Understanding Genotype–Phenotype Correlations of Desmoplakin Splice Site Variants

Evidence provided by Pascolini et al. [1], supports our previous observation that splice site variant DSP:c.273 + 5G > A is disease-modifying by reducing the total dose of desmoplakin and aggravating the phenotype in combination with a nonsense variant on the other allele [2]. Pathogenic DSP variants have previously been associated with phenotypes like palmoplantar keratoderma, woolly hair or epidermolysis bullosa and/or dilated, non-compaction or arrhythmogenic cardiomyopathy. This is a good example of human pathology providing new pointers of modulatory elements of splicing, which up until today remains mostly an enigma.

The unique case of DSP:c.273 + 5G > A teaches us new lessons. Firstly, functional minigene assays fall short in establishing human desmoplakin pathology. With this assay, only aberrantly spliced products, with either partial or full retention of intron 2, were observed. While in contrast, none of these aberrantly spliced RNA products were detected in cells from two affected patients. In patient cells, only native splicing was detected, although western blot revealed a reduced protein dose from this allele. Generally, minigene assays are artificial, simplified versions of a gene, and due to the enormous size of the DSP gene (332 kDa protein), only a small portion of this gene could be inserted into the plasmid vector. Other non-included segments of DSP could possess regulatory elements that influence the outcome of splicing. The splicing of most introns seems to involve a choice between multiple possible splice sites, both real and pseudo sites [3]. This variant seems to affect this choice by decreasing the recognition of the real site. Hence, the native splice site is weakened, and other pseudo sites are strengthened. As a result, the fine balance of isoforms produced by alternative and normally spliced exons is disturbed.

The majority of reported DSP variants lack functional data to support their pathogenicity [4]. In the ClinVar database, over 65 variants in donor and acceptor splice sites spanning the 24 exons of the DSP gene have been reported. Besides the DSP:c.273 + 5G > A variant modifying exon/intron 2 splicing, only a few of these were functionally investigated (Table 1). Variant c.423 − 1G > T was shown to cause exon 4 skipping with a subsequent premature termination codon (PTC) by means of RNA sequencing of patient-derived cells. However, no protein data were available to further support this. Variant c.939 + 1G > A caused retention of intron 7 and introduction of a PTC, which induced nonsense mediated mRNA decay (NMD) and no protein expression. Lastly, variant c.2876_2877 + 3del abrogating the donor splice site of exon 20 caused read through of intron 20 and a PTC, and while RNA/protein studies were lacking, immunohistochemistry on the biallelic patient-derived biopsy showed that antibodies recognised the N-terminus and ROD domain of desmoplakin, but not the C-terminus.

Full or partial retention of introns frequently create PTCs, which will often but not always cause NMD, and thus mostly classify under the loss-of-function variants of DSP. Opposingly, skipping of exons will mostly lead to truncated protein variants, and both the former and the latter are disease causing. The exact location and type of variation in the splice site region may, to some degree, predict the outcome. The (DSP:c.273) + 5G > A variant now proves that it is capable of generating wild type transcripts, but other canonical 5′ splice site +2T > C variants are known to do the same in some genomic contexts [5]. Whether this happens for the DSP gene remains to be investigated.

To conclude, uncovering disease-modifying elements of DSP (splice site) variants requires the utilisation of patient-derived material (cells, tissue) for both RNA and desmoplakin protein analysis. For patients with the DSP:c.273 + 5G > A variant in cis in combination with a DSP nonsense variant in trans, frequent cardiac screening and subsequent clinical management are very important.

M.C.B.: feedback and editing. M.C.S.C.V.: research and writing.

The authors declare no conflicts of interest.