Brn3a regulation of TrkA/NGF receptor expression in developing sensory neurons

Neurotrophins and their Trk family receptor tyrosine kinases act on the development and function of the PNS. Numerous in vitro and in vivo studies have established the roles of each Trk receptor in neurite outgrowth, survival and differentiation (Bibel and Barde,2000; Huang and Reichardt,2001; Snider,1994). The differential expression of Trk receptors in subsets of sensory neurons indicates that Trk receptor expression and sensory neuron differentiation are closely related processes. For example, regulation of the expression of Trk family receptor is fundamental to the appropriate appearance and survival of sensory neurons.

TrkA (Ntrk1 — Mouse Genome Informatics), the prototype of the Trk gene family and the receptor for nerve growth factor(NGF), is required for the survival of nociceptive sensory and sympathetic neurons (Martin-Zanca et al.,1990; Smeyne et al.,1994). In the murine PNS, TrkA expression is confined to neural-crest-derived sensory neurons, including trigeminal and dorsal root ganglia, from early embryonic stages (E9.5) to adults. TrkA is also expressed in sympathetic neurons from E16.5 onwards(Martin-Zanca et al., 1990; Tessarollo et al., 1993). In the CNS, TrkA expression is described in only a subset of cholinergic neurons in the basal forebrain (Holtzman et al., 1992). Understanding the molecular mechanisms for TrkA expression is important to dissect the biological processes regulating the survival and differentiation of sensory and sympathetic neurons.

One approach to identify transcriptional regulators of TrkA is to search for known transcription factors that are expressed in sensory neurons during mouse embryonic development and whose loss-of-function lead to alterations of TrkA expression. The POU-homeodomain transcription factor Brn3a (Pou4f1— Mouse Genome Informatics) (Gerrero et al., 1993; He et al.,1989; Ninkina et al.,1993; Xiang et al.,1993) stands out as an attractive candidate. Brn3a is expressed early in emerging sensory neurons (Artinger et al., 1998; Fedtsova and Turner, 1995) and mutation of this gene in mice leads to abnormal development and loss of neurons that express TrkA, TrkB (Ntrk2 — Mouse Genome Informatics) and TrkC (Ntrk3 — Mouse Genome Informatics)(Eng et al., 2001; Huang et al., 2001; Huang et al., 1999; McEvilly et al., 1996; Xiang et al., 1996). However,questions remain about why and how TrkA becomes downregulated in Brn3a-null sensory neurons. Whether Brn3a is a requisite transcriptional regulator of the TrkA enhancer or an indirect mediator of TrkA expression is unresolved.

Previously, we identified a TrkA minimal enhancer that confers specific expression of β-galactosidase transgenes in PNS sensory neurons(Ma et al., 2000). Through mutational analysis in transgenic mice, multiple important, cis-elements were identified. These results indicate that the tightly regulated TrkAgene is controlled by multiple transcription factors acting in concert(Lei et al., 2001; Ma et al., 2000). However,these studies failed to identify canonical Brn3a-binding sequences in this enhancer. If Brn3a directly regulates TrkA transcription, the binding sites on the TrkA enhancer remain unidentified.

In the present study, we have used genetic, biochemical and transgenic approaches to examine whether Brn3a is a direct or indirect regulator of TrkA transcription.

Brn3a^+/— mice and Bax^+/— mice were crossed to generate nine different genetic combinations with mixed genetic background: Brn3a^+/+/Bax^+/+; Brn3a^+/+/Bax^+/—; Brn3a^+/+/Bax^-/-; Brn3a^+/—/Bax^+/+; Brn3a^+/—/Bax^+/—; Brn3a^+/—/Bax^-/-; Brn3a^-/-/Bax^+/+; Brn3a^-/-/Bax^+/— and Brn3a^-/-/Bax^-/-. Four genotypes were used for experiments (see results). Mouse husbandry followed the guidelines of UTSW IACRAC.

We designed novel PCR primers to genotype Brn3a-mutant mice:wild-type 5′ primer, 5′-CTTGGCTTCCACTCAGCATCTGGAGC-3′;wild-type 3′ primer, 5′-CTGTATTCAGTGGAGAGAAGTGGAAACGG-3′;NEO primer, 5′-GATTCGCAGCGCATCGCCTTCTATCG-3′. PCR conditions were:94°C, 1 minute; 55°C, 1 minute; 72°C, 1 minute; 35 cycles. The size of the wild-type and mutant bands are ∼600 bp and ∼450 bp,respectively. The PCR primers and conditions for Bax genotyping were as described (Deckwerth et al.,1996): Bax exon primer 5′-TGATCAGAACCATCATG-3′; Bax intron primer 5′-GTTGACCAGAGTGGCGTAGG-3′; and Bax neo/pgk primer 5′-CCGCTTCCATTGCTCAGCGG-3′. PCR primers for β-galactosidase genotyping were: 5′-AACTGGAAGTCGCCGCGCCACTGGTGTGGG-3′ and 5′-TGAACTGCCAGCTGGCGCAGGTAGCAGAGC-3′. PCR conditions were:94°C, 30 seconds; 65°C, 30 seconds; 72°C, 30 seconds; 31 cycles. Generation of transient transgenic embryos and X-gal staining of the transgenic embryos were performed as described(Ma et al., 2000).

Escherichia coli BL21 competent cells were transformed with an expression plasmid for GST-Brn3a POU-domain fusion protein(Xiang et al., 1995). 1 liter cell culture at OD₆₀₀=0.6 was induced to express the fusion protein with 1 mM IPTG for 4 hours. The cell pellet was resuspended in lysis buffer and sonicated five times on ice for 10 seconds. The supernatant was purified with Glutathone-Sepharose 4B beads (Pharmacia). Cells transformed with the control plasmid expressing only GST were processed simultaneously.

EMSA was performed as described (Gruber et al., 1997) with minor modifications. Briefly, 10,000 to 40,000 cpm of ³²P-labeled oligonucleotide probe was mixed with 1 μl(0.3 μg) purified GSTBrn3a fusion protein, 5 × gel-shift buffer and 1μg poly(dI-dC) (Boehringer). Finally, water was added to a total volume of 20 μl. The mix was incubated at 25°C for 30 minutes. For competition assays, different amounts of cold oligonucleotides were mixed with radiolabeled probes. Electrophoresis was done using 5% polyacrylamide gels in 0.5×TBE buffer at 25°C for 3 hours. Gels were then dried and exposed with Kodak X-ray film.

DNase I footprint assays were performed as described(Brenowitz et al., 1995) with minor modifications. Either the 5′ or 3′ end of the TrkAminimal-enhancer fragment was labeled with [γ-³²P]-ATP. The labeled fragment was purified with phenol/chloroform extraction and ethanol precipitation. Different amounts of either purified, GST-Brn3a fusion protein or GST control protein was mixed with the labeled fragment and treated with graduated amounts of DNase I for 2 minutes at 25°C. The reaction was stopped by adding three volumes of cold ethanol, precipitated and run on sequencing gel. Sequencing reactions primed from the starting position of the labeled end were loaded onto the same gel to identify the protected sequences.

In situ hybridization was performed as described(Lei et al., 2001). The templates for making antisense in situ probes were TrkA (454 bp, extracellular domain) and NCAM (a 420 bp coding sequence). Sections were processed simultaneously for comparison.

Immunohistochemistry was performed as described(Ma et al., 2002) using the following antibodies and dilutions: monoclonal NeuN (1:500) (Chemicon); rabbit anti-NF200 (1:1000) (Sigma); rabbit anti peripherin (1:1000) (Chemicon); and rabbit anti TrkA (1:1000) (Advanced Targeting System). Antigen retrieval was performed as described by the manufacturer (BioGenex, CA) to detect TrkA signal. Sections were processed simultaneously with the same antibody for comparison.

Affymetrix microchip assay and analysis followed the manufacturers instructions (Affymetrix, CA). The source of labeled probe was trigeminal ganglia prepared from E13.5 Brn3a-null, heterozygous and wild-type littermates.

Brn3a-knockout mice exhibit extensive sensory neuron death in late gestation. During this process, expression of Trk family genes is noted to decrease (Huang et al., 1999; McEvilly et al., 1996). To determine whether Brn3a-mediated effects on TrkA expression are direct or secondary to unhealthy neurons that are in the process of perishing, we sought to uncouple neuronal survival from Trk-receptor expression and neurotrophin requirement.

The proapoptotic gene Bax is required to mediate apoptosis in sensory neurons (Deckwerth et al.,1996). Although the nociceptive sensory neurons are normally absent in the TrkA^-/- mice at birth, these neurons survive in TrkA^-/-/Bax^-/- mice(Patel et al., 2000). We reasoned that if we placed the Brn3a mutation(Xiang et al., 1996) in the Bax-null background, sensory neurons would be rescued, thus permitting analysis of TrkA expression in the absence of cell death. The genetic cross between Brn3a-null and Bax-null mice yields nine different genetic combinations (see Materials and Methods). To simplify our studies, we used the following four genotypes in our experiments: Brn3a^+/—/Bax^+/— (control); Brn3a^-/-/Bax^+/— (Brn3a null); Brn3a^+/—/Bax^-/-(Bax null); and Brn3a^-/-/Bax^-/- (double null). The Brn3a^+/—/Bax^+/— mice used as controls appear to be phenotypically normal with respect to sensory function, fertility and longevity.

To examine TrkA expression in trigeminal ganglia, we first performed in situ hybridization using a TrkA-specific probe with each of the four analyzed genotypes at postnatal day 0 (P0)(Lei et al., 2001; Martin-Zanca et al., 1990). As indicated in Fig. 1A1-B4, TrkA transcripts were reduced in the two genotypes that lack Brn3a (Brn3a-null and double-null pups). Thus, loss of Brn3a results in the downregulation of TrkA expression,which is unrelated to TrkA functional (trophic) dependence (n=3 for all genotypes). As shown in Fig. 1C1-C4, NCAM mRNA is present at similar level in all four genotypes.

We next examined trigeminal ganglia from P0 pups of the four genotypes using Nissl staining and neuron-specific-antibody immunostaining (peripherin,NeuN and NF-200; Fig. 2A1-B4and data not shown). No apparent difference in staining intensity was observed between the four genotypes, indicating the presence of apparently differentiated sensory neurons (n=3 for all genotypes). As anticipated, Bax-null mice have increased numbers of sensory neurons(Lonze et al., 2002; Patel et al., 2000) and Brn3a-null mice have decreased numbers of sensory neurons(Huang et al., 1999). Assessment of TrkA protein was consistent with the data obtained for mRNA expression. Control and Bax-null trigeminal ganglia have similar TrkA staining intensity (Fig. 2C1,C2,D1,D2) whereas both Brn3a-null and double-null ganglia exhibit reduced TrkA-specific staining(Fig. 2C3,C4,D3,D4). We confirmed the specificity of the TrkA antibody by performing immunostaining using sections of dorsal root ganglion from control and TrkA-null animals. Only weak background signals were detected in TrkA-null sections whereas strong, specific TrkA signals were observed in control sections (data not shown). We also stained trigeminal neurons from these mice with a Brn3a polyclonal antibody (Fedtsova and Turner,1995). Both Brn3a-null and double-null neurons lost the Brn3a signal in the nucleus whereas control and Bax-null neurons maintained specific, nuclear staining (Fig. 2E1-E4). This confirms the results of in situ hybridization studies (Fig. 1): that loss of Brn3a contributes to the loss of TrkA expression in the absence of cell death.

Another independent genetic method to assess whether TrkAexpression was dependent on Brn3a function was employed by crossing transgenic mice containing a β-galactosidase gene driven specifically by the TrkA minimal enhancer in nociceptive neurons(Ma et al., 2000) with control, Bax-null, Brn3a-null, and double-null mice. In the absence of Brn3a, sensory neurons in trigeminal ganglia have apparently reduced β-galactosidase signal compared to either control or Bax-null ganglia. These results are consistent with the immunohistochemistry and in situ hybridization experiments(Fig. 3). Thus, in vivo, the TrkA minimal enhancer is significantly attenuated in the absence of functional Brn3a, as assayed by β-galactosidase staining.

In situ hybridization and immunohistochemical approaches discern the qualitative loss of TrkA expression in Brn3a-null sensory neurons, which is most apparent around P0 (Figs 1, 2)(Huang et al., 1999). To obtain more quantitative information and gain better insight into the timing of Brn3a requirement for TrkA expression, we analyzed the gene-expression profile of E13.5 trigeminal ganglia from wild-type,heterozygous and Brn3a-null mice using Affymetrix gene-expression assays (Table 1). Consistent with preceding data at P0 and the following transgenic analysis (see below;Figs 5, 6), TrkA mRNA was already downregulated two-fold in E13.5 Brn3a-null trigeminal ganglia compared to controls (Table 1). By contrast, the expression of multiple housekeeping and nervous-system-specific genes was unchanged in Brn3a-null embryos at E13.5 (Table 1). Therefore, as early as E13.5, before the onset of sensory apoptosis in Brn3a-null mice, loss of Brn3a attenuates TrkA expression.

Collectively, these results demonstrate that TrkA expression is reduced but not absent in the context of Brn3a mutation, regardless of whether the neurons are dying (Brn3a-null) or alive (double-null). In addition, the genetic data indicate that Brn3a acts upstream of TrkA and lead to the hypothesis of a direct transcriptional control of the TrkA minimal enhancer.

To address whether the genetic interaction between Brn3a and TrkA is direct, we searched the minimal enhancer sequence of TrkA for consensus Brn3a-binding sites, either GCAT(A/T)A(T/A)T(A/T)AT (Gruber et al.,1997) or (A/G)CTCATTAA(T/C)(Xiang et al., 1995). No canonical Brn3a-binding sites could be identified. We next scrutinized the full, 2 kb promoter/enhancer region upstream of the TrkA gene and found no consensus Brn3a-binding sequences (data not shown).

We then turned to an empirical approach using purified GSTBrn3a DNA-binding-domain-fusion proteins (Xiang et al., 1995) to perform gel-shift assays with six, overlapping,100 bp DNA fragments that cover the entire 457 bp minimal enhancer(Fig. 4A). Only the two,outlying, DNA fragments at the 5′ and 3′ ends of the minimal enhancer were shifted by the GST-Brn3a fusion protein(Fig. 4B), indicating the presence of novel Brn3a-binding sites in these two regions. As controls, GST protein alone did not cause any shift whereas an anti-GST antibody disrupted the interaction of F1 and F6 with the GSTBrn3a protein (data not shown). To identify the core Brn3a-binding sequences in the two 100 bp fragments, we performed DNase I footprint analyses (Ausubel et al., 1995) with the entire 457 bp minimal enhancer using the GST-Brn3a fusion protein. As shown in Fig. 4C and D, sequences protected by the GST-Brn3a fusion protein were present in the 5′ and 3′ sequences that were included in the gel-shifted fragments(Fig. 4A,B). The length of both protected DNA sequences is <40 bp, providing more accurate information about the core binding sites of Brn3a. These two protected sequences compete successfully with the corresponding 100 bp sequences for GST-Brn3a binding in a gel shift assay (Fig. 4E,F,left panels), providing further evidence of Brn3a-specific DNA sequences in the TrkA minimal enhancer.

To pinpoint the Brn3a core-binding sequences in the minimal enhancer, we compared the two protected sequences. A similar A/T-rich sequence, T(A/T)ATT,that resembles the consensus Brn3a core binding site was found in both fragments (Fig. 4E,F, red shadows, top panels). To determine the importance of the core T(A/T)ATT sequence for Brn3a binding, we mutated these sequences in both protected oligonucleotides (Fig. 4E,F,red, top panels) and tested them in gel-shift assays(Fig. 4E,F, lower panels). Although wild type oligonucleotides successfully competed for the binding of GST-Brn3a fusion proteins (Fig. 4E,F, lanes 1-4), the mutant oligonucleotides lost the ability to compete with wild-type sequences or to bind the GST-Brn3a fusion protein(Fig. 4E,F, lanes 5-8). These data indicate the presence of two novel, specific Brn3a-binding sites in the TrkA minimal enhancer that could mediate the effect of Brn3aon TrkA transcription.

We have previously defined several cis elements in the TrkAminimal enhancer that are required for transcription through site-directed mutagenesis and transgenic assays (Ma et al., 2000). We therefore used similar methods to determine the importance of the newly defined Brn3-binding sites for in vivo enhancer function. The 5′ and 3′ core sequences defined in Fig, 4E,F (boxed sequences)were subjected to mutagenesis and then to transient transgenic analysis(Ma et al., 2000).

As previously described, the wild type TrkA minimal enhancer drives strong and specific lacZ expression in trigeminal ganglia in >50% of lacZ positive embryos at E13.5 (Fig. 5, E13.5 and Fig. 6A). The variability is related to transgene positional effects(Ma et al., 2000). When the minimal enhancer harbors the 5′ Brn3a-site mutation, two out of two(100%) blue E13.5 embryos exhibited normal expression in trigeninal ganglia(Fig. 5 and Fig. 6B). Mutation of the 3′ Brn3a site resulted in only one out of four blue embryos maintaining normal trigeminal expression (Fig. 5, Fig. 6C for normal expression, and data not shown).

We next tested the functional consequences of mutating both the 3′and 5′ Brn3a-binding sites in the enhancer. Of ten blue E13.5 embryos,only one exhibited appreciable expression in the trigeminal ganglion (data not shown), the remaining nine embryos either had extremely weak, or undetectable trigeminal-ganglion expression (Fig. 5, Fig. 6D, and data not shown). These results are consistent with the microarray experiments indicating that by E13.5, mutation of both Brn3a sites significantly compromised the function of the TrkA minimal enhancer in trigeminal ganglia. Mutation of each site alone has either no detectable (5′ site)or weak (3′ site) effect on TrkA-enhancer function.

We extended our transient transgenic analysis to E17.5 with the minimal enhancer bearing mutations in either one or both Brn3a-binding sites. As predicted, the wild-type enhancer drives β-galactosidase expression strongly and specifically in nociceptive neurons in trigeminal ganglia at E17.5 (Fig. 5 and Fig. 6E)(Ma et al., 2000). Neither mutation alone (either 5′ or 3′) appreciably affected the activity of the TrkA minimal enhancer at E17.5(Fig. 5; Fig. 6F,G). Finally, to test if these two Brn3a sites in the minimal enhancer are redundant for expression in the trigeminal ganglion at E17.5, transient transgenic embryos were generated with the minimal enhancer bearing both the 5′ and 3′ sites mutations. Of eight blue embryos generated, seven showed either greatly reduced or no detectable expression in trigeminal ganglia(Fig. 5; Fig. 6H,I, and data not shown). The remaining blue embryo expressed β-galactosidase comparably to wild type and single-mutant embryos (data not shown). We also generated TrkA minimal enhancer transgenic P0 pups bearing both Brn3a-binding-site mutations, with similar results. Double mutations clearly affected enhancer function in trigeminal ganglia(Fig. 5 and data not shown),with only three out of fifteen blue pups maintaining normalβ-galactosidase expression in trigeminal ganglia. Taken together, our transgenic results indicate that these two Brn3a sites are important for maintaining TrkA enhancer function after E13.5, although they are functionally redundant.

The differentiation of sensory neurons has been closely linked to the specific expression of Trk family receptors in subtypes of sensory neurons(Liebl et al., 1997; Ma et al., 2000; Snider, 1994). The molecular programs that specify sensory neurons are largely unknown. Studies in other areas of the nervous system, such as motor neurons in spinal cord(Tanabe and Jessell, 1996),telencephalic neurons (Nieto et al.,2001; Schuurmans and Guillemot, 2002) and sympathetic neurons(Pattyn et al., 1997; Pattyn et al., 1999), show that spatial and stage-dependent interaction of multiple transcription factors is the driving force for the specification of neuronal subtypes. We undertook to understand the transcriptional regulation of Trk genes to derive clues about the molecular mechanisms of sensory-neuron specification.

Several transcription factors have been shown to function in sensory and sympathetic gangliogenesis. For example, the bHLH transcription factors,neurogenin1, neurogenin2 and mash1, are required for normal development of sensory neurons (Anderson,1999; Anderson et al.,1997). Because of their early expression, these factors are likely to affect neural-precursor proliferation rather than the differentiation and survival of postmitotic neurons (Ma et al., 1999). It is unclear whether these transcription factors participate in regulating expression of TrkA.

Pou-domain proteins exert crucial effects on neural development. In the PNS, Brn3a has received considerable attention because of its striking pattern of expression and the dramatic phenotypic consequences of Brn3aablation in mice, which includes loss of sensory neurons(He et al., 1989; Huang et al., 1999; McEvilly et al., 1996; Xiang et al., 1996). A key,unresolved question has been the underlying mechanistic requirements for Brn3a by sensory neurons. Although putative Brn3a-target genes have been proposed(Smith et al., 1997a; Smith et al., 1998; Smith et al., 1999; Smith et al., 1997b), the relevance of these genes in vivo remains to be tested.

Brn3a-mutant sensory neurons from trigeminal ganglia have decreased expression of TrkA, beginning at E13.5. Because Brn3a expression appears at E9, the observed downregulation of TrkA at E13.5 and subsequent neuronal death in Brn3a-null mice could be caused by neuronal defects other than direct regulation of TrkA by Brn3a (Eng et al.,2001; Huang et al.,1999). Even if TrkA is regulated by Brn3a, it is unclear whether this is direct regulation, mediated by Brn3a binding to TrkA cis-regulatory sequences, or indirect regulation, mediated by other factors. To investigate these possibilities, we sought to dissociate neuronal survival from Trk-receptor expression and function.

The use of Bax-null mice permitted sensory neurons to survive in the absence of TrkA function because Bax is required for neurotrophin-deprived neuronal apoptosis (Deckwerth et al.,1996). Furthermore, the sensory neurons rescued from apoptosis by crossing into a Bax-null background are known to express normal sensory neuron markers and have been utilized to uncouple apoptosis from other biological events, such as peripheral axonal outgrowth and elaboration(Deckwerth et al., 1996; Lonze et al., 2002; Patel et al., 2000). Using a similar approach, we were able to discern the appearance of sensory neurons in the absence of functional Brn3a, which supports the model that Brn3ais not required for early specification and differentiation of sensory neurons. Instead, these neurons were present but had reduced expression of TrkA transcripts and proteins. These results are consistent with transcriptional regulation of TrkA expression by Brn3a.

The two Brn3a sites in the TrkA enhancer are not identical to binding sequences described previously(Gruber et al., 1997; Xiang et al., 1995). Analysis in vitro indicates that they have lower affinity than consensus Brn3a sites,although they are well conserved between mice and humans (data not shown)(Gruber et al., 1997). As expected, mutation of both sites did not completely ablate reporter-gene expression in sensory neurons. A reduced, but apparent, level of enhancer activity still exists. Considering possible enhancing and suppressing effects of genomic sequences surrounding the transgene-insertion sites, the mutagenized enhancers might exhibit either near normal or much weaker activity. However, the trend of reduced activity should be maintained if a significantly large number of transgenic embryos are analyzed. Our transient transgenic analysis fits this prediction nicely.

How does binding affinity in vitro relate to function in vivo? The physiological environment provided by chromatin and available cofactors in a given cell nucleus will impact on the accessibility of a transcription factor to its DNA-binding site and, thus, the physiological affinity. In Caenorhabditis elegans, Drosophila and mammals, there are many examples that low affinity sites in vitro are important for transcription in vivo. Among these are the FoxA transcription factor PHA-4 in C. elegans (Gaudet and Mango,2002), bicoid and dorsal morphogen in Drosophila(Driever et al., 1989; Jiang and Levine, 1993), and HNF4 in the regulation of erythropoietin expression and Elf-1 in lymphoid-specific-gene expression in mammals(John et al., 1996; Makita et al., 2001). In the case of HNF4-mediated regulation of erythropoietin, transcriptional initiation is mediated by a high affinity interaction with the retinoid acid receptors,which is then supplanted by low affinity binding of HNF4(Makita et al., 2001). Similarly, Brn3a is not required for initiation of TrkA transcription but is required for its maintenance, beginning at E13.5. Because TrkA is still expressed in Brn3a-null sensory neurons, although at a lower level,other transcription factors (Lei et al.,2001) may also function to regulate TrkA expression. These factors may interact with Brn3a to regulate TrkA expression in wild-type neurons.

Transcriptional cofactors also affect the activity of a transcription factor. UNC86, a Brn3a ortholog in C. elegans, interacts with the LIM-type homeodomain protein MEC-3 and the two proteins function synergistically to activate the mec-3 promoter in vitro (Xue, 1992;Xue, 1993; Lichtsteiner, 1995). In addition, both proteins are required for the fate specification of touch cells in C. elegans through heterodimeric binding to consensus cis-elements(Duggan et al., 1998). Recent experiments demonstrate further examples of how interactions between the Oct-1 POU-domain protein and SNAP190 cofactor affect DNA binding(Hovde et al., 2002). It remains interesting to identify Brn3a cofactors that are important for TrkA expression and the development of mammalian sensory neurons.

It is interesting to note that no significant differences in TrkAmRNA levels were detected in Brn3a wild-type and heterozygous embryos(Table 1). This finding is consistent with our recent observation that autoregulation of the Brn3a locus leads to suppression of haplotype insufficiency, and results in similar levels of Brn3a mRNA in trigeminal neurons that contain either one or two copies of the Brn3a gene(Trieu et al., 2003). It is also consistent with studies that show no detectable phenotype in Brn3a heterozygotes (Eng et al.,2001; Huang et al.,1999) and validates the use of heterozygotic controls in the present study.

Brn3a-null sensory neurons have aberrant axonal arborization at E13.5 (Eng et al., 2001). At this time TrkA protein levels are minimally reduced(Huang et al., 1999). Lack of TrkA/NGF signaling leads to defective innervation of peripheral targets by nociceptive neurons (Patel et al.,2000). Considering that Brn3a might regulate multiple target genes in sensory neurons, it is likely that the sensory axonal defects observed in Brn3a-null mice are the consequences of altered gene expression at many loci, including TrkA. Identification of additional Brn3a target genes should shed light on these complex neuronal phenotypes.

Dr M. Xiang kindly provided the GST-Brn3a POU-domain fusion protein expressing plasmid. S. Mckinnon, J. Schklair and Z. Jorai provided excellent help with genotyping and maintenance of the mice. S. Hall generated the transgenic embryos. Y. J. Liu and P. Houston assisted with histology and immunohistochemistry. Dr H. Ren provided expert help with the gelshift experiment. LFP is supported by the NINDS and by the Christopher Reeve Paralysis Foundation Consortium.