In vertebrates with mutations in the Notch cell-cell communication pathway, segmentation fails: the boundaries demarcating somites, the segments of the embryonic body axis, are absent or irregular. This phenotype has prompted many investigations, but the role of Notch signalling in somitogenesis remains mysterious. Somite patterning is thought to be governed by a "clock-and-wavefront" mechanism: a biochemical oscillator (the segmentation clock) operates in the cells of the presomitic mesoderm, the immature tissue from which the somites are sequentially produced, and a wavefront of maturation sweeps back through this tissue, arresting oscillation and initiating somite differentiation. Cells arrested in different phases of their cycle express different genes, defining the spatially periodic pattern of somites and controlling the physical process of segmentation. Notch signalling, one might think, must be necessary for oscillation, or to organize subsequent events that create the somite boundaries. Here we analyse a set of zebrafish mutants and arrive at a different interpretation: the essential function of Notch signalling in somite segmentation is to keep the oscillations of neighbouring presomitic mesoderm cells synchronized.

The zebrafish provides an important model for vertebrate inner ear development. The otic placode becomes visible at approximately 16 hours (at 28.5 degrees C) and forms a vesicle with a lumen by cavitation at approximately 18 hours. Two otoliths appear in the lumen by 19.5 hours, and at about 24 hours the first sensory hair cells are seen, grouped in two small patches, one beneath each otolith, corresponding to future maculae. Staining with fluorescent phalloidin reveals 10-20 hair cells in each macula by 42 hours; between 3 days and 7 days the numbers grow to approximately 80 per macula. Neurons of the statoacoustic ganglion are first visible by staining with HNK-1 antibody at about 24 hours. Serial sections and time-lapse films show that the neuronal precursors originate by delamination from the ventral face of the otocyst; the peak period of delamination is from 22 hours to 30 hours. The system of semicircular canals forms between 42 hours and 72 hours by outgrowth of protrusions from the walls of the otocyst to form pillars of tissue spanning the lumen. Three further clusters of hair cells also become visible in this period, forming the three cristae. Thus, by the end of the first week, all key components of the ear are present. Subsequent growth produces thousands more hair cells; additional neurons probably derive from proliferation of neuronal precursors within the ganglion. Although the timetable is species-specific, the principles of inner ear development in the zebrafish seem to be the same as in other vertebrates.

Four genes - deltaA, deltaB, deltaC and deltaD - coding for homologues of the Notch ligand Delta have been discovered in zebrafish (Haddon et al., 1998b). We report here the cDNA sequence and expression pattern of deltaC. Its closest relatives are deltaB and Xenopus X-Delta-2. Unlike deltaA, deltaB, and deltaD, deltaC is not expressed in the majority of nascent primary neurons; but it is strongly expressed in the early retina, where it precedes other delta genes. It is also expressed in cranial ganglia, in sensory epithelia including ear and lateral line, and in scattered epidermal cells. In the mesoderm, expression is visible by 50% epiboly; it is seen subsequently in the tail bud, in stripes in the presomitic mesoderm and in the posterior half of each somite. There is expression also in notochord, blood vessels and pronephros.

Drosophila fringe and its homologues in vertebrates code for glycosyltransferases that modify Notch, altering the sensitivity of this receptor protein to its ligands Delta and Serrate and, thereby, playing an essential part in the demarcation of tissue boundaries. We describe the isolation and characterization of three zebrafish (Danio rerio) fringe homologues: lunatic fringe (lfng), radical fringe (rfng), and manic fringe (mfng). In addition to the sites previously described (Prince et al.

No abstract

Each sensory hair cell in the ear is normally surrounded by supporting cells, which separate it from the next hair cell. In the mind bomb mutant, as a result of a failure of lateral inhibition, cells that would normally become supporting cells differentiate as hair cells instead, creating sensory patches that consist of hair cells only. This provides a unique opportunity to pinpoint the functions for which supporting cells are required in normal hair cell development. We find that hair cells in the mutant develop an essentially normal cytoskeleton, with a correctly structured hair bundle and well-defined planar polarity, and form apical junctional complexes with one another in standard epithelial fashion. They fail, however, to form a basal lamina or to adhere properly to the adjacent non-sensory epithelial cells, which overgrow them. The hair cells are eventually expelled from the ear epithelium into the underlying mesenchyme, losing their hair bundles in the process. It is not clear whether they undergo apoptosis: many cells staining strongly with the TUNEL procedure are seen but do not appear apoptotic by other criteria. Supporting cells, therefore, are needed to hold hair cells in the otic epithelium and, perhaps, to keep them alive, but are not needed for the construction of normal hair bundles or to give the hair bundles a predictable polarity. Moreover, supporting cells are not absolutely required as a source of materials for otoliths, which, though small and deformed, still develop in their absence.

The membranous labyrinth of the inner ear, with its three semicircular canals, originates from a simple spheroidal otic vesicle. The process is easily observed in Xenopus. The vesicle develops three dorsal outpocketings; from the two opposite faces of each outpocketing pillars of tissue are protruded into the lumen; and these paired ‘axial protrusions’ eventually meet and fuse, to form a column of tissue spanning the lumen of the outpocketing like the hub of a wheel, with a tube of epithelium forming the semicircular canal around the periphery. Each axial protrusion consists of epithelium encasing a core of largely cell-free extracellular matrix that stains strongly with alcian blue. In sections, at least 60% of the stainable material is removed by treatment with Streptomyces hyaluronidase. When Streptomyces hyaluronidase is microinjected into the core of a protrusion in vivo, the protrusion collapses and the corresponding semicircular canal fails to form. Hyaluronan (hyaluronic acid) in the core of the protrusion therefore seems to be essential in driving the extension of the protrusion. Autoradiography with tritiated glucosamine indicates that the hyaluronan-rich matrix is synthesised by the epithelium covering the tip of the protrusion; the basal lamina here appears to be discontinuous. These findings indicate that the epithelium of the axial protrusion propels itself into the lumen of the otocyst by localised synthesis of hyaluronan. Hyaluronan may be used in a similar way in the development of other organs, such as the heart and the secondary palate.

Mechanosensory hair cells in the sensory patches of the vertebrate ear are interspersed among supporting cells, forming a fine-grained pattern of alternating cell types. Analogies with Drosophila mechanosensory bristle development suggest that this pattern could be generated through lateral inhibition mediated by Notch signalling. In the zebrafish ear rudiment, homologues of Notch are widely expressed, while the Delta homologues deltaA, deltaB and deltaD, coding for Notch ligands, are expressed in small numbers of cells in regions where hair cells are soon to differentiate. This suggests that the delta-expressing cells are nascent hair cells, in agreement with findings for Delta1 in the chick. According to the lateral inhibition hypothesis, the nascent hair cells, by expressing Delta protein, would inhibit their neighbours from becoming hair cells, forcing them to be supporting cells instead. The zebrafish mind bomb mutant has abnormalities in the central nervous system, somites, and elsewhere, diagnostic of a failure of Delta-Notch signalling: in the CNS, it shows a neurogenic phenotype accompanied by misregulated delta gene expression. Similar misregulation of delta; genes is seen in the ear, along with misregulation of a Serrate homologue, serrateB, coding for an alternative Notch ligand. Most dramatically, the sensory patches in the mind bomb ear consist solely of hair cells, which are produced in great excess and prematurely; at 36 hours post fertilization, there are more than ten times as many as normal, while supporting cells are absent. A twofold increase is seen in the number of otic neurons also. The findings are strong evidence that lateral inhibition mediated by Delta-Notch signalling controls the pattern of sensory cell differentiation in the ear.