We have generated double mutant mice deficient in pairs of two different Trk receptors and have analysed the effects on survival and differentiation of dorsal root ganglion (DRG), inner ear cochlear and vestibular sensory neurons. In most combinations of mutant trk alleles, the defects observed in double compared to single mutant mice were additive. However, double homozygous trkA/;trkB/ DRG and trkB/;trkC/ vestibular neurons showed the same degree of survival as single trkA/ and trkB/ mice, respectively, suggesting that those neurons required both Trk signaling pathways for survival. In situ hybridisation analysis of DRG neurons of double mutant mice revealed differential expression of excitatory neuropeptides. Whereas calcitonin-gene-related peptide expression correlated with the trkA phenotype, substance P expression was detected in all combinations of double mutant mice. In the inner ear, TrkB- and TrkC-dependent neurons were shown to at least partially depend on each other for survival, most likely indirectly due to abnormal development of their common targets. This effect was not observed in DRGs, where neurons depending on different Trk receptors generally innervate different targets.

During vertebrate development, many peripheral neurons depend for their survival on neurotrophins synthesised by the tissues that they innervate (Davies, 1994a). The effects of neurotrophins are mediated by the Trk family of receptor tyrosine kinases (Barbacid, 1993). To understand more precisely the function of neurotrophins during nervous system development, we and others have recently generated and analysed germline targeted mice deficient in either Trk receptors or neurotrophins (for reviews see Davies, 1994b; Klein, 1994; Snider, 1994). Mutant mice showed lesions of peripheral sensory neurons specific for the deleted receptor or ligand. Mutant mice deficient in either TrkA receptors or its high affinity ligand NGF, suffered from loss of dorsal root ganglion (DRG) sensory neurons responsive to temperature and pain (Crowley et al., 1994; Smeyne et al., 1994). Mice deficient in TrkB receptors or BDNF also had lesions in DRG neurons (Klein et al., 1993; Ernfors et al., 1994a; Jones et al., 1994). In addition, BDNF/ mutants were found to have reduced numbers of inner ear vestibular and nodose neurons (Ernfors et al., 1994a; Jones et al., 1994). trkC/ or NT-3/ mutant mice were found to lack DRG proprioceptive neurons, which resulted in abnormal movements and postures (Ernfors et al., 1994b; Farinas et al., 1994; Klein et al., 1994). In addition, NT-3/ mice were found to lack the majority of trigeminal and inner ear cochlear neurons.

At least two important conclusions could be drawn from the defects observed in these mice. First, sensory neurons subserving different functions require different Trk signaling pathways for their survival. Thus, neurotrophin dependence was confirmed to be modality specific (Snider, 1994), as previously suggested (Davies, 1987). Second, mutations of a single trk or neurotrophin gene causes dramatic cell loss of specific peripheral neurons, demonstrating that, in these neurons, one Trk receptor cannot compensate for the lack of another Trk receptor. Despite these important conclusions, many questions still remain unanswered. For example, do neu-rotrophins act sequentially during development (Davies, 1994c) so that the same population of neurons might be affected by the loss of two different Trk receptors? Can Trk receptors functionally compensate for each other, either directly within the cell, or indirectly via trophic interactions? Are the surviving sensory neurons in Trk receptor-deficient mice somehow affected by the mutation and, if so, do they show changes in expression of certain differentiation markers? In order to address these questions genetically, we have generated double mutant mice deficient in pairs of Trk receptors by intercrossing single mutant mice lacking one of the three Trk receptors. We have focused on sensory neuron survival and differentiation to be able to derive some general conclusions on functional interactions between Trk signaling pathways.

Histology and cell counting

For the inner ear histology, entire temporal bone primordia were dissected from newborn mice and fixed in 4% paraformaldehyde in 0.1 M sodium phosphate (pH 7.2) for 6-12 hours, dehydrated in ethanol, embedded in paraffin, serially sectioned at 8 μm and stained with 0.1% cresyl violet. For DRG histology and in situ hybridisation analysis, whole mouse bodies were mounted in tissue freezing medium (Tissue-Tek) and transversally sectioned at 14 μm in cranial direction on a cryostat. The correct ganglia levels were identified by carefully mapping the foramina intervertebralia beginning at the sacrum and by comparing the spinal cord morphology. For neuronal counts, series of every fifth section through the L4 ganglia were immersion fixed in 4% paraformaldehyde and counter stained with toluidine blue. All neurons having visible nuclei were counted. The raw counts were corrected according to Abercrombie (Abercrombie, 1946). Neuronal counts of cochlear and vestibular neurons were done essentially as described (Schimmang et al., 1995). Briefly, nuclei were counted at 200× magnification in 4 randomly chosen fields in every section of 8 μm thickness. Up to about 80 fields per ganglion in sections that were 40 μm apart were analysed. No corrections were done here, since neuronal size (max. 25 μm) was always smaller than the intersection interval, making it impossible to count the same neuron twice.

In situ hybridisation

In situ hybridisation analysis with oligonucleotide probes was performed as previously described (Dagerlind et al., 1992). The syn-thesised 48-mer oligonucleotide against substance P was complementary to nucleotides 145-192 of the rat preprotachykinin mRNA (Krause et al., 1987) and the 44-mer probe against CGRP was complementary to nucleotides 13-56 in the transcripts from the α-CGRP/calcitonin and β-CGRP genes encoding the mature α- and β-CGRP peptides (Amara et al., 1985). The synthesised 48-mer oligonucleotide probe against GAP-43 was complementary to nucleotides 70-117 of the rat GAP-43 gene (Karns et al., 1987). The probes were labelled at the 3′ end using 35S-labelled deoxyadenosine-alpha(thio)triphosphate and terminal deoxynucleotidyl transferase (Amersham), then hybridised to the sections without pretreatment for 16-18 hours at 42°C. The hybridisation buffer contained 50% formamide, 4× SSC, 1× Denhardt’s, 1% sarcosyl (N-lauroylsarcosine; Sigma), 0.02 M phosphate buffer, 10% dextran sulfate, 250 μg/ml yeast tRNA, 500 μg/ml salmon sperm DNA and 200 mM DTT. Following hybridisation, the sections were washed several times in 1× SSC at 55°C, dehydrated in ethanol and dipped in NTB2 nuclear track emulsion (Kodak). After 3-4 weeks, the sections were developed in D-19 developer (Kodak) and cover-slipped. Control sections were hybridised with a 20-fold excess of cold oligonucleotide probe. SP expression was quantified by counting cells on sections from in situ hybridisation experiments. Cells were scored as positive, when the number of silver grains was 4-times over background (on average, 900 cells were counted).

Immunohistochemistry

Immunohistochemistry was carried out using the ABC Vectastain kit (Vector Labs) on 25 μm cryosections. Sections were incubated in TBS solution (50 mM Tris-HCl buffer pH 7.5, containing 0.1% sheep serum, 0.1% BSA, 0.1% Triton), quenched in 3% H2O2, blocked with serum and left overnight at 4°C in TBS solution containing 2-4 μg/ml of a mouse anti-200K neurofilament (Boehringer Mannheim) and anti-β-tubulin antibodies. After incubation with a secondary biotinylated antibody and the ABC reagent, peroxidase was reacted with 0.05% diaminobenzidine tetrahydrochloride and 0.003% hydrogenperoxide.

Survival of dorsal root ganglion and inner ear neurons in trk double mutant mice

Double heterozygous mice carrying mutant alleles of two different trk genes were crossed and the offspring analysed at postnatal day 1 (P1), since none of the possible combinations of double homozygous mutant mice (trkA/;trkB/, trkA/; trkC/, and trkB/;trkC/) survived to later stages. To assess the effects of combined trk mutations on dorsal root ganglion (DRG) neurons, we calculated neuron numbers in serial sections of lumbar level 4 DRGs (Table 1). As recently described (for review see Snider, 1994), each of the single trk mutant mice showed reduced survival of DRG neurons to varying degrees, with loss of TrkA receptors being more severe (73%), than loss of either TrkB (20%) or TrkC (17%) receptors. Double mutant trkA/;trkC/ mice had phenotypes that were most consistent with the sum of the individual phenotypes (93% combined DRG neuron loss compared to 90%, the sum of individual trkA and trkC phenotypes). Likewise, the combination of trkB and trkC mutant alleles was found to have an additive effect. However, combinations involving trkA and trkB mutant alleles did not result in significantly increased neuron loss compared to the single trkA mutant phenotype (78% combined neuron loss versus 73% single trkA mutant mice). The difference between the 93% reduction in the trkA/;trkC/ phenotype compared to the 78% reduction in the trkA/; trkB/ phenotype was highly significant. Consistently, the number of DRG neurons in trkA/;trkB/ mice was three times higher than in trkA/;trkC/ mice (Table 1). These results suggest that a subset of DRG neurons requires both TrkA- and TrkB-mediated signals for survival. In contrast, TrkC-positive neurons do not seem to require functional TrkA or TrkB receptors. The small dosage effect in trkA+/;trkC/ as compared to single trkC/ mice is therefore likely to occur through an indirect mechanism.

Table 1.

Numbers of neurons and substance P (SP) expression in L4 dorsal root ganglia of wild-type and trk mutant mice

Numbers of neurons and substance P (SP) expression in L4 dorsal root ganglia of wild-type and trk mutant mice
Numbers of neurons and substance P (SP) expression in L4 dorsal root ganglia of wild-type and trk mutant mice

We next analysed the effects of combined trk mutations on the survival of inner ear sensory neurons, contained in the cochlear and vestibular ganglia. Here, we confined our analysis to trkB/trkC double mutant mice, since expression of trkA in the inner ear is limited to a short period during embryogenesis well before these neurons become dependent on neurotrophins for their survival (Schecterson and Bothwell, 1994). Consistent with these observations, trkA/ single mutant mice do not appear to have reduced numbers of inner ear sensory neurons (data not shown). Analysis of cochlear ganglia from trkB/;trkC/ double mutant mice revealed an overall 61% reduction in the numbers of neurons, indicating that the effects of the two mutations were primarily additive (Table 2). In contrast, vestibular neuron survival was not further reduced in trkB/;trkC/ versus trkB/ mice. This was reminiscent of trkA/;trkB/ DRG neurons and suggests that for a subset of vestibular neurons both TrkB and TrkC signaling pathways are required. Surprisingly, double mutant mice that were homozygous for trkB and heterozygous for trkC (trkB/;trkC+/) showed a dramatic reduction of cochlear neurons (53%) similar to the trkC/ (51%) and trkB/;trkC/ phenotypes (61%). Likewise in vestibular ganglia (although in the reverse combination), mice heterozygous for trkB and homozygous for trkC (trkB+/;trkC/) showed a much greater effect (41% reduction) than single trkC/ (16%) or double heterozygous mice (8%). This indicated that in inner ear sensory neurons, TrkB and TrkC receptors are able, to some extent, to directly or indirectly compensate for each other. This was in contrast to DRG neurons, where such dramatic dosage effects were not observed (see Table 1)

Table 2.

Neuron numbers in cochlear and vestibular ganglia of wild-type and trkB, trkC mutant mice

Neuron numbers in cochlear and vestibular ganglia of wild-type and trkB, trkC mutant mice
Neuron numbers in cochlear and vestibular ganglia of wild-type and trkB, trkC mutant mice

The trkA knockout phenotype correlates with expression of calcitonin-gene-related peptide in DRG neurons

To further analyse the surviving sensory neuron population in double mutant mice, we performed in situ hybridisation assays on DRG neurons using an oligonucleotide probe recognising the mRNA transcripts encoding calcitonin gene-related peptide (CGRP) (Rosenfeld et al., 1983). Consistent with CGRP expression in the rat (Ju et al., 1987), approximately 40-45% of wild-type lumbar DRG neurons were found to contain CGRP-specific transcripts (Fig. 1). Analysis of the single mutant trk mice revealed that CGRP expression was specifically absent in trkA/ mice, but did not seem to be significantly reduced in trkB/ or trkC/ DRGs. This correlation between the trkA knockout phenotype and CGRP expression was also observed in DRGs of double mutant mice. In all combinations of trk genotypes containing two mutant trkA alleles, we failed to detect CGRP expression (Fig. 1). Identical results were obtained by immunohistochemical analysis of CGRP expression using a CGRP-specific antiserum (data not shown). Double mutant mice heterozygous for trkA and homozygous for either trkB or trkC showed normal CGRP expression indicating that CGRP expression was neither critically dependent on the presence of two functional trkA alleles nor was it influenced by the presence or absence of TrkB- or TrkC-dependent neurons. Together, these results indicate that CGRP expression is limited to TrkA-expressing DRG neurons.

Fig. 1.

Expression of calcitonin-gene related peptide (CGRP) in neonatal L4 DRG neurons. In situ hybridisation analysis was performed using a CGRP-specific 35S-labelled oligonucleotide probe on transverse cryosections from either wild-type mice (+/+) or germline targeted mice carrying the indicated combinations of mutant trk alleles. The specificity of the signals was controlled in sections from wild-type mice by hybridisation in the presence of excess unlabelled oligonucleotide probe (Control). Note the specific absence of CGRP expression in all combinations containing two mutant trkA alleles. The white bar scale in the +/+ panel corresponds to 100 μm.

Fig. 1.

Expression of calcitonin-gene related peptide (CGRP) in neonatal L4 DRG neurons. In situ hybridisation analysis was performed using a CGRP-specific 35S-labelled oligonucleotide probe on transverse cryosections from either wild-type mice (+/+) or germline targeted mice carrying the indicated combinations of mutant trk alleles. The specificity of the signals was controlled in sections from wild-type mice by hybridisation in the presence of excess unlabelled oligonucleotide probe (Control). Note the specific absence of CGRP expression in all combinations containing two mutant trkA alleles. The white bar scale in the +/+ panel corresponds to 100 μm.

Substance P expression is detectable in all combinations of trk mutant alleles

CGRP-positive DRG neurons often co-express another excitatory neuropeptide, substance P (SP) (for a recent review see Hökfelt et al., 1994). CGRP and SP are thought to synergistically modulate sensory stimuli to the spinal dorsal horn including nociceptive reflexes (Wiesenfeld-Hallin et al., 1984). As shown in Fig. 2 and Table 1, in situ hybridisation analysis using a SP-specific oligonucleotide probe revealed expression in 39% of DRG neurons of wild-type mice. Analysis of single homozygous mutant mice showed a slightly increased proportion of SP-positive neurons in trkB/ and trkC/ mice, and reduced, but detectable SP expression in trkA/ mice. Moreover, SP expression could be detected in all combinations of double mutant mice including a very small population in trkA/;trkC/ mice, despite the 90% reduction of DRG neurons. Quantitation of SP expression revealed that the majority of SP-positive neurons were TrkA-dependent (Table 1). In contrast to CGRP expression, however, 16% of TrkA-independent neurons expressed SP. Most of this TrkA-negative, SP-positive neuron population seemed to be TrkC-dependent, possibly proprioceptive neurons, since 14% of the surviving DRG neurons in trkA/;trkB/ mice still expressed SP. To visualise the reduced size of the ganglia, Fig. 2 depicts adjacent sections hybridised to an oligonucleotide probe specific for mRNA encoding the neuronal growth cone protein GAP-43 (Karns et al., 1987). At this developmental stage, GAP-43 is expressed by all neurons, independently of the trk genotype, and thus gives a reliable measure of the actual size of the ganglia.

Fig. 2.

Expression of substance P (SP) and GAP-43 in neonatal L4 DRG neurons. In situ hybridisation analysis was performed on adjacent sections using 35S-labelled oligonucleotide probes. Control panel indicates SP expression in the presence of excess unlabelled oligonucleotide probe. Note that SP expression is detectable in all single mutant and all combinations of double mutant mice. This includes trkA/;trkC/mutant DRGs despite their >90% loss of neurons as visualised by the GAP-43 in situs. The white bar scale in the +/+ panel corresponds to 100 μm.

Fig. 2.

Expression of substance P (SP) and GAP-43 in neonatal L4 DRG neurons. In situ hybridisation analysis was performed on adjacent sections using 35S-labelled oligonucleotide probes. Control panel indicates SP expression in the presence of excess unlabelled oligonucleotide probe. Note that SP expression is detectable in all single mutant and all combinations of double mutant mice. This includes trkA/;trkC/mutant DRGs despite their >90% loss of neurons as visualised by the GAP-43 in situs. The white bar scale in the +/+ panel corresponds to 100 μm.

TrkB and TrkC receptors functionally interact in inner ear development

Our recent analysis of single trkB/ and trkC/ mice had revealed that TrkB and TrkC receptors are necessary for maintenance of afferent target innervation (Schimmang et al., 1995). We therefore analysed neurofilament- and β-tubulinstained sections of inner ear sensory epithelium from trkB/trkC double mutant mice to determine, if TrkB and TrkC receptors functionally interact in afferent innervation of the inner ear. Longitudinal sections through the organ of Corti of wild-type (Fig. 3A), double heterozygous (not shown), and single trkB/ mice (Fig. 3B) revealed thick neurofilament staining within the sensory epithelium at the level of inner hair cells. Surprisingly, but correlating with the dramatic effects on neuronal survival, mice homozygous for trkB and heterozygous for trkC (trkB/;trkC+/) did not show neurofilament staining (Fig. 3C) comparable to trkB/;trkC/ mice (Fig. 3D). Like in the cochlea, although in reverse combination, trkB+/;trkC/ mice show merely fractionated neurites in vestibular sensory epithelium (Fig. 3G), even though double heterozygous (not shown) or trkC/ mice do not show significantly altered neurofilament staining (Fig. 3F). Similar staining patterns in both cochlea and vestibular systems were observed with an antibody against β-tubulin (data not shown), indicating that the observed defects were not caused by down-regulation of neurofilament expression but rather represented morphological changes.

Fig. 3.

TrkB and TrkC receptors functionally interact in maintaining inner ear sensory innervation. Longitudinal sections through the sensory epithelium at the basal turn of the cochlea showing inner hair cells of P1 wild-type (A), trkB/single mutant (B), trkB/;trkC+/ (C), and trkB/;trkC/ double mutant mice (D) were stained with a monoclonal antibody against neurofilaments. In wild-type (A), heterozygous (not shown) and trkB/single mutant mice (B) thick afferent nerve fibres (small arrows) enter the sensory epithelium (indicated by large arrowheads). trkB/;trkC+/ (C) and trkB/;trkC/ double mutant mice (D) only show thin immunoreactive fibres, which do not enter and innervate the sensory epithelium. Anti-neurofilament stained sections through the utricular maculae of wild-type (E), trkC/ single mutant (F), trkB+/;trkC/ (G), and trkB/;trkC/ (H) double mutant mice. In wild-type (E), heterozygous (not shown), and trkC/ mice (F) nerve fibers (small arrows) reach and enter the sensory epithelium (indicated by large arrowheads). Only thin fractionated neurites are observed in trkB+/;trkC/ (G), or trkB/;trkC/ double mutant mice (H). Bar scale in A corresponds to 75 μm.

Fig. 3.

TrkB and TrkC receptors functionally interact in maintaining inner ear sensory innervation. Longitudinal sections through the sensory epithelium at the basal turn of the cochlea showing inner hair cells of P1 wild-type (A), trkB/single mutant (B), trkB/;trkC+/ (C), and trkB/;trkC/ double mutant mice (D) were stained with a monoclonal antibody against neurofilaments. In wild-type (A), heterozygous (not shown) and trkB/single mutant mice (B) thick afferent nerve fibres (small arrows) enter the sensory epithelium (indicated by large arrowheads). trkB/;trkC+/ (C) and trkB/;trkC/ double mutant mice (D) only show thin immunoreactive fibres, which do not enter and innervate the sensory epithelium. Anti-neurofilament stained sections through the utricular maculae of wild-type (E), trkC/ single mutant (F), trkB+/;trkC/ (G), and trkB/;trkC/ (H) double mutant mice. In wild-type (E), heterozygous (not shown), and trkC/ mice (F) nerve fibers (small arrows) reach and enter the sensory epithelium (indicated by large arrowheads). Only thin fractionated neurites are observed in trkB+/;trkC/ (G), or trkB/;trkC/ double mutant mice (H). Bar scale in A corresponds to 75 μm.

We next analysed cresyl-violet-stained sections of inner ear sensory epithelia from trkB/trkC double mutant mice to determine if TrkB and TrkC receptors would also interact in target tissue formation. Abnormal sensory epithelium differentiation was previously observed in cochlea of single trkC/ mice (Schimmang et al., 1995) and vestibular organ of BDNF/ mice (Ernfors et al., 1995). Transversal sections through the basal turn of the cochlea showed the three compartments, scala vestibuli, scala media and scala tympani (Fig. 4A,B). The scala media of trkB/;trkC+/ (Fig. 4B) and of trkB/;trkC/ mice (not shown) appeared larger in size, because of the reduced thickness of the sensory epithelium. Upon higher magnification, wild-type (Fig. 4C), single or double heterozygous (not shown), and trkB/ mice (Fig. 4E) show normally developed organ of Corti with the typical arrangement of hair cells. In contrast, trkB/;trkC+/ mice (Fig. 4D) had a much thinner, less stratified sensory epithelium, comparable to trkC/ (Schimmang et al., 1995) and almost as severe as trkB/;trkC/ mice (Fig. 4F).

Fig. 4.

Functional interaction between TrkB and TrkC receptors in the development of cochlear sensory epithelium. Cresyl violet stained transversal sections through the basal turn of the cochlea of P1 wild-type (A) and trkB/;trkC+/ (B) mice. 1, scala vestibuli; 2, scala media; 3, scala tympani; CG, cochlear ganglion. Note the reduced thickness of the sensory epithelium causing the scala media to appear enlarged. (C-F) High magnification of organ of Corti primordium of wild-type (C), trkB/;trkC+/ (D), trkB/ single (E), and trkB/;trkC/ double mutant mice (F). The sensory epithelium is indicated by arrowheads. Note that the epithelia in panels D,F are thinner, less stratified, and lack the typical arrangement of hair cells seen in wild-type (C), and trkB/ mice (E). Bar scale corresponds to 200 μm in A,B, and 100 μm in C-F.

Fig. 4.

Functional interaction between TrkB and TrkC receptors in the development of cochlear sensory epithelium. Cresyl violet stained transversal sections through the basal turn of the cochlea of P1 wild-type (A) and trkB/;trkC+/ (B) mice. 1, scala vestibuli; 2, scala media; 3, scala tympani; CG, cochlear ganglion. Note the reduced thickness of the sensory epithelium causing the scala media to appear enlarged. (C-F) High magnification of organ of Corti primordium of wild-type (C), trkB/;trkC+/ (D), trkB/ single (E), and trkB/;trkC/ double mutant mice (F). The sensory epithelium is indicated by arrowheads. Note that the epithelia in panels D,F are thinner, less stratified, and lack the typical arrangement of hair cells seen in wild-type (C), and trkB/ mice (E). Bar scale corresponds to 200 μm in A,B, and 100 μm in C-F.

A similar situation was observed in sensory epithelia of the vestibular organ. The utricular macula of newborn wild-type (Fig. 5A) and trkC/ mice (Fig. 5B) is already quite well differentiated into dark-stained supporting cells and lightly stained hair cells. In both trkB+/;trkC/ (Fig. 5C) and trkB/; trkC/ mice (Fig. 5D) the epithelium is reduced in thickness and hair cells appear less well differentiated. Likewise, the crista ampullaris of newborn wild-type (Fig. 5E) and trkC/ mice (Fig. 5F) is well differentiated showing a thick epithelium with mature appearing hair cells. In contrast, cristae of trkB+/;trkC/ (Fig. 5G) and trkB/;trkC/ mice (Fig. 5H) are smaller in size, their sensory epithelium is thinner and hair cells appear less well differentiated. Therefore, the dramatic cell loss observed in trkB/;trkC+/ cochlear and trkB+/;trkC/ vestibular ganglion neurons correlates with abnormal differentiation of their target tissues.

Fig. 5.

Functional interaction between TrkB and TrkC receptors in the development of vestibular sensory epithelia. Cresyl violet stained sections through utricular macula of P1 wild-type (A), trkC/ single mutant (B), trkB+/;trkC/ (C), and trkB/;trkC/ double mutant mice (D). Note that the sensory epithelium is much thinner and less stratified in panels C,D. Cresyl violet stained sections through the center of cristae ampullaris of P1 wild-type (E), trkC/ single mutant (F), trkB+/;trkC/ (G), and trkB/;trkC/ double mutant mice (H). Note that the size of the cristae and the thickness of the sensory epithelium are reduced in panels G,H. Bar scale corresponds to 100 μm.

Fig. 5.

Functional interaction between TrkB and TrkC receptors in the development of vestibular sensory epithelia. Cresyl violet stained sections through utricular macula of P1 wild-type (A), trkC/ single mutant (B), trkB+/;trkC/ (C), and trkB/;trkC/ double mutant mice (D). Note that the sensory epithelium is much thinner and less stratified in panels C,D. Cresyl violet stained sections through the center of cristae ampullaris of P1 wild-type (E), trkC/ single mutant (F), trkB+/;trkC/ (G), and trkB/;trkC/ double mutant mice (H). Note that the size of the cristae and the thickness of the sensory epithelium are reduced in panels G,H. Bar scale corresponds to 100 μm.

Subpopulations of sensory neurons require two different Trk receptors for survival

In the present report, we have focused on the analysis of sensory neuron survival and differentiation in double mutant mice deficient in pairs of different Trk receptors. Consistent with the apparent specificity of phenotypes observed in single trk mutant mice, the effects on sensory neuron survival in trkA/;trkC/ and trkB/;trkC/ DRG and trkB/;trkC/ cochlear neurons were largely additive. These results confirmed initial conclusions from single knockout mice, that those neuron populations develop independently from each other, express a single Trk receptor species and depend on this Trk signaling pathway at some time point during development (Snider, 1994). Since, however, the analysis was confined to the newborn stage, it is still possible that the kinetics of neuronal death differs between the single and double mutant mice. Other combinations of mutant alleles such as trkA/;trkB/ in DRG or trkB/;trkC/ in vestibular neurons showed no significant increase in neuron loss over single trkA/ (DRG) or single trkB/ (vestibular ganglion) mutant mice. These data indicate that certain subpopulations of sensory neurons co-express and require two different Trk signaling pathways for survival. Interestingly, combined heterozygosity for pairs of two different mutant trk alleles did not cause significant reductions of neurons in newborn mice, although receptor expression in heterozygous mice was shown to be reduced by more than 50% (Klein et al., 1993, 1994). Alternatively, the requirement of both TrkB and TrkC receptor signaling pathways could be sequential and may not involve stable co-expression of both receptor pathways. Trigeminal sensory neurons have been shown to switch their neurotrophin specificity from BDNF/NT-3 to NGF during embryonic development (Davies, 1994c), presumably by changing neurotrophin receptor expression. If, in wild-type mice, certain vestibular sensory neurons switch between TrkB and TrkC receptor expression, i.e. between BDNFand NT-3 dependency during development, they could be vulnerable for either of the two mutations.

Trk receptors compensate for each other in inner ear development

One of the surprising findings of this study was the fact that trkB+/;trkC/and trkB/;trkC+/ double mutants showed inner ear defects comparable to the double homozygous phenotype, whereas the relevant individual mutations had no, or very minor effects on their own. The defects included loss of ganglion cells, lack of afferent target innervation and, most importantly, abnormal sensory epithelium development, the latter most likely being secondary to sensory deprivation. The results indicated that, in the inner ear, reduced levels of one receptor could be compensated for by functional alleles of a different Trk receptor. Based on the specificity of the single mutant phenotypes, we would postulate that the observed genetic interaction between TrkB and TrkC signaling pathways may be indirect and may not require co-expression or sequential expression of Trk receptors in the same cell.

We would like to propose a model (Fig. 6) in which ganglion cell degeneration influences development of the target, which in turn could affect different neuronal populations innervating the same target. As illustrated in Fig. 6A, DRG neurons requiring different Trk receptors generally innervate different targets. For example, TrkA- and TrkC-dependent neurons innervate skin and muscle, respectively. Consequently, the phenotype of trkA+/;trkC/ DRG neurons closely resembles the trkC/ phenotype, indicating no or very little interaction between TrkA and TrkC receptors. Fig. 6B illustrates the different situation in inner ear, where both TrkB- and TrkC-dependent neurons innervate the same target. In a trkB/; trkC+/ mutant mouse, loss of TrkB-dependent cochlear neurons and partial loss of target innervation (outer hair cells) may affect the release of BDNF and NT-3 from the sensory epithelium. The TrkC-dependent neuron population is indirectly affected by the primary genetic defect. However, as long as this neuron population expresses normal levels of TrkC receptors, it may be able to survive with reduced amounts of neurotrophins released by the affected target. Should TrkC receptor expression become reduced, such as in a heterozygous situation, it will no longer be able to receive sufficient trophic support and die. With both TrkB- and TrkC-dependent neurons disappearing, the sensory epithelium will also eventually degenerate. This model can be applied to both cochlear and vestibular sensory neurons, since both consist of TrkB- and TrkC-positive populations that innervate the same sensory epithelia. In addition, it seems that the inner ear system has to be very tightly controlled with respect to levels of neurotrophins and Trk receptors. Within the same sensory epithelium, there are great differences in innervation patterns. In the cochlea, several outer hair cells are innervated by a single type II, TrkB-positive cochlear neuron, whereas a single inner hair cell is innervated by up to ten type I, TrkC-positive cochlear neurons. This might explain why the inner ear system is more sensitive to changes in levels of neurotrophins and Trk receptors than other sensory systems, such as DRGs.

Fig. 6.

Model for indirect functional interaction in the inner ear between trkB and trkC mutant alleles. (A) TrkA- and TrkC-dependent DRG neurons innervate different targets and do not depend on each other in trk double mutant mice. Each population of neurons depends on specific neurotrophins secreted by the targets (arrows pointing down). The targets also depend on sensory neuron innervation for proper development (arrows pointing up). (B) TrkB- and TrkC-dependent inner ear sensory neurons innervate the same target and indirectly depend on each other due to abnormal development of the target. We hypothesise that partial lack of afferent innervation (e.g. in a trkB/ mouse) causes reduced release of neurotrophins. As a result, trkC+/ neurons, which have reduced levels of receptors, start to degenerate, having more adverse effects on the target. Finally, the sensory epithelium degenerates as in double homozygous mice.

Fig. 6.

Model for indirect functional interaction in the inner ear between trkB and trkC mutant alleles. (A) TrkA- and TrkC-dependent DRG neurons innervate different targets and do not depend on each other in trk double mutant mice. Each population of neurons depends on specific neurotrophins secreted by the targets (arrows pointing down). The targets also depend on sensory neuron innervation for proper development (arrows pointing up). (B) TrkB- and TrkC-dependent inner ear sensory neurons innervate the same target and indirectly depend on each other due to abnormal development of the target. We hypothesise that partial lack of afferent innervation (e.g. in a trkB/ mouse) causes reduced release of neurotrophins. As a result, trkC+/ neurons, which have reduced levels of receptors, start to degenerate, having more adverse effects on the target. Finally, the sensory epithelium degenerates as in double homozygous mice.

Comparison of trk and neurotrophin mutant phenotypes

Analysis of sensory neuron defects in trk and neurotrophin knockout mice indicates consistency between the relevant ligand/receptor pairs (Snider, 1994). For example, both trkA/ and NGF/ mice lacked approximately 70% of their DRG neurons (Crowley et al., 1994; Smeyne et al., 1994). The exception to the rule, however, have been NT-3/ mice, which generally displayed more cell loss (55 or 78%) than trkC/ mice (17-19%) (Ernfors et al., 1994b; Farinas et al., 1994; Klein et al., 1994; this study). This was interpreted as the ability of NT-3 to signal through either TrkA or TrkB receptors in the absence of its preferred TrkC receptor (Farinas et al., 1994). Consistent with this hypothesis, we have recently shown that NT-3 can, in fact, promote survival in culture of both NGF- and BDNF-dependent primary sensory neurons, derived from trkC/ mice (Davies et al., 1995). This clearly demonstrates a role for NT-3 in the absence of its preferred TrkC receptor. DRGs from trkB/;trkC/ mice suffered increased ganglion cell death (41% reduction), but did not seem to be as severely affected as DRGs from NT-3/ mice suggesting that NT-3 can signal through TrkA-dependent DRG neurons in the absence of functional TrkB and TrkC receptors. A similar situation was observed in cochlear neurons in trkB/trkC double homozygous mice (61% reduction) compared to newborn NT-3/ mice (85% reduction) (Farinas et al., 1994). A direct comparative analysis between trk and neurotrophin mutant mice should reveal whether or not these differences are significant. Interestingly, in vestibular ganglia, the BDNF mutation also appeared to have a more dramatic effect (87% reduction) (Jones et al., 1994), compared to the trkB/ (56% reduction) or the trkB/;trkC/ genotype (58% reduction). BDNF has not been shown to crossreact in vitro with TrkA or TrkC receptors (Ip et al., 1993) indicating that the differences between ligand and receptor knockouts may have a more general cause. One explanation might be the presence of non-catalytic isoforms encoded by the trkB and trkC genes (Klein et al., 1990; Middlemas et al., 1991; Tsoulfas et al., 1993; Valenzuela et al., 1993). These isoforms were not destroyed by the homologous recombination and may mediate some sort of signal transduction that could partially prevent or significantly delay cell death, thereby creating the apparent difference in phenotype.

Neuropeptide expression of DRG neurons is changed in trk knockout mice

The second surprising finding of this study was that CGRP expression strictly correlated with the trkA knockout genotype, whereas SP expression could be detected in all single and combinations of double mutant mice. These results indicate that, at least in lumbar DRGs, CGRP expression mainly defines nociceptive and thermoceptive neurons. CGRP expression has recently been analysed in NGF/ mice (Crowley et al., 1994). Interestingly, NGF/ mice were shown to have dramatically reduced levels of CGRP immunoreactivity, but to still contain some CGRP-positive DRG neurons, suggesting a partial rescue of TrkA-positive neurons by NT-3 or NT-4. In contrast to CGRP, a significant fraction of SP-positive neurons were independent of the trkA knockout genotype and represented to a large part TrkC-dependent, possibly proprioceptive neurons. This is particularly interesting since earlier studies have shown that most, if not all, SP-positive DRG neurons co-express CGRP (Gibbins et al., 1987; Hökfelt et al., 1994). Moreover, careful size histogram studies had revealed that SP and CGRP expression is limited to small and intermediate-sized neurons (Boehmer et al., 1989; Bowie et al., 1994). However, it was recently shown (Noguchi et al., 1994) that SP expression is induced in medium-sized and large-sized DRG neurons after axotomy (see below). It is therefore possible that a SP-positive, large-sized neuron population has escaped attention, especially since large neurons often have lower peptide levels. These neurons may correspond to the SP-positive, TrkA-negative population described here, and it is quite possible that such small populations can only faithfully be revealed in a mutant situation, when the major (TrkA-positive) neuron population has been removed. Alternatively, SP expression may be regulated by extrinsic factors and may differ in mutant mice. Evidence for plasticity of SP expression comes from nerve transection studies (Jessell et al., 1979) and studies using exogenous NGF (Kessler and Black, 1980; Otten et al., 1980; Fitzgerald et al., 1985; Lindsay and Harmar, 1989). In addition, it was shown that SP expression could be altered in neurons that were manipulated to inappropriately innervate different target tissues. For example, afferent fibres from muscle which normally do not contain SP immunoreactivity could be induced to express SP when they were surgically forced to reinnervate skin, clearly indicating a role for the target in regulating SP expression (McMahon and Gibson, 1987). In our system, we could speculate that in trkA/;trkB/ mutant mice, the remaining TrkC-positive proprioceptive neurons may inappropriately innervate target fields, such as skin, that contain factors which induce SP expression. Interestingly, CGRP expression could also be induced by NGF in adult sensory neuron cultures (Lindsay and Harmar, 1989), suggesting differences in the regulation of neuropeptide expression depending on age.

In summary, studies using mice carrying different combinations of mutant trk alleles will help to better characterise neuron populations with regard to their differentiation state and plasticity of gene expression and to derive general conclusions about redundancy of signaling pathways.

We would like to thank Alun M. Davies and Fernando Giraldez for helpful discussions, H åkan Aldskogius for expert advice on neuronal counts in DRGs, K.C. Kent Lloyd and Franca Casagranda for critically reading the manuscript. L. M. and T. S. are supported by longterm EMBO fellowships. This research was partially funded by a grant from Direccion General de Ciencia y Tecnologia (PB92/0621) and Fondo de Investigaciones Sanitarias (94/1405 to J. R).

Abercrombie
,
M.
(
1946
).
Estimation of nuclear populations from microtome sections
.
Anat. Rec
.
94
,
274
329
.
Amara
,
S.
,
Arriza
,
J.
,
Leff
,
S.
,
Swanson
,
L.
,
Evans
,
R.
and
Rosenfeld
,
M.
(
1985
).
Expression in brain of a messenger RNA encoding a novel neuropeptide homologous to calcitonin gene-related peptide
.
Science
229
,
1094
1097
.
Barbacid
,
M.
(
1993
).
Nerve growth factor: a tale of two receptors
.
Oncogene
8
,
2033
2042
.
Boehmer
,
C. G.
,
Norman
,
J.
,
Catton
,
M.
,
Fine
,
L. G.
and
Mantyh
,
P. W.
(
1989
).
High levels of mRNA coding for substance P, somatostatin and α- tubulin are expressed by rat and rabbit dorsal root ganglia neurons
.
Peptides
10
,
1179
1194
.
Bowie
,
D.
,
Feltz
,
P.
and
Schlichter
,
R.
(
1994
).
Subpopulations of neonatal rat sensory neurons express functional neurotransmitter receptors which elevate intracellular calcium
.
Neurosci
.
58
,
141
149
.
Crowley
,
C.
,
Spencer
,
S. D.
,
Nishimura
,
M. C.
,
Chen
,
K. S.
,
Pitts-Meek
,
S.
,
Armanini
,
M. P.
,
Ling
,
L. H.
,
McMahon
,
S. B.
,
Shelton
,
D. L.
,
Levinson
,
A. D.
and
Phillips
,
H. S.
(
1994
).
Mice lacking nerve growth factor display perinatal loss of sensory and sympathetic neurons yet develop basal forebrain cholinergic neurons
.
Cell
76
,
1001
1012
.
Dagerlind
,
A.
,
Friberg
,
K.
,
Bean
,
A.
and
Hökfelt
,
T.
(
1992
).
Sensitive mRNA detection using unfixed tissue: combined radioactive and non- radioactive in situ hybridization histochemistry
.
Histochemistry
98
,
39
49
.
Davies
,
A. M.
(
1994a
).
Role of neurotrophins in the developing nervous system
.
J. Neurobiol
.
25
,
1334
1348
.
Davies
,
A. M.
(
1994b
).
Tracking neurotrophin function
.
Nature
368
,
193
194
.
Davies
,
A. M.
(
1987
).
Molecular and cellular aspects of patterning sensory neurone connections in the vertebrate nervous system
.
Development
101
,
185
208
.
Davies
,
A. M.
(
1994c
).
Switching neurotrophin dependence
.
Curr. Biol
.
4
,
273
276
.
Davies
,
A.M.
,
Minichiello
,
L.
, and
Klein
,
R.
(
1995
).
Developmental changes in NT3 signalling via TrkA and TrkB in embryonic neurons
.
EMBO J
.,
14
,
4482
4489
.
Ernfors
,
P.
,
Lee
,
K.-F.
and
Jaenisch
,
R.
(
1994a
).
Mice lacking brain-derived neurotrophic factor develop with sensory deficits
.
Nature
368
,
147
150
.
Ernfors
,
P.
,
Lee
,
K.-F.
,
Kucera
,
J.
and
Jaenisch
,
R.
(
1994b
).
Lack of neurotrophin-3 leads to deficiencies in the peripheral nervous system and loss of limb proprioceptive afferents
.
Cell
77
,
503
512
.
Ernfors
,
P.
,
Van de Water
,
T.
,
Loring
,
J.
, and
Jaenisch
,
R.
(
1995
).
Complementary roles of BDNF and NT-3 in vestibular and auditory dev elopment
.
Neuron
14
,
1153
1164
.
Farinas
,
I.
,
Jones
,
K.
,
Backus
,
C.
,
Wang
,
X.-Y.
and
Reichardt
,
L.
(
1994
).
Severe sensory and sympathetic deficits in mice lacking neurotrophin-3
.
Nature
369
,
658
661
.
Fitzgerald
,
M.
,
Wall
,
P. D.
,
Goedert
,
M.
and
Emson
,
P. C.
(
1985
).
Nerve growth factor counteracts the neurophysiological and neurochemical effects of chronic sciatic nerve section
.
Brain Research
332
,
131
141
.
Gibbins
,
I. L.
,
Furness
,
J. B.
and
Costa
,
M.
(
1987
).
Pathway-specific patterns of the co-existence of substance P, calcitonin gene-related peptide, cholecystokinin and dynorphin in neurons of the dorsal root ganglia of the guinea-pig
.
Cell Tissue Res
.
248
,
417
437
.
Hökfelt
,
T.
,
Zhang
,
X.
and
Wiesenfeld-Hallin
,
Z.
(
1994
).
Messenger plasticity in primary sensory neurons following axotomy and its functional implications
.
Trends NeuroSci
.
17
,
22
30
.
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
49
.
Jessell
,
T.
,
Tsunoo
,
A.
,
Kanazawa
,
I.
and
Otsuka
,
M.
(
1979
).
Substance P: depletion in the dorsal horn of rat spinal cord after section of the peripheral processes of primary sensory neurons
.
Brain Research
168
,
247
259
.
Jones
,
K.
,
Farinas
,
I.
,
Backus
,
C.
and
Reichardt
,
L.
(
1994
).
Targeted disruption of the BDNF gene perturbs brain and sensory neuron develoment but not motor neuron development
.
Cell
76
,
989
999
.
Ju
,
G.
,
Hökfelt
,
Brodin E.
,
Fahrenkrug
,
J.
,
Fisher
,
P.
,
Frey
,
P.
,
Elde
,
R.
and
Brown
,
J.
(
1987
).
Primary sensory neurons of the rat showing calcitonin-gene-related peptide immunoreactivity and their relation to substance P-, somatostatin-, galanin-, vasoactive intestinal polypeptide-, and cholecystokinin-immunoreactive ganglion cells
.
Cell Tissue Res
.
247
,
417
431
.
Karns
,
L.
,
Ng
,
S.-C.
,
Freeman
,
J.
and
Fishman
,
M.
(
1987
).
Cloning of complementary DNA for GAP-43, a neuronal growth-related protein
.
Science
236
,
597
600
.
Kessler
,
J. A.
and
Black
,
I. B.
(
1980
).
Nerve growth factor stimulates the development of substance P in sensory ganglia
.
Proc. Natl. Acad. Sci. USA
77
,
649
652
.
Klein
,
R.
(
1994
).
Role of neurotrophins in mouse neuronal development.FASEB J
.
8
,
738
744
.
Klein
,
R.
,
Conway
,
D.
,
Parada
,
L. F.
and
Barbacid
,
M.
(
1990
).
The trkB tyrosine protein kinase gene codes for a second neurogenic receptor that lacks the catalytic kinase domain
.
Cell
61
,
647
56
.
Klein
,
R.
,
Silos-Santiago
,
I.
,
Smeyne
,
R. J.
,
Lira
,
S. A.
,
Brambilla
,
R.
,
Bryant
,
S.
,
Zhang
,
L.
,
Snider
,
W. D.
and
Barbacid
,
M.
(
1994
).
Targeted disruption of trkC, the neurotrophin-3 receptor gene, eliminates Ia muscle afferents and results in loss of proprioception
.
Nature
368
,
249
251
.
Klein
,
R.
,
Smeyne
,
R. J.
,
Wurst
,
W.
,
Long
,
L. K.
,
Auerbach
,
B. A.
,
Joyner
,
A. L.
and
Barbacid
,
M.
(
1993
).
Targeted disruption of the trkB neurotrophin receptor gene results in nervous system lesions and neonatal death
.
Cell
75
,
113
122
.
Krause
,
J.
,
Chirgwin
,
J.
,
Carter
,
M.
,
Xu
,
Z.
and
Hershey
,
A.
(
1987
).
Three rat preprotachykinin mRNAs encode the neuropeptides substance P and neurokinin
.
Proc. Natl. Acad. Sci. USA
84
,
881
885
.
Lindsay
,
R. M.
and
Harmar
,
A. J.
(
1989
).
Nerve growth factor regulates expression of neuropeptide genes in adult sensory neurons
.
Nature
337
,
362
4
.
McMahon
,
S. B.
and
Gibson
,
S.
(
1987
).
Peptide expression is altered when afferent nerves reinnervate inappropriate tissue
.
Neurosci. Lett
.
73
,
9
15
.
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
53
.
Noguchi
,
K.
,
Dubner
,
R.
,
DeLeon
,
M.
,
Senba
,
E.
and
Ruda
,
M.
(
1994
).
Axotomy induces preprotachykinin gene expression in a subpopulation of dorsal root ganglion neurons
.
J. Neurosci. Res
.
37
,
596
603
.
Otten
,
U.
,
Goedert
,
M.
,
Mayer
,
N.
and
Lembeck
,
F.
(
1980
).
Requirement of nerve growth factor for development of substance P-containing sensory neurones
.
Nature
287
,
158
159
.
Rosenfeld
,
M.
,
Mermod
,
J.
,
Amara
,
S.
,
Swanson
,
S.
,
Sawchenko
,
P.
,
Rivier
,
J.
,
Vale
,
W.
and
Evans
,
R.
(
1983
).
Production of a novel neuropeptide encoded by the calcitonin gene via tissue-specific RNA processing
.
Nature
304
,
129
135
.
Schecterson
,
L.
and
Bothwell
,
M.
(
1994
).
Neurotrophin and neurotrophin receptor mRNA expression in developing inner ear
.
Hearing Research
73
,
92
100
.
Schimmang
,
T.
,
Minichiello
,
L.
,
Vazquez
,
E.
,
Jose
,
I.S.
,
Giraldez
,
F.
,
Klein
,
R.
, and
Represa
,
J.
(
1995
).
Developing inner ear sensory neurons require TrkB and TrkC receptors for innervation of their peripheral targets
.
Development
,
121
,
3381
3391
.
Smeyne
,
R. J.
,
Klein
,
R.
,
Schnapp
,
A.
,
Long
,
L. K.
,
Bryant
,
S.
,
Lewin
,
A.
,
Lira
,
S. A.
and
Barbacid
,
M.
(
1994
).
Severe sensory and sympathetic neuropathies in mice carrying a disrupted Trk/NGF receptor gene
.
Nature
368
,
246
249
.
Snider
,
W.
(
1994
).
Functions of the neurotrophins during nervous system development: what the knockouts are teaching us
.
Cell
77
,
627
638
.
Tsoulfas
,
P.
,
Soppet
,
D.
,
Escandon
,
E.
,
Tessarollo
,
L.
,
Mendoza
,
R. J.
,
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.
,
Maisonpierre
,
P.
,
Glass
,
D.
,
Rojas
,
E.
,
Nunez
,
L.
,
Kong
,
Y.
,
Gies
,
D.
,
Stitt
,
T.
,
Ip
,
N.
and
Yancopoulos
,
G.
(
1993
).
Alternative forms of rat trkC with different functional capabilities
.
Neuron
10
,
963
974
.
Wiesenfeld-Hallin
,
Z.
,
Hökfelt
,
T.
,
Lundberg
,
J. M.
,
Forssmann
,
W. G.
,
Reinecke
,
M.
,
Tschopp
,
F. A.
and
Fischer
,
J. A.
(
1984
).
Immunoreactive calcitonin gene-related peptide and substance P coexist in sensory neurons to the spinal cord and synergise in spinal behavioral responses of the rat
.
Neurosci. Lett
52
,
199
204
.