Blood vessels guide Schwann cell migration in the adult demyelinated CNS through Eph/ephrin signaling

Myelination, the evolutionary characteristic acquired by vertebrates to allow rapid and saltatory nerve conduction, is supported by two different glial cell types, oligodendrocytes in the central nervous system (CNS), and Schwann cells (SC) in the peripheral nervous system (PNS). These two cell types are mutually exclusive in physiological conditions. However, in demyelinating diseases or injury, this PNS/CNS segregation is compromised: SC can invade and repair the CNS [24, 30], while oligodendrocytes can myelinate peripheral nerve root axons [18]. Remyelination of CNS axons by SC protects axons, restores axonal conduction and even reverses neurological deficits [26], highlighting their potential to rescue the injured CNS [62]. These remyelinating SC arise either from the PNS [6, 43], or are generated from adult oligodendrocyte precursors cells (OPC) [59, 61]. However, SC remyelination of CNS axons is always restricted to the spinal root’s entry and exit zones (reviewed in [62]), and the presence of peripheral myelin has been frequently observed close to blood vessels (BV) [24]. These observations suggest that although SC migrate efficiently in vitro [11, 45] and in vivo [16] into the PNS, their survival and migration within the CNS are limited. The presence of peripheral myelin close to BV raises the possibility for BV to play a role in guiding SC movements within the CNS. While migration along BV has recently been described for different CNS cell types [15, 28, 58] including for SC along regenerating nerves [16], their role in SC invasion of the CNS has not been explored.

CNS white matter [9, 17, 32] inhibits SC migration. Among the molecules involved in various cell-type segregation and guidance within the CNS, the Eph/ephrin family is implicated in both developmental [41, 57] and pathological conditions [47]. In particular, EphrinB3 is expressed in myelin in brain and mouse spinal cord [12, 23, 55], and plays an important role in preventing neurite outgrowth [12], axonal regeneration [23] and OPC  differentiation [55]. EphrinB3 also acts as a repellent molecule in the guidance of axon tracts in the spinal cord during development [49]. Hence, these receptors could also mediate similar SC repulsion by myelin-associated EphrinB3. In addition, EphB is involved in SC sorting and migration in regenerating peripheral nerves [51]. Eph/ephrin signaling modulates cell–cell adhesion, which results in increased integrin-mediated adhesion of Eph/ephrin-expressing cells [5, 20] or, as in the case of EphB signaling, in cell sorting by re-localization of N-cadherin in SC [51]. Furthermore, EphrinB ligands control cell migration through positive adhesion to substrates such as collagen and fibronectin (FN) [54, 60], main components of perivascular extracellular matrix (ECM). Therefore, these cues could influence the capacity of SC to migrate within the CNS and/or interact with CNS myelin.

In spite of these observations, a role for BV in SC migration/recruitment in the CNS and for EphrinB3 in modulating SC interactions with CNS myelin and/or BV has not been investigated. Using ex vivo, in vivo and in vitro paradigms, we show that BV are a preferred substrate for SC migration into the CNS. Moreover, we establish that CNS myelin plays an essential role in SC exclusion from the CNS and demonstrate that this effect is partially mediated by EphrinB3. Finally, myelin-associated EphrinB3 modulates SC adhesion to the ECM component FN, via interactions with integrinβ1. This increased SC–FN adhesion to perivascular ECM overrules SC inhibition by myelin and promotes SC migration along BV, facilitating their arrival at the lesion. Thus, Eph/ephrin may guide SC within the CNS according to a dual mode of action, repulsing SC from CNS white matter on one hand, and favoring their interaction with BV on the other. These observations shed new light on the mechanism of SC invasion into the damaged CNS.

Animals 8-week-old C57Bl/6JR female mice were purchased from Janvier Labs (Rodent Research Models and Associated Services) and used for engraftment. SC were obtained from green fluorescent protein-tagged actin (GFP⁺) transgenic mice, Krox20^Cre/+R26R^YFP/+ mice, and TdTomato Krox20^Cre/flox,R26^mT/+ mice which were previously characterized [18, 44], and maintained at ICM and IBENS animal facilities. Animal experiments were performed according to European Community regulations, ICM and INSERM ethical committee (authorization 75-348; 20/04/2005) and were approved by the local Darwin ethical committee.

SC isolation and purification Sciatic nerves from GFP⁺ mice were isolated at postnatal day 15. Purification procedure was adapted from the previously described protocol [7]. Briefly, enzymatic dissociation was performed by incubation with trypsin 0.025% and collagenase (420 U/ml) for 10 min at 37 °C, followed by mechanical dissociation through different needle gauges. After ending dissociation with fetal calf serum (FCS), SC were seeded in FN-coated flasks, and expanded in Dulbecco’s modified Eagle medium, containing 10% heat-inactivated FCS serum, penicillin (100 mg/ml), streptomycin (100 U/ml), human recombinant Neu-differentiation factor ß (hrNDFß) (125 ng/ml), insulin (10 µg/ml) and forskolin (2 µg/ml). SC were purified by differential adhesion [37], and used at passage P2 or P3. Purification was controlled by immunocytochemistry for p75 and GFAP as markers of non-myelinating Schwann cells [33], and exclusion of the Thy1-2 marker of mouse fibroblasts [14]. For adhesion, migration and blocking receptor assays, SC were maintained in Sato serum-free medium [13] supplemented with hrNDFβ (125 ng/ml), and forskolin (2 µg/ml).

iDISCO whole-mount immunofluorescence and imaging Spinal cords were processed as described in the iDISCO protocol [52], including modifications described in the updated online protocol (https://​idisco.​info, Dec 2016). The primary antibody used was rabbit anti-RFP (1:1000, Rockland). Secondary antibodies used were donkey anti-rabbit Cy3 (1:800, Jackson Immunoresearch) and donkey anti-mouse IgG Cy5 (1:800, Jackson Immunoresearch) for intravascular staining. The cleared samples were imaged with a light sheet microscope (Ultramicroscope II; LaVision Biotec).

RNA transcriptome analysis For RNA preparations, SC from Krox20^Cre/+,R26^mT/+ (TdTomato) were freshly isolated from 2-week-old sciatic nerves. Nerves were cultured 4 days in vitro in the absence of growth factor to allow SC to de-differentiate and migrate out of the nerve. SC were then selected by FACS based on tdTomato expression. Library preparation and Illumina sequencing were performed at the IBENS genomic core facility. Briefly, (polyA +) mRNAs were purified from 250 ng of total RNA using oligo(dT). Libraries were prepared using the strand-specific RNA-Seq library preparation TruSeq Stranded mRNA kit (Illumina). Libraries were multiplexed by 6 in 1 high-output flow cells. A 75-bp read sequencing was performed on a NextSeq 500 device (Illumina). A mean of 94 ± 9.5 million reads passing Illumina quality filter was obtained for each of the six samples.

The analyses were performed using the Eoulsan pipeline [35], including read filtering, mapping, alignment filtering, read quantification, normalization and differential analysis. Before mapping, poly N read tails were trimmed, reads of ≤ 40 bases were removed, and reads with quality mean of ≤ 30 bases were discarded. Reads were then aligned against the Mus musculus genome from Ensembl version 84 using STAR [22]. Alignments from reads matching more than once on the reference genome were removed using the Java version of SamTools [39]. All overlapping regions between alignments and referenced gene were counted using HTSeq-count 0.5.3 [4]. The sample counts were normalized using DESeq 1.8.3 [3]. Statistical treatments and differential analyses were also performed using DESeq 1.8.3.

Data availability The RNASeq gene expression data and raw fastq files are available on the GEO repository (www.​ncbi.​nlm.​nih.​gov/​geo/​) under accession number: GSE107401 (accession password: mlkfkwoezxujvsn).

Myelin protein extract isolation Myelin was purified by sucrose gradient centrifugation [48]. Cerebral hemispheres of adult mice (3 months old) were homogenized on ice in 0.35 M sucrose and 5 mM EGTA, and the suspension was overlaid onto an equivalent volume of 0.85 M sucrose and 5 mM EGTA, and centrifuged at 100,000×g at 4 °C for 20 min. The myelin-containing fraction at the interface was collected, diluted threefold in distilled water, and centrifuged at 100,000×g at 4 °C for 30 min. After washing with distilled water, the isolated myelin pellet was resuspended in 20 mM Tris–HCl, aliquoted, and stored at – 20 °C.

Pre-clustering of recombinant EphrinB3–Fc Mouse EphrinB3–Fc fragments and human Fc were purchased from R&D Systems. The soluble forms of EphrinB3–Fc and its control Fc have low effect on receptor activation [19]; therefore, they were mixed with anti-mouse Fc–IgG and anti-human Fc–IgG (Alexa 555), respectively (ratio = 1:5), and incubated for 1 h at 37 °C prior to addition to SC [25].

Adhesion and spreading assays Adhesion and spreading in vitro assays were performed in 24-well dishes. Silicon strips on coverslips were used to separate two coated areas of each coverslip [8]. Surfaces were coated overnight at 37 °C with recombinant EphrinB3–Fc fusion at 10 µg/mL and Fc equimolar (as control) on each half, or myelin extract (100 µg/mL) and PBS buffer (as control). Before cell seeding, strips were removed and coverslips were washed carefully with PBS. 10⁵ SC were seeded in serum-free Sato medium to avoid proliferation, and allowed to adhere for 3 h. Data were always expressed as ratio in respect to the intra-coverslip control [12].

Survival assay GFP⁺SC were seeded on uncoated glass coverslips in normal medium. After overnight adhesion, medium was changed, adding Sato serum-free medium supplemented with clustered EphrinB3 at 10 µg/mL or Fc equimolar (as control), or with myelin extract (100 µg/mL) or PBS (as control). SC were incubated for 3 h or 24 h as specified in each experiment. After fixation in 4% paraformaldehyde (5 min), SC were immuno-stained for caspase 3 adding Hoechst dye to visualize all nuclei, and coverslips were mounted with fluoromount.

Migration assay SC were resuspended at 3 × 10⁶ cells/ml in Sato medium containing 0.8% low-melting point agarose (Sigma). One drop (1.5 μL) of this suspension was applied to the center of FN +EphrinB3, or FN +Fc-coated glass coverslips, which were placed at 4 °C for 1 min to allow the agarose to solidify. The cooled drop was covered with Sato medium with hrNDFß (125 ng/ml) and forskolin (2 µg/ml), and placed up to 6 h at 37 °C in the incubating chamber of a video-microscope (ZEISS).

Immuno-staining Cultured SC were fixed for 5 min in 4% paraformaldehyde prior to immuno-staining and mice were killed by trans-cardiac perfusion of PBS followed by cold 4% paraformaldehyde, and post-fixed in the same fixative for 1 h. Spinal cords were cryo-protected by immersion in 20% sucrose solution overnight, embedded in cryomatrix (Thermo Scientific), and frozen in cold isopentane at − 60 °C. Finally, they were sectioned with a cryostat at 12 µm (Leica Microsystems). Both cells and sections were washed, blocked in 5% BSA for 40 min and incubated with the primary antibodies. While cells were incubated 1 h at room temperature, slides were incubated overnight at 4 °C. For MOG staining, sections were incubated with absolute ethanol for 10 min followed by primary antibody, and then washed profusely. Primary antibodies were as follows: anti-EphA4 (1:50, AF641, R&D Systems); anti-EphA4-Tyr(602) (1:50, EP2731, ECM Biosciences); anti-EphB6 (1:50, SAB4503476, Sigma); anti-EphB1 (1:50, SAB4500776, Sigma; anti-Eph receptor B1 + Eph receptor B2 (phospho Y594) (1:50, ab61791, Abcam); anti-Ki67 (1:100, 556003, BD Biosciences); anti-cleaved caspase3 (1:500, 9661S, Cell Signalling); anti-GFP (1:400, GFP-1020, Aves); anti-MOG (1:20, mouse IgG1 hybridoma, clone C18C5; provided by C. Linnington, University of Glasgow, Glasgow, United Kingdom); anti-MBP (1:50, ab7349, Sigma); anti-Glut1 (1:100, 07-1401, Merck Millipore; and 1:400, MABS132, Sigma); anti-Fibronectin (1:600, F6140, Sigma); anti-CD31 (1:200, 553370, BD Pharmigen); anti-NF200 (1:200, N4142,Sigma), anti-p75 (1:100, 8238S, Ozyme), anti-CD13 (1:50, MCA2183, BioRad), anti-CD68 (1:400, MCA1957, BioRad), anti-CD11b (1:400, MCA74G, BioRad), anti-F8/40 (1:100, MCA497R, BioRad), anti-Collagen 4(1:400, ab19808, Abcam), anti-Olig2 (1:300, MABN50, Millipore); anti-sox10 (1:50, AF2864, R&D Systems); and anti-CD13 (1:200, 553370, BD Pharmingen). Next, cells or sections were washed and incubated with secondary antibodies and Hoechst dye for 1 h at room temperature. The excess secondary antibody was removed by several PBS washes, and coverslips/slides were mounted using fluoromount.

Electron microscopy For electron microscopy, mice were perfused with PBS followed by 4% paraformaldehyde/2.5% glutaraldehyde (Electron Microscopy Science) in PBS for 45 min. Dissected spinal cords were post-fixed with the same solution for 2 h, then sectioned into 60-µm slices with a vibratome and washed twice with PBS before enzyme immunolabeling. For DAB revelation, endogenous peroxidase was inhibited with a methanol/oxygen peroxide incubation, and washed and blocked by 5% BSA for 1 h. Sections were incubated with anti-GFP overnight at 4 °C, then washed and incubated with a secondary biotinylated antibody for 2 h at room temperature. After several washes with PB 0.1 M, sections were incubated with the ABC kit (VECTASTAIN^® ABC-HRP Kit, Vector Lab) containing peroxidase–anti-peroxidase for 40 min followed by a DAB/oxygen peroxide mix before stopping the reaction with distilled water. Samples were fixed in 2% osmium tetroxide (Sigma-Aldrich) 30 min, washed gently and incubated with 5% uranyl acetate for 30 min in the dark. After dehydration, samples were embedded in Epon resin 812. Ultra-thin sections (80 nm) were examined with a HITACHI 120 kV HT-7700 electron microscope.

Demyelinating lesions and grafts Wild-type mice were anaesthetized with a ketamine/xylazine mixture. Demyelination was induced by stereotaxic injection of lysolecithin (LPC) (1%, 0.5 µl) in PBS. LPC or PBS (in control animals) was injected into the dorsal funiculus of the spinal cord at the level of T8–T9 in the dorsal column of white matter using a glass micropipette. SC (10⁵/2 µL) were injected the same day, two vertebrates caudally (4 mm) in the same tract.

Western blotting SC (3 × 10⁵ cells/well) were lysed in RIPA buffer with Complete^® and Phosphostop^® inhibitor, and analyzed by electrophoresis in an SDS 4–20% MINI PROTEAN TGX gel. After electrophoresis, proteins were transferred electrophoretically to polyvinylidene difluoride membranes and probed with the following antibodies:anti-EphB6 (1:500, SAB4503476, Sigma), anti-EphB1 (1:500, SAB4500776, Sigma), anti-EphA4 (4 µg/mL, 37-1600, ThermoFisher), anti-p-EphB1 + 2 (1:300, ab61791, Abcam), anti-p-EphA4 (1:300, EP2731, ECM Biosciences), anti-Integrinβ1 (1:500, 550531, BD Pharmingen), anti-EphrinB3 (1:250, 1 µg/mL, AF395 R&D Systems), anti-MBP (1:1000, ab980, Millipore), anti-GAPDH (1:5000, MAB374, Millipore) and anti-Actin (1:50,000, A2228, Sigma). Peroxidase-conjugated anti-rabbit, anti-goat or anti-mouse IgG secondary antibodies (Jackson Immuno Research) were used at a dilution of 1:5000, 1:10,000 and 1:20,000, respectively, and anti-Rat biotinylated (1:100, Vector Labs) followed by peroxidase-conjugated streptavidin. Protein bands were visualized by chemoluminescence (ECL BioRad). Intensity of the bands was quantified with FIJI.

Neutralization of EphrinB3 epitopes in myelin extracts EphrinB3 epitopes in myelin extract proteins were neutralized by incubation with anti-EphrinB3 antibodies (AF395, R&D system and sc-271328, Santa Cruz Biotechnology, ratio: 1:1) for 2 h at room temperature prior to the addition to the cells [55].

Spinal cord live imaging LPC lesion followed by GFP⁺SC engraftment was performed in 60-day-old mice and terminally anesthetized for imaging 36 h later. Rhodamine-labeled BSL I (Vector Labs RL-1102) was injected at 2 mg/ml in the beating heart to label BV. After 5 min allowing dye circulation, spinal cords were dissected in ice-cold HBSS solution supplemented with 6.4 mg/mL D-(+)-glucose and bubbled for 30 min with bubbled with 95% O₂/5% CO₂. Spinal cord segments including lesion and graft sites were laid onto Millicell-CM slice culture inserts (Millipore) over culture medium (50% DMEM + Glutamax, 25% HBSS, 25% heat-inactivated horse serum, 5 mg/mL D-(+)-glucose, 20 mM Hepes, penicillin (100 mg/ml), streptomycin (100 U/ml), hrNDFß (125 ng/ml), and forskolin (2 µg/ml) in glass bottom plates, and then placed in an inverted Leica SP8X confocal microscope with an on-stage incubator, while streaming 95% O₂ and 5% CO₂ into the chamber. Spinal cords were imaged using a 25 × immersion objective at intervals of 15 min during 12 h with intermittent repositioning of the focal planes. Maximum intensity projections of the collected stacks (~ 60 μm at 2 μm step size) were compiled in FIJI.

Compelling evidence indicates that SC, whether recruited from the periphery or from resident progenitor cells, remyelinate CNS axons after demyelination [6, 43, 59, 61, 62], having an obvious impact on clinical recovery [24, 26]. However, in spite of SC presence within the CNS, the modalities of their recruitment after myelin injury remain elusive. To gain insights into this question, we studied SC behavior when confronted with CNS myelin and BV ex vivo and in vivo in demyelinating conditions. Using classic, live, and 3D imaging as well as electron microscopy, we provide solid evidences that SC use the vascular scaffold to migrate within the adult demyelinated CNS. This phenomenon is doubly modulated by CNS myelin, and in particular, the CNS myelin-specific component, EphrinB3, which negatively regulates SC adhesion to, and spreading on, myelin, while enhancing SC adhesion to perivascular ECM.

Communication and coordinated interaction between the vascular and nervous systems [29, 50] results in a functional neurovascular unit that contributes to wound healing [16], immune response [21] and embryonic development [40]. Recently, a new role in supporting long-distance migration of different kinds of cells within the nervous system was attributed to BV, both during development and adulthood [15, 58] as well as under pathological conditions [16, 28]. So far, despite the increasing number of cells guided by BV, the role of these structures in guiding SC within the CNS to participate in CNS repair has not been explored. While SC are known to promote endothelial cell migration and angiogenesis [31] and use BV as scaffold during peripheral nerve regeneration [16], our work demonstrates that this SC–BV interface extends to the CNS and is of relevance to their contribution to CNS repair.

We show that BV are a favorable substrate for exogenous SC and guide them towards demyelinating lesions. BV serve as scaffold for SC as soon as they leave the graft and along their path until arrival at the lesion. Along this path, SC are organized in chains going from one BV to another. At their arrival in the lesion, SC embedded in perivascular ECM become more randomly dispersed and this dispersal faithfully overlaps with BV expansion (Fig. 1c₂). While associated with BV at their arrival at the lesion at 3 dpi, SC detached from BV to contact and align with the demyelinated axons at 5 dpi. This change in substrate association may result from signals arising from the axons that trigger their differentiation into more mature SC as a first step to myelin repair. The absence of SC away from their narrow path of migration between the graft and lesion, as well as their progressive increase in number in the lesion, point to their specific recruitment by the lesion most likely involving attractant signals yet to be defined.

SC intervening in CNS repair are generated mainly from endogenous OPCs nested within perivascular niches while only a small proportion originate from PNS sources [6, 61]. Our model of grafting SC remotely (4 mm) from the lesion shows that once they are in the CNS, SC can travel long distances on BV to reach the lesion. Based on the cited work, long-distance migration along BV may then reflect the behavior of the majority of SC participating to endogenous remyelination. Moreover, the induction of demyelination in Krox20^Cre,Rosa^YFP mice showed that YFP⁺ SC were associated with BV in vicinity of the lesion, therefore, implying that YFP⁺ SC, irrespective of their PNS or CNS origin, use BV as scaffold to reach the lesion during spontaneous repair. Due to technical limitations and the difficulty to track a minor event, our work did not reveal the modalities of SC transgression between the PNS and CNS. However, the observation of SC migration over long distances on BV when ectopically placed in the CNS, and the presence of a vascular network forming bridges between the PNS and CNS, hints towards the possibility that PNS-derived SC can use this scaffold to transgress the PNS–CNS border. Future studies based on more specific transgenic tools and other lesion models are needed to resolve this question.

BV also support SC migration in the injured PNS. However, in these circumstances, SC create direct contacts via protrusion with endothelial cells to migrate from one nerve stump to the other or in vitro when grown in 3D, suggesting that these protrusions constitute mechanical means to propel SC migration along BV in a tight environment [16]. Although we found SC associated with BV along their migrating route, direct contact with endothelial cells or pericytes was never observed. Instead, SC were heavily embedded in the perivascular ECM, thus indicating that although SC share similar mechanisms to conquer the injured nervous system, some differences exist in their mode of migration between CNS and PNS, and BV guidance and perivascular ECM seems to prevail for their migration in the CNS. The observed differences may result from different molecular and cellular environment existing between PNS and CNS, including different degrees of confinement in which SC are placed.

We previously demonstrated that SC avoid and are repelled by myelin [10, 17]. Here, we confirm these data and show that SC, prevented from migrating directly through white matter, are somehow forced to migrate on BV. We showed previously that MAG, a CNS myelin component that prevents axonal regeneration in the CNS, is also inhibitory to SC migration and survival [17]. While MAG accounted only partially for the repulsive effect of myelin to SC, we identified EphrinB3 as another myelin component negatively regulating SC in contact with CNS myelin. Like MAG, EphrinB3/EphA4 receptor signaling has been implicated in axon pathfinding [49]. This suggests that myelin components exerting their inhibitory effect on axons are not exclusively directed against axons, but extend their inhibition to other neural components such as myelin-competent cells, preventing differentiation of oligodendrocyte progenitors into mature oligodendrocytes [36] as well as SC survival and migration (present data).

We first proved in vitro that SC-bound EphrinB3 is able to activate both EphA4 and EphB1 by phosphorylation, as well as to bind EphB6, which can be also trans-phosphorylated EphB1 [27]. Once bound to these receptors, EphrinB3 impairs the adhesion of these cells to myelin proteins, diminishing their process extension onto their substrate. Our in vitro data indicate that interfering with both EphB6 and EphA4 does not show additional improvement of this myelin–SC repulsion compared to single interference. Eph receptor signaling is not as straightforward as one receptor binding to one ligand. To trigger efficient activation, Eph receptors must cluster, homotypically (one subtype of Eph receptor) or heterotypically (involving the oligomerization of different Eph receptor subtypes) [56]. This lack of additive effect suggests that the mechanism of Eph/ephrin activation in SC might be mediated by heterotypic recruitment independent of the initial receptor activation.

Ephrin signaling not only induces repulsion but also modulates expression of adhesion molecules [5, 20, 51]. We show that myelin-associated EphrinB3 modulates SC adhesion and migration to ECM, in particular FN, and enhanced integrinβ1 expression, overruling SC inhibition by myelin in vitro and promoting their migration along BV to reach the demyelinated lesions in vivo. This event occurs in correlation with increased FN expression, among other ECM molecules, during BV remodeling in response to demyelination (data not shown Garcia-Diaz, unpublished data) [59], suggesting that the increased expression of ECM molecules by BV favors SC–BV interaction and subsequent migration along the vasculature in vivo.

Despite the present implication of EphA4 and EphB6 receptors in SC response to myelin, and their contribution to SC migration along CNS BV, the involvement of other Eph receptors in SC migration within the CNS should not be disregarded. Of note, EphB2, also expressed by SC and able to bind myelin-associated EphrinB3, mediates SC–SC interaction through N-cadherin re-localization to organize SC chain migration in the PNS [51]. This may be implicated in the SC chains forming bridges between BV (Fig. 1c₁ and f, Movie S1).

In addition, EphA4 is involved in SC–astrocyte repulsion [1]. Although the interactions between SC and astrocytes have not been explored in this study, the localization of the grafted SC in the vascular unit between the blood vessel wall and the astrocyte feet suggests that the repulsion exerted by astrocytes can help confine SC to the perivascular space, and thereby contribute to the mechanism of SC migration along BV. Interfering with EphA4 in SC could have altered this interaction and further allowed the grafted cells to escape from their perivascular path and intermingle more with the surrounding white matter parenchyma.

In conclusion, we used multiple in vivo and in vitro approaches to highlight a novel mechanism of guidance and migration of SC during the early events of CNS repair. We also provide strong evidences that the Eph/ephrin family regulates the complex interactions existing between SC, myelin and blood vessels. SC encountering myelin-associated EphrinB3 retract their processes failing to intermingle with white matter, and adhere preferentially to BV via activation of Integrinβ1. This dual effect, repulsing SC from CNS myelin and enhancing their attraction to basal lamina, directs their migration along CNS vasculature towards the lesion (Fig. 8). Lesions of white matter undergoing the formation and/or reshaping of the vasculature with increased expression of ECM adhesion molecules, in particular FN, further triggers SC mobilization throughout the lesion. While SC invasion of the CNS is not restricted to demyelinating diseases, future studies should indicate whether such mechanisms are of relevance for other clinical pathologies such as trauma as well as genetic and acquired myelinopathies.