Discussion
Normal pre-mRNA splicing depends on accurate recognition of exons and introns, so many variations that break the correct definition of introns and exons will lead to defects in pre-mRNA splicing, causing or modifying the severity of many genetic diseases[4]. It is currently clear that these variations affect splicing not only by disrupting classic splice sites (DS or AS), but also by altering other important RNA elements (splicing enhancers or silencers binding sites) or by creating new splice sites[4,5,22,23].
Traditionally, exonic variants outside the conserved splice site dinucleotide have been classified as missense, nonsense or synonymous variants when gene analysis is performed only at the DNA level. Currently, it is widely believed that RNA analysis should be performed to assess the pathogenicity of sequence variations, especially splicing abnormalities caused by variants affecting the splicing regulatory sequences. In many cases, however, RNA samples from the affected individual are unavailable, and mRNAs are difficult to detect due to the activation of the nonsense-mediated mRNA decay pathway[24]. Fortunately, a functional splicing assay based on the minigene assay was confirmed as an effective, reliable and relatively simple tool to functionally assay potential splicing[25], which has been extensively validated in our previous studies[20,21,26].
Single-base substitutions in the SLC4A1 , ATP6V1B1 ,ATP6V0A4 , FOXI1 , WDR72 and ATP6V1C2 gene, many of which are predicted to lead to missense, nonsense or synonymous variants, are often identified in patients with primary dRTA. Herein, we assumed that the pathogenic effect of some of these variants is the alteration of pre-mRNA splicing. Using bioinformatics tools, a total of 15 single-base variants in SLC4A1 , ATP6V1B1 andATP6V0A4 gene were selected and tested with a minigene assay. These variants were predicted to modify splicing regulatory sequences (disruption of ESEs or creation of ESSs), generate new splice sites or induce a significant reduction of splice site strength.  Minigene plasmids ligated to either WT or mutant genomic sequences were constructed. The constructs were transfected into HEK293T cells, and the mRNA was analyzed by RT-PCR and DNA sequencing. Consequently, 8 of them were found to cause exon skipping.
Transcript analysis by quantitative PCR of four variants of c.1765C>T in SLC4A1 , c.481G>T and c.1102G>A in ATP6V1B1 , and c.322C>T inATP6V0A4 gene, located in the internal position of the exon, have shown that they affect pre-mRNA splicing by causing a significant imbalance in the proportion of ESEs/ESSs. Many splicing regulatory sequences, including ESEs and ESS, can promote or inhibit the recognition of surrounding splicing sites by recruiting different protein factors, thus coordinating to regulate the correct splicing of exons[27]. Therefore, an improper ESE/ ESS ratio will prevent correct preRNA splicing. Variant c.1765C>T (p.Arg589Cys) in SLC4A1gene was reported to cause dRTA in the form of autosomal dominant inheritance[28]. In the previous study, the Arg509 was considered as a mutation hotspot, which is located in the intracellular domain between the 6th and the 7th transmembrane regions of the AE1 protein and at the apex of the cytoplasm of the 6th transmembrane region, and two other forms of the variant in the same site (p.Arg589His, p.Arg589Ser) have been confirmed to be pathogenic[29,30]. Variants at p. Arg589 site may lead to failure to completely transported to the cell membrane of AE1 protein, or the defective loading of the advanced structure after transferring to the cell membrane, resulting in structural and functional abnormalities[28-30]. In this study, online software HSF was used to predict that c.1765C>T could cause a significant imbalance in the proportion of ESEs/ESSs. Further in vitro minigene analysis revealed that this variant produced part of transcripts lacking exon 14, which did not change the open reading frame. We hypothesized that it not only generated a transcript containing c.1765C>T but also produced a transcript lacking exon 14, resulting in abnormal morphogenesis of the encoded protein and leading to a phenotype of dRTA.
Nonsense variant, a single-nucleotide substitution, is generally considered to generate a premature stop codon. However, many studies have found that nonsense variants can occasionally alter exonic elements and affect the splicing pathway, which can be demonstrated by the loss of correlation between phenotype and genotype[30,31]. The result of Zhu’s research has elucidated the detailed molecular mechanisms of exon skipping induced by nonsense mutations in theDMDgene[30]. In this context, variant c.484G>T in ATP6V1B1 was predicted as nonsense mutation, p.Glu162*, and the minigene splicing assay for c.484G>T showed that it prevents exon 6 inclusion with a subsequent frameshift and premature termination at the 22th codon in exon 7 in the mRNA probably by affecting a functional ESE site and/or creating a functional ESS site. The bioinformatics analysis predicted the loss of five ESE and the gain of two ESS motifs with HSF. Moreover, missense variant c.1102G>A (p.Glu368Lys) in ATP6V1B1also generated a part of aberrant transcripts that lack exon 11, resulting in a subsequent frameshift and premature termination at the 25th codon in exon 13. We hypothesize that this is probably due to the disruption of seven functional ESE sites and the generation of a functional ESS site, which binds hnRNPA1. Similar to c.484G>T in ATP6V1B1, variant c.322C>T in ATP6V0A4 was initially defined as nonsense variant (p.Gln108*) leading to an unable mRNA due to the nonsense-mediated mRNA decay. Our results showed that it influenced related ESEs and ESSs motifs and disturbed the normal splicing in vitro causing exon 6 skipping. Subsequently, the ligation of exons 5 and 7 would result in a lack of 42 amino acids and the production of a truncated protein. As a consequence, this mutated subunit A4 of apical H+-ATP protein would destroy the first topological domain of cytoplasmic and may reduce or abolish the transport activity of H+-ATPase.
Minigene assay allows us to categorize three exonic variants associated with dRTA, c.368G>T and c.370C>T inATP6V1B1 , c.1571C>T and c.1572G>A inATP6V0A4 , as splicing mutation. These variants located at or near to classical splice sites (DS or AS), may affect normal splicing either by decreasing the recognition efficiency of the 5′ss DS or 3′ss AS or by creating new splice sites. Sequence variations that occur in the conserved GT and AG dinucleotides flanking exons will cause splicing aberration, because the invariant GT and AG are required for the spliceosome to recognize 5’s and 3’ss, respectively[1,2,25]. Variants c.368G>T and c.370C>T in ATP6V1B1 involving the first and +3 nucleotide of exon 5, respectively, were initially reported as missense mutations p.Gly123Val and p.Arg124Trp, respectively.  Analysis of variants c.368G>T and c.370C>T in BDGP showed a reduction in the score of the 3’ splice site. The minigene analysis of these two mutations revealed that they certainly caused the skipping of the exon 5.
Variants c.1571C>T and c.1572G>A inATP6V0A4 , located at the -2 and the last nucleotide of exon 15, were identified as missense (p.Pro524Leu) and synonymous variant (p.Pro524Pro), respectively. Such substitution usually reduces the recognition of classic splicing sites. Our BDGP analysis further showed that c.1571C>T resulted in the reduction of recognition strength of donor splicing site, while c.1572G>A did not. But HSF predicted that c.1572G>A destroyed the WT donor sites. Consequently, the minigene analysis revealed that these two variants certainly altered normal splicing by increasing approximately 3.74% and 84.48% exon 15 exclusion compared with WT, respectively. The skipping of exon 15 results in a loss of 94 nucleotides with a subsequent frameshift and premature termination at the 13th in exon 16. Significantly, synonymous substitutions are generally considered to be benign from a protein perspective. But from the mRNA’s point of view, these substitutions may lead to abnormal splicing. Just as synonymous variant c.1572G>A of theATP6V0A4 in this study produced the aberrant transcript, the five synonymous variants were reclassified as splicing mutations using a minigene assay by Yohann Jourdy, which highlight that it is necessary to investigate the splicing impact of all synonymous substitutions to evaluate their clinical significance[2].
Among 8 variants, the PCR product of c.368G>T inATP6V1B1 was a unique transcript lacking exon 5, however, c.1765C>T in SLC4A1 , c.370C>T, c.481G>T and c.1102G>A in ATP6V1B1 , and c.322C>T, c.1571C>T and c.1572G>A in ATP6V0A4 gene, still produced one transcript with the size of the WT product except that one RT-PCR product was the transcript missing corresponding exon, so they probably have a double damaging effect: while the some part of the transcript is deleterious due to the skipping of the exon, which would cause loss of part of some important domains or premature termination of codons due to frameshift, the remaining mRNA is damaging due to the mutated amino acid change, which could result in non-functional protein, truncated polypeptide or nonsense mediated mRNA decay.
In addition, variants of c.1437C>G and c.1564G>A in SLC4A1 , c.481G>A inATP6V1B1 and c.52C>T, c.2190C>G inATP6V0A4 , were found to influence surrounding ESEs and ESSs motifs by the assessment of HSF or to cause the change of recognition strength of splicing site. However, results of our minigenes assay demonstrated that these variants did not affect pre-mRNA splicing, which was inconsistent with the predicted results of software, indicating that the results of the online software had a certain high sensitivity and low specificity. Also, we did not introduce these variants into cDNA and detect the cell surface expression and ion transport activity after AE1 or H+-ATPase mutations, which require further study.
In conclusion, we have performed an extensive analysis of exonic variants in SLC4A1ATP6V1B1 and ATP6V0A4associated with dRTA using bioinformatics tools and mini-genes. This study allowed to reclassify 8 previously presumed missense, nonsense or synonymous variants as splicing variations, which should be taken their pathogenicity into account. These variants either disrupt ESEs and produce ESSs, or interfere with the recognition of the splicing site of 3 ’ AS and a 5’ DS, resulting in exon skipping. This study emphasized the importance of assessing the effect of exon point mutations at the mRNA level in the dRTA, especially under the condition of the failure to obtain the patients’ RNA or kidney specimens, minigene assay may be a valuable tool.