Saturation editing of RNU4-2 reveals distinct dominant and recessive disorders
Article metadata
Article Date: 08 April 2026
Article URL: https://www.nature.com/articles/s41586-026-10334-9
Article Title: Saturation editing of RNU4-2 reveals distinct dominant and recessive disorders
Article Image: (none provided)
Summary
This study used saturation genome editing (SGE) to test 539 engineered variants across the 145-nt noncoding RNU4-2 (U4 snRNA) transcript in human HAP1 cells. The authors produced a high-resolution functional map that: (1) accurately separates known dominant ReNU syndrome variants from population variants (SGE AUC 0.93), (2) narrows the previously defined 18-nt critical region (CR) to two smaller hotspots — the T-loop (n.62–70) and Stem III (n.75–78) — and (3) reveals a separate set of depleted variants outside the CR that underlie a newly recognised recessive neurodevelopmental disorder (NDD) with distinct clinical and MRI features. The SGE function scores correlate with clinical severity for dominant ReNU variants and can be used to provide calibrated functional evidence (ACMG PS3/BS3) for variant interpretation within the CR. The work also shows limitations of in silico predictors (for example, CADD) and of a growth-based HAP1 assay, and highlights pleiotropy of a very short noncoding RNA where different regions produce dominant versus recessive disease mechanisms.
Key Points
- SGE of RNU4-2 in HAP1 cells scored 539 variants and achieved high replicate concordance (Pearson r ≈ 0.83–0.86).
- All 18 variants reported in ReNU syndrome were significantly depleted by SGE; function scores discriminate ReNU variants from population variants with ROC-AUC = 0.93.
- The original 18-nt critical region is refined to two narrower hotspots: T-loop (n.62–70) and Stem III (n.75–78), where most pathogenic dominant variants cluster.
- Function scores stratify disease severity: stronger depletion correlates with more severe developmental delay, intellectual disability and absent/limited speech.
- Many depleted variants outside the CR map to four structural regions important for protein or Sm-ring binding; biallelic variants in these regions cause a distinct recessive NDD with white matter changes and cerebellar atrophy.
- SGE outperforms CADD and RNA-binding ΔΔG predictions for RNU4-2 variant classification; CADD cannot reliably separate pathogenic from benign in this highly conserved snRNA.
- Calibrated SGE scores can provide strong PS3/BS3 evidence for clinical classification within the CR, but evidence outside the CR remains provisional (recommend PS3 supporting until orthogonal calibration).
- Assay limitations: growth-based readout in HAP1 cells may not capture tissue-specific splicing changes; haploid/diploid context does not fully separate dominant vs recessive effects.
Content summary
The authors designed an HDR library to introduce all single-nucleotide variants across RNU4-2 plus selected indels and performed SGE in LIG4-deficient haploid HAP1 cells using a unique upstream gRNA and a PAM-blocking edit to ensure locus specificity. Sequencing at day 4 and day 14 enabled calculation of function scores (log2 day14/day4 normalised to neutral insertions).
Results show depleted variants cluster rather than distribute evenly. All previously reported ReNU syndrome variants are depleted; many population-observed variants score neutral. The authors redefined the functional CR to two smaller regions corresponding to known U4/U6 structural elements and showed function scores within the CR correlate with phenotype severity and with the degree of splicing disruption measured in patients.
Unexpectedly, SGE identified many depleted variants outside the CR. Searching rare-disease cohorts uncovered 20 individuals with biallelic depleted variants (homozygous or compound heterozygous) who present a recessive NDD clinically distinct from ReNU syndrome. These recessive variants localise to structural elements required for protein binding and U4 biogenesis (for example, k-turn, Stem II central region, Sm binding site), and some map to positions equivalent to pathogenic variants in the minor spliceosome U4atac gene.
Diploid HAP1 experiments attenuated scores across the gene but did not neatly separate dominant from recessive variants. The authors therefore caution that functional scores alone cannot fully distinguish inheritance mode without orthogonal evidence. They also highlight that most ReNU cases share a recurrent insertion (n.64_65insT), but SGE shows many other variants have equal or greater functional impact, suggesting recurrence may reflect mutation rate or germline selection rather than unique functional severity.
Context and relevance
This paper provides a variant-effect map for the first noncoding snRNA shown to be a major contributor to neurodevelopmental disease. It both improves clinical interpretation of RNU4-2 variation and reveals a separate recessive syndrome, expanding the locus’ clinical spectrum. The results are directly relevant to clinical genetics, diagnostic laboratories and researchers working on splicing, noncoding RNAs and variant interpretation frameworks. The SGE approach and locus-specific design to overcome paralogue homology are widely applicable to other disease-linked snRNAs.
Why should I read this?
Short version: if you deal with genetic diagnoses, splicing or neurodevelopmental disorders, this paper saves you time and head-scratching. It tells you which RNU4-2 changes are actually damaging, tightens the critical region for dominant ReNU cases, and flags an entirely different recessive syndrome caused by biallelic hits elsewhere in the gene. In other words — fewer false alarms, better variant calls, and a new disease to be aware of. Read it if you want clinical-grade functional data for a tiny but clinically important RNA.
Author style
Punchy: this is a must-read for clinical labs and researchers focusing on splicing or noncoding variants. The work delivers actionable functional evidence that will change how RNU4-2 variants are classified and uncovers a distinct recessive disorder — an important advance with immediate diagnostic utility.