To date, several retrospective studies reported the proportion of patients with splicing variants in XLAS. Jais et al. (7) investigated genetic and clinical characteristics in 195 families with XLAS and revealed that 29 families (14.9%) had splice site variants. Bekheirnia et al. (8) also reported the proportion of each variant in 175 families with male XLAS patient and it revealed 13.7% of all families possessed splice site variants. Recently, our group investigated 269 Japanese families with XLAS and we found that splicing variants were detected in 18.2% of all families (Table 1) (10). The higher proportion of splicing variants in our study was thought to be the effect of active transcriptional analysis to detect deep intronic variants and exonic splicing variants other than canonical dinucleotides splice site (GU/AG) variants. Interestingly, among 71 male XLAS patients from 49 families with splicing variants in our study, only 35 patients (26 families) had variants in canonical dinucleotides splice site of COL4A5 gene. These findings highlight the importance of splicing variants in COL4A5 and demonstrate similarity to the other inherited diseases mentioned in previous section.

As described above there are strong genotype-phenotype correlations in males with XLAS; in particular, patients with missense or small in-frame variants (so-called non-truncating variants) have less severe phenotypes compared to patients with truncating variants (e.g., nonsense, small insertion, or deletion leading to out-of-frame sequences) (7–9). In addition, patients with splice site variants have shown intermediate severity, between the phenotypes associated with non-truncating and truncating variants.

Although splicing variants can be classified depending on their transcript pattern (i.e., in-frame or out-of-frame transcript), genotype-phenotype correlation analysis based on transcriptional analysis had not been conducted. Recently, we focused on this transcriptional difference in splicing variants and analyzed the kidney survival of patients with splicing variants. We found a significant difference between patients with in-frame splicing variants and those with out-of-frame splicing variants; the median kidney survival of patients with the in-frame splicing variants (n = 33) was 28 years, whereas it was 23 years for patients with the out-of-frame splicing variants (n = 32; P < 0.05) (10). This result demonstrates the value of transcriptional analysis for splicing variants in XLAS to estimate their kidney prognosis and enable genetic counseling.

Some exonic variants, which can be considered as missense or nonsense if transcriptional analysis is not conducted, affect splicing and may cause disease. In particular, single-base substitutions at the last nucleotide position in each exon are reported to likely affect splicing patterns (19, 20). However, no studies have addressed the characteristic of exonic variants in the COL4A5 gene which are likely to affect splicing. Therefore, we focused on the variants affecting the last nucleotide of exons in COL4A5 and conducted a comprehensive in vitro transcript analysis (22). We found 14 reported variants located in last nucleotide of any COL4A5 exon from the Human Gene Mutation Database (HGMD) and six novel variants from our cohort. All 14 variants in HGMD are reported as missense and most of them are glycine substitution, which is most common type of missense variant in COL4A5 (3). Furthermore, using an in vitro functional splicing analysis, 17 out of the total 20 variants showed aberrant splicing. In this study, we also conducted the genotype-phenotype correlation analysis of the splicing variants caused by substitution of last nucleotide in exons comparing to our previous report of missense variants. We found that the median age of developing end-stage kidney disease (ESKD) in cases with splicing variants was significantly worse than those with missense variants (27 vs. 40 years old, P < 0.01). From this result, we concluded that variants located in last nucleotide position of exons in COL4A5, even if they are glycine missense substitution, should be considered likely splicing variants and examined by transcriptional analysis.

In addition, the importance of synonymous or silent variants in abnormal splicing in COL4A5 should be considered. We analyzed COL4A5 transcripts in three patients clinically diagnosed XLAS with synonymous variants by using the in vitro functional splicing assay and analysis of patient mRNA. This revealed that all three cases showed aberrant splicing patterns (23, 24). Interestingly, among these three cases, one patient had both aberrant and normal transcripts by mRNA analysis, and they exhibited a milder phenotype. This finding suggested that synonymous variants in COL4A5 can affect splicing pattern and might show milder phenotype via producing mixture of both normal transcripts and aberrant splicing.

Although the canonical dinucleotides splice site (GU/AG) is important for correct splicing and variants in this region cause aberrant splicing, intronic variants outside this region may also influence splicing. Indeed, there are several non-canonical intronic variants around exon-intron boundaries that have been reported in genomic databases such as HGMD (14). However, the pathogenicity of those intronic variants were not all proven by transcript analysis. Recently, our group reported the results of transcript analysis for seven non-canonical intronic variants in COL4A5 (6 reported variants and one from our cohort) by using in vitro splicing analysis with or without in vivo RNA sequencing. Consequently, five variants were expected to cause aberrant splicing (by skipping the respective exon) while one variant was found less likely to alter the splicing pattern (15). From the above, we should carefully judge the pathogenicity of non-canonical intronic variants and transcript analysis is recommended to assess their influence on splicing.

It has been known that deep intronic variants in COL4A5 also can cause aberrant splicing. Although this type of variant can be confirmed by only mRNA analysis because vast majority of intronic substitutions are polymorphisms, several pathogenic variants in deep introns of COL4A5 have been reported (14, 25). While variants close to exon-intron boundary frequently cause exon skipping, deep intronic variants in COL4A5 show the pathogenicity by the creation of cryptic exon (26, 27).

Previously, Sanger sequencing was widely used for the genetic diagnosis of Alport syndrome. However, screening of all three Alport genes (COL4A3/COL4A4/COL4A5) by conventional Sanger sequencing is time-consuming because each gene contains ~50 exons with no hotspots. Therefore, targeted exome sequencing with Next Generation Sequencing (NGS) has become the first line screening method for the genetic analysis. However, it should be noted that targeted exome sequencing, which screen exons and exon–intron boundaries, cannot detect all types of variant.

For example, large deletion across exons and copy number variations (CNVs) could not be screened by direct sequence and multiplex ligation-dependent probe amplification (MLPA) is the only way to detect this type of variant. However, a paired analysis approach, comparing NGS data of patients and normal controls, recently enabled us to screen CNVs and we detected COL4A5 CNVs successfully with this method (28). Similarly, deep intronic variants causing aberrant splicing also could not be detected by standard sequencing for exons and exon-intron boundaries. Although whole genome sequencing (WGS) approach can screen deep intronic variants, it is difficult to detect pathogenic variants located in introns among the vast majority of non-pathogenic polymorphisms. Therefore, it is essential to carry out transcriptional analysis such as RNA sequencing to detect deep intronic variants.

Transcriptional analysis is important to assess the pathogenicity of intronic variants located outside canonical dinucleotides splice site (GU/AG). Variants located in canonical dinucleotide splice site can be diagnosed as pathogenic because of its critical effect on splicing (29), other intronic variants need to be assessed by transcriptional analysis to determine whether they cause aberrant splicing or not. In addition, it should be noted that there is a rare exception (<1%) of the type of canonical dinucleotide splice site (i.e., GC/AG etc.) and variants in this site do not always lead to exon skipping but sometimes partial deletion of exons or exonization of introns, which is important information to estimate renal prognosis for the evaluation of in-frame or out-of-frame deletions at the transcript levels (11, 14).

To detect pathogenic variants including splicing variants effectively, we conduct the genetic analysis for patients suspected as having XLAS in a stepwise manner, as shown in Figure 2. Briefly, targeted exome sequencing including all three Alport genes is performed first and then screening for copy number variations (CNVs) in the COL4A3/ COL4A4/COL4A5 genes is performed using paired analysis for patients in whom no pathogenic variants are detected by targeted exome sequencing. When paired analysis detects the possibility of CNVs in cases, MLPA is used to confirm CNVs. In addition, for patients with no obvious pathogenic variants, RNA sequencing using RT-PCR will be performed to detect aberrant splicing by intronic variants.

Although the frequency of patients with CNVs is lower than those with splicing variants, our screening for CNVs can be conducted using the data of NGS analysis therefore this step is placed earlier than RNA sequencing. In addition, as the screening for CNVs using NGS data may not sufficiently detect CNVs of small size (smaller than 1,000 bp), we add MLPA analysis for the patients who are strongly suspected of having Alport syndrome from their clinical findings even if any pathogenic variants in COL4A3/COL4A4/COL4A5 genes were not detected by NGS and RNA sequencing.

Peripheral leukocytes or hair roots have been traditionally used as a common source of mRNA for COL4A5 transcriptional analysis because of its accessibility. However, the expression level of COL4A5 mRNA in these samples is low and the nested PCR technique is required to amplify the targeted regions (14, 30). Although mRNA from patient kidney biopsies has abundant COL4A5 expression, the cDNA analysis for COL4A5 by using mRNA from kidney is not a common procedure since we now conduct a genetic analysis first approach for the diagnosis of Alport syndrome without performing a kidney biopsy. In contrast, patient urine samples have good accessibility and mRNA directly extracted from urine sediments is also abundant in COL4A5 expression and can be used for RT-PCR (10, 11, 14, 23).

Urine-derived cells have been shown to express podocyte markers, and podocytes are the main source of type IV collagen α3α4α5 in the glomerular basement membrane (31, 32). Sergio et al. reported that urine derived “podocyte-lineage” cells from a patient with Alport syndrome expressed all three Alport gene mRNA and could be used for RT PCR analysis (32). Our group also use this cultured urine derived cells as a main source of mRNA for RNA sequencing of Alport genes. Comparing to direct extraction of mRNA from urine sediment, mRNA from urine derived cultured cells is easy to handle because of its larger and stable amount of RNA.

In addition to transcript analysis using mRNA from patient derived samples, an in vitro functional splicing assay (minigene analysis) using an expression vector is also useful and can be applied for possible splicing variants in COL4A5. We routinely examine novel intronic variants or variants suspected of causing aberrant splicing by using a minigene assay, which is constructed to encode two cassette exons (A and B), an intervening sequence containing a multiple cloning site and a promoter region (Figure 3).

The analytical method of this assay is shown in Figure 3. Briefly, hybrid minigene constructs are created by inserting a test sequence fragment consisting of target gDNA region (exons) and the flanking introns into the multiple cloning site within an intervening intron between two exons (exon A and B) of the minigene construct. If a patient sample (or gDNA) is not available, the variant is introduced by site-directed mutagenesis using PCR. Then, hybrid minigenes are transfected into cultured human cells and total RNA was reverse- transcribed into cDNA and the PCR will be performed. Finally, PCR products are analyzed by means of electrophoresis and direct sequencing (Figure 3).

As described earlier, it is often hard to obtain samples for transcript extraction with high expression of target genes for inherited kidney diseases. However, the minigene assay has high flexibility because it does not require any mRNA or even gDNA samples from patients. Variants can be introduced to the minigene by using site directed mutagenesis allowing transcript analysis even if patient sample is not available (33, 34). We have analyzed several novel intronic variants in various inherited kidney diseases using this assay (35–40). As for intronic variants of COL4A5, several studies report using the minigene assay including our studies (11, 15, 41). In addition, most recently, our group applied this in vitro splicing assay for variants located in exons of COL4A5 to clarify the characteristic of exonic splicing variants (22, 24).