Canal Cristae Growth and Fiber Extension to the Outer Hair Cells of the Mouse Ear Require Prox1 Activity

Background The homeobox gene Prox1 is required for lens, retina, pancreas, liver, and lymphatic vasculature development and is expressed in inner ear supporting cells and neurons. Methodology/Principal Findings We have investigated the role of Prox1 in the developing mouse ear taking advantage of available standard and conditional Prox1 mutant mouse strains using Tg(Pax2-Cre) and Tg(Nes-Cre). A severe reduction in the size of the canal cristae but not of other vestibular organs or the cochlea was identified in the E18.5 Prox1Flox/Flox; Tg(Pax2-Cre) mutant ear. In these mutant embryos, hair cell differentiated; however, their distribution pattern was slightly disorganized in the cochlea where the growth of type II nerve fibers to outer hair cells along Prox1 expressing supporting cells was severely disrupted. In the case of Nestin-Cre, we found that newborn Prox1Flox/Flox; Tg(Nestin-Cre) exhibit only a disorganized innervation of outer hair cells despite apparently normal cellular differentiation of the organ of Corti, suggesting a cell-autonomous function of Prox1 in neurons. Conclusions/Significance These results identify a dual role of Prox1 during inner ear development; growth of the canal cristae and fiber guidance of Type II fibers along supporting cells in the cochlea.


Introduction
The mammalian inner ear is composed of the cochlea that mediates the auditory function, and the vestibule that mediates the gravitational and angular acceleration sensing. In mammals, six epithelial sensory patches found in the cochlear and vestibular regions of the inner ear mediate auditory and vestibular functions: the organ of Corti is the sensory patch found in the cochlea and three cristae and two maculae are the sensory patches of the vestibule. Each of these sensory patches includes mechanosensory hair cells and non-sensory supporting cells. Both of these cell types originate from epithelial progenitors that become specified as prosensory precursors. According to their position in the ear, these prosensory patches will give rise to the definite vestibular or cochlear sensory patches. Cells in those sensory patches ultimately assume final fates as either hair cells (e.g., inner and outer hair cells in the cochlea) or a variable number of non-sensory supporting cells (distributed between hair cells). While the molecular machinery governing the development of hair cells has received much attention [1,2] far less is known about the molecular basis of cell fate decision in supporting cells [3,4]. In the mammalian cochlea, at least five unique types of supporting cell can be identified: Pillar cells, Deiter's cells, Hensen cells, Claudius cells and inner sulcus cells [3,5].
Interestingly, in addition to the developing retina [15,20] and spinal cord [25], Prox1 expression was also identified in another sensory organ; i.e., the developing ear of zebrafish [26], chicken [27], and mice [28,29]. By taking advantage of available standard and conditional Prox1 mouse mutant strains [30,31], we have now determined that Prox1 is an important new player during the development of the mammalian vestibular and auditory systems. We demonstrate that in the canal cristae, lack of Prox1 function affects the overall growth of these vestibular sensory epithelia. In contrast, in the cochlea, absence of Prox1 disrupts stereotyped cellular organization and fiber guidance of Type II neurons apparently in a cell autonomous fashion.

Mice
Prox1 +/LacZ , Prox1 flox/flox , Atoh1, and Tg(Pax2-Cre) and Tg(Nes-Cre) mice have been previously reported [30,31,32,33,34,35,36]. The developmental stage of mouse embryos was determined by considering noon of the day the vaginal plug was detected in the pregnant dam as E0.5. All of the mouse experiments were approved by the Creighton University, University of Iowa, and St. Jude Children's Research Hospital Animal Care and Use Committees.

Detection of b-Galactosidase Activity
To detect b-gal activity, ears were dissected and X-gal staining was performed as described previously [37]. Whenever required, we enhanced the X-gal reaction using 2-photon photoactivation on whole mounts and sections [38]. In addition, we ran some ears without fixation to avoid any quenching of the b-galactosidase activity. Stained ears were mounted flat or alternatively, they were embedded in epoxy resin, sectioned (20 mm) and imaged using a compound lightmicroscope (Nikon Eclipse 800) and captured using a Coolsnap camera and Metamorph software. Some ears were processed for transmission electron microscopy and viewed in a Hitachi TEM as previously described [39]. Unfortunately, use of either Tg(Pax2-Cre) or Tg(Nes-Cre) leads to early postnatal lethality; therefore, we were not able to analyze the conditional mutant ear beyond P1.
Prox1 in situ hybridization. Whole mount in situ hybridization was performed using a riboprobe as previously described [15].

Immunohistochemistry
Primary antibodies were rabbit anti-b-gal (ICN), rabbit (Covance Research Products) anti-mouse Prox1 (Promega), rat anti-mouse btubulin (Sigma), Hoechst nuclear stain (Sigma), Myo VII (gift of T. Hasson, San Diego), Sox2 and BDNF (Invitrogen). Secondary antibodies were Alexa 488, 543, and 634-conjugated donkey antirabbit (Molecular Probes), Cy3-conjugated donkey anti-guinea pig (Jackson ImmunoResearch Laboratories), and Cy3-conjugated donkey anti-rat (Jackson ImmunoResearch Laboratories) were used Figure 1. The early onset of Prox1 expression is revealed by b-galactosidase expression and in situ hybridization. Whole mount bgalactosidase histochemcial reaction using X-Gal was performed in Prox1 heterozygous and nullizygous embryos. A. Starting at E11.0, a progressive upregulation of Prox1 is seen in the anterior (AC) and posterior (PC) canal cristae. B. By E13.5, expression is also detected in the horizontal canal crista (HC), the striolar region of the utricle (U), the canals and the endolymphatic duct (ED); expression in the saccule is barely detected (S). In the cochlea, upregulation of b-galactosidase expression is detected in the apex and decreases toward the base. Arrows indicate expression in anterior and posterior canal with their expression. C. Expression of b-galactosidase is identical in heterozygous and nullizygous mice with the exception that the signal is stronger in nullizygous mice. Faint b-galactosidase expression is also detected in the delaminating spiral ganglion neurons (SPG; C and insert in B,C). D. In situ hybridization shows at E14.5 expression in the canal cristae and the cochlea, but indicates a more prominent upregulation in the base at this stage. Only spiral ganglion sensory neurons are faintly positive for Prox1 in situ (SPG in D). E,F At postnatal stages, Prox1 expression remains in the canal cristae as revealed by in situ hybridization for Prox1 mRNA or X-Gal reaction, but does not show the extensive expression in the non-sensory parts of the canals as in earlier stages (insert in F). Bar, 100 mm. doi:10.1371/journal.pone.0009377.g001 predominantly on whole mounted microdissected sensory epithelia [40]. Sections and whole mounts were imaged using a confocal system (Zeiss LSM 510 or Leica SP5). Images were assembled into plates using CorelDraw software. Size of sensory epithelia was measured using ImagePro software on fully calibrated confocal images. PTI lipophilic tracers (NV Maroon) were used for afferent and efferent fibers [41]. Briefly, dyes were inserted into central targets or as small local injections and the fibers were filled with the diffusible dye, epithelia were microdissected and viewed with a confocal system (Zeiss LSM 510 or Leica SP5).

Quantification
In order to evaluate the qualitative effects of lack of Prox1 function on the growth of the vestibular epithelia we measured the length of the anterior canal crista and the utricle using the calibration setting of the Zeiss LSM 510 system in six flat mounted vestibular organs of Prox1 flox/flox ; Tg(Pax2-Cre) (E18.5 mutant) and Pax2-Cre (E18.5 control). Differences were evaluated for significance suing a T-test. We also counted the number of hair cells using Myo VII immunocytochemistry to identify hair cells and Hoechst nuclear staining to label the nuclei in three of these vestibular areas of control and mutant mice. Counting was done on flat mounts of three anterior canal cristae by grabbing a  confocal stack at 6 mm interval (slightly wider than the average nuclear diameter to avoid double counting). Shrinking or other counting artifacts should be equal but this procedure will slightly underestimate the total number of hair cells [42,43]. A nonparametric rank correlation test was used to assess statistical significance of cell counts.

Prox1 Expression in the Developing Inner Ear
Previous work using immunohistochemistry reported that Prox1 expression in the inner ear starts around E11.0 in three vestibular sensory patches and around E11.5 is highly expressed in the canal cristae and saccule of the sensory epithelia and weakly in the utricle [28]. Expression in the cochlea starts at around E14.5 in Pillar cells, Deiter's cells and outer hair cells, and also extends weakly to nonsensory parts of the ear [29].
In order to precisely compare the profile of Prox1 expression with the well known onset of hair cell proliferation [40,44] and differentiation [45,46] we took advantage of an available Prox1 heterozygous strain in which the b-galactosidase reporter gene was inserted in frame into the Prox1 genomic locus [31]. As shown in Fig. 1A, at E11.0 Prox1 expression was restricted to two X-gal positive patches corresponding to the anterior and posterior canal cristae. Two days later, an additional third patch of expression was seen in the region corresponding to the horizontal crista (Fig. 1B). Around E13.5 Prox1 expression also starts to be detected in what appears to be the striola region of the utricle and is barely detected in the saccule (Fig. 1B). It is only at around this stage that Prox1 expression starts to be detected in the cochlea where Prox1 upregulation begins broadly in the apex and expands toward the base (Fig. 1B). Prox1 expression is not restricted to sensory epithelia but is also found in the forming canals and the endolymphatic duct (Fig. 1B). In addition, Prox1 expression starts in the spiral ganglia around that time (Fig. 1C, insert). Prox1 in situ hybridization detects signal in the canal cristae but in the organ of Corti of the cochlear duct only at E14.5 (Fig. 1D). As indicated by X-gal staining (Fig. 1F) and in situ hybridization (Fig. 1E), Prox1 expression remains in the newborn canal cristae but is lost in the non-sensory part of the canal (insert in Fig. 1F).

Canal Cristae Are Smaller in Prox1-Null Embryos
Next, and in order to identify possible functional roles of Prox1 during the development of the ear, we characterized the inner ear of E14.5 Prox1-null embryos [31]. It was previously reported that Prox1-null embryos die at around E14.5 [31]. In agreement with the lack of Prox1 expression in developing sensory neurons at early developmental stages, no obvious phenotypic alterations were identified in the Prox1-null ears prior to E14.5 (Fig. 1B,C). This data indicated that Prox1 activity is not required for sensory neuron differentiation at these early stages.
As indicated above, high levels of Prox1 expression are detected in the developing canal cristae (Fig. 1). In agreement with this expression and as revealed by X-gal and Myo VII stainings [an early marker of hair cell differentiation; [47]], the size of the anterior canal cristae (AC) was clearly reduced in E14.5 Prox1-null embryos ( Fig. 2A-D, Fig. 3). The posterior canal cristae (PC) was similarly affected (data not shown) and the horizontal canal cristae (HC) was not as affected ( Fig. 2A-D). In addition to the high level of expression in the canal cristae, X-gal staining of E14.5 Prox1 heterozygous and nullizygous embryos confirmed that Prox1 expression was only transient and weak in the utricle ( Fig. 2A,B) and almost not detectable in the saccule (Fig. 1C). In situ hybridization verified that a weak but detectable signal persisted in the utricle at least until P1 (Fig. 1F) as previously described [29]. We determined that on average (N = 6), the size of the anterior canal cristae in Prox1-null embryos was 20% smaller (p,0.05; ttest) than in their heterozygous littermates (Fig. 3) (no differences in size were found between wild-type and Prox1 heterozygous littermates; data not shown). We also counted the number of hair cells and found that the anterior canal cristae of Prox1-null embryos had only about 605 (+265) hair cells compared to the control littermate that had about 913 (+278) hair cells (p,0.05).
To confirm and expand this observation indicating that removal of Prox1 activity affects the size of the vestibular sensory epithelia, we took advantage of a previously generated Prox1 conditional knock-out strain [30] to remove Prox1 activity from the inner ear in a time and tissue specific manner. To this end, Tg(Pax2-Cre) transgenic mice [33] were used to delete Prox1 from E9.0 onward in all cells of the ear, including all hair cells and sensory neurons [48]. Using this approach we also expected to overcome the early embryonic lethality of standard Prox1-null embryos [31]. Analysis of Prox1 flox/flox ;Tg(Pax2-Cre) conditional mutant embryos at E18.5 identified phenotypic alterations similar to those described in the E14.5 Prox1-null embryos; e.g., the size of the anterior cristae was significantly reduced (30% N = 6; p,0.05; T-test) (Fig. 2D, Fig. 3). Despite this size reduction, the overall shape and morphology of the cristae, and the formation of the non-sensory cruciate eminence (CE) were not affected in these mutant embryos (Fig. 2D, F). As indicated by Myo VII staining, no obvious gross morphological alterations were detected in the development and distribution of the vestibular hair cells of the canal cristae at these later stages (Fig. 2D, F). No obvious alterations in the distribution and morphology of supporting cells (indicated by Hoechst stained nuclei), or in the size of the utricle were identified in these conditional mutant embryos (Figs. 2,3). In E18.5 Prox1 heterozygous animals, expression as revealed with 2 photon photoactivation of the b-galactosidase reaction product [38], is found throughout all supporting cells of the canal cristae. In agreement with a recent report [29], at this stage Prox1 expression was also detected in some hair cells and non-sensory cells adjacent to the canal cristae (Fig. 2E9).
In summary, these initial results revealed that removal of Prox1 function from the developing ear resulted in a significant reduction in the size of the canal cristae.

Lack of Prox1 Function Results in Hair Cell Misalignment and Disrupted Type II Spiral Ganglion Cell Guidance
Previous work has shown that cell cycle exit of hair cells in the canal cristae starts around E11.5 [44]. Accordingly, Prox1 expression is detected prior and during cell cycle exit of hair cells and supporting cells of the canal cristae (Fig. 1). In contrast, in the cochlea Prox1 expression started to be detected in the cells of the apex at around E13.5; although, it was faintly expressed in cells near the base at this stage (Fig. 4A) and clearly is upregulated only after hair cells have exited the cell cycle [40].
Multiple rows of hair cells and supporting cells form initially as a short aggregate, but undergo convergent-extension movement to eventually form three rows of outer and one row of inner hair cells [49]. At around this stage of convergent extension, hair cells have already exited the cell cycle which progresses from the apex to the base of the cochlea between E11.5-E14.5 [40,44,45]. X-gal staining of Prox1 +/LacZ embryos and Prox1 in situ hybridization at different developmental stages revealed that in the cochlea, Prox1 expression progressed initially from the apex to the base (Fig. 1B,C;  Fig. 4A); a result suggesting that its expression is in cells that have already exited the cell cycle [45]. As shown in Fig. 4B-D, by E17.5, Prox1 expression is prominent throughout the cochlea and near the base is almost exclusively detected in the five supporting cells of the lesser epithelial ridge (the three rows of Deiter's cells and the two rows of pillar cells); only limited expression was seen in some outer hair cells and inner phalangeal cells (arrows in Fig. 4D, D9). This limited expression in inner phalangeal cells seen in the X-gal stained and photoactivated organ of Corti (Fig. 4C9, D9), is also observed when using Prox1 antibodies (Fig. 5A). At this stage, Prox1 expression in the apex is fainter and not organized into the five rows of supporting cells (Fig. 4E, E9; [29]. The prominent expression in supporting cells remained during postnatal stages, at least until P16 as shown by X-gal staining of Prox1 +/LacZ (Fig. 4F,G; [29]. In later stages, a faint Prox1 signal was also detected in sensory neurons (Fig. 4F). This signal was more prominent using in situ hybridization (Fig. 4H,I). We verified the expression of Prox1 as revealed by X-gal staining using in situ hybridization (inserts in Fig. 4A, C; Fig. 4H,I0 and immunocytochemistry. For unknown reasons, X-gal staining of Prox1 +/LacZ was easily lost after fixation in sensory neurons and could be demonstrated only in unfixed ears ( Fig. 1D and insert). We also verified the supporting cell and neuronal expression that was so obvious with in situ hybridization starting at E15.5 (Fig. 4A, insert; Fig. 4C, insert) with immunocytochemistry (Fig. 4J). Combined, all three techniques show a profound upregulation of Prox1 in supporting cells and sensory neurons (with the caveat of suppression of X-gal staining of Prox1 +/LacZ in sensory neurons following fixation).
In order to determine whether Prox1 expression in supporting cells (Fig. 4C,D, D9, J,K,L9) is an indication that its functional activity is required to control any developmental aspect of this cell type, we analyzed the cochlea of E18.5 Prox1 f/f ;Tg(Pax2-Cre) mutant embryos. Using this approach, Prox1 expression was extensively removed from the developing cochlea (Fig. 5A9). Although as discussed above (Fig. 2D, F), no obvious alterations in hair cell differentiation were observed, hair cells patterning was found to be occasionally disrupted. At this stage and as shown by Myo VII staining (Fig. 5B), wild-type hair cells exhibit the typical one row of inner hair cells and three parallel rows of outer hair cells. In the mutant littermates, inner and particularly outer hair cells appeared disorganized, misaligned, and containing extra rows near the apex (Fig. 5B9arrows). These results indicated that lack of Prox1 function did not affect hair cell differentiation (hair cell differentiation markers Myo VII and BDNF were normally expressed in the mutant hair cells; Figs. 5B, B9, 6); however, hair cell patterning was slightly defective. Light and electron microscopic radial sections confirmed the near normal development of hair cells and supporting cells but also some degree of disorganization of both cell types (Fig. 6).
These results indicate that Prox1 activity is not required for hair cell differentiation were further corroborated by the fact that Prox1 expression was not affected in E18.5 Atoh1-null embryos (Fig. 7) with defective hair cell differentiation [37,50]. These results also demonstrate that Prox1 expression is not dependent on Atoh1 or on hair cell differentiation consistent with recent reports, indicating autonomy of Prox1 expression from hair cell differentiation [4,51].
Together, these results suggested that in the organ of Corti, lack of Prox1 function promotes some limited phenotypic alterations in the overall patterning resulting in a slightly disorganized distribution of supporting and hair cells, including short extra rows of outer hair cells and misalignment of inner hair cells. Interestingly, direct comparison of Prox1 flox/flox ;Tg(Pax2-Cre) and wildtype littermates showed that the Prox1 in situ signal was somewhat weaker in the organ of Corti but unaltered in the sensory neurons (Fig. 4H,I). This could indicate that the disorganization of supporting cells may affect overall level of Prox1 expression.
Next, we analyzed whether these phenotypic alterations identified in supporting and hair cells affected nerve fiber growth. Previous work [52,53,54,55,56] showed that the stereotyped growth of Type II fibers toward outer hair cells is more advanced in the base and upper middle turn (Fig. 5C, D). Type II spiral ganglion cells extend first radially through the tunnel of Corti and then turn sharply toward the base to form three parallel rows in front of the three rows of Deiter's cells that are spaced between the three rows of outer hair cells (arrows Fig. 5C, D). We found that in Prox1 flox/flox ; Tg(Pax2-Cre) mutant littermates, the fibers also extended radially (arrows Fig. 5C9, D9). We found that all mutant nerve fibers extended beyond the first row of Deiter's cells (D1; Fig. 8B9, B0) and randomly turned at the 2 nd and 3 rd row of Deiter's cells (D2, D3; Fig. 9C,C9, C0). No obvious reduction in the density of the radial fibers was observed (Fig. 5C9, D9). These results were corroborated further by triple immunolabeling where supporting cells were identified by the use of Sox2 antibodies (Fig. 8). The normal organization of the Sox2-expressing supporting cells (green), BDNF-expressing hair cells (blue), and b-tubulin-expressing fibers (red) is shown in Fig. 8A-A0. In contrast, misaligned supporting and hair cells are seen in E18.5 Prox1 flox/flox ;Tg(Pax2-Cre) mutant littermates (Fig. 8B-B0, C-C0). These results and those obtained using electron microscopy (Fig. 7) suggested that in the absence of Prox1, the signaling mechanisms controlling where and to which direction fibers should turn is disrupted. Close comparison between wildtype (Fig. 5E), FGFR3 null mice (Fig. 5F) and E18.5 Prox1 flox/flox ;Tg(Pax2-Cre) show that type II afferents are disorganized in FGFR3 null mice (Fig. 5E.F), but that this disorganization is different from that seen in Prox1 flox/flox ;Tg(Pax2-Cre).
Next we investigated fiber growth in Prox1 flox/flox ;Tg(Nes-Cre) mice to evaluate possible spiral sensory neuron cell autonomous defects. Nestin (Nes), a neuronal stem cell marker, is expressed in developing sensory neurons of the ear [57]. We used a Tg(Nes-Cre) line [36] to conditionally delete Prox1. As seen in Fig. 9A, Prox1 expression was deleted in the spiral neurons but remained in the sensory epithelium (Fig. 9A, A9, A0). We traced the nerve fibers with lipophilic dyes or tubulin immunocytochemistry. Similar to what was observed in Prox1 flox/flox ;Tg(Pax2-Cre) pups, P1 Prox1 flox/flox ;Tg(Nes-Cre) mice showed severe disruption in the organization of Type II fibers (Fig. 9A,C,D,H). In the apex, where Type II fibers are growing out we could clearly identify that each fiber made an almost random turn to either the base or the apex (Fig. 9D,H) compared to the stereotyped decision of growth cones in wildtype (Fig. 9F,G), which always turn toward the base. Near the base, were Type II fibers have a longer trajectory at this stage, we find an intertwined mesh of fibers near the second and third row of Deiter's cells (Fig. 9C) instead of the very regular organization near all three rows of Deiter's cells (Fig. 9B). Given that our Prox1 antibody shows a clear, only somewhat interrupted signal in supporting cells (Fig. 9A9, A0, C,D), it seems that this disruption of fiber projection is predominantly due to the lack of Prox1 expression in sensory neurons in the Prox1 flox/flox ;Tg(Nes-Cre) conditional null mice, a signal which, according to our in situ hybridization data, is becoming increasingly prominent after E14.5 (Fig. 4H,I). Since the first Type II fibers are growing toward outer hair cells at around E16.5 [53,55], it appears that Prox1 upregulation coincides with the ability of Type II fibers to navigate their normal stereotyped trajectory. In the absence of Prox1 either in sensory neurons or in sensory neurons and supporting cells combined this ability is partially disrupted. However, Type II fibers may be able to reach the outer hair cells but extend beyond the first row of Deiter's cells thus might miss the first row of outer hair cells.
To better understand the inability of type II fibers to turn correctly, we investigated the outgrowth of fibers to the outer hair cells in E18.5 Prox1 flox/flox ;Tg(Nes-Cre) conditional null mice  (Fig. 10). We used different colored lipophilic dyes to trace small subsets of spiral sensory neurons from the cochlear nuclei [52]. To avoid confusion with the second fiber type that reaches the outer hair cells, the olivocochlear efferent system [58,59], we labeled these fibers with a differently colored lipophilic dye [60]. At this stage, only type II afferents grow to outer hair cells. In control mice all fibers navigated their way between supporting cells and turned invariably toward the base (Fig. 10. A-C). In contrast, in Prox1 flox/flox ;Tg(Nes-Cre) conditional null mice we found that the initial fiber growth was undirected, frequently stalled with branches in both directions or turned randomly toward the base or the apex (Fig. 10D-J). Absence of Prox1 protein disables recognition of directional signals during type II fiber growth.

Discussion
In this paper we identified Prox1 as a novel regulator of inner ear development and extent previous expression data [28,29]. We determined that this gene's activity is required for the proper growth of the canal cristae and correct fiber patterning of Type II afferents in the cochlea. Consistent with the transient low level of expression, no phenotypic alterations were identified in the gravistatic receptors organs (utricle and saccule).

Prox1 Regulates Canal Cristae Growth
During inner ear development, one of the earliest and more prolonged expression patterns of Prox1 was detected in the canal cristae. In this organ, the onset of Prox1 expression overlaps with that of other gene products such as Gata3 [61], Fgf10 [62], Foxg1 [63], Sox2 [64], Lmx1a [65] and Bmp4 [66], whose activities are essential for the formation and differentiation of the sensory epithelia. In general, gene inactivation of any of these factors resulted in the partial or total loss of the sensory canal cristae [67,68] or overgrowth [65]. In Prox1-null embryos the canal cristae did not exhibit any morphological alteration; however, their size was reduced. Accordingly, it could be speculated that Prox1 activity is necessary to maintain and expand the pool of neurosensory progenitor cells. Atoh1 is essential for hair cell differentiation [50] and Atoh1-null mice fail to differentiate hair cells and supporting cells [37]. Therefore, our finding that Prox1 expression remained normal in Atoh1-null ears, and that Prox1-null hair cells expressed typical hair cell markers eliminates the possibility that Prox1 was required for hair cell differentiation at the level of neurosensory progenitors. This does not rule out that misexpression of Prox1 in hair cells can result in their degeneration, as was recently shown for cochlea but not for vestibular hair cells [29].

Prox1 Regulates Fiber Guidance of Type II Spiral Neurons in a Cell Autonomous Way
Similar to what has been reported for the cell cycle kinase inhibitor p27 [45,69], the neurotrophin Bdnf [70,71] and the growth factor Fgf10 [62], Prox1 expression in the cochlea starts to be detected almost a day after hair cell precursors exited the cell cycle [40,45]. While Prox1 is not expressed in hair cell progenitor cells, it is expressed transiently in differentiating hair cells [29]. However, its continued expression in organ of Corti cells of Atoh1null mice [4,51], who have only hair cell precursors that fail to differentiate [37], indicates that at least the expression in supporting cells is not regulated by Atoh1 or other genes specifically expressed in differentiated hair cells (Fig. 7). Given that Prox1 expression persists at least until P26 in supporting cells [29], it is possible that this gene remains expressed after at least neonatal hair cell loss and its promoter could be used to drive molecular expression toward reconstitution of the a functional organ of Corti.
As previously reported [28], later during embryogenesis Prox1 expression is detected in the five supporting cells of the lesser epithelial ridge (Fig. 4). In these cells, lack of Prox1 function lead to subtle phenotypic alterations; e.g., defective alignment of hair cells and supporting cells (Fig. 5) However, major pathfinding defects Figure 10. Dye tracing reveals Type II fiber outgrowth problems in the organ of Corti of Prox1 flox/flox ; Tg(Nes-Cre) conditional mutant embryos. NV Maroon (green) was inserted into the cochlear nucleus and NV Orange (red) was inserted into the olivocochlear bundle to label a small population of afferents (green) and all efferents (red). Efferents show a similarly organized intraganglionic spiral bundles in wildtype (A-C) and Prox1 flox/flox ; Nes-Cre conditional mutant mice (D-J) and grow toether with afferents in radial fiber bundles (RF) to the organ of Corti. Note that at this stage only occasional efferents extent to outer hair cells. In contrast, type II afferents grow to the second or third row of outer hair cells (OHC) where they invariably turn toward the base (B,C). At this stage, none of the multiple type II afferents of Prox1 flox/flox ; Nes-Cre conditional mutant mice show this coordinated growth pattern. Instead, fibers grow randomly toward the base or apex but mostly seem to stall with multiple branches extending toward the base and the apex (F,G). IGSB, intraganglionic spiral bundle; OHC, outer hair cells; RF, radial fibers; SPG, spiral ganglion. Bar, 100 mm (A-D), 50 mm (E-H; inserts in A,A9, A0). doi:10.1371/journal.pone.0009377.g010 were identified in Type II spiral ganglion fibers. In this case, the turning of these fibers toward the base [52,55] was severely disrupted (Fig. 5,8,9,10). We found that in conditional null mutants fibers abnormally extended toward the second and third rows where they turned randomly instead of turning toward the base in front of each of the three rows of Deiter's cells. Radial fiber growth beyond the inner pillar cells was not affected. It is worth mentioning that pathfinding defects have been identified in the CNS of Prospero mutant flies [72].
While Prox1 is the first gene that plays a cell autonomous role in Type II pathfinding, at the moment it is not known how Prox1 affects fiber pathfinding of these neurons. It is known that Fgf8 and Fgf10 mediated activation of Fgfr1, 2b and 3 signaling participates in the differentiation of supporting cells of the lesser epithelial ridge [3,62,73,74,75,76], and Fgfr3 -null mice also exhibit short extra rows of outer hair cells [3,77] with some minor fiber disorganization that is clearly distinct from the Prox1 effects ( Fig. 5F,G), but where exactly Prox1 fits into these interactions remains to be determined.