THE JOURNAL OF EXPERIMENTAL ZOOLOGY 261:310-321(1992)

Molecular Mechanisms of Axon Guidance in the Developing Insect Nervous System ALLAN L. HARRELSON Division of Biological Sciences, University of Missouri-Columbia, Columbia, Missouri 65211 For many years scientists have employed a variety of species, including insects such as the grasshopper Schistocerca americana, the moth Manduca sexta, and the fruit fly Drosophilu melanoguster, as model organisms for the study of important questions in developmental neurobiology. This review addresses the question of axon outgrowth and synaptic specificity in the developing insect nervous system. The general problems that must be solved in neural differentiation are the same in insects as in more recently evolved species. A sufficient number of neurons must be produced in both the peripheral nervous system (PNS) and the central nervous system (CNS). Sensory neurons must differentiate to form receptive structures for light, air vibration, and airborne and liquid-phase molecules. This differentiation must be coupled with the projection of axons into appropriate regions of the CNS to synapse with networks of central neurons, which integrate sensory impulses. Central neurons must make the necessary connections within and between these central networks to allow accurate processing and integration of this incoming sensory information, leading to appropriate motor and neuroendocrine responses. Many of the same questions occur with whatever model system we use for our investigative efforts: How are nerve cells produced? How are they differentiated from one another t o become functionally distinct? How are they positioned appropriately and connected together to assemble functional circuits? The answers t o such questions are at present only fragmentary and encompass data from studies at the molecular, cellular, circuit, and organismal levels. Nevertheless, the pace of research in this area is rapidly accelerating, and work on insects has played a crucial role. This paper focuses on several recent molecular-level studies on axon outgrowth and guidance. For surveys of cellular-level data on this subject, the reader is referred to reviews by Bate ('781, Anderson et al. ('80), and Lnenicka and Murphey ('89). 0 1992 WILEY-LISS.INC.

Axon outgrowth comprises the events beginning when the growth cone (GC) is organized and initiates extension from the neuronal soma and ending when the growth cone makes contact with the final target cell(s) and ceases motility. Next, the process of synaptogenesis begins, resulting in morphologically specialized contacts for transmitter release and impulse transmission. The study of these processes has shown that numerous complex molecular interactions underlie these cellular events, and undoubtedly many gene products controlling these processes remain to be discovered and characterized.

PERIPHERAL NERVOUS SYSTEM The molecular cues responsible for accurate PNS axon ougrowth specificity are unknown at present. Presumably in the periphery they include some cue derived from the epithelium, since cellular studies have shown that the epithelium is sufficient to allow proper ingrowth of the PNS sensory neurons (Lefcort and Bentley, '87; Caudy and Bentley, '86, '87). The work of Nardi ('83) on epithelium in Munduca suggests that there may be gradients of adhesivity to which peripheral neurons respond. Unfortunately, no molecules have been identified so far that show a graded proximal/distal distribution corresponding t o the general direction of sensory axon growth. In general investigators have looked for candidates among antigens expressed on the cell surface and secreted from cells. One would expect extracellular matrix (ECM)molecules in the periphery to function in adhesion of growth cones and axons, and studies by Anderson and Tucker and other support this notion (Andersonand Tucker, '88, '89; Condic and Bentley, '89a-c). For example, enzymatic destruction of the peripheral ECM can disrupt sensory axon outgrowth (Condic and Bentley, '89a-c). Furthermore, Anderson and Tucker ('89)find that there are features of ECM that vary within the grasshopper limb and over development and that might conceivably function as guidance cues for GCs. It is important to distinguish, however, between manipulations that change the direction of GC

INSECT NEURAL DEVELOPMENT

movement and those that completely prevent axon outgrowth. ECM molecules may be permissive, providing a n adherent substrate that allows cells to extend their axons without imparting specific directionality. Moreover, the effectsof molecules in the ECM may be complex and conditional. Condic and Bentley ('89a-c) have found that peripheral sensory neurons will grow toward the CNS even after treatment of the ECM with certain protease enzymes (elastase, ficin, and papain) and its apparent loss (via scanning electron microscopy). In the grasshopper embryo, if digestion occurs before the identified Ti1 neuron axon grows out, then the growth cone still extends properly. On the other hand, with digestion of the ECM before the Ti1 axon contacts the more proximal F e l neuron, the Ti1 growth cone retracts partially or completely; the effect is dose dependent and reversible, since after time the Ti1 growth cone will reemerge and grow toward the CNS. This suggests that ECM cues can be used by the Ti1 GC for outgrowth but are not essential. If digestion occurs after Ti1 has contacted Fel, then the cell-cell contacts are maintained; interestingly, Ti1 neurons that have extended their growth cones after the basal lamina was digested with elastase will no longer respond to a new elastase treatment by retracting; they are resistant. This suggests that there may be multiple guidance cues and/or adhesion molecules and that, when one cue from the ECM is eliminated at a n early time, at later points its presence or absence no longer has any effect. While soluble tropic factors might conceivably play a role, they are clearly not essential for some PNS neurons. In the grasshopper appendage, sensory GCs migrate centrally but apparently without any necessary central tropic influence: limbs and antennae cultured in isolation from the CNS still show proper GC directionality (Sanes et al., '76; Berlot and Goodman, '84;Lefcorte and Bentley, '87). Recently scientists have begun to look for molecular correlates of axon outgrowth specificity in the neuronal cytoplasm of GCs. Within the neurons of the developing insect PNS, what molecules might function in the GC to allow outgrowth, adhesion, etc.? One obvious candidate is calcium. Numerous studies of GCs in vitro have correlated axon outgrowth and GC morphology and motility with intracellular calcium concentration (Kater et al., '88). Recently Bentley et al. ('91) have begun to address this issue in the developing grasshopper. Using calcium-sensitive fluorescent dyes and video microscopy, they found that calcium levels increase in sen-

311

sory neurons with the onset of axonogenesis, to levels in the range of 100-115 nM free calcium. Some pioneer neurons exhibit a shallow gradient of calcium, highest at the growth cone and lowest at the cell body. There was some tentative evidence that sensory axon GCs could transfer calcium to more proximal neurons t h a t had not yet initiated axonogenesis, possibly via transient gap junctions formed between the cells. This study suggests that calcium levels fluctuate in correlation to specific cellular events in pathfinding; however, the function of calcium in this process is not clear. The question remains whether the changes in calcium are a cause of guidance or a result of guidance mechanisms. In summary, in the PNS, despite the relatively simplified anatomy compared with the developing CNS, we have relatively few new candidate molecules for functional studies on axon outgrowth specificity. The cellular studies suggest a certain degree of redundancy in environmental cues. The demonstrated ability of peripheral sensory neurons to grow into the CNS without the aid of other sensory axons, suggesting a relatively uniform capability for proper pathfinding, makes it more difficult to pick a candidate molecular expression pattern to test for functional importance; i.e., possible guidance candidate molecules would be expected to show at most a gradient or repetition of expression relative to the proximal-distal or anterior-posterior axis of the limb or body wall. So far, no such candidates have been put forward.

CENTRAL NERVOUS SYSTEM Fasciclins Molecular studies on CNS axon outgrowth in insect are, on balance, more numerous and detailed than those on the PNS. Early studies suggested that some type of surface cue was involved in the ability of GCs to recognize their proper targets (for reviews, see Goodman et al., '82; Bastiani et al., '85). A number of investigations at the molecular level were undertaken to discover candidate guidance molecules (Bastiani et al., '87; Pate1 et al., '87; Snow et al., '88; Zinn et al., '88; Harrelson and Goodman, '88; Bieber e t al., '89; Snow et al., '89; Hortsch et al., '90a). The first three candidates to emerge from this project were proteins expressed on the surfaces of axon bundles (or fascicles) in the embryonic CNS and were therefore called fasciclins. The term fasciclin is an operational term, not meant to imply that the molecules are structurally similar or functionally related. The fasciclin 1cell surface glycoprotein (Fasl)has a complex expression pattern, associated with rapid

312

A.L. HARRELSON

alterations of protein synthesis in a restricted subset of neurons during the period of axon outgrowth, strongly suggestive of a pathfinding function (Bastiani et al., '87). In the grasshopper Schistocerca americana, it is expressed on a subset of cells in the neurogenic region during neuroblast formation and neurogenesis, although there is no exact correlation between the neuroectoderm and later expression in progeny neurons. In the early CNS, Fasl is expressed on a subset of commissural and longitudinal axon bundles. The protein is clearly expressed on only parts of the surfaces of some axons and is also expressed on the surfaces of some GCs and on filopodia. The Fasl protein is expressed on the intersegmental nerve and particular axon bundles in the segmental nerve as well as on parts of the cell bodies of some of the neurons giving rise to these bundles. Moreover, the Fasl protein occurs in the PNS, being present on all embryonic sensory neurons (and also on a subset of peripheral ectodermal cells). The complexity and transient nature of Fasl expression indicate a complex regulatory mechanism, lending credence to the idea that the molecule functions in cell recognition. The pattern of expression of the Fasl protein on identified neurons and axon bundles in the Drosophila embryo is generally similar to that of the grasshopper (Zinn et al., '88). For example, the intersegmental nerve and the aCC neuron are positive, while the pCC neuron is negative. However, the axon bundle homologies between grasshopper and Drosophila are not yet completely clear and require further study for more precise comparison of Fasl expression on individual nerve cells and fascicles. In general, the levels of the Fasl mRNA during different developmental stages correlate strongly with the presence of the protein, but little is known about its distribution in situ. Although the anatomical expression pattern of Fasl suggests a pathfinding function, its structure provides few clues to support or disprove this idea (Zinn et al., '88).The mature grasshopper Fasl protein has 638 amino acids, giving a molecular weight of about 70 kD (Zinn et al., '88). The protein has a hydrophobic signal sequence at the 5' end (the N terminus of the protein) and is anchored in the outer cell membrane by a glycophosphatidyl-inositol anchor. There are six potential N-glycosylationsites, but only two are conserved in the Drosophila homolog. The protein has an internal sequence repeat of about 150 amino acids, a domain that occurs three times and within which there is a region of 23-40 amino acids with highest sequence similarity. This domain structure is conserved in the Fasl form from

Drosophila, at a level of 75-90% within these domains. Overall, the grasshopper and Drosophila Fasl proteins have 48% amino acid sequence identity, but with no significant similarity to any other proteins. The gene for Fasl in D. melanogaster is located on the third chromosome, at the 89D cytogenetic position (Zinn et al., '88). Null mutations exist that make no detectable Fasl mRNA. Surprisingly, flies homozygous for the null mutation are viable, showing normal fertility and no gross anatomical defects in the CNS. They do have a behavioral defect, however: they show uncoordinated movement and are hesitant in stepping and patterned walking (Elkins et al., '90b). However, when flies are made doubly mutant for Fasl and the Drosophila abl homolog, there are major developmental defects in the CNS. Drosophila abl is a protein kinase expressed initially in the CNS on all axons, becoming restricted to the longitudinal segments. Fasl lab1 homozygous double mutants show gross defects in the developing CNS axon scaffold.The commissural tract axons are often lost in bundles where Fasl expression is normally strongest, and there is a variable loss of axons in longitudinal tracts between segments. In 80%of double mutants, the axon outgrowth by the RP1 neuron is disrupted, and first-instar larvae show uncoordinated motor behavior. Interestingly, there is no apparent morphological defect in the PNS or in other tissues, and viability is not reduced below simple abl homozygous mutations. The effect of abl mutations on Fasl was specific: mutations in other CNS genes such as neuroglian, FUSS,disco, Notch, the c-src homolog, torpedo, and others showed no ability to affect the Fasl phenotype. These genetic results suggest that the Fasl protein may have a pathfinding function, but by itself it may be redundant to other molecules expressed in the developing nervous system. On the other hand, it is likely thet Fasl has unique functions that are very subtle, affecting only a small set of cells in a detectable way. Like Fasl, the second fasciclin, Fas2, was initially isolated from grasshopper (Bastiani et al., '87). It is also a cell surface glycoprotein, but unlike Fasl it is expressed in the CNS primarily on longitudinal axons. Fas2 occurs initially on longitudinal cells that pioneer one of the main longitudinal bundles, at the time when their axons begin to fasciculate together to extend from segment to segment. Fas2 occurs on some neuronal cell bodies in its initial phase of expression, but soon becomes restricted to axonal segments and excluded from the surface of the cell soma. As embryonic development proceeds,

INSECT NEURAL DEVELOPMENT

Fas2 expression spreads to all longitudinal axon segments. Conversely,except for an early transient expression, the protein does not occur on commissural axon segments. The expression pattern suggests that this protein may be involved in selective fasciculation of specific longitudinal axon bundles and/or as a general label of longitudinal axon segments. Since many neurons have both commissural and longitudinal axon segments, the distribution of Fas2 suggests that neurons are capable of partitioning specific molecules to geometrically defined regions of their cell surfaces. Fas2 has been cloned and sequenced and shown to be a member of the immunoglobulin (lg) gene superfamily (Harrelson and Goodman, '88).It has five lg-like domains, two fibronectin-related domains, a putative transmembrane segment, and a small cytoplasmic domain. This domain identity and organization are similar t o several other proteins, such as N-CAM and L1, which are neural cell adhesion molecules in vertebrates (Williams,'87; Hunkapiller and Hood, '87, '89; Williams and Barclay, '88). When grasshopper embryos are cultured in vitro in the presence of monoclonal antibodies against the extracellular portion of Fas2, specific alterations in longitudinal axon outgrowth are observed, suggesting that Fas2 may function in the fasciculation of these axon bundles. Such a function would accord with its structure and developmental expression pattern. The third fasciclin, Fas3, cloned and characterized originally from D. melanogaster, is expressed in a complex, dynamic pattern in the developing embryo, more closely resembling the Fasl pattern than that of Fas2 (Patel et al., '87; Snow et al., '89). In the CNS, it occurs on a subset of axons different from Fasl but in about the same number of cells: five commissural fascicles show expression, and there is clearly regional expression on axon segments. For example, Fas3 is expressed on the cell body and axon of the RP1 neuron but only on the cell body of the RP2 neuron. Other expressing cells include a subset of dorsal glia near the segment border and patches of epidermis in the neurogenic region. There is also transient expression on about 20% of the neuroectoderm surface, on patches about three or four cells long and four cells wide. Interestingly, Fas3 is absent from the neuroblasts under these patches but is expressed on their ganglionmother cells and their progeny neurons. There is not a strong correlation of neuroectoderm expression and neuronal expression at the time of axonogenesis. Therefore, there is no direct relation between cell lineage and expression of Fas3, which

313

is similar to what is observed for Fasl. Outside the nervous system, Fas3 is expressed on epidermal stripes at segmental grooves and patches of epithelial cells near stomodeal and proctodeal invaginations. In addition, there is strong Fas3 expression in the visceral mesoderm around the gut and the luminal surface of the salivary gland epithelium. Although Fas3 has an embryonic expression pattern somewhat analogous to that of Fasl, the two molecules are not structurally related. Fas3 is a medium-sized cell surface glycoprotein of approximately 80 kD (Patel et al., '87). The Fas3 amino acid sequence has been deduced from the DNA sequence of the corresponding cDNA and is not similar to that of other fasciclins or any other known protein (Snow et al., '89). The function of Fas3 is not clear: Fas3 null mutations in Drosophila are recessive, and homozygous null mutations are viable. Nevertheless, studies on Fas3 protein expressed in cultured cells suggest that it may function as a homophilic, calcium-independent adhesion molecule (Snow et al., '89). Thus, like Fasl, Fas3 appears to have the potential to mediate specific axonal fasciculation, but the genetic evidence from Drosophila suggests that at best only a small subset of neurons are actually affected by loss of the Fas3 gene.

Other CNS cell surface and extra cellular molecules A number of other molecules have been identified that seem likely to have functions in insect CNS development and possibly in axonal pathfinding. Probably the most intensely studied is a protein called neuroglian (Bieber et al., '89). Neuroglian protein is expressed in the Drosophila embryo CNS on the surfaces of a substantial group of neurons, including identified cells such as the RPs. As with the fasciclins, axons of the peripheral nerve roots express neuroglian protein. While many longitudinal axon bundles are stained by antibodies against neuroglian, the protein is expressed on a more limited number of commissural bundles. Thus neuroglian has a general expression pattern on a larger group of cells than Fasl or Fas3 but occurs during the period of axon outgrowth and thus might have a pathfinding function. Such a role is also suggested by neuroglian expression in other CNS cells: glial cells also express neuroglian protein. Specifically, glia at the roots of the segmental and intersegmental nerves are stained by antineuroglian antibodies, as are longitudinal and midline glia in the CNS. The neuroglian expression is especially intense at regions of

314

A.L. HARRELSON

contact with labelled axons or neuronal cell bodies. In all cases, neuroglian expression occurs on the cell surface. The overall cellular expression of the neuroglian mRNA is similar t o the protein expression pattern. The structure of the neuroglian protein is also suggestive of a cell adhesion and/or recognition function in embryonic development.The protein appears on reducing polyacrylamide gels as three bands with approximate molecular weights of 180,167, and 155 kD, the middle band being the most intense. As for some other neural cell adhesion molecules, there appears to be significant glycosylation of neuroglian with 0-linked and N-linked sugars. Like Fas2, neuroglian is a member of the immunoglobulin gene superfamily. Molecular cloning experiments have revealed that the neuroglian has 1,216 amino acids and a predicted molecular weight of 145 kD. The sequence consists of a presumptive signal sequence, 1,115 extracellular amino acids, a potential transmembrane domain, and 85 amino acids in the cytoplasmic domain. Neuroglian has six Ig-related regions and five type three fibronectinlike domains. Molecular analysis by Hortsch et al. ('90a) indicates that differential mRNA splicing generates a nervous system-specificneuroglian form. This form has an identical extracellular sequence but a longer cytoplasmic domain and is specifically expressed in neurons and some peripheral support cells for sensory structures. Notwithstanding these suggestive anatomical and biochemical results, the precise function of neuroglian is not yet known. The neuroglian gene lies at 7F of the D. melanogaster X chromosome, and recessive lethal mutations are known. However, the homozygous lethal alleles produce embryos with no gross morphological defects in CNS development. Thus, like the fasciclins, neuroglian appears to have all the features of a molecule that would play a role in axon outgrowth, but the genetic experiments in Drosophila suggest that the precise neuroglian function is subtle. Neuroglian is related to another cell surface protein expressed in the developing insect nervous system, the Drosophila amalgam protein, which is also a member of the Ig gene superfamily (Seeger et al., '88). Like the fasciclins and neuroglian, amalgam is synthesized by nonneuronal cells as well as within the nervous system. It appears t o be expressed at a high level at midembryogenesis and during early pupation. The protein is first expressed transiently in embryonic mesoderm shortly after gastrulation and ends by stage 16 but is expressed on a subset of splanchnopleur derivatives (muscles, fat body,

etc.). At the time of neuroblast segregation, amalgam protein is expressed in a subset of the ventral midline cells of the CNS. Apparently no neuroblasts stain with antiamalgam antibodies, but possibly neural support cells are expressing the protein. Then, ganglion mother cells and/or neurons express the protein. Amalgam protein is expressed on apparently all CNS neuron cell bodies and axons starting at stage 13, when most are extending GCs. Only a small subset of peripheral neurons expresses amalgam. No peripheral axons are stained. At stage 13, a single cluster of PNS neuron cell bodies are amalgam positive, and the spiracle organis also positive. By stage 16, there are three rows of PNS cells positive along the dorsoventral axis of each segment. At this stage, cephalic sensory organs also begin to stain. The amalgam mRNA is apparent at 0-2 hr of embryogenesis, peaks at 6-8 hr and has a low level after 12 hr of embryonic development. Amalgam is structurally related to Fas2 and neuroglian but is considerably smaller. It has 333 amino acids, based on the cDNA sequence. The protein appears t o be extracellular and/or secreted, since the DNA data show a putative signal sequence and no evidence for a potential hydrophobic transmembrane domain for anchoring the protein to the membrane. It has three Ig-like domains: the first domain has V-type homology, the second and third domains are C2-type Ig domains; this is somewhat different from Fas2 and neuroglian, which lack the V-type Ig domain. All three amalgam Ig-related domains are coded by a single genomic exon, which is also different from many other members of the Ig gene superfamily (Williams,'87; Hunkapiller and Hood, '87, '89; Williams and Barclay, '88).The closest sequence relation with another protein is with N-CAM, which shows 25% identity over 228 amino acids. In a shorter stretch, there is greater similarity of Fas2,32% identity over 144 amino acids. There is as yet relatively little evidence for the function of the amalgam protein, but again its general structure and expression suggest the potential for a role in axon outgrowth. Not all adhesion proteins in insects are members of the Ig gene superfamily. Another protein expressed in the embryonicDrosophila CNS is called neurotactin (Piovant and Lena, 1988; Hortsch et al., '90b; Barthalay et al., '90; De La Escalera et al., '90).Drosophila neurotactin is a cell surface, transmembrane protein expressed early in development in the cellular blastoderm, then later in cells of the developing nervous system. Neurotactin synthesis in the nervous system begins in the neuroblasts shortly after they delaminate, as they increase their

INSECT NEURAL DEVELOPMENT

expression levels above the surrounding ectoderm. All central neurons appear to express neurotactin but only a subset of peripheral sensory neurons. The neurotactin protein has an apparent molecular weight on sodium dodecyl sulfate (SDS)-polyacrylamide reducing gels of 135 kD. In contrast, three different neurotactin mRNAs are observed during development, each with a different expression profile. Its cDNA sequence shows a carboxy terminal extracellular domain similar to that of serine esterases, thyroglobulin, and acetylcholinesterase. Specifically, this domain is similar to the Drosophila ECM protein glutactin (Olson et al., '90). Neurotactin function is unlikely t o be proteolytic, since critical protease amino acid residues are not found in neurotactin. However, Barthalay et al. ('90) have presented evidence that neurotactin is a heterophilic adhesion molecule, using the same type of assay system that indicates that the fasciclins can have adhesive functions. The effects of genetic lesions in the neurotactin gene are unknown, in part due to confounding effects of nearby loci. Because of its broad distribution in the developing CNS, neurotactin likely has no specific pathfinding function; however, because expression patterns and function do not always correlate, we should hold open the possibility that neurotactin functions in only a subset of cells. Finally, among the genes discovered and characterized in Drosophilu is one coding for another potentially interesting cell surface molecule, the chaoptin protein (Zipursky et al., '85). Chaoptin is a photoreceptor-specific cell surface glycoprotein that appears on the cell surface during axonogenesis. Its amino acid sequence has 41 tandemly repeated leucine-rich motifs, making it part of a gene family related by this leucine-rich sequence and not part of the Ig gene superfamily. Chaoptin mutants show some disorganization of eye structure, perhaps related to defective cellular attachments, but show apparently normal axonogenesis, suggesting that the protein does not function in axon outgrowth. However, recent results suggest that it will act as a cell adhesion molecule when expressed in Drosophila S2 cells (Krantz and Zipursky, '90). Thus, like so many other molecules, chaoptin appears to have the potential for causing cell adhesion, but in its absence axon outgrowth seems generally normal. A number of other molecules are expressed in the developing insect CNS on axons and cells at the time of pathfinding but for which there is little or no information on molecular structure and function. Nevertheless, some have especially provocative expression patterns, which may suggest

315

specific functions in recognition of cells by GCs and selective axonal fasciculation.For example, Carpenter and Bastiani ('91) recently reported the isolation and characterization of a protein expressed on glial cells in the grasshopper embryo CNS. Monoclonal antibodies against the antigen, called REGA-I , immunoprecipitate a single glycoprotein of apparent molecular weight of about 60 kD. The REGA-1 protein is found in a subset of glial cells found at specific locations in each segment. As with the fasciclins described earlier, the REGA-1protein is expressed in a regional fashion by these cells: that is, the cell bodies lack detectable REGA-1 levels, but lamellar processes extending into the developing neuropil show high levels of protein labelling. The expressing lamellopodia appear t o limit the anterior and lateral portions of the developing axon scaffolding "box" in each segment. A different glycoprotein from cricket, with an apparent molecular weight of about 185 kD and expressed in a different set of glial cells, has also been purified (Meyer et al., '87, '88). In another example of molecular specificity in the embryonic nervous system, Denbug et al., ('89) have discovered several interesting antigens in developing cockroach, which are expressed in complex patterns on the surfaces of CNS axons and cell bodies, in patterns reminiscent of the fasciclins. Several of these antigens are carbohydrate in nature, as determined by periodate digestion, and four are apparently glycolipids. The presence of glycolipids in subsets of developing axons is provocative but of no known functional relevance. A somewhat similar situation pertains in vertebrates, wehre subsets of sensory afferents express different carbohydrate epitopes, but their function is not known (Dodd and Jessell, '85, '86; Regan et al., '86). Presumably neural development and axonal pathfinding in the insect CNS require not only a great diversity of cell surface molecules but also specific ECM components. A promising candidate in the insects is a homolog to the vertebrate ECM protein laminin (Fessler et al., '87; Monte11 and Goodman, '88, '89). Laminin is a multidomain protein composed of three subunits, the a chain, the f3lchain, and the p2 chain. All three chains have been cloned from Drosophila, and much of the domain structure (inferred from the DNA sequence) is preserved between insects and the vertebrate species. For example, in the p l chain, domains I11 and V (epidermal growth factor-like repeats) and VI (a putative collagen-binding region) have about 55% similarity between mouse and Drosophila. In domains I11 and V, there is 90% similarity in regions

316

A.L. HARRELSON

similar to the epidermal growth factor repeat receptor-binding sequence. Drosophila laminin has a cell adhesion binding site sequence found in vertebrates (YSGSR). In domain VI of the p2 chain, there is 64% identity. All three chains in laminin have been cytogenetically mapped in Drosophila; however, as with many of the other cloned genes from the insects, the mutant phenotypes associated with these laminin chain loci have not yet allowed investigators to relate laminin function t o specific events in neuronal development. To summarize, recently nearly a dozen different molecules have been discovered and characterized by having their genes cloned and their protein expression patterns studied in detail. Generally they occur on neurons during development in locations where they could easily function in specific axon outgrowth. Several of these proteins are capable of causing cell adhesion in vitro ,suggesting an in vivo function in cell-cell recognition and/or adhesion. However, for only a few of them is there any evidence for a specific function in neural development, in particular in axon pathfinding.

SEGMENTATION AND HOMEOTIC GENES AND AXON OUTGROWTH Several studies on axon outgrowth specificity have examined not just cell surface molecules for clues to molecular mechanisms but also nuclear and other proteins. From studies in Drosophila, a subset of segmentation and homeotic genes has been found to act in the developing nervous system to control the production and morphology of neural cells (for review, see Akam, '87). The early pattern specification genes number near 100 and were identified on the basis of ectodermal development in the embryo and larva (see, e.g., Mayer and NussleinVollhard, '88). A large number of them appear to be proteins that regulate gene transcription, containing homeobox, zinc finger, and other sequence motifs associated with such proteins, but many of them produce different types of proteins, such as integral membrane proteins, and could conceivably be directly involved in axon outgrowth. Molecular cloning of these genes has led to production of specific antisera and antibodies against a number of the gene products. In many cases, in addition to expression in different regions of ectoderm, these proteins are expressed in subsets of cells in the developing nervous system and help to specify the position and lineage of CNS and PNS cells. Further investigation of these genes has provided evidence that some of them have a more direct role in regulation of axon outgrowth. Not only are they

expressed in neural cells but they have functions that appear to be independent of their roles in ectoderma1 pattern formation. In particular, investigators have shown that mutations in some of these genes cause alterations in where specific neurons send their GCs during outgrowth in the CNS. For example, the segmentation gene fushi taruzu (ftz) functions to determine pathfinding by specific neurons in the Drosophila embryo (Doe et al., '87). Doe et al. ('87) found that the Rz protein is expressed in a subset of ganglion mother cells, which are derived from eight or nine neuroblasts in each thoracic hemiganglion. In some cases, the expression of ftz protein begins the same in sibling ganglion mother cell progeny but differs after a few hours. This parallels what we know about the timing of neural determination for some sibling neurons. The ftz gene function has a direct effect on pathfinding by certain neurons, as shown by the CNS RP2 neurons, which normally express the ftz protein. In mutants lacking ftz expression in the CNS, the RP2 neuron shows abnormal pathfinding, and the effect is quite specific, since there are six other ftzexpressing neurons that show no apparent pathfinding alterations in the mutant. The mechanism of this pathfinding alteration is not completely clear, since it is theoretically possible that the change in pathfinding is due t o a lack of ftz protein expression in cells that furnish cues for the RP2 GC as well as a change in ftz expression in the RP2 neuron. Since the ftz protein appears to function in transcriptional control, the effect of the ftz mutations in the nervous system is presumably due to a change in expression of ftz-dependent genes that function in pathfinding. For example, the ftz protein might control the expression of cell surface genes that normally function in guidance of the RP2 GC. A similar investigation of the evenskipped (eve) segmentation gene was also reported by Doe et al. ('881, with corollary results. A third example of a segmentation gene that affects pathfinding is the Drosophila orthodenticle (otd) gene (Finkelstein e t al., '90). The predicted otd gene product is a homeobox-containingprotein, similar t o ftz and eve proteins, and is expressed in medial cells of the developing CNS. Null mutations in the otd gene disrupt axonal morphology in the developing CNS, resulting in abnormal fusion of commissural bundles and narrowing of the space between the longitudinal connectives. A number of identified midline cells in the CNS are missing in otd embryos, due to either cell death or alterations in their fate, causing them to have abnormal morphology. The extent to which there may be

INSECT NEURAL DEVELOPMENT

alterations in axon outgrowth due to otd remains a promising but open question. There are other segmentation genes that affect CNS pathfinding but that appear to act by a mechanism different from that directly affecting gene transcription. The patched gene product in Drosophila is a cell surhce protein with seven putative transmembrane loops and a short segment with sequence similarity t o growth hormone (Pate1 et al., '89b). The patched gene product is expressed in a subset of CNS cells and mutations in the patched gene cause defects in axon outgrowth in the embryo. Another group of pattern regulation genes is known to function in neural development, but they have their initial function not in anterior-posterior segmentation but instead in dorsal-ventral ectoderma1 specification. These genes regulate pattern for cells located in the ventrolateral domain, within the area defined by the zerknult gene (Mayer and Nesslein-Vollhard, '88).For example, among these are genes that control the development of the ventral midline cells in the embryo. The foremost of these is the Drosophila gene single minded ( s i n ) , which controls midline development (Crews et al., '88; Nambe et al., '90; Thomas et al., '88). Fly sim mutants are recessive late embryonic lethals. They are cell autonomous (i.e., cause mutant phenotypes only in the cells carrying the defective gene copy) and affect cells on the ventral midline, causing a loss of neuronal and nonneuronal cells lying along the ventral midline neuroepithelium (see Crews et al., this issue). The result of this cell loss is fusion of the longitudinal connectives into a single bundle and loss of all CNS commissures. The sim protein is first expressed at the cellular blastoderm stage just prior to gastrulation in a strip two cells wide running the length of the neurogenic region on each side at the border of the presumptive mesoderm (Crews et al., '88; Nambu et al., '90). These cells come to lie at the ventral midline after gastrulation, and they include the precursors of the MP neurons, the median neuroblast, the ventral unpaired median neurons, and midline ectodermal cells that give rise to glia. The sim protein is expressed in all these medial cells that delaminate from the ventral ectoderm.An antibody against the sim protein shows that it is localized in the nucleus. The sim protein sequence suggests that it is a member of the helix-loop-helixfamily of gene transcription proteins. Genetic studies show that alterations in sim function lead to changes in the expression of a number of genes in the midline cells. Another gene expressed in the ventral midline cells of the Drosophila embryo is called slit (Fbthberg

317

et al., '88). The slit gene product is initially expressed in the ectoderm at the cellular blastoderm stage in the presumptive ectoderm. After gastrulation, the slit protein is expressed in lateral ectoderm but not in the neurogenic region. Subsequently expression becomes quite strong in the ventral midline of the neuroepithelium. The protein is specifically expressed in the subset of glial cells associated with the longitudinal axon bundles and also at a lower level on the surfaces of some axons. The slit protein is similar to epidermal growth factor in sequence, with six adjacent and one noncontiguous epidermal growth factor-likerepeats. The phenotype of slit mutants is similar to that of sim: the commissures are not evident, and the midline is collapsed so that the longitudinal tracts are apparently fused together. A slightly different group of genes is expressed in the ectoderm between the areas of sim and Zen expression. These include Star, spitz, pointed, and rhomboid. Both the CNS and the PNS are derived from this region of the blastoderm. Mutations in any of these genes cause deletion of ventral cuticule elements and affect the morphology of the nervous system. For example, the rhomboid gene affects the development of the ventral mesectoderm. The rhomboid gene produces a transcript of about 2.5 kb, which codes for a protein with a predicted molecular weight near 40 kDa (Bier et al., '90). Mutants in rhomboid, as in sim and slit, lead to narrowing or collapse of the commissures and various degrees of fusion of the longitudinal axon bundles (Mayer and Nusslein-Vollhard, '88). The predicted amino acid sequence is like an integral membrane protein, but with no similarity to other known proteins. Rhomboid mRNA is expressed initially in a ventral-lateral strip of cells in the Drosophila embryo cellular blastoderm, overlapping with the expression of the sim protein in these cells and their progeny. Subsequent rhomboid mRNA expression is quite complex, including transient expression in peripheral cells, which are probably precursors to a subset of chordotonal organs (Bier et al., '90).As was indicated previously, rhomboid mutations delete this subset of chordotonal organs. Another locus that appears to disrupt development is thepolyhomeotic gene in Drosophila, which has a null phenotype where developing flies lack the normal CNS scaffolding (Smouse et al., '88).In each hemisegment, instead of a regular ladder-like pattern, there is a disorganized whorled mass of axons, isolating each segmental ganglia from one another. Since the polyhomeotic gene product regulates the expression of a number of other seg-

318

A.L. HARRELSON

mentation and homeotic genes, it is likely that p l y homeotic mutations disrupt axon pathfinding by altering the determination of normal CNS cell identities; presumably such identities control the expression of many genes and the outgrowth specificity of each neuron. The idea that segmentation, homeotic, and other ectodermal pattern-formation genes may function in nervous system development can be generalized beyond the insects. Pate1 et al. (’89a)have studied the expression of the engruiled segmentation gene in the CNS of a variety of arthropods and found a very similar pattern. In each case, engrailed protein is found in a defined subset of neuroblasts, ganglion mother cells, and neurons. To summarize, there is a rapidly increasing roster of identified proteins, glycolipids, and genes that are candidate regulators of axon outgrowth specificity in the insect CNS. We have experimental evidence for the function of a few of these molecules and genes, although even in these cases the data are complex and the functions depend on interactions with other genes. Nevertheless, these ongoing studies have generated much momentum, and new technologies such as P element-mediated mutagenesis screens with enhancer trap vectors promise to add much more data to this effort (see, e.g., Bier et al., ’90).Enhancer trap genetic screens can identify genes on the basis of their expression pattern, allowing one to “tag” loci on the basis of activation in different groups of developing neurons, possibly a very useful indicator for molecules involved in axonal pathfinding. It is likely, however, that the complexity of CNS development and the interplay between redundant molecular expression patterns and interdependent biochemical actions will make decisive functional tests very difficult at the molecular level.

CONCLUSIONS In light of current progress in this area, we can reach some qualified conclusions about specific axon outgrowth in the insects. One is that molecular mechanisms of axonal pathfinding are probably fairly different between the periphery and the CNS. This is suggested by the general capability of peripheral sensory axons to navigate independently of one another and the specific axon-axon interactions required for correct CNS pathfinding. Cellular ablation experiments support this idea but must be viewed with some caution-it is rarely possible to isolate CNS neurons from one another and to test their outgrowth in the same way as occurs in the periphery. Similarly, efforts to isolate guidance mol-

ecules have produced many more candidates for outgrowth specificity in the CNS than in the periphery, suggesting great divergence at the molecular level. Some caution is in order here, also, since this apparent molecular difference may be due in part to one’s viewpoint and not to radically different pathfinding mechanisms. That is, in the CNS, pathfinding molecules have been identified primarily on the basis of their expression on different subsets of cells; they have been tested mainly for their function in allowing an axon from one cell t o fasciculate with that of another. However, the problem of general directionality of axon growth (e.g., posterior growth vs. anterior extension) has no molecular correlate in the CNS, and it is this latter problem that is most relevant to the pathfinding behavior of PNS neurons. Thus there may be some molecular features, yet to be discovered, that are shared. Uncovering such functions and discovering new genes using techniques based on anatomical methods are more difficult than may originally have been thought. Expression patterns of many antigens in the developing insect nervous system may be somewhat misleading when extrapolated to encompass the antigens’functions. It is evident that there may be only a loose correlation between the anatomical expression pattern of a protein and its function in neural development. This is most clear from studies on the fasciclins. Fasl and Fas3 have extremely complex patterns of cellular synthesis, but, when their genes are mutated or removed, many of the neurons that normally express these proteins still manage to attain connectivity sufficiently normal for the embryo to survive and reproduce. Therefore, anatomical descriptions of the fasciclins and other proteins may be somewhat misleading in regard to any essential functions they may have. Similarly, expression patterns of functionally unknown antigens may not provide particularly useful clues to their specific in vivo functions. Another general conclusion we can reach is that much redundancy exists in developmental mechanisms at the molecular level. In several cases where candidate molecules have been identified, and demonstrated by cell aggregation assays to cause cell adhesion (e.g., Fasl; Elkins et al., 1990a, 1990b), their absence does not disrupt normal neural development to a degree detectable in the embryo. This has led to the supposition that many cell adhesion molecules are redundant, so that the loss of a single participant in the adhesion/recognition process is not enough to block normal cellular events from occurring. Needless to say, this result is somewhat troubling; it leads to an expanding roster of

INSECT NEURAL DEVELOPMENT

structurally unrelated genes, each conserved for many millions of years in evolution, that have greatly overlapping functions. This also poses some difficulty for the idea that axon outgrowth specificity is governed by differential adhesiveness, the notion that GCs migrate onto the most adhesive surface they can contact. In a system where individual cells may possess half a dozen different cell adhesion molecules, what allows a cell to determine which surface is the most adhesive, and what difference is necessary for such a determination? The answer to this question may be quite illuminating for this general area of neural development. With luck, more detailed investigations in different species will uncover subtle effects of purportedly redundant genes, which may be tolerable under conditions of laboratory culture (e.g., ofD. melunoguster) but would surely be eliminated by selection in the wild. A final conclusion evident from reviewing the literature is that we face several important questions for which obtaining answers will likely require experiments performed in real time, on embryo and cell cultures. In particular, given the multitude of interacting or independent proteins and the diversity of cell types in the developing embryo, we have great difficulty in deciding which molecular cues are decisive and which are simply redundant or permissive. For example, there is a wealth of data on the morphology of developing sensory neurons in grasshopper limb, but most has been obtained by observation of fixed preparations at the conclusion of experimental manipulations. The same is true of studies on axon outgrowth in the CNS. It is thus very difficult to seek molecular correlates t o GC pathfinding in these preparations and to make detailed hypotheses when we cannot observe the pathfinding process as it occurs, on a moment-tomoment basis. If the fasciclins and other cell surface molecules function in precise combinations on individual cells, as is likely, then we must investigate the dynamic aspects of their expression and function to understand how they work. Similarly, studies of genetic mutants in fasciclins and other genes have been illuminating, but in only a limited way, indicating some general functions and interactions with other genes. Observation of mutant phenotypes in living embryo CNS and PNS, as they develop, would provide some of the precision and specificity necessary for forming precise, testable hypotheses about molecular mechanisms.

LITERATURE CITED Akam, M. (1987) The molecular basis for metameric pattern in the Drosophila embryo. Development, 101:1-22.

319

Anderson, H.J., J.S. Edwards, and J . Palka (1980) Developmental neurobiology of invertebrates. Annu. Rev. Neurosci., 3 ~97-139. Anderson, H., and R.P. Tucker (1988)Pioneer neurones use basal lamina as a substratum for outgrowth in the embryonic grasshopper limb. Development, 104~601-608. Anderson, H., and R.P. Tucker (1989) Spatial and temporal variation in the structure of the basal lamina in embryonic grasshopper limbs during pioneer neurOne outgrowth. Development, 106:185-194. Barthalay, Y., R. Hipeau-Jacquotte,S. De La Escalera, F. Jimenez, and M. Piovant (1990) Drosophila neurotactin mediates heterophilic cell adhesion. EMBO J.,9~3603-3609. Bastiani,M.J., C.Q. Doe, S.L. Helfand, andC.S. Goodman(1985) Neuronal specificity and growth cone guidance in grasshopper and Drosophila embryos. Trends Neurosci., 8:257-266. Bastiani, M.J., A. Harrelson, P.M. Snow, and C.S. Goodman (1987) Expression of fasciclin I and I1 glycoproteins on subsets of axon pathways during neuronal development in the grasshopper. Cell, 48~745-755. Bate, C.M. (1978) Development of sensory systems in arthropods. In: Handbook of Sensory Physiology, Vol. IX. M. Jacobson, ed. Springer-Verlag, Berlin; pp. 1-53. Bentley, D., P.B. Guthrie, and S.B. Kater (1991) Calcium ion distribution in nascent pioneer axons and coupled preaxonogenesis neurons in situ. J. Neurosci., 11~1300-1308. Berlot, J., and C.S. Goodman (1984) Guidance of peripheral pioneer neurons in the grasshopper: An adhesive hierarchy of epithelial and neuronal surfaces. Science, 223:293-295. Bieber, A.J., P.M. Snow, M. Hortsch, N.H. Patel, J.R. Jacobs, Z. Traquina, R. Schilling, and C.S. Goodman (1989) Drosophila neuroglian: A member of the immunoglobulin gene superfamily with extensive homology of the vertebrate neural adhesion molecule L1. Cell, 59:447-460. Bier, E., L.Y. Jan, and Y.N. Jan (1990)Rhomboid,a gene required for dorsoventral axis establishment and peripheral nervous system development in Drosophila melanogaster. Genes Dev., 4:190-203. Carpenter, E.M., and M.J. Bastiani (1991)Developmental expression of REGA-1, a regionally expressed glial antigen in the central nervous system development in Drosophila melanogaster. Genes Dev., 4~190-203. Carpenter, E.M., and M.J. Bastiani (1991)Developmental expression of REGA-1, a regionally expressed glial antigen in the central nervous system of grasshopper embryos. J . Neurosci., 11:277-286. Caudy, M., and D. Bentley (1986) Pioneer growth cone steering along a series of neuronal and nonneuronal cues of different affinities. J . Neurosci. 6:1781-1795. Caudy, M., and D. Bentley (1987) Pioneer growth cone behavior a t a differentiating limb segment boundary in the grasshopper embryo. Dev. Biol., 119:454-465. Condic, M.L., and D. Bentley (1989a) Pioneer neuron pathfinding from normal and ectopic locations in vivo after removal of the basal lamina. Neuron, 3:427-439. Condic, M.L., and D. Bentley (198913)Removal of basal lamina in vivo reveals growth cone-basal lamina adhesive interactions and axonal tension in grasshopper embryos. J. Neurosci., 9~2678-2686. Condic, M.L., and D. Bentley (1989~) Pioneer growth cone adhesion in vivo to boundary cells and neurons after enzymatic removal of basal lamina in grasshopper embryos. J . Neurosci., 9~2687-2696. Crews, S.T., J.B. Thomas, and C.S. Goodman (1988)The Dro-

320

A.L. HARRELSON

sophila single-minded gene encodes a nuclear protein with Lefcort, F., and D. Bentley (1987) Pathfinding by pioneer neurons in isolated, opened and mesoderm-free limb buds of sequence similarity to the per gene product. Cell, 52:143-151. embryonic grasshoppers. Dev. Biol., 119:466-480. De La Escalera, S., E.-0. Bockamp, F. Moya, M. Piovant, and F. Jimenez (1990) Characterization and gene cloning of Lnenicka, G.A., and R.K. Murphey (1989) The refinement of invertebrate synapses during development. J . Neurobiol., neurotactin, a Drosophila transmembrane protein related to 20:339-355. cholinesterases. EMBO J., 9~3593-3601. Denburg, J.L., B.A. Norbeck, R.T. Caldwell, and J.A.M. Marner Mayer, U., and C. Nusslein-Vollhard (1988) A group of genes required for pattern formation in the ventral ectoderm of the (1989) Developmental stage-specific antigens in the nervous Drosophila embryo. Genes Dev., 2:1496-1511. system of the cockroach. Dev. Biol., 132:l-13. Dodd, J., and T.M. Jessell (1985)Lactoseries carbohydrates spec- Meyer, M.R., P. Brunner, and J.S. Edwards (1988) Developmental modulation of a glial cell-associated glycoprotein, 5B12, ify subsets of dorsal root ganglion neurons projecting to the in an insect, Acheta domesticus. Dev. Biol., 130:374-391. superficial dorsalhorn of rat spinal cord. J . Neurosci., Meyer, M.R., G.R. Reddy, and J.S. Edwards (1987) Immuno5:3278-3294. logical probes reveal spatial and developmental diversity in Dodd, J.,and T.M. Jessell (1986) Cell surface glyconjugates and insect neuroglia. J. Neurosci., 7:512-521. carbohydrate-binding proteins: possible recognition signals Montell, D.J., and C.S. Goodman (1988) Drosophila substrate in sensory neuron development. J. Exp. Biol., 124:225-238. adhesion molecule: sequence of laminin B1 chain reveals Doe, C.Q., Y. Hiromi, W.J. Gehring, and C.S. Goodman (1987) domains of homology with mouse. Cell, 53:463-473. Expression and function of the segmentation gene fushi tarazu Montell, D.J., and C.S. Goodman (1989) Drosophila laminin: during Drosophila neurogenesis. Science, 239:170-175. Sequence of B2 subunit and expression of all three subunits Doe, C.Q., D. Smouse, and C.S. Goodman (1988) Control ofneuduring embryogenesis. J. Cell Biol., 109:2441-2453. ronal fate by the Drosophila segmentation gene even-skipped. Nardi, J. (1983) Neuronal pathfinding in developing wings of Nature, 333:376-378. the moth Manduca sexta. Dev. Biol., 95163-174. Elkins, T., M. Hortsch, A.J. Bieber, P.M. Snow, and C.S. Goodman (1990a)Drosophila fasciclin I is a novel homophilic adhe- Olson, PF., L.I. Fessler, R.E. Nelson, R.E. Sterne, A.G. Campbell, and J.H. Fessler (1990) Glutactin, a novel Drosophila sion molecule that along with fasciclin I11 can mediate cell basement membrane-related glycoprotein with sequence simsorting. J. Cell Biol., 110:1825-1832. ilarity to serine esterases. EMBO J., 9:1219-1227. Elkins, T., K. Zinn, L. McAllister, EM. Hoffiann, and C.S. Goodman (1990b)Genetic analysis of a Drosophila neural cell adhe- Patel, N.H., E. Martin-Blanco, K.G. Coleman, S.J. Poole, M.C. Ellis, T.B. Kornberg, and C.S. Goodman (1989a) Expression sion molecule: Interaction of fasciclin I and Abelson tyrosine of engrailed proteins in arthropods, annelids, and chordates. kinase mutations. Cell, 60:565-575. Cell, 58:955-968. Fessler, L., K.G. Campbell, K.G. Duncan, and J.H. Fessler (1987) Drosophila laminin: characterization and localization. Patel, N.H., B. Schafer, C.S. Goodman, and R. Holmgren (1989b) The role of segment polarity genes during Drosophila neuroJ . Cell Biol., 105:2383-2391. genesis. Genes Dev., 3:890-904. Finkelstein, R., D. Smouse, T.M. Capaci, A.C. Spradling, and N. Perrimon (1990) The orthodenticle gene encodes a novel Patel, N.H., P.M. Snow, and C.S. Goodman (1987) Characterization and cloning of fasciclin 111: A glycoprotein expressed homeo domain protein involved in the developmentof the Droon a subset of neurons and axon pathways in Drosophila. Cell, sophila nervous system and ocellar visual structures. Genes 48:975-988. Dev., 4:1516-1527. Goodman, C.S., M. OShea, R. McCaman, and N.C. Spitzer (1979) Piovant, M., and P. Lena (1988)Membrane glycoproteins immunologically related to the human insulin receptor are associEmbryonic development of identified neurons: Temporal patated with presumptive neuronal territories and developing tern ofmorphological and biochemical differentiation. Science, neurones in Drosophila melanogaster. Development, 103: 204:1219-1222. 145-156. Goodman, C.S., J.A. Raper, S. Chang, and R. Ho (1982) Grasshopper growth cones: Divergent choices and labeled pathways. Regan, L., J. Dodd, S.H. Barondes, and T.M. Jessell (1986)Selective expression of endogenous lactose-binding lectins and Symp. SOC.Dev. Biol., 40:275-316. lactoseries glycoconjugates in subsets of rat sensory neurons. Harrelson, A.L., and C.S. Goodman (1988) Growth cone guidProc. Natl. Acad. Sci. USA, 83:2248-2252. ance in insects: Fasciclin I1 is a member of the immunoglobRothberg, J.M., D.A. Hartley, Z. Walther, and S. Artavanisulin gene superfamily. Science, 242:700-708. Tsakonas (1988) slit: An EGF-homologous locus of D. melaHortsch, M., A.J. Bieber, N.H. Patel, and C.S. Goodman (1990a) nogaster involved in the development of the embryonic central Differential splicing generates a nervous system-specificform nervous system. Cell, 55:1047-1059. of Drosophila neuroglian. Neuron, 4:697-709. Hortsch, M., N.H. Patel, A.J. Bieber, Z.R. Traquina, and C.S. Sanes, J.R., J.G. Hildebrand, and D.J. Prescott (1976) Differentiation of insect sensory neurons in the absence of their Goodman (1990b) Drosophila neurotactin, a surface glycopronormal synaptic targets. Dev. Biol., 52:121-127. tein with homology to serine esterases, is dynamically expressed during embryogenesis. Development, 110: 1327- 1340. Seeger, M.A., L. Haffley, and T.C. Kaufman (1988) Characterization of amalgam: a member of the immunoglobulin superHunkapiller, T., and L. Hood (1987) The growing immunoglobfamily from Drosophila. Cell, 55:589-600. ulin gene superfamily. Nature, 323:15-16. Hunkapiller, T.,and L. Hood (1989) Diversity of the immuno- Smouse, D., C.S. Goodman, A.P Mahowald, and N. Perrimon (1988)Polyhomeotic:A gene required for the embryonic develglobulin gene superfamily. Adv. Immunol., 44:l-63. opment of axon pathways in the central nervous system of Kater, S.K., M.P. Mattson, C. Cohan, and J . Connor (1988) CalDrosophila. Genes Dev., 2:830-842. cium regulation of the neuronal growth cone. Trends NeuroSnow, P.M., A.J. Bieber, and C.S. Goodman (1989) Fasciclin 111: sci., 11:317-323. A novel homophilic adhesion molecule in Drosophila. Cell, Krantz, D.E., and S.L. Zipursky (1990) Drosophila chaoptin, a 59:313-323. member of the leucine-rich repeat family, is a photoreceptor Snow, PM., K. Zinn, A.L. Harrelson, L. McAllister, J. Schilling, cell-specific adhesion molecule. EMBO J., 9:1969-1977.

INSECT NEURAL DEVELOPMENT M.J. Bastiani, G. Makk, and C.S. Goodman (1988) Characterization and cloning of fasciclin I and fasciclin I1 glycoproteins in the grasshopper. Proc. Natl. Acad. Sci., USA, 85: 5291-5295. Thomas, J.B., S.T. Crews, and C.S. Goodman (1988) Molecular genetics of the single-minded locus: A gene involved in the development of the Drosophila nervous system. Cell, 52: 133-141. Williams, A.F. (1987) A year in the life of the immunoglobulin superfamily. Immunol. Today, 8:298-303.

321

Williams, A.F., and A.N. Barclay (1988) The immunoglobulin superfamily-Domains for cell surface recognition. Annu. Rev. Immunol., 6:381-405. Zinn, K., L. McAllister, and C.S. Goodman (1988) Sequence and expression of fasciclin I in grasshopper and Drosophila. Cell, 53:577-587. Zipursky, S.L., T.R. Venkatesh, and S. Benzer (1985) From monoclonal antibody to gene for a neuron-specific glycoprotein in Drosophila. Proc. Natl. Acad. Sci, USA, 82:18551859.

No title

THE JOURNAL OF EXPERIMENTAL ZOOLOGY 261:310-321(1992) Molecular Mechanisms of Axon Guidance in the Developing Insect Nervous System ALLAN L. HARRELSO...
1MB Sizes 0 Downloads 0 Views