Growth Cone Guidance in the Zebrafish Central Nervous System

Growth cone guidance in the zebrafish central nervous system

John Y. Kuwada

University of Michigan, Ann Arbor, Michigan, USA

The accessibility and simplicity of the zebrafish embryo have allowed researchers to make a detailed characterization of pathfinding by identifiable growth cones. The growth cones follow precise cell-specific pathways to their targets. Analyses of pathfinding in mutant and experimentally manipulated wild type embryos have shown that growth cones accomplish this by interacting with specific cellular cues in their environment, many of which are likely to be redundant. ,I

Current Opinion in Neurobiology 1992, 2:31-35

Introduction and do not appear to encounter any organized set of channels or other cellular features that could potentially The zebrafish embryo is quickly being established as an guide them [6,7]. This opens up the possibility that spa- excellent vertebrate embryo for studying the processes of tially restricted subsets of the endfeet of neuroepithelial development with a combination of genetic, molecular, cells may express molecules which promote or direct and cellular techniques. One problem that is being fruit- axonal outgrowth di!Terentially. In fact, the endfeet of fully addressed is how growth cones navigate through the neuroepithelial cells may serve as part of an ensemble embryos to lind their targets. The clarity and simplicity of of redundant pathfinding cues for some follower growth the early embryo, combined with the availability of muta- cones in the zebrafish brain (see below). tions and the ability to ablate and transplant specific cells, Growth cones that are projected later in development have been particularly useful in elucidating the mecha- also follow stereotyped cell-specific routes within the nisms that direct growth cones along their pathways. This early scaffold of axons in the zebrafish embryo [4*]. review focuses on the recent analyses of pathfinding by The stereotyped pattern of axonal outgrowth seen in ze- growth cones in the central nervous system of early zebrafish embryos is similar to that found in the embryos bralish embryos, but also includes recent, relevant studies of a wide variety of animals including insects [8] and in other vertebrate and invertebrate embryos. Aspects of mammals [9’]. pathfinding by zebra&h motor neurons are discussed in another review (M Westerfield, this issue, pp 28-30).

Extrinsic cues guide growth cones

Pathfinding is precise and cell-specific Spinal cord The detailed description of pathlinding by identified The central nervous system of the early zebrafish embryo growth cones in the zebrafish embryo has indicated sites contains a simple scaffold of axonal tracts that are pio- where growth cones may potentially select their correct neered by the growth cones of speciiic classes of neu- pathway from a number of other pathways, and has rons in the spinal cord [1,2,3*] and small clusters of helped to determine potential sources of guidance infor- neurons in the brain [+,9,6]. These early growth cones mation. In the spinal cord, one such site is the floor plate. follow precise and cell-specific pathways to establish this The floor plate has been identified as a site where growth scaffold of tracts. In some cases, for instance for some cones make very precise turns in the rat spinal cord [lo*] classes of spinal intemeurons, pioneer growth cones in- as well as in other animals [11,12]. This site has also teract with a specific set of cells that act as intermediate been identified as the area where the axons of rat com- targets (see below). In other cases, such as the epiphysial missural neurons switch from expressing one axonal ad- neurons in the zebralish brain and Rohon-Beard neurons hesion molecule to another [ 131. In the zebrafish spinal in the medaka Jish spinal cord, pioneer growth cones cord, the growth cones of two classes of intemeurons extend along the endfeet of neuroepithelial cells found (CoPA and VeLD) exhibit cell-specific turns near the ven- on the superficial surface of the central nervous system, tral midline of the cord in the floor plate region (Fig. la>

Abbreviations Ml&-medial longitudinal fasciculus; nucPC-nucleus of the posterior commissure; PC--posterior commissure; TPOC-tract of the postoptic commissure.

@ Current Biology Ltd ISSN 09594388 31 32 Development

[3*]. Both the CoPA and VeLD neurons initially project Vem growth cones were similar in both cyc-2 embryos growth cones toward the ventral midline, but once near and embryos after laser ablation of the midline floor plate the midline their growth cones diverge. The VeLD growth cells. For example, CoPA growth cones followed an ip- cone turns posteriorly without crossing the midline to silateral ascending pathway, a ventral ascending pathway, pioneer a ventrolateral longitudinal tract, while the CoPA and a descending pathway, in both cyc-1 and laser-ab- growth cone crosses the midline and turns anteriorly, lated embryos as opposed to the normal contralateral eventually ascending in a dorsolateral longitudinal tract. ascending pathway. These effects on growth cones were Not only is the floor plate the site of the turns made by cell-specific, other classes of spinal neurons were not af- the growth cones, but it also emits a chemotropic signal fected. The findings demonstrate that the midline floor which attracts the growth cones of commissural neurons plate cells are not necessary for outgrowth by growth in both mammals [14] and chicks [15*] (see the review cones towards the ventral midline, but are necessaty for by M Tessier-Iavigne, this issue, pp 6&65). In light of normal error-free pathlinding at the midline. Thus, the these findings, it has been proposed that the floor plate midline floor plate cells appear to function as interme- serves as an intermediate target which provides multiple diate targets that provide multiple cell-specific guidance cell-specific guidance cues in the zebrafish spinal cord. cues. The floor plate cells may also serve as an intermediate target for reticulospinal axons in the zebrafish hindbrain. (a) Many reticulospinal neurons project an axon along a tract which is located just lateral to the floor plate region [ 171. These axons are also disarrayed in the hindbrain of cyc- I and floor plate ablated embryos (K Hatta et al: Sot Neurosci Abstr 1990, 16:310). FP- The llip side of the experiments in the spinal cord is (b) that a significant number of both CoPA and VeLD growth cones followed their normal pathways despite the ab- .--\ 0 r--b sence of the midline floor plate cells. The floor plate I ‘\ , \ I \ region usually consists of three neuroepithelial cells in ‘\ 8’ \ ._____ cross-section, one on either side of the midline cell [3*] (see [16-l for an alternative interpretation). Thus, the variable behavior of growth cones in the absence of the midline cell may be due to a quantitative decrease in a Fig. 1. (a) A schematic diagram showing the relationship of the CoPA and VeLD axons to the floor plate (FP) in a sideview of cue normally provided by the entire floor plate region. the early zebrafish spinal cord. Dashed lines indicate axons which This, however, appears not to be the case. First, ablation have crossed the ventral midline. Anterior is to the left. fb) Four of one of the cells lateral to the midline floor plate cells examples of the axonal trajectories of CoPA neurons in cyc-l em- induces no errors by CoPA growth cones (RR Bernhardt, bryos and in wildtype embryos following laser ablation of the N Nguyen and JY Kuwada, unpublished data). Second, midline floor plate cells. CoPA growth cones followed either normal or aberrant pathways. ablation of all the floor plate cells (the midline and lateral cells) in wildtype embryos, and neuroepithelial cells near the ventral midline in cyc-l embryos, did not pre- This hypothesis has recently been tested by examination vent CoPA growth cones from extending to the midline of pathfinding by the VeLD and CoPA growth cones in the along their normal route. Once at the midline their error absence of a morphologically and antigenically distinct rate was not significantly higher than that seen in unma- subset of floor plate cells, the midline floor plate cells nipulated cyc-I embtyos (RR Bernhardt and JY Kuwada, (RR Bernhardt and JY Kuwada: Sot Neurosci Abstr 1990, unpublished data). 16309; JY Kuwada and K Hat@ Sot Neurosci Abstr 1990, I6:309). These cells form a single row which straddles The variable trajectories followed in the absence of the the ventral midline. The midline floor plate cells were midline floor plate cells may indicate that a second source specifically eliminated either by laser ablation prior to ax- of pathfinding cues exist near the ventral midline. Two onal outgrowth or by the cyc 2 mutation. The cyc- 2 (bZG) likely candidates are the basal lamina under the floor mutation results in a pleiotropic recessive phenotype that plate and the notochord. CoPA growth cones cross the is 92% penetrant [16**]. In the mutant cord the pheno- midline by extending between the ventral surface of the type appears to be limited to the deletion of the midline floor plate cells and the basal lamina. The notochord floor plate cells, which never appear to develop. In both is located just ventral to the spinal cord and contacts mutant and laser ablated embryos, the absence of the the cord during the entire period growth cones are ap- midline floor plate cells did not prevent outgrowth to- proaching and crossing the midline [ 3*]. If guidance cues wards the ventral midline by the VeLD and CoPA growth are deposited on the basal lamina, then the source of the cones. Once near the midline, however, some growth secreted cue could either be the floor plate cells or the cones followed aberrant pathways while others followed notochord. Of the two cell types, the notochord is more correct pathways (Fig. lb). For example, CoPA growth likely to be a source as in cyc- I embryos the midline floor cones followed aberrant pathways in 57 % of cases in plate cells never appear to develop [16**]. cyc-l embryos, compared to 1% in wildtype embryos. If the notochord is another source of pathfinding cues, The range of aberrant pathways taken by the CoPA and then elimination of the notochord would be expected Growth cone guidance in the zebrafish central nervous system Kuwada 33 to induce errors in pathfinding, and elimination of both the notochord and floor plate would perhaps induce DVDT PC a higher rate of errors than would the elimination of the floor plate or notochord alone. In fact, disruption of the normal pattern of axons does occur in Xenopus and mouse embryos when the notochord and floor plate are missing [18,19]. Ultraviolet irradiation of the vege- TPOC 1 tal poles of Xenopus embryos prior to the first cleavage MLF produces embryos which lack both the notochord and fb) floor plate. Likewise, in the cat&l regions of the neuraxis of embryos a&ted with the Danforth’s short tail (Sd, DVDT PC DVDT PC DVDT PC mutation, the spinal cord is missing both the notochord and floor plate. In both cases some spinal axons are disarrayed, but it is not presently possible to ascribe these defects to either the missing floor plate or the notochord, or both, as they were both absent. I iEPCI 1..-I\,) MLF MLF-., : MLF _ .’ In the anterior region of the neuraxis of the Sd embryos the spinal cords are sometimes missing the notochord but not the floor plate. In these embryos the pattern of Fig. 2. (a) The pathway taken by the nucleus of the posterior commissure (nucPC) growth cones along the early scaffold of tracts commissural axons in cross sections of rostral regions in the brain. Only the portion of the scaffold in the diencephalon of the spinal cord appear identical to those taken from and midbrain is depicted. DVDT, dorsoventral diencephalic tract. sibling wildtype embryos. This demonstrates that the ini- MLF, medial longitudinal fasciculus. TPOC, tract of the posterior tial circumferential portions of commissural axons to the commissure. (b)-Three examples of the pathways taken by the nucPC growth cones in the absence of the TPOC. NucPC growth floor plate were not obviously perturbed. The trajectories ” cones follow either normal or aberrant pathways. For m&e de- of these axons once at the floor plate, however, are dif- tails see text. ficult to discern from cross sections of spinal cords. For example, mistakes such as turning ipsilaterally rather than contralaterally, or posteriorly rather than anteriorly, in the dent of the TPOC axons in the dorsolateral tegmentum, floor plate region by some but not all axons would be or are prevented from entering incorrect pathways. These diIEcult to detect. As both the floor plate and notochord alternatives have been tested by surgically preventing the are absent in cases in which pathfinding errors are ap- TPOC axons, which are projected by a cluster of neu- parent, and as it is unclear whether some commissural rons in the forebrain, from entering the midbrain. Fol- axons have abnormal trajectories in the absence of the lowing this manipulation, nucPC growth cones followed notochord alone, the notochord’s role in the guidance of either their normal longitudinal pathway in the dorsolat- growth cones in the floor plate region remains an open eral tegmentum (61%) or inappropriate pathways (39 %> question. (Fig. 2b). Growth cones following their normal pathway extended along the endfeet of neuroepithelial cells located in the dorsolateral tegmentum (AE3 Chitnis et al: Brain Sot Neurosci A&r 1991, 17:40). Nearly all the growth The growth cones of specific clusters of brain neurons cones following aberrant pathways followed other tracts, also follow precise cluster-specific pathways in the early and growth cones from cells within a single nucPC of- zebra&h brain. Pathfinding within the early scaffold of ten followed both normal and aberrant pathways. These tracts by the growth cones of one cluster, the nucleus results support the following conclusions. First, selec- of the posterior commissure (nucPC), located near the tive fasciculation is important for guiding nucPC growth forebrain/midbrain border, has recently been analyzed cones, as in the absence of the TPOC axons a signifi- experimentally [20**]. The midbrain contains two bilat- cant proportion of nucPC growth cones make mistakes. erally paired longitudinal tracts, the tract of the postop- Second, inhibition of nucPC growth cones along other tic commissure (TPOC) and the medial longitudinal fas tracts is not likely to be a dominant factor during nor- ciculus (MLF), connected by the posterior commissure mal pathfinding as nucPC growth cones can follow other (PC). NucPC growth cones extend ventrally along the tracts in experimental embryos. Third, nucPC growth PC to the anterior tegmentum where the PC intersects cones are not obligated to follow each other as they of- the TPOC and MLF. At this intersection nucPC growth ten take divergent pathways in the absence of the TPOC. cones, in principle, could turn anteriorly or posteriorly Fourth, pathhnding cues associated with the dorsolateral onto either the TPOC or the MLF, but they virtually always tegmentum independent of the TPOC axons, perhaps turn posteriorly to fasciculate with the TPOC axons in the located along a longitudinal strip of neuroepithelial end- dorsolateral tegmentum and follow these axons into the feet, also appear to play an important role in guiding hindbrain (Fig. 2a). nucPC growth cones as many can follow normal pathways despite the absence of the TPOC axons. Multiple The turning behavior of the nucPC growth cones could redundant cues therefore appear to act to guide growth occur because they selectively fasciculate with the TPOC cones in the brain as well as in the spinal cord of ze- axons, respond to longitudinal pathhnding cues indepen- brafish embryos. 34 Development

Multiple redundant mechanisms of growth pioneer axons in the grasshopper central nervous system cone guidance [X$291.

Multiple redundant cues that guide growth cones at a choice site may involve one or more molecules expressed by more than one cell or group of cells, or a number of Perspective molecules on a single cell or group of cells. Such multiple cues may occur in a variety of animals as the elimina- Experimental analyses of pathfinding by growth cones, tion of putative patiding cues can result in growth including those of the zebrafish embryo, clearly demon- cones following both normal and abnormal pathways in strate that growth cones actively select appropriate path- several different species. First, in the grasshopper cen- ways, and suggest that multiple redundant cues may be tral nervous system the G growth cones turn correctly, operative in insuring normal error-free pamding by but extend much less than normal or fail to make their these growth cones. Future investigations should iden- correct turn, following ablation of the axons which they tify the sources of putative redundant cues and help us normally turn onto [21]. Second, blocking the action of understand how multiple redundant cues may interact the fasciclin 2 adhesion molecule in the grasshopper cen- with growth cones. This is especially promising in the tral nervous system, by application of a specific antibody, simple and well characterized spinal cord of the zebrafish caused growth cones that normally express fasciclin 2 embryo. to either extend much less along an apparently normal pathway or extend along abnormal pathways [ 221. Third, mutations in several unc genes in the nematode lead to an increased number of pathfinding errors made by References and recommended reading identified neurons; but the same neurons can also have normal trajectories [23-l. Fourth, genetic suppression of Papers of particular interest, published within the annual period of re- putative guidepost cells in the Dro.qbilu wing does not view, have been highlighted as: . of special interest prevent normal pathlinding by wing neurons, but may . . of outstanding interest increase errors in pathfinding by these cells [ 241. 1. BERNHARDTRR, CWMIS AB, LINDAMERL, KUWADAJY: Identi- The existence of multiple redundant cues is also con- fication of Spinal Cord Neurons in Embryonic and Larval sistent with a decrease in axonal outgrowth following Zebrafish. J Camp Neural 1990, 302607-616. the elimination of pathfinding cues. In vitro, the growth 2. KUWADAJY, BE- RR, NGUYEN N: Development of cones of chick ciliary ganglion cells extend less along Spinal Neurons and Tracts in the Zebrafish Embryo. J Camp axons following the application of antibodies against sev- Neural 1990, 302:617A28. eral adhesion molecules normally found on the axons, 3. KLIWADAJY, BE RNHARDTRR, CHUTNEYAB: Pathfinding by Iden- but extension was not completely eliminated [ 251. In . tidied Growth Cones in the Spinal Cord of Zebrafish Em- J Neurosci viva, follower growth cones extend along apparently nor- bryos. 1990, 10:129+1308. A detailed account of the celkspecilic behaviors exhibited by identi- mal pathways, but the amount they extend decreases af- fied growth cones at the floor plate in the spinal cord of the zebrafish ter ablation of the appropriate pioneer axons in the em- embryo. This paper illustrates how the simplicity of this system has al- bryonic spinal cord of the Japanese medaka fish, and the lowed investigators to perform highly detailed experimental analyses of central nervous system of grasshopper embryos [7,21]. growth cone guidance mechanisms. Axons are similarly affected following antibody block of 4. CHITNIS AB, KUWADA JY: Axonogenesis in the Brain of Ze- axonal glycoproteins in the grasshopper central nervous brafish Embryos. J Neurcsci 1990, 10:1892-1905. system 1221. song with [5*] this paper describes the simple sctiold of tracts and the identifiable cell clusters that establish the scaffold in the early ze- Multiple cues have been directly demonstrated at the bra&h brain. The report also describes the stereotyped route taken by the nucPC growth cones within the brain scatfold and identifies a molecular level in two ways. First, the in vitro applica- potential site where these growth cones might choose among several tion of specific antibodies against adhesion molecules, pathways. j31-integti or n-cadherin, to chick ciliary neurons grown 5. Wuso~ SW, Ross IS, PARRET T, EASTER SJ: The Development on glia decreases, but does not prevent, neurite out- . of a Simple Scaffold of Axon Tracts in the Brain of the growth. Blocking both molecules simultaneously, how- Embryonic ZebraEsh Bracbydanio rerio. Develqiment 1990, ever, does prevent neurite outgrowth [26]. Second, 108:121-145. Drosophila embryos defective in the fal or abl gene Along with [4*] this paper describes the simple scaffold of tracts found in the early zebra&h brain. loci, which encode putative adhesion and signal trans- duction molecules, respectively, exhibit little disruption 6. WIISON SW, EASTER SJ: A Pioneering Growth Cone in the in axonal outgrowth. Extensive pathfinding errors are Embryonic Zebra&h Brain. Proc Natl Acud Sci LISA 1991, W&22932296 obselved, however, in double mutants defective in both genes [ 27**]. 7. KUWADA JY: CelI Recognition by Neuronal Growth Cones in a Simple Vertebrate Embryo. Science 1986, 233:740-746.

Although some growth cones appear to rely on the simul- 8. GCQDMAN CS, BAST~AN~MJ, tk3E CQ, DU L4C S, HELFAND Sl, taneous action of several guidance cues, not all growth KUWADA JY, ‘Ih0mu.s JB: Cell Recognition During Neuronal cones necessarily require multiple cellular cues. For ex- Development. Science 1984, 225:1271-1279.

ample, some follower growth cones follow incorrect 9. O’LEARY DDM, BICKNTSE JA, DE CARGOS DA, HEFFNER CD, pathways, or do not extend at all after ablation of other . KOESTER SE, Km LJ, TERASHIMAT: Target Selection by Growth cone guidance in the zebrafish central nervous system Kuwada 35

Cortical Axons: Alternative Mechanisms to Establish Ax- 20. CHITNISAB, KLIWADAJYz Ehmination of a Brain Tract Ifl- onaI Connections in the Developing Brain. Coki Spring Hurb . . creases Errors in PatMnding by Follower Growth Cones Sjwnp Quunt Bid 1990, LV:45+468. in the Zebrafish Embryo. Neuron 1991, 7:277-285. The diiferent ways in which mammalian cortical neurons establish their This study takes advantage of the detailed description of the behavior of axon trajectories are discussed in this review. Layer 5 neurons establish a specik set of growth cones at a potential choice site in the zebratish their mature trajectories by first projecting growth cones which follow brain. Elimination of the axons which these growth cones normaUy turn ceilspeciiic pathways within the cortex. Beyond this initial phase, ax onto at this site causes the growth cones to follow either aberrant or onogenesis involves the extension of collateral branches to potential correct pathways. This study illustrates the usefulness of the zebrafish targets followed by selective elimination of a subset of the branches. brain for analysis of growth cone guidance in the vertebrate brain. This also occurs in other organisms including chicks and leeches. 21. RARERJA, BAST!ANIMJ, GOODMANCS: Path6nding by Neuronal 10. BOVALENTAP, DODD J: Guidance of CommissuraI Growth Growth Cones in Grasshopper Embryos. Iv. The Effects of . Cones at the Floor Plate in Embryonic Rat Spinal Cord. Ablating the A and P Axons Upon the Behavior of the G Development 1990, lm435-447. Growth Cone. J Neurosci 1984, 4:23292345. A clear demonstration that the axons of commissurai growth cones mm 22. HARRELSoNAL, GCXIDMANCS: Growth Cone Guidance in In- precisely at the lateral border between the floor plate and the rest of sects: Fascichn II is a Member of the ImmunogIobuhn Su- the spinal cord. perfamily. Science 1988, 242:70&708. 11. YAGINUMAH, HOMMAS, KUNZ R, OPPENHEIMRW: Pathtind- 23. HEDGECOCKEM, Cu~orn JG, Hw DH: The Unc-5, ~nc-6, ing by Growth Cones of Commissural Intemeurons in the . . and Unc-40 Genes Guide Circumferentiai’Migrations of Pi- Chick Embryo Spinal Cord: a Light and Electron Microsopic oneer Axons and Mescdermai CeRs on the Epidermis in C. Study. J Camp Neural 1991, 304:78102. elegans Neuron 1990 4:6185. _,. -:, This paper describes in detail the effects of mutations in three genes 12. ROBER’IXA, DALEN, OTTERSENOP, STORM-MATFIISENJ: Develop- on navigation by identified growth cones and migration by mesodermal ment and Characterization of Commissural Intemeurones cells in the nematode. It clearly demonstrates with a multi-allele anaIysis in the Spinal Cord of Xenopus Iaeuis Embryos Revealed that mutations in these genes can cause pioneer growth cones to follow by Antibodies to GIycine. Development 1988, 103447461. variable pathways as weU as their normal ones. DODD J, MORTON SB, KARACXXEOSD, YAMAMOTOM, JESSELL 13. 24. SCHU~IGERM, PA~XA J: Genetic Suppression of Putative TM: Spatial Regulation of AxonaI Glycoprotein Expression Guidepost Ceils: Effect on Establishment of Nerve Path- on Subsets of Embryonic Spinal Neurons. Neuron 1988, ways in Drosophila Wigs. Deu Biol 1985, 108:3~10. 1:lO~lll. 25. CHANG S, RATHJENFG, RARERJA Extension of Neurites on 14. TESSIER-LWIGNEM, P~ACZEK M, LUMSDENAGS, DODD J, Axons Is Impaired by Antibodies Against SpeciIic Neural JESSEUTM: Chemotropic Guidance of Developing Axons CeII Surface Glycoproteins. J Cell Biol 1987, 104:355362. in the Mammalian CentraI Nervous System. Nature 1989, TOMAZ.ELLIKJ, NEUGEBAUERKM, BIXBY JL, LILIENJ, REICHARDT 336775-778. 26. LF: N-Cadhecin and Integrins: Two Receptor Systems that 15. YAGWUMAH, OPPENHEIMRW: An Experimental Analysis of In Mediate Neuronal Process Outgrowth on Astrocyte Sur- . Viva Guidance Cues Used by Axons of Spinal Intemeurons faces. Neuron 1988, 1:3343. in the Chick Embryo: Evidence for Chemotropism and Re- 27. EUUNS T, ZINN K, MCALLISTERL, HOFFMANNFM, GOODMAN lated Guidance Mechanisms. J Neurosci 1991, 11:25982613. . . CS: Genetic Analysis of a Drosophila Neural Cell Adhesion Spinal cord grafts rotated relative to the host spinal cord demonstrate Molecule: Interaction of Fasciclin I and Abelson TymsIne that commissural axons near the graft/host boundary are attracted to Kinase Mutations. Ceil 1990, c&565-575. the nearest set of Boor plate cells. This experiment uses the advanced molecular genetics of Drosqohih 16. HATrA K, KIMMELCB, Ho RR, WALKERC: The Cyclops Muta- to clarify the roles of putative pathiinding genes. Most importandy it . . tion Blocks SpeciEcation of the Fioor Plate of the Zebraiish demonstrates that the growth cones of neurons, e.g. RPl, which ex- Central Nervous System. Nature 1991, 3503339341. press both the fasciciin I (fas 1) adhesion molecule and the Abelson This important paper illustrates the power of the genetic approach for tyrosine kinase <& protein show no gross defects in mutants defec the anaIysis of the development of zebrafish embryos by characterizing tive in one of the genes, but cleady follow aberrant pathways in double me cycl~blb) mutation in zebraiish embryos. It demonstrates that in mutants. This vividly demonstrates that the fm I and &I genes interact. the spinal cord the mutation deletes the midline floor plate ceils. The Double mutants homozygous for the fm I and numerous other genes, elegant analysis of mosaic embryos demonstrates that the mutation acts including other adhesion molecule genes, however, show little pheno- cell autonomously and that floor plate cells can be induced by other typ=e.Interestingly 7 % of RPl neurons still follow normal pathways even Boor plate cells. in the fm I ubldouble murants.

17. KIMMELCB, POWELLSL, METCALFEWK: Brain Neurons which 28. BASTIANIMJ, DU IAC S, GooDMAN CS: Guidance of Neuronal Project to the Spinal Cord in Young Larvae of the Zebraiish. Growth Cones in the Grasshopper Embryo. I. Recognition J Conap NeuroI 1982, 205:112-127. of a SpeciEc Axonal Pathway by the pCC Neuron. J Neurosci 1986, 6:35183531. 18. BOVOLENTAP, DODD J: Perturbation of Neuronai Diieren- tiation and Axon Guidance in the Spinal Cord of Mouse 29. DU L4C S, BASTIANIMJ, GC0DMANCS: Guidance of Neuronal Embryos Lacking a Fioor Plate: Analysis of Danforth’s Short- Growth Cones in the Grasshopper Embryo. IL Recognition Tail Mutation. Deuelqxnent 1991, 1133625639. of a Speciiic Axonai Pathway by the aCC Neuron. J Neurosci 1986, 6:3532-3541. 19. CLARKEJDW, HOLDERN, S~FFE SR, STORM-MAIHISENJ: Neuro- anatomical and Functional Analysis of Neural Tube Forma- tion *in Notochordless Xenopus Embryos; LateraIity of the JY Kuwada, Department of Biology, Natural Science Building, University Ventral Spinal Cord is Lost. DevelqOment 1991, 112:49’+516. of Michigan, Ann Arbor, Michigan 48109.1048, USA