ABSTRACT
In Drosophila, the ability to analyze the development of individually identified neurons with a variety of imaging and biophysical techniques can be complemented by sophisticated genetics and molecular biology. This powerful combination is allowing the development and function of single neurons and their synaptic connections to be unraveled at an unparalleled level of resolution. In this article, we focus on a single, identified motoneuron – RP3 – arguably the best understood neuron in the fruitfly. Many events in the life history of RP3 are well characterized, including cell migration, axon outgrowth and pathfinding within the central nervous system, pathfinding in the periphery to its appropriate muscle target domain, the specific recognition of its muscle targets, the events of synapse formation and maturation, and its mature function in the locomotion of the fly larva. Genetic analysis has revealed mutations in a number of different genes which affect specific aspects of RP3 development from axon outgrowth to synapse formation.
INTRODUCTION
Invertebrate neurobiologists have worked with identified neurons for decades. Indeed, one of the most compelling reasons for working with “simpler” nervous systems is the consistency with which individual neurons can be uniquely identified from animal to animal as distinct and stereotyped elements of relatively simple and well defined neural networks. Embryologists too have taken advantage of the notion of identified cells, having worked with specific progenitor cells for a century or more. Over the past decade, these two approaches, the embryological and the neurobiological, have come together, with neurons being identified early in their development in embryonic insect nervous systems and the origin of these same neurons being linked, by means of cell lineage analysis, to specific divisions of identified progenitor cells, called neuroblasts. In principle, the complete developmental pathway from neuroblast to neuron to synapse can be defined, and the mechanisms controlling these events analyzed by a wide range of surgical, biochemical and molecular genetic manipulations.
The developmental analysis of identified neurons began more than a decade ago in the large embryo of the grasshop per (e.g., Bate, I 976a,b; Goodman and Spitzer, I 979; Goodman and Bate, 1981; Bate and Grunewald, I 981; Goodman et al., 1982; Bentley and Keshishian, 1982; Raper et al., I 983; Bentley and Caudy, 1983; Raper et al., 1984; Ball et al., 1985), an excellent system for cellular approaches including cell lineage analysis, intracellular dye injections, and a variety of different experimental manipulations.
Fortunately, the common blue-print that underlies the diversity of insect nervous systems has meant that the homo logues of cells and cell-cell interactions identified first in the grasshopper embryo could, almost without exception, be found in the embryo of Drosophila (e.g., Thomas et al., I 984; Goodman et al., 1984; Jacobs and Goodman, 1989; reviewed by Goodman and Doe, 1993). Moreover, the constant burgeoning of technical advances – from mono clonal antibodies, to enhancer trap lines, to reporter gene constructs – has provided powerful cell-specific probes that have enhanced our ability to study identified neurons, their processes, and their interactions in the developing nervous system of Drosophila, It is also possible in Drosophila to use the larva to study the physiology of neuromuscular synapses (Jan and Jan, 1976a,b), and the embryo to study the physiology of these same developing synapses as they form and mature (Broadie and Bate, 1993a-c). Thus, a complete range of genetic, molecular, embryological and physiological techniques can be brought to bear on the con struction of a neural network, from migration and process outgrowth to synapse formation and maturation.
This analysis is furthest along for an identified motor neuron called RP3, although here too, some gaps still remain in our knowledge. In this article, we review what is known about the development of RP3 and its synapse with muscles 6 and 7 in Drosophila. The ability to study individual motor neurons and their muscle targets has greatly aided the analysis of the function of genes and molecules involved in specific aspects of neuromuscular development and function.
IDENTIFICATION OF RP MOTOR NEURONS
At present, nearly 20 of the approx. 200 embryonic neurons on each side of each CNS thoracic and abdominal neuromere can be readily recognized from embryo to embryo and uniquely identified based on their cell body location and axon trajectory (reviewed by Goodman and Doe, 1993). Most of these cells are early-born neurons with cell bodies near the inner (dorsal) surface of the developing CNS. Thus, the cell bodies are accessible in dissected preparations and bracketed by a scaffold of dorsal axon tracts that provide a grid of spatial reference points for uniquely identifying these cells.
From this population of identified cells, the RP neurons, a cluster of five prominent motor neurons (RP1-5), are amongst the best characterized (Thomas et al., I 984; Jacobs and Goodman, 1989; Sink and Whitington, 1991a,b; Halpern et al., 1991; Goodman and Doe, 1993; Van Vactor et al., 1993). Finding memorable names for identified neurons is not always easy, and in this instance the acronym RP, picked late one night over a decade ago by M. B. and C. S. G., stands for Raw Prawn, an Australian phrase aptly describing the delicate and innocent appearance of two cells – RPI and RP2 – as they were first discovered in the embryo of the moth Manduca sexta and then by homology in the grasshopper and fruitfly (Thomas et al., 1984). RP3, RP4, and RPS were subsequently identified in Drosophila (Sink and Whitington, 1991a).
The RP neurons form a cluster of five somata lying in or just dorsal to the plane of the ‘CNS ladder’ (Fig. IA), a region bounded by the anterior and posterior commissures and the lateral borders of the longitudinal connectives. At maturity, all of the RP neurons send their axons out of the intersegmental nerve (ISN) into the periphery to innervate identified muscle targets. However, whereas four of these neurons (RPI, RP3, RP4, and RPS) extend their axons across the midline of the CNS and then out the next-posterior con tralateral ISN, RP2 extends its axon out the next-anterior ipsilateral ISN. In the periphery, the axons of RPI, RP3, RP4, and RPS switch to the segmental nerve (SN) and then to branch b of the segmental nerve (SNb) to innervate ventral domain muscles, whereas RP2 continues to extend in the ISN and innervates a dorsal domain muscle (Fig. I).
Apparently, the RPs do not form a single lineage group, but rather are descended from different neuronal precursor cells, called neuroblasts (NBs). For example, RP2 is known to arise from NB 4-2 (Doe, 1992), whereas the other RPs appear to be generated from one or more distinct NBs (C. Q. Doe, personal communication; N. H. Patel, personal communication). The precise lineage of RP3 is not yet known, although this should now be possible due to recent advances in cell lineage techniques (Udolph et al., 1993). Although most of our knowledge about RP3 begins shortly after its birth at the onset of its cell-specific migration, it should be possible in the future to study the life history of this neuron from birth to mature function.
RP3 MIGRATION AND AXON GUIDANCE TOWARDS THE MIDLINE
Active cell migrations appear to play only a limited role in determining cell positions in the developing fly CNS (Goodman and Doe. 1993). Most neurons are born and differentiate without migrating and are only passively displaced internally during successive rounds of NB division. Nevertheless, some specific migrations are known to occur for both neurons and glia. Of particular interest here, RP3 and RPI migrate during early neurogenesis to take up their characteristic medial dorsal positions in the CNS (Patel et al., 1987; Jacobs and Goodman, 1989). RP3 is born in a lateral position, many cell diameters away from its final position near the ventral midline. It begins to migrate towards the midline at about 9 hours of development (stage 12) just before extending its first axonal growth cone (Patel et al., 1987; Jacobs and Goodman, 1989) and continues migration as axonogenesis proceeds (9.5 hours; stage 13). At present, nothing is known about the mechanisms con trolling this cell-specific migration.
The RP neurons begin to display distinct cell surface properties about the time RP3 migrates toward its medial position and commences axonogenesis. The motor neurons RPI, RP3, Rf’4, and RPS express the homophilic cell adhesion molecule (CAM) fasciclin III (Fas III; Patel et al., 1987; Snow et al., 1989) on their cell bodies and begin extending axons around 9 hours of development (Patel et al., I 987; Halpern et al., 1991; H. Sink, unpublished results). Several of the same neurons (e.g., RPI and RP3) also express the homophilic CAM Fasciclin I (Fas I) on both their cell bodies and axons (Zinn et al., I 988; McAllister et al., 1991). A different, but overlapping subset of RP neurons express the homophilic CAM connectin (Nose et al., I 992; Meadows et al., unpublished data) during the same devel opmental period. In particular, RP3 expresses Fas III, Fas I, and Connectin on its cell body and axon during its medial migration and the onset of axonogenesis. Mutations in any one of the genes that encode these proteins does not affect early neurogenesis and, in particular, RP3 migration and axon outgrowth are normal in single mutants (Elkins et al., I 990; A. Nose et al., unpublished results). These observations suggest that if these CAMs do play a significant role in directing cell-specific migrations and axon outgrowth in the early CNS, then it is their combinatorial and overlapping expression that controls these events. Indeed, a double mutation in the fasl gene and in the abelson (abl) gene, which encodes a cytoplasmic tyrosine kinase, blocks RPI and RP3 axon extension across the CNS midline (Elkins et al., 1990). In the absence of crossing the midline, in the fasl; abl double mutant, the RP I and RP3 axons instead extend on their own side of the CNS and in many but not all cases, project out the ipsilateral !SN. The extension of the RPI and RP3 growth cones across the CNS midline is also blocked in mutations in the commissureless gene (Seeger et al., 1993), as described below.
RP3 PATHFINDING AND TARGET RECOGNITION
RP3 develops a prominent contralateral axon, which projects across the CNS midline and then out the ISN to innervate its specific muscle targets. RP3 first generates an axon during its medial migration at 9.5 hours of development (stage 13) (Patel et al., 1987). The axon projects within the anterior commissure to join the contralateral ISN (stage 14) (Jacobs and Goodman, 1989).
Jacobs and Goodman (1989) described the pattern of RP3 axon outgrowth in the CNS from reconstructions of serial electron micrographs. Subsequently, several groups (Halpern et al., 1991 ; Sink and Whitington. 1991b; Broadie and Bate. 1993a) described RP3 axon pathfinding within the CNS and into the periphery to its target muscles using intracellular injection of the fluorescent dye. Lucifer Yellow (LY). and with antibodies specific for small subsets of motor axons (i.e. Fas HI expression on RP3). Sink and Whitington (1991b) divided the sequence of axonogcncsis for the RPI. RP3. RP4. and RP5 motor neurons into live phases. Thus, although axonogcncsis for all four neurons is very similar up to the lime they contact and make their synaptic connections on specific ventral muscles, the actual sequence shown in Fig. 2. and described below, is specifically for RP3 (11. Sink and P. M. Whilington. unpublished results). First, during its soma migration, the growth cone of RP3 extends medially along the axon of its contralateral homologue in the anterior commissure to contact and wrap around the contralateral RP3 cell body (stage 12; Fig. 2A.B). Second, the RP3 axon grows posteriorly in the dorsal contralateral longitudinal connective, fasciculating with other RP axons within the dorsal RP tract (stage 13; Fig. 2C). Third, the RP3 axon leaves the longitudinal tract, turns laterally, and enters the ISN via the anterior ISN nerve root previously pioneered by the aCC growth cone (Fig. 2D). Once outside the CNS. the RP3 axon crosses over to the SN al the exit junction (Fig. 2E). before contacting the surfaces of the ventral muscles (stage 14). Fourth, having reached muscles 15 and 16. RP3 and the other RP axons diverges from the main SN into the SNb (Johansen ct al.. 1989) to contact the internal face of muscle 28 (Fig. 2F). This location where RP3 and other RP axons leave the main SN to form the SNb has been termed a choice point to denote that several alternate paths are available and that different growth cones make divergent choices when confronted with this choice (Van Vactor et al., 1993). Lateral to muscle 28. RP3’s axon grows between the most internal and intermediate of the three muscle layers (Bate. 1990). advancing laterally across (he ventral muscles (stage 15). During this period. RP3 axonal processes ramify over a restricted ventral muscle field (see below), contacting both target and nontarget muscles. Fifth. RP3’s inappropriate processes are withdrawn to generate its mature projection pattern onto its target muscles (Fig. 2G). the two ventral muscles 6 and 7 (stage 16).
In addition to its primary axon. RP3 develops additional processes within the C’NS. As RP3\s motor axon is ramifying within the ventral musculature (stage 15), the soma extends a collateral process in the ipsilateral longitudinal connective (Sink and Whitington. 1991b). Dendritic branches develop within the dorsal neuropile contralateral to the soma (Fie. 2E-G).
TARGET ABLATIONS LEAD TO SPECIFIC DEFECTS IN RP3 PATHFINDING AND TARGETING
Several studies (Sink and Whilington. 1991c: Chiba et al., 1993: Keshishian ct al.. 1993) have examined RP3 axono-genesis in the absence of ils target muscles (6 and 7). RP3’s muscle targets have been deleted using genetic, experimental and laser ablation techniques. In the absence of large! muscles 6/7. RP3 still diverges from SNb into the corred ventral muscle field, arguing against the release of chemoattractants by these specific large! muscles (although such chemoattractants might be released by the more general ventral domain of muscles and thus by the remaining ventral muscles). Without its normal targets. RP3’s axon ramifies over the normal held of ventral muscles and, within these muscles, apparently makes contacts at random and often innervates one or more incorrect targets (Fig. 3B). In the absence of only one of its targets (6 or 7). RP3 still can identify the remaining target, although not as reliably as when both targets arc present. In the absence of adjacent non-targcl muscles, which RP3 normally encounters either en route to its target or following target contact. RP3 still contacts and recognizes its muscle 6/7 target.
These observations suggest that (1) RP3 leaves the SN to enter the ventral muscles using cues from sources other than its specific target muscles, and (2) that once in the ventral muscle region. RP3 receives particular cues from muscle 6/7 which enable it to identify them as it’s targets. Moreover, contact between RP3 and its correct targets appears to play a decisive role in signaling the retraction of inappropriate processes. Thus, muscles 6 and 7 must possess unique characteristics that identify them as RP3’s primary targets and these characteristics arc maintained even if the number of ventral muscles, and therefore their relative positions, arc altered.
CERTAIN MUTATIONS LEAD TO SPECIFIC DEFECTS IN RP3 PATHFINDING AND TARGETING
Seeger et al., (1993) conducted a near saturation screen for mutations that affect the pattern of commissural and longitudinal axons in the developing CNS. This screen identified several genes required for the construction of longitudinal connectives [e.g.. longitudinals lacking (Iola), longitudinals gone (logo)] and axon commissures [e.g.. commissureless (comm) and roundabout (robo)]. These mutations identify genes that are required for the early stages of neuronal pathlinding and navigation within the developing CNS. including the navigation and path choices made by the RP3 growth cone. In particular, mutations in comm block the early navigation of the RP3 axon towards and across the CNS midline in the anterior commissure. In comm mutants. RP3’s contralateral projection is blocked and instead an axon projects ipsilaterally. eventually leaving the CNS in the ipsilateral ISN. switching to the SNb. and innervating the ipsilateral homologue of its target muscles (6 and 7), a phenotype intriguingly similar to that of the fasl: ahi double mutant mentioned above (Elkins et al., 1990). Thus, comm is one clement in a genetic hierarchy that specifies progressive choice points in the navigation of the RP3 axon within the CNS.
Van Vactor et al., (1993) conducted a further genetic screen of the second chromosome (approx. 40% of the genome) in Drosophila for mutations affecting pathlinding outside the CNS and the development of neuromuscular connectivity (Fig. ID.E). This screen identified several genes required for motor neuron pathlinding [e.g., beaten path (beat), short slop (shot) and stranded (sand)] and muscle target recognition [ e.g.. clueless (clu) and walkabout (wako)].
beat. sand, and shot are genes required for motor neurons to correctly navigate through specific choice points in both the ISN and SNb (Fig. ID). In particular, the SNb defects observed in these mutants suggest that these three genes control discrete steps in RP motor neuron pathlinding. In beat mutants. RP axons (like RP3) fail to separate from the ISN and SNa to pioneer SNb. Thus, beat is required for the RPs to enter the ventral muscle domain. In shot mutants. RPs succeed in pioneering SNb and just enter the ventral muscle domain, but usually do not succeed in contacting distal target muscles. In sand mutants. RP axons gel a little further and usually manage to navigate past muscles 28 and 14 before terminating. RP3 often reaches its target muscles, but more dorsally projecting RPs fail to reach their targets. Thus. beat. shot, and sand appear to define a hierarchy of genes that is required for proper pathlinding of the RP axons to leave the SN. enter the SNb. and then properly enter and explore the ventral muscles (Fig. ID).
Mutations in two genes — wako and clu — prevent RP neurons from recognizing their correct target muscles. In these mutations, all the RP axons have access to and contact with their target muscles, but still fail to form appropriate neuromuscular synapses (Figs IE, 4. 5). Unlike the pathlinding mutants beat. shot, and sand, mutations in wako and clu (in the alleles wako1. wako-. and clu1) appear to affect specifically the axons in the SNb and a small portion of the motor axons in SNa. while other axon trajectories arc normal (Van Vactor et al., 1993). In wako and du. RP axons in SNb ramify widely over the ventral muscles, occasionally even crossing the segment boundary (Figs 4. 5). and appear, in the light microscope, to form neuromuscular contacts at random among these muscles. In particular, (he normal RP3 target synaptic site on muscles 6 and 7 receives little or no innervation in these mutants and. instead. RP3 synapses at unpredictable locations on one or more adjacent ventral muscles (Fig. 5). However, as in wild-type development. RP3’s axon is always restricted within a subset of the ventral muscles and will only form synapses with muscles within this restricted field. Thus, both wako and clu are genes that are required for the RP axons in SNb to recognize their correct target muscles once they have entered the ventral muscle domain.
The SNb is unique among the peripheral nerve branches because most of the motor axons in this nerve (i.e. RPI. RP3. RP4. and RP5). and their muscle targets (i.e. 28. 7. 6. 14. 13 and 12 -listed in order from the ventral midline), have been individually identified (Fig. 1A-C). This has allowed us to follow individual axon trajectories in various mutant backgrounds or experimental conditions. RP3. in particular, has been studied by intracellular LY injection in mutant (e.g. wako) genetic backgrounds (Fig. 5B). For example, in embryos mutant for wako. RP3 fails to recognize its target in the synaptic cleft between muscles 6 and 7. and instead continues to grow over and past its target and even cross the segment boundary (Van Vactor et al., 1993). This mutant phenotype is intriguingly reminiscent of RP.3 behavior when its muscle targets have been ablated d ig. 3; Sink and Whitington. 1991C). In both cases. RP3’s axon wanders over and arborizes, apparently at random, with neighboring muscles — but only in a highly restricted ventral muscle set. These observations have led to the idea of “muscle target domains” (Van Vactor et al., 199.3). defined as a group of neighboring muscles sharing common signals that attract appropriate cohorts of motor neuron growth cones. In the case of RP.3. the muscle domain includes the normal muscle targets of the SNb motor neurons as well as the more lateral muscles 5 and 8 (Fig. 3: Van Vactor et al., 1993). This concept suggests that (I) groups of motor neurons and their target muscles share a common identity, and (2) specific genes (c.g. wako and cht) operate in pathways that subdivide these common identity groups so that individuals can be distinguished and synap-lically coupled according to unique identities (c.g.. RP3 with muscles 6/7).
RP3 SYNAPSES ON MUSCLES 6 AND 7
The RP.3 axon projects in SNb to innervate a pair of large, ventral, internal-longitudinal muscles (numbers 6 and 7 (Fig. 6); Crossley. 1978; Bate. 1990]. The first indication of synaptogenesis is the transient expression of the homophilic CAM Fas III (Snow et al., 1989) al the synaptic site along the medial lateral border between the two muscles termed the synaptic cleft — starting about 12 hours of development (Halpern et al., 1991). Fas III is also strongly expressed on the RP.3 growth cone prior to and during the early stages of neuromuscular contact (Fig. 7B; Halpern et al., 1991). In the absence of RP.3. Fas III is still expressed on its target muscles at the correct synaptic site (Broadie and Bate. 1993c). These observations lead to the suggestion that differential adhesion mechanisms might be involved in either neuromuscular recognition or the recognition of specific synaptic sites on the muscle target.
However, arguing against these two hypotheses is the observation that spatially restricted Fas III expression is neither necessary nor sufficient to mediate synaptogenesis between RP.3 and muscles 6/7. Null mutations in fas Hl are viable (Elkins et al., 1990) and do not appear to perturb the development of neuromuscular specificity (Keshishian et al., 199.3). Moreover, mutations in wako or chi do not perturb normal Fas 111 expression and yet they block RP3’s recog-nition of its target (Van Vactor et al., 1993). Thus, Fas III may work in some combinatorial way to establish or strengthen early synapses, or alternatively, it may not play a significant role in these events but rather may control some as yet undefined functions. Other proteins, such as ‘loll (Nose et al., 1992). are known to be expressed al the synaptic cleft of muscles 6/7 with a pattern similar to Fas III and may play a role in similar mechanisms.
Several studies (Johansen ct al.. 1989a; Halpern et al., 1991; Sink and Whitington. 1991b; Broadie and Bale. 1993a) have described the morphological differentiation of the neuromuscular junction (NMJ) between RP3 and muscles 6/7 using intracellular LY injections in RP3 (Fig. 6) and immunohistological observations al both the light and scanning electron microscope levels (Fig. 7). NMJ morphogenesis can be divided into six stages. First, morphological differentiation of the NMJ begins with three prominent RP3 growth cone processes in the synaptic cleft, two anterior and 1 posterior to the anterior lateral axon entry point (Fig. 7A; 12.5-13 hours). Second, during these initial stages of RP3 contact with its target. RP3 maintains exuberant processes over a variable range of muscles in the ventral muscle domain, often even extending beyond its target to contact muscles 30. 12. and 13. These processes arc distinct from growth cone filopodia, in that the processes arc both thicker and longer than filopodia, and perhaps should be termed ‘axon branches’ (Fig. 6A.B). Third, in early stage 16(13-14 hours), RP3 relines its terminal arbor in the synaptic cleft between muscles 6 and 7. Inappropriate axonal branches to non-target muscles are retracted and the anterior process within the cleft is also retracted, leaving the mature projection of RP3 processes restricted to the synaptic cleft posterior to the axon entry point (Fig. 6C,D). Fourth. RP3 develops presynaptic specializations in the form of varicosities or boutons (presynaptic transmitter release sites) al 14-15 hours (Figs 61). 7C). Fifth, a second motor neuron synapses al RP3’s pre-established synaptic site (15-16 hours; Fig. 7B). Sixth, during stage 17. the NM.I develops its mature morphology (Figs 61). 7C).
Broadie and Bate (1993a.b) described the accompanying physiological development of the NM.I between RP3 and muscle 6 with whole-cell patch-clamp techniques and divided synaptic differentiation into several progressive steps. First, during the initial stages of RP3/muscle 6 contact (12.5-13 hours), muscle 6 is electrically and dye-coupled to several adjacent muscles in the ventral muscle domain. Second, muscle uncoupling (13 hours) heralds the onset of a rapid period of differentiation: the muscle membrane rapidly develops its mature electrical properties, the contractile apparatus becomes functional, ami physiological receptors for the excitatory neurotransmitter, L-glutamate, are expressed in the muscle membrane. Third, glutamate receptors (gluRs) are initially homogeneously expressed in the muscle membrane (13–14 hours), but arc localized to the synaptic cleft soon after the refining of the presynaptic arbor (14–16 hours AEL). Fourth, the NMJ becomes functional and RP3 activity begins to drive contractions in its target muscles (14–15 hours). Coordinated peristaltic muscle movements soon develop. Fifth, during stage 17 (16 hours onwards), the density of gluRs in the postsynaptic receptor field is dramatically increased. Sixth, by the end of embryogenesis. RP3 generates rhythmic bursts of synaptic excitatory currents that drive the periodic contraction of muscles 6 and 7 during locomotory movement (Fig. 8).
RP3 plays no discernible role in the differentiation of muscle properties in its target muscles (Broadie and Bate. 1993d): in the absence of RP3. muscle 6 develops normal muscle contractile and electrical properties. In contrast. RP3 does induce postsynaptic specializations in muscle 6 (Broadie and Bate. 1993c). In the absence of RP3. muscle 6 expresses functional gluRs but fails to localize these receptors to the synaptic domain. Furthermore, the upregulation of functional gluR expression associated with late synaptic development also fails to occur in the absence of RP3. Thus. RP3 provides signals to its muscle target that induce the construction of the postsynaptic receptor field.
POSTEMBRYONIC DEVELOPMENT AND FUNCTION OF RP3
At hatching, the morphology of the NMJ between RP3 and its targets, like other NMJs, is relatively rudimentary (Fig. 7C). During postembryonic development, the NMJ is refined with higher order synaptic branching and the elaboration of presynaptic boutons (Fig. 7D: Keshishian et al., 1993: Atwood et al., 1993; Jia et al., 1993). In addition, as the lana grows rapidly during its four day life, the synapse must also grow accordingly to accommodate the enlarging muscle. Thus, though the basic neuromuscular connectivity is established during embryogenesis, synaptic development continues well into postembryonic life. In the embryo, neural pathfinding and morphological synaptogenesis proceed through largely activity-independent mechanisms, such as selective adhesion. For example. RP3 develops a normal terminal arbor on muscles 6 and 7 in the complete absence of neural electrical aclisity (Keshishian et al., 1993; K. Broadie and M. Bale, unpublished observations). However, postembryonic synaptic development proceeds through distinct activity-dependent mechanisms. Larval NMJs. including that of RP3 on muscles 6 and 7. show usedependent morphological plasticity (Budnik et al., 1990; Jia et al., 1993). such that increased electrical activity (e.g.. in the hy percxcilablc double mutant eag shaker) increases both the number of synaplic branches and boutons.
These NMJs show altered synaptic plasticity in memory mutants with a defective cyclic AMP (cAMP) cascade (Zhong et al., 1992). For example, dunce (due), which affects the cAMP-specific phosphodiesterase, increases the number of branches and boutons of RP3 on muscles 6/7. Likewise, rutabaga (rut), which affects adenylate cyclase, suppresses this Jnc-mediatcd effect. Thus, the cAMP messenger cascade affects morphological development. Similarly, the cAMP cascade plays a role in both synaptic facilitation and potentiation between RP3 and muscle 6 (Zhong and Wu. 1991) thus, both activity-dependent and cAMP-depcndent plasticity arc important elements of synaptic development as (hey pros ide mechanisms whereby the animal can selectively strengthen or weaken neuromuscular connections based simply on their use.
Broadie and Bate. 1993a. b; Van Vactor et al., 1993). Further studies ol neuromuscular development beyond stage 17 (including larval stages) are not covered in this timeline. (C) The range of time when defects arise in different mutants are indicated by black bars. The coinmissurele.w and l/abl mutant phenotypes are described by Seeger et al. (1993) and Elkins et al. (1990). respectively. I he other mutants arc described by Van Vactor et al. (1993). The approximate lime after egglaying is shown relative to embryonic stage as described by Campos-Ortega and Hartenstcin (1985). (Adapted from Broadie and Bale. 1993a. and Van Vactor et al., 1993).
Jan and Jan (1976a,b) first described the morphological and physiological characteristics of the mature larval NMJ between RP3 and muscles 6 and 7. In the larva, muscles 6 and 7 are among the largest longitudinal muscles and. working as a jointly innervated muscle unit, provide much of the force for the animal’s normal locomotory movements. RP3 is an excitatory motor neuron, which elicits muscle contraction via the excitatory transmitter L-glutamate (Figs 7E). 8; Jan and Jan. 1976b; Johansen et al., 1989b). The larva moves with a series of regular peristaltic muscle movements. Movement is controlled in each muscle with a brief burst of high frequency excitatory junctional currents (EJCs) at the NMJ (Fig. 8). Periodic bursts of neural action potentials can be recorded in the peripheral motor nerves (Budnik et al., 1990) and. in the case of the NMJ between RP3 and muscle 6. this activity triggers a similar pattern of periodic EJC bursts at the neuromuscular synapse.
CONCLUSION
In this article, we summarize the life history of a single identified motor neuron RP3 — and focus in particular on the events from its cell migration and axon outgrowth to the formation of its mature neuromuscular synapses (Fig. 9). The location of the RP3 cell body has made it possible to study its development throughout much of its embryonic and larval life. Likewise, the identity and development of RP3’s synaptic targets — muscles 6 and 7 — have been described throughout their development (Fig. 9A). Finally, the excitatory neuromuscular synapse of RP3 onto muscles 6 and 7 has been examined both morphologically and physiologically throughout its development and during its mature function in the locomotion of the fly larva. This detailed characterization of a single neuromuscular unit has proven invaluable in the analysis of the function of molecules and genes implicated in neuromuscular development and function Fig. 9C). In this review, we have briefly mentioned numerous genetic mutations, which have been isolated and/or analyzed using on gained from the resolution of working with individually identified neurons, muscles, and synapses. In the future, we hope to characterize these and other mutations and in so doing to gain an understanding of the molecular and genetic pathways which control the specificity, formation, maturation, and ongoing remodeling of identified synaptic connections in the fruitfly.
ACKNOWLEDGEMENTS
We thank Emma Rushton. Helen Skaer and Nathan Tublitz for critically reading earlier versions of this manuscript. The authors work was supported by an Oliver Gaily Studentship and AFCU scholarship to K. B.. the Hasselblad Foundation and Wellcome Trust to M. B.. an AC’S Postdoctoral Fellowship to D. V. V.. an H. H. M. I. Postdoctoral Fellowship to H. S.. and the Australian Research Council to P. M. W. C. S. G. is an Investigator with the I toward Hughes Medical Institute.