Pathogenic variants in autism gene KATNAL2 cause hydrocephalus and disrupt neuronal connectivity by impairing ciliary microtubule dynamics

Abstract

Enlargement of the cerebrospinal fluid (CSF)-filled brain ventricles (cerebral ventriculomegaly), the cardinal feature of congenital hydrocephalus (CH), is increasingly recognized among patients with autism spectrum disorders (ASD). KATNAL2, a member of Katanin family microtubule-severing ATPases, is a known ASD risk gene, but its roles in human brain development remain unclear. Here, we show that nonsense truncation of Katnal2 (Katnal2Δ17) in mice results in classic ciliopathy phenotypes, including impaired spermatogenesis and cerebral ventriculomegaly. In both humans and mice, KATNAL2 is highly expressed in ciliated radial glia of the fetal ventricular-subventricular zone as well as in their postnatal ependymal and neuronal progeny. The ventriculomegaly observed in Katnal2Δ17 mice is associated with disrupted primary cilia and ependymal planar cell polarity that results in impaired cilia-generated CSF flow. Further, prefrontal pyramidal neurons in ventriculomegalic Katnal2Δ17 mice exhibit decreased excitatory drive and reduced high-frequency firing. Consistent with these findings in mice, we identified rare, damaging heterozygous germline variants in KATNAL2 in five unrelated patients with neurosurgically treated CH and comorbid ASD or other neurodevelopmental disorders. Mice engineered with the orthologous ASD-associated KATNAL2 F244L missense variant recapitulated the ventriculomegaly found in human patients. Together, these data suggest KATNAL2 pathogenic variants alter intraventricular CSF homeostasis and parenchymal neuronal connectivity by disrupting microtubule dynamics in fetal radial glia and their postnatal ependymal and neuronal descendants. The results identify a molecular mechanism underlying the development of ventriculomegaly in a genetic subset of patients with ASD and may explain persistence of neurodevelopmental phenotypes in some patients with CH despite neurosurgical CSF shunting.

Keywords: autism; cerebrospinal fluid dynamics; ciliopathy; hydrocephalus; structural brain disorder.

Fig. 1.

Katnal2Δ17 mutant mouse creation with CRISPR Cas9-mediated knockout. (A) KATNAL2 is predicted to be a microtubule severing enzyme with a LisH domain and an AAA ATPase domain highly conserved between mice and humans. The mouse gene encodes five isoforms: S1, S2, L1, L2, and L3. The Katnal2Δ17 isoform contains a 17 bp deletion encoding a nonsense truncation in exon 2 shared among all long (L) splice variants. (B) Sequence of Katnal2 exon 2, the translated protein sequence, and the sgRNA “Kat2KOG4” used to generate the Katnal2Δ17 mice. (C) Sanger sequencing chromatograms of WT and Katnal2Δ17 KO locus confirm 17 bp deletion in exon 2, predicted to knock out all long isoforms.

Fig. 2.

Enlargement of the ventricular area in Katnal2Δ17 mutant mice. Nissl-stained sections from average wild-type (A), heterozygous (B), and Katnal2Δ17 homozygous (C) mice. We also observed some Katnal2Δ17 homozygous mice with more severe hydrocephalus (D and E). Quantitation of ventricular area in homozygous Katnal2Δ17 mice (n = 11) greatly exceeded that in Katnal2 WT (n = 10) and heterozygous mice (n = 10) (F). (G) Representative in vivo T2 MRIs in coronal, sagittal, and axial planes from a wild-type Katnal2^+/+ mouse and mutant Kantal2^Δ17/Δ17 littermate at 5 mo of age confirm communicating ventriculomegaly. (H) Representative dorsal and ventral views of 3D reconstructions from MRIs as in (G) further illustrate the degree of ventriculomegaly. Ant = anterior; Post = posterior. (I) Ventricular volumes (mm³) of Katnal2 WT (gray, n = 7) and Kantal2Δ17 mice (red, n = 8) at 4 and 5 mo of age were quantitated from MRI data. Means ± SEM; **P < 0.01 by the two-tailed t test. Means ± SEM; ****P < 0.0001 by one-way ANOVA.

Fig. 3.

Katnal2 expression in the developing brain. (A) UMAP clustering of mouse fetal brain cells (25), colored by cell type. (B) KATNAL2 expression (orange) in the cell clusters shown in (A). (C) Gene expression heat map showing KATNAL2 expression in the cell types shown in (A). (D) Multiplexed fluorescent RNAScope of Katnal2 on sagittal sections of E16.5 mouse brain. DAPI counterstain allows anatomical identification of lateral ventricle (LV), ventricular zone (VZ), subventricular zone (SVZ), intermediate zone (not labeled), and cortical plate (CP). A montage of 20× images for Ki67 (red) highlights proliferating VZ cells extending into the SVZ. Map2 mRNA (blue) marks more differentiated neurons of the SVZ and CP. Katnal2 mRNA (green) is most highly expressed in VZ and retains expression in both SVZ and CP. (Scale bar, 200 µm.)

Fig. 4.

Altered primary cilia, planar cell polarity, and defective CSF flow in Katnal2Δ17 mice. (A and B) 3D projection of P0 anteroventral (AV) region stained for ependymal primary cilia using anti-Arl13b (green) outlined by β-catenin (red). (C) Whole-mount images from the anterodorsal (AD) area of the lateral ventricle of 4-mo-old mice show basal body clusters expressing γ-tubulin (red) within the apical domains of ependymal cells outlined by β-catenin (green) in wild-type Katnal2^+/+ (Left) and mutant Kantal2^Δ17/Δ17 littermates (Right). The cluster orientation of cilia within the ependymal cells is indicated by white arrowheads. (D) Basal body cluster angles were plotted on circular maps for both the AD AV areas. Cluster angle distributions differed between Katnal2^+/+ (n = 4) and Kantal2^Δ17/Δ17 (n = 4) mice. (E) Schematic of OCT imaging of microbeads placed onto the AD and AV areas of the lateral wall of lateral ventricle (LV) explants. (F) Color-time lapse images from OCT recordings of representative wild-type Katnal2^+/+ (Left) and mutant Katnal2Δ17 littermates (Right) aged 4 mo. A temporal color code depicts particle trajectories over time. (G) Quantitation of OCT recordings that track flow from individual microbeads reveals decreased velocity of CSF flow in the AV area of mutant Kantal2^Δ17/Δ17 LV explants (n = 4) relative to Katnal2^+/+ controls (n = 3). (H) The AD area of LV explants did not display a change in flow speed. (I and J) Reversible ciliobrevin inhibition of CSF flow confirms requirement of motile cilia for CSF flow in both WT and mutant mice. Means ± SEM. ***P < 0.001; **P < 0.01 by the two-tailed t test. (n =4). Means ± SEM; ***P < 0.001 by the two-tailed t test. (Scale bar, 10 μm.)

Fig. 5.

Katnal2 loss leads to decreased frequency of mEPSCs. Whole-cell voltage clamp recording was performed on layer 2/3 pyramidal neurons. (A) Representative traces of mEPSCs recorded from P12 acute cortical slices indicate a higher frequency of events in a wild-type neuron than in Katnal2Δ17 knockout neurons (red). (B) Averaged mEPSC waveforms from all captured wild-type (black) and Katnal2Δ17 neurons (red) displayed nonsignificant increase in amplitude. (C) Peak-scaled events from control (black) and KO (red) cells indicate no change in mEPSC kinetics. (D) Quantitation of data reveals that Katnal2 knockout decreases mEPSC frequency with no alteration in mEPSC amplitude or mIPCS frequency or amplitude. (E) A mixture of rAAV2-retro-GFP (green) and rAAV8-mCherry-Chrimson (Red) was injected unilaterally into the medial prefrontal cortex of P0 mice. One week after injection, we found both mCherry-chrimson and GFP-positive cells (yellow) in the ipsilateral hemisphere and only GFP-positive callosal projection neurons in the contralateral hemisphere. (F) Higher magnification images of an acute brain slice demonstrated specific labeling of layer 5 pyramidal neurons in the contralateral prelimbic cortex. (G) Voltage responses to a train of current injections (Bottom) recorded in a layer 5 callosal projection neuron in the PFC revealed increased amplitude of the slow AHP in Katnal2Δ17 knockout (red) vs. wild type (black). The Inset shows individual cell data. (H) 30 superimposed action potentials in response to short (20 ms) high amplitude (270 pA) current injections. Action potential half-width was reduced in knockout neurons (red; the Inset shows individual cell data). Data are means ± SEM. *P < 0.05, ***P < 0.001, by the two-tailed t test.

Fig. 6.

Deleterious, heterozygous KATNAL2 variants are present in cases of neurosurgically shunted spontaneous human congenital hydrocephalus. (A) Flowchart depicting expansion of the clinical cohort used for whole exome sequencing in this study. (B) Rare damaging variants in KATNAL2 and KATNB1. (C) Representative brain MRI images of individuals KCHYD246-1 and KCHYD115-1 from our CH cohort who harbor heterozygous KATNAL2 pathogenic variants in the setting of comorbid NDDs (including autism spectrum disorder). (D) Evolutionary conservation of heterozygous variants in KATNAL2 and KATNB1 identifies several recurrent variants which may mark clinically significant residues. (E) In silico biophysical modeling showed that the D-mis variants in KATNAL2 (p.K233N, p.D349Y, p.T435P, and p.D440N) and KATNB1 (p.T189M) are predicted to impair protein function.

Fig. 7.

Novel Katnal2 point pathogenic variant mouse recapitulates ventriculomegaly. (A) Mapping of all deleterious heterozygous variants in KATNAL2 from the current cohort (orange) and SAFARI database (blue). (B) Representative Nissl stain of coronal brain sections from a wild-type and Katnal2 F244L point mutant mouse with quantitation of ventricular areas reveal Katnal2 F244L mutants have ventriculomegaly.