A Drosophila kinesin required for synaptic bouton formation ... - Nature

© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience

ARTICLES

A Drosophila kinesin required for synaptic bouton formation and synaptic vesicle transport Eunju Pack-Chung1,2, Peri T Kurshan1,2, Dion K Dickman1–3 & Thomas L Schwarz1,2 The morphological transition of growth cones to synaptic boutons characterizes synaptogenesis. Here we have isolated mutations in immaculate connections (imac; CG8566), a previously uncharacterized Drosophila gene encoding a member of the Kinesin-3 family. Whereas earlier studies in Drosophila implicated Kinesin-1 in transporting synaptic vesicle precursors, we find that Imac is essential for this transport. An unexpected feature of imac mutants is the failure of synaptic boutons to form. Motor neurons lacking imac properly target to muscles but remain within target fields as thin processes, a structure that is distinct from either growth cones or mature terminals. Few active zones form at these endings. We show that the arrest of synaptogenesis is not a secondary consequence of the absence of transmission. Our data thus indicate that Imac transports components required for synaptic maturation and provide insight into presynaptic maturation as a process that can be differentiated from axon outgrowth and targeting.

When a neuronal growth cone reaches its target field and contacts appropriate cells, the axon terminal undergoes synaptic morphogenesis. This transformation is a complex process by which the filopodial specializations of the growth cone and apparatus for axon extension and navigation are replaced by stable varicose nerve terminals specialized for releasing neurotransmitter1. At present, the molecular and cellular mechanisms that guide the transition of growth cones to mature synapses are not well defined. Growth cones and synapses are structurally distinct. Networks of actin cytoskeleton support the outer edge of the growth cone, and the extension of microtubules can direct axonal growth2. By contrast, mature synapses contain machinery for stable interactions with postsynaptic specializations, distinct cytoskeletal arrangements, and the apparatus for transmitter release and recycling. This apparatus includes synaptic vesicles and the active zone, an electron-dense membrane region that contains Ca2+ channels and protein complexes necessary for transmitter release. Thus, there is a distinction between the components of a growth cone and a mature synapse that parallels their distinct morphologies. Synaptic components are primarily synthesized in the neuronal cell body and therefore, to be available for forming synapses, need to be transported down the axon3. Synapse formation, however, can occur within 30 min of initial axodendritic contact4. To permit this rapid transition, synaptic materials are present in axons before the onset of synaptogenesis and are subsequently captured by incipient synaptic contacts1,5. Transported as aggregates of membrane-bound organelles, these vesicular structures, which are sometimes called ‘packets’, include synaptic vesicle proteins (synaptotagmin-I, synaptobrevin, SV2 and synapsin-1) and plasma membrane proteins (Ca2+ channels). These

cargos become immobilized at axo-dendritic contacts as part of a rapid process that leads to functional connectivity6,7. One transport organelle is known as a PTV (Piccolo-Bassoon transport vesicle) and is thought to provide many components of active zones1,8,9. Both the packets and the PTVs show bidirectional movement in axons6,8, but no motor has been shown to mediate this movement. These transport organelles occur in both peripheral and central synapses of vertebrates1,10, and analogous systems are probably needed at Drosophila neuromuscular junctions (NMJs). At this synapse, motor neuron growth cones transform to functional but immature NMJs within an hour of initial contact with target muscles11. Morphological maturation into synaptic boutons, however, requires 3–5 h. As in mammalian systems, synaptotagmin-I and other synaptic components are present in axons before they reach their targets. Bouton formation correlates with increases in synaptotagmin immunoreactivity, in clearand dense-cored vesicles, and in active zones12,13. Transport of these components during NMJ development is thus implied, although the motors that are required remain unknown. We now report the identification of a genetic locus encoding a kinesin that is required for presynaptic maturation. Mutations in the gene immaculate connections (imac; Flybase, CG8566) prevent nerve endings from transforming to synaptic boutons: growth cones become constricted but remain within contact fields as thin processes that lack varicosities or boutons. In addition to lacking synaptic and dense-core vesicle components, imac mutant nerve endings also contain very few active zones. Characterization of imac thus implicates a particular motor in the process of presynaptic maturation and provides insight into the relationships between motors and synapse formation and pre- and postsynaptic differentiation.

1Program in Neurobiology, Children’s Hospital, 300 Longwood Avenue, Boston, Massachusetts 02115, USA. 2Department of Neurobiology, Harvard Medical School, 220 Longwood Avenue, Boston, Massachusetts 02115, USA. 3Present address: Department of Biochemistry and Biophysics, University of California, San Francisco, 1550 4th St., San Francisco, California 94158, USA. Correspondence should be addressed to T.L.S. ([email protected]).

Received 2 April; accepted 7 June; published online 22 July 2007; doi:10.1038/nn1936

980

VOLUME 10

[

NUMBER 8

[

AUGUST 2007 NATURE NEUROSCIENCE

ARTICLES

a 4 mV 0.5 s

imac52

Control


b

c 12

12

13

13

6

6

7

7

d

+/+ at 14 h

f

imac170

e

imac102/102

+/+

g

h

imac116/116

i

12 12 13 13 6 7

6

+/+ at 21 h

imac170

+/+

7

j

k

l

imac 52

imac102

RESULTS imac mutants lack synaptic varicosities We performed a forward genetic screen to isolate genes affecting synapses. In this screen, the EGUF-hid method was used to produce homozygous mutant eyes in an otherwise heterozygous fly, and synaptic function was assessed by selecting for blindness and abnormal electroretinograms (ERGs)14,15. One complementation group consisted of eight independently derived alleles. Homozygous mutant eyes externally appeared normal and their ERGs showed robust responses to light (Fig. 1a); however, the ‘on-off transients’ of the ERG, which require synchronous synaptic transmission from the photoreceptors to second-order neurons, were absent in eyes that were homozygous for any of the mutant alleles (Fig. 1a; arrow). Homozygous flies of each allele died as unhatched late-stage embryos. The gross morphology of these embryos was comparable to that of a wild-type fly that has completed embryogenesis 20–22 h after egg-laying (AEL). The mutants, however, were paralyzed and lacked the coordinated muscle peristalsis required for hatching. This paralysis suggested a neurological deficit and we therefore examined the developmental pattern of individual neurons. Specifically, we explored the embryonic NMJ (Fig. 1b–l), a system in which the timing and anatomy of innervation and synaptogenesis have been described in detail11,13. For the intersegmental nerve b (ISNb), branches of which innervate the ventral longitudinal muscles on which we focused (muscles 6, 7, 11 and 12), the growth cones arrive by 12 h

NATURE NEUROSCIENCE VOLUME 10

[

NUMBER 8

imac170

[

AUGUST 2007

Figure 1 Defective photoreceptor synaptic transmission and motor neuron synapse formation in imac mutants. (a) imac mutants lacked ERG on-off transients. Responses to a light (red bar) in wild-type eyes (y,w; FRT42D GMR-hid 2R CL/FRT42D parental; EGUF/+) included a sustained component and transients (arrows) at light on and light off. The transients were absent in imac mutant eyes (y,w; FRT42D GMR-hid 2R CL/FRT42D imac52 or imac170; EGUF/+). (b) Wild-type NMJ showing the arrangements of muscles (6, 7, 12 and 13) and ISNb motor neurons (orange) at 14 h AEL. (c–e) NMJs of 14-h embryos labeled with anti-FasII. (c) Growth cones of wild-type (+/+) ISNb nerve terminals showed branching along the appropriate muscle boundaries. (d,e) Both imac102/102 and imac116/116 nerves reached the target muscles but appeared abnormal. Filopodia-like processes were present in the muscle area of some flies (open arrow). (f) As b, but at 21 h AEL. (g–l) ISNb motor neurons at 21 h stained for HRP, a pan-neuronal marker. Distinct synaptic varicosities (arrow) were present in wild type (g) that were absent in imac mutants (h–l). Occasionally, an ectopic projection, also lacking boutons, from the neighboring nerve was observed (i, arrowhead). Boundaries of the ventral longitudinal muscles 12, 13, 6 and 7 are marked by short lines. Unless otherwise noted, genotypes: +/+ (y,w; FRT42Diso), imac170 (y+/,w; FRT42D imac170/Df(2R)DAlk21), imac52 (y+/,w; FRT42D imac52/Df(2R)DAlk21), imac102 (y+/,w; FRT42D imac102/Df(2R)DAlk21), imac160 (y+/,w; FRT42D imac160/Df(2R)DAlk21). Scale bar, 10 mm.

AEL. These contacts rapidly mature and, by B14 h AEL (stage 16), the neurites extend along boundaries between adjacent muscle fibers (Fig. 1b,c). These contacts then transform into synapses that have characteristic fine branches with varicosities or boutons that resemble ‘beads on a string’ (Fig. 1f,g). In mutant embryos from this complementation group, including those homozygous for null alleles (see below), ISNb axons had contacted their muscles by 14 h AEL (Fig. 1d,e). Despite normal axonal outgrowth and targeting, the contacts lacked extended branches. Filipodia-like structures were sometimes observed (Fig. 1d, open arrow), but were not restricted to muscle boundaries. At 21 h AEL, the nerve was visible in its target regions, but the contacts had not transformed into mature synapses (Fig. 1h–l); the nerves lacked the bead-like boutons seen in wild-type embryos (Fig. 1g, arrow). Instead, the endings and their branches were more constricted than those at 14 h AEL and frequently retained some filipodia-like processes (Fig. 1i). The failure to form boutons was completely penetrant; it was observed in every abdominal segment of all genotypes examined (alleles imac52, imac102, imac160, imac116 and imac170 as homozygotes or in combination with Df(2R)DAlk21; Fig. 1). At times, axons at the boundary of muscles 12 and 13 were not observed, raising the possibility that some had retracted. Occasionally, we observed ectopic projections from neighboring nerves (Fig. 1i, arrowhead). These inputs form as a result of denervation of the target muscle fibers16 and were therefore consistent with the absence of synapses on these muscles. Unlike the mature boutons formed by ectopic branches in other mutants, however, those that formed in imac mutant embryos did not have boutons. Consequently, both imac160

981

ARTICLES

982


2R094

2R083

in Drosophila but contains a forkheadassociated (FHA) domain, a characteristic of the Kinesin-3 family of motors19. Imac shows 53C 54D most homology to Unc-104, a Kinesin-3 from Df(2R)DAIk21 Caenorhabditis elegans (51% amino acid identity), and the murine kinesins KIF1A + strand (53%) and KIF1Bb (51%; Supplementary +/+ – strand imac Fig. 1 online). Although the Drosophila genome predicts the existence of other g h b W58stop R911C members of the Kinesin-3 family, Imac is imac170 imac102 P12L the only member that includes a carboxy 172 G97S imac terminus pleckstrin homology (PH) domain, imac 52 1671 aa a feature shared with Unc-104, KIF1A and imac 102 KIF1Bb. imac170 has an early stop codon at 76 160 50 imac imac imac imac 116 Trp58 and therefore is likely to be a null allele. i j This allele was therefore used for most of the c d phenotypic analysis, either as a homozygote or over Df(2R)DAlk21 to avoid the effects of possible second-site mutations. Two of the 102 170 imac ;UAS -imac imac +/+ alleles contain amino acid substitutions in the conserved motor domain (Fig. 2b). One Figure 2 imac encodes a kinesin. (a) Genomic map of imac, adapted from Flybase, showing the PLP of these, imac52, has a serine substitution at markers (2R083 and 2R094) and deficiency line used to localize the gene. Sequencing of a candidate Gly97, an essential residue in the ATP-binding gene in the region identified imac as Klp53D (CG8566). (b) Predicted domain structure of Imac and the locations of mutations in four alleles with altered splice sites (red arrowheads), in one nonsense allele, site of all kinesin motors. The arrest of preand in three missense alleles (black arrowheads). Motor domain, blue; coiled coils, gray; FHA domain, synaptic bouton formation in the imac52 pink; PH domain, green; immunogen used, black bar. (c,d) Imac antibody staining of 21-h ventral nerve mutant was comparable to that in the null cords. (c) Imac immunoreactivity was enriched in the synapse-rich neuropil of the CNS and was also allele mutant, indicating that the developdetected in segmental nerves (arrowhead). (d) Lack of immunoreactivity in the imac170 mutant mental defect resulted from loss demonstrated specificity of the antibody. (e–j) Neuromuscular endings labeled with anti-HRP and anti– of an ATP-dependent motor function of synaptotagmin-I (SytI). Synaptic boutons and synaptotagmin-I were lacking in nerve endings of late-stage imac mutant embryos (g,h). Neuronal expression of a UAS-imac transgene enabled flies to hatch as firstImac (Fig. 1j). instar larvae with synaptic boutons (arrows) and concentrations of Syt1 at their terminals (i,j) similar to We generated polyclonal antibodies to a those of wild-type larvae (e,f). Genotypes: +/+ (y,w; FRT42Diso), imac170 (y+/,w; FRT42D imac170/ portion of the stalk domain of Imac Df(2R)DAlk21), imac102 (y+/,w; FRT42D imac102/Df(2R)DAlk21), imac102; UAS-imac (y+/,w; FRT42D (Fig. 2b). Immunostaining of wild-type 102 imac /Df(2R)DAlk21; elavGAL4/UAS-imac). Scale bar, 10 mm. embryos at 21 h AEL showed enrichment of Imac in the nervous system, in particular in the normal innervation and the occasional ectopic projection the synapse-rich regions (Fig. 2c). Imac staining was absent in homowere incapable of normal morphogenesis. Thus, axonal outgrowth zygous mutant embryos, verifying the specificity of the antiserum for and targeting of mutant motor neurons occurred but subsequent Imac (Fig. 2d). In whole-mount embryos, neural expression of Imac synapse formation was arrested. We named the gene immaculate was first detected at stages 11–12 and remained enriched in the nervous connections (imac) to indicate the complete lack of synaptic boutons system for the remainder of embryogenesis (Supplementary Fig. 2 online). The onset of Imac expression corresponds to the expression of in the mutant. many presynaptic proteins20. Thus, the temporal and spatial expression imac encodes a neuronal Kinesin-3 family member of Imac precludes a significant maternal contribution of the protein Polymerase chain reaction (PCR) length polymorphism (PLP) map- and instead is suggestive of a function late in neuronal development. ping17 placed imac between 53C and 54E (Fig. 2). A deficiency, Df(2R)DAlk21, that uncovered the region 53C7 to 53D (ref. 18) did Selective axonal transport defects in imac mutants not complement the imac alleles. In this region, a predicted gene The identification of Imac as a kinesin suggested that the arrest of encoding a kinesin homolog (FlyBase symbol Klp53D) was identified as synaptogenesis might result from a failure to transport necessary a plausible candidate gene. Sequencing of genomic DNA in all eight cargos. Homologs of Imac in C. elegans (Unc-104) and mouse alleles of imac identified point mutations that altered the predicted (KIF1A) are implicated in the transport of synaptic vesicle-associated protein product of this gene (Fig. 2b). Because these mutations were proteins21–23. In Drosophila neurons, by contrast, the conventional induced on an isozygous genetic background and because nearby genes kinesin (Kinesin-1 or KHC) has been reported to transport synaptic showed no sequence differences between one another or in comparison vesicles on the basis of the appearance of axonal aggregates of synaptic to the parental chromosome, the predicted kinesin was identified as vesicles in partial loss-of-function mutants of kinesin heavy chain Imac. This identification was confirmed by rescuing the embryonic (khc)24. We therefore examined the location of synaptic components lethality by expression of an imac cDNA transgene under control of in imac mutants. NMJs of wild-type embryos at 21 h AEL showed either a da-GAL4 or elav-GAL4 driver. With restored neuronal expres- clusters of various presynaptic components at synaptic boutons. For sion of the kinesin, mutant larvae emerged with synaptic boutons at example, synaptotagmin-I, a synaptic vesicle membrane protein, was their NMJs (Fig. 2i). restricted to discrete zones in the synaptic varicosities (Fig. 3a,b). Little The predicted Imac product (Fig. 2b; GenBank accession AF247761) or no synaptotagmin-I was observed along the axons of wild-type represents a kinesin isoform that has not previously been studied embryos. In imac mutants, however, synaptotagmin-I staining was

a

e

HRP

f

Sytl

VOLUME 10

[

NUMBER 8

[


ARTICLES

a

HRP

Sytl

b

c

HRP


HRP

Sytl

h

ANF-GFP

imac170

+/+

e

d

f

ANF-GFP

g

HRP

imac170

+/+

lacking both in the nerve endings and in the axon trunk (Fig. 3c,d). Synaptotagmin-I immunoreactivity was restored in the mutants by expressing imac cDNA in neurons (Fig. 2f,h,j). Similarly, terminal and axonal staining in the mutants was negligible with antibodies to the vesicular glutamate transporter (VGlut). Axon terminals can also release peptides that are stored in dense-core vesicles. These organelles are synthesized in the cell body and undergo kinesin-mediated, microtubule-dependent transport to the site of secretion25,26. To examine whether transport of dense-core vesicles was affected in imac mutants, we monitored the expression of a transgene encoding green fluorescent protein (GFP) fused to rat atrial natriuretic factor precursor (ANF). This transgene (UAS-ANF-GFP) undergoes processing, transport and release in a manner similar to other neuropeptides in Drosophila27. Robust expression of GFP puncta was observed at the synaptic terminals of control lines (Fig. 3f). We did not detect ANF-GFP in either the motor axons or their terminals in imac mutants (Fig. 3h). Thus, similar to the involvement of C. elegans Unc-104 in dense-core vesicle movement26, Imac probably mediates axonal transport of dense-core vesicles in Drosophila. Vesicle components accumulate in imac mutant cell bodies The lack of synaptic and dense-core vesicle components at the nerve terminals could be attributed to a failure to synthesize these proteins, an inability to export them from the cell bodies, or an inability to retain them at nascent synapses. To explore these possibilities, we examined the localization of the synaptic molecules in the CNS (Fig. 4). In the CNS of Drosophila embryos, the neuropil defines a cell body–free region that is enriched in synaptic contacts, axons and glia. The cortex, a region surrounding the neuropil, primarily contains the neuronal cell bodies. Cross-sections of the CNS can thus be used to monitor the concentrations of proteins in cell bodies relative to the axonal and synaptic regions in a broad sampling of neuronal types. Labeling the plasma membranes of neurons with antibodies to horseradish peroxidase (HRP), a neuronal membrane marker, revealed the overall architecture of the CNS (Fig. 4d). The density of membrane in the neuropil causes this structure to be more intensely labeled than the cortex. This pattern was not detectably altered in imac mutants, indicating that the mutants had no gross anatomical defect. Components of synaptic and dense-core vesicles were, however, markedly redistributed in imac mutants. Synaptotagmin-I immunoreactivity, for example, is normally intensely concentrated in the neuropil and scant in the cortex. In imac mutants this pattern was


[

NUMBER 8

[

AUGUST 2007

Figure 3 Synaptic and dense-core vesicle proteins are absent in imac nerve endings. (a–h) Neuromuscular nerve endings of 21-h embryos stained for the neuronal membrane marker HRP (a,c,e,g), and either synaptotagmin-I (b,d) or the dense-core vesicle marker ANF-GFP (f,h). The synaptic varicosities of wild-type NMJs had distinct clusters of synaptotagmin-I (b) and ANF-GFP (f), but these clusters were missing in imac170 mutants (d,h). Genotypes for visualization of dense-core vesicles: +/+ (w; FRT42D; elavGAL4/UAS–ANF-GFP) and imac170 (w; FRT42D imac170/Df(2R)DAlk21; elav-GAL4/UAS-ANFGFP). Muscle boundaries are marked by short lines. Scale bar, 10 mm.

reversed: synaptotagmin-I accumulated in cell bodies and was low in the neuropil (Fig. 4a). Synaptotagamin-I–GFP, expressed in the imac mutant nervous system by an elav-GAL4 driver, showed the same redistribution (Supplementary Fig. 3 online). VGlut, normally found at a few CNS synapses that are glutamatergic, was similarly redistributed from neuropil to cell bodies in imac mutants (Fig. 4b), as was cysteine string protein (Supplementary Fig. 3) and ANF-GFP (Fig. 4c). The concentration of the vesicular proteins in imac mutant cell bodies indicates that these proteins are synthesized in the mutants. The loss of these markers from axon tracts and synaptic regions indicates a defect in their transport from the cell bodies. Absence of immunoreactivity of these molecules in the peripheral axons of motor neurons (Fig. 3) likewise supports the idea that Imac has a direct role in the transport of synaptic materials. In addition, live-cell imaging of vesicle precursors in the segmental nerves of an imac mutant has revealed defects in the anterograde movement of these cargos (R. Barkus and W. Saxton, personal communication), further substantiating the idea that Imac has a motor function. Some axonal transport persists in imac mutants To determine whether Imac is involved in the transport of other proteins and organelles, we examined the distribution of various intracellular components. The normal axonal growth and guidance observed in imac mutants implies that the transport of post-Golgi vesicles with new membrane and cell-surface proteins persists. As shown above (Figs. 1 and 4), the neuronal membrane marker, HRP, did not reveal any significant differences between wild type and imac mutants. We also assayed post-Golgi membrane trafficking in imac mutants by using a fusion of the extracellular and transmembrane domains of CD8 to a cytoplasmic GFP, a construct that serves as a nonspecific reporter of the constitutive transport of membrane proteins to the cell surface, including axons and terminals28. The distribution of CD8-GFP was not affected in imac mutants (Fig. 4h). Similarly, Fasciclin II (FasII), a Drosophila homolog of vertebrate neural cell-adhesion molecules and an essential regulator of motor neuron growth and guidance29, remained concentrated in the axon tracts and synaptic regions in imac mutants (Fig. 1c–e). Another plasma membrane protein, syntaxin, is needed for exocytosis and addition of membrane to the cell surface30. No change in the distribution of syntaxin, visualized with the monoclonal antibody 8C3, was detected in imac-null embryos (Fig. 4f). These data are consistent with our observation that axon outgrowth and targeting proceeds normally in imac mutants, and they demonstrate that the vesicles required for membrane extension and axon targeting are not conveyed by Imac. In

983

ARTICLES

b

N

d

e

f

C

imac170

C N

c

SytI

ANF-GFP

h

HRP

LAP

i

Syntaxin

j

+/+

g

VGlut

imac170


+/+

a

HRP

CD8-GFP

HRP

addition, the data support the current model wherein mechanisms regulating membrane outgrowth are distinct from mature synaptic vesicle exocytosis28. Synaptic terminals are also enriched in mitochondria, an organelle transported by kinesin-1 (ref. 31). We visualized mitochondria using Mito-GFP, a mitochondrially targeted GFP variant32. Wild-type and imac mutant nerves had similar distributions of mitochondria, indicating their proper transport in motor neuron axons (Supplementary Fig. 3). Because of the absence of bead-like boutons in imac mutants, we examined the neuronal cytoskeletal component Futsch, a Drosophila protein with homology to the vertebrate microtubule-associated protein MAP1B (ref. 33). Loops of bundled, Futsch-containing microtubules are present in many wild-type synaptic boutons at NMJs of third-instar larvae and may contribute to the rounded structure of the bouton. By contrast, unbundled microtubules characterize growth cones, and Futsch is likely to contribute to the transition from unbundled to bundled microtubules. In embryos, however, the significance of Futsch at the NMJ is not known. In wild-type embryos, Futsch immunoreactivity was abundant in the axons of motor neurons and in synapses at the NMJ, although Futschimmunoreactive microtubule loops were discernible only occasionally in boutons (Fig. 4j). In imac mutants, Futsch was similarly detected in motor axons and at the NMJ, although the immunoreactivity was more discontinuous than in wild type. Consistent with the absence of

984

Figure 4 Synaptic vesicle proteins are selectively concentrated in cell bodies. (a–f) Ventral nerve cords of wild-type (top) and imac170 (bottom) 21-h embryos. (a–c) Immunoreactivity of synaptic vesicle proteins (synaptotagmin-I and VGlut) and a dense-core marker (ANF-GFP) were concentrated in the neuropil (N) region of wild-type CNS, but were redistributed in imac mutants with more in the cell body–rich area of the cortex (C) and less in the neuropil (N). (d–f) Several other proteins showed unaltered distribution in imac mutants, including the membrane marker HRP (d), the cytoplasmic endocytosis protein LAP (e) and the plasma membrane protein syntaxin (f). (g–j) ISNb of wild-type (top) and imac170 (bottom) 21-h embryos. (g,h) Post-Golgi transport of CD8-GFP to neuronal membranes showed labeling similar to anti-HRP in both wild-type (w; FRT42D; elavGAL4/UAS-CD8-GFP) and imac170 (w; FRT42D imac170/Df(2R)DAlk21; elav-GAL4/UAS-CD8GFP) embryos. (i,j) Immunoreactivity of a cytoskeletal protein, Futsch, was observed in both wild-type and imac170 nerve endings. Futsch loops were occasionally detected in wild-type boutons (j, top; arrowhead and inset). imac170 nerves showed fragmented Futsch immunoreactivity (j, bottom; inset). Scale bars, 10 mm.

boutons, no loops of Futsch immunoreactivity were seen in imac mutants. Although the distribution of Futsch suggests that changes have occurred in the cytoskeleton of imac mutants, the presence of Futsch at the endings indicates that this cytoskeletal protein is transported independently of Imac. This finding is consistent with the known movement of many cytoskeletal components by slow axonal transport34. In addition, the normal distribution of Futsch mitochondria and trans-Golgi–derived vesicle markers indicates that the failure to transport synaptic components cannot be secondary to microtubule defects. Likewise, the absence of boutons in imac mutants cannot be attributed to the absence of Futsch. We also examined the localization of two synaptic proteins that are cytoplasmic rather than organelle associated or cytoskeletal. Both proteins, LAP (also known as AP180) and endophilin, are involved in the endocytosis of synaptic vesicles35. Their distribution in the nervous system was enriched in, but not strictly limited to, the neuropil in both wild type and imac mutants (Fig. 4e and Supplementary Fig. 3 online). Thus, the mislocalizations observed in imac mutants seem to be specific to a subset of synaptic proteins. Few active zones form in imac mutants The components of the active zone may arrive at nascent synapses, at least in part, through PTVs8, although this vesicle class has not been described in Drosophila. To determine the extent of active-zone transport and assembly in imac mutants, we examined an active-zone marker, monoclonal antibody nc82, which has been shown to recognize Brp, a member of the ELKS/CAST/ERC protein family and a probable component of the T-bar, a dense body that projects back into the cytoplasm from the electron density at the plasma membrane36. Wild-type NMJs at 21 h AEL showed clusters of nc82 puncta in boutons (66.0 ± 3.0 puncta per ISNb ending on muscles 6, 7, 12 and 13; Fig. 5a,b). imac mutants showed only 11% of the puncta of wild type (7.5 ± 1.0; t-test,

VOLUME 10

[

NUMBER 8

[


ARTICLES P ¼ 1.3 1021; Fig. 5c) and these puncta were also fainter (62% of the average intensity in wild type; Fig. 5e). This reduction could not be explained by a decrease in the surface area of nerve-muscle contacts in the mutant: the amount of nc82 puncta per HRP area was only 23% of that of wild type (Fig. 5e). To verify that puncta in imac mutants were not being missed owing to a decrease in the intensity of the nc82 14-h AEL

15-h AEL

17-h AEL

13-h AEL

14-h AEL

15-h AEL

17-h AEL

21-h AEL

+/+

13-h AEL

21-h AEL

imac170

b

60 40

*

20

*

*

e

200 150 100

*

*

*

*

50 0

19 15 17 Time (h AEL)

13 h imac

21 h +/+

21 h imac

80 60 40 20

0 13

13 h +/+

120 100

% of wild type

d

80

HRP area (µm2)

c Number of puncta


a

immunoreactivity, we recounted puncta with a higher confocal gain setting that could not be used on wild-type NMJs because of saturation and consequent optical merging of puncta. Even with this bias toward overcounting in the mutant, only 24% of the puncta seen in wild type were observed. In addition, within the area counted, many of the puncta in imac mutants occurred along the axons rather than at the endings,

21

13

15 17 19 Time (h AEL)

21

0 Puncta

HRP area

Puncta Puncta /area intensity

Puncta

HRP area

Puncta Puncta /area intensity

Figure 5 Altered developmental distribution of the active-zone marker nc82 in imac mutants. (a,b) Time course of nc82 staining in ISNb of wild-type (a) and imac170 (b) embryos. ISNb endings onto muscles 6, 7, 12 and 13 were immunolabeled with anti-HRP (top, blue; bottom, white) and active zones were labeled with nc82 (top, yellow). Mutant nerve endings at the same time points did not develop boutons and contained few nc82 puncta; in particular, the large increase in nc82 fluorescence at 15 h AEL did not occur in imac mutants. Muscle boundaries are marked with short lines. Scale bar, 10 mm. (c,d) Quantification of the developmental time course of nc82 puncta (c) and HRP area (d) in ISNb nerve endings of wild-type (black, filled circles) and mutant (blue, open circles) embryos at 13–21 h AEL. (e) Quantification of nc82 distribution at 13 and 21 h AEL as a percentage of wild-type. The number of nc82 puncta, nc82 puncta intensity and HRP area in the mutant nerve endings were comparable to those in wild type at 13 h, but were markedly reduced at 21 h. Data are the mean ± s.e.m. (13 h, n ¼ 13 +/+, 10 imac170; 14 h, n ¼ 13 +/+, 7 imac170; 15 h, n ¼ 4 +/+, 4 imac170; 17 h, n ¼ 7 +/+, 6 imac170; 21 h, n ¼ 19 +/+, 20 imac170). *P o 0.05 by t-test.


[

NUMBER 8

[

AUGUST 2007

985

ARTICLES

a

b

c

d

N


e

Mu

M

imac170

+/+

imac170

Figure 6 Synaptic vesicles are rare and active zones are few in imac nerve profiles. (a–e) Electron microscopy cross-sections of 21-h embryonic nerve endings. (a) Typical wild-type terminal profile showing synaptic vesicles, mitochondrion (M) and active zone (boxed). (b) Representative micrograph of an imac170 nerve ending (N) that is clear, lacks synaptic vesicles, and lies underneath the basal lamina (arrowhead) adjacent to the muscle (Mu). (c) Representative micrograph of the occasional active zone with a T-bar encountered in mutant nerve profiles. These densities were not associated with synaptic vesicles. (d,e) Enlargements of boxed areas in a and c, respectively. Scale bar, 200 nm.

where they are normally located. In the CNS, the ratio of nc82 staining in the cell body region relative to the neuropil was also higher in imac mutants than in wild type (Supplementary Fig. 4 online). Because the shift in nc82 immunoreactivity was not as absolute as the difference observed for synaptic vesicle markers, we examined quantitatively the emergence of nc82 immunoreactivity at the NMJs of ISNb (Fig. 5). In wild type, a few puncta of nc82 immunoreactivity were detected at 13 and 14 h AEL, when the growth cone first contacts the muscle and processes begin to elongate along the muscle boundaries. At 15 h AEL, however, the number of puncta increased sharply, and modest further increases subsequently occurred until hatching (Fig. 5a,c). At 13 and 14 h AEL, imac embryos, like wild-type embryos, had only a few nc82 puncta. In imac mutants, however, the large increase at 15 h AEL did not take place. Instead, the number of puncta declined modestly for the remainder of embryonic development (Fig. 5b,c). Even taking into account the smaller surface area of the nerve-muscle contacts in imac mutants, nc82 puncta were sparser in imac mutants than in wild type (Fig. 5d,e). Despite their paucity, the presence of some nc82 puncta (but not synaptic vesicle proteins; Supplementary Fig. 4) in the mutants at early stages suggested that some active-zone protein can diffuse or be transported into the axons independently of imac, at least at this early stage. The lack of increase in nc82 staining, however, is an early phenotype of the mutants HRP a that is apparent before the stage at which varicosities fail to form.

To determine the nature of the mutant nerve-muscle contacts and to look for signs of active-zone formation or vesicle accumulation that might not have been detected by immunocytochemistry, we performed electron microscopy. In a given cross-section through a wild-type embryo, we identified on average ten NMJs, amongst which five active zones and three T-bars were seen. Identification of wild-type NMJs was relatively straightforward in that terminals contained uniformly sized synaptic vesicles (Fig. 6a,d). In imac mutants, however, most nervemuscle contacts lacked vesicles and were therefore recognized on the basis of the close apposition of the electron-lucent nerve ending to the muscle fibers and the presence of the basement membrane overlying the nerve ending (Fig. 6b,c). By using these criteria, 0–2 nerve-muscle contacts were typically identified per section. The absence of synaptic vesicles made identification of the nerves more difficult in imac mutants and consequently some contacts may have been missed. To systematize our characterization of these nerve-muscle contacts, we undertook serial electron microscopy of cross-sections of imac embryos (see Methods). Active-zone and T-bar counts were normalized to the measured surface area of the neurite to take into account the paucity of recognizable contacts in imac mutants and changes in their size. Active zones were indeed recognized in imac mutant nerve

b

GluRIIC

c

d HRP

Figure 7 Postsynaptic differentiation in imac mutants. (a–n) ISNb nerves of wild-type (a–f) and imac170 (g–n) 21-h embryos immunolabeled with anti-HRP (a,c,g,i) to visualize the nerve, anti-GluRIIC (b,d,h,j) to determine the extent of postsynaptic differentiation, and nc82 (e,k,m) to detect the presynaptic active zone. (c–f,i–n) Enlargements of boxed areas in a and g, respectively. Clusters of postsynaptic receptors were apparent near the neuronal membrane of both genotypes (a–d,g–j). These clusters (green) were directly opposite the presynaptic nc82 puncta (magenta) in wild-type embryos (f). In imac mutants, nc82 immunoreactivity was faint at the gain settings used for wild-type embryos (k,l and Fig. 5). When the gain was increased (m,n), however, most of the clusters of GluRIIC were found to lie adjacent to faint nc82 puncta (n). Muscle boundaries are marked by white lines. Scale bar, 5 mm.

986

e

f nc82

+/+

g

h

GluRIIC

i

nc82

j HRP

k nc82

m

NUMBER 8

nc82

GluRIIC

n nc82

[

GluRIIC

l

imac170

VOLUME 10

GluRIIC

[

nc82

GluRIIC



ARTICLES endings, but at B60% of the frequency encountered in wild type, when either the active-zone number or the active-zone area was normalized to the total neuronal surface area in the sections (Supplementary Table 1 online). imac mutants showed a greater reduction (17% of control) in T-bar frequency. In addition, imac T-bars were often smaller than wild-type T-bars. Occasionally, they were surrounded by diffuse electron-dense material that did not resemble membrane vesicles (Fig. 6e). On average, we counted 70 synaptic vesicle profiles per section of wild-type boutons. In all 120 sections examined from 29 mutant nerve-muscle contacts, we observed only one region that contained any synaptic vesicle-like structures (16 structures in total, equivalent to 0.13 vesicle profiles per nerve cross-section). Thus, as expected from the immunocytochemistry (Figs. 3 and 5), imac nerve endings were less extensive, rarely if ever contained synaptic vesicles, and had few active zones and very few T-bars. Transmission mutants and postsynaptic development The absence of synaptic vesicles in imac terminals indicated that synaptic transmission did not occur; thus, it was possible that the absence of boutons was an indirect consequence of the lack of synaptic transmission. We therefore examined the embryonic NMJ of syntaxin1AD229, a null mutation of an essential component of the SNARE complex. Null alleles of syntaxin1A lack both evoked and spontaneous release of synaptic vesicles37. At 21 h AEL, however, the NMJs of syntaxin1AD229 retained distinct synaptic boutons (Supplementary Fig. 5 online). The terminals of syntaxin1AD229 were less elaborate, as shown previously38, but the boutons contained nc82 puncta (Supplementary Fig. 5). Similarly, formation of active zones and vesicle clusters occur at syntaxin1A NMJs37,38. Therefore, morphological formation of boutons cannot require synaptic transmission and the failure of synaptogenesis in imac mutants cannot be attributed to this cause. The lack of complete presynaptic differentiation in imac mutants enabled us to determine whether postsynaptic development proceeds in its absence. Ultrastructural data revealed the presence of postsynaptic membrane densities opposite presynaptic specializations in imac mutants (Fig. 6c,e). To determine the degree of postsynaptic differentiation, we immunolocalized glutamate receptors (GluRs). In imac mutants, discrete puncta of the subunits GluRIIC (Fig. 7h) and GluRIIA (data not shown) were evident at sites where the processes of the motor neurons contacted the muscle fibers, although they were largely absent elsewhere on the muscle. Most of these GluRIIC puncta were directly opposite the remaining faint nc82-immunoreactive puncta39 (Fig. 7k–n). Thus, the muscle fibers were competent to respond to nerve contact with specialization of the postsynaptic membrane, despite the incomplete differentiation and abnormal morphology of the presynaptic endings in imac mutants. DISCUSSION Our studies of imac mutants have revealed that a molecular motor has a role in presynaptic maturation and provide insight into the formation of synapses. Specifically, our phenotypic analysis demonstrates, first, that Imac transports components required for the morphological transformation of axonal growth cones to mature boutons; second, that the membrane and proteins that support axon outgrowth, axon guidance and target recognition do not rely on the same kinesin for their transport as the one that supports synaptogenesis; and third, that postsynaptic differentiation does not depend on synaptic vesicle clustering, synaptic transmission or presynaptic bouton formation. In addition, the finding that Imac is essential for the transport of synaptic vesicle precursors suggests that this is their primary motor, rather than KHC, as previously implicated24. Other components of the


[

NUMBER 8

[

AUGUST 2007

synapse can be classified as either dependent or independent of Imac on the basis of their transport phenotype. Absence of synaptic boutons in imac mutants The phenotype of imac mutants is unprecedented despite extensive genetic analysis of the Drosophila NMJ as a model system for synapse development. Prior studies have uncovered molecules that are crucial to axon pathfinding40, and yet, despite mistargeting, axons in those mutants still form boutons if they reach a muscle. By contrast, mutation of imac affects all of the NMJs in the embryo that we have analyzed and prevents varicosity formation despite signs of appropriate outgrowth, guidance and nerve-target recognition. Many signaling pathways, including those mediated by electrical activity, Wingless, bone morphogenetic proteins, highwire and fasciclins, can regulate the growth and plasticity of these synapses41. These pathways influence bouton number and the size, length and branching of endings, but they do not prevent bouton formation outright. Likewise, mutants that affect synapse specificity in C. elegans (for example, syg-1 and syg-2) form structurally normal synapses42. It is interesting to compare the imac mutant with mutations altering the release of neurotransmitter at the synapse. In mammals and C. elegans, mutations in individual components of synaptic vesicles or the active zone, including mutations that prevent transmitter release, do not prevent growth cones from transforming into synapses and do not lead to the ultrastructural abnormalities observed in imac mutants1,43. In Drosophila, the most complete genetic blockade of transmitter release is provided by mutants of syntaxin1A, which is required for all vesicle fusions, both spontaneous and evoked. In a direct comparison of imac and syntaxin1A mutants, we found that the latter continued to form varicosities and had abundant active zones (as detected with the monoclonal antibody nc82), as previously reported37,38. Boutons similarly persist in other mutants that affect transmitter release (synaptotagmin-I, neuronal-synaptobrevin, dunc-13 and vha100-1)35 and vesicle recycling (for example, shibire, endophilin and synaptojanin)44. Therefore, the morphological phenotype of imac mutants cannot be due to a secondary consequence of the absence of transmission. Indeed, surprisingly few genes currently have a clear role in synaptogenesis and in the structural assembly of synapses. Our studies implicate, for the first time to our knowledge, a specific gene in the transition from growth cone to varicosities and will facilitate examination of a morphological process that is not well understood at the molecular level. Which kinesin transports synaptic vesicle components? Despite evidence from C. elegans and mammalian systems implicating Kinesin-3 motors (Unc-104 and KIF1A) in axonal transport of synaptic vesicle precursors21–23,45, mammalian Kinesin-1 motors have also been associated with synaptic vesicle transport46,47. In addition, KHC, a Kinesin-1, has been the predominant candidate for this transport in Drosophila24,48. Key evidence implicating KHC has been the presence in khc mutant larvae of axonal swellings filled with various vesicles and organelles, including synaptic vesicle markers. These accumulations, sometimes called ‘traffic jams’, are distinct from the imac mutant phenotype: that is, the categorical absence of synaptic vesicle markers from motor neuron axons and the persistent transport of other organelles. The absence of vesicles from axons in imac mutants would seem to indicate that no other motor can substitute for Imac during NMJ formation, although KHC is present in these motor neurons and transports mitochondria in the absence of Imac. The accumulation of synaptic vesicle proteins in axonal swellings in khc mutants

987

ARTICLES


might arise indirectly from the failure of other classes of transport, potentially snarling traffic in the axons. Alternatively, KHC-mediated transport may supplement Imac in older Drosophila, despite the essential role of Imac in transport of synaptic vesicles during de novo embryonic synaptogenesis. Active-zone transport PTVs are thought to carry many constituents of the mammalian active zone, but nothing is known about active-zone transport in Drosophila. Thus, the change in Brp (nc82) distribution and the ultrastructural phenotype of imac reported here represent an initial investigation into that process. These phenotypes are not as absolute, however, as those observed for synaptic and dense-cored vesicle markers; therefore, the delivery of active-zone proteins may be more complex. The paucity of these proteins at imac mutant nerve endings might occur because Imac has a direct role in their transport or because there may be a failure to trap or concentrate them at the terminal. On the one hand, the persistence of some Brp in axons and their endings, although sharply reduced at NMJs, suggests that transport or diffusion of Brp may persist but that its capture at synapses may be impaired by the lack of synaptic maturation. On the other hand, the reductions in nc82-immunoreactive puncta and active-zone counts were greater than could be explained by the decrease in size of the nerve-muscle contacts. In addition, the dearth of nc82 puncta was apparent by 15 h AEL, before the stage when boutons form. Thus, the failure to form sufficient active zones and T-bars is unlikely to be secondary to the failure to form boutons. Rather, it suggests that Imac is directly responsible either for the transport of a portion of the activezone proteins or for the transport of a protein needed to retain them at muscle contacts. Postsynaptic maturation In imac mutants, postsynaptic differentiation was assessed by two methods: first, the presence of postsynaptic densities by electron microscopy; and second, the clustering of glutamate receptors by immunocytochemistry (Figs. 6 and 7). By both criteria, specialization of the postsynaptic membrane at the site of nerve contact proceeded in the absence of synaptic vesicles and without the morphological transition to a synaptic bouton. Postsynaptic structural changes, as indicated by the clustering of receptors, are thought to be induced by contact with the nerve20. The importance of transmitter release in these processes, however, has been controversial38,49, in part because no mutants have been available that cleanly prevent both evoked release and the spontaneous release of synaptic vesicles without also compromising the survival of the axon. Axons in imac mutants lack components essential for neurotransmitter release including VGlut, which is required to load vesicles with glutamate, and indeed lack synaptic vesicles themselves (Figs. 3, 4 and 6). Despite this, glutamate receptor subunits cluster on the muscle beneath the neuronal membranes, indicating that vesicular transmitter release per se is not needed for postsynaptic differentiation. We cannot, however, exclude the possibility that non-vesicular glutamate induces the postsynaptic response or that other molecules are released from the growth cone38. Pre- and postsynaptic development cannot be completely independent because release sites and receptor clusters need to be aligned. In imac mutants, this alignment is preserved, despite the reduced number of active zones and diminished nc82 immunoreactivity (Fig. 7). Thus, the machinery involved in the matching of pre- and postsynaptic structures persists, at least in part, in imac mutants. Glutamate receptor clusters were not reduced in number or intensity as markedly as recognizable active zones in electron microscopy images or as the intensity of nc82 puncta in nerve

988

endings. Thus, the low concentration of nc82 immunoreactivity opposite most of the glutamate receptor clusters may represent incompletely assembled active zones. Our analysis of imac mutants has facilitated a dissection of the transport mechanisms that function during a crucial phase of neuronal development. The existence of a motor for presynaptic maturation that is distinct from that for axon outgrowth and guidance may reflect the different regulatory needs for distributing the molecules that mediate these events. New membrane, for example, is consistently brought to the growing tip of the axon, whereas synaptic precursors travel in both directions in the axon, scanning for signals from target cells that will determine where they will form a functional connection6,8. Further understanding of synaptogenesis will require identification of both the factors that regulate these motors and the particular cargos that alter the morphology of nerve endings. METHODS Drosophila stocks and transgenic imac constructs. A list of stocks and the methods for generating imac cDNA and antibody are provided in Supplementary Methods online. Immunocytochemistry. For 13–14-h embryos, dissection was performed on polylysine-coated slides. Later stage embryos, dissected with pulled-glass needles, were attached to Sylgard with Nexaband glue28 (Abbott Laboratories). For details, see Supplementary Methods. We used the following primary and secondary antibodies: mouse anti-Futsch (22C10, diluted 1:100; Developmental Studies Hybridoma Bank, DSHB); mouse anti-syntaxin (8C3, 1:50; DSHB); mouse anti–discs large (1:1,000; DSHB); mouse anti-FasII (1D4, 1:20; DSHB); mouse anti-Brp, (nc82, 1:100; DSHB); rabbit anti–synaptotagmin-I (1:4,000; N. Reist, Colorado State University); rabbit anti-GluRC (1:2,000; A. DiAntonio Washington University); rabbit anti-VGlut (1:10,000; A. DiAntonio); anti-LAP (1:200; B. Zhang, University of Texas); Cy5- or FITC-conjugated goat anti-HRP (1:100, Jackson Immunoresearch); and FITC-, Cy3- or Texas red–conjugated secondary antibodies (Jackson Immunoresearch). For immunolabeling with FasII antibody, we used biotinylated anti-mouse secondary antibody and the Vectastain ABC system (Vector Laboratories) and developed the slides with diaminobenzidine (Sigma). Imaging and analysis. Diaminobenzidine-processed slides were imaged with a Nikon microscope (Eclipse E800). Confocal images were acquired by using a laser scanning confocal microscope (LSM 510 META/NLO, Carl Zeiss MicroImaging). For ISNb confocal images, data were acquired as stacks of multi-tracked, separate channels and then projected to a single plane with LSM Imaging Analysis Software 3.2 (Carl Zeiss MicroImaging). We prepared images using Photoshop and Illustrator (Adobe). For details, see Supplementary Methods. For quantification of nc82 puncta, projections of stacked confocal images were analyzed with MetaMorph software (Molecular Devices). For a given data set (mutant and wild type analyzed together), the same threshold setting was used. We drew a region of interest (ROI) around the nerve profiles in muscles 6, 7, 12 and 13. The pixel area (in mm2) for the ROI was calculated for the channel representing HRP. For the same ROI, nc82 puncta were counted. The sum of pixel intensities (in arbitrary units) for the same ROI was also calculated for each of the channels. Data are reported as mean ± s.e.m. Two-tailed Student’s t-test was used for comparisons. P values of o 0.05 were designated as statistically significant. Electron microscopy. Late-stage (21 h AEL) mutant and wild-type embryos were fixed by injecting 5% glutaraldehyde in 0.05 M phosphate buffer into the posterior end and processed accordingly50. For details, see Supplementary Methods. Note: Supplementary information is available on the Nature Neuroscience website.

ACKNOWLEDGMENTS We thank W. Saxton, D. Van Vactor and J. Kaplan for comments on the manuscript; A. DiAntonio, R. Palmer, N. Reist, M. Higashi, and the Bloomington

VOLUME 10

[

NUMBER 8

[


ARTICLES


Stock Center for reagents and fly strains; A.Y.N. Goldstein and J. Salogiannis for experimental assistance; members of the Schwarz laboratory for discussions; HCNR and DDRC imaging cores for imaging and analysis assistance; A. Prokop and the HMS electron microscopy facility for electron microscopy help. This work was supported by a US National Research Service Award predoctoral fellowship (E.P.-C.), a Howard Hughes Medical Institute predoctoral fellowship (D.K.D.), and a grant from the US National Institutes of Health (RO1MH075058 to T.L.S.). COMPETING INTERESTS STATEMENT The authors declare no competing financial interests. Published online at http://www.nature.com/natureneuroscience Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions 1. Ziv, N.E. & Garner, C.C. Cellular and molecular mechanisms of presynaptic assembly. Nat. Rev. Neurosci. 5, 385–399 (2004). 2. Lee, H. & Van Vactor, D. Neurons take shape. Curr. Biol. 13, R152–R161 (2003). 3. Hannah, M.J., Schmidt, A.A. & Huttner, W.B. Synaptic vesicle biogenesis. Annu. Rev. Cell Dev. Biol. 15, 733–798 (1999). 4. Friedman, H.V., Bresler, T., Garner, C.C. & Ziv, N.E. Assembly of new individual excitatory synapses: time course and temporal order of synaptic molecule recruitment. Neuron 27, 57–69 (2000). 5. Roos, J. & Kelly, R.B. Preassembly and transport of nerve terminals: a new concept of axonal transport. Nat. Neurosci. 3, 415–417 (2000). 6. Ahmari, S.E., Buchanan, J. & Smith, S.J. Assembly of presynaptic active zones from cytoplasmic transport packets. Nat. Neurosci. 3, 445–451 (2000). 7. Kraszewski, K. et al. Synaptic vesicle dynamics in living cultured hippocampal neurons visualized with CY3-conjugated antibodies directed against the lumenal domain of synaptotagmin. J. Neurosci. 15, 4328–4342 (1995). 8. Shapira, M. et al. Unitary assembly of presynaptic active zones from Piccolo-Bassoon transport vesicles. Neuron 38, 237–252 (2003). 9. Zhai, R.G. et al. Assembling the presynaptic active zone: a characterization of an active zone precursor vesicle. Neuron 29, 131–143 (2001). 10. Sanes, J.R. & Lichtman, J.W. Development of the vertebrate neuromuscular junction. Annu. Rev. Neurosci. 22, 389–442 (1999). 11. Featherstone, D.E. & Broadie, K. Surprises from Drosophila: genetic mechanisms of synaptic development and plasticity. Brain Res. Bull. 53, 501–511 (2000). 12. Yoshihara, M., Rheuben, M.B. & Kidokoro, Y. Transition from growth cone to functional motor nerve terminal in Drosophila embryos. J. Neurosci. 17, 8408–8426 (1997). 13. Rheuben, M.B., Yoshihara, M. & Kidokoro, Y. Ultrastructural correlates of neuromuscular junction development. Int. Rev. Neurobiol. 43, 69–92 (1999). 14. Stowers, R.S. & Schwarz, T.L. A Genetic method for generating Drosophila eyes composed exclusively of mitotic clones of a single genotype. Genetics 152, 1631–1639 (1999). 15. Dickman, D.K., Horne, J.A., Meinertzhagen, I.A. & Schwarz, T.L. A slowed classical pathway rather than kiss-and-run mediates endocytosis at synapses lacking synaptojanin and endophilin. Cell 123, 521–533 (2005). 16. Chang, T.N. & Keshishian, H. Laser ablation of Drosophila embryonic motoneurons causes ectopic innervation of target muscle fibers. J. Neurosci. 16, 5715–5726 (1996). 17. Berger, J. et al. Genetic mapping with SNP markers in Drosophila. Nat. Genet. 29, 475–481 (2001). 18. Loren, C.E. et al. A crucial role for the Anaplastic lymphoma kinase receptor tyrosine kinase in gut development in Drosophila melanogaster. EMBO Rep. 4, 781–786 (2003). 19. Vale, R.D. The molecular motor toolbox for intracellular transport. Cell 112, 467–480 (2003). 20. Broadie, K.S. & Bate, M. Development of the embryonic neuromuscular synapse of Drosophila melanogaster. J. Neurosci. 13, 144–166 (1993). 21. Hall, D.H. & Hedgecock, E.M. Kinesin-related gene unc-104 is required for axonal transport of synaptic vesicles in C. elegans. Cell 65, 837–847 (1991). 22. Okada, Y., Yamazaki, H., Sekine-Aizawa, Y. & Hirokawa, N. The neuron-specific kinesin superfamily protein KIF1A is a unique monomeric motor for anterograde axonal transport of synaptic vesicle precursors. Cell 81, 769–780 (1995). 23. Zhao, C. et al. Charcot-Marie-Tooth disease type 2A caused by mutation in a microtubule motor KIF1Bb. Cell 105, 587–597 (2001).


[

NUMBER 8

[

AUGUST 2007

24. Hurd, D.D. & Saxton, W.M. Kinesin mutations cause motor neuron disease phenotypes by disrupting fast axonal transport in Drosophila. Genetics 144, 1075–1085 (1996). 25. Goldstein, L.S. & Yang, Z. Microtubule-based transport systems in neurons: the roles of kinesins and dyneins. Annu. Rev. Neurosci. 23, 39–71 (2000). 26. Jacob, T.C. & Kaplan, J.M. The EGL-21 carboxypeptidase E facilitates acetylcholine release at Caenorhabditis elegans neuromuscular junctions. J. Neurosci. 23, 2122–2130 (2003). 27. Rao, S., Lang, C., Levitan, E.S. & Deitcher, D.L. Visualization of neuropeptide expression, transport, and exocytosis in Drosophila melanogaster. J. Neurobiol. 49, 159–172 (2001). 28. Murthy, M., Garza, D., Scheller, R.H. & Schwarz, T.L. Mutations in the exocyst component Sec5 disrupt neuronal membrane traffic, but neurotransmitter release persists. Neuron 37, 433–447 (2003). 29. Lin, D.M. & Goodman, C.S. Ectopic and increased expression of fasciclin II alters motoneuron growth cone guidance. Neuron 13, 507–523 (1994). 30. Burgess, R.W., Deitcher, D.L. & Schwarz, T.L. The synaptic protein syntaxin1 is required for cellularization of Drosophila embryos. J. Cell Biol. 138, 861–875 (1997). 31. Glater, E.E., Megeath, L.J., Stowers, R.S. & Schwarz, T.L. Axonal transport of mitochondria requires milton to recruit kinesin heavy chain and is light chain independent. J. Cell Biol. 173, 545–557 (2006). 32. Pilling, A.D., Horiuchi, D., Lively, C.M. & Saxton, W.M. Kinesin-1 and dynein are the primary motors for fast transport of mitochondria in Drosophila motor axons. Mol. Biol. Cell 17, 2057–2068 (2006). 33. Roos, J., Hummel, T., Ng, N., Klambt, C. & Davis, G.W. Drosophila Futsch regulates synaptic microtubule organization and is necessary for synaptic growth. Neuron 26, 371–382 (2000). 34. Ma, D., Himes, B.T., Shea, T.B. & Fischer, I. Axonal transport of microtubule-associated protein 1B (MAP1B) in the sciatic nerve of adult rat: distinct transport rates of different isoforms. J. Neurosci. 20, 2112–2120 (2000). 35. Schwarz, T.L. Transmitter release at the neuromuscular junction. Int. Rev. Neurobiol. 75, 105–144 (2006). 36. Wagh, D.A. et al. Bruchpilot, a protein with homology to ELKS/CAST, is required for structural integrity and function of synaptic active zones in Drosophila. Neuron 49, 833–844 (2006). 37. Broadie, K. et al. Syntaxin and synaptohrevin function downstream of vesicle docking in Drosophila. Neuron 15, 663–673 (1995). 38. Featherstone, D.E., Rushton, E. & Broadie, K. Developmental regulation of glutamate receptor field size by nonvesicular glutamate release. Nat. Neurosci. 5, 141–146 (2002). 39. DiAntonio, A. Glutamate receptors at the Drosophila neuromuscular junction. Int. Rev. Neurobiol. 75, 165–179 (2006). 40. Dickson, B.J. Molecular mechanisms of axon guidance. Science 298, 1959–1964 (2002). 41. Marques, G. & Zhang, B. Retrograde signaling that regulates synaptic development and function at the Drosophila neuromuscular junction. Int. Rev. Neurobiol. 75, 267–285 (2006). 42. Shen, K., Fetter, R.D. & Bargmann, C.I. Synaptic specificity is generated by the synaptic guidepost protein SYG-2 and its receptor, SYG-1. Cell 116, 869–881 (2004). 43. Ackley, B.D. & Jin, Y. Genetic analysis of synaptic target recognition and assembly. Trends Neurosci. 27, 540–547 (2004). 44. Dickman, D.K., Lu, Z., Meinertzhagen, I.A. & Schwarz, T.L. Altered synaptic development and active zone spacing in endocytosis mutants. Curr. Biol. 16, 591–598 (2006). 45. Klopfenstein, D.R., Tomishige, M., Stuurman, N. & Vale, R.D. Role of phosphatidylinositol(4,5)bisphosphate organization in membrane transport by the Unc104 kinesin motor. Cell 109, 347–358 (2002). 46. Sato-Yoshitake, R., Yorifuji, H., Inagaki, M. & Hirokawa, N. The phosphorylation of kinesin regulates its binding to synaptic vesicles. J. Biol. Chem. 267, 23930–23936 (1992). 47. Takamori, S. et al. Molecular anatomy of a trafficking organelle. Cell 127, 831–846 (2006). 48. Miller, K.E. et al. Direct observation demonstrates that Liprin-a is required for trafficking of synaptic vesicles. Curr. Biol. 15, 684–689 (2005). 49. Saitoe, M., Schwarz, T.L., Umbach, J.A., Gundersen, C.B. & Kidokoro, Y. Absence of junctional glutamate receptor clusters in Drosophila mutants lacking spontaneous transmitter release. Science 293, 514–517 (2001). 50. Prokop, A. & Technau, G.M. in Cellular Interactions in Development: A Practical Approach (ed. Hartley, D.) 33–57 (Oxford Univ. Press, London and New York, 1993).

989