Lysine methylation is an important post-translational modification found on histones that is added and removed by histone lysine methyltransferases and demethylases, respectively. Lysine methylation occurs in a specific and well-regulated manner, and plays key roles in regulating important biological processes such as transcription, DNA damage and cell cycle. Regulation of the protein abundance of these methylation enzymes particularly by the ubiquitin-proteasome system has emerged as a key mechanism by which the histone methylation status of the cell can be regulated, allowing cells to respond rapidly to specific developmental and environmental cues. In my thesis, I focus on two histone lysine demethylases, KDM4A and PHF8, both of which appear to be regulated by E3 ligases; this regulation impacts their function in the cell. Chapter 2 shows that KDM4A is targeted for proteasomal degradation by the SCFFBXO22, and mis-regulation of KDM4A results in changes in global histone 3 lysine 9 and 36 (H3K9 and H3K36) methylation levels and impacts the transcription of a KDM4A target gene, ASCL2. Chapter 3 shows how PHF8 is targeted for proteasomal degradation by the APCCDC20 via a novel, previously unreported LxPKxLF motif on PHF8. I also found that similar to other APCCDC20 substrates like Cyclin B, PHF8 is an important G2-M regulator, loss of which results in cell cycle defects such as prolonged G2 and defective M phases. To further interrogate PHF8 biology, Chapter 4 describes the generation of a PHF8 conditional knockout mouse. PHF8 biology is interesting and relevant to human disease, as mutations are found in X-linked intellectual disability and autism. Complete loss of PHF8 by full body knockout in the mouse appears to be embryonically lethal, underscoring its key role in early development. This mouse model would allow us to extensively study the biochemistry and biology of PHF8 in the context of development and especially in brain function, where it is anticipated to play key roles. Overall, my dissertation work provides mechanistic and biological insights into how histone demethylases are dynamically regulated by the ubiquitin-proteasome system, providing an extra dimension to our understanding of how chromatin marks can be regulated.

Prostate cancer (PCa) is the most common cancer in American men. Although initial androgen deprivation therapy (ADT) confers a five year survival rate of 99%, the relapse of metastatic and drug resistant PCa (CRPC- Castration-Resistant PCa) continues to account for most deaths. How certain PCa cells develop into CRPC is the key question in the field. In addressing it, attention has focused on epigenetic factors that contribute to CRPC development. Herein we investigated the role and regulation of histone demethylase PHF8 during PCa neuroendocrine differentiation (NED) and progression into CRPC. We utilized bioinformatic analyses and biochemical approaches in PCa/CRPC cell line and mouse models to unravel the following results: First, we discovered that PHF8 post-transcriptionally clusters with cell cycle genes during NED and into CRPC via an AR/MYC/miR-22 regulatory axis. We showed that this axis is dysregulated in CRPC cells to allow enhanced cell proliferation and resistance to the clinical AR antagonist drug Xtandi® (enzalutamide). Second, we revealed that PHF8 is necessary for hypoxia induced NED by demethylating repressive H3K9me2 and H3K27me2, above maintaining active H3K4me3 on select NED genes. Importantly, we unveiled that PHF8 sustains HIF1α expression in CRPC cells via a regulatory role associated with full length AR. Third, we recapitulated the role of PHF8 in vivo by excising its floxed allele in the prostate of TRAMP mice -Transgenic Adenocarcinoma of the Mouse Prostate. We observed that KO of Phf8 lowered tumor burden in part by sustaining Ezh2 expression during NED transition into CRPC. In conclusion, our data implicates PHF8 in multiple oncogenic roles and regulations during PCa NED into CRPC. Our results lay a foundation for understanding the dynamics of histone modifying enzymes during PCa progression and hint at designing small molecule inhibitors against PHF8 as a novel CRPC therapeutic target.

Though antibodies as therapeutics are limited to extracellular targets, their repertoire of molecular interactions has particular relevance to the many intracellular cellular proteins for which small molecule screening has reached impasse. For such proteins there is little recourse to theory, since molecular recognition is, in practical terms, still not well understood. Here I apply antibody discovery to the lysine demthylases KDM4D and PHF8, two proteins difficult to inhibit selectively due to the similarity of their binding pockets to those of the larger family. With a selective, picomolar affinity antibody, dependent on residues distal to the KDM4D active site, I present what is likely the first example of allosteric inhibition of a KDM4 lysine demethylase, demonstrating that there is opportunity outside active sites oversubscribed with pan inhibitors. Antibody discovery for PHF8, however, was plagued by a familiar problem: antibodies that bound when their antigen was immobilised directly to a surface, but barely bound at all when it was free in solution. The common explanation is that the partial denaturation that accompanies immobilisation reveals epitopes unavailable in solution, but examining the problem in detail for the Plant Homeodomain of PHF8 revealed a connection to its rarely sampled conformations. The prominence these antibodies in the immune responses to PHF8, and to some extent KDM4D, motivates two hypotheses on their origin: either the states are very immunogenic or there is a connection between states of irreversible damage and those sampled reversibly, but rarely, by a protein in solution.