Determination of the pathogenicity of known COL4A5 intronic variants by in vitro splicing assay

X-linked Alport syndrome (XLAS) is a congenital renal disease caused by mutations in COL4A5. In XLAS cases suspected of being caused by aberrant splicing, transcript analysis needs to be conducted to determine splicing patterns and assess the pathogenicity. However, such analysis is not always available. We conducted a functional splicing assay using a hybrid minigene for seven COL4A5 intronic mutations: one was identified by us and six were found in the Human Gene Mutation Database. The minigene assay revealed exon skipping in four variants, exon skipping and a 10-bp insertion in one variant, and no change in one variant, which appeared not to be pathogenic. For one variant, our assay did not work. The results of all three cases for which transcript data were available were consistent with our assay results. Our findings may help to increase the accuracy of genetic test results and clarify the mechanisms causing aberrant splicing.


Results
The minigene assay revealed aberrant splicing in five of seven variants. We observed exon skipping in four variants (No. 1,No. 3,No. 5 and No. 7), both exon skipping and a 10-bp insertion in one variant (No. 2), and no change in one variant (No. 6) ( Table 1, Fig. 2 and Supplementary Fig. 2). Analysis of minigene construct No. 4 (intron 28-14T>A), containing introns 28-29, was not possible. The inserted sequence produced numerous nonspecific bands, making it difficult to assess the splicing pattern. Among three of the seven variants (No. 2, No. 5 and No. 7), our in vitro assay results were identical to the previously reported patient transcripts (Table 1). In these cases, mRNA samples were extracted from skin fibroblasts (No. 2), hair roots (No. 5) or peripheral blood leukocytes (No. 7). The results of in silico analysis are shown in Table 1. A marked decrease in the original acceptor site score was found for No. 2,No. 4 and No. 5. This suggests that the corresponding changes made it more difficult for the original acceptor site to function in exonic recognition. Novel splicing acceptor sites were created in No. 2,No. 4 and No. 7, suggesting the possibility of insertion or deletion. Indeed, No. 2 leads to a 10-bp insertion in addition to exon 9 skipping ( Supplementary Fig. 3). In three variants, reduced polypyrimidine tract scores were observed (No. 3,No. 5 and No. 6). The polypyrimidine tract is preceded by a branch point sequence and is important for downstream exonic recognition. A reduced polypyrimidine tract score may lead to impaired exonic recognition. However, minigene analysis showed no change in splicing pattern in No. 6, even though a reduced polypyrimidine tract score was observed. Of the six cases in which in vitro results were obtained, in four cases (No. 1, No. 2, No. 3 and No. 5) the results could be predicted to some extent by the in silico results, while in No. 6 and No. 7 prediction from the in silico results was difficult (Table 1). At present, this in silico analysis is insufficient to predict the precise splicing pattern. In all cases, no branch point alteration was observed.

Discussion
This is the first report comprehensively exploring COL4A5 splicing patterns in variants assumed to be pathogenic in intronic regions 10 to 40 bp upstream of exons. Although XLAS is a monogenic disease caused by COL4A5, pathogenic COL4A5 mutations are not detected in all clinically diagnosed XLAS patients. In such cases, intronic mutations or exonic synonymous mutations can cause aberrant splicing 4-6 . Moreover, 15-60% of pathogenic mutations cause genetic disease through pre-mRNA splicing abnormalities 22 . Therefore, a renewed focus should be placed on intronic variants that can cause splicing abnormalities.
Recently, splicing has been focused on as a target of treatment. Modifying the splicing pattern is an important goal of some molecular therapies 23 . Oligonucleotide-based therapies such as Nusinersen or Eteplirsen have been approved by the FDA 24,25 . Clarification of the involvement of splicing in disease pathogenicity can lead to the development of treatments. For XLAS, we recently reported the phenotype-genotype correlation restricted to cases with COL4A5 splicing abnormalities. These results suggested the possibility of oligonucleotide therapy leading to exon skipping in patients with XLAS 5 . Taken together, these results show that determining the involvement of splicing in the pathogenicity of inherited kidney diseases is important and may inform treatment approaches.
When nucleotide changes are detected in a gene, it is important to determine whether or not they are pathogenic. This is often difficult, particularly when variants are found in introns. To assess the pathogenicity of intronic variants, it is important to conduct transcript analysis. Tools for in silico analysis, such as the Human Splicing Finder (http://www.umd.be/HSF3/), can be used to predict the effect of transcriptional variants and may aid in predicting the disruption of the original consensus splice sites. We previously reported that, among 41 families with COL4A5 variants inducing aberrant splicing, 19 were completely compatible with the Human Splicing Finder results. Of the remaining families, 17 yielded data that enabled prediction of splicing defects, a precise novel splice site could not be predicted, and the remaining five families yielded insufficient data for the prediction of splicing abnormalities 5 . In this study, No. 6 showed decreasing polypyrimidine tract scores, but in vitro results showed no difference between this patient and the control. Therefore, the limitation of in silico analysis lies in accurately predicting splicing patterns.
The most accurate method to predict pathogenicity is to conduct transcriptional analysis using the patients' affected organ tissues. However, obtaining kidney specimens to observe tissue-specific gene expression is always www.nature.com/scientificreports www.nature.com/scientificreports/ difficult. In addition to renal tissue, peripheral blood lymphocytes, skin biopsies, hair roots and urine-derived cells from patients have also been reported as adjuncts for the diagnosis of AS 5,6,14,18,20,[26][27][28][29] . It may be possible to determine the pathogenicity of a mutation by examining the mRNA extracted from each sample. However, this approach is suboptimal as transcripts are not stable within tissues, and abnormal transcripts can be quickly degraded by nonsense-mediated mRNA decay 30 . Instead, it has been reported that it is effective to examine the expression of type IV collagen α5 chain in the hair root, which can be collected noninvasively, but the sensitivity and specificity have not been fully established 14 . Therefore, there remains a serious need for another system to assess the pathogenicity of variants of uncertain significance.
Recently, a splicing assay with a hybrid minigene approach has been established as a relatively fast and accurate means to identify splicing aberrations and to study their underlying functional mechanisms [10][11][12][13][14][15]31 . This  www.nature.com/scientificreports www.nature.com/scientificreports/ approach has been validated by reports of the same results being obtained from in vivo and in vitro studies [10][11][12]32 . In our study, in three of the seven patient transcript analyses, the obtained results were completely consistent with our in vitro assay results. In one case, our minigene system did not work, possibly because of the low original acceptor site score. In addition, the results of No. 5 minigene showed normal and skipping bands in the WT, but only skipping bands in a patient (Fig. 2d). The possibility of identifying abnormal splicing bands in the WT has already been reported in other COL4A5 studies with minigene assay 5,15 , which can lead to false positive results of aberrant splicing. Therefore, the in vitro results should be evaluated in comparison with controls, bearing in mind variation in splicing patterns among different tissues and differences between in vitro and in vivo conditions.
In silico analysis assisted the interpretation of in vivo or in vitro analysis results, but was insufficient to predict the splicing pattern by itself. Three intronic variants analyzed by COL4A5 minigene assay have been reported so far 5,14,15 . Among them, the coexistence of exon skipping and a 43-bp deletion was detected in one case. It was possible to predict exon skipping by in silico analysis from reduced branch point scores in that case; however, the 43-bp deletion could not be predicted. Furthermore, although exon skipping occurred in the remaining two cases, their splice site scores decreased only slightly, supporting the difficulty of predicting exon skipping by in silico analysis itself (Supplementary Table 1). However, although it is difficult to predict the splicing pattern, clarifying the mechanisms of aberrant splicing by in silico analysis allows us to understand how to correct the splicing pattern to weaken or ameliorate the pathogenicity. Combining in vivo and/or in vitro and in silico analyses can thus be a powerful tool for assessing pathogenicity and for the development of appropriate therapeutic approaches.
In conclusion, our splicing assay with hybrid minigene makes it possible to assess whether the mutation in question causes aberrant splicing. In addition, in silico tools can predict the aberrant splicing mechanisms. Our system is useful to increase the accuracy of genetic tests through determining the pathogenicity of intronic mutations and may help inform treatment discovery strategies.

Methods
In vitro splicing assay. To create hybrid minigene constructs, we used the H492 vector that we developed previously, which is based on the pcDNA 3.0 mammalian expression vector and contains a multicloning site (Invitrogen, Carlsbad, CA, USA) ( Supplementary Fig. 1) 10 . We cloned DNA fragments containing a couple of exons and introns around the target variant in the COL4A5 gene using classical restriction and ligation methods or In-fusion cloning methods, as shown in Table 1. As for No. 1-3 and No. 7, cloning was performed by classical restriction and ligation methods using NheI and BamHI. As for No. 4-6, we used infusion cloning methods with the HD Cloning Kit (Takara Bio Inc., Tokyo, Japan), in accordance with the manufacturer's instructions. As for No. 7, the patients' gDNA was available, so gDNA of the patients and WT was cloned. As for No. 1-6, because the patients' gDNA was not available (just the sequence was reported), we initiated cloning with WT gDNA and then introduced mutations by site-directed mutagenesis using PrimeSTAR mutagenesis basal kit (Takara Bio Inc.), in accordance with the manufacturer's instructions. The primers used are shown in Supplementary Table 1.
The hybrid minigenes were confirmed by sequencing and transfected into HEK293T and HeLa cells using Lipofectamine ® 2000 (Thermo Fisher Scientific, Waltham, MA, CA). Total RNA was extracted from cells after 24 h using the Rneasy ® Plus Mini Kit (QIAGEN, GmbH, Hilden, Germany). Total RNA (1 µg) was reverse-transcribed using RNA to cDNA EcoDry Premix (Double Primed) (Takara Bio Inc). PCR was performed using a forward primer corresponding to a segment upstream of exon A (YH307: 5′-ATTACTCGCTCAGAAGCTGTGTTGC-3′) and a reverse primer complementary to a segment downstream of exon B (YH308: 5′-CTGCCAGTTGCTAAGTGAG AGACTT-3′). PCR products were analysed by electrophoresis on an 1.5% agarose gel using ϕX174-Hae III digest marker and direct sequencing.
compliance with ethical standards. All procedures were reviewed and approved by the Institutional Review Board of Kobe University School of Medicine.

Data Availability
The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.