|Home | About | Journals | Submit | Contact Us | Français|
Atoh1 is a basic helix-loop-helix transcription factor necessary for the specification of inner ear hair cells and central auditory system neurons derived from the rhombic lip. We used the Cre-loxP system and two Cre-driver lines (Egr2Cre and Hoxb1Cre) to delete Atoh1 from different regions of the cochlear nucleus (CN) and accessory auditory nuclei (AAN). Adult Atoh1-conditional knockout mice (Atoh1CKO) are behaviorally deaf, have diminished auditory brainstem evoked responses and disrupted CN and AAN morphology and connectivity. In addition, Egr2; Atoh1CKO mice lose spiral ganglion neurons in the cochlea and AAN neurons during the first 3 days of life, revealing a novel critical period in the development of these neurons. These new mouse models of predominantly central deafness illuminate the importance of the CN for support of a subset of peripheral and central auditory neurons.
Deafness is one of the most common sensory abnormalities in humans, affecting an estimated 16.1% (29 million people) of the US population (Agrawal et al., 2008). The pathology leading to deafness can occur in the peripheral and/or central nervous system. Peripheral hearing impairment comprises the majority of cases and arises from gross cochlear malformations; dysfunction of cochlear hair cells, which detect auditory stimuli and transmit the signal to spiral ganglion neurons (SGNs); dysfunction of the constant ionic current flow within the cochlea necessary for normal hair cell transduction; and/or dysfunction of the auditory nerve, which contains SGN axons and relays the auditory signal to the brainstem cochlear nucleus (CN). Central hearing impairment arises from disruption of the central auditory pathways, which are comprised of two parts. The main pathway for sound perception runs from the CN through the lateral lemniscus to the nuclei of the inferior colliculus, which in turn relay auditory information to the medial geniculate bodies of the thalamus and on to the temporal lobes. The other pathway travels from the CN through the trapezoid body to the accessory auditory nuclei (AAN) of the brainstem, which are comprised of the superior olivary complex (SOC) and the nuclei of the trapezoid body (NTB). These nuclei are thought to function in the localization of the temporal and spatial origins of sounds (Willott, 2001). The etiologies of primary central deafness are varied; examples include brainstem malformations, environmental causes (such as fetal alcohol syndrome), dementia, CNS infection, stroke and trauma (Griffiths, 2002). Unfortunately, the pathogenesis of deafness caused by central defects is poorly understood, in part due to the absence of genetic animal models that selectively affect these components of the auditory pathway.
Atoh1, or Math1, is a basic helix-loop-helix transcription factor important for the development of cells in the peripheral and central portions of the auditory system. In the peripheral auditory system, Atoh1 is both necessary and sufficient to direct the differentiation and maintenance of hair cells of the cochlea (Bermingham et al., 1999).
In the central nervous system, Atoh1 is necessary for the specification of populations of neurons in the nuclei of the dorsal lateral lemniscus and CN, which derive mostly from the Atoh1-expressing dorsal region of the developing embryonic hindbrain known as the rhombic lip (Akazawa et al., 1995; Ben-Arie et al., 1996; Wang et al., 2005). We hypothesized that deletion of Atoh1 exclusively in regions of the developing central nervous system would create an excellent model of central deafness and would provide insight regarding the connectivity and maintenance of the central pathways.
To test our hypothesis, we used knowledge of the spatial embryonic origins of the CN (Marin and Puelles, 1995; Cambronero and Puelles, 2000; Cramer et al., 2000; Farago et al., 2006) to design an experimental paradigm using Cre-loxP technology to specifically ablate expression of Atoh1 in areas of the murine rhombic lip known to contribute neurons to the developing CN. We demonstrate that loss of Atoh1 expression in these hindbrain regions results in primary loss of neurons from the CN and AAN. This neuronal loss causes secondary disruption of both the peripheral and central auditory systems, possibly due to loss of trophic support for these neuronal populations.
The generation of Atoh1CreER-T2, Atoh1flox, Atoh1LacZ, Egr2Cre, Hoxb1Cre and ROSAR26R mice were described previously (Ben-Arie et al., 1996; Soriano, 1999; Voiculescu et al., 2000; Arenkiel et al., 2003; Machold and Fishell, 2005; Shroyer et al., 2007). All animals used for these experiments were maintained on mixed genetic backgrounds except for the Atoh1LacZ and ROSAR26R mice, which are congenic on the C57Bl/6J strain background.
Atoh1 conditional knockout (Atoh1CKO) mice were generated by first crossing either Egr2Cre mice or Hoxb1Cre mice with Atoh1LacZ mice to generate Egr2Cre/+; Atoh1LacZ/+ and Hoxb1Cre/+; Atoh1LacZ/+ double transgenic animals. These animals were mated with Atoh1flox/flox mice to generate triple transgenic mice of four genotypes: Cre+/+; Atoh1+/flox, CreCre/+; Atoh1+/flox, Cre+/+; Atoh1LacZ/flox and CreCre/+; Atoh1LacZ/flox. Only animals with Egr2Cre and Atoh1LacZ or Hoxb1Cre and Atoh1LacZ alleles lack Atoh1 expression in the Egr2Cre (Egr2; Atoh1CKO) or Hoxb1Cre (Hoxb1; Atoh1CKO) distributions. Mice of the other three genotypes (Cre+/+; Atoh1+/flox, CreCre/+; Atoh1+/flox, and Cre+/+; Atoh1LacZ/flox) are collectively referred to as “wildtype” because they display no abnormal phenotypes and are indistinguishable based on the testing reported here.
Acoustic startle responses of 2–10 month-old mice (6–15 mice of each genotype) were measured using the SR-Lab System (San Diego Instruments, San Diego, CA). Mice were placed in a Plexiglas cylinder and left undisturbed for 5 minutes, and 70 dB background white noise was played throughout the testing period. Test sessions consisted of six blocks of eight different trial types presented in pseudorandom order such that each trial type occurred once within a block of eight trials (average inter-trial interval = 15 seconds, range = 10–20 seconds). One trial type consisted of the startle stimulus alone, which was a 120 dB pure tone sound played for 40 ms. Three different “prepulse” trial types consisted of 20 ms sounds of 74, 78, or 82 dB presented 100 ms before the 120 dB startle stimulus. These three prepulse sounds were also presented separately without the startle stimulus. Finally, trials where no stimulus was presented were used to establish the startle baseline. The startle response, consisting of movement by the test mouse when the sounds were played, was recorded every 1 ms for 65 ms starting with the onset of the stimulus trial, and the maximum startle amplitude recorded during the 65-ms sampling window was used for comparison between mice of different genotypes.
Our methodology for measuring ABR, DPOAE, CAP and CM in mice has been described previously (Xia et al., 2007). Briefly, 6–7 week-old mice were anesthetized using ketamine (100mg/kg) and xylazine (5 mg/kg). Mice were placed on a heating pad to maintain normal body temperature throughout the test procedures. Different sets of animals were used for ABR/DPOAE (n = 6–9 per genotype) and CAP/CM (n = 3 – 4 per genotype) measurements; typical recording times ranged from 25–40 minutes. Acoustic stimuli were generated digitally, converted to analog signals, and then attenuated to the appropriate intensity according to our experimental design (RP2 and PA5, Tucker-Davis Technologies). Two different speaker systems were used: high frequency piezoelectric speakers for the ABR and DPOAE measurements (EC1, Tucker-Davis Technologies) and a supertweeter (Radio Shack) for the CAP and CM measurements. The speakers were connected to an ear bar inserted into the ear canal and calibrated from 4 to 95 kHz by a probe-tip microphone (type 8192, NEXUS conditioning amplifier, Bruel and Kjar, Denmark) inserted through the earbar. The tip of the microphone was within 3 mm of the tympanic membrane.
The ABR was measured from a needle electrode positioned at the ventral surface of the tympanic bulla referenced to an electrode placed at the vertex of the skull. A ground electrode was placed in the hind leg. The stimulus was a 5 ms sine wave tone pip of alternating polarity with cos2 envelope rise and fall times of 0.5 ms and a repetition time of 50 ms. The stimulus intensity ranged from 10 to 80 dB SPL in 10 dB steps. The frequency range studied was 4 to 90 kHz. Two hundred fifty ABR responses were sampled at each frequency over the 50 ms repetition time and averaged. Thresholds were calculated by interpolating the peak-to-peak voltages of the ABR waveforms over the range of stimulus intensities and determining when the ABR was four standard deviations above the noise floor. If no ABR response was detected at our equipment limit of 80 dB SPL, we arbitrarily defined the threshold to be 80 dB.
To measure the CAP, a surgical procedure was performed to open the tympanic bulla. The anesthetized mouse was secured rigidly in a head holder, a ventral incision was made, and the pinna was resected. The bulla was carefully opened medial to the tympanic annulus and the earbar was secured within the ear canal. The CAP was measured from the ball-ended tip of a Teflon-coated silver wire (0.003 inch diameter, AM Systems, Carlsborg, WA) advanced onto the round window membrane with a micromanipulator. The signal was referenced to a silver wire inserted under the skin near the vertex of the skull. The ground electrode was placed in the hind leg. The stimulus used to elicit the CAP was identical to that used to elicit the ABR, however only 24 repetitions were averaged at each stimulus intensity level.
The stimuli for eliciting DPOAEs were two sine wave tones of differing frequencies (F2 = 1.2*F1) of 1 second duration with F2 ranging from 4 to 90 kHz. The two tones were presented at identical intensities, which ranged from 20–80 dB SPL in 10 dB increments. The acoustic signal picked up by the microphone in the earbar was digitized at 200 kHz and the magnitude of the 2*F1-F2 distortion product determined by fast Fourier transform (FFT). The surrounding noise floor was also calculated by averaging 20 adjacent frequency bins around the distortion product frequency. DPOAE thresholds were calculated off-line by interpolating the data and identifying when the signal was > −5 dB SPL and greater than two standard deviations above the noise floor. If no DPOAE response was detected at our equipment limits of 80 dB SPL, we arbitrarily defined the threshold to be 80 dB.
The CM signal was measured from a silver wire placed on the round window membrane, as described for the CAP measurement. The stimulus was a 30 ms 6 kHz tone repeated every 1 second, and its intensity was ranged from 10–100 dB SPL. By measuring the speaker output with the probe tip microphone in the ear bar, FFT analysis demonstrated that all stimulus harmonics and noise at all other frequencies were at least 50 dB below the primary signal at all stimulus intensities. The CM signal measured by the bioamplifier was digitized at 200 kHz and the magnitude of the response at 6 kHz determined by FFT.
For embryonic tissue, pregnant dams were euthanized and embryos dissected into cold 1X PBS. Either whole embryos (E9.5) or dissected heads and brains (E16-P0) were immersion-fixed for 30 minutes-overnight at 4°C in either fresh 4% paraformaldehyde (PFA)/0.1M phosphate buffer or 10% neutral buffered formalin (NBF; Fisher). Mice 3 days of age and older were transcardially perfused with 10% NBF. Brains and heads were dissected at P3, while for older ages brains and cochlea were dissected and tissues were post-fixed overnight at 4°C. Following fixation, all tissues were washed three times in 1X PBS. Brains were stored in 70% EtOH for paraffin embedding or 30% sucrose/1X PBS for cryoembedding. Heads and cochlea were decalcified in 0.12M EDTA for 5 days, then transferred to 70% EtOH and embedded in paraffin.
For lipophilic dye injections, P0-P18 mice were anesthetized with 60 mM tribromoethanol (Avertin) and then transcardially perfused with cold 1X PBS followed by cold 4% PFA/0.1M phosphate buffer. Heads were stored in 0.4% PFA/0.1M PB at 4°C until the time of dye injection.
For paraffin sectioning, tissues were dehydrated and embedded in Paraplast (McCormick Scientific) or TissuePrep (Fisher). Six μm serial sections were cut on a Leica microtome and collected on Superfrost®/Plus slides (Fisher).
For cryostat sectioning, tissues were embedded in Tissue-Tek OCT (Sakura Finetek), serially-sectioned at 20–25 μm on a freezing microtome, and collected on Superfrost®/Plus slides.
Embryonic and P0 tissues were stained for β-galactosidase activity in wholemount preparations for 2–24 hours at 37° C. Tissues were then washed three times in 1X PBS and transferred to 70% EtOH for storage. Slides with cryostat sections of adult brain were stained in a similar fashion, then counterstained with Nuclear Fast Red (Vector Labs), dehydrated, and mounted with Cytoseal 60 (Richard-Allan Scientific).
Paraffin sections from brains of all ages were rehydrated and stained with Cresyl Violet, then dehydrated and coverslipped. Paraffin sections of heads and cochlea of all ages were rehydrated and sequentially stained with Mayer’s Hematoxylin and Eosin Y (Sigma), followed by dehydration and coverslipping.
For all histological analyses (Cresyl Violet, X-gal, immunochemistry and in situ hybridization), series of slides were processed to allow exact matching of the anteroposterior levels between wildtype, Egr2; Atoh1CKO and Hoxb1; Atoh1CKO brains.
Tissue sections from at least two brains of each genotype were blocked for 1–2 hours at room temperature (RT) in 1X PBS/0.3% Triton X-100 (PBST) with 2% normal goat or donkey serum (PBST-S). Slides were incubated at 4°C overnight in primary antibodies diluted in PBST-S: mouse anti-Cat-301 (MAB5284, Millipore) 1:500; chicken anti-choline acetyltransferase (AB15468, Chemicon) 1:2000. Goat anti-mouse (Jackson Immunoresearch) and donkey anti-goat (Jackson Immunoresearch) secondary antibodies conjugated to Cy3 were used at a 1:500 dilution in PBST-S and applied for 30 minutes at RT. For horseradish peroxidase staining (ChAT), HRP-conjugated goat anti-chicken secondary antibody from Vector Labs was used following manufacturer’s directions.
Goat polyclonal anti-RORα (SC-6062, Santa Cruz Biotechnology) was used on antigen retrieved tissue (Ino, 2004). Following fixation and prior to cryoembedding, brains were cut into 3–5 mm thick slices in the coronal plane and boiled for 3 minutes in 10 mM sodium citrate buffer, pH 6.0. A 1:2000 dilution of the primary antibody was used in PBST in 2% normal donkey serum. Diaminobenzidine labeling was performed using a Vectastain kit (Vector Labs) according to manufacturer’s instructions.
Tissue preparation and automated ISH were performed as previously described (Carson et al., 2002; Visel et al., 2004; Yaylaoglu et al., 2005) and as described online at http://www.genepaint.org/RNA.htm. Briefly, heads of embryonic mice were embedded in OCT and fresh frozen in a custom-made freezing chamber that allows stereotaxic alignment of the specimen. Serial sections were cut at 20 μm thickness on a freezing microtome. After PFA fixation and acetylation the slides were assembled into flow-through hybridization chambers and placed into a Tecan (Mannedorf, Switzerland) Genesis 200 liquid-handling robot, which executes a script that performs non-radioactive ISH in less than 24 hrs. Antisense probes for Atoh1 and VGLUT2 were generated from PCR-amplified cDNA clones and used for in vitro transcription of digoxygenin-labeled riboprobe using either T7 or SP6 RNA polymerase. Robotic ISH was performed according to a previously published protocol (Yaylaoglu et al., 2005). Hybridized antisense probe was detected by catalyzed reporter deposition (CARD) using biotinylated tyramide followed by colorimetric detection of biotin with avidin coupled to alkaline phosphatase (Carson et al., 2005; Yaylaoglu et al., 2005). At least two brains of each genotype were analyzed.
Pregnant dams obtained from matings of Atoh1CreER-T2 and ROSAR26R animals were intraperitoneally injected at E9.5 or E10.5 with 4mg (200 μL) of a 20mg/mL solution of tamoxifen (Sigma) dissolved in corn oil. Embryos were harvested on day E18.5 of gestation.
Wedges soaked with lipophilic dyes of different colors (Fritzsch et al., 2005) were injected into the ear (base, apex and vestibular organs) or the brainstem (ventral acoustic stria, cochlear nuclei) using appropriate landmarks (trigeminal nerve or facial nerve) of P0–P19 conditional knockout and wildtype mice. Brains and ears were incubated for 7–14 days (depending on age) at 36°C in 4% PFA. Ears were dissected and imaged as whole mounts to ensure application accuracy. Brains were viewed as whole mounts and subsequently embedded in 4% gelatin and hardened for 10 days in 10% PFA at 4°C. One hundred μm thick coronal sections were mounted in glycerol and viewed with a Zeiss LSM 510 or Leica SPE confocal microscope using appropriate filter settings. Image stacks along the Z-axis were taken and combined using Zeiss LSM or Leica software. Two or three brains or cochlea from each genotype were analyzed.
Spiral ganglion neuron (SGN) counts were done on 6 μm paraffin sections. Consecutive sections through the entire spiral ganglion were stained with Mayer’s Hematoxylin/Eosin Y, and all SGNs with a clear nuclear membrane were counted on every 5th section (every 30 μm) at 200 × magnification on a Zeiss Axioplan 2 microscope (Zeiss Instruments). Neurons and pyknotic nuclei of the medial nucleus of the trapezoid body (MNTB) were counted in a similar fashion, except every 10th (every 60 μm; P0, P3, P7) or every 20th (every 120 μm; adult) section was analyzed. A total of 3 or 4 cochlea and 4 MNTBs from 2 animals of each genotype at each age were counted.
Axonal counts were obtained from three ears of P17–19 wildtype (Egr2+/+; Atoh1LacZ/flox) and Egr2; Atoh1CKO mice. For this, ears were osmicated and tangential sections were taken parallel to the basal hook region and the middle turn (Postigo et al., 2002). Images were taken from the middle and basal cochlear turns using a Nikon E800 with a 60× oil immersion lens (NA 1.4) and all myelinated axons in a 100 μm stretch of these sections were counted on prints as previously described (Postigo et al., 2002).
To correct for overcounting of SGNs, digital photographs of all regions of the spiral ganglion were taken in a single mid-modiolar section using a Zeiss Axiocam and nuclear diameters were measured using Zeiss Axiovision software (Version 4.5). The Hendry method (Hendry, 1976) was used to correct for over-representation of nuclei in multiple sections. No corrections were done for axon counts.
All statistical analysis was done using SPSS version 11 for Mac OS-X (SPSS Inc.). ANOVA was used to compare ABR, acoustic startle, CAP, CM and DPOAE values across genotypes, followed by calculation of least-square differences for pair-wise comparisons. Cell and axon counts were compared using independent sample two-tailed t-tests.
The Egr2Cre line drives Cre expression specifically in rhombomeres 3 and 5 of the embryonic neural tube (Voiculescu et al., 2000), while the Hoxb1Cre line drives expression in the rhombencephalon distal to the posterior border of rhombomere 3 (Arenkiel et al., 2003). We verified these expression patterns and demonstrated that cells derived from these lineages populate different regions of the adult cochlear nucleus (CN) and accessory auditory nuclei (AAN) (SUPPLEMENTAL FIGURE 1A–H). Virtually all neurons of the anteroventral cochlear nucleus (AVCN) and AAN are descended from the Egr2 lineage, as are large neurons of the posteroventral (PVCN) and dorsal (DCN) cochlear nuclei (SUPPLEMENTAL FIGURE 1B–D). In contrast, the Hoxb1 lineage densely populates the PVCN and DCN but contributes very few neurons to the AVCN (SUPPLEMENTAL FIGURE 1F, G). The majority of the AAN is also derived from the Hoxb1 lineage with the exception of the medial nucleus of the trapezoid body (MNTB), where only about half of the neurons are labeled (SUPPLEMENTAL FIGURE 1H). These data suggest that the AVCN is derived almost solely from rhombomere 3, the PVCN from rhombomeres 3 and 4, the DCN from rhombomeres 4 and 5, the MNTB from rhombomeres 3 and 5, and the rest of the AAN from rhombomere 5. This is in good agreement with previous fate maps of the mouse CN (Farago et al., 2006) and chick superior olivary complex (Marin and Puelles, 1995).
We next used these Cre-driver alleles to delete a floxed allele of Atoh1 (Atoh1flox) (Shroyer et al., 2007) in transgenic mice. Cre expression in both of these lines precedes Atoh1 expression by 0.5–1 day, and we verified that Atoh1 expression was completely absent from the regions where Cre protein was produced (SUPPLEMENTAL FIGURE 1I–N). This suggests that the overall recombination efficiency was close to 100% in both lines. Egr2; Atoh1CKO animals were born in the expected Mendelian ratio, had a normal lifespan, and did not display any overt morphological or behavioral phenotypes. Hoxb1; Atoh1CKO animals were born in the expected Mendelian ratio, but roughly 50% of these animals died within 24–36 hours of birth.
We examined the CN of adult Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals for morphological abnormalities. Egr2; Atoh1CKO animals lack nearly the entire AVCN with the exception of a small portion of the posterior region, and both the DCN and PVCN are small compared to wildtype animals (FIGURE 1B, E, H; SUPPLEMENTAL FIGURE 2). In contrast, the AVCN is completely preserved in Hoxb1; Atoh1CKO animals, but the majority of the PVCN and DCN are lost (FIGURE 1C, F, I; SUPPLEMENTAL FIGURE 3). Cochlear root neurons are absent from adult Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals (SUPPLEMENTAL FIGURES 2, 3). This pattern of neuronal loss corresponds to the areas of highest Cre expression in these Cre-driver lines and substantiates the proposed rhombomeric origins of the CN (SUPPLEMENTAL FIGURE 1B, C, F, G).
We performed immunocytochemistry for RORα, which is expressed by cartwheel cells in the DCN (Ino, 2004), and Cat-301, a marker of the extrasynaptic neuronal surface expressed throughout the AVCN and PVCN (Lurie et al., 1997), to more accurately identify the missing regions of the disrupted CN of Atoh1CKO animals. RORα-positive and Cat-301-positive neurons are present in the CN of Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals (FIGURE 1J–O). These data demonstrate that no subdivision of the CN is completely lost in either of these Atoh1CKO lines.
Given the disruption of CN structure, we hypothesized that output from the CN would also be disrupted. To assess this possibility, we examined the size of the trapezoid body, or ventral acoustic stria, a fiber tract that projects from the CN to the brainstem AAN. Nissl-stained sections of and lipophilic dye injections directly into the trapezoid body demonstrate that it is of reduced caliber in Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals, but that this reduction is much more dramatic in Egr2; Atoh1CKO animals (single arrows in FIGURE 2A–C). In addition, dye injections into the trapezoid body failed to back-label CN neurons of Egr2; Atoh1CKO mice, suggesting that these fibers do not originate from or connect to the CN (FIGURE 2B). Finally, dye injections into the CN of Egr2; Atoh1CKO animals labeled cochlear afferents, but failed to label the trapezoid body or AAN (data not shown). These results suggest that the cellular disruption in the CN of Hoxb1; Atoh1CKO mice reduces the projection to the AAN, and completely ablates this projection in Egr2; Atoh1CKO animals.
The brainstem AAN, comprised of the superior olivary complex (SOC) and nuclei of the trapezoid body (NTB), are major targets of the CN. We therefore examined the morphology of these regions for possible secondary affects of CN disruption. The lateral superior olive (LSO), medial superior olive (MSO) and medial NTB (MNTB) are dramatically smaller in Egr2; Atoh1CKO compared to wildtype animals, while more subtle decreases in size of the lateral NTB (LNTB) and ventral NTB (VNTB) are also present (FIGURE 2D, E, G, H; SUPPLEMENTAL FIGURE 4). In contrast, the effects on the AAN of Hoxb1; Atoh1CKO animals are much less dramatic, with qualitative evidence of cell loss visible only in the LSO and MSO (FIGURE 2F, I; SUPPLEMENTAL FIGURE 4). The superior periolivary nucleus (SPN) is not affected in either Atoh1CKO line.
The olivocochlear bundle (OCB) is the main efferent pathway to the cochlea. In rodents, choline acetyltransferase (ChAT)-positive neurons of the LNTB, LSO and VNTB project in the OCB (Yao and Godfrey, 1998). Since neurons in these regions are lost in Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals, we sought to determine if olivocochlear neurons were affected. ChAT-positive neurons are found in the regions of the LNTB, LSO and VNTB of Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals, but are more densely packed and located more ventrally than in wildtype animals (FIGURE 2J–L). These data demonstrate that the efferent auditory connections to the cochlea are not dependent upon connectivity with the CN for their survival.
The AAN lie within the domains of Egr2Cre and Hoxb1Cre expression (SUPPLEMENTAL FIGURE 1D, H), opening the possibility that primary deletion of Atoh1 from this region could directly result in loss of neurons. Preliminary fate mapping data from our laboratory suggested that a population of SOC neurons that express the vesicular glutamate transporter VGLUT2 were derived from an early (E9.5 – E10.5) Atoh1-expressing lineage, and that these neurons were lost in Atoh1-null mice (HYZ, submitted). No Atoh1-lineal neurons labeled from E11.5 – E14.5 populated the AAN (data not shown). We used a tamoxifen-inducible system to fate map these cells in Atoh1CreER-T2; ROSAR26R mice and verified that a subset of neurons that populate the LSO and MSO were indeed descended from the Atoh1 lineage (FIGURE 2M). In situ hybridization for VGLUT2 on sections of the AAN from E18.5 Egr2; Atoh1CKO animals shows that these neurons are missing, suggesting a primary effect of Atoh1 loss on this neuronal population (FIGURE 2N, O). Neurons in other regions of the AAN are not descended from the Atoh1 lineage, suggesting that loss of neurons in these areas must be secondary to deletion of Atoh1 from the CN, LSO and/or MSO neurons.
We next sought to characterize the behavioral and physiological effects caused by disruptions of the central auditory system in Atoh1CKO mice. We used the acoustic startle response to test hearing in Egr2; Atoh1CKO and Hoxb1; Atoh1CKO mice. Mice were placed in a small Plexiglas chamber where a piezoelectric device detected their movement in response to sounds of varying intensities. Hoxb1; Atoh1CKO mice demonstrated no response to sounds of 120 dB (n=7, no stim vs 120 dB, t-test p = 0.36). Egr2; Atoh1CKO animals responded to sounds of 120dB (n=15, no stim vs. 120dB, t-test p = 0.006), but this response was greatly attenuated when compared to littermate controls (inter-genotype comparison, ANOVA p = 0.0002, least-squares difference for pairwise comparisons p = 0.00002 to 0.33 for Egr2; Atoh1CKO vs. other genotypes). Thus, Egr2; Atoh1CKO and Hoxb1; Atoh1CKO mice are behaviorally deaf.
To uncover the underlying auditory system defects, we performed electrophysiological testing of cochlear hair cell, auditory nerve and brainstem auditory pathway function. We measured distortion product otoacoustic emissions (DPOAEs), sounds generated secondary to non-linearities associated with outer hair cell force production, to examine forward and reverse transduction in Atoh1CKO mice. Hoxb1; Atoh1CKO animals had no measurable DPOAEs at any frequency or sound intensity tested, suggesting that defects in the peripheral auditory system are sufficient to explain their deafness. In contrast, DPOAEs and cochlear microphonics (CM), which are field potentials representing outer hair cell receptor potentials in response to acoustic stimuli, were normal in Egr2; Atoh1CKO mice (FIGURE 3A, B). This demonstrates that hair cell function is normal in Egr2; Atoh1CKO animals.
We next tested auditory nerve function in Egr2; Atoh1CKO mice by recording compound action potentials (CAPs), field potentials generated by the synchronized firing of SGNs in response to tone burst stimuli (FIGURE 3C, E). CAPs with normal waveform morphology (FIGURE 3E) but slight decreases in amplitude were found in Egr2; Atoh1CKO mice (FIGURE 3C, t-test *p = 0.05). Together with the normal hair cell function testing, these data suggest that defects in the peripheral auditory system are insufficient to explain the profound hearing impairment of Egr2; Atoh1CKO mice.
In contrast, we found that the auditory brainstem evoked response (ABR), consisting of synchronized field potentials generated by the auditory nerve and brainstem neurons in response to acoustic stimuli, had threshold elevations of 34–53dB (mean = 44dB) across all frequencies tested and showed loss of the later waveforms generated by the brainstem auditory pathways (FIGURE 3D, E). These findings show that sound information does not propagate beyond the auditory nerve. Together with the histological findings, these data demonstrate that CN dysfunction causes the deafness seen in Egr2; Atoh1CKO mice.
Atoh1 is necessary for the development of inner ear hair cells (Bermingham et al., 1999). To be certain that Atoh1 was not deleted from this cell population, we carefully analyzed the structure of Egr2; Atoh1CKO mouse cochleae (FIGURE 4). The overall structure of the cochlea is unchanged in Egr2; Atoh1CKO animals, where a continuous line of Atoh1-expressing hair cells is present in all turns of the cochlea at birth (FIGURE 4A, B). In addition, both inner and outer hair cell morphology is preserved throughout all cochlear turns, as demonstrated by representative sections from the mid-modiolar region of adult cochleae shown in FIGURE 4C–F. These data, combined with normal hair cell electrophysiological data, suggest that the hair cell population is not affected in Egr2; Atoh1CKO animals. The cochlear hair cell population of Hoxb1; Atoh1CKO mice, however, was not normal (data not shown); further studies of these abnormalities will be the subject of another report. Based on the potential complex interactions of primary peripheral and central effects on the auditory system of Hoxb1; Atoh1CKO mice, the rest of the present report will focus on further characterization of Egr2; Atoh1CKO animals.
We were surprised to find that SGNs and their axons were missing from P18 and adult Egr2; Atoh1CKO cochleae (FIGURE 5A–C). Quantification of SGN soma (means: 10,258 neurons in Egr2+/+; Atoh1+/flox vs. 7,094 neurons in Egr2; Atoh1CKO, t-test p = 0.01) and axon numbers (means: base, 70 axons in Egr2+/+; Atoh1+/flox vs. 48 axons in Egr2; Atoh1CKO, t-test p = 0.09; middle, 247 axons in Egr2+/+; Atoh1+/flox vs. 148 axons in Egr2; Atoh1CKO, t-test p = 0.004) in the cochlea and auditory nerve demonstrated a 30% reduction in Egr2; Atoh1CKO animals (FIGURE 5C). This reduction in SGN and auditory nerve fiber numbers probably accounts for the decreased CAP amplitude found in Egr2; Atoh1CKO mice (FIGURE 3C). SGN numbers in wildtype animals with one copy of the Egr2Cre or Atoh1LacZ allele were normal (Egr2+/+; Atoh1+/flox vs. Egr2Cre/+; Atoh1+/flox vs. Egr2+/+; Atoh1LacZ/flox; ANOVA p=0.21), demonstrating that this effect was specific to conditional deletion of Atoh1.
Atoh1 is expressed by postnatal SGNs, although its role in this cell population is unclear (Matei et al., 2005). To rule out a direct effect of Atoh1 deletion on SGNs, we fate-mapped Egr2-lineal cells in the cochlea of adult Egr2Cre; ROSAR26R animals (FIGURE 5D–D″). Double immunolabeling for β-galactosidase and a neuronal marker (TuJ1) demonstrated that the two antigens have non-overlapping expression patterns in the spiral ganglion. These data show that, as expected, Egr2Cre was expressed only by Schwann cells of the spiral ganglion. Although it has never been demonstrated in any region of the central or peripheral nervous systems, it is also possible that Atoh1 could be expressed by glial cells of the SG and is somehow necessary for their survival. Glial cells are present in SGs of adult and early postnatal Egr2; Atoh1CKO animals (FIGURE 6C, D), demonstrating that deleterious effects of Atoh1 loss on this cell population cannot explain SGN loss. Thus, loss of Atoh1 expression from the spiral ganglion is not the cause of the decreased SGN number seen in Egr2; Atoh1CKO cochleae.
We next examined the central auditory nerve projection using lipophilic dyes to determine if SGN axons projected appropriately to the CN. At P0, robust projections normally connect the apex (green) and base (red) to the DCN and VCN (FIGURE 5E, G). Axon bundles of the central projection to the VCN are reduced in size in P0 Egr2; Atoh1CKO animals, and the axons of the auditory nerve fibers defasciculate and project aberrantly in both the rostral and caudal dimensions (FIGURE 5F, H). These data, along with the SGN counts, suggest that the deletion of Atoh1-lineal cells from the CN of Egr2; Atoh1CKO animals causes secondary loss of SGNs and their axons, and aberrant projection of the remaining auditory nerve fibers.
If the neuronal loss in the peripheral and central auditory systems is in fact secondary to a failure in genesis of CN and SOC neurons, there should be a time period at which the SGN, LNTB, MNTB and VNTB neuronal populations are normal in appearance. We therefore examined the morphology of the SGN and AAN of Egr2; Atoh1CKO animals and counted the number of neurons in these structures at different postnatal times. SGs of Egr2; Atoh1CKO mice were normal in appearance and neuron number at P0, but by P3 the number of neurons dropped to 70% of the wildtype value (FIGURE 6A, B, E). Similarly, the LNTB, MNTB and VNTB of Egr2; Atoh1CKO animals appeared qualitatively normal at P0, but showed loss of neurons by P3 (FIGURE 6F–I). Quantification of MNTB neurons (chosen because the nucleus is easily identifiable at all postnatal ages) revealed no difference between wildtype and Egr2; Atoh1CKO animals at P0 (t-test, p=0.43). However, neuronal number decreased to 50% of the wildtype value at P3, and to ~30% at P7 and in adulthood (FIGURE 6J). In addition, there was a 7-fold increase in cell death (as measured by the number of pyknotic nuclei) in the MNTB of P3 Egr2; Atoh1CKO compared to wildtype animals, but no difference at P0 or P7 (FIGURE 6K). These data identify a critical period between P0 and P3 during which the SGN and MNTB neuronal populations require contact with CN neurons for their survival.
The most novel finding of our study is that the survival of a subset of neurons in the SG and AAN depends upon Atoh1-lineal neurons in the central auditory system during a well-defined developmental time period coincident with the first three days of postnatal life. This critical period is not concomitant with or an extension of any previously reported period of naturally occurring cell death in these populations. This observation adds to the evidence for a complex interdependence of peripheral and central auditory system neurons that changes during developmental time.
Our data suggest that the majority of AAN neuronal loss in Atoh1CKO animals is a secondary effect of disruption of the AVCN. Primary deletion of Atoh1 from the LSO and MSO should occur to a similar extent in Egr2; Atoh1CKO and Hoxb1; Atoh1CKO animals since these nuclei are derived from rhombomere 5, yet loss of neurons from AAN subnuclei not derived from the Atoh1 lineage (LNTB, MNTB, and VNTB) occurs only in Egr2; Atoh1CKO animals. In addition, the magnitude of neuronal loss from the LSO and MSO themselves is much greater in Egr2; Atoh1CKO than Hoxb1; Atoh1CKO animals. These findings are consistent with the fact that the AVCN provides the majority of afferents to the LNTB, LSO, MNTB and MSO (Harrison and Irving, 1966; Willott, 2001). These data are surprising given that they suggest that deafferentation of AAN neurons rather than loss of target is what leads to their demise; possible reasons for this are discussed in the next section.
Lesions of the AVCN and cochlear nerve cause substantial loss (up to 95%) of SGNs in adult animals, suggesting that the mature SGN population must maintain central connectivity to remain viable (Spoendlin, 1971; Leake et al., 1992; Sekiya et al., 2003). However, interpretation of these studies is difficult because damage to SGN axons and/or the cochlear blood supply occurs and probably contributes to the neuronal loss. No experiments prior to our own have attempted to assess the dependency of the mammalian SGN population on central connectivity during early postnatal development.
In contrast to peripheral lesions, the effects of central deafferentation depend upon the age of the animal. In adult animals, cochlear ablation results only in decreased soma size and dendritic arborization, but no cell loss, in the CN and AAN (Powell and Erulkar, 1962; Pasic et al., 1994). Cochlear ablation prior to the onset of hearing, however, causes neuronal death in the ipsilateral VCN and LSO (Trune, 1982; Moore, 1990, 1992), but not until the age of hearing onset. This suggests that a critical period concomitant with the initiation of sensory-evoked activity exists in these neuronal populations. The secondary neuronal loss that we observe in the AAN of Egr2; Atoh1CKO mice precedes the onset of hearing by more than one week, which occurs between P11 and P15 in mice (Friauf and Lohmann, 1999). This suggests that a mechanism independent of sensory-evoked activity is responsible for the secondary neuronal loss. We propose that this mechanism involves neurotrophin signaling.
Neuronal death in the SG and AAN of Egr2; Atoh1CKO mice could be secondary to the loss of trophic factors normally supplied by the AVCN. BDNF, Ntf3 and their receptors (Ntrk2 and Ntrk3, respectively) are expressed during embryonic and early postnatal development in the cochlea and in regions of the auditory brainstem (Ylikoski et al., 1993; Schecterson and Bothwell, 1994; Hafidi et al., 1996; Pirvola et al., 1997). Both neurotrophins are necessary for maintenance of the SGN population, as Ntf3 null mice lose about 85%, BDNF null mice lose 7–15%, and BDNF/Ntf3 double-null mice lose all of their SGNs (Jones et al., 1994; Ernfors et al., 1995; Rubel and Fritzsch, 2002). Combined disruption of the cognate BDNF and Ntf3 receptors (Ntrk2 and Ntrk3, respectively) also results in complete loss of SGNs (Silos-Santiago et al., 1997), and incomplete deficiencies of these receptors lead to variable amounts of regional SGN loss (Fritzsch et al., 1998). In the AAN, an unusual twist in this scenario is that the trophic support would have to occur in an anterograde rather than retrograde fashion. There is precedence for this phenomenon, as anterograde actions of Ntf3 have been described in the central nervous system, primarily in the visual system (von Bartheld et al., 1996; Wang et al., 2002; Butowt and von Bartheld, 2005; von Bartheld and Fritzsch, 2006).
Other trophic influences could also be involved in the maintenance of neurons in the SG and AAN. For instance, the ablation of the CN could negatively impact the propagation of spontaneous activity between these regions. Spontaneous activity in the developing auditory system begins during embryogenesis (Lippe, 1994; Jones et al., 2001), but its role in development and trophism in these cell populations remains unknown. However, spontaneous activity has been demonstrated to be important for the correct differentiation and maintenance of neuronal subclasses in other regions of the central nervous system (Hanson and Landmesser, 2004). Other forms of trophism stemming from direct neuronal contact might also be important for the maintenance of these cell populations. Further experiments are necessary to determine the relative contributions of these potential trophic interactions on the developing auditory system.
Our data provide further support for the evolutionary conservation of Atoh1 function in the organization of the auditory system as a whole. In Drosophila, atonal is necessary and sufficient for specifying sensory neurons found in chordotonal organs (CHOs), which act as proprioceptors in the fly cuticle (Jarman et al., 1993). An array of CHOs make up Johnston’s organ, a structure in the second antennal segment that is specialized for hearing; atonal-null flies lack this structure and are subsequently deaf. In mammals, there is recent evidence that Atoh1 may be expressed at low levels in a subset of SGNs (Matei et al., 2005), although the role this expression plays in this cell population is unknown. In contrast, Atoh1 is highly expressed by inner ear hair cells (which detect the auditory signal) and by neurons of the CN, LSO and MSO. These cell groups contribute to the afferent auditory pathway, require Atoh1 for their specification, and their presence is necessary for SGN survival. In addition, we show that brainstem AAN neurons depend upon the presence of Atoh1-lineal neurons in the CN for their maintenance. So, although the sites of Atoh1 expression have changed through evolutionary time, the gene is still required for the organization of both the peripheral and central auditory pathways.
In summary, we demonstrate that conditional deletion of Atoh1 from rhombomeres 3–5 results directly in deletion of neurons from the CN, LSO and MSO and in secondary loss of neurons in the SG and brainstem AAN. Thus, Egr2; Atoh1CKO and Hoxb1; Atoh1CKO mice represent novel model systems to examine perturbations of the central auditory system. Our findings may have relevance for understanding the extent to which malformations of and/or prenatal insults to the brainstem in human patients impair hearing, and may help to provide insight into why cochlear implants in these patients sometimes prove unsuccessful.
We thank Drs. Patrick Charnay (Egr2Cre) and Mario Capecchi (Hoxb1Cre) for generously sharing mouse lines with us; Dr. William Brownell, Dr. Sharyl Fyffe and members of the Zoghbi lab for thoughtful discussions of the data and critical reading of this manuscript; and Bobbie Antalffy, Nancy Ao, Gabrielle Schuster and Sukeshi Vaishnav for technical assistance. Mouse behavior testing was supported by the Baylor MRDDRC Neurobehavioral Core; in situ hybridization was done by the Baylor MRDDRC In Situ Hybridization Core (NIH HD024064); and confocal work was done at the Baylor MRDDRC Imaging Core, Case Western Reserve University Neurosciences Imaging Center, and at Creighton University in part through a gift from the Roy J. Carver foundation to BF. SMM is funded by NIH K08NS053419, JSO by NIH K08DC006671, BF by NIH R01DC005590, and HYZ is an investigator in the Howard Hughes Medical Institute.