Brn3a/Pou4f1 Regulates Dorsal Root Ganglion Sensory Neuron Specification and Axonal Projection into the Spinal Cord

Abstract

The sensory neurons of the dorsal root ganglia (DRG) must project accurately to their central targets to convey proprioceptive, nociceptive and mechanoreceptive information to the spinal cord. How these different sensory modalities and central connectivities are specified and coordinated still remains unclear. Given the expression of the POU homeodomain transcription factors Brn3a/Pou4f1 and Brn3b/Pou4f2 in DRG and spinal cord sensory neurons, we determined the subtype specification of DRG and spinal cord sensory neurons as well as DRG central projections in Brn3a and Brn3b single and double mutant mice. Inactivation of either or both genes causes no gross abnormalities in early spinal cord neurogenesis; however, in Brn3a single and Brn3a;Brn3b double mutant mice, sensory afferent axons from the DRG fail to form normal trajectories in the spinal cord. The TrkA⁺ afferents remain outside the dorsal horn and fail to extend into the spinal cord, while the projections of TrkC⁺ proprioceptive afferents into the ventral horn are also impaired. Moreover, Brn3a mutant DRGs are defective in sensory neuron specification, as marked by the excessive generation of TrkB⁺ and TrkC⁺ neurons as well as TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ double positive cells at early embryonic stages. At later stages in the mutant, TrkB⁺, TrkC⁺ and parvalbumin⁺ neurons diminish while there is a significant increase of CGRP⁺ and c-ret⁺ neurons. In addition, Brn3a mutant DRGs display a dramatic down-regulation of Runx1 expression, suggesting that the regulation of DRG sensory neuron specification by Brn3a is mediated in part by Runx1. Our results together demonstrate a critical role for Brn3a in generating DRG sensory neuron diversity and regulating sensory afferent projections to the central targets.

Introduction

To correctly perceive and distinguish different sensations, the sensory neurons of particular modalities must project accurately to specific targets in the central nervous system (CNS). Identifying the mechanisms by which sensory modality and central connectivity are specified and appropriately coordinated has been a major focus of study in neural development. Specific sensory neurons of the dorsal root ganglion (DRG) express trans-membrane receptor tyrosine kinases that transduce signals from neurotrophic factors. TrkA is expressed by many cutaneous nociceptive and thermoceptive neurons, TrkB by a subpopulation of cutaneous mechanoreceptive neurons, and TrkC by proprioceptive neurons (Gonzalez-Martinez et al., 2004; Molliver et al., 1995; Oakley et al., 2000; Snider, 1994). Previously it has been shown that Ngn1 and Ngn2 are required to generate TrkA⁺ and TrkB⁺ or TrkC⁺ DRG neurons, respectively (Ma et al., 1999). The response of these Trk receptors to trophic ligands is essential for neuronal survival (Bibel and Barde, 2000; Huang and Reichardt, 2003). Moreover, the Trk-mediated signal transduction plays important roles in controlling sensory axon projections and the expression of neuropeptides, and therefore is required for forming accurate circuits to the periphery and central nervous systems (Hellard et al., 2004; Hippenmeyer et al., 2004; Moqrich et al., 2004; Oakley et al., 2000; Patel et al., 2000; Patel et al., 2003).

The Brn3/Pou4f (Brn3a/Pou4f1, Brn3b/Pou4f2 and Brn3c/Pou4f3) genes encode three closely related transcription factors characterized by the presence of a DNA-binding POU domain (Gerrero et al., 1993; Turner et al., 1994; Xiang et al., 1995; Xiang et al., 1993). They play crucial roles in neuronal specification, differentiation and survival during development of both central and peripheral nervous systems. Thus, targeted disruption of Brn3a in mice results in neuronal loss and axon guidance defects in the trigeminal ganglion (TG), inner ear sensory ganglia, and brainstem (Huang et al., 2001; Huang et al., 1999; McEvilly et al., 1996; Xiang et al., 1996); Brn3b inactivation causes loss, mis-specification, and aberrant axon projections of retinal ganglion cells (Badea et al., 2009; Erkman et al., 1996; Erkman et al., 2000; Gan et al., 1999; Gan et al., 1996; Mu et al., 2004; Qiu et al., 2008; Wang et al., 2000; Xiang, 1998); and elimination of Brn3c leads to loss of all the cochlear and vestibular hair cells, resulting in complete deafness and profound deficit in the vestibular system (Erkman et al., 1996; Xiang et al., 1997; Xiang et al., 1998).

Brn3 factors are expressed in the DRG, trigeminal ganglion (TG) and dorsal spinal cord (Fedtsova and Turner, 1995; Gerrero et al., 1993; McEvilly et al., 1996; Ninkina et al., 1993; Xiang et al., 1996; Xiang et al., 1995). Previous studies have revealed critical roles of Brn3a in controlling sensory neuron development in the TG, including cell survival, transcriptional regulation of Trk receptors, and peripheral axonal projections (Dykes et al., 2010; Dykes et al., 2011; Eng et al., 2007; Eng et al., 2001; Eng et al., 2004; Huang et al., 2001; Huang et al., 1999; Lei et al., 2006). Microarray analysis (Eng et al., 2007; Eng et al., 2004) and locus-wide chromatin immunoprecipitation (Lanier et al., 2007) have identified a large number of candidate genes which might be effectors of development programs regulated by Brn3a in both TGs and DRGs, and defined functions of Brn3a during sensory neuron development in repressing early regulators of neurogenesis, repressing gene expression programs of other tissues, creating permissive conditions for major subtype-specifying genes, and activating specific genes essential for sensory functions. Indeed, a recent report has demonstrated that Brn3a is involved in specifying sensory neuron subtypes during TG development (Dykes et al., 2010). However, a detailed investigation of Brn3a function in subtype diversity of DRG sensory neurons and spinal cord interneurons is still missing.

To better understand the requirement of Brn3 factors for DRG and spinal cord development, we examined subtype specification of DRG and spinal cord sensory neurons as well as central projections of DRG afferents into the spinal cord in Brn3a and Brn3b single and double mutant mice. Although the dorsal interneurons of the spinal cord were generated without obvious defects in these mutants, we observed severe sensory neuron specification defects in Brn3a mutant DRGs that resulted in profound abnormal central projections into the spinal cord. There was a near complete absence of nociceptive afferents from the dorsal horn gray matter of Brn3a mutant spinal cords. These results suggest that Brn3a is required for proper specification of DRG sensory neuron subtypes, and for regulating their projections to the central targets.

Materials and Methods

Generation and maintenance of Brn3a and Brn3b single and double mutant mice

Mice carrying targeted null-mutations in Brn3a and Brn3b have been described previously (Gan et al., 1999; Xiang et al., 1996). In order to generate Brn3a;Brn3b double mutant mice, Brn3a^+/− mice were crossed with Brn3b^−/− animals and the cross between Brn3a^+/−Brn3b^+/− and Brn3a^+/−Brn3b^−/− mice produced Brn3a^−/−Brn3b^−/− double mutants. Both Brn3a^+/−Brn3b^+/− double heterozygotes and Brn3a^+/−Brn3b^−/− mutant mice are viable and fertile, whereas Brn3a^−/−Brn3b^−/− double mutant mice die shortly after birth.

Immunohistochemistry

All immunostaining was performed on cryosections following standard protocols. Whole mouse embryos younger than E14.5 were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C. The cervical, thoracic and lumbar segments were first dissected from E14.5 and older embryos, and then fixed. After fixation, samples were immersed in 30% sucrose (in PBS) at 4°C overnight and frozen in Tissue Tek O.C.T. compound (Sakura Finetek USA). 14 µm transverse sections were cut and collected onto Superfrost Plus microslides (VWR). For immunostaining, the following primary antibodies were used: mouse anti-Brn3a (Millipore, 1:200), goat anti-Brn3b (Santa Cruz Biotechnology, 1:400), rabbit anti-Brn3c (Xiang et al., 1995), rabbit anti-TrkA (Millipore, 1:1000), goat anti-TrkB (R&D systems, 1:1000), goat anti-TrkC (R&D systems, 1:1000), mouse anti-Isl1 (DSHB, 1:500), rabbit anti-Pax2 (Zymed, 1:200), mouse anti-Lhx1/5 (DSHB, 1:100), rabbit anti-Lbx1 (Gross et al., 2000), rabbit anti-CGRP (Millipore, 1:2000), goat anti-CGRP (AbD Serotec, 1:4000), rabbit anti-substance P (Protos Biotech Corporation, 1:2000), rabbit anti-parvalbumin (Swant, 1:2000), rabbit anti-c-ret (IBL, 1:50), rabbit anti-Runx1 (Cell signaling, 1:400), mouse anti-Runx3 (Abnova, 1:500), rabbit anti-active caspase-3 (BD Pharmingen, 1:200). The sections were blocked in 5% normal donkey serum and 0.25% Triton X-100 in PBS at room temperature for 1 hour, incubated with primary antibodies at 4°C overnight, followed by incubation with Alexa Fluor 488 or 594 labeled donkey anti-rabbit, mouse, or goat IgG secondary antibodies (Invitrogen, 1:1000). DAPI (Vector Laboratories) and Topro-3 (Invitrogen) were used for nuclear counter-staining. If needed, sections were first subjected to heat-induced antigen retrieval in DakoCytomation Target Retrieval Solution (Citrate, pH6.0). Images were captured with a Nikon Eclipse 80i microscope or a laser scanning Leica TCS-SP2 confocal microscope.

RNA in situ hybridization and EdU labeling

RNA in situ hybridization was carried out as described previously (Liu et al., 2001; Mo et al., 2004). Digoxigenin-labeled riboprobes were prepared following the manufacturer's protocol (Roche Diagnostics). The mouse Tlx3 probe was a gift from Dr. Qiufu Ma, and mouse Lmx1b probe was amplified by PCR from mouse embryonic spinal cord cDNA, with 5’ primer: CCGGAATTCAACGAGTCGTCCTGGCACGAGG, and 3’ primer: CCCAAGCTTCGGTGTAGGAGGCCATCATGCC, as described previously (Gray et al., 2004).

For EdU labeling, the Click-iT EdU labeling kit was purchased from Invitrogen. Timed pregnant mice were injected with EdU solution (160 µg/g body weight) and sacrificed 1 hr later. Embryos were collected, fixed and processed according to standard immunostaining protocol. EdU staining was performed according to the procedure provided by the kit.

DiI labeling

E18.5 mice were fixed in 4% paraformaldehyde in PBS at 4°C overnight. Crystals of a lipophilic fluorescent dye (DiI, Invitrogen) were placed onto the Th9-10 DRGs, after cutting off the ventral roots. The dyes were allowed to retrogradely label DRG axons for 14 days at 37°C. Then 14 µm cryosections were prepared and counter-stained with DAPI. Images were captured with a Nikon Eclipse 80i microscope.

Cell counting

To count immunoreactive DRG neurons, level-matched DRGs from thoracic segments at E11.5 and E12.5 were serially sectioned at 14 µm. Sections were co-immunostained with antibodies against specific cell markers and Isl1. 10–12 sections and 15 sections from level-matched thoracic DRGs per animal per marker were counted for E11.5 and E12.5, respectively. For E15.5 and E18.5 samples, 15–30 sections from level-matched thoracic DRGs per animal per marker were counted. To count TrkA/B and TrkA/C double immunereactive cells, sections were triple-immunostained with antibodies against TrkA, TrkB or TrkC, and Isl1 (for visualizing the neuclei). At least 3 embryos of each genotype were counted for all stages and all markers. To calculate the immunoreactive cells per mm², we measured the Isl1⁺ DRG cross-sectional area from each section using the NIH ImageJ software. For counting Isl1⁺ cells at E11.5, E12.5 and E18.5, serial sections from level-matched thoracic DRGs from control and mutant animals were stained. The Isl1⁺ cell number per ganglion was calculated as the total positive cells from all the sections. Only immunoreactive cells with a clear nucleus were counted.

Statistical analysis

All quantifications were based on at least 3 pairs of embryos. All data were tested for significance using two sample Student's t test with unequal variances. Differences were considered significant if the p value was less than 0.05. All results were expressed as the mean±s.d..

Results

Expression of Brn3 factors in DRG sensory neurons during mouse embryogenesis

We investigated the expression patterns of Brn3 factors in the developing DRG by immunostaining using antibodies that specifically recognize the three proteins. As early as E10.5, strong Brn3a expression was detected in all DRG neurons immunoreactive for the neuron marker Isl1, whereas only a few cells were found to weakly express Brn3b or Brn3c by this stage (Figs. 1A, C, and E). From E11.5 throughout embryonic development, Brn3a maintained its strong expression in most DRG neurons (Figs. 1B, G, I, K, and M). Meanwhile, Brn3b was expressed in increasingly more neurons, and by E15.5, these two proteins share a near completely overlapping expression pattern in the DRG (Figs. 1D, G, and I). On the other hand, Brn3c was seen in a small number of neurons that co-express Brn3a at all these embryonic times (Figs. 1F, K, and M). Interestingly, in Brn3a null mutant DRGs, we failed to detect Brn3b⁺ cells despite their normal presence in the spinal cord (Figs. 1H and J). This indicates that Brn3b expression is dependent on Brn3a in the DRG but not in the spinal cord. By contrast, Brn3c⁺ neurons were present in Brn3a null DRGs at E11.5 and E12.5 but had completely disappeared by E15.5 (Figs. 1L and N), suggesting that Brn3a is required for maintaining Brn3c expression but not for initiating its expression in the DRG.

Fig. 1.

Expression patterns of Brn3 proteins during mouse DRG sensory neuron development. (A–F) Transverse DRG sections from wild-type mouse embryos at indicated developmental stages were co-immunostained with antibodies against Isl1 (A–F) and Brn3a (A and B), Brn3b (C and D), or Brn3c (E and F). They were also counter-labeled with DAPI (blue). Isl1 was used as a pan-sensory maker for the DRG. Brn3a is strongly expressed in almost all Isl1-labeled cells at E10.5 and E11.5, whereas Brn3b and Brn3c are expressed weakly only in some of the Isl1-immunoreactive cells at these stages. Insets show Brn3b- (C and D) and Brn3c- (E and F) immunoreactive cells in corresponding outlined regions at a higher magnification and arrows point to representative Brn3b- or Brn3c-immunoreactive cells. (G–N) Co-expression of Brn3b or Brn3c with Brn3a in DRG sensory neurons and the regulation of their expression by Brn3a at E12.5 and E15.5. Brn3b is expressed in a subset of Brn3a-immunoreactive cells at E12.5 but in the majority of them at E15.5 (G and I). Brn3c is expressed in a subset of Brn3a-exrpessing cells at both E12.5 and E15.5 (K and M). In Brn3a^−/− mice, the expression of Brn3b is abolished in the DRG at E12.5 and E15.5 (H and J), but not in the spinal cord (H). The expression of Brn3c is largely unchanged in Brn3a^−/− DRGs at E12.5 (L), but becomes abolished at E15.5 (N). (O–Q) Transverse sections from E10.5 (O and P) and E11.5 (Q) embryos pulse-labeled with EdU were double-stained for Brn3a and EdU. None of the Brn3a-immunoreactive cells were co-labeled by EdU in the spinal cord (O and P) and only a few were colabeled in the DRG (P and Q). Arrows in P and Q point to representative co-localized cells. (R–T) Double-immunostaining between Brn3a and TrkA, TrkB or TrkC showed co-localization of Brn3a and these Trk receptors in E11.5 DRG neurons. Insets show corresponding outlined regions at a higher magnification. Abbreviations: drg, dorsal root ganglion; sc, spinal cord. Scale bar: A–F, P–T, 50 µm; G–O, 100 µm.

Immunostaining EdU-labeled DRG samples with an anti-Brn3a antibody showed that Brn3a was largely expressed in post-mitotic cells in the DRG at E10.5 and E11.5, although occasional EdU⁺ mitotic progenitors could be detected to express Brn3a (Figs. 1O–Q). By double-immunolabeling with anti-TrkA, anti-TrkB and anti-TrkC antibodies that mark cutaneous nociceptive/thermoceptive sensory neurons, a subpopulation of cutaneous mechanoreceptive neurons, and proprioceptive neurons, respectively, we found that Brn3a was expressed in all of them at E11.5 (Figs. 1R–T), the time when all three proteins are detected in the DRG. This raises the possibility that Brn3 factors, and Brn3a in particular, may play a role in the specification and/or differentiation of DRG sensory neurons during mouse embryogenesis.

Defective primary sensory projection patterns in the spinal cord of Brn3a single and Brn3a;Brn3b double mutant embryos

To determine whether Brn3 factors were involved in DRG sensory neuron development, we examined DRG sensory afferent projections to the spinal cord in Brn3a single and Brn3a;Brn3b double mutant mice by immunostaining with antibodies that specifically recognize different types of sensory projections. We did not detect any sensory defects in Brn3a^+/− or Brn3b^+/− heterozygous mice compared to the wild type. Therefore, we used both wild type and heterozygous animals as controls when analyzing phenotypes in homozygous mutants.

The axons of DRG sensory neurons enter the spinal cord via the dorsal root entry zone (DREZ) where they extend branches rostrally and caudally within the dorsal funiculus (Mirnics and Koerber, 1995; Ozaki and Snider, 1997). The proprioceptive sensory neurons project their axons into the medial half of the dorsal funiculus where they navigate toward the motoneurons in the ventral spinal cord. These projections can be recognized by antibodies against TrkC and parvalbumin. In contrast, the nociceptive and thermoceptive projections are positioned more laterally in the dorsal funiculus and terminate in the dorsal-most laminae of the cord. These axons can be recognized by the anti-TrkA antibody (Chen et al., 2006a; Yoshida et al., 2006). At E12.5 when many DRG sensory axons have reached the DREZ, both TrkA⁺ and TrkC⁺ axons were detected near the DREZ in Brn3a^+/− and Brn3a^−/− mice (Figs. 2A–D). In the heterozygous control mice, the overlapped region of TrkA⁺ and TrkC⁺ axon oval bundles was relatively small. TrkC⁺ bundles were located more medially compared to TrkA⁺ bundles (Fig. 2A, C, and E). In Brn3a^−/− mice, however, we noticed that the detected fluorescent signal of TrkC⁺ axons was slightly weaker, and TrkC⁺ axons extensively overlapped with TrkA⁺ axons. Moreover, the TrkA⁺ axons concentrated to a more-medial region (Fig. 2B, D, and F). As a result, the thick TrkA⁺/TrkC⁺ bundle deformed the normal shape of the spinal cord in the mutant (Figs. 2G and H). This sensory projection pattern change suggests that there may be mis-specification and/or improper differentiation of proprioceptive and nociceptive neuron subtypes in Brn3a^−/− DRGs.

Fig. 2.

Altered TrkA⁺ and TrkC⁺ central afferent projection patterns in E12.5 Brn3a mutant spinal cords at the forelimb level. The bundles of TrkA⁺ (red) and TrkC⁺ (green) afferent fibers are detected by immunostaining from lateral to medial regions of the dorsal horn edge in Brn3a^+/− spinal cords (A, C, and E). In Brn3a^−/− spinal cords, TrkA⁺ and TrkC⁺ fibers are narrowly located in a much smaller region, and extensively overlap with each other (B, D, and F). DAPI counter-staining (blue) shows a deformed region of dorsal horn (arrow head) in the Brn3a^−/− spinal cord (H, compared to G). Brn3a^+/− and wild-type embryos are indistinguishable in these axonal projection patterns. Scale bar: 25 µm.

At E15.5 when the TrkA⁺ afferent fibers were detected penetrating into the dorsal horn of the spinal cord in Brn3a^+/+Brn3b^+/− mice, we did not detect any obvious TrkA⁺ fibers within the dorsal horn of Brn3a^−/−;Brn3b^+/− mutant mice (Figs. 3E and F). The decreased TrkA⁺ fiber ingrowth in the Brn3a mutant is accompanied by a significant increase of TrkA⁺ axon bundles remained in the dorsal funiculus (Figs. 3C–F). The thickened dorsal funiculus could also be visualized by labeling with an antibody against the neuronal intermediate filament protein peripherin (Figs. 3A and B). There was no obvious difference between the wild type and Brn3a mutant mice in terms of the TrkC⁺ afferent projections into the ventral spinal cord (Figs. 3C–F). By E18.5, the latest time that was analyzed because of postnatal lethality of the Brn3a mutant mice, the TrkA⁺ afferents failed to penetrate into the dorsal horn grey matter in the mutant, and thick bundles of TrkA⁺ fibers were detected mostly outside the dorsal horn (Figs. 3I–L). The number of TrkC⁺ and parvalbumin⁺ proprioceptive afferents projecting to the ventral horn is also decreased in the Brn3a mutant (Figs. 3G–L). The severity of TrkA⁺ and TrkC⁺ afferent projection defects appeared to depend on the axial position, with more severe defects observed in cervical and lumbar regions of the mutant spinal cord (Fig. S1). We did not observe any obvious difference in primary sensory projection patterns between wild type and Brn3b mutant mice at all developmental stages examined (Figs. S2A and B; and data not shown). Moreover, Brn3a^−/− and Brn3a^−/−;Brn3b^−/− mutant animals exhibited essentially identical DRG afferent projection defects (Figs. S2C and D), consistent with the loss of Brn3b expression in Brn3a^−/− DRGs Figs. 1H and J).

Fig. 3.

Aberrant central projections of DRG sensory afferents in Brn3a mutant spinal cords. (A–R) Transverse thoracic spinal cord sections of indicated genotypes from E15.5 and E18.5 were immunostained with antibodies that recognize specific sensory afferents. (A–F) At E15.5, thicker peripherin- and TrkA-immunoreactive afferent fibers are found at the edge of the dorsal horn in Brn3a mutants. Boxes in C and D are magnified in E and F, respectively. The asterisk in E indicates TrkA⁺ afferents in the dorsal horn of the control spinal cord, where they are missing in Brn3a mutants (E and F). No obvious difference was observed for TrkC⁺ projections between the control and the mutant (C–F). (G–L) At E18.5, parvalbumin⁺ proprioceptive projections toward the ventral horn are largely reduced in the Brn3a mutant spinal cord (G and H). The TrkA⁺ nociceptive projections into the dorsal horn are also greatly reduced in the Brn3a mutant spinal cord. Instead, TrkA⁺ afferent fibers mostly stall at the edge of the dorsal horn, together with decreased TrkC⁺ projections (I–L). Boxes in I and J are magnified in K and L, respectively. Brackets in K and L indicate the thickness of dorsal horn layers with TrkA⁺ afferent innervations. (M–R) Nociceptive afferent fibers expressing the neuropeptide substance P are detected at the edge as well as in deep layers of the dorsal horn in Brn3a^+/− spinal cords; in Brn3a^−/− spinal cords, however, almost all of these fibers are located at the edge of the dorsal horn and absent from the deep layers (M–P). Boxes in M and N are magnified in O and P, respectively. Similar defects are seen for CGRP⁺ fibers (Q and R). All sections were counter-labeled with DAPI. Scale bar: A–D, G–J, M, and N, 200 µm; E, F, K, L, and O–R, 100 µm.

The defective central projection of TrkA⁺ nociceptive sensory afferents was also revealed by immunostaining using two additional markers. At E18.5, the neuropeptides SP (substance P)- and CGRP (calcitonin gene-related peptide)-expressing fibers project mainly to the lamina I and outer lamina II (II_o), and belong to “peptidergic” C- and Aδ fibers (Larsson, 2009; Snider and McMahon, 1998). These fibers are unmyelinated or thinly-myelinated and thought to play important roles in transmitting nociceptive information, such as acute and chronic pain sensations to the CNS (Willis, 1985). In Brn3a mutant spinal cords, we detected increased CGRP-immunoreactive fibers along the edge of the dorsal horn, and a marked reduction of them in deeper layers of the spinal cord (Figs. 3Q and R). Similarly, the SP-immunoreactive fibers were mostly accumulated along the dorsal edge and were almost eliminated from the deep layers of the dorsal spinal cord in Brn3a mutants (Figs. 3M–P). It has been shown previously that Brn3a inactivation causes a partial loss of SP⁺ cells in deep dorsal horn laminas (Xu et al., 2008), thereby contributing to the loss of SP⁺ fibers in the mutant spinal cord (Figs. 3N and P). However, the concentrated accumulation of SP⁺ fibers at the dorsal edge of the mutant cord clearly indicates a failure for them to extend into the deep layers (Figs. 3N and P). These results together thus suggest that the peptidergic-dependent circuit conducting a major class of nociceptive information does not form properly in Brn3a mutant mice.

The abnormalities of afferent projections in mutant spinal cords were confirmed by an independent tracing experiment. We applied the DiI lipophilic tracer dye to the level-matched thoracic DRGs in control and mutant animals to visualize the structural profiles of afferent axons in the spinal cord at E18.5. In Brn3a^+/+;Brn3b^+/− mice, we observed DiI-labeled sensory afferent axons projecting into the dorsal horn, intermediate zone and ventral horn of the spinal cord (Fig. 4A). In both Brn3a^−/−;Brn3b^+/+ and Brn3a^−/−;Brn3b^−/− mutant mice, there was a dramatic decrease of DiI-labeled afferent fibers projecting into the superficial and deep layers of the dorsal horn (Figs. 4A, C and D), most of which presumably represent cutaneous nociceptive sensory projections. In addition, we observed a reduction of proprioceptive projections to the ventral horn in the mutant (Figs. 4A, C and D). No obvious projection defects were observed in Brn3b mutant mice (Figs. 4A and B).

Fig. 4.

DiI labeling reveals impaired DRG sensory afferent projections into the spinal cord in Brn3a single and Brn3a;Brn3b double mutant spinal cords. (A–D) DiI-labeled DRG sensory afferent projections in the thoracic spinal cord of E18.5 embryos of the indicated genotypes. In Brn3a^+/+Brn3b^+/− and Brn3a^+/+Brn3b^−/− mice, DiI-labeled DRG sensory afferents extended robustly into the dorsal horn, and there were also labeled proprioceptive axon fibers projecting into the ventral horn (A and B). In Brn3a^−/−Brn3b^+/+ and Brn3a^−/−Brn3b^−/− mice, however, there was a drastic reduction of DiI-labeled afferent fibers in the dorsal horn and the proprioceptive axons that projecting toward the ventral horn were also greatly diminished (C and D). These projection patterns are indistinguishable between Brn3a^+/+Brn3b^+/− and Brn3a^+/+Brn3b^−/− (A and B), or Brn3a^−/−Brn3b^+/+ and Brn3a^−/−Brn3b^−/− (C and D) spinal cords. Scale bar: 200 µm.

Brn3a is required for segregation of TrkA/TrkB and TrkA/TrkC expression during DRG sensory neuron specification

The abnormal sensory projection patterns in the spinal cord of Brn3a single and Brn3a;Brn3b double mutant mice prompted us to examine the specification of sensory neurons in the DRG, where cell bodies of the primary afferents are located. Because there was a near complete loss of Brn3b expression in DRGs of Brn3a mutant embryos and there was no projection defect in Brn3b mutant mice, we analyzed only DRGs of Brn3a single mutant mice.

We initiated the investigation at E11.5, by immunohistochemistry using TrkA, TrkB and TrkC as molecular markers for DRG neuron subtypes. We counted the number of DRG neurons that were immunoreactive for these three markers as well as neurons co-expressing TrkA/TrkB and TrkA/TrkC double markers, and then determined the number of immunoreactive cells in one mm² cross-sectional area. At this stage, there was a 67% increase in the number of TrkA⁺ neurons in Brn3a^−/− compared to control DRGs (Brn3a^−/−, 2807±459/mm², n=3; control, 1679±285/mm², n=3). There was also an increase of TrkB⁺ cells (Brn3a^−/−, 4953±694/mm², n=3; control, 3400±354/mm², n=3) but a significant decrease of TrkC⁺ cells (Brn3a^−/−, 2256±384/mm², n=3; control, 4125±451/mm², n=3) (Fig. 5). In addition, we noticed that there were significantly more TrkA⁺/TrkB⁺ double positive cells in the mutant DRG (Brn3a^−/−, 1745±218/mm², n=3; control, 525±99/mm², n=3) (Figs. 5A, B and E). These data suggest that early specification of DRG sensory neurons may be defective in Brn3a mutant embryos, with abnormally increased number of TrkA⁺, TrkB⁺ and TrkA⁺/TrkB⁺ cells, and a decreased number of TrkC⁺ cells.

Fig. 5.

Early specification defects of DRG sensory neuron subtypes in Brn3a mutant mice. (A–D) Transverse DRG sections of similar forelimb levels from E11.5 Brn3a^+/− and Brn3a^−/− embryos were examined for the expression of TrkA, TrkB and TrkC by immunostaining. TrkA and TrkB, as well as TrkA and TrkC, are co-expressed by a small population of neurons in Brn3a^+/− DRGs (A and C). The number of TrkA expressing neurons is increased in Brn3a mutant DRGs (A–D), accompanied by an increase of TrkA/TrkB co-expressing neurons (A and B) and a decrease of TrkC-expressing neurons (C and D). (E) Quantification of TrkA⁺, TrkB⁺, TrkC⁺, TrkA⁺/B⁺, and TrkA⁺/C⁺ neurons in control and mutant DRGs. Control includes both Brn3a^+/+ and Brn3a^+/− DRGs, which are indistinguishable in neuron numbers. Data are shown as mean±s.d.; *p<0.05, **p<0.01, t test. Scale bar: 50 µm.

During normal development, the cells that express TrkA, TrkB or TrkC in the DRG become segregated by E12.5, with few cells co-expressing them with each other (Gross et al., 2000; Kramer et al., 2006). Accordingly, we examined the segregation of these proteins in Brn3a mutant DRGs at this time. We first assessed the total neuron number for level-matched thoracic DRGs from the mutant and wild-type, by quantifying Isl1-immunoreactive cells from serial sections. We did not observe any significant change in the total number of Isl1⁺ neurons in the mutant DRG (Fig. 6J). Similarly, there was no significant change in the number of TrkA⁺ neurons in Brn3a mutant DRGs (Fig. 6I). However, the number of TrkB⁺ cells increased by 2.6-fold (Brn3a^−/−, 2070±108/mm², n=3; control, 792±90/mm², n=3), and the number of TrkC⁺ cells also increased (Brn3a^−/−, 1089±44/mm², n=3; control, 846±61/mm², n=3). Notably, the number of TrkA⁺/TrkB⁺ (Brn3a^−/−, 474±27/mm², n=3; control, 72±6/mm², n=3), and TrkA⁺/TrkC⁺ (Brn3a^−/−, 680±36/mm², n=3; control, 94±15/mm², n=3) hybrid cells increased in the mutant by 6.6- and 7.2-fold, respectively (Figs. 6A–I). The increase of TrkA⁺/TrkC⁺ hybrid cells is consistent with the extensive overlap of TrkA⁺ and TrkC⁺ afferent fibers and the more medial location of TrkA⁺ fibers at the DREZ of Brn3a mutant spinal cords (Fig. 2F).

Fig. 6.

The expression of TrkA/TrkB and TrkA/TrkC in DRG sensory neurons are not properly segregated in Brn3a mutant mice at E12.5. (A–H) Forelimb-level transverse DRG sections from E12.5 Brn3a^+/− and Brn3a^−/− mice were immunostained with indicated antibodies. Sections in (A, B, E, and F) were counter-labeled with Topro-3. In Brn3a^−/− DRGs, TrkB⁺ and TrkC⁺ cells significantly increase (A, B, E, and F), so do the smaller-sized TrkC⁺ cells in particular (E and F). There is also a significant increase in the number of TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ co-expressing cells in Brn3a^−/− DRGs (C, D, G, and H). Arrow heads in D and H point to representative co-expressing cells and those in F point to representative smaller-sized TrkC⁺ cells. (I and J) Quantification of sensory neuron subtypes in control and Brn3a mutant DRGs. Total neuron number is not changed in Brn3a mutant DRGs at E11.5, E12.5 and E18.5, as measured by unchanged Isl1⁺ cells (J). The numbers of TrkB⁺, TrkC⁺, TrkA⁺/B⁺, and TrkA⁺/C⁺ neurons are all increased but TrkA⁺ cells remain unchanged in E12.5 Brn3a^−/− DRGs (I). (K and L) Distribution of cell cross-sectional areas of TrkB⁺ and TrkC⁺ neurons in E12.5 control and Brn3a^−/− DRGs. Areas were measured by the NIH Image J software and data were plotted using Microsoft Excel. In Brn3a^−/− DRGs, the peak of TrkC⁺ neuron sizes has shifted to the left compared to the control (L). There is no obvious change in the distribution of TrkB⁺ neuron sizes in Brn3a^−/− DRGs (K). Control includes both Brn3a^+/+ and Brn3a^+/− DRGs, which are indistinguishable in neuron number and size. Data are shown as mean±s.d.; **p<0.01, ***p<0.001, t test. Scale bar: 75 µm.

Aside from the increased Trk⁺ hybrid cells, we also noted an increase of smaller-sized TrkC⁺ cells in Brn3a mutant DRGs (Figs. 6E–H). Trk receptors are usually expressed in distinct populations of sensory neurons with different sizes. TrkA is expressed predominantly by small-diameter neurons, whereas TrkC is expressed in large-diameter neurons (McMahon et al., 1994; Mu et al., 1993; Wright and Snider, 1995). To quantify the observed changes in neuron size, we measured the cross-sectional areas for TrkB⁺ and TrkC⁺ cells from both control and mutant DRGs. We found that in Brn3a^+/+ and Brn3a^+/− DRGs, the TrkC⁺ cells had areas between 40–160 µm², with a peak at ~110 µm². In Brn3a^−/− DRGs, although the areas of TrkC⁺ cells were distributed in a similar range, there was a significant increase in TrkC⁺ cells with smaller sizes, peaking at ~60 µm² (Fig. 6L). The increased population of small-sized TrkC⁺ cells was mostly TrkA⁺/TrkC⁺ hybrid cells, which could be distinguished by immunostaining (Figs. 6G and H). We did not detect any obvious change in the size of TrkB⁺ cells in mutant DRGs (Figs. 6A–D and K). Thus, Brn3a inactivation prevents the proper segregation of TrkA/TrkB and TrkA/TrkC expression during DRG sensory neuron specification, with the abnormal generation of TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ hybrid cells that may cause aberrant sensory afferent projection to the mutant spinal cord.

Despite the increase of TrkB⁺ and TrkC⁺ neurons in Brn3a^−/− DRGs at E12.5, by E15.5, however, there were much fewer TrkC⁺ cells in the mutant DRG (Brn3a^−/−, 288±22/mm², n=3; control, 645±133/mm², n=3) and the number of TrkB⁺ or TrkA⁺ cells was the same between the mutant and control (Figs. 7D). Moreover, compared to E12.5, the number of TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ cells decreased by 6- and 19-fold, respectively, in the mutant DRG at this stage (Fig. 7D compared to Fig. 6I). This indicates that in the Brn3a mutant DRG, the anomalously formed TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ hybrid cells fail to maintain themselves over time. Indeed, there was a significant increase of active caspase 3- immunoreactive apoptotic cells in Brn3a mutant DRGs at E15.5 (Brn3a^−/−, 146±19/mm², n=3; control, 97±21/mm², n=3) (Figs. 7A and D), suggesting that the reduced cell number may in part result from increased apoptotic cell death in the mutant DRG.

Fig. 7.

Changes of DRG sensory neuron types in Brn3a mutant at E15.5 and E18.5. (A–C) Level-matched thoracic transverse DRG sections from control and Brn3a mutant embryos at E15.5 and E18.5 were immunostained with indicated antibodies. At E15.5, there are more active-caspase3⁺ cells in Brn3a^−/− DRGs (A). At E18.5 in Brn3a^−/− DRGs, the numbers of TrkB⁺, TrkC⁺ and parvalbumin⁺ cells are decreased, but the numbers of c-ret⁺, CGRP⁺ and c-ret⁺/CGRP⁺ cells are all increased (B and C). There is no difference in TrkA⁺ cell number between the control and mutant (B). (D–F) Quantification of sensory neurons expressing various markers in control and mutant DRGs at E15.5 (D) and E18.5 (E and F). Control includes both Brn3a^+/+ and Brn3a^+/− DRGs, which are indistinguishable in neuron numbers. Data are shown as mean±s.d.; *p<0.05, **p<0.01, ***p<0.001, t test. Scale bar: 100 µm.

Alteration of sensory neuron subtypes in the Brn3a mutant DRG

At E18.5 when the central afferent projection defects were most obvious in Brn3a^−/− spinal cords, we analyzed the differentiation of several sensory neuron subtypes in the mutant DRG using markers selectively expressed in nociceptive and proprioceptive sensory neurons. We determined the total number of neurons in the DRG at this stage by quantifying Isl1-immunoreactive cells from serial sections, and found no significant difference between Brn3a mutants and controls (Fig. 6J). However, the number of cells expressing proprioceptive proteins TrkC or parvalbumin was significantly decreased (TrkC: Brn3a^−/−, 267±51/mm², n=3; control, 396±6/mm², n=3; parvalbumin: Brn3a^−/−, 138±6/mm², n=3; control, 277±28/mm², n=3), and so was the number of neurons expressing the mechanoreceptive marker TrkB (Figs. 7B, C and E). On the other hand, the number of sensory neurons expressing the nociceptive marker TrkA remained unchanged in Brn3a mutant DRGs (Brn3a^−/−, 3544±221/mm², n=3; control, 3475±90/mm², n=3) (Fig. 7E), although the fluorescence intensity of TrkA labeling was generally weaker in the mutant.

CGRP and SP are two neuropeptides expressed in subpopulations of nociceptive neurons, distinguishing them from non-peptidergic nociceptive neurons that express c-ret (Molliver et al., 1997; Scott, 1992). Although the number of cells expressing SP was unchanged in the Brn3a mutant DRG at E18.5, we found a significant increase of CGRP⁺ (Brn3a^−/−, 1203±101/mm², n=3; control, 542±70/mm², n=3), c-ret⁺ (Brn3a^−/−, 2788±258/mm², n=3; control, 1675±215/mm², n=3), and CGRP⁺/c-ret⁺ cells (Brn3a^−/−, 673±57/mm², n=3; control, 221±33/mm², n=3) (Figs. 7C, E and F). In spite of the normal number, the SP-immunoreactive neurons failed to differentiate properly in the mutant since their afferent fibers were unable to project into the deep layers of the dorsal horn in Brn3a mutant spinal cords (Figs. 3M–P). These results suggest that in the absence of Brn3a, the specification and differentiation of both nociceptive and proprioceptive sensory neuron subtypes are severely affected in the DRG, with a decrease of neurons expressing proprioceptive markers and an increase of neurons expressing several peptidergic and non-peptidergic neuron markers.

Reduced Runx1 expression in the Brn3a mutant DRG

The Runt domain transcription factors Runx1 and Runx3 have been shown to play a critical role in regulating the specification of proprioceptive and nociceptive neurons and their projections (Chen et al., 2006a; Chen et al., 2006b; Inoue et al., 2002; Kramer et al., 2006; Marmigere et al., 2006; Yoshikawa et al., 2007). Given the extensive defects in proprioceptive and nociceptive neuron subtype specification and afferent projection in Brn3a null mutant embryos, we examined by immunostaining the expression of Runx1 and Runx3 proteins in Brn3a^−/− DRGs. At E13.5 and E15.5, there was a dramatic decrease of Runx1-immunoreactive cells in Brn3a^−/− DRGs compared to the control (Figs. 8A, B, E, and F); however, there was hardly any change in the number of Runx3-immunoreactive cells (Figs. 8C, D, G, and H). Thus, Brn3a appears to positively regulate Runx1 gene expression and Runx1 may in part mediate Brn3a function during DRG development.

Fig. 8.

Expression of Runx1 and Runx3 in Brn3a mutant DRGs at E13.5 and E15.5. (A–H) Level-matched thoracic transverse DRG sections from wild-type and Brn3a mutant embryos were immunostained with antibodies against Runx1 and Runx3 at E13.5 (A–D) and E15.5 (E–H). Sections were also counter-labeled with DAPI (blue). At both stages, there is a great loss of Runx1⁺ cells in Brn3a^−/− DRGs (A, B, E, and F). By contrast, there is no obvious difference in the number of Runx3⁺ cells between the wild-type and mutant at both stages (C, D, G, and H). Scale bar: A–D, 50 µm; E–H, 100 µm.

Normal early development of dorsal interneurons in Brn3a and Brn3b single and double mutant spinal cords

Given the expression of Brn3 genes in the dorsal spinal cord (Fedtsova and Turner, 1995; Ninkina et al., 1993; Turner et al., 1994), there is a possibility that inactivating Brn3a and Brn3b may result in improper differentiation of dorsal interneurons in the spinal cord, which in turn may cause sensory afferent projection defects indirectly. To determine whether Brn3 factors play such a role, we first investigated the spatial and temporal expression patterns of all three Brn3 proteins during mouse spinal cord development. At E10.5–11.5, anti-Brn3a and anti-Brn3b antibodies immunostained cells at the lateral margin but not within the ventricular zone in the dorsal spinal cord, whereas anti-Brn3c did not detect any positive cells (Figs. S3A–E). Double-immunostaining of spinal cord sections using anit-Brn3 antibodies and those specific to distinct interneuron subtypes indicated that Brn3a is expressed in dI1–dI3 and dI5 interneurons and Brn3b in dI2, dI3 and dI5 neurons (Figs. S3F–I and S4). Immunostaining with anti-Brn3a and anti-Brn3b antibodies combined with EdU labeling showed that Brn3a and Brn3b are exclusively expressed by post-mitotic neurons in the spinal cord at these stages (Figs. 1O and P; and data not shown).

Immunostaining with late-born neuron markers at E12.5 showed that Brn3a-expressing cells had a mutually exclusive pattern with respect to Pax2-expressing, dIL^A cells (Fig. S5A). In contrast, Brn3a was co-expressed with Lbx1 in some cells of the subventricular zone and more lateral regions of the dorsal spinal cord (Fig. S5C). Brn3b-expressing cells at this stage mainly resided in the region lateral to the subventricular zone. Many of them co-expressed Brn3a, and shared similar co-localization patterns with Pax2 and Lbx1 as those expressing Brn3a (Figs. S5B, D and I). At E15.5 when neurogenesis in the dorsal spinal cord has ceased, Brn3a- and Brn3b-expressing cells were located mainly in the deeper layers of the dorsal horn (Figs. S5E–H and J). A small number of Brn3a-expressing cells co-expressed Lbx1 whereas there were rarely any cells co-expressing Brn3b and Lbx1 (Figs. S5G and H). Compared to E10.5 and E12.5, there were fewer Brn3a and Brn3b co-expressing cells by E15.5, and the number of Brn3b-expressing cells was significantly less than that of Brn3a⁺ cells (Figs. S5I–K). These observations suggest that Brn3a and Brn3b are likely expressed in some late-born dIL^B interneurons, which also express Lbx1, but not in dIL^A neurons.

By immunostaining or RNA in-situ hybridization, we examined the expression of dorsal interneuron markers in Brn3a and Brn3b single and double mutant spinal cords. At E10.5 and E11.5, the expression of Isl1, Lhx1/5, Pax2, Lmx1b, and Tlx3 was grossly normal in terms of their patterns and cell numbers in both the single and double mutants compared to the control (Figs. 9, S6 and S7). Therefore, early neurogenesis and initial differentiation of dorsal interneurons appear to occur properly without any obvious defect in Brn3a and Brn3b single and double mutants. However, Brn3a has been shown to act downstream of Tlx1/3 to specify a small population of early-born SP⁺ neurons (Xu et al., 2008), suggesting a role for Brn3 factors in subtype specification and/or terminal differentiation.

Fig. 9.

Subtype specification of dorsal spinal cord interneurons in Brn3a and Brn3b single and double mutant embryos at E11.5. (A–H) Forelimb-level spinal cord sections of the indicated genotypes were co-immunostained with antibodies against Pax2 and Isl1 (A, C, E, and G) or Pax2 and Lhx1/5 (B, D, F, and H). No obvious difference was detected in the number of cells expressing these dorsal interneuron markers or in expression patterns of these markers between different genotypes. Sections were counter-labeled with DAPI (blue). Scale bar: 100 µm.

Discussion

In this study, we performed detailed analysis of DRG sensory neuron development in Brn3a and Brn3b single and double mutant mice. In Brn3a mutants, the expression of TrkA fails to segregate from TrkB and TrkC expression in DRG sensory neurons at early stages, and the specification of major nociceptive and proprioceptive sensory neuron subtypes is altered subsequently. Meanwhile, sensory afferent projections in the spinal cord are severely impaired. Nociceptive projections are largely devoid from designated layers in the dorsal horn and proprioceptive projections do not reach the ventral horn in the Brn3a mutant spinal cord. We have further shown that the expression of Runx1 is positively regulated by Brn3a during DRG sensory development, suggesting that it may be one of the effector genes mediating the Brn3a function. Our data have thus revealed an essential role for Brn3a in regulating the specification of DRG sensory neurons and their projections to the spinal cord.

Requirement for Brn3a in DRG sensory neuron subtype specification

During sensory neurogenesis, sensory neurons that belong to distinct functional modalities express the neurotrophin receptors TrkA, TrkB and TrkC (Anderson, 1999; Marmigere and Ernfors, 2007; Woolf and Ma, 2007). In the early stages of their expression, co-expression of TrkB and TrkC in a fraction of DRG sensory neurons (Farinas et al., 1998; Kramer et al., 2006), and similar co-expression of two or more of those markers in TG sensory neuron precursors (Dykes et al., 2010; Turner et al., 1994) have been observed. Subsequently, the fraction of hybrid cell populations decrease as sensory neurons adopt distinct identities as nociceptive, mechanoreceptive, and proprioceptive neurons. Brn3a is expressed strongly in most or all DRG neurons as early as E9.5 and appears to co-express in all TrkA-, TrkB- or TrkC-positive neurons during mouse embryogenesis (Fig. 1) (Huang et al., 1999). Thus, the timing of Brn3a expression coincides with the emergence of Trk⁺ neuron subtypes and implicates a role for Brn3a in sensory neuron subtype specification. This in fact is the case as we observed extensive alteration of Trk⁺ sensory neuron complements and their central projections in Brn3a^−/− DRGs. By contrast, Brn3b and Brn3c are expressed only weakly in subsets of DRG cells during the transition from neurogenesis to differentiation, and more importantly, the initiation and/or maintenance of their expression is dependent on Brn3a (Fig. 1). Thus, Brn3b and Brn3c might be dispensable for DRG development if their functions are completely redundant with that of Brn3a. This indeed seems to be the case as Brn3b single mutants do not have any obvious DRG sensory neuron defect and Brn3a and Brn3b compound mutants fail to exhibit a more severe phenotype than the Brn3a single mutant. It has been shown previously that an earlier onset of expression determines whether Brn3a or Brn3b plays a dominant role during development of the inner ear sensory ganglia and retinal ganglion cells (Huang et al., 2001; Xiang, 1998).

Our study shows that Brn3a is required by early DRG sensory neuron precursors for segregating TrkA/TrkB and TrkA/TrkC expression. The excessive TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ hybrid cell populations in Brn3a mutants are first seen at E11.5, at the beginning when all three neurotrophin receptor markers can be detected. By E12.5, there is a more than 6-fold increase in the number of TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ neurons. In addition, there is a marked increase of smaller-sized TrkC⁺ cell population in the Brn3a mutant DRG. All these changes indicate a severe defect in early DRG sensory neuron specification in the absence of Brn3a.

The slightly increased apoptotic cell number in Brn3a mutant DRGs at E15.5 suggests that a certain number of mis-specified sensory neurons may not have survived by this stage. Consistent with this, we did observe a persistent decrease of TrkB⁺ and TrkC⁺ cells from E12.5 to E18.5, following the transient increase from E11.5 to E12.5. At the same time, we did not find any significant change in the number of TrkA⁺ cells by E18.5, although the overall expression level of TrkA in the mutant seems decreased, as indicated by weakened TrkA immunofluorescence intensity. In agreement with our observation, in a previous study (McEvilly et al., 1996), the expression levels of TrkA, TrkB and TrkC mRNA were all found to decrease at P0 in Brn3a mutant DRGs. Interestingly, there appears to be no significant change in the number of TrkA⁺ neurons in the mutant despite reduced TrkA expression at both RNA and protein levels.

By E18.5, the maturation of nociceptive sensory neurons is marked by the expression of neuropeptides (Hippenmeyer et al., 2004; Oakley et al., 2000), notably CGRP, SP, etc.. Some TrkA⁺ DRG neurons switch from nerve growth factor (NGF)-dependent TrkA⁺ neurons to glial cell-derived neurotrophic factor (GDNF)-dependent c-ret⁺ neurons (TrkA⁻/c-ret⁺), while other TrkA⁺ neurons induce CGRP expression (TrkA⁺/CGRP⁺/c-ret⁻) (Molliver et al., 1997). Our data indicate that Brn3a is also involved in regulating this cell fate switch during late embryonic and perinatal stages. In the absence of Brn3a, the number of CGRP⁺ and c-ret⁺ neurons, which are peptidergic and non-peptidergic, respectively, are increased, accompanied by an increase of CGRP⁺/c-ret⁺ double-positive neurons. This suggests that rather than selectively determining one cell fate over the other, Brn3a may function in those TrkA⁺ neurons to suppress precocious expression of both CGRP and c-ret. The increased CGRP⁺ and c-ret⁺ neurons and perhaps other subtypes may compensate for the loss of TrkB⁺ and TrkC⁺ neurons due to apoptosis, thereby explaining the normal number of Isl1⁺ and TrkA⁺ cells in E18.5 Brn3a mutant DRGs.

Brn3a regulates central projections of DRG sensory afferents into the spinal cord

Shortly after we first detected defective sensory neuron specification in Brn3a mutant DRGs at E11.5, we noticed a marked central projection defect for TrkA⁺ and TrkC⁺ afferent fibers. The boundary of distinct projection patterns for TrkA⁺ and TrkC⁺ fibers is lost in Brn3a mutant spinal cords by E12.5, when the most significant sensory neuron mis-specification is observed in the mutant DRG. Neurotrophin receptors have long been thought to be involved in sensory neuron survival, differentiation, as well as axonal growth to both central and peripheral targets (Oakley et al., 2000; Patel et al., 2000; Patel et al., 2003; Tucker et al., 2001). Therefore, it is important for different DRG sensory neurons that belong to distinct functional modalities to express respective neurotrophin receptors. In a previous study, Moqrich et al. (Moqrich et al., 2004) demonstrated that by expressing rat TrkC from the TrkA locus in mice, the surviving presumptive TrkA-expressing neurons adopted a proprioceptive phenotype, including an adoption of proprioceptive central projection pattern. Thus it is likely that the altered central projection patterns for TrkA⁺ and TrkC⁺ axons in Brn3a mutant spinal cords may at least in part result from the aberrant co-expression of TrkA and TrkC receptors in a certain population of DRG sensory neurons.

At late embryonic stages, the central projection defects of Brn3a mutants are highlighted by a near complete loss of nociceptive afferent penetration into the spinal cord, as being demonstrated by both immunostaining of molecular markers and an independent DiI tracing assessment. Whether this phenotype results from a secondary effect of early projection defect observed at E12.5 or from other roles of Brn3a in regulating axonal growth is still unclear. Interestingly, it has been shown that in mouse embryos deficient for Brn3a, trigeminal axons fail to correctly innervate their peripheral targets, and there are abnormal branches of the trigeminal nerve (Eng et al., 2001). Here we demonstrate a profound defect in central projections of DRG sensory neurons in Brn3a mutants: a dramatic loss of nociceptive innervations in the dorsal gray matter of the spinal cord and a modest decrease of proprioceptive projections. This suggests somatosensory and motor dysfunction in the mutant animal, and may explain certain behavioral phenotypes of new-born Brn3a mutants, such as aberrant limb and trunk movements and impaired suckling (Xiang et al., 1996).

Our present work and that of McEvilly et al. (McEvilly et al., 1996) have both demonstrated that the severity of central projection defects of Brn3a mutant DRG neurons depends on the axial position, with the most severe defects observed in the cervical region. However, a normal pattern of spinal innervation was reported by McEvilly et al. (McEvilly et al., 1996) for the lumbar region in the mutant based on DiI tracing at P0, whereas our marker labeling clearly indicated an abnormal one at the same region. It is unclear what causes this discrepancy but it could result from different strain background and mutant allele used in these two studies.

Distinct roles of Brn3a in regulating sensory neuron specification in the DRG and TG

Several studies indicate that Brn3a controls survival and differentiation of trigeminal neurons by regulating expression of each of the three Trk receptors (Dykes et al., 2010; Huang et al., 1999; Lei et al., 2006; Ma et al., 2003). In particular, Brn3a is reported to bind to a minimal enhancer that regulates TrkA expression in embryonic trigeminal sensory neurons, suggesting that TrkA may be a direct transcriptional target of Brn3a in vivo (Lei and Parada, 2007; Ma et al., 2003). However, there is no strong evidence to support the same role for Brn3a in DRG sensory neurons. We were unable to detect any decrease of TrkA⁺ cells in Brn3a mutant DRGs at E18.5 by immunostaining, although the expression level seemed to reduce slightly judged by somewhat weakened fluorescence intensity (data not shown). All three Trk receptors are detected in the perinatal Brn3a mutant DRG, in contrast to a complete absence of Trk receptors in E17.5 and P0 mutant TGs (Huang et al., 1999). Instead of a progressive loss of TrkA⁺ neurons and a complete absence of TrkC expression in the TG of Brn3a mutant mice, there is a transient increase of DRG sensory neurons expressing both markers from E11.5 to E12.5 before they are down-regulated in the mutant. On the other hand, there is a transient increase of TrkB-expressing neurons in both the DRGs and TGs of Brn3a mutants (Huang et al., 1999), highlighting overlapping as well as distinct functions of Brn3a in DRG and TG sensory neuron development.

In sensory ganglia, Runx1 and Runx3 are suggested to repress TrkB expression but promote the expression of TrkA and TrkC in nociceptors and proprioceptors, respectively (Inoue et al., 2007; Kramer et al., 2006; Levanon et al., 2002b). A recent study indicates that in the Brn3a mutant TG, the expression of Runx1 is markedly diminished while the expression of Runx3 is not initiated by E12.5 and nearly absent at later stages (Dykes et al., 2010). Thus, Brn3a mutant TG neurons exhibit abnormal persistence of the co-expression of TrkA/TrkB and TrkB/TrkC at E12.5 (Dykes et al., 2010). We observed in Brn3a mutant DRGs a similar defect in segregation of TrkA⁺/TrkB⁺ and TrkA⁺/TrkC⁺ neurons. However, despite a severe downregulation of Runx1 expression similar to that in Brn3a mutant TGs, Runx3 expression appears unchanged in Brn3a mutant DRGs. Thus, Brn3a may control distinct molecules and pathways to discriminate TrkA, TrkB and TrkC expression in DRG neurons and the role of Brn3a during sensory neuron specification may be context-dependent. Indeed, the Runx3-dependent TrkC⁺ neurons in the TG are mechanoreceptive and derived from the trigeminal ectodermal placode (Lazarov, 2002; Senzaki et al., 2010). Such equivalent neurons may not exist in the neural-crest-derived DRG. On the other hand, the Runx3-dependent TrkC⁺ proprioceptive neurons in the DRG are unlikely to be present in the TG since the proprioceptive neurons of the trigeminal system are located in the mesencephalic trigeminal nucleus (Inoue et al., 2002; Lazarov, 2002; Levanon et al., 2002a; Nakamura et al., 2008).

The function of Brn3a in DRG sensory neuron specification appears to be mediated only in small part by Runx1. In both Brn3a and Runx1 mutant DRGs, there is a significant increase of TrkA⁺, TrkC⁺ and CGRP⁺ sensory neurons (Figs. 5–7) (Chen et al., 2006b; Yoshikawa et al., 2007). However, unlike in Runx1 mutant DRGs, the increased TrkA⁺ and TrkC⁺ neurons fail to maintain until late embryonic stages in Brn3a mutant DRGs. Moreover, at E17.5/E18.5, c-ret⁺ neurons significantly increase in Brn3a mutant DRGs while greatly decrease in Runx1 mutant DRGs; TrkB⁺, TrkC⁺ and parvalbumin⁺ neurons all decrease in Brn3a mutant DRGs but are either unchanged or increase in Runx1 mutant DRGs (Yoshikawa et al., 2007). Furthermore, Brn3a mutants display a much more severe defect in the DRG sensory afferent projection pattern (Figs. 2–4) (Chen et al., 2006b; Yoshikawa et al., 2007). Undoubtedly, additional key regulators remain to be identified that together with Runx1 may mediate the whole range of Brn3a functions in the specification of DRG sensory neuron subtypes.

Highlights.

Brn3a inactivation causes improper segregation of Trk receptor expression in the DRG.

Major sensory neuron subtypes are altered in the Brn3a mutant DRG.

Brn3a regulates sensory neuron specification in part through Runx1.

DRG central projections in the spinal cord are defective in Brn3a mutants.