The olfactory placodes of the zebrafish form by convergence of cellular fields at the edge of the neural plate

In vertebrates, some derivatives of the peripheral nervous system (PNS), including primary sensory neurons of the inner ear and the olfactory epithelium, develop from specialized tissues called placodes. Placodes are thought to arise from cells along the lateral edge of the anterior neural plate lying between neural and non-neural ectoderm (Landacre, 1910; Verwoerd and van Oostrom, 1979; Northcutt and Gans, 1983; Northcutt, 1996). Placodes of the PNS appear as distinct structures once the neural tube has formed, yet the morphogenetic movements leading to their formation are still unclear.

Although many studies have described the development of the olfactory sensory system after the appearance of the placodes (Farbman, 1988, 1992; Hansen and Zeiske, 1993), few studies have examined the processes leading to the formation of the olfactory placodes. The currently accepted hypothesis is that the olfactory placodes arise from groups of cells that separate from the lateral edge of the anterior neural plate through differential cell movements (Verwoerd and Van Oostrom, 1979; Farbman, 1992). Later, these isolated groups of cells proliferate to give rise to the olfactory placodes. Although fate-map studies in the chick support this hypothesis that the olfactory placodes arise from the neural ridge (Couly and Le Douarin, 1985), isolated groups of cells lying next to the anterior neural plate have not been described. Thus, it is still unclear whether a group of cells separates from the edge of the anterior neural plate to form each olfactory placode or whether the olfactory placodes develop within the anterior neural plate.

To understand the cellular events leading to placode formation, we have investigated the development of the olfactory placodes in the zebrafish, Danio rerio, where we can follow the fates of individual cells in the neural plate. In the zebrafish, the olfactory placodes appear as thickenings of the ectoderm by 17-18 hours after fertilization (h) and these thickenings later invaginate to form the naris, or opening of the nose, by 32 h (Hansen and Zeiske, 1993). Previously we have shown that the first axons projecting from the olfactory placodes reach the central nervous system (CNS) by 22 h. These early axons extend from pioneer neurons and establish the initial pathway later followed by the axons of the olfactory sensory neurons (Whitlock and Westerfield, 1998). The pioneer neurons arise in the anterior end of the neural plate from a distinct lineage that does not include olfactory sensory neurons (Whitlock and Westerfield, 1998). Thus, by the 4-somite stage (12 h), olfactory fates are already regionalized within the developing neural plate.

To understand how the placodes form, we injected single cells with lineage tracer dyes in the anterior end of the neural plate prior to the morphological appearance of the olfactory placode and followed the fates of these cells until after the olfactory placode formed. Surprisingly, we found that a large field of cells widely distributed along the edge of the anterior neural plate contributed to the olfactory placode. This field of cells is regionalized such that cells in specific anteroposterior locations give rise to specific parts of the developing olfactory organ. We found that cells adjacent and medial to this olfactory placode field give rise to the telencephalon. This telencephalic field is also regionalized along the anteroposterior axis, such that anterior olfactory bulb arises from cells adjacent to the region from which the first sensory neurons derive. Thus, in contrast to the hypothesis that the olfactory placodes arise from groups of cells that separate from the neural plate, we show here that the placodes arise from large fields of cells along the anterior-lateral neural plate. We suggest that they form by anterior convergence of these cellular fields.

Zebrafish embryos were obtained from natural spawnings of laboratory strains of wild-type fish (AB; Walker, 1998). Fish were staged by morphology (Kimmel et al., 1995) and developmental times are expressed as hours postfertilization at 28.5°C (h).

Embryos at the 4- to 5-somite stage (approx. 12 h) were embedded in agar with a window cut to expose the lateral edge of the neural plate. These blocks were mounted on a microscope slide and covered with physiological saline. Microelectrodes (approximately 300 MΩ resistance) were pulled (Sutter Instruments) and backfilled with a mixture of Rhodamine dextran dye (2%; Mr 3000, Molecular Probes) and Fluorescein dextran dye (2%: Mr 3000, Molecular Probes) in 0.2 M KCI. Cells were visualized with a 40× water immersion lens using a fixed-stage microscope (Zeiss). After penetration of a single cell, dye was passed into the cell by ‘ringing’ the electrode using capacitance compensation and an oscilloscope to measure the amplitude and duration of the ringing. Successful filling of the cell was judged by a rapid check of the Rhodamine fluorescence. Preparations for lineage tracing were allowed to develop to the desired age and then fluorescent images were collected on a confocal microscope (Zeiss) and reconstructed using the Voxel View and Voxel Math programs (Vital Images). Non-fluorescent images were obtained by using the anti-Fluorescein antibody (Boehringer Mannheim, 1:1000) coupled with the alkaline phosphatase Fast Red (Boehringer Mannheim, red) or NBT/BCIP (Boehringer Mannheim, blue). In general, only one side of the embryo was labeled in the lineage tracing experiment, except when labeling the pioneer neurons. To label cells in the region of the pioneer neurons, animals were mounted dorsal side up and approached from the anterior end. This made it possible to label cells on both sides of the neural plate so these preparations had two placodes to score per animal. The fate map was constructed from the results of 175 successful single cell labels out of 468 cells attempted representing an overall success rate of 37%, although on different days the success rate varied greatly. The average clone size was 2.8 (s.d.±1.0) cells with a range of 2-5 cells.

Cells at the edge of the neural plate and subsequently in the developing olfactory placode (Ekker et al., 1992; Akimenko et al., 1994) and in the developing forebrain (Papalopulu and Kintner, 1993) express the gene distal-less 3 (dlx3). Premigratory cranial neural crest cells express the forkhead 6 gene (fkh6; Odenthal and Nüsslein-Volhard, 1998). RNA probes generated with template for the zebrafish dlx3 and fkh6 genes were used for in situ hybridization. RNA in situ hybridization was performed using digoxigenin (DIG)-labeled RNA probes (Thisse et al., 1993) for the dlx3 gene and fluoresceinated probes for the fkh6 gene. The probe colors were also reversed to compare results. We found that it is better to label the weaker probe with DIG because it gives a stronger signal. The NBT/BCIP blue coloration was performed first to prevent ‘blueing’ in the subsequent red coloration. The reaction was stopped, the preparations rinsed once in dH2O, rinsed in high pH buffer (pH 8.2), rinsed once in dH2O, the probe stripped (0.l M glycine, pH 1.8), rinsed 3× in PBS, and then followed by the Fast Red reaction. After rinsing in PBS, the preparations were postfixed in 4% paraformaldehyde. Again the signal from the NBT/BCIP blue coloration was generally stronger than that from the Fast Red coloration.

We correlated the dlx3 expression domain with our lineage tracing fate map by combining lineage tracing labeling and in situ hybridization using the dlx3 mRNA probe. Embryos were fixed immediately after injecting a single cell, rinsed, heat blocked for 1 hour at 70°C, incubated in blocking solution (PBDT: 0.1 M phosphate buffer, 2% bovine serum albumin, 1% dimethylsulfoxide [DMSO] and 0.5% Triton X-100 with 2% normal goat serum) and processed with alkaline phosphatase anti-Fluorescein antibody and reacted with Fast Red. After the reaction was complete, the embryos were rinsed, postfixed and processed for dlx3 gene expression (see preceding section). In this way, we could align our single cell labeling with the expression domain of the dlx3 gene.

Embryos were collected and maintained at 28.5°C in embryo medium (13.7 mM NaCl, 0.54 mM KCI, 1.3 mM CaCl₂, 1.0 MgSO₄, 0.044 mM KH₂PO₄, 0.025 mM NaH₂PO₄, and 0.42 mM NaHCO₃ at Ph 7.2). They were allowed to develop until 6-(approx. 12 h), 10-(approx. 14 h), 14-(approx. 16 h), and 18-(approx. 18 h) somite stages and 20 embryos from each stage were fixed overnight at 4°C in 4% paraformaldehyde in 0.l M phosphate buffer with 0.15 mM CaCl₂ and 4% sucrose (pH 7.3). After rinsing 2× in PBST, embryos were preabsorbed in 2% normal goat serum (NGS) in PBDT (0.1 M phosphate buffer, 2% bovine serum albumin, 1% dimethylsulfoxide [DMSO] and 0.5% Triton X-100). Animals were incubated with anti-phosphorylated Histone H3, (Upstate Biotechnology, NY) 1:800 in PBDT+2%NGS overnight at 4°C, which labels cells in M phase (Hendzel et al., 1997), then rinsed 3× over 2 hours and placed in goat anti-rabbit secondary antibody, (Sternberger Monoclonals Inc.) at 1:200. After a series of washes, embryos were incubated in peroxidase-rabbit anti-peroxidase complex (Sternberger Monoclonals Inc.) overnight at 4°C. Embryos were then rinsed in 0.1 M phosphate buffer and 1% DMSO 3× for 1 hour. The coloration reaction was done using 0.02% DAB (diaminobenzidine) in 1:1 dH2O/PBS with 1% DMSO and 0.003% hydrogen peroxide. The preparations were processed through a glycerol series (30%, 50%, 70%, 90%, glycerol: 0.1 M phosphate buffer), mounted in 100% glycerol and viewed with Nomarski optics.

Initially, we used morphological landmarks visible in live embryos to define cellular fields and then we correlated the positions of these fields with the gene expression domains of dlx3 and fkh6. In the zebrafish, cells thought to form the olfactory and auditory placodes express dlx3 at early segmentation stages (Akimenko et al., 1994; Ekker et al.,1992). At the 4- to 5-somite stage (12 h), a time prior to the morphological differentiation of the olfactory placode, cells in a broad, lateral stripe extending around the anterior end of the neural plate to the posterior edge of the eyes express dlx3 (Fig. 1A,B, red). In addition, cells in the more medial part of the anterior neural plate express low levels of dlx3 (Figs 1A, 2). This difference in the intensity of dlx3 expression at the edge versus the center of the developing neural plate was observed in 1 μm confocal optical sections through the anterior neural plate (data not shown) suggesting that it is due to differences in the level of expression, not to the thickness of the layer of cells expressing dlx3 in these regions. As the olfactory placode becomes morphologically distinct (by 17 h) expression of dlx3 is localized within the region of the placode with weaker expression in the developing telencephalon.

Cells of the premigratory neural crest express the gene, fkh6 (Odenthal and Nüsslein-Volhard, 1998), and by the 4- to 5-somite stage, the anterior limit of fkh6 expression (Fig. 1A,B, blue) aligns with the posterior border of the broad dlx3 expression domain (Fig. 1A,B, red). The interface of these two expression domains also corresponds to the posterior border of the developing eye (Fig. 1A,B,D, pink arrow). Thus the locations of cells expressing dlx3 and fkh6 at the 4- to 5-somite stage can be predicted based on their positions relative to the developing eyes. Posterior to the developing eyes and lateral to the fkh6 domain, a strip of cells approximately one cell diameter wide expresses dlx3. This expression domain extends posteriorly and broadens in the area where the ear will form.

To define the region of the neural plate giving rise to the olfactory placode, we constructed a fate map of cells in the anterior neural plate (Fig. 3A) at the 4- to 5-somite stage (12 h: Fig. 1E, lateral view; Fig. 3B, dorsal view). In a lateral view at the 4- to 5-somite stage (12 h), the outline of the developing eye can be seen in the live embryo (Fig. 1D). Using this border as a landmark, we subdivided the field of cells extending from the anterior end of the neural plate into five regions where region I is the area at the anterior end of the neural plate and regions II-V lie medial to the developing eye field at progressively more posterior positions (Fig. 1E). Each region was approximately eight cell diameters along the anteroposterior axis. At 12 h (4-5 somites), we labeled single cells in the region extending from the anterior end of the neural plate to the posterior edge of the developing eye (Fig. 1D, arrow). We allowed the labeled embryos to develop until 50 h and examined the positions and morphologies of the clones resulting from the injected cells.

We scored the fates of the progeny of the injected cells and color-coded them according to fate on a diagram of the anterior neural plate (Figs 1E, 3B,C). The lateral edge of the neural plate (Fig. 3B, border between medial and lateral is depicted by the heavy black shading extending along the anteroposterior axis) gave rise to cells in the olfactory organ. This domain of cells always giving rise to fates in the olfactory organ extends an average of four cell diameters (cell diameter approx. 12 μm) medially from the edge of the developing eye (Fig. 1D), and strongly expresses dlx3 (Fig. 1C). This ‘olfactory organ domain’ (Fig. 3B, multicolor) extended posteriorly to the premigratory cranial neural crest domain, indicated by the anterior border of fkd6 expression (horizontal heavy black shading, Fig. 3B).

Cells in different regions of the olfactory domain of the neural plate contributed to characteristic regions of the olfactory organ. The anterior neural plate gave rise to the anterior, medial part of the olfactory organ (Fig. 3B,C, orange) whereas the lateral region, lying next to the eye, arose from more posterior cells in the neural plate (Fig. 3B,C, red). The most posterior region of the olfactory organ domain gave rise to the most dorsal region of the olfactory organ (Fig. 3B,C, green). This general spatial organization was consistent from embryo to embryo. There were no sharp borders and cells were somewhat intermixed with respect to their future fates (Fig. 3B), although this may result from slight differences when trying to match positions in different embryos.

Cells lying medially more than four cell diameters from the eye, (Fig. 2, line with double-head arrow) gave rise to progeny in the telencephalon and showed lower dlx3 expression. This telencephalic domain (Fig. 3B,C, yellow) was organized so that the most anterior region gave rise to the most ventral posterior region of the telencephalon near the postoptic commissure whereas the most posterior medial neural plate contributed to dorsal telencephalon (Table 1). Like the olfactory domain, the telencephalic domain in the anterior neural plate extended posteriorly to the premigratory cranial neural crest domain (Fig. 3B). This fate map demonstrates that the olfactory placode and the telencephalon develop from distinct cellular fields at and adjacent to the edge of the neural plate.

The injected cells in the lateral domain of the neural plate gave rise to clones of progeny primarily in the olfactory organ (Table 1). Whole-mount preparations were viewed from the ventral side (Fig. 4A) because this was the best orientation from which to see the entire olfactory organ at 50 h. Clones containing sensory neurons with axons that were the first to extend into the developing olfactory bulb, arose from progenitors in the area posterior to where the nose placode later arose (Table 1, region IV; Figs 1E, 3B,C, red cells). The cell bodies of these neurons lay laterally within the olfactory organ and their axons extended into the region of the developing olfactory bulb (Fig. 4B, arrowhead). The neurons in the anterior region of the olfactory organ at 50 h (Fig. 4C) arose primarily from region II (Figs 1E, 3B,C, orange; Table 1). Clones of cells in the dorsal olfactory organ came primarily from cells in region V (Figs 1E, 3B,C, green; Table 1). In general, the majority of labeled neurons (excluding pioneer neurons) in the olfactory organ had cell bodies and dendrites, but no axons by 50 h (51/64; Table 1). The few neurons with axons (Fig. 4A,B,D) had projections that turned dorsally in the developing olfactory bulb (Fig. 4D, arrowhead). With the exception of pioneer neurons, cells with axons always had dendrites. Thus, the common pattern was to extend dendrites first, then axons later (Fig. 4E). Occasionally we observed a cell type in the extreme ventral part of the nose, which we termed ‘ventral neuron’ (n=2; Table 1; Fig. 4F). The axons of these ventral neurons grew dorsally to meet the olfactory nerve and then abruptly turned posteriorly and extended in the anterior commissure (Fig. 4F, arrowhead). The most anterior part of region I of the dlx3 domain gave rise to structural elements of the nose and ethmoid plate, and neurons in the pituitary (Fig. 4G), but rarely (n=3/24) to olfactory neurons. Also, cells scattered throughout the lateral dlx3 expression domain produced support cells of the olfactory epithelium. These cells generally extended processes from the apical to basal borders of the olfactory epithelium and were fairly uniform in width throughout their length (Fig. 4H, arrow).

To determine how far medially the field of cells giving rise to the olfactory organ extended, we labeled single cells (Fig. 5A, red) more medially, in an area where the intensity of dlx3 gene expression is lower (Fig. 2, double-headed arrow; Fig. 5A). Cells in this medial domain gave rise to progeny that populated the telencephalon, including the developing olfactory bulb (Table 1). In region II (Fig. 1E), they gave rise to progeny in the ventral telencephalon near the anterior commissure, whereas cells labeled in region III (Fig. 1E) contributed to the ventral half of the developing olfactory bulb. Cells labeled medially in regions I, II, III and V of the neural plate lacked axons at 50 h (Fig. 5B, arrow). In contrast, cells labeled in region IV (Table 1, medial; Fig. 5C, arrow) produced clones with axons, allowing us to identify them positively as interneurons. These interneurons had cell bodies in the anterior developing olfactory bulb (Fig. 5C, arrow) and axons that extended in the anterior commissure to the contralateral side (Fig. 5C, arrowhead). With Nomarski optics, there was no apparent morphological border within the neural plate between cells giving rise to olfactory organ and cells giving rise to telencephalon. Rather, there was an apparent mixing of cells in a transition zone where the probability of labeling a cell giving rise to telencephalon increased as cells were labeled more than four cell diameters from the edge of the developing eye. However, a single labeled precursor never gave rise to a clone of mixed fates i.e. olfactory organ and telencephalic cells.

Previously we described pioneer neurons in the zebrafish olfactory system that arise from a specific region of the neural plate and establish the initial axon pathway to the CNS (Whitlock and Westerfield, 1998). We present here additional data showing the initial outgrowth of the axons of the pioneer neurons as they exit the placode and the sizes of clones generated from their precursors in the neural plate. Clones derived from single cells labeled in the region of the neural plate that gives rise to pioneer cells (Whitlock and Westerfield, 1998; Table 1, region II; Fig. 1E) included neurons with large cell bodies (Fig. 6, arrows) and axons extending into the developing olfactory bulb by 28 h (Fig. 6, arrowheads). In 22 h preparations, we observed initial axon outgrowth from the pioneer neurons (Fig. 6A). As a pioneer axon grew out to meet the telencephalon, it had a large simple growth cone (Fig. 6A, arrowhead) that lacked the fine processes evident later, after it entered the region of the developing olfactory bulb (see Fig. 6B,C, arrowheads). Upon closer inspection, the axons of these older cells were very ‘bushy’ in appearance (Fig. 6B,C, arrowheads). In all cases, the cell bodies of pioneer neurons were large, round and devoid of dendrites (Fig. 6, arrows) and extended a single axon out of the olfactory placode.

Pioneer neurons developed from neural plate precursors without further cell division. The cell depicted in Fig. 6A was labeled at 12 h and scored at 22 h after it had sprouted an axon. The two cells depicted in Fig. 6B,C resulted from labeling two adjacent cells at the 4- to 5-somite stage. When scored at 28 h there were still only two cells and the cells had extended axons. This was true for all pioneer precursors (n=9) labeled in this study (see Table 1). This indicates that the cells labeled at 12 h were postmitotic and differentiated directly into pioneer neurons. In contrast, neural plate cells that gave rise to telencephalic or olfactory sensory neurons produced clones with an average size of 2.8 cells (s.d.±0.99) indicating continued proliferation before differentiation.

Localized cell division is another mechanism that could potentially contribute to the formation of the olfactory placode. We examined patterns of cell division in the 6 hours (12-18 h) prior to the morphological appearance of the olfactory placodes. To do so, we collected and fixed animals at 12 h (4-6 somites), 14 h (10 somites), 16 h (14 somites) and 18 h (18 somites) stages, and then assayed them for cell division using an antibody that recognizes dividing cells. At 12 h, there is very little cell division in the anterior neural plate (Fig. 7A). At 14h (10 somites), there is cell division in the forming ventricle (Fig. 7B, arrow), but few cells label in the developing olfactory placode field (Fig. 7B, brown bracket). At 16 h (14 somites), dividing cells continue to be localized in the developing ventricles (Fig. 7C, arrow) and, in addition, the developing eyes (Fig. 7C, arrowhead). Still no localized cell division is observed in the region of the developing olfactory placode (Fig. 7C, brown bracket). With Nomarski optics, the olfactory placode is apparent at 17-18 h in the living embryo. At this time, cell division is observed within the formed olfactory placode (Fig. 7D, dashed outline). Therefore, from 12 h through 18 h, the time period preceding the formation of the olfactory placodes, we found no evidence of extensive cell division in the regions where the olfactory placodes will form.

Early studies of the developing olfactory placode suggested that the placode develops from ectoderm adjacent to the neural plate (Zwilling, 1940; Emerson, 1944, 1945). Transplantation experiments in amphibians further suggested that the olfactory placodes arise as a result of an inductive interaction between the neural plate and the overlying ectoderm (Haggis, 1956) with the placodes arising from the ectoderm rather than the neural plate. In contrast, more recent studies using serial reconstruction of staged mouse embryos (Verwoerd and Van Oostrum, 1979) suggested that the olfactory placodes arise from the neural plate rather than from overlying ectoderm.

In support of this hypothesis, chick-quail chimeras and transplantation experiments (Couly and Le Douarin, 1985, 1987) showed that the anterior-lateral region of the neural plate contributes to the olfactory placodes. Our data in zebrafish (Whitlock and Westerfield, 1998; present study) further support the hypothesis that the olfactory placodes arise from the neural plate and are not induced from overlying ectoderm as the neural tube forms. Moreover, our results show no evidence for distinct developmental events where neural plate gives rise to the telencephalon and adjacent non-neural ectoderm gives rise to olfactory placode. Rather, these two regions of the developing nervous system, one central and the other peripheral, arise from adjacent fields of cells within the neural plate.

Currently, the favored model for the origin of the olfactory placode (Farbman, 1992) proposes that a piece of tissue separates from the neural plate through differential cell movements and gives rise to the olfactory placode. In contrast, our data suggest that the olfactory placode develops through convergence of a field of cells within the neural plate rather than from a small piece of detached neural plate. By lineage tracing, we found that olfactory placode precursor cells occupy a long narrow band, about four cells wide, at the 4- to 5-somite stage (12 h) and, by 17-18 h, these cells have converged to form an obvious placodal structure. The field of olfactory placode precursor cells in the neural plate is continuous with, and adjacent to, the field of cells that later form the telencephalon. As the olfactory placode field converges anteriorly, it maintains regionalized cell fates such that cells in the anterior olfactory placode field contribute to the anterior medial part of the olfactory organ and posterior cells form the lateral part of the nose (Fig. 8A,B).

Previous time-lapse video analysis of zebrafish development (Karlstrom and Kane, 1996) revealed a significant anterior movement of cells during early segmentation stages in the developing head. Part of this movement is due to the CNS moving forward, and part to the cranial neural crest cells migrating forward to populate the frontal mass. We propose that the anterior convergence of the olfactory placode field is part of this movement, and that this morphogenesis brings about the formation the olfactory placodes.

Our fate map also shows that cells medial to the olfactory placode field along the anterior edge of the neural plate contribute to the telencephalon. Many cells that we labeled in this region produced clones of labeled cells within the developing olfactory bulb. The positions of cells in the medial anterior neural plate at 12 h are indicative of their future fates within the developing telencephalon. Thus, cells lying most posteriorly populate the dorsal telencephalon, whereas those lying very anteriorly populate the most ventral, posterior telencephalon (Fig. 8A,C). The final positions of these clones result from both movements of the posterior cells anteriorly (Karlstrom and Kane, 1996) and flexure of the brain (Kimmel et al., 1995) that brings cells lying ventral-anterior to a ventral-posterior position. These results suggest that the telencephalon (thought of as an anterior structure like the olfactory placode) originates from a field of cells medial to the olfactory field, which extends from the anterior part of the neural plate posteriorly to the cranial neural crest domain at the 4-somite stage. Although there is no morphologically apparent border separating the olfactory placode field from the telencephalic field, the fates that we observe suggest that, at 12 h, the cells in these fields are committed to a single fate. Indeed, we never observed clones containing mixed olfactory organ and telencephalic fates arising from a single cell, indicating strict cell fate restrictions even in the absence of a clear morphological border. Within both fields, cells at the same anteroposterior position (region IV, Fig. 8A) were the first to differentiate. In the lateral field, these cells gave rise to the first olfactory sensory neurons with axons and, in the medial field, they gave rise to the first interneurons in the telencephalon with axons. The axons of these first sensory neuron extended into the area of the developing olfactory bulb where the cell bodies of the interneurons lie (Fig. 8D). It is striking that the first olfactory sensory neurons and interneurons differentiate in the same position (region IV) in adjacent fields indicating the possibility of developmentally coordinated signals controlling differentiation of olfactory sensory neurons and their neuronal partners within the future olfactory bulb. These results show, for the first time, that in zebrafish the olfactory placode and olfactory bulb develop in parallel from adjacent fields of cells in the neural plate.

In principle, localized cell proliferation could contribute to the thickening of the developing olfactory placode. Our results, however, suggest that the olfactory placodes, which are apparent by 17-18 h, form with very little cell division. This implies that cell movement, rather than proliferation, is the major contributor to the initial formation of the olfactory placodes. This finding is consistent with previous work by Harris and Hartenstein (1991) who showed that when cell division is blocked in Xenopus at the beginning of gastrulation, development proceeds rather normally through neural induction and CNS differentiation, indicating that these events can take place in the absence of cell division. In zebrafish, cell cycles lengthen at the end of gastrulation (10 h) and become more variable (Kimmel et al., 1994) making it less likely that we would have found large numbers of dividing cells during early segmentation stages (12-18 h).

Our lineage tracing results also show that a specific cell type, the pioneer neurons, differentiates in the absence of cell division. Previous work in zebrafish has defined neurons generated by the end of gastrulation (10 h; cycle 16) as ‘primary neurons’ (Kimmel and Westerfield, 1990; Kimmel et al., 1994). The pioneer neurons, whose precursors were labeled at 12 h and had differentiated into neurons by 24 h in the absence of cell division, fit the criteria as being early differentiating primary neurons. In contrast, we found that individual olfactory sensory neuron precursors in the neural plate produced two to five cells in the placode when scored at 50 h, indicating that their differentiation is accompanied by cell division. This cell division occurs after the initial formation of the placode, 12-18 h, which was the time period when we observed very little cell division. Thus, at the 4- to 5-somite stage, the precursors of pioneer neurons are already postmitotic whereas olfactory sensory neuron precursors divide twice, on average, before differentiating. The olfactory placode fields converge to form the placodes and in doing so gather cells of different mitotic futures: the postmitotic pioneer neurons that will undergo axonogenesis and then die, and the sensory neuron precursors that will divide and differentiate.

We have shown that the olfactory placode arises from a continuous field of cells in the anterior neural plate extending posteriorly to the anterior limits of premigratory cranial neural crest. This is very surprising because previous work predicted that the lens placode should lie between the olfactory placode region and neural crest (Rudnick, 1944; Jacobson, 1963). In our studies, the lens appears to arise from non-neural ectoderm lying lateral to the field giving rise to the olfactory placode (K.W., unpublished data). Because the olfactory placodes arise from cellular fields extending along the anteroposterior neural plate, this makes the olfactory field continuous with the anterior limits of the premigratory cranial neural crest domain (Fig. 8A). A characteristic of vertebrate evolution is the ‘invention’ of neural crest, a cell population proposed to arise after the appearance of cephalochordates. Our findings of adjacent olfactory placode and neural crest domains suggest that they may have arisen from the same single ancestral domain. Recent studies using Branchiostoma floridae (amphioxus; a cephalochordate) have shown that cells at the edge of the neural plate express distal-less during development (Holland et al., 1996). This observation led to the proposal that Branchiostoma may have a preneural crest like population. An alternative is that, in amphioxus, distal-less gene function is involved in setting aside peripheral sensory structures, and that the subsequent (in evolutionary time) expression of genes such as fkh6 may have subdivided this domain at the edge of the neural plate to allow for the specialization of neural crest and its derivatives. Future work coupling gene expression and cell fates in both zebrafish and amphioxus will help us to better understand the developmental and evolutionary relationship between neural crest and sensory placodes.

Our results suggest a new mechanism, different from the previously proposed model, for olfactory placode development. Rather than arising from a patch of cells that detaches from the edge of the anterior neural plate, a field of cells converges anteriorly to form the olfactory placode in the absence of localized pockets of cell proliferation. In addition, our findings show that similar morphogenesis of more medial cells produces the telencephalon. Future work following cell movements in living embryos will help us unravel the complex cellular behaviors that produce a distinct peripheral sensory structure, the olfactory placode, and its central target, the olfactory bulb.

We would like to thank Charles Kimmel and John Ewer for constructive criticisms of the manuscript. K. E. W. would like to thank A. Farbman, L. Barlow, and her laboratory members for valuable discussions. In addition K. E. W. thanks S. H. Brown and Minnie for their unwavering support of the olfactory sciences. This research was funded by a NIH/NIDCD grant (RO3DC03222) to K. E. W., a March of Dimes Birth Defects Foundation Basil O’Connor Award to K. E. W., a NIH grant (PO1HD022486) to M. W., and the W. M. Keck Foundation.