Progressive Optic Neuropathy in Hydrocephalic Ccdc13 Mutant Mice Caused by Impaired Axoplasmic Transport at the Optic Nerve Head

Abstract

Purpose: Optic nerve head (ONH) atrophy is frequently associated with hydrocephalic conditions. Cerebrospinal fluid (CSF)-containing meninges form a subarachnoid space that terminates at the ONH, which physically impacts it. This study aims to characterize optic neuropathy in congenital hydrocephalic mice with genetic disruption of the Ccdc13 gene.

Methods: The ccdc13 germline knockout mice were generated. The hydrocephalus phenotype and subarachnoid space surrounding the optic nerve were evaluated using routine histology and Evans blue stain. Optic neuropathy was examined with immunohistochemistry and transmission electron microscopy (TEM). Axon transport was indicated by cholera toxin subunit B (CTB) fluorescence conjugate. Retinal function was evaluated by electroretinography (ERG), and Ccdc13 expression was revealed by a knock-in Gfp reporter.

Results: Ccdc13 mutant mice manifested hydrocephalus at birth. ONH displacement, or negative cupping, and enlarged subarachnoid space at the optic terminus occurred as early as 1 month after birth. Intraocular pressure (IOP) was normal. Optic neuropathy was first observed at the ONH, followed by a distal-to-proximal progression of optic nerve pathology indicated by alteration of axonal ultrastructure and deposition of unphosphorylated neurofilament heavy chain. Anterograde axonal transport was also hampered. Retinal ganglion cell (RGC) function was compromised as early as postnatal day 21 (P21), along with reduced neurofilament heavy chain expression. Optic neuropathy caused by disruption of Ccd13 was non-cell autonomous, stemming from hydrocephalus with presumed high intracranial pressure (ICP), which physically impacts the ONH by increasing the translaminar pressure gradient.

Conclusions: We provided knowledge of optic neuropathy from a congenital mouse model for hydrocephalus. The hydrocephalus in mice could damage the ONH by increasing the translaminar pressure gradient and negative cupping, leading to impairment in axoplasmic transport and RGC pathology. Our findings highlight the importance of the interplay between IOP and ICP in the development of glaucoma.

Cerebrospinal fluid (CSF) is a colorless fluid that circulates in the space of the ventricular system and the spinal cord in the central nervous system. CSF is rich in nutrients, hormones, and signaling molecules that are essential for maintaining brain function by clearing out toxic byproducts from the brain.¹ CSF is mainly produced by the choroid plexus of the ventricular system and flows from the lateral ventricles, entering the third ventricle through the foramen of Monroe, then the fourth ventricle through the central aqueduct, and finally leaves the ventricular system, entering the subarachnoid space.² Abnormalities in CSF secretion, flow, and absorption usually lead to hydrocephalus, with the expansion of the lateral ventricles and increased intracranial pressure.³ Accordingly, above 80% of infants with hydrocephalus had visual function deficits in a small study.⁴ 

As a part of the central nervous system, the optic nerve is also surrounded by CSF. The CSF enters the optic nerve periphery through the optic nerve sheath connecting to the meninges. CSF may also enter the optic nerve through the lymphatic pathway, playing a vital role in the nutrient delivery to the optic nerve and the clearance of metabolic waste.^5–8 On the other hand, clinical studies and experimental animal models suggest that perturbation in CSF secretion, flow dynamics, or composition could result in altered intracranial pressure (ICP), leading to changes in translaminar pressure gradients and displacement of the lamina cribrosa.^4,9–15 Specifically, if IOP is held constant and in the normal range, lowered ICP is a risk factor for normal tension glaucoma as it increases the likelihood of cupping, an outward displacement of the ONH.^13,16–18 Conversely, elevated ICP could also increase the risk for retinal ganglion cell (RGC) atrophy¹⁹ through inward displacement of ONH or negative cupping. 

Hydrocephalus is usually associated with high ICP,²⁰ transmitted from the opposite direction from the intraocular pressure (IOP) to the optic nerve head (ONH) through CSF, leading to nerve stress.^11,18,19 The mechanisms of optic nerve degeneration under hydrocephalic conditions are not well understood. A series of studies suggested that stasis of axoplasmic flow results in axon swelling and extracellular fluid leakage in artificially created intracranial hypertension models of rabbits and monkeys.¹⁹ This notion presumably holds true in general for hydrocephalic conditions with increased ICP; however, direct evidence from genetically engineered animals is still lacking. 

In contrast to primates or many other medium-sized animals, mice have the advantage of genetic manipulations, and many genetic mutants exhibit hydrocephalus. One class of definitive genetic causes of hydrocephalus is the ciliopathy genes, which are suggested to affect the ventricular choroid plexus secretion and CSF flow dynamics, including those coding for coiled-coil domain and transmembrane ciliary proteins.^21,22 On the other hand, the mouse ONH lacks the lamina cribrosa, the anatomic structure thought to be vulnerable to changes in translaminar pressure gradient.^19,23 Thus, whether hydrocephalic mice ONH atrophy is similar to primates or humans remains unknown. 

In this study, we created mutant mice with a disrupted ciliary gene, Ccdc13,²⁴ which exhibited progressive postnatal hydrocephalus. Very recently, Ccdc13 has been characterized as a novel ancestry-specific locus for primary open-angle glaucoma.²⁵ With this mouse model, we investigated the RGC and optic nerve phenotypes. We found an early-onset distal-to-proximal progression of optic neuropathy and impaired axon transport initiated at the site of ONH compounded with a reduction of RGC expression of a neurofilament subunit. 

Materials and Methods

Experimental Animals, Body Weight, and Survival Rate Measurement

The Ccdc13^a/+ transgenic mice were generated using the Cyagen service (Santa Clara, CA, USA). All mice were housed on a 12-hour light-dark cycle at the Guangdong Laboratory Animals Monitoring Institute, and the study was approved by the Zhongshan Ophthalmic Center Animal Care and Use Committee. All experiments conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Animal body weight was measured every 2 days from postnatal day 12 (P12) to P20. Survival rate was measured from P0 to P60, at which nearly all mutant animals died. 

Generation of Ccdc13 Mutant Alleles, Southern Blot, and PCR Genotyping

The “knockout-first” strategy was used to create the Ccdc13 gene-trap allele (Ccdc13^a/+, a/+) by homologous recombination in embryonic stem (ES) cells.²⁶ Briefly, a gene-trap construct was engineered with an Frt site and En2 splicing acceptor (SA) preceding a Gfp reporter, followed by a polyadenylation signal (p(A)), loxP, Pgk-neo, Frt, and loxP. Such a gene-trap cassette was placed between Ccdc13 exons 2 and 3. A third loxP was placed right after exon 3, which will be used for the generation of a conditional knockout allele. For the generating a Ccdc13 null allele (b/+), Sox2-Cre transgenic female mice (JAX Laboratory, 014094, USA) were crossed onto the Ccdc13^a/+ mice to delete exon3, leaving the Gfp reporter driven by the endogenous Ccdc13 promoter. 

For identification of the initially targeted Ccdc13^a/+ allele, Southern analysis was performed as described previously.²⁶ A 5′-genomic probe outside of the left arm of the targeting vector was used to identify 8.8 kb wild type and 13 kb Ccdc13^a/+ EconI-digested fragments, respectively. A 3′- probe outside of the right arm was used to identify 11 kb and 15 kb AhdI fragments from wild type and Ccdc13^a/+ alleles, respectively. 

For genotyping, mouse tail genomic DNA extraction and subsequent PCR reactions were prepared using the TIANGEN DNA Extraction kit (TIANGEN, DP304, China) and Rapid Taq Master Mix (Vazyme, P222-01, China), respectively, according to the manufacturer's instructions. The primers and protocols used for genotyping are provided in the Supplementary Materials. 

Tissue Preparation, Hematoxylin and Eosin Staining, Transmission Electron Microscopy, and Evans Blue Histology

Fixed frozen and vibratome sections and flat mount retinas were prepared essentially, as described in previous studies with some modifications.^22,26 Specifically, for vibratome sections, brains and retinas were fixed and embedded in 4% agarose and then cut into 100 µm thick slices using a vibratome machine (Leica, VT1000 S, Germany). Frozen sections were cut in 15 µm thick slices. 

For the preparation of the eye cups with optic nerves, skull bones were opened and removed first, followed by the removal of the brain tissue. The optic chiasm was cut off to separate it from the optic nerves. Eyeballs with optic nerves were carefully dissected, fixed, embedded in the Optimal Cutting Temperature compound (Tissue-Tek, 4583, USA), and stored at –80°C. Cryostat sections through the retina and optic nerve were prepared for immunohistochemistry. 

For plastic sectioning, brain tissue was fixed and embedded using a Technovit kit, according to the manufacturer’s instructions (Technovit, 14653, Germany), as previously described by Guo et al.²² The 5 µm plastic sections were cut with a microtome (Leica, RM 223, Germany) and stained with hematoxylin and eosin (H&E), according to the manufacturer’s instruction (UBIO, UI0402, China). 

For transmission electron microscopy (TEM), optic nerves were dissected after cardiac perfusion with 4% paraformaldehyde (PFA), fixed in 2% PFA, and 2.5% glutaraldehyde in PBS at 4°C, followed by a secondary fixation with OsO₄ in cacodylate buffer following a standard protocol, as described by Guo et al. (2022).²² The tissues were then embedded in EMBED812 resin (Electron Microscopy Sciences, 14120, USA), and ultrathin sections of 60 to 80 nm were cut using an ultramicrotome (Leica, EM UC7, Germany) and collected onto copper grids (Electron Microscopy Sciences, FCF200-Cu-50, USA). Electron microscopy images were acquired by HT-7700 electron microscope (Hitachi, Japan). 

For Evans blue injection, mice of different ages were anesthetized by intraperitoneal injection of 1% sodium pentobarbital solution at 5 µL per gram body weight (Sigma-Aldrich, P3761, USA). After being anesthetized, the mice were placed on a stereotaxic frame (Thinker Tech Nanjing Bioscience Inc., China), and a small burr hole was drilled over the right lateral ventricle at about 0.85 mm right and −1 mm posterior of the Bregma. Evans blue (Sigma-Aldrich, E2129, USA) was injected at a depth of 1.5 mm using a microinjection needle (Hamilton Company, 700, China). Evans blue was examined 10 minutes after the dye injection on fixed brain vibratome sections. For examining Evans Blue dye in the optic nerve subarachnoid space, optic nerve sections were prepared 1 hour after the dye injection. 

Immunostaining, Western blotting, and RT-qPCR

Immunostaining, Western blotting, and RT-qPCR followed standard protocols. Briefly, for immunostaining of PFA-fixed retinal frozen sections, sections were blocked with PBST containing 10% BSA and 0.1% Triton X-100 (Sigma-Aldrich, X100, USA) followed by incubation with primary antibodies and fluorescent dye-conjugated secondary antibodies. A modification made for staining of optic nerve sections is to elevate Triton X-100 concentration to 2% in the PBST buffer. 

Vibratome sections and flat mount retinas were blocked with PBST containing 0.5% Triton X-100 with elongated incubation with primary and secondary antibodies followed by extensive washes (each for about 30 minutes), respectively. Sections were mounted on slides in Fluoro-mount G medium (Southern Biotech, 0100-01, USA) and subjected to imaging with a Zeiss LSM880 confocal microscope. 

For Western blotting, the retina proteins were extracted with RIPA buffer added with protease inhibitors (Sigma-Aldrich, P7626, USA). Proteins were separated using 4% to 12% FuturePAGE (ACE Biotechnology, ET12412, China) and transferred to PVDF membranes (Millipore, IPVH00010, USA). Blotted membranes were probed with primary antibodies followed by incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies. The chemiluminescent signal from the HRP-catalyzed substrates’ reaction (Thermo Scientific, 34580, China) was imaged, and band densitometry was analyzed using NIH Image J software. Protein expression levels were first normalized to the β-actin signal, then to controls, and expressed as fold changes. The primary antibodies used for Western blot are listed in the Supplementary Materials. 

For RT-qPCR, total RNA was isolated using a total RNA Isolation Kit (Vazyme RC112-01, China) and reverse transcribed into cDNA (HiScript III RT SuperMix, Vazyme R323-01, China). PCR reaction was subsequently performed (ChamQ SYBR Color qPCR Master Mix, Vazyme Q411-02, China) using primers specific for each gene. Gene expression (Ct values) was normalized to Gapdh expression. Relative expression levels were calculated as fold change (=2^−ΔΔCt) and normalized to the controls. Gene primers and PCR protocols are given in the Supplementary Materials. 

Electroretinography

For full-field Flash electroretinography (ERG), the mice were weighed and dark-adapted overnight. The mice were anesthetized as described previously, and the pupils were dilated with Compound Tropicamide Eye Drops (SINQI, H20055546, China). Throughout the procedure, the mice were kept on a built-in heating plate of the CELERIS system (Diagnosys LLC, D430, USA) to maintain their body temperature at 37°C. ERG responses were recorded with photoconductive electrode stimulators placed directly on the cornea of both eyes with 20 mg/mL Hypromellose Eye Drops (ZOC, H20071448, China) applied to the cornea surface. Scotopic and photopic ERGs were recorded using strobe flash stimuli in 8 averaged responses of 0.003 to 10 cd.s/m² and 0.3 to 30 cd.s/m², respectively. 

For pattern ERG (pERG) testing, a photo-guide electrode stimulator was placed directly onto the ocular surface of one eye, centered on the pupil, to present a black and white reversed checkerboard pattern at a 1 degree square size. The photoconductive electrode on the other eye serves as the reference electrode. Exchanging electrodes would give recordings from both eyes. Visual stimuli were 98% contrast, mean luminance constant at 50 cd/ m², with spatial frequency at 0.05 cpd and temporal frequency at 1 hertz (Hz). A total of 200 pERG complete contrast reversals were repeated at least twice for each eye, and the responses were segmented, averaged, and recorded. The mean values of pERG were analyzed to assess P1-N2 amplitudes. 

Anterograde Cholera Toxin Subunit B Conjugate Tracing

For anterograde labeling of RGC axons, a microinjection needle (Hamilton Company, 700, China) was used to deliver 1 µL of 10 µg/mL Alexa Fluor 555 cholera toxin subunit B conjugate (CTB; Invitrogen, C34776, USA) to the mouse eye posterior chamber for at least 2 minutes. Injections were made in both eyes. The optic nerve and brain were harvested 2 weeks after injection for anterograde CTB transport evaluation. Flat mount retinas and vibratome brain sections were prepared as described previously. 

RNA-Sequencing Analysis

P5, P9, and P15 retinas were dissected, followed by total RNA isolation using FastPure Cell/Tissue Total RNA Isolation Kit (Vazyme, RC112-01, China). RNA library construction, sequencing, initial processing, and RNA reads quality control (QC) and trimming were described previously by Guo et al. 2019.²⁷ Trimmed reads were then aligned to the mouse genome NCBI37/mm9 using Tophat (version 2.0.11) to achieve unique mapped reads through the filtepmerr of NH tag of bam files using the BGI Company (China) services. 

Intraocular Pressure Measurement

The optimized rebound tonometer Tonolab (Vantaa, iCare Finland, Finland) was used to record the IOP. The P21, P35, and P56 mice were anesthetized, as stated previously. The IOP was measured at the center of the cornea 5 times consecutively within 2 minutes after the beginning of anesthesia, and the average IOP was recorded as the IOP of the mice. 

Statistics and Optical Imaging

Statistical analysis was performed using GraphPad Prism 9. All graph data are presented as mean ± standard deviation (SD). A one-tailed unpaired Student t-test was performed, and statistical significance was defined as P ≤ 0.05. Sample information and detailed statistics are given in the figure legends. Optical imaging was conducted on LSM 880 (Zeiss, Germany). 

Results

Targeted Inactivation of Ccdc13 Led to Congenital Hydrocephalus

To investigate the function of the cilia protein Ccdc13, we genetically inactivated the Ccdc13 gene in mice using a knockout-first strategy to create a gene-trap allele (Ccdc13^a), which was subsequently converted into a null allele carrying a Gfp reporter (Ccdc13^b) upon germline Cre excision, as described previously²⁶ (also see Materials and Methods; Fig. 1A, Supplementary Fig. S1). The gene-trap allele was confirmed by Southern blotting analysis using mouse tail genomic DNA (Fig. 1B). Western blotting further confirmed the loss of Ccdc13 protein in Ccdc13 homozygous mutants (Ccdc13^b/b; Fig. 1C). Ccdc13^b/b mice manifested severe hydrocephalus with reduced body size (Figs. 1D, 1E). Only 47% of Ccdc13^b/b mice survived to postnatal day 20 (P20), and 27% survived beyond P35 (Fig. 1F). H&E staining revealed significantly enlarged lateral ventricles from P1, a sign of congenital hydrocephalus (Fig. 1G). An examination of Ccdc13 expression using the knock-in Gfp reporter indicated Ccdc13 was strongly expressed in the choroid plexus and ependyma at P1 (Fig. 1H), suggesting dysfunctions of the choroid plexus and ependyma may have a role in the phenotype. 

Figure 1.

 

Targeted inactivation of Ccdc13 leads to congenital hydrocephalus. (A) Ccdc13 -targeting strategy (see Materials and Methods section). The Ccdc13 gene (Ccdc13⁺) has 16 exons, with protein codon starting at exon 2 (E2). A gene-trap vector comprising a Gfp reporter and a Pgk-neo cassette was placed between Ccdc13 E2 and 3 (Ccdc13^a) through homologous recombination in ES cells. The Frt and Loxp sites were arranged in such a way that allowed the conversion of the Ccdc13^a to null Ccdc13^b or conditional Ccdc13^c allele upon Flp and Cre excision, respectively. The restriction enzyme sites EcoNI and AhdI were used for Southern blotting analysis in (B) with 5′ and 3′ probes outside each targeting arm. SA, splicing acceptor; p(A), polyadenylation signal; GFP, green fluorescent protein; Pgk-neo, phosphoglycerate kinase promoter-driven neomycin gene; Flp, flippase; Frt, Flp recognition sequence. (B) Southern analysis of Ccdc13^a allele (Ccdc13^a) to confirm correct gene engineering. As expected, Southern blotting using 5′ probe with EcoNI digestion gave rise to 8.8 kb (for Ccdc13^+/+, +/+ for simplicity) and 13 kb (Ccdc13^a/+, a/+ for simplicity) bands, and using 3′ probe with AhdI digestion gave rise to 10.9 kb (for +/+) and 15.1 kb (for a/+). (C) Western blotting of P28 testis extracts using an anti-Ccdc13 antibody. (D) Representative wild type (+/+, top) and Ccdc13^b/b (b/b for simplicity) mice at P21 (bottom). The arrowhead points to the bumpy skull of a representative mutant mouse, and each grid of the ruler line is 1 cm. (E) A quantification of the body weight of “+/+” and “b/b” mice (n = 6; ***: P < 0.001, Student's t-test). (F) Survival curves with a total monitoring period of 60 days. There were 43 mice for each wild type and the mutant genotype. (G) H&E-stained sections showing the lateral ventricles at P1. (H) Gfp reporter (red) expression in Ccdc13^b/+ (b/+) ventricles and choroid plexus (CP; arrows) at P1 detected by an anti-GFP antibody. Arrowheads point to ventricles.

Enlarged Subarachnoid Space and Deformed Optic Nerve Head in the Ccdc13 Mutants

We next examined whether the hydrocephalus was due to obstruction of the flow path of the ventricular passages. Evans blue was injected into the lateral ventricle of the P3 wild type and mutant mice to monitor the CSF flow (Fig. 2A). We observed Evans blue in all four ventricles and spinal cord subarachnoid spaces (SaS) within 10 minutes after injection in both wild type and mutant mice (Fig. 2B). The mutant lateral ventricles appeared enlarged as expected on coronal vibratome sections (Fig. 2C). Thus, the results suggest the hydrocephalus manifestation in Ccdc13 mutant mice is not due to the obstruction of the CSF flow path. 

Figure 2.

 

Enlarged subarachnoid space of the ON sheath and deformed ONH in the Ccdc13 mutants. (A) Schematic illustration of the Evans blue injection into the mouse brain ventricles. (B, C), Cerebrospinal fluid (CSF) flow passages were not obstructed in P3 Ccdc13^b/b spinal cord and brain ventricles. White arrows in B indicate the subarachnoid space (SaS) of the spinal cord. (C) Note the enlarged lateral ventricles of the mutant brains. LV, lateral ventricle; 3rdV, third ventricle; Aq, aqueduct; 4thV, fourth ventricle. Arrowheads point to ventricles. (D) Representative pictures of the SaS of optic nerve sheath stained with Evans blue (red) at P14, P21, and P35. The white closed lines represent areas of the subarachnoid space/cavity (SaS). Micrographs were taken from sections of the optic nerve at about 400 µm posterior to the sclera. This distance was calculated by section thickness multiplied by numbers from the sclera position. ON, optic nerve. (E) Statistical analysis of the SaS areas with Student's t-test. Four consecutive sections from five optic nerves (total 20) of three mice for each wild type and the mutant genotype were counted for subarachnoid space areas. ns, nonsignificant; **, P < 0.1; *** P < 0.01. Error bars indicate mean ± SD. (F) Immunostaining of upNFH (green) and MBP (red) of P35 longitudinal eye sections through the optic disc. The optic nerve between choroid (c) and MBP (m) proximal margin is defined as “cmONH.” (G) Quantification of the height (h) and width (w) of the “cmNFH.” The cmNFH area is roughly a trapezoid, with the midline parallel to the upper and lower bases representing the width. There were six eyes from three mice. (H) P42 longitudinal eye sections stained for upNFH (green) and MBP (red). (I) Quantification of the height (h) and width (w) of the “cmNFH.” There were six eyes from three mice. Student's t-test, *, P < 0.5; **, P < 0.1; ***, P < 0.01. Error bars indicate mean ± SD.

The CSF enters the optic nerve through the subarachnoid space connected with the brain. The highly distended mutant ventricular space by the excess CSF might also occur in the CSF-containing subarachnoid space of the optic nerve sheath. Subarachnoid space varies markedly in different regions of the optic nerve. By tracing the injected fluorescent dye Evans blue, we compared multiple consecutive optic nerve cross sections at about 400 µm behind the globe (the optic nerve was myelinated at this region; see Figs. 2A, 2D). A significant increase of subarachnoid space in the optic nerve was found at the ages of P21 and P35 (see Fig. 2D), which was further confirmed by quantification (Fig. 2E). The distended subarachnoid space might also cause deformation at the adjacent ONH, as the enlarged ventricles did to the brain cortex. Thus, we examined a subdomain of the ONH between the choroid and proximal myelin boundary by performing immunolabeling of myelin basic protein (MBP) and unphosphorylated neurofilament (upNFH; Figs. 2F, 2H). Delineation of the choroid boundary was aided by upNFH staining, which also outlined choroid vessels (besides the nerve neurofilament) probably due to the use of secondary anti-mouse Ig antibody (see Figs. 2F, 2H). We measured the average height (h) and width (w) of this ONH subdomain as parameters to determine the mutant ONH deformation (see Materials and Methods, and Figs. 2F, 2H legends). Quantitative analysis demonstrated that the height of the mutant ONH was significantly lowered, whereas the width slightly expanded (Figs. 2G, 2I). The data together suggest a displacement of ONH in the Ccdc13 mutant mice. 

ONH Axonopathy in the Unmyelinated Portion of the Optic Nerve in Ccdc13^b/b Mutants

To investigate the consequence of an enlarged subarachnoid space and deformed ONH, we performed TEM to examine ultrastructural alterations of nonmyelinated ONH and myelinated portions of the optic nerve at P35 and P56. On TEM cross-sections of the P35 ONH, we found severe swelling of large numbers of axons with axoplasmic vacuoles (compare panels of Figs. 3A, 3B). Astroglia processes were also apparently increased (see Fig. 3A right panel). In contrast, the myelinated axons (immediately distal to the ONH) did not show apparent ultrastructural changes at this age (Figs. 3C, 3D). Axon density was not significantly altered either (Fig. 3E). We then examined the P56 optic nerve and found severely swollen and degenerating ONH axons compared with P35 (Figs. 3F, 3G). Degenerating axons were also seen in the myelinated portion at this age but were less severe than the ONH (Figs. 3H, 3I). In keeping with these findings, significant axon loss was observed (Fig. 3J). 

Figure 3.

 

Axonopathy of the nonmyelinated ONH and myelinated optic nerve in the Ccdc13^b/b mutants. (A–D) TEM imaging non-myelinated ONH on cross sections at P35. GP, glial processes. The boxed areas are magnified in B. B The arrowheads point to the dystrophic axons with large vacuoles. C TEM sections from a myelinated portion of the optic nerve. The boxed areas are magnified in D. D No obvious axon morphological changes were found in the myelinated portion of the P35 mutant nerves. (E) Quantification of axon density is presented as mean per 100 µm². Each data point represents axon density in one TEM optic nerve image as in C. A total of 15 images from 3 mice (5 each) were pulled together for quantification. Ns, nonsignificant, P > 0.05, Students’ t-test. (F–I) P56 optic nerve. Same arrangement as A to D. (J) Axon density quantification was performed as in E. ***, P < 0.001, Students’ t-test. (K) Wild type P56 TEM section. (L, M) Two mutant TEM mages with L at a similar location with K separating from M at about 1 mm. (N) A quantification of normal axons. Ten images from respective L and M locations of three nerves were quantified for the presence of normal axons; ***, P < 0.001, Students’ t-test.

To further elucidate whether a regional difference exists in axonopathy of the myelinated nerve, we compared TEM images immediately and 1 mm distal to ONH. Compared with the control (Fig. 3K), both areas showed axonal atrophy and degeneration (Figs. 3L, 3M), with the distal area being more severe both qualitatively and quantitively (Figs. 3L–N). A detailed look into pathological myelinated axons (Supplementary Fig. S2) showed several striking features, including aggregates of dark amorphous substances, swelling mitochondria with loss of cristae, and collapsing axolemma (see Supplementary Fig. S2E, S2F compared with S2D). Thus, the results suggest that ONH might be a primary site of mechanical stress, yet a distal-to-proximal progression of optic neuropathy may follow in the Ccdc13^b/b mutants. 

Distal-to-Proximal Progression of Neurofilament Alteration in the Ccdc13^b/b Mutants

Neurofilaments are pivotal to axon integrity and function. The upNFH is a sensitive indicator for axonal skeleton changes.²³ Thus, we first examined protein expression and localization of upNFH along with MBP in optic nerves from P35 to P56. Proximal-to-distal (refers to ONH) attenuation of upNFH staining was consistently observed in optic nerves of both control and mutant mice through their ages (Fig. 4). However, staining was diminished to a greater extent in the mutant compared to age-matched controls (see Figs. 4B, 4B1–B4; 4D, 4D1–D4; 4F, 4F1–F4). Accumulated upNFH puncta were observed in the mutant at P42 (see Fig. 4D, 4D1–D4 compared with Figs. 4C, 4C1–C4, white arrowheads). Myelin MBP also showed some puncta staining at this age of mutant nerves (see Figs. 4D2–D4, orange arrowheads, compared with Figs. 4C2–C4). Extensive swelling of axons and bead-like upNFH deposition advanced more proximally at P56, with relatively weaker expression still in the distal areas (see Fig. 4E, 4F; compare 4E1 and 4E2, with 4F1 and 4F2, respectively). 

Figure 4.

 

Distal-to-proximal alterations of unphosphorylated neurofilaments (upNFH) expression and localization. (A, B) Immunofluorescence of unphosphorylated neurofilaments (upNFH, green) and myelin basic protein (MBP, red) at P35. “1, 2, 3, and 4” denoted four imaging areas from proximal to distal optic nerves of wild type A and mutant B mice, which are magnified in (A1–4) and (B1–4), respectively. Dashed lines demarcate the boundary between non-myelinated (nM) and myelinated (M) areas. Attenuated upNFH staining was observed in distal optic nerves of mutant mice (B3, B4). (C, D) P42 optic nerve sections. (C1-4, D1-4) are magnified from the PMSF four boxed regions from C and D, respectively. The white arrowheads denote accumulated upNFH dots, whereas the orange arrowheads denote accumulated MBP. (E, F) P56 optic nerve sections stained for upNFH. Two imaging areas (1 and 2) were demonstrated as (E1, E2) and (F1, F2), respectively, for wild type and mutant nerves. Note the swollen axons and upNFH punctate deposition in F1 (white arrowheads) and the severely weakened upNFH expression in (F2).

Next, we examined whether the phosphorylated form of neurofilaments (pNFH) and neural-specific tubulin beta III (Tubb3) also changed in the Ccdc13 mutant nerves. Surprisingly, neither pNFH nor Tubb3 showed notable differences from the wild type at P35 or P42 (Supplementary Figs. S3A–S3L). At P56, when accumulated MBP deposition was observed (see Supplementary Figs. S3R, S3S), the mutant pNFH staining pattern remained comparable to that of the control (compare Supplementary Figs. S3M with S3Q). Thus, the unphosphorylated but not the phosphorylated NFH alteration showed a distal-to-proximal progression in Ccdc13^b/b mutant mice. 

Early Loss of RGC Unphosphorylated NFH and Malfunction in the Ccdc13^b/b Retina

We postulated that upNFH expression would also decline in intraretinal RGC axons. To determine if this was the case, P21, P35, and P56 retinas were stained for Rbpms (labeling RGCs) and upNFH proteins. Four concentric retinal areas were imaged and compared (Fig. 5F). In all ages examined, no significant difference in RGC numbers (indicated by Rbpms) was found between controls and the mutants (Figs. 5A–D, Supplementary Fig. S4A–S4G). In contrast, decreased upNFH was consistently found at all examined ages and areas (see Figs. 5A–D, Supplementary Figs. S4A–S4F). At P56, axon retraction bulbs were additionally observed with upNFH deposition (Fig. 5E). Western blotting analysis of P26 retinas further confirmed reduced upNFH but not pNFH proteins (Figs. 5G, 5H), along with a reduced level of NFH total mRNA (Fig. 5I). Consistent with the previous results of immunostaining, neither Rbpms nor Tubb3 was altered (Figs. 5G, 5J). 

Figure 5.

 

RGC axon integrity and function in the Ccdc13^b/b retina. (A–F) Immunostained flat mount retinas with anti-upNFH (green) and anti-Rbpms (red) antibodies. The retinas were divided into four concentric areas, and pictures were roughly taken from the four numbered areas of each genotype, as shown in F. Arrows in D point to axon retracting bulbs. Boxed areas were magnified in E. F Representative flat mount retinas stained with upNFH and Rbpms antibodies at P56. Numbers indicate imaged areas corresponding to A to D. (G) Western Blotting detection of retinal expression of upNFH, pNFH, Rbpms, Tubb3, and β-actin at P26. (H) Quantification of Western Blot signal of upNFH and pNFH proteins. There were four eyes from three mice. (I) RT-qPCR detection of total NFH mRNA. There were six eyes from three mice. (J) Quantification of Western Blot signal of Rbpms and Tubb3 proteins. There were four eyes from three mice. Fold change is relative to wild type control. Each data point is from retinal preparation from one eye. ***, P < 0.001, Students’ t-test. (K–M) pERG waveforms of P21 K, P35 L, and P56 M retinas. P1, positive 1; N2, negative 2. (N) Quantification of amplitude differences from P1 to N2. P21: n = 10 eyes from 5 mice for each genotype; P35: n = 12 eyes from 6 mice for wild type control (+/+), and n = 9 eyes from 5 mice for mutants (b/b); P56: n = 10 eyes from 5 mice for wild type control (+/+), and n = 9 eyes from 5 mice for mutants (b/b). Statistical powers were detected by the Student's t-test. *, P < 0.05; *** P < 0.001. Error bars indicate mean ± SD.

To further correlate NFH phosphorylation with the RGC function, we performed pERG analysis. The pERG P1 and N2 waves of the Ccdc13 mutant mice were significantly decreased from P21 to P56 (Figs. 5K–N). Taken together, these results suggest that altered upNFH expression and deposition are an early indicator of aberrant RGC and optic nerve function. 

Hampered Axon Transport of the Mutant Optic Nerve

The swollen ONH axons in the mutants suggest axoplasmic flow may be impeded. Thus, we examined the axonal transport by intravitreal injection of a CTB-conjugated fluorescent dye (Fig. 6A). We first examined the CTB transport along the optic tract 2 weeks after injection at P56. In flat mount wild type retinas, CTB staining was only weakly present in the retinal RGC layer (indicated by Rbpms; Figs. 6C–D). In contrast, more CTB was accumulated in the mutant retinas (Figs. 6F–H). Longitudinal sections through the optic nerves found the CTB puncta similar to that previously observed for NFH (Fig. 6J compared with 6I). Punctate CTB also appeared at the optic chiasm and accumulated more in the mutant optic tract (Fig. 6L compared with 6K). CTB staining in the lateral geniculate nucleus (LGN) and superior colliculus (SC) is comparable between wild type and mutant mice (Figs. 6M–P), except that the CTB in the mutant SC was generally weaker (Figs. 6O, 6P). 

Figure 6.

 

Impeded anterograde axon transport in the Ccdc13^b/b mutants. (A) Schematic illustration of CSF pathway_, intravitreal injection of CTB, and sectioned planes for (I–P). CBT was observed 2 weeks after intravitreal injection at P56. (B) The square box in the drawn flat mount retina represents the imaging areas in (C–H). C to H CTB (green) and Rbpms (red) staining showing intraretinal retention of CTB fluorescent dye and RGC density in flat mount retinas. OD, optic disc. (I, J) Obstructed CTB transport (grey, fluorescent signal) along the optic nerve indicated by the dotted accumulation. (K–P) CTB (red) in preoptic, thalamic, and midbrain coronal sections. K to L OC, optic chiasm; SCN, superior optic nucleus; 3v, third ventricle. Arrowheads point to OC. Note the dotted CTB staining of the optic nerve at the OC location. (M, N) Thalamic dorsal (d) and ventral (v) lateral geniculate nucleus (LGN). (O, P) Midbrain superior colliculus (SC). Note the attenuated CTB fluorescent intensity of the mutant SC. (Q, R) CTB (green) transport was observed 48 hours after intravitreal injection. Three consecutive sections through the optic disc region are shown for the control (+/+) and the mutant mice. Arrows point to the approximate transport front of the CTB 48 hours after injection.

We next performed anterograde CTB tracing for a shorter duration to visualize its transport rate. CTB was intravitreally injected at P51 and examined after 48 hours of injection when most of the dye was passing through the optic disc and ONH areas (Figs. 6Q, 6R). The CTB traveled distance was consistently shorter in the mutant nerve fibers compared with the wild type controls (Figs. 6Q, 6R, 3 consecutive sections shown for each genotype). Whereas severe hydrocephalus in the mutant mice prevented us from performing retrograde tracing, the above data combined suggests a generally impaired axonal transport. 

Activation of Optic Nerve Astrocytes and Microglia in Ccdc13^b/b Mice

Optic nerve glial components, including astrocytes, microglia, and oligodendrocytes, respond to various injuries.^28–32 GFAP, specifically labeling astroglia, was broadly enhanced in the mutant nerves at the examined ages of P35 and P56 (Supplementary Figs. S5A–S5F). The number of microglia/macrophages, as indicated by Iba1 labeling, appeared to be increased at P35 (see Supplementary Figs. S5G–S5I), although not reaching a level of statistical significance (P > 0.05; see Supplementary Fig. S5I). However, more ameboid microglia/macrophages with thickened processes appeared in the mutant nerves at P56 compared with the wild type (see Supplementary Figs. S5J–S5L). In contrast to the astro- and microglia, the oligodendrocytes whose survival and development depend on RGC axons^33,34 did not show significant changes despite attenuated Olig2 expression in many cells (see Supplementary Figs. S5M–S5R). Furthermore, the glial nuclei normally aligned to the axonal fascicles were disorganized in the mutant nerves revealed by H&E histology (see Supplementary Figs. S5S, S5T). These observations are considered a corollary of optic neuropathy under various injury conditions. 

The Optic Nerve Injury is Nonautonomous and Independent of the Retina

The above findings hinted at a hydrocephalus-associated optic nerve injury. However, possibilities exist such that Ccdc13 could function in the retina and underlie the injury, thus confounding the interpretation of our results. Hence, we investigated Ccdc13 expression at several time points during retinal development. RNA-sequencing analysis revealed insignificant Ccdc13 RNA reads from P5 to P15 (Supplementary Fig. S6A). Additionally, using the knock-in Gfp reporter surrogating for the Ccdc13 expression, very weak GFP staining was found in the outer neural blast layer (ONBL) of P1 retina, and barely in any cells of P28 retina (Supplementary Fig. S6B). Furthermore, no staining was detected at any examined ages of the optic nerves (Supplementary Fig. S6C, representative images from P56). A measurement of the IOP at postnatal ages from P21 to P56 did not show significant changes either (Supplementary Fig. S6D). 

To functionally exclude possible retinal contributions to the nerve atrophy, we performed full-field ERG to examine the photoreceptors’ function. As expected, neither photoreceptor “a” nor “b” wave was altered under scotopic or photopic conditions (Supplementary Figs. S7A–S7D). Additionally, pERG to measure RGC function in a mouse line with retina-specific deletion of Ccdc13 at P35 (see Materials and Methods section) did not show any alterations (Supplementary Fig. S7E). The results point to a nonautonomous optic nerve atrophy caused by hydrocephalus without primary injury to the retina. 

Discussion

The current study characterized hydrocephalus-associated optic neuropathy in genetically engineered mice with disruption of the Ccdc13 gene. Optic neuropathy features include physically displaced ONH with swelling axons, impeded axoplasmic transport, compromised RGC function with downregulation of NFH expression, and a distal-to-proximal progressing axonopathy of myelinated nerves. Related findings are the activation of optic nerve astrocytes and microglia. Many of these features are also present in primate models of chronic intracranial hypertension (increased ICP) and the DBA/2J mutant mice with elevated IOP.^19,35,36 In general, our findings combined with those reported by others suggest that altered translaminar pressure gradients at the ONH likely underlie pressure-related RGC/optic nerve pathophysiology, regardless of elevation or reduction in IOP or ICP. This will be discussed below. 

The ONH is vulnerable to perturbations in IOP and ICP in experimental glaucoma animal models and in patients with glaucoma.^{13–18,23,35} The subarachnoid space adjacent to the ONH is hydrostatically continuous with brain ventricles,^15,16 representing a CSF reservoir at the optic terminus. Elevated pressure in this portion of the CSF space due to the increased hydrocephalic ICP may strangle the underneath nerves, leading to the obstruction of axoplasmic flow and axon swelling.¹⁹ Meanwhile, nonmyelinated ONH shall be more sensitive to mechanical embarrassment than the ensuing myelinated optic nerve, thus increasing the injury risk. In line with these notions, the ONH in the Ccdc13 mutant develops earlier and more severe axonopathy than the trailing myelinated portion of the nerve based on TEM observation. The excessively distended CSF space at the optic terminus and the displaced ONH with a shortened length and expanded diameter in the Ccdc13 mutants further support mechanical stress as being the cause of optic neuropathy at ONH. Consistent with this, a positive correlation between optic nerve sheath diameter and ICP has also been reported in patients with hydrocephalus.³⁷ Thus, our study provided the first genetic disease model showing that ONH could be a major point of vulnerability to translaminar pressure perturbations, given the distended CSF space at the optic terminus caused by the hydrocephalic ICP. Notably, Ccdc13 was also reported as a novel genetic locus for primary open-angle glaucoma very recently.²⁵ 

It remains undetermined regarding the initiation or progression of “pressure-related” optic neuropathy. Although axonopathy is first observed at the ONH, late-developing neuropathy in the myelinated region of the mutant optic nerve appears distal to the ONH and progresses proximally toward it — both NFH immunostaining and the TEM ultrastructural analysis demonstrate such a pattern of progression. At first glance, this seems counterintuitive given the earlier involvement of ONH, and neuropathy would be expected to propagate distally from it. However, this may be understood in the context of impeded axonal transport, under which materials flow along the axons would decline more distantly from the injured site.³⁶ Thus, our data support a distal-to-proximal progression of optic neuropathy initiated at the ONH in the Ccdc13 mutant mice. 

The NFH puncta in the Ccdc13 mutant nerves have also been observed in a rat glaucoma model.²³ These aggregates may have originated from broken-down neurofilaments consisting of the upNFH due to impaired axonal transport and reduced NFH mRNA synthesis. On the other hand, the pNFH is largely intact in terms of protein levels and localization, which might be explained by the protein turnover rate, with the pNFH being more stable and long-lived.^38–44 Nonetheless, it would be reasonable to predict that pNFH will eventually be affected and further impact axon transport in the long term. 

It is interesting to note that reduced RGC NFH expression in the Ccdc13 mutants occurs as early as the reduced pERG. In contrast, RGC expression of Rbpms remains largely unaffected in the mutant. Additionally, microtubule protein expression and structure appear grossly normal in the Ccdc13 mutant axons, as indicated by Tubb3 expression and immunostaining. These observations combined suggest that NFH is an early sensitive indicator of RGC and axonal damages. 

Sectorial RGC loss is reported by many studies of ocular hypertension glaucoma, especially in the DBA/2J mice.^{35,36,45–48} However, this was not seen in the hydrocephalic Ccdc13 mutant, which has normal retinotopic patterns in the superior colliculus, as indicated by CTB. One likely explanation is that the short lifespan of the Ccdc13 mutant mice does not leave enough time for a full-blown RGC degeneration to become manifest. Additionally, the severe hydrocephalus conditions in the Ccdc13 mutant mice prevented us from injecting tracing dyes into the SC or LGN to assess retrograde transport directly. Nevertheless, we deem it reasonable to infer that axon transport is impaired bidirectionally, given the confirmed impediment to anterograde transport. 

Hydrocephalus is typically associated with increased ICP.^3,20 Although we were not able to measure the ICP or the CSF pressure directly at the ONH site due to the technical difficulties of performing such experiments on small animals, such as mice, the unblocked CSF path and distended subarachnoid space of the mutant optic nerve head leaves little theoretical room for not to be the case. Further supporting evidence for elevated ICP is also from the positive correlation of the optic nerve sheath diameter (reflected by the current study) with ICP consistently reported by several clinical studies.^37,49 Future technical development may allow us to directly measure ICP, particularly at the optic nerve terminus. 

Finally, several lines of evidence suggest that the optic nerve neuropathy of the Ccdc13 mutant is non-cell autonomous. These include barely detectable Ccdc13 expression in the retina, optic nerve, or vessels, specific Ccdc13 expression in the choroid plexus, and normal retinal ERG function and IOP of the Ccdc13 mutants. Nevertheless, we acknowledge that there are possibilities that Ccdc13 might also function in other tissues to affect the optic nerve, which is currently not found and requires further attention and examination. 

We propose a model of pressure-related optic neuropathy under different scenarios of altered translaminar pressure: (1) elevated IOP would press the laminar cribrosa outward, leading to a “positive cupping” (Fig. 7B); (2) conversely, abnormally increased ICP would lead to a “negative cupping” (Fig. 7C). The molecular events of optic neuropathy observed in this study are summarized in Figure 7D. Notably, despite the mouse ONH having a glial lamina instead of the primate lamina cribrosa with an extracellular matrix plate,^35,50 the ONH retains the same vulnerability to mechanical stress produced by an imbalance of IOP and ICP that acts across it. 

Figure 7.

 

A hypothetical optic cupping model under different scenarios of altered translaminar pressure. (A) Under normal CSF and IOP, translaminar pressure is maintained within a normal range with balancing the two forces from the intraocular and the optic nerve head. (B) Elevated IOP would press the laminar cribrosa outward leading to a “positive cupping.” (C) Abnormally increased ICP would lead to a “negative cupping.” (D) A summary of molecular events of optic neuropathy observed in the Ccdc13 mutant mice. IOP, intraocular pressure; CSF, cerebrospinal fluid; TP, translaminar pressure; CSFP, CSF pressure; ICP, intracranial pressure; GL, glial lamina. Asterisks indicate the optic disc cupping; the red lines represent lamina cribrosa. Up and down arrows indicate increase and decrease, respectively.

Acknowledgments

The authors thank Rong Ju from the State Key Laboratory of Ophthalmology at Zhongshan Ophthalmic Center for providing valuable suggestions in the preparation of the manuscript and giving helpful comments. 

Supported by the National Natural Science Foundation of China (NSFC: 32371015), Guangdong Provincial Natural Science Foundation (2022A1515012515), Guangzhou City-University Joint Foundation (202201020275), and Research funding from State Key Laboratory of Ophthalmology at Zhongshan Ophthalmic Center (303060202400373) to Chunqiao Liu, and National Science Foundation Youth 598 Program (32000553) to Dianlei Guo. 

Disclosure: M. Wu, None; X. Zhao, None; S. Peng, None; X. Zhang, None; J. Ru, None; L. Xie, None; T. Wen, None; Y. Su, None; S. Xu, None; D. Guo, None; J. Hu, None; H. Lin, None; T. Li, None; C. Liu, None