RESULTS
A total of 366 missense variants, distributed in 26 exons of SLC12A3, were collected for bioinformatics predictions. They were analysed in silico with BDGP for splice site prediction, and with ESE/ESS estimation algorithms integrated in HSF. It is well-known that regulatory elements are commonly exist in exons with weak splice sites (Wu, Zhang, & Zhang, 2005). Therefore, we sought to select variants located in exons that have a weak 5′ or 3′ splice site. Except for variants located in the first and last exon that could not be analyzed with the minigene approach, four potential splicing variants were selected following the screening criteria (BDGP and HSF score below 0.7 and -8, respectively). We also selected six variants within two bases of 5’ or 3’ ends of the exons, which were predicted to have a big influence on classic splicing sites according to BDGP. As a result, a total of 10 candidate variants (c.602G>A, c.602G>T, c.1452C>G, c.1567G>A, c.1667C>T, c.1925G>A, c.2548G>C, c.2549G>C, c.2755A>T and c.2863A>T) located in seven exons of the SLC12A3gene were included in this study and collected from literatures (Balavoine et al., 2011; Colussi et al., 2007; Fujimura et al., 2019; Jiang et al., 2015; Miao et al., 2009; Riveira-Munoz et al., 2007; Vargas-Poussou et al., 2011). These candidate variants involved the generation of ESSs and/or the destruction of ESEs, the activation of a cryptic splice site, or the alteration of canonical splice sites (Table 1).
Based on this bioinformatics data, we followed up with minigene splicing assays in vitro . Taking the corresponding WT minigenes of each exon (pSPL3 Ex5, pSPL3 Ex12, pSPL3 Ex13, pSPL3 Ex15, pSPL3 Ex21, pSPL3 Ex22, pSPL3 Ex24) as the template, we successfully constructed all candidate variants minigenes through site-directed mutagenesis (Figure 1). Results of minigene analysis indicated that 60% (6/10) of them effectively disturbed the normal pre-mRNA splicing as expected (Figure 2 A and C), and these were further confirmed by sequencing analysis. Among ten candidates selected by BDGP and HSF, eight variants (c.602G>A, c.602G>T, c.1452C>G, c.1567G>A, c.1667C>T, c.1925G>A, c.2548G>C and c.2549G>C) led to exon skipping and two variants (c.2755A>T and c.2863A>T) caused no exon skipping. Although two variants (c.1452C>G and c.1567G>A) resulted in exon 12 skipping, it was regrettable that the WT minigene of exon 12 (pSPL3 Ex12) completely caused exon skipping, possibly due to experimental technique.
1. Presumed missense variantsc.602G>A (p.Gly201Asp) andc.602G>T (p.Gly201Val) induced skipping of exon 5compared with the WT plasmids.
Missense variants c.602G>A (p.Gly201Asp) and c.602G>T (p.Gly201Val) both affected the G at the first nucleotide in exon 5 of SLC12A3. Bioinformatics predictions from BDGP showed that the variants reduced the score of the WT 3′acceptor splice site from 0.83 to 0.57 and 0.35 respectively (Table 1). To observe the effects of the two variants on pre-mRNA splicing, the WT (pSPL3 Ex5) and mutants (c.602G>A and c.602G>T) minigenes were transfected individually into HEK 293T and Hela cells. The result of the minigene assays revealed that two different electrophoresis bands were detected in the WT and the mutant (c.602G>A) lane: a larger band of 403 bp and a smaller band of 263 bp. Meanwhile, the mutant lane of c.602G>T revealed one unique band of 263 bp in line with skipping of exon 5 (Figure 2 A and C). Quantitative analysis of cDNA products obtained from HEK 293T and Hela cells indicated that there was an increase of the exon 5-skipping transcript of c.602G>A corresponded to the WT plasmids (93.04% versus 9.90% in HEK 293T, P<0.0001; and 90.02% versus 7.56% in Hela cells, P<0.0001; Figure 3).
2. Variantc.1667C>T (p.Pro556Leu) gave rise toskipping of exon 13.
Variant c.1667C>T (p.Pro556Leu), located at the third-to-last nucleotide position of exon 13, was predicted by HSF to make a significant alteration of ESE / ESS motifs ratio (-9), broking five ESEs (CGCC TG, CGCC TGG (ESE_ASF), CGCC TGG (ESE_ASFB) , GCC TGG, CC TGGT) and creating four ESSs (GCT TGG, GCT TGGT, CT TGGT, T TGGTA) (Table 1, Table S3). In addition, the splicing effect of exon 13 is affected by a weak acceptor site, and the BDGP score of this site is 0.13 (Table 1). As a result of minigene splicing assays, the WT lane showed two different electrophoresis bands, representing a mature mRNA and a defected mRNA without exon 13 (36.26% in HEK 293T and 32.96% in Hela cells). Mutant (c.1667C>T) also generated two different transcripts, including the larger one with the variant and the shorter one with exon 13 skipping compared with the normal transcript (Figure 2 A and C). The analysis of the PCR products suggested that the rate of transcript without exon 13 to the full transcripts was 69.46% in HEK 293T and 61.21% in Hela cells, respectively (P<0.0001, Figure 3).
3.Variantc.1925G>A (p.Arg642His) led to exon 15 skipping.
Variant c.1925G>A (p.Arg642His) caused by the last nucleotide substitution in exon 15, was demonstrated that it reduced the score of the WT donor site from 0.99 to 0.68 with BDGP and also broken the WT donor site with HSF (Table 1, Table S3), most probably affecting splicing. Taken together, we used the WT (pSPL3 Ex15) and control mutant minigene to examine the experimental effect of variant c.1925G>A. The final PCR analysis result revealed that the WT lane respectively showed 2 different fragments of 363 bp and 263 bp, in contrast the mutant lane just showed one unique fragment of 263 bp corresponding to the lack of exon 15 of mRNA (Figure 2 A and C). Therefore, variant c.1925G>A (p.Arg642His) disturbed the 5’DS and led to exon 15 skipping.
4. Variant c.2548G>C (p.Gly850Arg) prevented incorporation of exon 21 into the mature mRNA.
Variant c.2548G>C (p.Gly850Arg) was identified at the last nucleotide of exon 21 and was predicted to decrease the score of the WT 5’DS from 0.85 to 0.22 with BDGP (Table 1). Simultaneously, HSF predicted that Variant c.2548G>C could lead to the deletion of exon 21 by inactivating the splicing donor site (Table S3). Through the WT (pSPL3 Ex21) and mutant (c.2548G>C) minigene analysis, the satisfactory result showed that a short band of 263 bp was detected in the mutant lane and a long band of 365 bp was detected in the wild-type lane, indicating that the mutant only produced the exon 21-excluded transcript compared with the original exon 21-included transcript of the WT (Figure 2 A and C).
5. Variantc.2549G>C (p.Gly850Ala)resulted in skipping of exon 22.
Variant c.2549G>C (p.Gly850Ala) affected the G at the first nucleotide in exon 22. Bioinformatic analysis of BDGP revealed that the score of the acceptor site is 0.52, however, it could not be analyzed after the nucleotide substitution (Table 1). The minigene assays result was identified that cDNA products generated by the mutant and WT (pSPL3 Ex22) minigenes were different. The lane of WT demonstrated one fragment of 375 bp that contains exon 22, whereas mutant c.2549G>C generated two different fragments of 263 bp and 375 bp, respectively (Figure 2 A and C). Because the WT minigene only generated the transcript with exon 22, it was obvious that the mutant (c.2549G>C) resulted in skipping of exon 22.
6.Variants of c.2755A>T (p.Arg919Trp) and c.2863A>T (p.Ile955Phe) in exon 24did not alter pre-mRNA splicing .
The pre-mRNA splicing of exon 24 is regulated by a weak donor site, and the BDGP score of this site is 0.35 (Table 1). Variants c.2755A>T, identified at the 8th nucleotide of exon 24, was predicted by HSF to make a significant alteration of ESE / ESS motifs ratio (-8) (Table 1 and Table S3). Variant c.2863A>T was located at the 21th-to-last nucleotide position of exon 24 and was predicted to alter splicing with a HSF score of -10 (Table 1 and Table S3). However, analysis of the minigenes demonstrated that both the WT (pSPL3 Ex24) and mutants (c.2755A>T and c.2863A>T) just generate the exon 24-included transcript (Figure 2 B and D), indicating that the two variants did not alter pre-mRNA splicing.
7. The WT minigene of exon 12 (pSPL3 Ex12) andmutants(c.1452C>G and c.1567G>A) only produced the exon 12-excluded transcript.
Variant c.1452C>G (p.Cys484Trp), caused by the 9th nucleotide substitution in exon 12, was predicted to disturb pre-mRNA splicing with a HSF score of -9 (Table 1 and Table S3). Variant c.1567G>A (p.Ala523Thr) affected the G at the last nucleotide of exon 12, and was predicted to decrease the score of the donor site from 1 to 0.45 with BDGP (Table 1). Regrettably, the result of the minigene assays showed that a band of 263 bp (the exon 12-excluded transcript) was both detected in the WT and the mutant lane (Figure 2 B and D). Therefore, although these two mutants (c.1452C>G and c.1567G>A) resulted in exon 12 skipping, there was no significant difference from the WT minigene (pSPL3 Ex12).