The Journal of Neuroscience, December 1990, 70(12): 39473959 The Formation of Terminal Fields in the Absence of Competitive Interactions Among Primary Motoneurons in the Zebrafish Dennis W. C. Liua and Monte Westerfield Institute of Neuroscience, University of Oregon, Eugene, Oregon 97403 To make specific synaptic connections, projection neurons Evidence from studiesof several vertebrate speciessuggests that extend neurites to regions containing appropriate targets, the growth cones of spinal motoneurons grow directly to the then form synapses with the correct type and number of regions of their appropriate targets, and, if displaced experi- target cells. To investigate the mechanisms controlling this mentally, they will take alternative routes to reach appropriate process, we have studied the formation of motoneuronal targets (Landmesser,1980; Lance-Jonesand Landmesser,198 1; terminal fields in live zebrafish embryos. The primary mo- Fare1and Bemelmans, 1985). In the zebrafish, precisecell-spe- toneurons of the zebrafish are identifiable as individuals and cific pathfinding by the growth cones of identified pioneer mo- innervate neighboring but mutually exclusive territories. To toneurons has been observed directly in living embryos (Eisen study the first week of their development, which includes et al., 1986; Myers et al., 1986). These studies demonstrated embryonic and early larval stages, we labeled identified mo- that individual primary motoneurons extend axons directly to toneurons with fluorescent dyes and made sequential ob- cell-specific locations, where they branch into characteristic ar- servations of the axonal branches of individual neurons. We easof segmentedaxial muscle(Myers et al., 1986). More recent assessed the roles of competitive interactions and synapse studies have shown that zebrafish primary motoneurons follow elimination in the formation of specific synapses by identi- correct pathways in the absenceof either activity-dependent fied neurons that innervate neighboring territories. interactions between the motoneurons and their targets (Wes- Our results demonstrate that primary motoneurons estab- terfield et al., 1990) or competitive interactions among the mo- lish their cell-specific terminal fields primarily by directed toneurons (Eisen et al., 1989; Pike and Eisen, 1990). outgrowth of branches and formation of neuromuscular junc- Correct pathfinding can lead a growth cone to the proper target tions almost exclusively on appropriate muscle fibers, rather region, but pathfinding alone cannot account for the precise than by overproduction and selective elimination of inap- shapeand size of a neuron’s terminal field. Once a growth cone propriate branches. Retraction of the few branches that are reachesthe appropriate target region, there must be mechanisms inappropriately placed, though correlated in time with the that constrain its growth and determine its choice of synaptic ingrowth of branches from appropriate motoneurons, occurs partners. Growth cones may be inhibited from extending into independently of the influences of these other cells and a particular region, as seemsto be the case for the segmental when neuromuscular transmission is blocked. We suggest specificity of zebrafish primary motoneurons, which do not cross that, similar to the way in which they pioneer peripheral nerve the borders between muscle segments(Frost and Westerfield, pathways, primary motoneurons establish their cell-specific 1986; Myers et al., 1986; Westerfield, 1987). On the other hand, terminal fields using mechanisms that operate indepen- somemotor axons initially branch exuberently in target regions dently of activity and competition. The target or substrate and later eliminate redundant projections (Purves and Licht- interactions that are likely to instruct directed growth-cone man, 1985). The elimination of synapsesrequires regressive navigation may be similar to the interactions that determine events such as cell death or selective retraction of some of a the locations of territorial borders and that instruct the re- neuron’s branches. In embryonic mammals and birds, and in traction of misplaced branches. larval amphibians,many motoneuronsdie, including somewhose axons have reachedthe target region (Oppenheim, 198 la; Betz, Pathfinding by the growth conesof pioneering peripheral axons 1987). During later developmental stages,large numbersof neu- has been studied extensively in a variety of insects(Bentley and romuscular synapsesare eliminated (Purves and Lichtman, Keshishian, 1982; Goodman et al., 1982; Blair and Palka, 1985). 1980). Regressiveevents also occur in invertebrate nervous sys- In casesin which the final specificity of synaptic connections is tems in which they may play a role in establishing synaptic known, growth conesextend directly toward appropriate targets specificity (Murphey, 1986; Gao and Macagno, 1987). In most (Ball et al., 1985; Johansen et al., 1989; Myers et al., 1990). systems, regressiveevents appear to be regulated by activity- dependent competitive interactions among neurons (Oppen- heim, 198lb; Purves and Lichtman, 1985; seealso Betz et al., Received May 8, 1990; revised Aug. 6, 1990; accepted Aug. 10, 1990. 1990). This work was supported by NIH Grants NS21132, HD22486, and GM07257, To examine the relative roles of directed growth and synapse the Murdoch Foundation, and an RCDA to M.W. We wish to acknowledge Har- rison Howard and Sean Poston for expert photographic assistance and Judith elimination in the establishmentof preciseterminal fields, and Eisen. Charles Kimmel. and Janis Weeks for comments on the manuscriet. to assessthe influence of activity-dependent competitive inter- Correspondence should be addressed to Monte Westerheld at the above address. actions in regulating the shape and size of axonal arbors, we a Present address: Department of Genetics, University of Washington, Seattle, WA 98195. studied the formation of terminal fields by primary motoneu- Copyright 0 1990 Society for Neuroscience 0270-6474/90/123947-13%03.00/O rons in the zebrafish embryo. In the adult, these motoneurons 3948 Liu and Westerfield l Development of Motoneuronal Terminal Fields ROP w CaP Figure 1. Terminal fields of the CaP and RoP motoneurons are mu- tually exclusive by third day of development. A composite drawing of Figure 2. Individual motoneurons labeled with fluorescent dyes. A a 62-hr CaP motoneuron (thick lines) and a 66-hr RoP motoneuron Nomarski photomicrograph illustrates the labeling method in a sagittal (thin Zincs) labeled with Di-I in different animals. The segment bound- view of a live 19-hr zebrafish embryo with the spinal cord at the top aries are shown as slanted lines. Rostra1 is to the left and dorsal to the and the notochord below. A Di-I-filled micropipette has been advanced top in this and all subsequent figures. The dorsal part of the segment is through the spinal cord via a small opening in the dorsal skin and is not shown. The CaP axon courses through RoP territory in the region prepared to eject dye directly onto the cell body of a CaP motoneuron. of the horizontal septum (ITS, dotted he). The CaP motoneuron is the Scale bar, 20 pm. only primary motoneuron to contact the ventral portion of the muscle. Segment 13, right for CaP and 12 left for RoP. Scale bar, 25 pm. Materials and Methods Animals. Zebrafish embryos were obtained from a laboratory colony maintained on a 14-hr light, lo-hr dark cycle. Embryos were reared at 28.5”C and staged as previously described (Westerlield, 1989). All em- innervate adjacent, mutually exclusive muscle territories (Wes- bryo ages are reported as hours postfertilization (hr) when raised at terfield et al., 1986). In the embryo, primary motoneurons are 28.5%. The nit-l(b107) mutation is a recessive, lethal mutation that the first neurons to send growth cones out of the spinal cord. was induced with gamma rays (Westerfield et al., 1990). Homozygous We examined the formation of terminal fields by 3 of the pri- mutants are distinguished from wild-type embryos by their lack of spon- mary motoneurons, but concentrated our efforts on the CaP and taneous and touch-evoked movements. The muscles of nit-1 mutants respond to direct electrical stimulation but not to application of cho- RoP motoneurons (Eisen et al., 1986), because these neurons linergic agonists. The gal- I (b 1) viable recessive mutation (Streisinger innervate adjacent muscle territories and because the CaP growth et al., 198 1) affects pigmentation of the zebrafish, mutant embryos re- cone must extend through putative RoP territory to reach the main transparent for a much longer developmental period than wild region it ultimately innervates (Fig. 1). The CaP and RoP growth types. Labeling motoneurons. Individual motoneurons were identified with cones follow a common pathway out of the spinal cord, and differential interference contrast (DIC) microscopy in the spinal cords their axons are closely associated until they reach the horizontal of 17-30-hr embryos. The tips of blunt (0.5-5-pm tip diameter) micro- septum, where they then proceed along divergent paths. By pipettes were filled with a dilute solution (0.1-l%) of l,l’-dioctadecyl- labeling individual CaP and RoP motoneurons with long-lasting 3,3,3’,3’-tetramethylindocarbocyanine perchlorate (Di-I) or 3,3’-dioc- vital fluorescent dyes, we were able to observe them directly as tadecyloxacarbocyanine perchlorate (Di-0, Molecular Probes) in NJGdimethylformamide. Embryos were embedded in a warm (40°C) they established their