Delineating genetic alterations for tumor progression in the MCF10A series of breast cancer cell lines

Abstract

To gain insight into the role of genomic alterations in breast cancer progression, we conducted a comprehensive genetic characterization of a series of four cell lines derived from MCF10A. MCF10A is an immortalized mammary epithelial cell line (MEC); MCF10AT is a premalignant cell line generated from MCF10A by transformation with an activated HRAS gene; MCF10CA1h and MCF10CA1a, both derived from MCF10AT xenografts, form well-differentiated and poorly-differentiated malignant tumors in the xenograft models, respectively. We analyzed DNA copy number variation using the Affymetrix 500 K SNP arrays with the goal of identifying gene-specific amplification and deletion events. In addition to a previously noted deletion in the CDKN2A locus, our studies identified MYC amplification in all four cell lines. Additionally, we found intragenic deletions in several genes, including LRP1B in MCF10CA1h and MCF10CA1a, FHIT and CDH13 in MCF10CA1h, and RUNX1 in MCF10CA1a. We confirmed the deletion of RUNX1 in MCF10CA1a by DNA and RNA analyses, as well as the absence of the RUNX1 protein in that cell line. Furthermore, we found that RUNX1 expression was reduced in high-grade primary breast tumors compared to low/mid-grade tumors. Mutational analysis identified an activating PIK3CA mutation, H1047R, in MCF10CA1h and MCF10CA1a, which correlates with an increase of AKT1 phosphorylation at Ser473 and Thr308. Furthermore, we showed increased expression levels for genes located in the genomic regions with copy number gain. Thus, our genetic analyses have uncovered sequential molecular events that delineate breast tumor progression. These events include CDKN2A deletion and MYC amplification in immortalization, HRAS activation in transformation, PIK3CA activation in the formation of malignant tumors, and RUNX1 deletion associated with poorly-differentiated malignant tumors.

Figure 1. Figure 1 shows DNA copy number variation in chromosomes 5, 8, 9, 10, and 19.

The graph was generated using the Affymetrix Genome Browser. Genomic position is displayed on the x-axis and log2ratio (tumor hybridization intensity normalized by diploid HapMap reference samples) is on the y-axis. DNA copy number gains at 5q23.1-qter, 8q24.21, and 10q22.1 are seen in all four cell lines. Homozygous deletion at the CDKN2A locus in chromosome 9p21.3 occurs in all four cell lines. Note that the homozygous deletion (400 kb) is embedded within a hemizygous deletion region (4 Mb). 19q13-qter copy number gain is observed in the MCF10A, MCF10AT, and MCF10CA1h cell lines.

Figure 2. Intragenic RUNX1 deletions in the MCF10A series of cell lines.

Figure 2A illustrates intragenic DNA deletion and partial deletion in MCF10CA1a and MCF10CA1h, respectively. The deletions span the promoter regions of two of the transcripts. Note that the transcripts are oriented from right to left in this figure. Figure 2B summarizes altered transcripts detected in MCF10CA1a cells due to indicated genomic deletions. Deletion structure is illustrated here based on the data shown in Figure 2A. Three alternative transcripts of the RUNX1 gene are shown here. Truncated transcripts involving removal of exons 2–5 (truncated type 1) or exons 2–6 (truncated type 2). Figure 2C shows Western blot analysis of the RUNX1 protein in the MCF10A series of cell lines. RUNX1 protein expression was analyzed by Western blot using anti-RUNX1 antibody. The β-actin staining shows a similar level of protein loading in all lanes. Note the dramatic reduction of the RUNX1 protein in MCF10CA1a cells. MCF10CA1h cells show a reduced level of RUNX1 protein.

Figure 3. Expression of RUNX1 gene in primary breast tumors.

Figure 3A shows qRT-PCR analysis of the RUNX1 mRNA expressions in primary breast tumors and the adjacent normal samples. 25 of 29 our analyzed tumors have histological grade information. High histological grade tumors (n = 16) have significant reduction of RUNX1 expression compared to low/mid grade tumors (n = 9) or to the adjacent normal samples (n = 11). Data are summarized by box plots. The box represents gene expression values between 1^st and 3^rd quartiles; the line denotes median. Gene expression difference between high-grade (grade 3) and low/mid-grade (grade 1 plus grade 2) is significant (pvalue = 0.044, by t-test). Figure 3B shows RUNX1 mRNA expression in Ivshina breast microarray dataset. Ivshina breast dataset consists of 289 tumors (68 Grade 1, 166 Grade 2, and 55 Grade 3). Affymetrix U133A array expression data showed progressive reduction of RUNX1 gene (reporter: 210365_at) from Grade 1 to Grade 2 to Grade 3. The progressive reduction of RUNX1 expression in higher grade tumors is highly significant (p-value = 3×10⁻¹³ by linear regression analysis). Oncomine™ (Compendia Bioscience, Ann Arbor, MI) was used for this analysis.

Figure 4. Identification of PIK3CA mutation in the MCF10CA1h and MCF10CA1a cells.

Figure 4A shows a segment of the sequencing chromatogram of PIK3CA . MCF10A and MCF10AT cells have a wild type A-allele marked by an arrow. In contrast, MCF10CA1h and MCF10CA1a cells exhibit predominance of the mutant G-allele in addition to the wild type A-allele. The mutation results in an amino acid change from His to Arg at 1047 (H1047R). Figure 4B shows increased phosphorylation of AKT1 Ser473 in the MCF10CA1h and MCF10CA1a cells. An antibody targeting phosphorylated Ser473 in AKT1 recognized specifically the phosphorylated form of the protein, which was dramatically increased in cells containing the PIK3CA activating mutation, i.e. the MCF10CA1h and MCF10CA1a cells. β-actin served as a control for protein loading.

Figure 5. The effect of chromosome copy number gain on gene expression in the MCF10A series of cell lines.

For each cell line, genes are grouped into those showing copy number (CN) greater than 2 (CN >2, dashed lines) and those showing copy number equal to or less than 2 (CN ≤2, solid lines). Expression levels of genes are indicated along the X-axis. The Y-axis represents the cumulative distribution function (cdf), which describes cumulative probability of gene expression less than or equal to the threshold level indicated by the value on X-axis.The data depicted in this figure indicate show largedifference between the expression levels for genes with copy number gain (CN >2) versus those with no gain (CN ≤2) only in the two malignant cell lines, MCF10CA1h and MCF10CA1a.