LRP1 regulates peroxisome biogenesis and cholesterol homeostasis in oligodendrocytes and is required for proper CNS myelin development and repair

Low-density lipoprotein receptor-related protein-1 (LRP1) is a large endocytic and signaling molecule broadly expressed by neurons and glia. In adult mice, global inducible (Lrp1flox/flox;CAG-CreER) or oligodendrocyte (OL)-lineage specific ablation (Lrp1flox/flox;Pdgfra-CreER) of Lrp1 attenuates repair of damaged white matter. In oligodendrocyte progenitor cells (OPCs), Lrp1 is required for cholesterol homeostasis and differentiation into mature OLs. Lrp1-deficient OPC/OLs show a strong increase in the sterol-regulatory element-binding protein-2 yet are unable to maintain normal cholesterol levels, suggesting more global metabolic deficits. Mechanistic studies revealed a decrease in peroxisomal biogenesis factor-2 and fewer peroxisomes in OL processes. Treatment of Lrp1−/− OPCs with cholesterol or activation of peroxisome proliferator-activated receptor-γ with pioglitazone alone is not sufficient to promote differentiation; however, when combined, cholesterol and pioglitazone enhance OPC differentiation into mature OLs. Collectively, our studies reveal a novel role for Lrp1 in peroxisome biogenesis, lipid homeostasis, and OPC differentiation during white matter development and repair.


Introduction
In the central nervous system (CNS), the myelin-producing cell is the oligodendrocyte (OL). Mature OLs arise from oligodendrocyte progenitor cells (OPCs), a highly migratory pluripotent cell type (Rowitch and Kriegstein, 2010;Zuchero and Barres, 2013). OPCs that commit to differentiate along the OL-lineage undergo a tightly regulated process of maturation, membrane expansion, and axon myelination (Emery et al., 2009;Hernandez and Casaccia, 2015;Li and Yao, 2012;Simons and Lyons, 2013). Even after developmental myelination is completed, many OPCs persist as stable CNS resident cells that participate in normal myelin turnover and white matter repair following injury or disease (Fancy et al., 2011;Franklin and Ffrench-Constant, 2008).
LRP1 is a member of the LDL receptor family with prominent functions in endocytosis, lipid metabolism, energy homeostasis, and signal transduction (Boucher and Herz, 2011). Lrp1 is broadly expressed in the CNS and abundantly found in OPCs (Auderset et al., 2016;Zhang et al., 2014).
LRP1 is a large type-1 membrane protein comprised of a ligand binding 515 kDa a chain noncovalently linked to an 85 kDa b chain that contains the transmembrane domain and cytoplasmic portion. Through its a chain, LRP1 binds over 40 different ligands with diverse biological functions Lillis et al., 2008). LRP1 mediates endocytotic clearance of a multitude of extracellular ligands (May et al., 2003;Tao et al., 2016) and participates in cell signaling, including activation of the Ras/MAPK and AKT pathways (Fuentealba et al., 2009;Martin et al., 2008;Muratoglu et al., 2010). The LRP1b chain can be processed by g-secretase and translocate to the nucleus where it associates with transcription factors to regulate gene expression (Carter, 2007;May et al., 2002).
Here, we combine conditional Lrp1 gene ablation with ultrastructural and electrophysiological approaches to show that Lrp1 is important for myelin development, nerve conduction, and adult CNS white matter repair. Gene expression analysis in Lrp1-deficient OPCs identified a reduction in peroxisomal gene products. We show that Lrp1 deficiency decreases production of peroxisomal proteins and disrupts cholesterol homeostasis. Mechanistic studies uncover a novel role for Lrp1 in PPARg-mediated OPC differentiation, peroxisome biogenesis, and CNS myelination.

Results
In adult mice, inducible ablation of Lrp1 attenuates CNS white matter repair To study the role of Lrp1 in CNS myelin repair, we pursued a mouse genetic approach. Lrp1 global knockout through the germline results in embryonic lethality (Herz et al., 1992). To circumvent this limitation, we generated Lrp1 flox/flox ;CAG-CreER TM mice (Lrp1 iKO) that allow tamoxifen (TM)-inducible global gene ablation. As control, Lrp1 mice harboring at least one wild-type or non-recombined Lrp1 allele were injected with TM and processed in parallel (Figure 1-figure supplement 1). TM injection into P56 mice resulted in an approximately 50% decrease of LRP1 in brain without noticeable impact on white matter structure (Figure 1-figure supplement 1). One month after TM treatment, Lrp1 iKO and control mice were subjected to unilateral injection of 1% lysophosphatidylcholine (LPC) into the corpus callosum. The contralateral side was injected with isotonic saline (PBS) and served as control. Brains were collected 10 and 21 days after LPC/PBS injection and the extent of white matter damage and repair were analyzed (Figure 1a Because Mbp mRNA is strongly upregulated in myelin producing OLs and transported into internodes (Ainger et al., 1993), we used Mpb ISH to find the white matter lesion (Figure 1b). The section with the largest circumference of the intensely labeled Mbp + area was defined as the  (CC) and PBS into the contralateral side. Coronal brain sections (series of 6, each 120 mm apart) probed for Mbp by in situ hybridization (ISH). Brain sections containing the lesion center were identified and subjected to quantification. (c) Coronal brain sections through the Figure 1 continued on next page lesion center and subjected to quantification ( Figure 1c). The extent of white matter lesion, the outer rim of elevated Mbp labeling (white dotted line), was comparable between Lrp1 control and iKO mice ( Figure 1d). As shown in Figure 1c, the area that failed to undergo repair, the inner rim of elevated Mbp labeling (yellow solid line), was larger in Lrp1 iKO mice ( Figure 1c). Quantification of lesion repair revealed a significant decrease in Lrp1 iKO mice compared to Lrp1 control mice ( Figure 1e). As an independent assessment, serial sections were stained for Pdgfra, Plp1, and Mag transcripts and revealed fewer labeled cells within the lesion of iKO mice (Figure 1-figure supplement 2e). Together these studies indicate that in adult mice, Lrp1 is required for the timely repair of a chemically induced white matter lesion. When coupled with the broad expression of Lrp1 in different neural cell types , this prompted further studies to examine whether Lrp1 function in the OL lineage is important for CNS white matter repair.

OL-lineage specific ablation of Lrp1 impairs timely repair of damaged white matter
To determine the cell autonomy of Lrp1 in adult white matter repair, we generated Lrp1 flox/flox ; Pdgfra-CreER TM (Lrp1 iKO OL ) mice that allow inducible gene ablation selectively in OPCs in adult mice. At P56 Lrp1 iKO OL mice were injected with TM and one month later subjected to unilateral injection of LPC into the corpus callosum and PBS on the contralateral side. Lrp1 control mice, harboring at least one wildtype or non-recombined Lrp1 allele, were processed in parallel. Twenty-one days post LPC/PBS injection (21 DPI), brains were collected and serially sectioned (Figure 1f). Detection of the initial white matter lesion and quantification of the extent of white matter repair was assessed as described above (Figure 1b). The initial size of the LPC inflicted white matter lesion was comparable between Lrp1 control and iKO OL mice (Figure 1g and h). However, the extent of lesion repair was significantly decreased in Lrp1 iKO OL mice (Figure 1g and i). This demonstrates an OL-linage-specific role for Lrp1 in the timely repair of a chemically induced white matter lesion.

Lrp1 is important for proper CNS myelin development and optic nerve conduction
To examine whether Lrp1 is required for proper CNS myelin development, we generated Lrp1 flox/ flox ;Olig2-Cre mice (Lrp1 cKO OL ) ( Figure 2-figure supplement 1a). Lrp1 cKO OL pups are born at the expected Mendelian frequency and show no obvious abnormalities at the gross anatomical level (data not shown). LRP1 protein levels in the brains of P10, P21, and P56 Lrp1 control and cKO OL mice were analyzed by Western blot analysis and revealed a partial loss of LRP1b (Figure 2-figure supplement 1b). The partial loss of LRP1b in brain lysates of Lrp1 cKO OL mice is due to Lrp1 expression in several other neural cell types. Olig2-Cre mice express cre recombinase under the (d) Quantification of the initial lesion size (lesion out ) in Lrp1 control (n = 8) and iKO (n = 6) mice. (e) Quantification of white matter repair in Lrp1 control (n = 8) and iKO (n = 6) mice. The extent of repair was calculated as the percentile of (lesion out -lesion int )/(lesion out ) x 100. (f) Timeline in weeks showing when OL-lineage-specific Lrp1 ablation (Lrp1 flox/flox ;Pdgfra-CreER TM , Lrp1 iKO OL ) was induced, LPC injected, and animals sacrificed. (g) Coronal brain sections through the CC at 21 days post LPC injection of Lrp1 control and iKO OL mice. The initial lesion area is demarcated by a white dashed-line. A solid yellow line delineates the non-myelinated area. Scale bar = 200 mm. (h) Quantification of the initial lesion size in Lrp1 control (n = 4) and iKO OL (n = 4) mice. (i) Quantification of white matter repair in Lrp1 control (n = 4) and iKO OL (n = 4) mice. Results are shown as mean ±SEM, *p<0.05, **p<0.01, and ***p<0.001, Student's t-test. For a detailed statistical report, see Figure 1-source data 1. DOI: https://doi.org/10.7554/eLife.30498.002 The following source data and figure supplements are available for figure 1: Source data 1. Raw data and detail statistical analysis report. DOI: https://doi.org/10.7554/eLife.30498.005 endogenous Olig2 promoter, rendering mice haploinsufficient for Olig2. Loss of one allele of Olig2 has been shown to reduce Mbp mRNA expression in neonatal mouse spinal cord (Liu et al., 2007). Therefore, we examined whether the presence of the Olig2-Cre allele influences LRP1b, MAG, CNP, or MBP in P21 brain lysates. Quantification of protein levels revealed no differences between Lrp1 flox/+ and Lrp1 flox/+ ;Olig2-Cre mice (Figure 2-figure supplement 1c). However, LRP1b is reduced in P10 and P21 Lrp1 flox/flox ;Olig2-Cre mice, compared to Lrp1 flox/flox or Lrp1 flox/+ ;Olig2-Cre mice ( Figure 2-figure supplement 1d).
To examine whether Lrp1 cKO OL mice exhibit defects in myelin development, optic nerves were isolated at P10, the onset of myelination; at P21, near completion of myelination; and at P56, when myelination is thought to be completed. Ultrastructural analysis at P10 revealed no significant difference in myelinated axons between Lrp1 control (17 ± 6%) and cKO OL (7 ± 2%) optic nerves. At P21 and P56, the percentile of myelinated axons in the optic nerve of cKO OL mice (49 ± 4% and 66 ± 5%, respectively) is significantly reduced compared to controls (70 ± 2% and 88 ± 1%, respectively) ( Figure 2a and b). In Lrp1 cKO OL mice, hypomyelination is preferentially observed in intermediate to small caliber axons (Figure 2-figure supplement 1f-1n). As an independent assessment of fiber structure, the g-ratio was determined. At P10, P21, and P56 the average g-ratio of Lrp1 cKO OL optic fibers is significantly larger than in age-matched Lrp1 control mice (Figure 2c and To examine whether Lrp1 in OLs is required for nodal organization, optic nerve sections of P21 Lrp1 control and cKO OL mice were immunostained for sodium channels (PanNaCh) and the paranodal axonal protein (Caspr). Nodal density, the number of PanNaCh + clusters in longitudinal optic nerve sections is significantly reduced in Lrp1 cKO OL mice (Figure 2d and g). In addition, an increase in nodal structural defects, including elongated nodes, heminodes, and nodes in which sodium channel staining is missing, was observed in mutant nerves. Quantification revealed an increase of nodal structural defects from 13.7 ± 1.3% in Lrp1 control mice to 33.4 ± 2.9% in cKO OL optic nerves (Figure 2e and h). Sodium channel staining associated with large caliber (>1 mm) axons was increased and staining associated with small (<0.5 mm) caliber axons was reduced (Figure 2f  To assess whether structural defects observed in optic nerve of Lrp1 cKO OL mice are associated with impaired nerve conduction, we used electrophysiological methods to measure compound action potentials (CAPs) in acutely isolated nerves (Figure 2-figure supplement 2). Recordings revealed a modest but significant delay in a subpopulation of myelinated optic nerve axons. The observed changes in conduction in Lrp1 cKO OL nerves fit well with defects at the ultrastructural level and aberrant node assembly.
Conditional ablation of Lrp1 in the OL-lineage attenuates OPC differentiation CNS hypomyelination in Lrp1 cKO OL mice may be the result of reduced OPC production or impaired OPC differentiation into myelin producing OLs. To distinguish between these two possibilities, optic nerve cross-sections were stained with anti-PDGFRa, a marker for OPCs; anti-Olig2, to account for all OL lineage cells; and anti-CC1, a marker for mature OLs. No change in OPC density was observed, but the number of mature OLs was significantly reduced in Lrp1 cKO OL mice (Figure 3a and c). Optic nerve ISH for Pdgfra revealed no reduction in labeled cells in Lrp1 cKO OL mice, a finding consistent with anti-PDGFRa immunostaining. The density of Plp and Mag expressing cells, however, is significantly reduced in the optic nerve cross-sections and longitudinal-sections of Lrp1 cKO OL mice (Figure 3b and d). The optic nerve cross-sectional area is not different between Lrp1 control and cKO OL mice ( Figure 3-figure supplement 1a). Together these studies show that OPCs are present at normal density and tissue distribution in Lrp1 cKO OL mice, but apparently fail to generate sufficient numbers of mature, myelin-producing OLs.

Loss of Lrp1 attenuates OPC differentiation in vitro
To independently assess the role of Lrp1 in OL differentiation, we isolated OPCs from brains of Lrp1 control and cKO OL pups ( Figure 3e). OPCs were kept in PDGF-AA containing growth medium (GM) or switched to differentiation medium (DM) containing triiodothyronine (T3). Staining for the proliferation marker Ki67 did not reveal any change in OPC proliferation in Lrp1 cKO OL cultures after 1 or 2 days in GM (Figure 3-figure supplement 1d-1f). After 3 days in DM, the number of NG2 + and (c) Averaged g-ratio of Lrp1 control and cKO OL optic nerve fibers from four mice per genotype for each of the three time points. At P10, n = 488 myelinated axons for control and n = 261 for cKO OL ; at P21, n = 1015 for control and n = 997 for cKO OL ; at P56, n = 1481 for control and n = 1020 for cKO OL mice. (d) Nodes of Ranvier in P21 optic nerves of Lrp1 control and cKO OL mice were labeled by anti-PanNaCh (green, node) and anti-Caspr (red, paranode) staining. Scale bar = 1 mm. (e) Nodal defects detected include elongated node, heminode, and missing node (Na + channels absent). (f) Representative nodal staining categorized by axon diameter. (g) Quantification of nodal density in P21 Lrp1 control (n = 6) and cKO OL (n = 5) optic nerves. (h) Quantification of abnormal nodes of Ranvier in Lrp1 control (n = 6) and cKO OL (n = 5) optic nerves. (i) Quantification of nodes associated with large (>1 mm), intermediate (0.5-1 mm), and small caliber fibers (<0.5 mm) in Lrp1 control (n = 6) and cKO OL (n = 5) optic nerves. Results are shown as mean ±SEM, *p<0.05, **p<0.01, and ***p<0.001, Student's t-test. For a detailed statistical report, see    (Figure 3f and h). An abundant signal for LRP1b was detected in Lrp1 control lysate, but LRP1 was not detectable in Lrp1 cKO OL cell lysate, demonstrating efficient gene deletion in the OL linage ( Figure 3g). Moreover, a significant reduction in CNP, MAG, and PLP was detected in Lrp1 cKO OL cell lysates (Figure 3-figure supplement 1g and i). Importantly, halplosufficiency for Olig2 in cultures prepared from Lrp1 flox/+ and Lrp1 flox/+ ;Olig2-Cre pups, did not reveal any difference in MAG, PLP, or LRP1b protein ( Figure 3-figure supplement 1b and c). As LRP1 signaling is known to regulate ERK1/2 and AKT activity (Yoon et al., 2013), immunoblots were probed for pAKT(S473) and pErk1/2. When normalized to total AKT, levels of pAKT are reduced in Lrp1 cKO OL lysate, while pERK1/2 levels are comparable between Lrp1 control and cKO OL lysates (Figure 3-figure supplement 1h and j). Extended culture of Lrp1-deficient OLs in DM for 5 days is not sufficient to restore myelin protein levels. Compared to Lrp1 control cultures, mutants show significantly fewer MAG + , PLP + , and MBP + cells (Figure 3i and k) and immunoblotting of cell lysates revealed a reduction in total CNP, MAG, PLP, and MBP (Figure 3j and l). Collectively, our studies demonstrate a cell-autonomous function for Lrp1 in the OL lineage, important for OPC differentiation into myelin sheet producing OLs.

Lrp1 deficiency in OPCs and OLs causes a reduction in free cholesterol
While LRP1 has been implicated in cholesterol uptake and homeostasis in non-neural cell types (van de Sluis et al., 2017), a role in cholesterol homeostasis in the OL-lineage has not yet been investigated. We find that Lrp1 À/À OPCs, prepared from Lrp1 flox/flox ;Olig2-Cre, have reduced levels of free cholesterol compared to Lrp1 control OPCs (Figure 4a and b). Levels of cholesteryl-ester are very low in the CNS (Bjö rkhem and Meaney, 2004) and near the detection limit in Lrp1 control and Lrp1 À/À OPCs ( Figure 4c). Morphological studies with MBP + OLs revealed a significant reduction in myelin-like membrane sheet expansion in Lrp1 À/À OLs (Figure 4d and e), reminiscent of wildtype OLs cultures treated with statins to inhibit HMG-CoA reductase, the rate limiting enzyme in the cholesterol biosynthetic pathway (Maier et al., 2009;Paintlia et al., 2010;Smolders et al., 2010). To assess cholesterol distribution in primary OLs, cultures were stained with filipin. In Lrp1 control OLs, staining was observed on myelin sheets and was particularly strong near the cell soma. In Lrp1 À/À OLs, filipin and MBP staining were significantly reduced ( Figure 4f). Reduced filipin staining is not simply a reflection of smaller cell size, as staining intensity was decreased when normalized to myelin sheet surface area ( Figure 4g). Thus, independent measurements revealed a dysregulation of cholesterol homeostasis in Lrp1 À/À OPCs/OLs. Cellular lipid homeostasis is regulated by a family of membrane-bound basic helix-loop-helix transcription factors, called sterol-regulatory element-binding proteins (SREBPs). To assess whether Lrp1 deficiency leads to an increase in SREBP2, OLs were cultured for 3 days in DM and analyzed by immunoblotting. OL cultures prepared from Lrp1 flox/+ and Lrp1 flox/+ ;Olig2-cre pups showed very similar levels of SREBP2. In marked contrast, we observed a strong upregulation of SREBP2 in Lrp1 À/À cultures (Figure 4i and Figure 4-figure supplement 1a and b). Elevated SREBP2 in mutant cultures can be reversed by exogenous cholesterol directly added to the culture medium (Figure 4i and j). This shows the existence of LRP1-independent cholesterol uptake mechanisms in Lrp1 À/À OLs and a normal physiological response to elevated levels of cellular cholesterol. In Lrp1 control cultures, bath application of cholesterol leads to a small, yet significant decrease in SREBP2 ( Figure 4j). Given the importance of cholesterol in OL maturation (Krämer-Albers et al., 2006;Mathews et al., 2014;Saher et al., 2005), we examined whether the differentiation block can be rescued by bathapplied cholesterol. Remarkably, cholesterol treatment of Lrp1 À/À OLs for 3 days failed to augment PLP, MAG or CNP to control levels (Figure 4k-n). While cholesterol treated Lrp1 À/À OLs showed a modest increase in PLP, levels remained below Lrp1 controls. Moreover, prolonged cholesterol treatment for 5 days failed to increase PLP levels (Figure 4o and p) or the number of MBP + OLs in Lrp1 À/À cultures (Figure 4q and r). Although differentiation of Lrp1 À/À OLs cannot be 'rescued' by bath applied cholesterol, cells are highly sensitive to a further reduction in cholesterol, as shown by bath applied simvastatin (

Lrp1 deficiency impairs peroxisome biogenesis
To further investigate what type of biological processes might be dysregulated by Lrp1 deficiency, we performed transcriptomic analyses of OPCs acutely isolated from Lrp1 control and cKO OL pups. Gene ontology (GO) analysis identified differences in 'peroxisome organization' and 'peroxisome proliferation-associated receptor (PPAR) signaling pathway' (Figure 5-figure supplement 1a). Six gene products regulated by Lrp1 belong to peroxisome and PPAR GO terms, including Pex2, Pex5l, Hrasls, Ptgis, Mavs, and Stard10 ( Figure 5-figure supplement 1b). Western blot analysis of Lrp1 À/À OLs further revealed a significant reduction in PEX2 after 5 days in DM (Figure 5-figure supplement 1c and d). Because PEX2 has been implicated in peroxisome biogenesis (Gootjes et al., 2004), and peroxisome biogenesis disorders (PBDs) are typically associated with impaired lipid metabolism and CNS myelin defects (Krause et al., 2006), this prompted us to further explore a potential link between LRP1 and peroxisomes. To assess whether the observed reduction in PEX2 impacts peroxisome density in primary OLs, MBP + OLs were stained with anti-PMP70, an ATP-binding cassette transporter enriched in peroxisomes (Figure 5a). In Lrp1 À/À OLs, we observed reduced PMP70 staining ( Figure 5b) and a decrease in the total number of peroxisomes ( Figure 5c). Normalization of peroxisome counts to cell size revealed that the reduction in Lrp1 À/À OLs is not simply a reflection of smaller cells (Figure 5d). The subcellular localization of peroxisomes is thought to be important for ensuring a timely response to metabolic demands (Berger et al., 2016). This prompted us to analyze the distribution of peroxisomes in primary OLs. Interestingly, while the number of PMP70 positive puncta near the cell soma is comparable between Lrp1 control and Lrp1 À/À OLs, we observed a significant drop in peroxisomes along radial processes of MBP + OLs (Figure 5eh).

Combination treatment of cholesterol and PPARg agonist rescues the differentiation block in Lrp1-deficient OPCs
In endothelial cells, the LRP1-ICD functions as a co-activator of PPARg, a key regulator of lipid and glucose metabolism (Mao et al., 2017). Activated PPARg moves into the nucleus to control gene expression by binding to PPAR-responsive elements (PPREs) on numerous target genes, including Lrp1 (Gauthier et al., 2003). In addition, PPREs are found in genes important for lipid and glucose per mm 2 (g). For Lrp1 control and cKO OL OLs, n = 29 cells from three mice in each group. (h) Timeline in days showing when growth medium (GM) or differentiation medium (DM) with (+) or without (-) Chol was added and when cells were harvested. (i and k) Immunoblotting of OL lysates prepared from Lrp1 control and cKO OL cultures after 3 days in DM. Representative blots were probed with anti-LRP1b, anti-SREBP2, anti-b-actin, anti-PLP, anti-MAG, anti-CNP, and anti-GAPDH. (j, l-n) Quantification of SREBP2 (j), PLP (l), MAG (m), and CNP (n) in Lrp1 control and cKO OL cultures ± bath applied Chol. Number of independent immunoblots: anti-PLP and MAG, n = 3 per condition; anti-SREBP2 and anti-CNP, n = 4 per condition. (o) Immunoblotting of OL lysates prepared from Lrp1 control and cKO OL cultures after 5 days in DM ±bath applied Chol. Representative blots were probed with anti-LRP1b, anti-PLP/DM20, and anti-b-actin. (p) Quantification of PLP (n = 4 per condition) in Lrp1 control and cKO OL cultures ± bath applied Chol (q) Immunostaining of OLs after 5 days in DM ±bath applied Chol. Primary OLs stained with anti-MBP and Hoechst dye33342. Scale bar = 100 mm. (r) Quantification showing relative number of MBP + cells in Lrp1 control and cKO OL cultures (n = 3-5 per condition). Results are shown as mean values ± SEM, *p<0.05, **p<0.01, and ***p<0.001, 2-way ANOVA, post hoc t-test. For a detailed statistical report, see  (Fang et al., 2016;Hofer et al., 2017). In vitro, a 5-day treatment of Lrp1 control OPCs with pioglitazone, an agonist of PPARg, results in elevated LRP1 (Figure 6a and b) and accelerated differentiation into MBP + OLs (Figure 6c and d) (Bernardo et al., 2009). This stands in marked contrast to Lrp1 À/À cultures, where pioglitazone treatment fails to accelerate OPC differentiation (Figure 6c and d). Moreover, pioglitazone does not regulate PMP70 staining intensity in MBP + Lrp1 control or Lrp1 À/À OLs, nor does it have any effect on total peroxisome counts per cell (Figure 6e-i). However, pioglitazone leads to a modest but significant increase in the number of peroxisomes located in cellular processes of Lrp1 À/À OLs (Figure 6j and k). Treatment of Lrp1 control OPCs with the PPARg antagonist GW9662 blocks differentiation into MBP + OLs (Roth et al., 2003), but does not lead to a further reduction in MBP + cells in Lrp1 À/À OL cultures (Figure 6l and m). This suggests that in Lrp1 À/À OLs PPARg is not active.
Given LRP1's multifunctional receptor role, we asked whether simultaneous treatment with pioglitazone and cholesterol is sufficient to rescue the differentiation block of Lrp1 À/À OPCs (Figure 7a). This is indeed the case, as the number of MBP + cells in Lrp1 À/À cultures is significantly increased by the combination treatment (Figure 7b and c). Moreover, the size of MBP + Lrp1 À/À OLs increased (Figure 7d and f) and peroxisome counts are elevated (Figure 7e and g), however the anti-MBP staining intensity was only partially rescued ( Figure 7h). Quantification of peroxisome distribution in Lrp1 À/À OPC/OL cultures subjected to combo treatment revealed a marked increase in PMP70 + peroxisomes in OL processes (Figure 7i and j). Together, these findings indicate that LRP1 regulates multiple metabolic functions important for OL differentiation. In addition to its known role in cholesterol homeostasis, LRP1 regulates expression of PEX2 and thereby metabolic functions associated with peroxisomes.

Discussion
LRP1 function in the OL-lineage is necessary for proper CNS myelin development and the timely repair of a chemically induced focal white matter lesion in vivo. Optic nerves of Lrp1 cKO OL show fewer myelinated axons, thinning of myelin sheaths, and an increase in nodal structural defects. Morphological alterations have a physiological correlate, as Lrp1 cKO OL mice exhibit faulty nerve conduction. Mechanistically, Lrp1 deficiency disrupts multiple signaling pathways implicated in OL differentiation, including AKT activation, cholesterol homeostasis, PPARg signaling, peroxisome biogenesis and subcellular distribution. The pleiotropic roles of LRP1 in OPC differentiation are further underscored by the fact that restoring cholesterol homeostasis or activation of PPARg alone is not sufficient to drive differentiation. Only when cholesterol supplementation is combined with PPARg activation, is differentiation of Lrp1 À/À OPC into MBP + OLs significantly increased. Taken together, our studies identify a novel role for LRP1 in peroxisome function and suggest that broad metabolic dysregulation in Lrp1-deficient OPCs attenuates differentiation into mature OLs (Figure 8).
In the embryonic neocortex, LRP1 is strongly expressed in the ventricular zone and partially overlaps with nestin + neural stem and precursor cells (Hennen et al., 2013). In Lrp1 flox/flox neurospheres, conditional gene ablation reduces cell proliferation, survival, and negatively impacts differentiation into neurons and O4 + OLs (Safina et al., 2016). Consistent with these observations, Lrp1 cKO OL mice show reduced OPC differentiation in vivo. Studies with purified OPCs in vitro and OL-linage  Figure 6 continued on next page specific gene ablation in vivo, suggest a cell-autonomous role for Lrp1 in OPC maturation. Non-cellautonomous functions for LRP1 following white matter lesion are likely, since LRP1 is upregulated in astrocytes and myeloid cells near multiple sclerosis lesions (Chuang et al., 2016). Moreover, deletion of Lrp1 in microglia worsens the course of experimental autoimmune encephalomyelitis and has been proposed to promote a proinflammatory milieu associated with disease exacerbation (Chuang et al., 2016). Studies with Lrp1 iKO OL mice show that Lrp1 in the OL linage is necessary for the timely repair of a focal myelin lesion. This suggests that similar to OPCs in the developing brain, OPCs in the adult brain depend on Lrp1 for rapid differentiation into myelin producing OLs. Since white matter repair was analyzed by repopulation of the lesion area with Mbp + cells, additional studies, including electron microscopy, will be needed to demonstrate a requirement for Lrp1 in remyelination of denuded axons. Cholesterol does not cross the blood-brain-barrier (Saher and Stumpf, 2015) and CNS resident cells need to either synthesize their own cholesterol or acquire it through horizontal transfer from neighboring cells, including astrocytes (Camargo et al., 2017). In the OL-lineage cholesterol is essential for cell maturation, including myelin gene expression, myelin protein trafficking, and internode formation (Krämer-Albers et al., 2006;Mathews et al., 2014;Saher et al., 2005). Sterol biosynthesis is in part accomplished by peroxisomes. Specifically, the pre-squalene segment of the cholesterol biosynthetic pathway takes place in peroxisomes. However, cholesterol is only one of many lipid derivatives produced by this pathway (Faust and Kovacs, 2014). A drop in intracellular cholesterol leads to an increase in SREBPs, a family of transcription factors that regulate expression of gene products involved in cholesterol and fatty acid synthesis (Faust and Kovacs, 2014;Goldstein et al., 2006). In Schwann cells, SREBPs and the SREBP-activating protein SCAP are required for AKT/mTOR-dependent lipid biosynthesis, myelin membrane synthesis, and normal PNS myelination (Norrmén et al., 2014;Verheijen et al., 2009). In the OL linage blockage of SREBP inhibits CNS myelination (Camargo et al., 2017;Monnerie et al., 2017). Blocking of SREBP processing in primary OLs leads to a drop in cholesterol and inhibits cell differentiation and membrane expansion. This can be rescued by cholesterol added to the culture medium (Monnerie et al., 2017). In primary OLs, Lrp1 deficiency leads to activation of SREBP2, yet cells are unable to maintain cholesterol homeostasis, suggesting more global metabolic deficits. The cholesterol sensing apparatus in Lrp1-deficient OPCs appears to be largely intact, as bath applied cholesterol restores SREBP2 to control levels. Since SREBP2 can be induced by ER stress (Faust and Kovacs, 2014), reversibility by bath applied cholesterol suggests that Lrp1 cKO OL cultures upregulate SREBP2 due to cholesterol deficiency and not elevated ER stress (Faust and Kovacs, 2014). Significantly, restoring cellular cholesterol homeostasis in Lrp1 À/À OPCs is not sufficient to overcome the differentiation block, suggesting more widespread functional deficits.
Members of the PPAR subfamily, including PPARa, PPARb/d, and PPARg, are ligand-activated transcription factors that belong to the nuclear hormone receptor family (Michalik et al., 2006). PPARs regulate transcription through heterodimerization with the retinoid X receptor (RXR). When activated by a ligand, the dimer modulates transcription via binding to a PPRE motif in the promoter region of target genes (Michalik et al., 2006). PPARs-regulated gene expression controls numerous  4). (e-f) Primary OLs probed with anti-MBP and anti-PMP70. Scale bar = 10 mm. (g-i) Quantification of OL size in mm 2 (g), the number of PMP70 + puncta (h), and the intensity of MBP staining per cell (i). (j) Distribution of peroxisomes as a function of distance from the cell center in Lrp1 control and cKO OL OLs treated ±Pio. The number of PMP70 + peroxisomes between 6-15 mm in Lrp1 control and cKO OL cultures was subjected to statistical analysis in (k). Lrp1 control (n = 112 cells, three mice), Lrp1 control cultures with Pio (n = 180 cells, three mice), cKO OL (n = 60 cells, three mice), and cKO OL cultures with Pio. (n = 110 cells, three mice) (k). (l) Immunostaining of OLs after 5 days in DM ±GW9662, probed with anti-MBP and Hoechst dye33342. Scale bar = 50 mm. (m) Quantification of MBP + cells under each of the four different conditions (n = 3 per condition). Results are shown as mean values ± SEM, *p<0.05, **p<0.01, and ***p<0.001, 2-way ANOVA, post hoc t-test. For a detailed statistical report, see Figure 6-source data 1. DOI: https://doi.org/10.7554/eLife.30498.019 The following source data is available for figure 6: Source data 1. Raw data and detail statistical analysis report. DOI: https://doi.org/10.7554/eLife.30498.020  Figure 7 continued on next page biochemical pathways implicated in lipid, glucose and energy metabolism (Berger and Moller, 2002;Han et al., 2017). A critical role for PPARg in OL differentiation is supported by the observation that activation with pioglitazone or rosiglitazone accelerates OPC differentiation into mature OLs (Bernardo et al., 2009;Bernardo et al., 2013;De Nuccio et al., 2011;Roth et al., 2003;Saluja et al., 2001) and inhibition with GW9662 blocks OL differentiation (Bernardo et al., 2017). Deficiency for the PPARg-coactivator-1 alpha (PGC1a) leads to impaired lipid metabolism, including an increase in very long chain fatty acids (VLCFAs) and disruption of cholesterol homeostasis (Camacho et al., 2013;Xiang et al., 2011). In addition, PGC1a deficiency results in defects of peroxisome-related gene function, suggesting the increase in VLCFAs and drop in cholesterol reflects impaired peroxisome function (Baes and Aubourg, 2009). Following g-secretase-dependent processing, the LRP1 ICD can translocate to the nucleus where it associates with transcriptional regulators (Carter, 2007;May et al., 2002). In endothelial cells, the LRP1-ICD binds directly to the nuclear receptor PPARg to regulate gene products that function in lipid and glucose metabolism (Mao et al., 2017). Treatment of Lrp1 À/À OPCs with pioglitazone leads to an increase in peroxisomes in OL processes but fails to promote differentiation into myelin sheet producing OLs. In the absence of the LRP1-ICD, pioglitazone may fail to fully activate PPARg (Mao et al., 2017), but the observed increase in PMP70 + peroxisomes in OL processes of Lrp1 deficient cultures suggests that mutant cells still respond to pioglitazone. Because Lrp1 cKO OL cultures are cholesterol deficient and the LRP1-ICD participates in PPARg regulated gene expression, we examined whether a combination treatment rescues the differentiation block in Lrp1 deficient OPCs/OLs. This was indeed the case, suggesting that Lrp1 deficiency leads to dysregulation of multiple pathways important for OPC differentiation.
The importance of peroxisomes in the human nervous system is underscored by inherited disorders caused by complete or partial loss of peroxisome function, collectively described as Zellweger spectrum disorders (Berger et al., 2016;Waterham et al., 2016). PEX genes encode peroxins, proteins required for normal peroxisome assembly. Defects in PEX genes can cause peroxisome biogenesis disorder (PBD), characterized by a broad range of symptoms, including aberrant brain development, white matter abnormalities, and neurodegeneration (Berger et al., 2016). The genetic basis for PBD is a single gene mutation in one of the 14 PEX genes, typically leading to deficiencies in numerous metabolic functions carried out by peroxisomes (Steinberg et al., 1993). Mounting evidence points to a close interaction of peroxisomes with other organelles, mitochondria in particular, and disruption of these interactions may underlie the far reaching metabolic defects observed in PBD and genetically manipulated model organisms deficient for a single PEX (Fransen et al., 2017;Wangler et al., 2017). In developing OLs, Lrp1 deficiency leads to a decrease in peroxisomal gene products, most prominently a~50% reduction in PEX2, an integral membrane protein that functions in the import of peroxisomal matrix proteins. Mice deficient for Pex2 lack normal peroxisomes but do assemble empty peroxisome membrane ghosts (Faust and Hatten, 1997). Pex2 mutant mice show significantly lower plasma cholesterol levels and in the brain the rate of cholesterol synthesis is significantly reduced (Faust and Kovacs, 2014); brain size is reduced, cerebellar development impaired, and depending on the genetic background death occurs in early postnatal life (Faust, 2003). Mutations in human PEX2 cause Zellweger spectrum disorder but have no apparent impact on white matter appearance (Mignarri et al., 2012). In mice, CNP-Cre mediated ablation of Pex5 in OLs disrupts peroxisome function and integrity of myelinated fibers, but does not impair Distribution of peroxisomes as a function of distance from the cell center in Lrp1 control and cKO OL cultures with (+) or without (-) Pio and Chol combotreatment. The number of PMP70 + peroxisomes between 5-15 mm in Lrp1 control and cKO OL cultures was subjected to statistical analysis in (j). Lrp1 control cultures with vehicle (n = 210 cells, three mice), Lrp1 control cultures treated with Pio and Chol (n = 208 cells, three mice), cKO OL cultures treated wtih vehicle (n = 199 cells, three mice), and cKO OL cultures treated wtih Pio and Chol (n = 190 cells, three mice) (k). Results are shown as mean values ± SEM, *p<0.05, **p<0.01, and ***p<0.001, 2-way ANOVA, post hoc t-test. For a detailed statistical report, see Figure 7-source data 1. DOI: https://doi.org/10.7554/eLife.30498.021 The following source data is available for figure 7: Source data 1. Raw data and detail statistical analysis report. DOI: https://doi.org/10.7554/eLife.30498.022 CNS myelinogenesis (Kassmann et al., 2007). This suggests that defects in CNS myelinogenesis observed in Lrp1 cKO OL mice are likely not only a reflection of reduced peroxisome biogenesis or transport into internodes. Rather we provide evidence that Lrp1 deficiency in OPCs leads to dysregulation of additional pathways implicated in myeliogenesis, including AKT, SREBP2, and PAPRg. We propose that the combined action of these deficits attenuates OPC differentiation.
In sum, our studies show that Lrp1 is required in the OL lineage for proper CNS myelin development and the timely repair of a chemically induced white matter lesion in vivo. Mechanistic studies with primary OPCs revealed that loss of Lrp1 causes differentiation block that can be rescued by bath application of cholesterol combined with pharmacological activation of PPARg.

Materials and methods
Key resources

Mice
All animal handling and surgical procedures were performed in compliance with local and national animal care guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC). Lrp1 flox/flox mice were obtained from Steven Gonias  and crossed with Olig2-Cre (Schüller et al., 2008), CAG-CreER TM (Jackson Laboratories, #004682, Bar Harbor, ME), and Pdgfra-CreER TM (Kang et al., 2010) mice. For inducible gene ablation in adult male and female mice, three intraperitoneal (i.p.) injections of tamoxifen (75 mg/kg) were given every 24 hr. Tamoxifen (10 mg/ml) was prepared in a mixture of 9% ethanol and 91% sunflower oil. Mice were kept on a mixed background of C57BL/6J and 129SV. Throughout the study, male and female littermate animals were used. Lrp1 'control' mice harbor at least one functional Lrp1 allele. Any of the following genotypes Lrp1 +/+ , Lrp1 +/flox , Lrp1 flox/flox , or Lrp1 flox/+ ;Cre + served as Lrp1 controls.

Genotyping
To obtain genomic DNA (gDNA), tail biopsies were collected, boiled for 30 min in 100 ml alkaline lysis buffer (25 mM NaOH and 0.2 mM EDTA in ddH 2 O) and neutralized by adding 100 ml of 40 mM Tris-HCI (pH 5.5). For PCR genotyping, 1-5 ml of gDNA was mixed with 0.5 ml of 10 mM dNTP mix (Promega, C1141, Madison, WI), 10 ml of 25 mM MgCl 2 , 5 ml of 5X Green GoTaq Buffer (Promega, M791A), 0.2 ml of GoTaq DNA polymerase (Promega, M3005), 0.15 ml of each PCR primer stock (90 mM each), and ddH 2 O was added to a total volume of 25 ml. The following cycling conditions were used: DNA denaturing step (94˚C for 3 min) 1X, amplification steps (94˚C for 30 s, 60˚C for 1 min, and 72˚C for 1 min) 30X, followed by an elongation step (72˚C for 10 min) then kept at 4˚C for storage. The position of PCR primers used for genotyping is shown in Figure 1

Stereotaxic injection
Male and female mice at postnatal-day (P) 42-56 were used for stereotaxic injection of L-a-Lysophosphatidylcholine (LPC) (Sigma, L4129, Mendota Heights, MN) into the corpus callosum. Mice were anesthetized with 4% isoflurane mixed with oxygen, mounted on a Stoelting stereotaxic instrument (51730D, Wood Dale, IL), and kept under 2% isoflurane anesthesia during surgery. A 5ml-hamilton syringe was loaded with 1% LPC in PBS (Gibco, 10010023, Gaithersburg, MD), mounted on a motorized stereotaxic pump (Stoelting Quintessential Stereotaxic injector, 53311) and used for intracranial injection at the following coordinates, AP: 1.25 mm, LR:±1 mm, D: 2.25 mm. Over a duration of 1 min, 0.5 ml of 1% LPC solution was injected on the ipsilateral site and 0.5 ml PBS on the contralateral side. After the injection was completed, the needle was kept in place for 2 min before retraction. Following surgery, mice were treated with three doses of 70 ml of buprenorphine (0.3 mg/ml) every 12 hr. Brains were collected at day 10, and 21 post injection.

Histochemistry
Animals were deeply anesthetized with a mixture of ketamine/xylazine (25 mg/ml ketamine and 2.5 mg/ml xylazine in PBS) and perfused trans-cardially with ice-cold PBS for 5 min, followed by ice-cold 4% paraformaldehyde in PBS (4%PFA/PBS) for 5 min. Brains were harvested and post-fixed for 2 hr in perfusion solution. Optic nerves were harvested separately and post-fixed for 20 min in perfusion solution. Brains and optic nerves were cryoprotected overnight in 30% sucrose/PBS at 4˚C, embedded in OCT (Tissue-Tek, 4583, Torrance, CA), and flash frozen in powderized dry ice. Serial sections were cut at 20 mm (brains) and 10 mm (optic nerves) at À20˚C using a Leica CM 3050S Cryostat.
Serial sections were mounted onto Superfrost + microscope slides (Fisherbrand,Pittsburgh,PA) and stored at À20˚C.

In situ hybridization
Tissue sections mounted on microscope slides were post-fixed overnight in 4%PFA/PBS at 4˚C. Sections were then rinsed 3 times for 5 min each in PBS and the edge of microscope slides was demarcated with a DAKO pen (DAKO, S2002, Denmark). Sections were subsequently incubated in a series of ethanol/water mixtures: 100% for 1 min, 100% for 1 min, 95% for 1 min, 70% for 1 min, and 50% for 1 min. Sections were then rinsed in 2x saline-sodium citrate (SSC, 150 mM NaCl, and 77.5 mM sodium citrate in ddH 2 O, pH7.2) for 1 min, and incubated at 37˚C for 30 min in proteinase K solution (10 mg/ml proteinase K, 100 mM Tris-HCl pH8.0, and 0.5 mM EDTA in ddH 2 O). Proteinase digestion was stopped by rinsing sections in ddH 2 O and then in PBS for 5 min each. To quench RNase activity, slides were incubated in 1% triethanolamine (Sigma, 90278) and 0.4% acetic anhydride (Sigma, 320102) mixture in ddH 2 O for 10 min at room temperature, rinsed once in PBS for 5 min and once in 2X SSC for another 5 min. To reduce non-specific binding of cRNA probes, sections were pre-incubated with 125 ml hybridization buffer (10% Denhardts solution, 40 mg/ml baker's yeast tRNA, 5 mg/ml sheared herring sperm DNA, 5X SSC, and 50% formamide in ddH 2 O) for at least 2 hr at room temperature. Digoxigenin-labeled cRNA probes were generated by run-off in vitro transcription as described (Winters et al., 2011). Anti-sense and sense cRNA probes were diluted in 125 ml prehybridization buffer to~200 ng/ml, denatured for 5 min at 85˚C, and rapidly cooled on ice for 2 min. Probes were applied to tissue sections, microscope slides covered with parafilm, and incubated at 55˚C overnight in a humidified and sealed container. The next morning slides were rinsed in 5X SSC for 1 min at 55˚C, 2X SSC for 5 mins at 55˚C, and incubated in 0.2X SSC/50% formamide for 30 min at 55˚C. Sections were rinsed in 0.2X SSC at room temperature for 5 min then rinsed with Buffer1 (100 mM Tris-HCl pH7.5, and 1.5M NaCl in ddH 2 O) for 5 min. A 1% blocking solution was prepared by dissolving 1 g blocking powder (Roche, 11096176001, Switzerland) in Buffer1 at 55˚C, cooled to room temperature (RT), and applied to slides for 1 hr at RT. Slides were rinsed in Buffer1 for 5 min and 125 ml anti-Digoxigenin-AP antibody (Roche, 11093274910, 1:2500) in Buffer1 was applied to each slide for 1.5 hr at RT. Sections were rinsed in Buffer1 for 5 min, then rinsed in Buffer2 (100 mM Tris-HCl pH9.5, 100 mM NaCl, and 5 mM MgCl 2 in ddH 2 O) for 5 min, and incubated in alkaline phosphatase (AP) substrate (Roche, 11681451001, 1:50) in Buffer2. The color reaction was developed for 1-48 hr and stopped by rinsing sections in PBS for 10 min. Sections were incubated in Hoechst dye 33342 (Life technology, H3570, Pittsburgh, PA) for 5 min, air dried, mounted with Fluoromount-G (SouthernBiotech, 0100-01), and dried overnight before imaging under bright-field. The following cRNA probes were used, Pdgfra and Plp (DNA templates were kindly provided by Richard Lu (Dai et al., 2014)), Mag (Winters et al., 2011), and Mbp (a 650 bp probe based on template provided in the Allen Brain Atlas).

Quantification of lesion size and myelin repair
Serial sections of the corpus callosum, containing the LPC and PBS injection sites were mounted onto glass coverslips and stained by ISH with digoxigenin-labeled cRNA probes specific for Mbp, Mag, Plp, and Pdgfra. For quantification of the white matter lesion area, the same intensity cutoff was set by Image J threshold for all brain sections and used to measure the size of the lesion. The outer rim of the strongly Mbp + region (lesion out ) was traced with the ImageJ freehand drawing tool. The inner rim facing the Mbpregion (lesion in ) was traced as well. For each animal examined, the size of the initial lesion area (lesion out ) in mm 2 and remyelinated area (lesion out -lesion in ) in mm 2 was calculated by averaging the measurement from two sections at the lesion core. The lesion core was defined as the section with the largest lesion area (lesion out ). To determine remyelination, the ratio of (lesion out -lesion in )/(lesion out ) in percent was calculated. As an initial lesion depth control, criteria of lesion out area must cover the center of the corpus callosum in each serial section set. If a lesion out area was not located within the corpus callosum, the animal and corresponding brain sections were excluded from the analysis.

Transmission electron microscopy (TEM)
Tissue preparation and image acquisition were carried out as described by Winters et al. (2011).
Briefly, mice at P10, P21, and P56 were perfused trans-cardially with ice cold PBS for 1 min, followed by a 10 min perfusion with a mixture of 3% PFA and 2.5% glutaraldehyde in 0.1M Sorensen's buffer.
Brains and optic nerves were dissected and post-fixed in perfusion solution overnight at 4˚C. Postfixed brain tissue and optic nerves were rinsed and transferred to 0.1M Sorensen's buffer and embedded in resin by the University of Michigan Imaging Laboratory Core. Semi-thin (0.5 mm) sections were cut and stained with toluidine blue and imaged by light microscopy. Ultra-thin (75 nm) sections were cut and imaged with a Philips CM-100 or a JEOL 100CX electron microscope. For each genotype and age, at least three animals were processed and analyzed. For each animal, over 1000 axons in the optic nerve were measured and quantified by ImageJ. For each optic nerve, 10 images at 13,500x magnification were randomly taken and quantified to calculate the g-ratio and the fraction of myelinated axons. The inner (area in ) and outer (area out ) rim of each myelin sheath was traced with the ImageJ freehand drawing tool and the area within was calculated. We then derived axon caliber and fiber caliber (2 r) by the following: area in = r 2 p. The g-ratios were calculated as such: ffiffiffiffiffiffiffiffi ffi area in p ffiffiffiffiffiffiffiffiffiffi area out p . The g-ratio is only accurate if the compact myelin and axon outline can clearly be traced. Individual fibers with not clearly defined features were excluded from the quantification.

Optic nerve recordings
Compound action potentials were recorded as described elsewhere (Carbajal et al., 2015;Winters et al., 2011). Briefly, optic nerves were acutely isolated from P21 mice and transferred into oxygenated ACSF buffer (125 mM NaCl, 1.25 mM NaH 2 PO 4 , 25 mM glucose, 25 mM NaHCO 3 , 2.5 mM CaCl 2 , 1.3 mM MgCl 2 , 2.5 mM KCl) for 45 min at RT before transferring into a recording chamber at 37 ± 0.4˚C. Suction pipette electrodes were used for stimulation and recording of the nerve (Figure 2-figure supplement 2a). A computer-driven (Axon pClamp10.3 software) stimulus isolation unit (WPI, FL) was used to stimulate the optic nerve with 2 mA/50 ms pulses. The recording electrode was connected to a differential AC amplifier (custom-made). A stimulus artifact-subtracting pipette was placed near the recording pipette. A data acquisition system (Axon digidata 1440A, Axon pClamp 10.3, Molecular Devices, Sunnyvale, CA) was used to digitize the signals. Conduction velocity was calculated from the length of the nerve and the time to peak of each component of the CAP. Amplitudes were normalized to a resistance ratio of 1.7, as described (Fernandes et al., 2014). Raw traces were fitted with four Gaussian curves with Origin9.1 software for analysis of individual components of the CAP. Due to limitations in the resolution of individual peaks in short nerves, CAP recordings from nerves that were shorter than 1 mm in length were excluded from the analysis.

OPC/OL primary cultures and drug treatment
OPCs were isolated from P6-P9 mouse pups by rat anti-PDGFRa (BD Pharmingen, 558774) immunopanning as described (Mironova et al., 2016). For plating of cells, 5-7.5 Â 10 3 cells (for 12 mm cover glass) or 3-5 Â 10 4 (12-well plastic plate) were seeded onto PDL pre-coated surface. Primary OPCs were kept in a 10% CO 2 incubator at 37˚C. To maintain OPCs in a proliferative state, growth medium (20 ng/ml PDGF-AA (Peprotech, 100-13A, Rocky Hill, NJ), 4.2 mg/ml Forskolin (Sigma, F6886), 10 ng/ml CNTF (Peprotech, 450-02), and 1 ng/ml NT-3 (Peprotech, 450-03) in SATO) were added to the culture. To induce OPC differentiation, differentiation medium was constituted by adding (4.2 mg/ml Forskolin, 10 ng/ml CNTF, and 4 ng/ml T3 (Sigma, T6397) in SATO) to the culture. For drug treatment, all compounds were mixed with differentiation medium at the desired concentration, and the compound-containing medium was replaced every other day. Stock and working solutions including 20 mg/ml cholesterol (Sigma, C8667) in 100% EtOH were kept at RT and warmed up to 37˚C before use, then diluted in differentiation medium to 5 mg/ml; 10 mM pioglitazone (Sigma, E6910) in DMSO was kept at À20˚C and diluted in differentiation medium to 1 mM; 10 mM simvastatin (Sigma, S6196) in DMSO was kept at À20˚C and diluted in differentiation medium to 0.5 mM; 10 mM GW9662 (Sigma, M6191) in DMSO was kept at À20˚C and diluted in differentiation medium to 1 mM.

OPC staining and quantification
At different stages of development, OPC/OL cultures were fixed for 15 min in 4%PFA/PBS. Cells were rinsed three times in PBS and permeabilized with 0.1% Triton/PBS solution for three mins. Cells were then rinsed in PBS and incubated in blocking solution (3% BSA/PBS) for 1 hr at RT. Primary antibodies were prepared in blocking solution. For immunostaining, 35 ml were dropped onto a sheet of parafilm, the coverslips were inverted onto the primary antibody drop, and incubated overnight at 4˚C. The following day, coverslips were transferred back to a 24-well-plate and rinsed with PBS 3 times for 5 min each. Secondary antibody ±filipin (Sigma, F9765, 0.1 mg/ml) was prepared in blocking solution, 350 ml were added to each well, and the coverslips were incubated for 2 hr at RT. Coverslips then were rinsed in PBS three times for 5 min each and stained with Hoechst (1:50,000) for 10 s. Coverslips then were rinsed in ddH 2 O and mounted in ProLong Gold antifade reagent. For quantification in Figure 3, the percent of OL markers +/ Hoechst + cells was calculated from 10 images that were taken from randomly selected areas in each coverslip at 20X magnification with an Olympus IX71 microscope. For quantification in Figure 4 and after, the percent of OL markers+/ Hoechst + cells was calculated from 25 images that were taken from randomly selected areas in each coverslip at 10X magnification with a Zeiss Axio-Observer microscope. For single-cell intensity and size measurement in Figures 4-7, individual cell images were taken at 40X magnification with a Zeiss Axio-Observer microscope with Apotom.2. For quantification, the same intensity cutoff was set by Image J threshold to all cells and binary images were generated to define each cell outline. The individual cell outline was applied to original images to measure the intensity of filipin, MBP, or PMP70 staining per cell. For PMP70 puncta distribution analysis, the coordinates of each PMP70 + center were acquired by the process >find maxima function in ImageJ, the cell center coordinate was defined by point selection function, and the distance of each PMP70 + dot to the cell center was then calculated. The data were then binned from 1 to 25 mm at 1 mm divisions, and plotted. Primary antibodies included: rat anti-PDGFRa (BD pharmingen, 558774, 1:500), rabbit anti-CNPase (Aves Labs, 27490 R12-2096, Tigard, OR, 1:500), mouse anti-MAG (Millipore, MAB1567, 1:500), rat anti-MBP (Millipore, MAB386, 1:1000), chicken anti-PLP (Aves Labs, 27592, 1:500), mouse anti-GFAP (Sigma, G3893, 1:1000), chicken anti-GFAP (Aves Labs, GFAP, 1:500), rabbit anti-NG2 (Millipore, AB5320, 1:500), rabbit anti-LRP1b (Abcam, ab92544, 1:500), rabbit anti-PMP70 (Thermo, PA1-650, Waltham, MA, 1:1000).

Cholesterol measurement
OPCs were isolated by immunopanning as described above. OPCs bound to panning plates were collected by scraping with a Scraper (TPP, TP99002) in 250 ml of ice-cold PBS and sonicated in an ice-cold water bath (Sonic Dismembrator, Fisher Scientific, Model 500) at 50% amplitude three times for 5 s with a 5-s interval. The sonicated cell suspensions were immediately used for cholesterol measurement following the manufacturer's instructions (Chemicon, 428901). For colorimetric detection and quantification of cholesterol, absorbance was measured at 570 nm with a Multimode Plate Reader (Molecular Devices, SpectraMax M5 e ). Results were normalized to total protein concentration measured by DC TM Protein Assay according to the manufacturer's manual (Bio-Rad, 5000112).

Microarray and gene ontology analysis
OPCs were isolated by immunopanning as described above and RNA was isolated with the RNeasy Micro Kit (Qiagen, 74004, Germany). To compare Lrp1 control and cKO OL RNA expression profiles, the Mouse Gene ST2.1 Affymetrix array was used. Differentially expressed genes, with a p-value<0.05 set as cutoff, were subjected to gene ontology (GO) analysis. Go terms were quarried from Mouse Genome Informatics (MGI) GO browser. The fold enrichment was calculated by dividing the number of genes associated with the GO term in our list by the number of genes associated with the GO term in the database.

Statistical analysis
There was no pre-experimental prediction of the difference between control and experimental groups when the study was designed. Therefore, we did not use computational methods to determine sample size a priori. Instead, we use the minimum of mice per genotype and experimental treatment for a total of at least three independent experiments to achieve the statistical power discussed by Gauch, 2006). We used littermate Lrp1 control or Lrp1 cKO or iKO mice for comparison throughout the study. All independent replicas were biological replicas, rather than technical replicas. For each experiment, the sample size (n) is specified in the figure legend. Throughout the study, independent replicas (n) indicate biological replica. Technical replicas were used to control for the quality of each measurement and were averaged before quantification and the average value was used as (n = 1) biological replica. Unless indicated otherwise, results are represented as mean value ±SEM. For single pairwise comparison, Student's t-test was used and a p-value<0.05 was considered statistically significant. For multiple comparisons, two-way ANOVA followed by post hoc ttest were used. Numbers and R software (see source code file for details) were used for determining statistical significance and graph plotting. For detailed raw data and statistical report, see source data files for each figure. For image processing and quantification, ImageJ 1.47 v software was used for threshold setting, annotation, and quantification. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.