Development 122, 501-507 (1996) Printed in Great Britain © The Company of Biologists Limited 1996 DEV4634
501
Regulative interactions in zebrafish neural crest David W. Raible* and Judith S. Eisen Institute of Neuroscience, University of Oregon, Eugene, Oregon 97403-1254, USA *Author for correspondence at present address: University of Washington, Box 357420, Department of Biological Structure, Seattle, WA 98195-7420, USA (e-mail:
[email protected])
SUMMARY Zebrafish trunk neural crest cells that migrate at different times have different fates: early-migrating crest cells produce dorsal root ganglion neurons as well as glia and pigment cells, while late-migrating crest cells produce only non-neuronal derivatives. When presumptive earlymigrating crest cells were individually transplanted into hosts such that they migrated late, they retained the ability to generate neurons. In contrast, late-migrating crest cells transplanted under the same conditions never generated neurons. These results suggest that, prior to migration, neural crest cells have intrinsic biases in the types of derivatives they will produce. Transplantation of presumptive early-migrating crest cells does not result in production of
dorsal root ganglion neurons under all conditions, suggesting that these cells require appropriate environmental factors to express these intrinsic biases. When earlymigrating crest cells are ablated, late-migrating crest cells gain the ability to produce neurons, even when they migrate on their normal schedule. Interactions among neural crest cells may thus regulate the types of derivatives neural crest cells produce, by establishing or maintaining intrinsic differences between individual cells.
INTRODUCTION
cells within a forming tissue. For example, differentiated cells can influence the types of cells undifferentiated precursors may produce. This type of feedback interaction has been proposed to occur in the developing retina (Reh, 1992) and, during neural crest development, in the formation of peripheral ganglia (Shah et al., 1994). Alternatively, precursor cells can influence the fates of other types of precursor cells before they differentiate. For example, interactions between heterologous precursors occurs during sea urchin development, in which primary mesenchyme prevents secondary mesenchyme from forming skeleton (Ettensohn, 1992). A special form of regulative interactions occurs during lateral specification within invertebrate equivalence groups (reviewed by Greenwald and Rubin, 1992). In equivalence groups, cells make hierarchical fate choices based on signaling among group members. Although cells within the group are initially equivalent, signals from cells that assume the primary or default fate cause other members of the group to assume a secondary or alternate fate. When cells destined to follow the primary fate are removed, they are replaced by cells that would otherwise follow the secondary fate. In systems that undergo regulative development, a cell’s developmental potential is not simply defined by its fate. Cell fate represents what a cell will do in its usual environment as the normal outcome of development. In contrast, cell potential encompasses all the possible fates a cell may undertake given appropriate environmental conditions (Weiss, 1939; Slack, 1991). As development proceeds, cells become fate-restricted so that their progeny express only a subset of possible fates.
A basic premise of developmental biology is that cell fate decisions result from interplay between intrinsic factors and signals from the surrounding environment. The neural crest is a favorite system for studying cell fate specification since it begins as a population of cells that later forms diverse derivatives including neurons and glia of the peripheral nervous system, pigment cells and craniofacial mesenchyme, after it migrates from the neural tube (Horstadius, 1950; Weston, 1970; Le Douarin, 1982). Current models of neural crest development discuss the relative importance of intrinsic and extrinsic factors in determining cell fate (Weston, 1991; Anderson, 1994; Le Douarin et al., 1994). In some developing systems, interactions among cells within a population contribute to cell fate decisions, such that the cells inhibit their neighbors from following the same developmental pathway; such interactions can be considered regulative. In the work described in this paper, we provide evidence that regulative interactions among zebrafish neural crest cells play a role in neural crest cell fate decisions. Ideas about developmental regulation first arose from experiments by Driesch (1892), in which isolated sea urchin blastomeres each developed into a whole organism. Regulative phenomena also occur at the tissue level, such as in limb regeneration (Bryant et al., 1992) or, with respect to the neural crest, regeneration of the dorsal neural tube and associated crest after unilateral ablation (Scherson et al., 1993). At the cellular level, regulative interactions are involved in the differentiation of
Key words: Danio rerio, neural crest, cell fate, lateral specification, dorsal root ganglia
502
D. W. Raible and J. S. Eisen
Restrictions in fate are defined experimentally by following individual cells and identifying divisions after which progeny give rise to a limited set of derivatives. However, restrictions in fate do not necessarily imply restrictions in cell potential; that a cell’s presumptive fate is different from its potential was first recognized by Driesch (1892). Restrictions in potential can be identified experimentally by challenging cells with new environments and determining whether they change their developmental program. Zebrafish neural crest cells express tissue-specific markers and display characteristic cell behaviors, revealing that they have become specified, before reaching their final locations (Schilling and Kimmel, 1994; Raible and Eisen, 1994). They also undergo lineage restrictions to produce precursors that give rise to a single derivative type. Although fate restrictions are indicative of cell specification, they say nothing about restrictions in potential, since specification may be conditional (Davidson, 1990; Kimmel et al., 1991). In this paper, we suggest that regulative interactions among neural crest cells play a role in how they become specified. By transplanting individual cells from different neural crest subpopulations, we demonstrate that they have different intrinsic biases in the types of derivatives they will make. We find that after ablation of specific subpopulations of neural crest cells, remaining neural crest cells are able to compensate and generate derivatives they normally never produce. Although two defined populations of neural crest cells have different intrinsic biases in the types of derivatives they will produce, under appropriate conditions they can both produce the same derivative types, suggesting they initially have the same developmental potential. We propose a model where cell fate decisions are influenced by interactions among neural crest cells, and draw parallels to lateral specification in invertebrate equivalence groups. MATERIALS AND METHODS Animals Embryos were obtained from the zebrafish colony at the University of Oregon, and were staged by hours post-fertilization at 28.5°C (h; Kimmel et al., 1995). Chorions were removed with watchmaker forceps and living embryos were mounted for observation between coverslips held apart by spacers (Raible et al., 1992). When necessary, embryos were immobilized in a dilute solution of tricaine methylsulfonate (Sigma). Cell ablation Embryos were mounted in 1.2% agar so that neural crest cells could be visualized under Nomarski (DIC) optics (Raible et al., 1992). Premigratory neural crest cells were removed by aspiration with a pipette whose tip was manually broken to a diameter of about 20 µm. The suction pipette was inserted into the embryo through a hole produced manually with fine glass needles. Alternatively, neural crest cells were ablated by laser-irradiation as described previously (Eisen et al., 1989). Irradiated cells were observed for 5-10 minutes to ensure that they did not recover. Single-cell transplantation Transplants of single neural crest cells were performed essentially as described for individual motoneurons (Eisen, 1991; Eisen and Pike, 1991). Briefly, donor embryos were labeled at the 2-8 cell stage with lysinated rhodamine dextran (10×103 Mr; Molecular Probes). Labeled
donor and unlabeled host embryos were mounted side by side in agar. Individual neural crest cells were removed from segments 6-8 of donor embryos by gentle suction using a micropipette whose tip was manually broken to a diameter of about 20 µm. The micropipette was withdrawn from the donor embryo, monitored to ensure the removal of a single crest cell and inserted into the host embryo. The cell was then expelled with gentle pressure onto the dorsolateral aspect of the neural tube at the level of segments 5-8 of host embryos. The time of migration onset for the transplanted cell was established by monitoring host embryos at half-hour intervals. Progeny of transplanted cells were identified at 2 and 3 days of development. Intracellular labeling and antibody staining Neural crest cells were labeled by intracellular injection with lysinated rhodamine dextran (10×103 Mr; Molecular Probes) as described (Raible et al., 1992). Labeled cells were monitored using low light level, video-enhanced fluorescence microscopy and images were captured on a Macintosh IIci using the Axovideo program (Axon Instruments; Myers and Bastiani, 1991). For whole-mount antibody staining, embryos were fixed overnight in 4% paraformaldehyde at 4°C and rinsed several times with glassdistilled H2O. Embryos were incubated for 1 hour in blocking buffer (PBS with 1% BSA, 1% DMSO, 0.1% Triton-X 100), then incubated overnight in primary antibody at 4°C. Embryos were rinsed in wash buffer (PBS with 0.1% Triton-X 100), incubated overnight in horseradish peroxidase (HRP)-conjugated goat anti-mouse secondary antibody (Sternberger) at 4°C, rinsed in wash buffer, and incubated overnight in mouse PAP (Sternberger) at 4°C. Embryos were then rinsed in wash buffer followed by 0.1 M phosphate buffer, and then incubated in 50 µg/ml diaminobenzidine with 0.01% H2O2 in 50 mM phosphate buffer to develop the HRP reaction product. For antibody staining of sections, embryos were oriented in blocks of 1.5% agar in 5% sucrose and then incubated overnight in 30% sucrose. Cryostat sections were incubated for 30 minutes in blocking buffer, 2 hours in primary antibody, and 1 hour in fluorescein-conjugated secondary antibody (Cappell). The neuron-specific anti-Hu monoclonal antibody (Marusich et al., 1994) was obtained from Michael Marusich at the University of Oregon. The zn-5 monoclonal antibody (Trevarrow et al., 1990) was obtained from Ruth BreMiller at the University of Oregon.
RESULTS Intrinsic differences between neural crest cells are revealed by cell transplantation Zebrafish trunk neural crest cells at the same axial level that migrate at different times produce different derivatives (Raible and Eisen, 1994); this characteristic can be used to define two different cell types. There are fewer neural crest cells in zebrafish than in tetrapod embryos, with only 10-12 cells per trunk segment, but in other respects, zebrafish trunk neural crest cells are similar in the migration pathways they follow and the derivatives they make (Raible et al., 1992). Earlymigrating crest (EMC) cells constitute a group of 5-8 cells per hemisegment positioned on the dorsolateral aspect of the neural tube. At the level of somite 7 in the embryo, EMC cells begin to migrate between 16.5 and 18h on a medial path between the somite and neural tube, and generate all types of neural crest derivatives, including dorsal root ganglion (DRG) neurons, glial cells and pigment cells. In contrast, latemigrating crest (LMC) cells are positioned medially on the dorsal neural tube, begin to migrate on the medial pathway after 18h, and give rise to glial cells and pigment cells but do
Zebrafish crest cell interactions Table 1. Cell transplantation reveals intrinsic biases Cell type EMC LMC EMC Native EMC†
Host stage (h)
Migration times of transplanted cells (h)
Cells producing DRG neurons (%)
15-16 15-16 18-19 −
18.5-20.5 (19.25) 17.5-20.5 (19.1) 21-23.5 (22.1) 16.5-18 −
4/21 (19) 0/30 (0)* 0/39 (0)* 22/63 (35)
EMC cells and LMC cells were removed from hosts just before beginning to migrate. EMC cells from 16.5h donor embryos and LMC cells from 18h donor embryos were transplanted into hosts of ages indicated. Of 191 cells transplanted, 94 survived to generate neural crest derivatives. Cells recover 34.5 hours after transplantation before they begin to migrate, and this characteristic was used to examine the fates of EMC cells under two different environmental conditions. The range and average migration times (parenthesis) were determined by monitoring 8-10 transplanted cells for each experimental condition at half-hourly intervals to establish when cells first entered the medial migration pathway. As described for native non-neuronal clones (Raible and Eisen, 1994), transplanted non-neuronal clones consisted of pigment and glial cells. Native EMC cells that produce DRG neurons also often produce pigment or glial cells; this was true for 3 of the 4 transplanted EMC cells described in the first line of this table. *, significantly different from EMC cells transplanted into 15-16h hosts with P