Npc1 Acting in Neurons and Glia Is Essential for the Formation and Maintenance of CNS Myelin

Cholesterol availability is rate-limiting for myelination, and prior studies have established the importance of cholesterol synthesis by oligodendrocytes for normal CNS myelination. However, the contribution of cholesterol uptake through the endocytic pathway has not been fully explored. To address this question, we used mice with a conditional null allele of the Npc1 gene, which encodes a transmembrane protein critical for mobilizing cholesterol from the endolysosomal system. Loss of function mutations in the human NPC1 gene cause Niemann-Pick type C disease, a childhood-onset neurodegenerative disorder in which intracellular lipid accumulation, abnormally swollen axons, and neuron loss underlie the occurrence of early death. Both NPC patients and Npc1 null mice exhibit myelin defects indicative of dysmyelination, although the mechanisms underlying this defect are incompletely understood. Here we use temporal and cell-type-specific gene deletion in order to define effects on CNS myelination. Our results unexpectedly show that deletion of Npc1 in neurons alone leads to an arrest of oligodendrocyte maturation and to subsequent failure of myelin formation. This defect is associated with decreased activation of Fyn kinase, an integrator of axon-glial signals that normally promotes myelination. Furthermore, we show that deletion of Npc1 in oligodendrocytes results in delayed myelination at early postnatal days. Aged, oligodendocyte-specific null mutants also exhibit late stage loss of myelin proteins, followed by secondary Purkinje neuron degeneration. These data demonstrate that lipid uptake and intracellular transport by neurons and oligodendrocytes through an Npc1-dependent pathway is required for both the formation and maintenance of CNS myelin.


Introduction
Ensheathment of axons by myelin is an evolutionary feature of the vertebrate nervous system that is accomplished by the extended and specialized plasma membranes of oligodendrocytes in the CNS and Schwann cells in the PNS. Myelin contains at least 70% lipids by dry weight [1], and this high ratio of lipid to protein ensures the insulating properties of myelin to maximize the efficiency of nerve conduction. Among all the lipid species found in the myelin sheath, unesterified cholesterol is a major component [1]. In the mouse CNS, cholesterol in compact myelin represents ,78% of the total lipid pool [2], and the availability of cholesterol is the rate-limiting step for myelination [3]. Since the CNS is shielded by the blood brain barrier, cholesterol required for myelination comes entirely from local synthesis [2]. Both neurons and glia obtain the cholesterol they need either through endogenous synthesis or by uptake of lipoprotein particles produced and released within the CNS. That endogenously synthesized cholesterol is critical for CNS myelination in mice is demonstrated by deletion in oligodendrocytes of squalene synthase, the first dedicated enzyme of sterol synthesis [3]. These mutant mice exhibit delayed myelination, suggesting that exogenously supplied cholesterol also contributes to CNS myelin formation. However, whether cholesterol from exogenous sources is required for myelin synthesis, or just a compensatory source when endogenous synthesis is lacking in myelinating glia, has not been explored.
An essential component of the pathway through which cholesterol in lipoprotein particles is mobilized from the endolysosomal system is the Npc1 protein [4,5]. This multipass transmembrane protein resides in late endosomes and lysosomes [6][7][8][9], and functions cooperatively with the Npc2 protein to facilitate cholesterol efflux [10,11]. Loss of functional Npc1 disrupts intracellular lipid trafficking, and leads to the sequestration of unesterified cholesterol and glycosphingolipids in late endosomes and lysosomes [12]. Mutations in the human NPC1 gene cause Niemann-Pick type C disease (NPC), a fatal childhoodonset neurodegenerative disorder [13]. Mice with a null mutation in the Npc1 gene (Npc1 2/2 ) recapitulate the human disease, and exhibit progressive CNS neuropathology in which intracellular lipid accumulation, abnormally swollen axons, neuron loss and gliosis underlie the occurrence of ataxia and early death [5,14]. Notably, both NPC patients and Npc1 2/2 mice exhibit myelin defects indicative of dysmyelination, particularly in the forebrain [15][16][17][18][19]. However, the complex pathology resulting from Npc1 deficiency in both neurons and oligodendrocytes has limited the utility of these global null mutants to provide a detailed understanding of the contribution of exogenous cholesterol to CNS myelination.
Here we use mice with a conditional null allele of the Npc1 gene to achieve temporal and cell type specific deletion in order to define effects on CNS myelin. We show that deletion of Npc1 restricted to neurons unexpectedly recapitulates the dysmyelination phenotype of global null mutants. This effect is mediated by a block in maturation of oligodendrocyte lineage cells that is associated with decreased activation of Fyn kinase, an integrator of axon-glial signals that normally promote myelination. Furthermore, we show that deletion of Npc1 in oligodendrocytes triggers a similar, though less severe impairment of CNS myelination, as well as myelin protein loss and secondary neurodegeneration. Our analyses suggest that exogenous cholesterol entering cells and trafficking through an Npc1-dependent pathway is necessary for both the formation and maintenance of CNS myelin.

Global Npc1 Deficiency Leads to CNS Dysmyelination, Followed by Late Stage Loss of Myelin Proteins
To confirm the requirement of Npc1 for proper myelination in mice during early postnatal stages, we utilized mice with a floxed Npc1 allele (Npc1 flox ) [20]. Cre-mediated deletion yields a null allele that is functionally indistinguishable from the spontaneous null mutation found in Npc1 nih mice (Npc1 2/2 ) [5,20]. To generate mice with Npc1 deletion in the germline, Npc1 flox/flox mice were bred with transgenic mice expressing Cre recombinase under the control of the EIIa promoter [21]. Mice mosaic for the conditionally deleted allele were bred with mice carrying the Npc1 2 allele to generate compound heterozygotes of the conditionally deleted and null Npc1 alleles (Npc1 D/2 ). We also generated mice with Npc1 deletion in adults by using a tamoxifen-regulated Cre recombinase under the control of the cytomegalovirus (CMV) promoter (Cre-ER TM+ ) [22]. Cre-mediated deletion of Npc1 in adults was induced by tamoxifen injections at 6 weeks, an age at which myelination is complete. Mice with adult deletion (Npc1 flox/ 2 , Cre-ER TM+ ) have been shown to recapitulate most features of NPC neuropathology, and reach end-stage by ,22 weeks [23]. To determine the effect of the timing of Npc1 deletion upon myelination, we compared 7-week-old mice with germline deletion (Npc1 D/2 ), 22-week-old mice with adult deletion (Npc1 flox/2 , Cre-ER TM+ ) and age matched controls. Myelin basic protein (MBP, a standard marker for mature myelin [1]) and FluoroMyelin (a lipophilic stain for compact myelin) staining of sagittal midline brain sections revealed a dramatic reduction of myelin proteins and lipids in Npc1 D/2 mice, particularly in the forebrain ( Figure 1A, 1B). This striking pattern of regionally selective myelin defects is similar to that previously reported in Npc1 2/2 mice [14,15,17]. In contrast, Npc1 flox/2 , Cre-ER TM+ mice exhibited a staining pattern morphologically similar to that in controls ( Figure 1A, 1B). The difference in MBP staining patterns between Npc1 D/2 mice and Npc1 flox/2 , Cre-ER TM+ mice suggests that Npc1 is required in early postnatal stages for proper myelin formation. Further analysis of myelin-specific proteins demonstrated a decrease in MBP and CNP protein levels in Npc1 flox/2 , Cre-ER TM+ mice compared to littermate controls, particularly in the cortex ( Figure 1C, 1D). We conclude that myelin was properly formed in Npc1 flox/2 , Cre-ER TM+ mice during postnatal development, but that these mice exhibit loss of myelin proteins at later stages, particularly in the cerebral cortex, after Npc1 deletion at 6 weeks. Axonal loss could contribute to the late stage pathology in Npc1 flox/ 2 , Cre-ER TM+ mice, as evidenced by decreased neurofilament levels in these aged mutants ( Figure 1C). Taken together, our analysis suggests that lack of myelin in NPC mice is caused by dysmyelination at early postnatal days, followed by loss of myelin proteins at end stage.

Neuronal Deletion of Npc1 Leads to Blockade of Oligodendrocyte Maturation and Dysmyelination
We next sought to dissect the contribution of different CNS cell types to NPC dysmyelination. We started by deleting Npc1 specifically in neurons, using transgenic mice expressing Cre recombinase under the control of the Synapsin1 promoter (Syn1-Cre) [24]. We confirmed gene deletion by staining brain sections with filipin, a fluorescent dye that specifically marks accumulation of unesterified cholesterol [25]. NeuN and filipin co-staining verified that Npc1 flox/2 , Syn1-Cre + mice, but not Npc1 flox/+ , Syn1-Cre + controls [23], developed filipin-positive neurons throughout the brain, including brainstem and cortex ( Figure S1A). A subset of neurons remained filipin negative, possibly reflecting mosaic gene deletion. To further verify neuron-specific gene deletion, Syn1-Cre + mice were crossed to a Rosa reporter line that has been widely used to demonstrate gene deletion in both neurons and oligodendrocytes [26]. LacZ staining revealed widespread positive cells in many brain regions including the cortex, with minimal staining in the corpus callosum, where neuronal cell bodies are lacking ( Figure S1B). Co-staining with NeuN or Olig2 showed that these LacZ positive cells were neurons, and not oligodendrocyte lineage cells ( Figure S1C), further supporting the notion that we achieved neuron-specific deletion by using Syn1-Cre + mice.
The effect of Npc1 deficiency in neurons upon myelination was first evaluated by MBP immunofluorescence at 3 different ages. At postnatal day 16 (P16), myelination was actively occurring in the forebrain of Npc1 flox/+ , Syn1-Cre + controls, with abundant MBPpositive myelinating oligodendrocytes populating the cortex ( Figure 2B). In contrast, Npc1 flox/2 , Syn1-Cre + mutants exhibited a severe paucity of myelin in the same region, with most of the MBP positive cells exhibiting the morphology of pre-myelinating oligodendrocytes ( Figure 2B). At 7 weeks, myelination was completed in Npc1 flox/+ , Syn1-Cre + controls, but was greatly attenuated in the cortex of Npc1 flox/2 , Syn1-Cre + mutants. No recovery of myelination was observed in mutants aged to 16 weeks ( Figure 2B), which is end stage for these mice [23]. Similarly, FluoroMyelin staining revealed a paucity of compact myelin in the corpus callosum of Npc1 flox/2 , Syn1-Cre + mutants at 16 weeks ( Figure 2B, bottom panel). Although MBP staining was markedly

Author Summary
The myelin sheath in the central nervous system is a specialized extension of the oligodendrocyte plasma membrane that serves as an electrical insulator to ensure proper nerve conduction. To accomplish this, myelin is enriched in lipids, particularly unesterified cholesterol, which is an essential and limiting component for myelin formation. Here we determine the contribution of exogenously derived cholesterol to myelination by using a conditional null mutant of the mouse Npc1 gene. Npc1 encodes a transmembrane protein critical for mobilizing exogenously derived cholesterol from late endosomes and lysosomes, and is mutated in patients with Niemann-Pick type C disease, a degenerative disorder caused by impaired intracellular lipid trafficking. We show that mice lacking Npc1 in either neurons or oligodendrocytes exhibit a defect in myelin formation in selected brain regions, with an arrest in oligodendrocyte maturation. In addition, mice with Npc1 deficiency in oligodendrocytes, when aged, show progressive motor dysfunction with myelin breakdown and secondary Purkinje neuron loss. Taken together, our findings demonstrate the role of Npc1 in mediating reciprocal signaling between neurons and glia, and highlight the importance of exogenous cholesterol for CNS myelin formation and maintenance. Npc1 Is Required for CNS Myelination week-old Npc1 D/2 (middle), and 22-week-old Npc1 flox/2 , Cre-ER TM + mice following tamoxifen injections at 6 weeks (bottom). Ctx, cortex; CC, corpus decreased in the cortex of Npc1 flox/2 , Syn1-Cre + mutants, other brain regions exhibited a normal staining pattern, reminiscent of the selective defects in myelination observed after global germline deletion ( Figure 1A). Regional-specific dysmyelination was further supported by western blots showing decreased levels of myelinspecific proteins including CNP, MBP and MAG in cortex, but not brainstem of Npc1 flox/2 , Syn1-Cre + mutants ( Figure 2C). Electron microscopy confirmed that the density of myelinated nerve fibers in the corpus callosum was greatly reduced in Npc1 flox/ 2 , Syn1-Cre + mutants at 3 weeks ( Figure 2E). Notably, neurofilament protein levels in the cortex were similar between Npc1 flox/+ , Syn1-Cre + controls and Npc1 flox/2 , Syn1-Cre + mutants at P16 ( Figure 2C), and neurofilament immunofluorescence staining showed no significant axonal pathology ( Figure 2D). These data indicate that dysmyelination in the forebrain of Npc1 flox/2 , Syn1-Cre + mutants was not secondary to axonal loss.
To characterize the mechanism underlying dysmyelination in Npc1 flox/2 , Syn1-Cre + mutants, we assessed oligodendrocyte lineage cells at different stages of differentiation. At P16, Npc1 flox/2 , Syn1-Cre + mutants showed a significantly reduced number of CC1-positive mature oligodendrocytes in the forebrain ( Figure 3A, 3C) but a normal density of NG2-positive oligodendrocyte precursor cells (OPCs) ( Figure 3A, 3B). As previously reported for global null Npc1 mutants [17], this deficit of mature oligodendrocytes was not associated with evidence of increased apoptosis (data not shown). The paucity of mature oligodendrocytes was associated with a reduced number of cells in the corpus callosum expressing Sip1, a signaling protein implicated oligodendrocyte differentiation ( Figure 3C) [27]. These data indicated that Npc1 deficiency in neurons triggered a block of oligodendrocyte maturation, and prompted us to determine whether signals known to regulate oligodendrocyte maturation and myelination were perturbed in Npc1 flox/2 , Syn1-Cre + mutants. We first examined proteins that mediate signaling between axons and oligodendrocyte lineage cells including PSA-NCAM [28], Lingo1 [29] and Jagged1 [30], and found no differences between Npc1 flox/2 , Syn1-Cre + mutants and controls at P16 ( Figure S2A). Similarly, we found no evidence of astrocyte activation in the corpus callosum of Npc1 flox/2 , Syn1-Cre + mutants at P16 ( Figure S2B, S2C), consistent with prior studies showing that astrogliosis is limited to the thalamus of Npc1 2/2 mice at two weeks [31]. In contrast, activity of the non-receptor tyrosine kinase Fyn [32] was reduced in the cortex of Npc1 flox/2 , Syn1-Cre + mutants, as evidenced by decreased levels of the active form (phosphorylated at tyrosine 420) and concurrently increased levels of the inactive form (phosphorylated at tyrosine 531) ( Figure 3E). As oligodendroglial Fyn is an integrator of axonal signals that promote myelination [33], the decreased activity of Fyn in Npc1 flox/2 , Syn1-Cre + mutants suggests that Npc1 deficiency in axons leads to a disruption of axon-glial signaling that is crucial for oligodendrocyte differentiation and myelination.
Oligodendrocyte Deletion of Npc1 Results in a Similar, but Milder Dysmyelination Phenotype during Postnatal Development Next, we tested if Npc1 deficiency in oligodendrocyte lineage cells contributes to the pathogenesis of dysmyelination in NPC mice. To accomplish this, we used transgenic mice expressing Cre recombinase under the control of the CNP promoter (CNP Cre/+ ) [34]. In these mice, Cre is abundantly and specifically expressed in postmitotic oligodendrocytes. Co-staining for Cre and Olig2, a marker of both OPCs and postmitotic oligodendrocytes, verified that Cre was specifically expressed in a subset of Olig2 + oligodendrocyte lineage cells in various brain regions including brainstem and cortex ( Figure S3B). Filipin staining revealed minimal accumulation of unesterified cholesterol in Npc1 flox/2 , CNP Cre/+ mutants ( Figure S3A), a finding both consistent with a previous report showing no detectable cholesterol accumulation in oligodendrocytes of Npc1 2/2 mice [35] and indicative of the cell lineage specificity of this Cre line.

Mice with Npc1 Deletion in Oligodendrocytes Exhibit Myelin Protein Loss in Late Stages
As the Npc1 flox/2 , CNP Cre/+ mutants aged, they developed progressive motor deficits ( Figure 6C), although weight was not affected ( Figure 6A, 6B). This led us to examine myelin protein levels in 23-week-old Npc1 flox/2 , CNP Cre/+ mutants. We found decreased levels of myelin proteins not only in cortex, but also in brainstem and cerebellum ( Figure 7A), where myelination in early postnatal days was nearly normal ( Figure 4B). This suggested that myelin loss was taking place in several brain regions of the aged Npc1 flox/2 , CNP Cre/+ mutants. We found this was associated with only mild changes in the pattern of MBP staining ( Figure 7B). Interestingly, the total number of Olig2 + oligodendrocyte lineage cells in the cerebellar white matter was unchanged in aged mutants ( Figure 7C, 7D), suggesting that loss of Npc1 did not affect the survival of oligodendrocytes in adult mice. This loss of myelin proteins was associated with secondary neuron loss in the cerebellum. We detected Purkinje cell loss in anterior lobules of 23-week-old but not 7-week-old Npc1 flox/2 , CNP Cre/+ mutants, as demonstrated by calbindin staining of sagittal midline sections ( Figure 7E Figure 7F). Importantly, no filipin-positive Purkinje neurons were identified in these mice (not shown), supporting the conclusion that Purkinje cell loss was a consequence of non-cell autonomous toxicity. We conclude that Npc1 acts in oligodendrocytes both to promote normal myelination and to ensure the maintenance of myelin in the adult CNS.

Discussion
Here we used Npc1 conditional null mice to establish the critical role of Npc1 in both neurons and oligodendrocytes for proper CNS myelination. Our findings demonstrate that deletion of Npc1 in neurons alone is sufficient to recapitulate the dysmyelination   phenotype that occurs following global germline deletion. These mice display a severe phenotype, particularly in the forebrain, characterized by a lack of mature oligodendrocytes but a normal density of OPCs, indicating that Npc1 deficiency in neurons triggers an arrest of oligodendrocyte maturation. Our data also demonstrate that deletion of Npc1 in oligodendrocytes leads to similar but milder forebrain dysmyelination that largely recovers by 7 weeks, consistent with a delay rather than a block in myelination. Furthermore, we demonstrate that these oligodendrocyte-specific mutants develop ataxia as they age, and that this is associated with decreased myelin proteins and Purkinje cell loss in anterior cerebellar lobules, establishing the occurrence of second-ary neurodegeneration. Our results highlight the importance of Npc1 in both neurons and oligodendrocytes for the formation and maintenance of CNS myelin.
Significant effort has been devoted to defining the contribution of specific cell types to NPC neuropathology. Studies in chimeric mice, a conditional knock-out model, and several neuron-specific transgenic rescue experiments all demonstrate that neuronal loss can be a consequence of cell autonomous neurotoxicity [20,23,[37][38][39]. Furthermore, these analyses indicate that brain inflammation is a consequence rather than a driver of neuron loss [20,23,38,40]. The role of astroglial cells in NPC neuropathology has been more controversial. While in vitro data suggest that Npc1 deficient   astrocytes fail to fully support cultured neurons [41], both conditional knockout and transgenic rescue experiments failed to establish a significant role for astrocytes in pathogenesis [23,38]. A transgenic line that highly over-expresses Npc1 from the GFAP promoter does show some rescue [42], but the extent of cell type restricted expression in these mice remains incompletely defined. The effects of Npc1 deficiency restricted to oligodendrocytes had not been previously explored. As for effects on CNS myelin, prior transgene rescue experiments using the NSE promoter to drive Npc1 expression demonstrated partial rescue of myelination [39]. These findings are consistent with our observation that neuronal expression of Npc1 plays an important role in oligodendrocyte maturation and myelination. Finally, we note that aged, oligodendrocyte-specific null mutants show evidence of neuron loss. While prior studies firmly establish that neuronal deficiency of Npc1 is sufficient to mediate neurotoxicity [20,23], the data reported here indicate that non-cell autonomous pathways arising from oligodendrocytes also contribute to neuropathology.
Oligodendrocyte differentiation and myelination are highly dynamic processes controlled by both intrinsic factors and extrinsic mechanisms [43]. Recent studies of axon-glial communication have identified a series of axonal signals important for regulating myelination. Oligodendroglial Fyn, a Src family kinase, has been suggested to play a central role in integrating diverse axonal signals to initiate myelination [33]. Downstream signaling from activated Fyn kinase promotes oligodendrocyte survival, alters cytoskeleton polarity and increases the expression of myelin genes. Our analysis of neuron-specific Npc1 mutants reveals decreased Fyn activity and a regionally-restricted dysmyelination phenotype similar to that of Fyn knockout mice [44]. We suggest that Npc1 deficiency in neurons disrupts an axon-glial signal vital for promoting myelination. The axonal ligand responsible for oligodendroglial Fyn activation remains elusive. The requirement of Npc1 for Fyn activation raises the possibility that a lipid species, such as cholesterol or a sphingolipid, may contribute to this signal. Additionally, recent neuron-glial co-culture studies demonstrate the role of action potentials in stimulating myelination through Fyn-dependent mechanisms [45]. It is therefore also possible that defective Fyn activation results from decreased electrical activity of axons in Npc1 flox/2 , Syn1-Cre + mutants. Recently, a similar role in myelination has been demonstrated for neuron-restriction expression of the PI(3,5)P(2) phosphatase Fig4 [46], suggesting that defects in axon-glial signaling may underlie dsymyelination in several disorders.
Animal studies of cholesterol metabolism in myelinating glia have highlighted the importance of cell-autonomous production of cholesterol for myelin formation. Mice lacking oligodendroglial squalene synthase, an enzyme required for cholesterol synthesis, exhibit perturbed CNS myelination in early postnatal days [3]. Similarly, deletion of SCAP (SREBP-cleavage-activating protein) in Schwann cells, a protein that complexes with SREBP to regulate the expression of genes promoting cholesterol synthesis and lipoprotein uptake, leads to PNS hypomyelination [47]. It is notable that both mouse models partially recover at later stages, suggesting that myelinating glia have the capacity to overcome the lack of endogenous cholesterol production, probably through increased uptake. Here we present in vivo evidence indicating an important contribution of exogenous cholesterol to myelin synthesis. Our findings show that deletion of Npc1 in oligodendrocytes, which eliminates their utilization of cholesterol from the endocytosis of LDL or similar lipoprotein particles, leads to perturbed myelin formation in the CNS. Npc1 deficiency also impairs intracellular trafficking of sphingolipids [48] and endogenously synthesized cholesterol [49]. Nonetheless, the blockade of exogenous cholesterol utilization and the essential role that cholesterol plays in myelination leads us to favor the conclusion that the effects observed here are due to a disruption in the availability of exogenous cholesterol. As shown for other cell types [12], we speculate that the synthesis of endogenous cholesterol may be up-regulated in Npc1 deficient oligodendrocytes yet insufficient to overcome the lack of exogenous cholesterol, especially during the peak phase of myelination. This suggests that extracellularly-derived cholesterol is indispensible for normal CNS myelination.
Although Npc1 flox/2 , CNP Cre/+ mutants form myelin in the brainstem and cerebellum during postnatal development, these regions exhibit loss of myelin proteins in adults. Biochemical studies have shown that in the adult CNS, myelin production and cholesterol turnover decrease to very low levels [2]. It is therefore unlikely that the loss of myelin proteins in these adult mutants results from impaired access to exogenous cholesterol as a consequence of Npc1 deficiency. Rather, we speculate that latestage pathology stems from the unstable nature of the myelin sheath produced by mutant oligodendrocytes. Studies of cellular models of NPC have shown that cholesterol content is decreased in the plasma membrane of mutant cells [50,51]. This change may impact myelin by disrupting membrane fluidity, altering lipid rafts or modulating the function of membrane proteins, and thereby increase the vulnerability of aged mutants. Further analysis of the biochemical composition of the myelin sheath generated by Npc1deficient oligodendrocytes will help define the mechanism mediating late-onset loss of myelin proteins. Axonal degeneration and neuron loss in these mutants highlights the important role of oligodendrocytes in supporting neuron function and survival. Similar observations have been made in mice over-expressing alpha-synuclein in oligodendrocytes [52]. While this effect may be mediated in part through loss of myelin, other studies have shown that oligodendroglia support axons through metabolic pathways independent of myelination [53]. It is currently unclear which of these mechanisms accounts for Purkinje neuron loss in Npc1 flox/2 , CNP Cre/+ mutants.
In summary, the data reported here extend our understanding of the role of cholesterol metabolism in myelination, and demonstrate that exogenous cholesterol is needed by both neurons and oligodendrocytes for the formation and maintenance of CNS myelin. A characteristic feature of Npc1 deficient mice, both global nulls and cell-specific knockouts, is the regionally severe dysmyelination that occurs during early postnatal stages. Fate-mapping studies have established that OPCs originate from heterogeneous regions of the subventricular zone, under the influence of different signaling pathways [54]. We speculate that these regional differences in oligodendrocyte lineage cells lead to distinct responses to axonal signals or to the need for exogenously-derived cholesterol for proper myelination, contributing to severe dysmyelination particularly in the forebrain of Npc1 mutants. While the precise mechanism underlying this regional selectivity remains to be defined, our data establish a critical role for Npc1 in both myelin formation and maintenance. Our findings have important implications for understanding the pathogenesis of NPC disease and may also inform our knowledge of other dysmyelinating/demyelinating disorders.

Ethics Statement
Animal use and procedures were approved by the University of Michigan Committee on the Use and Care of Animals.

Tamoxifen Induction
Tamoxifen (Sigma) was dissolved in corn oil (Sigma) at 20 mg/ ml and stored at 220C in the dark. The stock solution was warmed to 37C before injection. 6-week-old mice were injected intraperitoneally with 3 mg tamoxifen per 40 g body weight for 5 consecutive days.

Phenotype Analysis
Motor function was measured using the balance beam test as described previously [20].

Western Blot
Brain lysates were homogenized in RIPA buffer (Thermo Scientific) containing Complete protease inhibitor cocktail (Roche) and phosphatase inhibitors (Thermo scientific) using a motor homogenizer (TH115, OMNI International). Samples were resolved by 4-20% Tris-glycine gradient gel and transferred to nitrocellulose membranes (BioRad) on a semidry transfer apparatus. Immunoreactivity was detected by Immobilon chemilluminescent substrate (Thermo Scientific). Antibodies used were rat anti- Immunoprecipitation 200 mg brain lysates were immunoprecipitated with 10 mg anti-Fyn antibody (FYN3, Santa Cruz) overnight at 4C, followed by incubation with 20 ml Protein A beads (Santa Cruz) for 1 h at 4C. The immunoprecipitates were then washed 4 times with protein lysis buffer before being boiled with 26 sample buffer at 100C for 5 min. For the subsequent western blot analysis, anti-Fyn (FYN3, Santa Cruz), Src pY418 and pY529 antibodies (Life technologies) were used to detect total Fyn and phosphorylation of Fyn at Y420 and Y531, respectively.

Histology
Mice were perfused with 0.9% normal saline followed by 4% paraformaldehyde. Brains were removed and post-fixed in 4% paraformaldehyde overnight. Brains were bisected, with the right hemisphere processed for paraffin embedding and the left hemisphere processed for frozen sections. Prior to freezing, brain tissue was cryoprotected in 30% sucrose for 48 hr at 4C. Brains were frozen in isopentane chilled by dry ice and embedded in OCT (Tissue-Tek). Frozen sections were prepared at 14 mm in a cryostat and used for LacZ staining and subsequent eosin counter staining or immunohistochemical staining for Olig2 (1:500, Millipore) and NeuN (1:500, Millipore). For filipin staining, frozen sections were first used for immunofluorescence staining for NeuN or Olig2, followed by incubation for 90 min in PBS with 10% fetal bovine serum plus 25 mg/ml filipin (Sigma). For FluoroMyelin staining, frozen sections were rehyrated in PBS for 20 min, incubated with FluoroMyelin solution (1:300, Life Technologies) at room temperature for 2 hours, and then cleared with four 30minute washes with PBS. Paraffin-embedded sections were prepared at 5 mm and used for staining with H&E staining or MBP (1:100, Abcam), SMI-31P (1:200, Covance), NG2 (1:100, Millipore), CC1 (1:200, Calbiochem), Calbindin (1:1000, Sigma), Sip1 (1:100, Santa Cruz) and GFAP (1:1000, Dako) immunofluorescence. For visualization of staining, secondary antibodies conjugated to Alexa Fluor 594 or Alexa Fluor 488 (Molecular Probes) were used and images were captured on a Zeiss Axioplan 2 imaging system. For NG2 and CC1 co-staining and Olig2 staining, images were captured on an Olympus FluoView 500 Confocal Microscope system. Quantification of CC1 + or Olig2 + cells was performed using NIH ImageJ software. Quantification of Purkinje cell loss was performed on H&E stained sections. Counts were normalized to the length of the Purkinje layer, as measured by NIH ImageJ software, and reported as Purkinje cell density.

Electron Microscopy
Mice were perfused with 0.9% normal saline followed by 3% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M Sorensen's buffer. The corpus callosum was removed and post-fixed in perfusion solution overnight, followed by fixation in 1% osmium tetroxide solution for 1 h at room temperature. After dehydration, tissues were embedded in epoxy resin. For transmission electron microscopy, ultrathin sections were cut, and images were captured on a Philips CM-100 imaging system at 10,5006 magnification.

Statistics
Statistical significance was assessed by unpaired Student's t test. Statistics were performed using the software package Prism 5 (GraphPad Software). P values less than 0.05 were considered significant.