Introduction

NP-C is an autosomal recessive disorder caused by pathogenic variants in either the NPC1 or NPC2 genes in ~95% and 5% cases respectively [1]. NPC1 encodes a transmembrane protein involved in transport of lipids, particularly cholesterol, from late endosomes or lysosomes [2]. The loss of its function leads to excessive accumulation storage of unesterified cholesterol, sphingomyelin, phospholipids and glycolipids in liver, spleen, and central nervous system. Subsequent neuropathological defects include Alzheimer’s-like neurofibrillary tangles, neuronal degeneration, neuroaxonal dystrophy, and demyelination with white matter tracts severely affected, especially in corpus callosum.

NP-C has a broad range of clinical symptoms and severity, from progressive fatal neonatal disorder to milder adulthood and later onset forms. Progressive neurodegeneration is a highlight of the disease and determines its severity. Neurological symptoms include motor development delay in early childhood; gaits, falls, clumsiness, cataplexy, and school problems in later childhood and adolescence; psychiatric illness and dementia in adult variants. Due to the predominating neuropsychiatric and cognitive symptoms, adult forms of NP-C are probably highly underestimated and patients are often misdiagnosed [3, 4]. According to recent publications, the prevalence of adult- and adolescent-onset forms of NP-C could be much higher than 1:100000 for the “classical” incidence with early severe phenotype [1, 5, 6].

In this article we present the detailed clinical description, biochemical data and functional analysis of the novel complex allele (CA) c.[2727 C > T;2793 C > T] in the NPC1 gene of two dizygotic twins with an adult-onset form of NP-C.

Materials and methods

All variants are called according to NPC1 canonical transcript NM_000271.5 and the Human Genome Variation Society nomenclature. Exons are numbered according to reference sequence NG_012795.1.

Description of methods, patients clinical and biochemical data are available in Supplementary file 1.

Results

P1 and P2 (55 y.o. twins) were suspected for NP-C disease based on a characteristic set of clinical symptoms presented in Supplementary Table 1. Elevation of the metabolites Lysosphingomyelin-509, cholestan-3β, 5α, 6β-triol and 7-ketocholesterol were detected and subsequent genetic testing of the NPC1 and NPC2 genes was performed.

The next-generation sequencing of NPC1 and NPC2 exons revealed two compound heterozygous variants in NPC1: complete loss-of-function variant c.2196dup (p.Ala732fs*30), described previously [7], and rare synonymous variant c.2727 C > T (p.Cys909 = ) with very low allele frequency in gnomAD (1.06e−5). The validation of these variants by Sanger sequencing also revealed the presence of c.2793 C > T (p.Asn931 = ) polymorphism in homozygous state (gnomAD allele frequency 4.94e−1), located in the NPC1 exon 18 donor SS. The DNA analysis of patient’s kindreds demonstrated that they are heterozygous carriers either of c.2727 C > T or c.2196dup variants, confirming their trans-position (Fig. S1).

The RT-PCR performed on RNA samples from blood and fibroblasts showed that patients have an additional lower-molecular-weight band together with 552 b.p. band of WT product (Fig. 1A). Sanger sequencing revealed that this band presents a 74 b.p. deletion r.2722_2795del (p.Val908Tyrfs*32) in exon 18, caused by activation of cryptic exonic SS (Fig. 1B). Also, the presence of r.2727 T variant in chromatogram indicates that the c.2727 C > T variant is “leaky” and gives significant amount of full-length transcript isoforms. In order to confirm the NP-C diagnosis, we performed the filipin staining test. After cholesterol depletion, skin fibroblast cultures from both patients were challenged either with a medium containing 10% of FBS and 10% of LDLP-enriched human serum. Filipin staining showed increased intracellular accumulation of non-esterified cholesterol in perinuclear vesicles of both patients with a smaller number and size of fluorescent perinuclear vesicles and a lower overall level of fluorescence compared to the positive control cells, treated with U18666A, consistent with a “variant” pattern according to Vanier and Latour classification [8] (Fig. 1C).

Fig. 1: Patients’ mRNA analysis and filipin staining.
figure 1

A Visualization of PCR products from amplification of patients NPC1 cDNA including exons 16–20. B The Sanger sequencing chromatogram demonstrating 74 b.p. deletion r.2722_2795del (p.Val908Tyrfs*32) caused by cryptic SS activation. The presence of r.2727T variant in chromatogram indicates that the c.2727 C > T variant is “leaky” and gives significant amount of full-length transcript isoforms. C Filipin staining of patients’ fibroblasts demonstrate the “variant” pattern.

To validate the effect of c.2727 C > T variant on splicing, and also to estimate the proportion of transcript isoforms in the absence of nonsense-mediated mRNA decay and the second allele we decided to perform minigene assay. Furthermore, since the c.2793 C > T (p.Asn931 = ) polymorphism is located in the WT donor SS of the same exon 18, we hypothesized whether this c.2793 C > T variant could affect the balance between WT and mutant SS activity or even decrease the strength of WT site below a certain threshold required for the cryptic site activation.

The bioinformatic analysis with various tools demonstrated no or low impact of both variants on splicing (Fig. 2A). MaxEntScan algorithm integrated in HSF wasn’t able to predict the effect of c.2793 C > T due to the very low score of the native donor splice site, although the same algorithm stand-alone predicts a dramatic reduction of its strength −0.68 > −3.03.

Fig. 2: Functional analysis of the complex allele.
figure 2

A The results of bioinformatic analysis (HSF—Human Splicing Finder 3.1). B The structure of the minigene and location of variants. C The visualization of PCR products from the minigene assay. D The box plot representing the relative expression of wild type transcript isoform (WT) and the one harboring 74 b.p. deletion (DEL) in minigenes with various combinations of studied variants.

To assess the impact of these variants on normal splicing alone and their relationship with each other, we built 4 minigene constructs bearing various combinations of these variants.

17, 18, 19 exons of NPC1 gene were amplified from the patient’s DNA and cloned into pSpl3-Flu2 vector. The RT-PCR analysis of cells transfected with these minigenes demonstrated the same splicing pattern as was observed from amplification of the patients’ cDNA.

At the next step, using site directed mutagenesis, we built two additional minigenes, one of which contains c.2727 C > T but lacks c.2793 C > T and other one that lack both c.2727 C > T and c.2793 C > T. The visualization of splicing products showed that only the minigene containing both c.2727 C > T and c.2793 C > T gives a significant amount of aberrant transcripts with 74 b.p. deletion (Fig. 2C).

To make quantitative assessment, the real-time PCR technique was applied. Minigene with both variants in cis showed 58% reduction of WT isoform and the 4.33 fold more mutant isoform, compared to c.2727 C > T alone (Fig. 2D).

Discussion

The existence of CAs partially explains the phenotypic diversity for a number of genes [9,10,11,12]. In most described cases the functional effect of these variants is realized on the protein level, thus making it difficult to clarify their interactions and role in pathogenesis. Furthermore, the detection of CA itself is hampered by the limitations of common genetic screening methods e.g. filtering of frequent variants in the process of analyzing NGS data.

The CAs which alters splicing are extremely underrepresented in literature, although their effect could be predicted bioinformatically and their functional analysis is much easier compared to those which change the protein function. Furthermore, CAs affecting splicing should be located relatively close to each other on DNA sequence, making it easier for a researcher to suspect its interaction. Deep intronic variants in cis could strengthen the cryptic SSs or create the canonical dinucleotides, leading, for example, to pseudoexon inclusion [13]. Frequent or rare variants in the native SSs could change its strength and modify the effect of those variants, which activate the cryptic SSs or alters the motifs of splicing enhancers and promotes exon skipping.

Our results provide strong evidence that the frequent polymorphic variant c.2793 C > T, located in the donor SS of NPC1 exon 18 significantly affects the activity of cryptic SS created by c.2727 C > T variant. This data is supported by the bioinformatic analysis, which demonstrated low or no impact of this variant on splicing alone. The significant amount of normally spliced mRNA, observed in the minigene assay, in the presence of typical loss-of-function variant on the second allele correlates well with patient’s adult onset phenotype. Also, the microscopy examination of filipin staining test demonstrated the “variant” pattern, which is common for adult onset form of NP-C.

Overall, we identified the novel rare CA c.[2727 C > T;2793 C > T] in NPC1, characterized its deleterious effect on mRNA level and the role in NP-C pathogenesis. We highlight the necessity of analyzing the raw sequence data in cases where the identified variant leads to the cryptic SS activation, as any additional genetic variants in close proximity could significantly affect the ratio of transcript isoforms and lead to a misconception in genotype–phenotype correlations or even a wrong diagnosis.