Dissection of pleiotropic effects of variants in and adjacent to F8 exon 19 and rescue of mRNA splicing and protein function

Abstract

The pathogenic significance of nucleotide variants commonly relies on nucleotide position within the gene, with exonic changes generally attributed to quantitative or qualitative alteration of protein biosynthesis, secretion, activity, or clearance. However, these changes may exert pleiotropic effects on both protein biology and mRNA splicing due to the overlapping of the amino acid and splicing codes, thus shaping the disease phenotypes. Here, we focused on hemophilia A, in which the definition of F8 variants' causative role and association to bleeding phenotypes is crucial for proper classification, genetic counseling, and management of affected individuals. We extensively characterized a large panel of hemophilia A-causing variants (n = 30) within F8 exon 19 by combining and comparing in silico and recombinant expression analyses. We identified exonic variants with pleiotropic effects and dissected the altered protein features of all missense changes. Importantly, results from multiple prediction algorithms provided qualitative results, while recombinant assays allowed us to correctly infer the likely phenotype severity for 90% of variants. Molecular characterization of pathogenic variants was also instrumental for the development of tailored correction approaches to rescue splicing affecting variants or missense changes impairing protein folding. A single engineered U1snRNA rescued mRNA splicing of nine different variants and the use of a chaperone-like drug resulted in improved factor VIII protein secretion for four missense variants. Overall, dissection of the molecular mechanisms of a large panel of HA variants allowed precise classification of HA-affected individuals and favored the development of personalized therapeutic approaches.

Keywords: F8; FVIII; RNA; bioinformatic; haemophilia; hemophilia; missense; pleiotropic effect; prediction; splicing.

Figure 1

F8 exon 19 contains dense splicing regulatory elements (A) Crystal structure (PDB: 2R7E) and schematic representation of FVIII. The regions of FVIII interacting with other coagulation factors are reported: factor X (FX), activated factor IX (FIXa), von Willebrand factor (vWF), and phospholipids (PL). The involved residues are also shown. Grey bars are not in scale. (B) Sequence of F8 exon 19, with exonic and intronic nucleotides indicated by capital and lowercase letters, respectively. With the sequence we report the predicted binding sites for splicing factors (predicted through SpliceAid tool), the binding sites of the modified U7snRNAs, and the investigated variants (reported as nucleotide and amino acid changes). All variants previously analyzed are indicated by asterisks. Exon 19 is shown in red. The splicing pattern resulting from co-expression of modified U7snRNAs (U7^a, U7^b) is also shown (inset). Scores of the splice sites, predicted through the Splice Site Prediction tool, are reported at the exon boundaries.

Figure 2

Characterization of the impact of exon 19 variants on splicing and protein (A) Exon 19 splicing assays. Minigenes were transfected in HEK293T cells and the splicing pattern evaluated by RT-PCR followed by agarose gel electrophoresis. The levels of exon inclusion, expressed as percentage of all transcripts, were estimated by densitometric analysis of bands (ImageJ). (B) Western blot analysis of pull-down experiments. Results are presented as mean ± standard deviation (SD) of three independent experiments and a representative blot is shown (NE, nuclear extract input; C-, naked beads). Histograms report fold changes over the wild-type. (C) Expression of rFVIII missense variants. rFVIII antigen and activity levels in media from transiently transfected HEK293T cells were measured by ELISA and chromogenic assays, respectively. Results are expressed as percentage of wild-type rFVIII. Missense variants are grouped as specified in the results section. Specific activity (values within boxes) was calculated as the ratio between activity and antigen levels. Results are presented as mean ± SD of three independent experiments. ns, not significant; ^∗p < 0.1; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001.

Figure 3

Evaluation of the prediction power of in vitro and in silico analyses In vitro expressed variants were considered as correctly classified if the inferred severity (class) corresponded to that reported in the EAHAD and CHAMP databases (see Table 1 and 2). Based on REVEL tool recommendation and the predicted “pathogenicity” scores, we arbitrarily set three classes (A–C) with increasing degree of stringency to infer the impact of protein changes and thus the associated severe, moderate, mild, or negligible phenotypes. See Table 1 for REVEL scores.

Figure 4

Targeted correction of exon 19 variants (A) Analysis of splicing patterns in HEK293T cells expressing minigenes variants alone or in combination with U1^B. Results have been obtained by analysis of peaks from denaturing capillary electrophoresis of fluorescently labeled RT-PCR products (Figure S3). The amount of each transcript is represented as percent of the total. Gain was calculated as the difference between the percent of correct transcript (exon 19 inclusion) in treated (+) and not treated (−) groups. Electropherograms of each detected transcript is provided (lower panel). (B) FVIII antigen and activity levels, measured respectively by ELISA and chromogenic assays, in media from transiently transfected HEK293T cells with and without the addition of NaPBA. Results are expressed as percentage of untreated wild-type rFVIII. Specific activity (values within boxes) was calculated as the activity/antigen ratio. Results are presented as mean ± SD of three independent experiments. The increase, expressed as fold-change compared to the untreated cells, is reported above. ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001.