Functional classification of RUNX1 variants in familial platelet disorder with associated myeloid malignancies

Le résumé fourni par la source

Heterozygous, pathogenic germline variants of RUNX1 [ 1 ] are causative for familial platelet disorder with associated myeloid malignancies (RUNX1-FPD, FPDMM, FPD/AML; OMIM 601399; ORPHA 71290) [ 2 ]. RUNX1-FPD is characterized by incomplete penetrance and a broad spectrum of clinical phenotypes, even within affected families [ 3 , 4 ]. The clinical presentation includes thrombocytopenia, most frequently moderate, functional platelet defects, and a risk of ~44% [ 5 ] to develop a hematological malignancy, mainly myelodysplastic syndrome and acute myeloid leukemia, but rarely also T-cell acute leukemia [ 6 ]. Many RUNX1 variants are reported only in individual families [ 3 ], hence they have not been associated with RUNX1-FPD before and no functional data is available. In silico prediction tools (e.g., rare exome variant ensemble learner (REVEL) [ 7 ]) are frequently not convincing, especially for variants in the highly conserved and frequently affected Runt homology domain (RHD) (Supplementary Fig. 1 ). According to present classification rules [ 5 ], RUNX1 variants must be frequently classified as variants of uncertain significance (VUS). Lately, the ClinGen Myeloid Malignancy Variant Curation Expert Panel (MM-VCEP) has published adjusted ACMG/AMP guidelines for RUNX1 including recommendations how to integrate functional data in variant classification [ 5 ]. To functionally characterize nine previously reported RUNX1 variants and complement the recent MM-VCEP guidelines [ 5 ], we developed a set of functional assays addressing heterodimerization with CBFB, phosphorylation of RUNX1, and the ability of RUNX1 to activate transcription. For controls, 6 known pathogenic variants and wild-type RUNX1 were investigated in parallel (Fig. 1a, b ). We used a scoring system to ascertain the functional class of investigated variants and determine their clinical classification by applying the MM-VCEP guidelines integrating our functional data. For detailed information about variants investigated (including HGVS nomenclature regarding RUNX1b [NM_001001890.3] and its transfer to RUNX1c [NM_001754.5], genomic localization (GRCh38), ClinVar and dbSNP links, REVEL score, and available clinical information), materials, and methods please refer to Supplementary Data . Fig. 1: Functionally addressed processes, variants investigated, and representative results. a The ability of RUNX1 variants to form heterodimers with CBFB was analyzed using a flow-cytometry-based FRET assay. Phosphorylation of RUNX1 missense variants was quantified by means of western blotting. Transcriptional activation through RUNX1 variants was investigated using luciferase reporter assays. This figure was created using free, adapted images from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com . b Schematic overview of the investigated RUNX1 variant set and relevant RUNX1 domains based on Lam and Zhang [ 1 ]. The variant nomenclature refers to RUNX1 transcript variant 2 (NM_001001890.3), which encodes for isoform b of RUNX1 (453 amino acids). For variant denotations with respect to RUNX1 isoform c and additional information please refer to Supplementary Data . c Representative contour plots of HEK293T cells after cotransfection of RUNX1.YFP with CFP (upper panel) and CBFB.CFP (lower panel) indicating from left to right the gating strategy: (1) gating of YFP + CFP + -positive cells, (2) exclusion of false positive FRET + signals resulting from YFP excitation by the 405 nm laser, and (3) detection of FRET + signals after adjusting the gate following cotransfection of RUNX1.YFP with CFP that should be FRET – . Below the dot plot panels, the bar graph displays the amount of FRET + cells relative to wild-type RUNX1 (mean + standard deviation (SD); three biological replicates; one-way ANOVA; Dunnett’s post hoc test; * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001). d Representative western blot result. As indicated, the upper and lower RUNX1 band represents the phosphorylated and unphosphorylated form of RUNX1, respectively. The lower panel displays GAPDH which was analyzed as loading control. The bar graph shows the quantification of the amount of phosphorylated and Ser249-phosphorylated RUNX1 protein relative to wild-type RUNX1 (mean + SD; three biological and two technical replicates; one-way ANOVA; Dunnett’s post hoc test; * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001). e The left and right bar graph displays the firefly/renilla ratios relative to wild-type RUNX1 for the reporter constructs r ETV1 in HEK293T (mean + SD; three biological and three technical replicates) and r CSF1R in HEL (mean + SD; two biological and five technical replicates), respectively (one-way ANOVA; Dunnett’s post hoc test; * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001). Full size image The interaction of RUNX1 and its cofactor CBFB is essential for efficient and stable DNA binding of the resulting transcription factor complex. To determine the heterodimerization ability of RUNX1 variants, we performed a FRET assay using YFP and CFP fusion proteins in HEK293T cells (Fig. 1c , Supplementary Fig. 2 ). In comparison with the wild-type protein, two pathogenic RUNX1 variants (i.e., Leu40Alafs*80 and Arg139*) and the variant Gln158Pro failed to efficiently dimerize with CBFB. Variants Trp79Arg and Lys167Asn showed moderately reduced heterodimerization, whereas all other variants showed heterodimerization to at least 85% of wild-type activity. Phosphorylation influences the ability of RUNX1 to activate transcription and affects its stability [ 8 ]. We coexpressed RUNX1 or its variants with CBFB in HEK293T cells and quantified the proportion of phosphorylated RUNX1. Additionally, Ser249-phosphorylation of RUNX1 was investigated (Fig. 1d , Supplementary Fig. 3 A). Our data revealed reduced phosphorylation for all pathogenic missense controls (i.e., Lys83Glu, Arg139Gln, and Arg174Gln) and for six out of nine variants of interest (i.e., Trp79Arg, Leu117Gln, Arg118Gly, Gln158Pro, Lys167Arg, and Arg180Trp). Since RUNX1 CBFB complexes function as transcriptional activators, we used four independent luciferase reporters to examine the ability of RUNX1 variants to activate transcription in human nephrogenic HEK293T cells (i.e., r CSF1R , r ETV1 , r MYL9 , and r PDE4DIP ). We observed omitted transcriptional activation for all pathogenic RUNX1 variants and for three out of nine variants of interest (i.e., Trp79Arg, Gln158Pro, and Lys167Arg). The variants Gly60Cys and Arg205Gln mainly resembled the activity of wild-type RUNX1, whereas the remaining four variants (i.e., Leu29Ser, Leu117Gln, Arg118Gln, and Arg180Trp) showed varying levels of impaired transcriptional activation (Fig. 1e , Supplementary Fig. 4 ). To further evaluate the transcriptional activation ability of RUNX1 variants in a hematopoietic context, we additionally performed luciferase assays in the human erythroid leukemia cell line (HEL) using two independent reporter constructs (i.e., r ETV1 and r CSF1R , Fig. 1e , Supplementary Figs. 5 , 6 ). In line with the results for nephrogenic HEK293T cells, we detected substantially decreased transcriptional activation for all pathogenic controls and for three out of nine variants of interest (i.e., Trp79Arg, Gln158Pro, and Lys167Arg). After developing and applying individual functional assays comparable to previous investigations [ 9 ], we applied a scoring system to integrate results of individual assays for each RUNX1 variant (Fig. 2 ). Based on the recommendations of the MM-VCEP [ 5 ] and our functional data, we classified variants with ≥2 scores <20% as non-functional (PS3_strong), variants with ≥3 scores <60% or >140% as likely non-functional (PS3_moderate), variants with ≤1 scores <80% or >120% as functional (BS3_strong), variants with ≤2 scores <80% or >120% as likely functional (BS3_supporting), and variants with conflicting results not allowing final conclusions as uncertain . We identifie

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.