Download - Analysis of combinatorial variability reveals selective accumulation of the fibronectin type III domains B and D of tenascin-C in injured brain

Experimental Neurology 225 (2010) 60–73

Contents lists available at ScienceDirect

Experimental Neurology

j ourna l homepage: www.e lsev ie r.com/ locate /yexnr

Analysis of combinatorial variability reveals selective accumulation of the fibronectintype III domains B and D of tenascin-C in injured brain

Alexandre Dobbertin a,c,1, Stefan Czvitkovich b,1, Ursula Theocharidis a, Jeremy Garwood a,Melissa R. Andrews b, Francesca Properzi b, Rachel Lin b, James W. Fawcett b, Andreas Faissner a,⁎a Department of Cell Morphology and Molecular Neurobiology, Ruhr University of Bochum, 44780 Bochum, Germanyb Cambridge Centre for Brain Repair, University of Cambridge, Forvie Site, Cambridge CB2 2PY, UKc Université Paris Descartes, INSERM U686, Centre universitaire des Saints-Pères, 75006 Paris, France

⁎ Corresponding author. Ruhr University, DepartmMolecular Neurobiology, NDEF 05/593 UniversitätsGermany. Fax: +49 234 32 14313.

E-mail address: [email protected] These authors contributed equally to this work.

0014-4886/$ – see front matter © 2010 Published by Edoi:10.1016/j.expneurol.2010.04.019

a b s t r a c t
a r t i c l e i n f o
Article history:Received 22 December 2009Revised 23 April 2010Accepted 30 April 2010Available online 5 May 2010

Keywords:Cerebral cortexCNS injuryGlial scarAstrocyteTenascinFibronectin type III domainExtracellular matrixRegenerationTGFβ1Cytokine

Tenascin-C (Tnc) is a multimodular extracellular matrix glycoprotein that is markedly upregulated in CNSinjuries where it is primarily secreted by reactive astrocytes. Different Tnc isoforms can be generated by theinsertion of variable combinations of one to seven (in rats) alternatively spliced distinct fibronectin type III(FnIII) domains to the smallest variant. Each spliced FnIII repeat mediates specific actions on neuriteoutgrowth, neuron migration or adhesion. Hence, different Tnc isoforms might differentially influence CNSrepair. We explored the expression pattern of Tnc variants after cortical lesions and after treatment ofastrocytes with various cytokines. Using RT-PCR, we observed a strong upregulation of Tnc transcriptscontaining the spliced FnIII domains B or D in injured tissue at 2–4 days post-lesion (dpl). Looking at specificcombinations, we showed a dramatic increase of Tnc isoforms harboring the neurite outgrowth-promotingBD repeat with both the B and D domains being adjacent to each other. Isoforms containing only the axongrowth-stimulating spliced domain D were also dramatically enhanced after injury. Injury-induced increaseof Tnc proteins comprising the domain D was confirmed by Western Blotting and immunostaining of corticallesions. In contrast, the FnIII modules C and AD1 were weakly modulated after injury. The growth conerepulsive A1A2A4 domains were poorly expressed in normal and injured tissue but the smallest isoform,which is also repellant, was highly expressed after injury. Expression of the shortest Tnc isoform and ofvariants containing B, D or BD, was strongly upregulated in cultured astrocytes after TGFβ1 treatment,suggesting that TGFβ1 could mediate, at least in part, the injury-induced upregulation of these isoforms. Weidentified complex injury-induced differential regulations of Tnc isoforms that may well influence axonalregeneration and repair processes in the damaged CNS.

ent of Cell Morphology andstrasse 150, 44780 Bochum,

de (A. Faissner).

lsevier Inc.

© 2010 Published by Elsevier Inc.

Introduction

Tenascin-C (Tnc) is a glycoprotein present in the extracellularmatrix (ECM) of a variety of tissueswhere it playsmultiple roles in cellgrowth, migration and adhesion during development but also underconditions of tissue remodeling in the adult organism such as woundhealing and tumor growth (Bartsch, 1996; Faissner, 1997; Jones andJones, 2000; Joester and Faissner, 2001; Chiquet-Ehrismann andChiquet, 2003). In the central nervous system (CNS), Tnc is widelyexpressed at early stages of development, principally by astrocytes andradial glia, where it has a variety of functions such as influencingneurite outgrowth and guidance and in the development of glia

(Faissner, 1997; Meiners et al., 1999; Joester and Faissner, 2001;Garcion et al., 2001; Rigato et al., 2002). Tnc expression is down-regulated in most regions of the adult CNS but it is stronglyupregulated around lesion sites of CNS stabwound injuries, associatedwith a subset of GFAP positive astrocytes (McKeon et al., 1991; Laywellet al., 1992; Brodkey et al., 1995; Zhang et al., 1997; Tang et al., 2003).Tnc upregulation is also seen following other forms of CNS damage,such as in the dentate gyrus after unilateral entorhinal cortex lesion(Deller et al., 1997) or the hippocampus after injection of kainic acid(Niquet et al., 1995; Nakic et al., 1996).

Tnc is encoded by a single gene giving rise to a number ofalternatively spliced variants that differ in their number of fibronectintype III (FnIII) domains (see Fig. 1 for structural details). The smallestTnc variant contains a series of 8 constant FnIII repeats that are presentin all rodent Tnc molecules. Numerous larger isoforms of Tnc aregenerated by the insertion of up to six (in mice) and seven (in rats)additional alternatively spliced FnIII domains between the 5th and 6thconstant FnIII domains of the basic structure. In mice, 28 Tnc isoforms

mailto:[email protected]

http://dx.doi.org/10.1016/j.expneurol.2010.04.019

http://www.sciencedirect.com/science/journal/00144886

Fig. 1. (A) Schematic representation of the structural organization of the tenascin-Cmonomer. Tnc is produced as a hexamer and each monomer exhibits a multimodularcomposition with a serial arrangement of characteristic structural domains. The amino-terminal sequence or Tnc assembly domain (TA) contributes to the central knob, acysteine-rich structure that assembles six monomers via disulfide bridges to form thenative hexameric protein, the hexabrachion. This region is followed by 14.5 EGF-likerepeats (31 aa each) and 8 fibronectin type III (FnIII) modules (90 aa each) in thesmallest Tnc variant. The sequence is terminated by a domain (FG) homologous tofibrinogen-β and -γ (Joester and Faissner, 2001). Numerous larger isoforms of Tnc aregenerated by the insertion of up to seven additional alternatively spliced FnIII domainsbetween the 5th and 6th constant FnIII domains of the basic structure in rats. Thesealternatively spliced domains have been named A1, A2, A4, B, C and D and, recently, wehave discovered an additional spliced domain for the rat, called AD1. (B) The diagramillustrates the strategy used to measure expression of the various alternatively spliceddomains and their combinations. Primer pairs were used that bridge single domains,two domains, three domains, four domains and the whole alternatively spliced region.

61A. Dobbertin et al. / Experimental Neurology 225 (2010) 60–73

have so far been identified in the CNS which differ in the combinationof additional FnIII repeats (Joester and Faissner, 1999; von Holst et al.,2007). The alternatively spliced domains have been named A1, A2, A4,B, C and D and, recently, we have discovered an additional spliceddomain in rat Tnc, called AD1 (unpublished results) as it correspondsto the AD1 domain, previously only described in humans and chicken(Derr et al., 1997). Diverse functions have been ascribed to several ofthe spliced FnIII domains through studies using fusion proteins and

blocking antibodies. For example, a FnIII A1,A2,A4 peptide exhibitsgrowth cone repulsive properties, whereas the recombinant proteinsFnIII D, FnIII D6 and FnIII BD promote neurite outgrowth from varioustypes of neurons (Götz et al., 1996; Meiners et al., 1999, 2001; Rigatoet al., 2002). These results suggest large functional diversity of Tncisoforms. In vivo and in vitro studies demonstrate that Tnc can providepermissive as well as inhibitory cues for axonal growth and guidance(Faissner and Kruse, 1990; Lochter et al., 1991; Husmann et al., 1992;Götz et al., 1997; Meiners and Geller, 1997; Meiners et al., 1999).Interestingly, expression of Tnc isoforms and spliced FnIII domains isdifferentially regulated during development, and periods of increasedaxonal growth in the developing CNS are closely correlated withexpression of large isoforms (Prieto et al., 1990; Bartsch et al., 1992;Tucker, 1993; Joester and Faissner, 1999, 2001).

Given the potentially diverse actions of differentially spliced Tncmolecules, it is possible that the expression of different forms of Tncmight have functional significance in glial scarring and axonalregeneration. To date, it is not known which alternatively splicedTnc forms and FnIII domains are preferentially expressed after injury.Here, we examined the expression of Tnc isoforms and spliced FnIIIdomains in a cortical and basal forebrain injury model. In vitro,astrocytic expression of Tnc is increased by TGFβ1 and bFGF, both ofwhich are upregulated after CNS damage (Meiners et al., 1993;Mahleret al., 1997; Smith and Hale, 1997). However, the factors regulatingthe expression of specific isoforms in these cells are unknown. Thus,we have also examined the effects of injury-related cytokines on theproduction of Tnc isoforms by cultured astrocytes.

Materials and methods

Cytokines and primary antibodies

Recombinant cytokines: human Transforming Growth Factor β1(TGFβ1), human basic Fibroblast Growth Factor (bFGF), humanEpidermal Growth Factor (EGF), human Transforming Growth Factorα (TGFα), human Platelet Derived Growth Factor-AB (PDGF-AB) andmurine Vascular Endothelial Growth Factor (VEGF) were from R&DSystems (Abingdon, Oxon, UK) and from PeproTech Inc. (Rocky Hill,New Jersey). The rabbit polyclonal antibody (pAb) KAF14 was raisedagainst purified Tnc from postnatal mouse brains as describedpreviously (Faissner and Kruse, 1990). The monoclonal antibody(mAb) 578 recognizes specifically the FnIII domain D and mAb 19H12binds to an epitope located on the FnIII combination A1A2A4 (Götzet al., 1996). Mouse Tnc used in Western Blots was purified from P7–P14 mouse brains as described previously (Faissner and Kruse, 1990).Mouse anti-GFAP (Glial Fibrillary Acidic Protein)mAbwas from Sigmaand rabbit anti-GFAP pAb was from Dako.

Surgical procedures

Adult female Sprague Dawley rats (Charles River, Margate, UK)weighing approximately 200 g were anesthetized under halothane.3 mm deep lesions were generated unilaterally by lowering a sterilescalpel into the right-hand side of the cerebral cortex with the otherhemisphere serving as a control (Asher et al., 2000). Animals wereallowed to survive for 2 to 28 days after the operation before they wereterminally anesthetized and decapitated. Brains were either immedi-ately dissected or stored at−70 °C. For immunohistochemistryweusedadult male CD1mice (20–30 g). For cortical or basal forebrain lesions, aScoutenwire knife (Kopf Instrument, Harvard Apparatus, UK)was used.Animals were perfused and sacrificed 7 days post-lesion. Brains wereremoved and stored first in 4% paraformaldehyde and then in 30%sucrose. All procedures were conducted in compliance with the UKAnimals Scientific Procedures Act 1986.

62 A. Dobbertin et al. / Experimental Neurology 225 (2010) 60–73

Culture of astrocytes

The cells were prepared from the brains of newborn to 2 day-oldSprague–Dawley rats and dissociated as described previously (Dobber-tin et al., 2003). Finally, dissociated cellswere resuspended inDulbecco'sModified Eagle Medium (DMEM; Life Technologies) containing 10%fetal calf serum (FCS, Harlan Sera Lab, Loughborough, UK), 100 U/mlpenicillin, 100 µg/ml streptomycin and 2.5 µg/ml amphotericin B(Fungizone) (all from Sigma) (DMEM FCS 10%) and plated (1–2brains/flask) into 75 cm2 poly-D-lysine (Sigma)-coated tissue cultureflasks (Iwaki, Japan). The medium was replaced next day andsubsequently every third day with fresh DMEM FCS 10%. Between the8th and 12th days, cultures were shaken overnight to removemacrophages and progenitor cells. Adherent astrocytes were passagedinto poly-D-lysine-coated dishes (Nunc) or flasks and maintained inDMEMFCS 10% until confluence for about 4 days. Themediumwas thenreplaced with serum-free medium containing 6.25 µg/ml insulin,6.25 µg/ml transferrin, 6.25 ng/ml selenious acid, 1.25 mg/ml BSA,5.35 µg/ml linoleic acid (as ITS+, Collaborative Biomedical Products,Bedford, MA) and 20 µM of the DNA synthesis inhibitor, cytosine-1-β-Darabinofuranosid (Ara-C, Fluka, Dorset, UK). After 5–7 days in thismedium, astrocytemonolayerswere invariably freeof any contaminatingprogenitor cells, and in particular of any O2A cells as checked byimmunostaining. Remaining microglia-like cells were eliminated bytreatment with 10 mM L-leucine methyl ester (Sigma) for 30 min. Cellswere grown for an additional day in the same serum-free mediumwithout Ara-C. Astrocytemonolayers were then cultivated in serum-freemedium containing the cytokines or growth factors under investigationfor 1 to 4 days according to the experiment.

Cell counting

At the end of some experiments, cultures of astrocytes were fixed incoldmethanol (−20 °C) for 2 min. The nuclei of fixed cellswere stainedwith a solution of bis-benzimide (Sigma) diluted 1:5000 in PBS for30 min at room temperature. After washing and mounting in DABCO-glycerol, blue stained cell nuclei were counted under a fluorescencemicroscope.

Sample preparation for SDS-PAGE

Brain tissueA b1 mmrimof tissuewas dissected fromaroundCNS lesions, giving

approximately 10 mg of tissue. Equivalent regions of uninjured brainswere removed to provide control tissue. Scar or control tissues from twoanimals were pooled and homogenized in 300 µl extraction buffer(150 mMNaCl, 1% NP-40, 0.5% DOC, 0.1% SDS (Sigma), 50 mMTris–HClpH 7.5, 20 mM EDTA, complete protease inhibitor cocktail (Roche),10 µg/ml pepstatin A) until no tissue was visible anymore. Homoge-nized tissue was extracted for 1 h on ice. Extracts were centrifuged for20 min at 15,000 g at 4 °C. Supernatants were measured for proteincontent (Biorad) and 30 µg of extract was loaded per lane. As control,embryonic day 18 brains were homogenized and extracted using thesame procedures as above.

Astrocyte-conditioned mediumAfter 4 days cytokine treatment, supernatants were centrifuged to

remove cell debris, and a mixture of protease inhibitors (EDTA (5 mM),phenylmethylsulfonyl fluoride (PMSF; 1 mM; Sigma), iodoacetamide(0.1 mM) (Fluka), soybean trypsin inhibitor, aprotinin, leupeptin andpepstatin (all at 5 µg/ml and from Sigma)) were added, and super-natants were frozen on dry ice and stored at −80 °C. In someexperiments, defrosted supernatants were placed in columns (Milli-pore)with a cut off at 100 kDa and centrifuged for aminimumof 20 minat 1000 g at 4 °C to enrich for high molecular weight proteins. Spinning

was stopped when around 250 µl was left in the columns to give a 10times concentrated supernatant.

Electrophoresis and Western Blot

Proteinswere separated by SDS-PAGE 5% polyacrylamide gels underreducing conditions, before electrophoretic transfer (175 mA, 17 h)onto polyvinylidene difluoride (PVDF) membranes (Amersham). Blotswere blocked for 1 h in 5%non-fat driedmilk in TBS–0.05%Tween20. Allsubsequent steps were performed with this blocking buffer at roomtemperature. Blots were then incubated with the primary pAb KAF14(1:500) and then incubated with peroxydase-conjugated anti-rabbitIgG (dilution 1/10,000; Amersham) antibodies or with mAb 578(1:100) or mAb 19H12 (1:100) followed by peroxydase-conjugatedanti-rat Ig (1:1000) before exposure to the chemiluminescent substrateECL or ECL Plus (Amersham). No signal was observedwhen the primaryantibody was omitted.

Immunofluorescence labeling of brain lesions

Because some of our anti-Tnc antibodies were from rat, immuno-histochemistry for localization of its expression was performed inlesioned mouse brain. Microtome sections (30 µm) of lesioned mousecortex were cut from fixed brains and pretreated with PBS Triton 0.1%for 10 min, then incubated in PBS 3% BSA for 2 h before doubleimmunostaining with either the rabbit anti-Tnc pAb KAF14 (1:200)and a mouse anti-GFAP mAb (Sigma; 1:400) or with the rat anti-FnIIIdomain D mAb 578 (1:20) or anti-FnIII domains A1,2,4 mAb 19H12(1:20) and a rabbit anti-GFAP pAb (Dako; 1:400). Primary antibodieswere visualized with Alexa Fluor 488 goat anti-mouse or anti-rabbit(GFAP), Alexa Fluor 568 goat anti-rabbit (KAF14) secondary anti-bodies (Invitrogen) or with a biotinylated goat anti-rat secondary Aband then streptavidin 568 (Invitrogen) (for mAb 578 and 19H12stainings), all at 1:200.

ELISA detection of tenascin-C

Purified astrocytes (see above) were plated in 4-well dishes(Greiner) and grown for one day in serum-containing medium untilconfluency. Themediumwas replaced by serum-freemedium (DMEM/ITS) for 1 day. Then, the cells were treated with TGFβ1 (5 ng/ml) andbFGF (10 ng/ml) andHeparin (0.25 U/ml) for 4 days. Control cellswereleft without growth factors. The medium was removed and the cellswashed with PBS before lysis with H2O–EDTA 10 mM for three times15 min on ice. The matrix was washed with PBS and blocked with PBSBSA 0.25% for 45 min at 37 °C. In some conditions chondroitin sulfateswere removed by incubating the matrix with chondroitinase ABC(ChABC, Sigma, 50 mU/ml) for 3 h at 37 °C. After this treatment thematrix was washed with PBS and blocked for 1 h with 5% non-fat driedmilk in PBS 0.05% Tween 20 at 37 °C. In this buffer the polyclonalantibodyKAF14(1:250)and themonoclonal antibody578(1:100)wereincubated overnight at 4 °C. After washing four times for 10 min withPBS-Tween 20 the secondary HRP-coupled antibodies (anti-rabbit,1:5000; anti-rat, 1:10,000 in PBS-Tween) were applied for 90 min at37 °C. The ABTS substrate (1 mg/ml in acetate-phosphate buffer) wasadded and the color reaction stopped after 30 min with 0.6% SDS. Thecolored product was quantified with an ELISA reader at 405 nm.

RNA isolation

The brains were immediately dissected in RNase-free conditions. A8–10 mgpieceof tissue containing the lesionwasexcised fromthe brainwhile a similar-sized piece was dissected from the same location on thecontralateral cortex. The tissue was then frozen in dry ice beforehomogenization in guanidinium thiocyanate according to the methoddescribed by Chomczynski and Sacchi (1987). After two rounds of


ethanol precipitation, purified RNA was resuspended in DEPC-treatedwater and quantified by spectrophotometry. For the astrocytes, afterremoval of the culture medium, total RNA from astrocytes was isolatedusing the RNeasy mini kit (Qiagen) according to the manufacturer'sprotocol. In brief, cells were directly lysed in the culture dish and thelysate applied to anRNeasymini column.After binding andwashing, thepurified RNA was eluted with RNase-free water. To eliminate genomiccontamination the samples were treated with DNase I.

Semi-quantitative RT-PCR analysis

For the reverse transcription of RNA from astrocytes, 1 µg of totalRNA was reverse transcribed using 0.4 µg random hexamer primersand the first strand cDNA synthesis kit (MBI Fermentas) in a totalvolume of 40 µl according to the manufacturer's protocol. For RNAfrom brain tissue, 0.8 µg RNA was reverse transcribed using 0.4 µgrandom hexamer primers (Promega) and SuperScript III ReverseTranscriptase (Invitrogen) in a total volume of 20 µl.

Amplification of isoform-specific Tnc transcripts was performedusing 2 µl of diluted reverse transcriptase reaction per 20 µl of PCRreaction with 300–500 nM of the appropriate sense and antisenseprimers (see Table 1 and Fig. 1 for the description and sequences of theprimers). The reaction conditions were as follows: 60 mM Tris/HCl pH8.8, 15 mM (NH4)2SO4, 2 mM MgCl2, 0.2 mM dNTPs and 0.1 µl ofHotStarTaq polymerase (5 U/µl) (Qiagen, France) per reaction. Cyclingwasperformedusing a thermocycler (Mastercycler, Eppendorf) startingwith a 15 min activation step at 95 °C, followed by 30–35 cycles (up to41 cycles for the amplication of domainsA1, A2, A4) of 30 s at 94 °C, 30 sat 55 °C and 60 to 150 s (according to the PCR fragment length) at 72 °C,and a final extension step at 72 °C for 10 min. Rat integrin α7 (senseprimer: 5′-TCCATTAAGAACTTGCTGCTCA-3′ and antisense: 5′-ACTGA-CACAGGATGTCCATCAG-3′) and F3/contactin (sense: 5′-GAAGGT-GAAGGAAGGGAAGG-3′; antisense: 5′-CAACGTTCGCGATGTAGAGA-3′)were amplifiedwith 35 cycles of 30 s at 94 °C, 30 s at 57 °C and 1 min at72 °C. Quantification of final PCR products from 1% agarose gel picturesused the Tina 2.0 software (France). Relative RNA concentrations werequantified using the housekeeping gene, glyceraldehyde 3-phosphatedehydrogenase (GAPDH)whichwasamplifiedusing the rat forward: 5′-CCTTCATTGACCTCAACTACATGG-3′ and reverse:5′-CCAGTAGACTCCAC-GACATACTCA-3′ primers (amplicon size=193 bp) and 25 cycles toremain in a linear range (30 s 94 °C, 30 s at 59 °C and 1 min at 72 °C). Nogenomic DNA could interferewith the amplifications of combinations of

Table 1Tenascin-C primers used for amplification of specific FnIII domains or combinations.Nucleotides corresponding to Tnc FnIII domain sequences of primers are in bold. Sense(s) and antisense (as) primers start respectively at the first and last nucleotide of thecorresponding fibronectin type III domain chosen for amplification, except for Caswhich starts at position 162 of the domain C (length 273 nucleotides), for A4as whichstarts at position 37 of the domain A4 and for AD1s and AD1as which start at 15 and 191of the domain respectively. In some primers, restriction sites for KpnI (GGT ACC) and forXbaI (TCT AGA) were included and are indicated in italics.

Name Oligonucleotide sequence (5′→3′)

7s cgg ggt acc gct ctg gat ggt cca tct g8as c tag tct aga tgt tgt gaa gat ggt ttg g5s cgg ggt acc gaa att gat gca ccc aag gA1s cgg ggt acc gaa gaa gtg cct tcc ctg gA2as c tag tct aga tgt caa gac ctc aac aga gA4as cag tca cag aca agc ctBs cgg ggt acc gcc aga gaa cct gaa att ggBas c tag tct aga tgt ggt agc cgt ggt act gAD1s gtt ggg cat gct aat ctt tag cAD1as acc aag cct gtg atg tga gct tCs cgg ggt acc gag gcc ttg ccc ctt ctg gCas ggt tcc tga aag tgt gaa ttc cDs cgg ggt acc gaa gct gaa ccg gaa gtt gDas c tag tct aga tgt tgt tgc tat ggc act g6as c tag tct aga tgt gat tag agt ccc cga g

spliced domains such as BD as these are all separated from each one byan intron. For the other amplifications including GAPDH and singlespliced domains, no significant signal was detected when the reversetranscription step was omitted. To set up non-saturating conditions forthe PCR, a range of 3 dilutions of the reverse transcriptase reactionweresystematically amplified in parallel in each case. The results presentedcorrespond to non-saturating PCR conditions obtained for reversetranscriptase reaction diluted by a factor 3 (except for A1,A2,A4 PCRwith a dilution factor 2).

Results

RT-PCR analysis of the differential regulation of Tnc isoforms after CNS injury

In order to study the regulation of Tnc mRNA and its splice variantsafter CNS injury, we examined tissue dissected from around lesions incerebral cortices 2, 4, 7 and 15 days post-lesion (dpl) and in controltissue by semi-quantitative RT-PCR analysis. To determine the expres-sion levels of all Tnc variants, we used a primer pair amplifying theconstitutively expressed domains 7 and 8 (Fig. 1 and Table 1). Weobserved a strong upregulation of Tnc transcripts in lesions at 2–4 dpl ofabout 6 times compared to uninjured tissue, before returning to basallevels by 15 dpl (Fig. 2A). Consistent with Tnc downregulation in CNSmaturation, overall Tnc mRNA levels in control adult brains were low(Joester and Faissner, 1999, 2001).

The size range of splice variants amongst the Tnc transcripts wasinvestigated using oligonucleotide primers for the constitutive FnIIIdomains 5 and 6 which flank the splice site for all of the isoformvariations (see Fig. 1). Eight bands were observed after RT-PCRcorresponding to the transcripts, which contained different numbersof additional spliced FnIII domains from the shortest (devoid of anyadditional FnIII exon between FnIII domains 5 and 6) to the largestisoform (containing all 7 additional FnIII inserts). In the normal cerebralcortex, the strongest signal corresponded to the shortest isoform(Fig. 2B). Tnc isoforms comprising 1 and 2 spliced FnIII domains weremuch less expressed than the shortest isoform,whereas larger isoformswere barely detectable, in accordance with previous results indicatingdownregulation of large isoforms in CNS maturation (Joester andFaissner, 1999).After injury, the shortest isoformremained theprincipalformexpressed. In addition to shortest isoform, anupregulation of otherisoforms could be observed too. Isoforms comprising a single spliceddomain were highly expressed at 2 and 4 dpl, while larger isoformswere much less detected in comparison. However, it should be notedthat the amount of large isoforms might be underestimated due to apossibly less efficient PCR amplification than for short isoforms and,therefore, it is not possible to make exact quantitative comparisons fordifferent sized isoforms. Nevertheless, despite a probable loweramplification rate of large isoforms, the signal corresponding to isoformscontaining 5 spliced domains wasmuch greater compared to signals forisoforms comprising 3 and 4 additional FnIII domains at 2 and 4 dpl. Theintensity of PCR products containing 5 inserted FnIII domains wasequivalent (considering its size-linked higher ethidium bromidefluorescence) to the signal from forms with 2 spliced FnIII repeats at4 dpl, thus indicating a greater relative abundance of forms with 5spliced FnIII domains than forms with 3 and 4 additional FnIII inserts atthese time points.

In contrast, it is possible tomake a quantitative comparison betweencontrol and lesioned tissue for each isoform size. Isoforms comprising 0and 2 inserted FnIII domainswere augmented at 2–4 dpl by about 4 and5 fold respectively (Fig. 2B).Moreover, we observed amore pronouncedincrease in mRNA expression of short Tnc isoforms mRNA bearing oneadditional spliced FnIII module with a 10 fold peak increase at 4 dplbefore returning to basal levels by 7–15 dpl (Fig. 2B), revealing subtleregulations among the different Tnc variants. Although not quantifiable,because of low expression in controls, larger isoforms containing 3 to 7additional spliced FnIII domains, and in particular isoforms comprising 5

Fig. 2. (A) Semi-quantitative RT-PCR analysis of time course changes in the expression of overall Tnc transcripts following cerebral cortex injury. RNA was prepared from controluninjured and injured cerebral cortex tissue dissected at 2, 4, 7 and 15 dpl. Total levels of overall Tnc transcripts were assessed by PCR amplification of the constitutive FnIII domains7 and 8 with the primer pair 7S and 8AS. GAPDH served as an internal control for relative amounts of RNA between the different samples. Left: Image of a gel with correspondingGAPDH amplification representative of 4 independent experiments. Right: The relative amount of total Tnc transcripts was quantified by densitometry. Results are expressed as themean±SEM fold changes in mRNA levels for Tnc in the lesioned tissue relative to equivalent unlesioned tissue, assigned a value of 1 (no change). The data of 4 independentexperiments, corresponding to 4 different RNA extracts were pooled. Significance of the changes were evaluated using Student's t test (*pb0.05; **pb0.01). (B) Regulation of thedifferent sized Tnc isoforms expressed after cerebral cortex injury. Left: Size range of Tnc isoforms was assessed by the amplification of the region flanking the splice site between theconstitutive FnIII domains 5 and 6. The positions of amplification products representing 0 to 7 FnIII inserts are depicted on the left of the image. An FnIII exon is 273 bp. The smallestband (532) corresponds to FnIII domains 5 and 6 without insertion of any additional spliced FnIII domain. Each additional FnIII domain adds 273 bp, i.e. +1 FnIII=805 bp, +2FnIII=1078 bp, and so on, up to the largest isoform containing all 7 FnIII domains (2443 bp); M=molecular weight marker; image of gel representative of 4 independentexperiments. Right: The relative amount of each sized Tnc isoform was quantified by densitometry. Results are expressed as above. Only small isoforms with no (open circles), one(open squares) or two (filled losanges) spliced FnIII domains were quantifiable. For the smallest isoform quantification was performed on nonsaturated images. Isoformwith no FnIIIinsert: pb0.01 for 2, 4 and 7 dpl, ns at 15 dpl; 1 insert: pb0.01 for 2, 4 and 15 dpl, pb0.05 for 7 dpl; 2 inserts: ns for all.


spliced FnIII domains, were transiently increased after injury at 2 to7 dpl, but remained underrepresented in comparison to forms with noor 1 FnIII insert (Fig. 2B).

Next, we used specific primer pairs (see Fig. 1 and Table 1) to studythe expression profile of specific FnIII domains or combinations afterinjury. Individual alternatively spliced FnIII domainswere detected onlyat low levels after RT-PCR of control mRNA (Fig. 3), in agreement withour results from uninjured tissue, indicating that the smallest isoformlacking all variable FnIII repeats is predominantly expressed in the adultuninjured cerebral cortex. By amplification of the single FnIII domain D(amplification of FnIII domain D or combination D6), we observed animportant increase of TncmRNA levels containing the FnIII module D inlesioned cortices, with a 4 fold peak increase at 4 dpl, before signalsreturned to control levels by 15 dpl (Fig. 3A). Isoforms comprising thedomain B were even more augmented after injury reaching a 15 foldpeak increase at 4 dpl (Fig. 3A). In contrast, global levels of isoformscontaining the FnIII domain C or AD1 remained almost unchanged afterinjury. Isoforms comprising the growth cone repulsive A1A2 domains,although transiently increased after injury, required more PCRamplification rates than the other spliced domains to be detected,suggesting that they were poorly expressed (Fig. 3A). Similar resultswere obtained for the A1A2A4 combination, while A1A4 was almostundetectable even after 41 PCR cycles (data not shown). Tnc isoformspossessing one alternatively spliced FnIII insert carry either the domain

DorA1 but noneof the other spliced FnIII domains are found exclusivelyinserted between the constant FnIII domains 5 and 6 (Joester andFaissner, 1999). To explore the regulation of the Tnc isoform bearingonly the single alternatively spliced domainD,we used FnIII 5 sense andFnIII D antisense primers and analysed the smallest PCR productobtained (see Fig. 1), which corresponds to the isoform where D isflanked by the constant FnIII domains 5 and 6. This PCR product wascalled 5D.Weobserved a dramatic increase of this formby about 13 foldat 2–4 dpl (Fig. 3A). The isoform containing only A1 as additional FnIIIinsert (smallest band after amplification with FnIII A1 sense and FnIII 6antisense primers and called A16; Fig. 1) was not detected at the sameamplification rate (data not shown), indicating that the isoform withonly D is the most frequent amongst the one spliced FnIII domain-containing Tnc isoforms after injury as observed previously duringdevelopment and in adult normal brain (Joester and Faissner, 1999). Inaddition, these results show that the isoformbearingonlyD accounts formost of the injury-induced upregulation of Tnc variants comprising onespliced FnIII domain.

We then analysed FnIII repeats for combinatorial expression byemploying primers flanking several spliced FnIII domains (see Fig.1 andTable 1). First, by using B sense and D antisense primers, we observed adramatic upregulation of isoforms containing the neurite outgrowth-promoting pair BDwith a peak augmentation of over 25 fold in lesionedtissue at 4 dpl as compared to unlesioned control (Fig. 3B).

Fig. 3. Regulation of specific alternatively spliced FnIII domains or combinations after CNS injury. RT-PCR analysis was performed with RNA prepared from uninjured and injuredcerebral cortex dissected at 2, 4, 7 and 15 dpl, using the specific primer pairs described in Table 1 and Fig. 1B. (A) Left: Images of the gels corresponding to PCR amplification of FnIIIdomains or combinations A1A2, B, AD1, C, D/D6 (D and D6 pooled) and 5D. Amplification with primers of FnIII 5 and Dmodules yielded different bands, the shortest one (called 5D)corresponding to the Tnc isoform containing only the domain D. Amplifications were performed with 35 cycles, except for A1A2 with 41 PCR cycles. Right: Densitometry: results areexpressed as the mean±SEM fold changes in mRNA levels of the different variants in the lesioned tissue relative to equivalent unlesioned tissue, assigned a value of 1 (no change).Data of 3 independent experiments were pooled, except for D/D6 (4 experiments). Significance of the changes were evaluated using Student's t test: for B: pb0.01 at 4 dpl andpb0.05 at 2 and 7 dpl; for D/D6: pb0.01 for all; for 5D pb0.01 for 2 and 4 dpl and pb0.05 for 7 dpl; for A1A2 pb0.01 at 4 and 7 dpl. (B) Left: Amplification with primers of FnIII B andD modules yielded three bands, the shortest one corresponding to the adjacent BD combination, an intermediate band corresponding to BCD (and not to BAD1D as the presence ofAD1 implies that of C) and the largest band is BAD1CD. Right: Densitometric analysis. Results are expressed as above. BD (filled circles), BCD (open circles), BAD1CD (filled squares). 4independent experiments were performed. For BD: pb0.01 for all; for BCD pb0.01 at 4 dpl and pb0.05 at 15 dpl; for BAD1CD pb0.01 at 7 dpl. (C) Left: Amplification with primers ofFnIII B and C domains produced two bands, the shortest one corresponding to the adjacent BC combination and the largest band is BAD1C. Amplification with the primer pair for Cand D produced a single band corresponding to the adjacent FnIII pair CD. Right: Densitometric analysis. Results are expressed as above. BC (filled circles), BAD1C (open circles), CD(open squares). 3 independent experiments were performed. For CD: pb0.05 at 4 and 15 dpl; for BC pb0.01 at 7 dpl and pb0.05 at 2 and 4 dpl; for BAD1C pb0.05 at 2 and 7 dpl.


Concomitantly, combinations where B and D are interrupted by C orAD1C (BCD; BAD1CD) were unchanged at 2–4 dpl and were slightlyincreased later at 7–15 dpl. Isoforms containing BC, CD or BAD1Ccombinations were increased by about 4 fold at 4 dpl but to a muchlesser extent than BD (Fig. 3C). Otherwise, isoforms bearing only theA1D repeat, a frequent combination amongst 2 FnIII spliced modules-containing Tnc variants (Joester and Faissner, 1999), was poorlyexpressed in control and injured brain (data not shown). Althoughamplifications with different primers cannot be used for quantitativecomparisons between different PCR reactions because of variations inthe PCR efficiency, a qualitative comparison of the intensities of thebands suggests that, among the domains and combinations tested,forms containing B, D, BD or CD are the most expressed after injury.Using a primer pair amplifying from FnIII domain AD1 to domain 6 we

obtained a single product corresponding to isoforms containing theAD1CD combination, suggesting that the expression of AD1 impliesthose of C andD. In contrast, amplification of domains 5 toAD1yielded 4bands according to the presence of A1, A2, A4 and B domains (data notshown). Altogether, these results demonstrate a differential expressionof Tnc isoforms and of alternatively spliced FnIII domains after corticalstab injury, with a strong increase of both the FnIII B and D domain andof the Tnc isoform containing only the spliced FnIII domain D. An evenmore spectacular increase was observed for Tnc forms comprising theneurite outgrowth-promoting BD combination where B and D areadjacent.

These findings prompted us to analyse the mRNA expression of twoTnc receptor molecules involved in neurite outgrowth promotion, F3/contactin and the integrin α7, which bind to the combination BD and


domain D, respectively (Rigato et al., 2002; Mercado et al., 2004). BothF3/contactin and integrin α7 mRNA were clearly present in the adultinjured cortex and were unchanged compared to control uninjuredtissue (data not shown).

Western Blot and immunostaining analysis of Tnc isoform variabilityafter CNS injury

To determine if differential regulation of the various Tnc isoformscould be observed at the protein level, we performed quantitativeWestern Blot analysis of cortical lesions. Injured tissue was dissectedfrom around cortical lesions in adult rats. Western Blots for Tnc displayseveral bands because the sizes of the Tncmonomers vary dependingonthe number (0 to 7 in rats) of alternatively spliced FnIII domains thatcompose the different isoforms, and because of glycosylation. Thetheoretical size of Tnc protein monomers ranges from 170 kD for theshortest up to 240 kD for the largest variant, but increases further due toN-glycosylation. Potential N-glycosylation sites of Tnc are moreabundant within the spliced FnIII domains, in particular in A1 and B(Gulcher et al., 1989; Faissner and Kruse, 1990; Joester and Faissner,2001; Rigato et al., 2002). Recombinant proteins formed of spliceddomains, such as the combination BD, are extensively glycosylatedwhen expressed in eucaryotic cells (Rigato et al., 2002).

First, we estimated total Tnc amounts using a rabbit polyclonalantibody (KAF14) recognizing all isoforms. Tnc immunoreactive bandswere detected with molecular weights (MW) ranging from about250 kD (shortest isoform without any FnIII insert) to 320 kD (largestisoform) (Fig. 4). It shouldbenoted that isoformsdiffering in sizeof theirpolypeptide chains might not be resolved as distinct entities due toheterogeneity of glycosylation.Nevertheless, large isoforms that containmore than 2 or 3 alternatively spliced inserts are likely tomigrate at thelevel of the smeary band observed at around 270–320 kD. This smearmight also include highly glycosylated smaller Tnc isoforms comprisingonly one or two additional FnIII domains, resulting from extensiveattachment of N-linked sugars to the spliced FnIII domains, asmentioned above. The intense lower band at about 250 kD correspondsmainly to the shortest isoform without any additional FnIII insert asconfirmed by its absence when the blots were probed with anti-FnIII Dantibody (Fig. 4).

Thus, themarked increase in the intensityof the smearybandaround270–320 kD, observed in lesions at 7–14 dpl (Tnc upregulation was

Fig. 4. Western Blot analysis of the post-lesional expression of Tnc and FnIII domains Dand A1A2A4 in rat cerebral cortex lesions. The left lanes are from control brain (Ct) anddissected from around cortical lesions at 7, 14 and 28 dpl. Lane E is a protein extractfrom E18 brain, lane Tnc is purified mouse tenascin-C (1 μg). Equal amounts of proteinextracts (30 μg per lane) were resolved and blots were immunolabeled with pAbKAF14, mAb 578, mAb 19H12. The blots presented are representative of threeindependent experiments.

observed already at 4 dpl; not shown) before returning to control levelsby 28 dpl, reflects an increase of various Tnc isoforms ranging fromthose containing one or two FnIII inserts to the largest ones (Fig. 4), inaccordance with the relative injury-induced increase of thecorresponding transcripts. The shortest isoform, migrating around250 kDwas already present at a high level in control uninjured cerebralcortex and was slightly increased after injury. This correlates with thesubstantial amounts of transcripts for the shortest Tnc variant found inuninjured cortex and their relativelymoderate increase after injury. Theshortest isoform was predominant in uninjured tissue but this was nolonger the case after injury. Weak signals under 250 kD are eitherunspecific or correspond to the shortest isoformwith less glycosylationor to proteolytic fragments of Tnc.

To examine the protein expression of specific alternatively splicedTnc molecules after cortical lesions, we used two rat monoclonalantibodies directed either against the FnIII domain D (mAb 578) or theFnIII combinationA1A2A4 (mAb19H12) forWesternBlotting. In controlextracts, FnIII domain D was almost undetectable. The amount of Tncproteins comprising the domain D (visible as a smear extending from300–320 kD for long isoforms to about 260–270 kD for shorter forms)was markedly increased 7 and 14 dpl before returning to control levelsat 28 dpl (Fig. 4). As expected no signal was detected at the position(250 kD) corresponding to the shortest isoform without any FnIIIalternative insert. This is consistentwith our RT-PCR results indicating astrong injury-induced upregulation of transcripts for Tnc variantscontaining the D domain, in particular when it is the unique additionaldomain or when it is adjacent to the domain B, as revealed by the verystrong increase of transcripts comprising the BD pair. In contrast, thedomain combination A1A2A4 remained undetectable in uninjured andin injured cerebral cortex (Fig. 4). These results indicate that the injury-induced differential regulations of Tnc isoforms we observed at themRNA level are reproduced at the protein level with, in particular, thestrong increase of D-containing isoforms after injury.

We also examined expression of Tnc and its isoforms by immuno-histochemistry at 7 dpl in mice. Tnc was just detectable abovebackground in unlesioned tissue but was significantly increased 7 dplafter cortical injury, mainly in the lesion core (Fig. 5A). Similarly, FnIII

Fig. 5. Immunofluorescence labeling of Tnc and the FnIII domain D after cortical lesions.Microtome sections of lesioned brains (7 dpl) were double labeled for Tnc (A) and GFAP(B). Labeling of the FnIII domain D is shown in C with corresponding control withoutprimary antibody in D. After lesion, both Tnc and FnIII domain D-contaning variantswere clearly increased and mostly expressed in the lesion core. Approximate extent ofthe lesion is demarcated by the dotted line. The scale bars represent 200 µm.


domain D-containing variants were elevated 7 dpl at the lesion site,whereas only background levels were detected in control uninjuredtissue (Figs. 5C andD). The stainingwith themAb578 (J1tn2) against thedomain D overlapped the staining obtained with polyclonal anti-Tncantibodies (batch KAF14), although it was slightly less pronounced, inparticular at the bottom of the lesion track. Polyclonal anti-Tnc antibodystaining was observed throughout the lesion down to the subventricularzone. Immunostaining for the FnIII A1A2A4 epitope did not reveal anyprotein expression at 7 dpl (data not shown). Around the injuries anastroglial reaction as reportedpreviouslywas seen,with a strong increasein GFAP immunoreactivity (Fig. 5B). The distribution of Tnc stainingappeared to be more restricted than the GFAP labeling, as noticedpreviously (Laywell et al., 1992), suggesting that only a subpopulation ofreactive astrocytes, at the lesion core, are induced to express Tnc. Sameresults were obtained in nigrostriatal injuries (data not shown).

RT-PCR analysis of the differential regulation of Tnc isoforms incytokine-treated astrocytes

Astrocytes are a predominant source of Tnc in the normal andinjured CNS (McKeon et al., 1991; Bartsch et al., 1992; Bartsch, 1996;Niquet et al., 1995; Deller et al., 1997; Hirsch and Bähr, 1999; Tang et al.,2003). In order to identify factors that might regulate Tnc isoformexpression in the injured CNS, we investigated the effects of severalinjury-related cytokines and growth factors on the expression ofdifferent Tnc isoforms in purified cultured astrocytes by RT-PCR andby Western Blotting.

Fig. 6. (A) Semi-quantitative RT-PCR analysis of the influence of various cytokines on the exptreated for 24 h with 1, 5 or 10 ng/ml of TGFβ1, TGFβ1 (5 ng/ml)+bFGF (10 ng/ml), bFGFmethods. Left: Total levels of overall Tnc transcripts were assessed by PCR amplification of trepresentative of at least 4 independent experiments. Right: Densitometric analysis. For eaclevels for Tnc in cytokine-treated astrocytes relative to control untreated astrocytes, assignedexpression of different sized transcripts. Left: Size range of Tnc isoformswas assessed by the a5 and 6. The positions of amplification products representing 0 to 3 FnIII inserts are depicteeach sized Tnc isoform was quantified by densitometry. Results are expressed as above. Onisoforms were difficult to detect. For the smallest isoform quantification was performed on ndifferent RNA extracts were pooled. Significance of the changes were evaluated using Stud

First, to study the influence of cytokines on the regulation of TncmRNA and of its splice variants, we conducted semi-quantitative RT-PCR analysis on RNA isolated from pure confluent astrocytes treatedfor 24 h with various cytokines in serum-free condition. Using aprimer pair amplifying the constitutively expressed domains 7 and8 (Fig. 1), we observed that TGFβ1 at 5 and 10 ng/ml induced an about2 fold upregulation of the overall Tnc transcripts in cultivatedastrocytes (Fig. 6A). The combination of 5 ng/ml of TGFβ1 and10 ng/ml of bFGF caused a similar 2 fold increase of the overall Tnctranscripts. In contrast, treatments of the astrocytes with 10 ng/ml ofbFGF alone or with EGF were without effect on the overall Tnc mRNAlevels (Fig. 6A). Also, PDGF and VEGF at 10 ng/ml had no significantinfluence on the amount of the overall Tnc mRNA (data not shown).

The size range of splice variants amongst the Tnc transcripts wasinvestigated using oligonucleotide primers for the constitutive FnIIIdomains 5 and 6 which flank the splice site for all of the isoformvariations (Fig. 1). Transcripts corresponding to the shortest isoformwithout any FnIII insert were the most abundant in astrocytescultivated in control conditions while mRNA of isoforms containing1 and 2 FnIII inserts were less expressed (Fig. 6B). Larger isoformswere hardly detected. These results are consistent with our previousobservations indicating that astrocytes in culture express relativelyfew of the large transcripts (Garwood et al., 2004). However, largeisoforms might be underestimated due to a possibly less efficient PCRamplification. After treatment with TGFβ1, astrocytes displayed adramatic increase (about 7–8 fold) of the mRNA expression of thesmallest Tnc variant. The effect of TGFβ1 was already observed at1 ng/ml, reached a maximum at 5 ng/ml and was not greater at

ression of overall Tnc transcripts in cultivated astrocytes. RNA was prepared from cells(10 ng/ml) or EGF (10 ng/ml) in serum-free conditions as described in Materials andhe constitutive FnIII domains 7 and 8 with the primer pair 7S and 8AS. The gel image ish cytokine treatment, results are expressed as the mean±SEM fold changes in mRNAa value of 1 (no change). (B) Analysis of the influence of the cytokines on the astrocyticmplification of the region flanking the splice site between the constitutive FnIII domainsd on the left of the image. M=molecular weight marker. Right: The relative amount ofly small isoforms with no, one or two spliced FnIII domains were quantifiable as largeronsaturated images. The data of at least 4 independent experiments, corresponding to 4ent's t test (*pb0.05; **pb0.01).


10 ng/ml in the presence or not of bFGF. In comparison bFGF and EGFexerted no significant influence (Fig. 6B). Tnc isoforms containing 1 or2 additional FnIII inserts were also increased by TGFβ1 but to a muchlesser extent by about 4–5 and less than 2 fold, respectively. Thecombination of 10 ng/ml of bFGF and 5 ng/ml of TGFzβ1 induced astronger increase by over 7 fold of the 1 FnIII insert-containingisoforms than 5 ng/ml of TGFβ1 alone, while 10 ng/ml of bFGF alonehad no significant effect, suggesting a potentiation of the TGFβ1 effectby bFGF. EGF and the other cytokines tested, including PDGF and VEGF(not shown), were without effect on these forms. Two close bandswere detected at the size corresponding to the isoforms comprising 1FnIII insert. In fact, only the upper band is specific for forms with oneadditional FnIII domain andwas quantified, whereas the lower band isan amplicon of parts of the FnIII domains 5 and C (von Holst et al.,2007) whose specificity is unknown.

Fig. 7. RT-PCR analysis of the regulation of specific spliced FnIII domains or combinations bytreated for 24 h with the different cytokines in serum-free condition as above, using thecorresponding to PCR amplification of FnIII domains or combinations A1A2, B, AD1, C, D/D6A1A2 with 40 PCR cycles. Right: Each PCR product was quantified by densitometry. Results acytokine-treated astrocyte relative to control untreated astrocytes, assigned a value of 1 (nbands, the shortest one corresponding to the adjacent BD combination, an intermediate banResults are expressed as above. (C) Left: Amplificationwith primers of FnIII B and Cmodules pthe largest band is BAD1C. Amplification with the primer pair for C and D produced a singleare expressed as above. Data are from at least 4 independent experiments. Significance of t

We then used primer pairs (Fig. 1 and Table 1) for individual repeatsor combinations of FnIII inserts to study the influence of variouscytokines on their astrocytic expression. mRNA levels of Tnc isoformscontaining the domain D (amplification of D or D6) were increased byaround 50% after treatment of the astrocytes with 5 ng/ml of TGFβ1withno synergistic actionof bFGFandTGFβ1. bFGF, EGF, PDGFandVEGFalone had no significant influence on mRNA levels of D-containing Tncisoforms (Fig. 7A). The FnIII domain B was also modulated by TGFβ1with a nearly 2 fold increase but the greatest effect was obtained withthe combination of TGFβ1+bFGF which induced a nearly 3 foldupregulation (Fig. 7A). In contrast, global levels of isoforms containingthe FnIII domain C or AD1 were not significantly augmented (less than30%) by TGFβ1, with or without bFGF, and by all other cytokines tested.Isoforms comprising the growth cone repulsive A1A2 domains,although augmented by TGFβ1 plus bFGF, required more PCR

cytokines in astrocytes. RT-PCR analysis was performed with RNA prepared from cellsspecific primer pairs described in Table 1 and Fig. 1B. (A) Left: Images of the gels(D and D6 pooled) and 5D. Amplifications were performed with 34 cycles, except for

re expressed as the mean±SEM fold changes in mRNA levels of the different variants ino change). (B) Left: Amplification with primers of FnIII B and D modules yielded threed corresponding to BCD and the largest band is BAD1CD. Right: Densitometric analysis.roduced two bands, the shortest one corresponding to the adjacent BC combination andband corresponding to the adjacent FnIII pair CD. Right: Densitometric analysis. Resultshe changes were evaluated using Student's t test (*pb0.05; **pb0.01).

Fig. 8. Western Blot analysis of the influence of various cytokines on Tnc isoformexpression by astrocytes. (A) The cells were treated for 96 h with the indicatedconcentrations of TGFβ1, or with 5 ng/ml of TGFβ1 and 10 ng/ml of bFGF, or with10 ng/ml of bFGF or EGF in serum-free condition. Concentrated supernatants weremeasured and for each treatment a 30 µg protein was separated and blots were probedwith pAb KAF14. (B) Cells were treated for 96 h with increasing concentrations up to20 ng/ml of TGFβ1, bFGF or TGFα. The blots were probed with pAb KAF14, mAb 578 ormAb 19H12. Note that the separation is higher than in (A) allowing the visualization ofdoublets at 300–320 kD and 240–260 kD as well as bands around 280 kD. The blots arerepresentative of at least 3 independent experiments.


amplification cycles than the other spliced domains to be detected,suggesting that theywerepoorly expressed in astrocytes (Fig. 7A). Sameresults were obtained for A1A2A4 combination, while A1A4was almostundetectable evenafter40PCR cycles (datanot shown). TheTnc isoformcontainingonly the single spliceddomainD(analysis of the smallest PCRproduct obtained after PCR with FnIII 5 sense and FnIII D antisenseprimers and called 5D) was strongly increased by about 3.5 fold aftertreatment of the astrocytes with 5 ng/ml of TGFβ1, and no furtherincrease was observed on treatment with TGFβ1 (5 ng/ml)+bFGF(10 ng/ml) (Fig. 7A). The isoform containing only A1 was not detectedat the same amplification rate (data not shown), indicating that theisoform with only D accounts for most of the TGFβ1-inducedupregulation of Tnc variants comprising one spliced FnIII domain inastrocytes.

We looked in more detail at some other specific FnIII combinations.First, using B sense and D antisense primers (Fig. 1 and Table 1), weobserved an almost 2.5 fold increase of isoforms containing the neuriteoutgrowth-promoting FnIII pair BD after treatment of the astrocyteswith TGFβ1, and a 4 fold increase on treatment with TGFβ1 and bFGF,revealing some synergistic effects of bFGF and TGFβ1 on the expressionof BD (Fig. 7B). bFGF and EGF at 10 ng/ml induced a moderate nonsignificant upregulation of BD. Isoforms containing the CD combinationwere also increased, but to amuch lesser extent thanBD, by about40% inTGFβ1 treated cells and by 70% when a combination of TGFβ1 and bFGFwas applied (Fig. 7C). Expression of BAD1C was only enhanced inresponse to the treatment with both TGFβ1 and bFGF as compared tountreated astrocytes, while BC remained unchanged in all conditionstested. Tnc variants comprising only the repeat A1D were poorlyexpressed. The qualitative comparison of the band intensities suggeststhat, among the domains and combinations tested, forms containing B,D, BD or CD are the most intensely expressed after TGFβ1 and TGFβ1plus bFGF treatments of astrocytes. Using a primer pair amplifying fromFnIII domain AD1 to domain 6 we obtained a single productcorresponding to isoforms containing the AD1CD combination, suggest-ing that the expression of AD1 implies the one of C and D. In contrast,amplification of domains 5 to AD1 yielded 4 bands according to thepresence of A1, A2, A4 and B domains (data not shown). These resultsdemonstrate a differential astrocytic regulation of alternative Tncisoforms in response to cytokine treatment in vitro.

Western Blot and ELISA analysis of Tnc isoform variability in cytokine-treated astrocytes

To study the influence of cytokines on astrocytic Tnc expression atthe protein level, we examined astrocyte-conditioned medium for totalTnc, FnIII domain D- and A1A2A4-containing variants after 96 h oftreatment with various cytokines in serum-free condition by WesternBlot analysis. Total Tnc amounts were visualized using the pAb KAF14.Tnc monomers migrated as two major bands at 240–260 kD and 300–320 kD, a typical pattern we observed previously for astrocytes(Garwood et al., 2004). In dependence of the migration distance duringelectrophoresis, bands occasionally divided into doublets. The signal at240–260 kD, which corresponds to short isoforms, including thesmallest one without any FnIII insert, was progressively elevated byTGFβ1 at concentrations starting from 0.5 to 5 ng/ml, but was lessincreased at higher concentrations (Figs. 8A and B). This observationimplies that the optimal concentration of TGFβ1 ranges around5 ng/ml.The combination of 5 ng/ml of TGFβ1 with 10 ng/ml of bFGF provedslightlymore efficient than TGFβ1 alone, indicating aweak potentiationof the TGFβ1 effect by bFGF (Fig. 8A). These results closely parallel theregulatory patternweobserved at themRNA level. The Tncbandat 300–320 kD, which represents mainly larger isoforms, was also increased byTGFβ1, but to amuch lesser degree than the 240–260 kD band (Figs. 8Aand B). Again, a weak potentiation of the TGFβ1 effect by bFGF could beobserved. bFGF and EGF/TGFα induced only a very limited increase ofthe intensity of the high molecular weight band and had no significant

influence on the 240–260 kD band (Figs. 8A and B). Consistent with thefact that they bind to the same receptor, EGF-R (Junier, 2000), EGF andTGFα exhibited identical effects.

Subsequently, we explored the influence of cytokines on theexpressionof isoforms containing the FnIII domainDor the combinationA1A2A4, using the rat mAbs 578 or 19H12, respectively (Fig. 8B).Isoforms comprising the FnIII domain Dwere hardly detected in controlcultures. After treatment of astrocytes with TGFβ1, we observed adramatic increase in FnIII domain D-containing Tnc isoforms sized 300–320 kD and around 240–260 kD. Again, the effect of TGFβ1 peaked at5 ng/ml, in agreement with our observations at the mRNA level. Theinducing capacity of TGFβ1 on the expression of FnIII D appeared morepronounced at the protein than at the mRNA level. This may reflect theaccumulation of the protein in themediumover the 96 h culture period.Similar to total Tnc expression, EGF/TGFα and bFGF induced a verymoderate elevation of solely the FnIII D domain-containing isoformsmigrating at 300–320 kD and exerted no influence on smaller forms(Fig. 8B). No FnIII A1A2A4 epitope was detected in the control orconditioned media, which implies that this combination is poorlyexpressed, in accordance with our RT-PCR results (data not shown).None of the factors tested had any significant influence on cell number(data not shown). The absence of any cell proliferation is presumablydue to contact inhibition in our confluent astrocytemonolayer cultures.

We have observed that Tnc forms comprising the domain D areincreased in the conditioned medium of TGFβ1/bFGF-treated astro-cytes. To measure if domain D-containing Tnc isoforms are enrichedwithin the extracellular matrix derived from TGFβ1 (5 ng/ml)+bFGF(10 ng/ml)-stimulated astrocytes, we used enzyme-linked immuno-sorbent assay (ELISA). First, we measured the level of overall Tncisoforms detected in the matrix by using the pAb KAF14. We observedan about 2 fold increase of Tnc in the matrix deposited by astrocytescultivated for 96 h in the presence of TGFβ1/bFGF compared to


unstimulated cells (Fig. 9). Chondroitin sulfate proteoglycans (CSPG)are known to be modulated by TGFβ1 or bFGF and to interact withvarious matrix components including Tnc and, thus, we compared theresults obtained on chondroitinase ABC-treated to untreatedmatrix. Nosignificant difference was observed, indicating that chondroitin sulfateswere not interfering on Tnc detection. Then, we used the mAb 578 todetect specifically the D-containing Tnc isoforms present in the matrix.We observed a significant 3 to 4 fold increase of the amount of D-containing Tnc isoforms detected in the matrix deposited by astrocytescultivated for 96 h in the presence of TGFβ1/bFGF compared tounstimulated cells. This greater increase of D-comprising Tnc forms(about 3–4 fold) than total Tnc (2 fold) is in accordancewith our RT-PCRand Western Blot results.

Altogether, these results suggest that TGFβ1+bFGF could mediate,at least in part, the injury-induced accumulation of D-containing Tncisoforms.

Discussion

Expression of the Tnc isoforms and FnIII domains after CNS injury

Tnc has been implicated in many cellular processes, including celladhesion, migration, proliferation and axon growth and guidance.Importantly, the alternatively spliced FnIII repeats between FnIIIdomains 5 and 6 have specific functions, and may thus play specificroles in these processes. We have therefore investigated the pattern ofexpressionof Tnc splice variants after CNS injury, and found that there isa change in expression of the different alternatively spliced FnIII repeatsand resulting Tnc isoforms. The shortest isoform dominated in theresting adult CNS. After injury, the relative abundance of larger formscontaining alternatively spliced FnIII domainswas augmented,whereas,in comparison, the shortest variant without FnIII insert appeared lessmodulated. As a consequence the shortest isoform without FnIII insertwasno longer thepredominant form in injured tissue, as observed at theprotein level, although it remained expressed at a high level.

Fig. 9. ELISA analysis of the Tnc content in the extracellular matrix derived fromTGFβ1+bFGF-treated astrocytes. Confluent astrocytes were prepared as de-scribed in Materials and methods and were treated for 96 h with 5 ng/ml of TGFβ1and 10 ng/ml of bFGF in serum-free condition. At the end of the culture the cellswere lysed and, after washing, the astrocyte-deposited matrix attached to theplastic dish was treated (+) or not (−) with the chondroitinase ABC to removechondroitin sulfates. Then, the ELISA analysis was performed using either pAbKAF14 or mAb 578. Results are expressed as the mean±SEM fold changes in theELISA signal for total Tnc or Tnc variants comprising the domain D in the differentconditions relative to the condition corresponding to untreated astrocytes andwithout chondroitinase ABC treatment of the matrix (set at 1). Data are from 6(total Tnc) and 5 (FnIII D domain) independent experiments. Significance of thechanges was evaluated using the Student's t test (*pb0.001 for Tnc and pb0.01 forFnIII D domain).

Interestingly, large Tnc isoforms are also found in regions of thedeveloping CNS in which axon growth and tissue remodeling isoccurring (Tucker, 1993; Joester and Faissner, 2001). Recent studieshave shown that the subventricular zone also responds to CNS lesions(Richardsonet al., 2007) andTnc is expressed in theadult subventricularzone (Gates et al., 1995). Yet, it is presently not knownwhether corticallesions have an impact on the Tnc isoform composition of thesubventricular zone. In a previous study, we have shown an upregula-tion of large Tnc variants in neural stem cells overexpressing theneurogenic transcription factor Pax6 and cultivated as neurospheres(von Holst et al., 2007), but it is not known whether Pax6 is induced inthe adult subventricular zone after injury.

Amajor point is our demonstration that the expression of the splicedFnIII domains is differentially regulated after injury, with strongincreases of the domains B and D, whereas the expression of thedomains AD1 and C was almost unchanged. Our most spectacularobservation is the 25 fold injury-induced increase of the spliced FnIIIpair BD-containing forms. Interestingly, this injury-induced expressionof BD-containing isoforms is reminiscent of a developmental expressionas the prevalence of variants containing adjacent FnIII domains B and Dwere shown to be high in the embryonic hippocampus beforedecreasing towards postnatal ages (Rigato et al, 2002). However,despite the upregulation of both domains, the formation of the FnIII BDcombination appeared more due to the increase in FnIII B than FnIII Drepeats with the FnIII domain D being already more abundant in theuninjured cortex, based on our RT-PCR findings. This supports findingsby Joester and Faissner (1999) who showed that in the adult brain, thedomain D is almost the only additional domain represented in mRNATnc variants. Nevertheless, our results at the protein level demonstratethat D-containing isoforms are poorly found in undamaged adult brainbut are dramatically elevated after injury in the form of both long andshort isoforms. In addition, our RT-PCR data indicate that isoformscontaining only the single domain D are strongly increased after CNSinjury. Interestingly, FnIII domain Dwas shown to be present inmost ofthe splice variants in the developing brain (Joester and Faissner, 1999;Rigato et al., 2002).

Tnc transcripts comprising A1A2 or A1A2A4 combinations wereincreased after injury but their detection required more amplificationcycles than the other spliced domains and, at the protein level, A1A2A4was undetectable in control and injured tissue, suggesting that variantswith A1A2/A1A2A4 combinations are poorly expressed. Transcriptscontaining the combination of FnIII domains A1A2 were also shown tobe relatively rare during hippocampal development (Rigato et al., 2002),in accordance with our results showing very few A1A2A4-containingTnc proteins in rat E18 brain.

Regulation of the Tnc isoforms and FnIII domains expression in astrocytesby cytokines

TGFβ1 is strongly induced by activated macrophages/microglia andreactive astrocytes after CNS injury and is involved in glial scarformation andextracellular remodeling (Smith andHale, 1997; Flanderset al., 1998). TGFβ1 increased overall Tnc production by astrocytes, inaccordance with previous observations by Smith and Hale (1997).Stimulation of Tnc expression by TGFβ1 has also been observed in othertissues and cell types such asfibroblasts (Jones and Jones, 2000) or bonecells (Mackie et al., 1998). Here, we demonstrate that TGFβ1 alsodifferentially regulates the astrocytic expression of Tnc splice variantsand alternative FnIII repeats. First, TGFβ1 induced a more prominentincrease of small, in particular isoforms with no or one spliced domain,than large Tnc isoforms in astrocytes, as was observed in other cells ortissues such as fibroblasts or lung (Tucker et al., 1993; Zhao and Young,1995). Second, we showed that TGFβ1 induced increases in domain D-,as well as B-, containing Tnc forms expressed by astrocytes and also avery strong upregulation of BD-containing forms, but little change inexpressionof theAD1andCdomains. A strong increasewas revealed for


single domain D-containing isoforms. On a protein level, we observed avery strong increase of FnIII domain D both in short and long Tnc afterTGFβ1 treatment. Moreover, we observed a strong increase of D-containingTnc isoforms in theextracellularmatrix derived fromTGFβ1/bFGF-treated astrocytes. Our mRNA and protein data therefore arelargely in agreement. bFGF is another cytokine that is upregulated afterCNS injury (Logan et al., 1992; Smith et al., 2001). In contrast to TGFβ1,bFGF induced a moderate but specific increase of long Tnc isoforms inastrocytes as reported before by others in vivo and in vitro (Tucker et al.,1993; Meiners et al., 1993; Mahler et al., 1997; Smith and Hale, 1997;Suzuki et al., 2002). Compared toTGFβ1 the influence of bFGFon specificspliced FnIII domains wasminimal, except for BD that was increased by75%. However, we observed that bFGF slightly potentiated the TGFβ1-induced mRNA upregulation of isoforms comprising the BD combina-tion and the B domain. Synergistic effects of bFGF and TGFβ1 weredescribed previously for the upregulation of overall Tnc protein inastrocytes and fibroblasts (Tucker et al., 1993; Smith and Hale, 1997).Thus, our results suggest that TGFβ1 in cooperation with bFGF could, atleast partly, mediate the injury-induced upregulation of Tnc and of theB-, D- and BD-containing isoforms.

Role of the different Tnc isoforms in axonal regeneration

With regard to stimulation of regeneration, the enhancedexpression of the alternatively spliced domain D after CNS injury isof particular interest as it was shown to promote neurite outgrowthfrom various neuron types including cerebral cortical neurons (Götzet al., 1996, 1997; Meiners et al., 1999, 2001; Mercado et al., 2004). Apeptide promoting neurite elongation was identified within thedomain D (Meiners et al., 2001) but sustained process elongation wasobserved when FnIII domain D was associated with the spliced FnIIIdomain B or the constitutive FnIII module 6 (Rigato et al., 2002). TheFnIII repeat pair BD has been shown to promote neurite outgrowthfrom embryonic hippocampal neurons through an interaction withthe cell adhesion molecule F3/contactin, but only when B and D arenot interrupted by domain C (Rigato et al., 2002), whereas the neuriteoutgrowth-promoting effect of the D6 pair on cerebellar granule andcerebral cortical neurons was shown to be mediated by the α7β1integrin (Mercado et al., 2004). Thus, the strong injury-inducedincrease of the expression of isoforms containing the BD or D6 (as FnIIID and 6 are obligatorily adjacent in Tnc molecules) combinationsmight help to create an environment promoting axonal regrowth.

In contrast to the domain D, the expression of domain C remainedalmost unchanged after injury. In choice assays the C domain has beenshown to provide permissive neurite guidance cues capable ofovercoming the barrier to neurite advance formed by the FnIIIcombination A1A2A4, the shortest Tnc variant and also by variouschondroitin sulfate proteoglycans (Meiners et al., 1999; Liu et al., 2005).The lack of a substantial increase of isoforms comprising theC domain inthe lesioned region could, therefore, contribute to the fact that, despite amassive increase of the neurite extension promoting pair BD,regeneration is very limited in the CNS, as axons fail to be guided acrossthe inhibitory terrain of the glial scar characterized by the accumulationof inhibitory chondroitin sulfate proteoglycans (Rhodes and Fawcett,2004). In addition, isoforms comprising the growth cone repellingdomains A1A2A4 (Götz et al., 1996), although expressed at a low level,were increased after injury. This could also contribute to impede axonalregrowth. Moreover, the smallest Tnc variant without FnIII insertsremained strongly expressed at lesions andwas even slightly increased.This might also contribute to create less favorable environment forregrowth as the smallest Tnc isoform was shown to be repulsive toadvancing growth cones, a property attributed to the epidermal growthfactor (EGF) domains, that is strongly attenuated in the large isoforms(Dörries et al., 1996; Götz et al., 1996; Meiners et al., 1999). Theexpression in injuries of Tnc isoforms with growth cone deflecting andboundary forming properties is consistent with the observation that,

after entorhinal lesion, axons failed to invade the Tnc rich denervatedouter molecular layer of the rat fascia dentata, although axons presentwithin the molecular layer were sprouting (Deller et al., 1997). Also,terminal sprouting of mossy fibres proved impeded in territories withincreased Tnc immunoreactivity of kainate-treated hippocampus(Niquet et al., 1995). Boundaries formed by Tnc are also observedduring development (Faissner and Steindler, 1995; Treloar et al., 2009).In summary, the Tnc isoforms' expression profile observed after injurysuggests complex effects with on one side neurite outgrowth-promoting properties associated, in particular with the strong D andBD upregulations, and, on the other hand, growth cone repellant effectsdue to thehighexpressionof isoformswithout spliced FnIII domainsandto the relatively poor expression of domain C.

The rate of local axonal regrowthwill not depend only on the ratio ofaxon growth stimulatory and inhibitory Tnc isoforms but also on theirrespective localization. Indeed, localized expression of specific Tncisoforms and spliced FnIII domains might define neuronal micro-environments as permissive or inhibitory for axonal regeneration andguidance. Astrocytes are heterogeneous and their properties vary withthe distance to the lesion (Hoke and Silver, 1994) and might thereforeproduce different Tnc variants with the distance to the lesion site.Another source of local diversity among the Tnc variants produced afterinjury could be provided by reactive oligodendrocyte precursors orNG2-positive cells which develop after injury (Levine et al., 2001) andwere shown to produce Tnc (Garwood et al., 2004). It would be ofinterest to determine whether individual cells in vivo express differentTnc isoforms after injury. This would have to be achieved by in situhybridization analysis with probes specific for the different isoforms.Unfortunately, so far it was not possible to construct specific probesdistinguishing the different Tnc variants, presumably because of cross-reactions. In fact, any alternatively spliced FnIII domain is shared byseveral isoforms. In addition, the influence of the Tnc isoforms onregeneration might depend on their mode of presentation to cells andon their interactions with other extracellular matrix componentspresent in CNS injuries such as the chondroitin sulfate proteoglycans,neurocan and phosphacan, or diverse heparan sulfate proteoglycans(Meiners and Geller, 1997; Milev et al., 1997; Asher et al., 2000;Garwood et al., 2001; Dobbertin et al., 2003). Interestingly, thesemolecules were shown to interact with various Tnc domains and mightmask different active sites for neurite growth and guidance within theTnc molecule. Hence, if it is conceivable that accumulation of D-containing Tnc variants at the lesion site exert locally a positiveinfluence onaxonal sprouting, it is alsopossible that anybeneficial effectof these Tnc variants might be obliterated by extracellular matrixcomponentsmasking the axon growth permissive domain D of Tnc. Tnceffects on neuronal process regeneration may also depend on the ageand type of the neurons. For example, Tnc has been shown to promotethe elongation of axons from embryonic rat hippocampal neurons butdoes not stimulate the axon outgrowth from chicken retinal neurons invitro (Faissner, 1997).Otherwise, axonal regeneration in response toTncmight depend on the neuronal expression level for the different Tncreceptors. The integrin α7β1 which binds the FnIII domain D (Mercadoet al., 2004) was shown to be elevated in damaged peripheral neurons,where it was connected to successful regrowth of peripheral injuredaxons, but it was not increased in CNS neurons (Werner et al., 2000).Thus, it is conceivable that a lack or low expression of integrin α7β1 inCNS neurons might limit their axonal regeneration in response to theexposure of Tnc isoforms containing the domain D. Consistent withthese findings, α7β1 was expressed, but not increased, in our stabwound model. Similarly, we observed that the adhesion molecule F3/contactin, which links FnIII pair BD (Rigato et al., 2002),was present butunaltered after CNS injury. However, further experiments dedicated tothe expression of F3/contactin on injured neuronswould be required tounravel the role of this Tnc receptor in regeneration.

Our results should help to further orientate in vivo investigations tomanipulate and evaluate the respective roles of specific Tnc isoforms in


CNS injuries. Beyond their influence on axonal growth, different Tncisoforms and spliced FnIII domains have been shown to modulate cellmigration, survival and proliferation of various CNS cell types such asoligodendrocyte precursors and astrocytes (Joester and Faissner, 2001;Garwood et al., 2004). Changes in Tnc isoforms in the damaged tissueprobably therefore will have multiple effects.

Acknowledgments

The work presented in this study has been supported by the GermanResearch Council (DFG, SPP-1048 “Molecular and Cellular Basis of CNSRepair” Fa159/11-1,2,3). Ursula Theocharidis is supportedby theGermanResearch Council (DFG, GRK-736 “Development and Plasticity of theNervous System”). Stefan Czvitkovich was funded by a Wellcome TrustPhD studentship and by a grant from the Christopher and Dana Reevefoundation. Funding is also acknowledged from the Medical ResearchCouncil, The Henry Smith Charity and Action Medical Research.

References

Asher, R.A., Morgenstern, D.A., Fidler, P.S., Adcock, K.H., Oohira, A., Braistead, J.E., Levine,J.M., Margolis, R.U., Rogers, J.H., Fawcett, J.W., 2000. Neurocan is upregulated ininjured brain and in cytokine-treated astrocytes. J. Neurosci. 20, 2427–2438.

Bartsch, S., Bartsch, U., Dorries, U., Faissner, A., Weller, A., Ekblom, P., Schachner, M.,1992. Expression of tenascin in the developing and adult cerebellar cortex. J.Neurosci. 12, 736–749.

Bartsch, U., 1996. The extracellular matrix tenascin-C: expression in vivo and functionalcharacterization in vitro. Prog. Neurobiol. 49, 145–168.

Brodkey, J.A., Laywell, E.D., O'Brien, T.F., Faissner, A., Stefansson, K., Dorries, H.U.,Schachner, M., Steindler, D.A., 1995. Focal brain injury and upregulation of adevelopmentally regulated extracellular matrix. J. Neurosurg. 82, 106–112.

Chiquet-Ehrismann, R., Chiquet, M., 2003. Tenascins: regulation and putative functionsduring pathological stress. J. Pathol. 200, 488–499.

Chomczynski, P., Sacchi, N., 1987. Single step method of RNA isolation by acidguanidium thiocyanate phenol chloroform extraction. Anal. Biochem. 162,156–159.

Deller, T., Haas, C.A., Naumann, T., Joester, A., Faissner, A., Frotscher, M., 1997. Up-regulation of astrocyte-derived tenascin-C correlates with neurite outgrowth in therat dentate gyrus after unilateral entorhinal cortex lesion. Neuroscience 81,829–846.

Derr, L.B., Chiquet-Ehrismann, R., Gandour-Edwards, R., Spence, J., Tucker, R.P., 1997.The expression of tenascin-C with the AD1 variable repeat in embryonic tissues, celllines and tumors in various vertebrate species. Differentiation 62, 71–82.

Dobbertin, A., Rhodes, K.E., Garwood, J., Properzi, F., Heck, N., Rogers, J.H., Fawcett, J.W.,Faissner, A., 2003. Regulation of RPTPbeta/phosphacan expression and glycosami-noglycan epitopes in injured brain and cytokine-treated glia. Mol. Cell. Neurosci.24, 951–971.

Dörries, U., Taylor, J., Xiao, Z., Lochter, A., Montag, D., Schachner, M., 1996. Distincteffects of recombinant tenascin-C domains on neuronal cell adhesion, growth coneguidance, and neuronal polarity. J. Neurosci. Res. 43, 420–438.

Faissner, A., 1997. The tenascin gene family in axon growth and guidance. Cell TissueRes. 290, 331–341.

Faissner, A., Kruse, J., 1990. J1/tenascin is a repulsive substrate for central nervoussystem neurons. Neuron 5, 627–637.

Faissner, A., Steindler, D., 1995. Boundaries and inhibitory molecules in developingneural tissues. Glia 13, 233–254.

Flanders, K.C., Ren, R.F., Lippa, C.F., 1998. Transforming growth factor-βs inneurodegenerative disease. Prog. Neurobiol. 54, 71–85.

Garcion, E., Faissner, A., French-Constant, C., 2001. Knockout mice reveal a contributionof the extracellular matrix molecule tenascin-C to neural precursor proliferationand migration. Development 128, 2485–2496.

Garwood, J., Rigato, F., Heck, N., Faissner, A., 2001. Tenascin glycoproteins and thecomplementary ligand DSD-1-PG/phosphacan-structuring the neural extracellularmatrix during development and repair. Restor. Neurol. Neurosci. 19, 51–64.

Garwood, J., Garcion, E., Dobbertin, A., Heck, N., Calco, V., ffrench-Constant, C., Faissner,A., 2004. The extracellular matrix glycoprotein tenascin-C is expressed byoligodendrocyte precursor cells and required for the regulation of maturationrate, survival and responsiveness to platelet-derived growth factor. Eur. J. Neurosci.20, 2524–2540.

Gates, M.A., Thomas, L.B., Howard, E.M., Laywell, E.D., Sajin, B., Faissner, A., Götz, B.,Silver, J., Steindler, D.A., 1995. Cell and molecular analysis of the developing andadult mouse subventricular zone of the cerebral hemispheres. J. Comp. Neurol. 361,249–266.

Götz, B., Scholze, A., Clement, A., Joester, A., Schuette, K., Wigger, F., Frank, R., Spiess, E.,Ekblom, P., Faissner, A., 1996. Tenascin-C contains distinct adhesive, anti-adhesive,and neurite outgrowth promoting sites for neurons. J. Cell. Biol. 132, 681–699.

Götz, M., Bolz, J., Joester, A., Faissner, A., 1997. Tenascin-C synthesis and influence onaxonal growth during rat cortical development. Eur. J. Neurosci. 9, 496–506.

Gulcher, J.R., Nies, D.E., Marton, L.S., Stefansson, K., 1989. An alternatively spliced regionof the human hexabrachion contains a repeat of potential N-glycosylation sites.Proc. Natl. Acad. Sci. U. S. A. 86, 1588–1592.

Hirsch, S., Bähr, M., 1999. Immunocytochemical characterization of reactive optic nerveastrocytes and meningeal cells. Glia 26, 36–46.

Hoke, A., Silver, J., 1994. Heterogeneity among astrocytes in reactive gliosis. Perspect.Dev. Neurobiol. 2, 269–274.

Husmann, K., Faissner, A., Schachner, M., 1992. Tenascin promotes cerebellar granulecell migration and neurite outgrowth by different domains in the fibronectin typeIII repeats. J. Cell. Biol. 116, 1475–1486.

Joester, A., Faissner, A., 1999. Evidence for combinatorial variability of tenascin-Cisoforms and developmental regulation in the mouse central nervous system. J.Biol. Chem. 274, 17144–17151.

Joester, A., Faissner, A., 2001. The structure and function of tenascins in the nervoussystem. Matrix Biol. 20, 13–22.

Jones, P.L., Jones, F.S., 2000. Tenascin-C in development and disease: gene regulationand cell function. Matrix Biol. 19, 581–596.

Junier, M.P., 2000. What role(s) for TGFα in the central nervous system? Prog.Neurobiol. 62, 443–473.

Laywell, E.D., Dörries, U., Bartsch, U., Faissner, A., Schachner, M., Steindler, D., 1992.Enhanced expression of the developmentally regulated extracellular matrixmolecule tenascin following adult brain injury. Proc. Natl. Acad. Sci. U. S. A. 89,2634–2638.

Levine, J.M., Reynolds, R., Fawcett, J.W., 2001. The oligodendrocyte precursor cell inhealth and disease. Trends Neurosci. 24, 39–47.

Liu, H.Y., Nur-E-Kamal, A., Schachner, M., Meiners, S., 2005. Neurite guidance by the FnCrepeat of human tenascin-C: neurite attraction vs. neurite retention. Eur. J.Neurosci. 22, 1863–1872.

Lochter, A., Vaughan, L., Kaploni, A., Prochiantz, A., Schachner, M., Faissner, A., 1991. J1/tenascin in substrate-bound and soluble form displays contrary effects on neuriteoutgrowth. J. Cell. Biol. 113, 1159–1171.

Logan, A., Frautschy, S.A., Gonzales, A.M., Baird, A., 1992. A time course for the focalelevation of synthesis of basic fibroblast growth factor and one of its high-affinityreceptors (flg) following a localized cortical brain injury. J. Neurosci. 12,3828–3837.

Mackie, E.J., Abraham, L.A., Taylor, S.L., Tucker, R.P., Murphy, L.I., 1998. Regulation oftenascin-C expression in bone cells by transforming growth factor-beta. Bone 22,301–307.

Mahler, M., Ben-Ari, Y., Represa, A., 1997. Differential expression of fibronectin,tenascin-C and NCAMs in cultured hippocampal astrocytes activated by kainate,bacterial lipopolysaccharide or basic fibroblast growth factor. Brain Res. 775,63–73.

McKeon, R.J., Schreiber, R.C., Rudge, J.S., Silver, J., 1991. Reduction of neurite outgrowthin amodel of glial scarring is correlated with the expression of inhibitory moleculeson reactive astrocytes. J. Neurosci. 11, 3398–3411.

Meiners, S., Geller, H.M., 1997. Long and short splice variants of human tenascindifferentially regulate neurite outgrowth. Mol. Cell. Neurosci. 10, 100–116.

Meiners, S., Marone, M., Rittenhouse, J.L., Geller, H.M., 1993. Regulation of astrocytictenascin by basic fibroblast growth factor. Dev. Biol. 160, 480–493.

Meiners, S., Mercado, M.L.T., Nur-e-Kamal, M.S.A., Geller, H.M., 1999. Tenascin-Ccontains domains that independently regulate neurite outgrowth and neuriteguidance. J. Neurosci. 19, 8443–8453.

Meiners, S., Nur-e-Kamal, M.S.A., Mercado, M.L.T., 2001. Identification of a neuriteoutgrowth-promoting motif within the alternatively spliced region of humantenascin-C. J. Neurosci. 21, 7215–7225.

Mercado, M.L., Nur-e-Kamal, A., Liu, H.Y., Gross, S.R., Movahed, R., Meiners, S., 2004.Neurite outgrowth by the alternatively spliced region of human tenascin-C ismediated by neuronal α7β1 integrin. J. Neurosci. 24, 238–247.

Milev, P., Fischer, D., Haring, M., Schulthess, T., Margolis, R.K., Chiquet-Ehrismann, R.,Margolis, R.U., 1997. The fibrinogen-like globe of tenascin-C mediates itsinteractions with neurocan and phosphacan/protein-tyrosine phosphatase-zeta/beta. J. Biol. Chem. 272, 15501–15509.

Nakic, M., Mitrovic, N., Sperk, G., Schachner, M., 1996. Kainic acid activates transientexpression of tenascin-C in the adult rat hippocampus. J. Neurosci. Res. 44, 355–362.

Niquet, J., Jorquera, I., Faissner, A., Ben-Ari, Y., Represa, A., 1995. Gliosis and axonalsprouting in the hippocampus of epileptic rats are associated with an increase oftenascin-C immunoreactivity. J. Neurocytol. 24, 611–624.

Prieto, A.L., Jones, F.S., Cunningham, B.A., Crossin, K.L., Edelman, G.M., 1990. Localizationduring development of alternatively spliced forms of cytotactin mRNA by in situhybridization. J. Cell. Biol. 111, 685–698.

Rhodes, K.E., Fawcett, J.W., 2004. Chondroitin sulphate proteoglycans: preventingplasticity or protecting the CNS? J. Anat. 204, 33–48.

Richardson, R.M., Sun, D., Bullock, M.R., 2007. Neurogenesis after traumatic brain injury.Neurosurg. Clin. N. Am. 1, 169–181.

Rigato, F., Garwood, J., Calco, V., Heck, N., Faivre-Sarrailh, C., Faissner, A., 2002. Tenascin-C promotes neurite outgrowth of embryonic hippocampal neurons through thealternatively spliced fibronectin type III BD domains via activation of the celladhesion molecule F3/Contactin. J. Neurosci. 22, 6596–6609.

Smith, G.M., Hale, J.H., 1997. Macrophage/microglia regulation of astrocytic tenascin:synergistic action of transforming growth factor-beta and basic fibroblast growthfactor. J. Neurosci. 17, 9624–9633.

Smith, C., Berry, M., Clarke, W.E., Logan, A., 2001. Differential expression of fibroblastgrowth factor-2 and fibroblast growth factor receptor 1 in a scarring andnonscarring model of CNS injury in the rat. Eur. J. Neurosci. 13, 443–456.

Suzuki, S., Li, A.J., Ikemoto, M., Imamura, T., 2002. Expression of tenascin-C longisoforms is induced in the hypothalamus by FGF-1. Neuroreport 13, 1041–1045.


Tang, X., Davies, J.E., Davies, S.J., 2003. Changes in distribution, cell associations, andprotein expression levels of NG2, neurocan, phosphacan, brevican, versican V2, andtenascin-C during acute to chronic maturation of spinal cord scar tissue. J. Neurosci.Res. 71, 427–444.

Treloar, H.B., Ray, A., Dinglasan, L.A., Schachner, M., Greer, C.A., 2009. Tenascin-C is aninhibitory boundary molecule in the developing olfactory bulb. J. Neurosci. 29,9405–9416.

Tucker, R.P., 1993. The in situ localization of tenascin splice variants and thrombos-pondin 2 mRNA in the avian embryo. Development 117, 347–358.

Tucker, R.P., Hammarback, J.A., Jenrath, D.A., Mackie, E.J., Xu, Y., 1993. Tenascin expressionin themouse: in situ localization and induction in vitro by bFGF. J. Cell Sci. 104, 69–76.

von Holst, A., Egbers, U., Prochiantz, A., Faissner, A., 2007. Neural stem/progenitor cellsexpress 20 tenascin C isoforms that are differentially regulated by Pax6. J. Biol.Chem. 282, 9172–9181.

Werner, A., Willem, M., Jones, L.L., Kreutzberg, G.W., Mayer, U., Raivich, G., 2000.Impaired axonal regeneration in alpha7 integrin-deficient mice. J. Neurosci. 20,1822–1830.

Zhang, Y., Winterbottom, J.K., Schachner, M., Lieberman, A.R., Anderson, P.N., 1997.Tenascin-C expression and axonal sprouting following injury to the spinal dorsalcolumns in the adult rat. J. Neurosci. Res. 49, 433–450.

Zhao, Y., Young, S.L., 1995. TGF-beta regulates expression of tenascin alternative-splicing isoforms in fetal rat lung. Am. J. Physiol. 268, L173–L180.