CRYO-ELECTRON MICROSCOPY SINGLE PARTICLE STUDIES OF HUMAN CANCER TARGETS: UBIQUITIN-SPECIFIC PROTEASE 7 (USP7), USP28, AND KEAP1-CULLIN3-RBX1 E3 LIGASE MACHINERY

The following work describes the methodology and materials used to study three human protein complexes involved in the etiology and progression of cancer. The first, ubiquitin-specific protease 7 (USP7) is an isopeptidase that employs a unique auto-regulatory mechanism. The second is another ubiquitin-specific protease, USP28, which forms higher order states in solution. Lastly, the third case was a protein complex that utilizes an oxidation-sensitive dimeric protein, Keap1, and two components of an E3 ligase – Cul3-Rbx1. Each of these studies involved overcoming unique challenges for cryo-EM sample optimization. Not all yielded the quality of data that would result in high-resolution (< 6 Å) densities. Despite this, new information was discovered about each system.

USP7 has a unique mechanism of intramolecular regulation that stems from a hypothesized tethered-rheostat, whereby the c-terminal distal domains activate the catalytic domain via a hypothetical wide degree of conformational movement. My cryo-EM work, done in collaboration with the Wen Jiang lab, is the first comprehensive structural data that provides structural evidence for the movement of the tethered-rheostat. The particle set showed a great degree of conformational heterogeneity, even after a strategy was employed with a chemically-modified ubiquitin substrate to ameliorate these issues. The data showed that during the ubiquitin-bound state, after the release of a hypothetical substrate, but prior to the release of mono-ubiquitin, the HUBL4-5 domains do not remain engaged with the catalytic domain. This information suggests a change to existing models of catalysis.

Additionally, the structural model built from the cryo-EM density has revealed an interfacial region between domains that were previously not thought to interact. This interfacial region between the TRAF domain and HUBL1-3 represents a candidate location of binding for a mixed, non-competitive inhibitor of USP7 previously identified in the lab. Enzyme kinetics, DSF, and Glide molecular docking experiments all yielded data that corroborate this idea.

Structural studies on USP28 have been difficult as the multi-domain enzyme adopts oligomers in solution and is generally not amenable to crystallographic analysis. Prior to the work described herein, the only structural data were a solution NMR structure describing a few alpha-helical motifs in the N-terminus. During my graduate studies, two articles were published of the USP28 catalytic domain crystallographic structure. Both corroborated the existence of a dimer. The USP28 catalytic domain migrates during analytical gel filtration assays with the apparent molecular weight of a tetramer. Furthermore, glutaraldehyde crosslinking experiments show the catalytic domain appears to adopt a tetrameric state, like the USP25 tetramer. The USP25 tetramer was published alongside the USP28 catalytic domain dimer, concluding that a USP28 tetrameric state was not observed. Upon cryo-EM data collection and single particle analysis, it was observed that the compositional heterogeneity of the dataset was too great for any meaningful reconstruction. Although, the dataset appeared to how the presence of the E. coli GroEL chaperone complex. Co-expression experiments confirmed that the GroEL chaperone complex migrates with USP28 throughout the purification and may be useful for purifying USPs for structural studies.

Currently, our lab has a single-angle X-ray scattering (SAXS) model of the Keap1-Cul3 E3 ligase complex. But, the field does not fully agree on the molecular stoichiometry or the overall structure-function of this oxidation sensor – E3 ligase complex. It is hypothesized that Keap1 forms a dimer through its BTB domain, and a single Cul3 molecule then binds this dimer. The oxidation state of Keap1 cysteines appears to be critical to the interaction, but the field remains uncertain about which residues are responsible for the interaction with the Cul3-Rbx1 E3 ligase. To better understand this interaction and to obtain structural information to corroborate the SAXS model, recombinant Keap1 and Cul3-Rbx1 were purified and their interaction was tested by ITC, gel filtration assay, and a new technique called mass photometry.

It was found that the Keap1 Cys151 residue is not the oxidation sensor critical to the interaction, contrary to what some in the field anticipated. Additionally, it was found that under oxidative conditions, WTKeap1 could not form a complex with Cul3-Rbx1. The complex was successfully purified and was measured by SDS-PAGE, gel filtration assay, and mass photometry, and then used for cryo-EM single particle analysis. Full data collection and analysis has not yet been completed. It is anticipated that like the data from mass photometry, analytical SEC, and cryo-EM single particle analysis will show the complex appears to show a 1:1 Keap1-Cul3 stoichiometry, as opposed to the anticipated 2:1 ratio.