The Role of Pioneer Neurons in Guidance and Fasciculation in the CNS of Drosophila

Development 124, 3253-3262 (1997) 3253 Printed in Great Britain © The Company of Biologists Limited 1997 DEV1201

Targeted neuronal ablation: the role of pioneer neurons in guidance and fasciculation in the CNS of Drosophila

A. Hidalgo† and A. H. Brand* The Wellcome/CRC Institute, and Department of Genetics, Cambridge University, Tennis Court Road, Cambridge, CB2 1QR, UK *Author for correspondence †Present address: Department of Genetics, Cambridge University, Downing Street, Cambridge CB2 3EH, UK

SUMMARY

Although pioneer neurons are the first to delineate the axon formation, (2) the interaction between two pioneers is pathways, it is uncertain whether they have unique necessary for the establishment of each fascicle and (3) pathfinding abilities. As a first step in defining the role of pioneer neurons function synergistically to establish the pioneer neurons in the Drosophila embryonic CNS, we longitudinal axon tracts, to guide the fasciculation of describe the temporal profile and trajectory of the axons of follower neurons along specific fascicles and to prevent four pioneer neurons and show that they differ from pre- axons from crossing the midline. viously published reports. We show, by targeted ablation of one, two, three or four pioneer neurons at a time, that (1) Key words: pioneer neurons, cell ablation, CNS, Drosophila, axon no single pioneer neuron is essential for axon tract pathway, neuron

INTRODUCTION P pioneer neurons, or that the timing of axon outgrowth is crucial. For example, the A and P neurons may follow cues on The first neurons to extend their axons, the ‘pioneer’ neurons glial cells that the G neuron cannot recognise, or that are not (Bate, 1976), navigate in an environment devoid of other present at the time the G axon grows out. axons. Subsequently, the axons from later differentiating It is not clear whether an individual neuron can act as a neurons, the ‘follower’ neurons, contact the axons of the pioneer neuron (Bastiani et al., 1986), or whether pathways are pioneers and fasciculate with them to form the mature axon pioneered by a group of cells acting in concert at a given time bundles. Pioneer neurons may have intrinsic pathfinding (Raper et al., 1984). In the grasshopper embryo, the A/P pathway abilities that distinguish them from follower neurons. Alterna- is pioneered by several neurons rather than by an individual tively, they may simply be the first neurons to extend their pioneer (Raper et al., 1984). Only when all of the P pioneer axons and, in their absence, follower neurons might be able to neurons are ablated does the follower G axon stall (Raper et al., establish the axonal pathways independently. A functional 1984). This indicates that there is not a functional hierarchy study of the cellular roles of pioneer neurons would aid in the among the P neurons, but that they act together as a group of further molecular analysis of axon guidance. For instance, if pioneer cells required for the formation of the A/P fascicle. pioneer neurons are unique cell types, they might express Similarly, in the grasshopper antenna and the limb bud, axon specific adhesion, signalling or receptor molecules that are not pathways are pioneered by pairs of neurons (Bate, 1976). found in the follower neurons. Experiments to ablate the pioneer neurons of the longitudi- Cell ablation has been carried out in several different nal axon pathway have been carried out both in grasshopper organisms to address the question of whether pioneer neurons (Bastiani et al., 1986) and in Drosophila (Lin et al., 1995). The behave as a unique cell type. There is evidence from vertebrates MP1/dMP2 pathway in the grasshopper embryo is the earliest and invertebrates supporting the notion that pioneer neurons are longitudinal pathway to be formed, and is pioneered by the unique cells, and hence essential for axon guidance (du Lac et descending axons of MP1 and dMP2, along which the al., 1986; Eisen et al., 1986; Ghosh et al., 1990; Gong and ascending axon of pCC extends (Bate and Grunewald, 1981; Shipley, 1995; Klose and Bentley, 1989; Raper et al., 1984). For Jacobs and Goodman, 1989). The MP1 and dMP2 neurons in example, cell ablation in the grasshopper central nervous grasshopper embryos were ablated with a laser and the effects system (CNS) has revealed a hierarchy in neuronal pathfinding on the axon of pCC were analysed. Bastiani et al. (1986) ability: follower neurons cannot navigate in the absence of showed that pCC requires the other pioneers in order to extend pioneers (Raper et al., 1984). The A/P longitudinal pathway is its axon anteriorly. In their absence, the pCC growth cone either pioneered by two A and three P neurons. When these cells are stalls or follows the axon of its sibling neuron aCC, extending ablated, the follower G neuron is unable to extend its axon in towards the periphery. This suggests that pCC behaves in this pathway (Raper et al., 1984), indicating either that the G grasshopper as a follower neuron, and that there is an absolute neuron does not possess the pathfinding abilities of the A and requirement for the MP1 and dMP2 pioneer neurons in axon 3254 A. Hidalgo and A. H. Brand guidance. The consequences of ablating MP1 and dMP2 were that we use is non-invasive, cell-specific and cell-autonomous only studied at the time of pioneer axon extension and the effect (Hidalgo et al., 1995). We have selectively killed each of the on later follower neurons was not reported. pioneer neurons of the longitudinal tracts of the ventral nerve More recently, ablation of the MP1, dMP2, vMP2 and pCC cord of the embryonic CNS. We have also ablated several neurons in Drosophila was reported and the effects on follower pioneer neurons in combination. This has allowed us to address neurons analysed. Lin et al. (1995) reported that follower the following questions. (1) Are the pioneer neurons essential neurons were able to establish the longitudinal pathways in the for the formation of the longitudinal axon pathways? (2) Are absence of the pioneers. This contrasts with the result in pioneer neurons different from each other? (3) Do pioneer grasshopper, but might be explained by the random, mosaic neurons function in isolation or as a group of cells? Interest- nature of the ablation technique used by Lin et al. (1995). For ingly, our results differ dramatically from those that have pre- example, Bastiani et al. (1986) had previously shown that the viously been published (Lin et al., 1995). loss of a pioneer neuron in one segment can be rescued by a pioneer axon extending from a neighbouring segment. MATERIALS AND METHODS Similar discrepancies between grasshopper and Drosophila are found in pioneering the intersegmental nerve (ISN). In Flies were raised on standard Drosophila media at 25¡C. Embryonic grasshopper, the ISN, which extends from the CNS to the stages are as described by Campos-Ortega and Hartenstein (1985). periphery, is pioneered by the U neurons followed by aCC (du Lac et al., 1986). Ablation of the U neurons leads to stalling GAL4 expressing lines of the aCC axon (du Lac et al., 1986). In Drosophila, this nerve Lines that express GAL4 in a restricted cell- or tissue-specific pattern is pioneered by aCC and the U neurons extend along its axon were generated by enhancer detection (Brand and Perrimon, 1993). An enhancerless gene encoding the yeast transcriptional activator (Lin et al., 1995). Ablation of aCC causes some early defects GAL4 is inserted randomly into the Drosophila genome where, in pathfinding by the Us, but these are later corrected to form depending upon the site of integration, expression is directed by any a normal nerve (Lin et al., 1995). This suggests that aCC is not one of a diverse array of genomic enhancers. A second gene, con- required for the formation of the ISN in the fly embryo. taining GAL4-binding sites within its promoter, can then be intro- Experiments in different organisms have provided conflict- duced into this background where it will only be transcribed in those ing views on the roles of pioneer neurons (Chitnis and Kuwada, cells where GAL4 is expressed. Transgenic lines that drive GAL4 1991; Eisen, 1991; Ghosh et al., 1990; Gong and Shipley, 1995; expression in the pioneer neurons were identified by activation of a Keshishian and Bentley, 1983; Klose and Bentley, 1989; Pike UAS-tau-lacZ gene fusion (Hidalgo et al., 1995). In a screen of et al., 1992). For instance, there is evidence from zebrafish and approximately 1100 independent GAL4 insertion lines (E. Watkins, grasshopper showing that pioneer neurons are dispensable E. L. Dormand, A. Wade, R. Barbosa, S. Park, N. Keller, N. Sheard, because, after their elimination, follower neurons can find their N. Perrimon and A. H. B., unpublished data), we found two lines that express GAL4 primarily in the pioneer neurons (15J2, C544). The pathways normally (Chitnis and Kuwada, 1991; Eisen, 1991; third pioneer neuron line (MZ465) was a kind gift from J. Urban and Keshishian and Bentley, 1983; Pike et al., 1992). If the neurons G. Technau. The ftzNGAL4 line (Lin et al., 1995) was a kind gift from that pioneer the grasshopper limb bud nerve are ablated, a David Van Vactor. normal axon pathway is formed by follower neurons (Keshishian and Bentley, 1983). It has also been suggested that GAL4 expression in pioneer neurons a different environment (provided by the basal lamina and (1) Line 15J2 expresses GAL4 in both the vMP2 and dMP2 interneu- guidepost neurons and glia) is present at the time of pioneer rons (Fig. 1a,b,g,j). GAL4-driven expression of Tau-β-galactosidase outgrowth and that this, and not cell type, drives axon guidance. is detected at stage 12/1, when the pioneer growth cones first extend However, in cases where a structure is placed at an ectopic site, along the longitudinal tracts. Both vMP2 and dMP2 can be detected as a consequence of transplantation or mutation, pioneer at stage 14 (Fig. 1a,b) but, by stage 17 (Fig. 1j), expression is often neurons can still connect normally to the CNS despite having restricted to the dMP2 neurons. GAL4 expression can occasionally be to navigate in a region devoid of axons and other guidance cues, detected in pCC (on average, in 1 hemisegment per embryo at stage such as guidepost cells (Passani et al., 1991; Tix et al., 1989). 15; by comparison dMP2 is seen in 20 hemisegments and vMP2 in 13 hemisegments), in an unidentified cell located laterally within the In order to reconcile the results from different experimental neuropile and later in scattered cells, varying between segments, at systems, it has been suggested that distinct nerves within an stage 17. organism, and similarly between organisms, may use different (2) Line C544 expresses GAL4 in the MP1 interneurons and occa- mechanisms of axon guidance (Pike et al., 1992). It has also sionally in the midline glia (in 2-3 segments per embryo; Fig. 1c,h,k). been suggested that the focus on the “pioneering problem” Expression is driven rarely in the VUMs and in a few other unidenti- should shift from testing the roles of individual pioneer neurons fied cells. In this line, GAL4 expression in the MP1 neurons is usually to questioning the role played in axon guidance by multiple seen in only a small number of segments in each embryo. Expression elements (Chitnis and Kuwada, 1991; Ho and Goodman, 1982). is first detected at stage 11 in the cell bodies of the MP1s, before the For example, as for the A/P fascicle, pathways may be growth cones have started to extend, and persists throughout embryo- pioneered by several neurons rather than by an individual genesis. pioneer (Bastiani et al., 1986; Ho and Goodman, 1982; Jacobs (3) Line MZ465 expresses GAL4 in the pCC and aCC sibling neurons and in RP2 starting from about stage 12/1 and continuing and Goodman, 1989; Lin et al., 1995; Raper et al., 1984). through embryogenesis (Fig. 1d,f,i,l). Very rarely, expression is also In Drosophila, cell ablation has been technically difficult, detected in the MP2s. GAL4 expression in this line can be variable. limiting exploration of this problem. It is now possible to ablate Scattered cells, varying between segments, can be detected through- cells genetically by means of the GAL4 system (Brand and out the CNS, more frequently in older (stage 16-17) embryos. Perrimon, 1993; Brand et al., 1994; Hidalgo et al., 1995; Lin Expression of GAL4 in two or more pioneer neurons was achieved et al., 1995; Smith et al., 1996). The cell ablation technique by genetically combining these three GAL4 insertions (Fig. 2). The Targeted ablation of pioneer neurons 3255 resulting multiple insertion lines expressed GAL4 with increased reg- anteriorly to meet the homologous axons from the next anterior ularity (for MP1-MP2s see Fig. 2d, for MP2s-pCC see Fig. 2g and for segment (Fig. 2a,b,e; see also Jacobs and Goodman, 1989). MP1-MP2s-pCC see Fig. 2i). In general, patterns are less consistent Similarly, the axons of MP1 and dMP2 fasciculate and extend before stage 13, quite specific and regular between stages 13 and 16, posteriorly until they meet the homologous axons from the next and include more scattered cells at stage 17. posterior segment (Fig. 2e; Jacobs and Goodman, 1989). At UAS-tau-lacZ and UAS-ricin A stage 13, the ascending axons of vMP2 and pCC encounter the Pioneer neurons were visualised by driving expression of the reporter descending axons of dMP2 and MP1 to form the first joint lon- gene, UAS-tau-lacZ (Hidalgo et al., 1995). Targeted ablation was gitudinal pathway (Fig. 2e; Jacobs and Goodman, 1989). At achieved by expression of UAS-ricin A (Hidalgo et al., 1995; John, stage 14, the vMP2/pCC fascicle separates from the Davidson, Moffat, O’Kane, Halder, Gehring, Yoffe and A. H. B., MP1/dMP2 fascicle, to form two pathways that make contact unpublished data). only at the segment boundary (Lin et al., 1994): an outer fascicle composed of the axons of MP1 and dMP2, and an Immunohistochemistry and photography inner one consisting of pCC and vMP2 (Fig. 2c). It is possible Antibody staining reactions were carried out by standard methods that a minor axon from dMP2 also joins this pathway (not (Patel, 1994). Antibodies were diluted in PBS, 0.1% Triton X-100, β shown). and used at the following concentrations: rabbit anti- -galactosidase From this stage onwards, it had been thought that all of the 1:10000 (Cappel); mouse mAb anti-BP102 at 1:5-1:10 (Seeger et al., 1993; Patel, 1994; a gift from N. Patel); mouse mAb 1D4 anti- pioneer axons run in a combined MP1/MP2 pathway along the fasciclin II at 1:8 (Van Vactor et al., 1993; a gift from C. Goodman most medial tract of the three fascicles that express FasII (Lin and K. Broadie). Secondary antibodies were either directly conjugated et al., 1994). By following individual axons throughout to FITC (1:200), Texas Red (1:200) or horse radish peroxidase (HRP, embryogenesis, we have observed that the pathways traced by 1:300) (Jackson Labs), or were biotinylated (1:300) (Vector Labs) and these axons are dynamic and differ from previous reports (for detected after incubation for 1 hour in avidin and biotinylated horse example, see Lin et al., 1994). Between stages 14 and 15, the radish peroxidase (Vectastain Elite ABC kit, Vector Laboratories). axons of the pioneer neurons reassort. First, the two axon HRP was detected using diaminobenzidine (0.3 mg/ml in PBS, 0.1% fascicles move closer together (Fig. 2d). Next, these two tracts, Triton X-100; Sigma) as a substrate. HRP-stained embryos were which were fused at the segment boundary, separate (Fig. 2d). cleared overnight in 50% glycerol in PBS, 0.1% Triton X-100 and Finally, the axons change partners: dMP2 defasciculates from then mounted in 70% glycerol in PBS, 0.1% Triton X-100. All HRP images are of dissected ventral nerve cords. Dissection was carried MP1 (or is displaced by follower neurons) and fasciculates out after antibody staining, using tungsten needles. Transmitted light with pCC (Fig. 2f,g), and vMP2 defasciculates from pCC (or images were generated by DIC microscopy using a Zeiss Axiophot is displaced) and runs in a more ventral plane along the same microscope. For fluorescent detection, primary antibody concentra- longitudinal pathway (Fig. 2f). These patterns are maintained tions were doubled and embryos were mounted in Vectashield (Vector for the rest of embryogenesis. Labs). Fluorescent images were taken from whole-mount embryos on At stage 17, the axons of pCC and dMP2 run dorsally (Fig. a Biorad MRC600 confocal microscope and they are projections of 2g) and vMP2 ventrally (Fig. 2f), along the most medial FasII 20-30 1 µm optical sections through the ventral nerve cord. tract (Fig. 2h); MP1 runs dorsally along the middle tract (Figs 1k, 2d,i). Although only three FasII-expressing axon tracts can be seen in ventral views of each hemisegment, optical sec- RESULTS tioning by confocal microscopy reveals at least five FasII- expressing axon fascicles in the D/V axis of each hemisegment: Axon outgrowth and fasciculation of the pioneer the most medial tract consists of a dorsal and ventral fascicle, neurons as does the middle tract (E. L. Dormand and A. H. B., unpub- The ventral nerve cord of the Drosophila embryo consists of lished data). Our results contrast with previous reports, which two longitudinal axon bundles that run the length of the body assigned all of the pioneer axons, including MP1, to the most to the brain. In each segment, two commissural axon bundles medial FasII-expressing fascicle (which was then called the cross the midline and link the longitudinal connectives “combined MP1/MP2 pathway” (Lin et al., 1994). Therefore, (reviewed by Goodman and Doe, 1992). The longitudinal tracts we have renamed the most medial FasII-positive tract the of Drosophila and grasshopper are pioneered by four cells per “pCC/MP2 pathway”, and the middle tract in each hemiseg- hemisegment: the MP1, dMP2, vMP2 and pCC neurons (Bate ment the “MP1 pathway”. At present, we have not identified and Grunewald, 1981; Bastiani et al., 1986; Grenningloh et al., the neurons that pioneer the outermost FasII-positive fascicle. 1991; Jacobs and Goodman, 1989; Lin et al., 1994). We have characterised lines in which GAL4 is expressed in each of the Targeted ablation of pioneer neurons four pioneer neurons (Fig. 1). By driving expression of a UAS- It was previously reported that the pioneer neurons are not tau-lacZ reporter gene (Hidalgo et al., 1995), we can clearly required for the formation of the longitudinal tracts (Lin et al., visualise the axon morphology of each pioneer neuron through- 1995). Lin et al. (1995) used a line in which GAL4 transcrip- out development. By genetic crosses, we have generated flies tion is directed by the ftz neurogenic enhancer (ftzNGAL4) to that carry two or three GAL4 insertions and have used these to drive expression of diphtheria toxin in clones of cells, so that label unique combinations of pioneer neurons. We can thus a subset of the GAL4-expressing cells are killed (the so-called describe the relative positions of the pioneer axons with respect “Blue Death” method). In these experiments, the follower to each other at each stage of embryogenesis (Fig. 2). neurons still formed relatively normal axon tracts after pioneer The formation of the longitudinal tracts begins with the neurons and their followers were genetically ablated. extension of the pioneer axons at stage 12. At the onset of axon In contrast to these results, we find that the longitudinal outgrowth, the axons of vMP2 and pCC fasciculate and extend tracts are severely disrupted when we use the ftzNGAL4 driver 3256 A. Hidalgo and A. H. Brand to direct expression of UAS-ricin A in all of the GAL4-express- longitudinal pathways in the absence of pioneers. Another ing cells (see Materials and Methods). GAL4 is expressed from dramatic feature is the misrouting of axons across the midline stage 11 onwards in the MP2 neurons, and from stage 12 also (Fig. 3d,g). in pCC, MP1 and many lateral neurons (Fig. 3a,b). Hence the It cannot be concluded that these phenotypes are solely a phenotype should reflect the loss of these neurons. After cell consequence of ablating the pioneer neurons, however, as ablation, the longitudinal axons are completely lost at stage 14 ftzNGAL4 expresses in the glioblast and in a large number of (Fig. 3d; Table 1). These defects persist throughout embryo- follower neurons. For example, in wild-type embryos FasII is genesis, as detected with antibodies against both FasII (Fig. 3g) expressed in three longitudinal fascicles at stage 17 (Fig. 3f). and BP102 (Fig. 3h; compare to wild-type embryo in Fig. 7b), In the ftzNGAL4 line, GAL4 is expressed primarily in neurons suggesting that the follower neurons cannot establish the that extend their axons along the two inner tracts at stage 17.

Fig. 1. Expression of GAL4 in a dMP2,vMP2 b dMP2,vMP2 c MP1 dfpCC pCC each pioneer neuron of the longitudinal connectives. (a,b,g,j) GAL4-driven expression of Tau-β-galactosidase in dMP2 and vMP2 neurons. (a) Late stage 13 embryo. The axons of the two pioneers have just extended along contiguous segments, and the two vMP2 MP1 aCC pathways become prominent. pCC (b) The cell bodies of vMP2 and dMP2 lie in a ventral plane with vMP2 regard to the axons. dMP2 RP2 (g) Identification of the MP2s by double labelling with FasII (red) and anti-β-galactosidase dMP2 antibodies (green). Colocalisation appears as yellow, in the cell bodies of vMP2 and dMP2. FasII only stains the cell surface, so colocalisation does not cover the whole cell. (j) At stage 17, the axons of dMP2 and vMP2 run along the first, most medial, FasII g h i fascicle (arrow, colocalisation of axons in yellow). Staining as in g. (c,h,k) Expression of Tau-β- galactosidase in the MP1s. vMP2 MP1 (c) Stage 12 embryo, as the growth cones extend (arrow). (h) Identification of MP1 by colocalisation (in yellow) of FasII dMP2 (red) and anti-β-galactosidase (green) antibodies. Note also the pCC characteristic almond shape of these cells and their midline location. (k) At stage 17, the axons of the MP1s run along the vMP2 second FasII fascicle. Staining as j kMP1 l in h, colocalisation of axons shows up in yellow (arrow); arrowhead, cell bodies of MP1s. dMP2 (d,f,i,l) Expression of Tau-β- galactosidase in pCC, aCC and RP2 neurons. (d) Focal plane of pCC axons of pCC (arrow), of a stage 16 embryo. (f) Focal plane of the cell bodies. (i) Identification of pCC (arrowhead) by 3 colocalisation (in yellow) of FasII 2 3 (red) and anti-β-galactosidase 1 1 2 1 2 3 (green) antibodies. Note the characteristic shape of this cell and the ascending axon (arrow). (l) At stage 17, the axon of pCC (arrow) runs along the first, most medial, FasII fascicle, like the MP2s. In this and all remaining figures, all images are of dissected ventral nerve cords, with anterior up. Targeted ablation of pioneer neurons 3257

a c a ftzNGAL4/ b c +/+ d ftzNGAL4/ b d UAS-tau-lacZ UAS-Ricin

dMP2 MPs vMP2 MPs

vMP2

M 2 pCC dMP2 P P MP1 M 1 e d pCC dMP2 pCC vMP2

dMP2 MP2s MP2s+MP1 β−gal vMP2 β-gal fasII fasII f g h i e f +/+ g ftzNGAL4/ h UAS-Ricin 3 2 1 d 1

M 2

2 p

vMP2 C

+ d

M M

P P

1 2

β-gal fasII fasII vMP2+dMP2 pCC+dMP2 vMP2+dMP2 MPs+pCC BP102 Fig. 2. Profile of axonal patterns. (a,b) Expression of Tau-β- Fig. 3. Targeted ablation with ftzNGAL4. (a,b,e) Expression driven by galactosidase in dMP2, vMP2 and pCC, detected with anti-β- ftzNGAL4 in the ventral nerve cord of the CNS as detected by crossing galactosidase antibodies. The growth cones of vMP2 and pCC this line to flies carrying a UAS-tau-lacZ gene and visualising with anti- β β fasciculate and extend together at stage 12. (a) Focal plane of -galactosidase antibodies. (a) ftzNGAL4 drives expression of Tau- - galactosidase in the pioneer neurons from the onset of their outgrowth, axons showing the axons of vMP2/pCC (arrow) extending but also in many other cells even from this early stage (arrow). Stage 13 anteriorly, that of dMP2 (white arrow) extending posteriorly and embryo, in which the axons of the pioneer neurons from adjacent the cell body of pCC. (b) Focal plane of cell bodies. segments have just met. At this stage, this line also drives expression (c,e) Expression of Tau-β-galactosidase in dMP2 and vMP2. At transiently in the glioblast (data not shown). (b) At stage 14, ftzNGAL4 stage 13 (e), the axons of dMP2 (white arrow; white arrowhead clearly drives expression of Tau-β-galactosidase in all the pioneer showing cell body) and vMP2 (arrow; cell body, arrowhead) reach neurons of the longitudinal connectives (arrowhead, cluster of MP1, contiguous segments and lie very close to each other, making light dMP2, vMP2; white arrowhead, pCC), but still in many other cells contact. The first joint longitudinal pathway is thus formed. (c) At within the ventral nerve cord (arrow). (c) Staining with FasII antibody stage 14, two axon pathways separate: the pCC/vMP2 (arrow) and in a wild-type embryo of the same age, in which only the pioneer the MP1/dMP2 (white arrow) pathway. (Arrowhead, cell body of neurons are visualised (arrowhead, cluster of MP1, dMP2, vMP2 and vMP2; white arrowhead, cell body of dMP2.) (d) Expression of white arrowhead for pCC; compare this with b). (d) FasII expression in a stage 14 embryo in which all the pioneer neurons of the longitudinal GAL4 in the MP1s and MP2s. At stage 15, the two pathways, connectives have been ablated with ftzNGAL4. The phenotype is which were previously joined at the segment boundary, are fully characterised by a dramatic absence of axons along the longitudinal separated but closer to each other. The MP1 axon always lies pathways (arrows) and misrouting of axons across the midline (white furthest from the midline. (f) Expression of GAL4 in vMP2 and arrows). (e) Expression of Tau-β-galactosidase driven by ftzNGAL4 in dMP2 at stage 15. From this stage on, the axon of vMP2 runs a stage 17 embryo. Two axon tracts are clearly stained (arrows), and along the same longitudinal pathway as the axons of pCC and additionally some cells along a third outermost tract (arrowhead). dMP2, but in a ventral plane (arrowhead, vMP2 cell body). (f) Characteristic expression of FasII in a stage 17 wild-type embryo, (g) Expression of GAL4 in pCC and the MP2s, predominantly in revealing three fascicles in each hemisegment (arrows). Compare to the dMP2s, at stage 15. With this line, GAL4 is frequently not two fascicles visualised with ftzNGAL4 in e. (g) Expression of FasII in expressed in vMP2, whereas it is in the dMP2. The axons of pCC a stage 17 embryo following ablation with ftzNGAL4. The phenotype and dMP2 run together in the same dorsal plane (arrow). is characterised by the frequent complete absence of longitudinal axons (arrow), but one axon or fascicle can also remain (arrowhead). (h) Expression of GAL4 with a line that drives expression in dMP2 ftzNGAL4 is not expressed in all the fascicles detected by FasII. There and also in vMP2 at stage 17. A thin axon was observed running is also a very dramatic presence of axons crossing the midline along the along the MP1 pathway (arrows). We do not know the origin of commissures (white arrow). (h) Expression of BP102 in a stage 16 this minor axon. (i) Expression of GAL4 in dMP2, vMP2, MP1 embryo of genotype UAS-ricin A;ftzNGAL4. For a wild-type embryo, and pCC, showing two axon fascicles: the inner pCC/dMP2 see Fig. 7b. BP102 stains most CNS axons. After ablation of the GAL4- pathway and the outer MP1 pathway. expressing cells in this line, the longitudinal axons are missing (arrows). 3258 A. Hidalgo and A. H. Brand

Table 1. Penetrance and expressivity of ablation phenotypes. Penetrance Expressivity Line Stage Total embryos Mutant % Total hemisegs. Mutant % MZ465 14 14 14 100 256 164 64 17 14 12 85 296 68 22

C544 14 18 15 83 342 157 45 17 31 21 67 644 108 16

15J2 14 20 20 100 382 252 65 17 26 16 61 544 51 9.3

MZ465, 15J2 14 14 14 100 248 214 86 17 15 14 93 308 126 40

C544; 15J2 14 21 21 100 327 255 77 17 37 30 81 758 137 18

C544; MZ465,15J2 14 10 10 100 202 192 95 17 11 11 100 222 184 82

FtzNGAL4 14 26 26 100 452 393 86 17 19 19 100 384 373 97

Penetrance (frequency of phenotype): number of embryos in the total population exhibiting an abnormal phenotype after staining for Fasciclin II. Expressivity (severity of the phenotype): number of hemisegments that exhibit an abnormal phenotype in each mutant embryo.

Nonetheless, after ablation of these cells, the most lateral of Two pioneer neurons are required for the the FasII fascicles is missing in some hemisegments (Fig. establishment of each longitudinal pathway 3g,h). Defects in the outer fascicle may be due to the loss of Ablation of two or more pioneer neurons at a time was carried glial cells and/or follower neurons. Therefore, in order to out by genetic combination of two or three GAL4 insertions dissect the role of the pioneer neurons, it is necessary to ablate (Fig. 6). The loss of two pioneer neurons that extend along the only the pioneers, not the follower neurons or glial cells. same pathway markedly increases the severity of axonal defects, suggesting that the interaction between two pioneer Axon pathways can form in the absence of neurons is required for the establishment of each longitudinal individual pioneer neurons fascicle. We analysed the consequence of ablating each pioneer neuron After ablation of MP1 and both vMP2 and dMP2 , the alone, or in combination with the other pioneer neurons. Toxin MP1/dMP2 tract is lost at stage 14, whereas the pCC/vMP2 ablation is rapid, efficient and cell autonomous: the cells that fascicle can still form (Fig. 6b). Similarly, after ablation of pCC are targeted for ablation are killed, but adjacent cells continue and both dMP2 and vMP2, the pCC/vMP2 fascicle is lost to express the appropriate markers and exhibit wild-type mor- whereas the MP1/dMP2 tract forms (Fig. 6c). pCC or MP1 can phology (Hidalgo et al., 1995). For example, when the MP1 therefore navigate in the absence of the other pioneer neurons. neurons are killed, the neighbouring MP2 neurons can still be By stage 17, the follower neurons can form the three axon detected (Fig. 4b, stage 14). Similarly, ablation of dMP2 and tracts, implying that the following axons can still ﬁnd their way vMP2 leaves the MP1s intact, and killing pCC and aCC does when three of the four pioneer neurons have been killed, but not compromise the survival of the U neurons (data not shown). misrouting of axons and the loss of axon tracts become more Ablation of each pioneer neuron individually affects axon frequent (Fig. 6e,f). outgrowth in early embryogenesis (Fig. 5b-d), but these defects are generally masked as development proceeds, suggesting that Pioneer neurons function as a group in the follower neurons can grow relatively normally in the absence formation of longitudinal tracts of any single pioneer neuron (Fig. 5f-h). When defects are seen Ablation of MP1, pCC, dMP2 and vMP2 eliminates all FasII- later in development, at stage 17, they are speciﬁc to the positive axons at stage 14 (Fig. 7a), and dramatic phenotypes pathway along which the ablated pioneer would extend. First, are still seen at stage 17 (Fig. 7c-h). These phenotypes are killing the MP1 neuron can disrupt the MP1 pathway, resulting never observed after ablation of single pioneer neurons, or in defasciculation, or axon misrouting within the longitudinal pairs of pioneers. The penetrance of this phenotype is complete tract (individual axons, or the entire fascicle, can deviate from and the expressivity is high (see Table 1). The elimination of the correct pathway and fasciculate with adjacent axons; Fig. all pioneer neurons causes thinning of the longitudinal tract and 5f, Table 1). Second, ablation of pCC can eliminate the pCC breaks in the axon bundles (Fig. 7c-h), fusion of fascicles (Fig. pathway at stage 14 and lead to axon misrouting across the 7g), misrouting of axons across the midline (Fig. 7c-h) and the midline at stage 17 (Fig. 5c,g). Third, ablation of both dMP2 disorganisation and broadening of commissures (Fig. 7c,d). and vMP2 can affect both the MP1 and dMP2/pCC pathways The loss of discrete longitudinal axon bundles and the fusion at stage 14, and causes occasional misrouting across the of axons into a single fascicle infers that pioneer neurons are midline at stage 17 (Fig. 5d,h). required to keep follower neurons growing along distinct tracts. Targeted ablation of pioneer neurons 3259

Occasionally, individual fascicles can form (Fig. 7f), suggest- two pathways: an outer one formed by the axons of MP1 and ing that the follower axons are also guided by cues extrinsic to dMP2, and an inner one formed by pCC and vMP2. Subse- the pioneer neurons. quently, the fasciculation patterns change: dMP2 defascicu- While we cannot rule out that the loss of unidentified lates, or is displaced, from MP1 and fasciculates with pCC and neurons might contribute to these phenotypes, when we kill the vMP2 defasciculates, or is displaced, from pCC and runs along pioneer neurons individually, or in combinations of two or the same longitudinal pathway but in a more ventral plane. three, the phenotypes are specific to the fascicles pioneered by These patterns are maintained for the rest of embryogenesis. those neurons. The combined line also expresses GAL4 in the Hence, in relation to the FasII fascicles at stage 17, pCC and midline glia in some segments, so axon misrouting across the dMP2 run dorsally and vMP2 ventrally along the first FasII midline might result from the loss of these glial cells. However, tract, which we therefore rename the ‘pCC/MP2 pathway’. ablation of the GAL4-expressing cells in line ftzNGAL4, MP1 runs along the second tract, which we therefore rename which does not express in the midline glia, also causes mis- the ‘MP1 pathway’. We do not know which neurons constitute routing across the midline (Fig. 3d,g). Therefore, the follower the third FasII fascicle. axons may be prevented from crossing the midline because of their adhesion to the pioneer neurons. Pioneer neurons are required as a functional group We have analysed the effect on the formation of the longitudinal tracts of ablating each pioneer neuron alone, or in combi- DISCUSSION nations. At the end of embryogenesis, FasII is expressed in three tracts (Grenningloh et al., 1991; Lin et al., 1994). We For some time, the question of whether pioneer neurons are have shown that the pioneer neurons described to date con- essential for axon guidance has remained obscure. This is due tribute to only two of these tracts. The pioneers of the third to the fact that different experiments yielded contradictory fascicle are as yet unknown. As FasII is not expressed on cell results. For instance, in grasshopper ablation of all P pioneer bodies at stage 17, we cannot identify the origin of the FasII neurons prevents extension of follower neurons (Raper et al., axons at this stage. 1984), whereas ablation of pioneers of the MP1 and MP2 Previously when the pioneer neurons had been ablated in pathways in Drosophila was reported to have no effect on Drosophila (Lin et al., 1995), the pattern of FasII-expressing followers (Lin et al., 1995). We argue that some of these dis- axons at stage 17 was found to be relatively normal. It was crepancies were due to the experimental approach (Lin et al., concluded that follower axons can largely recover from the loss 1995). Using a different cell ablation method, we show that of pioneer neurons. However, the ablation method used in this pioneer neurons are necessary for the formation of the longi- work was mosaic: not all the pioneer neurons from every tudinal pathways. We show that the pioneer neurons direct the segment of each embryo were ablated. To prevent transient outgrowth of follower neurons along the ipsilateral pathways, expression of diphtheria toxin, Lin et al. (1995) inserted the restricting axon extension to specific fascicles and preventing lacZ gene and transcriptional terminator (flanked by FRT growth across the midline. Our experiments suggest that this sequences) between the GAL4 UAS and the toxin coding requirement is not absolute and that pioneer neuron-indepen- sequence. β-galactosidase, but not diphtheria toxin, is dent cues may be provided, for example, by glial cells. expressed in a GAL4-dependent fashion. To induce toxin expression and cell ablation, mRNA encoding FLP-recombi- Fasciculation and defasciculation of pioneer axons nase is injected into individual embryos. Recombination We describe the pathways followed by individual pioneer catalysed by FLP at FRT sites excises lacZ, and allows neurons during the course of embryogenesis. Previous descrip- expression of diphtheria toxin in clones of GAL4-expressing tions relied on electron micrographs and FasII expression cells. Those cells that do not receive FLP mRNA or protein, or patterns (Grenningloh et al., 1991; Jacobs and Goodman, 1989; in which the recombinase does not work efficiently, survive Lin et al., 1994). In both cases the descriptions were incom- and continue to express β-galactosidase. plete. First, the EM studies covered only a limited time window Interestingly, it had previously been shown in grasshoppers (Jacobs and Goodman, 1989). Second, FasII is not expressed that the loss of pioneer neurons in one segment can be rectified on the axons of all pioneer neurons before stage 14. Third, by extending pioneer axons from adjacent segments. Bastiani using FasII as a marker, it is not possible to trace the axons et al. (1986) carried out elegant experiments in which they back to individual cell bodies between stages 15 and 17 ablated the MP1 and MP2 neurons from one or more adjacent (Grenningloh et al., 1991; Lin et al., 1994). Using the GAL4 segments. Only when pioneer neurons from at least two system to drive expression of the microtubule-associated adjacent segments are ablated are the follower axons (in this reporter protein, Tau-β-galactosidase (Callahan and Thomas, case pCC) affected. This implies that in Drosophila, as in 1994; Hidalgo et al., 1995), we have been able to trace the grasshopper, the pioneer axons from neighbouring segments development of each pioneer neuron, both individually and in can extend along segments in which the pioneer neurons have relation to the other pioneers, throughout embryogenesis. been ablated to eventually establish a normal pathway. This At the onset of growth cone extension, the axons of vMP2 explanation is consistent with the fact that Lin et al. (1995) and pCC fasciculate together and extend anteriorly to meet the observe a repair of axon defects as embryogenesis proceeds. homologous axons from the next anterior segment (Jacobs and We have ablated the pioneer neurons in every segment using Goodman, 1989). The posteriorly extending axons of MP1 and a method that is efficient, non-invasive and cell autonomous dMP2 meet the homologous axons from the next posterior (Hidalgo et al., 1995). We used the same GAL4 line as Lin et segment. At stage 13, all axons contact one another along a al (1995; ftzNGAL4) to ablate all of the pioneer neurons, single longitudinal axon tract. At stage 14, these axons trace causing a dramatic loss of longitudinal axons throughout 3260 A. Hidalgo and A. H. Brand embryogenesis. The high penetrance and expressivity of the function as a group to pioneer the A/P pathway in grasshopper phenotypes at later stages implies that the nerve cord does not (Raper et al., 1984). All three P neurons must be ablated to recover from the loss of the pioneer neurons. As ftzNGAL4 is disrupt the growth of the follower neurons. expressed not only in the pioneer neurons, but also in many lateral neurons, and transiently in the glioblast, the ablation Pioneer neurons restrict follower axons to the phenotype may result from ablating both neurons and glial ipsilateral fascicles cells. Consequently, we used more specific GAL4 lines to Ablation of all four pioneer neurons leads to axon defascicu- ablate the pioneer neurons individually and in combinations. lation, the loss or thinning of axon fascicles, breaks in the We have shown, first, that ablation of each pioneer neuron longitudinal axon bundles, fusion of axons into a single individually can cause defects in early embryogenesis, but that fascicle, and misrouting of axons across the midline. The these defects are generally masked by the time that the nerve cord is formed. Despite early guidance defects, follower a b c d neurons can grow normally in the absence of any one pioneer neuron. Second, two pioneers are required to pioneer each longitudinal fascicle. After ablation of pairs of pioneer neurons, an axon fascicle is missing at stage 14 . However, the longitudinal tracts still form relatively normally by stage 17. Although follower neurons grow more reliably if all of the pioneer axons are present, they can still usually find their correct pathway in the absence of one or two pioneers. Third, the loss of all pioneer pCC neurons inhibits the normal extension of follower neurons. The MP1 phenotype in stage 17 embryos is more severe than when only two or three pioneers are ablated. Although we cannot rule out a cumulative effect of deleting the earliest extending neurons, these results are consistent with the role of the P neurons, which +/+ No MP1 No pCC No dMP2+vMP2 a e f g h vMP2

dMP2 MP1

+/+ No MP1 No pCC No dMP2+vMP2 Fig. 5. Effects of ablating one pioneer neuron as visualised with the marker FasII. Upper row, stage 14 embryos; lower row, stage 17 embryos. (a,e) FasII expression in wild-type embryos. Note the cell Fig. 4. Cell autonomy of targeted ablation. (a) A cluster of cells bodies of the pioneer neurons and the two axon pathways that fuse at comprising the MP1s (arrow), dMP2s (feathered arrow) and vMP2s the segment boundary at stage 14 (arrow, pCC/dMP2 pathway; white (white arrow) in one segment of a normal embryo, as visualised with arrow, MP1/dMP2 pathway). At stage 17, FasII reveals three axon FasII antibodies. These cell bodies lie in a ventral position, therefore, tracts (arrows). (b) Ablation of the MP1s can cause defects at stage their axonal projections are out of focus in this image. FasII is a cell 14. This embryo has lost the MP1 neurons, but the axon pathways surface protein involved in cell adhesion and it normally stains most are established normally (arrows). (f) Ablation of MP1s can cause strongly areas of cell-cell contact. The MP1s are characterised by misrouting of axons from the second FasII fascicle (arrows), along their almond shape and the intense stripe of FasII expression at the which the MP1 axon normally runs. (c) Ablation of pCC can cause interface between the two MP1s at the midline (arrow). loss of the pCC/vMP2 pathway at stage 14 (black arrow), whereas (b) Following the ablation of the MP1s with a GAL4 line driving the MP1/dMP2 pathway remains (white arrow). In this line, aCC and expression of the toxin ricin A exclusively in these cells, the MP1s RP2 are also ablated. (g) Later on, loss of pCC can cause misrouting are specifically lost as detected with FasII (arrow). Note, in of axons that normally run along the first FasII fascicle to cross the particular, the absence of FasII staining in the stripe that normally midline (arrow). (d) Ablation of dMP2 and vMP2, which can cause lies along the midline (arrow). The absence of MP1s implies that cell damage to the MP1/dMP2 (white arrow) and the pCC/vMP2 (black ablation is efficient and specific. The neighbouring MP2s remain arrow) pathways. (h) Later on, loss of the MP2s can cause misrouting intact (white arrow, vMP2; feathered arrow, dMP2). This means that of axons across the midline (arrow). Despite these specific defects, in cell ablation is restricted to the cells in which GAL4 is expressed. all these cases (f,g,h) at stage 17 most embryos look normal in most Both images are from stage 14 embryos. of their segments. Targeted ablation of pioneer neurons 3261

a b c

pCC MP1 MP1

pCC

+/+ No MP1,MP2s No pCC,MP2s d e f

Fig. 7. Ablation of all pioneer neurons. (a) At stage 14, axon pathways are completely absent (arrows) as detected with FasII. (b- d) Phenotypes detected with BP102 antibodies in wild-type (b) and after ablation of all pioneer neurons (c,d) in stage 16 embryos. (b) Characteristic thick BP102-positive longitudinal bundles (arrows) +/+ No MP1,MP2s No pCC,MP2s and commissures (arrowhead). (c,d) Following ablation of pCC, Fig. 6. Ablation of two or more pioneer neurons at a time. Upper MP1, dMP2 and vMP2, the longitudinal bundles are missing or very row, stage 14 embryos; lower row, stage 17 embryos. All phenotypes thin (arrows) and the commissures are thicker than normal, broken or are detected with FasII antibodies. (a,d) Wild-type embryos. disorganised (arrowheads). (e-h) Phenotypes detected with FasII (a) Black arrow, MP1/dMP2 pathway; white arrow, pCC/vMP2 expression following ablation of pCC, MP1, dMP2 and vMP2 in pathway. (d) Arrows, three FasII fascicles. (b,e) Ablation of MP1, stage 17 embryos. These examples illustrate the variability of the dMP2 and vMP2. At stage 14, the MP1 pathway is lost (black phenotype and the features commonly found: breaks or loss of arrows), whereas the pCC/vMP2 pathway remains (white arrow) longitudinal tracts (arrows), one axon fascicle remaining (e, g and h), although it can be damaged. Later on (e), this causes misrouting of fusion of remaining axons into one fuzzy, irregular tract (white axons affecting the second FasII fascicle (arrow) or directing them arrows in g) and misrouting of axons across the midline across the midline (arrowhead). (c,f) Ablation of pCC, dMP2 and (arrowheads). vMP2. (c) At stage 14, the pCC/vMP2 pathway is lost (white arrow), whereas the MP1/dMP2 pathway remains (black arrow). (f) Later on, this causes breaks (arrow) and misrouting across the midline (arrowhead) in the most medial, FasII fascicle. Defects are also Axon defasciculation, the fusion of axons into one fascicle noticeable in the second FasII fascicle. and the misrouting of axons into different fascicles suggest that the pioneer neurons direct the fasciculation of axons along the distinct fasciclin II-positive tracts. This is consistent with experiments in grasshopper showing that the follower G neuron breaks and loss of longitudinal tracts reveal the requirement for fasciculates specifically with the P, not the A, neurons (Raper pioneer neurons in the guidance of followers. However, the et al., 1984). Hence, it supports the notion of selective fascic- thinning of the longitudinal bundle implies that there are still ulation, by which follower neurons read specific cues along enough cues to allow the formation of a longitudinal pathway. different pioneer axons. These cues may reside on as yet unidentified neurons, on the Finally, the misrouting of axons across the midline implies extracellular matrix or on the longitudinal glia, which overlie that one role of pioneer neurons is to keep follower axons the longitudinal axon tracts. Hence, the requirement for pioneer growing along the longitudinal pathways. In this context, it is neurons in axon guidance is not absolute. intriguing that the longitudinal pioneer neurons are among the 3262 A. Hidalgo and A. H. Brand few neurons that extend their axons exclusively along ipsilat- du Lac, S., Bastiani, M. J. and Goodman, C. S. (1986). Guidance of neuronal eral pathways, whereas most neurons first project contralater- growth cones in the grasshopper embryo. II. Recognition of a specific axonal ally (Bossing et al., 1996). Diffusible signals and local cell-cell pathway by the aCC neuron. J. Neurosci. 6, 3532-3541. Eisen, J. S. (1991). Motoneuronal development in the embryonic zebrafish. interactions, for example with the longitudinal glia, may Development Supplement 2, 141-147. determine the projection pathways of the pioneer neurons. It Eisen, J. S., Myers, P. Z. and Westerfield, M. (1986). Pathway selection by has been suggested that attractive cues emanating from the growth cones of identified motoneurons in live zebra fish embryos. Nature midline cells may guide commissural axons to cross the ventral 320, 269-271. midline, and that repulsive cues may prevent longitudinal Ghosh, A., Antonini, A., McConnell, S. K. and Shatz, C. J. (1990). Requirement for subplate neurons in the formation of thalamocortical axons from crossing (Keynes and Cook, 1995; Seeger et al., connections. Nature 347, 179-181. 1993). The fact that follower neurons cross the midline in the Gong, Q. and Shipley, M. T. (1995). Evidence that pioneer olfactory axons absence of pioneers implies that these neurons are not repelled regulate telencephalon cell cycle kinetics to induce the formation of the by the midline. It is possible that the pioneer axons alone olfactory bulb. Neuron 14, 91-101. Goodman, C. S. and Doe, C. Q. (eds) (1992). Embryonic development of the respond to the repulsive midline signals and that follower Drosophila central nervous system. In Development of Drosophila neurons are guided by fasciculation with pioneer axons and by melanogaster, pp. 1131-1206. Cold Spring Harbour Laboratory Press. other local cues. Alternatively, the repulsive signals may no Grenningloh, G., Rehm, E. J. and Goodman, C. S. (1991). Genetic analysis longer be present when the follower axons grow out. of growth cone guidance in Drosophila: fasciclin II functions as a neuronal In conclusion, the pioneer neurons function together as a recognition molecule. Cell 67, 45-57. Hidalgo, A., Urban, J. and Brand, A. H. (1995). Targeted ablation of glia group of cells that are necessary for the formation of the lon- disrupts axon tract formation in the Drosophila CNS. Development 121, gitudinal axonal pathways. We have shown that the pioneer 3703-3712. neurons organise followers into distinct axon fascicles. A Ho, R. K. and Goodman, C. S. (1982). Peripheral pathways are pioneered by highly disorganised axon pathway can form in the absence of an array of central and peripheral neurones in grasshopper embryos. Nature all pioneer neurons, suggesting that guidance cues either on 297, 404-406. Jacobs, J. R. and Goodman, C. S. (1989). Embryonic development of axon glial cells, neurons or the extracellular matrix, can be read by pathways in the Drosophila CNS. II. Behavior of pioneer growth cones. J. the follower axons. This favours a view whereby a combina- Neurosci. 9, 2412-2422. tion of signals (long and short range) and cell-cell contacts Keshishian, H. and Bentley, D. (1983). Embryogenesis of peripheral nerve (neuron-neuron and neuron-glia) provides guidance cues to pathways in grasshopper legs. III. Development without pioneer neurons. both pioneer and follower neurons. Dev. Biol. 96, 116-124. Keynes, R. and Cook, G. M. W. (1995). Axon guidance molecules. Cell 83, 161-169. We would like to thank Jo Urban, Gerd Technau, Kim Kaiser, David Klose, M. and Bentley, D. (1989). Transient pioneer neurons are essential for Van Vactor, Norbert Perrimon, Nipam Patel and Corey Goodman for formation of an embryonic peripheral nerve. Science 245, 982-984. fly stocks and antibodies. Many thanks to Michael Bate for comments Lin, D. M., Auld, V. J. and Goodman, C. S. (1995). Targeted neuronal cell on the manuscript and for helpful discussions throughout the course ablation in the Drosophila embryo: pathfinding by follower growth cones in of this work. A. H. was supported by an EC Human Capital and the absence of pioneers. Neuron 14, 707-715. Mobility Postdoctoral Fellowship. This work was funded by the Lin, D. M., Fetter, R. D., Kopczynski, C., Grenningloh, G. and Goodman, Wellcome Trust. C. S. (1994). Genetic analysis of fasciclin II in Drosophila: defasciculation, refasciculation and altered fasciculation. Neuron 13, 1055-1069. Passani, M. B., Peinado, A., Engelman, H., Baptista, C. A. and Macagno, E. R. (1991). Normally unused positional cues guide ectopic afferents in the REFERENCES leech CNS. J. Neurosci. 11, 3868-3876. Patel, N. H. (1994). Imaging neuronal subsets and other cell types in whole Bastiani, M. J., du Lac, S. and Goodman, C. S. (1986). Guidance of neuronal mount Drosophila embryos and larvae using antibody probes. In Drosophila growth cones in the grasshopper embryo. I. Recognition of a specific axonal melanogaster: Practical Uses in Cell and Molecular Biology (ed. L. S. B. pathways by the pCC neuron. J. Neurosci. 6, 3518-3531. Goldstein and E. A. Fyrberg), pp. 445-487. San Deigo: Academic Press. Bate, C. M. (1976). Pioneer neurones in an insect embryo. Nature 260, 54-55. Pike, S. H., Melancon, E. F. and Eisen, J. S. (1992). Pathfinding by zebrafish Bate, C. M. and Grunewald, E. B. (1981). Embryogenesis of an insect motoneurons in the absence of normal pioneer axons. Development 114, 825- nervous system II: A second class of neuron precursor cells and the origin of 831. the intersegmental connectives. J. Embryol. Exp. Morph. 61, 317-330. Raper, J. A., Bastiani, M. J. and Goodman, C. S. (1984). Pathfinding by Bossing, T., Udolph, G., Doe, C. Q. and Technau, G. M. (1996) The neuronal growth cones in grasshopper embryos. IV. The effects of ablating Embryonic central-nervous-system lineages of Drosophila melanogaster. 1. the A and P axons upon the behavior of the G growth cone. The J. Neurosci. Neuroblast lineages derived from the ventral half of the neuroectoderm. Dev. 4, 2329-2345. Biol. 179, 41-64. Seeger, M., Tear, G., Ferres-Marco, D. and Goodman, C. S. (1993). Brand, A. H., Manoukian, A. S. and Perrimon, N. (1994). Ectopic expression Mutations affecting growth cone guidance in Drosophila: genes necessary in Drosophila. In Drosophila melanogaster: Practical Uses in Cell and for guidance toward and away from the midline. Neuron 10, 409-426. Molecular Biology (ed. L. S. B. Goldstein and E. A. Fyrberg), pp. 635-654. Smith, H. K., Roberts, I. J. H., Allen, M. J., Connolly, J. B., Moffat, K. G. and San Diego: Academic Press. O’Kane, C. J. (1996). Inducible ternary control of transgene expression and Brand, A. H. and Perrimon, N. (1993). Targeted expression as a means of cell ablation in Drosophila. Development, Genes and Evolution 206, 14-24. altering cell fates and generating dominant phenotypes. Development 118, Tix, S., Bate, M. and Technau, G. M. (1989). Pre-existing neuronal pathways 401-415. in the leg imaginal disks of Drosophila. Development 107, 855-862. Callahan, C. A. and Thomas, J. B. (1994). Tau-β-galactosidase, an axon Van Vactor, D., Sink, H., Fambrough, D., Tsoo, R. and Goodman, C. S. targeted fusion protein. Proc. Natl. Acad. Sci. USA 91, 5972-5976. (1993). Genes that control neuromuscular specificity in Drosophila. Cell 73, Chitnis, A. B. and Kuwada, J. Y. (1991). Elimination of a brain tract increases 1137-1153. errors in pathfinding by follower growth cones in the zebrafish embryo. Neuron 7, 277-285. (Accepted 1 July 1997)