RNU4-2-Related Autosomal Recessive Neurodevelopmental Disorder: Research Report

Scope. This disorder is caused by biallelic variants in RNU4-2 (hgnc:10193). RNU4-2 is a non-protein-coding gene that makes the U4 small nuclear RNA (snRNA), a core part of the major spliceosome. This is a different disease from the dominant, usually de novo ReNU syndrome, even though both come from the same 145-nucleotide gene.

Where the evidence comes from. Almost everything known rests on three studies, all from 2026:

• Rius et al. 2026, Nat Genet (PMID:41951959): the disease-defining clinical cohort.
• De Jonghe et al. 2026, Nature (PMID:41951737): the companion saturation-editing study that found the disease.
• Hayashi et al. 2026, J Hum Genet (PMID:41408479): an independent Japanese/international series.

All three are already in references_cache/. Quotes below are copied from those cached files. A kb/disorders/RNU4-2-Related_Autosomal_Recessive_Neurodevelopmental_Disorder.yaml entry already exists and already uses most of this material.

1. Disease information

Overview. This is a moderate-to-severe syndromic neurodevelopmental disorder (NDD) caused by homozygous or compound heterozygous RNU4-2 variants. The variants lie outside the 18-nucleotide ReNU critical region. Its hallmark is a distinctive pattern of white-matter change on brain MRI.

• Rius et al.: "Here we describe a new recessive NDD associated with homozygous and compound heterozygous variants in RNU4-2." (PMID:41951959)
• Rius et al.: "we have characterized the phenotype associated with biallelic variants in RNU4-2 as a moderate-to-severe syndromic neurodevelopmental disorder with distinct white matter abnormalities." (PMID:41951959)
• De Jonghe et al.: "We show that these variants cause a new recessive neurodevelopmental disorder that is distinct from ReNU syndrome." (PMID:41951737)

Identifiers

Synonyms: recessive RNU4-2-related (or -associated) NDD; recessive RNU4-2 syndrome; biallelic RNU4-2 neurodevelopmental syndrome.

Data source. The data are aggregated cohort data from individual patients in rare-disease sequencing programmes, not EHR data. The programmes are Genomics England 100,000 Genomes Project and NHS GMS, CaRDinal, SeqOIA/Auragen (PFMG 2025), UDN, Broad CMG and GREGoR. Controls came from UK Biobank and All of Us.

2. Etiology

Cause. The disease is caused by biallelic loss-of-function variants in RNU4-2. The variants fall in parts of U4 that bind other spliceosome components:

• "We identify 38 individuals with biallelic variants outside the 18-nucleotide ReNU syndrome region that cluster within other functionally important elements of U4: Stem II, the k-turn and the Sm protein binding site." (PMID:41951959)
• "These biallelic variants map to regions of U4 that are important for binding to other spliceosome components: the Stem II region of interaction with U6, the Sm protein binding site that is essential for snRNA biogenesis and stability, and the k-turn structure in the 5′ stem loop that binds to SNU13/15.5k." (PMID:41951959)

Genetic risk factors

• Parental consanguinity. "Eight individuals from five families, all of whom had homozygous variants, had consanguineous parents." (PMID:41951959)
• High mutation rate at the locus. "RNU4-2, like other snRNA genes, has a substantially elevated mutation rate" (PMID:41951959). The same paper notes that negative selection keeps every variant very rare. This lowers the chance of homozygosity in outbred populations.

Modifier genes. None established. RNU4-1, the paralog, is upregulated in affected people but does not compensate fully (see section 6).

Environmental, protective and gene–environment factors. None reported. The disease is purely genetic.

3. Phenotypes

The figures below come from 31 individuals with detailed data (21 male, 10 female; median age at last follow-up 10 years, range 6 weeks to 32 years) in Rius et al. 2026, Table 1 (PMID:41951959). All suggested HP IDs were checked against cache/hp/terms.csv.

Onset. "Most individuals had infantile-onset phenotypes (n = 16; 51.6%), while ten (32.2%) had congenital or neonatal onset and five (16.1%) had childhood onset." (PMID:41951959)

Severity relative to ReNU syndrome. The two available sources differ:

• Hayashi et al. describe their biallelic cases as milder: "individuals with biallelic variants exhibited developmental delay and intellectual disability that were similar to but milder than those with monoallelic variants, notably lacking extracerebral organ involvement" (PMID:41408479). That cohort had only 4 individuals from 2 families.
• Rius et al. report GDD, speech delay and seizure frequencies similar to ReNU, plus extracerebral involvement (genital, skin, dental).

Quality of life. No validated QoL instrument data exist. Lifelong dependence can be inferred from the high rates of nonverbal status and non-ambulation.

4. Genetic and molecular information

Gene. RNU4-2 (hgnc:10193) encodes the U4 snRNA. Alignment with its paralog RNU4-1 shows mismatches at "only 4 of the 145 nt" (PMID:41951737).

Variant spectrum (Rius et al.)

• 35 unique variants in 38 included individuals: 7 homozygous variants (13 individuals) and 31 compound heterozygous variants (25 individuals).
• Variants seen in both the homozygous and compound heterozygous state: n.127C>T, n.119A>G and n.7G>C.
• The broader search found 43 individuals from 33 families. "Fifteen individuals from 11 families had homozygous variants, and 27 individuals from 22 families had compound heterozygous variants." Five were later excluded because their SGE scores were similar to those in the population.

Variant types. Mostly single-nucleotide variants, plus small indels such as n.7_8insA and n.43_44insT. This is a noncoding RNA, so "missense" and "nonsense" do not apply.

Structural regions. The SGE paper defines these coordinates (PMID:41951737):

The ReNU critical region (T-loop n.62–70, Stem III n.75–78) is spared in this disorder. Population variants are instead enriched in the 3′ stem loop (n.85–117): 7 of 14 UK Biobank variants versus 0 of 35 case variants, P = 5.0×10⁻⁴.

Equivalent RNU4ATAC variants. Thirteen of the 35 variants (37.1%) sit at nucleotides where the equivalent RNU4ATAC variant is pathogenic or likely pathogenic. Five exact equivalents are listed in the SGE paper: n.32G>A, n.45G>C, n.46G>A, n.119A>G and n.122T>G. For n.119A>G, the RNU4ATAC equivalent n.117A>G is ClinVar ID 1525441. n.119A>G was homozygous in two individuals and compound heterozygous in three (PMID:41951737).

ACMG/AMP classification

• Of the 35 variants: 1 pathogenic, 8 likely pathogenic, the rest VUS.
• Per individual: 10/38 (26.3%) had P/LP variants on both alleles, 4/38 had one P/LP allele, and 24/38 (63.2%) had VUS on both alleles (PMID:41951959).
• No phenotype frequency differed between the P/LP–P/LP group and everyone else.

Using the SGE function scores clinically

• A variant counts as significantly depleted (deleterious) when its function score is below −0.302.
• The SGE authors recommend that "PS3 supporting evidence be assigned to significantly depleted variants outside the CR" (PMID:41951737).
• A normal score does not show a variant is benign: "normal SGE scores should not be used as evidence of benignity in a recessive context" (PMID:41951959).
• Only 20 of 35 case variants (57.1%) were SGE-significant.

Population frequency

• SGE-depleted variants outside the critical region are carried by 0.12% of UK Biobank and 0.094% of All of Us participants (PMID:41951737).
• Only 11 UK Biobank participants were biallelic; none had a neurodevelopmental phenotype.
• Only one UK Biobank / All of Us individual was homozygous for an SGE-depleted variant (n.31T>G, score −0.730). That person reported cognitive and functional difficulties suggestive of possible ID (PMID:41951737).

Other genetic features

• Origin: germline, inherited from carrier parents.
• Parental origin of alleles: maternal and paternal alleles had similar SGE scores (P = 0.095).
• Heterozygous carriers: carriers of depleted variants outside the critical region (n = 592 in UK Biobank) showed no differences in fluid intelligence, developmental diagnoses or education.
• Epigenetics and chromosomal abnormalities: none reported. The disease is not caused by copy-number change.

5. Environmental information

Not applicable. No environmental, lifestyle or infectious contributors are reported.

6. Mechanism and pathophysiology

Causal chain

1. Biallelic germline variants arise in RNU4-2 in Stem II, the 5′-stem-loop k-turn, the Sm site or the terminal stem loop. These variants disrupt U4's binding to U6, SNU13/15.5K or the Sm ring. This step is inferred from structure, from yeast U4 mutagenesis and from in vitro SNU13 binding data, as cited in PMID:41951737.
2. → This leads to reduced steady-state U4 (RNU4-2) RNA. This is shown in blood RNA-seq: mean 314 TPM versus 1,645 TPM in controls, P = 0.012 (PMID:41951959). Impaired biogenesis or stability is the presumed route, which is inferred, especially for Sm-site variants.
3. → This is accompanied by compensatory upregulation of RNU4-1 (1,952 vs 346 TPM). The RNU4-2:RNU4-1 ratio shifts from 5.4 in controls to 0.2 (P = 0.003). Rius et al.: "Our data suggest that RNU4-1 cannot fully compensate for loss of RNU4-2." The reason is not known. Their three candidate explanations are functional divergence, insufficient upregulation, or tissue-specific failure in the brain.
4. → The net result is partial loss of function of U4-containing complexes (U4/U6 di-snRNP and U4/U6·U5 tri-snRNP) and the major spliceosome. Cell-fitness loss is shown in haploid HAP1 cells by SGE. The specific splicing defect has not been demonstrated:
5. No ReNU-type 5′ splice-site signature appears in lymphocyte RNA-seq.
6. No excess intron retention appears in blood, unlike RNU4ATAC disease.
7. The authors suggest the defect is subtle, because a global defect across U2-type introns would be nonviable.
8. → In the nervous system this results in abnormal development and white-matter integrity: dilated perivascular spaces, white-matter volume loss, cerebellar atrophy and a thin corpus callosum. Serial imaging shows these can be progressive. The cell-level link between U4 loss and these findings is unknown. Oligodendrocyte or glial vulnerability is a hypothesis only.
9. → This produces the clinical picture: GDD/ID, absent speech, hypotonia, spasticity and ataxia, and seizures.
10. Branch: an extracerebral, multisystem arm (genital, skin, dental, growth). Its mechanism is not studied.

How it differs from ReNU syndrome

• ReNU variants (T-loop and Stem III) mis-position the U6 ACAGAGA box. This alters 5′ splice-site selection, a dominant mechanism.
• In ReNU, RNU4-2 RNA is raised (2,699 TPM) rather than reduced.
• "Together, these data suggest that dominant and recessive RNU4-2 disorders are driven by different molecular processes and are consistent with a loss-of-function mechanism in the recessive condition." (PMID:41951959)

Suggested ontology terms (checked against the local caches)

• GO:
• GO:0005687 U4 snRNP
• GO:0046540 U4/U6 x U5 tri-snRNP complex
• GO:0000244 spliceosomal tri-snRNP complex assembly
• GO:0000387 spliceosomal snRNP assembly
• "mRNA splicing, via spliceosome" is not in the local cache; look it up before use.
• CL: CL:0000540 neuron; CL:0000128 oligodendrocyte; CL:0000127 astrocyte; CL:0000121 Purkinje cell (relevant given cerebellar atrophy; speculative); CL:0000542 lymphocyte (the tissue in which RNA changes were measured).
• Model and assay context: the SGE assay was run in HAP1-LIG4-KO cells; evidence from it is IN_VITRO.

Omics

• Blood RNA-seq from the Genomics England NGRL (n = 7,826; 3 biallelic cases).
• Lymphocyte RNA-seq (3 biallelic cases, 19 ReNU, 20 controls).
• No proteomic, metabolomic, single-cell or spatial data exist.

7. Anatomical structures affected

• Primary: brain.
• Cerebral white matter: periventricular and deep, peritrigonal.
• Cerebellum (UBERON:0002037): both hemispheres and vermis.
• Corpus callosum (UBERON:0002336).
• Lateral ventricles (ventriculomegaly).
• "Cerebral white matter" is not in the local UBERON cache; look it up before use.
• Secondary: eye (strabismus), male external genitalia, skin and nails, teeth, skeleton, endocrine system, and growth.
• Subcellular: nucleus, specifically the spliceosomal snRNPs (GO:0005687, GO:0046540).
• Laterality: the white-matter changes are bilateral and symmetric (for example, "peritrigonal region bilaterally").

8. Temporal development

• Onset: congenital or neonatal in 32%, infantile in 52%, childhood in 16%. Hypotonia is often neonatal.
• Course: a static neurodevelopmental disorder in which imaging can progress. Two siblings imaged serially from infancy at first showed only ventriculomegaly. Follow-up "revealed progressive pathology, including perivascular space dilation, white matter volume loss and cerebellar atrophy." Perivascular dilation also increased between 32 months and 7 years in another individual (PMID:41951959).
• Within families: siblings were largely concordant, but expressivity varied. In three sibships, one child walked and the other did not.
• Duration: lifelong. The oldest reported individual was 32. No remission has been described.
• Critical period: early childhood, when white-matter change appears to evolve. This is not formally established.

9. Inheritance and population

• Inheritance: autosomal recessive (HP:0000007), homozygous or compound heterozygous.
• Penetrance: not established. Rius et al. note that some included individuals may be false positives and that some excluded individuals may be true cases.
• Expressivity: variable, including within sibships.
• Anticipation and germline mosaicism: none reported.
• Founder effects: none reported.
• Consanguinity: important. All five consanguineous families carried homozygous variants.
• Frequency: much rarer than ReNU syndrome. "while we identified 61 individuals with ReNU syndrome in 8,841 individuals (0.69%) with previously undiagnosed NDD in the Genomics England 100,000 Genomes Project, we only identified seven (0.08%) with biallelic variants in the same cohort." (PMID:41951959)
• Case enrichment: 6/5,386 undiagnosed NDD trios versus 0/4,776 non-NDD trios (P = 3.3×10⁻³).
• Population prevalence and incidence: unknown. The authors expect higher frequency in populations with more consanguinity.
• Carrier frequency: about 0.1% carry SGE-depleted variants outside the critical region (UK Biobank 0.12%, All of Us 0.094%). This is a ceiling on the frequency of deleterious alleles, not a confirmed carrier rate.
• Sex: 21 male and 10 female in the clinical cohort. This is probably ascertainment, not biology; a recessive autosomal locus should affect both sexes.
• Geography: cases from the UK, France, Germany, the USA, Australia, Turkey and Pakistan, plus a Japanese/Israeli/Malaysian/Brazilian series in Hayashi et al. No endemic clustering has been reported.

10. Diagnostics

• Genetic testing is required. RNU4-2 is noncoding and is missed by standard exome capture.
• Whole-genome sequencing (NCIT:C101294) is the recommended test.
• Extended-exome designs that capture snRNA genes also work. PMID:42502410 reports WGS reanalysis of 1,578 unsolved probands that found RNU4-2 cases, and recommends extended-exome screening.
• Variant calling must allow for paralog and pseudogene homology with RNU4-1.
• Phasing: use parental testing, or read-backed phasing in IGV when parents are unavailable.
• Variant interpretation:
• Use the SGE function scores at supporting strength (PS3_supporting) for significantly depleted variants outside the critical region.
• Check whether the equivalent RNU4ATAC nucleotide is pathogenic.
• Check the structural region.
• Imaging: MRI (NCIT:C16809). Marked periventricular perivascular-space dilation with a "tightly packed microcystic" appearance should prompt RNU4-2 analysis: "The presence of these changes on MRI, particularly the marked periventricular perivascular space dilation, should prompt sequencing and analysis of variants in RNU4-2." (PMID:41951959)
• Candidate RNA biomarker (not yet validated): the RNU4-2:RNU4-1 expression ratio in rRNA-depleted blood RNA-seq. "it could be a useful diagnostic biomarker for the recessive disorder, if validated in additional cohorts." (PMID:41951959)
• Facial gestalt: GestaltMatcher separates biallelic cases from ReNU cases (biallelic–biallelic pairs P = 1.8×10⁻⁵; ReNU–biallelic pairs P = 0.37).
• EEG: obtain one when seizures occur. No characteristic pattern has been reported.
• Differential diagnosis:
• ReNU syndrome. Distinguished by a heterozygous T-loop or Stem III variant, the 5′ splice-site signature, raised RNU4-2 RNA, myopathic facies and no cerebellar atrophy.
• Recessive RNU2-2 disorder.
• RNU4ATAC disorders (MOPD I, Lowry–Wood, Roifman).
• RNU12 disorder.
• Other leukodystrophies and metabolic disorders with enlarged perivascular spaces, such as mucopolysaccharidoses.
• Hypomyelinating leukodystrophies.
• Screening: there is no newborn screening. Carrier and prenatal testing are possible once the familial variants are known.

11. Prognosis

• No mortality or survival data exist. Individuals up to 32 years old have been reported.
• Disability is substantial: about half are nonverbal, and 28% are non-ambulatory after age 5.
• Epilepsy is mostly controllable: 13 of 17 treated individuals were on monotherapy, 3 were pharmacoresistant, and none had status epilepticus.
• Prognostic factors are not established. More severe perivascular dilation went with lower mean SGE scores, but not significantly (P = 0.056, n = 5 per group).

12. Treatment

No disease-specific treatment exists. Management is symptomatic.

• Antiseizure medication: NCIT:C15986 Pharmacotherapy, with NCIT:C264 Anticonvulsant Agent as the agent class.
• Therapies: physical therapy (NCIT:C15302), occupational therapy (NCIT:C121351) and speech-language therapy (NCIT:C159273), including AAC for nonverbal patients.
• Supportive care (NCIT:C15747) for feeding difficulties, plus ophthalmology for strabismus, and urology or endocrine follow-up.
• Genetic counselling: NCIT:C15240.
• Experimental: no clinical trials were found. Because the disease is loss of function, restoring or augmenting U4 (for example, upregulating RNU4-1) is a conceivable future approach. This is speculative, and the SGE authors frame their data only as a resource for "therapeutic development."
• Pharmacogenomics: none known.

13. Prevention

• Primary prevention: genetic counselling. The recurrence risk is 25% for carrier couples.
• Carrier, prenatal and preimplantation testing: possible once both familial variants are identified. Carrier screening is especially relevant in consanguineous families.
• Tertiary prevention: seizure control, orthopaedic care for spasticity, and nutritional support.
• Vaccines and public-health measures: not applicable.

14. Other species

No naturally occurring animal disease has been reported, and there is no OMIA entry. U4 snRNA is deeply conserved. The RNU4-2 upstream region "is both unique and poorly conserved across species" (PMID:41951737). Yeast U4 mutagenesis implicates the same regions (stem II, k-turn, Sm site) as essential for function. Copy-number and paralog structure differ between species, so a one-to-one ortholog mapping needs care. No NCBI Gene orthology details were checked here.

15. Model systems

• In vitro: saturation genome editing in HAP1-LIG4-KO haploid human cells. All possible SNVs across the 145-nt transcript plus 6 nt 3′ (539 variants) were tested, with a cell-fitness readout (PMID:41951737).
• Strength: it found the recessive disease and gives variant-level function scores.
• Limitations:
  • It does not separate dominant from recessive effects; a diploid HAP1 screen did not either.
  • It does not measure the splicing mechanism.
  • It reports nothing about brain cells.
  • It is less sensitive for recessive variants.
• Patient-derived samples: lymphocyte cultures and whole blood (RNA-seq).
• Animal models: none reported (no mouse, zebrafish or fly).
• Gaps: neural iPSC models, organoids, or oligodendrocyte or glial models are needed to find the splicing defect and explain the white-matter pathology. Rius et al. say a splicing defect "may require additional samples sequenced at higher depth or sequencing of disease-relevant cell types and tissues."

Points for the curator

• No MONDO or OMIM identifiers exist yet. Keep the broadMatch to MONDO:0100500 until MONDO adds a class.
• Two sources disagree on severity. Hayashi et al. (n = 4) describe the disorder as milder and without extracerebral involvement; Rius et al. (n = 31) report extracerebral features. Record this as a disagreement rather than resolving it.
• No splicing signature has been shown. The loss-of-function claim rests on reduced RNU4-2 RNA and on analogy with RNU4ATAC disease, not on any demonstrated splicing defect. The KNOWLEDGE_GAP and HUMAN_MODEL_MISMATCH discussion kinds both fit here.
• Two terms need lookup before binding. "Dilated perivascular spaces" (HPO) and "mRNA splicing, via spliceosome" (GO) are not in the local caches. Look both up rather than writing an ID.

Sources

• Rius R et al. Biallelic variants in the noncoding RNA gene RNU4-2 cause a recessive neurodevelopmental syndrome with distinct white matter changes. Nat Genet 2026. PMID:41951959. nature.com/articles/s41588-026-02554-6; preprint at medRxiv 2025.08.13.25333306; Crick record
• De Jonghe J et al. Saturation editing of RNU4-2 reveals distinct dominant and recessive disorders. Nature 2026. PMID:41951737. nature.com/articles/s41586-026-10334-9
• Hayashi Y et al. Monoallelic and biallelic RNU4-2 variants in neurodevelopmental disorders. J Hum Genet 2026. PMID:41408479. nature.com/articles/s10038-025-01442-x; PubMed
• Whole-genome discovery of pathogenic snRNA variants and efficient extended-exome screening. PMID:42502410 (cached)
• Background on dominant ReNU syndrome: Chen Y et al. 2024 Nature (PMID:38991538); Greene D et al. 2024 Nat Med (PMID:38821540); dismech ReNU Syndrome page
• Related recessive snRNA disorder: RNU2-2 medRxiv preprint
• Oxford bibliographic record: ndm.ox.ac.uk publication 2404058

Reference Validation

Checked with linkml-reference-validator 0.4.0.

All extracted references resolved successfully.

Term Validation

Checked with linkml-term-validator 0.4.5, through the ols: adapter.

Terms the report names something else

These identifiers resolve, so nothing about them looks wrong, and the ontology calls them something unrelated to what the report calls them. That usually means the identifier is not the one the sentence needs:

• HP:0031936 (1 mention) - the report calls it "28/30; median first steps at 2.5 years"; HP calls it Delayed ability to walk
• HP:0001272 (1 mention) - the report calls it "12/24 (50%); OR 22.0 vs ReNU"; HP calls it Cerebellar atrophy
• HP:0000486 (1 mention) - the report calls it "15/28 (53.6%); eye involved 77.4%"; HP calls it Strabismus

Terms whose name is worth a second look

The report's name for these is recognisably related to the term's own name without being one of them. A loose paraphrase reads the same way as a citation of the wrong sibling term - and so does a related synonym, which the ontology records precisely because it names something adjacent rather than the same thing - so these are listed rather than judged:

• NCIT:C16809 (1 mention) - the report calls it "Imaging: MRI"; NCIT calls it Magnetic Resonance Imaging**, and lists "MR Imaging" among its other names
• NCIT:C15302 (1 mention) - the report calls it "Therapies: physical therapy"; NCIT calls it Physical Therapy**