Multiomic Discovery of Neopeptides from Non-Canonical ORFs on SF3B1 Mutation-Driven lncRNAs in Leukemia

Most of the transcriptome is populated by non-coding genes such as long non-coding RNAs (lncRNAs) that are a highly abundant yet critically understudied component in human diseases. Conventionally believed to be non-coding, recent works have shown that lncRNAs can undergo translation from non-canonical open reading frames (ncORFs) to give rise to neopeptides, yielding a novel regulatory dynamic with broad implications for cancer diagnostics and treatments. In chronic lymphocytic leukemia (CLL), a mature B cell malignancy, the RNA splicing gene SF3B1 is recurrently mutated and is associated with a worse clinical prognosis. Mutant SF3B1 is known to drive widespread aberrant RNA splicing and promote the production of neopeptides. However, whether SF3B1 mutation can regulate ncORF translation by impacting lncRNA biogenesis remains elusive. Gaining a better understanding of SF3B1mutation-driven translatable lncRNAs and their functional implications could yield novel diagnostic markers and therapeutic strategies.

To this end, using the isogenic Nalm-6 SF3B1 WT and K700E cell lines as our model, we carried out an integrative multiomics analysis encompassing both short and long-reads sequencing (ISO-seq), ribo-seq, and mass spectrometry (LC/MS). Through ISO-seq, we first detected a total of 137,000 lncRNAs genome-wide, consisting of both protein coding and noncoding genes including known oncogenes, and with 61% of identified lncRNAs being novel isoforms. Subsequent quantification using short reads RNA-seq revealed that >500 lncRNAs were significantly upregulated in the SF3B1 K700E Nalm-6 cell line (FDR<0.1). Upregulation of 183 of these novel lncRNAs was also confirmed in two separate CLL SF3B1 mutant cell lines (HG-3 and MEC-1). Through splicing analysis, we also identified ~10,000 novel splicing junctions mainly encompassing intron retention and alternative 3’ splicing events, confirming that splicing defects induced by SF3B1 mutation contribute to aberrant lncRNA biogenesis.

We next investigated the possibility of SF3B1 mutation-driven ncORF translation. Through ribo-seq, nearly all the identified lncRNAs were found to contain ORFs longer than 20 amino acids with 8% showing overlapping ribosomal footprints (>10 reads covering over 50% of ORF length), suggesting active translation. Encouraged by this observation, we queried 2.4 million ORF sequences from 137,000 lncRNAs in our LC/MS data from Nalm-6 cells. To minimize false positives, we devised a novel pipeline where the ORF list was divided into smaller subsets and each of them was individually searched against the MS spectra. The ORF hits gathered from all the subsets were then filtered in a final pass to eliminate false positives. This resulted in 800 peptide hits that did not resemble any known proteins, indicating that they originated from ncORFs. Using AlphaFold, we were able to predict that 73% of our lncRNA-derived peptides folded into novel proteins harboring functional amino acid motifs, indicative of functional significance. Furthermore, 132 peptides could be detected in publicly accessible proteomic data from over 100 CLL patients. 3 of those peptides significantly correlated with disease aggressiveness in multivariate regression analyses (p<0.05, HR: 1.1-1.42), considering mutations in SF3B1, TP53, and ATM deletion. These results strongly imply that lncRNA derived peptides have a significant role in CLL biology.

To experimentally confirm our multiomic analysis, we selected 10 translatable lncRNAs that were upregulated in Nalm-6 K700E and at least in one CLL cell line with SF3B1 mutation. We next validated our hits using the RNA-targeting CRISPR CasRx platform to systematically identify lncRNAs essential to the survival of SF3B1 mutant cells. From this screening, we confirmed that the knockdown of 2 lncRNAs (EMC3-AS1 and WAKMAR2) can specifically inhibit the growth of SF3B1 mutant cells without affecting the wild type cells.This finding highlights the essential role of SF3B1 mutation driven-lncRNAs and reveals a potential synthetic lethality in targeting these lncRNAs.

Our study demonstrates the significance of translatable lncRNAs as a major outcome of SF3B1 mutation with potential disease implication. Future work will include more thorough investigation of the cellular pathways affected by these lncRNAs and their protein products, leading to possible therapeutics applications.