Probing the Interfaces of Epigenetic Complexes: Efforts Towards Elucidating and Targeting Critical Protein:Protein and Protein:lncRNA Interactions of Lysine-Specific Demethylase 1 (KDM1A/LSD1)

The post translational modification (PTM) of histone proteins is a highly dynamic process that is utilized in the control of gene transcription. This epigenetic process involves enzymatic ‘writers’ and ‘erasers’ which place or remove chemical modifications to the unstructured tails of histone proteins which protrude out from the nucleosomal core. In a highly dynamic manner, each PTM is spatiotemporally regulated and combinations of PTMs at a gene promotor or enhancer region leads to transcriptional enhancement or repression. The gene targets as well as selectivity and specificity of epigenetic enzymes is regulated by the multimeric complexes each enzyme is co-opted. Each complex contains a unique set of coregulatory proteins with RNA and DNA binding domains and PTM ‘reader’ domains to direct the catalytic machinery to a specific subset of genes. The coregulatory proteins also affect the specificity and selectivity of the enzyme through mechanisms which are only beginning to be explored.

Our interest is in elucidating the role of coregulatory proteins and lncRNA with respect to lysine-specific demethylase 1 (LSD1/KDM1A). A flavin-dependent mono-and di-demethylase of H3K4me1/2 and H3K9me1/2, KDM1A has been implicated in many different multimeric enzymatic complexes which, in some cases such as the REST and NuRD complexes, function on opposing pathways. This disparity in the downstream outcome being coordinated by the same enzyme highlights the need to understand not only epigenetic enzymes, but to consider the complexes as a whole towards therapeutic targeting.

The specific aims of my thesis were to (a) interrogate the role of individual and multiple coregulatory partners in enzyme selectivity and specificity (b) establish tools to study the mechanisms of biochemical and biophysical of protein:protein and protein:lncRNA interactions and (c) elucidate key characteristics of protein:protein and protein:lncRNA interfaces towards targeted disruption. To this end, I have utilized cloning and mutagenesis methods to heterologously express and purify coregulatory partners of KDM1A in E. coli. I chose coregulatory partners found in a common catalytic core as well as several additional coregulatory proteins from a stable KDM1A-containing 5-mer complex. I have produced multiple constructs for four of these proteins to allow for multiple affinity purification routes as well as for future binding studies. I have further expressed each of these constructs and have made significant efforts towards the purification of each construct based on solubility.

I furthermore established HDX-MS and SELEX protocol in our lab as tools to allow us to explore the dynamics of these epigenetic interactions. I further demonstrated and confirmed that there is no hotspot along the binding interface between KDM1A and CoREST, but that CoREST stabilizes the apical end of the KDM1A tower domain via HDX-MS with the highest change in deuterium uptake, over 20%, long KDM1A TαA residues 440-451.

I also made significant efforts towards elucidating the interaction between KDM1A and HOTAIR. Firstly, I established an RNA radiolabeled EMSA assay for the lab which allowed us to test the binding of HOTAIR to KDM1A. With this assay, we saw that CoREST286-482, specifically the linker region (residues 293-380), must be bound to KDM1A for HOTAIR to bind and that the dissociation constant was unchanged at 1.710.38 µM and 1.29±0.34 µM, respectively. Further, I confirmed that the first 320 nt of domain 4 of HOTAIR (nt 1500-1820) contain the critical binding and that the dissociation constant was slightly higher at 2.97±0.96 µM.

I have also optimized SHAPE-MaP and crosslinking strategies to explore the binding interface between KDM1A:CoREST286-482 and lncRNA. I determined that there were 83 nt that displayed at least a 1.5-fold change in SHAPE reactivity of HOTAIR D4 due to the presence of KDM1A:CoREST286-482. I also utilized a free-energy based secondary structure model to establish a secondary structure for HOTAIR D4 based on my SHAPE-MaP data. I noted that 44% of the significant nt were confined to a stretch of RNA (nt 1538-1610, 1779-1844) that is predominantly dsRNA. Further usage of photochemical crosslinking strategies revealed a propensity for G:C paired nt to be crosslinked to KDM1A:CoREST286-482. A similar nt sequence around these paired nt suggests a binding motif. The implications of these results is discussed herein.