Steroid Receptors Reprogram FoxA1 Occupancy through Dynamic Chromatin Transitions

Abstract

The estrogen receptor (ER), glucocorticoid receptor (GR), and forkhead box protein 1 (FoxA1) are significant factors in breast cancer progression. FoxA1 has been implicated in establishing ER-binding patterns though its unique ability to serve as a pioneer factor. However, the molecular interplay between ER, GR, and FoxA1 requires further investigation. Here we show that ER and GR both have the ability to alter the genomic distribution of the FoxA1 pioneer factor. Single-molecule tracking experiments in live cells reveal a highly dynamic interaction of FoxA1 with chromatin in vivo. Furthermore, the FoxA1 factor is not associated with detectable footprints at its binding sites throughout the genome. These findings support a model wherein interactions between transcription factors and pioneer factors are highly dynamic. Moreover, at a subset of genomic sites, the role of pioneer can be reversed, with the steroid receptors serving to enhance binding of FoxA1.

Figure 1. Binding patterns of FoxA1, ER, and GR in hormone treated MCF-7 breast cancer cells

ChIP-Seq binding profiles are shown for ER, GR, and FoxA1 for untreated cells or cells treated with E2 or Dex. A. Venn diagram represents the number of ER+E2 binding sites that overlap with FoxA1+E2 or FoxA1 untreated binding sites. B. Venn diagram demonstrates the overlap of ChIP-seq GR+Dex induced binding patterns with FoxA1+Dex and FoxA1 untreated binding patterns. Each group identified has been notated with a lower case letter. C. Heat map represents the binding intensity of FoxA1 untreated, FoxA1+E2, and ER+E2 at the specific binding groups characterized by the Venn diagram and labeled by lower case letters. The heat map is presented as the number of reads per 10⁶ sequences with the position of the reads in a 2 kb region flanking the center of the peak. Group “a”, “b”, and “c” is an example of 1,000 binding sites from the individual groups and highlighted in yellow. D. The heat map represents the ChIP-seq tag density of FoxA1 untreated, FoxA1+Dex, and GR+Dex peaks at the binding patterns identified in the Venn diagram and labeled by a lower case letter. The heat map is presented as the number of reads per 10⁶ sequences with the position of the reads in a 2 kb region flanking the center of the peak. Group “h”, “i”, and “j” is an example of 1,000 binding sites from the individual groups and highlighted in yellow. E. Three histograms represent the average tag count of all ER+E2 ChIP-seq samples, FoxA1 untreated ChIP-seq samples, and FoxA1+E2 ChIP-seq samples at the specific binding groups of “d”, “e”, and “f” over a 2 kb region. F. The three histogram represent the average tag count of all FoxA1 untreated, FoxA1+Dex, and GR+Dex ChIP-seq samples over a 2 kb distance for sites identified at groups “k”, “l”, and “m”. See also Figure S1.

Figure 2. Positioning analysis and differential hormone regulation of FoxA1 in MCF-7 breast cancer cells

The average location of FoxA1 in relation to ER and GR peaks over a 10 kb range. A–B. Cumulative distribution function plotted to determine the location of all FoxA1 (untreated and E2 treated sites) in relation to all ER+E2 binding sites (broken line) and ERE motif (solid line) over a 500 bp range (A) and 10 kb range (B). C–D. The distance of all FoxA1 (untreated and Dex) peaks in relation to GR+Dex binding sites (broken line) and GRE motif (solid line) over a 500 bp range (C) and 10 kb rang (D) has been plotted as a cumulative distribution function. E. The percentage of FoxA1 peaks within a 10 kb range of the closest ER peak. FoxA1 sites identified in group “f” (black), “e” (grey), and “a” (white). (Expanded view) The percentage of FoxA1 peaks within a 100 bp range of the closest ER peak. F. The percentage of FoxA1 peaks closest to a GR peak within a 10 kb distance has been calculated and separated into sites identified in the individual binding groups for “m” (black), “l” (grey), and “h” (white). (Expanded view) The distance has been decreased to 100 bp and the percentage calculated for sites identified in individual groups. G. Scatterplot shows the genome-wide changes in FoxA1 binding in untreated compared with E2 treated cells. Green points donate sites induced by E2, red points donate sites lost by E2 and blue points donate sites unchanged by E2 treatment. Sites gained or lost by E2 have a 2-fold change in tag density over the background of the E2 treatment. H. The genome-wide changes in FoxA1 binding in untreated compared with Dex treated cells are displayed in the scatterplot. Green points donate sites induced with Dex, red points donate sites lost by Dex and blue points donate sites unchanged with Dex. Sites gained or lost by Dex have a 2-fold change in tag density over the back ground of the Dex treatment. See also Figure S2.

Figure 3. Supervised clustering analysis reveals different FoxA1 binding modules in MCF-7 cells

A. Supervised cluster analysis of ER binding sites and FoxA1 binding patterns found to be gained or lost with E2 treatment with a 2-fold change in tag density reveals three specific clusters with cluster 3 representing ER-induced FoxA1 sites. The heat map represents the number of reads per 10⁶ sequences and the position of the reads in a 1 kb region flanking the peak. B. Histogram represents the average tag count per bp over a 2 kb range of FoxA1 and ER at the three identified clusters. C. De novo motif analysis has been conducted on sites identified as ER-induced FoxA1 sites and compared to the FoxA1 classical cluster omitted from the supervised clustering analysis. D. Supervised clustering analysis of GR+Dex peaks and FoxA1 untreated and Dex peaks at FoxA1 sites gained or lost by Dex treatment reveals four specific clusters with cluster 1 representing GR-induced FoxA1 sites. The heat map represents the number of reads per 10⁶ sequences and the position of the reads in a 1 kb region flanking the peak. E. Histogram represents the average tag count per bp over a 2 kb range of FoxA1 and GR at the four identified clusters. F. De novo motif analysis for sites identified as GR-induced FoxA1 sites are compared to the FoxA1 classical cluster. See also Figure S3.

Figure 4. Chromatin accessibility altered at hormone induced FoxA1 binding sites

A. Box plot analysis comparing cells treated with E2 or left untreated show there are significant changes in DHS at the ER-induced FoxA1 sites (cluster 3) compared with the FoxA1 classical cluster. Genomic region illustrating changes in DHS at a specific ER-induced FoxA1 site (UCSC browser shot). B. Analysis of DHS-seq in cells treated with either Dex or left untreated demonstrate a significant increase in chromatin remodeling at GR-induced FoxA1 sites (cluster 1) compared with the FoxA1 classical cluster. A genomic region identified at a GR-induced FoxA1 site illustrating changes in (UCSC browser shot). P values are determined by the Two-sample Kolmogorov Smirnov test. See also Figure S4.

Figure 5. Single-molecule imaging analysis demonstrates a short residence time for FoxA1 under all treatment conditions

Top panel represents the bound fractions and the average residence time for FoxA1 in (A) untreated cells, (B) E2 treated cells, (C) Dex treated cells, for ER in (D) untreated cells, (E) E2 treated cells, and for GR in (F) Dex treated cells. Pie charts (A–F) represent percentage of molecules unbound (white), bound at the fast short-lived fraction (green), and bound at the slow long-lived fraction (blue). The average residence time of fast short-lived and slow long-lived fraction is presented next to their representative fraction. G. Box plot represents the dwell time for all molecules in the fast short-lived fraction (green box) and the slow long-lived fraction (blue box) for FoxA1 in untreated, E2, and Dex treated cells, ER in untreated, and E2 treated cells, and GR in Dex treated cells. The number of molecules in the long-lived fraction is presented as n. P value represents a Two-sample Kolmogorov Smirnov test defined by the brackets. See also Figure S5.

Figure 6. Lack of DHS protection at the FoxA1 motif with in FoxA1 binding site in MCF-7 breast cancer cells

Footprint analysis in MCF-7 cells for (A) FoxA1 untreated cells, (B) treated with E2, (C) treated with Dex, (D) for ER treated with E2 (E), and for GR treated with Dex. The top panel represents the DNase I cleavage with the observed profile designated in red and the expected profile designated in green. The observed profiles represent the average raw DNase cut counts over the cognate motif element identified by the ChIP-seq peaks. The expected profiles represent the average of DNA hexamer frequencies from the naked DNA cut counts using the hexamers centered at each bp position. The bottom panel represents the log ratio of observed versus expected profiles. The black broken line denotes where the observed cut count levels match the expected. The red broken line is the mean of the log ratio of observed over expected cut counts at the motif region extended by 2 bp in both directions. The difference between the black and red broken lines represents the footprint depth. See also Figure S6.