An integrated bioinformatics approach identifies elevated cyclin E2 expression and E2F activity as distinct features of tamoxifen resistant breast tumors

Abstract

Approximately half of estrogen receptor (ER) positive breast tumors will fail to respond to endocrine therapy. Here we used an integrative bioinformatics approach to analyze three gene expression profiling data sets from breast tumors in an attempt to uncover underlying mechanisms contributing to the development of resistance and potential therapeutic strategies to counteract these mechanisms. Genes that are differentially expressed in tamoxifen resistant vs. sensitive breast tumors were identified from three different publically available microarray datasets. These differentially expressed (DE) genes were analyzed using gene function and gene set enrichment and examined in intrinsic subtypes of breast tumors. The Connectivity Map analysis was utilized to link gene expression profiles of tamoxifen resistant tumors to small molecules and validation studies were carried out in a tamoxifen resistant cell line. Despite little overlap in genes that are differentially expressed in tamoxifen resistant vs. sensitive tumors, a high degree of functional similarity was observed among the three datasets. Tamoxifen resistant tumors displayed enriched expression of genes related to cell cycle and proliferation, as well as elevated activity of E2F transcription factors, and were highly correlated with a Luminal intrinsic subtype. A number of small molecules, including phenothiazines, were found that induced a gene signature in breast cancer cell lines opposite to that found in tamoxifen resistant vs. sensitive tumors and the ability of phenothiazines to down-regulate cyclin E2 and inhibit proliferation of tamoxifen resistant breast cancer cells was validated. Our findings demonstrate that an integrated bioinformatics approach to analyze gene expression profiles from multiple breast tumor datasets can identify important biological pathways and potentially novel therapeutic options for tamoxifen-resistant breast cancers.

Figure 1. Venn diagram showing overlap between three sets of differentially expressed genes.

The numbers of differentially expressed genes, which were either over-expressed or under-expressed in tamoxifen-resistant compared to sensitive tumors, are 275 for GSE6532, 130 for GSE9195, and 252 for GSE9893.

Figure 2. Expression level of Cyclin E2 mRNA in breast tumors from different intrinsic subtypes.

The mean expression level of Cyclin E2 was determined in breast tumors of different intrinsic subtypes as described in Materials and Methods. The Log2 transformed mean expression level is relative to the mean expression level in all tumors in the breast tumor compendium.

Figure 3. Results of clustering and survival analysis.

(A) Clustering results of untreated Luminal A and Luminal B breast tumors from the compendium gene expression profile using principle components of the DE genes identified in GSE6532 (left), GSE9195 (middle) and GSE9893 (right). Green: luminal A tumors; Pink: luminal B tumors; (B) Kaplan-Meier estimation of survival for stratified ER+ tumors using the DE genes identified in GSE6532 (left), GSE9195 (middle) and GSE9893 (right); the number in the parentheses is the number of ER+ tumors in each cluster. (C) Kaplan-Meier estimation of survival for stratified Luminal A tumors using the DE genes identified in GSE6532 (left), GSE9195 (middle) and GSE9893 (right); the number in the parentheses is the number of Luminal A tumors in each cluster.

Figure 4. Phenothiazines inhibit proliferation and down-regulate Cyclin E2 expression in tamoxifen-resistant MCF-7 breast cancer cells.

(A) Cells were treated for five days with increasing doses of the three phenothiazine compounds in the absence or presence of 1 µM 4-hydroxytamoxifen (4OHT) as indicated. Cell viability was determined by methylene blue staining and expressed as % of vehicle treated control cells. (B) A BrdU assay was carried out after 48 hr of treatment with 5 µM of each phenothiazine drug. (C) Cyclin E2 mRNA levels were determined by QPCR following 24 hr treatment with 5 µM of each phenothiazine as indicated. All data represent the mean +/− SEM from three independent determinations. *, P<0.05, **, P<0.01.

Figure 5. Prochlorperazine inhibits growth of BT474 cells.

Cells were treated 5 µM of prochlorperazine for 5 days and cell proliferation was measured by methylene blue staining (A) or for 2 days and cyclin E2 mRNA levels were by QPCR (B).