Brain ventriculomegaly in Down syndrome mice is caused by Pcp4 dose-dependent cilia dysfunction

Abstract

Down syndrome is a leading cause of congenital intellectual disability caused by an additional copy of the chromosome 21. Patients display physiological and morphological changes affecting the brain and its function. Previously we showed that Ts1Cje and Ts2Cje, Down syndrome mouse models carrying overlapping trisomic segments of different length, show similar ventriculomegaly and neurogenesis dysfunction leading to the hypothesis of a cause-consequence relationship between these phenotypes. However, we here discovered that Ts1Rhr Down syndrome model, carrying an even shorter trisomic segment, was sufficient to trigger ventricular enlargement and ependymal cilia beating deficiency without affecting neurogenesis. We further found that Pcp4 gene on the Ts1Rhr trisomic segment is expressed in ependymal cells, and its resumption to two copies rescued both ventricular enlargement and cilia dysfunction in Ts1Rhr mice. This work underlines a Pcp4-dependent ciliopathy in Down syndrome brain affecting cerebrospinal fluid flow.

Introduction

Down syndrome, caused by trisomy of human chromosome 21 (HSA21) occurs with an incidence of one in 700 to 1000 live births (1). Symptoms and phenotypes vary in frequency and severity in the Down syndrome population (2). They include intellectual disability, craniofacial and brain dysmorphology, congenital heart defects, megakaryocytic leukemia and early onset Alzheimer’s disease.

Brain morphology is affected both in children (3) and adults (4) with Down syndrome with an overall decreased size of specific brain regions such as prefrontal cortex or hippocampus. Additionally, cerebrospinal fluid (CSF) and brain ventricle volumes are increased in persons with Down syndrome who developed dementia (5,6) as well as in a significant part of the Down syndrome population without dementia (7,8). CSF volume increase suggests abnormal homeostasis that could be directly involved in developmental and functional deficits observed in Down syndrome human brain, such as decreased neurogenesis (9).

The low accessibility to brain structures is a barrier to further investigate the origin of these phenotypes in human. To circumvent this limitation, mouse models for Down syndrome are now widely used. The most common models were generated by targeting mouse chromosome 16 (MMU16) in its region of synteny to HSA21. Ts65Dn, Ts1Cje and Ts1Rhr are major Down syndrome models and have been studied intensively over the past decade (10). Interestingly Ts65Dn harboring a large trisomy (∼86 known protein coding genes from miR155 to Znf295), Ts1Cje carrying a shorter but overlapping trisomy (∼66 known protein coding genes from Sfrs15 to Znf295) and the even shorter Ts1Rhr trisomy but still overlapping with the two previous models (∼33 known protein coding genes from Cbr3 to Fam3b) all show Down syndrome-like learning and memory deficits associated with electrophysiological abnormalities (11–15).

Using these animal models, we previously described the co-occurrence of ventriculomegaly and decreased neurogenesis in Ts1Cje and Ts2Cje mice (equivalent to Ts65Dn) (16). This led to the hypothesis that cause-consequences relationships could exist due to the central role played by the CSF in brain homeostasis and that candidate genes are located within the ∼66 genes triplicated in Ts1Cje model. Here, using a comparative study, we challenged this hypothesis and used a gene copy number resumption approach to investigate genes and mechanism leading to ventriculomegaly in Down syndrome mice.

Results

Ts1Rhr fragment is sufficient to trigger ventriculomegaly but not neurogenesis deficit

In order to narrow down the genetic region of interest, we here conducted a comparative study between Ts1Cje and the shorter (∼33 genes) Ts1Rhr trisomy (Fig. 1D) (12,14).

Figure 1

Enlarged brain ventricles in Ts1Cje and Ts1Rhr mice. (A) Raw longitudinal slices and reconstructed 3D images showed enlarged lateral ventricles (LV) in Ts1Cje (Cje) and Ts1Rhr (Rhr) compared to their WT littermates. Dorsal (D3V) and ventral (V3V) third ventricles as well as forth ventricle (4V) also appeared slightly enlarged in both Down syndrome models but did not reach statistical significance. Calculation of normalized volumes (defined as the ratio ventricular volume/brain volume) (B) confirmed significant enlargement of LV from 1.43 ± 0.14% in WT to 2.27 ± 0.28% in Ts1Cje (P = 0.028, N = 5) and from 1.48 ± 0.09% in WT to 2.42 ± 0.30% in Ts1Rhr (P = 0.008, N = 9). D3V, V3V and 4V were not significantly different from WT littermates in either Ts1Cje or Ts1Rhr. Whole brain volume (C) was significantly increased from 477.7 ± 4.2mm³ in WT to 519.8 ± 7.0 mm³ in Ts1Rhr but was not significantly different from WT in Ts1Cje samples. Values for normalized volumes and whole brain volumes are expressed as mean ± standard error of the mean. Statistical significance was assessed using two-tailed unpaired t-test with significance set (*) at P < 0.05 and (**) at P < 0.01. (D) Ts1Rhr region is sufficient to trigger ventriculomegaly in mouse but not sufficient to induce neurogenesis deficit arguing against a cause-consequence relationship between these phenotypes. Pcp4, located within Ts1Rhr region is a relevant candidate for ventriculomegaly.

We first investigated whether Ts1Rhr trisomy was sufficient to trigger brain morphological abnormalities. MRI imaging showed that lateral brain ventricle volumes were significantly larger in both Ts1Rhr and Ts1cje mice than in their respective WT littermates at 3 months old (Fig. 1A and B, Supplementary Material, Videos S1–S4). The extent of this enlargement was comparable in both models: +58.5% in Ts1Cje and +63.8% in Ts1Rhr (P = 0.028 and P = 0.008 respectively; Fig. 1B). Other (third and fourth) ventricles in Ts1Cje and Ts1Rhr showed trends of enlargement but this did not reach statistical significance. Whole brain volume was significantly larger in Ts1Rhr mice than in their WT littermates (P = 9.34E-5) but this was not the case in Ts1Cje (P = 0.698) (Fig. 1C).

We then investigated neurogenesis in the adult brain using a 5-bromo-2’-deoxyuridine (BrdU) integration experiment within the main neurogenic regions: the subventricular zone (SVZ, Supplementary Material, Fig. S1A) and the dentate gyrus (DG, Supplementary Material, Fig. S1B). Neurogenesis was significantly decreased in Ts1Cje brains (39.1% and 42.7% drop in SVZ and DG respectively; Student’s t-test SVZ: P = 0.037; DG: P = 0.025) whereas no significant differences were observed between Ts1Rhr and their WT littermates (Supplementary Material, Fig. S1C and D).

Cerebral aqueduct is not significantly affected in Down syndrome mice

Stenosis/closure of connections between the different brain ventricles is one of the main causes for ventriculomegaly (17). The foramen of Monro (connecting the lateral ventricles to the third ventricle) was clearly identifiable in every Ts1Rhr sample using MRI images (Fig. 1A, Supplementary Material, Videos S1–S4). As the aqueduct of Sylvius (connecting the third to the fourth ventricle) was difficult to identify by MRI, we investigated its morphology using standard histology techniques (Fig. 2A) and morphometric measurements (Supplementary Material, Table S1). Ventricular area was increased in the Ts1Rhr samples compared to their WT counterparts (P = 0.005 and P = 0.008 at bregma and bregma -1.8 mm coordinates respectively). The aqueduct of Sylvius was clearly identifiable in all samples and appeared similar in Ts1Rhr and WT brains, even at the bregma -3.8mm coordinate where its diameter is the thinnest (2450 ± 896 µm² for WT vs 2030 ± 246 µm² for Ts1Rhr, P = 0.194). In order to further confirm this observation, we investigated the Dp(16)1Yey model carrying a duplication of the complete region of synteny on MMU16 (18), including the Ts1Rhr fragment (Fig. 1D). In our breeding colony 35.4% of the trisomic mice developed severe hydrocephalus and died by 6 weeks of age (Fig. 2B–D, Supplementary Material, Table S2). Even in samples affected by severe hydrocephalus we did not observe a significant stenosis of the aqueduct of Sylvius (Fig. 2D).

Figure 2

Aqueduct of Sylvius is not closed in Ts1Rhr nor Dp(16)1Yey mice. (A) MRI imaging and hematoxylin/eosin stained serial sections at 4 different rostro-caudal coordinates confirmed enlargement of the lateral ventricles. The aqueduct of Sylvius in Ts1Rhr samples appeared highly similar to their WT littermates, even at coordinates where its diameter is at thinnest (bregma -3.8mm and enlargement box). (B–D) Dp(16)1Yey (Dp16) model mice showed a somewhat stronger phenotype. A significant part (18 out of 48) of Dp16 offspring showed abnormal morphology at 4 weeks of age with small body size associated with dome-shaped skull (D). Dissection revealed drastically enlarged cerebrum leading to aberrant positioning of the cerebellum. Serial sectioning confirmed a severe hydrocephalus associated with a clear decrease in cortical thickness even though no closure of the aqueduct of Sylvius was observed. Bleeding was observed in the cortico-cerebellar space as well as on the ventro-lateral part of the mid-brain in the animal presenting severe hydrocephalus. This phenotype was associated with a short lifespan (4–6 weeks). (B) No such malformation was seen in any WT littermates. (C) The rest of the Dp16 population displayed milder ventriculomegaly. Scale bars correspond to: (A) 300 µm in whole section pictures and 100 µm in aqueduct enlargement boxes and (B–D) 1 cm (mouse and brain pictures), 150 µm (cerebral aqueduct) and 2 mm (whole section).

Measurements of cortical thickness and hippocampal area suggested that in Ts1Rhr mice these structures were also enlarged in particular in the medial part of the brain (Supplementary Material, Table S1). The larger whole brain volume is thus not only due to an increase in ventricular volume but also to an overall increase in volume of various brain structures. These observations would however require more optimized volumetric quantifications to be confirmed.

Pcp4 is a candidate for CSF generation and/or flow dysfunction

Abnormalities in CSF production or flow can result in enlarged ventricles (19–21). We assumed that Pcp4 (Purkinje cell protein 4, also known as PEP-19) would be an interesting candidate due to its localization within the Ts1Rhr region (Fig. 1D) and because its expression has been shown in the ventricular zone of the embryonic diencephalon at E10.5 (22). Whether Pcp4 is expressed in structures related to CSF production or flow at post-natal stages and in the adult brain however remained unknown. Anti-PCP4 immunohistochemistry in adult WT brain sections showed dense expression in the cerebellar Purkinje cells and in the CA2 pyramidal cells (Fig. 3A). In addition, we identified for the first time to our knowledge a strong expression of PCP4 in the choroid plexus and the ependymal cells from the ventricular wall. In order to confirm these observations and to investigate the potential involvement of Pcp4 in CSF flow we generated a knockout mouse line (Supplementary Material, Fig. 2A–C). Pcp4^-/- homozygous KO mice survived to adulthood and did not show obvious morphological abnormalities. In adult Pcp4^-/- brain no protein was detected by immuno-labeling, confirming both the loss of functional PCP4 protein and the specificity of the signal detected in the WT section (Fig. 3A, Supplementary Material, Fig. 3). The low signal observed in the cerebellar granular cell layer and the olfactory bulbs in the Pcp4^-/- sample was confirmed as non-specific using negative controls (Supplementary Material, Fig. 4). MRI imaging of Pcp4-null mouse brain showed that their lateral ventricles were significantly smaller in volume compared to their WT littermates (P = 0.013)(Fig. 3B and C).

Figure 3

PCP4 is expressed in choroid plexus and ventricular wall ependymal cells; its loss leads to decreased ventricular volume. (A) Anti-PCP4 immunohistochemistry show strong and specific expression in cerebellar Purkinje cells and CA2 pyramidal cells. Dense labeling is also observed in the peripheral layer of the choroid plexus, corresponding to epithelial cells, as well as in the ependymal cell layer from the ventricular walls. Intensity of the signal was comparable in all these structures and totally absent in Pcp4^-/- samples. (B) Raw longitudinal MRI sections and reconstructed 3D images showed mildly smaller lateral ventricles (LV) in Pcp4^-/- samples compared to their WT littermates. (C) Calculation of normalized volumes (defined as the ratio ventricular volume/brain volume) confirmed a significant decrease (P = 0.013). Dorsal (D3V) and ventral (V3V) third ventricles as well as fourth ventricle (4V) were not significantly different from WT. (D) The whole brain volume was also not significantly different between WT and Pcp4^-/- samples (P = 0.529). Data were acquired from 5 adult mice (3 months old) per genotype. Values are expressed as mean ± standard error of the mean. Statistical significance was assessed using two-tailed unpaired t-test with significance (*) set at P < 0.05. Scale bars in (A) correspond to 200 µm (cerebellum); 100 µm (hippocampal CA2); and 25 µm (choroid plexus and ependymal cells).

Pcp4 resumption rescues ventriculomegaly in Ts1Rhr

We next assessed whether Pcp4 was playing a role in Ts1Rhr ventriculomegaly. For that purpose, we designed a subtractive approach by crossing Pcp4^+/- and Ts1Rhr mouse lines to genetically normalize Pcp4 copy number and its expression level (Supplementary Material, Fig. 2D). MRI imaging confirmed a significant ventriculomegaly in Ts1Rhr samples (Fig. 4A and B). Interestingly lateral ventricles were significantly smaller in Ts1Rhr;Pcp4^+/+/- than in Ts1Rhr samples (P = 0.023) and rather close to that of WT. Significant differences were also observed in dorsal third ventricle volume (larger in Ts1Rhr compared to Pcp4^+/-, P = 0.011) and ventral third ventricle volume (larger in Ts1Rhr than in WT and Pcp4^+/-, P = 0.035 and P = 0.028 respectively). Pcp4 resumption however did not rescue Ts1Rhr whole brain enlargement (Fig. 4C). Pcp4 deficiency led to a decrease in brain ventricles volume in Pcp4-null mice, a phenotype opposite to that seen in the Ts1Rhr model (Fig. 3B and C).

Figure 4

Resumption of Pcp4 gene rescues ventricular enlargement and cilia beating dysfunction in Ts1Rhr. (A) Raw longitudinal MRI sections and reconstructed 3D images confirmed enlarged lateral ventricles (LV) in Ts1Rhr whereas both Ts1Rhr:Pcp4^+/+/- and Pcp4^+/- samples appeared closer to WT (N = 13 per group). (B) Calculation of normalized volumes (ratio ventricular volume/brain volume) confirmed significantly enlarged LV in Ts1Rhr (from 1.59 ± 0.09% in WT to 3.45 ± 0.56% in Ts1Rhr, P = 2.9E-4). Ts1Rhr:Pcp4^+/+/- and Pcp4^+/- were not significantly different from WT but differed significantly from Ts1Rhr (2.21 ± 0.09%; P = 0.023 and 1.73 ± 0.10%; P = 8.4E-4, respectively). The dorsal third ventricle (D3V) and ventral third ventricle (V3V) were mildly enlarged in Ts1Rhr (D3V: p_{Rhr_vs_Pcp4+/-}=0.011; V3V: p_{Rhr_vs_WT}=0.035, p_{Rhr_vs_Pcp+/-}=0.028). (C) Whole brain volume enlargement in Ts1Rhr was not rescued in Ts1Rhr:Pcp4^+/+/- mice (WT: 486.3 ± 4.8 mm³; Ts1Rhr: 529.8 ± 7.6 mm³; Ts1Rhr:Pcp4^+/+/-: 525.5 ± 4.7 mm³ respectively). Cilia beating frequency (CBF) (D) and angle (CBA) (E) were significantly decreased in Ts1Rhr brains and Pcp4 resumption reverted this deficit leading to both higher CBF and CBA in Ts1Rhr:Pcp4^+/+/- samples (CBF: n > 300 cells per animal from N = 5 per group, CBA: n > 80 cells per animal from N = 4 per group). Plotting the normalized lateral ventricle volume as a function of CBF (F) revealed a clear group separation with Ts1Rhr:Pcp4^+/+/- and WT clustering close to each other and Ts1Rhr being further apart. Values are expressed as mean ± standard error of the mean. Statistical significance was assessed using One-way ANOVA and Tukey’s post-hoc test with significance (*) set at P < 0.05 and (**) at P < 0.01.

Pcp4 dose-dependent cilia dysfunction underlies ventriculomegaly in Down syndrome mice

PCP4 regulates calmodulin and intracellular calcium concentration (23,24) and motile cilia beating in ependymal cells is highly dependent on calcium-regulated mechanisms, either through calmodulin (25) or cAMP (26). It was thus of interest to assess cilia function in our model mice. For that purpose, we acquired MRI images and investigated the functional characteristics of motile cilia in a new set of mice (Fig. 4D–F, Supplementary Material, Videos S5–S7). Both cilia beating frequency (Fig. 4D) and cilia beating angle (Fig. 4E) were significantly decreased in Ts1Rhr mice (p_{Rhr_vs_WT}=5.4E-5 and p_{Rhr_vs_WT}=0.012 respectively, N = 5 per genotype). Pcp4 resumption to two copies was able to significantly revert this decrease by setting back both beating frequency and angle to a level slightly, yet significantly higher than in WT (p_{Rhr;Pcp4+/+/-_vs_WT}=0.021 and p_{Rhr;Pcp4+/+/-_vs_WT}=0.017 respectively, N = 5 per genotype). Plotting the volume of the lateral ventricles as a function of the cilia beating frequency showed a clear separation by a group of genotype, Ts1Rhr;Pcp4^+/+/- clustering close to the WT littermates whereas Ts1Rhr samples were more distant (Fig. 4F).

Discussion

To date, as no systematic ultrasonic or MRI screening in the Down syndrome population has been reported, it is difficult to estimate the exact prevalence of ventriculomegaly. Several targeted studies however reported phenotypes ranging from mild ventriculomegaly in adults and young adults with Down syndrome (7,8) to more severe hydrocephalus (27). Low birth-weight infants with Down syndrome were recently shown to be a high-risk population for ventriculomegaly (28). Genes and mechanisms responsible for this phenotype however remained unexplained.

Through a comparative study using model mice for Down syndrome, the present work allows us to refine regions of interest for morphological and functional phenotypes affecting the brain. First, we observed a consistent ventriculomegaly of equivalent severity in Ts2Cje, Ts1Cje and Ts1Rhr (16, present study). Interestingly, the phenotype was more severe in the larger Dp(16)1Yey trisomy. In this model one third of the mice showed severe hydrocephalus. Though in the few reports involving the Dp(16)1Yey available so far hydrocephalus has not been mentioned, in the equivalent model Dp1Tyb severe hydrocephalus is contributing to post-natal lethality (29 and comments available in the online version). So far, studies dealing with the Dp(16)1Yey model were conducted on 129Sv, C3H, C3Sn or mixed C57BL/6J backgrounds whereas the Dp1Tyb were maintained on C57BL/6J for more than 6 generations. Our Dp(16)1Yey colony was backcrossed for more than 10 generations on C57BL/6J background. Advanced backcrossing and subsequent allelic composition is thus likely affecting the outcome in these mice. Interestingly, a significant number of mutations leading to primary cilia dyskinesia in mouse lead to hydrocephalus or ventriculomegaly on C57BL/6J but not on 129Sv or mixed C57BL/6J;129Sv backgrounds (30). In a similar way to these knockout models, hydrocephalus affecting Dp(16)1Yey mice is thus likely sensitive to genetic composition. Important genetic modifiers in the C57BL/6J background are certainly involved in the susceptibility of Dp(16)1Yey model to develop hydrocephalus. Taken together, these results indicate that the gene(s) responsible for ventriculomegaly are located within the Ts1Rhr region (from Cbr3 to Fam3b). These observations also suggest that the genetic fragment specific to Dp(16)1Yey (from Rbm11 to Ncam2) contains additional aggravating factors.

Furthermore, in the present study, we provide evidence that Ts1Rhr fragment is sufficient to trigger ventriculomegaly but not sufficient to induce abnormalities in neurogenesis. These results allow us to reject our previous hypothesis of a cause-consequence relationship between neurogenesis deficit and ventricular enlargement (16). The serine/threonine kinase Dyrk1a, located in the Ts1Rhr region, was so far a strong candidate for deficits in neurogenesis through a deregulation of the Sox2 pathway (31,32). In the present study, we however show that despite carrying an additional copy of Dyrk1a, Ts1Rhr mice do not show significant changes in adult neurogenesis. Neurogenesis deficit is thus induced by, or at least necessitates genes located in the Ts1Cje-specific genetic fragment (i.e. outside the Cbr3 to Fam3b interval).

Additionally, we report here for the first time to our knowledge a new non-neuronal expression and function for Pcp4. Reports have so far focused on its role in neuronal physiology (33) and its potential involvement in cerebellar impairment in Down syndrome (34). More importantly, the present study defines cilia beating deficit as a leading cause for ventriculomegaly in Down syndrome and Pcp4 as a key player in that deficient process. Among the main causes of ventriculomegaly and/or severe hydrocephalus, the role played by cilia dysfunction has been widely documented in human as well as in model mice (30). In model mice with mutations affecting cilia beating, the ciliopathy is sometimes associated with cerebral aqueduct stenosis (19) but can also lead to ventriculomegaly without significantly affecting the aqueduct (35). In a similar way the cerebral aqueduct was normal in our Ts1Rhr mice, but we observed Pcp4 dose-dependent ventriculomegaly and cilia beating deficiency suggesting a cause-consequence relationship. PCP4 has the ability to trigger calcium extrusion (24). Increase and decrease in [${\text{Ca}}^{2\text{+}}$]_I are known to lead respectively to increase and decrease in cilia beating frequency (36). In the case of Down syndrome, excessive ${\text{Ca}}^{2\text{+\hspace{0.17em}}}$extrusion/buffering by Pcp4 overexpression could lead to a drop in [${\text{Ca}}^{2\text{+}}$]_i, and subsequent AMPc-dependent and calmodulin-dependent decrease in cilia beating frequency and cilia beating angle. Further investigation is required to clarify the exact molecular mechanism by which Pcp4 regulates motile cilia activity.

Materials and Methods

Mouse lines and genotyping

Ts(12;16)1Cje, referred to as Ts1Cje, was maintained by crossing carrier males with C57BL/6J females. Dp(16Cbr1-ORF9)1Rhr, referred to as Ts1Rhr, and Dp(16)1Yey were obtained from Jackson Laboratory (www.jax.org; date last accessed Januray 12, 2017, Stock numbers 005383 and 013530 respectively) and maintained the same way. Genotyping for Ts1Cje was performed by PCR amplification as previously described (37). Ts1Rhr and Dp(16)1Yey PCR genotyping were performed as described by the Jackson Laboratory. Genotype from Pcp4 knockout mice was determined using the following specific PCR primers: Pcp4-R1: GCTGCACTTAGGCACAAATC; Pcp4-KO-F2: AGCAACAG GTTTCCTTGTGG; Pcp4-WT-F2: GAAT GCCT CTCATTGGTTGG amplifying PCR products of 596 bp for the WT allele and 423 bp for the mutant allele (Supplementary Material, Fig. S2).

Generation of Pcp4 knockout mice

Pcp4 heterozygous knockout mice (Pcp4^+/-) were obtained by using in-vivo Cre/loxP recombination (Supplementary Material, Fig. S2A). Briefly, we inserted loxP sites upstream and downstream from Pcp4 exon one. Heterozygous flox-neo females were mated to Tg(EIIa-cre)C5379Lmgd males (38) resulting in the deletion of both Pcp4 exon one and neomycin resistance gene. Amplification and sequencing of a 423bp fragment overlapping Pcp4 exon one confirmed its deletion (Supplementary Material, Fig. S2B and C).

Magnetic resonance imaging - MRI

MRI recordings were performed using a vertical-bore 9.4 T Bruker AVANCE 400WB imaging spectrometer (Bruker Biospin). In vivo scanning from 3 to 4.5 months old adult mice was performed under 1.5-2% isoflurane anesthesia. Images were acquired using a two-dimensional multislice spin echo sequence using the following parameters: field of vision = 1.60 cm, acquisition matrix 133 x 133, slice thickness = 0.12 mm, time of repetition/time of echo = 12,313.9/60.0 ms. Sixty longitudinal sections were acquired in a stack covering the entire brain. Raw image stacks were then processed using NIH ImageJ (http://rsb.info.nih.gov/ij/; date last accessed Januray 12, 2017) and Photoshop Elements 10. Representative images were selected as the closest sample to the group’s average. We used the ratio between the ventricle and whole brain volumes for normalization. A Ts1Cje/Ts1Rhr comparison experiment was conducted on 3 months old mice (Ts1Cje: N = 5 per group; Ts1Rhr: N = 9 per group). Pcp4 resumption was assessed using thirteen animals per group of genotype (aged 3 to 4.5 months). Pcp4 homozygous knockout MRI analysis was conducted on 4 months old Pcp4^-/- mice and WT littermates (N = 5 per group).

In-vivo BrdU labeling and immuno-detection

Three months old adult mice were administered daily 5-bromo-2’-deoxyuridine (BrdU; 100 mg/kg body weight in PBS, Sigma-Aldrich) intraperitoneal injections for eight consecutive days. Twenty-four hours after the last BrdU injection brains were collected. Equivalent frozen sections (30 µm thickness) from Ts1Cje, Ts1Rhr and their WT littermates were selected according to common landmarks (39). Slides were incubated in Retrievagen (#550524, BD Biosciences) and blocked using "Mouse-on-mouse" (MKB-2213, Vector Laboratories) and Avidin blocking (SP-2001, Vector Laboratories). Sections were then incubated with mouse anti-BrdU antibody (1:200, #555627, BD Biosciences) in Biotin blocking solution (SP-2001, Vector Laboratories). The signal was revealed using biotinylated anti-mouse IgG antibody (1:200, BA-2000, Vector Laboratories) followed by "VECTASTAIN Elite ABC kit for HRP revelation" (PK-6100, Vector Laboratories) and metal-enhanced chromogen diaminobenzidine tetrachloride (PI34065, Pierce). Series of Z-stacked images (twelve pictures, spaced 1 μm) were acquired using a BZ-8000 light microscope (Keyence). The images were then processed using NIH ImageJ software for particle counting. A total of twelve dentate gyrus (DG) and six sub-ventricular zone (SVZ) images were counted for each animal in a blinded manner (from a total of N = 3 animals for Ts1Cje and WT littermates and N = 4 animals for Ts1Rhr and WT littermates).

Histological analysis and immunohistochemistry

Brains from adult mice (3.5–4 months old) were collected, fixed in 4% PFA and paraffin embedded. For cerebral aqueduct detection, 6 µm thickness coronal sections were prepared from three WT samples with standard-sized ventricles and three Ts1Rhr samples with enlarged-ventricles (detected using MRI). Sections were deparaffinized, rehydrated and hematoxylin-eosin stained. For PCP4 immunohistochemistry, 6 µm thickness sagittal sections were prepared from WT and Pcp4^-/- samples (N = 2 per genotype). Sections were deparaffinized, rehydrated, incubated in RetrievagenA (#550524, BD Biosciences) and blocked using Avidin/Biotin blocking kit (SP-2001, Vector Laboratories). Sections were then incubated with anti-PCP4 rabbit polyclonal IgG (1:200, Santa-Cruz sc-74816, Santa-Cruz) and revealed using biotinylated goat anti-rabbit IgG antibody (1:250, Vector BA-1000, Vector Laboratories) followed by "VECTASTAIN Elite ABC" (PK-6100, Vector Laboratories) and NovaRed substrate kits (SK-4800, Vector Laboratories). Images were acquired using a BZ-8000 light microscope (Keyence). Morphometric measurements were obtained from 2 sections for each brain coordinate and each animal (from three Ts1Rhr mice and three WT littermates). Whenever appropriate, both right and left sides were measured in each section (e.g. lateral ventricles at bregma +0.0mm, or hippocampus at bregma −1.8 and −2.2 mm). For Dp(16)1Yey morphological and histological study four random animals, aged 4 weeks old were taken from the same litter. Samples were processed for hematoxylin/eosin staining as described above. Pictures were acquired prior to PFA perfusion for gross head morphology, post-PFA fixation (for whole brain gross morphology). Genotype for these four samples was checked at the end of the experimental process.

Cilia beating frequency and angle measurements

Prior to cilia beating investigation, MRI was recorded from adult mice (9 ± 1 months old, N = 5 per group of genotype) in order to calculate ventricular volume. Brains were then collected and surface from lateral ventricles were dissected in perfusion solution (⁠${\text{HCO}}_3^{-}$ solution: 121 mM NaCl, 4.5 mM KCl, 1 mM MgCl₂, 1.5 mM NaHCO₃, 1.5 mM CaCl₂, 1.5 mM NaHEPES, 5 mM HEPES and 5mM glucose) and mounted on Cell-Tak coated cover slips (Becton-Dickinson Labware). Coverslips were transferred to a 37 °C heated perfusion chamber (volume: 20 μl, perfusion rate: 200 μl/min) mounted on a differential interference contrast microscope (E600-FN, Nikon) connected to a high-frequency acquisition camera (FASTCAM-512PCI, Photoron). Videos were acquired at a 500 Hz sampling rate. Data were then analysed using DippMotion 2D (Ditect). Cilia beating frequency and angle were measured as previously described (40).

Cilia beating frequency data were obtained from a total of 1666 cells from five WT mice, 1823 cells from five Ts1Rhr mice and 2066 cells from five Ts1Rhr;Pcp4^+/+/- mice. Cilia beating angle data were obtained from a total of 337 cells from four WT mice, 468 cells from four Ts1Rhr mice and 331 cells from four Ts1Rhr;Pcp4^+/+/- mice. Data were averaged for each mouse and submitted to One-way ANOVA followed by Tukey’s post-hoc test.

Gene expression analysis by qRT-PCR

Brains were collected from adult mice (3 months old mice, N = 5 animals per group of genotype). Total RNA was prepared from one hemisphere per sample and quality-checked using Agilent 2100 Bioanalyzer (Agilent Technologies). cDNA was obtained by reverse transcription using Superscript III First-Strand Synthesis SuperMix (Invitrogen). Pcp4 expression level was assessed using Taqman MGB-FAM Mm00500973_m1 probe kit and normalized using Gapdh MGB-VIC probe kit #4352339E (Applied Biosystems). Quantitative measurement of cDNA level was done in triplicates on Applied Bioscience ABI-7900HT RealTime qPCR thermocycler (Applied Biosystems) and analysed using manufacturer’s Sequence Detection System Ver2.4 software (Applied Biosystems).

Statistical analysis

Experiments comparing either Ts1Cje or Ts1Rhr to their respective WT littermates were submitted to two-tailed unpaired t-test using KyPlot v2.0 software (KyensLab Inc.). Multi-group data generated for Pcp4 copy number resumption assessment (Ts1Rhr:Pcp4^+/+/-) were submitted to One-way ANOVA, followed by Tukey’s pair-wise comparison post-hoc test using KyPlot v2.0 software. Statistical significance was set at P < 0.05. No animal was excluded from any study. Experiments were conducted in a blinded manner except for the histological analysis in Ts1Rhr in which we used animals with the largest ventricles to check the appearance of the cerebral aqueduct.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We are grateful to BSI Research Resource Center for the support to generate Pcp4 knockout mice, maintaining mice and for the help with gene expression analysis. We also thank the other members of the laboratory for neurogenetics in RIKEN BSI for helpful comments and advice.

Conflict of Interest statement. None declared.

Funding

This work was supported in part by a grant from RIKEN Brain Science Institute [to K.Y.] and Grant-in-Aid for Scientific Research [22390078 to K.Y., 22791015 and 24591535 to S.A., and 26860269 to M.R.].