Previous studies using transfected cells have indicated that the mammalian receptor tyrosine kinase trkB binds the neurotrophins brain-derived neurotrophic factor, neurotrophin-3 and neurotrophin-4. However, most studies demonstrating that these neurotrophins prevent the death of embryonic neurons and have specific neuronal receptors have been performed with chick neurons. In order to explore the possibility that trkB is the molecular entity representing the high-affinity receptor for brain-derived neurotrophic factor on embryonic chick neurons, we cloned and expressed a chick trkB cDNA. In situ hybridisation results indicate that the distribution of trkB mRNA in the peripheral nervous system of the developing chick embryo correlates well with the structures known to respond to brain-derived neurotrophic factor. Binding studies performed with a cell line stably transfected with the ctrkB cDNA indicate a dissociation constant for brain-derived neurotrophic factor of 9.9 10 10 M, which is distinctly higher than that found on primary chick sensory neurons (1.5×10−11 M). When binding of brain-derived neurotrophic factor was determined in the presence of other neurotrophins, neurotrophin-3 was found efficiently to prevent the binding of brain-derived neurotrophic factor to both the ctrkB cell line and embryonic sensory neurons. In vitro, neurotrophin-3 at high concentrations completely blocked the survival normally seen with brain-derived neurotrophic factor. Thus, unlike previous cases of receptor occupancy by heterologous neurotrophins (which resulted in agonistic effects), the interaction between the brain-derived neurotrophic factor receptor and neurotrophin-3 on sensory neurons is antagonistic.

One of the major biological functions of the neurotrophins is the control of neuronal survival during the phase of naturally occurring cell death (Barde, 1989). Until now, four members of this gene family have been identified: nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3) and neurotrophin-4 (NT-4) (for review, see Chao, 1992). While the four neurotrophins have been shown to prevent the death of a variety of embryonic neurons either in vitro or in vivo, the vast majority of these studies have been performed using avian embryonic neurons. Binding studies have also been performed using radiolabelled neurotrophins and neurons isolated from the peripheral nervous system of the chick embryo. With all neurotrophin-responsive neurons studied so far, two distinct binding sites have been identified. The low-affinity binding sites for neurotrophins (Kd of 0.8-2.3×10−9 M) are found on neurons and non-neuronal cells and are not ligand specific: for example, binding of labelled NGF to these sites can be prevented by BDNF and unlabelled NGF equally well (Rodríguez-Tébar et al., 1990). The same characteristics have been observed with the p75 lowaffinity NGF, or neurotrophin, receptor (p75LANR) expressed on fibroblasts (D. Johnson et al., 1986; Radeke et al., 1987), which binds all neurotrophins with similar affinities (Rodríguez-Tébar et al., 1990, 1992; Hallböök et al., 1991). Until now, the exact biological function played by p75LANR is not understood (see Meakin and Shooter, 1992 for review). The high-affinity sites on neurons bind their cognate ligands with a Kd of 1.7-2.3×10−11 M (Sutter et al., 1979; Rodríguez-Tébar and Barde, 1988; Rodríguez-Tébar et al., 1992) and are thought to be the signal transducing elements. In contrast to the low-affinity binding sites, the high-affinity sites are characterised by their ligand specificities (Rodríguez-Tébar et al., 1990, 1992; Dechant et al., 1993).

In addition to their interaction with p75LANR, the neurotrophins are now known also to interact with a subfamily of receptor tyrosine kinases known as trk-receptors designated trk, trkB and trkC (see Barbacid et al., 1991 for review). Much of the available evidence for these interactions is derived from experiments performed with cell lines and indicate that trk interacts with NGF and to a lesser degree with NT-3 (Cordon-Cardo et al., 1991; Hempstead et al., 1991; Kaplan et al., 1991a,b; Klein et al., 1991a), trkB with BDNF, NT-3 and NT-4 (Glass et al., 1991; Klein et al., 1991b; Soppet et al., 1991; Squinto et al., 1991; Berkemeier et al., 1991; Klein et al., 1992; Ip et al., 1993) and trkC with NT-3 (Lamballe et al., 1991; Tsoulfas et al., 1993; Valenzuela et al., 1993). These results indicate that, although there is a preferred neurotrophin ligand for each individual trk receptor (Ip et al., 1993), there is not a simple one-to-one relationship between receptor and ligand, and that trkB shows the most complex pattern of ligand interactions. In all cases investigated so far, binding of a neurotrophin to a recombinant trk receptor on fibroblast cell lines is followed by tyrosine phosphorylation of the receptor (Berkemeier et al., 1991; Glass et al., 1991; Kaplan et al., 1991b; Klein et al., 1991a,b; Soppet et al., 1991; Ip et al., 1993; Valenzuela et al., 1993; Tsoulfas et al., 1993). It has been noted, however, that, depending on the cellular context, the degree to which NT-3 interacts with trkB can vary markedly, the lesser degree of phosphorylation seen with NT-3 in PC12 cells transfected with trkB being thought to result from an increased ligand specificity, compared with fibroblasts (Ip et al., 1993). Several lines of evidence indicate that members of the trk-family are very likely to be involved in neuronal high-affinity binding and intracellular signalling following binding. Firstly, the trk-family members are the only neurotrophin receptors known to be expressed in neurons with a ligand-specific, intrinsic signal transducing capacity. Secondly, they express a pattern of ligand specificity that is reminiscent of that seen with neuronal high-affinity receptors. Thus, trkC, like NT-3 high-affinity receptors, is highly ligand specific (Lamballe et al., 1991; Rodríguez-Tébar et al., 1992; Tsoulfas et al., 1993; Valenzuela et al., 1993), whereas trk, like neuronal NGF receptors, does not interact significantly with BDNF, but with NT-3 (Rodríguez-Tébar et al., 1990; Cordon-Cardo et al., 1991; Dechant et al., 1993). Thirdly, the expression patterns of some of the trk-family members, as assessed by in situ hybridisation in rodents, are suggestive. Especially trk exhibits a convincing correlation between trk-expressing and NGF-responsive neurons (Klein et al., 1989, 1990b; Martin-Zanca et al., 1990; Carroll et al., 1992; Holtzman, et al., 1992; Merlio et al., 1992).

In order to gain insight into the signal transduction of neurotrophins on their physiological target cells, it is desirable to study the molecular composition of the neurotrophin-binding sites on embryonic neurons. Given that several populations of BDNF-responsive neurons have been studied in chick and quail embryos (Lindsay et al., 1985; Davies et al., 1986; Kalcheim et al., 1987; Hofer and Barde, 1988; Oppenheim et al., 1992), we were interested in studying by in situ hybridisation the expression of chick trkB (ctrkB) during the development the chick nervous system. Also, as most experiments describing the characteristics of neuronal receptors for neurotrophins have been performed with chick neurons but all trk-receptors cloned and characterised in mammalian species only, it was of interest to determine the binding properties of BDNF using ctrkB expressed in a cell line.

Cloning and sequencing

The ctrkB cDNA was cloned from an adult chick brain lambda gt10 cDNA (Couturier et al., 1990) using a rat trkB fragment encoding both the transmembrane and tyrosine kinase domain obtained by PCR and corresponding to bp 1120-2434 of the rat sequence (Middlemas et al., 1991). The amplified fragment was cloned, sequenced and radiolabelled using a random hexamer priming kit (Pharmacia-LKB). 9×105 phage clones were transferred to nylon membranes (Hybond N, Amersham-Buchler) and the filters hybridised in the following hybridisation mixture: 5× SSC, 5× Denhardt’s solution, 40% formamide and the radiolabelled probe (2×106 counts/minute/ml) at 38°C for 18 hours. The filters were subsequently washed in 6× SSC, 0.1% SDS at 55°C for 30 minutes. From the primary screening, 11 clones were isolated and the cDNAs from 8 phages were cloned into pBluescript (Stratagene). The sequence of the inserted cDNAs was then determined by dideoxynucleotide sequencing with Taq-Polymerase on a sequencing automate (Type 373A, Applied Biosystems) using various oligonucleotide primers (synthesised on a 381A Applied Biosystems DNA synthesiser) derived from the previously analysed sequence. The sequence reported in Fig. 1 is a result from sequencing on both strands for the entire sequence.

Fig. 1.

Nucleotide and protein sequence of chick trkB. Nucleotide sequence the first 2699 base pairs (from the 5′ end) of the longest cDNA insert isolated from an adult chick brain library. The sequence contains an open reading frame of 818 amino acids. Computer based analysis of the protein sequence predicted a single membrane spanning domain (=) and a tyrosine kinase domain (start and stop indicated by arrows).

Fig. 1.

Nucleotide and protein sequence of chick trkB. Nucleotide sequence the first 2699 base pairs (from the 5′ end) of the longest cDNA insert isolated from an adult chick brain library. The sequence contains an open reading frame of 818 amino acids. Computer based analysis of the protein sequence predicted a single membrane spanning domain (=) and a tyrosine kinase domain (start and stop indicated by arrows).

Probe design and preparation

Two probes spanning either the extracellular portion of chick trkB (nucleotides 587 to 1298) or approximately 600 bases of the 3′ untranslated region (see Fig. 2) were selected for in situ hybridisation and northern blot analysis. Preliminary in situ experiments run with a riboprobe spanning the entire tyrosine kinase domain of chick trkB resulted in a labelling with high background levels; this probe was therefore not employed routinely for in situ hybridisation studies. Riboprobes were prepared by linearising plasmids and transcribing the template in the presence of either 32P-UTP (northern blot analysis) or 35S-UTP (in situ hybridisation) according to standard procedures. Following transcription, the template DNA was removed by DNase A treatment (6 U, 15 minutes at 37°C) and the probe precipitated with 2.5 M ammonium acetate and ethanol. The specific activity of the probes employed was consistently in the range of 5×109 counts/minute/μg. A cDNA probe corresponding to the tyrosine kinase domain of trkB (nucleotides 1470 to 1856) was also used for determining the trkB transcripts coding for a full-length catalytic receptor. This cDNA probe was labelled by random prime labelling in the presence of 32P-dCTP to a specific activity of 4×109 counts/minute/μg.

Fig. 2.

Northern blot of brain and DRG RNA probed with ctrkB probes. The chick trkB cDNA is divided into 3 domains: the extracellular region, the tyrosine kinase region and the 3′ untranslated region. Probes used for RNA blot analysis were derived from the sequences underlined. Blots A, B and C contains poly(A)+ enriched RNAs from E10 chick brain. Panel D contains total RNA from dorsal root ganglia. A has been hybridised with a cRNA probe derived from the extracellular sequence, B has been probed with a cDNA probe spanning the tyrosine kinase domain and C and D with a riboprobe arising from the 3′ untranslated region. 28S and 18S are the ribosomal markers. Four bands ranging in size from 9.0 kb to 5.0 kb are shown in A. This probe identifies also a 1.6 kb message, not shown in this figure. B and C show that a 9.0 kb transcript can be detected with probes spanning the tyrosine kinase domain or the 3′ untranslated sequence. This band is also present in total RNA from dorsal root ganglia (D)

Fig. 2.

Northern blot of brain and DRG RNA probed with ctrkB probes. The chick trkB cDNA is divided into 3 domains: the extracellular region, the tyrosine kinase region and the 3′ untranslated region. Probes used for RNA blot analysis were derived from the sequences underlined. Blots A, B and C contains poly(A)+ enriched RNAs from E10 chick brain. Panel D contains total RNA from dorsal root ganglia. A has been hybridised with a cRNA probe derived from the extracellular sequence, B has been probed with a cDNA probe spanning the tyrosine kinase domain and C and D with a riboprobe arising from the 3′ untranslated region. 28S and 18S are the ribosomal markers. Four bands ranging in size from 9.0 kb to 5.0 kb are shown in A. This probe identifies also a 1.6 kb message, not shown in this figure. B and C show that a 9.0 kb transcript can be detected with probes spanning the tyrosine kinase domain or the 3′ untranslated sequence. This band is also present in total RNA from dorsal root ganglia (D)

Northern blot analysis

Total RNA was extracted from embryonic chick tissues according to the guanidium thiocyanate phenol-chloroform method (Chomczynski and Sacchi 1987). Subsequently, poly(A)+ enriched RNA was prepared from brain RNA by conventional oligo(dT) chromatography (Sambrook et al., 1989). 5 μg of poly(A)+-enriched brain RNA and 5 μg total RNA from E8 dorsal root ganglia were electrophoresed under denaturing conditions on formaldehyde agarose gel (Sambrook et al., 1989). RNA was transferred to nylon filters by vacuum blotting and UV-fixed. The methylene blue staining patterns of the filters indicated that the amount of poly(A)+ RNA loaded in the different lanes was similar. The northern blots analysed with the riboprobes were prehybridised with 50% formamide, 4× SSC, 4× Denhardt, 1% SDS, 50 μg/ml tRNA for 7 hours at 65°C. Hybridisation was performed overnight at 65°C in the same mixture containing 32P-labelled probe (5×106 counts/minute/ml), followed by a 20 minutes wash in the hybridisation mixture at 65°C and two washes of 30 minutes in 0.1% SDS, 0.1× SSC. The northern blots analysed with the cDNA probe were prehybridised with 50% formamide, 7× SSC, 50 mM sodium phosphate buffer pH 7.4, 5× Denhardt, 0.25% SDS, 100 μg/ml salmon sperm DNA and 40 μg/ml tRNA for 7 hours at 42°C. Hybridisation was performed in the same conditions, overnight with 5×106 counts/minute/ml of 32P-labelled probe, followed by a 15 minutes wash at 42°C with 0.1% SDS, 2× SSC and 15 minutes at 55°C with 0.1% SDS, 0.1× SSC. Blots were air dried and exposed to X-ray films (Fuji) with intensifying screens for 2-10 days.

In situ hybridisation

Eggs were collected from embryonic day 3.5 (E3.5) of incubation (or stage 23 of Hamburger and Hamilton, 1951) to day 14. The embryos were either fixed as a whole by immersion (until E7), or, alternatively, the tissues of interest were dissected out and fixed in 4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4. The tissues were subsequently cryoprotected in sucrose solutions, frozen in liquid nitrogen cooled isopentane, cut and placed onto polylysine-coated slides as previously described (Biffo et al., 1990).

For in situ hybridisation, the method of Wilkinson et al. (1987) with small variations was used. Following washes and proteinase K digestion (0.3 μg/ml), sections were dehydrated, and the riboprobe was added to a final concentration of 5×105 counts/minute per slide (in 100 μl). Hybridisation was performed overnight at 62°C. Stringency washes were performed as previously described (Wilkinson et al., 1987) and included a treatment with RNaseA at 20 μg/ml in 0.5 NaCl, 20 mM Tris-HCl pH 8.0, 2 mM EDTA for 30 minutes at 37°C. Following dehydration sections were exposed to NTB2 (Kodak) emulsion at 4°C for up to 2 weeks, successively developed, counterstained with hematoxylin and mounted.

Neurotrophins

For the preparation of radioligand, recombinant mouse BDNF was produced in a Vaccinia virus expression system as previously described for NT-3 (Götz et al., 1992). In the inhibition assays, BDNF and NT-3 derived from an E. coli expression system were used as native ligands (kind gift from R. Lindsay). NGF was isolated from the submandibular glands of adult male mice as described (Suda et al., 1978). Recombinant Xenopus NT-4 was purified using the procedure described for NT-3 from the supernatant of A293 cells stably transfected with a Xenopus prepro-NT-4 cDNA (Hallböök et al., 1991) cloned in a pCMV expression vector. The specific biological activity of the ligand proteins was analysed in quantitative survival assays using embryonic sensory neurons as described (Lindsay et al., 1985; Götz et al., 1992). Halfmaximal survival effects were obtained with concentrations of BDNF, NT-3 and NT-4 of 30-40 pg/ml.

Cell cultures

The human embryonic kidney cell line A293 (Graham et al., 1977) was grown in DMEM (Gibco) containing 10% fetal calf serum (Boehringer Mannhein). A stably transfected cell line expressing chick trkB was obtained by transfecting the ctrkB cDNA (bp 1-2698) subcloned into the HindIII site of the eucaryotic expression vector Rc/CMV (Stratagene), which drives the expression of inserted DNA from the human cytomegalovirus promoter (Andersson et al., 1989). 20 μg of this expression vector construct (pCMV-ctrkB) was transfected into A293 cells using the method of Chen and Okayama (1987) together with 1 μg of the plasmid pSV2-pac that confers puromycin resistance to transfected cells (Vara et al., 1986). After 10 days of selection with 1 μg/ml puromycin (Sigma), puromycin-resistant cell clones were subcloned and clonal cultures were further grown in the presence of 1 μg/ml puromycin. Fourteen different clones were analysed for ctrkB expression by northern blot analysis of total RNA probed against a ctrkB fragment. RNA blots from all transfected clones, but not from parental A293 cells contained a band of the expected size. The clone with the highest expression level of chick trkB mRNA was chosen for further analysis (A293-ctrkB). For binding assays, the parental A293 and A293-ctrkB cells were removed from the plastic dishes by gentle trypsinization (0.01% trypsin, 5 minutes) centrifuged and resuspended in Krebs-Ringer-Hepes buffer containing 5 mg/ml BSA and 0.1 mg/ml horse heart cytochromeC (binding buffer, Rodríguez-Tébar and Barde, 1988). The number of living cells was determined using the trypan blue exclusion method and a hemocytometer. Primary sensory neurons from E8 to E10 chick embryos were isolated as described (Rodríguez-Tébar and Barde, 1988). Briefly, the ganglia (dorsal root or nodose ganglion) were dissected and collected in Gey’s buffer (Gey and Gey, 1936) on ice. After dissection, the ganglia were treated with 0.02% trypsin for 15 minutes in the same buffer at 37°C. Trypsinization was stopped by the addition of 10% horse serum and the ganglia were dissociated by gentle trituration in a wide-bore glass pipette. The cell suspension was subsequently incubated on culture dishes at 10% CO2 at 37°C. During this time, non-neuronal cells settle on the plastic dish and the neurons are collected by aspiration of the supernatants. This procedure results in a high enrichment of neurons in the suspension. The supernatant was subsequently centrifuged (1000 g, 5 minutes) and the cell pellet resuspended in binding buffer. The number of cells in the suspension was determined as described above. For the in vitro survival assays, nodose ganglia were dissected from E8 chick embryos as described (Lindsay et al., 1985). The trypsinized ganglia (20 minutes, 0.1% trypsin) were preplated for 3 hours as described above. Neurons were plated on polyornithine-laminincoated 48-well plates (Costar) at a density of 2000 cells/well in Ham’s F14 medium (Gibco) supplemented with 10% horse serum (Boehringer Mannheim). Neuronal survival was determined by counting phase-bright cells with neuronal morphology in duplicate wells after 24 hours.

Radiolabelling of BDNF

Radiolabelling of BDNF was performed using a modification of the lactoperoxidase-labelling protocol described for NT-3 (Rodríguez-Tébar et al., 1992). 10 μg of Vaccinia-derived mouse BDNF were incubated in a sodium phosphate buffer (0.3 M, pH 7.2) for 1 hour on ice in a total volume of 40 μl containing 0.4 μg lactoperoxidase (Sigma), 50 μM H2O2 and 1.5 mCi 125I (100 mCi/ml, Amersham). After this incubation, the reaction was stopped by dilution to a final volume of 1 ml followed by an overnight incubation at 4°C. The incorporation of 125I into BDNF was determined by TCA-precipitation. Typically, 88-95% of the total radioactivity were incorporated into the protein resulting in a radiolabelled ligand with a specific activity of 150-200 counts/minute/pg. The biological activity of the radiolabelled material was identical to the native factor in an in vitro survival assay with embryonic sensory neurons (40 pg/ml). The radiolabelled material was found to be stable for at least 2 weeks at 4°C, but was used within 10 days following iodination.

Binding experiments

All binding experiments were performed in suspension at 4°C in a water bath with gentle shaking. 500 μl of cell suspension containing 2.5-10×105 cells/ml were incubated in 50 ml plastic tubes (Falcon) either in the absence (total binding) or presence of a 100-to 500-fold excess of native ligand for 45 or 60 minutes (unspecific binding). Subsequently, the radioligand was added and the suspension incubated for a further 60 minutes. Free radioactivity was determined by measuring total activity in the cell suspensions assuming pseudo-first order conditions in the samples. Bound radioactivity was separated from free by rapid centrifugation (3000 g, 1 minute) of 4 samples (100 μl each) in 700 μl plastic tubes (Eppendorf) followed by complete removal of the supernatant. The tubes were inverted and air-dried. The bound radioactivity was determined by counting the tips of the tubes in a gamma counter with an efficiency of 78% (Canberra Packard) until the values reached a 1% sigma level. The tetraplicate samples were averaged. All values referring to bound radioligand (fmoles/106 cells or percent of binding) represent specific binding determined by subtraction of unspecific binding from total binding. For the steadystate binding experiments, unspecific binding for each individual point of the binding curve was determined. In the inhibition experiments, 100% binding refers to specific binding in the absence of heterologous ligand, 0% binding represents the unspecific binding. The samples in the inhibition experiments including heterologous ligands were incubated in the presence of these ligand for 60 minutes prior to addition of the radioligand. Binding data were analysed using a curve fitting computer program (Grafit, Erythacus Software).

Cloning of a chick trkB cDNA

Five independent clones were identified with a probe encoding the tyrosine kinase domain of rat trkB. Restriction analysis revealed that these clones contained fragments of the same cDNA. The largest insert had a size of approximately 4 kb. The first 2699 base pairs starting at the 5′-end of the clone were sequenced on both strands (Fig. 1). The sequence contained an open reading frame of 2454 bp. The bases upstream of the first methionine match the consensus sequence for translation initiation sites (Kozak, 1987). The 818 amino acid protein encoded in the open reading frame showed a clear homology to rat trkB (76.8% overall identity on amino acid level) and to a lesser degree to human trkA and to rat trkC (51% and 57.1%). The homology to rat trkB 61.2% identical to its rat homologue, whereas chick and rodent BDNF are 92.4% identical (Isackson et al., 1991). In the intracellular domain, only a small number of amino acids were found to be exchanged between chick and rat (97% identity) indicating that the cDNA encodes the chicken homologue of the mammalian receptor tyrosine kinase trkB (chick trkB, ctrkB).

Northern blot analysis

Northern blot analysis with a riboprobe spanning the extracellular domain of chick trkB revealed at least 5 transcripts in chick brain (Fig. 2A). Their size was estimated to be 9.0, 8.1, 6.3, 5.0 and 1.6 kb, the 1.6 kb mRNA being more obvious on total RNA samples than on poly(A)+-enriched RNA (data not shown). Analysis of mRNA expression in non-neuronal tissues showed a specific enrichment of the 5.0 kb band in mesenchymal derivatives (data not shown). In contrast, the riboprobe derived from the 3′ untranslated region (3′-probe) of chick trkB detected predominantly the 9.0 kb mRNA (Fig. 2C). In order to address the question if the 9.0 kb mRNA detected with the 3′-probe also contained the sequence corresponding to the tyrosine kinase domain of ctrkB, a blot was hybridised with a cDNA probe spanning this domain (Fig. 2B). Like the 3′-probe, this probe detected the 9.0 kb transcript (compare Fig. 2B and 2C). A transcript of the same size was detected in total RNA prepared from embryonic day 8 dorsal root ganglia neurons (Fig. 2D).

In situ hybridisation

Unless indicated otherwise, the data in the following were obtained using the 3′-probe and are summarized in Table 1 for the peripheral nervous system.

Table 1.

Expression of ctrkB in the peripheral nervous system of the developing chick embryo

Expression of ctrkB in the peripheral nervous system of the developing chick embryo
Expression of ctrkB in the peripheral nervous system of the developing chick embryo

While ctrkB transcripts were detected early in most sensory ganglia, no expression was detected at any stage in BDNF non-responsive sympathetic and ciliary neurons. Unambiguous expression of chick trkB m-RNA was first seen in the developing trigeminal ganglion at E 3.5 (stage 23 of Hamburger and Hamilton, 1951). Subsequently, the levels of trkB mRNA increased and at E5, expression was confined to the large neurons in the ventral region of the ganglion. Little or no expression was seen in the early embryo in the dorsal portion of the ganglion. However, at later stages, trkB mRNA expression became evident also in the most dorsal part of the trigeminal ganglion although more diffuse and weaker compared with the strongly labelled, large neurons located in the distal most part of the ganglion. The expression of trkB in the dorsal part of the trigeminal ganglion was presumably associated with small sized neurons and/or the satellite cells surrounding them (Fig. 3A,B). In the nodose ganglion, transcripts could also be detected at E4. In contrast to the trigeminal ganglion, the localization of the labelled neurons did not reveal any particular segregation (Fig. 3C,D). About 50% of the neurons were trkB positive in some sagittal sections through the ganglion, but trkB-positive neurons were always found in close association with cells not expressing trkB. The developing acoustic and vestibular ganglion was found to be highly labelled at all the stages examined. In this ganglion, essentially all neurons contained ctrkB-mRNA (Fig. 3E) as revealed by the antisense 3′-probe (see Fig. 3F for the sense control). The developmental pattern of trkB mRNA expression in the dorsal root ganglia (DRG) was examined in detail, as these ganglia represent the main source of neurons utilised in both neuronal survival experiments and binding studies. At mid-thoracic level, ctrkB-transcripts were already detected at E3.5 and a substantial expression was maintained throughout the period studied (Fig. 4A-D). The vast majority of the labelled neurons (about 30% of the total population) was concentrated in the ventral and lateral part of the ganglion and represented mostly large-size neurons (Fig. 4C). ctrkB-positive cells were identified in the ventral horn of the spinal cord, especially in the areas containing motoneurons, but the levels of expression of ctrkB mRNA were substantially lower as compared with the sensory ganglia (Fig. 4E). In the retina, no evidence for ctrkB-positive ganglion cells was found between E6 and E12, although scattered positive cells were detected mainly in the developing inner nuclear layer (not shown). In contrast to the labelling observed with the 3′-probe, that detected with the 5′-probe also revealed non-neuronal structures. However, the neuronal labelling detected with the 5′-probe was completely overlapping with that observed with the 3′-probe. Non-neuronal structures detected with the 5′-probe included restricted sites of the ependyma and various mesenchymal cells, for example those surrounding the spinal cord (Fig. 4D). No such labelling was detected with the 3′-probe.

Fig. 3.

trkB mRNA expression in the trigeminal, nodose and acoustic ganglion. (A,B) Bright- and dark-field images of the E6 trigeminal ganglion (t). Large trkB-positive neurons are located in the ventral portion of the ganglion. More diffuse labelling is also visible in the most dorsal part of the ganglion, but no labelling is seen around the trigeminal nerve (arrows) or in the medial portion of the ganglion. (C,D) Bright- and dark-field image of the E4.5 nodose ganglion (ng). C is a higher magnification of D: note that the positive neurons are a subset of the total neuronal population and are scattered throughout the ganglion. (E,F) Dark-field image of the developing (E 6) acoustic ganglion with antisense (E) or sense (F) probe. Abbreviations: c, developing cochlea; g, ganglion. Scale bar shown in A is 65 μm for A, B and D and 32 μm for C, E and F.

Fig. 3.

trkB mRNA expression in the trigeminal, nodose and acoustic ganglion. (A,B) Bright- and dark-field images of the E6 trigeminal ganglion (t). Large trkB-positive neurons are located in the ventral portion of the ganglion. More diffuse labelling is also visible in the most dorsal part of the ganglion, but no labelling is seen around the trigeminal nerve (arrows) or in the medial portion of the ganglion. (C,D) Bright- and dark-field image of the E4.5 nodose ganglion (ng). C is a higher magnification of D: note that the positive neurons are a subset of the total neuronal population and are scattered throughout the ganglion. (E,F) Dark-field image of the developing (E 6) acoustic ganglion with antisense (E) or sense (F) probe. Abbreviations: c, developing cochlea; g, ganglion. Scale bar shown in A is 65 μm for A, B and D and 32 μm for C, E and F.

Fig. 4.

trkB mRNA expression in the spinal cord and in the dorsal root ganglia. Bright-(A) and dark-field (a) images of the dorsal portion of a sagittal section through a E4.5 embryo (the spinal cord, sc, is indicated), showing the expression of trkB mRNA as detected with the 3′-probe. Note the labelling in the dorsal root ganglia (large arrows), which is absent from the sympathetic chain (s) and the spinal nerves (small arrow) and from nonneuronal derivatives. B and C show sections of DRG at E6 and E14. (D) Dark-field image of the E6 spinal cord showing the pattern of labelling as detected with the 5′-probe. Note that the DRG are labelled (small arrow) as well a conspicuous ring of mesenchymal cells surrounding the spinal cord (large arrow). (E) Higher magnification of the ventral horn (vh) shows presumptive motoneurons labelled for trkB mRNA. Scale bar shown in A is 170 μm for A, a, D, 65 μm for B, C and 32 μm for E.

Fig. 4.

trkB mRNA expression in the spinal cord and in the dorsal root ganglia. Bright-(A) and dark-field (a) images of the dorsal portion of a sagittal section through a E4.5 embryo (the spinal cord, sc, is indicated), showing the expression of trkB mRNA as detected with the 3′-probe. Note the labelling in the dorsal root ganglia (large arrows), which is absent from the sympathetic chain (s) and the spinal nerves (small arrow) and from nonneuronal derivatives. B and C show sections of DRG at E6 and E14. (D) Dark-field image of the E6 spinal cord showing the pattern of labelling as detected with the 5′-probe. Note that the DRG are labelled (small arrow) as well a conspicuous ring of mesenchymal cells surrounding the spinal cord (large arrow). (E) Higher magnification of the ventral horn (vh) shows presumptive motoneurons labelled for trkB mRNA. Scale bar shown in A is 170 μm for A, a, D, 65 μm for B, C and 32 μm for E.

Binding of BDNF: ctrkB cell line versus sensory neurons

The binding of radioiodinated BDNF to E8-E9 neurons isolated from DRGs was compared with that to a cell line stably expressing the full-length ctrkB cDNA described above. In steady-state binding assays, 125I-BDNF was found to bind to A293-ctrkB cells with a Kd of 9.9±1.2×10−10 M (Fig. 5). The number of receptors per cell was determined to be 2.4±0.1×105 as calculated by Scatchard transformation of the binding data. Although the A293 parental cells types also appear to bind BDNF at high concentration, no significant binding was detected at low ligand concentrations and no saturation was observed in the range of concentrations of radioligand that we have tested (up to 3×10−9 M). We next investigated whether the A293-ctrkB cell line displays significant numbers of high-affinity binding sites for BDNF. Steady-state binding experiments were performed at low ligand concentrations (below 3.5×10−11 M), using sensory neurons as a point of comparison (Fig. 6). At these concentrations no binding to parental A293 cells could be measured (data not shown). High-affinity BDNF-binding sites (3600±500 receptors/cell, Kd 1.5±0.5×10-11 M) were detected only on sensory neurons. No indication of saturable high-affinity binding was found for A293-ctrkB cells. Analysis of the binding data shown in Fig. 6 indicates that the specific binding seen with the A293-ctrkB cells can be attributed to the BDNF-binding sites described in Fig. 5, which are present in very large numbers on these cells (see above). The limit of detection for high-affinity binding sites in this assay was estimated to be about 200 receptors/cell corresponding to about 0.1% of the total receptor population on A293-ctrkB cells. As neuronal high-affinity sites for BDNF are characterised by a high ligand specificity, we then compared the binding of 125I-BDNF to A293-ctrkB cells and embryonic DRG neurons in the presence of various concentrations of heterologous neurotrophins. In these experiments, we used 125I-BDNF at a concentration of 3×10−11 M as at this low ligand concentration, significant binding is seen with both the ctrkB cells and sensory neurons (see above). Furthermore, the binding obtained with primary neurons is largely specific for high-affinity receptors because the contribution of the low-affinity binding sites to the total binding at this concentration is small, due to positive cooperativity of the BDNF binding to these sites (see Rodríguez-Tébar et al., 1990 and 1992 for discussion). In these experiments, differences were found between ctrkB and sensory neurons in all cases tested. Thus, the inhibition of 125I-BDNF binding by NGF was more effective on neurons than on A293-ctrkB cells: whereas 50% reduction of BDNF binding to DRG neurons was achieved with 2× 10−8 M NGF (see also Rodríguez-Tébar et al., 1990), no significant reduction was observed for ctrkB cells throughout the concentration range tested (Fig. 7A). Conversely, while inhibition of BDNF binding to DRG neurons by 50% occurred at 1.1×10−9 M NT-3, the same decrease was obtained with an approximately 10-fold lower concentration (1.3×10−10 M) on ctrkB cells (Fig. 7B). Xenopus NT-4 resulted in an efficient and complete inhibition of BDNF binding to ctrkB cells, 50% inhibition being observed at 7.3×10−11 M (Fig. 7C). Also with DRG neurons, inhibition by NT-4 was detected already at very low concentrations, but this inhibition of 125I-BDNF binding was incomplete (Fig. 7C). The inhibition curve is best fitted by a 2-site model for BDNF receptors on DRG neurons, in which NT-4 blocks 50% of BDNF binding to the first site at a concentration of 1.1×10−10 M. The second BDNF-binding site (representing about 25% of the total number of high-affinity BDNF receptors on E8 DRG neurons) is resistant to inhibition by NT-4.

Fig. 5.

Binding of 125I-BDNF to A293-ctrkB and A293 parental cells. Cells (400.000 cells/ml for A293-ctrkB and 480.000 cells/ml for A293) were incubated at 4°C with various concentrations of ligand. The results are mean of 4 determinations±s.d. Linear transformation of the binding data (insert) reveals a Kd of 9.9±1.2×10−10 M and 2.4±0.12×105 binding sites per cell.

Fig. 5.

Binding of 125I-BDNF to A293-ctrkB and A293 parental cells. Cells (400.000 cells/ml for A293-ctrkB and 480.000 cells/ml for A293) were incubated at 4°C with various concentrations of ligand. The results are mean of 4 determinations±s.d. Linear transformation of the binding data (insert) reveals a Kd of 9.9±1.2×10−10 M and 2.4±0.12×105 binding sites per cell.

Fig. 6.

Binding of 125I-BDNF to DRG neurons and A293-ctrkB cells. 125I-BDNF in a low concentration range (0.25-3.5×10−11 M) was incubated with either transfected A293 cells (330.000 cells/ml) or E8 DRG (400.000 cells/ml). The results are means of quadruple samples±s.d. (A) Saturable high-affinity binding of 125I-BDNF was detected with DRG neurons (Kd 1.5±0.5×10−11 M and 3600±500 receptors per cell). No saturable high-affinity binding was obtained with A293-ctrkB cells. The binding to these cells measured at low concentrations of radioligand (solid line) fits the curve predicted for the low-affinity binding sites on these cells (dashed curve, calculated from the data obtained in Fig. 5). (B) No indication for a saturable binding site was found by Scatchard transformation of the binding data for A293-ctrkB.

Fig. 6.

Binding of 125I-BDNF to DRG neurons and A293-ctrkB cells. 125I-BDNF in a low concentration range (0.25-3.5×10−11 M) was incubated with either transfected A293 cells (330.000 cells/ml) or E8 DRG (400.000 cells/ml). The results are means of quadruple samples±s.d. (A) Saturable high-affinity binding of 125I-BDNF was detected with DRG neurons (Kd 1.5±0.5×10−11 M and 3600±500 receptors per cell). No saturable high-affinity binding was obtained with A293-ctrkB cells. The binding to these cells measured at low concentrations of radioligand (solid line) fits the curve predicted for the low-affinity binding sites on these cells (dashed curve, calculated from the data obtained in Fig. 5). (B) No indication for a saturable binding site was found by Scatchard transformation of the binding data for A293-ctrkB.

Fig. 7.

Inhibition of 125I-BDNF Binding to A293-ctrkB cells and high-affinity BDNF sites on DRG neurons by heterologous neurotrophins. All experiments were performed at a 3×10−11 M concentration of 125I-BDNF. A293-ctrkB cells and embryonic DRG neurons were preincubated with various concentrations of heterologous ligand for 60 minutes, followed by addition of radioligand for another 60 minutes. For each data point, the bars represent the mean of a tetraplicate determination±s.d. (A) Inhibition by NGF: binding of radiolabelled BDNF to A293-ctrkB cells was not significantly inhibited by NGF. An inhibition constant (Ki) of 2×10−8 M was found for NGF and DRG neurons. (B) Inhibition by NT-3: 50% inhibition of 125I-BDNF binding was obtained at 1.3×10−10 M NT-3 for A293-ctrkB and 1.1×10−9 M for DRG neurons. (C) Inhibition by NT-4: binding of radioiodinated BDNF to A293-ctrkB cells could be completely blocked by Xenopus NT-4 with a Ki of 7×10−11 M. 75% of the BDNF binding to DRG neurons was inhibited with an Ki 1.1×10−10 M by NT-4. The remaining 25% of BDNF receptors could not be blocked by the concentrations of NT-4 tested.

Fig. 7.

Inhibition of 125I-BDNF Binding to A293-ctrkB cells and high-affinity BDNF sites on DRG neurons by heterologous neurotrophins. All experiments were performed at a 3×10−11 M concentration of 125I-BDNF. A293-ctrkB cells and embryonic DRG neurons were preincubated with various concentrations of heterologous ligand for 60 minutes, followed by addition of radioligand for another 60 minutes. For each data point, the bars represent the mean of a tetraplicate determination±s.d. (A) Inhibition by NGF: binding of radiolabelled BDNF to A293-ctrkB cells was not significantly inhibited by NGF. An inhibition constant (Ki) of 2×10−8 M was found for NGF and DRG neurons. (B) Inhibition by NT-3: 50% inhibition of 125I-BDNF binding was obtained at 1.3×10−10 M NT-3 for A293-ctrkB and 1.1×10−9 M for DRG neurons. (C) Inhibition by NT-4: binding of radioiodinated BDNF to A293-ctrkB cells could be completely blocked by Xenopus NT-4 with a Ki of 7×10−11 M. 75% of the BDNF binding to DRG neurons was inhibited with an Ki 1.1×10−10 M by NT-4. The remaining 25% of BDNF receptors could not be blocked by the concentrations of NT-4 tested.

Neuronal survival response elicited by BDNF in the presence of heterologous neurotrophins

Using cultured neurons isolated from the nodose ganglion of E8 chicks, about 50% of the neurons survive in the presence of BDNF (when used at saturation), whereas NT-3 supports the survival of about 25% of these neurons (Hohn et al., 1990; Götz et al., 1992). The effects of BDNF and NT-3 are additive (Götz et al., 1992; Fig. 8A). In a binding experiment with nodose ganglion neurons (data not shown), NT-3 used at 4×10−9 M prevented the binding of 125I-BDNF (3×10−11 M) to E8 nodose neurons as observed with the DRG neurons (Fig. 7B). In view of these results indicating that NT-3 at elevated concentrations is able to prevent the binding of BDNF to the high-affinity BDNF receptors on sensory neurons and on A293-ctrkB cells, we analysed the BDNF-induced survival response of nodose ganglion neurons in the presence of high concentrations of NT-3. At concentrations high enough to completely saturate the highaffinity BDNF receptors with NT-3 (20 μg/ml or 7.7×10−7 M, Fig. 8A), NT-3 supported the survival of 28±1% of the neurons. These effects on neuronal survival are not higher than those seen at 1 ng/ml NT-3 (23±4%), indicating that the occupancy of the BDNF receptors by NT-3 does not result in a productive interaction as assessed by neuronal survival. In addition, the effects of BDNF were completely blocked by this high concentration of NT-3 (20 μg/ml). Low concentrations of BDNF (0.1 ng/ml) support the survival of 34±1% of the nodose neurons, and these effects are additive with those of NT-3 when used at 1 ng/ml (65±6%, see Fig. 8A). At 0.1 ng/ml of BDNF in the presence of 20 μg/ml NT-3 only 28±2% of the neurons survived, the same number of neurons that is supported by NT-3 alone (Fig. 8A). This indicates that binding of NT-3 to high-affinity BDNF receptors on nodose ganglion neurons is not sufficient to induce neuronal survival of BDNF-responsive neurons in culture, and that NT-3 interacts in an antagonistic manner with neuronal high-affinity BDNF receptors.

Fig. 8.

In vitro survival assay with embryonic nodose ganglion neurons. Neurons prepared from E8 nodose ganglia were plated either in the presence or absence of neurotrophins as indicated. The number of surviving neurons was determined after 24 hours. Results are means of duplicate determinations. (A) NT-3 does not induce the survival of BDNF responsive neurons at concentrations high enough to saturate the high-affinity BDNF receptors. (B) NT-4 effects on nodose ganglion neurons are smaller than BDNF effects although both effects are not additive. The neurotrophins were added at a final concentration of 1 ng/ml. Higher concentrations did not lead to higher survival.

Fig. 8.

In vitro survival assay with embryonic nodose ganglion neurons. Neurons prepared from E8 nodose ganglia were plated either in the presence or absence of neurotrophins as indicated. The number of surviving neurons was determined after 24 hours. Results are means of duplicate determinations. (A) NT-3 does not induce the survival of BDNF responsive neurons at concentrations high enough to saturate the high-affinity BDNF receptors. (B) NT-4 effects on nodose ganglion neurons are smaller than BDNF effects although both effects are not additive. The neurotrophins were added at a final concentration of 1 ng/ml. Higher concentrations did not lead to higher survival.

Finally, in view of the binding of NT-4 to ctrkB and to the majority of BDNF high-affinity binding sites on neurons, we also tested the effects of NT-4 on nodose neurons (Fig. 8B). At concentrations of 1 ng/ml, NT-4 induced the survival of 30±3% of E8 nodose ganglion neurons. The effects of NT-4 were additive to those of NT-3 (51±4% versus 26±1% with NT-3 alone), but not to those of BDNF (44±1% BDNF plus NT-4, all factors at 1 ng/ml). This indicates that both NT-4 and BDNF induce neuronal survival via the same type of receptor, consistent with the results obtained with ctrkB. However, the effects of BDNF alone (51±4%) were found to be larger than the effects of NT-4 (30±3%) when used at saturating concentrations, suggesting that the effects of NT-4 might be mediated by a subpopulation of high-affinity BDNF receptors.

In this study, we isolated a full-length cDNA clone encoding chick trkB, currently the most likely functional BDNF receptor candidate. This clone allowed us to study the expression pattern of the chick ctrkB gene in neuronal populations known to respond to BDNF during development and to examine the binding of 125I-BDNF to ctrkB-expressing cells in comparison to BDNF-responsive neurons.

Northern blot analysis

Five different mRNAs coding for the extracellular domain of chick trkB were revealed by northern blot analysis using a 5′-probe corresponding to this domain. One of these mRNAs (the largest transcript, 9 kb in length) was found to cross-hybridise with a probe spanning the tyrosine kinase domain of chick trkB. From this analysis, it appears likely that, in the chick, as observed before in rodents (Klein et al., 1990a; Middlemas et al., 1991), the trkB gene encodes for several transcripts, including several lacking the tyrosine kinase catalytic domain. It is worth noting that the shortest mRNA detected in the brain by northern blot analysis is 1.6 kb in length, i.e. substantially shorter than the sequence necessary to code for a full-length receptor. The 9 kb transcript was also detected by a 3′-probe corresponding to the non-translated sequence of trkB. In situ hybridisation results of experiments performed with this probe showed a much more restricted pattern of labelling as compared to the 5′-probe. In particular, whereas both probes labelled neuronal structures equally well, only the 5′-probe labelled also nonneuronal structures (ependyma, choroid plexus and various mesenchymal derivatives). The combination of the in situ data with the northern blot analysis performed with domain-specific probes leads to the conclusion that a tyrosine kinase containing trkB sequence is expressed in neurons, whereas truncated forms are contained in non-neuronal cells, as has been described before for rodents (Klein et al., 1990a; Middlemas et al., 1991). However, our analysis cannot rule out the presence of truncated receptors in neurons as well.

In situ hybridisation

In general, a very good correlation was found between the pattern of expression of trkB, particularly in developing peripheral ganglia, and the localization of neurons known to respond to BDNF. Thus, chick trkB expression starts very early during embryogenesis, before and during the period of naturally occurring neuronal death in peripheral sensory ganglia. This is consistent with the observation that BDNF plays a role early during gangliogenesis (Kalcheim et al., 1987) and prevents cell death in vivo in DRG and nodose ganglia (Hofer and Barde, 1988). Within the DRG and trigeminal ganglia, the location of trkB-positive neurons, found mostly in the larger lateroventral population, is complementary to that of the smaller, mediodorsal neurons, enriched in NGF-responsive neurons (Hamburger et al., 1981; Verge et al., 1992). Complementarity has also been recently reported for the expression of trk and trkB mRNA in the spinal ganglia of the rat (Carroll et al., 1992). In vitro studies using DRGs and, especially, trigeminal ganglia, have indicated that the survival effects of NGF and BDNF are additive or complementary, suggesting different subpopulations of responsive neurons (Lindsay et al., 1985, Davies et al., 1986). Finally, trkB is not expressed at detectable levels in peripheral ganglia such as ciliary and sympathetic ganglia, known not to respond to BDNF (Lindsay et al., 1985). The sympathetic neurons have also been shown to display only low affinity, but not high-affinity BDNF-binding sites (Rodríguez-Tébar and Barde, 1988).

In the CNS, the low levels of trkB mRNA seen over the anterior part of the spinal cord fit with recent results indicating that the addition of BDNF during development prevents the death of a portion of spinal motoneurons (Oppenheim et al., 1992). In fact, only one exception to the correlation between BDNF responsiveness and trkB expression was noted: the retinal ganglion cells. In vitro, embryonic retinal ganglion cells isolated from the chick (or rat) retinae have been shown to be dependent on BDNF for survival (J. E. Johnson et al., 1986; Rodríguez-Tébar et al., 1989). trkB expression in the retinal ganglion cell layer could not be detected in vivo between E6 and E12, although a small number of labelled cells were occasionally seen in the inner nuclear layer. In view of the bulk of our results indicating an excellent correlation between trkB expression and BDNF responsiveness, this result is surprising. It should be noted however that while BDNF responsiveness has been observed in vitro, it remains to be demonstrated that BDNF acts on retinal ganglion cells in vivo. If BDNF also acts specifically on retinal ganglion cells in vivo, the possibilities should be considered that this might be through receptors other than the ctrkB receptor studied here, or that ctrkB levels of expression substantially below those found on other BDNF-responsive cells are sufficient for a BDNF response.

BDNF receptors on A293-ctrkB cells versus DRG neurons

As expected from similar studies in rodents (Klein et al., 1991b; Soppet et al., 1991; Squinto et al., 1991), BDNF binds specifically to A293 cells stably transfected with a cDNA encoding ctrkB. A dissociation constant of 9.9 ×10−10 M was determined, which is very close to that reported for BDNF binding to NIH-3T3 fibroblasts transfected with murine trkB, 1.8×10−9 M (Soppet et al., 1991). This result indicates that the differences between the affinities of trkB receptors expressed on cell lines and BDNF-binding sites on sensory neurons of the chick cannot be simply explained by species differences. No evidence for significant numbers of BDNF receptors on A293-ctrkB cells comparable to the high-affinity binding sites on sensory neurons was obtained. The number of high-affinity BDNF sites on DRG neurons was higher (about 10-fold) than in two previous studies (Rodríguez-Tébar and Barde, 1988; Rodríguez-Tébar et al., 1990). This difference is likely to result from two reasons: the use of recombinant BDNF (as opposed to BDNF isolated from pig brain before) and lactoperoxidase-induced iodination (rather than the iodinated Bolton-Hunter reagent in previous studies). No differences in affinity or specificity upon challenge with other neurotrophins could be observed. At present, the molecular basis for the increase in the number of 125I-BDNF-binding sites is not clear. Our results thus indicate that the highaffinity binding characteristics of BDNF on peripheral sensory neurons cannot be completely accounted for by ctrkB expressed in a non-neuronal context. Since the discovery of the trk-family members as neurotrophin receptors, much discussion has followed as to whether or not the expression of a trk-family member alone can lead to high-affinity receptors, represented, for example, by trk homodimers (Jing et al., 1992), or if co-expression of other molecules such as p75 LANR are necessary (Battleman et al., 1993). While the emphasis has been placed so far on the discrepancies, it is striking to note that in all studies, the vast majority (90% or more) of the neurotrophin-binding sites formed by recombinant trk tyrosine kinases is of the lowaffinity type. The question thus arises as to whether lowaffinity trk receptors also exist in abundance on neurons of the peripheral nervous system, and we feel that the available evidence suggests that this is not the case. While it is clear that these neurons express large numbers of low-affinity neurotrophin receptors (in the order of tens of thousand per neuron), the molecular entity explaining them is much more likely to be p75LANR than any member of the trk-family. Firstly, on sensory neurons, low-affinity 125I-NGF binding can be prevented by unlabelled BDNF and NGF equally well (Rodríguez-Tébar et al., 1990), a typical feature of p75LANR, but not of trk. Secondly, binding of NGF to lowaffinity binding sites on chick sensory neurons (or PC12 cells) is completely prevented by antibodies directed against p75LANR (Weskamp and Reichardt, 1991). The inhibition of BDNF binding by other neurotrophins also revealed quantitative differences between ctrkB and high-affinity BDNF binding on sensory neurons. Qualitatively, however, the similarities are obvious: BDNF binding to both A293-c trkB cells and chick DRG neurons is inhibited best by NT-4 followed by NT-3. NGF did not prevent BDNF binding to A293-ctrkB cells, even at the very high concentrations used to prevent BDNF binding to sensory neurons (10,000-fold excess). In general, these results are reminiscent of what has been observed with recombinant rodent trkB receptors (Klein et al., 1991b; Squinto et al., 1991; Ip et al., 1993). While NT-3 prevents binding of BDNF to both A293-ctrkB cells and sensory neurons, a 10-fold higher concentration of NT-3 is needed to prevent BDNF binding to neurons than to A293-c trkB cells. This is in line with the findings of Ip et al. (1993), indicating that the cross-linking of BDNF to trkB expressed in transfected fibroblast can be prevented by NT-3 much more efficiently than when BDNF is crosslinked to trkB on rat cortical neurons. Although the results of previous studies suggested that NT-3 is simply a poorer agonist of trkB when the latter is expressed on PC12 cells rather than on fibroblasts (Ip et al., 1993), our study with embryonic neurons reveals that NT-3 acts as an antagonist: when used at concentrations sufficient to prevent the interaction of BDNF with its receptor (either on neurons or on A293-ctrkB cells), NT-3 completely blocks the survival of BDNF-responsive nodose neurons. It is interesting to note that the interaction of NT-3 with heterologous neurotrophin receptors can be either antagonistic (this study, BDNF receptors) or agonistic: the occupancy of the high-affinity NGF sites on E11 sympathetic neurons by NT-3, which occurs at concentrations similar to those necessary to occupy the BDNF receptors, leads to the survival of these neurons, and not to a blockade of the effects of NGF (Dechant et al., 1993). Thus the interaction of NT-3 on sympathetic neurons with trk (the most likely NGF receptor) is agonistic, whereas that of NT-3 with ctrkB on sensory neurons is antagonistic.

Another indication for differences between BDNF receptors on neurons versus A293-ctrkB cells was revealed by the inhibition experiments performed with Xenopus NT-4. Again, while the A293-ctrkB cells behave like fibroblasts transfected with rodent trkB (50% inhibition of BDNF binding by almost equimolar concentrations of either NT-4 or BDNF, Klein et al., 1992; Ip et al., 1993), the sensory neurons display a small number of BDNF high-affinity binding sites that can be detected in spite of the presence of NT-4. The most likely explanation for this phenomenon is heterogeneity of the high-affinity BDNF receptors on embryonic neurons. Such a heterogeneity in high-affinity neurotrophin-binding sites has already been proposed for NGF and NT-3 (Weskamp and Reichardt, 1991; Rodríguez-Tébar et al., 1993). The incomplete inhibition of BDNF binding obtained with NT-4 is also reflected in the in vitro survival assay with sensory neurons. At saturating concentrations, the effects of NT-4 and BDNF are not additive, but the effects of NT-4 used alone are clearly smaller than those of BDNF. It thus appears that NT-4 is able to recruit only a fraction of the BDNF receptors to induce the survival of sensory neurons.

What could explain the differences in affinity and neurotrophin specificity between ctrkB and BDNF receptors on peripheral sensory neurons? It is possible that when expressed in such neurons, ctrkB only exists in dimeric form, thought to have a higher affinity (Jing et al., 1992). Also, other forms of ctrkB might be found in sensory neurons as a result of subtle alternative-splicing events not detected by northern blot analysis (as has been observed with for example agrin, Ruegg et al., 1992). In this context, it is interesting to note that all trkB cDNAs from rodents and chick were so far isolated from the adult CNS, and it remains to be shown that the sequence of the extracellular domain of ctrkB is identical in CNS and sensory neurons. Other possible mechanisms modifying ctrkB-binding properties might result from post-translational, neuron-specific modifications of the ctrkB protein or protein-protein interactions. p75LANR is a natural candidate to suggest as it represents the major source of low-affinity neurotrophin binding on neurons (see above). What is needed to change ctrkB from a receptor having many characteristics of a high-affinity BDNF receptor on peripheral neurons to one having all of them is an interesting question for the future.

We thank Dr Alfredo Rodríguez-Tébar (Madrid) for many helpful discussions and suggestions. We thank Mrs Claudia Cáp for her help with DNA synthesis and sequencing and Dr Marc Ballivet (Geneva) for his generous gift of the chick brain library.

Andersson
,
S.
,
Davis
,
D. N.
,
Dahlbäck
,
H.
,
Jörnvall
,
H.
and
Russell
,
D. W.
(
1989
).
Cloning, structure and expression of the mitochondrial cytochrome P-450 sterol 26-hydroxylase, a bile acid biosynthetic enzyme
.
J. Biol. Chem
.
264
,
8222
8229
.
Barbacid
,
M.
,
Lamballe
,
F.
,
Pulido
,
D.
and
Klein
,
R.
(
1991
).
The trk family of tyrosine kinase receptors
.
BBA Rev. Cancer
1072
,
115
127
.
Barde
,
Y.-A.
(
1989
).
Trophic factors and neuronal survival
.
Neuron
2
,
1525
1534
.
Battleman
,
D. S.
,
Geller
,
A. I.
and
Chao
,
M. V.
(
1993
).
HSV-1 vector-mediated gene transfer of the human nerve growth factor receptor p75hNGFR defines high-affinity NGF binding
.
J. Neurosci
.
13
,
941
951
.
Berkemeier
,
L. R.
,
Winslow
,
J. W.
,
Kaplan
,
D. R.
,
Nikolics
,
K.
,
Goeddel
,
D. V.
and
Rosenthal
,
A.
(
1991
).
Neurotrophin-5: a novel neurotrophic factor that activates trk and trkB
.
Neuron
7
,
857
866
.
Biffo
,
S.
,
Verhaagen
,
J.
,
Schrama
,
L.
,
Schotman
,
W.
,
Danho
,
K.
and
Margolis
,
F. L.
(
1990
).
B-50/GAP43 expression correlates with process outgrowth in the embryonic mouse nervous system
.
Eur. J. Neurosci
.
2
,
487
499
.
Carroll
,
S. L.
,
Silos-Santiago
,
I.
,
Frese
,
S. E.
,
Ruit
,
K. G.
,
Milbrandt
,
J.
and
Snider
,
W. D.
(
1992
).
Dorsal root ganglion neurons expressing trk are selectively sensitive to NGF deprivation in utero
.
Neuron
9
,
779
788
.
Chao
,
M. V.
(
1992
).
Neurotrophin receptors: A window into neuronal differentiation
.
Neuron
9
,
583
593
.
Chen
,
C.
and
Okayama
,
H.
(
1987
).
High-efficiency transformation of mammalian cells by plasmid DNA
.
Mol. Cell. Biol
.
7
,
2745
2752
.
Chomczynski
,
P.
and
Sacchi
,
N.
(
1987
).
Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction
.
Anal. Biochem
.
162
,
156
159
.
Cordon-Cardo
,
C.
,
Tapley
,
P.
,
Jing
,
S.
,
Nanduri
,
V.
,
O’Rourke
,
E.
,
Lamballe
,
F.
,
Kovary
,
K.
,
Klein
,
R.
,
Jones
,
K.
,
Reichardt
,
L. F.
and
Barbacid
,
M.
(
1991
).
The trk tyrosine kinase mediates the mitogenic properties of Nerve Growth Factor and Neurotrophin-3
.
Cell
66
,
173
183
.
Couturier
,
S.
,
Erkman
,
L.
,
Valera
,
S.
,
Rungger
,
D.
,
Bertrand
,
S.
,
Boulter
,
J.
,
Ballivet
,
M.
and
Bertrand
,
D.
(
1990
).
Alpha5, alpha3, and non-alpha3: three clustered genes encoding nicotinic acetylcholine receptor-related subunits
.
J. Biol. Chem
.
265
,
17560
17567
.
Davies
,
A. M.
,
Thoenen
,
H.
and
Barde
,
Y.-A.
(
1986
).
The response of chick sensory neurons to brain-derived neurotrophic factor
.
J. Neurosci
.
6
,
1897
1904
.
Dechant
,
G.
,
Rodríguez-Tébar
,
A.
,
Kolbeck
,
R.
and
Barde
,
Y.-A.
(
1993
).
Specific high affinity receptors for neurotrophin-3 on sympathetic neurons
.
J. Neurosci
.
13
,
2610
2616
.
Gey
,
G. O.
and
Gey
,
M. K.
(
1936
).
The maintenance of human normal cells and tumor cells in continous culture. I. Preliminary report: Cultivation of mesolimbic tumors and normal tissue and notes on methods of cultivation
.
Am. J. Cancer
27
,
45
76
.
Glass
,
D. J.
,
Nye
,
S. H.
,
Hantzopoulos
,
P.
,
Macchi
,
M. J.
,
Squinto
,
S. P.
,
Goldfarb
,
M.
and
Yancopoulos
,
G. D.
(
1991
).
TrkB mediates BDNF/NT-3-dependent survival and proliferation in fibroblasts lacking the low affinity NGF receptor
.
Cell
66
,
405
413
.
Götz
,
R.
,
Kolbeck
,
R.
,
Lottspeich
,
F.
and
Barde
,
Y.-A.
(
1992
).
Production and characterization of recombinant mouse neurotrophin-3
.
Eur. J. Biochem
.
204
,
745
749
.
Graham
,
F. L.
,
Smiley
,
J.
,
Russell
,
W. C.
and
Nairu
,
R.
(
1977
).
Characteristics of a human cell line transformed by DNA from human adenovirus type 5
.
J. Gen. Virol
.
36
,
59
77
.
Hallböök
,
F.
,
Ibanez
,
C. F.
and
Persson
,
H.
(
1991
).
Evolutionary studies of the Nerve Growth Factor family reveal a novel member abundantly expressed in Xenopus ovary
.
Neuron
6
,
845
858
.
Hamburger
,
V.
and
Hamilton
,
H. L.
(
1951
).
A series of normal stages in the development of the chick embryo
.
J. Morph
.
88
,
49
92
.
Hamburger
,
V.
,
Brunso-Bechtold
,
J. K.
and
Yip
,
J. W.
(
1981
).
Neuronal death in the spinal ganglia of the chick embryo and its reduction by nerve growth factor
.
J. Neurosci
.
1
,
60
71
.
Hempstead
,
B. L.
,
Martin-Zanca
,
D.
,
Kaplan
,
D. R.
,
Parada
,
L. F.
and
Chao
,
M. V.
(
1991
).
High-affinity NGF binding requires coexpression of the trk proto-oncogene and the low-affinity NGF receptor
.
Nature
350
,
678
683
.
Hofer
,
M. M.
and
Barde
,
Y. -A.
(
1988
).
Brain-derived neurotrophic factor prevents neuronal death in vivo
.
Nature
331
,
261
262
.
Hohn
A.
,
Leibrock
J.
,
Bailey
,
K.
and
Barde
,
Y. -A.
(
1990
).
Identification and characterization of a novel member of the nerve growth factor family
.
Nature
344
,
339
341
.
Holtzman
,
D. M.
,
Li
,
Y.
,
Parada
,
L. F.
,
Kinsman
,
S.
,
Chen
,
C.-K.
,
Valetta
,
J. S.
,
Zhou
,
J.
,
Long
,
J. B.
and
Mobley
,
W. C.
(
1992
).
p140trk mRNA marks NGF-responsive forebrain neurons: evidence that trk gene expression is induced by NGF
.
Neuron
9
,
465
478
.
Ip
,
N. Y.
,
Stitt
,
T. N.
,
Tapley
,
P.
,
Klein
,
R.
,
Glass
,
D. J.
,
Fandl
,
J.
,
Greene
,
L. A.
,
Barbacid
,
M.
and
Yancopoulos
,
G. D.
(
1993
).
Similarities and differences in the way neurotrophins interact with the Trk receptors in neuronal and nonneuronal cells
.
Neuron
10
,
137
149
.
Isackson
,
P. J.
,
Towner
,
M. D.
and
Huntsman
,
M. M.
(
1991
).
Comparison of mammalian, chicken, and Xenopus brain-derived neurotrophic factor coding sequences
.
FEBS Lett
.
285
,
260
264
.
Jing
,
S.
,
Tapley
,
P.
and
Barbacid
,
M.
(
1992
).
Nerve growth factor mediates signal transduction through trk homodimer receptors
.
Neuron
9
,
1067
1079
.
Johnson
,
D.
,
Lanahan
,
A.
,
Buck
,
C. R.
,
Sehgal
,
A.
,
Morgan
,
C.
,
Mercer
,
E.
,
Bothwell
,
M.
and
Chao
,
M.
(
1986
).
Expression and structure of the human NGF receptor
.
Cell
47
,
545
554
.
Johnson
,
J. E.
,
Barde
,
Y.-A.
,
Schwab
,
M.
and
Thoenen
,
H.
(
1986
).
Brain-derived neurotrophic factor supports the survival of cultured rat retinal ganglion cells
.
J. Neurosci
.
6
,
3031
3038
.
Kalcheim
,
C.
,
Barde
,
Y.-A.
,
Thoenen
,
H.
and
LeDouarin
,
N. M.
(
1987
).
In vivo effect of brain-derived neurotrophic factor on the survival of developing dorsal root ganglion cells
.
EMBO J
.
6
,
2871
2873
.
Kaplan
,
D. R.
,
Hempstead
,
B. L.
,
Martin-Zanca
,
D.
,
Chao
,
M. V.
and
Parada
,
L. F.
(
1991a
).
The trk proto-oncogene product: A signal transducing receptor for nerve growth factor
.
Science
252
,
554
558
.
Kaplan
,
D. R.
,
Martin-Zanca
,
D.
and
Parada
,
L. F.
(
1991b
).
Tyrosine phosphorylation and tyrosine kinase activity of the trk proto-oncogene product induced by NGF
.
Nature
350
,
158
160
.
Klein
,
R.
,
Parada
,
L. F.
,
Coulier
,
F.
and
Barbacid
,
M.
(
1989
).
trkB, a novel tyrosine protein kinase receptor expressed during mouse neural development
.
EMBO J
.
8
,
3701
3709
.
Klein
,
R.
,
Conway
,
D.
,
Parada
,
L. F.
and
Barbacid
,
M.
(
1990a
).
The trkB tyrosine protein kinase gene codes for a second neurogenic receptor that lacks the catalytic kinase domain
.
Cell
61
,
647
656
.
Klein
,
R.
,
Martin-Zanca
,
D.
,
Barbacid
,
M.
and
Parada
,
L. F.
(
1990b
).
Expression of the tyrosine kinase receptor gene trkB is confined to the murine embryonic and adult nervous system
.
Development
109
,
845
850
.
Klein
,
R.
,
Jing
,
S.
,
Nanduri
,
V.
,
O’Rourke
,
E.
and
Barbacid
,
M.
(
1991a
).
The trk proto-oncogene encodes a receptor for nerve growth factor
.
Cell
65
,
189
197
.
Klein
,
R.
,
Nanduri
,
V.
,
Jing
,
S.
,
Lamballe
,
F.
,
Tapley
,
P.
,
Bryant
,
S.
,
Cordon-Cardo
,
C.
,
Jones
,
K. R.
,
Reichardt
,
L. F.
and
Barbacid
,
M.
(
1991b
).
The trkB tyrosine protein kinase is a receptor for brain-derived neurotrophic factor and neurotrophin-3
.
Cell
66
,
395
403
.
Klein
,
R.
,
Lamballe
,
F.
,
Bryant
,
S.
and
Barbacid
,
M.
(
1992
).
The trkB tyrosine protein kinase is a receptor for neurotrophin-4
.
Neuron
8
,
947
956
.
Kozak
,
M.
(
1987
).
At least six nucleotides preceeding the AUG initiator enhance translation in mammalian cells
.
J. Mol. Biol
.
196
,
947
950
.
Lamballe
,
F.
,
Klein
,
R.
and
Barbacid
,
M.
(
1991
).
trkC, a New member of the trk family of tyrosine protein kinases, is a receptor for neurotrophin-3
.
Cell
66
,
967
979
.
Lindsay
,
R. M.
,
Thoenen
,
H.
and
Barde
,
Y.-A.
(
1985
).
Placode and neural crest-derived sensory neurons are responsive at early developmental stages to brain-derived neurotrophic factor
.
Dev. Biol
.
112
,
319
328
.
Martin-Zanca
,
D.
,
Barbacid
,
M.
and
Parada
,
L. F.
(
1990
).
Expression of the trk proto-onkogene is restricted to the sensory cranial and spinal ganglia of neural crest origin in mouse development
.
GenesDev
.
4
,
683
694
.
Meakin
,
S. O.
and
Shooter
,
E. M.
(
1992
).
The nerve growth factor family of receptors
.
Trends Neurosci
.
15
,
323
331
.
Merlio
,
J.-P.
,
Ernfors
,
P.
,
Jaber
,
M.
and
Persson
,
H.
(
1992
).
Molecular cloning of rat trkC and distribution of cells expressing messenger RNAs for members of the trk family in the rat central nervous system
.
Neuroscience
51
,
513
532
.
Middlemas
,
D. S.
,
Lindberg
,
R. A.
and
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
.
Oppenheim
,
R. W.
,
Qin-Wei
,
Y.
,
Prevette
,
D.
and
Yan
,
Q.
(
1992
).
Brain-derived neurotrophic factor rescues developing avian motoneurons from cell death
.
Nature
360
,
755
757
.
Radeke
,
M. J.
,
Misko
,
T. P.
,
Hsu
,
C.
,
Herzenberg
,
L. A.
and
Shooter
,
E. M.
(
1987
).
Gene transfer and molecular cloning of the rat nerve growth factor receptor: a new class of receptors
.
Nature
325
,
593
597
.
Rodríguez-Tébar
,
A.
and
Barde
,
Y.-A.
(
1988
).
Binding characteristics of brain-derived neurotrophic factor to its receptors on neurons from the chick embryo
.
J. Neurosci
.
8
,
3337
3342
.
Rodríguez-Tébar
,
A.
,
Jeffrey
,
P. L.
,
Thoenen
,
H.
and
Barde
,
Y.-A.
(
1989
).
The survival of chick retinal ganglion cells in response to brain-derived neurotrophic factor depends on their embryonic age
.
Dev. Biol
.
136
,
296
303
.
Rodríguez-Tébar
,
A.
,
Dechant
,
G.
and
Barde
,
Y.-A.
(
1990
).
Binding of brain-derived neurotrophic factor to the nerve growth factor receptor
.
Neuron
4
,
487
492
.
Rodríguez-Tébar
,
A.
,
Dechant
,
G.
,
Götz
,
R.
and
Barde
,
Y.-A.
(
1992
).
Binding of neurotrophin-3 to its neuronal receptors and interactions with nerve growth factor and brain-derived neurotrophic factor
.
EMBO J
.
11
,
917
922
.
Rodríguez-Tébar
,
A.
,
de la Rosa
,
E. J.
and
Arribas
,
A.
(
1993
).
Neurotrophin-3 receptors in the developing chicken retina
.
Eur. J. Biochem
,
211
,
789
794
.
Ruegg
,
M. A.
,
Tsim
,
K. W. K.
,
Horton
,
S. E.
,
Kröger
,
S.
,
Escher
,
G.
,
Gensch
,
E. M.
and
McMahan
,
U. J.
(
1992
).
The agrin gene codes for a family of basal lamina proteins that differ in function and distribution
.
Neuron
8
,
691
699
.
Sambrook
,
J.
,
Fritsch
,
E. F.
and
Maniatis
,
T.
(
1989
).
Molecular cloning: A labotatory manual
2nd ed.
Cold Spring Harbor Laboratory Press
.
Soppet
,
D.
,
Escandon
,
E.
,
Maragos
,
J.
,
Middlemas
,
D. S.
,
Reid
,
S. W.
,
Blair
,
J.
,
Burton
,
L. E.
,
Stanton
,
B. R.
,
Kaplan
,
D. R.
,
Hunter
,
T.
,
Nicolics
,
K.
and
Parada
,
L. F.
(
1991
).
The neurotrophic factors Brain-Derived Neurotrophic Factor and Neurotrophin-3 are ligands for the trkB tyrosine kinase receptor
.
Cell
65
,
21
.
Squinto
,
S. P.
,
Stitt
,
T. N.
,
Aldrich
,
T. H.
,
Davis
,
S.
,
Bianco
,
S. M.
,
Radziejewski
,
C.
,
Glass
,
D. J.
,
Masiakowski
,
P.
,
Furth
,
M. E.
,
Valenzuela
,
D. M.
,
Distefano
,
P. S.
and
Yancopoulos
,
G. D.
(
1991
).
trkB encodes a functional receptor for Brain-Derived Neurotrophic Factor and Neurotrophin-3 but not NGF
.
Cell
65
,
1
20
.
Suda
,
K.
,
Barde
,
Y.-A.
and
Thoenen
,
H.
(
1978
).
Nerve growth factor in mouse and rat serum: correlation between bioassay and radioimmunoassay determinations
.
Proc. Natl. Acad. Sci. USA
75
,
4042
4046
.
Sutter
,
A.
,
Riopelle
,
R. J.
,
Harris-Warrick
,
R. M.
and
Shooter
,
E. M.
(
1979
).
Nerve growth factor receptors: Characterization of two distinct classes of binding sites on chick embryo sensory ganglia cells
.
J. Biol. Chem
.
254
,
5972
5982
.
Tsoulfas
,
P.
,
Soppet
,
D.
,
Escandon
,
E.
,
Tessarollo
,
L.
,
Mendoza-Ramirez
,
J.-L.
,
Rosenthal
,
A.
,
Nikolics
,
K.
and
Parada
,
L. F.
(
1993
).
The rat trkC locus encodes multiple neurogenic receptors that exhibit differential response to neurotrophin-3 in PC12 cells
.
Neuron
10
,
975
990
.
Valenzuela
,
D. M.
,
Maisonpierre
,
P. C.
,
Glass
,
D. J.
,
Rojas
,
E.
,
Nuñez
,
L.
,
Kong
,
Y.
,
Gies
,
D. R.
,
Stitt
,
T. N.
,
Ip
,
N.
and
Yancopoulos
,
G. D.
(
1993
).
Alternative forms of rat TrkC with differerent functional capabilities
.
Neuron
10
,
963
974
.
Vara
,
J. A.
,
Portela
,
A.
and
Jimenez
,
A.
(
1986
).
Expression in mammalian cells of a gene from Streptomyces alboniger conferring puromycin resistance
.
Nucl. Acids Res
.
14
,
4617
4624
.
Verge
,
V. M. K.
,
Merlio
,
J.-P.
,
Grondin
,
J.
,
Ernfors
,
P.
,
Persson
,
H.
,
Riopelle
,
R. J.
,
Hökfelt
,
T.
and
Richardson
,
P. M.
(
1992
).
Colocalization of NGF binding sites, trk mRNA, and low-affinity NGF receptor mRNA in primary sensory neurons: Responses to injury and infusion of NGF
.
J. Neurosci
.
12
,
4011
4022
.
Weskamp
,
G.
and
Reichardt
,
L. F.
(
1991
).
Evidence that biological activity of NGF is mediated through a novel subclass of high affinity receptors
.
Neuron
6
,
649
663
.
Wilkinson
,
D. G.
,
Bailes
,
J. A.
,
Champion
,
J. E.
and
McMahon
,
A. P.
(
1987
).
A Molecular analysis of mouse development from 8 to 10 Days post coitum detects changes only in embryonic globin expression
.
Development
99
,
493
500
.