Kdm5b mouse model

To experimentally investigate the potential additive dosage effect of Kdm5b LoF, we performed behavioral tests, morphological measurements and brain differential gene expression analysis in WT, HET and HOM Kdm5b LoF mice. A mouse Kdm5b LoF allele (Mouse Genome Informatics ID: 6153378) was generated previously⁴⁵ using CRISPR/CAS9 mediated deletion of coding exon 7 (ENSMUSE00001331577), leading to premature translational termination due to a downstream frameshift. Breeding of testing cohorts was performed on a C57BL/6NJ background. Mice were housed in specific pathogen‐free mouse facilities with a 12-hour light–dark cycle (lights on at 7:30), an ambient room temperature of 21 °C and 55% humidity at the Research Support Facility of the Wellcome Sanger Institute. They were in mixed genotype cages (2–5 mice), and housed in individually ventilated cages (GM500, Tecniplast) containing Aspen chip bedding and environmental enrichment (Nestlets nesting material and cardboard play tunnels, Datesand). Food and water were provided ad libitum.

We applied a battery of behavioral tests commonly used to study mice for signs of perturbed neurodevelopment, including light–dark box (adapted from Gapp et al.⁸²), Barnes maze probe trial and new object recognition (Supplementary Information). We assessed a cohort of 25 WT, 34 HET and 15 HOM Kdm5b mutant male mice at 10 weeks of age. Behavioral tests were carried out between 9:00 and 17:00, after 1 hour of habituation to the testing room. Experimenters were blind to genotype; mouse movements were recorded with an overhead infrared video camera for later tracking using automated video tracking (EthoVision XT 11.5, Noldus Information Technology). We also measured mouse cranial length and width, skull height and transitional vertebrae phenotype with X-ray whole-body radiography for 15 Kdm5b^+/+, 12 Kdm5b^+/− and 9 Kdm5b^−/− mice (Supplementary Information).

All statistical analyses of mouse data were performed using R (v.4.1.3). Data were first transformed to achieve normality, using Box–Cox transformation (MASS package v.7.3–55) for behavioral data (λ limit = −2, 2) or quantile normalization (qnorm function) for X-ray data. Testing for genotype effect was performed using a double generalized linear model (dglm package v.1.8.5). The type of object used for new object recognition had a small (6%) and significant (P = 0.036) effect; therefore, it was used as a covariate for Box–Cox transformation and dglm. For visualization purposes, residual values were calculated from the linear model and z-scores were calculated relative to WT.

We also performed differential gene expression analysis for the Kdm5b mouse to assess the impact of the Kdm5b HET and HOM mutations on brain gene expression. RNA-seq was done for whole-brain embryonic tissue, and for the FC, HIP and CB of adult WT, HET and HOM Kdm5b mice. Differential gene expression and log₂ fold changes were obtained, and P values for differences in gene expression were calculated. A P threshold of 0.10 was used to identify significant differences between WT and mutant samples. In addition, Gene Ontology (GO) enrichment analysis was performed to identify functionally enriched terms in the DEGs (with a 5% FDR threshold). In all analyses, the background consisted of only genes expressed in the tissue studied. GO terms with more than 1,000 genes were excluded from the analysis. The European Nucleotide Archive (ENA) accession numbers for the RNA-seq sequences reported are listed in Supplementary Table 17. Further details on Kdm5b mouse RNA extraction, sequencing, data processing and analyses can be found in the Supplementary Information.