Visit the Carroll lab website
Nuclear receptor transcription
We are interested in defining the genomic and molecular features of oestrogen receptor (ER)-mediated transcription in breast cancer cells. We are specifically interested in understanding how these events and the machinery involved cause breast cancer cells to grow.
Oestrogen receptor is the defining feature of luminal breast cancers, where it functions as a transcription factor to induce cell cycle progression. ER is also the target of most endocrine therapies, including tamoxifen and aromatase inhibitors, which are effective treatments. However, some women can develop resistance to these drugs and in many cases, ER simply gets switched back on again, despite the presence of the drug. ER transcriptional activity requires a number of co-factors and co-operating transcription factors that possess enzymatic activity to alter chromatin structure, the outcome of which determines transcriptional activity. It is currently known that a number of ER co-factors can either assist in transcription (including SRC-1 and AIB-1) or are involved in gene repression by tamoxifen (including N-CoR and SMRT).
In addition, it is now known that ER requires proteins called pioneer factors to be able to maintain its association with DNA. Two of these proteins are FoxA1 and GATA3. FoxA1 is required for all ER-DNA interactions and in the absence of FoxA1, ER does not associate with DNA, switch genes on or cause cells to grow. Importantly, FoxA1 is also required for growth of cells that have acquired resistance to standard therapies, such as tamoxifen. Therefore, FoxA1 constitutes an attractive drug target for women with drug resistant breast cancer. Recent discoveries show that both GATA3 and FoxA1 are mutated in a significant fraction of women with breast cancer, but we do not currently know what the functional consequences of these changes are.
Characterisation of the role of GATA3 in ER biology
We are interested in identifying and characterising the role of the putative pioneer factor GATA3 in regulating ER activity. Using transcription factor mapping techniques (ChIP-seq), we have identified all the GATA3 binding events in a breast cancer cell line. This reveals high overlap with ER, supporting the hypothesis that GATA3 is intimately involved in ER function. For the first time, we have specifically removed GATA3 and assessed the impact on ER function in breast cancer cells. Unlike FoxA1, which is required for all ER-DNA interactions, we find that loss of GATA3 results in inhibition of some ER binding events, and also results in the reprogramming of novel ER binding events not normally seen in the presence of GATA3. These new ER-DNA interaction regions only observed in the absence of GATA3 correlate with changes in the genes that are regulated by ER, showing that the novel ER binding events are functionally relevant for breast cancer cells. Since GATA3 is mutated in more than 10% of all breast cancers, we believe that alterations in GATA3 sequence fidelity may be impacting ER binding capacity and the target genes that are regulated by ER. We are currently exploring what specific impact GATA3 mutations have on breast cancer cell growth and drug response.
Genomic analysis of ER function in primary breast cancer
All ER genomic studies to date have been limited to breast cancer cell line models, yet they have revealed extraordinary features about ER biology. We have now been able to extend genomic transcription factor mapping experiments into frozen primary breast cancer samples, by performing ER ChIP-sequencing in luminal breast cancer material. The data confirm that ER ChIP-seq can be performed in primary breast cancer samples and that the ER binding events accurately represent the binding sites in the cell lines. However, there are significant numbers of ER binding events that are acquired in tumours with a poor clinical outcome and in metastatic material that originated from an ER positive breast cancer. The novel ER binding events correlate with genes that have predictive value in independent breast cancer cohorts. We can model these events using drug sensitive or resistant cell line models, where ER binding events are dynamic and can be reprogrammed with growth factor stimulation. The reprogrammed ER binding events are mediated by changes in the pioneer factor FoxA1. We are currently exploring what enables changes in FoxA1, since these mediate the changes in ER binding events and subsequently influence the transcriptome. In addition, in close collaboration with the Caldas laboratory, we are embarking on a large scale transcription factor mapping experiment to identify ER and FoxA1 binding events by ChIP-seq, in ~200 primary breast cancers with detailed transcriptomic information and clinical follow up.
The role of androgen receptor in breast cancer
We previously showed that a subset of breast cancers, called molecular apocrine tumours, are driven by the male hormone receptor androgen receptor (AR) instead of ER. Unexpectedly, AR simply substitutes for ER and gets recruited to the same sites in the genome and subsequently regulates the same genes normally switched on by ER.
Figure 1: A collaboration between the Carroll and Neal laboratories has provided insight into how the male protein androgen receptor (AR) functions in breast cancer. Protein-DNA mapping experiments showed that in breast cancer cells, androgen receptor behaves less like it does in prostate cancer and instead it mimics the oestrogen receptor (ER). The data show a region of a chromosome and the binding sites of AR in the breast (purple) and prostate cancer (blue) contexts and ER in the breast cancer cells (red).
We had also made the observation that FoxA1 was mediating AR-DNA interactions, which paralleled the events normally seen between FoxA1 and ER. More recently we have explored the requirement for FoxA1 in AR-DNA binding capacity in molecular apocrine breast cancer cells. We find that specific loss of FoxA1 results in a reprogramming of AR to new regions in the genome. This is similar to what is observed in prostate cancer models and also what we observed between GATA3 and ER in breast cancer. Therefore, in molecular apocrine breast cancer, AR can mimic ER to regulate the genes normally controlled by ER, but its ability to move around the genome in the absence of FoxA1 is closer to what is observed in the prostate context. We are currently exploring the underlying mechanisms that govern AR reprogramming in the absence of FoxA1 and the potential impact that FoxA1 mutations may have on this process in breast cancer.