Myelin Extracellular Leaflet Compaction Requires Apolipoprotein D Membrane Management to Optimize Lysosomal‐Dependent

Received: 4 September 2017 | Revised: 9 November 2017 | Accepted: 17 November 2017 DOI: 10.1002/glia.23274

RESEARCH ARTICLE

Myelin extracellular leaflet compaction requires apolipoprotein D membrane management to optimize lysosomal-dependent recycling and glycocalyx removal

1Instituto de Biología y Genetica Molecular- Departamento de Bioquímica y Biología Abstract Molecular y Fisiología, Universidad de To compact the extracellular sides of myelin, an important transition must take place: from mem- Valladolid-CSIC, Valladolid, Spain brane sliding, while building the wraps, to membrane adhesion and water exclusion. Removal of the 2Instituto de Neurociencias de Castilla y negatively charged glycocalyx becomes the limiting factor in such transition. What is required to ini- Leon, IBSAL, Universidad de Salamanca, tiate this membrane-zipping process? Knocking-out the Lipocalin Apolipoprotein D (ApoD), essential Salamanca, Spain for lysosomal functional integrity in glial cells, results in a specific defect in myelin extracellular leaflet Correspondence compaction in peripheral and central nervous system, which results in reduced conduction velocity Maria D Ganfornina, Instituto de Biología y and suboptimal behavioral outputs: motor learning is compromised. Myelination initiation, growth, Genetica Molecular, c/Sanz y Fores 3, intracellular leaflet compaction, myelin thickness or internodal length remain unaltered. Lack of Universidad de Valladolid-CSIC, 47003 Valladolid, Spain. ApoD specifically modifies Plp and P0 protein expression, but not Mbp or Mag. Late in myelin matu- E-mail: [email protected] ration period, ApoD affects lipogenic and growth-related, but not stress-responsive, signaling pathways. Without ApoD, the sialylated glycocalyx is maintained and ganglioside content remains Funding information high. In peripheral nervous system, Neu3 membrane sialidase and lysosomal Neu1 are coordinately Junta de Castilla y Leon (JCyL), Grant Number: VA180A11-2, MICINN-MINECO, expressed with ApoD in subsets of Schwann cells. ApoD-KO myelin becomes depleted of Neu3 and BFU2011-23978, and BFU2015-68149-R; enriched in Fyn, a kinase with pivotal roles in transducing axon-derived signals into myelin proper- European Regional Development fund; ties. In the absence of ApoD, partial permeabilization of lysosomes alters Neu1 location as well. JAEPre – CSIC fellowship; JCyL fellowship, Exogenous ApoD rescues ApoD-KO hypersialylated glycocalyx in astrocytes, demonstrating that Grant Number: call#EDU/1883/2013; European Social Fund; Operational Pro- ApoD is necessary and sufficient to control glycocalyx composition in glial cells. By ensuring lysoso- gramme for Castilla y Leon; Consejería de mal functional integrity and adequate subcellular location of effector and regulatory proteins, ApoD Educacion (JCyL) guarantees the glycolipid recycling and glycocalyx removal required to complete myelin compaction.

KEYWORDS extracellular leaflet, gangliosides, lysosome, motor learning, myelin compaction

1 | INTRODUCTION followed by the construction of a few turns of incipient myelin. Signal- ing continues through the process of growth and maturation of myelin Myelinating glial cells have the important task of building and maintain- (reviewed by Baron & Hoekstra, 2010; Jacob, Lebrun-Julien, & Suter, ing large expansions of their membranes that enwrap axons and 2011; Pereira, Lebrun-Julien, & Suter, 2012; Quarles, Macklin, & Mor- provide the basis for saltatory nerve impulse propagation, thus condi- ell, 2006; Simons & Trotter, 2007; Snaidero & Simons, 2017; White & tioning the function of neural circuits. Complex glia-axon communica- Kramer-Albers, 2014). tion initiates the first contact with axons and induces glial cell polarity, Myelin maturation requires the exclusion of water-rich media from large compact myelin domains (Bakhti, Aggarwal, & Simons, 2014), Diego Sanchez and Maria D. Ganfornina contributed equally to this work. which results in intimate membrane appositions recognized by electron ...... This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. VC 2017 The Authors GLIA Published by Wiley Periodicals, Inc.

Glia.2017;1–18. wileyonlinelibrary.com/journal/glia | 1 2 | GARCIA-MATEO ET AL. microscopy as the major dense line (MDL, condensation of the intracel- diseases known so far, independently of the gene affected, are associ- lular space) and intra-period line (IPL, condensation of the extracellular ated to leukodystrophy or myelination problems (Nave & Werner, space). The complex process of myelination is a robust one, with many 2014; Renaud, 2012), and lysosomal-dependent membrane recycling regulatory “checkpoints” and redundant systems to safeguard the final has important implications for the correct location of Plp in CNS com- product: a healthy and properly myelinated axon. Knocking-out impor- pact myelin domains (Baron & Hoekstra, 2010; Trajkovic et al., 2006). tant structural or regulatory players often have less-than-expected We have recently discovered that the Lipocalin Apolipoprotein D effects. For example, functional myelination still occurs in specific brain (ApoD), already known to be required for adequate myelin manage- areas in the absence of Fyn, a kinase of the SLK family with pivotal ment after PNS injury (Ganfornina et al., 2010; Garcia-Mateo et al., roles in transducing and integrating axon-derived signals (reviewed by 2014), is targeted to lysosomes in astrocytes and neurons, and is Kramer-Albers & White, 2011; White & Kramer-Albers, 2014). required for lysosomal membrane stability and pH homeostasis (Pas- Myelin compaction of the already formed myelin layers can be dis- cua-Maestro, Diez-Hermano, Lillo, Ganfornina, & Sanchez, 2017). The sociated in two phases: intracellular and extracellular leaflet compaction. pH distribution of the subset of ApoD-positive lysosomes is in the While the former is well understood thanks to many studies on the role range of optimal pH for neuraminidases, also coincident with that of of myelin basic protein (Mbp) in myelin assembly (at cellular, biochemical secretory lysosomes. A lysosomal population with these properties and biophysical levels; reviewed by Aggarwal, Yurlova, & Simons, 2011; would be the perfect candidate to manage glycocalyx remodeling and Bakhti et al., 2014), the compaction of myelin extracellular sides remains myelin membrane recycling. This unexpected relationship of ApoD elusive. Experimental manipulations of key candidate proteins (protein with such a subset of lysosomes, and the evidences that myelin built in zero, P0, in peripheral nervous system [PNS] and proteolipid protein, Plp, an ApoD-KO background must have different properties altering mye- in central nervous system [CNS]), thought to mediate the required adhe- lin recognition and digestion by phagocytic cells (Garcia-Mateo et al., sion between extracellular leaflets, either produced major alterations in 2014; Pascua-Maestro et al., 2017), led us to study the role of ApoD in several myelination steps, affecting both intracellular and extracellular the development and maturation of myelin. leaflet compaction (Giese, Martini, Lemke, Soriano, & Schachner, 1992), By analyzing myelin formation and maintenance from postnatal or yielded less effects than expected on myelin compaction. Lack of Plp day 3 (P3) to aged mice (P630) we have found that lack of ApoD spe- does not prevent the formation of compact myelin (Klugmann et al., cifically alters the process of myelin extracellular leaflet compaction in 1997), and compact myelin regions are also found in P0-KO nerves both CNS and PNS without major defects in intracellular leaflet com- coexisting with severe defects in myelin sheaths (Giese et al., 1992), paction or other myelin properties. The functional consequences of this revealing again the robustness of the process. On the other hand, experi- specific alteration of the terminal phase of myelin maturation are mental manipulation of the glycocalyx of oligodendrocytes in culture explored, as well as the signaling pathways altered throughout myelina- modifies the adhesion of myelin particles (Bakhti et al., 2013). Currently, tion. We conclude that ApoD ultimately controls the removal of the the working hypothesis in the field is that removal of the hydrophilic gly- sialic-rich hydrophilic glycocalyx, by maintaining functional integrity of cocalyx is required, as a limiting factor, to allow for protein-protein inter- lysosomes. A detailed analysis of the mechanism in the PNS reveals actions between apposing myelin membranes (Bakhti et al., 2014; Bakhti that the proper localization (and hence activity) of both lysosomal et al., 2013). Within this paradigm, the role of P0 or Plp proteins is to Neu1 and plasma membrane Neu3, as well as of the membrane-bound maintain the “membrane zip” locked, not to initiate the “zipping” process Fyn kinase, depend on ApoD. Finally, we demonstrate that itself. However, no single mutant so far has been described that specifi- lysosomally-located ApoD is necessary and sufficient to revert the cally alters myelin IPL (Bakhti et al., 2014). What is then controlling the glycocalyx remodeling needed for the hypersialylated glycocalyx phenotype in ApoD-KO glial cells. final step in myelin compaction? Sialidases must play an important role, since their activity not only would reduce cell surface hydrophilicity, 2 | MATERIALS AND METHODS but also the negative charge of sialic acid in myelin glycoproteins and glycolipids, thus decreasing repulsion between apposed membranes. 2.1 | Animals Myelin-related problems have been associated to genetic disorders affecting lysosomal neuraminidase 1 (Neu1), its interacting partners or ApoD-KO mice, generated by homologous recombination (Ganfornina its subcellular location (sialidosis, OMIM 256550, galactosialidosis, et al., 2008), were maintained in positive pressure-ventilated racks at OMIM 256540, or GM1-gangliosidosis, OMIM 230500, and references 25 6 18C with 12 hr light/dark cycle, fed ad libitum with standard within), and lysosomal Neu4 is able to restore lysosomal phenotypes in rodent pellet diet (Global Diet 2014; Harlan Inc., Indianapolis, IN), and human fibroblasts (Seyrantepe et al., 2004). However, no myelin dys- allowed free access to filtered and UV-irradiated water. To avoid function has been reported for Neu4 (OMIM 608527) or for the potential maternal effects of ApoD, and to generate wild-type (WT) plasma-membrane-bound Neu3 (OMIM 604617), which has been asso- and ApoD-KO mice of homogeneous genetic background, the experi- ciated instead to axonal growth (Da Silva, Hasegawa, Miyagi, Dotti, & mental cohorts used in this study are the F1 generation of homozygous Abad-Rodriguez, 2005) or regeneration upon injury (Kappagantula crosses of ApoD 2/2 and ApoD 1/1 littermates born from heterozy- et al., 2014). Whether sialidase deficits result specifically in myelin gous crosses of an ApoD-KO line backcrossed for over 20 generations outer leaflet compaction is not known. Curiously, all lysosomal storage into the C57BL/6J background. GARCIA-MATEO ET AL. | 3

Experimental procedures were approved by the University of Val- Teased nerve fibers were obtained from 4% formaldehyde fixed ladolid Animal Care and Use Committee, following the regulations of mouse sciatic nerves as described (Krinke, Vidotto, & Weber, 2000). the Care and the Use of Mammals in Research (European Commission Cultured primary astrocytes attached to poly-L-lysine (Sigma, St. Directive 86/609/CEE, Spanish Royal Decree ECC/566/2015). Louis, MO)-treated coverslips were fixed with 4% formaldehyde, washed in PBS, and blocked and permeabilized with 0.1% Tween-20, 2.2 | Behavioral analyses 1% non-immune calf serum. No permeabilization was used when labeling with MAA-Lectin. Open field tests were performed with a MIR-100 infrared digital cam- We used the following primary antibodies: Rabbit polyclonal anti- era and the Activity Monitor (v. 5.0) acquisition and analysis program Mbp (Abcam, Cambridge, United Kingdom); Mouse monoclonal anti- (Med Associates). Locomotor behavior was explored during a 5-min Mag (SC Biotechnology, Dallas, TX); Goat polyclonal anti-ApoD (SC Bio- session in a 27 3 27 cm arena. technology); Rabbit anti-S100 (Zymed, South San Francisco, CA); Rabbit Rotarod analysis was performed with an Ugo Basile (Varese, Italy) polyclonal anti-Neu1 (SC Biotechnology); Rabbit polyclonal anti-Neu3 apparatus. The protocols used for training and test sessions are (SC Biotechnology); Mouse monoclonal anti-165 kDa neurofilament described in Figure 2e. (DSHB, Iowa City, IA); Rabbit polyclonal anti-P0 (SC Biotechnology); Rat monoclonal anti-Plp (kind gift of Dr. B. Zalc); Mouse monoclonal anti- 2.3 | Primary astrocyte cell cultures Fyn (SC Biotechnology); Rat monoclonal anti-Lamp2 (DSHB); Mouse ApoD-KO and WT neonatal (1–2 days old) mice were used for primary monoclonals anti-gangliosides GM1–2b, GD1a, GD1b, or GT1b (DSHB); astrocytes cultures as described (Bajo-Graneras,~ Ganfornina, Martin- Biotin-conjugated MAA-Lectin (EY Labs, San Mateo, CA). The secondary Tejedor, & Sanchez, 2011a; Pascua-Maestro et al., 2017). Cell extracts antibodies used were: HRP-conjugated Donkey anti-Goat IgG (SC Bio- from cerebral cortices were obtained by a combination of enzymatic technology); Alexa 488/594-conjugated Goat anti-mouse/rat/rabbit and mechanical dissociation, resuspended in Dulbecco’smodifiedEagle antibodies or streptavidin (Jackson Immunoresearch, West Grove, PA). medium with 10% fetal bovine serum, 1% L-glutamine, 100 U/ml peni- HRP development was performed with 3,30-diaminobenzidine cillin, 100 U/ml streptomycin, 0.25 lg/ml amphoterycin B plated on (DAB) (0.03%) and H2O2 (0.002%) in 50 mM Tris, pH 8.0, in parallel for 8 – culture flasks and incubated at 37 Cin5%CO2 with 90% 95% humid- all sections probed with the same antibody. Sections were mounted ity. The culture medium was replaced weekly. Cell cultures were used with coverslips and Eukitt after dehydration and clearing with xylene. > for experiments after two subculture steps, when 99% of cells are Fluorescent immunohistochemistry sections were mounted with Ever- ~ astrocytes (Bajo-Graneras et al., 2011a). BriteTM mounting medium with 4’,6-diamidino-2-phenylindole, and Human ApoD purified from breast cystic fluid (Ruiz, Sanchez, sealed with CoverGripTMCoverslip Sealant (Biotium, Fremont, CA). Correnti, Strong, & Ganfornina, 2013) or recombinant human ApoD Luxol Fast Blue (LFB) staining was carried out in paraffin sections from Escherichia coli (ProSpec) were added (10 nM) to the cell cultures. incubated in LFB solution at 378C overnight, washed in 95% ethanol, – For long-term treatment (7 days), media was replaced every 48 72 hr differentiated in 0.005% lithium carbonate followed by 70% ethanol, with freshly added ApoD. and mounted after dehydration and clearing in Eukitt. The process was performed in parallel for all sections probed. 2.4 | Histochemistry, immunochemistry, and immunocytochemistry 2.5 | In situ hybridization

Nerve and brain samples were fixed by immersion in 4% formaldehyde ApoD digoxigenin-labeled riboprobes (sense and antisense) were syn- 8 overnight at 4 C. The tissues were washed thoroughly in phosphate thesized from a fragment of the mouse cDNA clone comprising the buffered saline (PBS), and either embedded in paraffin following stand- coding sequence and part of the 30-UTR (Sanchez, Ganfornina, & Marti- ard procedures or cryoprotected and frozen in Tissue-Tek (Sakura, Tor- nez, 2002). In situ hybridizations were performed at 638Con30mm l rance, CA). Paraffin sections (4 m), performed with a rotary cryostat sections with the 1 mg/ml riboprobe. Non-specific hybridiza- microtome (Microm, Wayzata, MN), or Cryostat (Microm) sections tion of the ApoD riboprobe was evaluated with a sense probe. AP- (10 lm) were mounted on polysine slides (Menzel-Gläser) and dried or conjugated mouse anti-digoxigenin IgG (Roche, Basel, Switzerland) was stored at 2208Crespectively. used to develop the riboprobe signal. Paraffin sections were dewaxed in xylene and rehydrated through ethanol series in PBS. Endogenous peroxidase was inactivated with 2.6 | Myelin preparations H2O2 (0.9%) for 5 min in the dark before horseradish peroxidase (HRP) immunohistochemisty. Paraffin and cryostat sections were blocked in Myelin preparations were performed as described (Garcia-Mateo et al., 1% normal goat/bovine serum. When needed, the sections were per- 2014). In brief, myelin was isolated from whole brain or pools of sciatic meabilized with Triton X-100 (0.25%) or Tween-20 (0.2%). Labelings nerves of 3-month-old mice by sucrose density-gradient centrifugation, with Biotin-conjugated Maackia amurensis (MAA)-Lectin and Mabs according to Norton & Poduslo method (1973). We used six independ- anti-gangliosides were performed in cryostat sections in the absence of ent CNS myelin preparations per genotype (each preparation contain- detergents. ing pools of 3 brains). Two independent PNS myelin preparations per 4 | GARCIA-MATEO ET AL. genotype were used, each with 24–32 nerves. Myelin protein content transferred to polyvinylidene difluoride (PVDF) membranes using stand- was determined with MicroBCA Protein Assay (Pierce) and monitored ard procedures. Membranes were blocked in 5% milk/2% bovine serum by sodium dodecyl sulfate polyacrylamide gel electrophoresis and Coo- albumin and exposed to the following antibodies: Rabbit polyclonal anti- massie blue staining. Mbp (Abcam); Mouse monoclonal anti-Mag (SC Biotechnology); Goat polyclonal anti-ApoD (SC Biotechnology); Mouse monoclonal anti-Fyn 2.7 | Electron microscopy methods (SC Biotechnology); Rabbit polyclonal anti-Neu3 (SC Biotechnology); Rabbit polyclonal anti-phospho Erk (Cell Signalling, Danvers, MA). HRP- Sciatic nerves or pellets of CNS myelin preparations were fixed in 2% conjugated Mouse monoclonal anti-b actin (Sigma) was used to normalize formaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer pH protein loads. Membranes were developed with enhanced chemilumines- 7.4 overnight at 48C. Tissues were then dehydrated through a graded cence reagents (Millipore, Burlington, MA), and the signal visualized with series of ethanol, and embedded in Epoxy EMbed-812 resin (Electron a digital camera (BioRad, Hercules, CA). The integrated optical density of Microscopy Sciences). Ultrathin sections were obtained with an Ultra- immunoreactive protein bands was measured in images taken within the cut E ultramicrotome (Reichert/Leica), contrasted with uranyl acetate linear range of the camera, avoiding signal saturation. and lead citrate, and analyzed using a JEOL JEM-1011 HR electron We used three independent protein preparations per genotype microscope with a CCD Gatan ES1000W camera with iTEM software. and age. For CNS samples, 3 cerebella were pooled in each preparation. For PNS samples, pools of 6 nerves (P10 and older ages) or 12 nerves 2.8 | Image acquisition and analysis (P3) were used. Three experiments (technical replicas) were performed Cells and tissues were visualized and photographed with an Eclipse 90i with each sample set unless noticed. (Nikon, Minato, Tokyo, Japan) fluorescence microscope equipped with a DS-Ri1 (Nikon) digital camera. 2.10 | Quantitative real-time polymerase chain The images were acquired under the same conditions of illumina- reaction tion, diaphragm and condenser adjustments, exposure time, background correction and color levels. Mouse tissues used for mRNA expression studies were stored at Confocal images were obtained with a 633 oil immersion objective 2808C, and RNA was extracted with RNeasy Lipid Tissue Mini Kit (Qia- (HCX PL Apo CS NA 5 1.4; Leica, Wetzlar, Germany) attached to a con- gen, Hilden, Germany) using QIAzol Lysis Reagent (Qiagen). RNA confocal DMI 6000B microscope with a TCS SP5 confocal system (Leica) centration was measured with a Nanodrop spectrophotometer, and the equipped with acousto-optical beam-splitter (AOBS) and acousto-optic RNA quality assessed by agarose electrophoresis. Following DNAse tunable filter systems. Fluorophores were excited with WLL laser (Leica) treatment, 500 ng of total RNA were reverse-transcribed with Prime- and a 405 line (Leica) controlled by LAS AF software (Leica). Emissions Script (Takara Bio Inc., Otsu, Japan) using Oligo-dT primers and random were collected with the AOBS system and three spectral detectors. Laser hexamers. The resulting cDNA was used as a template for quantitative power and detection gains were set by scanning control samples labeled real-time polymerase chain reaction (RT-qPCR) using SybrGreen (SYBR with secondary antibody alone. We ensured to obtain similar dynamic Premix Ex Taq kit, Takara). The primers used for RT-qPCR are shown in ranges in our images, and adjusted gain and offset using look-up tables. Table 1. Rpl18 was used as the reference gene. Images were stored as 1024 3 1024 pixels and 8-bit TIFF files. Z-series (xyz scan) were performed and the optimal value of the Primer Name Sequence step size was calculated for the wavelength used to fulfill the Nyquist Mouse Rpl18-Forward 5-TTCCGTCTTTCCGGACCT theorem. The optical section thickness was 0.772 mm, and scanning Mouse Rpl18-Reverse 5-TCGGCTCATGAACAACCTCT was performed with a 1.0 Airy unit pinhole size. Mouse ApoD-Forward 5-GAAGCCAAACAGAGCAACG Images were processed with a Gaussian Blur filter (Sigma [Radius]: 1.00), to facilitate object detection, and analyzed with tools of the FIJI Mouse ApoD-Reverse 5-TGTTTCTGGAGGGAGATAAGGA software. Mouse Fasn-Forward 5-GCTGCTGTTGGAAGTCAGC

Mouse Fasn-Reverse 5-AGAAGAAAGAGAGCCGGTTG 2.9 | Immunoblot analysis Mouse Jun-Forward 5-AGAAGAAGCTCACAAGTCCG

Tissues were homogenized in lysis buffer (1% Nonidet P-40, 0.1% SDS, Mouse Jun-Reverse 5-TTCTTTACAGTCTCGGTGGC 10% Glycerol, 1% sodium deoxycholate, 1 mM dithiothreitol, 1 mM eth- Mouse Mbp-Forward 5-TCCGACGAGCTTCAGACCATCCA ylenediaminetetraacetic acid (EDTA), 100 mM 4-(2-hydroxyethyl)-1- piperazineethanesulfonic acid (HEPES), 100 mM KCl, 1 mM NaF, 10% Mouse Mbp-Reverse 5-ACCGCTAAAGAAGCGCCCGA complete protease inhibitors [Roche] in PBS), centrifuged after 30 min at Mouse Mag-Forward 5-TCTACCCGGGATTGTCACTG

48C, and the supernatant stored at 2808C. Protein concentration was Mouse Mag-Reverse 5-CGTCCTCACTTGACTCGGAT estimatedwithMicro-BCAproteinassay(Pierce,Waltham,MA).Tissue Mouse Plp-Forward 5-TGGCCCTCTTTGTGACTATG homogenates or myelin preparations were diluted in Laemmly-SDS Mouse Plp-Reverse 5-CCAACTACCACCATGAGGTT buffer. 10–20 mg of total protein/lane were electrophoresed and GARCIA-MATEO ET AL. | 5

The mRNA transcription levels were assessed with the DDCT 3.2 | Lack of ApoD results in deficient extracellular method (Livak & Schmittgen, 2001) using normalization to Rpl18 leaflet myelin compaction, concurring with minor for each condition. Statistically significant differences of gene tran- aging-related alterations in myelin thickness scriptional changes were evaluated with a Mann–Whitney U test Electron microscopy (EM) analyses of myelin sheaths and their relation- (Yuan, Reed, Chen, & Stewart, 2006) using DCT of each replica ship to axon diameter in early (P3), adult (P90) and aged (P630) sciatic (calculated by subtracting the average CT of the reference gene nerves reveal an apparently normal myelin wrapping of axons in the for each sample). The statistical level of significance was set at absence of ApoD (Supporting Information, Figure S2). Only minor P < 0.05. signs of demyelination (increased g-ratio, Supporting Information, Three independent mRNA preparations were performed per Figure S2b) are observed in aged ApoD-KO mice, altering slightly the genotype and age, with pools of 3 cerebella or 6 nerves (with the axon diameter/myelin thickness relationship (Supporting Information, exception of P3 and P10, where 12 nerves were used). Four Figure S2c). experiments (technical replicas) were performed with each sample However, a clear and consistent difference is observed at high set. resolution in ApoD-KO myelin, which is enhanced in the aged mice (Figure 1): the extracellular leaflet of PNS myelin, revealed in EM as the | 2.11 Statistical analysis IPL, appears as two distinct and separated membranes (Figure 1a–c,

Statistical analyses were performed with SPSS v.19 (IBM) and Sigma- arrows) resulting in significant enlargement of the period between Plot v.11.0 (Systat, San Jose, CA) softwares. A P value <0.05 was used MDL. The defect is observed in mature myelin and not during the initial as a threshold for significant changes. The tests used for each experi- phases of myelin formation, since P3 myelin does not differ between genotypes (Figure 1a,d). No differences in the condensation of the ment are stated in figure legends. intracellular space is observed at P3 or P90, and only a slight decrease in electron density is observed in the MDL of aged nerves (Figure 1c). 3 | RESULTS CNS myelin preparations from adult mouse brains show the same defects (Figure 1e,f). 3.1 | ApoD expression increases during myelination As mentioned above, the manipulation of obvious candidates to be and keeps growing throughout aging in PNS and CNS key myelin proteins triggering extracellular leaflet compaction (P0 and We have previously analyzed ApoD expression in the nervous system Plp), precluded a clear dissociation between phases of myelin compac- under stress, damage or pathological conditions (Bajo-Graneras~ et al., tion. After a normal early myelination phase, the specificity of ApoD- 2011a; Ganfornina et al., 2008; Ganfornina et al., 2010; Garcia- KO compaction defect, particularly evident in the extracellular leaflet Mateo et al., 2014; Li et al., 2015; Sanchez et al., 2015), and found of myelin in both CNS and PNS, gives ApoD a novel and unique func- that it is up-regulated in glial cells in both CNS and PNS (reviewed tion within the myelin maturation process. by Dassati, Waldner, & Schweigreiter, 2014). In the healthy brain, the major cell type expressing ApoD is the mature oligodendrocyte (Sup- 3.3 | Motor learning is compromised in the absence of porting Information, Figure S1a and Ganfornina et al., 2005; Navarro, ApoD Del Valle, & Tolivia, 2004; Ong et al., 1999; Provost et al., 1991; Does the defective ApoD-KO myelin maturation and the resulting Soreq et al., 2017), and the oligodendrocyte-specific transcriptome is altered structure have functional consequences? We already know that particularly altered in ApoD-KO mice (Bajo-Graneras~ et al., 2011b). ApoD-KO mice have age-dependent functional defects in the PNS, Also, Schwann cells are the major site of expression of ApoD in the with a 29% average reduction of motor nerve conduction velocity in healthy adult sciatic nerve (Supporting Information, Figure S1b and 40 week old mice (Ganfornina et al., 2010). “ ” Garcia-Mateo et al., 2014), where a dynamic salt and pepper pat- Recent reports have demonstrated that ongoing adult oligodendro- “ ” tern of expression suggests a steady state idle mode, ready for the cyte production and differentiation are required for learning motor expression boost needed upon injury. Since all antecedents point to skills, and that learning conversely promotes oligodendrocyte precursor important physiological functions of this lipid binding protein in mye- proliferation and maturation (McKenzie et al., 2014; Xiao et al., linating cells, we set to analyze its expression along myelin lifetime. 2016). To test whether myelin-related functional impairments are We analyzed sciatic nerve and cerebellum to sample both PNS and also present in the CNS of ApoD-KO mice, we evaluated their perform- CNS, from postnatal day 3 (P3) to aged mice (P630; Supporting ance in a motor task learning paradigm that requires accurate timing of Information, Figure S1c–f). In agreement with antecedents (de Magal- signal processing mainly in the cerebellum and its input and output haes, Curado, & Church, 2009; Loerch et al., 2008; Ong et al., 1999), circuits. ApoD mRNA and protein increase over aging (reviewed by Dassati Two independent cohorts were evaluated with a rotarod training et al., 2014). Here we show that a significant increase in ApoD and test paradigm that is performed along a 3 days period (Figure 2e). expression (at P25) coincides with the final phases of myelin This protocol was applied to a set of mice at P21 (Figure 2a), and to maturation. another group at P140 (Figure 2d), ages at which peripheral nerve 6 | GARCIA-MATEO ET AL.

FIGURE 1 Lack of ApoD results in altered compaction of the extracellular leaflet of myelin in both CNS and PNS. (a–c) Electron microscopy analysis of myelin sheaths in sciatic nerve ultrathin sections at P3 (a), P90 (b), and P630 (c). Calibration bars: 20 nm. Arrows point to the IPL, which is altered in P90 and P630 ApoD-KO nerves. Arrowheads point to the MDL. Only a slight decrease in electron density is observed in the MDL of aged nerves. (d) Sciatic nerve myelin periodicity measured as the distance between consecutive MDLs (mean 6 SD of 150 measurements/age/genotype from three independent mice; 5 measurements/fiber). No genotype differences occur early in the myelination process (P3), but abnormally large MDL distances are observed in adult and aged ApoD-KO mice. (e) Electron microscopy analysis of ultrathin sections of CNS myelin preparations at P90 shows the same compaction defects. Three levels of magnification are shown. Calibration bars: Black 50 nm; white 20 nm. Abnormal IPL pointed by arrow. MDL pointed by arrowhead. (f) MDL periodicity in CNS myelin preparations from P90 mice (N 5 14 myelin particles/genotype). Statistical differences assayed by unpaired Student’s t test. *P < .05; **P < .01 conduction velocity is not altered yet. None of the groups showed dif- 3.4 | Myelin sheaths maturing in the absence of ApoD ferences in weight gain over postnatal development or in general loco- have reduced extracellular leaflet-related protein motor activity assessed by an open field test (Supporting Information, expression, with minor modifications in intracellular Figure S3). myelin proteins Lack of ApoD results in significantly lower scores than WT control mice in the rotarod accelerated test sessions, but performance The structural and functional alterations of ApoD-KO myelin must be diverged during the constant speed training sessions. This genotype- accompanied of biochemical differences with WT myelin sheaths. We dependent low performance was particularly evident in the P140 explored the expression of the most abundant myelin proteins and cohort, clearly showing age-dependent alterations in the motor learn- found that Plp and P0, both with known implications in the compaction ing process itself. of the extracellular leaflet of myelin, are down-regulated, as evidenced To evaluate whether long-term storage of motor programs was from two different quantitation approaches: immunohistochemistry affected, we exposed the first cohort to two recall protocols (Figure and qRT-PCR for Plp (Figure 3a,b) and immunohistochemistry and Coo- 2b,c): a first one at P140 (less demanding, starting rod speed: massie staining of myelin preparations for P0 (Figure 3c,d). 4 rpm) and a second one, four weeks later, with the same difficulty However, Mbp, with functions performed from the intracellular level than the one experienced at P21. No differences between side of myelin sheaths, and Mag, preferentially located in uncompacted genotypes were observed in the recall sessions, suggesting that myelin domains, show either unaltered expression in CNS throughout long-term storage of the motor program necessary to master the life (mRNA relative levels measured by qRT-PCR, Figure 4a), or timing task has been successful (compare initial proficiency in Figure 2a differences in transcriptional levels in PNS (Figure 4b) that do not result and 2c, arrows). in major changes of net protein levels (Figure 4b,c). Other properties of Thus, the process of outer leaflet compaction needs to be com- the global structure of myelinated fibers in the PNS, like internodal dis- pleted in order to have adequately timed CNS circuit activity for motor tance, are not altered in the absence of ApoD (Figure 4d). task learning, and to be able to maintain normal axonal conduction The data obtained in PNS suggest that the transcriptional machin- velocities in peripheral nerves upon aging. Both CNS and PNS func- ery might be responding to an ApoD-dependent altered homeostasis in tional parameters require ApoD. Schwann cells, but the stability and low turnover of both Mag and Mbp GARCIA-MATEO ET AL. | 7

FIGURE 2 ApoD-KO mice show age-dependent motor learning disabilities without alterations in long-term motor program storage. (a) Rotarod performance of young (cohort 1, P21) ApoD-KO mice is lower than WT controls starting from the last training session, suggesting a deficiency in learning the motor task. (b and c) The same mouse cohort tested at P21 was subjected to two recall sessions. A low-demand recall (low speed) at P140 and a standard recall four weeks later. A better performance in the first training session of Recall 2 compared with first training at P21 (arrows) demonstrates the adequate storage of a long-term motor memory. (d) When the motor task is learned at P140 (cohort 2) ApoD-KO mice show a stronger phenotype. (e) Motor learning protocols. A standard rotarod sequence of 6 training sessions at 21 rpm without acceleration, and 3 tests (acceleration starting at 21 rpm), was performed at P21 or P140. Each session consists of two (white ovals) or three (black ovals) trials on the rod separated by 5 min waiting time. Cohort 1, tested at P21, was subjected to two recall sequences at low starting speed (4 rpm at P140), or standard speed (21 rpm at P168). Statistical differences were assessed by ANOVA followed by Tukey post-hoc test. *P < .05 proteins lead to unchanged final protein levels (see Discussion). In the compaction. The responses of myelinating glia to injury and stress are CNS, ApoD does not produce quantitative changes in intracellular mye- also well known (reviewed by Glenn & Talbot, 2013; Simons, Misgeld, lin proteins expression, though we observe a clear alteration in Mbp & Kerschensteiner, 2014; van der Knaap & Bugiani, 2017). spatial distribution (Figure 4b) that might be related to the slightly We evaluated a key stress-response pathway put forward upon modified appearance of the MDL (Figure 1e). injury (Arthur-Farraj et al., 2012) by measuring cJun transcript levels Thus, ApoD influence on the expression of myelin proteins is par- during postnatal development in both cerebellum and sciatic nerve ticularly prominent for those contributing to the process of extracellular samples (Figure 5a). No genotype-dependent differences were leaflet compaction, and produces none or only subtle alterations observed at any stage. We then monitored the state of the two major in myelin proteins located in the intracellular side in compact pathways regulating myelin growth: (1) the nutrient-sensing mTORC1- myelin domains or in the uncompacted myelin regions. Other global dependent lipogenic switch controlling Fasn expression and required structural parameters determining the length of myelin sheaths remain for the massive lipid biosynthesis during myelin growth (Lebrun-Julien unaltered. et al., 2014; Norrmen et al., 2014) and (2) the ERK-mediated pathway, which, in addition to early roles in myelinating glia differentiation trig- 3.5 | ApoD affects lipogenic and growth-related, but gered by axon-derived signals (Newbern et al., 2011), responds to growth-promoting signals and controls the number of myelin wraps not stress-responsive, signaling pathways during the and myelin thickness in later maturation phases (Ishii, Furusho, & Ban- late phase of myelin maturation sal, 2013). Early in the postnatal period (P3), neither Fasn transcript Intercellular and intracellular signaling in the different phases of myelin level nor the activated pERK1/2 protein change in PNS. However, a formation have been widely studied, from the continuous axon-glia significant decreased activity of both pathways are observed at P25, communication, to the different transcription patterns and downstream when completion of myelin compaction is taking place (Figure 5b,c). pathways controlling wrapping initiation, myelin sheath growth and Also, Fasn expression does not differ in ApoD-KO CNS samples (Figure 8 | GARCIA-MATEO ET AL.

FIGURE 3 ApoD influences the expression of extracellular leaflet-related myelin proteins. (a) Immunohistochemical quantification of Plp protein expression in cerebellar sections at P25. Plp protein expression is significantly reduced in ApoD-KO CNS white matter. (b) mRNA expression assessed by qRT-PCR shows a significant decrease at P25 without changes at earlier times (P3). (c) Immunohistochemical quantification of P0 protein expression in adult sciatic nerve (P90) reveals a significant decrease in P0 expression in the absence of ApoD. (d) Coo- massie blue staining of myelin protein preparations show a decreased relative expression of P0. Two independent pools of sciatic nerves per genotype were used (n 5 24–32 nerves/genotype/pool). Arrow points to an invariable protein used for protein loading normalization. Statistical differences assessed by unpaired Student’s t test in a, c and d, and by Mann–Whitney U test in b. *P < .05; **P < .01. Calibration bars: 20 lm [Color figure can be viewed at wileyonlinelibrary.com]

5b), but the ERK-mediated pathway behaves as in PNS, with a sus- functional impairment and lower conduction velocity of fibers in tained down-regulation starting at P25 (Figure 5d). ApoD-KO mice. Interestingly, ApoD genotype-dependent differences Our results demonstrate the influence of ApoD in growth-related in LFB staining of sciatic nerves appear at late stages of myelin matura- signaling responses that are either specific to PNS (the lipogenic switch tion (P10 and P25), but not at P3 when the initial phases of myelination control), or common to both CNS and PNS (signaling governing myelin are taking place (Supporting Information, Figure S4). wraps growth). ApoD loss starts altering signaling only after the first The outer surface glycocalyx of myelinating cells is initially abun- phases of myelin assembly have taken place. dant and negatively charged, due to sialylated glycolipids and glycoproteins, favoring displacement of myelin sheaths as they wrap axons. 3.6 | ApoD is required for the process of myelin However, this hydrophilic cover needs to be extensively removed in glycocalyx removal and conditions gangliosides order to promote the final adhesion of the two apposed membranes content and distribution (Bakhti et al., 2014; Bakhti et al., 2013). The most direct cause of the anomalous hydrophilicity and uncompacted outer leaflet in ApoD-KO The structural and functional abnormalities caused by the lack of ApoD myelin would be a deficit in glycocalyx removal. in both CNS and PNS myelin predict that basic biochemical properties To test this prediction, Maackia amurensis lectin was used to assay of myelin sheaths should be altered. Luxol Fast Blue staining reveals a sialic acid content in sciatic nerves. A clear increase is observed in much fainter signal in both PNS and CNS myelin in ApoD-KO mice ApoD-KO nerves (Figure 6b). (Figure 6a), showing clear genotype-dependent biochemical differen- Since no alterations are observed in the content of the abundant ces. The low retention of the colored anions of the sulfonated copper glycoprotein Mag (Figure 4), which is highly sialylated (Sedzik, Jastrzeb- phthalocyanine salt (Kiernan, 2010) indicates that ApoD-KO myelin is ski, & Grandis, 2015), we studied the most sialylated species of ganglio- less hydrophobic than WT myelin, and that the dye has less interac- sides in the nervous system (Gong et al., 2002; Lunn et al., 2000; Vajn, tions with membrane phospholipids (Pearse, 1955). This property is Viljetic, Degmecic, Schnaar, & Heffer, 2013). We used antibodies spe- coherent with aberrant water presence in the outer leaflet layer of cific for GM1–2b and GD1a (belonging to the A-series of ganglioside uncompacted myelin, and would undoubtedly contribute to the biosynthesis, with 1 and 2 sialic acid moieties), and for GD1b and GT1b GARCIA-MATEO ET AL. | 9

FIGURE 4 ApoD has minor influences in proteins functionally related to myelin inner leaflet compaction, and does not alter structural parameters like internodal length. (a) mRNA expression of Mag and Mbp genes in CNS (cerebellum) and PNS (sciatic nerve) across ages from P3 to P630. No genotype-dependent changes are observed in CNS. Timing variations are observed in the PNS transcription of both Mag and Mbp genes. (b) Representative examples of immunolocalization of Mbp and Mag proteins in CNS (corpus callosum) and PNS (sciatic nerve). Altered distribution of Mbp in CNS myelin is observed without quantitative changes. Double-labeling with the neuronal marker neurofilament protein (NF) is shown in the last panel. (c) ApoD-KO does not show significantly differences in Mbp or Mag protein levels. Two representative immunoblots are shown. (d) Immunolocalization of Mbp and Mag proteins in single myelinated axons from “teased” sciatic nerve preparations. Linear distance between consecutive nodes of Ranvier were quantified in Mag-labeled teased nerves. No differences were found between genotypes in internodal length. Statistical differences assessed by Mann–Whitney U test (a) and unpaired Student’s t test (d). *P < .05. Calibration bar: 20 lm(bandd)and100lm(d)

(from the B-series, with 2 and 3 sialic acid residues). ApoD causes 3.7 | Adequate subcellular localization of lysosomal alterations in all four major glycolipids (Figure 6c,d), with the three and membrane sialidases and of Fyn kinase depend on species most represented in myelin (compared with axons) showing ApoD significantly higher amounts in ApoD-KO nerves (GM1–2b, GD1b, and GT1b, Figure 6d), and with altered spatial distribution in GD1a To further understand the mechanism of myelin membrane remodeling and GT1b (Figure 6c), the most sialylated of the A and B series preceding the final step of extracellular leaflet compaction in the PNS, respectively. we analyzed the expression and subcellular localization of sialidases, in These results demonstrate that ApoD is required for proper glyco- charge of removing sialic acid moieties from glycolipids and glycopro- lipid management during myelin membrane compaction. All previous teins (reviewed by Miyagi & Yamaguchi, 2012). Our recent discovery stages of myelination initiation and wrapping of axons are unaltered, that ApoD traffics from plasma membrane to lysosomes in astrocytes but in the last phase of myelin maturation sialic acid moieties are not and neurons (Pascua-Maestro et al., 2017) led us to test lysosomal and removed properly from the abundant glycolipids of myelin membranes. membrane sialidases. Lysosomal Neu1 has glycoproteins as preferential The altered final distribution of these important myelin components substrates. Neu3, located at the plasma membrane, acts primarily on must be the reflection of disturbances in the process of myelin mem- glycolipids and its expression is boosted as part of the acute response brane recycling underlying the composition changes needed to transi- to nerve injury, contributing to myelin degradation (Kappagantula et al., tion from myelin sheaths growing and sliding to build the wraps, to the 2014). final phase of membrane apposition, adhesion and compaction of the As in astrocytes (Pascua-Maestro et al., 2017), ApoD is present in extracellular leaflet. Schwann cells lysosomes, as evidenced by co-localization with Lamp2 10 | GARCIA-MATEO ET AL.

FIGURE 5 Nutrient sensing and growth-related pathways, but not stress/injury-triggered pathways, are altered during postnatal myelin development in ApoD-KO mice. (a) Jun mRNA expression was evaluated as a readout of the stress-responsive JNK pathway in PNS (sciatic nerve) and CNS (cerebellum) samples at early (P3) and late (P25) phases of myelin maturation. No genotype-dependent changes are observed. (b) Fasn mRNA expression was evaluated as a readout of the nutrient-sensing lipogenic Akt/mTORC1 pathway as in A. Lack of ApoD significantly reduces Fasn age-related induction in PNS. (c) pERK1/2 protein levels were quantified and normalized to actin levels, to evaluate the activity of the growth promoting ERK pathway in sciatic nerves at P3 and P25. No activation with age was observed in ApoD- KO nerves. (d) pERK1/2 levels in cerebellum from P3 to P630. Band densitometry analysis and a representative immunoblot are shown. Levels of activated ERK decrease in ApoD-KO cerebellum starting at P25, without changes during the early phases of myelination. Statistical differences were assessed by Mann–Whitney U test (a and b), unpaired Student’s t test (c) and by ANOVA on ranks followed by Tukey post-hoc test (d). *P < .05

(Figure 7a–c). In normal conditions, ApoD scattered and stochastic pat- either it gets down regulated or its membrane location is reduced in tern of expression indicates that Schwann cells turn on ApoD expres- ApoD-KO fibers. sion when undergoing tasks that do not take place simultaneously In oligodendrocytes, membrane trafficking and sorting are known along the nerve (Garcia-Mateo et al., 2014). Lysosomal Neu1 is co- to take place through specialized endosome and lysosome-mediated expressed with ApoD (Figure 7d,e) in the same subset of Schwann cells recycling paths (Feldmann et al., 2011), and the key regulatory kinase and, as expected, also co-localizes with ApoD (Figure 7f–k). Further- Fyn has been proposed to regulate the lysosomal-mediated traffic more, Neu1 must respond to the same signals, as demonstrated by the determining adequate location of Plp in CNS myelin (White & Kramer- co-regulation triggered by injury (Supporting Information, Figure S5a,b). Albers, 2014). Fyn-specific antibody labeling reveals that while Neu3 Plasma membrane Neu3 is also co-expressed with ApoD in decreases, Fyn increases in teased ApoD-KO nerves (Figure 8b). The Schwann cells (Figure 7m,n), and it is abundant in myelin membrane, as labeling pattern of Fyn suggests a myelin sheath location, and the com- evidenced in teased nerve preparations (Figure 7l). Upon injury, the plementary behavior of Neu3 and Fyn suggests a traffic defect. increased levels of Neu3 also extensively co-localizes with ApoD (Sup- To test these ideas further, we examined whole protein extracts porting Information, Figure S5c). Therefore, there must be common sig- and myelin preparations by immunoblot (Figure 8c,d). Though no nals coordinating the expression of ApoD with both Neu1 and Neu3. ApoD-dependent changes are observed in total Neu3 protein expres- In astrocytes, ApoD is targeted to a subset of lysosomes particu- sion from P3 to P630 (Figure 8c), Neu3 is completely depleted in adult larly vulnerable to oxidative stress, and the lack of ApoD results in lyso- PNS myelin preparations (Figure 8d). Total Fyn expression peaks at somal partial permeabilization and alkalinization (Pascua-Maestro et al., P10, as expected (Kramer, Klein, Koch, Boytinck, & Trotter, 1999), and 2017). Using teased nerve preparations to isolate single fibers, we dem- this up-regulation is significantly enhanced in the absence of ApoD onstrate that the characteristic Neu1 punctate labeling (concentrated (Figure 8c). Total Fyn returns to low, ApoD-independent, levels at later in perinuclear and paranodal cytoplasm-rich regions of the Schwann ages. However, when testing for its presence in myelin preparations, cells) disappears in ApoD-KO nerves (Figure 8a), a result compatible Fyn is clearly enriched in ApoD-KO PNS myelin (Figure 8d), mirroring with the lysosomal leakage that occurs in the absence of ApoD. When the behavior of Neu3. This is in agreement with the increased labeling analyzing Neu3 expression pattern (Figure 8b), the results suggest that in the teased nerve preparations (Figure 8b). GARCIA-MATEO ET AL. | 11

FIGURE 6 Hydrophobicity, sialic acid and gangliosides content are altered in the absence of ApoD. (a) Luxol Fast Blue staining of paraffin sections of sciatic nerves and corpus callosum at P100. Reduced myelin hydrophobicity is reflected in the anomalous staining of ApoD-KO myelin, both in PNS and CNS. (b) Sialic acid content in WT and ApoD-KO sciatic nerves was quantified by Maackia amurensis lectin labeling in cryostat sections. A significant increase is observed in ApoD-KO nerves, while the perineurial labeling is not dependent on genotype. (c and d) Antibodies specific for the four most sialylated membrane glycolipids were used in sciatic nerve cryostat sections. Significantly increased levels are observed for the three ganglioside species most represented in myelin (GM1–2b, GD1b, and GT1b). Clear alterations in spatial distributions are also observed for GD1a and GT1b, constituting the terminal branches of A and B-series of ganglioside biosynthetic pathways. Statistical differences assayed by unpaired Student’s t test (b) and by ANOVA followed by Tukey post-hoc test (d). *P < .05 and **P < .01. Calibration bars: 20 lm [Color figure can be viewed at wileyonlinelibrary.com]

These results provide a strong indication that ApoD, by helping to 3.8 | Stable presence of ApoD in the lysosome is preserve lysosomal membrane integrity, provides the necessary condi- required to rescue ApoD-KO hypersialylated tions to maintain an adequate activity of lysosomal sialidases and, at glycocalyx in astrocytes the same time, controls the adequate lysosomal-mediated traffic of The data presented so far demonstrate that ApoD is required for the Neu3 and Fyn to and from myelin membrane, which is ultimately orchestrated actions leading to glycocalyx removal, the limiting factor required for myelin complete compaction. 12 | GARCIA-MATEO ET AL.

FIGURE 7 ApoD is located in Schwann cells lysosomes, and its expression is co-regulated with both lysosomal and membrane neuraminidases. (a–c) Co-localization of ApoD with the late endosome-lysosome compartment marked with Lamp-2. Representative Immunofluores- cence images from sagittal paraffin sections of wild type adult sciatic nerves. (d–e) The characteristic pattern of scattered ApoD-positive Schwann cells coincides with the pattern of lysosomal Neu1. (f–k) Co-localization of ApoD with lysosomal Neu1 shown in paraffin sections (f–h, immunofluorescence) and teased nerves preparations (i–k, confocal imaging). (l) Membrane Neu3 shows labeling in myelin sheaths (high magnification confocal imaging in teased nerves preparation). (m and n) ApoD and Neu3 are also co-expressed in subsets of Schwann cells (low magnification confocal images in paraffin sections). Calibration bars: 5 lm(k),10lm(a–c, f–l), 20 lm(dande,mandn) for extracellular leaflet myelin compaction. However, is ApoD suffi- 4 | DISCUSSION cient? Can we revert a defective glycocalyx by supplying ApoD to the cell? To answer this question we turned to the well-characterized pri- Our data brings an unexpected key element to the already complex mary astrocyte cultures, to first test if glycocalyx control is a general process of myelin biogenesis: ApoD, a lipid binding protein expressed cell-biological function of ApoD in all glial cells expressing it. by myelinating glia, whose function in the lysosomal compartment is As in myelinating glia, ApoD-KO astrocyte membranes have signifi- required for the functional integrity of these intracellular organelles cantly higher sialic acid content, as revealed by Maackia amurensis lec- (Pascua-Maestro et al., 2017). tin labeling in non-detergent conditions (Figure 9a,b). Paraquat- These data strongly support the following model. When ApoD is triggered oxidative stress, which results in partial lysosomal membrane missing, glial lysosomes are partially permeabilized and two primary con- permeabilization and lysosomal alkalinization (Pascua-Maestro et al., sequences follow: (1) lysosomal enzymes delocalize and/or are out of 2017), further increases sialic acid content (Figure 9b) as expected if theiroptimalpHforefficientactivityand(2)lysosome-myelintraffic both processes are causally linked. Treatment of ApoD-KO astrocytes becomes dysfunctional. These primary defects may result in inefficient with exogenous human ApoD (hApoD) during a 7-day period (with desialylation of glycoproteins by lysosomal Neu1, and inefficient traffic of replacement every 48–72 hr) completely rescues the hypersialylated membrane Neu3, leading to an accumulation of sialylated glycolipids glycocalyx phenotype (Figure 9a,c). No rescue is obtained after a short (gangliosides). The inability to remove the hydrophilic glycocalyx from the 24 hr treatment with hApoD (Figure 9b), as could be expected if ApoD outer surface of myelin results in a very specific (only the IPL is affected), had direct enzymatic activity on the cell glycocalyx. but generalized compaction defect that persists throughout life. To test whether the presence and stability of ApoD in lysosomes These defects in myelin glycocalyx removal eventually condition is required for glycocalyx remodeling, we performed the treatment the expression of Plp and P0, preventing completion of the extracellu- with recombinant bacterial hApoD. This protein is also endocytosed lar leaflet compaction process. Our data support that two separate and targeted to lysosomes, but it is rapidly degraded (Pascua-Maestro processes, the removal of repulsive structures and the expression of et al., 2017), in contrast with the fully glycosylated purified native adhesive proteins to “lock the zip” (as proposed by Bakhti et al., 2014), hApoD. Co-localization with Lamp-2 at the end of the treatment (Fig- can be dissociated. Thus, glycocalyx removal precedes Plp or P0 “sta- ure 9d) shows that only the native form is maintained in the lysosomal pler” function, and becomes a limiting factor for the completion of compartment. Bacterial hApoD has minor effects in the cell glycocalyx compact myelin. (Figure 9c), coherent with its brief and intermittent presence in Various functional consequences derive from the ApoD-KO “unfin- lysosomes. ished” myelin: (1) peripheral nerves decrease their conduction velocity In summary, by maintaining lysosomal functional integrity ApoD is by 40 weeks of age, without alterations in compound action potential necessary and sufficient for glial cells to remodel their glycocalyx, a amplitude or duration (Ganfornina et al., 2010) and (2) cerebellum- process that in myelinating glia is the limiting factor for extracellular dependent motor learning is compromised, without general locomotor leaflet compaction. deficits or alterations in other cognitive processes in the young adult GARCIA-MATEO ET AL. | 13

FIGURE 8 Lack of ApoD results in lysosomal Neu1 de-localization and altered traffic of membrane Neu3 and Fyn kinase. (a) Neu1 punctate labeling decreases in ApoD-KO nerves. Areas enriched in lysosomes, located either perinuclear (arrows) or in cytoplasmic regions of the parano- des (arrowheads), are pointed for comparison. (b) Neu3 labeling, which is apparent along the entire myelin coverage (arrowheads point to paran- odes), decreases while Fyn kinase increases in ApoD-KO myelin sheaths. Representative Immunofluorescence images from teased nerve preparations are shown in a and b. (c) Immunoblot analysis of whole protein extracts from WT and ApoD-KO sciatic nerves, from P3 to P630. The total amount of Neu3 protein in nerve extracts does not change with genotype. Total Fyn expression peaks at P10, and the absence of ApoD enhances this up-regulation. (d) Myelin preparations from adult sciatic nerves are analyzed by immunoblot. Neu3 signal, greatly enriched in WT PNS myelin, disappears in ApoD-KO. The opposite pattern is observed for Fyn kinase, which becomes enriched in ApoD-KO myelin preparations. Quantitation of two replicas of two independent sets of myelin preparations is shown. Neu-3 signal in ApoD-KO lanes was indis- tinguishable from background in all cases. Statistical differences assayed by ANOVA followed by Tukey post-hoc test (b) and unpaired Student’s t test (d). *P < .05. Calibration bars in a and b: 20 lm [Color figure can be viewed at wileyonlinelibrary.com]

(e.g., hippocampus-dependent object recognition tasks, Sanchez et al., Fyn kinase is also altered by the myelin recycling defect and gets 2015). Cognitive and motor deficits do appear later upon further aging anomalously enriched in PNS compact myelin domains. This finding is in ApoD-KO (Ganfornina et al., 2008; Ganfornina et al., 2010; Sanchez coherent with a specific alteration in lysosome-dependent (versus recy- et al., 2015), suggesting defective myelin vulnerability to age- cling endosome-dependent) myelin remodeling, and supports the pro- dependent deterioration. posal made by White and Kramer-Albers (2014) for the CNS: Fyn could 14 | GARCIA-MATEO ET AL.

FIGURE 9 Hypersialylation of ApoD-KO astrocytes membrane is rescued when ApoD is stably present in lysosomes. Sialic acid content in WT and ApoD-KO astrocyte membranes monitored by Maackia amurensis lectin labeling in the absence of detergents. Representative fluorescence images (a) and fluorescence quantitation in cell populations (b and c) are shown. (a) Sialic acid labeling increases significantly in ApoD-KO astrocytes. Treatment with exogenous hApoD for 7 days (with replacement every 48–72 hr) reverts sialic acid content to control levels. (b) Oxidative stress increases sialic acid content in both WT and ApoD-KO astrocytes (gray bars). Short term treatment with hApoD followed by a 24 hr waiting period does not modify the high levels of sialic acid in ApoD-KO astrocytes. (c) Quantitative analysis of hApoD dependent rescue. A complete reversion of the phenotype is achieved after 7-day continuous treatment with fully-glycosylated purified hApoD (gray bar), but not with recombinant bacterial hApoD (bhApoD). (d) hApoD, but not bhApoD, is present in astrocytes lysosomes at the end of the treatment, though both are internalized early after each addition (1 hr post addition is shown). Statistical differences assayed in b and c by ANOVA followed by Tukey post-hoc test. Only pairwise comparisons not significantly different are pointed (n.s.). Calibration bars: 20 lm (a), 10 lm(d) be a mediator between axonal signals and the lysosome-mediated traf- 2010), nor we observe signs of axonal degeneration or damage in our fic leading to a correct location of proteins in compact myelin. Particu- EM analysis. Thus, the functional consequences of ApoD deficiency are larly, Plp recycling is regulated by the small GTPase RhoA in response restricted to the extracellular leaflet compaction defect, leaving to axon signals (Kippert, Trajkovic, Rajendran, Ries, & Simons, 2007; ganglioside-enriched hydrophilic spaces that alter the insulating proper- Trajkovic et al., 2006), and Fyn kinase is known to function upstream ties of myelin. of RhoA (reviewed by Baron & Hoekstra, 2010; White & Kramer- Nevertheless, we also observe secondary modifications in the Albers, 2014). Furthermore, the effect of ApoD on Fyn subcellular membrane distribution of other components, like GD1a ganglioside, localization in the PNS suggests that this membrane-bound kinase does mainly present in axons, or the effects on Mbp spatial distribution in traffic to and from myelin membrane, and predicts changes in down- CNS myelin, probably resulting in subsequent alterations of the intra- stream events, depending on its compartmentalization in membranes cellular leaflet of myelin in aged mice. with different lipid-based structures (Kramer et al., 1999). Interesting aspects of the signaling mechanisms controlling myelin The fact that ApoD is not present in peroxisomes (Pascua-Maestro biogenesis and maintenance are also revealed by our work. A feedback et al., 2017), also supports the specificity of the lysosome-dependent mechanism must exist that senses when myelin compaction has been traffic defects observed. Unlike in ApoD-KO mice, peroxisomal defec- completed, and informs the system to switch from a “maturation” to a tive Schwann cells build myelin sheaths with unaltered thickness or “maintenance” program. The expression and signaling profile that we periodicity (Kleinecke et al., 2017), and a dysfunction of peroxisomes observe in the absence of ApoD, with halted growth promoting and secondarily produces functional deficits in axons, with alterations in the lipogenic pathways downstream of Nrg1 III/ErB3 (Glenn & Talbot, amplitude of compound action potentials recorded both ex vivo and in 2013; Norrmen et al., 2014), might be the “bar-code” of a Schwann cell vivo. This deficit is not present in ApoD-KO nerves (Ganfornina et al., that has not received the “myelin-is-ok” signal. In this situation, Fyn GARCIA-MATEO ET AL. | 15 kinase is augmented in myelin membranes (rich in detergent resistant cells or cell parts (e.g., synapses), “marking” them for phagocytosis. domains) after an enhanced expression peak at P10, a time point when Shahraz et al. (2015) were able to control the extent of macrophage LFB staining starts revealing the hydrophobicity defect in ApoD-KO inflammatory response and oxidative outburst by using soluble poly- nerves. However, this enrichment cannot override the primary defect. sialic chains as competitors of the sialylated glycocalyx of cellular debris The decrease in ERK1/2 activated form indicates that Fyn, possibly from dying neurons. “trapped” in the incompletely compacted myelin domains, is not able to Upon peripheral nerve injury, Schwann cells strongly induce ApoD promote the downstream activation of the ERK pathway. Yet, Fyn expression. Without ApoD, myelin clearance, axonal regeneration and over-expression and enrichment in myelin could be, at least in part, injury resolution are delayed (Ganfornina et al., 2010). Myelin phagocy- responsible for an altered oligodendroglial maturation program, result- tosis is impaired in two well differentiated processes (Garcia-Mateo ing in the observed transient alteration in the transcription profiles of et al., 2014): myelin-macrophage recognition (phagocytosis initiation) Mbp and Mag (reviewed by White & Kramer-Albers, 2014) that are and myelin degradation rate (phagocytosis resolution). ApoD also con- later equalized with WT levels, without significant alterations in protein trols macrophage recruitment and activation, and the expression of amounts. Also, the aberrant distribution of Mbp might be a down- pro-inflammatory and pro-myelinolitic factors (TNFa, cPLA2, lysophos- stream effect of Fyn dysregulation, given its known role in the local phatidic acid, arachidonic acid, and Galectin-3). All of them get exacer- control of Mbp translation (reviewed by Kramer-Albers & White, bated in the absence of ApoD, preventing nerve regeneration and 2011). injury resolution. Our current understanding of ApoD role in lysosome- The glycocalyx removal defects bringing about an uncompacted mediated control of myelin glycocalyx provides new insights for under- IPL in ApoD-KO mice occur in both CNS and PNS. However, as it is standing nerve injury progression: it would take longer to desialylate the case for other known molecular processes, our study reveals that and opsonize a hypersialylated myelin and, once incorporated into the PNS and CNS myelinations differ in mechanistic details (e.g., lipogenic phagocytic cell, suboptimal lysosomal function would hinder myelin switch controlling Fasn expression) that require further study. degradation by macrophages (Garcia-Mateo et al., 2014) as well as astrocytes (Pascua-Maestro et al., 2017). 4.1 | Functional consequences of ApoD glycocalyx In summary, in addition to the role for ApoD in myelin compaction, management in health and disease our data suggest that hypersialylated cells in the nervous system would condition phagocytosis efficiency not only upon injury or disease A major conclusion revealed by our findings is that a lipid binding pro- (where myelin phagocytosis is a limiting factor for axonal regeneration), tein of the Lipocalin family, typically described as lipid transporters, ulti- but in normal conditions. Our plastic brain, where phagocytosis of mately controls the glycocalyx of cells thanks to the maintenance of excess neurons or synapses is paramount to normal development and healthy lysosomal membranes, which conditions sialidases location and influences whole life learning and memory, requires a fine control of activity. An excessive glycocalyx results in a functionally suboptimal the glycocalyx. ApoD contribution to such a control opens new myelin cover, both in CNS and PNS. research avenues to explore in the future. When hyperglycosylation is the primary defect in a myelinating cell, as it is the case for one of the human mutations in P0 (D32N), a 5 | CONCLUSIONS severe demyelination takes place mainly due to altered traffic of the hyperglycosylated P0 protein (Prada et al., 2012). Our study of ApoD in the process of myelin biogenesis supports the When lysosomal dysfunction is the primary defect, it is not surpris- following conclusions: ing that myelin defects follow, as it is the case in all lysosomal storage diseases known so far (Nave & Werner, 2014; Renaud, 2012). Thus, 1. A clear dissociation exists between two related, but independent, our reports on the cellular and molecular mechanisms of ApoD function phases of myelin compaction: intracellular-leaflet compaction should have important implications for the understanding of lysosomal (MDL) and extracellular-leaflet compaction (IPL). storage diseases. They open the possibility for ApoD-based therapies 2. IPL formation specifically requires an adequate lysosomal- for lysosomal diseases of diverse etiology. dependent myelin membrane recycling and remodeling. Other pathological conditions in which cell membranes are hyper- 3. ApoD-dependent lysosomal functional integrity is essential for glycosylated might have two distinct outcomes. For cells like astro- myelin glycocalyx management: (1) by optimizing lysosomal siali- cytes, dealing with oxidative stress protective responses in the brain, a dase activity (Neu1) and (2) by regulating myelin membrane traffic proper regulation of glycocalyx extent might contribute to their resil- that determines the distribution of membrane sialidase Neu3 and ience. However, for phagocytic cells (microglia, Schwann cells after the membrane-bound Fyn kinase. injury, or infiltrating macrophages), hypersialylated membranes might represent a problem. 4. ApoD loss has consequences for growth promoting and lipogenic Nomura et al. (2017) have recently shown that when activated signaling cascades, integrated in a dynamic neuron-glia communi- microglia are presented with dying or stressed neurons, they produce cation required throughout life. sialidases. Desialylation of membranes enables Galectin-3 (also pro- 5. Knocking-out ApoD neither alters the myelination initiation produced by phagocytic microglia and Schwann cells) to bind and opsonize cess nor prevents MDL compaction. It precludes extracellular 16 | GARCIA-MATEO ET AL.

leaflet compaction throughout life and results in sub-optimal mye- Sciences of the United States of America, 110(8), 3143–3148. https:// lin sheets, which impinges on behavioral outputs. doi.org/10.1073/pnas.1220104110 1220104110 [pii] Baron, W., & Hoekstra, D. (2010). On the biogenesis of myelin mem- 6. Hypersialylation in the absence of ApoD is a mechanism that can branes: Sorting, trafficking and cell polarity. FEBS Letters, 584(9), be generalized to ApoD-expressing glial cells, and targeting ApoD 1760–1770. https://doi.org/10.1016/j.febslet.2009.10.085 S0014– to lysosomes is sufficient for glial cells to control their glycocalyx. 5793(09)00877-1 [pii] Da Silva, J. S., Hasegawa, T., Miyagi, T., Dotti, C. G., & Abad-Rodriguez, We need healthy lysosomes for the biogenesis of a healthy myelin and J. (2005). Asymmetric membrane ganglioside sialidase activity for myelin management upon injury or disease, and ApoD contributes specifies axonal fate. Nature Neuroscience, 8(5), 606–615. https://doi. significantly to both goals. org/nn1442 [pii] 10.1038/nn1442 Dassati, S., Waldner, A., & Schweigreiter, R. (2014). Apolipoprotein D takes center stage in the stress response of the aging and degen- ACKNOWLEDGMENT erative brain. Neurobiology Aging, 35(7), 1632–1642. https://doi. We thank C. Perez-Segurado for technical assistance and L. Brieva org/10.1016/j.neurobiolaging.2014.01.148 S0197–4580(14)00170-5 for help with EM analysis. This work was supported by grants to M. [pii] D. G. and D. S.: Junta de Castilla y Leon (JCyL) grant VA180A11-2, de Magalhaes, J. P., Curado, J., & Church, G. M. (2009). Meta-analysis of age-related gene expression profiles identifies common signatures of and MICINN-MINECO grants BFU2011–23978 and BFU2015– aging. Bioinformatics, 25(7), 875–881. https://doi.org/10.1093/bioin- 68149-R, co-financed by European Regional Development fund. N. formatics/btp073 btp073 [pii] – G- M. was supported by a JAEPre CSIC fellowship. RPM was sup- Feldmann, A., Amphornrat, J., Schonherr, M., Winterstein, C., Mobius, ported by a JCyL fellowship to young researchers (call#EDU/1883/ W., Ruhwedel, T., ... Kramer-Albers, E. M. (2011). Transport of the 2013), financed by the European Social Fund, Operational Pro- major myelin proteolipid protein is directed by VAMP3 and VAMP7. Journal of Neuroscience, 31(15), 5659–5672. https://doi.org/10.1523/ gramme for Castilla y Leon, and managed by Consejería de Educa- JNEUROSCI.6638-10.2011 31/15/5659 [pii] cion (JCyL). Ganfornina, M. D., Do Carmo, S., Lora, J. M., Torres-Schumann, S., Vogel, M., Allhorn, M., ... Sanchez, D. (2008). Apolipoprotein D is involved CONFLICT OF INTERESTS in the mechanisms regulating protection from oxidative stress. Aging Cell, 7(4), 506–515. https://doi.org/10.1111/j.1474-9726.2008. The authors declare that they have no competing interests. 00395.x ACE395 [pii] Ganfornina, M. D., Do Carmo, S., Martinez, E., Tolivia, J., Navarro, A., ORCID Rassart, E., & Sanchez, D. (2010). ApoD, a glia-derived apolipoprotein, Maria D. Ganfornina http://orcid.org/0000-0002-8567-4826 is required for peripheral nerve functional integrity and a timely response to injury. Glia, 58(11), 1320–1334. https://doi.org/10.1002/ glia.21010 REFERENCES Ganfornina, M. D., Sanchez, D., Pagano, A., Tonachini, L., Descalzi-Can- Aggarwal, S., Yurlova, L., & Simons, M. (2011). Central nervous system cedda, F., & Martinez, S. (2005). Molecular characterization and myelin: Structure, synthesis and assembly. Trends in Cell Biology, 21 developmental expression pattern of the chicken apolipoprotein D (10), 585–593. https://doi.org/10.1016/j.tcb.2011.06.004 S0962– gene: Implications for the evolution of vertebrate lipocalins. Develop- 8924(11)00117-6 [pii] mental Dynamics, 232(1), 191–199. https://doi.org/10.1002/dvdy. Arthur-Farraj, P. J., Latouche, M., Wilton, D. K., Quintes, S., Chabrol, E., 20193 Banerjee, A., ... Jessen, K. R. (2012). c-Jun reprograms Schwann cells Garcia-Mateo, N., Ganfornina, M. D., Montero, O., Gijon, M. A., Murphy, of injured nerves to generate a repair cell essential for regeneration. R. C., & Sanchez, D. (2014). Schwann cell-derived Apolipoprotein D Neuron, 75(4), 633–647. https://doi.org/S0896-6273(12)00583-1 controls the dynamics of post-injury myelin recognition and degrada- [pii] 10.1016/j.neuron.2012.06.021 tion. Frontiers in Cell Neuroscience, 8, 374. https://doi.org/10.3389/ Bajo-Graneras,~ R., Ganfornina, M. D., Martin-Tejedor, E., & Sanchez, D. fncel.2014.00374 (2011a). Apolipoprotein D mediates autocrine protection of astro- Giese, K. P., Martini, R., Lemke, G., Soriano, P., & Schachner, M. (1992). cytes and controls their reactivity level, contributing to the functional Mouse P0 gene disruption leads to hypomyelination, abnormal maintenance of paraquat-challenged dopaminergic systems. Glia, 59 expression of recognition molecules, and degeneration of myelin and (10), 1551–1566. https://doi.org/10.1002/glia.21200 axons. Cell, 71(4), 565–576. https://doi.org/0092-8674(92)90591-Y Bajo-Graneras,~ R., Sanchez, D., Gutierrez, G., Gonzalez, C., Do Carmo, S., [pii] Rassart, E., & Ganfornina, M. D. (2011b). Apolipoprotein D alters the Glenn, T. D., & Talbot, W. S. (2013). Signals regulating myelination in early transcriptional response to oxidative stress in the adult cerebel- peripheral nerves and the Schwann cell response to injury. Current lum. Journal of Neurochemistry, 117(6), 949–960. https://doi.org/10. Opinion in Neurobiology, 23(6), 1041–1048. https://doi.org/S0959- 1111/j.1471-4159.2011.07266.x 4388(13)00129-3 [pii] 10.1016/j.conb.2013.06.010 Bakhti, M., Aggarwal, S., & Simons, M. (2014). Myelin architecture: Zip- Gong, Y., Tagawa, Y., Lunn, M. P., Laroy, W., Heffer-Lauc, M., Li, C. Y., pering membranes tightly together. Cellular and Molecular Life Scien- ... Sheikh, K. A. (2002). Localization of major gangliosides in the ces, 71(7), 1265–1277. https://doi.org/10.1007/s00018-013-1492-0 PNS: Implications for immune neuropathies. Brain, 125(Pt 11), 2491– Bakhti, M., Snaidero, N., Schneider, D., Aggarwal, S., Mobius, W., Jansh- 2506. off, A., ... Simons, M. (2013). Loss of electrostatic cell-surface repul- Ishii, A., Furusho, M., & Bansal, R. (2013). Sustained activation of ERK1/ sion mediates myelin membrane adhesion and compaction in the 2 MAPK in oligodendrocytes and schwann cells enhances myelin central nervous system. Proceedings of the National Academy of growth and stimulates oligodendrocyte progenitor expansion. Journal GARCIA-MATEO ET AL. | 17

of Neuroscience, 33(1), 175–186. https://doi.org/10.1523/JNEURO- McKenzie, I. A., Ohayon, D., Li, H., de Faria, J. P., Emery, B., Tohyama, SCI.4403-12.2013 33/1/175 [pii] K., & Richardson, W. D. (2014). Motor skill learning requires active – Jacob, C., Lebrun-Julien, F., & Suter, U. (2011). How histone deacetylases central myelination. Science, 346(6207), 318 322. https://doi.org/10. control myelination. Molecular Neurobiology, 44(3), 303–312. https:// 1126/science.1254960 346/6207/318 [pii] doi.org/10.1007/s12035-011-8198-9 Miyagi, T., & Yamaguchi, K. (2012). Mammalian sialidases: Physiological – Kappagantula, S., Andrews, M. R., Cheah, M., Abad-Rodriguez, J., Dotti, and pathological roles in cellular functions. Glycobiology, 22(7), 880 C. G., & Fawcett, J. W. (2014). Neu3 sialidase-mediated ganglioside 896. https://doi.org/10.1093/glycob/cws057 cws057 [pii] conversion is necessary for axon regeneration and is blocked in CNS Navarro, A., Del Valle, E., & Tolivia, J. (2004). Differential expression of axons. Journal of Neuroscience, 34(7), 2477–2492. https://doi.org/10. apolipoprotein d in human astroglial and oligodendroglial cells. Journal 1523/JNEUROSCI.4432-13.2014 34/7/2477 [pii] of Histochemistry and Cytochemistry, 52(8), 1031–1036. https://doi. Kiernan, J. A. (2010). Why is lithium carbonate used for myelin staining org/10.1369/jhc.3A6213.2004 52/8/1031 [pii] with Luxol fast blue? Biotechnic & Histochemistry, 85(5), 325. https:// Nave, K. A., & Werner, H. B. (2014). Myelination of the nervous system: doi.org/10.3109/10520291003694502 Mechanisms and functions. Annual Review of Cell and Developmental – Kippert, A., Trajkovic, K., Rajendran, L., Ries, J., & Simons, M. (2007). Rho Biology, 30, 503 533. https://doi.org/10.1146/annurev-cellbio- regulates membrane transport in the endocytic pathway to control 100913-013101 plasma membrane specialization in oligodendroglial cells. Journal of Newbern, J. M., Li, X., Shoemaker, S. E., Zhou, J., Zhong, J., Wu, Y., ... Neuroscience, 27(13), 3560–3570. https://doi.org/27/13/3560 [pii] Snider, W. D. (2011). Specific functions for ERK/MAPK signaling dur- 10.1523/JNEUROSCI.4926-06.2007 ing PNS development. Neuron, 69(1), 91–105. https://doi.org/10. – Kleinecke, S., Richert, S., de Hoz, L., Brugger, B., Kungl, T., Asadollahi, E., 1016/j.neuron.2010.12.003 S0896 6273(10)00988-8 [pii] ... Kassmann, C. M. (2017). Peroxisomal dysfunctions cause lysoso- Nomura, K., Vilalta, A., Allendorf, D. H., Hornik, T. C., & Brown, G. C. mal storage and axonal Kv1 channel redistribution in peripheral neu- (2017). Activated microglia desialylate and phagocytose cells via ropathy. Elife, 6, https://doi.org/10.7554/eLife.23332 e23332 [pii] neuraminidase, galectin-3, and mer tyrosine kinase. Journal of Immu- – Klugmann, M., Schwab, M. H., Puhlhofer, A., Schneider, A., Zimmermann, nology, 198(12), 4792 4801. https://doi.org/10.4049/jimmunol. F., Griffiths, I. R., & Nave, K. A. (1997). Assembly of CNS myelin in 1502532 jimmunol.1502532 [pii] the absence of proteolipid protein. Neuron, 18(1), 59–70. https://doi. Norrmen, C., Figlia, G., Lebrun-Julien, F., Pereira, J. A., Trotzmuller, M., org/S0896-6273(01)80046-5 [pii] Kofeler, H. C., ... Suter, U. (2014). mTORC1 controls PNS myelina- Kramer-Albers, E. M., & White, R. (2011). From axon-glial signalling to tion along the mTORC1-RXRgamma-SREBP-lipid biosynthesis axis in – myelination: The integrating role of oligodendroglial Fyn kinase. Cellu- Schwann cells. Cell Reports, 9(2), 646 660. https://doi.org/10.1016/j. lar and Molecular Life Sciences, 68(12), 2003–2012. https://doi.org/ celrep.2014.09.001 S2211-1247(14)00769-4 [pii] 10.1007/s00018-010-0616-z Norton, W. T., & Poduslo, S. E. (1973). Myelination in rat brain: Method – Kramer, E. M., Klein, C., Koch, T., Boytinck, M., & Trotter, J. (1999). Com- of myelin isolation. J Neurochemistry, 21(4), 749 757. partmentation of Fyn kinase with glycosylphosphatidylinositol- Ong, W. Y., Lau, C. P., Leong, S. K., Kumar, U., Suresh, S., & Patel, S. C. anchored molecules in oligodendrocytes facilitates kinase activation (1999). Apolipoprotein D gene expression in the rat brain and light during myelination. Journal of Biological Chemistry, 274(41), 29042– and electron microscopic immunocytochemistry of apolipoprotein D 29049. expression in the cerebellum of neonatal, immature and adult rats. – Krinke, G. J., Vidotto, N., & Weber, E. (2000). Teased-fiber technique for Neuroscience, 90(3), 913 922. https://doi.org/S0306-4522(98)00507- peripheral myelinated nerves: Methodology and interpretation. Toxi- 7 [pii] cologic Pathology, 28(1), 113–121. https://doi.org/10.1177/ Pascua-Maestro, R., Diez-Hermano, S., Lillo, C., Ganfornina, M. D., & 019262330002800114 Sanchez, D. (2017). Protecting cells by protecting their vulnerable Lebrun-Julien, F., Bachmann, L., Norrmen, C., Trotzmuller, M., Kofeler, H., lysosomes: Identification of a new mechanism for preserving lysoso- Ruegg, M. A., ... Suter, U. (2014). Balanced mTORC1 activity in oli- mal functional integrity upon oxidative stress. PLoS Genetics, 13(2), godendrocytes is required for accurate CNS myelination. Journal of e1006603. https://doi.org/10.1371/journal.pgen.1006603 – Neuroscience, 34(25), 8432–8448. https://doi.org/10.1523/JNEURO- PGENETICS-D-16 02748 [pii] SCI.1105-14.2014 34/25/8432 [pii] Pearse, A. G. (1955). Copper phthalocyanins as phospholipid stains. Jour- – Li, H., Ruberu, K., Munoz, S. S., Jenner, A. M., Spiro, A., Zhao, H., ... Gar- nal of Pathology and Bacteriology, 70(2), 554 557. ner, B. (2015). Apolipoprotein D modulates amyloid pathology in Pereira, J. A., Lebrun-Julien, F., & Suter, U. (2012). Molecular mechanisms APP/PS1 Alzheimer’s disease mice. Neurobiology Aging, 36(5), 1820– regulating myelination in the peripheral nervous system. Trends in 1833. https://doi.org/10.1016/j.neurobiolaging.2015.02.010 S0197– Neuroscience, 35(2), 123–134. https://doi.org/10.1016/j.tins.2011.11. 4580(15)00110-4 [pii] 006 S0166–2236(11)00193-7 [pii] Livak, K. J., & Schmittgen, T. D. (2001). Analysis of relative gene expres- Prada, V., Passalacqua, M., Bono, M., Luzzi, P., Scazzola, S., Nobbio, L. A., sion data using real-time quantitative PCR and the 2(-Delta Delta C ... Grandis, M. (2012). Gain of glycosylation: A new pathomechanism (T)) Method. Methods, 25(4), 402–408. https://doi.org/10.1006/meth. of myelin protein zero mutations. Annals of Neurology, 71(3), 427– 2001.1262 S1046–2023(01)91262-9 [pii] 431. https://doi.org/10.1002/ana.22695 Loerch, P. M., Lu, T., Dakin, K. A., Vann, J. M., Isaacs, A., Geula, C., ... Provost, P. R., Villeneuve, L., Weech, P. K., Milne, R. W., Marcel, Y. L., & Yankner, B. A. (2008). Evolution of the aging brain transcriptome and Rassart, E. (1991). Localization of the major sites of rabbit apolipo- synaptic regulation. PLoS One, 3(10), e3329. https://doi.org/10.1371/ protein D gene transcription by in situ hybridization. Journal of Lipid journal.pone.0003329 Research, 32(12), 1959–1970. Lunn, M. P., Johnson, L. A., Fromholt, S. E., Itonori, S., Huang, J., Vyas, A. Quarles, R. H., Macklin, W. B., & Morell, P. (2006). Myelin formation, A., ... Sheikh, K. A. (2000). High-affinity anti-ganglioside IgG antibod- structure and biochemistry. In S. Brady, G. Siegel, R. W. Albers, & D. ies raised in complex ganglioside knockout mice: Reexamination of Price (Eds.), Basic Neurochemistry: Molecular, Cellular and Medical GD1a immunolocalization. Journal of Neurochemistry, 75(1), 404–412. Aspects. (pp. 51–71). Amsterdam, Netherlands: Elsevier, Inc. 18 | GARCIA-MATEO ET AL.

Renaud, D. L. (2012). Lysosomal disorders associated with leukoence- https://doi.org/S2211-1247(16)31684-9 [pii] 10.1016/j. phalopathy. Seminars in Neurology, 32, (1), 51–54. https://doi.org/10. celrep.2016.12.011 1055/s-0032-1306386 Trajkovic, K., Dhaunchak, A. S., Goncalves, J. T., Wenzel, D., Schneider, Ruiz, M., Sanchez, D., Correnti, C., Strong, R. K., & Ganfornina, M. D. A., Bunt, G., ... Simons, M. (2006). Neuron to glia signaling triggers (2013). Lipid-binding properties of human ApoD and Lazarillo-related myelin membrane exocytosis from endosomal storage sites. Journal of lipocalins: Functional implications for cell differentiation. Febs Journal, Cell Biology, 172(6), 937–948. https://doi.org/jcb.200509022 [pii] 280(16), 3928–3943. https://doi.org/10.1111/febs.12394 10.1083/jcb.200509022 Sanchez, D., Bajo-Graneras,~ R., Del Cano-Espinel, M., Garcia-Centeno, R., Vajn, K., Viljetic, B., Degmecic, I. V., Schnaar, R. L., & Heffer, M. (2013). Garcia-Mateo, N., Pascua-Maestro, R., & Ganfornina, M. D. (2015). Differential distribution of major brain gangliosides in the adult Aging without Apolipoprotein D: Molecular and cellular modifications mouse central nervous system. PLoS One, 8(9), e75720. https://doi. in the hippocampus and cortex. Experimental Gerontology, 67,19–47. org/10.1371/journal.pone.0075720 PONE-D-13–24906 [pii] – https://doi.org/10.1016/j.exger.2015.04.003 S0531 5565(15)00113- van der Knaap, M. S., & Bugiani, M. (2017). Leukodystrophies: A pro- 8 [pii] posed classification system based on pathological changes and patho- Sanchez, D., Ganfornina, M. D., & Martinez, S. (2002). Expression pattern genetic mechanisms. Acta Neuropathology, https://doi.org/10.1007/ of the lipocalin apolipoprotein D during mouse embryogenesis. Mech- s00401-017-1739-1 10.1007/s00401-017-1739-1 [pii] – – anisms of Development, 110(1 2), 225 229. https://doi.org/ White, R., & Kramer-Albers, E. M. (2014). Axon-glia interaction and S0925477301005780 [pii] membrane traffic in myelin formation. Frontiers in Cell Neuroscience, Sedzik, J., Jastrzebski, J. P., & Grandis, M. (2015). Glycans of myelin pro- 7, 284. https://doi.org/10.3389/fncel.2013.00284 – teins. Journal of Neuroscience Research, 93(1), 1 18. https://doi.org/ Xiao, L., Ohayon, D., McKenzie, I. A., Sinclair-Wilson, A., Wright, J. L., 10.1002/jnr.23462 Fudge, A. D., ... Richardson, W. D. (2016). Rapid production of new Seyrantepe, V., Landry, K., Trudel, S., Hassan, J. A., Morales, C. R., & oligodendrocytes is required in the earliest stages of motor-skill Pshezhetsky, A. V. (2004). Neu4, a novel human lysosomal lumen sia- learning. Nature Neuroscience, 19(9), 1210–1217. https://doi.org/10. lidase, confers normal phenotype to sialidosis and galactosialidosis 1038/nn.4351 nn.4351 [pii] cells. Journal of Biological Chemistry, 279(35), 37021–37029. https:// Yuan, J. S., Reed, A., Chen, F., & Stewart, C. N. Jr. (2006). Statistical anal- doi.org/10.1074/jbc.M404531200 M404531200 [pii] ysis of real-time PCR data. BMC Bioinformatics, 7(85), https://doi.org/ Shahraz, A., Kopatz, J., Mathy, R., Kappler, J., Winter, D., Kapoor, S., ... 1471-2105-7-85 [pii] 10.1186/1471–2105-7–85 Neumann, H. (2015). Anti-inflammatory activity of low molecular weight polysialic acid on human macrophages. Science Reports, 5, 16800. https://doi.org/10.1038/srep16800 srep16800 [pii] SUPPORTING INFORMATION

Simons, M., Misgeld, T., & Kerschensteiner, M. (2014). A unified cell bio- Additional Supporting Information may be found online in the sup- logical perspective on axon-myelin injury. Journal of Cell Biology, 206 porting information tab for this article. (3), 335–345. https://doi.org/jcb.201404154 [pii] 10.1083/ jcb.201404154 Simons, M., & Trotter, J. (2007). Wrapping it up: The cell biology of myelination. Current Opinion in Neurobiology, 17(5), 533–540. https://doi. How to cite this article: García-Mateo N, Pascua-Maestro R, org/S0959-4388(07)00100-6 [pii] 10.1016/j.conb.2007.08.003 Perez-Castellanos A, Lillo C, Sanchez D, Ganfornina MD. Myelin Snaidero, N., & Simons, M. (2017). The logistics of myelin biogenesis in extracellular leaflet compaction requires apolipoprotein D mem- the central nervous system. Glia, 65(7), 1021–1031. https://doi.org/ brane management to optimize lysosomal-dependent recycling 10.1002/glia.23116 and glycocalyx removal. Glia. 2017;00:1–18. https://doi.org/10. Soreq, L., Rose, J., Soreq, E., Hardy, J., Trabzuni, D., Cookson, M. R., ... Ule, J. (2017). Major shifts in glial regional identity are a transcrip- 1002/glia.23274 tional hallmark of human brain aging. Cell Reports, 18(2), 557–570.

Myelin Extracellular Leaflet Compaction Requires Apolipoprotein D Membrane Management to Optimize Lysosomal&#8208;Dependent

Myelin Extracellular Leaflet Compaction Requires Apolipoprotein D Membrane Management to Optimize Lysosomal‐Dependent