Activity and Centromere Targeting of a Maize Retrotransposon Integrase

Abstract

Though closely related in their structure and activities, retroviruses and long terminal repeat (LTR) retrotransposons vary significantly in their target site selection. The mechanisms and host co-factors targeted by the integrase protein (IN) that guides this targeting specificity, have been well characterized in many of the retroviruses and several of the LTR retrotransposons of yeast. The mechanisms guiding many of the retrotransposons have yet to be determined, however. In this dissertation I worked to determine whether centromeric histone H3 (cenH3) nucleosomes are the target guiding centromere-specific retrotransposon (CR) IN to the centromeres of their host’s genomes. The determination of the mechanism of CR centromere targeting has been elusive and could provide some answers to long-standing questions about centromere evolution. I leveraged Google Deepmindʻs AlphaFold2 and AlphaFold2-Multimer to predict structures of the protein and to model protein-protein interactions with close attention to the C-terminal domain (CTD) of the CR1, CR2, CR4, and CR5 sub-family members, the latter two having lost centromere specificity in more recent integration events. I found that the non-centromere targeting CR sub-family members contained surface exposed substitutions disrupting a hydrophobic pocket in the SH3 domain and an additional substitution in the CR motif disrupting an additional hydrophobic surface. The alpha-helical CR motif was predicted to act as an arginine anchor, a common feature of chromatin interacting proteins, binding to the acidic patch formed by the H2A-H2B dimer. Through these findings I developed a model that the CR IN interacts directly with the chromatin through the H2A-H2B acidic patch via the CR motif and that the SH3 domain may interact with an additional co-factor guiding centromere-specificity or that the presence of cenH3 on the nucleosome impacts the accessibility of the acidic patch.I next worked on the generation of bioactive CR IN from E. coli. I found that the CR IN is highly insoluble under a large range of conditions in E. coli and requires solubilization of inclusion bodies and refolding of the purified IN to produce bioactive protein. I employed a fractional factorial method by flash dilution to quickly identify refolding additives that allowed the production of soluble IN. With these refolding conditions I produced soluble IN as high as 1.0 mg/mL (17.56 µM) in high salt buffers. Refolded CR IN catalyzed both 3’-end processing and strand transfer as validated by Nanopore sequencing. To our knowledge this is the first reported case of an active plant retrotransposon in vitro. Finally, I studied whether cenH3 nucleosomes are targeted by CR IN. I performed a series of pulldown experiments and found that the CR IN interacted directly with the H2A-H2B dimer as was modeled in the protein-protein interaction models. In vitro nucleosome reconstitution followed by electromobility shift assays and demonstrated that CR IN CTD bound both H3 and cenH3 nucleosomes. Strand transfer assays with naked or nucleosome-bound (H3 and cenH3) target DNAs revealed that CR2 IN is only active on nucleosomal target DNAs while CR5 IN is highly active on naked DNA targets. Moreover, CR5 IN binds to naked DNA in a concentration dependent manner. Taken together, cenH3 nucleosomes alone do not appear to be the specific target of the CR IN although the nucleosome architecture does play a role in the interactions of CR IN with target DNA. It is likely that an additional centromeric co-factor is necessary for specific integration at the centromere.