CLN7 disease, also known as variant late-infantile neuronal ceroid lipofuscinosis, is a lysosomal storage disorder caused by biallelic defects in MFSD8. The gene encodes major facilitator superfamily domain-containing protein 8, also called the CLN7 protein, a 518-amino-acid lysosomal transmembrane protein whose precise function remains incompletely understood but is considered essential for normal neuronal homeostasis. Current evidence suggests that MFSD8 may function as a chloride channel involved in lysosomal enzyme trafficking. Classic CLN7 disease typically begins between the ages of two and seven years and can involve seizures, motor impairment, progressive mental regression, myoclonus, speech impairment, vision loss and personality disorders. More broadly, neuronal ceroid lipofuscinoses are estimated to occur in one to four people per million live births, with frequencies varying by genetic subtype and population.

The proband was the first and only child of healthy, non-consanguineous Russian parents. She was born at 38 weeks after an uncomplicated pregnancy, apart from a first-trimester threatened miscarriage. Her birth weight was 3,170 g, her length was 52 cm, and her Apgar scores were 8 and 8. Early motor and speech development was initially within normal limits: she walked independently at 13 months and used phrase speech by two years and four months. At approximately four years of age, subtle cognitive and behavioral changes appeared, including reduced attention and memory. Her first atonic seizure occurred during a febrile illness at four years and ten months. Recurrent afebrile atonic seizures followed, and myoclonic seizures began at five years and two months.

As the illness progressed, the child developed progressive intention tremor, gait ataxia, dysarthria and motor incoordination. From the age of six, her neurological deterioration accelerated. She lost expressive speech, regressed in fine and gross motor skills and lost control of her bladder and bowels. By six and a half years, she could no longer walk independently; later, she could not sit or feed herself. A neurological examination at six years and eight months showed severe developmental regression, absent speech, generalized hypotonia with brisk deep tendon reflexes and ankle clonus, marked upper-limb, head and trunk tremor, truncal ataxia, impaired fixation and complete loss of independent ambulation. She was underweight, with a weight of 21 kg and a BMI of −2.9 standard deviations, and had an acquired head circumference of 50 cm, or −1.1 standard deviations, despite a height of 132 cm, or +2.64 standard deviations.

Brain magnetic resonance imaging revealed diffuse cerebral and cerebellar atrophy. Video electroencephalography showed diffuse background slowing with generalized spike/polyspike-and-slow-wave discharges and a posterior photoparoxysmal response during intermittent photic stimulation. Ophthalmological examination found only subtle optic disc pallor, without clinically significant retinal degeneration at the time of evaluation. By eight years and two months, the patient was bedridden, had lost the ability to swallow independently and required gastrostomy feeding. She also developed severe sleep disturbances requiring pharmacological treatment, while her seizures became more frequent and could be terminated only with injectable midazolam. The clinical course therefore combined childhood-onset epilepsy, progressive developmental regression, ataxia, loss of motor and speech function, and cerebral and cerebellar atrophy.

The initial diagnostic evaluation did not reveal a cause. Routine metabolic investigations, lysosomal enzyme assays, echocardiography, nerve-conduction studies, serum creatine kinase analysis and whole-exome sequencing were all unremarkable. Trio whole-genome sequencing subsequently identified the homozygous deep intronic MFSD8 variant c.999-556C>G, with each parent carrying a heterozygous copy. The variant was absent from gnomAD version 4.1.1 and had not previously been reported in ClinVar or the literature. Its intronic location meant that its effect could not be established from DNA sequence alone, particularly because noncoding variants cannot usually be interpreted from the nucleotide change without functional evidence. The investigators therefore tested the variant at the RNA level to determine whether it altered MFSD8 splicing.

SpliceAI predicted that c.999-556C>G would create cryptic splice sites with a delta score of 0.73 for the acceptor and 0.67 for the donor, resulting in inclusion of a 98-nucleotide pseudoexon. To evaluate that prediction, the researchers isolated RNA from cultured skin fibroblasts of the proband and her heterozygous father. They amplified the region spanning exons 6 through 12 by reverse-transcription polymerase chain reaction and analyzed the products through targeted next-generation sequencing, using RNA from a healthy donor fibroblast culture as a control. Targeted sequencing confirmed inclusion of the predicted pseudoexon at chromosome 4 positions 127,922,520–127,922,617 in RNA from the proband and her father. Both individuals had a canonical transcript and an aberrant transcript containing the pseudoexon.

Pseudoexon inclusion caused a frameshift and generated the premature termination-codon allele p.(Ile334Valfs*113). Among the MFSD8 transcript molecules detected in the targeted amplicon from the homozygous proband, 84% contained the pseudoexon. In the heterozygous father, the aberrant isoform accounted for 12% of the detected transcripts. These figures describe relative proportions within the analyzed RNA amplicons, rather than absolute transcript abundance. The lower proportion in the father is compatible with partial degradation of the premature-termination-containing RNA by nonsense-mediated decay, although the contribution of that mechanism could not be directly quantified because no informative heterozygous exonic variants were available for allele-specific analysis. Treating the father’s fibroblasts with a nonsense-mediated-decay inhibitor increased the relative abundance of the pseudoexon-containing transcript by approximately 1.7-fold, but it did not restore the fraction to the roughly 50% expected for a heterozygous genotype.

An independent minigene assay reproduced the splicing defect in an engineered system. The construct carrying the mutant sequence generated predominantly the aberrant transcript containing the 98-nucleotide pseudoexon, which accounted for 84.3% of transcripts, together with a minor canonical population of 15.5%. By contrast, the wild-type construct generated only the canonical transcript. The mutant construct’s continued production of some canonical RNA indicates that the variant strongly promotes aberrant splice-site recognition without completely abolishing recognition of the normal splice sites. The abundance of canonical transcripts in the minigene system cannot, however, be extrapolated directly to the endogenous MFSD8 locus or used to infer the amount of functional MFSD8 protein remaining in the patient. In combination with the clinical concordance and the findings in patient-derived RNA, the minigene result supports a causal relationship between c.999-556C>G and the abnormal splicing event. The authors consequently classified it as the causative mutation and considered other variants found by whole-genome sequencing unlikely to be disease-causing.

The case illustrates why deep intronic variants can escape routine diagnosis. Exon-focused sequencing may miss pathogenic changes located within introns, while DNA analysis alone may not establish what a noncoding variant does to RNA processing. The findings show how genome sequencing can identify a candidate regulatory change and how transcript analysis can directly test its effect at the endogenous gene. The report nevertheless remains a single-patient case: it does not establish how frequently this mechanism occurs in CLN7 disease, how it is distributed across populations or whether the same variant produces the same clinical course in other people. It nevertheless adds a distinct deep intronic allele to an MFSD8 spectrum already known to include missense, nonsense, frameshift, splice-site and copy-number variants. The absence of prominent retinal degeneration in this patient, despite severe neurological progression, also documents variability within the clinical presentation of CLN7 disease. A patient-specific antisense oligonucleotide has previously been developed for pseudoexon inclusion in another MFSD8-related case, and the clearly defined splicing defect reported here may provide a measurable molecular endpoint for individualized therapeutic studies, but no treatment was tested in this case report.

Subject of Research: Functional characterization of a deep intronic MFSD8 variant associated with CLN7 disease

Article Title: Case Report: A novel deep intronic MFSD8 splice-altering variant causes CLN7 disease through pseudoexon inclusion

Article References: Akimova, D., Voskanyan, A., Orlova, A., & Skoblov, M. (2026). Case Report: A novel deep intronic MFSD8 splice-altering variant causes CLN7 disease through pseudoexon inclusion. Frontiers in Genetics, 17, Article 1980180. https://doi.org/10.3389/fgene.2026.1980180

Image Credits: AI Generated

DOI: 10.3389/fgene.2026.1980180

Keywords: MFSD8, CLN7 disease, deep intronic variant, pseudoexon inclusion, RNA splicing, whole-genome sequencing, rare disease genetics, neuronal ceroid lipofuscinosis, patient-derived fibroblasts, minigene assay, Case, Report