PRPF8 Defects Cause Missplicing In Myeloid Malignancies

Recent application of whole exome next generation sequencing has led to the discovery of a new class of somatic mutations affecting various components of the spliceosomal machinery in myeloid neoplasms. All affected spliceosomal genes, except for LUC7L2 (which might function in the U1snRNP-related pathway) encode U2snRNP associated proteins, but the molecular consequences of individual mutations remain unclear.

Based on 3 index cases where somatic PRPF8 mutations (M1307I, A687P and G1750E) were observed, we have screened a much larger cohort of patients with MDS and related conditions for the presence of acquired lesions affecting this gene. Somatic PRPF8 mutations and heterozygous deletions affecting the corresponding locus were found in 9/352 and 24/450 cases, respectively. Del17p13 was associated with haploinsufficient PRPF8 expression, but down-modulation of PRPF8 was also found in about 10% of additional cases with diploid 17p. PRPF8 mutations were mutually exclusive with other spliceosomal mutations, in particular with SF3B1. Serial studies of clonal architecture revealed that somatic PRPF8 mutation is an early event.

PRPF8 encodes the largest and evolutionarily most conserved spliceosomal protein. It is essential for pre-mRNA splicing and required in all tissues. Germline mutations in human PRPF8 are associated with an autosomal dominant form of retinitis pigmentosa type 13.

In total, around 1/2 of the PRPF8 mutant and del17p deletion cases were found in pAML and sAML. Accordingly, MDS with PRPF8 lesions conveyed poor prognosis as shown by KM statistics. Phenotypically, PRPF8 defects correlated with the presence of ring sideroblasts (83% of mutants and 65% of deletion cases), likely overseen in the context of a high blast count. Additionally, most of the cases also displayed pseudo Pelger-Huet anomaly.

In in vitro experiments, decreased expression levels of PRPF8 due to experimental knockdown by lentiviral shRNA were associated with increased cellular proliferation and clonogenicity, suggesting that PRPF8 has a tumor suppressor role in myeloid malignancies.

When we studied the consequences of PRPF8 defects in yeast models, mutations in PRPF8 abrogated a block in the second step of splicing introduced by an experimental intron mutation. RNA-sequencing of primary mutant and deletion cases as well as engineered shRNA knockdown cell lines showed that loss of PRPF8 function or mutation resulted in a global missplicing defect. Some of the most misspliced genes included those involved in mitochondrial function such as NDUFAF6, SFXN2, RPS24, and SLC25A19 and, perhaps most significantly, GATA1 which was mispliced in samples with either low or mutant PRPF8 as well as in the knockdown K562 cells. Similarly, analysis in patients with RS and defective PRPF8 showed common expression profiles of genes involved in mitochondrial electron transport chain complexes. These changes may correspond to ring sideroblasts associated with PRPF8 defects while missplicing of RNA helicases (frequently mutated in MDS, see other abstract from this group), may explain the mechanisms of lost tumor suppressor function.

In sum, we describe here another important somatic spliceosomal mutation associated with myeloid neoplastic transformation and unique phenotypic features.