FOXA1 binding sites with high FAIRE-chip signals correlate with gene expression in breast cancer. Distribution of FOXA1 binding sites with different FAIRE enrichments relative (within 20 kb of the TSS) to genes co-expressed or non-co-expressed with FOXA1 in primary breast tumors (van de Vijver et al. 2002; Wang et al. 2005). ** Indicates a statistically significant difference between the distribution of FOXA1 binding sites near correlated and non-correlated genes (P < 0.01).

High FAIRE enrichment at FOXA1 binding sites correlate with a shift toward active histone marks. (A) H3K9ac and H3K4me2 ChIP-chip signals at FOXA1 binding sites from MCF7 or specific to LNCaP cells were analyzed. Data represent mean ± SEM of signals derived from MAT analysis of the ChIP-chip data. *** Indicates a statistically significant difference (P < 0.001) between FOXA1 sites from MCF7 with high FAIRE versus low FAIRE enrichment. (B) ChIP-qPCR experiments were performed to monitor H3K9me1 and me2 levels in MCF7 cells at the indicated categories of FOXA1 recruitment regions from MCF7 cells or specific to LNCaP cells (LNCaP only). Data show relative enrichments compared to negative control regions. Data are mean ± SEM from at least three independent experiments. Statistical significance of the difference between FOXA1 recruitment sites with high versus low FAIRE enrichments is shown (** indicates P < 0.01).

FAIRE-chip signal from MCF7 cells at FOXA1 binding sites. FAIRE-chip signal from MCF7 cells within high confidence FOXA1 recruitment sites from MCF7 cells (lying outside of 1 kb promoters) or specific to LNCaP cells is indicated. A set of randomly selected Forkhead motif-containing regions not bound by FOXA1 in MCF7 cells (Unbound FKHR) as well as randomly selected regions were also used. Data represent mean ± SEM of signals derived from MAT analysis of FAIRE-chip data. *** Indicates a statistically significant difference (P < 0.001).

Cell-specific activity of conserved FOXA1 recruitment sites. (A) Comparison of FOXA1 binding sites identified in MCF7 and LNCaP cells by ChIP-chip (Lupien et al. 2008). (B) FOXA1 binding sites shared between MCF7 and LNCaP and harboring low FAIRE enrichment in MCF7 cells were selected. These sites were then classified relative to their FAIRE enrichment levels (low, medium, or high) in LNCaP cells. (C) Histogram indicating the percentage of sites defined in (B) that also recruits AR in LNCaP cells. ** Indicates a statistically significant difference (P < 0.01) between FOXA1 binding sites with high and low FAIRE enrichment in LNCaP cells. (D) FOXA1 binding sites shared between MCF7 and LNCaP and harboring low FAIRE enrichment in LNCaP cells were selected. These sites were then classified relative to their FAIRE enrichment levels (low, medium, or high) in MCF7 cells. (E) Histogram indicating the percentage of sites defined in (B) that also recruits ESR1 in MCF7 cells. *** Indicates a statistically significant difference (P < 0.001) between FOXA1 binding sites with high and low FAIRE enrichment in MCF7 cells.

Correlation between selective activity of common FOXA1 binding sites and cell-type-specific gene regulation. Examples of genes associated with FOXA1 binding sites common to MCF7 and LNCaP cells, but regulated in a cell-type-specific manner. Individual probe level signal for FAIRE-chip and ChIP-chip of the indicated factors or histone marks within the FOXA1 binding sites is shown. Transcriptional regulation by dihydrotestosterone (DHT) in LNCaP cells (Wang et al. 2007) or by estradiol (E2) in MCF7 cells (Carroll et al. 2006) was determined using a t-test (P < 5 × 10⁻³) and is indicated by blue and red bars, respectively. (A) Example of a gene regulated by DHT in LNCaP, but not by E2 in MCF7. (B) Example of a gene regulated by E2 in MCF7, but not by DHT in LNCaP. Note that shared FOXA1 binding sites with selective activities are often associated with cell-type-specific ones in the vicinity of target genes. This is consistent with the clustering of transcription factor recruitment regions nearby regulated genes (Chan and Song 2008; Krum et al. 2008).

Comparison of FOXA1 recruitment sites within MCF7, LNCaP, and DLD1 cells. (A) Venn diagram showing the overlap between FOXA1 recruitment sites identified in three different cell lineages, i.e., breast (MCF7), prostate (LNCaP), and colon (DLD1) cells. Only high confidence binding sites were used here (FDR1). (B) Histogram showing the percentage of high confidence FOXA1 binding sites (FDR1) from each cell-line that is also found in at least one of the other two cells at FDR1 or 20. (C) FAIRE-qPCR experiments were performed in MCF7, LNCaP, and DLD1 cells. CCND1 proximal promoter, used as a positive control (Eeckhoute et al. 2006), and 20 FOXA1 binding sites shared between the three cell-lines were analyzed. Relative enrichments compared to negative control regions are shown. Data are mean ± SD from two or three independent experiments. * Indicates statistically significant differences (P < 0.05) between FAIRE enrichments in DLD1 and both MCF7 and LNCaP cells.

FOXA1 is required but not sufficient for high FAIRE enrichment at bound regulatory elements. (A) FOXA1 binding sites were divided into tertiles and the average FAIRE-chip signal (based on MAT scores, see Methods) for each subset of sites (low, medium, and high FAIRE) was calculated. Each subset was comprised of 645 sites. FAIRE-chip enrichments at FOXA1 LNCaP-specific sites were also analyzed. Data represent mean ± SEM of signals derived from MAT analysis of FAIRE-chip data. (B) FAIRE-qPCR experiments were performed in MCF7 cells to monitor FAIRE enrichment at the indicated categories of FOXA1 recruitment regions (at least 8 different sites were analyzed for each subset). Relative enrichment compared to negative control regions is shown. Data are mean ± SD from three independent experiments. (C) Signals from FOXA1 ChIP-chip in MCF7 cells within FOXA1 binding sites with low, medium or high FAIRE-chip enrichments. LNCaP-specific sites were also analyzed. Data represent means ± SEM of signals derived from MAT analysis of the FOXA1 ChIP-chip data from Lupien et al. (2008). (D) FAIRE-qPCR experiments performed as in B. Decrease in FAIRE enrichments triggered by FOXA1 silencing is shown. Data are mean ± SD from three independent experiments. *** and * Indicate a statistically significant difference between FOXA1 recruitment sites with high and low FAIRE enrichments (P < 0.001 and P < 0.05, respectively).