ã European Neuroscience Association European Journal of Neuroscience, Vol. 11, pp. 3539±3551, 1999 Axotomy results in major changes in BDNF expression by dorsal root ganglion cells: BDNF expression in large trkB and trkC cells, in pericellular baskets, and in projections to deep dorsal horn and dorsal column nuclei G. J. Michael1, S. Averill1, P. J. Shortland1, Q. Yan2 and J. V. Priestley1 1 Neuroscience Section, Division of Biomedical Sciences, St. Bartholomew's and The Royal London School of Medicine and Dentistry, Queen Mary and West®eld College, London, E1 4NS, UK 2 Amgen Inc., Amgen Centre, Thousand Oaks, California 91320±1789, USA Keywords: autocrine, neurotrophin, rat, sciatic, ultrastructure Abstract Brain derived neurotrophic factor (BDNF) is normally expressed by a small number of predominantly trkA-expressing dorsal root ganglion cells. Using immunocytochemistry and in situ hybridization, we have examined the effect of sciatic nerve section on the expression of BDNF in the adult rat. Following axotomy there was a long lasting (4-week) increase in BDNF mRNA and protein in large-diameter, trkB- and trkC-expressing dorsal root ganglion cells. By 2 days postaxotomy, expression of BDNF mRNA had increased from 2% of trkB cells to 50%, and from 18% of trkC cells to 56%. In contrast, BDNF expression in most trkA cells was unchanged, although was increased in the small population of medium- and large-sized trkA cells. Following axotomy, BDNFimmunoreactive terminals appeared in the central axonal projections of large-diameter cells, including the deep dorsal horn and gracile nucleus. Neuropeptide Y was also upregulated following axotomy and was coexpressed with BDNF in the cell bodies and central terminals of the large cells. Ultrastructural analysis in lamina IV of the spinal cord revealed that BDNF terminals in these central projections establish synaptic contacts. Immunoreactivity at 4 weeks was also observed in pericellular baskets that contained calcitonin gene-related peptide (CGRP) and surrounded trkA- and trkB-expressing cells in L4 and L5 lumbar ganglia. These baskets are likely to arise from local, highly immunoreactive, BDNF/CGRP/trkA-expressing cells. Our results identify several novel targets for BDNF and imply that it acts locally in both autocrine and paracrine modes, as well as centrally in a synaptic mode, to modulate the response of somatosensory pathways in nerve injury. Introduction Adult dorsal root ganglion (DRG) cells are not only acted on by a variety of target-derived and locally-derived trophic factors but also make the neurotrophin brain-derived neurotrophic factor (BDNF). Brain-derived neurotrophic factor is synthesized by small nociceptive DRG cells that express the nerve growth factor (NGF) receptor trkA and respond to NGF with increased BDNF synthesis (Apfel et al., 1996; Cho et al., 1997; Michael et al., 1997). The BDNF is packaged into dense-cored vesicles and anterogradely transported, together with neuropeptides such as calcitonin gene-related peptide (CGRP), into axon terminals in the dorsal horn of the spinal cord (Zhou & Rush, 1996; Michael et al., 1997). The function of BDNF in trkA cells is not known, but one possibility is that it is released from central terminals to act on second-order neurons in the spinal cord. Brain-derived neurotrophic factor applied to the spinal cord induces c-fos expression (Bennett et al., 1996) and is also likely to have acute effects such as phosphorylation of NMDA receptors (Lin et al., 1998). Brain-derived neurotrophic factor may therefore be responsible for the changes in spinal cord excitability seen in in¯ammation. Increased production of NGF is thought to play a key role in Correspondence: Professor J. V. Priestley, as above. E-mail: j.v.priestley@qmw.ac.uk Received 4 February 1999, revised 17 May 1999, accepted 26 May 1999 in¯ammatory pain (McMahon et al., 1997) and one of its effects is to increase BDNF expression (Cho et al., 1997). In addition to increasing in trkA cells in response to NGF, BDNF synthesis is greatly increased following nerve damage. Dorsal root crush (Ernfors et al., 1993), sciatic nerve crush (Ernfors et al., 1993; Sebert & Shooter, 1993; Tonra et al., 1998) and sciatic nerve transection (Karchewski et al., 1995; Tonra et al., 1998) all lead to large increases in BDNF mRNA. Brainderived neurotrophic factor is thus unusual amongst the products of DRG cells in being dramatically increased in response to both in¯ammation and axotomy. Because long-lasting changes in spinal cord excitability also occur with nerve injury (for review see King & Thompson, 1995), one possibility is that BDNF plays a role in such changes. Alternatively, BDNF may be released locally to act as an autocrine survival factor for axotomised adult DRG cells (Acheson et al., 1995). However, the cell types which upregulate BDNF after axotomy are not known. In order to clarify some of these issues we have used in situ hybridization and immunocytochemistry to examine the effects of axotomy on BDNF expression in DRG cell bodies and central terminals. A preliminary report of some of this work has appeared in abstract form (Averill et al., 1997). 3540 G. J. Michael et al. Materials and methods washed brie¯y in PBS and then mounted in PBS/glycerol (1 : 3) containing 2.5% 1,4 diazobicyclo (2,2,2) octane (DABCO, antifading agent; Sigma-Aldrich, Dorset, UK). Tissue preparation A total of 44 adult male Wistar rats (200±400 g body weight) were processed for BDNF immunocytochemistry or in situ hybridization. Thirty-six of these underwent unilateral section of the left sciatic nerve. The sciatic nerve was exposed under pentobarbitone anaesthesia (Sagatal, RhoÃne MeÂrieux Ltd, UK; 40 mg/kg i.p.) and then ligated 20 mm distal to the obturator tendon and cut. After recovery from anaesthesia, which was uneventful, rats survived for a further 2, 4, 7, 14 or 28 days before perfusion ®xation. All experiments were carried out according to UK Home Of®ce regulations. Rats were anaesthetized with sodium pentobarbital (60 mg/kg) and perfused through the ascending aorta with 30 mL vascular rinse followed by 300 mL 4% paraformaldehyde in 0.1 M phosphate buffer. Left and right L4 and L5 dorsal root ganglia, lumbar spinal cord and caudal medulla were dissected out, post®xed for 1.5±2 h and frozen after cryoprotection in 15% sucrose. Cryostat sections (8±12 mm) were then processed for either immuno¯uorescence or in situ hybridization. In addition, four of the 7-day sciatic section animals were perfused with 4% paraformaldehyde, 0.05% glutaraldehyde in 0.1 M phosphate buffer and processed for electron microscopic immunocytochemistry. Immuno¯uorescence Sections were stained using previously-described single- or dualcolour immuno¯uorescence or indirect tyramide signal ampli®cation (TSA) ¯uorescence procedures (Michael et al., 1997). Incubations consisted of 1 h in 10% normal serum followed by 18±36 h in primary antibody and 3 h in developing secondary antisera. For BDNF labelling, an af®nity-puri®ed rabbit antibody raised against recombinant human BDNF was used at 1 : 2000±1 : 5000 with the TSA procedure. For double labelling, this antibody was combined with one of the following: rabbit or sheep (Af®niti, Exeter, UK) CGRP polyclonal antisera (1 : 2000), rabbit galanin polyclonal antiserum (1 : 2000; Af®niti, Exeter, UK), rabbit neuropeptide Y (NPY) polyclonal antiserum (1 : 2000; Af®niti, Exeter, UK), rabbit polyclonal trkA antiserum (code-labelled RTA, used at 1 : 4000), rabbit polyclonal tyrosine hydroxylase antiserum (TH, 1 : 2000; Chemicon International, Harlow, UK), N52 mouse monoclonal antibody against the 200-kDa neuro®lament subunit (1 : 600; Sigma-Aldrich, Dorset, UK). The characteristics and staining speci®city of all these markers have been reported previously (BDNF, Michael et al., 1997; Yan et al., 1997; RTA, Clary et al., 1994; Averill et al., 1995; CGRP, Averill et al., 1995; galanin and NPY, Thompson et al., 1998; TH, Thompson & Majithia, 1998). Controls for double labelling included reversing the order of the primary antisera, as well as omitting the ®rst or second primary antiserum. The two sets of antisera were applied sequentially and this normally involved BDNF TSA followed by indirect labelled immuno¯uorescence. For full details of double labelling using the TSA procedure see Shindler & Roth (1996) and Hunyady et al. (1996). Secondary reagents used for indirect immuno¯uorescence included both FITC- and TRITC-labelled antirabbit IgG and antisheep IgG af®nity puri®ed antisera (Jackson Immuno Research, Pennsylvania, USA; 1 : 100 dilution). Tyramide signal ampli®cation labelling was carried out using biotinylated goat antirabbit IgG (1 : 400, Vector Laboratories, Peterborough, UK) and Vectastainq Elite peroxidase reagent (Vector Laboratories, Peterborough, UK) followed by biotinyl tyramide (NEN Life Science Products, Hounslow, UK; TSA-indirect kit) and ExtrAvidin-FITC (1 : 500, Sigma-Aldrich, Dorset, UK). After incubation in secondary reagents, sections were In situ hybridization Oligonucleotide probes complementary to bases 273±306 of the rat BDNF sequence (Timmusk et al., 1993), bases 124±157 of the rat trkA sequence (Meakin et al., 1992), bases 2213±2246 of the rat trkB sequence (Middlemas et al., 1991) and bases 1099±1132 of the rat trkC sequence (Valenzuela et al., 1993) were hybridized to cryostat sections using standard procedures (Michael et al., 1997). The probes were labelled at the 3¢ end with 35S-dATP (Dupont-NEN, Hounslow, UK) and terminal transferase (Promega, Southampton, UK) to speci®c activities of approximately 5000 Ci/mmol. Sections were acetylated in 0.25 M acetic anhydride/0.1 M triethanolamine/PBS for 10 min, dehydrated in ethanols (70±100%) and delipidated with chloroform. Hybridizations were carried out overnight at 37 °C using probe concentrations of 2 nM. Hybridization buffer consisted of 4 3 standard saline citrate (SSC: 1 3 SSC is 150 mM sodium chloride, 15 mM sodium citrate, pH 7.0), 50% deionized formamide, 0.04% Ficoll-400, 0.04% polyvinylpyrrolidone, 0.04% bovine serum albumin, 10% dextran sulphate, 0.1% SDS, 20 mM dithiothreitol (DTT), 20 mg/mL yeast tRNA, 100 mg/mL sheared salmon sperm DNA and 10 mg/mL poly adenylate. Following hybridization, sections received two 15-min washes at room temperature (RT) in 2 3 SSC, two at 50 °C in 1 3 SSC and one at 50 °C in 0.2 3 SSC. Sections were washed for a further 2 h at RT in 1 3 SSC, were dehydrated through ethanols, dipped in autoradiographic emulsion (LM1; Amersham Pharmacia Biotech UK, Bucks, UK) and exposed for 4±8 weeks. Following development, slides were counterstained with toluidine blue, dehydrated and coverslipped. Some sections were processed for immuno¯uorescence followed by in situ hybridization, as described previously (Michael & Priestley, 1996; Michael et al., 1997). Standard indirect immuno¯uorescence was performed as described above, except that antisera were diluted in diethylpyrocarbonate (DEPC)-treated PBS containing 0.5 mM DTT and 100 units/mL RNasin (Promega, Southampton, UK) in addition to 0.2% Triton-X-100 and 0.1% sodium azide. After immunostaining, sections were processed as for single BDNF in situ hybridization except that developed sections were mounted in PBS/glycerol instead of being toluidine-blue counterstained and dehydrated. Electron microscopic immunocytochemistry Staining was carried out using standard pre-embedding procedures (Priestley et al., 1992). The spinal cord was dissected out, post®xed for 2.5 h in the same ®xative and immersed in PBS. Sections (40 mm) were cut using a vibratome and were pretreated with 1% sodium borohydride in PBS for 30 min before immunostaining. Sections were incubated for 30 min in 10% normal goat serum and then transferred to BDNF polyclonal antibody (1 : 1000) for 12 h at 4 °C. Primary antibody was subsequently revealed using 1 : 400 biotinylated goat antirabbit IgG (Vector, UK) and Vectastainq Elite peroxidase reagent (Vector Laboratories, Peterborough, UK). Sections were then developed with a solution containing 0.05% 3,3¢-diaminobenzidine (DAB), 0.04% (NH4)2SO4.NiSO4 and 0.01% H2O2 in 0.1 M phosphate buffer pH 7.3. Unless otherwise stated, incubations were carried out at room temperature and antisera were diluted in PBS. Stained sections were then contrasted in OsO4 (1%) and uranyl acetate (1%), dehydrated and ¯at-embedded in Durcupan (Fluka, Sigma-Aldrich, Dorset, UK). After light microscopic examination, areas of interest were processed further for electron microscopy. Sections were examined on a Joel 1200EX electron microscope. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 Axotomy upregulates BDNF in large DRG cells 3541 Imaging and quanti®cation Sections were viewed on a Leica DMRB epi¯uorescence microscope using Y3 (TRITC), L4 (FITC) and polarization ®lter blocks combined with bright-®eld and/or dark-®eld illumination. Immunostaining and in situ hybridization was documented by photography using Ilford TMAX ®lm or a Hamamatsu digital camera (C4742±95) with an image size of 1280 3 1024 pixels. Figures were generated digitally from the Hamamatsu images or from 35 mm negatives using a Nikon LS-1000 scanner at 900±1300 pixels/inch. Plates were composed using Adobe Photoshop and printed on a Sony UP-D8800 graphics printer at 300 pixels/inch. Contrast was optimized by adjusting grey levels and sharpening, but images were not otherwise manipulated. The proportion of BDNF-expressing DRG cells was determined by counting the number of labelled and unlabelled neuronal pro®les in dorsal root ganglion sections. For in situ hybridization, cells were identi®ed as positively labelled when epipolarized illumination revealed silver grains clustered over the cell body. This criterion has previously been shown to correlate well with image analysis in which labelled cells are de®ned as having the mean of the background labelling plus two times the standard deviation of that labelling (Michael & Priestley, 1999). In dual immuno¯uorescence or in situ hybridization plus immuno¯uorescence sections the percentage of BDNF-expressing cells expressing a second marker was assessed by switching between FITC, TRITC and/or polarization ®lter blocks. At least 250 labelled DRG cells were examined for each marker and were counted on randomly-chosen sections. Statistical signi®cance was assessed using Mann±Whitney U nonparametric comparisons. Using Visilog image analysis software, the cell size and level of expression of BDNF mRNA was assessed in trkA-immunoreactive and -immunonegative DRG cells as previously described (Michael et al., 1997). Images were captured directly off the microscope at 25 3 magni®cation using a Grundig FA87 digital camera with integrating framestore. Cells were then outlined manually using a computer mouse and the cell size and area within each cell that was occupied by silver grains was calculated. At least 300 cells of each type were counted. For analysis of BDNF expression by trkB- or trkC-expressing cells, serial sections that had been processed for BDNF, trkB or trkC in situ hybridization were examined. Only cells that were present in both a trk and a BDNF serial section were analysed and more than 150 trkB cells and 250 trkC cells were counted. Results Effect of nerve section on BDNF expression by dorsal root ganglion cells Sciatic nerve section led to a large and sustained increase in BDNF expression by lumbar DRG cells (Fig. 1). Brain-derived neurotrophic factor immunoreactivity and BDNF mRNA were both rapidly increased ipsilateral to the nerve section, but differed in their time course and in the number of labelled cells. Thus the number of DRG cells showing BDNF immunoreactivity increased from 13 to 40% at 4 days after axotomy and was then maintained at this level (Table 1). A similar but more rapid increase in expression of BDNF mRNA occurred, with the numbers of labelled DRG cells increasing from 29 to 51% by 2 days survival (Table 1). An increase in the number of cells expressing BDNF mRNA was also seen in the contralateral lumbar ganglia, but was smaller and of slower time course (Table 1). At all time points analysed, both ipsilateral and contralateral to the sciatic nerve section, BDNF immunoreactivity was present in some small DRG cells. However, the increased BDNF expression that occurred after axotomy appeared to be mainly in medium- and largesized cells (Fig. 1). This impression was con®rmed by quantitative analysis of the size of labelled cells and by double labelling using the antineuro®lament antibody N52, a widely-used marker for largediameter cells. For both BDNF immunoreactivity (Fig. 2) and BDNF mRNA (data not shown), axotomy led to a sustained shift in the cell size distribution such that proportionally more labelled cells were large-sized. Consistent with this cell-size shift, an increase was also seen in the expression of BDNF mRNA (Fig. 3a and b) or protein by N52-immunoreactive cells. For example, at 4 days after sciatic nerve section, the percentage of N52-immunoreactive cells that were also BDNF-immunoreactive was 33% (compared with 20% on the contralateral side) and the percentage of BDNF-immunoreactive cells that were also N52-immunoreactive was 58% (compared with 45% on the contralateral side). Image analysis further revealed that the increase in BDNF mRNA expression by large cells represented not only an increase in the numbers of cells but also an increase in labelling intensity. Thus sciatic section led to an increase in the number of silver grains over medium-sized and large cells but little change in labelling of small cells (Fig. 4a±c). Neurotrophin receptors and BDNF expression following axotomy BDNF has previously been shown to be regulated in trkA cells by NGF (Apfel et al., 1996; Michael et al., 1997). We therefore examined the type of neurotrophin receptors in DRG cells that express BDNF immunoreactivity (not illustrated) or BDNF mRNA (Fig. 3) following axotomy. In situ hybridization was used to localize the receptors trkB (selective for BDNF and neurotrophin-4/5) and trkC (selective for neurotrophin-3) and immuno¯uorescence was used to localize trkA. The axotomy-induced increase in the number of large DRG cells that express BDNF was re¯ected in a major shift in the trk pro®le of BDNF-expressing cells. Only two days after axotomy the percentage of trkB cells that express BDNF mRNA had increased to 50% (compared with 2% on the contralateral side) and the percentage of trkC cells that express BDNF mRNA had increased to 56% (compared with 18% on the contralateral side). In contrast there was little change in the percentage of BDNF-immunoreactive trkA cells at any time points (Table 2). Because trkA and trkC are coexpressed in some medium-sized DRG neurons (Kashiba et al., 1995; Wright & Snider, 1995), image analysis was carried out in order to quantify BDNF mRNA expression in different sized trkA cells. This con®rmed that BDNF expression shows little change in the small cells that represent the main population of trkA-expressing cells. However, BDNF expression did increase in medium- and largesized trkA cells (Fig. 4d±f). BDNF immunoreactivity in pericellular baskets In addition to cell bodies, BDNF immunoreactivity was present in axons. Such axons were most prominent in ®bre bundles within ganglia and the staining appeared increased after nerve section (Fig. 1a and b). However, at 4 weeks nerve section, BDNF immunoreactivity was also seen in pericellular nerve baskets (Fig. 5) with 1±4 baskets present per section. Double labelling revealed that such baskets were immunoreactive for CGRP (Fig. 5a± d) but not tyrosine hydroxylase, and that they surrounded both BDNF-immunoreactive (Fig. 5e and i) and nonimmunoreactive (Fig. 5a and c) cells. In order to determine the trk pro®le of cells possessing baskets, BDNF immunocytochemistry was combined with trkA immuno¯uorescence or with trkB in situ hybridization. Baskets were most prominent around small- to medium-sized cells and about half of these were trkA-immunoreactive (Fig. 5e±h). In addition, baskets with ®ne axons were seen around occasional large cells and these included trkB-expressing cells (Fig. 5i±l). Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 3542 G. J. Michael et al. BDNF expression and neuropeptide expression by dorsal root ganglion cells Following axotomy, the neuropeptides galanin and NPY are strongly upregulated by DRG cells. We therefore carried out double labelling in order to determine the relationship between this increased peptide expression and the increased BDNF expression. Seven days after sciatic section, the number of galanin- and NPY-immunoreactive cells was greatly increased in L4 and L5 ganglia. As previously reported (Villar et al. 1989; Wakisaka et al. 1991), galanin immunoreactivity was present in both small- and large-diameter cells but the NPY expression was con®ned mainly to medium- and large-sized cells. Brain-derived neurotrophic factor immunoreativity was generally not coexpressed with galanin in small cells but showed extensive coexpression with galanin or NPY in large cells (Fig. 6). This was particularly striking for NPY, with 56% of BDNF-immunoreactive cells showing NPY immunoreactivity and 59% of NPY immunoreactive cells expressing BDNF. FIG. 1. BDNF mRNA and protein are increased by sciatic axotomy. Micrographs show lumbar ganglia axotomised by sciatic nerve section (b, d and f) and their contralateral control ganglia (a, c and e). (a±d) Low- and high-magni®cation micrographs showing BDNF immun¯uorescence 4 days after nerve section. The number of immunostained cells is greatly increased in the axotomised ganglia, with immunoreactivity now present in large-diameter cells (asterisks in d). Increased axonal labelling is also evident (small arrows in b). A small number of highly immunoreactive small- to medium-sized cells are visible (short arrows in a,b) but are present on both the contralateral and axotomised sides. (e and f) Dark-®eld micrographs showing BDNF in situ hybridization 2 days after nerve section. A dramatic increase in labelling is evident on the axotomised side. Scale bars, 200 mm (a and b) and 100 mm (c±f). Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 Axotomy upregulates BDNF in large DRG cells 3543 BDNF immunoreactivity in the central projections of dorsal root ganglion cells BDNF protein is axonally transported into the central arbors of primary afferents (Zhou & Rush, 1996; Michael et al., 1997). In order to determine the effects of axotomy on the distribution of centrallytransported BDNF, immunoreactivity was examined in the spinal cord and dorsal column nuclei at various time points after unilateral sciatic TABLE 1. The effect of sciatic nerve section on the percentage of DRG cells expressing BDNF immunoreactivity or BDNF mRNA BDNF immunoreactivity Control 2-day Ipsilateral Contralateral 4-day Ipsilateral Contralateral 7-day Ipsilateral Contralateral 14-day Ipsilateral Contralateral 28-day Ipsilateral Contralateral BDNF mRNA (%) (n) (%) (n) 13.2 6 1.5 (3) 28.5 6 3.3 (4) 11.7 6 0.2 10.9 6 0.9 (3) (3) 50.7 6 6.4* 34.4 6 1.7 (3) (3) 39.5 6 0.6* 11.3 6 1.5 ² (3) (3) 47.9 6 1.7* 38.4 6 4.1 (3) (3) 37.6 6 2.8* 13.0 6 0.9 ² (3) (3) 53.0 6 2.1* 41.9 6 4.2* (3) (3) 32.9 6 6.0* 15.0 6 0.3 ² (3) (3) 52.0 6 0.0 43.3 6 0.1 38.9 6 2.1* 17.8 6 2.65³ (3) (2) Not determined Not determined ³ ³ (2) (2) Data are presented as mean 6 SEM; (n) is the number of animals analysed. *Ipsilateral signi®cantly different from controls at P = 0.05; ²ipsilateral signi®cantly different from contralateral at P = 0.05; ³too few animals to determine signi®cance level. nerve section. In the lumbar cord contralateral to the nerve section, BDNF-immunoreactive axons and terminals were concentrated mainly in the super®cial dorsal horn (laminae I and II) and with only isolated axons penetrating more deeply (Fig. 7b). In contrast, on the axotomised side many terminals were evident in the deep dorsal horn, particularly in the reticular region of lamina IV (Fig. 7a). This staining was concentrated in the central projection zones of the sciatic nerve, namely the medial and central dorsal horn at L4 and L5 levels, and was evident as early as 4 days postlesion and was particularly prominent at 7 days. Many of the immunoreactive axons in lamina V displayed prominent varicosities, and EM examination of selected ®bres showed that such varicosities represent sites of synaptic contact with adjoining dendrites (Fig. 7e±g). In the dorsal column nuclei a change in BDNF immunostaining was also seen and was even more striking than that in the spinal cord. Ipsilateral to a sciatic nerve section there were numerous immunoreactive axons and terminals in the gracile nucleus, but on the contralateral side there was very little BDNF immunoreactivity (Fig. 8a). Neuropeptide Y immunoreactivity was also increased ipsilateral to the lesion side (Fig. 8b), and double labelling showed that many of the BDNF terminals in the gracile nucleus coexpressed NPY (Fig. 8c and d). These changes were evident at 4 days postlesion and were particularly prominent at 7 days. Discussion Several studies have reported that BDNF expression by DRG cells is increased following nerve injury (Ernfors et al., 1993; Sebert & Shooter, 1993; Cho et al., 1998; Tonra et al., 1998); however, the cell types involved have not previously been characterized in detail. In this study we have shown that sciatic axotomy leads to an increase in BDNF mRNA in large-diameter DRG cells and to a major change in the expression pattern of BDNF protein both within dorsal root ganglia FIG. 2. Cell size distribution of BDNFimmunoreactive pro®les within normal L4/L5 dorsal root ganglia (CONTROL) and in ganglia 2, 4, 7, 14 and 28 days after sciatic nerve section. By 4 days the normal cell size distribution has become shifted towards the large cells and this is maintained at later time points. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 3544 G. J. Michael et al. FIG. 3. BDNF mRNA after axotomy is present in many trkB (and trkC) -expressing DRG cells and in some trkA cells. Ipsilateral lumbar ganglia, 2 day sciatic nerve section. (a-d) Two serial sections processed for BDNF (b) and trkB (d) mRNAs using in situ hybridization, also immunostained for the large-cell marker N52 (a and c). The micrographs show a region of ganglion that contains predominantly large-diameter cells. In (a) and (b), most of the BDNF-expressing cells (asterisks) can be seen to be N52-immunoreactive. Panels (c) and (d) show that trkB-expressing cells (asterisks) are also N52-immunoreactive. Comparison of (a and b) and (c and d) reveals that many of the BDNF-expressing cells coexpress trkB. Four such cells are indicated by the long arrows. The short arrows indicate two BDNF-expressing cells that are present in the serial section (c and d) but that do not express trkB. In a third serial section (not illustrated) these cells were seen to express trkC. Stars indicate small cells that are not N52 immunoreactive. (e and f) Combined trkA immuno¯uorescence (e) and BDNF in situ hybridization (f). A cell is visible that shows double labelling, for both trkA and BDNF (long arrow). However, cells single-labelled for either trkA (asterisks) or BDNF (short arrows) are also present. Scale bars, 50 mm. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 Axotomy upregulates BDNF in large DRG cells 3545 FIG. 4. Axotomy leads to increased BDNF expression by medium- and large-sized DRG cells. Q-sum (cumulative-sum) plots showing the effect of axotomy on BDNF mRNA expression by small (< 25 mm diameter), medium (25±37.5 mm diameter) and large (> 37.5 mm diameter) DRG cells. Expression of BDNF mRNA was quanti®ed by measuring the percentage of the cell body (PERCENT AREA) that was covered by silver grains. In (a±c), all BDNF-expressing cells were analysed; in (d±f) only BDNF-expressing cells that were also trkA-immunoreactive were analysed. Note that, in both cases, axotomy leads to an increased BDNF expression by medium- and large-sized DRG cells but not by small cells. The data shown is at 2 days after nerve section. Similar results were obtained 7 days after nerve section. and in their central projections. These changes suggest that BDNF is likely to play a key role in regulating the response of somatosensory pathways to injury, with likely autocrine and paracrine actions within dorsal root ganglia and synaptic actions in spinal cord and dorsal column nuclei. These various possibilities will be discussed in turn. BDNF expression by trkB- and trkC-expressing DRG cells Sciatic axotomy produced a rapid and long lasting (4-week) increase in BDNF expression which had a time course similar to that previously reported for sciatic nerve crush (Sebert & Shooter, 1993). Brain-derived neurotrophic factor mRNA was signi®cantly increased by 2 days whilst an increase in BDNF immunoreactivity was not apparent until 4 days. This may represent a delay in BDNF translation but it is also possible that there is an early increase in BDNF protein that subsides by 2 days. Our immunocytochemical and in situ hybridization analysis revealed that the increased BDNF expression is not generalized but rather is con®ned mainly to medium- and largediameter cells. Similar results have recently been reported by Cho et al. (1998). Such cells have previously been shown to express trkB and/or trkC receptors (McMahon et al., 1994; Kashiba et al., 1995; Wright & Snider, 1995) and, consistent with this, we observed a major increase in BDNF expression by these cell types. Thus 50% of trkB cells and 56% of trkC cells contained BDNF mRNA following sciatic nerve section. Values on the contralateral side (2% of trkB cells and 18% of trkC cells) were similar to those previously reported for normal lumbar ganglia (Kashiba et al., 1997), indicating that our axotomy results represent a genuine increased expression by trkB and trkC cells. The analysis of trkB or trkC and BDNF coexpression TABLE 2. The effect of sciatic nerve section on the percentage of trkAimmunoreactive cells that also express BDNF immunoreactivity trkA cells also expressing BDNF (%) 4 day sciatic Ipsilateral Contralateral 7 day sciatic Ipsilateral Contralateral 14 day sciatic Ipsilateral Contralateral 28 day sciatic Ipsilateral Contralateral (%) (n) 43.3 6 2.5* 34.0 6 3.7 (4) (4) 30.7 6 1.6* 29.9 6 1.6 (3) (3) 26.3 6 1.6³ 24.5 6 1.5 (2) (2) 31.2 6 3.6³ 23.3 6 3.7 (2) (3) Data are presented as mean 6 SEM; (n) is the number of animals analysed. *Not signi®cantly different from contralateral ganglion at P = 0.05; ³too few animals to determine signi®cance level. involved serial sections and was carried out at 2 days. This was in order to avoid complications that might arise due to changed trk expression at longer time points (Ernfors et al., 1993). However, our data on cell sizes and on BDNF coexpression with a large-cell marker (N52) indicate that the increased expression by trkB and trkC cells is likely to be maintained throughout the time points analysed. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 3546 G. J. Michael et al. In contrast to the trkB and trkC cells, trkA cells as a population showed little change in BDNF expression. Brainderived neurotrophic factor increased in medium- and large-sized trkA cells, a cell group that is likely to coexpress both trkA and trkC (McMahon et al., 1994; Wright & Snider, 1995). However, there was no increase in the main population of small trkA cells. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 Axotomy upregulates BDNF in large DRG cells 3547 The changes after axotomy thus contrast starkly with the massive BDNF increase in trkA cells that occurs with peripheral in¯ammation and/or NGF treatment (Apfel et al., 1996; Cho et al., 1997; Michael et al., 1997). Our results thus show that BDNF synthesis is regulated separately in the small- and largesized DRG neurons and increases in each subgroup under different circumstances. The factors that control BDNF expression in large cells are not known, but a variety of neuropeptides and neuropeptide receptors show a similar response to axotomy (see HoÈkfelt et al., 1994). One possibility is that BDNF increases in large cells due to the loss of a target-derived factor acting through trkB or trkC. Neurotrophin-3 (NT3), for example, has been shown to reduce the axotomy-induced increase in NPY (Ohara et al., 1995) and is a candidate negative regulator of BDNF expression. Alternatively, BDNF could be regulated positively by injury-induced factors and/or depolarizing events associated with the nerve injury. Given the wide-ranging effects that BDNF is likely to have (discussed below), it will be important to dissect the molecular events that control its synthesis. The BDNF gene consists of ®ve exons, four of which (I-IV) give rise to alternative transcripts with their own tissue- and stimulusspeci®c promoters (Timmusk et al., 1993, 1995). However, the expression pattern of exons I±IV has not been analysed in DRG cells. Local autocrine and paracrine effects The function of the BDNF synthesized by DRG cells is not known, but local effects within the ganglia (Acheson et al., 1995; Apfel FIG. 6. BDNF immunoreactivity after axotomy is expressed by large DRG cells that also contain the neuropeptides neuropeptide Y (NPY) and galanin (gal). Ipsilateral lumbar ganglia, 4-day sciatic nerve section. Double-labelled sections showing that many BDNF-immunoreactive large cells also contain NPY- (a and b) or galanin- (c and d) immunoreactivity (asterisks indicate double-labelled cells). However, small cells that upregulate galanin (arrows in d) mainly do not express BDNF. Scale bars, 50 mm. FIG. 5. Four weeks after nerve section, BDNF-immunoreactive axons form pericellular baskets. (a±d) Double labelling for BDNF- and CGRP-immunoreactivities showing that BDNF baskets also contain CGRP. Panels (a) and (b) show a complex basket (thin arrow) that exhibits strong BDNF and CGRP immunostaining. Nearby is a small CGRP cell (thick arrow) that lacks BDNF immunostaining. Panels (c) and (d) show a simpler BDNF/CGRP basket that envelops a nonimmunoreactive cell (asterisk) and adjoins a double-labelled cell (thick arrow). In the vicinity is a double-labelled axon (thin arrow) together with several CGRP single-labelled axons and a CGRP-immunoreactive basket (star). (e±h) Double labelling for BDNF- and trkA-immunoreactivities showing that BDNF baskets surround both trkA-negative (e and f) and trkA-positive (g and h) cells. Panels (e) and (f) show a BDNF basket (thin arrow) that surrounds a BDNF immunoreactive cell (asterisk). The basket is trkA-immunoreactive although the cell is not. In the vicinity are numerous trkA-immunoreactive cells, some of which show slight BDNF staining. Panels (g) and (h) show a BDNF basket that surrounds a trkA-immunoreactive cell (asterisk). (i±l) Double labelling for BDNF immunoreactivity and trkB in situ hybridization showing that BDNF baskets surround both trkB-negative (i and j) and trkB-positive (k and l) cells. Panels (i) and (j) are low-magni®cation micrographs showing trkB-expressing cells (thick arrows), several of which are BDNF-immunoreactive. However, the two cells which possess BDNF baskets (thin arrows) are not trkB-labelled. Panels (k) and (l) are high-magni®cation micrographs showing two trkB-expressing cells (thick arrows), one of which possesses a ®ne BDNF-immunoreactive basket (thin arrow). A BDNF-immunoreactive cell that lacks trkB is also visible. Scale bars, 50 mm (a±f and i±l), 20 mm (g and h). Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 3548 G. J. Michael et al. FIG. 7. Brain-derived neurotrophic factor immunoreactivity is increased in the deep dorsal horn. Seven-day sciatic nerve section. (a±d) Immuno¯uorescence micrographs showing the lumbar spinal cord on the axotomised (Axot) and contralateral control (Contralat) sides. Immunoreactivity is abundant in the super®cial dorsal horn (a and b) and is increased on the axotomised side in the deep dorsal horn (arrows in a). In high magni®cation micrographs (c and d), immunoreactive axons are visible on both sides but are much more numerous on the axotomised side. (e) Light micrograph showing a BDNF axon in deep dorsal horn, axotomised side. Osmium counterstained section, processed for pre-embedding electron microscopic immunocytochemistry. The same axon is shown in the electron micrographs of (f) and (g). The axon has prominent varicosities (arrows) and winds around two bundles of myelinated axons. Longitudinally-cut large myelinated axons are also visible. The varicosity indicated with a double arrow can be seen also in (f) and (g). (f) Low-magni®cation electron micrograph of the area delineated by the box in (e). Arrows indicate the BDNF-immunoreactive axon, asterisks indicate the two bundles of myelinated axons and Mx indicates a longitudinally-cut large myelinated axon. (g) High magni®cation electron micrograph showing the varicosity (t) identi®ed by the double arrow in (e) and (f). It makes a synapse (thick arrow) with a dendrite (d). Scale bars, 100 mm (a and b), 20 mm (c and d), 10 mm (e), 5 mm (f) and 0.5 mm (g). et al., 1996) as well as actions within the spinal cord (Zhou & Rush, 1996; Michael et al., 1997) have been postulated. In vitro, a BDNF autocrine loop has been shown to mediate the survival of a subpopulation of adult DRG cells (Acheson et al., 1995). However, Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 Axotomy upregulates BDNF in large DRG cells 3549 FIG. 8. BDNF immunoreactivity is increased in the dorsal column nuclei after sciatic nerve section and coexists with NPY. The dorsal column nuclei are shown double labelled for BDNF and NPY. Both BDNF (a) and NPY (b) are increased in the gracile nucleus (Gr) on the axotomised side. In addition, numerous ®ne NPY-immunoreactive axons are visible bilaterally in the nucleus of the solitary tract (Sol). High magni®cation images of the gracile nucleus (c,d) show that many individual axons (arrows) are double labelled for BDNF and NPY. Scale bars, 20 mm. previous anatomical studies have shown that in vivo there is virtually no coexpression of BDNF and trkB in normal (Kashiba et al., 1997) or NGF-treated (Michael et al., 1997) animals. We now show that nerve section leads to a rapid upregulation of BDNF within trkB and trkC cells, and to the formation of BDNF-containing pericellular baskets that envelop DRG cells. Brain-derived neurotrophic factor Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 3550 G. J. Michael et al. therefore may indeed also act in vivo to protect adult DRG cells, being upregulated and released from novel sites following axotomy. One possibility is action via an autocrine loop, as suggested originally by Acheson et al. (1995). DRG neurons in culture may in fact resemble the axotomy situation, because both large and small cells express BDNF in vitro (Barakat-Walter, 1996). Studies on cultured hippocampal and cortical cells indicate that neurons can release BDNF from secretory granules concentrated in their somatodendritic domain (Goodman et al., 1996; Haubensak et al., 1998). Thus BDNF may act on trkB-expressing DRG cells following release from their own cell bodies (autocrine action) or from those of neighbouring trkC cells (paracrine action). Because truncated trkB receptors are highly expressed by satellite cells (McMahon et al., 1994) and have also recently been shown to mediate signal transduction (Baxter et al., 1997), satellite cells may also be affected. Another possibility is that BDNF is released from axonal varicosities within the ganglia. McLachlan & Hu (1998) have recently shown that CGRP-containing pericellular baskets form around axotomised DRG cells. We now show that many of these baskets also contain BDNF and that they surround a range of cell types, including trkA- and trkB-expressing cells. We did not determine the origin of the baskets, but did observe some strongly immunoreactive BDNF/CGRP and BDNF/trkA cells, consistent with the McLachlan & Hu (1998) data showing that baskets derive from adjoining cells. It is possible that the BDNF is upregulated in preexisting baskets, but the parallel appearance of CGRP- and substance P-immunoreactive baskets (McLachlan & Hu, 1998) is most easily explained by local sprouting. It has been proposed that the stimulus for both CGRP and sympathetic baskets is NGF, possibly presented by p75 receptors following release from satellite cells (Ramer & Bisby, 1997; Zhou et al., 1999). We have shown previously that trkA-expressing DRG cells are able to synthesize both BDNF and CGRP, possibly packaged in the same dense-cored vesicles (Michael et al., 1997). After axotomy a small subpopulation of these cells may produce baskets and release both CGRP and BDNF locally. In the CNS, BDNF has been shown to modify neuronal excitability by a variety of postsynaptic mechanisms (e.g. Lin et al., 1998). Brain-derived neurotrophic factor may therefore have similar actions in dorsal root ganglia, acting alongside CGRP and noradrenalin to modify DRG responses and possibly contributing to paraesthesias and/or changes in pain threshold. Central synaptic effects In addition to changes within lumbar ganglia, we observed that nerve section led to an increase in BDNF-immunoreactive terminals in deep dorsal horn and dorsal column nuclei. Small-diameter DRG cells have previously been shown to axonally transport BDNF into their central axonal arbors (Zhou & Rush, 1996; Michael et al., 1997). Our data shows that the upregulated BDNF in large-diameter neurons is similarly transported, and is consistent with recent studies showing a major increase in BDNF anterograde transport (Tonra et al., 1998) and central terminal labelling (Cho et al., 1998) following sciatic nerve section. In addition to BDNF we observed upregulation in large cells of the neuropeptides NPY and galanin, with striking double labelling of gracile nucleus terminals for BDNF and NPY. Neuropeptides such as substance P, CGRP and NPY, that are upregulated by large cells, may modify the excitability of their target neurons and possibly contribute to neuropathic sensations (Noguchi et al., 1995). Our results suggest that BDNF may have a similar role. Neurons in dorsal column nuclei and deep dorsal horn gain a BDNF input and, at least in the dorsal horn, this is likely to be mediated via synaptic contacts. TrkB receptors are widely expressed in the brainstem and spinal cord (Bradbury et al., 1998; King et al., 1999) and, as in hippocampus, may have acute effects on membrane currents (e.g. Lin et al., 1998) as well as long-lasting effects on gene expression (Croll et al., 1994). Brain-derived neurotrophic factor has an unusual combination of features as regards DRG cells. It is one of a relatively small number of molecules that are upregulated following both nerve section and in¯ammation, it remains the only trophic factor that has unequivocally been shown to be synthesized by DRG cells, and it is the only one that is thought to be synaptically released. It is striking that both large and small cells can upregulate BDNF and that it generally coexists with a neuropeptide. Brain-derived neurotrophic factor is thus in a unique position to modify the properties of second-order neurons, re¯ecting changes in peripherally-derived trophic factors and possibly acting together with a coreleased neuropeptide. Intrathecal BDNF has been reported to be hypoalgesic (Siuciak et al., 1994) but further studies are required to dissect the functional role of primary afferent BDNF. Acknowledgements This work was supported by the Medical Research Council (UK). We thank Professor J. M. Polak and Dr D. O. Clary for provision of rabbit CGRP and trkA antibodies and Dr Von King for statistical analysis. We also gratefully acknowledge the assistance of Mr J. Manston and Mr C. Pelling with digitizing of electron micrographs. Abbreviations BDNF, brain-derived neurotrophic factor; CGRP, calcitonin gene-related peptide; DAB, diaminobenzidine; DABCO, 1,4 diazobicyclo (2,2,2) octane; DEPC, diethylpyrocarbonate; DRG, dorsal root ganglion; DTT, dithiothreitol; FITC, ¯uoroscein isothiocyanate; NGF, nerve growth factor; NPY, neuropeptide Y; RT, room temperature; SSC, standard saline citrate; TH, tyrosine hydroxylase; TRITC, tetramethylrhodamine isothiocyanate; TSA, tyramide signal ampli®cation. References Acheson, A., Conover, J.C., Fandl, J.P., DeChlara, T.M., Russell, M., Thadani, A., Squinto, S.P., Yancopoulos, G.D. & Lindsay, R.M. (1995) A BDNF autocrine loop in adult sensory neurons prevents cell death. Nature, 374, 450±453. Apfel, S.C., Wright, D.E., Wiideman, A.M., Dormia, C., Snider, W.D. & Kessler, J.A. (1996) Nerve growth factor regulates the expression of brainderived neurotrophic factor messenger-RNA in the peripheral nervoussystem. Mol. Cell. Neurosci., 7, 134±142. Averill, S., McMahon, S.B., Clary, D.O., Reichardt, L.F. & Priestley, J.V. (1995) Immunocytochemical localization of trkA receptors in chemically identi®ed subgroups of adult rat sensory neurons. Eur. J. Neurosci., 7, 1484±1494. Averill, S., Michael, G.J., Shortland, P.J. & Priestley, J.V. (1997) BDNF increases in large diameter dorsal root ganglion cells and in their central projections following peripheral axotomy. Soc. Neurosci. Abstr., 23, 327. Barakat-Walter, I. (1996) Brain-derived neurotrophic factor-like immunoreactivity is localized mainly in small sensory neurons of rat dorsal-root ganglia. J. Neurosci. Meth., 68, 281±288. Baxter, G.T., Radeke, M.J., Kuo, R.C., Makrides, V., Hinkle, B., Hoang, R., MedinaSelby, A., Coit, D., Valenzuela, P. & Feinstein, S.C. (1997) Signal transduction mediated by the truncated trkB receptor isoforms, trkB.T1 and trkB.T2. J. Neurosci., 17, 2683±2690. Bennett, D.L.H., French, J.S., Priestley, J.V. & McMahon, S.B. (1996) The effects of BDNF on c-fos and NOS expression in dorsal horn neurones of the adult rat spinal cord. Soc. Neurosci. Abstr., 22, 993. Bradbury, E.J., King, V.R., Simmons, L.J., Priestley, J.V. & McMahon, S.B. (1998) NT-3, but not BDNF, prevents atrophy and cell death of axotomised spinal cord projection neurons. Eur. J. Neurosci., 10, 3058±3068. Cho, H.J., Kim, S.Y., Park, M.J., Kim, D.S., Kim, J.K. & Chu, M.Y. (1997) Expression of mRNA for brain-derived neurotrophic factor in the dorsal root ganglion following peripheral in¯ammation. Brain Res., 749, 358±362. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551 Axotomy upregulates BDNF in large DRG cells 3551 Cho, H.J., Kim, J.K., Park, H.C., Kim, J.K., Kim, D.S., Ha, S.O. & Hong, H.S. (1998) Changes in brain-derived neurotrophic factor immunoreactivity in rat dorsal root ganglia, spinal cord, and gracile nuclei following cut or crush injuries. Exp. Neurol., 154, 224±230. Clary, D.O., Weskamp, G., Austin, L.A.R. & Reichardt, L.F. (1994). TrkA cross-linking mimics neuronal responses to nerve growth factor. Mol. Biol. Cell, 5, 549±563. Croll, S.D., Wiegand, S.J., Anderson, K.D., Lindsay, R.M. & Nawa, H. (1994) Regulation of neuropeptides in adult rat forebrain by the neurotrophins BDNF and NGF. Eur. J. Neurosci., 6, 1343±1353. Ernfors, P., Rosario, C.M., Merlio, J.P., Grant, G., Aldskogius, H. & Persson, H. (1993) Expression of mRNAs for neurotrophin receptors in the dorsal root ganglion and spinal cord during development and following peripheral or central axotomy. Mol. Brain Res., 17, 217±226. Goodman, L.J., Valverde, J., Lim, F., Geschwind, M.D., Federoff, H.J., Geller, A.I. & Hefti, F. (1996) Regulated release and polarized localization of brain-derived neurotrophic factor in hippocampal-neurons. Mol. Cell. Neurosci., 7, 222±238. Haubensak, W., Narz, F., Heumann, R. & Lessmann, V. (1998) BDNF-GFP containing secretory granules are localized in the vicinity of synaptic junctions of cultured cortical neurons. J. Cell Sci., 111, 1483±1493. HoÈkfelt, T., Zhang, X. & Wiesenfeld-Hallin, Z. (1994) Messenger plasticity in primary sensory neurons following axotomy and its functional implications. Trends Neurosci., 17, 22±30. Hunyady, B., Krempels, K., Harta, G. & Mezey, E. (1996) Immunohistochemical signal ampli®cation by catalyzed reporter deposition and its application in double immunostaining. J. Histochem. Cytochem., 44, 1353± 1362. Karchewski, L.A., Wetmore, C. & Verge, V.M.K. (1995) Alterations in BDNF expression in lumbar DRG following injury. Soc. Neurosci. Abstr., 21, 298. Kashiba, H., Noguchi, K., Ueda, Y. & Senba, E. (1995) Coexpression of trk family members and low-af®nity neurotrophin receptors in rat dorsal root ganglion neurons. Mol. Brain Res., 30, 158±164. Kashiba, H., Ueda, Y., Ueyama, T., Nemoto, K. & Senba, E. (1997) Relationship between BDNF- and trk-expressing neurones in rat dorsal root ganglion: an analysis by in situ hybridization. Neuroreport, 8, 1229±1234. King, V.R., Michael, G.J., Joshi, R.K. & Priestley, J.V. (1999) TrkA, trkB and trkC mRNA expression by bulbospinal cells of the rat. Neuroscience, 92, 935±944. King, A.E. & Thompson, S.W.N. (1995) Brief and prolonged changes in spinal-injury excitability following peripheral injury. Sem. Neurosci., 7, 233±243. Lin, S.Y., Wu, K., Levine, E.S., Mount, H.T.J., Suen, P.C. & Black, I.B. (1998) BDNF acutely increases tyrosine phosphorylation of the NMDA receptor subunit 2b in cortical and hippocampal postsynaptic densities. Mol. Brain Res., 55, 20±27. McLachlan, E.M. & Hu, P. (1998) Axonal sprouts containing calcitonin generelated peptide and substance P form pericellular baskets around large diameter neurons after sciatic nerve transection in the rat. Neuroscience, 84, 961±965. McMahon, S.B., Armanini, M.P., Ling, L.H. & Phillips, H.S. (1994) Expression and coexpression of trk receptors in subpopulations of adult primary sensory neurons projecting to identi®ed peripheral targets. Neuron, 12, 1161±1171. McMahon, S.B., Bennett, D.L.H., Michael, G.J. & Priestley, J.V. (1997) Neurotrophic factors and pain. In Jensen, T.S., Turner, J.A. & WiesenfeldHallin, Z. (eds), Progress in Pain Research and Management, Vol. 8. IASP Press, Seattle, pp. 353±379. Meakin, S.O., Suter, U., Drinkwater, C.C., Welcher, A.A. & Shooter, E.M. (1992) The rat trk protooncogene product exhibits properties characteristic of the slow nerve growth factor receptor. Proc. Natl Acad. Sci. USA, 89, 2374±2378. Michael, G.J., Averill, S., Nitkunan, A., Rattray, M., Bennett, D.L.H., Yan, Q. & Priestley, J.V. (1997) Nerve growth factor treatment increases brainderived neurotrophic factor selectively in trkA-expressing dorsal root ganglion cells and in their central terminations within the spinal cord. J. Neurosci., 17, 8476±8490. Michael, G.J. & Priestley, J.V. (1996) Combined immunocytochemistry and in situ hybridization. In Henderson, Z. (ed.), In Situ Hybridization Techniques for the Brain. John Wiley & Sons, Chichester, pp. 111±118. Michael, G.J. & Priestley, J.V. (1999) Differential expression of the mRNA for the vanilloid receptor subtype 1 in cells of the adult rat dorsal root and nodose ganglia and its downregulation by axotomy. J. Neurosci., 19, 1844± 1854. Middlemas, D.S., Lindberg, R.A. & Hunter, T. (1991) TrkB, a neural receptor protein-tyrosine kinase: evidence for a full-length and two truncated receptors. Mol. Cell. Biol., 11, 143±153. Noguchi, K., Kawai, Y., Fukuoka, T., Senba, E. & Miki, K. (1995) Substance P induced by peripheral nerve injury in primary afferent sensory neurons and its effect on dorsal column nucleus neurons. J. Neurosci., 15, 7633± 7643. Ohara, S., Tantuwaya, V., DiStefano, P.S. & Schmidt, R.E. (1995) Exogenous NT-3 mitigates the transganglionic neuropeptide Y response to sciatic nerve injury. Brain Res., 699, 143±148. Priestley, J.V., Alvarez, F.J. & Averill, S. (1992) Pre-embedding electron microscopic immunocytochemistry. In Polak, J.M. & Priestley, J.V. (eds), Electron Microscopic Immunocytochemistry. Oxford University Press, Oxford, pp. 89±121.. Ramer, M. & Bisby, M. (1997) Reduced sympathetic sprouting occurs in dorsal root ganglia after axotomy in mice lacking low-af®nity neurotrophin receptor. Neurosci. Lett., 228, 9±12. Sebert, M.E. & Shooter, E.M. (1993) Expression of messenger-RNA for neurotrophic factors and their receptors in the rat dorsal root ganglion and sciatic nerve following nerve injury. J. Neurosci. Res., 36, 357±367. Shindler, K.S. & Roth, K.A. (1996) Double immuno¯uorescent staining using two unconjugated primary antisera raised in the same species. J. Histochem. Cytochem., 44, 1331±1335. Siuciak, J.A., Altar, C.A., Wiegand, S.J. & Lindsay, R.M. (1994) Antinociceptive effect of brain-derived neurotrophic factor and neurotrophin-3. Brain Res., 633, 326±330. Thompson, S.W.N. & Majithia, A.A. (1998) Leukemia inhibitory factor induces sympathetic sprouting in infect dorsal root ganglia in the adult rat in vivo. J. Physiol.(Lond.), 506, 809±816. Thompson, S.W.N., Priestley, J.V. & Southall, A. (1998) Gp130 cytokines, leukemia inhibitory factor and interleukin-6 induce neuropeptide expression in intact adult rat sensory neurons in vivo: time course, speci®city and comparison with sciatic nerve axotomy. Neuroscience, 84, 1247±1255. Timmusk, T., Lendahl, U., Funakoshi, H., Arenas, E., Persson, H. & Metsis, M. (1995) Identi®cation of brain-derived neurotrophic factor promoter regions mediating tissue-speci®c, axotomy-induced and neuronal activityinduced expression in transgenic mice. J. Cell Biol., 128, 185±199. Timmusk, T., Palm, K., Metsis, M., Reintam, T., Paalme, V., Saarma, M. & Persson, H. (1993) Multiple promoters direct tissue-speci®c expression of the rat BDNF gene. Neuron, 10, 475±489. Tonra, J.R., Curtis, R., Wong, V., Cliffer, K.D., Park, J.S., Timmes, A., Nguyen, T., Lindsay, R.M., Acheson, A. & DiStefano, P.S. (1998) Axotomy upregulates the anterograde transport and expression of brain-derived neurotrophic factor by sensory neurons. J. Neurosci., 18, 4374±4383. Valenzuela, D.M., Maisonpierre, P.C., Glass, D.J., Rojas, E., Nunez, L., Kong, Y., Stitt, T.N., Ip, N.Y. & Yancopoulos, G.D. (1993) Alternative forms of rat TrkC with different functional capabilities. Neuron, 10, 963±974. Villar, M.J., Cortes, R., Theodorsson, E., Wiesenfeld-Hallin, Z., Schalling, M., Fahrenkrug, J., Emson, P.C. & HoÈkfelt, T. (1989) Neuropeptide expression in rat dorsal-root ganglion-cells and spinal- cord after peripheral-nerve injury with special reference to galanin. Neuroscience, 33, 587±604. Wakisaka, S., Kajander, K.C. & Bennett, G.J. (1991) Increased neuropeptideY (NPY) -like immunoreactivity in rat sensory neurons following peripheral axotomy. Neurosci. Lett., 124, 200±203. Wright, D.E. & Snider, W.D. (1995) Neurotrophin receptor mRNA expression de®nes distinct populations of neurons in rat dorsal root ganglia. J. Comp. Neurol., 351, 329±338. Yan, Q., Rosenfeld, R.D., Matheson, C.R., Hawkins, N., Lopez, O.T., Bennett, L. & Welcher, A.A. (1997) Expression of brain-derived neurotrophic factor protein in the adult rat central nervous system. Neuroscience, 78, 431±448. Zhou, X.F., Deng, Y.S., Chie, E., Xue, Q., Zhong, J.H., McLachlan, E.M., Rush, R.A. & Xian, C.J. (1999) Satellite-cell-derived nerve growth factor and neurotrophin-3 are involved in noradrenergic sprouting in the dorsal root ganglia following nerve injury in the rat. Eur. J. Neurosci., 11, 1711± 1722. Zhou, X.F. & Rush, R.A. (1996) Endogenous brain-derived neurotrophic factor is anterogradely transported in primary sensory neurons. Neuroscience, 74, 945±951. Ó 1999 European Neuroscience Association, European Journal of Neuroscience, 11, 3539±3551