A novel Agbl5 mutant mouse model exhibited domed head and reduced lifespan.

(A) Schematic representation of the knock-out/knock-in strategy to create the novel Agbl5 mutant (Agbl5M1) allele. A region between start codon and intron 8 was replaced with a tdTomato reporter cassette using CRISPR/CAS9 system. (B) RT-PCR using primers targeting a region spanning exon 6 and exon 8 of Agbl5 confirmed the absence of Agbl5 transcripts in brain, eye, spinal cord, spleen, and testis of Agbl5M1/M1 mice. NC, negative control. (C) The Kaplan-Meier survival curve showed that the Agbl5M1/M1 mice hardly survived more than 50 days (p<0.0005, log-rank (Mantel-Cox) test). (D-E’) Representative images of P50 wild-type (D, D’) and Agbl5M1/M1 (E, E’) mice. Different from the wild-type mice (D, D’), the mutant (E, E’) developed dome-shaped head as indicated with red lines (D’, E’).

Histological and flow assessment of the cerebrospinal fluid (CSF) pathway.

Hematoxylin-Eosin staining on a serial coronal brain sections of P46 wild-type (WT, A-C) and Agbl5M1/M1 (D-F) mice. Markedly enlarged lateral ventricles (LV, D, closed arrowhead), the third ventricles (D, open arrowheads, D’), and the fourth ventricle (F) were observed in Agbl5M1/M1 mice compared to the corresponding part in wild-type mice (A, B and C, arrowheads), while the size of the aqueduct was less affected in the mutant (B vs. E, arrowheads). A’ and D’ are the close view of ventral 3rd ventricles that were indicated in dashed boxes in A and D respectively. The dorsal and ventral 3rd ventricles are pointed with open arrowheads in red and purple respectively in A and D. (G-N) Flow assessment of CSF in mice of P30 intraventricularly injected with Evans Blue. Whole brains were fixed in 4% PFA 20 min later after injection to allow ink distribution and diffusion in the brain of wild-type (G-J) and mutant mice (K-N). Tissue sections showed that diffused ink was clearly observed in the lateral ventricles (LV, red arrowheads in G, K), the third dorsal (d3V, red arrowheads in H, L) and ventral ventricles (v3V, purple arrowheads in H, L), the fourth ventricles (4thV, red arrowheads in J, N) and aqueducts (Aq, red arrowheads in I, M) of both wild-type (G-J) and mutant (K-N) mice, indicative of communicating hydrocephalus. Scale bars: 2 mm for A-N; 200 μm for A’, D’.

Aberrant ependymal multicilia and basal body positioning in Agbl5M1/M1 hydrocephalic mice.

(A-B) Representative images of the whole-mount lateral walls of LVs from P45 wild-type (WT, A) or Agbl5M1/M1 (B) immunostained with the ciliary marker, acetylated tubulin (Act-Tub). While multicilia in WT ependyma evenly cover the ventricle surface and point to the same direction, mutlicilia bundles were scattered with many Act-Tub positive cilia lying on the cell surface. (C-F) Scanning electron microscopy analysis of LV walls from wild-type (C, E) and Agbl5M1/M1 (D, F) mice of P30. In the wild-type LV, ependymal cells were covered with evenly distributed cilia bundles in a uniformed direction (C), while in the mutant mice, cilia only in the middle of ependymal cell surface remain (D). (E, F) Higher magnification images showed that the length of remaining ependymal cilia in Agbl5M1/M1 mice (F) is similar to that in wild-type animal (E). (G) Sequential images of ciliary beating in wild-type (upper row) and Agbl5M1/M1 (lower row) mice. (H-L, N-R) Whole-mount lateral walls of LVs from P45 wild-type (H-L) or Agbl5M1/M1 (N-R) immunostained with the centriolar distal appendage marker CEP164 (I, O) with actin network labeled with phalloidin (H, K, N, Q). While BBs are clustered and polarized in the wild-type ependymal cells (I, J), those in the mutant (O, P) are often diffused. The actin networks are largely disrupted in the mutant ependymal cells (N, Q, vs. H, K for wild-type). (K-L’, Q-R’) Z-projection views of apical actin network around BB in wild-type (K, L) and Agbl5M1/M1 (Q, R) ependymal cells and respective orthogonal views (L’, R’). The mutant ependymal cells lack the compact actin networks even around clustered BBs (R, R’). (M, S) Quantification of ependymal cells with differently distributed BBs in wild-type (M) and Agbl5M1/M1 mice (S). Scale bar, A-D, H-J, N-P, 20 µm; E-F, K-L’, Q-R’, 2 µm; G, 5 µm.

Expression of genes promoting multiciliogenesis is not impaired in Agbl5M1/M1 ependyma.

(A-B’) Immunofluorescence analysis revealed that tdTomato signals can be detected in heterozygous Agbl5M1 (Agbl5WT/M1) brain (B, B’), but not in the wild-type control (A, A’). The tdTomato signals were localized in the ependymal cells but largely devoid from the subventricular zone (arrowhead). At the dorsal-lateral region of the LV, the tdTomato signals extend to 2-3 layers (arrow). (C, D) Lateral ventricles from P7 wild-type and Agbl5M1/M1mice were immunostained with Foxj1, a marker of multiciliation. (E) Quantification showed that the number of Foxj1-positive cells in individual LV walls of the mutant mice (n=5) is increased compared to that in the wild-type mice (n=5). (F-K) Representative images of the dorsal (F, I), lateral (G, J) and middle (H, K) walls of LV from P17 wild-type (F-H) or Agbl5M1/M1 (I-K) mice that were co-immunostained with S100β (red) and GT335 (green) with nuclei visualized with DAPI (blue). At this age, ependymal cells in all walls of LV are normally S100β-positive (F-H). In contrast, in Agbl5M1/M1 mice (I-K) many cells in the ependymal cell layer are not immunoreactive for S100β despite the presence of multicilia (arrows). (L) Quantitative analysis showed that the number of S100β-positive ependymal cells normalized to the length of ventricle walls is significantly reduced in the dorsal and the lateral walls in mutant mice. D, dorsal wall; L, lateral wall; M, middle wall. Error bars represent SEM. *, p< 0.05; **, p<0.01, ***, p<0.001; student’s t-test. Scale bars, A, B, 75 µm; A’, B’, 25 µm; C, D, 50 µm; F-K, 10 µm.

The glutamylation level is increased in ependymal multicilia of Agbl5M1/M1 mice.

(A) A schematic representation shows the enzymes involved in tubulin polyglutamylation and modifications recognized by GT335 and polyE antibodies respectively. (B) Immunoblotting of LV from mice of different ages showed that compared to the wild-type, the immunosignals of GT335 but not that of polyE are increased in Agbl5M1/M1 mice at all ages examined. (C-J) Lateral ventricles of P7 wild-type (C-F) and Agbl5M1/M1(G-J) mice stained with GT335 (green) and DAPI. Representative images show that the intensity and length of GT335 immunosignals in ependymal cilia are increased in all three (H, dorsal; I, lateral; J, middle) walls of LV in the mutant mice compared with the respective walls in wild-type LVs (D, dorsal; E, lateral; F, middle). While the number of multicilia tufts are comparable between the wild-type and mutant mice (K), the length of GT335 signals are increased in Agbl5M1/M1ependymal cilia (L). (M-T) LVs of wild-type (M-P) and Agbl5M1/M1 (Q-T) mice co-immunostained with acetylated-tubulin (Ac-Tub, green) and Arl13b (red) with nuclei visualized by DAPI staining. The intensity of acetylated tubulin immunosignals in ependymal cilia are reduced in all three walls of the LV mutant mice (R-T) compared to those of wild-type mice (N-P). The length of ciliary Ac-tub in the lateral wall is also reduced in the mutant (U). (V) Quantification showed that compared to that of the wild-type mice, the length of Arl13b signal in ependymal multicilia of Agbl5M1/M1 mice were not changed. Letters in blue: D, dorsal wall; L, lateral wall, M, middle wall. Error bars represent SEM, student’s t-test. Scale bars, C, G, M, Q, 100 µm; D-F, H-J, N-P, R-T, 10 µm.

The initially formed ependymal multicilia in Agbl5M1/M1 mice are motile.

(A-B) Images of SiR-tubulin labeled whole-mount LVs from P15 wild-type (A) and Agbl5M1/M1 (B) mice show that ependymal multicilia are initially formed in the mutant. (C) Sequential images of ciliary beating of P15 wild-type (upper row) and Agbl5M1/M1 (lower row) showed that the multicilia of wild-type ependymal cells beat in similar direction, while that of mutant are asynchronously. White and yellow open arrowheads indicate respective beating directions of multicilia of two cells; the closed arrowhead points to multicilia of an individual cell beat in opposite directions. (D) Bundled Agbl5M1/M1multicilia largely beat at the frequency similar to that of wild-type (n=30 for each animal). Error bars represent SD. (E) Histograms of beating angles for each animal, represented in polar coordinates. The area of each wedge is proportional to the percentage of angles in the corresponding angle range. (F, G) Whole-mount LVs from P15 wild-type (F) and Agbl5M1/M1 (G) mice were co-immunostained with Centrin (BB marker) and β-Catenin (cell boundary marker). (H-I) Traces of the intercellular junction labeled with β-Catenin of ependymal cell shown in F and G respectively. The purple arrows show the vectors drawn from the center of the apical surface to that of the BB patch. (J) Diagram showing the measurement of BB patch displacement. (K) Quantification showed that BB patches in Agbl5M1/M1 ependymal cells are not properly displaced (n= 198 for wild-type; n=253 for the mutant), p<0.001, student’s t test. (L) Histogram of the distribution of BB patch angles in ependymal cells of WT (blue) and Agbl5M1/M1 (orange), (n=138 for WT; n=119 for the mutant), p<0.001, Watson’s 2-sample U2 test. Scale bars, A-C, 5 µm; F, G, 20 µm.

Targeted disruption of CP domain alone in Agbl5 did not cause hydrocephalus, despite the increased glutamylation in ependymal cilia.

(A) Schematic representation of the knock-out/knock-in strategy to create a second Agbl5 mutant (Agbl5M2) allele that resembles the one used in previous studies (Wu et al., 2017). (B) RT-PCR using primers targeting deleted region in Agbl5M2 allele confirmed the absence of Agbl5 transcripts in brain, eye, and testis in Agbl5M2/M2mice. NC, negative control. (C-D) Similar to that in Agbl5WT/M1mice, tdTomato immunosignal is also detected in ependymal cells of P7 AgblM2heterogenous mice (D, arrows). (E-F) Hematoxylin-Eosin staining of coronal sections of brains from 3-month old wild-type (E) and Agbl5M2/M2 (F) mice, where no enlarged ventricles were observed. (G) Immunoblotting assay showed that compared with that of wild-type mice, the tubulin glutamylation level was increased in the brain of both Agbl5 mutants. (H-O) Immunostaining showed that the ciliary GT335 signals in both lateral (J) and middle (K) walls of the LV in Agbl5M2/M2 mice are increased compared with that in respective walls of the wild-type (H, I). (L-O) The ciliary acetylated-tubulin (Act-Tub) signals are reduced in both lateral (N) and middle (O) walls of the LV in Agbl5M2/M2 mice compared with respective walls of the wild-type (L, N). L, lateral wall; M, middle wall. Scale bars, C, D, 25 µm; E, F, 500 µm; H-O, 50 µm.