We have assessed the impact of progesterone (P4) on breast cancer cell lines and have found PR ligands to be antiproliferative. We showed that PR activation, directly impacts ER chromatin binding, by displacing ER from regulatory elements adjacent to cell cycle target genes and sequestering the ER complex at different cis-regulatory elements. As such, PR has the capacity to 'hide' ER in the genome. It has been known for many years, that PR ligands (i.e. Megace) can have antitumour activity in ER+ breast cancer patients, but the explanation for this response was unknown, until our discoveries. Mechanistically, we could show that PR competes with ER for shared, rate-limiting co-factors and that activation of PR forces ER to be redistributed throughout the genome. The findings from this part of the proposal have resulted in three independent clinical trials (PIONEER, run in UK, PEARL, run in the UK and WINPRO, run in Australia).
We have used CRISPR technologies to engineer the frequent GATA3 mutation in MCF-7 cells. We selected clonal cell lines and screened >500 individual clones. Unexpectedly, this GATA3 mutant clonal cell lines has dramatically different growth properties, with overall decreased growth potential, in contrast to what was expected. We could validate this in large patient sample cohorts, where GATA3 mutations were linked with better overall patient survival. This paradoxical set of observations suggested that a frequent breast cancer mutation was protective, a finding that was at odds with the original hypothesis. We have characterised the GATA3 mutant cell line and have found that the mutant GATA3, which possesses an additional ~65 amino acids of protein sequence on the C-terminus, is able to occupy distinct places within the genome. The mutant-specific GATA3 binding sites correlated with novel ER and H3K27Ac marks, suggesting that the mutant GATA3 was able to redistribute ER to new sites in the genome, that became transcriptionally active. This work is currently under review at Cell Reports. We have found an unexpected functional connection between GATA3 and TET2. TET2 is the enzyme that converts DNA methylation to 5hmC and there is literature suggesting that 5hMC can modulate gene expression and we hypothesise that the connection between GATA3 and TET2 may be intimately involved in this. This body of work has been completed and is currently under review at Cell Reports.
FOXA1 is the archetypal ER pioneer factor. The theory was the FOXA1 occupied poised enhancers that were demarcated by H3K4me1/me2 modification. We discovered an interaction between FOXA1 and MLL3, the methyltransferase that deposits the histone marks at enhancer elements. We could show that MLL3 is recruited to ER enhancers, via FOXA1, where it contributes to deposition of methyl groups to form H3K4me1 and H3K4me2 marks. This work was published in Cell Reports (Jozwik et al). During the course of the work, a manuscript was published which suggested that FOXA1 binding could be influenced by hormones, such as estrogen (E2) or glucocorticoids. To assess this we conducted a comprehensive analysis of FOXA1 binding in two model systems, using two distinct FOXA1 antibodies and a minimum of three, well matched biological replicates. When conducted at that technical level, we observed that 99% of FOXA1 binding sites did not change in the presence or absence of E2 and that this conclusion was observed in multiple cell line models, using multiple antibodies. This work was published in Cell Reports (Glont et al).
We have established a method called qPLEX-RIME. The method was published in Nature Comms (Papachristou et al). We identified STAT3 as a consistent physical interactor of ER in ER+ cell line models. Detailed analysis revealed that the cytokine IL6 (which activates STAT3) can induce phospho-STAT3 expression within ER+ cancer epithelial cells. By exploiting a novel in vivo metastatic system called MIND (intraductal) we could show that IL6 treatment results in increased metastatic burden. Mechanistically, we showed that STAT3 is recruited specifically to FOXA1/ER binding sites in the genome, but surprisingly, STAT3 binding is independent of both ER and FOXA1. Our results showed that ER targeted drugs do not influence STAT3 activity, suggesting that standard-of-care treatments will not affect the metastatic potential of STAT3 activity. This work was published in Cancer Cell (Siersbaek et al). Our genome-wide CRISPR screens revealed a role for ARID1A in the response to three different anti-growth agents. ARID1A was the most essential gene for tamoxifen and fulvestrant to work, but was the the number one 'hit' in JQ1 (BET inhibitor) treated cells, but in the opposite direction, where loss of ARID1A sensitised cells to JQ1. We identified the underlying mechanistic explanation for this, whereby ARID1A recruits HDAC1 to maintain a deacetylated state. Loss of ARID1A results in decreased HDAC1 binding and a gain in acetylation, which is subsequently read by BRD (BET) proteins. This work was published in Nature Genetics (Nagarajan et al).
