Introduction
Fabry disease (OMIM no. 301500) is an X-linked lysosomal storage disorder resulting from insufficient α-galactosidase A (GLA; GenBank accession no. 2717) activity caused by mutations in the
GLA gene at Xq22. An enzymatic defect in GLA protein results in the accumulation of globotriaosylceramide and related glycosphingolipids throughout the body [
1,
2]. Typically, hemizygous males with little (< 1% of normal) or no GLA activity suffer from characteristic abnormalities in multiple organs, namely neurological (acroparesthesia), cutaneous (angiokeratomas), renal (proteinuria, kidney failure), cardiovascular (cardiomyopathy, arrhythmia), and cerebrovascular (stroke) symptoms. This phenotype is known as the "classic type". In contrast, some patients have a somewhat preserved level of enzyme activity, and only display predominant cardiac or renal symptoms later in life; these phenotypes are sub-classified as the “cardiac type” [
3,
4] and “renal type” [
5], respectively.
To date, 1086 variants of the
GLA gene have been registered in the Human Gene Mutation Database (HGMD) Professional (released in 2021.2) (
https://portal.biobase-international.com/hgmd/pro/start.php). Among these, approximately 5% correspond to splicing changes that may cause aberrant splicing. Most intronic variants of the
GLA gene registered in HGMD are located at canonical splice sites (dinucleotides at each end of the intron, usually GT at the 5′-end and AG at the 3′-end). Generally, these variants within canonical splice sites cause aberrant splicing because they impact highly conserved sequences that determine exon–intron boundaries [
6]. Although several reported variants are located outside these canonical splice sites, their pathogenicity is not yet confirmed.
Recently, various bioinformatics tools have been developed to predict the possible pathogenic effects of missense and intronic variants [
6,
7]; however, these predictions are not always correct. Presently, transcriptional analysis of patient samples is the most appropriate method for identifying splicing aberrations. However, RNA from affected organs is not always available. Moreover, it is often difficult to analyze mRNA obtained from peripheral leukocytes owing to low content and fragility. As an alternative, minigene splicing analyses have been developed [
8‐
14]. In this study, we performed a functional minigene splicing assay on the reported
GLA non-canonical intronic variants to examine possible splicing defects.
Discussion
In this study, we demonstrated that all reported non-canonical splice site variants in GLA were associated with aberrant splicing. As the pathogenicity of non-canonical splice variants is quite complex to interpret, all candidate variants in this study were classified as having uncertain significance according to the ACMG guidelines. Furthermore, in silico tools often cannot predict the possibility of aberrant splicing in intronic variants that do not involve canonical splice sites. Therefore, validating splicing effects with an in vitro splicing assay is useful for confirming pathogenicity and determining associations with clinical phenotypes.
Patients with nonsense canonical splice sites, or frameshift variants, exhibit protein-truncating defects and generally show a classic phenotype [
22]. Variant nos. 1 and 2 generated transcripts that had partial deletions in adjacent exons, resulting in protein-truncating variants. Although the study reporting variant no. 1 did not describe a phenotype, it was known that the male patient harboring variant no. 2 showed a classical phenotype [
23]. Variant nos. 4–6 are located near exon 4, and the minigene assay revealed that they all exhibited complete exon 4 (92 bp) skipping. Patients harboring variant nos. 4 and 6 had classic phenotypes [
24,
25], whereas the patient with variant no. 5 exhibited a later-onset phenotype [
26]; however, this report did not specify sex, and thus it cannot be determined whether this was a heterozygous female or hemizygous male patient.
The minigene assay for variant no. 10 revealed whole exon 5 (162 bp) skipping, resulting in an in-frame deletion; furthermore, patients with variant no. 10 reportedly showed a classic phenotype [
27]. Riera et al. investigated
GLA sequence conservation patterns and revealed that amino acid residues 253–271, located near the active site, were highly conserved regions among multiple species [
28]. Notably, variant no. 10 caused the deletion of residues 214–267, which may explain the classic phenotype. Variant no. 11 generated a transcript with 2 bp of intron 4 included, resulting in a frameshift mutation, another reasonable explanation for the classic phenotype [
29]. The minigene assay for variant nos. 12 and 13 led to partial inclusion (36 bp) of intron 5, which initially appeared to be an in-frame insertion but was later found to create a stop codon in the fourth amino acid downstream of exon 5. Therefore, patients with these variants reportedly exhibited the classic phenotype [
30]. For variant no. 12, the minigene assay results differed from those of the RT-PCR using the patients’ mRNA, possibly because of the hybrid minigene construction. Recent research indicates that splicing regulatory elements are typically within 200–300 nucleotides upstream and/or downstream of the regulated exon [
31]. Therefore, considering intron 5 is a short intron, 140 nucleotides downstream of exon 5 could have been included when making the fragment in this study.
Variant nos. 3 and 7–9 were deep intronic variants that created cryptic exons, which were in the vicinity of the variants. These variants caused protein-truncating defects; the cryptic exon resulting from variant no. 3 consisted of 115 bp nucleotides, leading to frameshift insertion, whereas those resulting from variant nos. 7–9 contained premature-termination codons. The clinical phenotype, however, differed between the variants. Patients with variant nos. 3 and 7 exhibited the classic type [
32,
33], while variant no. 9 caused a milder cardiac phenotype [
34]. The direct RNA sequence of variant no. 9 revealed that the cryptic exon inclusion ratio (PSI) was 0.42, suggesting that approximately 60% of the transcripts were still normally spliced. This generation of normal and abnormal transcripts may, thus, reduce disease severity [
35]. In addition, the morality ratios of normal and aberrantly spliced transcripts for variant nos. 3 and 7–9 differed from those in HEK293T and HeLa cells. Moreover, different cell types have different αGal A activity cutoffs [
36]. Dai et al. [
37] reported another deep intronic variant (c.639+1326C>T) detected in a male patient with Fabry disease who had the renal phenotype with a mild αGal A deficiency (75% of normal control). The mRNA extracted from his peripheral blood lymphocytes was subjected to RT-PCR, which revealed normal transcripts and two types of abnormally spliced transcripts (both were truncating types). The aberrantly spliced transcripts encoded deficient αGal A with about 25% of wild-type αGal A activity. The high residual αGal A activity was considered to be caused by normal transcript expression (a quarter of the total transcripts) and activity level of αGal A encoded by the abnormally spliced transcripts.
Notably, variant no. 9, which causes the cardiac type, has a high prevalence, especially in Taiwan (1 in 2810 people) [
38]. Interestingly, Chiang et al. reported that the rates of variant carriers did not differ among healthy controls, people with type 2 diabetes, and people with cardiac disease in Taiwan [
39]. A minigene assay was conducted in seven types of human cell lines, which revealed that variant no. 9 influenced alternative splicing in a tissue-specific manner [
39]. Recent studies also found modulation of alternative splicing using antisense RNAs [
40] and amiloride [
41]. Thus, it is important to correctly identify mutations that cause aberrant splicing and verify the underlying mechanisms. Moreover, minigene analysis of single-nucleotide substitutions in introns with high allele frequencies, which were considered SNPs, showed normal splicing. This further confirms that minigene analysis correctly determines splicing abnormalities.
This study had several limitations. First, except for variant no. 9, the validated variants were not from our cohort but rather from reported cases. These are often less descriptive regarding clinical data, such as the Mainz Severity Score Index, which is widely used to indicate the clinical severity of Fabry disease [
42]. Second,
in vivo validation using patient samples was insufficient other than that for no. 9. Finally, as the hybrid plasmids used for the minigene splicing assay were constructed artificially, the assay did not always imitate the
in vivo splicing reaction. However, several previous studies performed transcriptional analyses via RT-PCR using patient mRNA, which mostly produced results consistent with ours [
32‐
34].
In conclusion, our study revealed and confirmed splicing defects in 13 variants of the GLA gene. This analysis of splicing aberrations established correlations with pathogenicity and may contribute to elucidating tissue-specific splicing mechanisms.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.