A hypothetical model for lateral inhibition and boundary formation. Left, Cells where Notch is active are yellow, those that adopt the primary cell fate are blue, and the red line denotes the boundary formation created by Fringe. Right, Lineage decisions. The sensory organ precursor cell divides into 2 daughter cells, 1 of which expresses Numblike (vertical shading). This cell follows the primary cell fate.
Model proposed for how fate determination occurs to generate a population of cochlea hair cells and supporting cells from a common precursor. The continued presence of Notch 1 and Jagged 1 in the supporting cell is speculative. Nbl indicates Numblike.
Weir J, Rivolta MN, Holley MC. Notch Signaling and the Emergence of Auditory Hair Cells. Arch Otolaryngol Head Neck Surg. 2000;126(10):1244-1248. doi:10.1001/archotol.126.10.1244
Recent insights into the mechanisms that determine a hair cell's fate have emerged from studies on invertebrate sensory organs and the avian inner ear. These mechanisms have important implications for our understanding of the possible therapeutic management of sensorineural deafness. This article reviews the current state of our knowledge regarding mammalian auditory hair cell fate specification.
Data were obtained from the MEDLINE database and data presented at the Molecular Biology of Hearing and Deafness Meeting (Bethesda, Md, October 1998). Articles reporting information about cell fate specification and Notch and its ligands were selected.
Main Outcome Measures
Data pertaining to cell fate mechanisms, Notch and its ligands, and application to hearing were extracted.
The Notch/ligand mechanism is responsible for the specification of the hair cell phenotype.
Major progress has been made in understanding this fundamental process, and its application to hair cell determination is only now being realized. Possible applications could involve the "switching" of supporting cells to hair cells, thus replenishing those hair cells damaged in sensorineural hearing loss.
THERE IS currently an explosion in the number of new genes being discovered that are responsible for sensorineural deafness,1,2 the most common sensory disorder in the Western world. In the next millennium, many of the genes that regulate the development and function of the auditory system will be discovered. Of key importance is the appreciation of the mechanisms that are involved in specifying the auditory hair cell phenotype. This will provide a basis for understanding the possible ways in which auditory hair cells may be regenerated, thus relieving a handicap that afflicts so many people. Ultimately, the therapeutic management of sensorineural deafness due to hair cell dysfunction will require either the maintenance, repair, or regeneration of hair cells. The fate of any specific cell requires the interaction of many different processes, including cell proliferation, migration, growth, and differentiation. Both intrinsic cell-autonomous, and extrinsic short- and long-range signals are required to guide cells through distinct developmental pathways (Table 1). Indeed, many of these pathways are probably required for the development of the inner ear. This article reviews the Notch signaling pathway, which is an intrinsic signaling pathway thought to be involved in specifying the fate of an auditory hair cell. There is also evidence that the Notch pathway is present in the vertebrate vestibular system,3 and it is likely that the mechanisms of cell fate specification are the same in both systems.
The Notch pathway can be deployed in 3 different types of developmental processes4,5 (Figure 1): (1) lateral inhibition, which is a cell contact–mediated signaling system between neighboring cells; (2) lineage decisions, which are decisions made between 2 daughter cells, leading one to adopt a different fate from its sibling; and (3) boundary formation, which occurs between 2 distinct areas of cells, such as sensory and nonsensory cells. All are critical processes in determining the exact position of the hair cell within the organ of Corti. One mechanism of cell fate specification broadly used by multicellular organisms is that mediated by the Notch receptor Notch and its ligands, Delta-Serrate and Jagged6,7 (Table 2). There are many homologs of Notch and of its ligands in the vertebrate body, which are expressed in varying combinations. If the components of the Notch signaling system are genetically altered in any way, the consequences can be drastic. Three human diseases have been linked to alterations of the Notch 1, Notch 3, and Jagged 1 genes. These include the T cell acute lymphoblastic leukemias and/or lymphomas8; a developmental disorder, Alagille syndrome9,10; and a late-onset neurological disorder known as cerebral autosomal dominant arteriopathy, with subcortical infarcts and leukoencephalopathy.11 Their nature points to the broad activity of Notch in humans. The challenge remains, therefore, to identify not only where but which cell fate decisions the Notch signaling pathway controls and in particular how this occurs in the inner ear.
In broad terms there are several principles that apply to Notch signaling. First, both Notch and its ligands Delta and Serrate are integral membrane proteins that transmit signals between adjacent cells from the same developmental background. Second, the effect of Notch activation on gene expression is immediate and direct. Third, there is regulation of the ligand in the Notch-activated cell, giving rise to feedback loops that correlate the fates of adjacent cells and influence the fine detail of the spatial pattern of differentiation.12 These 3 principles will be discussed with respect to developmental processes in the inner ear.
Prior to lateral inhibition, the steps of axis orientation and compartmentalization in the otocyst appear to be crucial for subsequent cell fate specification.13 Much of the work on Notch signaling in vertebrates has been on the control of neurogenesis.14 In an equivalent group of cells, activation of Notch by a Delta-expressing cell prevents the cell in which Notch has been activated from differentiating into a neuron. Thus, from a common pool of progenitor cells, one group differentiates into neurons while the other group remains as a progenitor pool.15 Essentially, activation of the Notch receptor by the ligand prevents that cell from differentiating along its default pathway.
Once Notch has been activated, its intracellular domain is cleaved by a protease.16 This domain, Notch IntraCellular Domain (NICD), then interacts with a transcriptional complex including suppressor of hairless SU(H) in Drosophila and RBP-Jκ/KBF2/CBF1 in mammals, collectively known as the CSL proteins.17 The Notch/CSL protein complex drives the transcription of a group of related genes known as enhancer of split E(spl) in Drosophila and Hes (homolog of hairy and enhancer of split) in mammals18 that repress the cell from undergoing differentiation. Concomitantly, by preventing differentiation, they also repress Delta expression. Thus, lateral inhibition appears to be regulated through the Notch, RJBk, Hes pathway, which has a fast track to the nucleus. Furthermore, Notch signaling through the expression of Hes can also upregulate the expression of Notch itself,19 thus providing positive feedback regulation of Notch, which would serve to reinforce the cell's responsiveness to Notch ligands expressed on surrounding cells.
The expression of messenger RNA (mRNA) for Notch 1 and Jagged 2 in the mammalian cochlea has recently been analyzed.20 At embryonic day 13, Notch 1 is distributed widely throughout the cochlear epithelium, but Jagged 2 does not appear until embryonic day 15. Their expression is localized to a narrow band that runs from the base to the apex of the cochlea, and this band of expression also migrates from the neural edge of the epithelium to the abneural edge (from the inner hair cells to the outer hair cells). By postnatal day 3, Jagged 2 is found only in hair cells. It would appear that the mechanism of lateral inhibition is critical for hair cell specification and depends, as it does in neural development, on the activity of Notch, its ligands, and their modulators.
Studies on the chick basilar papilla (equivalent to the auditory epithelium) have also shown a similar downregulation of expression of Serrate 1 (close homology to Jagged 1) and Notch 121 in hair cells. Results from our laboratory using an in vitro preparation from the Immortomouse (Ludwig Institute, London, England) have corroborated these findings. We found that the protein expression of Notch 1 and Jagged 1 was down-regulated in putative cochlear hair cells from embryonic day 13 to postnatal day 7, while the mRNA expression of Jagged 2 remained constant.22 It has been postulated that the continuing presence of Jagged 2 in the hair cells is to keep the supporting cells in their "inhibited" state.23 The data are not entirely consistent, since Lewis et al24 suggest that Notch 1 is present in adult cochlear hair cells. Further work will be required to provide a better understanding of the gene expression patterns, especially as the mammalian organ of Corti is not a perfect mosaic and this simplified model of lateral inhibition is probably far more complex.
During development of a cell, decisions relating to its lineage are made. This process, at least in part, relies on the gene Numb.25 Numb protein is associated with the membrane and asymmetrically localized in the shape of a crescent in the sensory organ precursor of Drosophila prior to cell division. After division, it is preferentially segregated into one of the daughter cells and indirectly determines its own fate. It does this by antagonizing Notch so that the daughter cell without Numb is the only one where Notch activity is operative. Furthermore, Numb is able to generate asymmetry in the subsequent division of the daughter cell. In the mouse, a Numb gene that shares a high degree of sequence homology with fly Numb has been found.26 Mouse Numb protein is also associated with the membrane prior to cell division, but unlike fly Numb it can be distributed either symmetrically or asymmetrically in the daughter cells. This difference may reflect the different strategies the fly and the mouse have adopted during evolution to fulfill neurogenesis.
In the fly, except in the eye, all the neural precursors undergo asymmetric cell division to generate neurons. In the mouse, neural precursors undergo symmetric cell division to increase the size of the precursor pool before undergoing asymmetric cell division to generate neurons.27 There are no published data on the presence of Numb in the inner ear. However, we have found that the mouse protein Numblike, a gene with significant sequence similarity to Drosophila Numb, is present in nascent cochlear hair cells in vitro and that it continues to be present postnatally.
The differential activation of Notch-mediated signaling is influenced by Numb at each successive step in the sensory organ precursor lineage. By antagonizing Notch, Numb creates a bias between daughter cells. This mechanism may be necessary to increase the reliability of Notch-mediated signaling in situations requiring rapid decision making, when there may be insufficient time to activate transcription-based feedback mechanisms. The mechanism by which Numb antagonizes Notch is not yet known, although one possibility is that Numb interferes with the interaction of Notch with RBP-Jκ.28
In the development of the dorsoventral boundary in the Drosophila wing margin, Notch is activated at the interface between 2 separate fields of cells and is involved in keeping the 2 populations distinct.29 In order to create this bias in Notch signaling the secreted protein, Fringe is required,30 which binds to one of the ligands and prevents it from activating Notch. In the cochlea during embryonic development, Lunatic Fringe, the murine homolog of Drosophila Fringe, is restricted to the supporting cells under the differentiating inner and outer hair cells.31 The exact mechanism of action of this gene at this particular site and its effect on Notch is not yet known, but it seems likely that it does play some role in defining the region of sensory epithelium and in specifying the fate of the sensory cells.
There are still many questions unanswered, and the precise mechanism of how one set of cells become hair cells and another set become supporting cells has yet to be fully determined. Certainly it can be argued that the decision is not simply one between hair cells and supporting cells but rather a more precise decision-making process between the type of supporting cell (Deiter, Hensen, or Pillar cells) and the type of hair cell (inner and outer). Within this complex pattern of Notch signaling, the subdivision of Notch-dependent processes into the 3 types is certainly an oversimplification. Nevertheless, we propose a simplified model in an attempt to bring together the concepts discussed in this review (Figure 2). From a group of equivalent cells in the prosensory cluster, 2 distinct phenotypes emerge: the hair cell and the supporting cell. Just prior to the terminal mitosis of cells in the prosensory cluster, we postulate that the protein Numblike is symmetrically distributed among the progeny. At the point of terminal mitosis, Notch 1 and one of its ligands, Jagged 1, may be switched on by one of the prosensory genes, such as Math 1.32 Thus, 2 equivalent cells express Numblike, Notch 1, and possibly Jagged 1 or another of the many ligands, such as Delta. In the mouse, the expression of Delta 1 in the inner ear appears to be limited to the nascent hair cells as well as some nonsensory areas.33
The next stage is unclear, but the hypothesis is that in 1 of the 2 cells (cell B), Numblike is down-regulated. In the other cell (cell A), Numblike continues to be expressed. It then binds to Notch 1 within the same cell and possibly blocks it. This sets up a bias in the signaling potential of the Numblike-bearing cell. This bias may be amplified by feedback regulation, with the subsequent up-regulation of the ligand in cell A. The ligand activates Notch 1 in cell B, which then activates Hes via RBP-Jκ, with the subsequent down-regulation of the prosensory genes. This in turn has the added affect via the regulatory feedback loop of up-regulating Notch 1 and downregulating the ligand in cell B. The down-regulation of the ligand in cell B results in the added downregulation of Notch 1 in cell A. As the prosensory genes are not deactivated in cell A, they continue with their function, which may well be the activation of differentiation, with the concomitant expression of BRN3.1, a gene critical for hair cell differentiation.34 In this way the fate of the cell has already been determined before BRN3.1 is switched on. Cell A follows the primary fate and becomes a hair cell, and cell B follows a secondary fate and becomes a supporting cell. However, for some reason as yet unexplained, Jagged 1 expression appears to be downregulated in the nascent hair cells as development progresses and is replaced by another ligand—Jagged 2.
There are still many factors other than the ones described that can alter the outcome of signaling, and as the direct consequences of Notch activation appear to be at a transcriptional level, there is still much to be ascertained as to how these responses are elicited. Understanding these responses will be an important step toward unraveling the many functions performed by Notch. Likewise, the place of Notch within the larger context of cell signaling needs to be determined, as does its expression within adult vertebrate differentiated cells. Perhaps, as some investigators have hypothesized, Notch is required to maintain mature cells in a specific differentiated state. With respect to sensorineural deafness, the presence of Notch and its ligands not only appears to be critical in the specification of the hair cells but also may be necessary for their continued well-being.
The Notch signaling system has fundamental clinical implications because it provides a mechanism for converting existing supporting cells into hair cells. In the mouse Jagged 2 knockout, for example, it was found that there was an increase in the number of inner and outer hair cells.20 Furthermore, in the chick basilar papilla, after drug-induced hair cell damage, Notch 1 is increased in the support cells during hair cell regeneration.3 This is the first evidence that Notch may be an important factor for hair cell regeneration. Activating this system in the mammalian ear provides a very real potential for stimulating hair cell replacement. Other genes that regulate cell proliferation, such as the cyclin-dependent kinase inhibitor p27 (KIP1), have also been identified in the mammalian auditory epithelium,35,36 and they are also possible targets for manipulation. Our ability to regenerate or repair damaged hair cells therapeutically will almost certainly be dependent on our understanding of the many signaling pathways involved in fate specification. Pharmaceutical companies and research groups are now investing optimistically in this exciting field, and we should expect to see major progress in the near future.
Accepted for publication April 11, 2000.
Corresponding author/reprints: Matthew C. Holley, PhD, Physiology Department, School of Medical Sciences, University of Bristol, University Walk, Bristol BS8 1TD, England (e-mail: M.C.Holley@bristol.ac.uk).