Background: Hearing restoration through hair cell regeneration will require revealing the dynamic interactions between proliferation and differentiation during development to avoid the limited viability of regenerated hair cells. Pax2-Cre N-Myc conditional knockout (CKO) mice highlighted the need of N-Myc for proper neurosensory development and possible redundancy with L-Myc. The late-onset hair cell death in the absence of early N-Myc expression could be due to mis-regulation of genes necessary for neurosensory formation and maintenance, such as Neurod1, Atoh1, Pou4f3, and Barhl1. Results: Pax2-Cre N-Myc L-Myc double CKO mice show that proliferation and differentiation are linked together through Myc and in the absence of both Mycs, altered proliferation and differentiation result in morphologically abnormal ears. In particular, the organ of Corti apex is re-patterned into a vestibular-like organization and the base is truncated and fused with the saccule. Conclusions: These data indicate that therapeutic approaches to restore hair cells must take into account a dynamic interaction of proliferation and differentiation regulation of basic Helix-Loop-Helix transcription factors in attempts to stably replace lost cochlear hair cells. In addition, our data indicate that Myc is an integral component of the evolutionary transformation process that resulted in the organ of Corti development. Developmental Dynamics 242:132–147, 2013. © 2012 Wiley Periodicals, Inc.
Hearing loss is a major epidemiologic problem worldwide that greatly affects hundreds of millions of peoples' quality of life. Two major forms of hearing loss exist: conductive and sensorineural. While both pose problems to individuals, sensorineural hearing loss (SNHL) is irreversible due to the damage of sensory hair cells in the inner ear. SNHL varies from mild to profound thus allowing for the success of modern day cochlear implants that function through remaining auditory neurons. Despite massive improvements, cochlear implants are still unable to fully recapitulate normal hearing (Roehm and Hansen, 2005; Chang and Fu, 2006; Bent et al., 2009; Gifford and Revit, 2010; Peterson et al., 2010; Renton et al., 2010; Colletti et al., 2011). Thus, while cochlear implants improve patients' auditory function and quality of life, the only permanent and perfect hearing restoration treatment is the regeneration of the damaged or lost organ of Corti to reform the “world's best hearing aid” (Puligilla and Kelley, 2009).
The organ of Corti is composed of hair cells, supporting cells, and neurons, all of which need to be properly formed and maintained for hearing to occur. In patients with short-term hearing loss, it is possible that only hair cells would need to be replaced; however, in patients with long-term hearing loss, the entire organ of Corti may be lost (Izumikawa et al., 2008). Regardless, therapy may require the ability to recapitulate critical developmental steps of the organ of Corti in vivo to manipulate a naive stem cell, stem-cell-like cell, or an induced post-mitotic inner ear cell towards each individual cell fate (Kopecky et al., 2011). To accomplish this, it may be necessary to elucidate more critically the embryonic developmental molecular network that initially formed the organ of Corti. Many of the early developmental steps that define the otic placode, inner ear axis specification, organ of Corti patterning, and even the neurosensory precursor populations, remain ill-defined.
The initial precursor population size of the organ of Corti is unknown; however, the end-product of development is roughly 15,000 hair cells in humans, divided into three rows of outer hair cells and one row of inner hair cells surrounded by several types of supporting cells. This organization is uniform along the several millimeters of the organ of Corti that stretches for over two complete turns from base to apex forming a sound frequency distribution map along its length. The stereotyped pattern of the organ of Corti across all mammals indicates that development is under tight genetic control and is not a stochastic process. To date, numerous studies have revealed that this developmental process requires diffusible factors that define the boundaries of the organ of Corti (Morsli et al., 1998; Pirvola et al., 2000; Pauley et al., 2003; Pan et al., 2011; Groves and Fekete, 2012) and a set of transcription factors that, first, determine the proliferation of the precursor populations, second, differentiate subpopulations of the precursor pool into hair cells, third, form cell–cell interactions to stabilize the cell fate of hair cells and supporting cells, and, last, provide long-term support to maintain the viability of this fragile and complex organ of Corti (Fritzsch et al., 2011). Each of these steps is interconnected to the steps before it as mouse models with mutations in upstream genes produce cumulative effects of downstream abnormalities. Therefore, perhaps the most important step for not only the proper development of the inner ear, but also for hair cell regeneration, is the highly complex modulation of proliferation, arguably the least understood part of ear development. Our current study focuses on the molecular interaction between the proliferation of neurosensory precursor populations, differentiation of these neurosensory precursors into hair cells, and how this delicate interaction may affect long-term maintenance of the organ of Corti.
Proliferation is the act of guiding a cell through the highly redundant and tightly regulated cell cycle. It is regulated through multiple cell cycle checkpoints and through both genetic and epigenetic mechanisms. Manipulation of proliferation throughout the ear in many species has been attempted through either exogenous mitogens such as the EGFs and FGFs (Zheng et al., 1997; Montcouquiol and Corwin, 2001; Witte et al., 2001) or through direct cell cycle regulation (Chen et al., 2003; Mantela et al., 2005; Sage et al., 2006; Oesterle et al., 2011). Various studies have been able to force continued proliferation of embryonic precursor cells or restart later proliferation of either supporting cells or, in rare cases, hair cells (Liu et al., 2012). Neither method was able to consistently provide long-term success as a common outcome of hair cells formed after manipulation of proliferation was cell death (Chen et al., 2003; Mantela et al., 2005; Sage et al., 2006; Weber et al., 2008; Oesterle et al., 2011). Other manipulations simply proved ineffective after a certain stage in development (White et al., 2006). Perhaps a node integrating many of these upstream signals (i.e., EGFs and FGFs) that synergize into important cell cycle control (i.e., Cyclins, pRB, and E2Fs) may mitigate these side effects and could achieve late induction of proliferation of organ of Corti cells. Indeed, the Myc node is upstream of many cell-cycle-regulating genes and possibly could provide insight into the underexplored balance between proliferation and differentiation during development (Conacci-Sorrell and Eisenman, 2011; Young et al., 2011). Unlike any other proto-oncogene or cell-cycle regulatory gene previously analyzed (Pauley et al., 2006; Laine et al., 2007; Rocha-Sanchez et al., 2011), the Myc's are bHLH transcription factors that bind specifically to E-boxes that are part of the promoters of target genes. Three bHLH transcription factors, Neurogenin1, Neurod1, and Atoh1, are essential for the differentiation of neurons and hair cells in the ear (Fritzsch et al., 2010) and these class II bHLH genes bind to a nearly identical class of E-boxes as the Mycs. Furthermore, immediately downstream to Myc are Inhibitors of DNA binding (Ids), which are known to directly inhibit class II bHLH transcription factors (Fritzsch et al., 2006; Jones et al., 2006) through interactions with the E-proteins needed for E-box binding of class II bHLH factors (Bhattacharya and Baker, 2011). Additionally, two genes, Eya1 and Six1, are sufficient for neuronal and hair cell formation in addition to bHLH expression (Ahmed et al., 2012a, 2012) and may also regulate Myc expression. Thus, the Mycs appear to be a key node that balances proliferation and differentiation and, therefore, manipulations of the Mycs may not only affect the level of proliferation in the ear, but also the onset of differentiation. Until recently, Myc function in the ear had been unstudied.
We have previously explored the effect of N-Myc in the inner ear using Pax2-Cre to conditionally delete N-Myc (Kopecky et al., 2011). A parallel study (Dominguez-Frutos et al., 2011) confirmed that N-Myc mutants had a severe reduction in size, abnormal morphogenesis, including a truncated cochlea with multiple rows of hair cells, a fused utricle, saccule, and base of the cochlea, and the loss of the horizontal canal (Kopecky et al., 2011). N-Myc and L-Myc are mostly co-expressed and are detected as early as embryonic day (E) 8.5 (Dominguez-Frutos et al., 2011). N-Myc and L-Myc are subsequently restricted to prosensory and then exclusively expressed in the sensory epithelia (Dominguez-Frutos et al., 2011; Kopecky et al., 2011). Despite the aberrant morphology, the Pax2-Cre N-Myc conditional knockout (CKO) mice developed both cochlear as well as vestibular hair cells; however, cochlear hair cells were progressively lost starting around postnatal day (P) 14, whereas vestibular hair cells were retained until at least 9 months of age (Kopecky et al., 2011). These mice were deaf and suffered from partially compensated ataxic gait (Kopecky et al., 2012a). From these initial studies, it was apparent that N-Myc and L-Myc but not C-Myc were expressed strongly in the inner ear and that this co-expression may have a potential redundant function during development. Furthermore, expression of both N-Myc and L-Myc in hair cells after birth coupled with the late loss of cochlear hair cells suggested a potential late-onset effect of N-Myc and L-Myc in long-term hair cell maintenance; an observation that could have significant translational impacts (Kopecky et al., 2011). We, therefore, tested this hypothesis by knocking out N-Myc and L-Myc in hair cells by using the previously described Atoh1-Cre (Matei et al., 2005) and the Pax2-Cre (Ohyama and Groves, 2004). Atoh1 upregulation will be delayed than that of Pax2. Atoh1 is specifically expressed in the differentiated hair cells whereas Pax2 is expressed in the pro-sensory region of the otic placode as early as E8.5. Delayed deletion of N-Myc and L-Myc in hair cells using Atoh1-Cre would test whether the hair cell loss seen in the Pax2-Cre N-Myc CKO mice was due to a late-onset effect of the Mycs specifically in hair cells. However, the late expression of N-Myc and L-Myc in hair cells likely does not play a direct role in long-term hair cell viability as there is no loss of cochlear hair cells in delayed deletion (Kopecky et al., 2012c). There is also no additional loss of cochlear hair cells in Atoh1-Cre N-Myc L-Myc double conditional deletion mouse even with the administration of ototoxic drug (Kopecky et al., 2012c). This suggested that the survival of cochlear hair cells is possibly due to the balance between proliferation and differentiation that is set up by very early expression of N-Myc and L-Myc during development. We test this hypothesis by knocking out N-Myc and/or L-Myc with Pax2-Cre and, subsequently, assessing important markers of both proliferation and differentiation throughout the time course of hair cell formation.
In the present study, we consider the relative compensatory abilities of N-Myc and L-Myc during development in conditional knock outs of N-Myc and/or L-Myc using Pax2-Cre, the overall loss of proliferation in the Pax2-Cre N-Myc L-Myc double (d) CKO mice, and how the loss of both N-Myc and L-Myc affect the balance between proliferation and differentiation. Our data indicate that the function of L-Myc can be compensated by that of N-Myc but L-Myc cannot compensate the function of N-Myc. However, the Pax2-Cre N-Myc L-Myc dCKO mice have a more severe cochlear phenotype than in the previously described Pax2-Cre N-Myc single CKO mice and are never viable, thus limiting postnatal investigations of long-term hair cell viability.
In this report, we compared the roles of N-Myc and L-Myc using different combinations of mutant alleles by the Pax2-Cre-mediated recombination such as Pax2-Cre N-Myc f/+ L-Myc f/f, Pax2-Cre L-Myc f/f, Pax2-Cre N-Myc f/f L-Myc f/+, and Pax2-Cre N-Myc f/f L-Myc f/f mice. However, heterozygosity of N-Myc and L-Myc do not result in a phenotype. Likely, they do not enhance any effect while combined with the homozygosity of other Myc gene (i.e. no difference between Pax2-Cre N-Myc f/+ L-Myc f/f and Pax2-Cre L-Myc f/f). We, therefore, termed Pax2-Cre N-Myc f/+ L-Myc f/f or Pax2-Cre L-Myc f/f mice as “L-Myc CKO,” Pax2-Cre N-Myc f/f L-Myc f/+ mice as “N-Myc CKO,” and Pax2-Cre N-Myc f/f L-Myc f/f mice as “dCKO” for the simplicity of the genotype in the rest of the report. Pax2-Cre N-Myc f/f mice as described in Kopecky et al. (2011) and Dominguez-Frutos et al. (2011) will be still written out to avoid confusion. N-Myc f/+ L-Myc f/+ mice without Cre are used as “control.”
Combined Loss of N-Myc and L-Myc Results in a More Disrupted Inner Ear Than Loss of Either N-Myc or L-Myc Alone
We generated three-dimensional reconstructions using confocal microscopy allowing for non-destructive optical sectioning of high-resolution images and easy three-dimensional manipulations and quantification of control, L-Myc CKO, N-Myc CKO, and dCKO ears from E11.5 to E18.5 to assess the development of the inner ear in the absence of either N-Myc or L-Myc or both. From E11.5 to E18.5, control development was identical to that previously described using the scanning Thin-Sheet Laser Imaging Microscopy (sTSLIM) technique (Kopecky et al., 2012b). At E11.5, the control inner ear was an otocyst with all axes defined. By E12.5, the ventral elongation of the cochlea became apparent. At this age, the dorsal vestibular canals began forming and the endolymphatic duct was prevalent. From E13.5 to E16.5, the inner ear grew and obtained a more mature form such that by E16.5, all sensory recesses were able to be uniquely identified with the formation of the ductus reuniens and utriculosaccular foramen (Fig. 1, row 1). In the L-Myc CKO developmental series, we knocked out L-Myc; however, no noticeable morphologic changes were apparent compared to age-matched controls (Fig. 1, row 2). However, when we assessed the N-Myc CKO developmental series, where the inner ear developed in the absence of N-Myc, there were dramatic developmental abnormalities as early as E11.5 (Fig. 1, row 3). The inner ear was drastically reduced in size and both the ventral and dorsal compartments were disrupted. No prominent saccular recess formed and the cochlear elongation was stunted, with a prominent circularization at the apex, sometimes completely disconnected from the middle turn, as seen in E18.5. Neither the ductus reuniens nor the utriculosaccular foramen formed, and thus, there was no morphologically distinct utricle, saccule, and base of the cochlea, but rather a continuum of these epithelia. Despite the lack of effect when L-Myc was knocked out alone in the L-Myc CKO mice, when L-Myc was knocked out in conjunction with N-Myc in the dCKO mice, the loss of L-Myc appeared to have an additive effect (Fig. 1, row 4). When both Mycs were knocked out, there were increased developmental disruptions in the vestibular canals. In the Pax2-Cre N-Myc f/f mice, the horizontal canal was absent, but neither the anterior nor posterior canals were noted to be distorted (Kopecky et al., 2011). Here, in the dCKO mice, in addition to the absence of horizontal canal, the anterior canal was always found to be affected. The anterior canal appeared much thinner with an impartially reabsorbed segment. In summary, the L-Myc CKO mice had no noticeable variation from control littermates, the N-Myc CKO mice had an identical phenotype to the Pax2-Cre N-Myc f/f mice, and the dCKO mice had the most severe defects.
In addition to assessing overall morphogenetic changes, three-dimensional reconstructions allowed quantification of inner ear endolymphatic space changes (Fig. 2). Overall, there was minimal endolymphatic space change in early development despite the vast morphological changes and overall growth of the control inner ear (Fig. 1). After E15.5, the endolymphatic space greatly increased coinciding with the formation of the ductus reuniens and utriculosaccular foramen (Fig. 2). Similar findings were noted in the L-Myc CKO ears that were nearly identical to control with minimal change until after E13.5 and increased after E15.5. These values were consistent between the confocal-based three-dimensional reconstructions and previous reconstructions using sTSLIM (Kopecky et al., 2012b). However, while compared between the N-Myc CKO and the control, a noticeable decrease in endolymphatic space was observed at each time point starting from E10.5 to E18.5 (Fig. 2). Despite the reduction in volume in the N-Myc CKO mice, there was little endolymphatic space increased after E15.5, which remained reduced compared to control littermate (Fig. 2). Lastly, we quantified the volumetric changes that occurred in the dCKO ears. The overall endolymphatic space in the dCKO ears significantly decreased (P < 0.05) compared to both control and L-Myc CKO ears during the entire embryonic development (Fig. 2). It was also reduced in comparison to the N-Myc CKO ears (Fig. 2). Therefore, it is obvious that the loss of L-Myc alone had minimal effect on overall morphogenesis of the ear; however, the loss of L-Myc in combination with N-Myc produced a greater effect than the loss of N-Myc alone.
Consistent with the pattern of the other mutant ears, there was minimal change in endolymphatic space until E15.5 in dCKO ears, which after E15.5 had increased but the relative ratio remained reduced compared to control. Interestingly, in all of the four mouse lines, endolymphatic space began to increase at the same time, suggesting that this increase is unrelated to the loss of N-Myc or L-Myc and is not dependent upon the proper formation of the inner ear, rather most notably on the formation of the ductus reuniens and utriculosaccular foramen.
N-Myc Is More Important Than L-Myc for the Formation and Functionality of the Inner Ear
At P0, all areas expressing Pax2 (inner ear, kidneys, and cerebellum) were substantially reduced in size in both the N-Myc CKO and dCKO mice compared to the L-Myc CKO mice, which itself showed only a small reduction in size compared to control (Fig. 3). Areas not expressing Pax2, such as the forebrains, were not reduced in size in any of the mice. In addition to the added size reductions, the dCKO mice were 100% lethal shortly after birth whereas some N-Myc CKO and all control and L-Myc CKO mice were viable. The reason why all the dCKO mice die is unclear but indicates a crucial compounding effect that likely is compensated for by either of the Myc transcription factors in the single knockout mice.
There were a number of defects observed while comparing the control and the dCKO mice. Most importantly, the cochlea was abnormally formed, notably the extreme base and apex were affected most (compare Fig. 4A and D). The base lost its high-frequency hook region (compare Fig. 4A’ and D’) and the apex was truncated and contained a ball-like formation of cells (compare Fig. 4A’’ and D’’) as shown with the three-dimensional reconstructions. Recognizable hair cells were found in newborn dCKO shown with Myo7a immunohistochemistry (Fig. 4B, B’, F, F’). Similar to the Pax2-Cre N-Myc f/f mice (Dominguez-Frutos et al., 2011; Kopecky et al., 2011), the dCKO mice had a fusion of the utricle and saccule as well as a partial fusion or abnormal tapering of the base of the cochlea (compare Fig. 4B’ and F’). The sections from the apex of the dCKO mice revealed a vestibular-like sensory epithelium sitting on periotic mesenchyme with a uniform set of vestibular-like hair cells (Fig. 4C, C’, E–E”). The hair cells at the most apical tip resembled type 1 and type 2 vestibular hair cells (black arrow in Fig. 4E and E’) and those closer to the middle turn resembled more cochlear type hair cells (white arrow in Fig. 4E and E’’’) and sat on periotic space comparable to a scala tympani.
While dCKO mice were lethal shortly after birth precluding behavioral studies, three N-Myc CKO mice survived up to P21 and were assessed for long-term functionality of remaining hair cells using auditory brainstem response (ABR). No responses could be elicited upon administration of the click stimuli or at any frequency tested (data not shown) consistent with Pax2-Cre N-Myc f/f mice previously described (Kopecky et al., 2012a). However, when quantifying the same parameters for the L-Myc CKO mice (Fig. 5A), statistically significant differences (P < 0.05) were noted in almost all ABR responses (except at 24 and 28 kHz). Behavioral gait performance was implemented using the Noldus Catwalk System to assess vestibulo-motor performance (a combined function of inner ear and cerebellum) through a number of gait performance assessments. Specifically for gait performance, we measure regularity, which quantitatively accounts for how consistently a mouse walks within a normal footfall pattern as we previously described (Kopecky et al., 2012a). While there was no gross morphological defect, the L-Myc CKO mice had significantly lower gait regularity compared to control littermate (P < 0.015; Fig. 5B). Three N-Myc CKO mice were also used for the gait performance assessment but these mice were unable to walk in the Catwalk system due to severe ataxia (data not shown). The N-Myc CKO mice contrasted with the L-Myc CKO mice as N-Myc CKO mice had severely disrupted hearing (no response) and balance (ataxic) consistent with their noted abnormalities (Kopecky et al., 2012a), whereas L-Myc CKO mice had some defect in both hearing and gait (Fig. 5) despite their near normal development (Fig. 1-3).
Proliferation is Severely Truncated After the Loss of Both N-Myc and L-Myc
In both the N-Myc CKO and dCKO mice, the growth of the entire inner ear was stunted with abnormal neurosensory differentiation (Fig. 1-3). The dCKO mice were not viable past P0, but the P21 N-Myc CKO were viable but were completely deaf despite initial formation of hair cells, consistent with previous data showing the loss of inner ear cochlear hair cells after P14 in the absence of early embryonic expression of N-Myc (Kopecky et al., 2011, 2012a). We have shown that this loss was not a result of late-onset effects of N-Myc or L-Myc (Kopecky et al., 2012c) and must be due to an early insult, possibly due to a disruption in the timing between proliferation and differentiation. To this end, proliferation was assessed at E13.5 with the thymidine analog 5′-Ethynyl-2′-deoxyuridine (EdU), which was injected into pregnant dams 8 hr prior to sacrifice. At E13.5, cells at approximately 15% of the distance from the extreme base toward the apex (termed “base”) along the presumptive organ of Corti were assessed for proliferation markers EdU as well as CyclinD1. In the control, EdU (Fig. 6A) and CyclinD1 (Fig. 6B) were seen throughout the entire basal region. At approximately 50% of the distance from base to apex (termed “middle”), EdU (Fig. 6C) and CyclinD1 (Fig. 6D) were seen only in the more basal portion of the middle turn whereas cells showed no signs of proliferation towards the apex. This suggested that in the control cochlea, organ of Corti precursor cells continued proliferating in the base and to a lesser degree in the middle, but not in the apex. This was consistent with apical cell-cycle exit around E12, middle turn cell-cycle exit at E13.5, and basal cell-cycle exit at E14.5 (Matei et al, 2005). On the other hand, in the dCKO mice, EdU (Fig. 6E, G) and CyclinD1 (Fig. 6F, H) were seen only in the base, which was severely reduced, and were absent in the middle turn compared to control. This showed that at E13.5, only cells in the base were continuing proliferation whereas cells in the middle turn and apex had likely exited the cell cycle (Fig. 6). Together, it appeared that in the absence of both N-Myc and L-Myc, organ of Corti cells exited the cell cycle approximately 1 day earlier than control organ of Corti cells.
In the Absence of N-Myc and L-Myc, Numerous Molecular Pathways Are Affected Including Differentiation and Cell Maintenance Genes Neurod1, Atoh1, Pou4f3, and Barhl1
Reduction in proliferation-associated genes in absence of both N-Myc and L-Myc
Proliferation in the inner ear is driven in part by N-Myc and L-Myc whereas differentiation of neurons and hair cells are a result of Neurod1 and Atoh1 upregulation, respectively, and the hair cell sub-types are fine-tuned by the interaction of these transcription factors (Jahan et al., 2010). Both N-Myc and L-Myc potentially regulate the timing of differentiation through two interdependent mechanisms, cell-cycle regulation and inhibition of differentiation. We assessed molecular changes in various genes of both of these pathways in the absence of Myc at E11.5 through E15.5 and at P0. In dCKO mice, N-Myc and L-Myc mRNA levels were drastically reduced as shown through qRT-PCR. In their absence, pRb, E2F4, and p27Kip1 (Table 1) were also reduced. Interestingly, CyclinD2 and E2F2 showed minimal change (Table 1). However, Id 2–3, which interact with bHLH class II E-proteins such as Atoh1 and Neurod1, were affected, with Id2 showing the greatest downregulation while Id3 was nearly unaffected (Table 1).
Timing of differentiation and maintenance genes are disrupted by altered proliferation
The qRT-PCR data from E11.5 to P0 mice revealed that Neurod1 mRNA levels were reduced in dCKO mice compared to control littermates (Table 1). Atoh1 mRNA levels in dCKO mice were notably reduced from E12.5 through E15.5 compared to control; however, at E14.5 and E15.5, Atoh1 levels in dCKO increased but never reached to control levels, suggesting a delayed and partial upregulation in the absence of N-Myc and L-Myc. At P0, Atoh1 levels were minimally reduced (Table 1). These qRT-PCR expression data of Atoh1 mRNA levels were validated through in situ hybridization (Fig. 7). At E13.5, no Atoh1 in situ expression was detected in the dCKO mice (Fig. 7B) whereas it was strongly expressed in the vestibular system in control (Fig. 7A). At E14.5, Atoh1 is expressed throughout the vestibular epithelia as well as in the basal-middle turn and extending towards the apex in the control (Fig. 7C). Atoh1 was restricted only in the vestibular epithelia in the E14.5 dCKO littermates (Fig. 7D), suggesting about a 1 day delay in Atoh1 upregulation. At P0, Atoh1 was expressed in all sensory epithelia including the cochlea in both control (Fig. 7E) and dCKO (Fig. 7F). Downstream hair cell maintenance genes of Atoh1 include Pou4f3 and Barhl1, were reduced in the dCKO mice detected by qRT-PCR (Table 1). A number of other organ of Corti specifying and defining genes were likewise downregulated, including Fgf8, Pax2, Gata3, and Sox2 (Table 1).
Summary of Control and Pax2-Cre N-Myc L-Myc Mutant Mice
Pax2-Cre N-Myc f/+ L-Myc f/+ mice were crossed with N-Myc f/f L-Myc f/f mice to assess functional redundancy of N-Myc and L-Myc and to help elucidate the potential mechanism behind initial hair cell formation followed by subsequent loss initially described in Pax2-Cre N-Myc f/f mice. In the absence of L-Myc, N-Myc appeared to be able to nearly fully compensate as L-Myc CKO mice developed normally with no apparent morphologic or neurosensory defects; however, these mice performed worse when assessed for hearing and balance performance, suggesting an incomplete ability of N-Myc to compensate for L-Myc loss. In the absence of N-Myc, L-Myc was unable to compensate for the loss as inner ears were severely reduced in size, with morphologic abnormalities, and viable mice could not hear. In the absence of both N-Myc and L-Myc, inner ears were more severely affected with a further reduction in endolymphatic space, additional cochlear and vestibular defects, and a 100% lethality rate shortly after birth. These dCKO mice stopped noticeable cell division in the middle turn of the cochlea around E13.5, approximately one full day earlier than control mice (Figs. 6, 8). This premature cell-cycle exit resulted in approximately one day delay in Atoh1 upregulation (Figs. 7, 8). Altered proliferation and differentiation resulted from the dramatic reduction in mRNA levels of N-Myc and L-Myc, which correlated with the altered timing and expression levels of Neurod1 and Atoh1, possibly through interaction with the IDs (Table1). These altered levels of Atoh1 corresponded with reduced levels of cell maintenance genes Pou4f3 and Barhl1. The apparent transformation of the apical tip into an almost vestibular macula-like epithelium and the complete loss of the basal hook region indicate that it is in particular the start and the stop of the organ of Corti hair cell formation that is affected by the absence of Myc signaling.
N-Myc Can Partially Compensate for the Loss of L-Myc Whereas L-Myc Cannot Compensate for N-Myc Loss
Both N-Myc and L-Myc are co-expressed strongly in the inner ear, with both detectable by E8.5 (Dominguez-Frutos et al., 2011; Kopecky et al., 2011). The expression domains of N-Myc and L-Myc mostly overlap, suggesting that there is possible redundancy between these two Myc genes expressed in the ear. By crossing Pax2-Cre N-Myc f/+ L-Myc f/+ with N-Myc f/f L-Myc f/f, four unique phenotypes are generated, a control, one that knocked out L-Myc completely, one that knocked out N-Myc completely, and one that knocked out both. In assessing the L-Myc knockouts, there seems to be no obvious morphologic abnormalities as these mice developed nearly identically to control, even if combined with N-Myc heterozygosity (Fig. 1). However, in the absence of L-Myc in Pax2-expressing regions at P0, the inner ear, kidney, and cerebellum are slightly reduced in size. Furthermore, functional studies of L-Myc CKO mice at P21 showed a slight reduction in auditory and vestibular functioning (Fig. 5). In contrast, N-Myc CKO mice have severe morphologic abnormalities throughout development, are reduced in size, and have a complete loss of hearing. While these abnormalities are worse in the absence of both N-Myc and L-Myc, it is obvious that the loss of L-Myc alone has only minimal effect on overall development and functionality whereas loss of N-Myc greatly alters the formation of the inner ear. Thus, N-Myc appears to play a much more pervasive role in inner ear development whereas L-Myc seems to play only an auxiliary function that can be mostly overcome through N-Myc expression. The neonatal death of the dCKO mice indicates that certain crucial aspects of development are non-redundant suggesting a limited but essential function of L-Myc, which remains to be determined as it likely is masked by N-Myc as long as this gene is normally expressed.
Proper Timing and Expression Levels of bHLH Transcription Factors Necessary for Differentiation Are Dependent on Proliferation
The roles of the Mycs throughout the body and their interactive cascades regulating proliferation are ancient (Young et al., 2011) and have been well studied (Conacci-Sorrell and Eisenman, 2011). However, the Mycs have been minimally studied in the inner ear and this is the first study attempting to elucidate the molecular networks involved with Myc signaling in the inner ear. Importantly, our data suggest that the alteration of proliferation through the Mycs leads to disruption of cellular differentiation. In the absence of N-Myc and L-Myc, proliferation is greatly reduced and downstream Id2 and CyclinD1 are also reduced. Contrary to initial expectations, while this downregulation of Id2 may lead to a disinhibition of class II bHLH transcription factors (Fritzsch et al., 2006), such an effect would be compromised due to the downregulation of both Neurod1 and Atoh1 expression. Neurod1 is expressed as early as E9.5 in the otic vesicle but is restricted to the vestibulo-cochlear ganglion by E10.5 and is essential for differentiation of neurons and interacts with Atoh1, necessary for hair cell formation (Jahan et al., 2010). Neurod1 levels never reach control levels in the dCKO mice. Atoh1, which is necessary for hair cell differentiation and forms a feedback loop with Neurog1 (Bermingham et al., 1999; Matei et al., 2005; Raft et al., 2007; Kirjavainen et al., 2008; Pan et al., 2011), is downregulated until E13.5 in dCKO mice after which Atoh1 slightly increases but remains below control levels. This transient downregulation of Atoh1 coincides with decreased levels of downstream hair cell maintenance genes Pou4f3 and Barhl1. Pou4f3 (Xiang et al., 2003; Clough et al., 2004; Chellappa et al., 2008; Masuda et al., 2012) and Barhl1 (Chellappa et al., 2008) are both well known to play roles in long-term hair cell maintenance (Pauley et al., 2008). It has been previously shown that level and timing of Atoh1 (Jahan et al., 2010; Pan et al., 2012) and expression of Pou4f3 and Barhl1 are all necessary for proper hair cell formation and long-term maintenance. The Pou4f3 promoter region is only partially characterized but is known to contain multiple binding sites including several E-boxes (Masuda et al., 2011). While we cannot say with certainty whether the progressive hair cell loss after initial hair cell formation is due to the delayed Atoh1 upregulation or the subsequent reduction of Pou4f3 and Barhl1, it is clear that by altering N-Myc and L-Myc, molecular pathways initiating and maintaining differentiation are disrupted, showing that in the ear, proliferation and differentiation are interconnected, at minimum through the Myc node, possibly by limiting the expression levels of Pou4f3 and Barhl1 expression, crucial genes for long-term hair cell maintenance. Cochlear hair cell death could also potentially disrupt the endocochlear potential (Cable et al., 1993; Tran, 2002) resulting in failure of formation of the ductus reuniens. It is conceivable that this results in a change in endocochlear potential to approximate the vestibular endopotential, potentiating hair cell loss.
N-Myc and L-Myc Regulate the Development of the Inner Ear Through a Partially Understood Molecular Network
In the absence of N-Myc and L-Myc, a number of abnormalities result as the end-product of inner ear development, two of which are a reduction in proliferation and a progressive loss of hair cells; these abnormalities likely are related to the interaction between proliferation and differentiation. However, a much more complicated picture arises when the remainder of the morphologic and neurosensory abnormalities associated with the dCKO mice is considered. Beyond the issue of long-term viability, the reduction of the high-frequency end of the basal “hook” region in the cochlea and the transformation of the apex into an unusual organization with hair cells of a mixed vestibular/cochlear phenotype indicate that the Myc bHLH transcription factors are likely affecting not only proliferation but also play into the cell fate decision-making process, possibly through interfering with Atoh1 signaling, a gene known to affect directly hair cell type differentiation in interaction with Neurod1 (Jahan et al., 2010).
Alternatively, the truncated development may lead to altered interaction of proliferating/postmitotic cells that results in the observed changes on the basal and apical part of the organ of Corti. To sort out the possible mechanisms by which some of these defects arose, we assessed several candidate genes. Fgf8 is expressed in the prospective otic placode by E8 and in the ventral otic placode by E9 and is important for the induction of the otic placode as its loss results in abnormal otic placode morphology (Ladher et al., 2005; Hertzano et al., 2010). It is later expressed in inner hair cells (Jacques et al., 2007; Jahan et al., 2010, 2012). Fgf8 is progressively downregulated in the dCKO shown by the qRT-PCR data (Table 1) as well as previously with in situ hybridization in Pax2-Cre N-Myc f/f mice (Kopecky et al., 2011). Another gene, Pax2, is expressed by E8.5 in the otic placode and in the medial sensory and non-sensory components of the cochlea at birth. Pax2 initially defines the pre-otic field and is induced by Fgf signaling and is dependent on Shh (Riccomagno et al., 2005; Bouchard et al., 2010). The dCKO mice have greatly reduced levels of Pax2. Another gene early expressed in the otic placode is Gata3 and in its absence, the otocyst remains cystic (Karis et al., 2001; Duncan et al., 2011). In the absence of N-Myc and L-Myc, Gata3 is downregulated. Another gene, Sox2 is expressed in the ventral neural region at E9.5 and later in hair cells and supporting cells of the cochlea (Dabdoub et al., 2008; Nichols et al., 2008). Sox2 mutants do not form a defined prosensory domain and have disorganization of hair cells. Sox2 inhibits Atoh1 expression but promotes Prox1 expression (Kiernan et al., 2005; Dabdoub et al., 2008; Mak et al., 2009). In the absence of N-Myc and L-Myc, Sox2 is also downregulated, indicating alterations in the fate determination of sensory epithelia. These data suggest that prosensory specification is changed as a direct consequence in the absence of Myc TFs. Future work needs to establish the interaction of Mycs on the promoter regions of these various factors to provide a rational basis for how the observed molecular changes tie into the obvious morphological changes. Additionally, while this appears to be a cell-autonomous effect, conclusive findings may require the use of inducible ear specific Cre lines or at minimum a ubiquitously expressed Cre to test how the timing of Myc deletion affects expression of these genes. However, targeted inducible lines with exclusive ear expression are not yet available and ubiquitously expressed lines such as Ros26ER-Cre may themselves have confounding effects due to widespread deletion of Mycs and a likely limited viability.
Mycs Are Necessary for the Formation of Therian Novelties in the Cochlea
Comparative data show that among tetrapods, only therians have high-frequency hearing above 10 kHz in the hook region (Muller et al., 2005) and have lost the lagena at the apical tip of the cochlear duct, instead showing a coiled cochlea (Webster et al., 1992; Fritzsch et al., 2011; Duncan and Fritzsch, 2012). Our data suggest that Myc-mediated proliferation and hair cell fate regulation may play a role in this evolutionary process. Clearly, the premature termination of proliferation in dCKO mice results in a complete loss of the basal hook region, the high-frequency mapping part, unique to the therian organ of Corti. Likewise, there is vestibular-like transformation at the apical tip of the shortened cochlear duct that has multiple hair cells sitting on top of periotic mesenchyme, typical for vestibular maculae, instead of a periotic space, which is an essential feature of the organ of Corti. Interestingly, disassociation of the scala tympani leads a vestibular-like basal turn in the Lmx1a mutant (Nichols et al., 2008), similar to what apparently occurs in the apex in the dCKO mice. Data are now needed to establish the evolutionary recruitment of the Mycs to the growing cochlea, which likely happened in the therian ancestor after their split from monotremes that lack both the high-frequency reception (Fay, 1988) and retain the ancestral lagenar macula at the tip of their cochlear duct (Fritzsch, 1987; Jorgensen and Locket, 1995).
N-Myc and L-Myc are mostly co-expressed in the inner ear. The early conditional embryonic loss of both genes using Pax2-Cre results in a severely disrupted ear representing a more profound phenotype compared to the loss of N-Myc alone. The combined losses of both N-Myc and L-Myc not only reduces proliferation and growth of the ear but also leads to subsequent alteration of differentiation, most notably through the reduction and delay of Atoh1 and downstream hair cell maintenance factors Pou4f3 and Barhl1. Morphologic changes are most striking in the base and the apex of the developing cochlea. Importantly, the loss of L-Myc alone has minimal effect on inner ear development suggesting a greater importance of N-Myc in mammalian inner ear development. The current study in conjunction with previously published studies detailing N-Myc and L-Myc in the inner ear illustrates the importance of proliferation control to proper neurosensory differentiation early in development. Any attempt to form hair cells, regardless of precursor source, may ultimately need to account for the interaction of proliferation genes such as the Mycs with differentiation genes such as Atoh1 and possibly other upstream genes such as Eya1 and Six1. Increased efforts should be made to more fully elucidate the molecular networks balancing these two interdependent processes for a therapeutically safe application in organ of Corti reconstitution.
Mice and Genotyping
A double conditional knockout line was generated by combining Pax2-Cre mice (Ohyama and Groves, 2004) with N-Myc floxed mice (Jackson Labs, Bar Harbor, ME; B6.129-N-Myctm1Psk/J) and L-Myc floxed mice (donated by Dr. Eisenman). Tail biopsies were used for genomic DNA and polymerase chain reactions for genotyping were performed using the following primers (N-Myc: IMR6727 5′gtcgcgctagtaagagctgagatc 3′ IMR6729 5′ cacagctctggaaggtgggagaaagttgagcgtctcc 3′ Cre: 1 5′ cctgttttgcacgttcaccg 3′ 2 5′ atgcttctgtccgtttgccg 3′ IMR42 5′ ctaggccacagaattgaaagatct 3′ IMR43 5′ gtaggtggaaattctagcatcatcc 3′). Animal care and usage was in accordance with the approved University of Iowa Institutional Animal Care and Use Committee (IACUC) guidelines for the use of laboratory animals in biological research (ACURF nos. 0804066 and 1103057).
The Noldus Catwalk System® consists of a walkway floor containing a sheet of glass encased with a light. P21 mice were placed in a corridor 60 cm long by 10 cm wide and allowed to move freely. A mounted camera captured a 40-cm × 10-cm field from below the walkway. As the mouse entered the field of view, a run was initiated and stored on a computer. If the mouse exited the field of view within 10 sec and there was less than a 60% variation in speed, the run was compliant. Numerous compliant runs were acquired for each mouse and were termed a trial. Between each trial, the walkway barrier was removed and the glass was cleaned using standard glass cleaner. All mice were tested under similar conditions. After each trial, the acquired data were classified. Classification is the semi-automated process of entering which limb induced the recorded print. Data points for each mouse are an average of three compliant runs. The catwalk system has many gait performance parameters. Regularity measures the “normalcy” of gait and how consistent the contralateral limbs are in contact with the glass walkway.
Auditory Brainstem Response (ABR)
Mice were anesthetized with 0.025mL/g of the anesthetic Tribromoethanol (Avertin®) and after a surgical level of anesthesia was induced, needle electrodes were inserted subcutaneously in the vertex, slightly posterior to the pinna, and in the contralateral hind-limb. A speaker was placed 10 cm from the pinna of the left ear and computer-generated stimuli were given in an open field environment in a soundproofed chamber. Clicks were presented and responses were averaged across 512 presentations using Tucker-Davis Technologies (Alachua, FL) System hardware running BioSig® Software. Recorded signals were band-pass and notch-filtered (300 Hz–5 kHz, 60 Hz). The sound level was decreased in 3-dB steps from a 96-dB sound pressure level until there was no noticeable response. Identical set-up but a tone pip program was given from 2 kHz through 28 kHz incrementally with a stepwise decrease in amplitude at each frequency.
Mice were injected intraperitoneally (IP) with greater than 0.025 mL/g of the anesthetic Tribromoethanol (Avertin®) and after ocular and pedal reflexes ceased, 4% paraformaldehyde (PFA) was pumped continuously with a 30-gauge needle into the left ventricle. The right atrium was opened to facilitate clearing. After fixation, heads were hemi-dissected, placed in 4% PFA in a multiwell plate, and covered with parafilm for long-term storage.
Epoxy Resin Sections
Dissected ears were incubated for 2 hr with 2.5% glutaraldehyde and washed with 0.1M phosphate buffer three times over a period of 2 hr. Ears were reacted in 1% osmium tetroxide for approximately 1 hr, or until tissue was dark, and rinsed with 0.1M phosphate buffer. Samples were dehydrated with an ascending graded ethanol series and then incubated with a 1:1 100% ethanol and propylene oxide mixture. Samples were mixed with Epon 812 and propylene oxide overnight and then oriented and embedded with Epon 812 and placed in an incubator for 24 hr at 60°C. Ears were sectioned at 1 μm with a Leica (Bannockburn, IL) Ultratome and sections were retrieved from water bath and placed on heated slides in sequential order. They were allowed to bake onto the glass slide and stained for 1 min with Stevenel's Blue and imaged. Stained slides were rinsed with dH20 and allowed to dry. Sections were coverslipped using DMX mounting solution and allowed to dry overnight. Samples were imaged with a Nikon (Melville, NY) Eclipse 800 microscope and captured with Image-Pro.
Pregnant dams were injected intraperitoneally with 100 µg EdU/gram body weight of mouse. After 8 hr, the mothers were sacrificed and pups were fixed and stored in 4% PFA. Dissected ears were then reacted with Click-iT reaction cocktail according to manufacturer's instructions (Invitrogen C35002).
Ears were dissected in 0.4% PFA and defatted overnight in 70% ethanol. They were removed from ethanol and blocked for 2 hr in blocking solution (5% normal goat serum, 0.1% Triton X-100, in Phosphate buffered saline, PBS). Myo7a (1:200; Proteus Biosciences, Ramona, CA; 25-6790) and CyclinD1 (1:200; Thermo Scientific, Waltham, MA; RM-9104) were diluted in the blocking solution and incubated for 3 days at 4°C on a shaker. Ears were washed with PBS three times for 2 hr each and Alexa Fluor 633 anti-mouse or anti-rabbit secondary antibodies were diluted in blocking buffer and incubated for 2 days, covered with aluminum foil, in 4°C. Ears were incubated in Hoechst nuclear stain for one hour and then washed in PBS and mounted on a glass-slide in glycerol. Images were captured with the TCS SP5 multiphoton confocal microscope.
Ears were decalcified in 10% EDTA for 4 days with fresh EDTA changed daily. Ears were rinsed with PBS at least three times with each wash lasting about 3 hr. Ears were put into 70% EtOH overnight. Ears were at this point put in 100% EtOH for 2 hr and then placed 2 days in a 100% rhodamine isothiocyanate and EtOH mixture until tissue was very lightly stained. One hundred percent rhodamine EtOH mixtures were removed and clearing solution was added to 100% EtOH to obtain a 1:1 ratio. Clearing solution was made with five parts methyl salicylate to three parts benzyl benzoate (MSBB). The 1:1 100% EtOH to MSBB was replaced with 100% MSBB overnight at room temperature. Ears were then put in two changes of MSBB and imaged.
Ears were placed on a glass slide and Dow Corning (Midland, MI) Grease was applied to the glass slide to make a retainer for MSBB. Ears were placed in the MSBB bath and spacers were added to ensure absence of compression stresses after coverslip application. Using a TCS SP5 confocal microscope, Z-stacks of 3–5-μm increments were obtained.
Segmentation and three-dimensional renderings
Z-stacks were loaded into Amira Version 5.4. All endolymphatic spaces were segmented in each section. Segmented images were resampled (2 × 2 × 1), surface generated (unconstrained), smoothed (20 iterations), and the surface was viewed. Quantitative measurements were obtained through the three-dimensional measurement functions within the Amira software. Protocol can now be viewed in full detail (Kopecky et al., 2012d).
Embryos from E11.5 through E15.5 were retrieved from the uterus in an RNAse-free environment. Embryos were placed into RNAlater for storage. P0 pup ears were immediately dissected and placed into RNAlater. RNAlater fixed inner ears were then dissected in RNAlater and homogenized and RNA was extracted using the Qiagen RNeasy Plus Mini Kit (Qiagen; Valencia, CA) followed by cDNA synthesis with Invitrogen (Carlsbad, CA) Super Script III. Due to the inability to dissect out sensory epithelia from the inner ear at E11.5, we used the entire inner ear for all ages (a dissected ear using only sensory epithelia later would provide biased results from earlier entire ears). cDNA was then mixed with a master mix consisting of Roche (Mannheim, Germany) Master Mix, water, a Roche Universal Probe Library Probe designed as intron spanning from the Roche Website (https://www.roche-applied-science.com/sis/rtpcr/upl/index.jsp?id=UP030000), and primer pairs (from Integrated DNA Technologies, IDT; Coralville, IA). All primer pairs and probes were verified with serial RNA dilutions with standard curves created to assess reverse transcription inhibition. Similar dilutions were performed with cDNA to quantify the minimal concentration of cDNA for each reaction. Ninety-six-well plates with three technical replicates were run on the Roche Light Cycler 480 housed in the Carver Center for Genomics at the University of Iowa. For each recorded gene and time point, at least three biological replicates were run following the established MIQE protocol (Bustin et al., 2009). Statistical analysis of qRT-PCR data was performed through either the Roche Light Cycler 480 SW 1.5 software or through Microsoft Excel. Fold change was calculated by normalizing the difference of target Cp values to reference Cp values of both positive control and unknown. For Student's t-tests, statistical significance was set at the 0.05 level unless indicated otherwise.
In Situ Hybridization
In situ hybridization was performed using the Atoh1 anti-sense RNA probe (provided by Huda Y. Zoghbi) labeled with digoxigenin. The plasmids containing the cDNAs were used to generate the RNA probe by in vitro transcription. Mice were perfused and fixed in 4% PFA after anaesthetized with Avertin. The ears were dissected in 0.4% PFA and dehydrated and rehydrated in graded methanol series and then digested briefly with 20 μg/ml of Proteinase K (Ambion, Austin, TX) for 15–20 min according to the stages of the mice. Then the samples were hybridized overnight at 60°C to the riboprobe in hybridization solution containing 50% (v/v) formamide, 50% (v/v) 2× saline sodium citrate (Roche), and 6% (w/v) dextran sulphate. After washing off the unbound probe, the samples were incubated overnight with an anti-digoxigenin antibody (Roche Diagnostics GmbH) conjugated with alkaline phosphatase. After a series of washes, the samples were reacted with nitroblue phosphate/5-bromo, 4-chloro, 3-indolil phosphate (BM purple substrate, Roche Diagnostics), which is enzymatically converted to a purple-colored product. The ears were mounted flat in glycerol and viewed in a Nikon Eclipse 800 microscope using differential interference contrast microscopy and images were captured with Metamorph software.
Heads were hemisected and ears removed from the head. Lateral images of the ear were taken and width (anterior to posterior) and height (ventral to dorsal) measurements were acquired. Cerebellum cross-sectional area from the hemisected head was also measured identical to the inner ear. The forebrain was measured from the hemisected head and the cross-sectional area of the midline was measured. Lastly, kidneys were removed from the retroperitoneal space and cross-sectional height from inferior to superior pole and width from medial to lateral were assessed.
We thank Dr. Peter Santi and his lab, notably Shane Johnson and Heather Schmitz, for their assistance in developing the protocol for sTSLIM image processing, which we adapted for confocal imaging, as well as their guidance in Amira usage. We thank Dr. Thomas Schimmang and his lab for their helpful discussions of the Pax2-Cre N-Myc CKO mice. We thank Drs. T. Ohyama and A. Groves for providing the Tg(Pax2-Cre) line. We thank Jackson Lab for providing the N-Myc f/f mice and Dr. Knoepfler for donating them. We also thank Dr. Eisenman for donating the L-Myc f/f mice. The Leica TCS SP5 confocal microscope was purchased in part with a grant from the Roy J. Carver foundation. Grant funding for Bernd Fritzsch was provided through the NIH and NIDCD R01-DC055095590. Funding for Benjamin Kopecky was provided by the CTSA, grant UL1RR024979. We thank the NIH P30 for grant DC010362, making the ABR equipment and the Noldus Catwalk System available as part of the supported core facilities. We thank the University of Iowa Carver College of Medicine, Medical Scientist Training Program, and Office of the Vice President for Research for support.