ABSTRACT
Neurotrophins signal through members of the trk family of tyrosine kinase receptors and are known to regulate several neuronal properties. Although initially characterized by their ability to prevent naturally occurring cell death of subsets of neurons during development, neurotrophins can also regulate the proliferation and differentiation of precursor cells. Here we report a novel involvement of neurotrophins in early development of the neural tube. We demonstrate that a functional trkB receptor is expressed by motor neuron progenitors in the ventral neural tube and that treatment of ventral neural tube explants with the trkB ligand Brain-Derived Neurotrophic Factor (BDNF) leads to a significant increase in the number of motor neurons. The only BDNF expression detectable at this stage is by a subset of ventrally projecting interneurons in the dorsal neural tube; ablating this region in vivo leads to a reduction of motor neuron numbers. This loss can be prevented by simultaneous treatment with BDNF. We propose that BDNF produced by dorsal interneurons stimulates proliferation and/or differentiation of motor neuron progenitors after anterograde axonal transport and release in proximity to the trkB-expressing motor neuron precursors, thereby coordinating development between dorsal and ventral regions of the neural tube.
INTRODUCTION
Two signalling centres involved in establishing dorsoventral pattern in the developing vertebrate neural tube have been identified (Tanabe and Jessell, 1996). Signals emanating from axial mesoderm, the notochord, induce ventral cell types as shown by embryonic manipulations and explant studies. Grafting pieces of ectopic notochord results in the induction of ectopic ventral cell phenotypes (Van Straaten et al., 1988; Yamada et al., 1991; Ericson et al., 1992), whereas the in vivo elimination of notochord leads to their absence (Van Straaten et al., 1991; Yamada et al., 1991; Ericson et al., 1992), accompanied by an expansion of the expression domains of genes normally restricted to the dorsal half of the neural tube into the ventral neural tube (Goulding et al., 1993). The signalling molecule Sonic hedgehog (SHH) has been implicated in these processes. It is expressed in the early notochord and its amino-terminal cleavage product is sufficient to induce ventral cell types, such as floor plate cells and motor neurons (Echelard et al., 1993; Roelink et al., 1994; Marti et al., 1995; Roelink et al., 1995), the induction of different cell types possibly requiring different concentrations (Roelink et al., 1995). Moreover, the role of SHH has been confirmed by antibody blockade (Marti et al., 1995; Ericson et al., 1996) and knock-out experiments (Chiang et al, 1996). Dorsal cell phenotypes, as represented by neural crest and commissural interneurons, can be induced in vitro by members of the Bone Morphogenetic Protein (BMP) family that are expressed early in the epidermal ectoderm flanking the neural plate (Liem et al., 1995).
Although BMPs and SHH have been identified as morphogens elaborating dorsoventral cell pattern in the neural tube, it is not known whether these molecules act at long-range in gradients that counteract each other or as short-range inducers acting indirectly through the induction of other signals.
Although progress is being made in understanding how the central nervous system becomes regionally specified, little is known about the intermediate steps between the earliest induction of precursor cells and the first appearance of differentiated cell types. For example, even though motor neurons can first be identified by their expression of the LIM homeo-domain gene Islet-1 (Isl-1) (Karlsson et al., 1990; Ericson et al., 1992), there are no known markers of motor neuron progenitor cells.
One class of secreted molecules involved in controlling the very early development of the vertebrate nervous system is the neurotrophin family. In higher vertebrates this consists of four members: NGF (Nerve growth factor), BDNF (Brain-derived neurotrophic factor), NT-3 (Neurotrophin-3), and NT-4/5 (Neurotrophin-4/5) (see Snider, 1994, for review). Although these molecules were initially characterized by their ability to prevent naturally occurring cell death of subsets of neurons during development, evidence has recently accumulated which suggests that neurotrophins can also regulate the proliferation and differentiation of precursor cells or support the survival of neuroblasts (Sieber-Blum, 1991; Kalcheim et al;, 1992; DiCicco-Bloom et al., 1993; Gaese et al., 1993; Averbuch-Heller et al., 1994; ElShamy and Ernfors, 1996; ElShamy et al., 1996). Neurotrophins initiate signal transduction by binding to members of the trk-family of transmembrane tyrosine kinase receptors. NGF signals via trkA, BDNF, NT-4/5 via trkB, whereas the main receptor for NT-3 is trkC (see Snider, 1994).
Here, we provide evidence that BDNF produced by ventrally projecting interneurons in the dorsal neural tube controls the generation of motor neurons by acting on trkB-expressing motor neuron progenitor cells in the ventral neural tube.
MATERIALS AND METHODS
Chick embryos
Fertilised chick embryos were incubated to the appropriate Hamburger and Hamilton (1951) stages at 38°C and 50% humidity.
In situ hybridisation and immunhistochemistry
Whole-mount in situ hybridisation was performed with antisense digoxigeninor FITC-labelled riboprobes as described by Heyman et al. (1995). The riboprobe spanning the extracellular region of trkB was derived form a PstI/PvuII subclone of chicken trkB (Dechant et al., 1993), the probe for the tyrosine kinase domain transcribed from the SspI-digested chicken trkB clone. Since the probe spanning the extracellular domain of trkB resulted in stronger signals compared to the tyrosine kinase domain probe, the extracellular probe was routinely used. To demonstrate the presence of a full length catalytic receptor, in situ hybridisations using the tyrosine kinase probe were performed with HH16, 19, 21 and 23 embryos, giving essentially the same results as the extracellular probe. The BDNF probe covered the entire coding region of chick BDNF (Herzog et al., 1994). The probe for chick Isl-1 (Tsuchida et al., 1994) was provided by T. Jessell. After staining, embryos were embedded in a gelatin/albumin mixture and transverse sections cut at 50 μm on a vibratome. For antibody staining, the embryos developed for whole-mount in situ hybridization were equilibrated in 30% sucrose overnight, embedded in Tissue Tec and 15 μm sections cut on a cryostat. Sections were collected on subbed slides and incubated overnight with the primary antibody (rabbit anti-LH2; generous gift from T. Jessell; rabbit anti-ISL-1/2, kindly provided by T. Edlund). Detection was performed using an ABC kit (Vector). Sections were examined on a Zeiss Axiophot.
BrdU labelling
HH 17 embryos were injected with approx. 5 μl of a 10 mg/ml 5-bro-modeoxyuridine (BrdU) solution and incubated for a further 3 hours. Embryos were fixed and in situ hybridization performed as described above. Then, embryos were sectioned on a cryostat and sections collected on subbed slides. Sections were incubated overnight with the anti-BrdU antibody (Bioscience Products AG). After incubation with a FITC-labelled secondary antibody, analysis was performed by confocal microscopy.
Notochord grafts
Small areas of the egg shell and the vitelline membrane were removed. To enhance visual contrast, diluted India ink was injected underneath the embryo and few drops of Howard’s ringer put onto the embryo to prevent dehydration. Using tungsten needles, an incision was made between hindbrain neural tube and the flanking mesoderm in HH9/10 embryos. Pieces of donor notochord dissected from HH11/12 embryos were inserted in the cut. Eggs were sealed and incubated further for 24 or 48 hours in the incubator. Embryos were collected, fixed and prepared for in situ hybridisation.
Axon tracing
Embryos were fixed in 3,5% paraformaldehyde, embedded in 20% gelatine and transverse sections were cut on a vibratome at 100 μm. DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethyl indocarbocyanine)-coated glass injection needles were briefly poked into the dorsal aspect of the hindbrain sections and labelled sections kept in fixative overnight at room temperature. Analysis was performed using laser scanning confocal microscopy.
Retrograde labelling and hindbrain dissociation
For retrograde labelling, the ventral hindbrain was lesioned in pinned-down HH17-18 embryo heads in the presence of a local pool of Rhodamine-conjugated dextran (Molecular probes). Heads were incubated in F12 medium at 38°C for 1.5-2 hours. For dissociation, heads were incubated in dispase for 20 minutes and the hindbrains dissected. Hindbrains were washed in Calcium/Magnesium-free medium containing 5 mM EDTA and dissociated using a fire-polished pipette. After several washes with F12 medium, cells were seeded on polylysine/laminin-coated dishes and fixed with paraformaldehyde after a 1-hour incubation at 38°C, 5% CO2. Detection of mRNAs and proteins was performed as describe above.
Neural tube explants
Rhombomere 3 was dissected from HH11 embryos with tungsten needles. The dorsal halves of the rhombomeres containing the pre-sumptive BDNF expression domain were then cut off and discarded. Rhombomere 3 was chosen because its boundaries are easy to identify in HH11 chick embryos and therefore provide landmarks that allow reproducible dissection of similar-sized pieces of the neural tube. The ventral halves were embedded in collagen gels and incubated in serum-free F12 medium containing SATO supplement (Bottenstein and Sato, 1979) with or without recombinant human BDNF (100 ng/ml; Regeneron) in 5% CO2 for 24 hours. Explants were fixed and whole-mount stained using anti-Isl antibody. After staining, explants were embedded in paraffin and stained nuclei counted in 5 μm sections.
Neural tube ablations and TUNEL staining
The dorsal quarter to half of HH11-12 hindbrains was cut off unilaterally with tungsten needles and the embryos incubated until they had reached HH17-19. Embryos were then fixed, stained for ISL proteins and the number of motor neurons counted on 6 μm wax sections in an area of the developing trigeminal motor nucleus on both sides of the hindbrain. The number of motor neurons on the lesioned side was expressed as a percentage of the number of motor neurons present on the unlesioned control side. BDNF-treated embryos were injected with 20 μl of a 0.25 μg/μl solution of recombinant human BDNF in ringer/5% chick serum once per day, a dose of neurotrophin known to result in neurotrophin concentrations in the embryo in a picogram/ml range (Ockel et al., 1996). TUNEL staining (Gavrieli et al, 1992) was performed approx. 24 hours after the ablations on 10 μm wax sections and apoptotic cells counted on 13-19 sections/embryo within the ablated area.
RESULTS
trkB expression in the early ventral neural tube
Using probes spanning the extracellular and tyrosine kinase domain, we mapped the expression pattern of chick trkB (Dechant et al., 1993) in the early neural tube. trkB expression starts at around HH9 with weak spots in the ventral hindbrain (not shown); by HH10, trkB is expressed in the ventral half of rhombomere 2 (Fig. 1A) with weaker and more ventrally restricted expression rostrally and caudally (not shown). At HH12, trkB becomes more restricted to the ventral region (Fig. 1B), and by HH13, expands into stripes on either side of the floor plate starting in the caudal midbrain and extending the entire length of the hindbrain, fading away towards the spinal cord and becoming restricted to the ventricular zone (see Fig. 1C).
trkB mRNA expression in the developing neural tube.(A) Transverse section through hindbrain rhombomere 2 in an HH10 embryo. trkB is expressed in the ventral half of rhombomere 2, but not in the floor plate.(B) Section through rhombomere 2 at HH12 showing that trkB expression is restricted to a more ventral area in the neural tube.(C)Transverse section at the level of the rostral hindbrain of a HH16 embryo showing trkB is expressed (in longitudinal stripes) on either side of the floor plate and with thickening of the neuroepithelium becomes restricted to the ventricular half of the neuroepithelium. (D) Proliferating BrdU+ cells are present within the trkB expression domain. Transverse section through the rostral hindbrain of a HH17 embryo showing that 30 minutes after a pulse with BrdU, BrdU+ cells are among the trkB mRNA-expressing cells (dark regions lateral of the midline). Scale bars, 60 μm.
trkB mRNA expression in the developing neural tube.(A) Transverse section through hindbrain rhombomere 2 in an HH10 embryo. trkB is expressed in the ventral half of rhombomere 2, but not in the floor plate.(B) Section through rhombomere 2 at HH12 showing that trkB expression is restricted to a more ventral area in the neural tube.(C)Transverse section at the level of the rostral hindbrain of a HH16 embryo showing trkB is expressed (in longitudinal stripes) on either side of the floor plate and with thickening of the neuroepithelium becomes restricted to the ventricular half of the neuroepithelium. (D) Proliferating BrdU+ cells are present within the trkB expression domain. Transverse section through the rostral hindbrain of a HH17 embryo showing that 30 minutes after a pulse with BrdU, BrdU+ cells are among the trkB mRNA-expressing cells (dark regions lateral of the midline). Scale bars, 60 μm.
Subsequently, the level of trkB expression increases and extends from the caudal midbrain caudally through the spinal cord at HH17 (not shown). Expression is also detected in the ventral midbrain (not shown). trkB expression is still detectable at HH25/26, the latest time point in this study (not shown). The observed expression patterns of trkB were the same - in the stages chosen for comparison - when using probes spanning either the tyrosine kinase domain or the extracellular region of trkB, indicating the presence of a full-length catalytic receptor.
trkB and Isl-1 expression at the same dorsoventral level in the neural tube
Both spatial and temporal characteristics of trkB expression in the chick neural tube are similar to those of the homeobox gene Isl-1, the earliest known marker of young postmitotic motor neurons (Ericson et al., 1992): Isl-1 expression starts at HH13 in the caudal midbrain and the hindbrain (not shown) and increases and expands into the entire spinal cord from approximately HH15 to HH17 (Ericson et al., 1992). In order to correlate the expression patterns of trkB and Isl-1, we performed double in situ hybridisation in HH18 embryos: strong Isl-1 expression spans approximately three-quarters of the neuroepithelium from the basal (pial) surface, leaving an area of weak or no expression towards the apical (ventricular) surface (Fig. 2A). trkB expression co-localizes at the same dorsoventral level as the Isl-1 expression and overlaps with it on the apical-basal axis (Fig. 2B). Immunohistochemical detection of ISL-1/2 reveals the presence of ISL+ cells within the trkB expression domain (Fig. 2C). Moreover, dissection and dissociation of HH17-18 hindbrains into a single cell suspension, followed by the sequential detection of both trkB mRNA and ISL-1/2 protein demonstrates the existence of a small proportion of cells expressing both trkB and ISL-1/2 (Fig. 2D). Therefore, trkB-expressing cells might represent proliferating progenitor cells giving rise to postmitotic ISL-1/2+ motor neurons, a possibility that is further suggested by the presence of proliferating, BrdU+ cells in the trkB domain (Fig. 1D).
Spatial and temporal overlap of trkB and Isl-1 expression and control of both trkB and Isl-1 expression by signals from the notochord. (A)Transverse section through the rostral hindbrain of a HH18 embryo showing Isl-1 mRNA expression (red) lateral to the floor plate. Expression is strongest in the outer three quarters of the neural tube and very weak towards the ventricular surface.(B)Simultaneous detection of trkB (blue) and Isl-1 (red) mRNA in a HH18 embryo reveals that both genes are expressed at the same level of the dorsoventral axis andthat the expression domains show a spatial overlap. Note the lateral patches of trkB expression in the neural tube, which are only present in rhombomeres 2 and 3. (C)Transverse section through the hindbrain of a HH16 embryo (showing one side of the ventral neural tube only) demonstrating that motor neurons expressing ISL-1/2 protein (brown) are present within the trkB expression domain (blue). (D) HH17-18 hindbrains were dissociated and double-stained for trkB mRNA (upper panel) and ISL-1/2 protein (lower panel). The arrows label a single cell expressing both trkB and ISL-1/2. (E) 24 hours after notochord grafting, trkB mRNA (blue) is already induced, whereas Isl-1 mRNA (red) is hardly detectable in the neural tube. The asterisk labels the grafted notochord, the dotted line highlights the midline of the neural tube. (F) 48 hours after notochord grafting the induction of trkB mRNA (blue) is paralled by strong expression of Isl-1 mRNA (red). An asterisk labels the grafted notochord; endogenous Isl-1 expression on the contralateral control side is indicated by the arrow. Scale bars in A, B,E,F, 60 μ m; C,D, 30 μ m.
Spatial and temporal overlap of trkB and Isl-1 expression and control of both trkB and Isl-1 expression by signals from the notochord. (A)Transverse section through the rostral hindbrain of a HH18 embryo showing Isl-1 mRNA expression (red) lateral to the floor plate. Expression is strongest in the outer three quarters of the neural tube and very weak towards the ventricular surface.(B)Simultaneous detection of trkB (blue) and Isl-1 (red) mRNA in a HH18 embryo reveals that both genes are expressed at the same level of the dorsoventral axis andthat the expression domains show a spatial overlap. Note the lateral patches of trkB expression in the neural tube, which are only present in rhombomeres 2 and 3. (C)Transverse section through the hindbrain of a HH16 embryo (showing one side of the ventral neural tube only) demonstrating that motor neurons expressing ISL-1/2 protein (brown) are present within the trkB expression domain (blue). (D) HH17-18 hindbrains were dissociated and double-stained for trkB mRNA (upper panel) and ISL-1/2 protein (lower panel). The arrows label a single cell expressing both trkB and ISL-1/2. (E) 24 hours after notochord grafting, trkB mRNA (blue) is already induced, whereas Isl-1 mRNA (red) is hardly detectable in the neural tube. The asterisk labels the grafted notochord, the dotted line highlights the midline of the neural tube. (F) 48 hours after notochord grafting the induction of trkB mRNA (blue) is paralled by strong expression of Isl-1 mRNA (red). An asterisk labels the grafted notochord; endogenous Isl-1 expression on the contralateral control side is indicated by the arrow. Scale bars in A, B,E,F, 60 μ m; C,D, 30 μ m.
Induction of trkB expression by signals from the notochord
Differentiation of floor plate cells and ventral neural tube-derived neurons, e.g. motor neurons, is inducible by signals from the notochord, presumably via the signalling molecule SHH (Van Straaten et al., 1988; Yamada et al., 1991; Roelink et al., 1994; Marti et al., 1995; Roelink et al., 1995; Chiang et al., 1996). To assess if trkB expression in the ventral neural tube is influenced by the same signals, we grafted pieces of notochord adjacent to the hindbrain neural tube of HH9/10 embryos. After only 24 hours, we observed induction of trkB expression that is restricted, as endogenous expression, to the ventricular half of the neuroepithelium, accompanied by a barely detectable induction of Isl-1 close to the pial surface (Fig. 2E). Between 24 and 48 hours after grafting, the induction of trkB expression is paralleled by a strong induction of Isl-1 expression (Fig. 2F). The more dorsal extension of the ectopic Isl-1 domain compared to the ectopic trkB domain in Fig. 2F is probably a secondary effect resulting from the dorsal migration of postmitotic branchiomotor neurons in the hindbrain towards their dorsally located exit points (Heaton and Moody, 1980; Covell and Noden, 1989; Simon et al., 1994). These results indicate that expression of trkB and Isl-1 are both controlled by signals from the notochord. The induced expression of trkB precedes and parallels the later induction of Isl-1, further supporting the possibility that trkB expression identifies motor neuron progenitors.
Expression of the trkB ligand BDNF in the dorsal neural tube
Next, we examined the expression pattern of the trkB ligand, BDNF (Leibrock et al., 1989), which has previously been shown to be localised to the dorsal region of the spinal neural tube (Kahane et al., 1996). First detectable at HH13 in the caudal hindbrain (rhombomere 6, Fig. 3A), BDNF expression extends both rostrally and caudally to include the entire length of the hindbrain between HH15/HH16, fading away in the anterior spinal cord. At later stages, the level of BDNF expression increases (Fig. 3B) and becomes detectable in the dorsal spinal neural tube. Later, at about HH23, the BDNF expression level decreases and is not detectable after HH24/25 (not shown). BDNF is expressed across the entire apical-basal thickness of the dorsal neuroepithelium. Cells in the dorsal-most region of the neural tube adjacent to the roof plate, however, do not express BDNF. These results show that the expression of both the trkB receptor and its ligand BDNF overlap temporally for an extended period of development from approximately HH13 to HH23.
Expression of BDNF mRNA by interneurons in the dorsal neural tube. (A,B) Transverse sections through rhombomere 6 of HH13 (A) and HH19 (B) embryos, showing expression of mRNA (blue) in the dorsal neural tube. BDNF is expressed throughout the entire apical-basal width of the neuroepithelium; the doral-most region of the neural tube does not express BDNF. (C) Transverse section through the hindbrain of a HH21 embryo showing the dorsal region of one side of the neural tube only (the ventricle is to the left). A subpopulation of LH-2+ interneurons (brown) overlaps in the outer neural tube with the BDNF expression domain (blue). (D) Transverse section through the hindbrain of a HH18 embryo labelled with DiI in the dorsal neural tube showing that axons of ventrally projecting interneurons (inter) have already reached the floor plate and established contralateral projections. inter, location of the interneuron cell bodies; IV, fourth ventricle; noto, notochord; roof, roof plate. Scale bars in A,C, 30 μm; B,D, 60 μm.
Expression of BDNF mRNA by interneurons in the dorsal neural tube. (A,B) Transverse sections through rhombomere 6 of HH13 (A) and HH19 (B) embryos, showing expression of mRNA (blue) in the dorsal neural tube. BDNF is expressed throughout the entire apical-basal width of the neuroepithelium; the doral-most region of the neural tube does not express BDNF. (C) Transverse section through the hindbrain of a HH21 embryo showing the dorsal region of one side of the neural tube only (the ventricle is to the left). A subpopulation of LH-2+ interneurons (brown) overlaps in the outer neural tube with the BDNF expression domain (blue). (D) Transverse section through the hindbrain of a HH18 embryo labelled with DiI in the dorsal neural tube showing that axons of ventrally projecting interneurons (inter) have already reached the floor plate and established contralateral projections. inter, location of the interneuron cell bodies; IV, fourth ventricle; noto, notochord; roof, roof plate. Scale bars in A,C, 30 μm; B,D, 60 μm.
Ventrally projecting interneurons are located in the dorsal BDNF expression domain
How could BDNF produced only by cells in the dorsal neural tube reach trkB-expressing target cells that are restricted to the ventral tube? One possibility is that BDNF might be produced by dorsal relay interneurons, whose axons project circumferentially around the marginal neural tube to join ipsilateral and contralateral fascicles beside the floor plate (Yaginuma et al., 1990; Shiga et al., 1991). To test this possibility, BDNF and LH-2 (a homeodomain protein marking interneurons with ventrally projecting axons in the dorsal neural tube; Xu et al., 1993) were sequentially detected by in situ hybridisation and immunohistochemistry. LH-2+ interneurons are localized in the mantle zone of the dorsal hindbrain (Fig. 3C), overlapping with the more ventricular domain of BDNF expression, suggesting that BDNF might be synthesized by both young radially migrating cells and possibly also by a subpopulation of mature axon-bearing cells.
To demonstrate that BDNF is expressed by dorsal interneurons, these cells were retrogradely labelled from the ventral hindbrain in HH17-18 embryos. Labelled hindbrains were dissected, dissociated into a single cell suspension and BDNF mRNA detected. The presence of BDNF in a subpopulation of retrogradely labelled cells clearly demonstrates expression of BDNF by ventrally projecting interneurons in the dorsal neural tube (Fig. 4).
Expression of BDNF mRNA by ventrally projecting interneurons. After retrograde labelling of interneurons from the ventral neural tube, hindbrains were dissected and dissociated into single-cell suspensions. Subsequently, BDNF mRNA (A,C) as well as the retrograde label (B, D) were detected. A,B and C,D show corresponding images. Arrows mark retrogradely labelled cells expressing BDNF. Scale bar in A for all panels: 30 μm.
Expression of BDNF mRNA by ventrally projecting interneurons. After retrograde labelling of interneurons from the ventral neural tube, hindbrains were dissected and dissociated into single-cell suspensions. Subsequently, BDNF mRNA (A,C) as well as the retrograde label (B, D) were detected. A,B and C,D show corresponding images. Arrows mark retrogradely labelled cells expressing BDNF. Scale bar in A for all panels: 30 μm.
In another set of experiments, we performed DiI labelling in the dorsal aspect of the hindbrain at different embryonic stages to reveal the developing projection pathways of dorsal interneurons. The axonal projections of the interneurons follow a circumferential pathway and reach the ventral midline at HH15 (not shown). Labelling at HH18 revealed that commissural axons have already crossed the midline and established contralateral projections (Fig. 3D).
BDNF stimulates the generation of motor neurons
To test for possible roles of BDNF in motor neuron generation, we used an in vitro explant assay. ISL-1/2+ cells were quantified in ventral halves of HH11 rhombomere 3 containing the trkB-expressing cell population (Fig. 5A). In control explants cultured in the absence of BDNF, an average of 118±7 s.e.m. (n=9) ISL-1/2+ cells was present. In contrast, explants cultured in the presence of BDNF contained an average of 163±8 s.e.m. (n=8) ISL-1/2+ cells, representing an approx. 40% increase compared to the control value (Fig. 5B).
BDNF increases the number of motor neurons generated in explants of the ventral neural tube. (A) Section through a collagen gel-embedded explant of ventral rhombomere 3 after 24 hours of culture, stained for ISL-1/2. (B) Culturing explants of ventral rhombomere 3 in the presence of 100 ng/ml BDNF leads to an approx. 40% increase in the number of motor neurons compared to control explants cultured without BDNF. Scale bar, 30 μm.
BDNF increases the number of motor neurons generated in explants of the ventral neural tube. (A) Section through a collagen gel-embedded explant of ventral rhombomere 3 after 24 hours of culture, stained for ISL-1/2. (B) Culturing explants of ventral rhombomere 3 in the presence of 100 ng/ml BDNF leads to an approx. 40% increase in the number of motor neurons compared to control explants cultured without BDNF. Scale bar, 30 μm.
Dorsal neural tube regulates motor neuron generation in the ventral neural tube
To explore a possible influence of the dorsal neural tube on the development of motor neurons in the ventral neural tube, we performed ablation experiments. Using a similar experimental strategy, it has recently been shown that, after ablation of the dorsal neuroepithelium in the chick hindbrain, regeneration of identified interneuronal populations, particularly those in the dorsal aspect of the hindbrain, occurs only to a limited extent (Diaz and Glover, 1996). We ablated the dorsal hindbrain unilaterally in HH11-12 embryos and compared the number of motor neurons between operated and unoperated control sides after approximately 40 hours further development: we found a 17-40% reduction (average 22%±3.8 s.e.m.) in motor neuron number between the operated and control sides in 7/10 embryos (Fig. 6A). Three embryos showed no significant reduction.
Unilateral ablation of the dorsal hindbrain leads to a reduction in the number of motor neurons on the ablated side that can be prevented by treatment with BDNF. (A) Transverse section through the rostral hindbrain of a HH17 embryo whose dorsal region including the presumptive BDNF expression domain had been ablated unilaterally at HH11. As revealed by staining for ISL-1/2, the number of motor neurons is reduced on the ablated side compared to the non-operated control side. (B) Transverse section through the rostral hindbrain of an approx. HH18 embryo stained for ISL-1/2 demonstrating that treatment of unilaterally ablated embryos with BDNF prevents the reduction in motor neuron numbers shown in A. (C) Dorsal view of a hindbrain of a HH17 embryo unilaterally ablated at HH11 and stained 40 hours later for MafB mRNA, a gene expressed in the very dorsal region of the neural tube (I. McKay and A.L., unpublished data). Rostral is up. Even 40 hours after the operation the ablated region has not regenerated, as revealed by the lack of MafB expression on the ablated side. Arrows indicate sites of the dorsal ablation. Scale bars in A,B, 60 μm; C, 100 μm.
Unilateral ablation of the dorsal hindbrain leads to a reduction in the number of motor neurons on the ablated side that can be prevented by treatment with BDNF. (A) Transverse section through the rostral hindbrain of a HH17 embryo whose dorsal region including the presumptive BDNF expression domain had been ablated unilaterally at HH11. As revealed by staining for ISL-1/2, the number of motor neurons is reduced on the ablated side compared to the non-operated control side. (B) Transverse section through the rostral hindbrain of an approx. HH18 embryo stained for ISL-1/2 demonstrating that treatment of unilaterally ablated embryos with BDNF prevents the reduction in motor neuron numbers shown in A. (C) Dorsal view of a hindbrain of a HH17 embryo unilaterally ablated at HH11 and stained 40 hours later for MafB mRNA, a gene expressed in the very dorsal region of the neural tube (I. McKay and A.L., unpublished data). Rostral is up. Even 40 hours after the operation the ablated region has not regenerated, as revealed by the lack of MafB expression on the ablated side. Arrows indicate sites of the dorsal ablation. Scale bars in A,B, 60 μm; C, 100 μm.
Treating ablated embryos with 5 μg recombinant BDNF per day, a neurotrophin dose previously shown to result in low, physiological concentrations of neurotrophin proteins within tissues of chick embryos (Ockel et al., 1996), resulted in complete rescue of the number of motor neurons on the ablated side (106% ±2.2 s.e.m. of control side; n=7) (Fig. 6B). The ablated region did not regenerate during the survival period as revealed by the lack of restoration of dorsal gene expression (mafB) on the ablated side of the hindbrain (Fig. 6C). To determine whether BDNF acts as a survival factor for motor neuroblasts, we quantified the number of apoptotic cells as revealed by terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labelling (TUNEL) staining (Gavrieli et al., 1992). There was no significant change in the number of apoptotic cells in control sides as compared to ablated sides (1.2±1.4 and 0.95±1.2 s.d./section, resp.; n=4), suggesting that the major role of BDNF is in the control of proliferation and/or differentiation of trkB-expressing motor neuron precursors rather than in promoting their survival.
DISCUSSION
Our data suggest that BDNF stimulates the proliferation and/or differentiation of motor neuron precursors, a new function of BDNF different from its potential survival-promoting activity on postmitotic motor neurons (Snider, 1994; McKay et al., 1996). BDNF, the only trkB ligand known in chick, is expressed by relay interneurons in the dorsal neural tube and is most likely secreted in the ventral region. A functional trkB receptor is expressed by motor neuron progenitor cells in the ventral neural tube. trkB represents the first known marker identifying motor neuron precursors.
Although not exclusively labelling motor neurons, most ISL-1/2+ cells in the ventral neural tube are motor neurons (Ericson et al., 1992). Expression of the homoebox gene Isl-1 is the earliest marker so far identified for postmitotic motor neurons, and the importance of Isl-1 for motor neuron development has been demonstrated in Isl-1 knock out mice that are unable to generate motor neurons (Pfaff et al., 1996). Four lines of evidence suggest that trkB-expressing cells in the neural tube are precursor cells for motor neurons and indicate that trkB signalling is an important regulator of motor neuron development. First, expression of trkB and Isl-1 show a spatial and temporal overlap: trkB cells occupy the ventricular region of the shared dorsoventral domain whereas Isl-1 cells occupy the mantle region, suggesting strongly that the onset of trkB expression precedes that of Isl-1 in the same cells. A small subset of motor neurons co-express trkB and ISL-1/2, suggesting that trkB becomes downregulated at around the time of the final mitosis. On the longitudinal axis, trkB is expressed in ventral midbrain, hindbrain and spinal cord, and therefore exactly matches the region of the neuraxis that generates motor neurons. If trkB acts to regulate the development of motor neuron progenitor cells, it would be expected to be expressed prior to the appearance of the earliest postmitotic motor neurons and marker genes such as Isl-1. This is exactly what we observed. The first postmitotic motor neurons in the hindbrain trigeminal motor nucleus (located in rhombomeres 2 and 3) are born at HH12 (Heaton and Moody, 1980; Covell and Noden, 1989). This is paralleled by Isl-1 expression around the same stage (not shown). trkB is expressed in the ventral region of rhombomere 2 at HH9 and is expressed throughout the presumptive motor column of the hindbrain by HH10.
Second, our data show that expression of both trkB and Isl-1 is controlled by notochord-derived signals. Induction of ventral phenotypes in the neural tube has been shown to be induced by signals from notochord and floor plate (Van Straaten et al., 1988; Yamada et al., 1991), most likely by the signalling molecule SHH, initially expressed by the notochord, later by both notochord and floor plate (Roelink, et al., 1994; Marti et al., 1995; Roelink et al., 1995). Grafting pieces of notochord as an additional source of ventralizing signals leads to the induction of ectopic trkB expression after 24 hours, and by 48 hours, the ectopic expression of trkB was paralleled by a very strong induction of Isl-1. Therefore, just as for the endogenous expression, trkB expression precedes the appearance of Isl-1 after ectopic induction by grafted notochord tissue. Moreover, the relatively small domain of trkB expression (presumably reflecting a rather small number of expressing cells) compared with the large domain of Isl-1 expression (reflecting a large number of expressing cells) is consistent with the notion of precursor cells being present in small numbers relative to the number of differentiated cells they generate.
Third, the in vitro explant assays revealed an effect of BDNF on the generation of motor neurons. Culturing ventral neural tube explants containing the trkB-expressing progenitor cells in the presence of BDNF lead to a approx. 40% increase in the number of motor neurons compared to control explants.
Finally, ablation of the dorsal neural tube, including the BDNF-expressing cells, results in a reduction in the number of motor neurons. This reduction is prevented by treatment with physiological levels of exogenous BDNF. Therefore, the ablation experiments demonstrate that the dorsal neural tube normally expressing BDNF is necessary for normal motor neuron development and that exogenous BDNF is sufficient to substitute for the loss of the dorsal neural tube. Therefore, reduction in motor neuron numbers after dorsal ablation cannot be explained, for example, by an enhanced dorsalization of the residual neural tube on the operated side. This explanation is also ruled out by the failure of the ablated side to restore dorsal gene expression. We have attempted to directly demonstrate an involvement of BDNF/trkB by blocking endogenous BDNF using trkB-Ig fusion proteins. So far, however, this has resulted in a small but statistically not significant reduction in motor neuron numbers, most likely due to ineffective tissue penetration by the fusion proteins (data not shown).
Although the in vitro assay and ablation systems do not enable us to elucidate the mechanism of BDNF action on motor neuron precursors, the known functions of neurotrophins during early development of the nervous system suggest two possibilities. First, BDNF might act as a mitogen on motor neuron progenitor cells. BrdU-incorporating cells are located within the trkB expression domain, and the observed increase in the number of ISL-1/2+ cells could be explained by a direct mitogenic effect of BDNF on these cells. BDNF is possibly only one of several mitogens acting on the precursors. Alter-natively, BDNF might influence the adoption of a motor neuronal fate by trkB-expressing precursor cells. If motor neuron fate in the hindbrain is determined during the last cell division of progenitors, as in the mammalian cortex (McConnell and Kaznowski, 1991), this role would also be consistent with the presence of BrdU+ cells among the trkB-expressing cells. Interestingly, NT-3 has been suggested to stimulate the differentiation of motor neurons from neuro-epithelial cells (Averbuch-Heller et al., 1994; Roelink et al., 1995). Since the expression of the NT-3 receptor trkC, however, colocalizes with postmitotic motor neurons and not with their precursors during early chick embryogenesis (unpublished data), NT-3 is unlikely to affect differentiation or proliferation of precursor cells. Rather, it might influence other aspects of motor neuron development.
Although it might also be expected that BDNF acts as a survival factor for motor neuron progenitors, we have seen little evidence for this: if endogenous BDNF were present in limiting amounts one would expect to see apoptosis within the trkB expression domain, which has not been observed (Homma et al., 1994). In addition, we were unable to detect an increase in the number of apoptotic cells in the ventral neural tube after ablations in the dorsal neural tube.
The question remains as to why knock-out mice lacking trkB, BDNF, Neurotrophin-4 (NT-4), and both BDNF and NT-4 do not show any apparent reduction in the number of motor neurons (Klein et al., 1993; Ernfors et al., 1994; Jones et al., 1994; Conover et al., 1995; Liu et al., 1995). One possibility is the existence of compensatory mechanisms in the mice substituting for the missing activities. Our data, however, provide another explanation. Motor neurons go through a period of naturally occurring cell death leading to the elimination of approximately 50% of initially generated cells (see Oppenheim, 1991, for review). Cell counts in the knock-out animals, however, have been performed after the cell death period. If the lack of trkB signalling leads to a non-production of 50% or so of the initial number of motor neurons, this absence would be masked by a correspondingly lower number of motor neurons being eliminated. Experiments addressing this question in knock-out animals are in progress.
One of the most interesting aspects of our data is the manner in which BDNF may be provided to the trkB-expressing cells. BDNF mRNA is detected in the dorsal neural tube throughout the entire hindbrain and spinal cord (our data and Kahane et al., 1996), overlapping temporally with trkB expression, during the first 2 days of motor neuron generation in vivo, i.e. embryonic day 2-4. A subpopulation of ventrally projecting interneurons is located in the dorsal BDNF expression domain. Dorsal interneurons with ventral projections are among the first neurons to be born and to differentiate in the dorsal neural tube (Yaginuma et al., 1990; Shiga et al., 1991). Following their birth at the ventricular surface, the interneurons migrate radially towards the pial surface, starting to grow their ventrally projecting axons as they move (Holley, 1982; Wentworth et al., 1984; Leber and Sanes, 1995). These axons reach the level of the trkB-expressing cells at around the time the first motor neurons are born (Hollyday and Hamburger, 1977; Yaginuma et al., 1990). It has recently been demonstrated that neurotrophins can be axonally transported in an anterograde fashion by neurons (Von Bartheld et al., 1996) and that they can also be released along the length of neurites (Thoenen, 1995; Blochl and Thoenen, 1996). Our ablation studies show that the dorsal neural tube is required for motor neurons to develop in normal numbers. We therefore suggest that BDNF produced by both young and more mature interneurons in the dorsal neural tube is transported anterogradely down to the ventral neural tube and released from growth cones or along the axons in proximity to the trkB-expressing motorneuron progenitor cells, exerting its influence locally on motor neuron development (Fig. 7). This novel mode of interaction between cells located at opposite poles of the neural tube would allow a coordination of neuronal development between remotely situated subpopulations of cells.
BDNF is produced by young as well as more mature, axon-bearing interneurons (inter; red) in the dorsal neural tube and may be released in the area of the trkB-expressing motor neuron progenitor cells (trkB; blue) after anterograde axonal transport. BDNF most likely influences proliferation and/or differentiation of the precursors, thereby regulating the generation of motor neurons (mn; green) in the ventral neural tube. fp, floor plate; noto, notochord.
BDNF is produced by young as well as more mature, axon-bearing interneurons (inter; red) in the dorsal neural tube and may be released in the area of the trkB-expressing motor neuron progenitor cells (trkB; blue) after anterograde axonal transport. BDNF most likely influences proliferation and/or differentiation of the precursors, thereby regulating the generation of motor neurons (mn; green) in the ventral neural tube. fp, floor plate; noto, notochord.
ACKNOWLEDGEMENTS
We thank Y. Barde for BDNF and trkB clones; T. Jessell for the Isl-1 clone and anti-LH-2 antiserum; T. Edlund for anti-ISL-1/2 antiserum; Regeneron for recombinant BDNF; M. Nishizawa for the chick MafB clone; A. Wizenmann for help with confocal microscopy; A. Graham, I. McKay, and A. Wizenmann for helpful suggestions and discussions; I. McKay for sharing unpublished data; N. Adams for computer drawing and help with computing. S. J. was an EMBO long term fellow and is currently supported by the EC. A. L. is an International Research Scholar of the Howard Hughes Medical Institute.