Decoding cancer-relevant splicing networks in CD19 and MST1R

Abstract

Alternative splicing is a highly complex cellular mechanism that enhances the protein-coding capacity of higher eukaryotic genomes. Like any complex process, it is prone to errors. These can lead to health-related issues. Nowadays, erroneous splicing is even considered one of the hallmarks of cancer. In this work, we investigated two different cancer-related splicing events. In MST1R proto-oncogene, skipping of exon 11 results in a pathological splicing isoform that leads to progression and metastasis in cancer. In CD19, aberrant splicing of exon 2 has been associated with failure of the CAR-T cell therapy targeting CD19 (CART-19). In both cases, we used a high-throughput minigene reporter assay to assess splicing changes upon thousands of different point mutations in the corresponding minigene regions. To decipher the splicing-effective mutations, linear regression-based in silico modeling was applied. As a result, the complete cis-regulatory landscape was revealed. In the case of MST1R, we also found that heterogeneous nuclear ribonucleoprotein H (HNRNPH) is an important trans-regulator. Using individual-nucleotide resolution UV crosslinking and immunoprecipitation (iCLIP) as well as synergy analyses, we uncovered HNRNPH binding sites and showed that the protein regulates MST1R splicing in a switch-like cooperative manner. Interestingly, for CD19, we found ~200 mutations that alter CD19 splicing and thus potentially predispose patients to relapse. In addition, we identified almost 100 novel cryptic splicing isoforms that most likely encode non-functional CD19 proteins. This, in turn, could lower CD19 protein levels and affect the long-term success of CART-19. In our analysis of trans-regulatory proteins, we found some RNA-binding proteins that significantly affect splicing isoforms relevant to CART-19 resistance. We also analyzed a previously reported cryptic CD19 splicing isoform in more detail. The isoform lacks classical splice sites and instead has repetitive sequence at both alleged splice sites. Using a splicing reporter and direct long-read RNA sequencing, we were able to demonstrate that the putative isoform is in fact an artifact caused by reverse transcription. Our results highlight the need for further validation of RNA junctions that do not exhibit the characteristics of classical splice sites. Overall, this work not only demonstrates the importance of mutations that alter splicing in cancer, but also provides new insights into splicing regulatory networks and methodological challenges. In addition, we present potential prognostic markers that are important for assessing the risk of CART-19 resistance.