The Experimentally Obtained Functional Impact Assessments of 5' Splice Site GT>GC Variants Differ Markedly from Those Predicted

Abstract

Introduction: 5' splice site GT>GC or +2T>C variants have been frequently reported to cause human genetic disease and are routinely scored as pathogenic splicing mutations. However, we have recently demonstrated that such variants in human disease genes may not invariably be pathogenic. Moreover, we found that no splicing prediction tools appear to be capable of reliably distinguishing those +2T>C variants that generate wild-type transcripts from those that do not. Methodology: Herein, we evaluated the performance of a novel deep learning-based tool, SpliceAI, in the context of three datasets of +2T>C variants, all of which had been characterized functionally in terms of their impact on pre-mRNA splicing. The first two datasets refer to our recently described “in vivo” dataset of 45 known disease-causing +2T>C variants and the “in vitro” dataset of 103 +2T>C substitutions subjected to full-length gene splicing assay. The third dataset comprised 12 BRCA1 +2T>C variants that were recently analyzed by saturation genome editing. Results: Comparison of the SpliceAI-predicted and experimentally obtained functional impact assessments of these variants (and smaller datasets of +2T>A and +2T>G variants) revealed that although SpliceAI performed rather better than other prediction tools, it was still far from perfect. A key issue was that the impact of those +2T>C (and +2T>A) variants that generated wild-type transcripts represents a quantitative change that can vary from barely detectable to an almost full expression of wild-type transcripts, with wild-type transcripts often co-existing with aberrantly spliced transcripts. Conclusion: Our findings highlight the challenges that we still face in attempting to accurately identify splice-altering variants.

5′ splice site GC>GT and GT>GC variants differ markedly in terms of their functionality and pathogenicity

“…In both cases (i.e., HESX1 c.357+2T>C and SPINK1 c.194+2T>C), the FLGSA‐derived data were in perfect agreement with the in vivo data. Fourth, in terms of the rate of GT>GC variants generating wild‐type transcripts, the FLGSA‐derived data agreed well with data obtained not only from disease‐causing variants but also from BRCA1 variants analyzed in their natural genomic sequence contexts (Chen et al, 2020; Findlay et al, 2018; Lin et al, 2019). Fifth, the FLGSA‐derived data correlated well with predictions made by SpliceAI, a recently developed artificial intelligence‐based splicing prediction tool (Chen et al, 2020; Jaganathan et al, 2019).…”

“…Fourth, in terms of the rate of GT>GC variants generating wild‐type transcripts, the FLGSA‐derived data agreed well with data obtained not only from disease‐causing variants but also from BRCA1 variants analyzed in their natural genomic sequence contexts (Chen et al, 2020; Findlay et al, 2018; Lin et al, 2019). Fifth, the FLGSA‐derived data correlated well with predictions made by SpliceAI, a recently developed artificial intelligence‐based splicing prediction tool (Chen et al, 2020; Jaganathan et al, 2019). Finally, the FLGSA assay preserves better the natural genomic sequence context of the studied variant as compared to the commonly used minigene assay (Wu et al, 2017; Zou et al, 2016).…”

“…In another recent study, the functional effects of over 4,000 BRCA1 variants were analyzed by means of saturation genome editing (Findlay et al, 2018). Of these variants, 12 were noted to be of the +2T>C type (Chen et al, 2020); 25% ( n = 3) of these +2T>C variants generated wild‐type transcripts, a proportion not inconsistent with our estimated 15–18% rate. By contrast, we know very little about 5′ splice site GC>GT (or +2C>T) variants.…”

“…A significant proportion of the genes with a genomic size of <8 kb were not expressed in the transfected HEK293T cells due to cell‐specific effects or for other reasons. Nonetheless, as we recently noted, the bottom line of the FLGSA assay is that (a) the genes were selected using a procedure that did not take into consideration the gene's physiological function and expression location; (b) the genes used for the final analysis were those that underwent normal splicing in the context of their specified reference messenger RNA sequences; and (c) the generation (or not) of the wild‐type transcripts from the variant alleles was observed under the same experimental conditions as for the wild‐type allele (Chen et al, 2020). Consequently, any differences in terms of functional impact on splicing between wild‐type and variant alleles could be directly attributable to the variant under study.…”

Data_Sheet_1.docx

“…As shown in Figure 3B, all five in silico algorithms provided by the Alamut suite under default conditions predicted that the variant allele would exhibit significantly reduced splicing potential as compared to the wild-type allele. We further employed the recently developed SpliceAI (Jaganathan et al, 2019) to predict the impact of the two splice-site variants on splicing and, as previously described (Chen et al, 2020), we focused our analysis exclusively on the Delta scores of donor loss although other scores may provide clues as to the nature of the resulting aberrantly spliced transcripts of splice-altering variants. Both variants were predicted to have a high probability of altering splicing, the Delta scores of donor loss being 0.91 and 0.99 for EXT1 c.1284 + 1G > C and EXT2 c.1173 + 2dupT, respectively (Delta scores range from 0 to 1, with 1 indicating the highest probability of altering splicing).…”