Discovery of UBE2B and RNF144b As Potential Candidates for Selective Degradation of NPM1c in AML

Introduction: Nucleophosmin, NPM1, is an essential protein that shuttles between the nucleus and cytoplasm but primarily localize in the nucleolus¹. In 30-35% of AML patients, NPM1 is mutated in the C-terminus disrupting the nucleolar localization signal and creating a new nuclear export signal leading to a predominant cytoplasmic localization of the mutant protein (NPM1c)¹. A small fraction of NPM1c resides in the nucleus and is recruited to chromatin to drive aberrant gene transcription of HOX and other genes thus contributing to leukemogenesis in addition to the cytoplasmic function^2-3. In NPM1c AML, either CRISPR KO or dTag-mediated degradation of the NPM1c allele leads to cell growth arrest and apoptosis^2-4. Thus, selective targeting via degradation of NPM1c could have therapeutic potential in this disease segment. We utilized co-IP and applied induced proximity assays⁵ to test the ability of E3 ligases to selectively degrade NPM1c.

Methods: OCIAML3 cells were lysed with IP Lysis buffer and NPM1wt and NPM1c specific antibody were used for IP. After overnight incubation, protein A dynabeads were added to the reaction followed by 1x SDS buffer and 10 minutes heating at 95c. Induced Proximity Assay was based on Poirson et al⁵ with some modifications. 293T/17 cells were infected with lentivirus containing wildtype or mutant NPM1-eGFP-ABI1cs-IRES-mCherry and selected with blasticidin. The top 10% of eGFP-expressing cells were sorted and transduced with lentivirus containing PYL1-UBE2B or PYL1-SPOP and selected with puromycin. 20,000 cells were treated with ABA or DMSO for 24-48 hours. For rescue experiment, cells were pre-treated with 1 µM MG-132 for 1 hour prior to 24-hour ABA treatment.

Results: We first measured NPM1 protein levels +/- proteasome inhibitor MG132 in OCIAML3 cells. NPM1c accumulated more rapidly than NPM1wt with MG132 treatment supporting our hypothesis that NPM1c is more accessible to its natural E3 ligase(s) for proteasome degradation than NPM1wt. To test native interaction of NPM1 (wt or mutant) with WWP1⁶ and RNF144b⁷, two E3 ligases identified from the literature to interact with NPM1, we first measured their expression in OCI-AML3 cells. We observed that RNF144b is expressed at very low levels unlike WWP1 expression. We used co-IP and observed no binding between endogenous NPM1c and WWP1. As molecular glues can bridge an E3 ligase to a target protein and force its degradation, we tested the ability of RNF144b to selectively degrade NPM1c by exploiting the induced-proximity system to chemically enforce interactions between two proteins⁴. Addition of the plant hormone abscisic acid (ABA) brings together the ABI1cs- and PYL1cs-fused proteins. We also used UBE2B and SPOP as positive controls, which were shown to have broad ability to degrade proteins using this system. Strikingly, after 24-hour treatment with ABA, we observed degradation of only NPM1c but not NPM1wt by RNF144b and UBE2B while degradation of both forms was observed with induced proximity to SPOP. Prolonged ABA treatment resulted in deeper degradation of NPM1c but not NPM1wt protein levels by RNF144b and UBE2B.

To validate that degradation of NPM1c is due to proteasomal activity, we pre-treated the cells with MG-132 and observed that NPM1c degradation is rescued in all lines. Next, we tested if selective degradation of NPM1c is restricted to the cytosol due to easier accessibility for E2/E3 ligases. We used cellular fractionation of ABA-treated PYL1-UBE2B cells and observed selective degradation of NPM1c in both the cytoplasm and the nucleus. This suggests that either NPM1wt is not accessible to E2/E3 even in nucleus due to nucleolar localization, or that UBE2B is driving selective degradation of NPM1c due to some other mechanism that remains to be identified.

Conclusion: Our data demonstrates the feasibility of selectively degrading NPM1c over NPM1wt and identified RNF144b and UBE2B as potential E2/E3 ligase candidates for a molecular glue degrader screen. Our discovery opens the door for development of an NPM1c-selective degrader that can have significant clinical benefit in NPM1c-AML patients with a strong therapeutic index by not degrading NPM1wt.

References:

1. Falini B. et al. Leukemia. 2009.

2. Brunetti L. et al. Cancer Cell. 2018.

3. Uckelmann HJ, et al. Cancer Discov. 2023.

4. Wang XQD. et al. Cancer Discov. 2023.

5. Poirson J. et al. Nature. 2024.

6. Jing Y. et al. J Exp Clin Cancer Res. 2021.

7. Du W. et al. J. Biol. Chem. 2010.

Aryal:AstraZeneca: Current Employment. Frost:AstraZeneca: Current Employment. Huang:AstraZeneca: Current Employment. Seong:AstraZeneca: Current Employment. Drew:AstraZeneca: Current Employment, Current equity holder in publicly-traded company. Tavana:AstraZeneca: Current Employment, Current equity holder in publicly-traded company.