Estrogen and insulin-like growth factor-I (IGF-I) independently down-regulate critical repressors of breast cancer growth

Abstract

Although estrogen receptor alpha (ERα) and insulin-like growth factor (IGF) signaling are important for normal mammary development and breast cancer, cross-talk between these pathways, particularly at the level of transcription, remains poorly understood. We performed microarray analysis on MCF-7 breast cancer cells treated with estradiol (E2) or IGF-I for 3 or 24 h. IGF-I regulated mRNA of five to tenfold more genes than E2, and many genes were co-regulated by both ligands. Importantly, expression of these co-regulated genes correlated with poor prognosis of human breast cancer. Closer examination revealed enrichment of repressed transcripts. Interestingly, a number of potential tumor suppressors, for example, B-cell linker (BLNK), were down-regulated by IGF-I and E2. Analysis of three down-regulated genes showed that E2-mediated repression occurred independently of IGF-IR, and IGF-I-mediated repression occurred independently of ERα. However, repression by IGF-I or E2 required common kinases, such as PI3K and MEK, suggesting downstream convergence of the two pathways. In conclusion, E2 and IGF-I co-regulate a set of genes that affect breast cancer outcome. There is enrichment of repressed transcripts, and, for some genes, the down-regulation is independent at the receptor level. This may be important clinically, as tumors with active ERα and IGF-IR signaling may require co-targeting of both pathways.

FIG. 1

Global gene transcription patterns initiated by E2 and IGF-I treatment in vitro. (A) Affymetrix profiles were taken of MCF-7 cells in serum-free medium (SFM) with or without the presence of either E2 (10nM) or IGF-I (100ng/mL) for 3hr or 24hr. Shown are genes found to be significantly (p<0.01) induced (yellow) or repressed (blue) following ligand stimulation (fold change>1.5). (B) Venn diagrams showing the extensive and significant (p<0.001) overlap of target genes regulated by both E2 and IGF-I at 3hr and 24hr. (C) Quantitative real-time RT-PCR (qRT-PCR) was used to calculate relative mRNA expression as ligand-mediated fold change compared to the untreated SFM control. MCF-7 cells were starved overnight in SFM and then treated with either E2 or IGF-I for 3hr or 24hr. RNA of select genes from each section of the Venn diagrams was measured by qRT-PCR. The data are an average of three replicates ± standard error of the mean (SEM).

FIG. 2

An activated E2/IGF-I co-regulated gene signature in clinical breast tumors is associated with poor prognosis. (A) Kaplan-Meier analysis of the van de Vijver tumor profile data set comparing the differences in risk among three groups of patients when only tumors classified as ER-positive were considered. Green line, tumors having an “activated” co-regulated gene signature (i.e. significant correlation, p<0.01, to the pattern of up- and down-regulation by E2 and IGF-I observed in vitro); pink line, “deactivated” tumors that have low similarity (i.e. anti-correlation, p<0.01) to the E2/IGF-I pattern; and yellow line, tumors that were neither similar nor dissimilar to the co-regulated gene expression pattern (p>0.01). The log-rank test evaluated whether there are significant differences among any of the three groups. (B) An E2/IGF-I co-regulated gene expression signature was defined and the 226 ER-positive tumors from the van de Vijver data set were analyzed. Tumors having high expression of induced genes (yellow) and low expression of repressed genes (blue) were classified as “activated” (black box outline). These tumors were found to correlate with the luminal B subtype (red) of ER-positive breast cancers. In contrast, tumors classified as “deactivated” correlated with the luminal A subtype (blue).

FIG. 3

Enrichment of repressed targets within the E2/IGF-I co-regulated gene set. (A) Genes found to be induced (yellow) by E2 at both 3hr and 24hr or repressed (blue) by E2 at both time points were identified. Microarray data was then used to examine expression of these genes in response to IGF-I treatment followed by examination of overlap between E2 and IGF-I regulation of targets within the induced and repressed groups. (B) Probe sets within the co-regulated group of genes (p<0.05; fold>1.1 for induction, fold<0.9 for repression) were analyzed at both 3hr and 24hr and divided into one of four groups—induced by both E2 and IGF-I, repressed by both ligands, induced by E2 but repressed by IGF-I, and repressed by E2 but induced by IGF-I. ↑ = induced, ↓ = repressed. (C) Quantitative real-time RT-PCR (qRT-PCR) was used to calculate relative mRNA expression as ligand-mediated fold change compared to the untreated SFM control. MCF-7 cells were starved overnight in SFM and then treated with either E2 (10nM) or IGF-I (100ng/mL) for 3hr or 24hr. The data are an average of three replicates ± standard error of the mean (SEM).

FIG. 4

Repression of target genes by E2 and IGF-I is independent at the receptor level but relies on common downstream kinases. (A and B) MCF-7 cells were starved overnight in SFM and then pre-treated for 30min with either the IGF-IR inhibitor BMS (1μM) or the anti-estrogen ICI (100nM). Following pre-treatment, cells were stimulated with E2 (10nM) or IGF-I (100ng/mL) for 3hr (A) or 24hr (B). RNA was isolated and qRT-PCR was performed. The data are an average of three replicates ± standard error of the mean (SEM). (C) MCF-7 cells were starved overnight in SFM and then pre-treated for 30min with either BMS (1μM), ICI (100nM), or the combination of BMS+ICI. Cells were then simultaneously treated with E2+IGF-I for 3hr. RNA was isolated and qRT-PCR was performed. The data are an average of three replicates ± standard error of the mean (SEM). (D) MCF-7 cells were maintained in SFM overnight and then pre-treated for 30min with either LY-294002 (20μM), U0126 (10μM), or Gö6983 (0.25μM). Following pre-treatment, cells were administered E2 (10nM), IGF-I (100ng/mL), or the combination of E2+IGF-I for 24hr. RNA was isolated and qRT-PCR was performed. The data are an average of three replicates ± standard error of the mean (SEM).

FIG. 5

Estradiol and IGF-I independently repress BLNK, a regulator of MCF-7 cell growth. (A) MCF-7 cells were starved overnight in SFM, and then treated with E2 (10nM), IGF-I (100ng/mL), or the combination of E2+IGF-I for 48hr. Cells were harvested and lysed in 5% SDS. Western blot was performed using antibodies directed against BLNK and β-actin, which was used as a loading control. (B) MCF-7 cells were transiently tranfected with BLNK siRNA (dashed lines) or non-specific control siRNA (solid lines), and then incubated in medium containing 5% charcoal-stripped serum (top graph) or 5% fetal bovine serum (FBS; bottom graph). Cell growth was assessed by MTS assay on days 0, 1, 2, and 4. The assay was performed with biological quintuplicates, and each point represents the average value ± SEM. The very bottom panel represents a Western blot showing BLNK protein levels from MCF-7 cells that were cultured for 72hr in medium containing 5% FBS or 5% charcoal-stripped serum (CSS). (C) MCF-7 cells were transfected with the BLNK −0.8kb promoter and maintained in SFM or treated with E2 (10nM), IGF-I (100ng/mL), or E2+IGF-I for 24hr. Luciferase reporter assays were performed. The data represent relative luciferase units (RLU) and are an average of three replicates ± SEM. (D) ChIP assays for RNA polymerase II and IgG (negative control) in MCF-7 cells maintained in SFM or after treatment with E2 (10nM), IGF-I (100ng/mL), or E2+IGF-I for 1hr. Quantitative real-time PCR was performed with primers designed around the BLNK transcriptional start site. The data is represented as % input and is an average of three independent experiments ± SEM.