The signals that induce the organ of Corti and define its boundaries in the cochlea are poorly understood. We show that two Notch modifiers, Lfng and Mfng, are transiently expressed precisely at the neural boundary of the organ of Corti. Cre-Lox fate mapping shows this region gives rise to inner hair cells and their associated inner phalangeal cells. Mutation of Lfng and Mfng disrupts this boundary, producing unexpected duplications of inner hair cells and inner phalangeal cells. This phenotype is mimicked by other mouse mutants or pharmacological treatments that lower but not abolish Notch signaling. However, strong disruption of Notch signaling causes a very different result, generating many ectopic hair cells at the expense of inner phalangeal cells. Our results show that Notch signaling is finely calibrated in the cochlea to produce precisely tuned levels of signaling that first set the boundary of the organ of Corti and later regulate hair cell development.https://doi.org/10.7554/eLife.19921.001
The mammalian cochlear duct is a sound transducer of exquisite mechanical sensitivity. Its sensory organ, the organ of Corti, contains mechanosensory inner and outer hair cells, surrounded by different types of supporting cells (Kelley et al., 2009; Basch et al., 2016). The organ of Corti is bounded by two populations of non-sensory epithelium, the inner and outer sulci. These three domains of the cochlear duct are induced in response to a gradient of canonical BMP signaling established by the expression of BMP4 in the future outer sulcus (Ohyama et al., 2010). High levels of BMP signaling promote outer sulcus differentiation, low or absent BMP signaling leads to formation of the inner sulcus, while intermediate levels of BMP signaling position the cochlear prosensory domain which will differentiate into the organ of Corti (Ohyama et al., 2010). Inner hair cells, which are the first cells to differentiate in the organ of Corti, arise at the boundary of the prosensory domain and the future inner sulcus, also known as Kölliker’s organ, in response to inducing signals (Groves and Fekete, 2012). Recent work has identified several candidate signals that initiate hair cell development, including the canonical Wnt signaling pathway, that promote hair cell differentiation (Shi et al., 2010; Jacques et al., 2012; Shi et al., 2014; Jansson et al., 2015) and other signals, such as Shh, that hold hair cell differentiation in abeyance until they are down-regulated (Bok et al., 2013; Tateya et al., 2013).
A central question in understanding the mechanism of signaling gradients is how a continuously changing, ‘analogue’ gradient of signaling is translated into a small number of distinct, ‘digital’ domains, and how the boundaries between these domains are established. The Notch signaling pathway has been implicated in the establishment of boundaries between developmental territories (Bray, 2006; Artavanis-Tsakonas and Muskavitch, 2010). For example, a zone of Notch signaling is established between the dorsal and ventral halves of the Drosophila wing imaginal disc (Baker, 2007). Here, the dorsal half of the imaginal disc expresses Notch, the glycosyltransferase enzyme Fringe and the Notch ligands Delta and Serrate, while the ventral territory expresses only Delta and Notch. Fringe proteins modify Notch receptors and ligands to increase the level of Notch signaling by Delta ligands and to attenuate Notch signaling by Serrate ligands (Rana and Haltiwanger, 2011; LeBon et al., 2014). Accordingly, the action of Fringe in the wing imaginal disc serves to attenuate Serrate-Notch signaling in the dorsal region of the disc (Rana and Haltiwanger, 2011), but permits a sharp boundary of Notch signaling at the boundary between dorsal and ventral halves in response to Serrate and Delta signals (Fortini, 2000). The situation in vertebrates is complicated by the presence of multiple Delta homologues (Dll1, 3 and 4) and two Serrate homologues, Jag1 and Jag2. Current evidence suggests that Fringe modification of Notch receptors tends to enhance signaling by Dll1 and Dll4 ligands and attenuate signaling by Jag1 and Jag2 (Hicks et al., 2000; LeBon et al., 2014).
We found that two Fringe genes, Lfng and Mfng, have an extremely dynamic expression pattern in the cochlea as the progenitor cells of the organ of Corti - the so-called prosensory domain – begin to differentiate into hair cells and supporting cells. Strikingly, we find these two genes are expressed transiently precisely where inner hair cells will differentiate, at the boundary between the prosensory domain and the non-sensory region of the cochlear duct known as Kölliker’s organ. Lfng and Mfng expression subsequently diverge as hair cells and their surrounding supporting cells differentiate. Our observations suggest that Notch signaling may act to first position the boundary between the future organ of Corti and Kölliker’s organ, and then subsequently regulate the correct formation of inner hair cells and their neighboring supporting cells. To test this, we systematically inactivated Lfng and Mfng, Notch receptors, Notch ligands, and other regulators of the Notch pathway in the developing cochlea. We find Notch signaling controls two sets of decisions at the edge of the organ of Corti. The first decision restricts the first differentiating inner hair cells and their associated supporting cells, the inner phalangeal cells, to the boundary with Kölliker’s organ. We find this fate decision is regulated by Fringe activity, requires low levels of Notch signaling and is extremely sensitive to changes in signaling strength. The second decision regulates the proportion of hair cells and supporting cells through previously characterized forms of lateral inhibition (Lewis, 1991, 1998; Kiernan, 2013). This fate decision does not require Fringe activity, requires higher levels of Notch signaling, and is much less sensitive to small changes in signaling strength. Our results suggest that qualitatively different forms of Notch signaling regulate different fate decisions during organ of Corti development.
Previous studies reported that Lunatic Fringe (Lfng) is expressed in the cochlea in Kölliker’s organ with the Notch ligand Jag1 before the formation of the first inner hair cells (Morsli et al., 1998; Murata et al., 2006; Ohyama et al., 2010; Basch et al., 2011). As the first hair cell progenitors differentiate near the base of the cochlea, they express Atoh1 and Manic Fringe (Mfng; Cai et al., 2013, 2015). To examine the changes in expression of these genes during cochlear development and hair cell differentiation, we examined the developing mouse cochlea between E13.5 and E15.5. We performed in situ hybridization for Lfng, Mfng and Atoh1 in adjacent serial sections (Figure 1A) and examined Jag1 expression in Lfng-GFP transgenic reporter mice from the GENSAT project (Gong et al., 2003; Geschwind, 2004; Heintz, 2004; Schmidt et al., 2013) in which GFP is expressed under control of a bacterial artificial chromosome containing the Lfng locus. We also examined Jag1 expression in Atoh1-GFP knock-in mice in which GFP is fused to the coding region of Atoh1 (Shroyer et al., 2007). The Atoh1-GFP fusion protein is expressed a little later than Atoh1 mRNA (Cai et al., 2013), but also provides a reliable indicator of differentiating hair cells.
We observed striking dynamic changes in the expression of Lfng, Mfng, Jag1 and Atoh1 in the cochlea between E13.5 and E15.5 as differentiation proceeds in a basal-apical direction (Figure 1A–C; summarized in Figure 1D). Jag1 is initially expressed in Kölliker’s organ with Lfng at E13.5 (Ohyama et al., 2010), and the expression of both genes then shifts into the differentiating prosensory domain and ultimately to supporting cells over the next 48 hr (Morsli et al., 1998; Murata et al., 2006). At E14.5, Lfng is down-regulated in much of Kölliker’s organ in the basal turn of the cochlea and is restricted to a column of differentiating Atoh1-expressing cells at the boundary of Kölliker’s organ and the prosensory domain (Figure 1A–C). Differentiating hair cell progenitors are observed in the mid-basal prosensory domain at the boundary of Kölliker’s organ starting at E13.5, where they express Atoh1 (Cai et al., 2013, 2015). Prior to E14.5, Manic Fringe (Mfng) was not detected in the cochlear duct, but at E14.5 and E15.5, it is co-expressed with Atoh1 in differentiating hair cells (Figure 1A; Cai et al., 2015).
Due to the basal-apical gradient of differentiation in the cochlea, we found that the most accurate picture of these dynamic expression changes could be obtained by examining serial sections along the length of the E15.5 cochlear duct. At this age, the apex of the cochlea contains the most immature cells, while the base of the cochlea represents the most mature differentiating hair cells and supporting cells (Figure 1A). We observed that Lfng and Mfng transiently co-localize at the site of Atoh1 expression at the boundary of Kölliker’s organ and the prosensory domain. We confirmed this co-localization of Lfng and Mfng by performing fluorescent in situ hybridization for Mfng in Lfng-GFP mice (Figure 1B). As hair cell and supporting cell differentiation continues, Fringe gene expression diverges once more, with Lfng and Jag1 becoming restricted to supporting cells (Morsli et al., 1998; Murata et al., 2006), and Mfng being restricted to hair cells. A more complete apical-basal series of serial in situ hybridization images are shown in Figure 1—figure supplement 1. The dynamic gene expression changes occurring along the apical-basal axis of the cochlea at E15.5 are summarized in Figure 1D.
The position of the transient stripe of Lfng- and Mfng-expressing cells at the border of Kölliker’s organ and the prosensory domain suggested these cells give rise to inner hair cells and/or inner phalangeal cells. To test this, we performed a fate mapping experiment with Lfng-CreER transgenic mice generated with the same bacterial artificial chromosome used to produce the Lfng-GFP transgenic line . We mated Lfng-CreER transgenic mice with Ai3 ROSA Cre reporter mice (Madisen et al., 2010) and administered a single dose of tamoxifen at E14. Since approximately 6–12 hr typically elapse between tamoxifen administration and the onset of recombination (Cai et al., 2013; Gridley and Groves, 2014), this time point allows us to map the fate of Lfng-expressing cells in the apical and middle regions of the cochlea that we observe at the prosensory domain border. Mice were sacrificed four days later at E18.5, and sections and whole mounts of the cochlea were stained for the presence of hair cells, pillar cells and the EGFP reporter. In the most apical regions of the cochlea, we observed most EGFP-labeled cells in Kölliker’s organ and very few in the inner hair cell region (Figure 2A,B; Figure 2—figure supplement 1), consistent with the expression of Lfng in this region between E12 and E14 (Morsli et al., 1998; Ohyama et al., 2010). In middle turn regions of the cochlea, we saw many EGFP-labeled inner hair cells and inner phalangeal cells (Figure 2A,B; Figure 2—figure supplement 1). In many cases, labeled inner hair cells were found adjacent to labeled inner phalangeal cells in whole mounts and sections. However, we observed very little labeling of other organ of Corti cell types in apical or middle turn regions (Figure 2A,B), suggesting that the strongly Lfng-expressing cells we observed at the border of the prosensory domain at E14-15 are progenitors of both inner hair cells and inner phalangeal cells. In basal regions of the cochlea, we now observed many labeled Deiters’ cells, some labeled inner pillar cells as well as labeled inner hair cells and inner phalangeal cells (Figure 2A). These results are consistent with the onset of expression of Lfng in all supporting cell types except outer pillar cells in the basal region of the cochlea at E15.5 (Figure 1A). We also occasionally saw labeled outer hair cells (Figure 2A,B), suggesting that some Deiters’ cell progenitors may differentiate into hair cells as Notch-mediated lateral inhibition establishes the precise pattern of hair cells and supporting cells in the organ of Corti. The transition from middle to basal turns of the cochlear duct was characterized by a gradual spreading of EGFP labeling into the Deiters’ cell and outer hair cell region (Figure 2A,B and Figure 2—figure supplement 1).
Our data showing that Lfng and Mfng are transiently expressed together at the boundary of Kölliker’s organ suggested they play a role in the differentiation of inner hair cells and their associated inner phalangeal cells at this boundary. We examined the cochleas of neonatal Lfng;Mfng double mutant mice and single Lfng or Mfng mutants. Lfng mutant mice have previously been shown to have no detectable cochlear phenotype (Zhang et al., 2000). Mfng mutant mice also had no cochlear phenotype (Figure 2C). In contrast, Lfng;Mfng double mutant mice showed significant numbers of supernumerary inner hair cells compared to controls (19.5 inner hair cells/100 µm compared to 12.3/100 µm in wild type controls; Figure 2C; Figure 3B). In contrast, we observed no significant changes in outer hair cell number (43.5 cells/100 µm versus 40.6 cells/100 µm in wild type controls; Figure 2C; Figure 3C). We were unable to detect expression of the third mammalian Fringe homologue, Radical Fringe (Rfng) in the cochlea, and Lfng;Mfng;Rfng triple homozygous mutants showed no significant difference in the number of supernumerary inner hair cells compared to Lfng;Mfng double mutant animals (Figure 3B,C).
The appearance of supernumerary hair cells is typically thought to occur by a failure of Notch-mediated lateral inhibition, whereby supporting cells trans-differentiate into hair cells, or their progenitors differentiate into hair cells instead of supporting cells (Lewis 1998; Kiernan, 2013). Since Fringe proteins act to modify Notch receptors and ligands to change their signaling properties (Rana and Haltiwanger, 2011), we examined the expression of supporting cell markers in our mutants. Surprisingly, we continued to observe p27kip1-expressing supporting cells lying beneath the supernumerary inner hair cells, which is not predicted by a classical lateral inhibition model (Figure 2C). To identify the type of supporting cells beneath the inner hair cells, we used antibodies to the glutamate/aspartate transporter GLAST encoded by the Slc1a3 gene and a nervous system-specific fatty acid binding protein, FABP7, both of which label inner phalangeal cells (Furness and Lehre, 1997; Saino-Saito et al., 2010; Zilberstein et al., 2012). We found both markers continued to be expressed in Lfng;Mfng mutants, and that supernumerary inner phalangeal cells were found adjacent to supernumerary inner hair cells (Figure 2D). Thus, in the absence of Fringe activity, both inner hair cells and their associated inner phalangeal cells were duplicated at the boundary of Kölliker’s organ and the prosensory domain. This suggests that the defects we observe in these mutant mice do not result from a failure of Notch-mediated lateral inhibition that has been previously demonstrated to distinguish between hair cells and supporting cells (Kiernan, 2013). Instead, it appears that the column of inner hair cell and inner phalangeal cell progenitors that transiently express Lfng and Mfng normally inhibit their neighbors in Kölliker’s organ from also adopting an inner hair cell/inner phalangeal cell fate.
Modification of Notch receptors by Fringe proteins can cause changes in the strength of Notch signaling (Haltiwanger and Stanley, 2002). Differentiating hair cells express three Notch ligands, Dll1, Dll3 and Jag2 (Lanford et al., 1999; Morrison et al., 1999; Hartman et al., 2007; Maass et al., 2015), of which Dll1 and Jag2 act as conventional Notch ligands, whereas Dll3 appears to act only by inhibiting Notch receptors in the same cell (Ladi et al., 2005; Geffers et al., 2007; Chapman et al., 2011). To determine whether the loss of Fringe proteins led to a reduction in Notch signaling delivered by differentiating hair cells, we compared the Lfng;Mfng phenotype to the cochlear phenotype seen in Jag2 mutants. Jag2 mutant mice display supernumerary inner hair cells (Lanford et al., 1999; Zhang et al., 2000; Figure 3A,B), and we also observed supernumerary inner phalangeal cells in these mice on the basis of p27kip1, GLAST and FABP7 immunostaining (Figure 3A). The Jag1 ligand is also expressed in the cochlear duct as the first hair cells differentiate (Figure 1C), and later in supporting cells (Morrison et al., 1999; Woods et al., 2004). Jag1 heterozygous mice also have increased inner hair cells (Kiernan et al., 2007; Moayedi et al., 2014), and we observed a similar phenotype of duplication of inner hair cells and inner phalangeal cells in Jag1 heterozygous mice (Figure 3A,B). We also confirmed that loss of both Lfng and Mfng led to a reduction in Notch signaling directly, by comparing N1ICD staining in Lfng;Mfng embryos to wild type controls (Figure 4A). In contrast to other regions of Lfng;Mfng mutant embryos such as the hindbrain (Figure 4A, inset), Lfng;Mfng mutants showed a reduction in N1ICD staining in the E15.5 cochlear duct.
We next asked whether the observed inner hair cell and inner phalangeal cell duplication phenotype could be recapitulated by a more general reduction in Notch signaling. To do this, we examined two mouse mutants that displayed a partial loss of Notch signaling in the cochlea. Pofut1 is an O-fucosyltransferase that adds fucose moieties to Notch receptors and ligands (Takeuchi and Haltiwanger, 2010), and whose activity generates the O-fucose substrate that is necessary for further N-glycosylation of Notch receptors by Fringe proteins (Shi and Stanley, 2003; Shi et al., 2005). Pofut1 null mutant mice die between E9.5 and E10.5 with an axial phenotype reminiscent of a Notch loss of function (Shi and Stanley, 2003), although different tissues can display differential sensitivities to loss of Pofut1 (Schuster-Gossler et al., 2009). Conditional deletion of Pofut1 in the cochlea with Pax2-Cre transgenic mice leads to a reduction, but not a complete loss of Notch signaling, as revealed by the presence of cleaved Notch1-intracellular protein in the nuclei of Pofut1 mutant supporting cells and the enhancement of the Pofut1 mutant phenotype by additional Notch inhibition with gamma secretase inhibitors in organ culture (Figure 3—figure supplement 1A,B). This incomplete loss of Notch signaling suggests that at least some Notch receptors lacking O-fucose glycans can still be chaperoned to the surface of progenitor cells in the Pofut1 mutant cochlea. Pofut1 conditional mutant mice also exhibited both supernumerary inner hair cells and inner phalangeal cells (Figure 3A,B), suggesting that this phenotype can be caused by a general reduction in Notch signaling. To confirm this, we also examined transgenic mice carrying a Cre-inducible dominant-negative form of MAML1, a nuclear co-factor responsible for the formation of transcriptional complexes between the Notch intracellular domain and the RBPJ co-factor (Kovall, 2008). Embryos derived by mating Pax2-Cre and Cre-inducible dnMAML1 mice (Tu et al., 2005) also showed only a partial Notch loss of function (Figure 3—figure supplement 1A,B) and displayed both supernumerary inner hair cells and inner phalangeal cells (Figure 3A). We found that Lfng;Mfng mutants, Jag2 mutants, Pofut1 conditional mutants and dMAML1 mutants all showed a significant increase in inner hair cells and the presence of extra inner phalangeal cells (Figure 3A,C), but we only observed an increase in outer hair cells in Jag2 mutants (Figure 3B; Lanford et al., 1999).
Our results suggested that the level of Notch signaling required to restrict the number of inner hair cells and inner phalangeal cells to a single row during cochlear development is very sensitive to manipulation, such that even a partial reduction causes duplication of the inner hair cell/inner phalangeal cell region. However, the persistence of inner phalangeal cells in our mutant lines (Figure 2B; 3A) suggest that Notch-mediated lateral inhibition between hair cells and supporting cells still occurs in our various mouse models when Notch signaling is reduced but not abolished. To test this, we examined Notch1 conditional null mutant mice which have been shown to produce a significant loss of lateral inhibition between hair cells and supporting cells in the cochlea (Kiernan et al., 2005). As expected, we observed a very large increase in both inner and outer hair cells, and a great reduction of many kinds of supporting cells (Figure 3A). Significantly, we saw no evidence for inner phalangeal cells in Notch1 conditional mutants on the basis of p27kip1, GLAST or FABP7 staining (Figure 3A), suggesting that the cochlear phenotype caused by a strong Notch loss of function is due to a combination of an increase in inner hair / inner phalangeal progenitor specification and the subsequent differentiation of these ectopic progenitors exclusively into inner hair cells.
The supernumerary inner hair cells and inner phalangeal cells observed when Notch function is reduced could arise by a change in cell fate within the cochlear duct, or alternatively by the prolonged proliferation of progenitors at the border of the prosensory domain and Kölliker’s organ. Indeed, previous studies have observed a small amount of prolonged proliferation of cochlear progenitors in Notch pathway mutants (Kiernan et al., 2005; Murata et al., 2009; Tateya et al., 2011). To distinguish between these possibilities, we administered EdU to pregnant female mice three times a day between E14.5 and E17.5 and collected embryos for analysis at E18.5. We performed these experiments in wild type, Pofut1 and dnMAML1 conditional mutant mice (Figure 4). We did not observe any significant differences in EdU labeling of outer hair cells, Deiters’ cells, pillar cells, inner hair cells or inner phalangeal cells (Table 1). We observed a modest but significant increase in EdU incorporation in dnMAML1 border cells immediately adjacent to inner phalangeal cells (66 versus 59 labeled cells; p=0.014). To test at what stage the supernumerary inner hair cells and inner phalangeal cells arose in Notch pathway mutants, we examined Atoh1 expression at the leading edge of hair cell differentiation in the apex of wild type and Pofut1 mutant cochleas at E15.5. We consistently observed doublets of Atoh1-expressing cells in this region of Pofut1 mutants, whereas single Atoh1-expressing cells were always observed at the leading edge of hair cell differentiation in wild type cochleas (Figure 4—figure supplement 1). These results suggest that the supernumerary inner hair cells are induced at the same time as their normal counterparts when Notch signaling is reduced in the cochlea.
To further demonstrate that the restriction of inner hair cell and inner phalangeal cell numbers was more sensitive to changes in Notch signaling than that required for the lateral inhibition of supporting cell fate by hair cells, we established an in vitro cochlear culture system in which we attenuated Notch signaling to different degrees. We used double transgenic mice carrying a GLAST-dsRed transgene (Regan et al., 2007) to label inner phalangeal cells (Figure 3—figure supplement 1C) and Atoh1-GFP reporter mice (Shroyer et al., 2007; Cai et al., 2013, 2015). We isolated cochleas from E14.5 double transgenic mice and cultured them for three days in the presence of different doses of the gamma secretase inhibitor DAPT or Notch1 blocking antibodies (Wu et al., 2010; Maass et al., 2015). We demonstrated the differential degree of Notch inhibition by measuring levels of Notch-responsive Hes and Hey genes in our cultures. We observed a modest down-regulation of Hes1 and Hes5 at intermediate doses of DAPT or anti-Notch1 antibodies, but no significant down-regulation of Hey1, Hey2 or Heyl (Figure 5B,E). However, at high concentrations of either inhibitor, all genes were significantly down-regulated with the exception of Hey2, which is much less sensitive to changes in Notch signaling in the cochlea (Doetzlhofer et al., 2009). At high concentrations of either inhibitor, we observed significant increases in both outer and inner hair cells, and a loss of inner phalangeal cells as revealed by the absence of GLAST-dsRed fluorescence (Figure 5A,D). This result is consistent with a loss of lateral inhibition between hair cells and supporting cells (Kiernan, 2013). In contrast, when we applied 100-fold lower concentrations of Notch1 antibody or 25-fold lower doses of DAPT to the cultures, we saw a smaller but still significant increase in the number of inner hair cells, but no significant increase in outer hair cell numbers (Figure 5C,F). Moreover, we continued to observe expression of the GLAST-dsRed transgene at these doses (Figure 5A,D), suggesting that inner phalangeal cells do not trans-differentiate into hair cells, but remain adjacent to the supernumerary inner hair cells.
The Notch signaling pathway acts at multiple stages to regulate the development of the inner ear. It was first proposed to regulate the production of cochleo-vestibular ganglion neurons, and later of hair cells and supporting cells by lateral inhibition (Lewis, 1991; Haddon et al., 1998; Lewis 1998). Notch signaling can also regulate the production of sensory patches though inductive signaling between Jag1 and Notch, leading to the up-regulation of Sox2 (Eddison et al., 2000; Hartman et al., 2010; Pan et al., 2010; Neves et al., 2011, 2013; Pan et al., 2013). In this study, we identify a new and unexpected role for Notch signaling in the ear: the positioning of inner hair cells and their associated inner phalangeal cells at the boundary of Kölliker’s organ and the prosensory domain. We show that this developmental decision is exquisitely sensitive to changes in Notch signaling. We observed a striking duplication of both inner hair cells and inner phalangeal cells in five different mutant mouse models and two organ culture systems that lower, but do not eliminate, Notch signaling in the cochlea. Unlike previous studies of Notch signaling in the cochlea, these phenotypes do not resemble defects in Notch-mediated lateral inhibition that is known to occur between hair cells and supporting cells (Kiernan, 2013). Only in conditions where Notch signaling is greatly reduced - such as in Notch1 mutants or at high doses of Notch blocking antibodies or gamma secretase inhibitors – do we see a conversion of duplicated inner phalangeal cells into inner hair cells, suggestive of a failure of lateral inhibition between hair cells and supporting cells. Our data, together with the careful analysis of Notch1 activation in the developing cochlea performed by Murata and colleagues (Murata et al., 2006) lead us to propose a new model for the development of the organ of Corti that involves two phases of Notch signaling. Because hair cell differentiation proceeds in a basal-apical direction over a period of several days (Chen et al., 2002; Lee et al., 2006; Cai et al., 2013), we discuss the timing of these events with respect to the basal region of the cochlea (Figure 6).
During the outgrowth of the cochlear duct, Jag1 and Lfng are expressed in cochlear progenitor cells (Morsli et al., 1998; Ohyama et al., 2010). Lfng activity is known to attenuate Jag1-Notch signaling between cells (Hicks et al., 2000; Rana and Haltiwanger, 2011; LeBon et al., 2014) which explains why these cochlear progenitor cells have been reported to receive low amounts of Notch signal (Figure 6A; Murata et al., 2006). We propose that starting at approximately E13, the cochlear duct receives hair cell-inducing signals that reach a peak spanning the boundary of Kölliker’s organ and the prosensory domain (Figure 6B). There is no consensus as to the identity of these signals; they may be positively acting signals such as Wnts (Shi et al., 2010; Jacques et al., 2012; Shi et al., 2014; Jansson et al., 2015) or may reflect the withdrawal of inhibitors of hair cell induction, such as Shh, which is expressed in the neurons invading the cochlear duct in this region and is cleared from the neurons in the same basal-apical gradient as hair cell differentiation (Bok et al., 2013; Tateya et al., 2013). In response to these hair cell-inducing signals, a column of prosensory cells adjacent to Kölliker’s organ begins to up-regulate early hair cell genes (Atoh1, Mfng, Jag2, Dll1), while still maintaining expression of Notch1, Lfng and Jag1 (Figure 6B, green cell). This single column of differentiating hair cell progenitors sends Notch signals to its neighbors in Kölliker’s organ that prevent them from responding to hair cell-inducing signals (Figure 6B). Our Lfng-CreER fate mapping data show that this column of cells will ultimately form inner hair cells and their neighboring inner phalangeal cells (Figure 2A,B).
We suggest this interaction sets the boundary of the organ of Corti and establishes a single row of inner hair cells and inner phalangeal cells. In our study, we experimentally perturbed this interaction by either reducing the amount of Notch signaling received by Kölliker’s organ cells (Pofut1 mutants or dnMAML1 mutants) or by reducing the amount of Notch signaling delivered by the adjacent column of prosensory cells (Jag1 and Jag2 mutants). Our observation that the duplication of this column is caused by reduction, but not elimination of Notch signaling in several mouse mutants and in cell culture (Figures 3 and 5) suggests this phase of Notch signaling involves moderate or low levels of Notch activity. We explain the effects of these perturbations on our model below.
Pofut1 mutants or dnMAML1 mutants cause a general reduction in Notch signaling received by the Kölliker’s organ cells. As a result, they can now respond to hair cell inducing signals, leading to the formation of a second column of cells that express hair cell genes (Atoh1, Mfng, Jag2, Dll1) and maintain expression of Notch1, Lfng and Jag1. This second column of cells ultimately differentiates into a second row of inner hair cells and inner phalangeal cells. Mutating Jag1 or Jag2 reduces the available pool of Notch ligands and hence reduces the amount of Notch signaling delivered to Kölliker’s organ, also causing a duplication phenotype. These results are summarized in Figure 7. Although we did not examine Dll1 conditional mutants in our study, the co-expression of Dll1 with Jag2 in hair cell progenitors, together with previous reports of Dll1 loss of function in the cochlea (Kiernan et al., 2005; Brooker et al., 2006) suggest these mutants will also have a duplicated row of inner hair cells and inner phalangeal cells.
Fringe proteins are known to regulate Notch signaling by making Notch receptors more sensitive to Delta-type ligands and less sensitive to Jagged/Serrate-type ligands (Haltiwanger and Stanley, 2002). We show that Lfng has a very dynamic expression pattern (Figure 1 and Figure 1—figure supplement 1) during cochlear development, but that it coincides with Mfng and Atoh1 precisely where inner hair cells and inner phalangeal cells differentiate at the boundary of the organ of Corti. We confirmed this observation by lineage tracing with Lfng-CreER mice (Figure 2A,B, Figure 2—figure supplement 1). Lfng;Mfng double mutants show a reduction in Notch signaling at this stage of overlap and display a duplication of inner hair cells and inner phalangeal cells (Figure 2C,D). These results are consistent with the column of progenitor cells transiently expressing Lfng, Mfng, Notch1, Atoh1, Dll1, Jag1 and Jag2 sending less Notch signal to their neighbors in Kölliker’s organ in Lfng;Mfng double mutants. How can we explain this result?
Previous studies suggest that when Notch ligands and receptors are present in the same cell, they attenuate each other’s activity, a phenomenon known as cis-inhibition (del Alamo and Schweisguth, 2009; Becam et al., 2010; del Álamo et al., 2011; LeBon et al., 2014). Fringe proteins have been shown to modulate cis-inhibition: in vertebrates, Lfng and Mfng proteins increase cis-inhibition between Notch and Delta ligands, but decrease cis-inhibition between Notch and Jagged ligands (LeBon et al., 2014). We therefore predict that the prosensory cell column adjacent to Kölliker’s organ that expresses Jag1, Jag2, Dll1, Notch1, and Lfng and Mfng experiences cis-inhibition (Figure 6B, red inhibitory lines), leading to only a moderate amount of Notch signal delivered to Kölliker’s organ. Loss of Lfng and Mfng is predicted to cause more cis-inhibition between Notch1 and Jag1 and Jag2, and less cis-inhibition between Notch1 and Dll1. As a result of these combined changes, this column of cells will deliver less total Notch signal in Lfng/Mfng double mutants (see Figure 4A), and thus the adjacent cell column in Kölliker’s organ also differentiates into inner hair cells and inner phalangeal cells. We summarize the effects and mechanism of mutating different components of the Notch pathway in Figure 7.
Our results suggest that the first phase of Notch signaling at the boundary of the organ of Corti establishes a single column of cells destined to form both inner hair cells and inner phalangeal cells. As this column of Atoh1, Lfng, Mfng, Jag2 and Dll1-expressing progenitors develops, we propose that a second fate decision between an inner hair cell and an inner phalangeal cell is achieved by conventional hair cell-supporting cell lateral inhibition mediated through Dll1 and Jag2 (Figure 6C, green cells; Hicks et al. (2000); Van de Walle et al., 2011). Our current data, previous expression studies (Chen et al., 2002; Cai et al., 2013) and previous Atoh1 lineage tracing showing inner phalangeal cells derived from Atoh1-expressing cells (Yang et al., 2010) suggests this entire column of cells initially expresses early hair cell genes such as Atoh1 and Mfng. Since we only observe a conversion of inner phalangeal cells to inner hair cells when Notch signaling is strongly reduced (in Notch1 mutants or at high doses of Notch inhibitors; Figures 3 and 5), we suggest this second phase of Notch signaling involves high levels of Notch signaling and is resistant to subtle manipulation of Notch signals. This sequence of organ of Corti differentiation is concluded as inner hair cells begin to express Fgf8, which causes cells lateral to the inner hair cells (Figure 6C, green hair cell) and inner phalangeal cells (Figure 6C, yellow cell) to differentiate as pillar cells (Mueller et al., 2002; Jacques et al., 2007).
This model for formation of the neural border of the organ of Corti explains the observations reported here, together with data from previous studies. The moderate level of Notch signaling delivered at the border of the prosensory domain and Kölliker’s organ has been visualized in previous studies (Murata et al., 2006; Basch et al., 2011; Liu et al., 2013; Figure 4A). Indeed, Murata and colleagues have observed Atoh1-expressing cells adjacent to cells containing activated Notch1 at the boundary of Kölliker’s organ (Murata et al., 2006). We show that reducing this moderate Notch signaling leads to the formation of additional columns of cells fated to form inner hair cells and inner phalangeal cells. Although this phenotype can be observed in a number of previous studies in which Notch signaling was reduced or particular Notch ligands or Notch targets were mutated (Lanford et al., 1999; Zheng et al., 2000; Zine et al., 2001; Kiernan et al., 2005; Brooker et al., 2006; Tateya et al., 2011; Abdolazimi et al., 2016), the duplication of inner phalangeal cells in these studies was not identified due to the lack of specific markers for these cells. It may be necessary to re-evaluate these phenotypes in light of the present study.
Our model can also explain why mutation of Lfng alone is able to rescue the inner hair cell phenotype, but not the outer hair cell phenotype of Jag2 mutant mice (Zhang et al., 2000). Since Lfng normally enhances cis-inhibition between Delta ligands and Notch receptors expressed in the same cell (LeBon et al., 2014), we predict that loss of Lfng would partially relieve cis-inhibition of Dll1 expressed at the boundary and allow it to deliver more Notch signal to compensate for the loss of Jag2. We predict that Mfng mutants would rescue Jag2 mutant mice in a similar manner. Although Fringe genes are often expressed in overlapping patterns, no obvious synergy has been reported in the brain, axial skeleton, limbs or cranial nerves for mutants in multiple family members (Moran et al., 2009). However, an interaction between Lfng, Rfng and Mfng has been observed in the maturation of marginal zone B cell precursors (Tan et al., 2009; Song et al., 2016), suggesting that in at least some cases, the combined effects of these three enzymes on Notch signaling in either cis or trans cannot be achieved by either alone.
We found no strong evidence for the supernumerary inner hair cells and inner phalangeal cells being caused by an increase in proliferation. Our EdU labeling comparisons between wild type, Pofut1 and dnMAML1 mutant mice showed no significant increase in labeling of inner phalangeal cells, inner hair cells or any other organ of Corti cell type. A previous study described significant increases in proliferation in the progenitors of pillar cells, Deiters’ cells and Hensen’s cells in Dll1+/−;Jag2−/− and Dll1hyp/−;Jag2−/− mutant embryos, but not single mutants for either ligand (Kiernan et al., 2005). We infer from this that the reduction in Notch signaling in all our Notch pathway mutants (with the exception of Notch1 conditional mutants) is insufficient to trigger aberrant proliferation, and that our phenotype is due solely to a disruption in the boundary between the prosensory domain and Kölliker’s organ.
It is now well established that the strength of Notch signaling in vertebrates can vary according to both the identity and post-translational modification of the activated Notch receptor and the identity of the activating ligand (for example, Ong et al., 2006; Yamamoto et al., 2012; Lee et al., 2013; Van de Walle et al., 2013; LeBon et al., 2014; Petrovic et al., 2014; Gama-Norton et al., 2015). The developing organ of Corti is an excellent system to study the strength of Notch signaling, as its stereotyped pattern of hair cells and different supporting cell types allows extremely small irregularities in cell numbers to be detected easily. Our data provide an example of how rapid dynamic changes in both Notch ligands (Jag1, Jag2, and Dll1) and regulatory Fringe proteins at the boundary of the organ of Corti can quickly transition between Notch signaling states that set the boundary (moderate Notch signaling; regulated by Fringe proteins) and then distinguish between hair cells and supporting cells (high Notch signaling; no requirement for Fringe proteins). Moreover, our observation that the early phase of Notch signaling is extremely sensitive to changes in signaling strength suggests that the Hes or Hey genes activated in this phase will have mostly low-affinity binding sites for the Rbpj transcriptional activator in their promoters. Indeed, low levels of Notch activation and Hes/Hey gene expression are observed at this boundary (Murata et al., 2006; Basch et al., 2011; Tateya et al., 2011) and Hes5, whose promoter contains low-affinity Rbpj binding sites (Ong et al., 2006), is up-regulated at the boundary of Kölliker’s organ as hair cells differentiate (Tateya et al., 2011). Furthermore, Hes5 mutant mice develop supernumerary inner hair cells and supporting cells at the onset of hair cell differentiation (Zine et al., 2001).
Our results provide additional evidence for the notion that organ of Corti development is an iterative process proceeding in the neural-abneural direction, with inner hair cells and their associated supporting cells differentiating first, which then release inducing signals such as Fgf8 that then drive the differentiation of the adjacent cells into pillar cells. It is likely that these short-range signals interact with and modify other patterning signals in the cochlear duct, such as FGF10 and FGF20 (Huh et al., 2012; Urness et al., 2015), although signals that promote specific cell fates in the outer hair cell region, such as outer hair cells and Deiters’ cells, remain to be identified. We do not know the nature of signals that limit the size of the organ of Corti on the abneural side of the cochlear duct, but given the proximity of a strong source of BMP4 in the future outer sulcus (Morsli et al., 1998; Ohyama et al., 2010), it is likely that this factor may play a role. Nevertheless, our data reveal how the deployment of the Notch pathway-modifying enzymes Lfng and Mfng at the boundary of the future organ of Corti modulate levels of Notch signaling to restrict hair cell-inducing signals to a precise location at this boundary. Since Lfng and Mfng expression subsequently diverge, with Mfng restricted to hair cells (Cai et al., 2015) and Lfng restricted to supporting cells (Zhang et al., 2000), it will be of interest to determine whether these enzymes continue to play any role in modulating Notch signaling in the mature organ of Corti.
Double homozygous Lfng and Mfng mutant embryos were generated by crossing Mfng homozygous mutant mice (B6(FVB)-Mfngtm1.1Cfg/Mmucd; RRID:MGI:5615604) obtained from the Functional Glycomics Consortium; http://www.functionalglycomics.org and available at the MMRRC, stock number 031948-UCD) with heterozygous Lfng mice (B6;129S1-Lfngtm1Grid/J; RRID:IMSR_JAX:010619) from the Jackson Laboratory (stock number 010619), and crossing the resulting compound mutant offspring. Triple homozygous Lfng;Mfng;Rfng embryos were obtained from Dr. Susan Cole, Ohio State University. Inner ear-specific dominant negative Mastermind-like (dnMAML1) mice, were generated by crossing Pax2-Cre mice (Ohyama and Groves, 2004) with a mouse carrying a human dnMAML construct downstream of a floxed PGK-neo-tpA cassette targeted to the ROSA26 locus (Tu et al., 2005). Mice homozygous for conditional alleles of Pofut1 (Shi et al., 2005) were crossed with Pax2-Cre mice that were also heterozygous for a null mutation in Pofut1. Pax2-Cre mice are available from the MMRRC (stock number: 010569-UNC; RRID:MGI:4438962). Glast-DsRed reporter mice (Regan et al., 2007) were a kind gift from Ben Deneen at Baylor College of Medicine with the permission of Jeffrey Rothstein. They were crossed to Atoh1A1GFP/A1GFP reporter mice (Shroyer et al., 2007); RRID:IMSR_JAX:013593) to generate the embryos used in organotypic cultures. At least four cochleas were analyzed for each genotype at each age. The following primers were used for genotyping:
Pax2-Cre allele: Cre1F (GCCTGCATTACCGGTCGATGCAACGA), Cre1R (GTGGCAGATGGCGCGGCAACACCATT) yield a 700 bp band.
Pofut1 floxed, deleted and wild type allele: Forward primer GGG TCA CCT TCA TGT ACA AGT GAG TG and reverse primer ACC CAC AGG CTG TGC AGT CTT TG yield a 960 bp floxed allele band, and either a 700 bp wild type band or a 300 bp deleted allele band.
Manic fringe deleted and wild type allele: Forward primer GTG CTG AAG CAG AGG CCA TG and reverse primer CAA GGT GAA GGA GCC CAG TT yield a 370 bp band for the deleted allele; forward primer GGC CCT CTC TCA CAT GGA TTT T and reverse primer TCT ACC TCC AAG CAC TAA GG yield a 444 bp band for the wild type allele
Lunatic fringe deleted and wild type allele: Forward primer CCA AGG CTA GCA GCC AAT TAG and reverse primer GTG CTG CAA GGC GAT TAA GTT yield a mutant band of 450 bd; forward primer GAA CAA ATA TGG CCA TTC ACT CCA and reverse primer GGT CGC TTC TCG CCA GGG CGA yield a wild type band of 450 bp.
Cochleas from stage E14.5 embryos were collected in PBS and incubated in calcium-magnesium free PBS containing dispase (1 mg/ml) and collagenase (1 mg/ml) for 8 min at room temperature as previously described (Doetzlhofer et al., 2009) to free the cochlear duct from surrounding condensed mesenchyme tissue. Embryonic cochlear explants were cultured on SPI black membranes (SPI Supplies) in DMEM-F12 (Invitrogen) with N2 supplement and Fungizone. All cultures were maintained in a 5% CO2/20% O2 humidified incubator for 48 or 72 hr. Cultures were treated with function blocking antibodies against IgD or against the Notch1receptor (Wu et al., 2010) provided by Genentech or with DMSO and the gamma secretase inhibitor IX (DAPT). For cell counts, three cochleas were counted per condition for each replicate, with at least four independent biological replicates performed. Cultures were excluded from analysis if they showed signs of contamination or if the cochlear explant failed to attach or displayed grossly abnormal morphology.
E14.5, E16.5 or P0 inner ears were fixed in 4% paraformaldehyde in PBS overnight at 4°C, sunk in 30% sucrose in PBS at 4°C, incubated in OCT at room temperature for 1 hr, and frozen in liquid nitrogen. Digoxigenin-labeled antisense ribroprobes to mouse Lunatic fringe, Manic Fringe and Atoh1 were synthesized using standard protocols (Stern, 1998). In situ hybridization was performed as recently described (Cai et al., 2015). Fluorescent in situ hybridization was performed according to a modified protocol (Denkers et al., 2004). Slides were incubated in DEPC-PBS with 3 µl H202/10 ml for 15–30 min, washed three times in DEPC-PBS for 5 min each, dried and then incubated in hybridization buffer containing the FITC- and/or DIG-labeled probes overnight at between 65–70°C. The following day, slides were washed for 10 min in 0.2x SSC at 65–70°C, and then twice more for 25 min each in 0.2x SSC at 65–70°C. Slides were then washed at room temperature for 5 min in MABT and then blocked for 30–6- minutes in MABT +20% sheep serum +2% Roche Blocking Reagent. Slides were then overnight at 4°C with 1:500 anti-DIG-HRP or anti-FITC-HRP antibodies in blocking buffer. The following day, the slides were washed three times in MABT for 5 min each, then incubated for 30 min in Tyramide working solution (TSA plus kit, Perkin-Elmer: 10 µl of stock in 500 µl of diluent) and the reaction monitored after 30 min by fluorescence microscopy until the desired staining was observed. If a second color is desired, slides were then washed and the antibody staining and Tyramide amplification repeated with anti-DIG-HRP or anti-FITC-HRP antibodies..
Antibodies used in this study were anti-p27Kip1 (NeoMarker/ThermoFisher; RRID:AB_1959178), anti-myosin-VI (Proteus; RRID:AB_10013626), anti FABP7 (a kind gift from Drs Yuji Owada and Nobuko Tokuda, Yamaguchi University, Japan), anti GLAST (Militenyl Biotec; RRID:AB_10829302), Nerve Growth Factor Receptor (p75, Advanced Targeting Systems; RRID:AB_171798), anti Jagged1 (Santa Cruz; RRID:AB_649685) and anti cleaved Notch1 (Cell Signaling; RRID:AB_331612). Alexa 594-conjugated phalloidin (Thermo Fisher; RRID:AB_2315633) was used to label actin in hair cell stereocilia. Secondary antibodies used were Alexa-594 or Alexa-488 (Thermo Fisher) and anti-rabbit HRP (Thermo Fisher). DAB staining was done using a DAB kit (Vector labs). For anti-p27Kip1 and Jagged1 staining, sections were boiled for 10 min in 10 mM citric acid pH 6.0. For anti-GLAST and anti-N1ICD staining, paraffin sections were boiled for 10 min in a pressure cooker with unmasking solution (Vector labs) as described in Morimoto et al. (2010). The signal was enhanced using an ABC kit (Vector Labs) and TSA kit (Perkin Elmer).
Timed pregnant females were injected stating at E14.5 with 1 mg of EdU diluted in PBS for every 20 grams of weight. Injections were performed three times a day every 4 hr for three consecutive days. At E18.5, embryos were collected, fixed in 4% PFA and cryosectioned. EdU detection was performed according to the manufacturer’s instructions with a ClickiT EdU Alexa Fluor 594 system (Thermo Fisher). Sections were also stained with Myosin VI antibodies and counterstained with DAPI. Positive cells from controls and mutant embryos were counted blind and significance was established using a modified Wald test (May and Johnson, 1997).
Images were taken using a Zeiss Axiophot microscope, or a Zeiss Axio Observer fluorescent microscope with an Apotome structured illumination attachment. Images were processed using Axiovision software then further processed in Adobe Photoshop. Inner and outer hair cells were counted in dissected or cultured cochleas per 200 mm using Axiovision and Adobe Photoshop software. Cochlea and cochlea culture lengths were measured using Axiovision software.
For RNA extraction, three cochlear cultures were pooled and total RNA was isolated by using a QIAGEN RNeasy Micro kit. cDNA was synthesized by using SuperScript III Reverse Transcription Reagents (Invitrogen). qPCR was performed with a Power SYBR Green kit (Applied Biosystems) and gene-specific primer sets on a StepOnePlus real-time PCR system (Applied Biosystems). Three technical replicates were performed for each qPCR reaction. At least four biological replicates of three cochleas each were performed for each genotype or culture condition. Relative gene expression was analyzed by using the ΔΔCT method (Livak and Schmittgen, 2001) and compared with Student two-tailed t test. cDNA from untreated cochlear explants was used as a calibrator, and a ribosomal gene (L19) was used as endogenous references. Gene-specific primer sets used are: Atoh1: F 5'-ATGCACGGGCTGAACCA-3’; R 5'-TCGTTGTTGAAGGACGGGATA-3’, L19: F 5'-GGTCTGGTTGGATCCCAATG-3’; R 5'-CCCGGGAATGGACAGTCA-3’, Hes5: F 5'-GCACCAGCCCAACTCCAA-3’; R 5'-GGCGAAGGCTTTGCTGTGT-3’, Hey1: F 5'-CACTGCAGGAGGGAAAGGTTAT-3’; R 5'-CCCCAAACTCCGATAGTCCAT-3’, Hey2: F 5' AAGCGCCCTTGTGAGGAAA-3’; R 5'-TCGCTCCCCACGTCGAT-3’, Heyl: F 5'-GCGCAGAGGGATCATAGAGAA-3’; R 5' TCGCAATTCAGAAAGGCTACTG-3’, Hes1: F 5'-GCTTCAGCGAGTGCATGAAC-3’; R 5'-CGGTGTTAACGCCCTCACA-3’.
Patterning signals and proliferation in Drosophila imaginal discsCurrent Opinion in Genetics and Development 17:287–293.https://doi.org/10.1016/j.gde.2007.05.005
Where hearing starts: the development of the mammalian cochleaJournal of Anatomy 228:233–254.https://doi.org/10.1111/joa.12314
A role of receptor Notch in ligand cis-inhibition in DrosophilaCurrent Biology 20:554–560.https://doi.org/10.1016/j.cub.2010.01.058
Notch signalling: a simple pathway becomes complexNature Reviews Molecular Cell Biology 7:678–689.https://doi.org/10.1038/nrm2009
Notch signalling: receptor cis-inhibition to achieve directionalityCurrent Biology 19:R683–684.https://doi.org/10.1016/j.cub.2009.07.025
Immunocytochemical localization of a high-affinity glutamate-aspartate transporter, GLAST, in the rat and guinea-pig cochleaEuropean Journal of Neuroscience 9:1961–1969.https://doi.org/10.1111/j.1460-9568.1997.tb00763.x
Notch signal strength controls cell fate in the haemogenic endotheliumNature Communications 6:8510.https://doi.org/10.1038/ncomms9510
Divergent functions and distinct localization of the Notch ligands DLL1 and DLL3 in vivoThe Journal of Cell Biology 178:465–476.https://doi.org/10.1083/jcb.200702009
GENSAT: a genomic resource for neuroscience researchThe Lancet Neurology 3:82.https://doi.org/10.1016/S1474-4422(03)00658-6
Overview of genetic tools and techniques to study Notch signaling in miceMethods in Molecular Biology 1187:47–61.https://doi.org/10.1007/978-1-4939-1139-4_4
Modulation of receptor signaling by glycosylation: fringe is an O-fucose-beta1,3-N-acetylglucosaminyltransferaseBiochimica et Biophysica Acta (BBA) - General Subjects 1573:328–335.https://doi.org/10.1016/S0304-4165(02)00400-2
Dll3 is expressed in developing hair cells in the mammalian cochleaDevelopmental Dynamics 236:2875–2883.https://doi.org/10.1002/dvdy.21307
Making sense of Wnt signaling-linking hair cell regeneration to developmentFrontiers in Cellular Neuroscience 9:66.https://doi.org/10.3389/fncel.2015.00066
Regulation of cell fate and patterning in the developing mammalian cochleaCurrent Opinion in Otolaryngology & Head and Neck Surgery 17:381–387.https://doi.org/10.1097/MOO.0b013e3283303347
Notch signaling during cell fate determination in the inner earSeminars in Cell & Developmental Biology 24:470–479.https://doi.org/10.1016/j.semcdb.2013.04.002
Notch signalling pathway mediates hair cell development in mammalian cochleaNature genetics 21:289–292.https://doi.org/10.1038/6804
Notch signalling and the control of cell fate choices in vertebratesSeminars in Cell & Developmental Biology 9:583–589.https://doi.org/10.1006/scdb.1998.0266
Expression of Delta1 and Serrate1 (Jagged1) in the mouse inner earMechanisms of Development 84:169–172.https://doi.org/10.1016/S0925-4773(99)00066-0
Notch-Hes1 pathway contributes to the cochlear prosensory formation potentially through the transcriptional down-regulation of p27Kip1Journal of Neuroscience Research 87:3521–3534.https://doi.org/10.1002/jnr.22169
Mapping of notch activation during cochlear development in mice: implications for determination of prosensory domain and cell fate diversificationThe Journal of Comparative Neurology 497:502–518.https://doi.org/10.1002/cne.20997
Fringe benefits: functional and structural impacts of O-glycosylation on the extracellular domain of Notch receptorsCurrent Opinion in Structural Biology 21:583–589.https://doi.org/10.1016/j.sbi.2011.08.008
Localization of fatty acid binding proteins (FABPs) in the cochlea of miceAnnals of Anatomy - Anatomischer Anzeiger 192:210–214.https://doi.org/10.1016/j.aanat.2010.06.007
BAC transgenic mice and the GENSAT database of engineered mouse strainsCold Spring Harbor Protocols, 2013, 10.1101/pdb.top073692, 23457350.
Beta-catenin up-regulates Atoh1 expression in neural progenitor cells by interaction with an Atoh1 3' enhancerJournal of Biological Chemistry 285:392–400.https://doi.org/10.1074/jbc.M109.059055
β-Catenin is required for hair-cell differentiation in the cochleaJournal of Neuroscience 34:6470–6479.https://doi.org/10.1523/JNEUROSCI.4305-13.2014
Canonical Notch signaling is dispensable for early cell fate specifications in mammalsMolecular and Cellular Biology 25:9503–9508.https://doi.org/10.1128/MCB.25.21.9503-9508.2005
Lunatic, Manic, and Radical Fringe Each Promote T and B Cell DevelopmentThe Journal of Immunology 196:232–243.https://doi.org/10.4049/jimmunol.1402421
Notch signaling is an important regulator of type 2 immunityThe Journal of Experimental Medicine 202:1037–1042.https://doi.org/10.1084/jem.20050923
Specific notch receptor-ligand interactions control human TCR-αβ/γδ development by inducing differential Notch signal strengthThe Journal of Experimental Medicine 210:683–697.https://doi.org/10.1084/jem.20121798
Math1 regulates development of the sensory epithelium in the mammalian cochleaNature Neuroscience 7:1310–1318.https://doi.org/10.1038/nn1349
Tanya T WhitfieldReviewing Editor; University of Sheffield, United Kingdom
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "Fringe proteins fine-tune Notch signaling to set the boundary of the organ of Corti and establish sensory cell fates" for consideration by eLife. Your article has been favorably evaluated by Marianne Bronner (Senior Editor) and three reviewers, one of whom, Tanya Whitfield, is a member of our Board of Reviewing Editors. The following individual involved in review of your submission has agreed to reveal his identity: Fernando Giraldez (Reviewer #2).
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
This is an interesting study that addresses the role of Fringe family proteins in regulating Notch signalling in the developing organ of Corti of the mammalian inner ear. Notch signalling is known to pattern the organ of Corti, and Fringe proteins are also known for their role in fine-tuning Notch signalling in other systems. The new finding here is that regulation of Notch signalling by Fringe plays a role in the generation of inner hair cells and their associated supporting cells, the inner phalangeal cells, in the cochlear duct. A surprising result is that a partial loss of Notch signalling results in both supernumerary hair cells and supernumerary supporting cells. This differs from the classical model for lateral inhibition in the developing cochlea, where a disruption of Notch signalling is predicted to result in supernumerary hair cells at the expense of supporting cells. The data are clear and well-presented, with careful and appropriate quantitation, use of controls and statistical analysis.
All three reviewers were positive towards the work and found it interesting, novel and potentially important. They praise the careful approach and high quality of the data presented. However, there was consensus that the manuscript needed revision in places, especially with respect to the focus on cis-inhibition. The revisions must therefore include:
1) Improved clarity of the explanations of cis-inhibition, and the relative roles of Fng proteins and DSL ligands in regulating Notch activity.
2) The argument that levels of Notch activity differ in different cell types and regions of the cochlear duct is critical to the model. Further experimental support for the claim that Notch signalling activity levels differ both over time and in different cell types in the developing cochlear duct is required (see comments from reviewers 2 and 3).
3) Further experimental support for cis-inhibition and the effects of Fng on cis-inhibition in the developing cochlear duct is required. If this cannot be provided, the argument for cis-inhibition should be presented as an untested model and the overall focus on cis-inhibition in the manuscript should be toned down.
4) Strengthen the claim that Lfng marks developing inner hair cells, as suggested by reviewer 3.
5) Addressing additional comments in the following reviews is at the authors' discretion.
1) The authors show that they can phenocopy the Fng mutant phenotype with the Jag2 mutant. The information presented in the Introduction indicates that Fng acts differentially on Dl-N and Jag/Ser-N interactions, whereas in the first paragraph of the subsection “Reduction in Notch signaling leads to the formation of supernumerary inner hair cells and inner phalangeal cells”, it is stated that Fng can potentiate signalling through Notch from both Dll1 and Jag2. It would therefore be interesting to know whether a Dll1 phenotype also phenocopies the Fng and Jag2 results. I appreciate that this may be beyond the scope of the current study, but would the authors predict a similar or a different result?
2) If Fng proteins are required for precision fine-tuning of Notch signalling, why is the system so robust to perturbations in one or the other Fng, and a phenotype is only seen when the functions of both genes are lost?
3) I am not really sure why the authors propose that their findings argue for a new role for Notch (Discussion, first paragraph); an interpretation could be that this is just another illustration of lateral induction, albeit in a restricted zone of the cochlea. Can the authors provide more information as to why they do not think this is the case?
4) Is extension of the cochlea normal? Are there more cells overall, or are they packed in a different way due to altered shape of the organ?
5) In trying to get to grips with the terminology here, I find that the terms 'lateral inhibition' (involving activation of Notch signalling) and 'cis-inhibition' (involving attenuation of Notch signalling) are potentially confusing/ambiguous and difficult to follow.
For example, the following statement is made in the Introduction: 'In vertebrates, Lfng and Mfng proteins regulate receptor-ligand interactions in cis in the same way as they do in trans: they increase cis-inhibition between Notch and Δ ligands, while attenuating cis-inhibition between Notch and Serrate/Jagged ligands (LeBon et al. 2014)'.
The wording 'in the same way' here is confusing, because it appears to be the opposite to the information given in the second paragraph of the Introduction, where it is stated that Fng proteins increase the level of Notch signalling by Δ. If Fng increases cis-inhibition between N and Dl (later in the aforementioned paragraph), given the information presented, that should lead to a decrease in Notch signalling activity in the Notch-expressing cell.
Likewise, the statement that Fng proteins 'attenuate Notch signaling by Serrate/Jagged ligands' (Introduction, second paragraph) appears to be contradictory to that later in the aforementioned paragraph where Fng proteins function by 'attenuating cis-inhibition between Notch and Serrate/Jagged ligands' (which in other words would potentiate Notch signalling via Ser/Jag).
I appreciate that the terms 'lateral inhibition' and 'cis-inhibition' are now entrenched in the literature, but I, for one, find the explanations above contradictory and feel that more clarification is needed. If the authors can make sure that their text is completely unambiguous that would be helpful.
6) I don't see how the descriptions of Fng activity in the Introduction (second paragraph) are illustrated in the summary diagram (Figure 6). If Fng potentiates Dl-N cis-inhibitory interaction, but attenuates Jag-N cis-inhibitory interaction, the dotted arrow from Fng to Jag-N in the green cell should be drawn as an inhibitory line (-|), not an arrow, and the cis-inhibitory interactions between Dl and N should be stronger (thicker lines) in the green cell than those in the white cell, whereas the diagram shows the opposite. These apparent discrepancies make it hard to follow the argument.
7) In the legend to Figure 6, the statement 'As a result, these cells do not respond to hair cell-inducing signals' seems to contradict the sentence at the beginning of the section describing panel B, which states that these cells are responding to hair cell-inducing signals, and have been shaded green to indicate this. If the sentence refers to the neighbouring cells, the 'these' is ambiguous, and the sentence does not follow logically from the previous one. If the authors mean that the green cells have responded to hair cell-inducing signals, but have not yet received instruction to undergo final differentiation into hair cells (as in C), this needs clarifying further.
The paper by Basch et al. is a very interesting analysis of the function of Fringe proteins in hair cell patterning in the ear. The main and certainly interesting aspect of the work is the disclosure of a very specific function of Fng proteins in cochlear development and the exploration of their association with Notch signalling. The experiments are extremely careful and beautifully presented, and there is no doubt about the consistency of the various phenotypes and treatments. Data are of high quality and involve a variety of approaches including lineage tracing, phenotypic analysis, organ culture, etc., all making it a robust piece of work. The authors propose an interesting model that involves Notch activity levels and cis- and trans- modes of ligand operation to accommodate the results and with those of the literature. However, it is here where the work requires some improvement.
1) Figure 1 shows that the expression of Lfng and Mfng are restricted to sensory patch. Although the expression patterns are overlapping, they are quite dissimilar. This suggests that modulation of Notch is different in the two cell types. Could it be that Lfng and Mfng have different effects on different ligands? Is it possible that Lfng maintains a low and constant Notch activity and Mfng favours Dll signalling?
No doubt about the specificity of the combination of Lfng+Mfng for the correct development of the inner row. However, the correlation with Notch signalling levels remains obscure. Lfng+Mfng phenotype gives an expanded inner cell row, and this is paralleled by the loss of Jag2, but also of Jag1. Jag1 is expressed all throughout prosensory development and in supporting cells, but Jag2 is expressed along with Dll1 only in hair cells. How this is matched?
2) Lineage tracing. Those are interesting experiments and show that almost the complete sensory epithelium derives from Lfng expressing progenitors (probably also from Jag1-positive ones). However, genetic tracing shows the common origin of the sensory epithelium, but not much about the restricted expression. In this view, the genetic tracing of Manic cells would have been of interest, because it seems that the uniqueness of the effects may come from the co-expression.
3) Fng LOF is mimicked by low Notch. That is the main line of the argument along the paper. But from here, I would expect the authors to consider how this could happen by showing the effects of Fng on ligands or on Notch activity would have been informative. Neither it is clear to me whether the levels of Notch affect the expression of modulators or vice versa. We do not really know how Notch operates outside lateral inhibition and all possibilities should remain open.
4) As to the levels of Notch activation, indeed, the observations fit well with the Murata paper. But nevertheless, the different conditions generated by the experiments in the paper open some questions about what are the levels of Notch in the different regions of the sensory epithelium. In this sense, conditions are not strictly comparable. Mfng and Lfng show overlapping but different expression domains, and in MAM and Pofut CKO or DAPT experiments, the decrease of Notch is ubiquitous. Can this be assessed specifically for Fng mutants? Please comment on that.
5) The data in the bar diagrams of Figure 5 indicate that there is an inverse correlation between Notch activity, Hes/Hey expression and the production of hair cells. And it shows the differential sensitivity of inner and outer hair cells to Notch. This is in good correspondence with MAMl and Pofut experiments. But what about in Fng mutants? It would have been interesting to see Hes/Hey levels in Fng mutants.
6) Cis-inhibition: The experiments say little about cis-inhibition. "Lfng activity attenuates Jag1-Notch signaling between cells (Hicks et al. 2000; Rana and Haltiwanger 2011) and promotes Jag1-Notch cis-inhibition in the same cell, with the result…" I am not sure how the second sentence follows the first and, moreover, whether it is a general true at all. Is there a good evidence for cis-inhibition in ear sensory cells? The results from Daudet's lab indicate rather the opposite for Dll1.
The argument on cis-inhibition continues in the second and third paragraphs of the Discussion, quoting LeBon 2014 as if an ear paper…: "we hypothesize that this column of cells only delivers a moderate level of Notch signaling, due to active cis-inhibition between the Notch ligands and Notch receptors that can occur in the same cell". However, the experiments do not provide any evidence of that occurring in the ear…
In summary, the results are indeed sound and of general interest, but as to the model, it is unclear that the function of Fng on Notch and cis-inhibition are sufficiently substantiated. Leaving aside cis-inhibition, which actually I see out of the paper, the model still needs a link between Fng, ligands and Notch in the cochlea. Something that may give a hint of what are the effects of Fng on Notch, Notch ligands or Fng. The main evidence for the model is correlative, the parallel between phenotypes. The main question, in my view, is what is the connection between both.
The organ of Corti is characterized by strikingly precise rows of hair cells, although how this precision is established is not known. This manuscript by Basch et al., proposes a novel form of Notch signaling that sets up the boundary of the initial row of inner hair cells, in which Fringe proteins, modulators of Notch signaling, are an essential component. The authors show that Lfng and Mng are initially precisely positioned at the location of the first inner hair cell row. In their absence, multiple rows form, along with a duplication of supporting cells. By comparing the results of minor reductions in Notch signaling to more robust loss of Notch, the authors suggest that establishment of the initial row of hair cells is inherently different in that it requires lower levels of Notch signaling, and is therefore more susceptible to mild reductions in Notch signaling. Overall the data is well-performed and presented, and the authors analyze quite a number of different Notch mutants to determine differences. The authors put forward an interesting hypothesis as to why the hair cell genesis is initially restricted to a single row. However, it is not clearly established that a lower level of Notch signaling is required for boundary formation – while milder defects in Notch signaling may first lead to an expansion of the sensory domain, it may be that the non-sensory regions (or supporting cells outside the hair cell regions) are simply more sensitive to loss of Notch signaling, and thus boundary formation is lost first.
The authors suggest that cis-inhibition mediated by Lfng and Mfng expression results in moderate levels of Notch activation in nearby Kolliker's cells (Figure 6B), and that this is important for setting up the boundary. However, in this case the prediction would be that deleting both fringes would lead to higher levels of Notch, as cis-inhibition would be largely relieved. However, this does not seem to be the case, as new hair cells develop, suggesting Notch activity is reduced in their absence. Since the model indicates that moderate levels of Notch activity are important for setting up this boundary, it would be important to show directly how levels of Notch are regulated at the boundary.
It is not clear that severe loss of Notch signaling shows an inherently different phenotype than milder forms. Is it not more likely that the severe loss obscures the initial milder phenotype? After all, there is a dramatic loss of boundary formation in the severe cases of loss of Notch signaling. While additional supporting cells are present in the milder cases, these may be induced by the excess hair cells, which themselves also convert to hair cells in the case of more severe reductions in Notch.
If the regulation of inner hair cell formation to a single row is not via lateral inhibition, what type of signaling is it? It would seem that even at the proposed lower levels the cell expressing the ligand is inhibiting nearby cells (in Kolliker's organ) from adopting the hair cell fate, and therefore this would fit the definition of lateral inhibition (see Figure 6B).
The authors show that Lfng likely initially marks the inner hair cell using fate-mapping, although given that they have reporters for both Lfng and Atoh1-it seems relatively straightforward to show that these both overlap during differentiation, further strengthening the hypothesis that Lfng and Mfng mark the inner hair cell boundary. It would also be interesting to look at the onset of these factors-Does Lfng mark the inner hair cell prior to Atoh1? Or vice versa?
The suggestion that Lfng modulates Jag1 levels initially (Figure 6A) has not been demonstrated previously-is there any evidence of this from the Lfng loss of function? If not this part of the model should be modified.https://doi.org/10.7554/eLife.19921.014
- Andrew K Groves
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank Hongyuan Zhang, Alyssa Crowder, Huiling Li and Bridgett McNulty for excellent technical assistance. We thank Jeffrey Rothstein for GLAST-dsRed transgenic mice, Huda Zoghbi for Atoh1A1GFP knock-in mice, Warren Pear for dnMAML1 mutant mice and Pamela Stanley and Shaolin Shi for Pofut1 conditional mutant mice. We thank Doris Wu for the Lfng probe and Yuji Owada and Nobuko Tokuda for their gift of FABP7 antibodies. This work was supported by DC006185 (AKG)
Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal experiments in this study were carried out in accordance with the Institutional Animal Care and Use Committee protocol (AN4956) at Baylor College of Medicine.
- Tanya T Whitfield, Reviewing Editor, University of Sheffield, United Kingdom
© 2016, Basch et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.