Brn3c null mutants at birth (P0)

Vestibular ganglia are smaller in Brn3c null mutants (Fig. 1b) than in control littermates (Fig. 1a), but larger than in BDNF or trkB null mutants of the same age [13, 15]. The reduction in apparent size of the vestibular ganglia is in agreement with quantitative data published previously [3, 13]. These data suggest a loss of 80–85% of vestibular sensory neurons in BDNF and trkB null mutants [13] and of 77% of vestibular sensory neurons in P4 Brn3c null mutants [3]. Thus, the size reduction in the vestibular sensory ganglion could be compatible with a loss of BDNF production in the immature hair cells.

However, the Brn3c null mutants show only a reduced density of afferent and efferent fibers to all vestibular sensory epithelia. There is no specific loss of all afferent and efferent innervation to the canal crista (Fig. 2a), a hallmark of both BDNF and trkB null mutations [13, 15]. In fact, the reduction of fibers seems to be rather uniform throughout a given sensory epithelium with the crista innervation being qualitatively no more reduced than the innervation of the utricle and saccule. No loss in specific areas, comparable to that in BDNF null mutant mice, is apparent in the saccule or utricle of Brn3c null mice. Similar patterns of innervation are obtained using acteylated tubulin immunocytochemistry (Fig. 2b).

Consistent with the finding of Xiang et al. [3] of only a 29% loss of spiral sensory neurons at P4, our data show little difference in the pattern of innervation of the cochlea in P0 Brn3c null mutants (Fig. 2c,2d). No selective loss of spiral neurons is observed in Brn3c null mutants in the basal turn, a feature of either NT-3 or trkC loss [6, 9, 14]. Likewise, the innervation of the apex (Fig. 2f) shows no detectable abnormality in overall pattern of innervation compared to control animals (data not shown), an indication that BDNF could be expressed in the apex [6]. In addition, the pattern of efferent innervation shows no deviation from normal either (Fig. 2e), whereas they show the same pattern of loss as afferent fibers in neurotrophin null mutants [16]. These data suggest that the spiral sensory neurons develop qualitatively normal at least until P0 and therefore allow normal pathfinding of efferents. Most interestingly, there is no increase in radial fiber spacing in the apex, a specific problem of BDNF null mutants [6, 13].

However, there is one qualitative difference not recognized in any single neurotrophin null mutant. Afferents reach all three rows of outer hair cells in the basal turn of control wildtype littermates (Fig. 2c), but both afferent and efferent outgrowth is disorganized to outer hair cells in Brn3c null mutants (Fig. 2d) and does not show any clear organization into three distinct longitudinal fiber bundles paralleling the three rows in the outer hair cell region. These data suggest that fiber organization in the outer hair cell region is partly disrupted in Brn3c null mutant.

Brn3c null mutants at P7/8

Reduction of vestibular innervation is more pronounced than at P0. Loss of afferents and, even more clearly of efferents, is most pronounced in the utricle (Fig. 3a,3b) and saccule (data not shown). In contrast, the innervation to the canal cristae, while reduced, has changed little from P0. In particular the horizontal canal receives a rather dense afferent and efferent innervation in Brn3c null mutants.

The cochlea at P8 shows more qualitative deviations from the normal pattern of innervation. For example, all three rows of outer hair cells now receive both afferent and efferent innervation in the base of control littermates (Fig. 3c). However, no fibers extend to the outer hair cell region in Brn3c null mutants (Figs. 3d,3e,3f). In control animals there is dense innervation of inner hair cells, whereas Brn3c null mutants show a curious aggregation of fibers near focal spots around the habenula perforata. The apex shows less pronounced deviations from normal (Fig. 3f,3g). However, as in the base, afferents and efferents extend sparingly to the outer hair cell region. There is formation of an inner spiral bundle communicating both afferents and efferents between the focal points where fibers appear to pass through the habenula perforata (Fig. 3f,3g). As in newborn animals, efferent fibers closely follow the pattern of innervation displayed by afferents (data not shown).

The limited qualitative effect of Brn3c null mutations on fiber loss is particularly obvious in the P8 Brn3c null mutant apical turn (Fig. 3g). Immunocytochemical and histochemcial data show the presence of only a few immature hair cells in the cochlea (Fig. 4), which amounted to about 1% (Table 1). Not a single immature hair cell forms in the apex of the cochlea. In contrast, more immature hair cells are found in the appropriate position of hair cells in the vestibular endorgans (Table 1). In fact, the vestibular end organs show no overt morphological defect other than the reduced presence of these immature hair cells. Nerve fibers can be traced into the sensory epithelium of P0 and P8 Brn3c null mutants past the habenula perforata and there is a normal location of spiral sensory neurons in Rosenthal's canal in the apex of P8 Brn3c null mutants (Fig. 3d,3g). The cochlea shows other defects as well. For example, the spiral capillary is not found underneath the organ of Corti but in the spiral limbus nearby (Fig. 4). The lack of immature hair cells seems to affect also the formation of pillar cells (undifferentiated almost throughout the cochlea; Fig. 4) and Deiters' cells (not recognizable based on their morphology), a suggestion consistent with recent molecular data on the interaction of hair cells and supporting cells [17, 18].

Comparison of numbers of immature hair cells in the vestibular and cochlear areas (Fig. 4) with the numbers of surviving vestibular and cochlear sensory neurons does not show any correlation (Table 1; Fig. 5). Only a few undifferentiated hair cells were detected by MyoVIIa immunostaining and only in the basal region of the cochlea. And yet only about 50% of cochlear sensory neurons are lost overall with a less severe loss in the apex in which we could not detect any immature hair cells with MyoVIIa immunocytochemistry. This does not exclude the possibility that even less differentiated hair cells may form in the apex.

The central projection of DiI and DiA traces fibers to the cochlear nuclei (Fig. 6) where they show a clear segregation of apical turn afferents (Fig. 6, green) and basal turn afferents (Fig. 6, red) in both the entering cochlear nerve (Fig. 6a) and the ventral cochlear nucleus (VCN) (Fig. 6b). Qualitative comparable central projections are found in wildtype control littermates (data not shown) and suggest that the central projection to cochlear nuclei is at least until P8 in mice not critically dependent on auditory information or other hair cell-mediated influences, in agreement with recent suggestions [5, 19].

6 months Brn3c null mutants

Very few fibers (both afferent and efferent) to the vestibular epithelia remain (Fig. 7a), thus displaying a further reduction of the pattern of innervation found in P8 Brn3c null mutants. Notably, there are still a number of fibers extending from the very few vestibular sensory neurons to the canal epithelia, in particular the horizontal canal (Fig. 7a,7b). Only very few fibers extend to the utricle and saccule with the reduction of efferents to these epithelia being almost complete (Fig. 7a and data not shown). Comparable to control mice, there is some transcellular labeling with DiI to cells apparently contacted by the afferent fibers in the Brn3c null mutant utricle (Fig. 7b). However, these cells could not be clearly identified as immature hair cells based on their morphology.

The cochlea, most prominently the apex, receives many afferent fibers, which form an interrupted inner spiral bundle (Fig. 7c). Some of the radial fiber bundles are myelinated (Fig. 7d). Only an occasional fiber extends beyond the inner spiral bundle into the area of the outer hair cells (Fig. 7d). Even at this late stage in maturation there are still numerous spiral neurons present in the apex (Fig. 7c,7d).

In summary, these tracing data suggest that mature hair cells are not necessary to direct fiber outgrowth and to maintain some fibers at least up to 6 months in the apex of the cochlea. Importantly, the spatio-temporal loss of innervation does not follow the pattern of loss known from neurotrophin and neurotrophin receptor null mutations as it starts in Brn3c mutant mice after birth.

In situhybridization confirms neonatal neurotrophin expression

We processed four ears of P0 and P8 Brn3c null and control littermates for expression of BDNF and NT-3 using the riboprobes previously described [7]. We did not analyze neurotrophin expression in 6-month old animals as neurotrophins are downregulated in neonates [20] and no expression can be found even in normal animals [4]. In the canal cristae we expected, and found, only a BDNF signal. This signal was very strong over hair cells of the P0 control animals (Fig. 8b). In contrast, only a faint signal was detected in the Brn3c null mutant (Fig. 8a) and the few silver grains were located predominantly over the area of the immature hair cells. The BDNF signal in the utricle was clearly centered on the hair cells in the control animal but some aggregation of silver grains appeared also in the Brn3c null mutants (data not shown).

In the saccule, we directly compared the BDNF (Fig. 8c,8d) and NT-3 signal (Fig. 8e,8f) of the Brn3c null mutant (Fig. 8c,8e) and control littermates (Fig. 8d,8f). A strong BDNF signal was found over the hair cells of control animals and an occasional immature hair cell also showed a BDNF signal in the Brn3c null mutants. The NT-3 signal was more diffuse in both the control and the Brn3c null littermates but showed some slightly above background signal in the area of the sensory epithelium even in the Brn3c null mutant.

In the cochlea, the BDNF signal was very weak in the base in both the control (Fig. 9b) and the Brn3c null littermates (Fig. 9a). However, there was a surprisingly strong NT-3 signal over the organ of Corti in the apex of Brn3c null littermates (Fig. 9c) that closely matched the signal found in the control littermates (Fig. 9d). The NT-3 signal in the area of the cochlea that corresponds to the IHC's persisted at least until P8 in the apex of Brn3c null mice (Fig. 9e,9f).

Immature hair cells and limited expression of neurotrophins rescue afferent projection until P8

The neurotrophins have long been implicated as the mediators for target mediated cell death. In essence, this theory of regulation of neuronal connections via neurotrophic support implies that only limited quantities of neurotrophins are released by the target tissue to support only properly connected sensory neurons by providing only these with a critical amount of neurotrophins [21]. Past experiments, which eliminated one or more neurotrophins entirely, only partially tested the basic assumption of this theory in vivo. For example, in BDNF heterozygotic animals there is a small decrease in the number of sensory neurons in the ear [13] but no change in the overall pattern of innervation. Moreover, all of the neurotrophin/neurotrophin receptor null mutants studied to date eliminate expression in both the peripheral target and the brain simultaneously and achieve their effects in late embryos. In contrast, Brn3c is barely expressed during embryogenesis in the CNS [22] thus arguing that the survival of ear primary neurons depends crucially on the periphery.

We suggest that limited expression of neurotrophins in the immature target leads to retention of primary neurons and their afferent fibers in newborn and early postnatal Brn3c null mice. In agreement with this suggestion, our in situ data show a severe reduction in expression of BDNF (Figs. 8, 9). Despite this reduction, some, apparently biologically significant low levels of neurotrophins are apparently expressed in all sensory epithelia in embryonic Brn3c null mutants. It appears that even these low levels of neurotrophins in the vestibular sensory epithelia can sustain normal fiber outgrowth and a limited maintenance until P0. The comparatively high level of NT-3 expression in Brn3c null mutant cochlea, which has been shown to be the neurotrophin most prominently supporting spiral sensory neurons [6, 9, 14], is in agreement with the normal development of cochlear innervation in newborn Brn3c null mutants.

Since BDNF appears to be exclusively produced by hair cells [6, 7, 20], the lack of terminal differentiation of hair cells appears to be accompanied by only limited expression of BDNF, as is the case in the canal cristae (Fig. 8). Past research has shown that in BDNF and trkB null mutants this innervation to canal cristae is lost before birth [6, 13]. That even these low levels of expression are significant is clear from the fact that all canal sensory epithelia receive an innervation by both afferents and efferents at birth and later (Figs. 2, 3). The situation in the utricle, saccule and cochlea is less clear as both BDNF and NT-3 are expressed [6, 7]. In fact, it appears that the levels of NT-3 expression hardly differ from those in control littermates (Figs. 8, 9). The latter is particularly obvious in the apex of the cochlea. In fact, in the cochlea of embryos and neonates as well as in the saccule and utricle of embryos, NT-3 is predominantly expressed in supporting cells [6]. This expression of NT-3 in supporting cells in embryos and neonates may even preserve in Brn3c null mutants the numerous spiral sensory neurons in the apex of 8-day old animals (Figs. 3, 4).

Long term survival of apical sensory neurons is unrelated to differentiated hair cells and likely is independent of neurotrophins

Past work has shown that BDNF and NT-3 appear to shift in their expression into inner hair cells around birth or are lost in the ear and other parts of the nervous system [6, 10, 20, 23]. It has been suggested that other neurotrophins may come into play in the ear [11]. However, the apex of Brn3c null mutants, which retains most of the spiral neuron afferents at 6-month, shows no formation of mature hair cells at any time surveyed here (P0, P8, P10). Thus the factor(s) cannot be released by differentiated hair cells but there is an unexplored possibility of generalized expression of neurotrophins at low levels in the undifferentiated epithelia. However, known neurotrophins are largely absent in the adult cochleae [4] and significant amounts of neurotrophin expression appear to develop only in postnatal animals in the CNS [24]. Such expression in the CNS could possibly offset to some extent the peripheral loss of neutrophins in targeted null mutants. This would work only in cases in which the loss of primary neurons would not already be completed long before birth as is the case in the ear [6, 13].

It therefore remains unclear what supports the many afferents in the apex of the Brn3c null mutants until 6 months of age or longer, unknown peripheral neurotrophic substances or neurotrophic support provided by the CNS. Comparable long term retention of apical spiral sensory neurons was described for the deaf white cat [25] and may also be the case in humans with congenital deafness.

Growth of fibers to the cochlea does not require mature hair cells

Formation of radial fibers that bring peripheral processes of spiral neurons to the organ of Corti seems to be rather normal, even in the apex of the cochlea which does not even develop immature hair cells recognizable by Myo VIIa. This does not preclude that even less differentiated hair cell precursors may form in the apex or that those precursors have died before birth. Interestingly, the growth to outer hair cells is most affected in Brn3c null mutant cochleae. Instead of extending radially through the tunnel of Corti to outer hair cells, afferents appear to stall and extend in longitudinal directions as inner spiral bundles (Figs. 3, 7). It has been shown that the differentiation of pillar cells depends on activation of Fgfr3 [26] by FGF's, probably Fgf8, a factor produced by developing and mature inner hair cells [18, 27]. The formation of supporting cells also appears to depend on the proper expression of various bHLH factors such as Hes 1 and 5 which appear to be regulated by the Notch signaling pathway [17, 28, 29]. It is possible that the apparent inability of fibers to extend along outer hair cells is related to the lack of differentiation of supporting cells in the absence of mature hair cells. Probing for the proper expression of supporting cell specific markers in Brn3c null mutants is necessary to further evaluate these suggestions.

Previous work has suggested that efferent fibers, derived from facial branchial motoneurons, follow afferents and for that reason reflect in detail all connectional deviations of afferents [16, 30–32]. Our data agree with this scenario but extend it. Specifically, efferent fibers may suffer the same loss of their molecularly unknown neurotrophic support as do afferents and therefore show the same spatio-temporal profile of loss (Fig. 3). Alternatively, efferent fibers can sustain established connections only as long as afferent fibers are present. More information on the actual molecular support of efferent fibers in the ear is needed before those two possibilities can be distinguished.

The onset of cochleotopic projections has been investigated in detail in only very few mammals [19]. These data suggest that a cochleotopic projection develops before onset of hearing. Our data show that at least a cochleotopic projection in the cochlear nerve and the cochlear nuclei develops even in Brn 3c null mutants (Fig. 6). Those data are also consistent with the formation of a crude cochleotopic projection in NT-3 null mutants [6, 14]. Those mutants have a much more severely reduced density of innervation in the developing cochlea [6, 14] than the reduction we found in Brn3c null mutants at P0. In this context, it would be interesting to see whether the recently generated transgenic mouse in which BDNF is expressed under NT-3 promoter control [33] could actually perform an even more pronounced rescue in a Brn3c null mutant background.

Relevance of our findings to other mutations with hair cell loss

Vahava et al. [12] and Frydman et al. [34] have demonstrated that mutation in a single human allele of Pou4f3 (Brn3c) results in a late onset of high frequency hearing loss. Thus far no data are available that correlate the late onset of hearing loss with morphological defects in the cochlea. In fact, an investigation of Brn3c heterozygotic mice showed no additional effect on hearing loss due to haploinsufficiency in these mice [35]. Our data on Brn3c null mouse mutants nevertheless suggest that the high to low frequency progressive hearing loss in humans may correlate with the unknown factor(s) that mediate apical spiral neuron survival into adulthood. Quantitative data on human Pou4f3 cochleae is needed to verify that spiral neuron loss is more pronounced in the basal, high frequency area of these hearing-deficient patients. Similar data are also needed for other congenitally deaf children [36] and in the deaf white cat [25]. Further work on the recently available mutant mice with specific absence of all hair cells [37], outer hair cells in the base [38] or apex [39] could further help to clarify the role of hair cells in forming and maintaining afferent and efferent innervation in these model mice.

Breeding and genotyping

Brn3c mice were bred as previously described [22]. Mice were genotyped and Brn3c null mutants were raised to a specific age (P0, N = 8; P7, N = 6; P8, N = 6; 6 months old, N = 2). The animals were deeply anesthetized with pentobarbital and transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Ears were postfixed in the same fixative for at least one day prior to dissection.

Tracing of nerve fibers

DiI tracing from the brainstem was performed as previously described [41]. Small filter stripes soaked with saturated DiI were implanted into the efferent fiber bundle near the floor plate or into the ascending or descending afferents in the alar plate [42]. After appropriate diffusion time of the dye, the ears were dissected and mounted flat for visualization in an epifluorescent microscope. Images were taken on black and white film or acquired with a cooled CCD camera and processed using ImagePro software (Media Cybernetics).

Ears were subsequently reacted for acetylated tubulin to reveal the pattern of innervation in addition to the DiI tracing with a different technique as previously described [14]. Briefly, dissected ears were incubated with 1:500 anti-acetylated tubulin antibodies (Sigma, St. Louis) followed by secondary antibodies conjugated to HRP. Dissected inner ears were reacted with DAB and H₂O₂ for HRP distribution and subsequently viewed as whole mounts.

After photographing as whole mounts, the ears were embedded in epoxy resin and sectioned to reveal the distribution of fibers inside the sensory epithelia in more detail.

Cochleotopic projection was evaluated by inserting DiI soaked filter strips into the base and DiA soaked filter strips into the apical turn of 2 P0 and 4 P8 Brn3c null mutants and a similar number of wildtype littermates. After appropriate diffusion time [43], the brains with the attached cochlear nerve were embedded in gelatin, hardened in 4% PFA over night, sectioned coronally on a vibratome (100 μm thickness) and viewed with a Biorad Radiance 2000 confocal system attached to a Nikon Eclipse 800 microscope. Image stacks were collapsed to view the entire projection in one section in one focal plane.

In situhybridization

Four ears each of Brn3c null mutants (P0 and P8) and control littermates were embedded in paraffin, sectioned and probed for the presence of BDNF and NT-3 mRNA using the in situ technique previously described [7]. Sections were lightly counterstained and viewed with bright and dark field microscopy. Images were acquired with a CCD camera and displayed as false color images as previously described [18].

Immunocytochemistry of hair cells and quantification of hair cells and neurons

Immunostaining of P7 cochlear whole mounts using the hair cell specific-myosin VII (MyoVIIa) antibody was performed as previously described [2]. Images of labeled organs of Corti were then acquired using a SPOT digital camera (Diagnostic Instruments Inc.) and the number of hair cells was scored from acquired images. To determine the number of hair cells or neurons in vestibular endorgans or sensory ganglia, serial sections were stained with Cresyl Violet and images were similarly acquired and scored. Every other section was scored for each sensory epithelium and every forth section was scored for each ganglion. Only neurons or cells with a clear nucleus and nucleoli were counted and 4–6 samples were counted for each epithelium and ganglion. All data were tested for significance using two-sample Student's t-test with unequal variances.