Metabolic Dysfunctions Driven By PRPF8 Explain Cases with Ring Sideroblasts without SF3B1 Mutations: Similarity/ Differences of Genotype/ Phenotype Association

Abstract

SF3B1 mutations (SF3B1MT) mostly identify a group of MDS with ringed sideroblasts (RS), isolated erythroid dysplasia and favorable prognosis. Besides SF3B1, we and others have demonstrated that other genes might be involved in RS formation either due to convergent metabolic pathways or erythroid driver genes. MDS subtypes characterized by multilineage dysplasia lacking SF3B1MT and presenting with RS are paradigmatic examples of such a possibility. In addition, RS in AML is often unrecognized by hematopathologists due to the focus on the presence of blasts as a predominant feature and/ or the reduction of erythroid precursors inherent to blast infiltration. However, when we previously studied defects in PRPF81, a central but unappreciated target of somatic lesions in MDS, we found a correlation with the RS phenotype. To follow up on this initial finding, we compiled a well-annotated cohort of patients seen at the Cleveland Clinic to assess phenotype/ genotype association of PRPF8MT and RS. NGS results including PRPF8 mutational status of 3,545 cases with myeloid neoplasia were reviewed. Calls were retained after assessment of pathogenicity by the ACMG and VarSome classifiers. Excluding cases with concomitant SF3B1MT, we found PRPF8MT (median VAF 38% [3-79]) in 78 (2%) patients. In total, 83% of the mutations were missense, whereas 15% were truncating. Homology sequence between yeast and humans denoted that hotspots mapped to highly conserved amino acids. D1598 and C1594, localizing in the U6-snRNA interacting domain (1442-1601), were mutated in 14% and 5% of the patients, respectively. T418S mapped in a central domain of several SFs (PROCN) and was found in 9% of the cases, whereas the F1099fs alteration was detected in 5% of our cohort. Prussian blue stain showed the presence of RS in 25% of PRPF8MT. Since PRPF8 maps to the proximal regions of 17p, we also noticed that three informative cases carried somatic PRPF8MT and 17p11.2 deletion or loss of chr.17 (1/2), all showing the presence of RS. We thus screened a more complete cohort of chr.17 abnormalities (88% in the context of a complex karyotype) to investigate their contribution to RS formation. While 66% had RS, the high blast count may have concealed their presence in cases lacking this feature because of their predominant AML or high risk MDS. Notably, we also found 8 additional patients in whom SNP-A showed cryptic losses such as del17p11.2 and 17p13.1 (n=7) or UPD (n=2) containing elevated RS. In this cohort, TP53 was the most commonly mutated gene (49%) followed by ASXL1 and DNMT3A (10% each), BCOR/L1 and TET2 (7.5%). In regard to other SFMT, no patient had SRSF2MT, three had U2AF1S34/Q157, two carried somatic DDX41c.139-5C>T/A225D (with -5/del5q) and one patient had a LUC7L2G48* without -7/del7p. While the coincidental presence of marrow RS in TP53MT lacking SF3B1MT has been previously demonstrated2, no study has investigated whether RS might be a result of shared metabolic feedback mechanisms common to SFMT. We thus analyzed RNA splicing and gene expression (GE)3 of patients with single PRPF8MT (n=34) and SF3B1MT (n=281) finding overlapping splicing features in the a-hemoglobin synthesis HBA2 gene. Upregulation of HBA2 has been described in SF3B1MT patients4,5 to correlate with impaired erythrocyte maturation and its accumulation in early erythrocytes. Interestingly enough, given our data on chr.17 abnormalities, we also found that cases with single TP53MT (n=52) displayed high covariance features for HBA2, a possible culprit in cases carrying RS but unmutated for SF3B1 and PRPF8. Indeed by applying gene-set overrepresentation analysis to PRPF8MT and TP53MT GE profiles, we clustered the top 5 categories (FDR< .001) in erythrocytes differentiation (KLF1, EPB42), homeostasis and development (ALAS2, GATA1, RHAG, SLC4A1) suggesting that these genes might inversely correlate with HBA2 expression and be responsible for the formation of RS. Our study demonstrates that PRPF8MT is another molecular lesion phenocopying SF3B1MT via a compensatory metabolic mechanism responsible for the failure of terminal erythroid differentiation. In analogy, given the proximal genomic location of PRPF8 and TP53 we do not exclude that this commonality might be involved in TP53MT with RS as well. These likely phenocopying mechanisms may enable therapeutic opportunities (luspatercept) in cases lacking SF3B1MT presenting with RS and transfusion-dependent anemia.