A multi-omics approach to reveal critical mechanisms of activator protein 1 (AP-1) in cancer

Abstract

Activator protein-1 (AP-1) is a transcriptional factor complex that mainly consists of Fos and Jun family members. Jun proteins can form both homodimers and heterodimers, whereas Fos family members only form heterodimers with the Jun family. Fra-1 is a member of the Fos family proteins and plays a vital role in breast cancer progression. It is overexpressed in triplenegative breast cancer (TNBC) and related to poor clinical outcome [1]. Fra-1 is a potential therapeutic target for TNBC. As Fra-1/AP-1 is a transcriptional factor, and its transcriptional activity is controlled by coregulators, I investigated its interactome, cistrome and transcriptome with multi-omics methods in this thesis, in order to provide a comprehensive understanding of its activity.

In Paper I, we reported the Fra-1 associated interactome, and identified 118 endogenous proteins that interacted with Fra-1. The most enriched one was DDX5. We investigated the cistrome and transcriptome data of both Fra-1 and DDX5 and found an extensive overlap, indicating a high degree of association between them. Further, we described that DDX5 promoted cell growth in TNBC cells, and enhanced Fra-1-related cell proliferation through enhancing the transcriptional activity of Fra-1/AP-1. Using immunohistochemistry on a breast cancer tissue microarray, we determined that DDX5 was overexpressed in basal-like breast cancer samples compared to non-basal-like tumors, similar to Fra-1. Also DDX5 was associated with poor clinical outcome of breast cancer patients. We suggested that DDX5 is a coactivator of Fra-1 and may be of interest for theraputical targeting.

Besides DDX5, we found that PARP1 was among the 118 endogenous proteins that interacted with chromatin-bound Fra-1 in TNBC cells. The PARP1 inhibitor olaparib is used in the clinic for BRCA-mutated TNBC breast cancer patients. In Paper II, we explored the interaction between Fra-1 and PARP1, and demonstrated that PARP1 downregulated Fra-1 expression, and reduced Fra-1/AP-1 transcriptional activity. We found that Fra-1 can mediate resistance towards olaparib treatment, and that silencing or inhibiting Fra-1 sensitized cells towards olaparib treatment. Additionally, we determined gene expressions modified by PARP1 and that a significant fraction of these were dependent on Fra-1 expression. Next, we described that also PARP1 was overexpressed in basal-like breast cancer compared to non-basal-like patients, and that high PARP1 expression indicated a poor clinical outcome in breast cancer patients. All in all, we suggested that Fra-1 inhibition may overcome olaparib resistance in TNBC patients.

Coregulators can modulate AP-1 transcriptional activity, and there is a well known but incompletely understood crosstalk between AP-1 and nuclear receptors, including estrogen receptors (ERs). A majority of breast tumors express ERα and estrogen drives the growth of these tumors. They are successfully treated with the ERα modulator tamoxifen. However, many develop resistance to this treatment, and the crosstalk between ERα and AP-1 has been reported to play a vital role in this resistance. On the contrary, estrogen signaling protects against colorectal cancer development. In the colon, ERβ is the predominant ER. We explored the transcriptional regulation, genome-wide binding activity of ERβ, and its impact on AP-1 transcription in presence and absence of tamoxifen, in Paper III. De novo motif analysis of ChIP-seq data generated with validated ERβ antibody in colon cancer cell lines with exogenous expression of ERβ revealed that ERE was the most enriched motif, followed by AP-1, indicating significant tethering by ERβ to AP-1 in colon cells. The TCF and KLF motifs were specific for colon and have not been described before in relation to ERβ cistromes (mostly performed in breast cancer cells). We demonstrated that ERβ bound and regulated tumor suppressors and oncogenes, exemplified by CST5 and LRP6, which indicated the molecular underpinnings of its anti-tumorigenic role in colon cancer.

As ER/AP-1 crosstalk plays an important role in the function of both ERα and ERβ. The two receptors appear to have reverse effects in cancer development, but also significant cell- and tissue-specific effects. To better understand the differencies and similarities between the homologues ERα and ERβ, and the respective role of AP-1 crosstalk, a model where the two receptors can be studied individually in the exact same estrogen responsive cell context is needed. No suitable cell line with endogenous expression of both receptors exist, and ERβ is rarely or never expressed in cell lines. Thus, in order to provide a suitable model, in Paper IV, we used CRISPR/Cas9 to establish a estrogen-responsive MCF7 cell model which had ERβ (only) expression. In presence of Tet, these cells express no ER, and in absence of Tet, they express ERβ (only). We based this model on previously generated MCF7 Tet-Off ERβ inducible cells. Cultured with Tet, these MCF7 Tet-Off Mock cells express only ERα, similar to the parental MCF7 cells, which was our control in this study. MCF7 cells with no ER did not grow. Inducing ERβ the cells grew at a low rate, but ERα (only)-expressing cells grew faster than ERβ (only)-expressing cells. Upon E2 stimulation, ERα and ERβ showed opposite responses. Estrogen increased proliferation of ERα (only) expressing cells while decreasing proliferation of ERβ (only) expressing cells. However, ERβ (only)-expressing cells migrated faster than ERα (only)-expressing cells, but E2 treatment reduced migration of ERβ (only) cells. The transcriptome data further indicated that ERα and ERβ to some extent both regulated a proportion of genes. But that ERβ also uniquely modulated the gene expression profiles, especially genes related to ‘negative regulation of cell proliferation’, whereas ERα mediated ‘positive regulation of cell proliferation’. Combined with cistrome data, we detailed this regulation, and exemplified how ERβ uniquely binds chromatin and regulates ANXA9 which is related to migration.

In conclusion, this thesis demonstrated the mechanism of Fra-1/AP-1 with multi-omics methods. We further characterized two proteins, DDX5 and PARP1, in TNBC, suggested that they played important roles in the Fra-1 signaling pathway, and their interaction provided novel strategies for TNBC therapy. Furthermore, we explored the interaction between ERβ and AP- 1, through analysis of genome-wide binding sites of ERβ in colon cancer cells, and elucidated the specific roles of ERα and ERβ in the same cellular context.