|Home | About | Journals | Submit | Contact Us | Français|
Salla disease and infantile sialic acid storage disease are autosomal recessive lysosomal storage disorders caused by mutations in the gene encoding sialin, a membrane protein that transports free sialic acid out of the lysosome after it is cleaved from sialoglycoconjugates undergoing degradation. Accumulation of sialic acid in lysosomes defines these disorders and the clinical phenotype is characterized by neurodevelopmental defects, including severe central nervous system (CNS) hypomyelination. In this study we used a sialin deficient mouse to address how loss of sialin leads to the defect in myelination. Behavioral analysis of the sialin-/- mouse demonstrates poor coordination, seizures and premature death. Analysis by histology, electron microscopy and Western blotting, reveals a decrease in myelination of the CNS, but normal neuronal cytoarchitecture and normal myelination of the peripheral nervous system. To investigate potential mechanisms underlying CNS hypomyelination, we studied myelination and oligodendrocyte development in optic nerves. We found reduced numbers of myelinated axons in optic nerves from sialin-/- mice, but grossly normal appearing myelin on the axons that were myelinated. Migration and density of oligodendrocyte precursor cells were normal; however, there was a marked decrease in the number of postmitotic oligodendrocytes and an associated increase in the number of apoptotic cells during the later stages of myelinogenesis. These findings suggest that a defect in maturation of cells in the oligodendrocyte lineage leads to increased apoptosis and underlies the myelination defect associated with sialin loss.
Sialic acids are amino sugars that play an important role in nervous system development and function. As negatively charged terminal residues of glycan chains, sialic acids have been implicated in electrostatic-based intermolecular interactions that mediate cell-cell recognition, cell adhesion and intercellular signaling (Vyas and Schnaar, 2001; Sampathkumar et al., 2006). Modulation of sialic acid content in glycoproteins and glycolipids (gangliosides) is crucial for normal neurodevelopment, and requires tightly regulated expression and efficient down-regulation of sialic acid containing macromolecules (Rosner, 2003).
A primary pathway for catabolism of sialoglycoconjugates is lysosomal degradation. Once these macromolecules are trafficked to the lysosome, sialic acid residues are sequentially removed through hydrolysis of the terminal glycosidic linkages by acid sialidases (neuraminidases). The liberated free sialic acid is then exported from the lysosome through the sialic acid transporter, sialin. Mutations in the gene encoding sialin lead to the recessive allelic lysosomal storage disorders, Salla disease and infantile sialic acid storage disease (ISSD) (Verheijen et al., 1999). Biochemical studies have shown a direct correlation between sialin transport activity and severity of disease phenotype (Morin et al., 2004; Wreden et al., 2005; Myall et al., 2007; Ruivo et al., 2008). Mutations that produce a functional but less active transporter, as found in Salla disease, show a less severe phenotype than mutations with complete loss of function, typical of ISSD (Aula et al., 2000).
In both Salla disease and ISSD the nervous system is predominantly affected with varying degrees of developmental delay in motor and cognitive skills, epilepsy and premature death. The small number of neuropathology studies have consistently identified cytoplasmic vacuoles typical of lysosomal storage disorders and hypomyelination as prominent features (Autio-Harmainen et al., 1988; Pueschel et al., 1988; Mancini et al., 1991; Lemyre et al., 1999). Clinical imaging studies also indicate a defect in white matter formation (Haataja et al., 1994; Morse et al., 2005). However, the limited number and descriptive nature of these studies leave many unanswered questions regarding the progression of cellular and molecular pathophysiology associated with the loss of sialin.
To identify potential mechanisms underlying the pathology of these disorders, we have characterized a sialin deficient mouse. Through behavioral and neuropathological analyses, we show that the sialin-/- mouse strain has a phenotype consistent with the free sialic acid storage disorders. Our observations reveal poor coordination, seizures, a failure to thrive, and premature death associated with loss of sialin expression. In addition to prominent vacuolar lesions, our histological characterization demonstrates a marked decrease in myelin throughout the central nervous system with normal appearing myelin in the peripheral nervous system. Using ultrastructural and molecular characterization of myelinogenesis in the sialin-/- mice, we further find that there is normal migration and proliferation of oligodendrocyte precursor cells (OPCs), but a reduction in mature myelin producing oligodendrocytes that is likely a consequence of oligodendrocyte lineage apoptosis. Finally, we find a delay in the developmentally regulated reduction in expression of polysialylated neural cell adhesion molecule, PSA-NCAM, providing a potential molecular mechanism for the impaired myelination and reduction in oligodendrocyte number.
All experimental procedures were approved by the Stanford Institutional Animal Care and Use Committee. Three male mice heterozygous for the sialin gene (B6; 129S5-Slc17a5tm1Lex) were obtained from the Mutant Mouse Regional Resource Centers (MMRRC; mmrrc.org). These mice were originally generated from 129S5/SvEvBrd-derived embryonic stem cells by Lexicon Genetics Incorporated through use of targeted homologous recombination. Specifically, the 104 nucleotides beginning immediately after the first nucleotide of the coding sequence were replaced with an IRES domain followed by a sequence coding for a beta-galactosidase-neomycin fusion protein. Subsequent to procurement from the MMRRC heterozygous male mice were crossed to C57BL(Thy1.2) female mice and the colony was maintained by sequentially crossing two generations of mice heterozygous for the mutation in sialin for every cross out to C57BL females.
For PCR genotyping sense and anti-sense oligonucleotide primers, their location, and predicted fragment lengths were as follows: Sialin knockout allele: 5’-GCAGCGCATCGCCTTCTATC -3’ and 5’-GCTAAGCGGAACCTGGCG -3’; 450 bps; wild-type sialin allele: 5’-GCTGGTGACACACATCTTGC -3’ and 5’-CCGCTTCGGTCTGCCGG -3’; 322 bps.
Total RNA was isolated using TRIzol (Invitrogen) or PureLink Micro-to-Midi Total RNA Purification System (Invitrogen), complementary DNA templates were prepared from 5-7ug of total RNA using random primers (Amersham, Buckinghamshire, UK) and 200 units of SuperScript II reverse transcriptase (Invitrogen Corporation, Carlsbad, CA) according to the instructions of the manufacturers. Sense and anti-sense oligonucleotide primer pairs used for PCR amplification of sialin cDNA fragments and the predicted product sizes were as follows: Exon 1 - Exon 4 (5’-AAACGACGATGAGGAGAGCTC-3’ and 5’-GCGTGCATAGCTGGAAACGT-3’; 521 bps); Exon 5 - Exon 11 (5’-CTGGACTTACGTCTTCTATC-3’ and 5’- GATACAGAAGACAGTCTGCC-3’; 711 bps); Exon 6 - Exon 11 (5’- ACTCACAAGACAATCTCCCA-3’ and 5’-TCAGTTTCTGTGTCCGTGGT-3’; 728 bps); Exon 10 - Exon 11 (5’-GTATGCTGGCATCCTCTTGG-3’ and 5’- GATACAGAAGACAGTCTGCC-3’; 126 bps); Exon 11 (5’- TGGCAGACTGTCTTCTGTAT-3’ and 5’- TCAGTTTCTGTGTCCGTGGT-3’; 103 bps). For transferrin receptor cDNA fragment amplification the sense oligonucleotide was 5’-TGGGAACAGGTCTTCTGTTG-3’, the antisense was 5’-TGCAGTCCAGCTGGCAAAGA-3’, and the predicted product size was 120 bps.
Hindpaws and forepaws of three week old mice were dipped into blue ink and red ink respectively and the mice were placed at one end of a cardboard tube, (7.6 cm diameter x 93.3 cm length) with a clean sheet of white paper placed on the floor to record the footprints. The end where the mice were placed was covered, the other end was left uncovered, and the mice were allowed to walk freely toward the open end. The paper was removed and the average stride length and variability of stride length were determined based on the distances between sequential left hindprints measured over a 25 cm segment of the paper. Coefficient of variation (CV) was calculated by normalizing the variance in stride lengths to the mean for each animal analyzed.
Mice were anesthetized with isoflurane and rapidly decapitated. The brain, optic nerves, cervical spinal cord, and sciatic nerves were dissected out and placed in ice cold fixative (2% paraformaldehyde / 3% glutaraldehyde / 0.1 M sodium cacodylate / 0.05% CaCl2). Within three hours the tissue was cut into 3 mm sections, fixed overnight at 4°C in the same fixative and then washed with 0.1 M cacodylate. The tissue was incubated with 2% OsO4 for 2 hrs at room temperature, washed with water, progressively dehydrated in ethanol:water mixtures and then embedded in Epon resin. Sections were stained with toluidine blue for light microscopy evaluation. For transmission electron microscopy, ultrathin sections (50 nm) were stained with 4% uranyl acetate, then 2.5% lead nitrate for five minutes each at room temperature. The sections were observed and images captured using a JEOL 1010 transmission electron microscope. Bright-field images were taken with a Nikon Eclipse E1000 equipped with a Diagnostic Instruments digital camera.
Tissue was harvested as above, immediately frozen on dry ice and stored at -80°C for later use. Subsequently samples were placed on ice and brought up in phosphate buffer saline (PBS) containing protease inhibitors (in μg/ml: 2 aprotinin, 1 leupeptin, 2 antipain, 10 benzamidine, 35 phenylmethanesulfonyl fluoride, 1 chymostatin, 1 pepstatin) and 1mM EDTA. The tissue was minced with scissors, homogenized and sonicated. Samples (2-20 μg) were subjected to SDS-PAGE and Western blotting. Primary antibodies were used as follows: rat anti-myelin basic protein (1:2000, Chemicon), mouse monoclonal anti-beta-actin (1:10000, Sigma), mouse monoclonal anti-neurofilament-86 (1:4000, Sigma), and mouse anti-polysialic acid-NCAM clone 2-2B (1:6000, Chemicon). HRP-conjugated secondary antibodies (Pierce) were used at 1:10,000. Proteins were detected using an ECL Western blotting detection system (Amersham Biosciences) and exposure of the blot to autoradiography film (Midsci). Scanned images of the films were generated and band intensities were measured using ImageJ software.
Optic nerves and brains were dissected out and fixed with ice-cold 4% paraformaldehyde in PBS overnight at 4°C and then cryoprotected in 30% sucrose in PBS. Optic nerves were embedded in OCT (Tissue-TEK) and cut into 10 μm sections using a cryostat. Brains were cut into 25 μm or 40 μm sections using a freezing microtome. Sections were stained with cresyl violet or subjected to immunostaining. For immunostaining tissue sections (free floating or mounted on slides) were blocked and permeabilized in 5% BSA/ 3% horse serum/ 0.2% Triton X-100 followed by overnight incubation at 4°C with the indicated primary antibodies. Dilutions for primary antibodies were as follows: rabbit anti-Olig2 (1:500, Chemicon), rabbit anti-NG2 (1:500, Chemicon), mouse anti-APC/CC1 (1:500, EMD Biosciences), rat anti-MBP (1:200, Chemicon), mouse anti-NF68 (1:400, Sigma), rabbit anti-cleaved caspase-3 (1:1000, Cell Signaling Technology), rabbit anti-sialin (1:3600, Alpha Diagnostic), and mouse anti-Caspr/paranodin clone K65/35 (1:200, NeuroMab). Primary antibodies were detected with Alexa Fluor dye conjugated (1:1000, Molecular Probes) or Rhodamine Red-X conjugated (1:1000, Jackson ImmunoResearch) secondary antibodies. Following antibody incubations, some optic nerves were stained with FluoroMyelin Red Fluorescent Myelin Stain (1:300, Molecular Probes) for 20 minutes at room temperature. For samples in which total cell counts were determined, nuclei were counterstained for 5 minutes with 100 nM 4’,6’-diamidino-2-phenylindole dihydrochloride (DAPI) (Molecular Probes). Coverslips were mounted with MOWIOL anti-fading medium. Confocal images were taken with a Leica TCS SPE Spectral confocal microscope, epifluorescence images with a Nikon Eclipse E800 microscope equipped with a Nikon digital camera, and bright-field images with a Nikon Eclipse E1000 equipped with a Diagnostic Instruments digital camera.
All cell counts were done blind to genotype by an investigator not involved in sample preparation. Two high power (63x objective) images were taken for each optic nerve and the number of cells expressing the marker of interest in each image was determined by manual counting. Cell migration was assessed by counting the number of Olig2+ cells in separate images taken at the chiasmal, middle and retinal segments of the optic nerve. To determine the number of apoptotic cells per mm2, 10 μm thick longitudinal optic nerve sections immunonstained for activated caspase-3 were viewed through a 20x objective on an epifluorescence microscope and all cells expressing the antigen were manually counted. Bright-field images were then taken of the entire nerve at 4x magnification using Spot Advanced software and the total area was determined using ImageJ software.
To count the number and determine the length of individual myelin segments, z-series of confocal images (taken at 1 μm steps) were analyzed. Image J software was used for length measurements.
Data were expressed as mean ± SEM. At least three pairs of sialin-/- and control littermates were used for each experiment. All groups were compared using two-tailed unpaired t test.
While the biochemistry of sialin and the clinical picture of the free sialic acid storage disorders are well described, a mechanistic link from sialin function to the clinical phenotype is lacking. To address this issue, we analyzed a sialin deficient mouse (http://www.informatics.jax.org/external/ko/lexicon/2361.html). These mice were generated using standard homologous recombination to replace the first coding exon of the sialin gene with an IRES-β-gal-neo gene (Fig. 1A). We obtained heterozygous male mice from the Mutant Mouse Regional Resource Centers and established our own breeding colony. The birth rates of wild-type, heterozygous and homozygous mutant animals from heterozygous crosses (29:46:26, n=203 animals from 28 litters) were consistent with Mendelian distributions, implying that there is no in utero lethality associated with complete sialin deficiency.
Sialin is encoded by 11 exons with some suggestion of variable splicing (Verheijen et al., 1999). Since only the first exon was deleted, we sought to determine whether an alternatively spliced isoform of sialin is expressed in sialin-/- mice. We analyzed sialin mRNA expression by RT-PCR using oligonucleotide primers derived from several different exon pairs. No sialin transcript was detected in the sialin-/- mice and a level approximately half of that in wild-type was present in the heterozygous mice (Fig. 1B). The absence of sialin expression in the sialin-/- mice was also confirmed by immunohistochemical analysis. Immunostained coronal brain sections of heterozygous mice show sialin immunoreactivity in the granule cell layer and hilar neurons of the dentate gyrus that is not present in the sialin-/- mouse hippocampus (Fig. 1C).
As early as postnatal day 3 (P3), sialin-/- mice could be identified by their smaller size and underdeveloped features. Sialin-/- mice failed to increase in size (Fig. 1D), developed a severe tremor and uncoordinated gait, appeared weak, and typically died during the third postnatal week. Throughout their observed lifespan, wild-type and heterozygous mice were grossly indistinguishable and were grouped together as controls for all analyses. To quantify gait abnormalities in the sialin-/- mice we analyzed their footprint pattern as they walked down a cylindrical tube (Fig. 1E). The sialin-/- mice tended to stay at the entrance of the tube and took longer than control littermates to walk the length of the tube. The stride length for the sialin-/- mice was, on average, approximately two-thirds that of their littermate controls and had greater variability. During the footprint analysis studies, handling-induced tonic-clonic seizures were observed in the sialin-/- animals, but never in littermate controls, consistent with the increased incidence of epilepsy in patients with the free sialic acid storage disorders (Varho et al., 2002).
Neuropathological studies of tissue from Salla disease and ISSD patients have identified widespread neuronal storage, axonal spheroids, myelin loss and cerebellar Purkinje cell loss (Autio-Harmainen et al., 1988; Pueschel et al., 1988; Mancini et al., 1991; Lemyre et al., 1999). If the sialin-/- mouse is an appropriate model for the human disorders, then similar findings should be present in these animals. On gross examination, the brains of the sialin-/- mice were notably smaller, show decreased brainstem bulk and have thinner optic nerves than control littermates (Fig. 2A). Light microscopic examination of cresyl violet stained sections from the forebrain of P21 mice demonstrated normal neuronal cytoarchitecture including neocortical and hippocampal lamination, but reduced numbers of cells in the corpus callosum of sialin-/- mice compared to control littermates (Fig. 2B). Prominent clear cytoplasmic structures consistent with vacuoles were evident in neurons of the cerebellum and spinal cord of sialin-/- mice (Fig. S1A). No such structures were found in tissue from the wild-type or heterozygous animals.
To define further the histological abnormalities we used electron microscopy. In addition to the neuronal vacuolization (Fig. S1B), a reduction in the density of axons that are myelinated in the ventral white matter of the spinal cord and in the optic nerve was evident (Fig. 2C). The myelin structures that were present in the tissue from the sialin-/- mice were relatively normal in appearance. In the sciatic nerve, myelin density and structure were similar in control and sialin-/- mice. Ultrastructural examination of the optic nerve also showed abnormal swellings containing electron dense material, typical of axonal spheroids, in both myelinated and unmyelinated axons (Fig. S1C). Similar pathological findings were seen in cerebellar and spinal cord axons.
To investigate further the myelination defect in sialin-/- mice, we examined the expression of myelin basic protein (MBP), a major structural protein of central and peripheral myelin, using quantitative Western blotting and immunostaining. Consistent with the histological analysis, the expression level of MBP was similar in the sciatic nerves from control and sialin-/- mice (Fig. 3A and 3B), while MBP levels in sialin-/- mouse cervical spinal cord samples were less than half those from control animals (p ≤ 0.001). The relative reduction in MBP expression in brain samples from sialin-/- mice was even greater with levels reaching only ~10% of controls (p ≤ 0.01). In contrast, levels of NF68, an axonal protein, were similar in control and sialin-/- animals in the peripheral and central nervous system (spinal cord and brain). These results suggest that a reduced level of myelin is specific to the CNS and is not secondary to axonal loss.
To determine whether there is regional variability to the myelination defect, we immunolabeled coronal brain sections with an antibody against MBP (Fig. 3C). While dense MBP staining was seen in all white matter structures of the control brain, we found a near complete absence of MBP immunofluorescence in the brains of sialin-/- mice. The MBP staining that was present in the sections from the sialin-/- mice occurred as isolated clusters of brightly stained elongated structures that appeared to originate from single cells. These myelin segments were present in the corpus callosum, deep layers of the cortex, more dorsal aspects of striatum and the lateral olfactory tracts. Immunolabeling for NF68 showed the axons were grossly intact in the sialin-/- brain (Fig. 3D), again suggesting minimal axonal loss in the sialin-/- mice.
To evaluate more fully the oligodendrocytes in the sialin-/- mice we assessed whether the myelin segments formed by sialin-/- oligodendrocytes were of normal number and length. We found number of myelin segments originating from individual MBP+ cell bodies (range of 18-41) in the striatum (Fig. S2A) and lengths of myelin segments (range of ~70-200 μm) labeled by MBP in the corpus callosum (Fig. S2B) to be consistent with published values for the CNS (Butt et al., 1994; Bjartmar, 1996; Murtie et al., 2007).
The optic nerve is a discrete central nervous system white matter tract in which all axons are myelinated in orderly and well characterized stages (Miller, 2002; Raff, 2007). Since loss of sialin has a profound effect on optic nerve myelination, we anticipated that examining myelin formation in this structure in sialin-/- mice might provide insight into underlying cellular and molecular pathophysiological mechanisms. As a first step, we analyzed the time course of optic nerve myelination. Myelination of the mouse optic nerve starts at around P7 with OPC differentiation into postmitotic, myelin protein producing cells and continues over the first few postnatal weeks (Pernet et al., 2008). We assessed myelination at P7, P15, and P21 by immunostaining optic nerves with an antibody against MBP and by using the lipophilic dye Fluoromyelin Red to identify compact myelin (Watkins, 2008).
MBP expression was evident at P7 in optic nerves of control and sialin-/- mice, but to a much lesser extent in the sialin-/- mice, suggesting a delay in the onset of myelination. Further, fine linear MBP+ structures suggestive of axonal ensheathment were more abundant in the optic nerves from P7 control mice. Consistent with the MBP staining, Fluoromyelin Red staining of the optic nerves from the P7 mice was faint, but stronger in the tissue from the control animals. At P15 and P21 we saw an increase in density of MBP and Fluoromyelin Red staining in optic nerves from control and sialin-/- mice, but at each time point, we found less MBP immunofluorescence and Fluoromyelin Red staining in optic nerves from sialin deficient mice compared to controls (Fig. 4A).
To assess MBP expression quantitatively during optic nerve development, we performed Western blots on P7, P15 and P21 optic nerve samples (Fig. 4B and 4C). Consistent with the optic nerve staining, we found age-related increases in MBP expression in control and sialin-/- mice. At P7, MBP expression was not readily detectable by Western blotting. At P15 and P21, MBP expression levels were greater than twice as high in control compared to the sialin-/- mice (P15: p ≤ 0.01, P21: p ≤ 0.05). Similar levels of NF68 were present in optic nerves from control and sialin-/- mice at each age analyzed. Throughout this time period MBP (and NF68) levels in control and sialin-/- mouse sciatic nerves were indistinguishable (data not shown), indicating that the defect in myelination is specific to the CNS.
A final stage of myelinated fiber maturation is formation of distinct domains along the axon including the nodes of Ranvier and flanking paranodal regions (Poliak and Peles, 2003). The paranode is a region of axo-glial septate-like junctions that is thought to attach the myelin sheath to the axon and to restrict lateral diffusion of axonally expressed channels involved in saltatory conduction (Poliak and Peles, 2003). To determine whether this late stage of myelination is reached in the sialin-/- mice, we analyzed distribution of the axonal protein, contactin-associated protein (Caspr), in optic nerves from P21 mice. Caspr, a cell adhesion glycoprotein related to neurexins, is initially expressed along the length of the axon and is redistributed to the paranodal junctions during myelin maturation (Einheber et al., 1997; Menegoz et al., 1997). Unlike the persistently diffuse pattern of Caspr localization seen in many myelin mutants (Dupree et al., 1999; Rasband et al., 1999; Mathis et al., 2001), we found paired clusters of Caspr protein in the optic nerves of sialin-/- mice indicating that these animals are able to form mature paranodal structures (Fig. 5A). As expected there were far fewer paired Caspr clusters in the optic nerves of the sialin-/- mice, but similar numbers of unpaired clusters (Fig. 5B). We also found a broader distribution in cluster length (Fig. 5C). Identification of nodes (Fig. S3A) and heminodes (Fig. S3B) on electron microscopic examination of optic nerves from the sialin-/- mice further demonstrates that an advanced stage of myelin maturation is achieved as indicated by the Caspr staining. These findings suggest that, although myelination of the optic nerve is reduced in the sialin-/- mice throughout development, the myelin that does form is mature and relatively normal in structure and organization.
The decreased myelination in the sialin-/- mouse could be caused by a decrease in number of mature oligodendrocytes or by an inability of oligodendrocytes to produce myelin. A decrease in the number of mature oligodendrocytes could in turn be due to a defect in migration, proliferation, differentiation or survival of cells in the oligodendrocyte lineage.
Optic nerve OPCs are born in the floor of the third ventricle, migrate into the optic chiasm and proliferate as they migrate along the optic nerve toward the retina (Small et al., 1987; Ono et al., 1997). OPCs can be visualized by immunolabeling with the transcription factor Olig2, which predominantly labels nuclei of oligodendrocyte lineage cells in white matter tracts (Lu et al., 2000; Zhou et al., 2000; Dimou et al., 2008). In the mouse, Olig2+ OPCs are first detected in the optic nerve at embryonic day 17.5 (E17.5) and reach the optic nerve head by P4 (Pernet et al., 2008). To determine whether hypomyelination in the sialin-/- mouse could be caused by a defect in OPC development, proliferation or migration, we immunostained P7 optic nerves for Olig2 and quantified number of Olig2+ cells in the chiasmal, middle and retinal portions of the nerves. Olig2+ cells were evenly distributed throughout the optic nerves of control and sialin-/- P7 mice and the density of Olig2+ cells in the optic nerves of control and sialin-/- mice were not significantly different (Table 1). These findings suggest that development, migration, and proliferation of optic nerve OPCs are essentially normal in the sialin-/- mice and not underlying causes for the myelination defect.
To determine whether a defect in oligodendrocyte differentiation could be contributing to the reduction in myelination, we next examined expression of Olig2 along with CC1, a marker of postmitotic oligodendrocytes (Bhat et al., 1996), in optic nerves from P7, P15 and P21 mice (Fig. 6A and S4). As noted above, P7 Olig2+ cell densities were similar in the control and sialin-/- mice, but at P15 and to a greater extent at P21 Olig2+ cell densities are reduced in optic nerves from the sialin-/- mice compared to those from control mice (Fig. 6B). As expected, CC1+ cell density in optic nerves from control mice continuously increases from P7 to P21. By comparison, the CC1+ cell density is lower in the sialin-/- optic nerves at P7 and increases, but to a lesser extent from P7 to P15. There is no change in the CC1+ cell density in the sialin-/- nerves between P15 and P21. The reduction in Olig2+ cells in the sialin-/- mice compared to controls is accounted for by the difference in CC1+ cells, indicating that the density of OPCs (Olig2+/CC1- cells) is similar in the optic nerves from control and sialin-/- mice. This is further supported by the similar staining of P21 optic nerves from sialin-/- and control littermate mice with an antibody to the OPC marker NG2 (Fig. S5).
The analysis of Olig2+ and CC1+ cell densities suggests that loss of sialin leads to delayed or impaired differentiation of cells in the oligodendrocyte lineage, or to selective loss of the more mature cells. This is supported by the cytoarchitecture of oligodendrocytes in the sialin-/- mouse optic nerve. In the optic nerves from P21 control mice the CC1+ cells have elongated cell bodies and are present in chains oriented along the long axis of the nerve, while CC1+ cell bodies in P21 sialin-/- mouse optic nerves are typically found in isolation and have rounder cell bodies. This rounder morphology of CC1+ cell bodies in sialin-/- optic nerve is reminiscent of the pattern seen in immature, P7 control nerves (Fig. S4).
To assess further the health of the surviving oligodendrocytes, we performed electron microscopy on longitudinal sections of P21 optic nerve. Consistent with the immunostaining we found that sialin-/- oligodendrocytes typically appeared in isolation rather than in long chains of cell bodies seen in control nerves. Similar to oligodendrocytes in control mice (Fig. S3C), the oligodendrocytes in the sialin-/- mice (Fig. S3D and Fig. S3E) had typical clumped chromatin adjacent to the nuclear envelope, dark cytoplasm, and prominent rough endoplasmic reticulum (Peters, 1991). In contrast to sialin-/- cerebellar and spinal cord neurons in which vacuoles were common (see Fig. S1) only rare oligodendrocytes had vacuoles (Fig. S3E).
Presumably to ensure axons are fully myelinated, oligodendrocytes are produced in excess. During normal myelination of the rodent optic nerve approximately 50% of postmitotic oligodendrocytes undergo apoptosis within 2-3 days of differentiation (Barres et al., 1992; Trapp et al., 1997). We wondered whether the decrease in CC1+ cell densities in sialin-/- mouse optic nerve might be due to enhanced apoptosis at this stage. To identify apoptotic cells, we immunostained optic nerves with an antibody to activated caspase-3. In the nerves from P7 animals we found a number of cells with faintly labeled MBP+ extensions expressing activated caspase-3 (data not shown), consistent with previous reports (Ueda et al., 1999). The numbers of activated caspase-3+ cells in optic nerves from control and sialin-/- mice were not statistically different, suggesting that the rates of apoptotic cell death are similar at this time point. Since the number of optic nerve Olig2+ cells peaks at P10 (Pernet et al., 2008) and the relative reduction in the density of CC1+ cells in the sialin-/- nerves was greater at later time points, we immunolabeled for activated caspase-3+ cells in P15 optic nerves (Fig. 7A). We found more than twice as many activated caspase-3+ cells in the optic nerves of sialin-/- mice compared to control (Fig. 7B). Although MBP staining was too dense at this stage to demonstrate colabeling of individual cells with activated caspase-3, the increased number of apoptotic cells correlates with the increased rate of Olig2+ cell loss between P15 and P21, suggesting that the apoptotic cells are of the oligodendrocyte lineage.
It has been suggested that, once differentiated, oligodendrocytes require axonal contact to survive (Barres and Raff, 1999; Barres, 2008). Without this interaction by three days the oligodendrocytes undergo apoptosis. The heavily sialylated PSA-NCAM is a cell-surface adhesion protein that has been postulated to inhibit myelination. PSA-NCAM is downregulated at the onset of myelination and its overexpression in vitro leads to a delay in oligodendrocyte maturation and myelin formation (Charles et al., 2000; Franceschini et al., 2004; Fewou et al., 2007). We wondered whether loss of sialin leads to impairment of the endosomal-lysosomal pathways responsible for PSA-NCAM downregulation. To test this we examined quantitatively PSA-NCAM expression levels in P7, P15, and P21 optic nerves. In samples from both wild-type and sialin-/- mice we found that PSA-NCAM expression levels were progressively reduced from P7 to P21 in both genotypes (Fig. 8A). However, the extent of the reduction was less in the sialin-/- mice such that at P21 the level was nearly twice as high in the optic nerves from sialin-/- mice compared to control (Fig. 8B). While other mechanisms are undoubtedly involved, the impaired downregulation of PSA-NCAM likely contributes to the delayed myelin formation in the sialin-/- mice.
Although the genetic and biochemical basis of Salla disease and ISSD have been well characterized, meaningful advances in our understanding of the pathophysiology of these diseases have been hindered by their rarity and the lack of an animal model. We have examined the behavioral and neuropathological phenotype of sialin deficient mice to determine whether they appropriately reflect the free sialic acid storage disorders. We found that the mice express many of the cardinal features of these disorders including marked CNS hypomyelination and are thus an appropriate model in which to identify pathophysiological mechanisms and to investigate potential treatments of these disorders.
The effect of sialin loss on development of the nervous system appears remarkably specific. In stark contrast to the normal gross CNS neuronal cytoarchitecture and PNS myelination, the sialin-/- mouse shows a severe CNS myelination defect. Our ultrastructural analysis of optic nerves demonstrates a reduction in number of myelinated axons in P21 sialin-/- mice when essentially all optic nerve axons are myelinated in control animals. The myelin that is present appears grossly normal with a thickness similar to that in littermate controls.
Myelination of the CNS is a complex, multistep process that begins with the specification of proliferating, migratory OPCs, followed by differentiation of these cells into postmitotic oligodendrocytes that ensheath axons and ultimately form compact multilamellar myelin membranes (Baumann and Pham-Dinh, 2001). While a defect at any one of these steps or defects in multiple steps could underlie the impaired myelination associated with loss of sialin, the normal complement of optic nerve OPCs suggests that the primary defect occurs during or after post-mitotic differentiation.
As myelin matures there is precise matching of surviving oligodendrocytes with the axons that require myelination (Barres et al., 1993; Barres and Raff, 1994). Our data indicate that the reduction in myelin corresponds to a reduction in the number of these mature myelinating oligodendrocytes. The relative reduction in post-mitotic oligodendrocyte number is evident as early as P7 and is more pronounced at P15. Between P15 and P21 the number of oligodendrocytes increases in the control animals, whereas there is no increase in sialin-/- mice. Interestingly, the myelin content (as indicated by MBP expression levels and Fluoromyelin Red staining) in the sialin-/- optic nerves increases between P15 and P21 even though the postmitotic oligodendrocyte count does not. This suggests that in the sialin-/- mice, although the number of mature oligodendrocytes that are formed is reduced, the few cells that are present are robustly producing myelin. The impression that sialin-/- mice can form mature myelin is further supported by our finding that normal appearing paranodal structures, including Caspr clusters, are present in these animals.
During normal development of the rodent optic nerve, it has been estimated that between P4 and P10 about 50% of oligodendrocytes undergo apoptotic cell death as a result of a competition for survival signals that are provided by astrocytes and axons (Barres et al., 1992; Trapp et al., 1997). Could this process be enhanced in the sialin-/- mice and contribute to the reduction in oligodendrocytes? Although at P7 (when this process is peaking in wild-type animals) we find similar numbers of activated caspase-3+ cells in control and sialin-/- mouse optic nerves, at P15 we find more than double the number in the optic nerves from the sialin-/- mice. While the density of MBP expression in P15 nerves prohibited identification of individual cells colabeled with activated caspase-3, we suspect that the apoptotic cells in the optic nerves from P15 sialin-/- mice are oligodendrocytes. Two factors support our suspicion. First, it is very likely that the apoptotic cells are of the oligodendrocyte lineage as a corresponding decrease in Olig2+ cells occurs with the increase in apoptosis. Since Olig2+/CC1- cell counts and NG2 staining are very similar in the optic nerves of sialin-/- and control mice it is likely that the apoptotic cells are not OPCs but rather oligodendrocytes. Second, as mentioned, apoptosis of a significant portion of oligodendrocytes occurs as part of the normal developmental process, suggesting that these cells can be readily induced to undergo programmed cell death.
The simplest explanation for the decrease in numbers of myelinating oligodendrocytes is that the there is pathological increase of the normal process of apoptosis mediated competitive cellular pruning. Could loss of sialin lead to a generalized lysosomal defect that enhances apoptosis of oligodendrocytes? Since apoptosis is inhibited by lysosomal dependent processes including autophagy (Ferraro and Cecconi, 2007) and growth factor receptor signaling (Sweeney and Davis, 2002), loss of sialin might enhance oligodendrocyte apoptosis. However, other lysosomal storage disorders in which lysosomal function is likely to be equally impaired do not have hypomyelination as a prominently reported component of the pathological phenotype (Cherqui et al., 2002; de Geest et al., 2002; Barranger, 2007).
If a generalized defect in lysosome function does not fully explain the hypomyelination of the sialin-/- mice, could a specific alteration in the metabolism of sialic acid containing molecules explain these defects? Our finding that PSA-NCAM downregulation is impaired suggests an intriguingly simple mechanism for the myelination defect. It has been shown that in vitro myelination can be accelerated by the addition of a sialic acid cleaving neuraminidase and that overexpression of PSA-NCAM leads to a delay in oligodendrocyte maturation and myelin formation (Charles et al., 2000; Franceschini et al., 2004; Fewou et al., 2007). Although alterations in PSA-NCAM expression might not lead directly to increased apoptosis of oligodendrocytes, a delay in the rate at which newly generated oligodendrocytes contact axons and form mature myelin could reduce survival of newly differentiated oligodendrocytes (Barres and Raff, 1999; Barres, 2008).
Could sustained PSA-NCAM expression alone explain the complex phenotype of the free sialic acid storage disorders? It is unlikely as metabolism of gangliosides, the major sialic acid-bearing conjugates in the vertebrate brain (Holian et al., 1971), is influenced by loss of sialin function (Pitto et al., 1996). The expression of specific gangliosides is highly regulated during neurodevelopment and overall abundance increases during stages of neurogenesis, axon elongation and myelination (Holian et al., 1971; Rosner, 2003). Mice double mutant for two critical ganglioside-specific glycosyltransferase genes (Siat9 and Galgt1), are unable to synthesize the major class of brain gangliosides (including GM1, GD1a, GD1b, and GT1b) and show severe white matter pathology (Yamashita et al., 2005). GD1a and GT1b are found in axons and are thought to interact with myelin-associated glycoprotein (MAG), a protein expressed in the periaxonal myelin membrane of oligodendrocytes. It has been suggested that the hypomyelination in the ganglioside deficient mice is due to loss of these gangliosides as functional binding sites for MAG during the initiation of myelination (Sheikh et al., 1999). If loss of sialin leads to altered ganglioside expression profiles it follows that this crucial MAG-ganglioside interaction might not occur.
In summary, we have demonstrated the validity of the sialin-/- mouse as a model for the free sialic acid storage disorders. We have provided evidence that an increase in the apoptotic death of cells in the oligodendrocyte lineage occurs in these mice and have identified delayed downregulation of PSA-NCAM as a potential upstream event. These mice can now be used to dissect the specific molecular mechanism (or mechanisms) underlying the hypomyelination characteristic of these disorders.
This work was supported by funding from the NIH (NS050417 and NS045634 to RJR and NS065664 to LMP), the March of Dimes (to RJR), and a Howard Hughes Research Training Fellowship for Medical Students (to LMP). LMP is in the Medical Scientist Training Program at Stanford University School of Medicine. We thank Ben Barres, Ben Emery, Trent Watkins, Craig Garner, Marion Buckwalter, and members of the Reimer lab for helpful comments. We thank Anita Briley and Isabel Parada for invaluable assistance with electron microscopy and histological studies.