Rapid Communication

NT-3 Delivered by an Adenoviral VectorInduces Injured Dorsal Root Axons toRegenerate Into the Spinal Cord of Adult RatsYi Zhang,1 Paul A. Dijkhuizen,2 Patrick N. Anderson,1* A. Robert Lieberman,1and Joost Verhaagen21Department of Anatomy and Developmental Biology, University College London, Gower Street,London, England2Netherlands Institute for Brain Research, Amsterdam-ZO, The Netherlands

Sensory axons interrupted in the dorsal roots of adultmammals are normally unable to regenerate into thespinal cord. We have investigated whether the intro-duction of a neurotrophin gene into the spinal cordmight offer an approach to otherwise intractablespinal root injuries. The dorsal roots of the 4th, 5th,and 6th lumbar spinal nerves of adult rats weresevered and reanastomosed. Fourteen to nineteendays later, adenoviral vectors containing either theLacZ or NT-3 genes were injected into the ventralhorn of the lumbar spinal cord, resulting in strongexpression of the transgenes in glial cells and motorneurons between 4 and 40 days after injection. Whendorsal root axons were transganglionically labelledwith HRP conjugated to cholera toxin subunit B, 16 to37 days after dorsal root injury, large numbers oflabelled axons could be seen to have regenerated intothe cord, but only in those animals injected withvector carrying the NT-3 gene. The regenerated axonswere found at the injection site, mainly in the greymatter, and had penetrated as deep as lamina V. Genetherapy with adenoviral vectors encoding a neuro-trophin has therefore been shown to be capable ofenhancing and directing the regeneration of a subpopu-lation of dorsal root axons (probably myelinated Afibres), into and through the CNS environment. J.Neurosci. Res. 54:554–562, 1998.r 1998 Wiley-Liss, Inc.

Key words: gene therapy; neurotrophin; axon regen-eration, dorsal root injury

INTRODUCTIONInjury to dorsal roots, particularly those connected

to the brachial plexus or forming part of the cauda equina,is a serious clinical problem that is difficult to treat. Thisrelates to the fact that, in adult mammals following dorsal

root injury, the central processes of dorsal root ganglion(DRG) neurons regenerate successfully in the peripheralnervous tissue of the dorsal roots but stop abruptly at thedorsal root entry zone (DREZ) where CNS glial cells arepresent (Reier et al., 1983; Carlstedt, 1985) and axonalgrowth is apparently inhibited. The regenerating axonseither turn and grow back along the root or form swollenend bulbs abutting DREZ astrocytes (Reier et al., 1983;Carlstedt, 1985; Liuzzi and Lasek, 1987; Stensaas et al.,1987). The regeneration of the central processes of DRGneurons can be enhanced by also interrupting theirperipheral processes (Richardson and Issa, 1984; Richard-son and Verge, 1987; Chong et al., 1996, and unpublishedobservations), following which small numbers of axonssucceed in growing back into the cord; however most ofthe axons are still unable to reenter the spinal cord andthose that succeed, penetrate no further than the superfi-cial dorsal horn (Chong et al., unpublished observations).

It is becoming clear that neurotrophins can stimu-late axonal regeneration even within the non-permissiveenvironment of the spinal cord (Schnell et al., 1994;Oudega and Hagg, 1996; Grill et al., 1997). Neuro-trophin-3 (NT-3) is a trophic factor for large diameter 1aprimary afferent neurons (McMahon et al., 1994; Wrightand Snider, 1995; Oakley et al., 1997) whose centralprocesses normally end on motor neurons deep in thegrey matter of the ventral horn. In adult mammals, motor

Contract grant sponsor: Medical Research Council; Contract grantsponsor: Netherlands Organisation for Scientific Research.

Paul Dijkhuisen and Yi Zhang contributed equally to this work.

*Corresponding author: Dr. Patrick Anderson, Department of Anatomyand Developmental Biology, University College London, GowerStreet, London WC1E 6BT, England. E-mail: p.anderson@ucl.ac.uk

Received 5 June 1998; Revised 11 August 1998; Accepted 19 August1998

Journal of Neuroscience Research 54:554–562 (1998)

r 1998 Wiley-Liss, Inc.

neurons express very low levels of NT-3 (Schectersonand Bothwell, 1992 and see Fig. 1a) and can thereforeoffer little trophic or tropic stimulation to regeneratingdorsal root axons. A source of NT-3 in the deep greymatter of the spinal cord might therefore be expected toenhance the regeneration of 1a afferents, but it is ex-tremely difficult to provide a continuous source by directapplication of the neurotrophin. In vivo gene therapy,however, offers a potential solution to this problem. Wereport here the successful deployment of an adenoviralvector to deliver NT-3 into glial cells and motor neuronsof the spinal cord and to promote the regeneration ofNT-3 sensitive dorsal root axons and their extensiveingrowth into the spinal cord.

MATERIALS AND METHODSSurgical Procedures

Adult female Sprague-Dawley rats (200–300 g)were anaesthetised with halothane and oxygen. The leftlumbar spinal roots were exposed by a hemilaminectomy.The L4, 5, and 6 dorsal roots were identified under anoperating microscope and individually transected about15 mm from their entrance into the cord. Each severedroot was then reanastomosed under an operating micro-scope by using 10/0 microsutures. Bovine fibrinogen wasused to ensure additional stability of the ligation. In someanimals ventral roots L4, 5, and 6 were also sectioned toprevent labelling of motor neurons by retrograde tracer(see below). The skin and muscle incisions were closed inlayers. In order to enhance the regenerative potential ofthe injured dorsal roots, which contain the centrallydirected axons of the primary sensory neurons withperipheral processes in the sciatic nerve, the left sciaticnerve was exposed at mid-thigh level and crushed severaltimes with watchmakers’ forceps. This was done in allanimals immediately following the dorsal root surgery.

Construction, Evaluation and Injectionof Adenoviral Vectors

Adenoviral vectors were constructed according tomethods described by Dijkhuizen et al. (1997). Thevectors contained either NT-3 cDNA (Ad-NT-3) or LacZcDNA (Ad-LacZ) in both cases under control of the CMVpromoter. Adenoviral vectors were purified by doublecaesium chloride density gradient and dialysed againstvirus buffer (10 mM Tris pH 7.5, 135 mM NaCl, and 1mM MgCl2 with 10% glycerol). Concentrations of theviral vectors are expressed as plaque forming units(pfu)/ml. The bioactivity of Ad-NT-3 was tested onembryonic DRG explants in vitro (Dijkhuizen et al.,1997). Following infection with 53106 pfu Ad-NT-3,DRG explants displayed outgrowth similar to explants to

which 100 ng/ml NT-3 protein had been added. Forinjection, the vectors were diluted with virus buffer to aconcentration of 23107 pfu/µl.

Fourteen to nineteen days after dorsal root injury,animals were again deeply anaesthetised with halothaneand the L4,5 segments of spinal cord were exposed by afurther laminectomy. The dura was opened using finemicrosurgical scissors. A fine glass cannula with an outerdiameter of 70 µm, fixed in a stereotaxic frame andattached via a polyethylene tube to a 10 µl Hamiltonsyringe, was used to deliver unilateral injections ofadenoviral vector. Each animal was given two injectionsof Ad-NT-3 or Ad-LacZ or a mixture of both. The twoinjection sites were immediately to the left of the centralvein, at a depth of 1.5 mm below the surface of the cordand about 5 mm apart along the rostro-caudal axis of theL4,5 segments. The adenoviral vectors in virus buffer (seeabove) were injected slowly and the needle was left inplace for one minute after completing the injection beforebeing withdrawn. The muscle and skin incisions wereclosed with sutures. The left sciatic nerve was thenre-exposed and crushed several times with watchmakers’forceps at or just above the level of the original crush tofurther enhance the regenerative response of the injureddorsal root axons. The animals were divided into threegroups. Group one animals (n58) were injected with 0.5µl of Ad-LacZ containing 13107 pfu. Group two animals(n54) were injected with 0.5 µl of Ad-NT-3 containing14107 pfu. Group three animals (n53) were injectedwith 0.5 µl of a mixture of Ad-NT-3 and Ad-LacZcontaining 7.53106 Ad-NT-3 pfu and 2.53106 Ad-LacZpfu. In addition, in a preliminary experiment to assess theefficiency of infection under these conditions, identicalinjections of Ad-NT-3 or Ad-LacZ were made in a groupof animals (n54) without dorsal root or sciatic nerveinjury. All animals received an injection of dexametha-sone (5 mg/kg) intramuscularly before the virus injectionwas begun, and daily injections of the same dose ofdexamethasone over the following three days, to suppressthe immune response provoked by the virus. The effectsof dexamethasone are described in detail elsewhere(Hermens and Verhaagen, 1998). Briefly, dexamethasonestrongly inhibits but does not completely eliminate amacrophage response, extends the period of transgeneexpression in glial cells but not neurons, and does so tothe same extent with both Ad-LacZ and Ad-NT-3 vectors.

In Situ HybridisationFour to 12 days after Ad-NT-3 injection (n54),

animals were killed by anaesthetic overdose with Sagatal(60–90 mg/kg). The lumbar spinal cord was quicklyremoved and frozen on dry ice. Frozen transverse sec-tions cut at a nominal thickness of 25 µm were collectedon slides and fixed with 4% paraformaldehyde in 0.1 M

NT-3 Induces Dorsal Root Regeneration 555

phosphate buffer saline (PBS). Standard in situ hybridisa-tion with a digoxygenin labelled cRNA probe was carriedout as described previously (Dijkhuizen et al., 1997).

X-gal StainingFour to forty days (n55) after Ad-LacZ injection,

animals were deeply anaesthetized with halothane andperfused transcardially with 4% paraformaldehyde in 0.1M PBS at pH 7.4 and at room temperature. The lumbarspinal cord was removed, post-fixed for 2 hr in the samefixative, cryoprotected overnight in 20% sucrose, andfrozen on dry ice. Cryosections were cut at a nominalthickness of 25 µm and reacted with X-gal (5 mMK4Fe(CN)6, 5 mM K3Fe(CN)6, 2 mM MgCl2, and 1mg/ml 5-bromo-4-chloro-3-indolyl-D-Galactoside) for 3hr at 37°C.

Application of CT-HRP Tracer to the Sciatic Nerveand Visualisation of HRP

Sixteen to thirty-seven days after virus injection theanimals were re-anaesthetized with halothane and 0.5 µlof 1.4% cholera toxin subunit-B-HRP conjugate (CT-HRP) was injected into the left sciatic nerve using a 5 µlHamilton syringe. Two to three days after CT-HRPinjection the animals were lightly anaesthetised withhalothane and overdosed with Sagatal (60–90 mg/kg) byintraperitoneal injection. The animals were perfusedtranscardially with 100 ml of 0.1 M PBS (pH 7.4)followed by 500 ml of 1% paraformaldehyde and 1.25%glutaraldehyde in 0.1 M PBS. The L4, 5, and 6 segmentsof spinal cord were identified by tracing the sciatic nerveinto the dorsal rami and into the dorsal horn. A block ofspinal cord encompassing the L4–6 segments was re-moved and postfixed for 2 hr in the same fixative and thenimmersed in 30% sucrose in 0.1 M PBS at 4°C overnight.Transverse sections were cut on a freezing microtome atthickness settings of 40–50 µm and collected into 0.1 MPBS. The free-floating sections were processed for thevisualisation of HRP using the tetramethyl benzidinemethod (Mesulam, 1982). Sections were mounted seri-ally on clean gelatin-coated slides and left to air dry. Thesections were then dehydrated, mounted under a cover-slip in DPX (BDH), and viewed by brightfield anddarkfield microscopy.

RESULTSEfficiency of Infection

Preliminary experiments, to test the efficiency ofinfection with the adenoviral vectors and to determine theexpression pattern of the infected genes, were conductedon animals without dorsal root or sciatic nerve injury.

Between 4 and 40 days after injection of Ad-NT-3 orAd-LacZ into the ventral horn, many glial cells and somemotor neurons within 2.5 mm of the injection sitesexpressed high levels of the transgenes as shown by insitu hybridisation in animals injected with Ad-NT-3 andX-gal histochemistry in animals injected with Ad-LacZ(Fig. 1b–d). Expression of both NT-3 and LacZ wasmaximal between 4 and 12 days after injection and haddeclined but was still strong at 40 days (data not shown).

Efficacy of the NT-3 Transgene in PromotingIngrowth of Regenerating Dorsal Root Axons

Having established that the vectors infect ventralhorn cells with high efficiency and that expression of thetransgene remains high for at least 6 weeks, similarinjections were made into the spinal cords of rats 14–19days after transection and reanastomosis of the left L4–6dorsal roots and crush of the left sciatic nerve. Between16 and 47 days after injection of vectors containingAd-LacZ (group 1) Ad-NT-3 (group 2) or a mixture ofboth (group 3) and 2 to 3 days after injection of CT-HRPinto the left sciatic nerve, transverse sections of the L4–6segments of the cord, processed for visualisation of HRP,were examined by bright and dark field microscopy. In allanimals in groups 2 and 3 (which received Ad-NT-3 or amixture of Ad-NT-3 and Ad-LacZ) numerous transgangli-onically labelled axons were present in the dorsal rootscentral to the transection site (Figs. 2b, 3) and in thedorsal horn at the level of the injections (Figs. 2c, 3). Thiswas observed at 19 days (n52), 26 days (n51), 33 days(n52), and 39 days (n52) after adenovirus vectorinjection.

Many of the labelled axons that could be followedfrom the root into the cord appeared to avoid the DREZand penetrate the cord along the surface of blood vessels(Figs. 2d, 3) and from its dorsal and dorsolateral surfaces.The labelled axons were often seen to be concentrated inthe vicinity of the injection tract (Fig. 2e) and extendedfor several hundred micrometers up to a maximum of1,100 µm into the dorsal horn and to a limited extent intothe lateral white column. In the dorsal horn they formedplexuses in laminae II-IV and some extended as far aslamina V. However, although labelled axons regeneratedinto regions which contain the dendrites of motor neu-rons, they did not grow into the ventral horn itself (Figs.2e, 3) even though motor neurons and glia in that regionwere infected. Labelled axons did not grow rostrallyalong the dorsal column, in keeping with previousevidence that myelinated tracts in the CNS are usuallyinhibitory to regenerating axons (Caroni et al., 1988;Schnell et al.,1994; but see the recent work of Davies etal., 1997). The extent of reinnervation of the dorsal hornand of penetration of axons into the grey matter was

556 Zhang et al.

Fig. 1. a–d: Transverse sections of rat spinal cords processedfor the visualization of NT-3 mRNA by in situ hybridisation(a,b) X-gal histochemistry (c,d). The spinal cords illustrated arefrom animals 4 days (b,c) and 12 days (d) following Ad-NT-3(b) injection or Ad-LacZ injection (c,d). a: Absence of NT-3message on the uninjected side of the spinal cord. b: Many cells,

including both large putative motor neurons and smaller glialcells, were infected by Ad-NT-3 and expressed high levels ofNT-3 mRNA. c,d: Both motor neurons and glial cells expresshigh levels of Lac-Z transgene at 4 days (c) and 12 days (d)following Ad-LacZ injection. VH, ventral horn. Scale bar5500µm for all panels.

Fig. 2. a–f: Transverse sections of the dorsolateral part of thespinal cord showing regenerating axons transganglionicallylabelled with HRP in the dorsal root and/or dorsal horn 19 days(b–e) or 27 days (f) following injections of a mixture ofAd-NT-3 and Ad-LacZ (b–e) or Ad-LacZ (f) into the left ventralhorn of the segments of L4,5 spinal cord. The left L4,5, and 6dorsal roots were sectioned and reanastomosed 17 days beforeinjection with Ad-NT-3 (b–f) and 19 days before injection withAd-LacZ (f). The diagram in a shows the location andorientation of the photomicrographs. In b numerous transgangli-onically labelled axons are present in the dorsal root and someof them have grown into the spinal cord. Note that almost all the

labelled axons in the dorsal root appear to be arrested at theDREZ. c: HRP-labelled axons growing directly into the cord. d:Labelled axons growing into the cord apparently around bloodvessels. e: HRP-labelled axons are seen extending into the deepparts of the dorsal horn (lamina V). In adjacent sections thenarrow tract of axons extending ventrally from the dorsalsurface of the cord was seen to be close to and aligned with theinjection needle tract. f: Labelled axons are present in the dorsalroot, but all appear to be arrested at the DREZ; no labelledaxons can be seen growing into the cord. DR, dorsal root; DH,dorsal horn; DC, dorsal column. Scale bars5100 µm.

558 Zhang et al.

Figure 3. (Legend appears on page 560.)

NT-3 Induces Dorsal Root Regeneration 559

assessed by measuring, in all animals, the area of cordoccupied by labelled axons and the distance from thepoint of entry to the tips of the most ventrally locatedlabelled axons; the results of this semi-quantitative analy-sis are summarised in Table 1. These measurements weremade several months after the material was prepared,when some fading of HRP reaction product had occurred.Thus the figures probably underestimate both the extentof ingrowth and depth of penetration of regeneratingaxons.

In contrast, very few or no labelled axons grew intothe spinal cord in the group 1 animals injected withAd-LacZ (Fig. 2f) at 16 days (n51), 19–20 days (n52),27 days (n51), 33–34 days (n52), 40 days (n51), or 47days (n51) after virus injection. The small numbers offibres that did enter the cord, in one such animal,penetrated no deeper than lamina I and II of the dorsalhorn (Table 1). Clearly, therefore, in the absence of NT-3expression in the ventral horn, dorsal root fibres do notregenerate into the cord in large numbers and the few thatenter do not penetrate beyond the superficial dorsal horn.

DISCUSSIONThese results demonstrate, for the first time, that an

increased level of NT-3 expression, resulting from theintroduction of the NT-3 gene into spinal cord cells,promotes the regeneration of the central axons of primarysensory axons into the cord and deep into the grey matterof the dorsal horn. The observation that regeneratingaxons were concentrated within 0.8 mm of the area foundto show NT-3 transgene expression indicates that over-expression of NT-3 not only acts as a trophic stimuluspromoting regeneration into the cord, but also acts as atropic stimulus, attracting what are assumed to be NT-3-sensitive axons. Direct infusions of NGF have also beenshown to attract peripheral axons (Conner and Varon,1995; Oudega and Hagg, 1996). The NT-3 produced bydelivery of adenoviral vectors to target areas within thespinal cord is apparently able to overcome the non-permissive or inhibitory environmental influences whichnormally prevent regenerating axons from growing intothe spinal cord. This adds to the growing number ofobservations that NT-3 can promote axonal regenerationin the spinal cord (Schnell et al.,1994; Grill et al., 1997).However, most of the regenerating axons did not growinto the cord through the DREZ, which together withobservations on the limited ingrowth into the cord ofaxons of fetal DRG neurons transplanted to the site ofadult DRGs (Kozlova et al., 1995) perhaps indicates thestrength of the inhibitory influence found in this region(Pindzola et al., 1993). The common association betweenregenerating dorsal root axons and blood vessels is notsurprising in the light of previous observations of closeassociation between blood vessels and regenerating axonsgrowing into (Kozlova et al., 1997) and within the CNS(Campbell et al., 1992; Curtis et al., 1993).

The regenerating axons did not spread widely in thegrey matter away from the injection sites. The latterobservation suggests that the tropic effect of the NT-3expressed immediately around the injection sites was toostrong or that other inhibitory influences may be at workin the spinal grey matter (Keynes and Cook, 1995; Hokeand Silver, 1996). An obvious example of possible

Fig. 3. (See page 559.) Camera lucida drawings of labelledregenerating dorsal root axons in the dorsal left quadrant of thelumbar spinal cord from the same animal illustrated in Fig.2b–e. The four drawings are of transverse sections encompass-ing approximately 500–600 µm of the 1.1–1.5 mm length ofcord analysed in a series of 27 sections. The dorsal root (DR),dorsal horn (DH), and dorsal column (DC) are identified, as aresome of the blood vessels (shaded) with the walls of which,small bundles of regenerating axons appear to be closelyassociated. Note that the regenerating axons penetrate the cordover a considerable rostrocaudal length and along a widemediolateral extent and although this material is not suitable forcounting individual axons, the number of such axons is clearlyvery large. Note also that the axons remain largely within thedorsal horn, which theypenetrate to a depth of approximately 1mm. Scale bar5500 µm.

TABLE I. Assessment of Regenerative Ingrowth of Dorsal RootAxons Into the Spinal Cord in Groups 1–3*

Area of DH withregeneratingaxons µm2

Depth ofpenetrationof DH by

regeneratingaxons (µm)

Group 1 (Ad-LacZ) N5 8 0–35.7 0–129Group 2 (Ad-NT-3) N5 3 1346 49 7156 65Group 3 (Ad-LacZ and Ad-NT-3)

N 5 3 1586 85 7536 313

*The extent of regenerative growth of L4–6 dorsal root axons into thespinal cord following transection and reanastomosis was assessed inevery animal in all three groups, with the exception of one animal of thefour in group 2 in which fading of HRP reaction product at the time ofanalysis made assessment impossible, reducing the number included inthis analysis to 3. In all of these animals CT-HRP had been injected intothe sciatic nerve to label regenerating dorsal root axons. Serialtransverse sections of the lumbar cord were examined and for eachanimal the section showing the maximal extent of invasion of the cordby regenerating axons was selected and the area of regenerating axonswithin and adjacent to the dorsal horn was outlined and measured. Alsomeasured was the maximum depth to which regenerating axons hadgrown from their point of entry. The results for animals in groups 2 and3 were averaged and are shown in the Table as means6 standarddeviations. In group 1 seven of the eight animals did not display ameasurable degree of axonal ingrowth into the cord and one animalshowed a modest ingrowth. The results for this group are thereforepresented as a range rather than as an average. DH, dorsal horn.

560 Zhang et al.

inhibitory influences is provided by the expression of thechemorepulsive protein semaphorin D/collapsin I byadult motor neurons (Giger et al., 1996) and althoughcollapsin I does not inhibit the growth of NT-3-sensitiveaxons during development (Pueschel et al., 1996) othersemaphorins or members of other families of inhibitorymolecules could be present in the ventral horn. Despitethe fact that the regenerating sensory axons did not extendthroughout their normal target region but were confinedto areas around the injection sites, they reached areas ofthe spinal grey matter which contain the dendrites ofmotor neurons and were consequently in a position wherereformation of contacts with such dendrites might resultin the restoration of spinal reflexes.

Clearly, in order to use gene therapy technology torestore the function of injured dorsal roots several furthersteps will be required. First, vectors carrying NGF andpossibly BDNF will be required in addition to thosecarrying NT-3, in order to promote the regeneration ofother classes of primary afferent fibre into the cord. It willbe interesting to see if NGF-sensitive axons regeneratinginto the cord selectively reinnervate the substantia gelati-nosa. Second, a vector with less cytotoxity than theadenovirus will be required since this virus elicits aninflammatory response in CNS tissue and some motorneurons were undoubtedly lost under the conditions ofthe present experiment (Byrnes et al., 1995; Hermens andVerhaagen, 1998). Adeno-associated virus vectors andlentiviral vectors seem particularly promising in thisrespect (Kaplitt et al., 1994; McCown et al., 1996;Blomer et al., 1997). None the less, our work hasdemonstrated that the problem of dorsal root repair isamenable to gene therapy with viral vectors carryingneurotrophin genes.

ACKNOWLEDGMENTSWork in the laboratory of P.N. Anderson and A.R.

Lieberman was funded by the Medical Research Counciland in the laboratory of J. Verhaagen by the NetherlandsOrganisation for Scientific Research.

REFERENCES

Blomer U, Naldini L, Kafri T, Trono D, Verma IM, Gage F (1997):Highly efficient and sustained gene transfer in adult neuronswith a lentivirus vector. J Virol 71:6641–6649.

Byrnes AP, Rusby JE, Wood MJA, Charlton HM (1995): Adenovirusgene transfer causes inflammation in the brain. Neuroscience66:1015–1024.

Campbell G, Lieberman AR, Anderson PN, Turmaine M (1992):Regeneration of adult rat CNS axons into peripheral nerveautografts: Untrastructural studies of the early stages of axonalsprouting and regenerative axonal growth. J Neurocytol 21:755–787.

Carlstedt T (1985): Regenerating axons form nerve terminals atastrocytes Brain Res 347:188–191.

Caroni P, Schwab M (1988): Two membrane protein fractions from ratcentral myelin with inhibitory properties for neurite growth andfibroblast spreading. J Cell Biol 106:1281–1288.

Chong MS, Woolf CJ, Turmaine M, Emson PC, Anderson PN (1996):Intrinsic versus extrinsic factors in determining the regenerationof the certral processes of rat dorsal root ganglion neurons: Theinfluence of a peripheral nerve graft. J Comp Neurol 370:97–104.

Conner JM, Varon S (1995): Effects of exogenous nerve growth factorupon sympathetic sprouting into the hippocampal formation.Exp Neurol 136:123–135.

Curtis R, Green D, Lindsay RM, Wilkin GP (1993): Up-regulation ofGAP-43 and growth of axons in rat spinal cord after compres-sion injury. J Neurocytol 22: 51–64.

Davies SJA, Fitch MT, Memberg SP, Hall AK, Raisman G, Silver J(1997): Regeneration of adult axons in white matter tracts of thecentral nervous system. Nature 390:680–683.

Dijkhuizen PA, Hermens WTJMC, Teunis MA, Verhaagen J (1997):Adenoviral vector-directed expression of neurotrophin-3 in ratdorsal root ganglion explants results in a robust neurite out-growth response. J Neurobiol 33:172–184.

Giger RJ, Wolfer DP, De Wit GM, Verhaagen J (1996): Anatomy of ratsemaphorin III/collapsin-1 mRNA expression and relationshipto developing nerve tracts during neuroembryogenesis. J CompNeurol 735:378–392.

Grill R, Murai K, Blesch A, Gage FH, Tuszinsky MH (1997): Cellulardelivery of neurotrophin-3 promotes corticospinal axonal growthand partial functional recovery after spinal cord injury. JNeurosci 14:5560–5572.

Hermens WTJMC, Verhaagen J (1998): Suppression of inflammationby dexamethasone prolongs adenoviral vector mediated trans-gene expression in the facial nucleus of the rat. Brain Res Bull47: in press.

Hoke A, Silver J (1996): Proteoglycans and other repulsive moleculesin glial boundaries during development and regeneration of thenervous system. Prog Brain Res 108:149–163.

Kaplitt MG, Leone P, Samulski RJ, Xiao X, Pfaff DW, O’Malley KL,During MJ (1994): Long-term gene expression and phenotypiccorrection using adeno-associated virus vectors in the mamma-lian brain. Nature Genet 8:148–154.

Keynes RJ, Cook GM (1995): Repulsive and inhibitory signals. CurrOpin Neurobiol 5:75–82.

Kozlova EN, Rosario CM, Stromberg I, Bygdeman M, Aldskogius H(1995): Peripherally grafted human foetal dorsal root ganglioncells extend axons into the spinal cord of adult host rats bycircumventing dorsal root entry zone astrocytes. NeuroRep6:269–272.

Kozlova EN, Seiger A, Aldskogius H (1997): Human dorsal rootganglion neurons from embryonic donors extend axons into thehost rat spinal cord along laminin-rich peripheral surroundingsof the dorsal root transitional zone. J Neurocytol 26:811–822.

Liuzzi FJ, Lasek RJ (1987): Astrocytes block axonal regeneration inmammals by activating the physiological stop pathway. Science237:642–645.

McCown TJ, Xiao X, Li J, Breese GR, Samulski RJ (1996): Differen-tial and persistent expression patterns of CNS gene transfer byan adeno-associated virus (AAV) vector. Brain Res 713:99–107.

McMahon SB, Armanini MP, Ling LH, Phillips HS (1994): Expressionand coexpression of Trk receptors in subpopulations of adultprimary sensory neurons projecting to identified peripheraltargets. Neuron 12:1161–1171.

NT-3 Induces Dorsal Root Regeneration 561

Mesulam MM (1982): ‘‘Tracing Neural Connections With HorseradishPeroxidase.’’ IBRO Handbook Series: Methods in the Neurosci-ences. New York: Wiley-Interscience.

Oakley RA, Lefcort FB, Clary DO, Reichardt LF, Prevette D,Oppenheim RW, Frank E (1997): Neurotrophin-3 promotes thedifferentiation of muscle spindle afferents in the absence ofperipheral targets. J Neurosci 17:4262–4274.

Oudega M, Hagg T (1996): Nerve growth factor promotes regenerationof sensory axons into adult rat spinal cord. Exp Neurol140:218–229.

Pindzola RR, Doller C, Silver J (1993): Putative inhibitory extracellu-lar matrix molecules at the dorsal root entry of the spinal cordduring development and after root and sciatic nerve lesions. DevBiol 156:34–48.

Pueschel AW, Adams RH, Betz H (1996): The sensory innervation ofthe mouse spinal cord may be patterned by differential expres-sion of and differential responsiveness to semaphorins. Mol CellNeurosci 7:419–431.

Reier PJ, Stensaas LJ, Guth L (1983): The astrocytic scar as animpediment to regeneration in the certral nervous system. In

Kao CC, Bunge RP, Reier PJ (eds): ‘‘Spinal Cord Reconstruc-tion.’’ New York: Raven Press, pp 163–165.

Richardson PM, Issa VMK (1984): Peripheral injury enhances centralregeneration of primary sensory neurones. Nature 309:791–793.

Richardson PM, Verge VMK (1987): Axonal regeneration in dorsalspinal roots is accelerated by peripheral axonal transection.Brain Res 411:406–408.

Schecterson LC, Bothwell M (1992): Novel roles for neurotrophins aresuggested by BDNF and NT-3 mRNA expression in developingneurons. Neuron 9:449–463.

Schnell L, Schneider R, Kolbeck R, Barde YA, Schwab M (1994):Neurotrophin-3 enhances sprouting of corticospinal tract duringdevelopment and after adult spinal cord lesion. Nature 367:170–173.

Stensaas LJ, Partlow LM, Burgess PR, Horch KW (1987): Inhibition ofregeneration: The ultrastructure of reactive astrocytes andabortive axons terminals in the transition zone of the dorsal root.Prog Brain Res 71:457–468.

Wright DE, Snider WD (1995): Neurotrophin receptor mRNA expres-sion defines distinct populations of neurons in rat dorsal rootganglia. J Comp Neurol 351:329–338.

562 Zhang et al.