Splice Variants and Spliceosome Disorders

Scope

Splicing goes wrong in disease in three distinct ways, and this project keeps them apart because they call for different pathographs:

1. The machine is broken (trans). A variant in a spliceosome component, an snRNA gene or a core protein, changes how many introns across the genome are handled. The disease mechanism is a pattern of mis-splicing, and the open question is usually which affected transcripts produce the phenotype.
2. One transcript is broken (cis). A variant in or near a splice site, or a deep-intronic variant that creates a pseudoexon, mis-splices a single gene. This is a variant-level fact on an ordinary gene entry, not a separate disease.
3. Splicing is the treatment. Splice-switching antisense oligonucleotides and small molecules correct or redirect splicing of a single transcript.

The first group is the core of the project. The other two are tracked so their modelling stays consistent.

Artifacts

Group 1: spliceosome component disorders

snRNA genes

These genes are short, noncoding and multi-copy, so variants are missed by standard exome capture and the copy that matters is often the one expressed in brain. Dominant disease variants cluster in a few nucleotides of one structural element.

Spliceosome protein genes

Covered as subtypes, not entries

None of these needs a stub.

MONDO has no term for the recessive RNU4-2 disorder or for U4/U6 snRNA-gene retinitis pigmentosa, so both entries leave disease_term unbound and map a broader MONDO class (Mendelian neurodevelopmental disorder; retinitis pigmentosa) as skos:broadMatch, with the searches recorded in each mapping_justification. ReNU Syndrome lists both as differential diagnoses pointing at the new entries.

Not yet curated

No spliceosome-component stubs remain in stubs/.

Group 2: cis splice-altering variants

These belong on the gene's own entry, as genetic[].variants with the splicing consequence stated, and not as new disease entries. Use the noncoding-variant-impact skill for deep-intronic and regulatory variants, and keep the physical change, the splicing outcome and the evidence for that outcome separate. No audit of how consistently the KB records these has been done yet.

Group 3: splice-switching therapy

Entries with an oligonucleotide treatment whose oligonucleotide_mechanism is SPLICE_MODULATION_EXON_SKIPPING or SPLICE_MODULATION_EXON_INCLUSION: Spinal Muscular Atrophy, Duchenne Muscular Dystrophy, Dravet syndrome, Ataxia Telangiectasia, Neuronal Ceroid Lipofuscinosis 7, Leber Congenital Amaurosis 10, Usher Syndrome Type 2 and Stargardt Disease, among others. Treatments that use this mechanism should conform to antisense_oligonucleotide_therapy.

Open questions

• Which mis-spliced transcripts cause the phenotype? Not answered for any major-spliceosome snRNA disorder. RNU5B-1 is recorded as a KNOWLEDGE_GAP with two proposed experiments: splicing in isogenic stem-cell-derived neurons, and the level and spliceosome incorporation of the mutant U5B-1.
• Neomorphic or loss of function? For dominant snRNA variants, the mutant RNA may redirect splice-site choice, or be degraded by snRNA quality control, or poison spliceosomes broadly. The RNU5B-1 entry carries both models as competing mechanistic_hypotheses, and ReNU Syndrome has a matching gap.
• Why are the phenotypes tissue-restricted? Mutations in core, ubiquitous machinery give craniofacial dysostoses (EFTUD2, SF3B4, SNRPB, TXNL4A), isolated retinopathy (PRPF31, PRPF8, SNRNP200) or neurodevelopmental disease (RNU4-2, RNU5B-1, RNU2-2). This is recorded as a KNOWLEDGE_GAP on spliceosome_component_dysfunction. The Eftud2 work shows the P53 branch contributes but is not sufficient: deleting Trp53 in neural crest did not rescue development.
• Is RPE failure primary in the PRPF retinopathies? Prpf8 knock-in mice show an RPE phagocytosis defect that patient-derived RPE did not reproduce, while retinal organoids degenerate without any RPE. Recorded as a HUMAN_MODEL_MISMATCH on PRPF8-Related_Retinopathy.

Progress

• spliceosome_component_dysfunction module: component variant, then mis-splicing of a sensitive transcript subset (the key conformance target), an optional P53 branch, failure of splicing-sensitive cells, and tissue-restricted disease.
• Spliceosomopathies grouping: 24 members defined by a single CONFORMS_TO_MODULE criterion, three of them through the nested RNU4ATAC Spectrum Disorder grouping.
• Curated from stubs: PRPF8, SNRNP200, U2AF2, PRPF19, CWC27, SNRPE (Hypotrichosis 11) and LSM7. RNU12 cerebellar ataxia was folded into the existing SCAR33 subtype.
• New entries without a MONDO term: RNU4-2-related autosomal recessive neurodevelopmental disorder and U4/U6 snRNA gene-related retinitis pigmentosa, previously listed only as differential diagnoses of ReNU. Neither shows a splicing defect in patient blood, so each conforms to the module at its variant node only (and, for the retinopathy, the photoreceptor node).
• Wired to the module: SNIP1, DHX16, WBP4 and SRRM2. Craniofacial Microsomia is left out, because its SF3B2 node also carries FOXI3, GBX2 and SHROOM3.
• Not every member conforms at the mis-splicing node. Mis-splicing is not shown for Hypotrichosis 11 (the founding study found none in patient lymphocytes) or LSM7 (no splicing assay yet), and for PRPF19 it rests on fly ortholog knockdown only.

Possible next steps

• Test which LSm ring LSM7 variants disrupt: the entry proposes a patient-cell experiment that separates splicing from mRNA decay.
• Bind disease_term on the recessive RNU4-2 and U4/U6 snRNA retinitis pigmentosa entries once MONDO has terms for them.
• Audit how consistently Group 2 splice variants are recorded on gene entries.