SEARCH

SEARCH BY CITATION

Keywords:

  • Drosophila larvae;
  • locomotion;
  • motor circuit;
  • neural development;
  • optogenetics

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

How are functional neural circuits formed during development? Despite recent advances in our understanding of the development of individual neurons, little is known about how complex circuits are assembled to generate specific behaviors. Here, we describe the ways in which Drosophila motor circuits serve as an excellent model system to tackle this problem. We first summarize what has been learned during the past decades on the connectivity and development of component neurons, in particular motor neurons and sensory feedback neurons. We then review recent progress in our understanding of the development of the circuits as well as studies that apply optogenetics and other innovative techniques to dissect the circuit diagram. New approaches using Drosophila as a model system are now making it possible to search for developmental rules that regulate the construction of neural circuits.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

Progress in the past decade or two has greatly advanced our knowledge of the development of individual neurons (Sanes et al. 2006). For example, we know a lot about how cell fate is determined by the action of transcription factors and how axon trajectory is determined by the activity of various guidance cues (Shirasaki & Pfaff 2006; Kolodkin & Tessier-Lavigne 2007). Despite this wealth of knowledge about single neurons, little is known about how neuronal circuits, composed of a population of neurons, develop and function to generate specific behavior. This is partly due to the complexity of the neural circuits and the need for technology that allows researchers to deal with this complexity. However, new techniques such as optogenetics, genetically coded calcium indicators and connectomes are now making it feasible to dissect the connectivity and dynamics of the circuits and their relevance to behavior (Lichtman et al. 2006; Deisseroth 2006; Peron & Svoboda 1992; Venken et al. 2008).

The Drosophila embryonic and larval nervous system has been an invaluable model system for studying the genetic program for neural development and function. A great advantage of this system is the accessibility of the component neurons, motor neurons and sensory neurons, which allows identification of the cells, axons and dendrites (Goodman & Doe 2011; Jan & Jan 2011). This accessibility, combined with the strong molecular genetics available in this species, has enabled researchers to unravel basic principles underlying neural development across species, including molecular mechanisms of neural specification, axon guidance, dendrite formation and synaptogenesis (Doe 1996; Jan & Jan 1993; Kolodkin & Tessier-Lavigne 2007). With recent advancement in tools for circuit analyses, the time has come to go one step further and elucidate the developmental rules that govern the construction of neural circuits. In this review, we focus on the sensory-motor circuits that regulate larval peristaltic locomotion, which are composed of motor neurons, interneurons and sensory feedback neurons. First we summarize what is known about the development of motor neurons and sensory neurons. We then review recent studies of the function and development of circuits, including those that have applied optogenetics to this system. There is a huge body of literature concerning the development of the Drosophila sensory-motor system, so we did not try to cover the entire literature. Rather, our intention was to provide a brief guide, which we hope will be helpful for students and researchers who are new to the field and are interested in circuit development.

Larval locomotion and its regulation by the nervous system

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

Drosophila larvae exhibit various patterns of motion including peristalsis, bending, turning and feeding (Green et al. 1993). Among these patterns, peristaltic crawling has been extensively examined (Berrigan & Pepin 1990; Cattaert & Birman 2010; Fox et al. 2005; Cheng et al. 2004; Nishimura et al. 2001; Scantlebury et al. 2012; Gomez-Marin & Louis 2012; Iyengar et al. 2007; Lahiri et al. 2005; Gershow et al. 2001; Keene & Sprecher 2006). Forward propagation of local contraction from posterior segments to anterior segments leads to forward peristaltic locomotion (Fig. 1A). The larva has 10 segments (three thoracic segments, T1–T3, and seven abdominal segments, A1–A7) and specialized structures at the anterior and posterior ends (Demerec 2011; Keshishian et al. 2012). The body-wall musculature in each abdominal segment is composed of approximately 60 muscles (approximately 30 in each hemi-segment) which are layered below the epithelium (Figs 1B–D, 2). External muscles are mostly oriented along the dorso-ventral (or “transverse”) axis, whereas internal muscles are oriented along the antero-posterior (or “longitudinal”) axis. Since the muscles contract along their longer axis, contraction of transverse and longitudinal muscles lead to circumferential and longitudinal contraction of the segment, respectively. During larval locomotion, the body fluids and the surrounding transverse and longitudinal muscles function as a hydrostatic skeleton: a decrease in volume in particular segments by muscle contraction increases the volume in other segments. The muscle contraction initiates in the rear-most segment and proceeds to more anterior segments, pushing the body forward. When the peristalsis reaches the head, the head is moved forward and then anchored on the crawling surface with the mouth hooks. Then, the next cycle of peristalsis initiates in the posterior end.

image

Figure 1. Larval locomotion and its neural regulation. (A) Schematics showing larval peristaltic locomotion. A larva moves forward by contracting the body wall muscles in successive segments from posterior to anterior. (B–D) Basic architecture of the Drosophila larval neuromuscular system. (B) Motor neurons (red circles), in each neuromere of the ventral nerve cord (VNC), innervate muscles in the corresponding segment in the body wall (red arrows). The wave of muscular contraction is generated by the wave of motor activity in the VNC. (C) Cross-sectional view of the body-wall muscles. Contraction of transverse muscles (green) leads to circumferential contraction of the body wall, whereas that of longitudinal muscles (other colors) leads to longitudinal contraction. (D) In many experiments, larvae are dissected along the dorsal midline and flatted to expose the muscles and the VNC. Note: dorsal is up and anterior is to the left in this and the following figures unless otherwise stated.

Download figure to PowerPoint

image

Figure 2. Neuromuscular connectivity and myotopic map. Domains of muscles (rectangles) and the innervating motor neurons are shown in matched colors. The boxes in the left show correspondence between the muscle domains (top) and the innervating motor nerves (bottom). Muscles are numbered according to Bate (1990). Dendrites of motor neurons that innervate different muscle domains arborize in distinct positions (shown in ovals) in the ventral nerve cord (VNC) forming the myotopic map. While muscles are arrayed in a segment of the body wall, the dendrite map is shifted anteriorly to be parasegmental in the VNC. Projection of a majority of intersegmental nerves (ISN, ISNb and ISNd) motor neurons also cross the segment boundary (SB). Ventro-lateral internal muscles 14 and 30 are beneath muscles 6, 7, 12 and 13 and are not addressed in this figure. Muscle 25 and 18 (grey) are innervated by the transverse nerve (TN) and the ISN, respectively. AC, anterior commissure; PC, posterior commissure. Modified from Landgraf et al. (2003a).

Download figure to PowerPoint

When larvae crawl on a flat surface (e.g., an agar plate), they move by performing a series of forward peristalsis alternating with brief pausing, head swinging and turning. The pattern of locomotion is rhythmic and highly stereotypic. During peristalsis, the muscle contraction seamlessly travels from the posterior to anterior segments in approximately 1 s. After completion of one cycle of peristalsis, another wave of muscular contraction immediately follows, generating rhythmic motion. For this behavior to occur, motor neurons in each segment have to be sequentially activated along the body axis in a highly coordinated manner (Fig. 1B). Electrophysiological recordings have indeed revealed rhythmic bursts of activity in motor neurons that occur concurrently with locomotive waves (Fox et al. 2005). In general, rhythmic movements in animals are controlled by interplay between neural networks called central pattern generators (CPGs), which provide the timing of motor discharge, and sensory feedback. This conceptual framework appears to apply to the neural circuits underlying the larval locomotion. Although the identity of CPGs responsible for larval locomotion is currently unknown, there is evidence that the patterned motor outputs can be generated by central circuits in the absence of sensory feedback (Fox et al. 2005; Hughes & Thomas 2001). However, as detailed below, when the function of feedback sensory neurons is compromised, the pattern of peristalsis is altered. Thus, the coordinated pattern of peristaltic locomotion is regulated by the action of neural circuits including motor neurons, sensory feedback neurons and interneurons in the CPGs.

Anatomy and development of motor neurons

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

Neuromuscular connectivity

The intricate pattern of motor output depends on the precise connectivity between motor neurons and muscles. As described above, there are 30 muscles in each abdominal hemi-segment of the body wall. These muscles are innervated in a highly stereotypic manner by approximately 40 motor neurons in the ventral nerve cord (VNC), which send their axons through one of the six branches of the peripheral nerves: intersegmental nerves (ISN, ISNb and ISNd), segmental nerves (SNa and SNc), and a transverse nerve (TN) (Fig. 2) (Keshishian et al. 2012; Landgraf & Thor 2003b). A majority of ISN, ISNb and ISNd motor neurons innervate muscles in the next posterior segment (thus the name “intersegmental”), whereas motor neurons of the segmental nerves (SNa and SNc) innervate muscles in the same segment. There is an anatomical correspondence between the choice of the peripheral branch and domains of the target muscles. ISN, ISNb and ISNd motor neurons innervate internal muscles in the dorsal (and dorso-lateral), ventro-lateral and ventral domains, respectively. SNa and SNc motor neurons innervate external muscles in the lateral and ventral domains, respectively (Fig. 2). Accordingly, the choice of the peripheral branch by motor neurons is crucial for matchmaking between motor neurons and muscles. Several transcription factors have been identified as regulating this process (reviewed in Landgraf & Thor 2003b). For example, the homeobox transcription factor Even-skipped (Eve) is necessary and sufficient for projection to the dorsal branch ISN (Landgraf et al. 2003a). Eve regulates the dorsal projection partly by suppressing the Unc-5 receptor (see below) (Labrador et al. 2008). Projection to ventral branches including ISNb and ISNd, is regulated by combined action of homobox proteins Nkx6 and Hb9 (Broihier & Skeath 1995). Even-skipped represses Nkx6 and Hb9, and vice versa, suggesting that mutual inhibition between these transcription factors determines the dorsal versus ventral projection of the motor neurons (Broihier & Skeath 1995). The ventrally projecting motor neurons are further specified by combined action of transcription factors including the LIM (lin-11/Isl-1/mec-3)-homeodomain-containing Islet and Lim3 and the POU (Pit-1/Oct-1/unc-86)-homeodomain containing Drifter, to make a choice between ISNb and ISNd (Thor et al. 2002; Certel & Thor 2001). Vertebrate orthologues of some of these transcription factors are also implicated in motor neuron specification, suggesting that the molecular mechanism of motor specification is evolutionarily conserved (Thor & Thomas 2004).

Several axon guidance molecules that directly regulate branch selection by motor neurons have been identified (reveiwed in Ruiz-Canada & Budnik 1997). Forward genetic screening conducted in the 1990s identified several mutants that show defects in motor axon projection, including beaten path and sidestep (Vactor et al. 2009; Sink et al. 1998). A recent study showed that Beaten path and Sidestep are a ligand-receptor pair for guiding axons to their targets (Siebert et al. 2002). Beaten path functions as a receptor expressed in motor neurons, whereas Sidestep functions as a ligand expressed on intermediate targets along the peripheral pathway (Siebert et al. 2002). Receptor tyrosine phosphatases DLAR and DPTP69D are required for ISNb and ISNd motor neurons to exit the common ISN pathway and extend towards the target muscles (Desai et al. 1950; Krueger et al. 2011). A heparan sulfate proteoglycans, Syndecan, has been identified as a ligand for DLAR (Fox & Zinn 2008; Johnson et al. 2010). Netrins and its receptors, Frazzled and Unc-5, also controls dorsal (ISN) versus ventral (ISNb) projection, by mediating attraction and repulsion to specific pathways (Labrador et al. 2008). Axon repulsion mediated by Semaphorin-1b (Sema-1b) is also required for the ISNb motor neurons to defasciculate from the ISN pathway and enter the target region (Winberg et al. 2011; Yu et al. 1998; Terman & Kolodkin 2007).

When motor neurons reach the target region by navigating through peripheral pathways, they finally find and synapse with their specific target muscle. Target recognition molecules, transmembrane or secreted proteins expressed on specific muscles, play critical roles in the choice of the partner muscle among a number of potential targets in the vicinity (reviewed in Nose 2010; Ruiz-Canada & Budnik 1997; Sanes & Yamagata 2006). Known target recognition molecules include attractive cues such as Fasciclin3 (Fas3) (Chiba et al. 2010), Connectin (Con) (Nose et al. 2010) and Capricious (Caps) (Shishido et al. 1996; Kurusu et al. 1996), as well as repulsive cues such as Wnt4 (Inaki et al. 2010), Toll (Rose et al. 2002; Inaki et al. 2011) and Sema-2a (previously termed SemaII) (Winberg et al. 2011). Motor neurons integrate information provided by multiple attractive and/or repulsive cues in determining which muscles to form synapse on (Winberg et al. 2011; Rose & Chiba 2000; Kurusu et al. 1996). When searching for the target muscles, presynaptic motor neurons extend numerous filopodia and contact a number of muscles in the target region including the non-target muscles. Postsynaptic muscles also extend numerous filopodia, called myopodia, which contact multiple motor neurons (Ritzenthaler et al. 2011; Kohsaka et al. 2009; Kohsaka & Nose 2001). Thus, matchmaking between motor neurons and target muscles appears to be a mutual recognition process in which both pre- and post-synaptic cells seek each other. A target recognition molecule, Caps, is localized at the tips of myopodia where initial contacts between motor neurons and muscles often occur (Kohsaka & Nose 2001). Thus, local and mutual interaction at the tips of myopodia might be crucial for target selection.

The basic pattern of neuromuscular connectivity is established by the end of embryogenesis and the pattern is maintained throughout larval life with minor changes in soma position, axon projection and dendrite morphology (Hoang & Chiba 2007; Kim et al. 1996). However, there is a dramatic increase in the size of the muscles during the larval period. To accommodate this change and maintain synaptic efficiency, the neuromuscular junctions expand, generating new branches and synaptic boutons (Schuster et al. 2000; Zito et al. 2007; Ruiz-Canada & Budnik 1997). This property makes this synapse an ideal model system to study the molecular mechanisms of synaptic growth and plasticity (reviewed in Ruiz-Canada & Budnik 1997; Collins & Diantonio 2004). Another important and well-characterized feature of the synapse is the homeostatic regulation that allows stable levels of synaptic activity (Davis 2011). The larval neuromuscular synaptic terminals are classified into larger type I endings, smaller type II endings and minor type III endings. Type I endings release glutamate, the main excitatory transmitter at this synapse and are further divided into type-I big (type-Ib) and type-I small (type-Is). Type-Ib motor neurons have bigger boutons, project to a single muscle, and are low-threshold, whereas type-Is motor neurons have smaller boutons, project to groups of muscles and are high-threshold (Choi et al. 2010; Schaefer et al. 2009). It has therefore been suggested that type-Ib and type-Is neurons are specialized for precise and powerful movements, respectively.

Dendrites and the myotopic map

Motor neurons receive information from the upstream central circuits connecting to their dendrites. Thus, the topology of motor neuron dendrites could be important in understanding the information flow in motor circuits. Comprehensive single cell analysis has been conducted to analyze the arrangement of motor neuron dendrites (Landgraf et al. 2011,1997). These studies revealed that dendrites of motor neurons form a map: dendrites of motor neurons innervating different muscle domains arborize in distinct regions in the neuropile along the anterior-posterior and medio-lateral axis (Fig. 2) (Landgraf et al. 1997; Kim et al. 1996; Mauss et al. 2005). Along the anterior-posterior axis, dendrites of motor neurons innervating the dorsal internal, dorso-lateral internal, ventral internal and external muscles occupy overlapping but distinct positions in the central nervous system (CNS) (Landgraf et al. 1997). A medio-lateral map was observed among internal muscles: dendrites of motor neurons targeted to ventral muscles extend more medially than those of more dorsally-targeted motor neurons (Kim et al. 1996; Mauss et al. 2005). Arrangement of the myotopic map likely reflects organization of the upstream neural circuits that control motor neurons.

What are the molecular and cellular mechanisms for drawing this map? Mauss et al. (2005) showed that the medio-lateral map is formed in part by the action of secreted factors expressed in the midline: the arrangement of dendrites is regulated by Netrin-Frazzled mediated attraction and Roundabout (Robo)-Slit mediated repulsion. Slit, a repellant secreted by the midline, regulates the positioning of dendrites at particular distances from the midline, and individual dendritic responses to Slit are mediated by cell-specific expression of Robo receptors. Similarly, Netrin, an attractant secreted by the midline, regulates the positioning of dendrites via the Frazzled receptor expressed on motor neurons. These findings suggest that the dendritic map can be generated in a genetically determined manner. There is evidence, however, that neural activity is also important for the growth of dendritic arbors (Tripodi et al. 2002). Blocking synaptic inputs to motor neurons induces overgrowth of the dendrites of motor neurons. Conversely, increasing the inputs induces a reduction in the arbors of the dendrites. Thus, the dendrites of motor neurons not only provide a spatially organized interface upon which upstream interneurons can connect, but also function as a variable device for adjusting motor output.

Anatomy and development of sensory feedback neurons

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

The roles of sensory neurons in larval locomotion

In many motor circuits, sensory feedback modulates the activity of the CPG to ensure that the final motor output meets the demand of the environment (Friesen & Cang 2006; Rossignol 1999). Recent behavioral and electrophysiological studies have shown that Drosophila larval locomotion is no exception to these rules (Caldwell et al. 2002; Hughes & Thomas 2001; Song et al. 2009; Cheng et al. 2004). In each abdominal hemi-segment, there are 43 sensory neurons, which are divided into three major types: external sensory (es) neurons, chordotonal (cho) neurons, and multidendritic (md) neurons (Fig. 3A) (Jan & Jan 2011). md neurons are further classified, based on the dendritic morphologies, into bipolar dendritic (bd) neurons and four classes of dendritic arborization (da) neurons (I–IV) (Grueber et al. 1983). The bd neurons have a simple linear morphology, whereas class I–IV da neurons show increasing complexity in their dendritic arbor. Among these, bd and class I da neurons, which have relatively simple dendritic arbors, have been shown to be particularly important for normal peristaltic locomotion. When the function of these neurons is transiently inhibited by temperature-sensitive Shibire (Shibirets) (Kitamoto 2009), larval locomotion dramatically slows down (Hughes & Thomas 2001; Song et al. 2009). The dendrites of bd neurons extend linearly across the width of each segment and the class I da neurons extend simple dendritic arbors along the longitudinal axis (Fig. 3A). Thus, these neurons are well suited to sense the contraction and relaxation of the body-wall segment during locomotion. It has thus been suggested that these neurons function as proprioceptors that feedback the information about the muscle contraction. Since the speed of peristalsis is greatly reduced in the absence of feedback, the feedback information appears to be essential for rapid propagation of the wave. TRPN1/NompC, a member of TRP channels expressed in these neurons has been implicated in the regulation of this sensory feedback (Cheng et al. 2004).

image

Figure 3. Anatomy and development of sensory projections. (A) Schematic diagram of the anatomy of sensory neurons. Cell bodies of the eight types of sensory neurons are shown in different colors and shapes. Dendritic morphology of two class I (da) neurons and a bd neuron (ddaD, ddaE) and two bipolar dendritic (bd) neurons (dbd and vbd) are also shown. (B) Central projection of sensory neurons and its regulation by guidance cues. (top) Each type of sensory neuron sends axons to distinct positions in the ventral nerve cord (VNC) Some of the external sensory (es) and class IV dendritic arborization (vbd) extend axon branches across the midline. Other sensory projections are ipsilateral. While a majority of sensory neurons extend their axons through a ventral nerve root, the dorsal bd (dbd) neurons send their axons through a dorsal nerve root (Schrader & Merritt 2010; Grueber et al. 2002). (bottom) The Slit-Robo system guides the sensory projection along the medio-lateral axis. While cho neurons that express both Robo and Robo3 terminate in a lateral position with a lower level of Slit, other sensory neurons (such as class I da and bd neurons) that express only Robo project to more medial positions with a higher level of Slit. Similarly, the Sema2a gradient regulates the dorso-ventral positioning. The ventral projection of class IV da neurons is dependent on the expression of PlexB in these neurons and the presence of a Sema2a gradient in the VNC. (C) Terminal formation by sensory neurons. (top) Position and morphology of terminal arbors of putative sensory feedback neurons (dbd and class I md) and class IV da neurons. (bottom) Regulation of terminal formation by Sema2b-PlexB mediated attraction. The terminal formation by cho neurons is dependent on PlexB in the neurons and sema2b on the target longitudinal tracts. Asterisks denote the midline. Grey ovals (in B) and lines (in C) indicate Fas2-positive bundles, generally used as a frame of reference of the VNC (CI, central intermediate; DM, dorsal median; VL, ventral lateral; Landgraf et al.1997 ,b). Modified from Zlatic et al. 2010; Grueber et al. 2002.

Download figure to PowerPoint

Temporal inhibition of the other sensory neurons does not show a dramatic effect on larval forward locomotion (Hughes & Thomas 2001). However, some sensory neurons may regulate larval locomotion in a subtler manner. The class IV md neurons function as polymodal nociceptors that are necessary for behavioral responses to noxious heat or noxious mechanical stimuli. The pickpocket (ppk) gene, which encodes a member of the Degenerin/Epithelial Sodium Channel (DEG/ENaC) family, is required for mechanical nociception but not thermal nociception in these neurons (Zhong et al. 1998). In the mutants of ppk, the larvae move faster than wild-type larvae, with fewer pauses and turn movements (Ainsley et al. 2003). Activation of the class IV da neurons induces a rolling behavior, which the larvae use to escape from the attack of parasitoid wasps (Hwang et al. 2007). Another type of sensory neuron, cho neurons, are stretch receptors and detect the vibration of the body segments (Wu et al. 2011). Mutants with no or defective cho neurons exhibit increased duration of turning and reduced duration of linear locomotion (Caldwell et al. 2002). In these mutants, the speed of locomotion is also reduced. We also observed that temporal inhibition of cho neurons with Shibirets results in a small but significant decrease in the speed of locomotion (A.F., H.K. and A.N., unpubl. data, 2011). Thus, class IV da neurons and cho neurons may regulate the motor circuits through their respective sensory roles.

Central projection

Whereas motor neurons extend their dendrites within the dorsal side of the neuropile, most of the sensory neuron projections terminate in the ventral side of the neuropile. Thus, there is clear segregation between sensory input and motor output within the CNS (Fig. 3B) (Landgraf et al. 1999). Each class of sensory neurons, representing different modalities, projects axons to distinct areas within the CNS to make connections to second-order neurons (Fig. 3B) (Merritt & Whitington 2009; Schrader & Merritt 2010; Zlatic et al. 2010; Grueber et al. 2002). The connectivity between sensory neurons and their target neurons in the CNS must be important in understanding how sensory input modulates motor activity. We therefore focus here on the central projection of md and cho neurons, which are implicated in motor control. Development of dendrites of sensory neurons has been recently reviewed (Jan & Jan 1993).

Fasciclin2 (Fas2)-positive axon fascicles have been used as a reference framework to map the terminals of sensory neurons (Fig. 3B) (Landgraf et al. 1999). All cho neurons project to a middle region of the neuropile and extend their terminals along the Fas2-positive longitudinal tract Cl (Zlatic et al. 2010; Wu et al. 2003). Class II–IV da neurons terminate in neighboring positions in the ventro-medial neuropile; class II, class III and class IV innervate the most lateral, an intermediate and a medial position, respectively (Grueber et al. 2002). In contrast to the ventral projection of the cho and class II–IV da neurons, the bd and class I da neurons, which convey major sensory feedback information, project to a dorso-medial region near the dendrites of motor neurons (Merritt & Whitington 2009; Schrader & Merritt 2010; Grueber et al. 2002). After reaching their respective terminal regions, the sensory neurons form arbors with characteristic morphology (Fig. 3C) (Merritt & Whitington 2009; Schrader & Merritt 2010; Grueber et al. 2002). An interesting correlation has been observed between the geometry of the dendritic arbor and the polarity of the terminals in two class I da neurons, ddaD and ddaE: ddaD, which have anteriorly oriented dendritic extensions, project axon terminals anteriorly, whereas ddaE, which have posteriorly oriented dendritic arbors, do so posteriorly (Grueber et al. 2002).

How do sensory neurons form class-specific terminals in discrete regions in the neuropile?

As is the case for dendritic formation in motor neurons, the organization of sensory terminals is regulated in part by the gradient of axon guidance cues laid down in the embryonic CNS. Different classes of sensory neurons express different combinations of Robo receptors and the differential expression, in particular that of Robo3, is critical for the medio-lateral positioning of the terminals of md and cho neurons (Fig. 3B bottom) (Zlatic et al. 2010; Grueber et al. 2002). Likewise, Sema-Plexin (Sema-Plex) signaling regulates the positioning along the dorso-ventral axis (Zlatic et al. 1999). A dorso-ventral gradient of secreted Sema-2a and heterogeneous distribution of membrane-bound Sema-1a provide repulsive cues for incoming class IV da neurons expressing the receptors PlexA and PlexB. Expression of PlexA and PlexB is necessary and sufficient for the sensory neurons to terminate in the ventral position that expresses a lower level of Semas. Sema-Plex signaling also regulates the final branching of sensory terminals, by mediating attraction rather than repulsion (Fig. 3C bottom) (Wu et al. 2003). Sema-2b expressed on the target longitudinal tract and the receptor PlexB expressed in the cho neurons are required for sensory neurons to form axon terminals along the correct tract. In sema-2b or PlexB mutants, larval response to vibration is also impaired (Wu et al. 2003).

Development of locomotor behavior

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

In many animal species, embryos move before the animals hatch or are born (Marder & Rehm 2008). Drosophila embryos also show peristaltic movements similar to those observed in the mature larvae while in the eggshell (Suster & Bate 2001; Pereanu et al. 2012; Crisp et al. 2007). The embryonic period of Drosophila lasts for about 21–22 h after egg laying at 25°C (Fig. 4). The first contractions of the body-wall musculature appear approximately 7 h before hatching. They are brief and weak, occur within a single hemi-segment, and are called “twitches”. The twitches become stronger and more frequent during the subsequent hours and begin to propagate along multiple neighboring segments. However, the waves are often unilateral and, if bilateral, are asymmetric. These initial movements are myogenic in origin and can occur in the absence of neural activity (Crisp et al. 2007). At approximately 3.5 h before hatching, a different pattern of motor activity emerges. Embryos exhibit bursts of uncoordinated motor activity, which are intermitted by relatively quiescent periods. These episodic activities are dependent on neural activity but do not require sensory input. They are thus generated by the developing central circuits (Crisp et al. 2007). Sensory inputs, however, do influence the frequency of the episodes (Crisp et al. 2007, 2008). Shortly after the onset of the episodic activities, partial peristalsis travels along several segments. Then, the first complete peristaltic wave emerges approximately 1 h after the first episodic bursts. The waves are initially uncoordinated and very slow but gradually become faster and reach a mature pattern several hours after hatching. These sequential changes in embryonic motions suggest that the circuits are wired as they produce the precocious movements.

image

Figure 4. Embryonic development of locomotor behavior. Summary of sequential events during the development of locomotor behavior in embryos. See text for details. Modified from Crisp et al. (2007, 2008).

Download figure to PowerPoint

Is the activity produced by the developing circuits necessary for their proper wiring? Evidence suggests that precocious activity is important for proper development of motor output. When synaptic transmission is reversibly blocked in all neurons during the period of episodic activity, the emergence of mature peristalsis is delayed (Crisp et al. 2008). Artificial and continuous activation of all neurons using channelrhodopsin (ChR2) also causes a delay in the appearance of the coordinated waves, suggesting that patterned neural activation is important. Blocking neural activity in a later period did not have such influence. These observations suggest the presence of “sensitive period” for the activity-dependent modulation of the network. Interestingly, this period coincides with the period during which synapses mature in the circuits. As described above, dendrites of motor neurons grow and elaborate a number of branches during this period in a manner dependent on presynaptic activity (Tripodi et al. 2002). Action potentials of motor neurons are first produced in this period (Baines 2006). Sensory neurons also elaborate their terminal branches (Schrader & Merritt 2010). It is likely that activity-dependent maturation of synaptic connectivity within the circuits is critical for the maturation of the network.

Optogenetic dissection of motor circuits

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

Despite the wealth of knowledge obtained on the development of motor and sensory components of the circuits, little is known about the identity and function of the interneuron populations involved in locomotion. Although it has been shown that motor neurons receive excitatory input(s) from cholinergic neurons and inhibitory input(s) form GABAergic and glutamatagic neurons (Rohrbough & Broadie 2009; Baines 2006), the identities of these upstream interneurons are unknown. This is partly due to the fact that it has been difficult to visualize and manipulate neural activity in the complex neuropile of the VNC. However, recent advances in optical techniques are changing this. Genetically coded calcium indicators, such as GCaMP (Nakai et al. 1997; Tian et al. 1999) and Cameleon (Miyawaki et al. 1995), have been successfully used to visualize the activity of neurons in other systems in Drosophila, such as the olfactory system (Wang et al. 1993; Venken et al. 2008). We applied this technique to the larval CNS and observed a wave of motor activity in the VNC of the third instar larvae, concomitant with the muscular wave of contraction (Fig. 5) (H.K. and A.N. unpubl. data, 2010). We have also used calcium imaging to visualize emerging motor activity in the developing circuits in the embryos (Y.I., H.K. and A.N., unpubl. data, 2011). Optogenetic tools such as ChR2 and halorhodopsin (NpHR) have also been successfully used in this system (Schroll et al. 2010; Hwang et al. 2007; Zhang et al. 2011; Crisp et al. 2007; Pulver et al. 2007; Inada et al. 2007; Venken et al. 2008). These light-driven probes provide superior temporal and spatial resolution compared with previously available tools such as Shibirets, making it possible to precisely control the activity of specific neurons in the functioning circuits. Taking advantage of these optogentics tools, Inada et al. (2007) inhibited the activity of motor neurons in a specific region at a specific time during peristaltic locomotion of dissected larvae and found that activation of motor neurons at the forefront of the wave is necessary for the wave to propagate (Fig. 6). This unexpected finding suggests that motor neurons are part of the CPGs that generate the wave of locomotor activity in the CNS.

image

Figure 5. Calcium imaging of the wave-like activity of motor neurons. (A) Diagram showing the experimental scheme. GCaMP was expressed in motor neurons and the fluorescence signal in multiple segments (enclosed by squares) was recorded. (B) Time course of the average fluorescence intensity in the color-matched square in (A). Note that the peak of fluorescence intensity travels from posterior to anterior like a wave.

Download figure to PowerPoint

image

Figure 6. Optogenetic dissection of motor circuits. Transient inhibition of motor neurons at the forefront of the wave was used to distinguish the following two alternative models of the motor circuits. (A) Activity propagation of the upstream central pattern generator (CPG) proceeds independently of the activity of motor neurons. If so, the activity should proceed, even if the activity of motor neurons is inhibited. (B) Activity propagation of the CPG depends on the activity of motor neurons. If so, the wave may be temporarily halted upon optical inhibition of motor neurons. Experimental results support the latter model, suggesting that motor neurons are part of the CPGs that generate the wave. Modified from Inada et al. (2007).

Download figure to PowerPoint

These new tools now allow investigators to identify and study the function of interneuron populations in this system. Using a collection of GAL4 lines, we expressed GCaMP in different subsets of interneurons and searched for those that show either wave-like or oscillating activity and thus could be involved in locomotion. We successfully identified several distinct interneuronal populations, including a class of pre-motor inhibitory neurons, which show activity patterns correlated with larval locomotion (H.K. and A.N. unpubl. data, 2010). Optogenetic manipulations of these interneurons are revealing their roles in locomotion (H.K. and A.N. unpubl. data, 2011).

Conclusions and future prospects

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

The study of the Drosophila motor system is at a turning point. Researches on motor and sensory neurons were the pioneering studies of neural development. However, most of these studies focused only on the function and connectivity of individual neurons. Now, with ongoing identification of the interneuronal population, systems-level analyses of the nervous system are becoming feasible. For example, one may perturb neural activity in a population of neurons in the circuits by optogenetics while recoding the activity of another population by calcium imaging or patch-clamp recordings (Rohrbough & Broadie 2009; Choi et al. 2010; Baines 2006; Pulver et al. 2007; Schaefer et al. 2009). Serial section electron microscopy is also being used to analyze the connectome of the larval VNC (Cardona et al. 2003; Chklovskii et al. 1995). Such systematic experimental analyses, combined with theoretical analyses, will elucidate the input-output relationship of the component neurons and allow comprehensive understanding of the structural and functional connectivity of the circuits. Being composed of several hundreds of identifiable neurons in each segment, Drosophila VNC provides an ideal “mesoscopic” model system, in which one can relate the knowledge about the microscopic phenomena (e.g. molecules, synapses or cells) to the knowledge about macroscopic phenomena (e.g. activity of the entire circuits or behavioral output). Once the basic architecture of the circuits is understood, one can ask the following questions about neural development. (i) How do individual neurons assemble to construct functional circuits? (ii) How do unique functions of the circuits emerge? (iii) How do hard-wired and activity-dependent processes contribute to the development of the circuits? (iv) What are the roles of sensory inputs in the construction of the central circuits? (v) How do the central synapses develop and how does synaptic development relate to the functional development of the circuits? We anticipate that ‘general principles’ on the development of functional circuits will be discovered by analyses of mesoscopic circuits in the coming years.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References

We thank members of Nose lab for discussion and Kasumi Shibahara for the help in illustration. This article is supported by a Grant-in-Aid for Scientific Research on Innovative Areas “Mesoscopic Neurocircuitry” (No. 22115002) and for Scientific Research on Innovative Areas “Comprehensive Brain Science Network” (No. 221S0003), of The Ministry of Education, Culture, Sports, Science, and Technology, Japan, and for Scientific Research (B) (No. 23300114) of Japan Society for the Promotion of Science (JSPS), to A.N. and Grant-in-Aid for Young Scientists (B) (No. 21700344) of Japan Society for the Promotion of Science (JSPS) to H.K.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Larval locomotion and its regulation by the nervous system
  5. Anatomy and development of motor neurons
  6. Anatomy and development of sensory feedback neurons
  7. Development of locomotor behavior
  8. Optogenetic dissection of motor circuits
  9. Conclusions and future prospects
  10. Acknowledgments
  11. References