Nuclear Export Signal Recognition by CRM1 Carrying the Oncogenic E571K Mutation and Structure-Based NES Prediction

Nuclear-cytoplasmic trafficking is an essential cellular process in eukaryotes that maintains and regulates the spatial distribution of key cellular processes in the nucleus and the cytoplasm. CRM1 is the major nuclear export mediator, which facilitates nuclear export of hundreds of protein and RNA molecules. CRM1 is essential to the survival of eukaryotic cells and is overexpressed or mutated in a variety of cancers. This involvement of CRM1 in cancer cell survival has led to the development of novel small molecule inhibitors such as XpovioTM or Selinexor, which has been approved for treatment of advanced and relapsed multiple myeloma and is also in many clinical trials for a variety of cancers. In addition, the point mutant CRM1(E571K) is found in a variety of tumors and is highly prevalent in some B-cell lymphomas. A deep understanding of how CRM1 recognizes the diverse NESs in hundreds to thousands of cargos is needed to understand the role of CRM1 in cancer pathogenesis and the mechanism of action of its inhibitors. My work has focused on expanding our foundational knowledge of CRM1-NES interactions in the presence and absence of the oncogenic E571K mutation. Previous studies measured the affinities of CRM1 interactions with 22 different nuclear export signals (NESs) from protein cargos using a low-throughput differential photobleaching assay. However, hundreds of very diverse CRM1 cargos and their NESs have yet to be characterized, limiting our understanding of the biological roles of CRM1 in disease. I developed a high-throughput, quantitative fluorescence polarization assay to measure the affinity of >100 NES peptides for WT CRM1 and for the oncogenic CRM1(E571K) to understand how the mutation affects NES binding. The large body of CRM1-NES affinity data was also used to develop a new structure-based NES prediction method. The CRM1 oncogenic mutation E571K is highly prevalent in some subtypes of B-cell lymphomas and can drive tumors in mouse B-cell models, but the mechanism of tumorigenesis is unclear. Using structural and biophysical approaches, I studied the recognition of 27 diverse NESs by CRM1(E571K) and showed that while most cargos are unaffected by the mutation, a small subset of highly charged NESs have greater than 10-fold affinity changes for the cancer mutant. To study whether these affinity changes cause nuclear export defects in cells, I used CRISPR/Cas to generate generated HEK 293 cells with either monoallelic CRM1WT/E571K or biallelic CRM1E571K/E571K. HEK 293 cells with CRM1WT/E571K or CRM1E571K/E571K had decreased proliferation and cells with homozygous CRM1E571K/E571K had obvious cell cycle defects. The eIF4E-Transporter which binds 10-fold weaker to CRM1(E571K) is mislocalized in HEK 293 cells with CRM1(E571K) and in chronic lymphocytic leukemia patient cells burdened with the CRM1 mutation. Additionally, I solved crystal structures of CRM1 bound to covalent KPT inhibitors, which showed that the mutation site, position 571 of CRM1, is located far from the bound drugs and thus unlikely to substantially affect their ability to inhibit CRM1. In order to understand the effects of the E571K on CRM1 activity and further illuminate possible cancer mechanisms, it is crucial to identify many more NESs across the entire human proteome. Therefore, NES prediction is of great interest, but existing sequence-based approaches give high false positive rates. My work also shows that may NES-like peptides are not accessible in the full-length cargos to to CRM1, in ways that many existing NES predictors cannot identify. To improve NES prediction, our collaborators in the Grishin Lab utilized CRM1-NES crystal structures as templates to develop a structure-based NES predictor and incorporated full-length protein context considerations to accurately identify novel NESs that are likely to be functional. I contributed to this work by expanding the CRM1-NES structural dataset and I helped to interpret modeling outliers to iteratively improve the CRM1-bound NES models. I also assisted the Grishin lab in using this structure-based predictor to map somatic mutations found in cancer patients that may disrupt nuclear export of the NES-containing proteins. Overall, I have added >100 new CRM1-NES binding affinity measurements and 16 CRM1 (WT or E571K) crystal structures to the field. These results have shown how the oncogenic E571K mutation binds very differently to small subset of highly charged cargos, and provided a robust dataset to develop a new structure-based NES prediction tool. Future work will build on this foundation to clearly illustrate how nuclear export defects lead to oncogenesis, and potentially give key insights into improving use of CRM1-targeted therapeutics.