromosome 8 are in agreement with previous studies on chromosome 8 abnormalities in breast cancer (45). The region 8p11-12 harbors ∼21 potential oncogenes and is frequently amplified in breast cancer (46). The results of spectral karyotyping and aCGH were in concurrence in nearly all instances thereby substantiating the results from both approaches.

A careful analyses of the aCGH findings revealed that almost all the subchromosomal genomic gains and losses were in those chromosomes that were identified as aberrant by the spectral karyotyping method. Thus, combining both these approaches in our cell system has allowed us to identify large-scale genomic aberrations and ploidy changes along with regions of subchromosomal abnormalities that might occur in the transformation of a normal cell to a cancerous cell. Overall, our spectral karyotyping and aCGH results indicate that a hallmark of progression toward breast cancer involves gradual accumulation of chromosomal aberrations in normal breast cells.

We further correlated genomic alterations with gene expression profiles of normal and malignant MCF10 cell lines. Regions within the MCF10CA1a cell line identified for gains and losses by aCGH were examined for changes in levels of gene expression (Table 5; Supplementary Table S4). The majority of the ∼1,000 genes investigated showed a lack of direct correlation between the chromosomal aberrations and its expression (Table 5). Previous studies also reported a weak, or in some cases, a complete lack of correlation between genomic gains and losses and gene expression levels in cancer cells (47). Despite this, we identified several genes in these regions that are related to breast cancer and show large changes in gene expression in the malignant MCF10CA1a versus the MCF10A breast epithelial cells (Supplementary Table S4).

Our results predict that progression to cancer is a combination of genomic instability as well as gene deregulation. Chromosomal aberrations such as amplifications, deletions, and complex chromosomal translocations are a hallmark of solid tumors and are generally believed to occur via telomere dysfunction and the breakage-fusion-bridge mechanism (48). Changes in gene copy number that occur due to the chromosomal aberrations lead to a disruption of the normal expression pattern of the genes, therefore causing transformation of a normal cell toward a malignant phenotype.

In conclusion, our combined approach employing spectral karyotyping, aCGH, and microarray gene expression analysis identified several important chromosomal regions and genes with structural and functional alterations that might be involved in the progression and development of breast cancer. Importantly, several of these genomic regions that have been shown previously to be amplified or deleted in breast cancer (49). Although the expression levels of many genes agreed with previously published results in breast cancer, there were some significant differences. This could relate to the multiple changes and likely variability in gene expression that occur during cancer progression. Thus, a variety of combinations of expression changes in key regulatory factors may lead to the same end of a highly malignant tumor.

We also report a lack of correlation between chromosomal aberrations in cancer and expression patterns of genes within those regions. Other mechanisms of gene regulation are likely involved including epigenetic processes such as methylation (50), histone modification (50), and expression of noncoding RNAs (50). Methylation of CpG residues and histone tail modifications are faithfully propagated from one cell generation to another under normal circumstances and any change might lead the cell to become metastatic. In cancer cells, methylation of cytosine in the CpG dinucleotides occurs in the promoter regions, which leads to gene inactivation (50). Several genes that are important in cell cycle control, DNA repair, and apoptosis have been found to undergo hypermethylation, thereby promoting the development of a cancerous phenotype (50). Modification of histone tails such as acetylation, deacetylation, and methylation are also known to influence gene expression (50), and in cancer, abnormal histone modifications that occur have shown to have an effect on gene transcription (50). Recent investigations also show a role of microRNAs in repression of certain important genes in cancer (50). Our studies, therefore, add to the growing evidence that amplification or deletions of genomic regions are not the sole mechanism for altering gene expression in progression toward cancer.

No potential conflicts of interest were disclosed.

Grant support: NIH grant GM-072131 (R. Berezney).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.