PHF5A-mediated exon recognition is selectively required for GSC viability and tumor formation.
Major finding: PHF5A-mediated exon recognition is selectively required for GSC viability and tumor formation.
Mechanism: PHF5A loss disrupts recognition of a subset of 3′ splice sites and induces G2–M arrest in GSCs.
Impact: Targeted inhibition of PHF5A/U2 snRNP activity impairs GBM initiation and maintenance in mice.
Efforts to develop targeted therapeutics against glioblastoma multiforme (GBM) have largely focused on analysis of tumor-initiating GBM stem cells (GSC). Recent studies using functional genetic screens have characterized proteins that are differentially required for the growth of GSCs compared with neural stem cells (NSC) and other untransformed cell types. Using a genome-wide RNA interference screen in patient-derived GSCs, Hubert and colleagues identified PHD finger protein 5A (PHF5A), which encodes a component of the U2 small nuclear ribonucleoprotein (snRNP) splicing complex, as a gene specifically required for the viability of GSCs but not normal NSCs. PHF5A depletion in GSCs induced RNA splicing defects, particularly exon skipping and intron inclusion events that introduced aberrant in-frame stop codons, in a subset of genes containing 3′ splice sites with cytosine-rich tracts, suggesting that PHF5A-mediated exon recognition is necessary for GSC growth. Consistent with this idea, this subset of genes was enriched for genes involved in cell-cycle progression, and PHF5A knockdown induced GSC cell death via G2–M cell-cycle arrest in premetaphase mitosis. Furthermore, PHF5A downregulation in GSCs inhibited GBM tumor formation and impaired tumor maintenance, resulting in enhanced survival of tumor-bearing mice. This requirement of PHF5A for GSC viability was dependent on its regulation of splicing activity as part of the U2 snRNP complex, as depletion of other spliceosome proteins or treatment with small-molecule U2 snRNP/splicing factor 3B inhibitors mimicked the effects of PHF5A loss. Intriguingly, expression of MYC was sufficient to enhance the sensitivity of NSCs to U2 snRNP inhibition, suggesting that oncogenic signaling increases the dependence of GSCs on PHF5A for proper exon recognition. These results establish PHF5A as a critical regulator of cancer-specific RNA splicing and suggest that suppression of PHF5A/U2 snRNP activity may be an effective therapeutic strategy in GBM.