Disruption of the mouse Jhy gene causes abnormal ciliary microtubule patterning and juvenile hydrocephalus

SUMMARY

Congenital hydrocephalus, the accumulation of excess cerebrospinal fluid (CSF) in the ventricles of the brain, affects one of every 1,000 children born today, making it one of the most common human developmental disorders. Genetic causes of hydrocephalus are poorly understood in humans, but animal models suggest a broad genetic program underlying the regulation of CSF balance. In this study, the random integration of a transgene into the mouse genome led to the development of an early onset and rapidly progressive hydrocephalus. Juvenile hydrocephalus transgenic mice (Jhy^lacZ) inherit communicating hydrocephalus in an autosomal recessive fashion with dilation of the lateral ventricles observed as early as postnatal day 1.5. Ventricular dilation increases in severity over time, becoming fatal at 4-8 weeks of age. The ependymal cilia lining the lateral ventricles are morphologically abnormal and reduced in number in Jhy^lacZ/lacZ brains, and ultrastructural analysis revealed disorganization of the expected 9+2 microtubule pattern. Rather, the majority of Jhy^lacZ/lacZ cilia develop axonemes with 9+0 or 8+2 microtubule structures. Disruption of an unstudied gene, 4931429I11Rik (now named Jhy) appears to underlie the hydrocephalus of Jhy^lacZ/lacZ mice, and the Jhy transcript and protein are decreased in Jhy^lacZ/lacZ mice. Partial phenotypic rescue was achieved in Jhy^lacZ/lacZ mice by the introduction of a bacterial artificial chromosome (BAC) carrying 60-70% of the JHY protein coding sequence. Jhy is evolutionarily conserved from humans to basal vertebrates, but the predicted JHY protein lacks identifiable functional domains. Ongoing studies are directed at uncovering the physiological function of JHY and its role in CSF homeostasis.

INTRODUCTION

The cerebral spinal fluid (CSF) within the ventricles of the brain functions to cushion, nourish and cleanse the brain tissues, and acts as a carrier for key growth factors and signaling molecules (Davson and Segal, 1996; Rodriguez, 1976; Wood, 1983). CSF is continually produced by the choroid plexus tissue within the ventricles, and flows from the lateral and third ventricles through the narrow aqueduct of Sylvius into the fourth ventricle (Welch, 1963; Worthington and Cathcart, 1966; Yamadori and Nara, 1979). CSF flow is accomplished by the movement of cilia on the surface of the ependymal cells lining the ventricles, and likely also by pulsatile changes in intracranial blood flow (Bhadelia et al., 1997; Greitz, 1993; Quencer et al., 1990). From the fourth ventricle, the fluid exits the brain, moving into the spinal canal and the subarachnoid space. CSF production is balanced by its removal; once thought to occur only through the arachnoid villi into the bloodstream, it is now believed that both the lymphatic system and the ventricular ependymal cells play a role in CSF uptake (Koh et al., 2006; Oi and Di Rocco, 2006; Weller et al., 1992).

An increase in the amount of CSF within the closed space of the brain results in the disease known as hydrocephalus. Congenital hydrocephalus is one of the most common human developmental disorders, and may present nonsyndromically or in conjunction with a number of developmental malformations. Population-based studies estimate an incidence of 0.4 to 0.8 in 1000 children born with hydrocephalus in the absence of any neural tube defect (Blackburn and Fineman, 1994; Christensen et al., 2003; Schrander-Stumpel and Fryns, 1998; Stoll et al., 1992). Physiologically, hydrocephalus may be caused by 1) overproduction of CSF, 2) failure to resorb CSF, or 3) a blockage that prevents flow through the ventricles. Hydrocephalus is categorized as either communicating, in which there is no physical blockage to CSF flow, or noncommunicating, in which CSF flow is blocked, usually at the aqueduct. As the volume of CSF in the ventricles increases, the pressure on surrounding brain tissues leads to neuronal cell death. Left untreated, progressive tissue destruction leads to physical and cognitive decline and eventual death (Gadsdon et al., 1979).

Congenital hydrocephalus, in which the disease presents at or shortly after birth, is often associated with subarachnoid hemorrhage, where residual blood products are believed to occlude the arachnoid (Rudas et al., 1998). It has been estimated that 60% of the cases of congenital hydrocephalus in human infants are acquired, with an identifiable pathology, while the other 40% have no known cause and are likely genetic in origin (Haverkamp et al., 1999). One known genetic cause of congenital hydrocephalus in humans is mutation of the gene encoding the L1CAM cell adhesion protein, although studies show a strong heritable component to the development of hydrocephalus (Castro-Gago et al., 1996; Chalmers et al., 1999; Chow et al., 1990; Chudley et al., 1997; Hamada et al., 1999; Koh and Boles, 1998; Munch et al., 2012; Portenoy et al., 1984; Rosenthal et al., 1992; Verhagen et al., 1998; Vincent et al., 1994).

The study of animal models has uncovered numerous mechanisms that may underlie the development of hydrocephalus. For example, abnormal differentiation of the arachnoid layer in mice deficient in the forkhead gene Foxc1 leads to communicating hydrocephalus due to a decrease in CSF absorption (Green, 1970; Kume et al., 1998). Deletion of a genomic segment encompassing the human FOXC1 gene has been associated with hydrocephalus in four human patients (Kume et al., 1998). Abnormal development of the subcommissural organ, overproduction of CSF by the choroid plexus, ependymal denudation or disorganization, and defects in ependymal cilia have also been shown to cause hydrocephalus in mouse and rat models (Banizs et al., 2005; Batiz et al., 2006; Blackshear et al., 2003; Davy and Robinson, 2003; Feng et al., 2009; Ibanez-Tallon et al., 2004; Jones and Bucknall, 1988; Krebs et al., 2004; Lechtreck et al., 2008; Lindeman et al., 1998; Sapiro et al., 2002; Tullio et al., 2001; Wilson et al., 2010).

Enhancer trapping, the random integration of a minimal promoter-reporter transgene, can localize genes active in specific tissues or at specific times during development (Allen et al., 1988). Insertional mutagenesis during the generation of transgenic mice is common, and enhancer traps may confer such mutations when they not only “trap”, but also inactivate, a gene or its regulatory elements. A promoter mapping transgene generated in our laboratory inadvertently functioned as an enhancer trap, conferring widely varying lacZ expression patterns across independent transgenic lines. One of these lines showed lacZ expression in the embryonic and adult brain, and as homozygotes, the mice developed a rapidly progressing juvenile hydrocephalus. These Juvenile hydrocephalus mice (formally Jhy^{Tg(Dlk1-lacZ)1Jvs}, here abbreviated Jhy^lacZ) show overt doming of the skull by one week after birth, and die or must be sacrificed by 4-8 weeks of age. Histological analysis shows the open aqueduct indicative of communicating hydrocephalus, and electron microscopy identified altered microtubule organization of the ependymal cilia. Mapping of the transgene integration site found that it disrupted the previously unstudied gene 4931429I11Rik (now named Jhy). The integration and an associated deletion remove a splice acceptor site within the Jhy gene, suggesting reduced function of Jhy is causative for the phenotype. The Jhy gene is conserved across vertebrates, but has no homology to other genes in the mouse genome, and carries no identifiable functional domains. This manuscript provides a detailed descr iption of the Jhy^lacZ/lacZ mouse phenotype; characterization of the Jhy gene is ongoing.

METHODS

Jhy^lacZ mice

The Juvenile hydrocephalus transgene (Jhy^lacZ) was generated from the promoterless lacZ vector pNASS-β (Clontech, Mountain View, CA). A genomic fragment from −248 bp to +50 bp relative to the Dlk1 transcriptional initiation site (Osoegawa et al., 2000) was isolated and cloned into the vector. The transgene (-248Dlk1lacZ), including the Dlk1 promoter, SV40 splice donor-acceptor site, lacZ gene, and polyadenylation signal, was then excised, gel purified, and injected into FVB/N fertilized mouse eggs with the assistance of the University of Illinois at Chicago Transgenic Production Service. Offspring were genotyped by PCR using primers spanning the junction between the Dlk1 promoter (OL410, 5′-GGGGACGACAGTACGAAAAGGC-3′) and the lacZ gene (OL411, 5′-GTATCGGCCTCAGGAAGATCGC-3′). DNA for genotyping was obtained from tail clips, digested by proteinase K overnight, phenol:chloroform extracted, and ethanol precipitated. Twelve transgenic lines were established on an FVB/N background and analyzed for lacZ expression. The Jhy^lacZ line is maintained by intercrossing Jhy^lacZ/+ animals. Jhy^lacZ mice were genotyped using two sets of primers that amplified either the wild type allele (OL1408, 5′-AAGCTCTGTCTGGGCTCACAATCT-3′; OL1409, 5′-TGACCTCTTGGGACATTGCCTGAT-3′) or the Jhy^lacZ allele (OL1420, 5′-TCCAGTTCAACATCAGCCGCTACA-3′; OL1395, 5′-TTAGAAGCGTTCGTTTGTGCAGCC-3′). PCR was performed in an Eppendorf Mastercycler (Hauppauge, NY) as follows: 95°C for 5 min; 35 cycles of 94°C for 30 sec, 62°C for 30 sec, and 72°C for 1 min; and a final step of 72°C for 10 min.

β-galactosidase assay for lacZ expression

Brains were dissected at postnatal day 0.5 (P0.5) and P5, testes were dissected from adult mice, and both were immediately fixed in 4% paraformaldehyde in 1× PBS for 1 hour on ice. Following fixation, tissues were washed in wash buffer (2 mM MgCl₂, 0.01% deoxycholic acid, 0.02% NP-40, in PBS) three times for 30 minutes each at room temperature (RT). Expression of lacZ was detected as β-galactosidase staining using the detection solution (1 mg/ml ×-gal in DMF, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl₂, 0.01% deoxycholic acid, 0.02% NP-40, in PBS), incubating overnight in the dark at RT. The next day, tissues werewashed three times in 1× PBS, dehydrated in solutions of increasing ethanol concentration, and infused with paraffin wax using a Shandon Citadel 1000 tissue processor (Fisher Scientific, Pittsburgh, PA, USA). Processed tissues were embedded in blocks of paraffin using a Shandon Histocentre 2 embedding machine (Fisher Scientific, Pittsburgh, PA). Embedded tissues were cut into 10 μm sections using a microtome (Leica Microsystems RM2125, Bannockburn, IL, USA) and background stained with 10% eosin for 1 minute. Coverslips were applied using Permount (Fisher Scientific, Pittsburgh, PA) and images were taken on a Leica MZFLIII dissecting microscope (Wetzlar, Germany).

Magnetic resonance imaging

MRI was performed at the University of Chicago Magnetic Resonance Imaging & Spectroscopy Research Laboratory. Jhy^+/+ and Jhy^lacZ/lacZ mice at P6 and P13 were sacrificed by CO₂ inhalation approximately one hour prior to imaging. Imaging was performed using a Bruker scanner equipped with a 4.7T magnet and a custom-made birdcage coil for juvenile mice. Coronal T2-weighted axial spin echo images were collected using the parameters: repetition time 3000 ms, echo time 60 ms, array size 256×256, field of view 2.0 cm.

Genomic localization and sequence analysis

The BD GenomeWalker Universal Kit (Clontech, Mountain View, CA) was used to map the integration site of the Jhy^lacZ transgene. The GenomeWalker protocol was followed with the following specifications. A library created from DraI digestion of genomic tail DNA was used for mapping the distal transgene integration boundary. Following the purification of digested DNA and ligation of adaptors, nested PCR was performed with the following primers: outer (OL834, 5′-CCCGTTTTTCCCGATTTGGCTACATGACA-3′; AP1, 5′-GTAATACGACTCACTATAGGGC-3′) and inner (OL836, 5′-CAACCGCTGTTTGGTCTGCTTTCTGACAA-3′; AP2, 5′-ACTATAGGGCACGCGTGGT-3′). The proximal boundary of the transgene integration was mapped using the following primers: OL1368, 5′-AACACTTTCATGGATCCTGAGGACA-3′; OL1618, 5′-GTAACCGTGCATCTGCCAGTTTG-3′. PCR products were sequenced by the University of Illinois at Chicago Research Resources Center DNA Services facility (Chicago, IL). Cross-species Jhy cDNA and protein alignments were performed in Geneious v5.6.6 (Biomatters Ltd) using Geneious and ClustalW alignments, respectively, with the default parameters.

Quantitative RT-PCR

Total RNA was extracted from P0.5 mouse brains using LiCl-urea precipitation (Auffray and Rougeon, 1980), and DNA contamination was removed using TURBO DNA-free (Ambion, Austin, TX). Reverse transcription was performed with 1 μg of total RNA using Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA). The cDNA was diluted 1:10, and 2 μl was used for PCR analysis. Control reactions were performed without reverse transcriptase. Quantitative RT-PCR (qRT-PCR) was performed using the DNA Engine Opticon 2 (MJ Research, Hercules, CA) with the following primers: Jhy, OL1366, 5′-AGCCAACAACACTAACAGGGAAGA-3′ and OL1524, 5′-TCCAGTTGGGATCATATCGGAGGT-3′ that span exons 2-3; Bsx, OL1325, 5′-CAGAACCGGCGGATGAAGCATAAA-3′ and OL1326, 5′-ACTTCGTCCTCGGGCTCAGTAA that lie within exon 3; Crtam, OL1114, 5′-AGCACACGTGGTATGGAAGAAGGA and OL1115, 5′-TTTGGCTCCCGAGTATAACTGCGT that span exons 8-10; β-actin, OL1112, 5′-TCTTGGGTATGGAATCCTGTGGCA-3′ and OL1113, 5′-TCTCCTTCTGCATCCTGTCAGCAA-3′. For all primer pairs qRT-PCR was performed using SYBR GreenER (Invitrogen, Carlsbad, CA) under the following conditions: 50°C for 2 min; 95°C for 10 min; followed by 40 cycles of 94°C for 15 sec, 62°C for 30 sec, 72°C for 15 sec. A melting curve ranging from 60°C to 95°C was run after each reaction to test for primer dimerization or the formation of multiple products.

Tg^BAC rescue mice

The BAC clone RP23-273J15 was obtained from the Children’s Hospital Oakland Research Institute (Oakland, CA). BAC DNA was linearized with BspEI and prepared for injection by phenol:chloroform extraction, ethanol precipitation, and resuspension in microinjection buffer (10 mM Tris-HCl pH 7.5, 0.1 mM EDTA, 30 μM spermine, 70 μM spermidine, 100 mM NaCl). The linearized BAC was microinjected into FVB/N fertilized mouse eggs with the assistance of the University of Illinois at Chicago Transgenic Production Service and 38 offspring were produced. Pups were genotyped for the BAC from tail DNA using the primer set: OL1183, 5′-CGACTCAAGCCTTCGCGAAAGAAA-3′; OL1561, 5′-CGTGTTTGAAGTGATCAGCGGCTT-3′. Five of the 38 transgenic mice were positive for the BAC, and one expressing transgenic line was generated on the FVB/N background. Formally Tg(RP23-273J15)1Jvs, this line is here abbreviated Tg^BAC.

Scanning electron microscopy

Jhy^+/+, Jhy^lacZ/+, and Jhy^lacZ/lacZ littermate P5 brains were dissected and fixed in Karnovsky’s fixative (2.5% glutaraldehyde, 2.0% paraformaldehyde, 0.1 M sodium cacodylate buffer pH 7.4) for 4 days at RT. Dissecting the left and right hemispheres away from the midbrain and removing the hippocampus exposed the lateral ventricles. Brains were washed 3 times in PBS at RT, then dehydrated in a graded ethanol series on ice with the 100% ethanol solution containing molecular sieves. The preparations were dried in a Bal-Tec CPD 030 Critical Point Dryer (Bal-Tec AG, Fürstentum, Liechtenstein), coated with gold/palladium using a DentonVacuum Desk IV Sputter Coater (Denton Vacuum LLC, Moorestown, NJ) set at 48% power, and imaged directly on a JEOL 5600 LV scanning electron microscope (JEOL, Peabody, MA). The accelerating voltage was set at 15 kV, with the higher magnification images taken at 20 kV.

Transmission electron microscopy

Jhy^+/+ and Jhy^lacZ/lacZ brains were dissected at P5 and immersed in Karnovsky’s fixative for seven days. The lateral ventricles were isolated and further fixed in Karnovsky’s for an additional 24 hours. The tissue was washed in three changes of 0.1 M sodium cacodylate buffer pH 7.4 and immersed in 1% osmium tetroxide in 0.1 M sodium cacodylate buffer pH 7.4 for two hours at RT. The tissue was again washed in 0.1 M sodium cacodylate buffer pH 7.4 and then dehydrated using increasing concentrations of ethanol. The tissue was infiltrated overnight with 1 part 100% ethanol and 1 part Spurr’s resin, and two additional changes with pure Spurr’s resin were made over a 24-hour period. Finally, the tissue was embedded with Spurr’s resin in Eppendorf tubes, and polymerized at 70°C for 48 hours. The polymerized blocks were trimmed and faced off at 1 μm sections using a Reichert-Jung Ultracut E Ultramicrotome (Leica Microsystems, Buffalo Grove, IL). The 1 μm sections were placed on glass slides, dyed with 1% toluidine blue, and viewed to determine the region to be used for imaging. Next, 80 nm sections were cut and picked up onto 200 hex mesh nickel parlodion carbon coated grids and allowed to completely dry. The sections were stained for 5 minutes in lead citrate, followed by three washes in deionized water. Once dried, the sections were stained for 45 minutes with uranyl acetate (4% aqueous), followed by three washes in deionized water. The sections were dried, and stained for an additional 5 minutes with lead citrate, followed by three washes in deionized water, and allowed to completely dry. The sections were viewed using a JEOL 1200EX transmission electron microscope (JEOL, Peabody, MA, USA), and areas of interest were photographed.

Statistical analysis

Two-tailed Student’s t-tests were used to calculate the significance of data obtained in qRT-PCR experiments. Standard error of the mean was used to calculate error bars.

Immunofluorescence

P5 mouse brains were fixed in Bouin’s fixative for 18-24 hours at RT, then cleared and stored in 70% ethanol. Tissues were then dehydrated, embedded in paraffin and sectioned with a microtome to 8 μm. Standard immunofluorescence techniques were used, with variations as follows. For JHY detection, sections were dewaxed with xylene and rehydrated through a series of ethanol washes. Heat induced antigen retrieval (0.3% Sodium Citrate, 0.05% Tween-20 pH 6.0) was performed followed by washing in PBS. Slides were blocked using 5% donkey normal serum, 1% BSA, 0.5% Tween-20, 0.75% glycine in PBS for 1 hour at RT, and incubated at 4°C overnight with primary antibody against JHY (1:10; Sigma-Aldrich, HPA039612). The sections were then incubated with Alexa Fluor 647-conjugated secondary antibody for 2 hours at RT (1:250; Jackson ImmunoResearch Laboratories, West Grove, PA). Finally, the slides were DAPI stained (D1306; Life Technologies, Grand Island, NY.) and mounted using Vectashield mounting medium (H1000; Vector Laboratories, Burlingame, CA).

RESULTS

Enhancer trapping identifies a gene required for CSF balance

An experiment was designed to identify regulatory elements of the mouse Delta-like 1 (Dlk1) gene (Fay et al., 1988; Laborda et al., 1993; Okamoto et al., 1997; Smas and Sul, 1993). Varying lengths of the Dlk1 upstream region were linked to a lacZ reporter gene and injected into fertilized FVB/N mouse eggs, generating 12 independent transgenic lines. When analyzed for lacZ expression during midgestation, embryos from the different lines showed widely varying patterns of expression, none of them resembling the well characterized pattern of Dlk1 (data not shown). Enhancer trapping can occur unintentionally, when a transgene that is believed to carry an enhancer either has no enhancer, or it is overwhelmed by stronger regulatory elements at the site of integration. Analysis of the Dlk1 upstream region subsequently found that no developmental enhancers lie within the transgene sequences; the Dlk1-lacZ transgenes functioned instead as enhancer traps (Rogers et al., 2012; Yevtodiyenko et al., 2004).

One transgenic line, examined as whole mount embryos at embryonic day 11.5 (e11.5), showed lacZ expression in a small region on the dorsal aspect of the head (Fig. 1A-C). Histologic analysis identified the expressing structure as the epiphysis, the roof of the diencephalon that will form the pineal gland (Fig. 1C). Intercrossing the transgenic animals showed that mice homozygous for the transgene develop progressive and ultimately fatal juvenile hydrocephalus (Fig. 1D, E). None of the other transgenic lines exhibited any degree of hydrocephalus, suggesting the phenotype results from an insertional mutation generated by the transgene, rather than expression of the transgene itself. The transgene integration therefore trapped, and likely altered the function of, a gene required for proper CSF balance in the developing mouse brain. The transgenic mouse line has therefore been named Jhy^lacZ for Juvenile hydrocephalus.

Figure 1. Jhy^lacZ/lacZ mice develop hydrocephalus.

A) Jhy^lacZ/lacZ embryo at e11.5 showing lacZ expression in the epi physis of the diencephalon. B) Enlarged view of lacZ expression in A. C) Saggital section of diencephalon lacZ expression in e11.5 Jhy^lacZ/lacZ embryo. D) Jhy^+/+ (left) and Jhy^lacZ/lacZ littermates (right) at 3 weeks of age, showing doming of the skull and ataxia in the Jhy^lacZ/lacZ mouse. E) Hemorrhaging beneath the skull is visible in a 3 week old Jhy^lacZ/lacZ mouse (top) compared to a Jhy^+/+ littermate (bottom). F, G) Hematoxylin/eosin stained sections of Jhy^+/+ (F) and Jhy^lacZ/lacZ (G) brains at P5, showing ventricular dilation.

Jhy^lacZ/lacZ mice show early and progressive hydrocephalus

Heterozygous Jhy^lacZ/+ mice appear normal in all aspects, and showed no evidence of hydrocephalus. Intercrossing Jhy^lacZ/+ animals gave offspring that appeared normal at birth, and all genotypes were found in the expected ratios. As the Jhy^lacZ/lacZ homozygous mice grew, however, 63% began to display the domed head indicative of hydrocephalus between 2 and 5 weeks after birth (Fig. 1D, E; Fig. 2A). The enlargement of the head progresses rapidly, and as the disease worsens the mice become ataxic and begin to lose weight, likely from motor deficits that interfere with feeding (Fig. 1D). Intracranial hemorrhaging can often be observed externally in the albino FVB/N mice, though this occurs relatively late in the course of the disease and appears to be secondary to the hydrocephalus (Fig. 1E). Histological analysis of the brains of Jhy^lacZ/lacZ animals at P5 shows dramatically enlarged lateral ventricles and loss of neuronal tissue (Fig. 1F, G). All of these early onset hydrocephalic mice have died, or been sacrificed for humane reasons, by 4-8 weeks of age (Fig. 2B). A small percentage of Jhy^lacZ/lacZ mice do not show the skull doming of early hydrocephalus, but begin to display the related ataxia and weight loss at 5-6 weeks of age (Fig. 2C). The lack of a skull phenotype is expected in these late onset animals, as the fusion of the anterior frontal cranial suture between 3 and 6 weeks of age prevents skull enlargement beyond this point (Bradley et al., 1996). These late onset mice were followed until 12 weeks of age and all survived to this time point, suggesting a milder course of the disease. Overall, 74% of Jhy^lacZ/lacZ mice develop hydrocephalus, the majority as juveniles.

Figure 2. Jhy^lacZ/lacZ hydrocephalus is rapid and progressive.

A) Jhy^lacZ/lacZ mice (squares) develop outwardly visible hydrocephalus with a penetrance of 74%, with most mice affected by 5 weeks of age. No Jhy^lacZ/+ mice (diamonds) developed hydrocephalus or died during the 12 week monitoring period. B) Most Jhy^lacZ/lacZ mice die by 4-8 weeks of age. C) Early onset hydrocephalus occurs in 63% of Jhy^lacZ/lacZ mice, with an additional 11% developing a late onset milder form of the disease.

Jhy^lacZ/lacZ mice show normal brain patterning

Juvenile hydrocephalus can be secondary to abnormal brain development or patternin g, when malformations may physically obstruct CSF flow (Partington, 2001; Pattisapu, 2001). Developmental malformation syndromes are characterized by gross defects in brain structure, or by more subtle disorganization of the cortical layers. To more closely analyze the neuronal development of Jhy^lacZ/lacZ mice, brains were examined by histological analysis at e18.5, P0.5 and P5. Jhy^lacZ/lacZ brains appear structurally normal at e18.5 with normal ventricular size and shape and open communication between the ventricles (data not shown). Serial sectioning was then performed through entire Jhy^+/+ and Jhy^lacZ/lacZ mouse brains at P0.5, prior to the onset of hydrocephalus (Fig. 3A-H). No changes were seen in comparison to Jhy^+/+ littermates, and all existing brain structures appeared correctly patterned. By P1.5 mild hydrocephalus is evident in many Jhy^lacZ/lacZ brains (data not shown), and at P5 the lateral and third ventricles of Jhy^lacZ/lacZ mice are significantly larger than their Jhy^+/+ littermates (Fig. 1F, G). The aqueduct and fourth ventricle do not appear enlarged at any stage observed, and the aqueduct remains patent at P0.5, as visualized in transverse and coronal sections (Fig. 3I and data not shown). These data indicate that Jhy^lacZ mice suffer from communicating hydrocephalus, in the absence of any developmental malformation.

Figure 3. Jhy^lacZ/lacZ brains show normal development and patterning.

A-H) Coronal sections of Jhy^+/+ and Jhy^lacZ/lacZ brains at P0.5 show normal morphology and patterning. I) Transverse section of Jhy^lacZ/lacZ brain at P0.5 shows a patent aqueduct. The inset shows an enlarged image of the aqueduct. J-M) Magnetic resonance imaging of Jhy^+/+ and Jhy^lacZ/lacZ brains. J, K) MRI at P6 shows significant dilation of the lateral ventricles of the Jhy^lacZ/lacZ mouse (K) as compared to a Jhy^+/+ littermate (J). L, M) MRI at P13 shows more pronounced ventricular dilation with loss of brain tissue. N-Q) Adult testis (N, O) shows reduced numbers of mature sperm, while kidney (P, Q) shows normal development.

To examine the progression of the disease in the intact animal, magnetic resonance imaging (MRI) was used to view the brains of Jhy^+/+ and Jhy^lacZ/lacZ littermates (Fig. 3J-M). At P6, dilation of the lateral ventricles can be seen in Jhy^lacZ/lacZ brains as compared to age matched Jhy^+/+ brains (Fig. 3J, K). At P13, the ventricular dilation in Jhy^lacZ brains has increased, with a corresponding loss of brain tissue and thinning of the overlying cortex (Fig. 3L, M). The dorsal region of the third ventricle is noticeably dilated at P13 (data not shown); the ventral region of the third ventricle and the fourth ventricle were not visualized in this analysis.

Jhy^lacZ/lacZ hydrocephalus is nonsyndromic

Hydrocephalus, in both mouse models and human patients, may occur as one component of certain generalized ciliopathies (Baker and Beales, 2009; Lee, 2011). Individuals suffering from these ciliary dysfunction syndromes display other cilia-related phenotypes that include situs inversus, infertility and recurrent respiratory infections (Ferkol and Leigh, 2012). No Jhy^lacZ/lacZ mice were found to display situs inversus, and they showed no evidence of respiratory infections during several years of maintenance. Histological analysis was performed on all major organ systems of Jhy^+/+ and early onset Jhy^lacZ/lacZ mice at 4 weeks of age (Fig. 3N-Q and data not shown). Particular attention was paid to those tissues involved in ciliary-related disease syndromes, such as the lungs, kidney and testis. With the exception of the testes, no alterations were found in any of the tissues examined. Overall testis structure was normal in Jhy^lacZ/lacZ mice, but there appeared to be reduced numbers of mature, flagellated sperm in the lumen of many tubules (Fig. 3N, O). As one example of an unaffected tissue, the structure of the kidney was unaltered in Jhy^lacZ/lacZ mice as compared to Jhy^+/+ (Fig. 3P, Q). The effects of the Jhy^lacZ mutation are therefore limited to the brain and the testes.

Fertility can be difficult to analyze in early lethal mutations, and most early onset Jhy^lacZ/lacZ mice are quite debilitated by breeding age. No early onset Jhy^lacZ/lacZ mice have produced pregnancies, but two of the late onset animals have given offspring. The longest surviving hydrocephalic Jhy^lacZ/lacZ mouse, a male, fathered a litter of 4 pups with an FVB/N female at 2 months of age. One female Jhy^lacZ/lacZ animal gave birth to a litter of 6 pups at 3 months of age when crossed to a Jhy^lacZ/+ male. The Jhy^lacZ/lacZ offspring of this Jhy^lacZ/lacZ female showed no phenotypic differences from Jhy^lacZ/lacZ pups of a Jhy^lacZ/+ female. Jhy^lacZ/lacZ mice are therefore fertile, but males may show subfertility due to reduced sperm numbers, and most animals of both sexes do not breed due to the rapid progression of the hydrocephalus phenotype.

The Jhy^lacZ reporter is expressed in the brain

The expression pattern of an enhancer trap transgene integrated into the mo use genome typically reflects that of a gene or genes near the integration site. The integration of the Jhy^lacZ transgene within the Jhy gene suggests that lacZ may serve as a proxy for the expression of Jhy and/or other genes in the region. To further characterize the transgene expression pattern, lacZ expression was analyzed at P0.5 and P5. At P0.5, lacZ expression is visible in the pineal gland (Fig. 4A, B, E, F), in the hypothalamus (Fig. 4A, B, M, N), and in the ependymal cells of the aqueduct of Sylvius (Fig. 4Q, R). At P5, expression is similarly observed in the pineal (Fig. 4C, D, G, H), the hypothalamus (Fig. 4C, D, O, P) and the ependyma of the aqueduct of Sylvius (Fig. 4S, T). At this stage expression is also seen in the choroid plexus of the third ventricle but not the choroid plexus of the lateral or fourth ventricles (Fig. 4K, L). Expression of the lacZ reporter within the developing brain is consistent with a causative role for the transgene integration in the hydrocephalus of Jhy^lacZ/lacZ mice. The choroid plexus and ependymal cells both play key roles in CSF balance, with the choroid responsible for producing the bulk of the CSF, while ependymal cilia regulate CSF flow and maintain a patent aqueduct (Bruni, 1998).

Figure 4. Jhy^lacZ transge ne reporter expression in the brain.

Jhy^lacZ transgene expression was examined in Jhy^+/+ and Jhy^lacZ/lacZ mice at P0.5 and P5. A-D) Expression of the lacZ reporter is observed in the pineal gland and hypothalamus at P0.5 and P5. Dilation of the lateral ventricles can be seen in Jhy^lacZ/lacZ mice at P5. E-H) Higher magnification of the pineal gland lacZ expression, and lack of expression in the subcommissural organ. I-L) Expression of lacZ is seen in the choroid plexus of the third ventricle at P5, but not in the choroid plexus or ependyma of the lateral ventricles. At P5 dilation of the third ventricle is seen in Jhy^lacZ/lacZ mice. M-P) Higher magnification of the hypothalamus lacZ expression. Q-T) Expression of lacZ in the ependyma lining the aqueduct of Sylvius. Structures are indicated as follows: A, aqueduct of Sylvius; C, choroid plexus; H, hypothalamus; L, lateral ventricle; P, pineal gland; S, subcommissural organ; 3, third ventricle.

Ependymal cilia are morphologically abnormal in Jhy^lacZ/lacZ mice

The motile cilia of the ependymal cells lining the ventricular surfaces play a role in maintaining proper CSF flow through the ventricular system. Defects in ciliary structure or function underlie several mouse models of hydrocephalus as well as certain human cases (al-Shroof et al., 2001; Batiz et al., 2006; Davy and Robinson, 2003; Feng et al., 2009; Ibanez-Tallon et al., 2004; Sapiro et al., 2002; Tullio et al., 2001; Wessels et al., 2003). Scanning electron microscopy (SEM) was used to inspect the developing cilia of P5 Jhy^lacZ/lacZ mice and their Jhy^+/+ littermates. The P5 time point was chosen with the goal of observing any ciliary changes that may be causative for the hydrocephalus, while avoiding the effects of morphological damage and neuronal loss found at later stages. Opening the ventricles of Jhy^+/+ animals invariably caused breakage at the attachment point that joins the ventricular walls (Fig. 5A, arrow). Early ventricular enlargement in Jhy^lacZ/lacZ brains prior to attachment leads to a loss of this adhesion point (Fig. 5B). Ependymal cilia develop along the lateral ventricular walls beginning at birth in a caudorostral-ventrodorsal direction. In Jhy^+/+ animals, large portions of the caudal ventricular walls are covered with ciliary tufts, while the cilia are less dense across more rostral areas (Fig. 5A). The cilia display a consistent directional orientation suggestive of the coordinated motion required for laminar flow (Fig. 5C, E). In P5 Jhy^lacZ/lacZ animals, cilia development has progressed equally far across the lateral ventricular wall, but at low magnification the cilia appear less dense than in comparable ventricular regions of Jhy^+/+ animals (Fig. 5B, D). A t hig her magnification, the cilia of Jhy^lacZ/lacZ animals are sparser, shorter and of irregular length, and oriented in random directions, suggesting a loss of coordinated movement (Fig. 5F). Similar data were obtained from all regions analyzed across the ciliated areas of the ventricles, regardless of the density of the developing cilia at each region.

Figure 5. Jhy^lacZ/lacZ ependymal cilia exhibit structural abnormalities.

A-B) SEM was used to visualize the ultrastructure of the ependymal cilia in P5 Jhy^+/+ (A) and Jhy^lacZ/lacZ (B) lateral ventricles. Caudal is to the left. C-D) Higher magnification images of the ependymal cilia in P5 Jhy^+/+ (C) and Jhy^lacZ/lacZ (D) lateral ventricles. E-F) High magnification images of the ependymal cilia in P5 Jhy^+/+ (E) and Jhy^lacZ/lacZ (F) lateral ventricles. G-I) Representative TEM images of Jhy^+/+ 9+2 (G), and Jhy^lacZ/lacZ 9+0 (H) and 8+2 (I) axonemes. J) TEM image showing partial displacement of a peripheral ring microtubule doublet towards the center of the axoneme in a Jhy^lacZ/lacZ brain. Magnification for all TEM images is 100,000X. Corresponding regions of the lateral ventricle were examined in each mouse (n = 3 each for Jhy^+/+ and Jhy^lacZ/lacZ), and representative images were taken with scale bars embedded at 75× (A, B), 2,000× (C, D), and 8,000× (E, F).

Transmission electron microscopy (TEM) was used to analyze the ultrastructure of Jhy^lacZ/lacZ lateral ventricular ependymal cilia in P5 brains. Normal motile cilia display a 9+2 arrangement of microtubule doublets, and this pattern was observed in 99% of the Jhy^+/+ ependymal cilia (Fig. 5G and Fig. 8D). In Jhy^lacZ/lacZ mice, only 7% of cilia displayed the normal 9+2 pattern, with 71% of cilia having a 9+0 microtubule arrangement, and 19% having an 8+2 pattern (Fig. 5H, I and Fig. 8D). The microtubule pairs found in the center of Jhy^lacZ/lacZ 8+2 cilia do not display the characteristic singlet morphology of the central pair of normal 9+2 cilia. Rather, they show the asymmetric A/B structure of a peripheral doublet that has moved to the center (Fig. 5I). In some axonemes this aberrant doublet is found positioned between the center of the axoneme and an open position in the outer ring (Fig. 5J). Although further investigation is needed, this pattern suggests a 9+0 ciliary axoneme in the process of reorganizing to form an 8+2 axoneme. All other axonemal components visualized, including the dynein arms and radial spokes, appeared normal (Fig. 5H, I).

Figure 8. Transgenic rescue of Jhy^lacZ/lacZ mice.

A) The RP23-273J15 BAC clone spans the region 80 kb proximal to 20 kb distal to the Jhy gene, while the integrated Tg^BAC has been truncated to contain 60-70% of the protein coding region of Jhy. The white boxes represent the Bsx, Jhy and Crtam genes, the arrowhead indicates the Jhy^lacZ integration, the arrows indicate the direction of gene transcription, the grey box indicates the end of the truncated portion of Tg^BAC that cannot be defined due to a lack of SNPs. B) The incidence of hydrocephalus is reduced from 74% in Jhy^lacZ/lacZ mice (diamonds) to 37% in Tg^BAC/+;Jhy^lacZ/lacZ mice (squares). C) Image of Jhy^lacZ/lacZ (left) and Tg^BAC/+;Jhy^lacZ/lacZ (right) littermates showing phenotypic rescue by Tg^BAC. D) Quantification of TEM analysis of ependymal ciliary microtubule arrangement in Jhy^+/+ (black bars), Jhy^lacZ/lacZ (white bars) and Tg^BAC/+;Jhy^lacZ/lacZ (grey bars) at P5 (n=620 for Jhy^+/+, 147 for Jhy^lacZ/lacZ, and 117 for Tg^BAC/+;Jhy^lacZ/lacZ), * denotes p ≤ 0.05 relative to Jhy^+/+, ** denotes p ≤ 0.05 relative to Jhy^lacZ/lacZ.

The Jhy^lacZ transgene integrated into an uncharacterized gene, 4931429I11Rik

Enhancer trapping with a reporter gene allows the bacterially-derived lacZ gene to be used as a tag to identify the transgene integration site. The GenomeWalker Universal Kit was used to identify the integration site of the Jhy^lacZ transgene. An adaptor-tagged restriction library was generated from Jhy^lacZ/lacZ genomic DNA, which served as a template for PCR using primers within the transgene against primers corresponding to the adaptors. Amplification yielded a junction fragment containing the 5′ end of the Jhy^lacZ transgene, along with endogenous flanking sequence that placed the integration on mouse chromosome 9 at 40.9 Mb, within exon 5 of the uncharacterized gene 4931429I11Rik (Fig. 6A, B). No function has been reported for this gene, and therefore the name Jhy has been approved by the Mouse Nomenclature Committee to reflect its presumed role in the Jhy^lacZ/lacZ phenotype. The Jhy^lacZ transgene integrated at the beginning of Jhy exon 5, in the same orientation as the gene (Fig. 6B). To characterize the integration boundary of the 3′ end of the transgene, a series of primers were designed within the Jhy gene flanking the integration site. Detailed PCR analysis determined that the integration generated a 513 bp deletion that removes 425 bp of intronic sequence plus the exon 5 splice acceptor site and the first 88 bp of Jhy exon 5 (Fig. 6B, C).

Figure 6. The Jhy^lacZ transgene disrupts the uncharacterized gene 4931429I11Rik (Jhy).

A) The Jhy^lacZ transgene integrated within the Jhy gene on mouse chromosome 9. Boxes denote the Jhy gene and flanking genes Bsx and Crtam, the arrowhead indicates the Jhy^lacZ integration, the arrows indicate the direction of transcription and the primers used for qRT-PCR are denoted with half-arrows. B) The Jhy^lacZ transgene integrated at the beginning of Jhy exon 5, generating a deletion that removes 425 bp of intron 4 and the first 88 bp of exon 5. The black boxes represent Jhy exons, the grey box the Jhy^lacZ transgene, the dotted line the intronic sequence deleted, and the bracket indicates the entire deletion. (In this view the chromosomal orientation of the Jhy gene has been rotated for clarity.) C) Jhy exons 4 and 5 are juxtaposed to show the exon 5 sequence deleted. D) qRT-PCR analysis of the expression levels of Jhy (white bars), Bsx (black bars), and Crtam (hatched bars) in P0.5 Jhy^lacZ/+ and Jhy^lacZ/lacZ brain in comparison to Jhy^+/+ levels (n ≥ 7 for all genotypes), * denotes p ≤ 0.05.

The Jhy gene is conserved across most vertebrate species, with all species analyzed carrying conserved syntenic regions also encompassing Bsx and Crtam. At the nucleotide level, the mouse Jhy cDNA has 72% identity to the human cDNA, 56% to the chicken, and 52% to the zebrafish. In the mouse, Jhy has 9 exons that would generate an mRNA of 3068 nucleotides, and encode a protein of 770 amino acids with a molecular weight of 87 kDa. At the protein level, conservation between mouse and human is 66%, between mouse and chicken 27%, and between mouse and zebrafish 24% (Supplementary Fig. 1). Analysis of the predicted JHY protein revealed no known functional domains, and beyond the single ortholog in all vertebrates there are no obvious paralogs in any species. The biological role of Jhy is therefore still to be determined.

Jhy mRNA and protein are reduced in Jhy^lacZ/lacZ mice

The autosomal recessive inheritance pattern of the Jhy^lacZ/lacZ phenotype suggests that the integration generated a loss of function or hypomorphic allele of Jhy. Quantitative RT-PCR (qRT-PCR) was used to analyze Jhy expression levels in the brains of Jhy^+/+, Jhy^lacZ/+ and Jhy^lacZ/lacZ mice at P0.5. Jhy expression levels were reduced to 85% of wild type in Jhy^lacZ/+ brains and to 41% in Jhy^lacZ/lacZ brains (Fig. 6D). The primers used for this analysis span Jhy exons 2-3, detecting all transcripts containing the 5′ end of the Jhy gene. The transcriptional termination signal of the Jhy^lacZ transgene is expected to disrupt all transcripts at the exon 5 integration point, but the actual transcripts produced from this allele are currently unknown. Should transcription proceed beyond the transgene integration site, the deletion of the first 88bp of exon 5 means that a fully wild type transcript cannot be produced.

The Brain specific homeobox (Bsx) and Cytotoxic and regulatory T cell molecule (Crtam) genes are closely linked to Jhy, with Bsx located 17 kb proximal and Crtam located 9 kb distal (Fig. 6A). Transgene integrations can cause regional changes in gene expression, and therefore expression of Bsx and Crtam were similarly examined by qRT-PCR in P0.5 brains. Levels of Crtam were not significantly altered in Jhy^lacZ/+ or Jhy^lacz/lacZ brains, but Bsx levels were reduced to 77% of wild type in Jhy^lacZ/+ and to 67% in Jhy^lacz/lacZ (Fig. 6D). These data suggest a long-range effect of the Jhy^lacZ transgene, and potential functional linkage of Jhy and Bsx expression.

To explore the levels of JHY protein in Jhy^lacZ mice, a commercial antibody generated against the human JHY homolog (C11ORF63) was used (Sigma-Aldrich, St. Louis, MO) (Ivliev et al., 2012). Ivliev, et al carried out a bioinformatics screen for proteins potentially expressed in ciliated cells, identifying C11ORF63 among 74 novel predicted ciliary proteins (Ivliev et al., 2012). Ciliary expression of these proteins was verified with data from the Human Protein Atlas, a high-throughput analysis of protein expression data across human tissues (Uhlen et al., 2010). This study had explored C11ORF63 expression in seve ral ciliated tissues using a peptide antibody directed against the exon 3-encoded region of the predicted C11ORF63 protein. This antibody was obtained and tested against P5 mouse brain, and found to recognize the mouse JHY protein as well (Fig. 7A). This was not surprising, as the C11ORF63 peptide antigen bears 87% identity to the homologous mouse sequence. In Jhy^+/+ P5 mouse brain sections, expression of JHY was seen in the ependymal cells lining the lateral ventricles (Fig. 7A), and scattered cells within deeper layers, but this protein signal was significantly reduced in Jhy^lacz/lacZ animals (Fig. 7B).

Figure 7. Jhy^lacZ/lacZ mice express reduced levels of JHY protein.

A) JHY protein localizes to the ependymal cells of the lateral ventricle in Jhy^+/+ P5 mouse brain using an antibody directed against the human JHY homolog C11ORF63. B) Jhy^lacZ/lacZ brain shows significantly reduced levels of JHY protein.

The Jhy^lacZ phenotype can be rescued by a Jhy transgene

The most definitive way to verify that a specific gene underlies a particular phenotype is to cure the phenotype by restoring the candidate gene. Bacterial artificial chromosomes (BACs) are ideal for transgenic rescue, since their large size makes it possible to isolate a gene and its regulatory elements on a single clone. Rescue of the Jhy^lacZ phenotype was performed using the BAC clone RP23-273J15 that spans the region from 105 kb proximal to 65 kb distal to the Jhy^lacZ integration site (Fig. 8A) (http://bacpac.chori.org). The RPCI-2 3 library is derived from the C57BL/6 mouse strain, allowing the use of strain-specific polymorphisms to distinguish the endogenous Jhy gene from the BAC Jhy gene, and endogenous from BAC-derived transcripts. The BAC clone was linearized and microinjected into FVB/N embryos and a single transgenic mouse line was obtained that both transmitted and expressed the BAC Jhy gene. Formally named Tg(RP23-273J15)1Jvs, this line is abbreviated here as Tg^BAC. The transgene was crossed into the Jhy^lacZ line to generate animals heterozygous for Tg^BAC and homozygous for Jhy^lacZ (Tg^BAC/+;Jhy^lacZ/lacZ).

Analysis of these mice revealed that a single copy of Tg^BAC yielded a 50% rescue of Jhy^lacZ/lacZ mice, with the incidence of hydrocephalus reduced from 74% in Jhy^lacZ/lacZ mice to 37% in Tg^BAC/+;Jhy^lacZ/lacZ mice (Fig. 8B). Phenotypically, Tg^BAC/+;Jhy^lacZ/lacZ animals displayed reduced doming of the skull, though subtle changes were still observed in some animals (Fig. 8C). Further analysis of the Tg^BAC integration, however, revealed that the BAC transgene had been truncated within the Jhy gene, likely during integration into the genome. Direct sequencing of PCR products across regions containing single nucleotide polymorphisms between FVB/N (Jhy^lacZ background) and C57BL/6 (Tg^BAC background) strains were used to determine that Tg^BAC was truncated at a position between 419 bp into Jhy exon 5 and 2,420 bp into Jhy intron 5 (Fig. 8A). Interestingly, this point is just slightly 3′ to the position of the Jhy^lacZ insertion in the endogenous Jhy gene. Between 60 and 70% of the Jhy protein coding sequence remains on Tg^BAC (Fig 8A).

Jhy expression in Tg^BAC/+;Jhy^lacZ/lacZ P0.5 brains was assayed by qRT-PCR using the Jhy exon 2-3 primers and was found to average 113% of wild type levels (data not shown). The Tg^BAC transgene therefore restores the leve l of an, albeit truncated, Jhy mRNA. Bsx expression, which is decreased to 77% of wild type in Jhy^lacZ/lacZ brains, was unchanged from this level in Tg^BAC/+;Jhy^lacZ/lacZ mice; this is expected since Bsx is not included on the truncated Tg^BAC (Fig. 8A). The increase in Jhy transcription conferred by the Tg^BAC transgene was sufficient to enact partial rescue of the ciliary phenotype as well; 32% of ependymal cilia displayed the normal 9+2 microtubule configuration in Tg^BAC/+;Jhy^lacZ/lacZ cilia, compared to only 7% of cilia in Jhy^lacZ/lacZ (Fig. 8D). These results suggest that Tg^BAC produces a truncated JHY protein that retains partial function.

DISCUSSION

Jhy is a new candidate gene for hydrocephalus

A transgenic insertional mutation on mouse chromosome 9 led to the development of juvenile hydrocephalus, and the subsequent discovery of the Jhy gene. None of the known murine hydrocephalus mutations map to this genomic region, indicating that Jhy^lacZ is a new model for juvenile hydrocephalus. Jhy^lacZ/lacZ mice develop communicating hydrocephalus during early postnatal development, with the earlie st cases visible by histology at P1.5 (Fig. 1 and data not shown). Grossly visible doming of the skull is often observed by 2 weeks of age. These early onset hydrocephalus cases have died or must be sacrificed by 4-8 weeks of age (Fig. 2B). The onset and progression of the Jhy^lacZ hydrocephalus varies, however, and a small percentage of animals instead develop a late onset, milder form of the disease (Fig. 2C). These mice show symptoms developing at 5-6 weeks of age, and the animals survive beyond a 12 week monitoring period.

The homozygous recessive inheritance pattern of the Jhy^lacZ phenotype suggests the disruption of a gene involved in CSF balance, with reduction or loss of function of this gene causative for the phenotype. Genomic analysis identified the locus disrupted by the transgene as carrying the uncharacterized gene 4931429I11Rik, now named Jhy. The Jhy gene can be located in all well-annotated vertebrate genomes, from zebrafish through humans. The expression pattern of the Jhy^lacZ reporter gene suggests activity in the pineal gland, the hypothalamus, the choroid plexus of the third ventricle, and the ependymal cells within the aqueduct of Sylvius. The ependymal cells and choroid plexus in particular play key roles in the production and flow of CSF, suggesting altered expression in these tissues might underlie the Jhy^lacZ/lacZ phenotype. While these experiments did not detect lacZ expression in the ependyma of the lateral ventricles, preliminary analysis of JHY protein by immunofluorescence indicates that JHY is present in these cells (Fig. 4, Fig. 7).

Ciliary defects in Jhy^lacZ/lacZ hydrocephalus

The most striking aspect of the Jhy^lacZ/lacZ phenotype is the disorganization and altered morphology of the ependymal cilia lining the lateral ventricles (Fig. 5). By SEM, the ependymal cilia are sparser, shorter and show a loss of dire ctional orientation that suggests they may be nonfunctional (Fig. 5A-F). By TEM, only 7% of Jhy^lacZ/lacZ cilia show the normal 9+2 microtubule organization of motile cilia, with most having instead a 9+0 or 8+2 configuration (Fig. 5G-J, Fig. 8D). Normal nonmotile cilia display a 9+0 confirmation, indicating that the Jhy^lacZ/lacZ 9+0 cilia are almost certainly immotile, but little is known about the potential motility of 8+2 cilia. The unique central pair morphology of 9+2 cilia is believed necessary for ciliary motility, with the central pair controlling the direction of the stroke. The abnormal doublet morphology of the central microtubules in the Jhy^lacZ/lacZ 8+2 cilia suggest these cilia may also be immotile. Loss of ciliary motility leading to a failure to maintain CSF flow is therefore the predicted mechanism underling the Jhy^lacZ/lacZ hydrocephalus. The JHY protein is not found in the ciliary proteome database (www.ciliome.com), but a recent bioinformatic-based study identified the predicted protein of the human homolog of Jhy (C11ORF63) in the cilia of adult human lung and oviduct (Ivliev et al., 2012).

Studies involving mouse models of hydrocephalus show a requirement for proper cilia structure and function in regulating CSF flow. Only two existing hydrocephalic models, however, show disruption of ciliary microtubule patterning. Hsf1 (Heat shock transcription factor 1)-null mice display reduced ciliary beat amplitude and frequency, and develop chronic sinusitis and communicating hydrocephalus (Takaki et al., 2007). Ultrastructural studies of Hsf1 respiratory epithelium showed that approximately 10% of cilia were structurally abnormal, with deletion and/or disorganization of the central pair and outer doublet microtubules. The abnormal cilia development in Hsf1 mutants was predicted to result from a secondary decrease in Hsp90 (Heat shock protein 90), which facilitates tubulin stability and polymerization (Takaki et al., 2007). Mice mutant for the Stumpy gene develop hydrocephalus and polycystic kidney disease (Town et al., 2008). The ependymal cells of Stumpy mutant animals contain intact basal bodies, but lack ciliary axonemes entirely. Localiz ation of STUMPY protein within the axonemes suggests it may be involved in the intraflagellar transport of proteins required for cilia outgrowth.

Regional gene expression is altered, and JHY protein is lost, in Jhy^lacZ/lacZ mice

Jhy^lacZ/+ mice expressing 85% of the wild type level of Jhy do not develop hydrocephalus, while Jhy^lacZ/lacZ mice expressing 41% of the wild type level develop hydrocephalus with a penetrance of 74%. These numbers suggest a threshold for Jhy expression, where the development of hydrocephalus occurs between 41-85% of wild type Jhy expression. The incomplete penetrance of the Jhy^lacZ/lacZ hydrocephalus may therefore be due to stochastic events that subtly alter Jhy expression levels in Jhy^lacZ/lacZ mice. In considering expression from the Jhy^lacZ allele, however, it is important to consider the transcripts that could be produced. The integration of a transcriptional termination codon suggests that Jhy^lacZ transcripts should contain at most the first 34% of the wild type transcript, which would encode 42% of the full length protein. Deletion of the first 88 bp of exon 5 dictate that any transcript produced beyond the integration site cannot be wild type in nature. The most likely mechanism for reconstructing a near-full length transcript, splicing from exon 4 to exon 6, would generate an early stop codon. An antibody generated against the human JHY homolog C11ORF63 found JHY protein localized to the ventricular ependymal cells of Jhy^+/+ P5 mouse brains, while sections of Jhy^lacZ/lacZ brains showed a significant reduction in JHY protein (Fig. 7). These data support a Jhy loss of function mechanism for the Jhy^lacZ mutation. Future studies will further characterize the protein products made from both Jhy^+/+ and Jhy^lacZ alleles.

Restoring a truncated version of the Jhy gene, roughly the same portion of the transcript as lies upstream of the Jhy^lacZ integration, provides significant rescue of the hydrocephalus phenotype (Fig. 8). Tg^BAC/+;Jhy^lacZ/lacZ animals show near-wild type levels of Jhy transcript when analyzed using primers span ning exon 2-3, and the phenotypic rescue suggests the mechanism is likely to be increased levels of a truncated protein. The JHY protein may therefore be modular in nature, where key domains within the remaining portion of the protein can provide partial function. There is some precedence for the rescue of ciliary defects, particularly structural defects, with gene fragments in other systems. A construct expressing just 35% of the Chlamydomonas centriolar protein Bld10p gave partial rescue of the Bld10p deletion phenotype (Hiraki et al., 2007). Bld10p is a component of the radial spokes of the centriolar cartwheel, and the truncated protein allowed the assembly of largely normal flagella, but with shorter spokes and 8 rather than the expected 9 microtubule triplets.

Expression analysis showed decreased levels of Bsx in Jhy^lacZ/lacZ brains, and no change in the expression of Crtam. Interestingly, the Bsx expression pattern overlaps with that predicted for Jhy in pineal gland and hypothalamus, though not in choroid plexus or ependyma (McArthur and Ohtoshi, 2007). There is no demonstrated role for Bsx in CSF balance or the development of hydrocephalus, and mouse models carrying a deletion of the Bsx gene are fully viable but display altered feeding and nursing behaviors (McArthur and Ohtoshi, 2007; Sakkou et al., 2007). Crtam levels are not altered in Jhy^lacZ/lacZ mice, and deletion of this gene displays no neural phenotype (Yeh et al., 2008). Taken together, these expression and rescue data provide strong support for altered function of the Jhy gene as causative for the Jhy^lacZ/lacZ hydrocephalus. It remains formally possible that altered expression of Bsx may contribute in part to the Jhy^lacZ/lacZ phenotype, and its absence from Tg^BAC may underlie the lack of full rescue. Ongoing experiments will generate a targeted deletion of the Jhy gene, along with additional Tg^BAC animals, to more clearly define all contributory factors.

Ependymal- and choroid plexus-based models for Jhy^lacZ hydrocephalus

At least two models can be proposed for the mechanism of the Jhy^lacZ hydrocephalus. First, loss of ciliary-mediated CSF flow may cause fluid accumulation in the ventricles resulting in hydrocephalus. Jhy appears to func tion in the ventricular ependymal cells to regulate cilia development, with loss of Jhy causing abnormally patterned ciliary microtubules. Hydrocephalus in the Jhy^lacZ mice arises as early as P1.5, suggesting the initiating event must precede this age. Previous analysis has shown that the ependymal cilia of the lateral ventricles develop between birth and P10, while the cilia lining the aqueduct of Sylvius are already present by P1 (Banizs et al., 2005; Spassky et al., 2005; Tissir et al., 2010). Our experiments find abundant cilia in the caudal region of the Jhy^+/+ lateral ventricles by P5, and these cilia are abnormal in Jhy^lacZ/lacZ mice. JHY protein is present in the ependymal cells of the lateral ventricles at P5, and expression of the Jhy^lacZ reporter suggests expression in the ependymal cells lining the aqueduct at P0.5 (Fig. 4R, Fig. 7). These data are consistent with loss of ciliary function in the aqueductal and/or ventricular ependyma being causative for Jhy^lacZ hydrocephalus.

Expression from the Jhy^lacZ transgene is also found in the choroid plexus of the third ventricle, yet histological analysis of the choroid plexus by light microscopy shows no obvious abnormalities (Fig. 3D, Fig. 4L and data not shown). The choroid plexus epithelium carries motile cilia that develop by P1, though these cilia contribute minimally to bulk CSF flow and instead appear to function in a signaling pathway that regulates CSF production (Banizs et al., 2005). In the Tg737^orpk mutant mouse, loss of normal cilia function on the choroid plexus leads to excessive CSF production and hydrocephalus (Banizs et al., 2005). Abnormal choroid plexus cilia might therefore initiate hydrocephalus through CSF oversecretion, with loss of ventr icular ependymal flow exacerbating the increased fluid load. Further investigation of the choroid plexus cilia will determine their role in the development of hydrocephalus in Jhy^lacZ/lacZ mice.

HIGHLIGHTS.

• Jhy^lacZ transgenic mice develop autosomal recessive communicating hydrocephalus.

• The Jhy gene produces a conserved protein with no known functional domains.

• Ependymal cilia in Jhy^lacZ brains are disorganized and randomly oriented.

• Jhy^lacZ cilia have abnormal 9+0 and 8+2 microtubule organization patterns.