Modulation of proteostasis by the ubiquitin-conjugating enzyme UBE2N in immunoproteasome-positive AML

Altered protein homeostasis through proteasomal degradation of ubiquitinated proteins is a hallmark of many cancers. Ubiquitination, coordinated by E1, E2, and E3 enzymes, involves up to 40 E2-conjugating enzymes in humans to specify substrate and ubiquitin linkages. In a screen for E2 dependencies in AML, UBE2N emerged as the top candidate. To investigate UBE2N's role in AML, we characterized an enzymatically-defective mouse model of UBE2N, revealing UBE2N's requirement in AML without significant impact on normal hematopoiesis. Unlike other E2s, which mediate Lysine-48 (K48) polyubiquitination and degradation of proteins, UBE2N exclusively synthesizes K63-linked chains, stabilizing or altering protein function. Proteomic analyses and a whole genome CRISPR-activation screen in pharmacologically and genetically UBE2N-inhibited AML cells unveiled a network of UBE2N-regulated oncoproteins. UBE2N inhibition reduced their protein levels, leading to increased K48-linked ubiquitination and degradation through the immunoproteasome and revealing UBE2N activity is enriched in immunoproteasome-positive AML. Furthermore, an interactome screen identified TRIM21 as the E3 ligase partnering with activated UBE2N in AML to modulate UBE2N-dependent proteostasis. In conclusion, UBE2N maintains proteostasis in AML by stabilizing oncoproteins through K63-linked ubiquitination and prevention of K48 ubiquitin-mediated degradation through the immunoproteasome. Thus, inhibition of UBE2N catalytic function suppresses leukemic cells through selective degradation of critical oncoproteins in immunoproteasome-positive AML. Overall design: MLL-AF9 transduced Ube2nWT or Ube2nC87S cells were treated in either 100% ethanol control or 50nM 4OHT for 48 hours in same density in triplicates. The cells were collected, and RNA was extracted using ZYMO RESEARCH Quick-RNA Miniprep kit (#R1055). After the confirmation of the RNA quality using Agilent 2100 Bioanalyzer, the libraries were prepared with polyA selection using the Truseq RNA Library Prep Kit, and the libraries were sequenced at an average depth of 30M paired-end 100bp nucleotide reads. After the quality of reads was examined using FastQC (v0.11.7, https://www.bioinformatics.babraham.ac.uk/projects/fastqc), paired-end reads were aligned against mouse mm10 genome (iGenome,https://support.illumina.com/sequencing/sequencing_software/igenome.html) using HISAT2 (v2.0.5,http://daehwankimlab.github.io/hisat2). The raw gene counts were calculated using featureCounts (v1.5.2, http://subread.sourceforge.net/) and normalized using edgeR (v3.16.5, https://bioconductor.org/packages/release/bioc/html/edgeR.html) in iGEAK (https://pubmed.ncbi.nlm.nih.gov/30841853/). Differentially expressed genes were predicted using limma/voom (v3.30.6, https://bioconductor.org/packages/release/bioc/html/limma.html).