Cellular phenotype is largely defined by its transcriptional state, which is in turn governed by transcriptional variability of individual genes and the gene regulatory network. The overall phenotype of a tissue or an organ is best characterized by the composition of various cell types and their respective transcriptional states. In the first part of the presentation, we will discuss a few projects that explore cellular states in healthy and tumor tissues associated with clinical outcome. In the second part, we will discuss deep learning approach to functionally characterize non-coding mutations and polymorphisms and applications to understanding human brain evolution and varying cancer risks across populations.
Cellular phenotype is largely defined by its transcriptional state, which is in turn governed by transcriptional variability of individual genes and the gene regulatory network. The overall phenotype of a tissue or an organ is best characterized by the composition of various cell types and their respective transcriptional states. In the first part of the presentation, we will discuss a few projects that explore cellular states in healthy and tumor tissues associated with clinical outcome. In the second part, we will discuss deep learning approach to functionally characterize non-coding mutations and polymorphisms and applications to understanding human brain evolution and varying cancer risks across populations.
Up-Frameshift Suppressor 3 Homolog B (UPF3B) is a nucleocytoplasmic shuttling protein central to the classical nonsense-mediated mRNA decay (NMD) pathway. While UPF3B interacts with spliced mRNAs and promotes their cytoplasmic surveillance, its precise role in NMD remains debated. Some studies suggest partial dispensability, whereas others demonstrate that depletion of UPF3B alongside its paralog UPF3A severely impairs NMD. The clinical importance of UPF3B is underscored by patient-derived sequencing data linking its loss-of-function mutations to neurodevelopmental disorders. These findings highlight the need for mechanistic insights into UPF3B’s contribution to NMD and gene regulation.
To address this, we generated mammalian UPF3B knockout (KO) cell lines using CRISPR/Cas9. Following stable Cas9 induction and sgRNA-directed editing, two UPF3B-KO clones were established with distinct deletions (55-bp and 5-bp). NMD activity was assessed using canonical substrates, including the PTC39 β-globin reporter, alongside transcriptomic profiling via RNA-Seq (STAR, Salmon, DESeq2) and isoform switching analyses (IsoformSwitchAnalyzeR).
UPF3B-KO clones displayed robust NMD impairment, evidenced by upregulation of classical NMD substrates and activation of NMD autoregulatory feedback. Differential expression analysis revealed significant induction of known NMD targets such as GADD45G, GPX1, and SELENOP. Transcript-level analysis identified 165 NMD-regulated and 153 novel PTC+ transcripts specifically stabilized in UPF3B-KO cells. Furthermore, isoform switching analyses uncovered increased expression of NMD-sensitive splice isoforms, including SAT1 and HNRNPA2B1, implicating UPF3B in splicing-associated quality control.
In summary, complete loss of UPF3B is sufficient to compromise NMD, alter transcriptome-wide gene expression, and promote the stabilization of unproductive splice variants. Our findings underscore UPF3B’s pivotal role in coupling alternative splicing with NMD, preventing aberrant isoform accumulation, and maintaining mRNA homeostasis.
