415RESEARCH ARTICLE

INTRODUCTIONTo form a myelin sheath, an oligodendrocyte initially extendsmultiple branching processes that survey the local environment forsuitable axons (Hardy and Friedrich, 1996; Kirby et al., 2006). Uponcontact with a target axon, oligodendrocyte processes coalesce toform a spreading sheet of membrane that begins to wrap the axon.The extracellular cues that govern this process and the intracellularmechanisms involved are poorly understood.

Laminin 2, which promotes oligodendrocyte maturation bysignaling through α6β1 integrins (Baron et al., 2005), has beensuggested to regulate process elaboration. Laminin-2-deficient micehave dysmyelinated and hypomyelinated axons (Chun et al., 2003).However, laminin 2 is not ubiquitous in myelinated CNS axon tracts(Colognato et al., 2002) and transgenic mice lacking β1 integrinexpression in oligodendrocytes exhibit no defects in CNSmyelination (Benninger et al., 2006). Thus, other ligand-receptorinteractions must also direct the changes in oligodendrocytemorphology required for myelination.

The axon-guidance cue netrin 1 (Ntn1) is a chemorepellent formigrating oligodendrocyte precursors (OPCs) in the embryonicspinal cord (Jarjour et al., 2003; Tsai et al., 2003). These cellsexpress the netrin 1 receptors Dcc, Unc5a and Unc5b, but not netrin1 itself. In the adult CNS, netrin 1 is expressed by myelinatingoligodendrocytes and is associated with non-compactedoligodendroglial membranes (Manitt et al., 2001). We therefore

investigated the possibility that netrin 1, expressed in the developingCNS and later by the oligodendrocytes themselves, might influencelate stages of oligodendrocyte differentiation.

The lamella elaborated by the tip of an extending oligodendrocyteprocess has been compared to a neuronal growth cone (Fox et al.,2006; Jarjour and Kennedy, 2004; Sloane and Vartanian, 2007).Interestingly, a number of the same intracellular signaling proteinshave been implicated in netrin-1-mediated axon guidance and thedevelopment of oligodendrocyte processes required for myelination(Fox et al., 2006; Jarjour and Kennedy, 2004). In both cases, thereorganization of the actin cytoskeleton requires activation of the Srcfamily kinase (SFK) Fyn (Meriane et al., 2004; Osterhout et al., 1999;Umemori et al., 1994) and involves Wiscott-Aldrich syndrome protein(N-WASP; Wasl – Mouse Genome Informatics) and Rho GTPases(Bacon et al., 2007; Liang et al., 2004; Shekarabi et al., 2005).

Here we provide evidence that netrin 1 and Dcc promote theextension of oligodendrocyte processes in vivo. We used in vitroassays to demonstrate that netrin 1 increases oligodendrocyteprocess branching. Furthermore, expression of Dcc and netrin 1 byoligodendrocytes promotes the formation of myelin-like membranesheets. Addressing the signaling mechanisms involved, we showthat the SFK Fyn is required for netrin-1-induced process branching,that netrin 1 recruits Fyn to Dcc in oligodendrocytes, increases SFKactivity, and promotes process extension and branching associatedwith a decrease in RhoA activity. Our findings reveal a novel rolefor netrin 1 and Dcc in activating a signaling cascade inoligodendrocytes that directs process remodeling.

MATERIALS AND METHODSAnimals and oligodendrocyte culturesSprague Dawley rat pups and pregnant BALB/c mice were obtained fromCharles River Canada (Montreal, Quebec, Canada). Ntn1–/+ and Dcc–/+ micewere obtained from Marc Tessier-Lavigne (Genentech, South San Francisco,

Netrin 1 and Dcc regulate oligodendrocyte processbranching and membrane extension via Fyn and RhoASathyanath Rajasekharan, K. Adam Baker, Katherine E. Horn, Andrew A. Jarjour, Jack P. Anteland Timothy E. Kennedy*

The molecular mechanisms underlying the elaboration of branched processes during the later stages of oligodendrocyte maturationare not well understood. Here we describe a novel role for the chemotropic guidance cue netrin 1 and its receptor deleted incolorectal carcinoma (Dcc) in the remodeling of oligodendrocyte processes. Postmigratory, premyelinating oligodendrocytes expressDcc but not netrin 1, whereas mature myelinating oligodendrocytes express both. We demonstrate that netrin 1 promotes processextension by premyelinating oligodendrocytes in vitro and in vivo. Addition of netrin 1 to mature oligodendrocytes in vitro evoked aDcc-dependent increase in process branching. Furthermore, expression of netrin 1 and Dcc by mature oligodendrocytes was requiredfor the elaboration of myelin-like membrane sheets. Maturation of oligodendrocyte processes requires intracellular signalingmechanisms involving Fyn, focal adhesion kinase (FAK), neuronal Wiscott-Aldrich syndrome protein (N-WASP) and RhoA; however,the extracellular cues upstream of these proteins in oligodendrocytes are poorly defined. We identify a requirement for Src familykinase activity downstream of netrin-1-dependent process extension and branching. Using oligodendrocytes derived from Fynknockout mice, we demonstrate that Fyn is essential for netrin-1-induced increases in process branching. Netrin 1 binding to Dcc onmature oligodendrocytes recruits Fyn to a complex with the Dcc intracellular domain that includes FAK and N-WASP, resulting in theinhibition of RhoA and inducing process remodeling. These findings support a novel role for netrin 1 in promoting oligodendrocyteprocess branching and myelin-like membrane sheet formation. These essential steps in oligodendroglial maturation facilitate thedetection of target axons, a key step towards myelination.

KEY WORDS: Oligodendroglia, Myelination, Myelin, Netrin, Integrin, Laminin, Autocrine, Mouse, Rat, FAK (Ptk2), N-WASP (Wasl)

Development 136, 415-426 (2009) doi:10.1242/dev.018234

Montreal Neurological Institute, Department of Neurology and Neurosurgery, McGillUniversity, Montreal, Quebec H3A 2B4, Canada.

*Author for correspondence (e-mail: timothy.kennedy@mcgill.ca)

Accepted 21 November 2008 DEVELO

PMENT

416

CA, USA) and Robert Weinberg (Harvard University, Cambridge, MA,USA), respectively. Fyn KO (Fyntm1Sor) and control F2 hybrid (B6129SF2/J)breeding pairs were obtained from Jackson Laboratories (Bar Harbor, ME,USA). All procedures were performed in accordance with the CanadianCouncil on Animal Care guidelines for the use of animals in research.

Cell cultureOPCs were derived from mixed glial cultures from the cerebral cortices ofpostnatal day 0 (P0) rat pups and grown in oligodendrocyte defined medium(OLDEM) as described previously (Armstrong, 1998; Jarjour et al., 2003),with 0.1% fetal bovine serum (FBS) to initiate differentiation. Cells wereseeded at 1.5�104 cells/chamber in 8-well chamber slides coated with10 μg/ml poly-L-lysine (Nalge Nunc, Rochester, NY, USA). Forimmunoprecipitation, western blots and GST-rhotekin pulldowns, cells wereplated at 1.5�106 cells/well in 6-well tissue culture dishes.

Mouse oligodendrocyte culturesCulturing of mouse OPCs was similar to that of rat OPCs, but 10% horseserum was used instead of 10% FBS in mixed glial culture media. Each T75flask of mixed glial culture required two to three pups. Newborn Ntn1–/– andDcc–/– mice were identified by distinct behaviors. The genotype ofindividual pups was confirmed by PCR.

AntibodiesPrimary antibodies used in this study were: rabbit polyclonal anti-netrin 1PN3 (Manitt et al., 2001), mouse monoclonal anti-Dcc (G97-449; BDBiosciences Pharmingen, San Jose, CA, USA), goat polyclonal anti-Dcc(Santa Cruz Biotech, Santa Cruz, CA, USA), rabbit polyclonal anti-myelinbasic protein (Mbp, Chemicon, Temecula, CA, USA), mouse monoclonalanti-Mbp (Chemicon), mouse monoclonal RIP antibody (Chemicon), mousemonoclonal anti-2�,3�-cyclic nucleotide 3� phosphodiesterase (Cnp,Sternberger Monoclonals, Lutherville, MD, USA), rabbit polyclonal anti-Mag (Chemicon), rabbit polyclonal anti-Nfm (Nefm – Mouse GenomeInformatics) (Chemicon), rabbit polyclonal anti-Fyn (Upstate CellSignaling, Charlottesville, VA, USA) used for western blots, rabbitpolyclonal anti-Fyn [gift of Dr Andre Veillete and described previously(Davidson et al., 1992)] used for immunoprecipitation and western blots,mouse monoclonal anti-FAK (BD Biosciences), rabbit polyclonal anti-N-WASP (Santa Cruz), and rabbit polyclonal anti-phospho-Src (CellSignaling), which recognizes the pY416 epitope in all SFK members.

For analyses of oligodendrocyte morphology in vivo, we crossed Ntn1 orDcc heterozygous mice, obtained embryonic day 18 (E18) embryos (plugdate taken as E1), fixed them in 4% paraformaldehyde and cut 18 μmsections of the spinal brachial enlargement with a cryostat. Sections werethen stained with anti-Cnp (1:250), anti-Nfm (1:250), anti-netrin PN3(1:100), visualized using Alexa 546- or Alexa 488-conjugated secondaryantibodies (Jackson ImmunoResearch, West Grove, PA, USA), and nucleistained with Hoechst. Images were obtained with a Magnafire CCD camera(Optronics, Goleta, CA, USA) and a Zeiss Axiovert 100 microscope(Toronto, Ontario, Canada).

Analysis of oligodendrocyte morphologyAnalyses of immature oligodendrocytes in vitro and in vivo were performedusing the NeuronJ plugin for ImageJ (NIH, Bethesda, MD, USA). Thelength of the longest process was measured from the base of the process toits tip (Fig. 3D; Fig. 4B). For Sholl analysis, the grid function in NorthernEclipse (Empix) was used to draw concentric circles 15 μm apart around thecell body of mature RIP-positive oligodendrocytes. The number ofintersections made by processes with each successive circle was counted.For studies using the β1 integrin subunit function-blocking antibody andmouse oligodendrocyte cultures, a Sholl analysis plugin was used withImageJ (starting radius, 1.02 cm; step size, 1.02 cm; end, 5.08 cm; thickness,0.02 cm). The Mbp-positive myelin-like membrane sheets were outlined inImageJ and the surface area reported in arbitrary units.

The pharmacological inhibitors PP2 and PP3 (Calbiochem) were used at2 μM to inhibit SFK activity. Purified hamster anti-rat CD29 (β1 integrin)monoclonal and purified hamster anti-IgM monoclonal antibodies were usedat 2 μM to investigate β1 integrin function.

Analysis of phospho-Src punctaThe number of phospho-Src-positive puncta was measured using ImageJ.The brightness and contrast of the images of individual oligodendrocyteswere modified to enhance puncta associated with extending processes;modifications made were consistent across all images. Images wereconverted to a binary format and then the number of particles countedautomatically. Staining in the cell body and major processes was notpunctate and thus counted as one particle, making variations under differentconditions a direct indicator of changes in the number of distally locatedpuncta. The number of puncta was divided by oligodendrocyte area tocontrol for variance in oligodendrocyte size.

Quantification and statistical analysesStatistical significance was calculated by ANOVA followed by a post-hocTukey test using Systat Software (San Jose, CA, USA). Analyses ofoligodendrocyte morphology in vitro used three independent experimentswith a minimum of 30 cells per condition. In vivo analysis of processextension and purification of oligodendrocytes from transgenic mice wereperformed using at least four pups of each genotype, derived from at leasttwo different litters.

ImmunoprecipitationCells in 6-well dishes were treated with netrin 1 for 5 minutes, then lysed inRIPA lysis buffer (10 mM sodium phosphate pH 7.2, 150 mM NaCl, 1%NP40, 0.1% SDS, 0.5% deoxycholate) and centrifuged at 13,000 rpm(13,800 g) for 7 minutes. Supernatant was pre-cleared with 30 μl proteinA/G beads (Santa Cruz) for 30 minutes, incubated with 1 μg/ml anti-Dcc(monoclonal) or anti-Fyn (rabbit polyclonal) for 1 hour, followed by additionof 30 μl protein A/G beads for 45 minutes.

GST pulldown assaysFusion proteins comprising the RhoA-binding domain of rhotekin (adownstream substrate of RhoA) or Pak-CRIB and glutathione-S-transferase(GST) were purified as described (Reid et al., 1996). Oligodendrocytes weretreated with netrin 1 for 24 hours. Cells were then lysed and protein purifiedas described (Ren and Schwartz, 2000; Shekarabi et al., 2005).

RESULTSOligodendrocytes express netrin 1 duringmyelination in the developing spinal cordNetrin 1, expressed by floor plate and neuroepithelial cells in theearly embryonic spinal cord, directs migrating OPCs away from theventral midline towards axons in the nascent white matter (Jarjouret al., 2003; Kennedy et al., 1994; Tsai et al., 2006; Tsai et al., 2003).Although not expressed by OPCs, netrin 1 is widely expressed byspinal interneurons, motoneurons and by most, if not all, maturemyelinating oligodendrocytes in dorsal and ventral white matter inthe adult rat and mouse CNS (Manitt et al., 2001).

To determine when netrin 1 is expressed by differentiatingoligodendrocytes, we performed a time-course analysis.Premyelinating, postmigratory oligodendrocytes within thecorticospinal tract at brachial, thoracic and lumbar levels wereexamined. Netrin 1 expression was not detected in oligodendrocytesat embryonic stages of development in mice (data not shown). Forpostnatal stages, netrin-1-expressing cells were identifiedimmunohistochemically. Oligodendrocytes in the developingdorsolateral white matter tracts were identified by expression of2�,3�-cyclic nucleotide 3� phosphodiesterase (Cnp), a marker thatlabels oligodendrocyte cell bodies and processes in vivo, and byexpression of the mature oligodendrocyte marker myelin basicprotein (Mbp). At P8, before myelination begins in the corticospinaltract, netrin 1 was not expressed by premyelinating oligodendrocytes(Fig. 1A-C). Netrin-1-expressing neuronal cell bodies and neural-epithelial cells (Fig. 1D,E), but not astrocytes (Fig. 1I), weredetected immediately adjacent to differentiating oligodendrocytes.

RESEARCH ARTICLE Development 136 (3)

DEVELO

PMENT

At P12, following the initiation of myelination in the ratcorticospinal tract (Schwab and Schnell, 1989), netrin 1 wasdetected in the cell bodies of a subset of oligodendrocytes doublelabeled with Cnp (Fig. 1F,G). At this stage of development, netrin 1immunoreactivity was associated with axons (Fig. 1I). By P22, largenumbers of mature myelinating oligodendrocytes expressing netrin1 were readily detectable throughout the nascent white matter of thespinal cord (Fig. 1J,K). These findings indicate that netrin 1 beginsto be expressed by oligodendrocytes during early stages ofmyelination.

Oligodendroglial expression of netrin 1 and Dccdoes not require neuronal contactThe timing of netrin 1 expression by oligodendrocytes suggestedthat expression might be regulated by axonal signals. To test this,purified rat oligodendrocyte progenitors were cultured in theabsence of neurons in conditions that promote differentiation.Immature oligodendrocytes in these cultures [4 days in vitro(DIV)] were defined as multipolar cells expressing Cnp and arethe equivalent of the premyelinating cells in vivo. These cells didnot express netrin 1. More mature oligodendrocytes, cultured for

6-8 DIV and identified by expression of Mbp and Cnp (Watanabeet al., 2006), extend highly branched processes that coalesce toform cytoplasmic sheets (Fox et al., 2006). These sheets, whichresemble unwrapped non-compacted myelin membrane (Knappet al., 1987), were immunopositive for netrin 1 (Fig. 2A). Thisstaining was present on cells that were not permeabilized,consistent with netrin 1 protein associated with the extracellularface of the plasma membrane. To verify that the cells were notpermeabilized, the absence of Mbp staining was used as anegative control (not shown). We conclude that in the absence ofneuronal contact in vitro, oligodendrocytes express netrin 1, andthat netrin 1 protein is associated with the surface of myelin-likemembrane sheets.

The netrin 1 receptor Dcc was detected along the processes ofimmature and mature oligodendrocytes. In mature oligodendrocytes,Dcc was associated with major branches and present in small punctaat the leading edge of myelin-like membrane sheets (Fig. 2B).Addition of recombinant Myc-tagged netrin 1 to matureoligodendrocytes revealed a preferential localization of exogenousligand at the branches and edges of sheets formed by the cells,similar to the distribution of Dcc (Fig. 2A�,B�).

417RESEARCH ARTICLERegulation of oligodendrocyte maturation

Fig. 1. Expression of netrin 1 by oligodendrocytesin vivo. (A,A�) Longitudinal section through a P8 ratspinal cord showing Mbp-immunopositiveoligodendrocytes (red) that do not express netrin 1,intermingled with netrin-1-expressing cells (green).(B,B�) At P8, netrin 1 immunoreactivity (green) wasdetected in the immediate environment surroundingthe cell bodies of Cnp-positive oligodendrocytes (red),but netrin 1 was not expressed by theoligodendrocytes themselves. (C,C�) An Mbp-positivenetrin-1-negative process, surrounded by netrin-1-expressing neuroepithelial cells (arrows) in P8 spinalcord. (D,D�) Longitudinal section through a P8 ratspinal cord showing axons positive for neurofilamentmedium polypeptide (Nfm), surrounded by netrin-1-expressing cells. (E,E�) Cell bodies of netrin-1-immunopositive neurons. (F) Cross-section of a P12 ratdorsal spinal cord showing Cnp-positive, myelinatingoligodendrocytes (red) expressing netrin 1 (green).(G-G�) Oligodendrocytes adjacent to the central canal(arrow) express netrin 1 in P12 spinal cord.(H,H�) Netrin 1 immunoreactivity (green) associatedwith a Cnp-immunopositive (red) myelinatingoligodendrocyte in the P12 rat spinal cord. (I-I�) By P12,netrin 1 expression (green) colocalizes with Nfm-immunoreactive axons (blue), but is not detected inGfap-positive astrocytes (red). (J,J�) Transverse sectionof a P22 thoracic spinal cord showing widespreadexpression of netrin 1 (green) by Cnp-positive,myelinating oligodendrocytes (red).(K-K�) Magnification of the boxed region from Jshowing Cnp-positive netrin-1-expressing cell bodies.A,F, 20�0.5 n.a. objective; G,J, 40�0.75 n.a.objective; D,E,I,K, confocal microscopy, 40�0.75 n.a.objective; B,C,H, confocal microscopy, 100�1.4 n.a.objective. Scale bars: 10μm in B�,C�,H; 20μm inA�,D,E,G,K; 40μm in F,I�,J.

DEVELO

PMENT

418

Netrin 1 and Dcc promote process extension bypremyelinating oligodendrocytes in vivoIn the developing spinal cord, as premyelinating oligodendrocytesmature into myelinating oligodendrocytes, netrin 1 is widelyexpressed by neurons and neuroepithelial cells, but not byoligodendrocytes (Fig. 1D,I; Fig. 3C). Mice lacking either netrin 1or Dcc die within hours of birth (Fazeli et al., 1997; Serafini et al.,1996). We therefore used E18 littermates to compare themorphology of postmigratory, premyelinating oligodendrocytes intissue sections from the spinal cords of wild-type and Ntn1 and Dccheterozygotes and knockout mouse embryos (Fig. 3A-C). At E18,Cnp-immunoreactive premyelinating oligodendrocytes in thedorsolateral spinal cord typically extend one or more processes (Fig.3D, dashed lines). We found the length of these processes in bothNtn1–/– and Dcc –/– mice to be significantly shorter than in wild-typelittermates (Fig. 3E,F). These findings provide evidence that at theend of precursor migration and at the initiation of oligodendrocytedifferentiation, netrin 1 in the local environment of a postmigratory,premyelinating oligodendrocyte promotes Dcc-dependent processextension in vivo, a phenomenon closely associated with thecapacity of an oligodendrocyte to contact target axons (Hardy andFriedrich, 1996; Kirby et al., 2006).

Interpretation of these findings could be confounded by twofactors. First, a subset of neurons in the spinal cords of Ntn1- andDcc-null mice exhibits defects in axon guidance. However, this isunlikely to exert a profound influence on the differentiation ofindividual oligodendrocytes as the majority of axons extendnormally in the absence of netrin 1 function and the nascent whitematter is well populated with axons. Second, a developmental delayin OPC dispersal in the Ntn1 and Dcc mutants might delay processextension. In order to directly address the mechanisms underlying

the apparent influence of netrin 1 on oligodendrocyte differentiationin vivo, we determined whether the putative aberrant processextension detected in Ntn1–/– and Dcc–/– mice could be replicated invitro in the absence of axonal influences or migration deficits.

Netrin 1 does not affect the differentiation ofimmature or mature oligodendrocytes in vitroDistinct changes in oligodendrocyte morphology accompanythe differentiation of oligodendrocytes in vitro. Immatureoligodendrocytes are characteristically multipolar cells that expressCnp but not Mbp (Fig. 4A,B). These cells differentiate into matureoligodendrocytes that elaborate myelin-like membrane sheets andexpress both Mbp and Cnp (Fig. 4A,D), and eventually myelin-associated glycoprotein (Mag). Immature cells (4-5 DIV) grown inthe presence of netrin 1 (100 ng/ml) for 24 hours showed a modest(4%) decrease in the ratio of Cnp- to Mbp-positive cells. In more-mature Mbp/Mag-positive cultures (6-8 DIV), the ratio of Mbp-positive cells to Mag-positive cells was not affected by the additionof netrin 1 (Fig. 4A). We conclude that the addition of netrin 1 invitro does not alter the acquisition of a mature phenotype, asassessed by the expression of standard markers.

Netrin 1 induces Dcc-dependent process extensionby immature oligodendrocytes in vitroThe effect of netrin 1 on process extension, as assayed by measuringthe length of the longest process, was investigated in Cnp-immunopositive immature oligodendrocytes in culture (Fig. 4B).Application of 100 ng/ml netrin 1 for 24 hours to immatureoligodendrocytes increased process length compared with controlcells (Fig. 4C). To determine whether Dcc is required for netrin-1-induced extension of oligodendrocyte processes in vitro, a Dccfunction-blocking antibody (Dccfb) was added for 24 hours toimmature oligodendrocytes treated with netrin 1. Consistent with thefindings obtained in vivo and described above (Fig. 3F), disruptionof Dcc function blocked the netrin-1-induced increase in processlength, but did not significantly alter process extension when addedalone (Fig. 4C). We conclude that the application of netrin 1promotes Dcc-dependent oligodendrocyte process extension, andthat these changes occur independently of defects in migration andaxon growth.

Netrin 1 promotes Dcc-dependent increases inoligodendrocyte branching and myelin-like sheetformationTo investigate roles for netrin 1 during later stages ofoligodendrocyte development, we characterized changes inoligodendrocyte process branching and in the capacity of these cellsto elaborate myelin-like membrane sheets in vitro. Matureoligodendrocytes cultured for 6-8 DIV were double labeled with RIPantibody (against Cnp) to identify major processes, and Mbpantibody to visualize extended myelin-like membrane sheets (Fig.4D, left). Oligodendrocytes grown in the presence of 100 ng/mlnetrin 1 for 24 hours exhibited a significant increase in the area ofMbp-positive sheets (Fig. 4E). The morphological complexity ofoligodendrocyte processes was quantified using Sholl analysis(Ricard et al., 2001) (Fig. 4D). Cells treated with 100 ng/ml netrin 1for 24 hours exhibited a significant increase in branching comparedwith the control (Fig. 4F,G). A dose-response analysis determined100 ng/ml netrin 1 to be optimal (Fig. 4F).

As described above, immunocytochemical analyses detected Dcc,but little, if any, netrin 1 associated with the branches ofoligodendrocyte processes. Both Nfb (netrin function-blocking

RESEARCH ARTICLE Development 136 (3)

Fig. 2. Expression of netrin 1 and Dcc by oligodendrocytes invitro. (A,A�) Netrin 1 protein (green) on the surface of myelin-likemembrane sheets of non-permeabilized mature rat oligodendrocytes inculture. Recombinant netrin 1, labeled using a Myc epitope tag,preferentially localizes to branches formed by the cells (arrow).(B,B�) Dcc (green) distributed along oligodendrocyte processes and atpuncta along the edge of Mbp-immunopositive (red) myelin-likemembrane sheets (arrow). A,B, 40�0.5 n.a. objective. Scale bars:40μm in A�; 20μm in B�.

DEVELO

PMENT

antibody) and Dccfb antibodies blocked the increase in myelin-likemembrane sheet formation and process branching induced by theaddition of exogenous netrin 1 (Fig. 4E,G), indicating that the netrin-1-induced changes in oligodendrocyte morphology are Dcc-dependent. Substantial netrin 1, but not Dcc, immunoreactivity wasdetected in association with the myelin-like membrane sheets(Fig. 2A); however, function-blocking antibodies applied in theabsence of netrin 1 did not alter sheet formation or processbranching (Fig. 4E,G). We therefore tested the hypotheses thatnetrin 1 made by oligodendrocytes does not exert an autocrineinfluence on the formation of myelin-like membrane sheets, and,alternatively, that the relatively short-term (24 hours) loss-of-function assays described above are too brief to reveal a role forendogenous netrin 1.

Autocrine netrin 1 regulates the formation ofmyelin-like membrane sheets, but not processbranching, in a Dcc-dependent mannerTo investigate an autocrine role for netrin 1, oligodendrocytes wereisolated from mixed glial cultures derived from P0 Ntn1–/– or Dcc–/–

mice (Fig. 5A). Application of netrin 1 for 24 hours to matureoligodendrocytes lacking Dcc did not increase process branching(Fig. 5B), indicating that oligodendrocyte processes require Dcc torespond to exogenous netrin 1. Furthermore, in agreement with thefunction-blocking antibody data described above (Fig. 4G), baselinelevels of process branching were not altered in oligodendrocytes

lacking either Dcc or netrin 1 (Fig. 5B,C). We conclude that netrin1 does not exert an autocrine effect on branching, and hypothesizethat this is a consequence of netrin 1 being sequestered to themyelin-like membrane sheets (Fig. 2A), thereby exposing theleading edges of processes to little, if any, endogenous netrin 1.

By contrast, oligodendrocytes lacking either netrin 1 or Dccexhibited a significantly reduced surface area of the myelin-likemembrane sheets compared with cells derived from wild-type orheterozygote littermates (Fig. 5D,E), thereby identifying anautocrine role for netrin 1 in sheet formation. This is consistent withour demonstration that the addition of exogenous netrin 1 increasesmyelin-like membrane sheet formation through a Dcc-dependentmechanism (Fig. 4E). These findings reveal differences between theresponse of oligodendrocytes to netrin 1 made by the cellsthemselves, which regulates myelin-like membrane sheet formation,and netrin 1 encountered in the local environment, which primarilyinfluences process branching.

Process elaboration induced by netrin 1 does notrequire β1 subunit-containing integrinsOligodendrocytes express αvβ1, αvβ3, αvβ5 and α6β1 integrins.Engagement of integrins, specifically α6β1, by extracellular matrix(ECM) components activates SFKs to regulate changes inoligodendrocyte morphology (Baron et al., 2005). Netrins aremembers of the laminin family of ECM proteins, and netrin 1 hasbeen proposed to function as a ligand for α3β1 and α6β4 integrins

419RESEARCH ARTICLERegulation of oligodendrocyte maturation

Fig. 3. Netrin 1 and Dcc regulate process outgrowth by immature oligodendrocytes in vivo. (A-C) Transverse sections of brachial spinal cordfrom E18 Ntn1–/– mice stained with antibodies to Cnp (green) and Nfm (red) to label premyelinating oligodendrocytes and neuronal cell bodies andaxons, respectively. (A) Arrows indicate Cnp-immunopositive premyelinating oligodendrocytes. (B) As premyelinating oligodendrocytes search forand initially contact axons, they extend one or two major processes (arrows). (C) Netrin 1 expression, determined by β-galactosidase (β-Gal) reporterexpression in a transgenic Ntn1+/– mouse, shown at the E18 spinal cord floor plate (arrow). Premyelinating oligodendrocytes are notimmunopositive for β-galactosidase, but are adjacent to cells expressing netrin 1. (D) Dashed lines illustrate the measurement of extendingprocesses of premyelinating oligodendrocytes in Ntn1+/+ and Ntn1–/– embryos. (E,F) Oligodendrocytes in the spinal cords of Ntn1–/– or Dcc–/– miceexhibit significantly shorter processes than wild-type or heterozygous littermates. *P<0.05, ***P<0.0001 versus control. The number of cellsanalyzed in each condition is indicated in parentheses. Error bars indicate s.e.m. A, 20�0.5 n.a. objective; B-D, 40�0.75 n.a. objective. Scale bars:40μm in A; 20μm in B-D.

DEVELO

PMENT

420

in pancreatic cells (Yebra et al., 2003). We therefore investigatedwhether netrin-1-mediated changes in the elaboration ofoligodendrocyte processes involve an interaction between netrin 1and integrins.

Application of a β1 integrin function-blocking antibody,which has been demonstrated to block fibronectin-inducedchanges in oligodendrocyte morphology (Liang et al., 2004),did not significantly alter the netrin-1-induced increase in

RESEARCH ARTICLE Development 136 (3)

Fig. 4. Netrin 1 induces Dcc-dependent process extension by immature oligodendrocytes, and Dcc-dependent process branching andmyelin-like membrane sheet extension by mature oligodendrocytes, in vitro. (A) Addition of netrin 1 (100 ng/ml, 24 hours) to cultured ratoligodendrocytes results in a 4% decrease in the ratio of immature cells expressing Cnp only (arrowheads) as compared with more mature cellsexpressing Cnp and Mbp. Netrin 1 did not change the ratio of cells expressing Mbp only versus Mbp and Mag. (B) Cnp-positive immatureoligodendrocytes are multipolar cells with one major process. Dashed red lines indicate examples of the processes measured. (C) The length of themajor process increased following the addition of netrin 1 (100 ng/ml) for 24 hours. Application of Dcc function-blocking antibody (Dccfb) toimmature oligodendrocytes blocks the netrin-1-induced increase in process extension, whereas Dccfb alone does not have a significant effect onprocess extension. (D) A mature Mbp- and RIP-positive oligodendrocyte extending both myelin-like membrane sheets and branched processes.Sheet area was quantified by tracing the Mbp-positive myelin-like membrane sheets (left, red outlines). Branching was quantified by measuring thenumber of intersections that processes made with concentric circles (right, red), which are numbered 1-5 to reflect increasing distance from the cellbody (and as labeled on the x-axis of bar charts displaying branching complexity). (E) Netrin 1 (100 ng/ml, 24 hours) increased the area of Mbp-positive sheets compared with the control. This effect was blocked by Dccfb. (F) Processes of mature oligodendrocytes exposed to netrin 1 for 24hours exhibited an increase in branching. Dose-response analysis indicated maximal branching at 100 ng/ml netrin 1. (G) Addition of Dcc function-blocking antibody together with netrin 1 prevented the netrin-1-dependent increase in branching, and decreased branching compared withcontrols. A, 20�0.5 n.a. objective; B,D,F, 40�0.75 n.a. objective. Scale bars: 40μm in A; 20μm in B,D,F. A.U., arbitrary units. *P<0.05, **P<0.005,versus control. Error bars indicate s.e.m. D

EVELO

PMENT

oligodendrocyte process branching (Fig. 6A). By contrast, thisantibody blocked HEK293T cell spreading on a fibronectinsubstrate (Fig. 6B), verifying its efficacy. Yebra and colleaguesidentified a 25 amino acid region within the C-terminus of netrin1 that binds α6β4 and α3β1 integrins, and hypothesized thatpotential interactions between netrin 1 and other integrins mightalso occur through this region (Yebra et al., 2003). To determinewhether such an interaction might contribute to netrin-1-inducedchanges in oligodendroglial morphology, we incubated cells witha peptide comprising the putative integrin-binding sequence thatfunctions as a competitive inhibitor of integrins binding netrin 1(Yebra et al., 2003). The netrin 1 peptide (20 μg/ml, a gift fromDr V. Cirulli, UCSD, CA, USA) did not affect process complexity(Fig. 6A), nor did it disrupt the netrin-1-dependent increase inbranching (Fig. 6A). We conclude that the novel role for netrin 1in regulating oligodendrocyte morphology occurs through Dccindependently of β1-containing integrins and the integrin-bindingregion at the C-terminus of netrin 1.

Netrin 1 binding to Dcc recruits Fyn to a complexcontaining FAKWe next assayed signaling proteins implicated as regulators ofcytoskeletal organization downstream of netrin 1 and Dcc in axonalgrowth cones to determine whether similar signaling complexesmight be engaged in oligodendrocytes. Oligodendrocytes expressthe SFKs Src, Fyn and Lyn (Colognato et al., 2004). Of these, onlymice lacking Fyn show defects in myelination (Sperber et al., 2001).Furthermore, Fyn activation has been implicated in the regulation ofprocess branching during the morphological maturation ofoligodendrocytes (Osterhout et al., 1999; Umemori et al., 1994). Inaxonal growth cones, application of netrin 1 recruits Fyn to the Dccintracellular domain, where it is activated by focal adhesion kinase(FAK; Ptk2 – Mouse Genome Informatics) (Liu et al., 2004). Todetermine whether Fyn might function downstream of Dcc inoligodendrocytes, cells derived from newborn (P0) rats wereallowed to mature in culture for 3-4 days or for 5-6 days until theyexpressed Mbp. Co-immunoprecipitation studies carried out using

421RESEARCH ARTICLERegulation of oligodendrocyte maturation

Fig. 5. Oligodendrocytes from mice lackingnetrin 1 or Dcc exhibit defects in processbranching and in myelin-like membrane sheetformation. (A) Representative examples of themorphology of oligodendrocytes derived from wild-type and netrin-1-deficient or Dcc-deficient mice.40�0.75 n.a. objective. Scale bar: 20μm.(B) Application of netrin 1 increased branching byoligodendrocytes from Dcc+/+ and Dcc+/–, but notfrom Dcc–/–, mice. (C) Oligodendrocytes from Dcc–/– orNtn1–/– mice exhibit no difference in branchingcomplexity compared with cells from wild-type andheterozygous littermates. (D) Ntn1–/– oligodendrocytesexhibit a significant decrease in myelin-like membranesheet area compared with cells derived from wild-typeand heterozygote littermates. (E) Dcc–/–

oligodendrocytes elaborate smaller myelin-likemembrane sheets than wild-type and heterozygouslittermates. A.U., arbitrary units. ***P<0.0001 versuscontrol. Error bars indicate s.e.m.

DEVELO

PMENT

422

two different antibodies against Fyn revealed an interaction betweenFyn and Dcc in both immature (4 DIV) and mature (6 DIV)oligodendrocytes (Fig. 7A). Five minutes following application ofnetrin 1, an increased amount of Fyn co-immunoprecipitated withDcc (Fig. 7A,D). A relatively minor Fyn-immunoreactive band ofslightly higher molecular weight was consistently detectedfollowing immunoprecipitation and might reflect Fynphosphorylation as a result of activation by netrin 1. Our findingsalso revealed a constitutive interaction between Dcc and FAK (Fig.7A), suggesting that application of netrin 1 to oligodendrocytesrecruits Fyn into a complex with FAK that is bound to theintracellular domain of Dcc, as has been reported for neurons (Li etal., 2004; Ren et al., 2004). Application of netrin 1 led to increasedphosphorylation of SFK tyrosine 416 (Y416), an event associatedwith kinase activation (Smart et al., 1981), in the SFK associatedwith Dcc (Fig. 7B,D). Analysis of SFK phosphorylation in whole-cell lysates did not detect a global change in phospho-Y416,indicating that the netrin-1-induced change is specific to SFKrecruited into a complex with Dcc (Fig. 7C). Using an antibodyspecific for the SFK Src, an interaction with Dcc was not detected(not shown). This provides evidence for specific recruitment of Fynto Dcc; however, we do not rule out that other SFKs might beinvolved in netrin 1 signaling in oligodendrocytes.

The netrin-1-induced increase in oligodendrocytebranching requires FynTo determine whether Fyn is required downstream of netrin 1 inoligodendrocytes, we isolated cells from mice lacking Fyn[Fyntm1Sor, Fyn knockout (KO)] and from wild types of a matchedgenetic background (B6129SF2/J, F2 hybrid) (Fig. 8A). Treatmentof Fyn KO cells with netrin 1 did not result in increased branching,in contrast to its effect upon cells isolated from control mice (Fig.8A,B). We conclude that Fyn is essential for these changes to occur.We therefore determined whether activation of SFKs is required topromote the morphological changes induced by netrin 1.

SFK activity is required for the netrin-1-inducedincrease in oligodendrocyte process length andbranchingImmunocytochemistry revealed SFK phospho-Y416immunoreactivity distributed within the oligodendrocyte cell bodyand proximal branches, and punctate staining within the distal

branches (Fig. 8C,D). Colocalization with Dcc was observed,consistent with our immunoprecipitation results (Fig. 7B). Todetermine whether changes in SFK phosphorylation occurred uponnetrin 1 stimulation, the relative number of SFK phospho-Y416-immunoreactive puncta was measured per unit area of the cell (Fig.8E). Treatment of mature oligodendrocytes with netrin 1 (100ng/ml) for 24 hours significantly increased the number of SFKphospho-Y416-positive puncta per unit area, consistent with anassociation between netrin 1 stimulation and increased SFK activity(Fig. 8F). Treatment with the SFK inhibitor PP2 (2 μM) (Hanke etal., 1996) blocked the netrin-1-induced increase in SFK phospho-Y416-positive puncta (Fig. 8F), whereas the inactive SFK inhibitoranalog PP3 (2 μM) had no significant effect on the number of punctaper unit area (Fig. 8F). Application of PP2 alone led to a decrease inthe relative number of puncta per unit area, consistent withconstitutive SFK activity contributing to the basal level of punctaobserved (Fig. 8F). To exclude the possibility that the increase inSFK phospho-Y416-positive puncta was secondary to increasedbranch formation, the same quantification was performed with cellsimmunostained for Fyn. No difference was found in the number ofFyn-immunopositive puncta per unit area in control and netrin-1-treated cells (Fig. 8G), consistent with the increase in SFK phospho-Y416-positive puncta resulting from an increase in SFK activity andnot a netrin-1-induced increase in the number of branches.

Our findings indicate that netrin 1 recruits Fyn to a complex withDcc, increasing SFK activity in oligodendrocytes. We then tested thehypothesis that SFK activation is required for the morphologicalchanges induced by netrin 1. SFK activity was assessed in bothimmature (4 DIV, Cnp-positive) and mature (6 DIV, Mbp-positive)oligodendrocytes. Treatment of oligodendrocytes with PP2 blockednetrin-1-induced process extension in Cnp-immunoreactiveimmature cells and the netrin-1-dependent increase in branching inMbp-expressing mature oligodendrocytes, whereas the inactiveanalog PP3 had no effect (Fig. 8H,I). Immature oligodendrocytestreated with PP2 alone showed a small but significant decrease inprocess length (Fig. 8H), which was not seen with PP3 treatment.Mature oligodendrocytes appeared less sensitive to the inhibition ofbasal SFK activity than immature cells, as PP2 alone did not affectbranching (Fig. 8I). We conclude that netrin 1 binding to Dcc resultsin the recruitment of the SFK Fyn, and that subsequent activation ofFyn is required for the netrin-1-induced changes in oligodendrocytemorphology.

RESEARCH ARTICLE Development 136 (3)

Fig. 6. Netrin 1 regulates oligodendrocyte morphology independently of integrins. (A) Application of β1 integrin function-blockingantibody (anti-β1, 2μM) or a peptide encoding the netrin 1 integrin-binding domain (peptide, 20μg/ml) failed to block the increase inoligodendrocyte process branching induced by netrin 1 (100 ng/ml). (B) The β1 integrin function-blocking antibody blocks the fibronectin-inducedincrease in the area (cell spreading) of HEK293T cells, whereas an anti-IgM control antibody does not. A.U., arbitrary units. *P<0.05, ***P<0.0001,versus control. Error bars indicate s.e.m.

DEVELO

PMENT

Netrin 1 inhibits RhoA but does not affect Cdc42or Rac1 in oligodendrocytesWe next asked what signals might act downstream of SFKs to triggerthe morphological changes induced by netrin 1 in oligodendrocytes.Members of the Rho family of small GTPases, including RhoA,Rac1 and Cdc42, regulate the elaboration and branching ofoligodendrocyte processes (Liang et al., 2004). The effect of netrin1 on Cdc42 and Rac1 activity was investigated using a GST-Pak-CRIB pulldown assay (Sander et al., 1998). Cell lysates of culturedmature oligodendrocytes were incubated with GST-Pak-CRIBfusion protein to quantify the levels of GTP-bound Cdc42 and Rac1.No significant change in the levels of GTP-bound Rac1 or Cdc42was detected in oligodendrocytes following application of netrin 1(Fig. 9B,C). We have reported that the Cdc42 effector protein N-WASP is recruited into a protein complex with the intracellulardomain of Dcc following the addition of netrin 1 to embryonic spinalcommissural neurons (Shekarabi et al., 2005). Unlike inoligodendrocytes, netrin 1 activates Cdc42 and Rac1 in commissuralneurons. In mature oligodendrocytes, N-WASP also constitutivelyco-immunoprecipitated with Dcc; however, addition of netrin 1 didnot significantly alter the amount of N-WASP associated with Dcc(Fig. 9D), which is consistent with netrin 1 not increasing theactivation of Cdc42 in oligodendrocytes.

Inhibiting RhoA, or its downstream effector Rho kinase(Rock1/2), in oligodendrocytes increases process extension (Lianget al., 2004; Miron et al., 2007; Wolf et al., 2001). Interestingly,activation of Fyn in oligodendrocytes leads to inactivation of RhoA(Wolf et al., 2001), and we therefore investigated the possibility thatnetrin 1 might regulate RhoA activity in oligodendrocytes. Levelsof GTP-bound RhoA in mature Mbp-positive oligodendrocytes wereassessed using a GST-rhotekin pulldown assay (Reid et al., 1996).Following a 5 minute treatment of mature oligodendrocytes with

netrin 1, a significant decrease in RhoA-GTP was detectedcompared with controls (Fig. 9A). We conclude that netrin-1-induced inhibition of RhoA, in the absence of altered Rac1 or Cdc42activity, promotes process elaboration by oligodendrocytes.

DISCUSSIONDeveloping oligodendrocytes extend and retract their processes oversubstantial distances, sampling the local environment to locateunmyelinated axons (Hardy and Friedrich, 1996; Kirby et al., 2006).As the cell matures, these processes branch and eventually formlamellae that ensheath target axons. Our findings indicate that netrin1 and Dcc regulate oligodendrocyte process extension, branchingand myelin-like membrane sheet formation, which are all essentialevents for initiating myelination.

Exogenous and autocrine netrin 1 contribute tooligodendrocyte maturationOur results support the conclusion that netrin 1, as expressed byneurons and neuroepithelial cells, when encountered by apremyelinating oligodendrocyte evokes Dcc-dependent extensionof the motile processes of the cell. In more mature cells, exogenousnetrin 1 promotes Dcc-dependent process branching and myelin-likemembrane sheet formation. Based on our findings in vitro and invivo, we hypothesize that by promoting the morphologicalmaturation of oligodendrocytes, netrin 1 facilitates the search forappropriate axonal targets. Interestingly, although exogenous netrin1 promotes changes in oligodendrocyte morphology, at later stagesof maturation oligodendrocytes themselves begin to express netrin1. We identify a selective, Dcc-dependent autocrine role for netrin 1in promoting the formation of myelin-like membrane sheets.Oligodendrocytes only begin to express netrin 1 in the developingspinal cord after myelination has begun. Our findings indicate that

423RESEARCH ARTICLERegulation of oligodendrocyte maturation

Fig. 7. Netrin 1 binding to Dcc recruits Fyn. (A) Increased interaction between Dcc and Fyn is detected upon addition of exogenous netrin 1(200 ng/ml, 5 minutes) to cultures of rat oligodendrocytes. This interaction was detected in both immature (4 DIV) and mature (6 DIV)oligodendrocytes and was observed when co-immunoprecipitations (IP) were performed with antibodies against Dcc or Fyn. A constitutiveinteraction between FAK and Dcc was also detected. IgG indicates immunoprecipitation with species-matched non-immune control IgG.(B) Increased active Src family kinase [phospho (p) SFK (pY416)] associated with Dcc in oligodendrocytes following application of netrin 1 (200ng/ml, 5 minutes). (C) No detectable change in SFK activity in whole-cell lysates of cultured oligodendrocytes treated with netrin 1 (200 ng/ml) foreither 5 minutes or 24 hours. (D) Quantification of the interaction between Fyn and Dcc, and between active SFK and Dcc. *P<0.05 versus control.Error bars indicate s.e.m.

DEVELO

PMENT

424

contact with axons is not essential for the initiation of netrin 1expression by these cells, but that it occurs coincident with myelin-like membrane sheet formation. We hypothesize that autocrineexpression of netrin 1 specifically promotes later stages ofmaturation, facilitating the formation of large myelin-like membranesheets by these cells. Netrin 1 expression by oligodendrocytes mightalso facilitate axon remodeling or prevent aberrant axonal sproutingat later stages of development.

Common signaling mechanisms in the regulationof oligodendrocyte processes and axonal growthcones by netrin 1The extending tips of oligodendrocyte processes in some waysresemble motile axonal growth cones (Fox et al., 2006; Jarjourand Kennedy, 2004), and increasing evidence suggests thatcommon regulators of actin nucleation function in neuronalgrowth cones and at the leading edge of oligodendroglial

RESEARCH ARTICLE Development 136 (3)

Fig. 8. SFK activity is required for netrin-1-induced changes in oligodendrocyte morphology. (A) Oligodendrocytes were obtained frommice lacking Fyn (Fyn KO) and treated with netrin 1 (100 ng/ml, 24 hours). Cells obtained from wild-type mice of a matched genetic backgroundwere used as controls. (B) Netrin 1 increased branching in control oligodendrocytes (F2 hybrid), but not in Fyn KO oligodendrocytes. (C,C�) Phospho-SFK [P-SFK (pY416); red] immunoreactivity colocalizes (yellow) with Dcc (green) within the oligodendrocyte cell body and proximal branches. Theconfocal section shown is adjacent to the substrate and the appearance of staining throughout the cell body is not indicative of staining in thenucleus. (D,D�) Portions of the images in C and C� have been magnified to illustrate punctate staining present within distal branches. Arrowsindicate colocalized enrichment of phospho-SFK (red) and DCC (green) at puncta. (E) Binary images of phospho-SFK immunostaining were used toquantify the number of phospho-SFK-positive puncta. (F) The relative number of phospho-SFK-positive puncta increased in the presence of netrin 1(100 ng/ml, 24 hour treatment). PP2 (2μM) blocked the netrin-dependent increase in puncta, and PP2 alone showed a significant decrease inpuncta compared with the control. PP3 (2μM) did not block the effects of netrin 1 and had no independent effect on the cells. (G) The number ofFyn-positive puncta per unit area was not significantly (n.s.) changed upon addition of netrin 1. (H) The netrin-1-dependent increase in processlength of Cnp-positive immature oligodendrocytes was blocked by the addition of PP2 (2μM). Addition of PP3 did not disrupt the effect of netrin 1on the cells. Oligodendrocytes treated with PP2 alone (2μM), but not PP3 alone, exhibit a small but significant decrease in process extensioncompared with the control. (I) Disruption of SFK activity in mature Mbp-positive oligodendrocytes prevented the increased branching observed inthe presence of netrin 1. PP2 and PP3 did not independently affect oligodendrocyte morphology. *P<0.05, **P<0.005, ***P<0.001, versus control.Error bars indicate s.e.m. A,E, 40�0.75 n.a. objective; C-D�, confocal microscopy, 40�0.75 n.a. objective. Scale bars: 20μm in A,E; 40μm in C�,D�. D

EVELO

PMENT

processes and sheets (Sloane and Vartanian, 2007). Here, wereport that the similarities between growing axons andoligodendrocyte processes include the recruitment of similarintracellular signaling proteins downstream of netrin 1: Fyn, FAK,N-WASP and Rho GTPases (Li et al., 2004; Liu et al., 2004; Renet al., 2004; Shekarabi and Kennedy, 2002). We demonstrate thatFyn is required for the netrin-1-dependent increase inoligodendrocyte process branching. Netrin 1 recruits Fyn to acomplex that includes Dcc and FAK, resulting in SFKphosphorylation and activation. Importantly, previous studieshave implicated Fyn, Rac1, Cdc42, RhoA and N-WASP in themechanism governing the cytoskeletal changes inoligodendrocytes that lead to myelination (Simons and Trotter,2007).

Activating Fyn results in RhoA inactivation inoligodendrocytes (Wolf et al., 2001). We demonstrate reducedlevels of GTP-bound RhoA upon netrin 1 stimulation, but foundno evidence for the activation of Cdc42 or Rac1. Interestingly,transgenic mice specifically lacking Cdc42 and Rac1 inoligodendrocytes exhibit normal oligodendroglial differentiation,but show eventual defects in myelin compaction (Thurnherr et al.,2006). Our findings support the hypothesis that in a backgroundof unchanging Rac1 and Cdc42 activity, modulation of RhoAfunction plays a key role in regulating the morphologicaldifferentiation of oligodendrocyte processes.

Netrin 1 activates a canonical signalingmechanism required for oligodendrocytematurationTo date, the candidate extracellular signals that might regulate theelaboration of oligodendroglial processes immediately precedingmyelination have been limited to ECM proteins, of which laminin 2and its receptor integrin α6β1 have been well characterized. Studiesin vitro have shown that integrin-dependent activation of Fynactivates Cdc42 and Rac1 and deactivates RhoA, leading to processoutgrowth (Liang et al., 2004; Osterhout et al., 1999). However,oligodendrocytes lacking the β1 integrin subunit mature andmyelinate normally, demonstrating that this pathway is not essentialin vivo (Benninger et al., 2006). Furthermore, laminin 2 is notubiquitously present in myelinating axon tracts in the CNS,indicating that other ligand-receptor complexes must trigger thesesignaling mechanisms independently of β1 integrin function. Ourresults show that netrin 1, acting through Dcc, activates anintracellular signaling pathway that is required for the morphologicalmaturation of oligodendrocytes. Crucially, the effects of netrin 1 donot require β1 integrin function and do not appear to act throughnetrin-binding integrins.

Our findings identify a novel mechanism regulatingoligodendrocyte morphology during later stages of differentiation.Interestingly, many oligodendroglial cells detected in multiplesclerosis lesions appear to have differentiated, but remain unable toelaborate myelin (Chang et al., 2002). A better understanding ofthe mechanisms that promote myelination will advance thedevelopment of therapeutics that aim to promote the recovery ofnervous system function.

We thank Simon Moore, Sarah-Jane Bull and Adriana Di Polo for comments onthe manuscript. S.R. and A.A.J. were supported by Multiple Sclerosis (MS)Society of Canada studentships and a Canadian Institutes of Health Research(CIHR) Neuroinflammation Strategic Training Program award, K.A.B. by aFonds de la Recherche en Santé du Québec (FRSQ) postdoctoral fellowship,and T.E.K. holds a FRSQ Chercheur Nationaux Award and is a KillamFoundation Scholar. The project was supported by grants (to T.E.K.) from theMS Society of Canada and the CIHR.

ReferencesArmstrong, R. C. (1998). Isolation and characterization of immature

oligodendrocyte lineage cells. Methods 16, 282-292.Bacon, C., Lakics, V., Machesky, L. and Rumsby, M. (2007). N-WASP regulates

extension of filopodia and processes by oligodendrocyte progenitors,oligodendrocytes, and Schwann cells-implications for axon ensheathment atmyelination. Glia 55, 844-858.

Baron, W., Colognato, H. and ffrench-Constant, C. (2005). Integrin-growthfactor interactions as regulators of oligodendroglial development and function.Glia 49, 467-479.

Benninger, Y., Colognato, H., Thurnherr, T., Franklin, R. J., Leone, D. P.,Atanasoski, S., Nave, K. A., Ffrench-Constant, C., Suter, U. and Relvas, J.B. (2006). Beta1-integrin signaling mediates premyelinating oligodendrocytesurvival but is not required for CNS myelination and remyelination. J. Neurosci.26, 7665-7673.

Chang, A., Tourtellotte, W. W., Rudick, R. and Trapp, B. D. (2002).Premyelinating oligodendrocytes in chronic lesions of multiple sclerosis. N. Engl.J. Med. 346, 165-173.

Chun, S. J., Rasband, M. N., Sidman, R. L., Habib, A. A. and Vartanian, T.(2003). Integrin-linked kinase is required for laminin-2-inducedoligodendrocyte cell spreading and CNS myelination. J. Cell Biol. 163, 397-408.

Colognato, H., Baron, W., Avellana-Adalid, V., Relvas, J. B., Baron-VanEvercooren, A., Georges-Labouesse, E. and ffrench-Constant, C. (2002).CNS integrins switch growth factor signalling to promote target-dependentsurvival. Nat. Cell Biol. 4, 833-841.

Colognato, H., Ramachandrappa, S., Olsen, I. M. and ffrench-Constant, C.(2004). Integrins direct Src family kinases to regulate distinct phases ofoligodendrocyte development. J. Cell Biol. 167, 365-375.

Davidson, D., Chow, L. M., Fournel, M. and Veillette, A. (1992). Differentialregulation of T cell antigen responsiveness by isoforms of the src-related tyrosineprotein kinase p59fyn. J. Exp. Med. 175, 1483-1492.

425RESEARCH ARTICLERegulation of oligodendrocyte maturation

Fig. 9. Treatment of mature oligodendrocytes with netrin 1inhibits RhoA activity (A) Levels of GTP-bound RhoA in ratoligodendrocytes were assessed using GST-rhotekin pulldown.Treatment of mature oligodendrocytes with netrin 1 (100 ng/ml, 5minutes) caused a significant decrease in RhoA-GTP compared withcontrols. (B,C) Similar treatment of cells resulted in no significantchange in the levels of GTP-bound Cdc42 or Rac1, as assessed by GST-Pak-CRIB pulldown. (D) Addition of netrin 1 to mature oligodendrocytes(100 ng/ml, 5 minutes) did not alter the amount of N-WASP recruitedto Dcc. *P<0.05 versus control. Error bars indicate s.e.m.

DEVELO

PMENT

426

Fazeli, A., Dickinson, S. L., Hermiston, M. L., Tighe, R. V., Steen, R. G., Small,C. G., Stoeckli, E. T., Keino-Masu, K., Masu, M., Rayburn, H. et al. (1997).Phenotype of mice lacking functional Deleted in colorectal cancer (Dcc) gene.Nature 386, 796-804.

Fox, M. A., Afshari, F. S., Alexander, J. K., Colello, R. J. and Fuss, B. (2006).Growth conelike sensorimotor structures are characteristic features ofpostmigratory, premyelinating oligodendrocytes. Glia 53, 563-566.

Hanke, J. H., Gardner, J. P., Dow, R. L., Changelian, P. S., Brissette, W. H.,Weringer, E. J., Pollok, B. A. and Connelly, P. A. (1996). Discovery of a novel,potent, and Src family-selective tyrosine kinase inhibitor: study of Lck- and FynT-dependent T cell activation. J. Biol. Chem. 271, 695-701.

Hardy, R. J. and Friedrich, V. L., Jr (1996). Progressive remodeling of theoligodendrocyte process arbor during myelinogenesis. Dev. Neurosci. 18, 243-254.

Jarjour, A. A. and Kennedy, T. E. (2004). Oligodendrocyte precursors on themove: mechanisms directing migration. Neuroscientist 10, 99-105.

Jarjour, A. A., Manitt, C., Moore, S. W., Thompson, K. M., Yuh, S.-J. andKennedy, T. E. (2003). Netrin-1 is a chemorepellent for oligodendrocyteprecursor cells in the embryonic spinal cord. J. Neurosci. 23, 3735-3744.

Kennedy, T. E., Serafini, T., de la Torre, J. R. and Tessier-Lavigne, M. (1994).Netrins are diffusible chemotropic factors for commissural axons in theembryonic spinal cord. Cell 78, 425-435.

Kirby, B. B., Takada, N., Latimer, A. J., Shin, J., Carney, T. J., Kelsh, R. N. andAppel, B. (2006). In vivo time-lapse imaging shows dynamic oligodendrocyteprogenitor behavior during zebrafish development. Nat. Neurosci. 9, 1506-1511.

Knapp, P. E., Bartlett, W. P. and Skoff, R. P. (1987). Cultured oligodendrocytesmimic in vivo phenotypic characteristics: cell shape, expression of myelin-specificantigens, and membrane production. Dev. Biol. 120, 356-365.

Li, W., Lee, J., Vikis, H. G., Lee, S. H., Liu, G., Aurandt, J., Shen, T. L., Fearon,E. R., Guan, J. L., Han, M. et al. (2004). Activation of FAK and Src arereceptor-proximal events required for netrin signaling. Nat. Neurosci. 7, 1213-1221.

Liang, X., Draghi, N. A. and Resh, M. D. (2004). Signaling from integrins to Fynto Rho family GTPases regulates morphologic differentiation ofoligodendrocytes. J. Neurosci. 24, 7140-7149.

Liu, G., Beggs, H., Jurgensen, C., Park, H. T., Tang, H., Gorski, J., Jones, K. R.,Reichardt, L. F., Wu, J. and Rao, Y. (2004). Netrin requires focal adhesionkinase and Src family kinases for axon outgrowth and attraction. Nat. Neurosci.7, 1222-1232.

Manitt, C., Colicos, M. A., Thompson, K. M., Rousselle, E., Peterson, A. C.and Kennedy, T. E. (2001). Widespread expression of Netrin-1 by neurons andoligodendrocytes in the adult mammalian spinal cord. J. Neurosci. 21, 3911-3922.

Meriane, M., Tcherkezian, J., Webber, C. A., Danek, E. I., Triki, I., McFarlane,S., Bloch-Gallego, E. and Lamarche-Vane, N. (2004). Phosphorylation of DCCby Fyn mediates Netrin-1 signaling in growth cone guidance. J. Cell Biol. 167,687-698.

Miron, V. E., Rajasekharan, S., Jarjour, A. A., Zamvil, S. S., Kennedy, T. E. andAntel, J. P. (2007). Simvastatin regulates oligodendroglial process dynamics andsurvival. Glia 55, 130-143.

Osterhout, D. J., Wolven, A., Wolf, R. M., Resh, M. D. and Chao, M. V.(1999). Morphological differentiation of oligodendrocytes requires activation ofFyn tyrosine kinase. J. Cell Biol. 145, 1209-1218.

Reid, T., Furuyashiki, T., Ishizaki, T., Watanabe, G., Watanabe, N., Fujisawa,K., Morii, N., Madaule, P. and Narumiya, S. (1996). Rhotekin, a new putativetarget for Rho bearing homology to a serine/threonine kinase, PKN, andrhophilin in the rho-binding domain. J. Biol. Chem. 271, 13556-13560.

Ren, X. D. and Schwartz, M. A. (2000). Determination of GTP loading on Rho.Methods Enzymol. 325, 264-272.

Ren, X. R., Ming, G. L., Xie, Y., Hong, Y., Sun, D. M., Zhao, Z. Q., Feng, Z.,Wang, Q., Shim, S., Chen, Z. F. et al. (2004). Focal adhesion kinase in netrin-1signaling. Nat. Neurosci. 7, 1204-1212.

Ricard, D., Rogemond, V., Charrier, E., Aguera, M., Bagnard, D., Belin, M.-F.,Thomasset, N. and Honnorat, J. (2001). Isolation and expression pattern ofhuman Unc-33-like phosphoprotein 6/collapsin response mediator protein 5(Ulip6/CRMP5): coexistence with Ulip2/CRMP2 in Sema3A- sensitiveoligodendrocytes. J. Neurosci. 21, 7203-7214.

Sander, E. E., van Delft, S., ten Klooster, J. P., Reid, T., van der Kammen, R.A., Michiels, F. and Collard, J. G. (1998). Matrix-dependent Tiam1/Racsignaling in epithelial cells promotes either cell-cell adhesion or cell migrationand is regulated by phosphatidylinositol 3-kinase. J. Cell Biol. 143, 1385-1398.

Schwab, M. E. and Schnell, L. (1989). Region-specific appearance of myelinconstituents in the developing rat spinal cord. J. Neurocytol. 18, 161-169.

Serafini, T., Colamarino, S. A., Leonardo, E. D., Wang, H., Beddington, R.,Skarnes, W. C. and Tessier-Lavigne, M. (1996). Netrin-1 is required forcommissural axon guidance in the developing vertebrate nervous system. Cell87, 1001-1014.

Shekarabi, M. and Kennedy, T. E. (2002). The netrin-1 receptor DCC promotesfilopodia formation and cell spreading by activating Cdc42 and Rac1. Mol. Cell.Neurosci. 19, 1-17.

Shekarabi, M., Moore, S. W., Tritsch, N. X., Morris, S. J., Bouchard, J. F. andKennedy, T. E. (2005). Deleted in colorectal cancer binding netrin-1 mediatescell substrate adhesion and recruits Cdc42, Rac1, Pak1, and N-WASP into anintracellular signaling complex that promotes growth cone expansion. J.Neurosci. 25, 3132-3141.

Simons, M. and Trotter, J. (2007). Wrapping it up: the cell biology ofmyelination. Curr. Opin. Neurobiol. 17, 533-540.

Sloane, J. A. and Vartanian, T. K. (2007). WAVE1 and regulation of actinnucleation in myelination. Neuroscientist 13, 486-491.

Smart, J. E., Oppermann, H., Czernilofsky, A. P., Purchio, A. F., Erikson, R. L.and Bishop, J. M. (1981). Characterization of sites for tyrosine phosphorylationin the transforming protein of Rous sarcoma virus (pp60v-src) and its normalcellular homologue (pp60c-src). Proc. Natl. Acad. Sci. USA 78, 6013-6017.

Sperber, B. R., Boyle-Walsh, E. A., Engleka, M. J., Gadue, P., Peterson, A. C.,Stein, P. L., Scherer, S. S. and McMorris, F. A. (2001). A unique role for Fyn inCNS myelination. J. Neurosci. 21, 2039-2047.

Thurnherr, T., Benninger, Y., Wu, X., Chrostek, A., Krause, S. M., Nave, K. A.,Franklin, R. J., Brakebusch, C., Suter, U. and Relvas, J. B. (2006). Cdc42 andRac1 signaling are both required for and act synergistically in the correctformation of myelin sheaths in the CNS. J. Neurosci. 26, 10110-10119.

Tsai, H. H., Tessier-Lavigne, M. and Miller, R. H. (2003). Netrin 1 mediatesspinal cord oligodendrocyte precursor dispersal. Development 130, 2095-2105.

Tsai, H. H., Macklin, W. B. and Miller, R. H. (2006). Netrin-1 is required for thenormal development of spinal cord oligodendrocytes. J. Neurosci. 26, 1913-1922.

Umemori, H., Sato, S., Yagi, T., Aizawa, S. and Yamamoto, T. (1994). Initialevents of myelination involve Fyn tyrosine kinase signalling. Nature 367, 572-576.

Watanabe, M., Sakurai, Y., Ichinose, T., Aikawa, Y., Kotani, M. and Itoh, K.(2006). Monoclonal antibody Rip specifically recognizes 2’,3�-cyclic nucleotide3�-phosphodiesterase in oligodendrocytes. J. Neurosci. Res. 84, 525-533.

Wolf, R. M., Wilkes, J. J., Chao, M. V. and Resh, M. D. (2001). Tyrosinephosphorylation of p190 RhoGAP by Fyn regulates oligodendrocytedifferentiation. J. Neurobiol. 49, 62-78.

Yebra, M., Montgomery, A. M., Diaferia, G. R., Kaido, T., Silletti, S., Perez,B., Just, M. L., Hildbrand, S., Hurford, R., Florkiewicz, E. et al. (2003).Recognition of the neural chemoattractant Netrin-1 by integrins alpha6beta4and alpha3beta1 regulates epithelial cell adhesion and migration. Dev. Cell 5,695-707.

RESEARCH ARTICLE Development 136 (3)

DEVELO

PMENT