Presynaptic Spectrin Is Essential for Synapse Stabilization

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Current Biology, Vol. 15, 918–928, May 24, 2005, ©2005 Elsevier Ltd All rights reserved. DOI 10.1016/j.cub.2005.04.030 Presynaptic Spectrin Is Essential for Synapse Stabilization

Jan Pielage,1 Richard D. Fetter,2 aptic nerve terminal. The basic unit of the spectrin skel- and Graeme W. Davis1,* eton is a heterotetramer composed of α- and β-Spectrin 1Department of Biochemistry and Biophysics subunits. These heterotetramers can interact with short Program in Neuroscience actin filaments to form a spectrin-actin filamentous net- University of California, San Francisco work that is localized to the plasma membrane [5]. San Francisco, California 94143 Spectrin can directly associate with membrane phos- 2 Laboratory of Neural Circuits and Behavior pholipids and may provide support for cell shape in this The Rockefeller University manner [5]. The spectrin skeleton has also been termed 1230 York Avenue a “protein accumulation machine” because of its in- New York, New York 10021 volvement in ion-channel, cell-adhesion-molecule, and adaptor-protein localization [6]. In the nervous system, the spectrin skeleton is abun- Summary dant both pre- and postsynaptically at both central and peripheral synapses [5, 7, 8]. However, little is known Background: Precise neural circuitry is established and regarding the function of spectrin in the postembryonic maintained through a regulated balance of synapse nervous system owing to the embryonic lethality of mu- stabilization and disassembly. Currently, little is known tations in most spectrin isoforms. In Drosophila, null about the molecular mechanisms that specify synapse mutations in a- or b-spectrin are late embryonic/early stability versus disassembly. first instar lethal [9, 10]. As a result, prior work has been Results: Here, we demonstrate that presynaptic spectrin restricted to the newly formed NMJ, where spectrin is is an essential scaffold that is required to maintain syn- present both pre- and postsynaptically [11]. Surpris- apse stability at the Drosophila neuromuscular junction ingly, no change in synapse morphology, synapse ultra- (NMJ). Loss of presynaptic spectrin leads to synapse structure, or receptor clustering was observed [11]. disassembly and ultimately to the elimination of the Similar conclusions were made in C. elegans [12]. NMJ. Synapse elimination is documented through light- These prior studies characterized the early develop- level, ultrastructural, and electrophysiological assays. mental requirements of the spectrin skeleton during These combined assays reveal that impaired neuro- synapse formation, but they were unable to assess the transmission is secondary to synapse retraction. We function of spectrin at mature synapses in the postem- demonstrate that loss of presynaptic, but not postsyn- bryonic nervous system. Here, we document the use of aptic, spectrin leads to the disorganization and elimina- transgenic dsRNA reagents that allow us to eliminate tion of essential synaptic cell-adhesion molecules. In spectrin protein selectively during postembryonic de- addition, we provide evidence of altered axonal trans- velopment from either the pre- or postsynaptic side of port and disrupted synaptic microtubules as events the synapse. We show that the presynaptic spectrin that contribute to synapse retraction in animals lacking skeleton is essential during postembryonic develop- presynaptic spectrin. ment for the maintenance of the Drosophila NMJ. Conclusions: Our data suggest that presynaptic spectrin functions as an essential presynaptic scaffold that Results may link synaptic cell adhesion with the stabilization of the underlying microtubule cytoskeleton. We have generated transgenic animals that allow tissue-specific expression of either a- or b-spectrin dsRNA Introduction with the GAL4/UAS expression system (see Experimen- tal Procedures). It has been clearly demonstrated that Throughout the nervous system, the development of transgenically expressed dsRNA can function cell au- neural circuitry involves exuberant synaptogenesis fol- tonomously in Drosophila and is highly sequence spe- lowed by the selective elimination of a subset of these cific [13–15]. We expressed either UAS-a-spectrin previously functional synapses [1–4]. Currently, we dsRNA or UAS-b-spectrin dsRNA in the embryo with know very little about the mechanisms that stabilize cell-type-specific GAL4 drivers and observed rapid, synaptic connections and equally little regarding the specific, and cell-autonomous knockdown of spectrin mechanisms that disassemble previously stable, func- mRNA and protein (data not shown). We observe effi- tional synaptic connections [2]. One way to approach cient knockdown of mRNA and protein in w6hrin this problem is to identify genes that when mutated or these cells. We next tested whether our ability to elimi- knocked down cause the inappropriate disassembly of nate spectrin protein with dsRNA can recapitulate the a previously stable synapse. This approach will identify late-embryonic/early-larval lethality observed in a- and genes necessary for synapse stabilization and may b-spectrin null mutations [9, 10]. Ubiquitous expression identify candidate genes involved in neurodegenera- of a- or b-spectrin dsRNA leads to lethality at early tive disease. stages of larval development. Thus, we nearly recapitu- Here, we demonstrate that the presynaptic spectrin late the null mutant lethal phenotype, and the slight dif- skeleton is essential for the stabilization of the presyn- ference in lethal phase can likely be attributed to the delay in dsRNA expression in comparison to the effects *Correspondence: [email protected] of a null mutation that blocks zygotic gene expression. Spectrin Stabilizes the Synapse 919

Figure 1. Selective Elimination of α- and β-Spectrin at the NMJ (A–C) A wild-type synapse at muscle 4 is stained with anti-β-Spectrin ([A], green) and anti-HRP ([B], red). The merged image is shown in (C). β-Spectrin staining is detected in the presynaptic motoneuron axon ([A] and [B], arrows) and in muscle. (D–F) Expression of UAS-b-spectrin dsRNA in the presynaptic neuron (sca-GAL4) results in the elimination of presynaptic β-Spectrin protein. Staining is no longer observed in the presynaptic motoneuron axon ([D], arrowhead) that can be identified by anti-HRP staining ([E], arrowhead). (G–I) Expression of UAS-b-spectrin-dsRNA in the postsynaptic muscle (BG57-GAL4) results in the elimination of β-Spectrin protein from the postsynaptic muscle ([G] and [H], asterisks), including sites where the presynaptic nerve terminal remains ([H], asterisk). (J and K) Synaptic stainings are shown at higher exposures. Presynaptic α-Spectrin staining can be observed after elimination of postsynaptic α- Spectrin. Presynaptic α-Spectrin staining is continuous with axonal staining prior to entry into the muscle field ([J] and [K], arrowheads). α-Spectrin staining is also contained within the presynaptic-membrane boundary defined by anti-HRP ([J] and [K], arrow).

Neuronal elimination of either α-orβ-Spectrin causes to visualize presynaptic spectrin. In Figures 1A–1I, lethality at late larval stages, with a small portion of ani- staining intensities are optimized to visualize protein mals surviving to early adult stages. Thus, both a- and levels at the wild-type NMJ that has high levels of b-spectrin are essential neuronal genes. spectrin expression. At longer exposure lengths, presynaptic spectrin immunoreactivity can be clearly ob- Analysis of the Postembryonic Synaptic served in the absence of postsynaptic spectrin protein Spectrin Skeleton (Figures 1J and 1K). Although we cannot formally rule Three spectrin genes are present in the Drosophila ge- out that some of this staining is postsynaptic, several nome [9, 10, 16]. At the larval NMJ, both α- and β-Spec- lines of evidence suggest that this staining is primarily trin are highly abundant and colocalize (Figure 1; see presynaptic. First, in these examples, α-Spectrin stain- also [11]) whereas βH-Spectrin is absent (data not ing at the nerve terminal does not extend beyond the shown). Postsynaptic α- and β-Spectrin can be ob- boundary of the presynaptic-membrane marker anti- served throughout the muscle and are concentrated to HRP (Figures 1J and 1K, arrows). In addition, the ob- the postsynaptic membranes surrounding the NMJ served α-Spectrin staining is continuous with staining (Figure 1). Presynaptic α- and β-Spectrin can be ob- that originates within the presynaptic axon prior to en- served in the motoneuron axon (Figure 1), and we pre- try into the muscle (Figures 1J and 1K, arrowheads). sent evidence below that this staining continues into the presynaptic boutons (Figures 1J and 1K). Loss of Presynaptic Spectrin Leads to Synapse Using the UAS-spectrin dsRNA approach, we can Disassembly and Elimination eliminate either α-Spectrin or β-Spectrin protein specif- Loss of presynaptic spectrin (either α-orβ-Spectrin) ically from either the pre- or postsynaptic side of the leads to presynaptic retraction and synapse elimination neuromuscular junction (Figures 1D–1I). Our ability to at the NMJ. We have previously published an assay for selectively eliminate postsynaptic spectrin enables us synapse disassembly at the Drosophila NMJ [17]. This Current Biology 920

assay is based on the observation that the gradual as- trin dsRNA presynaptically. We have compared the sembly of the postsynaptic muscle membrane folds number and severity of synapse retraction to both the (subsynaptic reticulum [SSR]) and proteins that localize wild-type and animals expressing the DN-Glued con- to the SSR requires the presence of the presynaptic struct presynaptically. It was previously shown that pre- nerve terminal [18, 19]. Therefore, the SSR and proteins synaptic expression of DN-Glued causes an increased that localize to this structure will be present only at rate of synapse retraction that is phenocopied by arp-1 sites where the nerve terminal resides or where it has knockdown [17]. We find that presynaptic elimination recently resided. Presynaptic retraction is more rapid of α-orβ-Spectrin leads to retractions at more than than the disassembly of the SSR (assessed at the light 60% of all synapses, far more than when DN-Glued is and EM levels in [17]). Thus, sites where SSR markers expressed presynaptically (Figure 2E). To assay the se- are present without opposing presynaptic markers rep- verity of synapse retraction, we quantified the number resent sites of presynaptic nerve-terminal retraction of bouton-shaped postsynaptic profiles within each that we have termed a “synaptic footprint” [17]. During synaptic footprint (Figure 2F). For example, the retrac- normal synapse development, a small number of re- tion in Figure 2B (arrow) includes eight boutons. In the traction events are observed, and these generally in- wild-type, synapse retractions never include more than clude only a few synaptic boutons within a larger, sta- two boutons. When DN-Glued is expressed presynapti- ble neuromuscular synapse (Figures 2E and 2F; [17]). cally, more than 90% of retractions encompass fewer Presynaptic expression of either a-spectrin dsRNA or than seven boutons. However, after the loss of presyn- b-spectrin dsRNA causes a dramatic increase in the aptic spectrin, a large fraction of retractions (>40%) in- number and extent of synaptic retraction events in clude at least seven boutons. Furthermore, 8%–11% of comparison to the wild-type (Figure 2). For example, in synapses analyzed show complete synapse elimination Figure 2B, much of the synapse shows wild-type oppo- (8% for b-spectrin and 11% for a-spectrin, n > 120 syn- sition of presynaptic Synapsin (green) with postsynap- apses; Figure 2F). Finally, we see a clear perturbation tic Discs-large (Dlg; red). However, at several sites of bouton morphology at many synapses lacking pre- within this NMJ, Dlg staining is no longer opposed by synaptic α-orβ-Spectrin, although the synapse grows presynaptic Synapsin staining, even though Dlg is clearly to near normal dimensions prior to the onset of syn- organized into a postsynaptic specialization (Figure 2B, apse retraction. These data suggest a role for spectrin arrow). We have used a battery of pre- and postsynap- in bouton formation as well as synapse stabilization. tic markers, including antibodies against synaptic vesicle proteins, presynaptic-membrane markers, cytoplasmic proteins, cell-adhesion molecules, and postsynaptic Electrophysiological Analysis of NMJ Lacking glutamate receptors. We observe similar retraction Presynaptic Spectrin events with each of these different markers (see Figure We have pursued experiments to directly correlate the S1 in the Supplemental Data available with this article extent of synapse retraction with altered synaptic func- online). Synapse disassembly always encompasses the tion. To do so, we fixed each synapse from which we distal-most portions of a synaptic branch and is always recorded and then stained for retraction events with the more severe distally than proximally. Thus, we conclude presynaptic active-zone marker nc82 and postsynap- that synapse retraction progresses distal to proximal tic Dlg. Synapses were categorized morphologically along a nerve-terminal branch and involves the simulta- for the extent of synapse retraction and then corre- neous disassembly of multiple synapse components lated with the electrophysiological data. Synapses were (Figure S1). This phenomenon of “synapse retraction” grouped into three categories: minor retractions pre- may be distinct from the neuronal-arborization degen- sent (<20% of the synapse is retracted), intermediately eration described in the Drosophila central nervous affected (20%–60% of the synapse is retracted), and system [4, 20]. severe retractions (>60% of the synapse is retracted). We not only find synapses with partial synapse re- Synapses of all three categories could be found within traction, we also frequently observe muscles that have single animals. organized postsynaptic Dlg staining and completely We find that altered synaptic efficacy is strongly lack opposing presynaptic markers (Figure 2C). Several correlated with the extent of synapse retraction. In ex- lines of evidence demonstrate that this represents complete synapse elimination. We observe these events with treme examples of morphologically defined synapse different presynaptic markers, including synaptic-vesi- elimination, both spontaneous and evoked transmitter cle, active-zone, and cytoplasmic markers (data not release are abolished, although passive membrane shown). Only one or two muscles within an animal will properties of the muscle remain within a wild-type range show an absence of presynaptic staining, and this is (n = 2; resting membrane potential = −69 mV and –73 Ω Ω never seen in wild-type animals stained in parallel. At mV; input resistance = 11 M and8M ). Further analy- sites lacking presynaptic antigens, Dlg remains highly sis demonstrates that the severity of synapse retraction organized into postsynaptic specializations. Finally, at after the loss of either presynaptic α-orβ-Spectrin cor- these sites of synapse elimination, we have observed relates with a decreased frequency of spontaneous discontinuous presynaptic membrane remnants, which vesicle release and decreased quantal content (Figures confirm that the presynaptic nerve terminal once re- 2G and 2H). However, the amplitude of spontaneous sided in opposition to these residual postsynaptic spe- miniature release events does not change with in- cializations (data not shown). creased severity of synapse retraction (Figures 2G and We have quantified the phenotype of synapse retrac- 2H). In addition, we show that there is no significant tion in animals expressing either a-spectrin or b-spec- change in the diameter of presynaptic vesicles at these Spectrin Stabilizes the Synapse 921

Figure 2. Loss of Presynaptic Spectrin Leads to Synapse Retraction and Elimination (A) A wild-type synapse is stained with anti- Synapsin (green) and anti-Dlg (red). Synap- sin opposes Dlg throughout the NMJ. (B) A synapse lacking presynaptic α-Spec- trin shows synapse-retraction sites where postsynaptic Dlg is no longer opposed by presynaptic anti-Synapsin staining (arrow). (C) An example of synapse elimination in which no Synapsin is observed opposed to postsynaptic Dlg that is clearly organized into a postsynaptic specialization. In this animal, other synapses stained effectively for Synapsin. (D) A synapse lacking presynaptic β-Spectrin is stained with the active-zone marker nc82 and anti-Dlg. Significant regions of the synapse lack nc82 despite robust postsynaptic Dlg staining. (E) The percentage of all muscle 6/7 synapses that show evidence of synapse retraction is plotted for each of the genotypes indicated: DN-Glued = elav-GAL4, DN-Glued/+ (n = 124 synapses); a-spectrin RNAi = elav- GAL4, 2× UAS-a-spectrin-dsRNA (n = 127); b-spectrin RNAi = sca-GAL4, 2× UAS- b-spectrin-dsRNA (n = 120); and wild-type (n = 130). Loss of presynaptic spectrin causes a significantly higher rate of synapse retraction in comparison to the wild-type (p < 0.001 for a- and b-spectrin) and DN- Glued (p < 0.001 for a- and b-spectrin). p val- ues were determined with a test of propor- tions analysis with a confidence interval of 0.01. (F) Synapses showing retractions are grouped according to the severity of the retraction and plotted as a percentage of the total number of synapses that show retraction in each of the genotypes indicated. Synapse retractions caused by loss of presynaptic spectrin are larger than those observed in the wild-type or when DN-Glued is driven presynaptically. Genotypes and samples sizes are as in (E). (G and H) Altered synaptic function parallels synapse retraction in animals lacking presynaptic spectrin. Synapses were categorized on the basis of the retraction phenotype and grouped into three categories (<20% retracted, 20%–60% retracted, and >60% retracted; see text for additional de- tail). Data are presented for synapses lacking presynaptic α-Spectrin (G) and presynaptic β-Spectrin (H). In both genotypes, increased severity of retraction correlates with a decrease in mepsp frequency and quantal content (* indicates a significant difference of p < 0.01). There is no difference in the average mepsp amplitude (p > 0.1). Error bars represent the SEM.

synapses, defined ultrastructurally (wt = 29.12 nm ± aptic spectrin is severely depleted prior to synapse re- 0.11; presynaptic a-spectrin dsRNA = 30.19 nm ± 0.61; traction. Therefore, the tight correlation between syn- p = 0.13; see Experimental Procedures for measure- apse retraction and synaptic efficacy suggests that ment). Because quantal content is impaired without a release changes observed in the absence of presynap- change in quantal size, these data indicate that the pre- tic spectrin are a secondary consequence of synapse synaptic release machinery is disassembled prior to the disassembly and most likely do not reflect a direct ac- loss of postsynaptic receptors [3, 17]. Finally, presyn- tivity of spectrin during the vesicle release process. Current Biology 922

Figure 3. Ultrastructural Analysis of Synapses Lacking Presynaptic α-Spectrin Cross-sections of synaptic boutons are shown. The postsynaptic muscle membrane folds are indicated in (A)–(C) (SSR). Active zones that include presynaptic t-bars are marked by asterisks. (A) A wild-type bouton is shown. (B) A bouton from an animal lacking presynaptic α-Spectrin shows depletion of presynaptic vesicles and a mild disruption of the postsynaptic SSR. (C) A separate bouton cross-section from the same genotype as (B) is shown in which the SSR is severely disrupted. The SSR is less compact and is invaded by muscle tissue. Two small-caliber boutons are present (b). Presynaptic vacuoles are indicated (arrows). (D) A higher magnification of a wild-type bouton that includes an active zone with a t-bar (asterisks) as well as a putative active zone without a t-bar (arrow; identified by increased pre- and postsynaptic electron density). Vesicles are highly concentrated to these active zones. (E) A portion of the image in (B) (bracket) is shown at a higher magnification. This region has a similar active-zone organization (arrow and asterisks) as in (D). Synaptic-vesicle density is severely decreased. A vacuole is present (feathered arrow). Note that the residual vesicles still cluster at the active zone with a t-bar (asterisk). The scale represents 400 nm in (A)–(C) and 200 nm in (D) and (E).

Loss of Postsynaptic Spectrin Does Not Cause is consistent with our light-level observation that pre- Synapse Retraction synaptic-membrane markers persist after the loss of We have determined whether postsynaptic spectrin is synaptic-vesicle antigens. In other examples, small-cal- required for synapse stabilization. We specifically elimi- iber presynaptic boutons are observed that contain nated postsynaptic α-orβ-Spectrin postembryonically vacuolar profiles (Figures 3C and 3E, feathered arrows). with muscle-specific GAL4 drivers (see Experimental In this example, the surrounding SSR is dramatically Procedures). We are able to efficiently eliminate muscle perturbed, with muscle folds becoming disorganized spectrin (Figures 1G, 1J, and 1K). Synapses lacking (Figure 3C). The perturbation of the SSR includes the muscle spectrin show obvious defects in the integrity invasion of the SSR by muscle-cell cytoplasm and of the subsynaptic muscle membrane folds (J.P. and mitochondria, events that are never observed in the G.W.D., unpublished data). However, we find no evi- wild-type. These data are consistent with the retraction dence of synapse retraction beyond that observed in of the presynaptic nerve terminal and then the gradual wild-type animals with our footprint assay (n = 120 syn- disassembly of the postsynaptic specialization. apses). Thus, spectrin is uniquely required in the presynaptic nerve terminal for synapse stabilization. Disrupted Presynaptic Microtubule Organization and Stability at Synapses Lacking Presynaptic Spectrin Ultrastructural Analysis of Synapse Retraction The presynaptic microtubule (MT) cytoskeleton can be at NMJ Lacking Presynaptic ␣-Spectrin visualized by staining for the Map1b-like protein Futsch Synaptic boutons lacking presynaptic α-Spectrin show [21, 22]. At wild-type synapses, a filament of Futsch a variety of ultrastructural phenotypes that are consis- staining extends the entire length of the nerve terminal. tent with synapses undergoing retraction. Synaptic Within the center of the nerve terminal, the Futsch-pos- vesicle density is often dramatically reduced (Figure itive MTs form a well-defined core that occasionally 3B). Although active zones can still be resolved, only a forms loop structures (Figure 4A; [22]). At synapses few synaptic vesicles remain clustered at these sites lacking presynaptic spectrin, the presynaptic MT cy- (Figures 3B and 3E, asterisks). The loss of vesicles at toskeleton is severely perturbed. The MTs pull back at sites where the presynaptic bouton membrane persists sites of presynaptic retraction on the basis of the ab- Spectrin Stabilizes the Synapse 923

asterisk). At these sites, spectrin protein levels are severely depleted or absent. These data demonstrate that the presynaptic spectrin skeleton is necessary to maintain the proper organization and integrity of the MT cytoskeleton throughout the nerve terminal. Spectrin is also present throughout the sensory and motor axons. Neuronal expression of either a- or b-spectrin dsRNA leads to minor changes in axonal microtubule organization, but it causes an axonal-blockage accumulation consistent with a disruption of microtubule-based axonal transport (Figures 4D–4F; [23]). Spectrin may participate in axonal transport via the reg- ulation of microtubule organization, or it could function directly to link cargo to motor-protein complexes [24, 25]. Disrupted axonal transport could contribute to the severity of synapse retraction observed in spectrin knockdown animals.

Spectrin Organizes and Stabilizes the Synaptic Cell- Adhesion Molecules Neuroglian and Fasciclin II A membrane-associated spectrin scaffold could orga- nize cell-adhesion molecules at the synaptic plasma membrane to control synapse stability. Neuroglian (Nrg) is a Drosophila L1 homolog that has been shown to be involved in axon pathfinding [26, 27]. Nrg, like L1, can be linked to the spectrin skeleton via Ankyrin [5, 28]. A role for Nrg at the NMJ has yet to be defined. Fasciclin II (Fas II) is a homophilic cell-adhesion molecule that is necessary, both pre- and postsynaptically, for the postembryonic growth and stabilization of the Drosophila NMJ [29]. We first addressed whether Nrg and Fas II are lost at sites of synapse retraction. We triple-labeled synapses with anti-HRP, anti-Dlg, and either anti-Nrg or anti-Fas II. In these experiments, we used an antibody that rec- ognizes the neuronal-specific isoform of Nrg. At wild- Figure 4. Loss of Presynaptic Spectrin Results in Disorganization and Retraction of Microtubules and Axonal Transport Defects type synapses, the cell-adhesion molecules are distrib- (A) A wild-type synapse stained with anti-Futsch (green) and anti- uted throughout the NMJ (Figures 5A–5C and 5G–5I). Dlg (red). The insets next to the image (from the region indicated However, after loss of presynaptic α-orβ-Spectrin, we by the arrow in [A]) show that presynaptic anti-Futsch staining (up- find sites where HRP is present and directly opposed per inset) extends into the terminal boutons of the synapse (arrow). to postsynaptic Dlg, but where Nrg or Fas II are absent (B) A synapse lacking presynaptic α-Spectrin is stained as in (A). (Figures 5D–5F and 5J–5L, arrows). Thus, cell-adhesion The presynaptic microtubules identified by Futsch staining are molecules are removed at sites of retraction prior to the thicker and do not extend to the distal end of the synaptic bouton chain (arrowhead). The insets show a higher-magnification view of removal of the presynaptic membrane. Because Fas II the terminal boutons indicated by the arrowhead in (B). staining will identify both pre- and postsynaptic protein, (C) A synapse lacking presynaptic α-Spectrin shows severely disor- these data suggest that trans-synaptic cell adhesion ganized presynaptic microtubules that fill an entire bouton (aster- mediated by Fas II is disrupted prior to the retraction isks). The insets show higher-magnification images of the region of the presynaptic membrane. These data indicate that indicated by the asterisk. the losses of Nrg and Fas II are not a secondary conse- (D–F) Axons of the larval peripheral nervous system are stained for microtubules (Futsch, red) and the synaptic-vesicle-associated quence of membrane removal, but could be early protein Synaptotagmin (green). (D) In wild-type axons, only very events in the destabilization of the presynaptic ter- few and small puncta of Synaptotagmin staining can be detected minal. Additional data support this model. By triple- (arrow). Presynaptic elimination of α-Spectrin (E) or β-Spectrin (F) labeling the NMJ for Fas II, the peri-active-zone protein results in large accumulations of Synaptotagmin-positive vesicles Nervous wreck (Nwk; [30]), and postsynaptic Dlg we in the axons, indicating defects in axonal transport ([E] and [F], were able to demonstrate that Fas II protein can be arrows). eliminated before the loss of other synaptic components (data not shown). Finally, because anti-HRP rec- sence of Futsch staining (Figure 4B, arrowhead). The ognizes additional proteins resident in the neuronal MTs are also perturbed throughout the nerve terminal, plasma membrane [31], it suggests some selectivity in even at sites that have intact pre- and postsynaptic the disruption of protein stability after loss of presynap- specialization and have yet to undergo synapse disas- tic spectrin. sembly. In these regions of the nerve terminal, the MTs We next analyzed Fas II organization at sites within appear thicker and disorganized (Figures 4B and 4C, retracting synapses that still show clear opposition of Current Biology 924

Figure 5. Loss of Presynaptic Spectrin Results in Disorganization and Elimination of the Cell-Adhesion Molecules Nrg and Fas II at Synapses Undergoing Retraction (A–C) A wild-type synapse from muscle 6/7 is shown stained for Nrg (A), HRP (B), and Dlg (C). All three makers colocalize throughout the NMJ. (D–F) A muscle 6/7 synapse lacking presynaptic β-Spectrin is stained as in (A)–(C). At sites within this synapse (arrows), presynaptic Nrg staining is absent from synaptic boutons that retain anti-HRP and Dlg staining. (G–I) Wild-type synapses, stained for Fas II (G), HRP (H), and Dlg (I), that colocalize throughout the NMJ. (J–L) A synapse lacking presynaptic α-Spectrin shows that Fas II staining is absent from synaptic boutons that retain anti-HRP and Dlg staining (arrows). (M–R) Fasciclin II is disorganized at semi-stable regions of synapses lacking presynaptic spectrin. (M–O) A wild-type synapse from muscle 6/7 is shown stained for Fas II (green), Nwk (red), and Dlg (blue). The higher magnifications (boxed area in [M]) show that Fas II is organized in a honeycomb-like network that surrounds presynaptic Nwk staining ([N] and [O], arrowheads; enlarged view of boxed area from [M]). (P–R) A muscle 6/7 synapse that lacks presynaptic β-Spectrin is stained as in (M)–(O). Regions of this synapse show simultaneous disassembly of the two presynaptic markers Fas II and Nwk ([P], arrow). At semi-stable regions of this synapse (boxed area in [P]), Fas II is no longer organized into a honeycomb-like structure. The staining becomes punctate and no longer completely surrounds presynaptic Nwk staining ([Q] and [R], arrowheads; enlarged view of boxed area from [P]). pre- and postsynaptic markers. These regions likely re- intact active zones show accumulations of vacuoles present sites that are in the early stages of synapse and multivesicular bodies indicative of ongoing nerve- disassembly, and we refer to these sites as “semi-sta- terminal retraction [17]. In the experiments presented ble” regions of the synapse. Consistent with this defini- here, we triple-labeled the synapse with anti-Fas II, tion of semi-stable regions of the synapse, prior serial anti-Nwk, and anti-Dlg. At wild-type synapses, Fas II is EM reconstructions provided evidence of graded lev- organized into a honeycomb-like network that is pre- els of synapse disassembly along the nerve-terminal sent both pre- and postsynaptically (Figures 5M–5O; length [17]. For example, regions of the synapse with [29, 32]). Fas II staining at the bouton periphery circum- Spectrin Stabilizes the Synapse 925

scribes the cytoplasmic marker Nwk (Figure 5N, arrow- embryo, and a severe deficit in synaptic function was heads), consistent with the pre- and postsynaptic observed. The deficit in synaptic function was attrib- membrane localization of Fas II. We then examined Fas uted to perturbed localization of synaptic proteins be- II staining at synapses that lacked presynaptic α-or cause synapse morphology and vesicle distribution at β-Spectrin and showed clear evidence of ongoing pre- the active zone were normal [11]. We show evidence of synaptic retraction (Figure 5P, arrow). Within semi-sta- impaired axonal transport, and this could be the cause ble regions of these synapses, Fas II staining is no of the altered protein distribution previously observed longer organized in a honeycomb-like pattern and in- in the embryo. We can only speculate as to why no stead appears to be broken into large puncta (Figures synapse retractions were observed in the embryo, even 5P–5R). In addition, Fas II at the bouton periphery no at the ultrastructural level. One possibility is that the longer clearly circumscribes the underlying anti-Nwk highly dynamic, newly formed synapse does not rely staining (Figure 5Q, arrowheads). These data are con- heavily upon the stabilizing scaffold provided by the sistent with the removal of Fas II from the plasma mem- presynaptic spectrin skeleton. This is consistent with brane in synapse regions that are likely to be in the the observation that synapse formation occurs in the early stages of disassembly. Similar observations can absence of the stabilizing cell-adhesion molecule Fas- be made for Neuroglian (data not shown). ciclin II. In fasciclin II null mutations, impaired synaptic Finally, it has been demonstrated genetically that Fas growth and synapse loss are only observed at later II is required both pre- and postsynaptically for synapse stages of development [29]. maintenance [29]. However, we find that loss of postsynaptic α-orβ-Spectrin does not lead to synapse dis- Functional Analysis of Synapse Elimination assembly. We therefore assayed whether loss of post- in the Absence of Presynaptic Spectrin synaptic spectrin affects the stability of synaptic Fas II. In our functional analysis of synapse elimination, we Although the organization of Fas II appears disrupted, it correlate the extent of synapse retraction with impaired is still concentrated at the synaptic plasma membrane, synaptic function. We observe a tight correlation be- indicating that homophilic adhesion remains intact tween the extent of synapse retraction (with active- throughout the synapse in the absence of postsynaptic zone markers) and physiological deficits. These data spectrin (Figure S2). In addition, Fas II staining still cir- are consistent with synapse disassembly gradually de- cumscribes the presynaptic marker Nwk at these syn- pleting the nerve terminal of functional active zones apses (Figure S2). These data imply a unique relation- (Figure S1). Importantly, because synapses lacking pre- ship between spectrin and Fas II that is required for synaptic spectrin (with only minor retractions) show synapse stabilization within the presynaptic nerve ter- normal neurotransmitter release, it argues that spectrin minal. One possibility is the existence of a neuronal- is not necessary for presynaptic vesicle fusion at the specific adaptor protein that could link the presynaptic Drosophila larval NMJ. These data differ from prior spectrin skeleton to Fas II. studies at mammalian central synapses with function- blocking antibodies that perturbed the spectrin-Synap- Discussion sin interaction and provided evidence that spectrin is necessary for vesicle release [42, 43]. Our data demon- Here, we demonstrate that the presynaptic spectrin strating a precise structural and functional correlation skeleton is essential for the stabilization of the Dro- in the absence of presynaptic spectrin also argue that sophila NMJ. On the basis of the data presented, we synapse retraction is not a secondary consequence of propose that the spectrin skeleton provides a stabiliz- impaired motoneuron health. A precise structural and ing bridge linking cell adhesion at the plasma mem- functional correlation would not be expected if altered brane to the underlying microtubule cytoskeleton. The motoneuron health were leading the way to cata- observed disruption of axonal transport after the loss strophic synapse elimination. of presynaptic spectrin may also contribute to synapse retraction by disrupting the retrograde transport of Presynaptic Spectrin Organizes and Stabilizes essential trophic signaling molecules [33]. Importantly, trans-Synaptic Cell Adhesion disruption of dynactin-mediated axonal transport has Synapse stabilization requires sustained trans-synaptic been shown to cause synapse retraction in Drosophila cell adhesion [2, 44]. At the Drosophila NMJ, the homo- [17] and has been implicated in motoneuron disease in philic cell-adhesion molecule Fasciclin II is necessary humans and mice [34–37]. Similarly, ßIV-spectrin mu- both pre- and postsynaptically for the maintenance of tant mice show deficits consistent with impaired func- the NMJ [29]. Prior genetic analysis has shown that the tion of the neuromuscular system [38–41]. These data NMJ is destabilized in fasciclin II null mutations, and highlight the utility of Drosophila for the identification synapse stability can only be restored by resupplying of genes involved in the stability of the neuromuscular fasciclin II simultaneously on both sides of the synapse system in other species. [29]. The role of Nrg during synapse stabilization has Our approach, using expression of transgenic dsRNA, yet to be determined. has allowed us to selectively perturb spectrin at a ma- Here, we provide evidence that the presynaptic spec- ture neuromuscular synapse. Prior work examining null trin skeleton is essential for the normal organization mutations in a- and b-spectrin at the embryonic NMJ and maintenance of both Fas II and Nrg at the synapse. did not reveal any evidence of synapse retraction at the In the absence of presynaptic α-orβ-Spectrin, Fas II light or ultrastructural level [11]. Changes in the distri- and Nrg are eliminated at sites of synapse retraction bution of synaptic proteins were observed in the prior to the retraction of the presynaptic membrane, in- Current Biology 926

dicating that the loss of cell adhesion is an early event not contain significant homology to any other Drosophila gene (and during the process of synapse disassembly (Figure 5). in particular to any other spectrin) to ensure specificity of the re- In addition, when we examine sites of presumed ongo- sulting dsRNA. We introduced XbaI restriction sites (underlined) to allow direct cloning into the pWIZ vector [13]. ing synapse disassembly, we find that the Nrg and Fas For a-spectrin, a 587 bp fragment starting at position 430 of the II proteins are grossly disorganized. The punctate ap- a-spectrin cDNA was amplified with the primers 5#-CAGTACTT pearance of Fas II in these regions and the eventual CAATCTAGATGCCGACGAGTTGGA-3# and 5#-AGCTTCTCTAGACG elimination of Fas II suggest that Fas II is destabilized CACAATGGCTAACTG-3#. For b-spectrin, a 544 bp fragment start- and internalized at sites of ongoing synapse retraction. ing at position 1687 of the b-spectrin cDNA was amplified with the # Because elimination of postsynaptic spectrin does not primers 5 -GGGCGGGTTCTAGAAGCCATCGAAACCGACATCTT CGC-3# and 5#-GGGCGGGTTCTAGAACTGCCAGAGTTTGCGGCT cause elimination of Fas II, additional motoneuron-spe- CTCC-3#. DNA fragments were amplified from wild-type genomic cific factors are likely to be involved in the proposed DNA and cloned into pWIZ according to [13]. Both constructs were spectrin-mediated stabilization of Fas II at the presyn- confirmed by sequencing. Transgenic flies were generated by stan- aptic nerve terminal. In Drosophila and vertebrate sys- dard methods. In each case, at least two independent transgene tems, Spectrin can be linked to Nrg (L1) in the plasma insertions were established for chromosomes 2 and 3. membrane via the adaptor protein Ankyrin [5, 28]. The association of Nrg (L1) to Ankyrin can be regulated by Immunocytochemistry Wandering third-instar larvae were dissected in HL3 saline and tyrosine phosphorylation, suggesting that this could be fixed with 4% formaldehyde/PBS for 7 min except for stainings for a mechanism to regulate synapse stability [45]. How- D-GluRIIA, in which larvae were fixed in Bouin’s fixative (Sigma) for ever, the function of Ankyrin at this NMJ has yet to be 1 min. Primary antibodies were applied at 4°C overnight. Primary determined. antibodies were used at the following dilutions: anti-Synapsin (3C11), 1:50; nc82, 1:50 (both antibodies gifts from E. Buchner); anti-Dlg (4F3), 1:50; anti-α-Spectrin (3A9), 1:50; anti-D-GluRIIA Spectrin Is Necessary for the Organization (8B4D2), 1:10; anti-Fasciclin II (1D4), 1:10; anti-Futsch (22C10), and Integrity of the Presynaptic 1:50; anti-Neuroglian (BP104), 1:10 (all provided by the Develop- Microtubule Cytoskeleton mental Studies Hybridoma Bank, Iowa); rabbit polyclonal anti-Dlg, The retraction of the microtubule cytoskeleton is a 1:5.000 (gift from V. Budnik); rabbit polyclonal anti-α-Spectrin, β common underlying feature of synapse elimination at 1:500; rabbit polyclonal anti- -Spectrin, 1:500; rabbit polyclonal anti-β -Spectrin, 1:500 (all spectrin antibodies were gifts from R. both the vertebrate and Drosophila NMJ [1, 17]. Retrac- H Dubreuil); rabbit polyclonal anti-Synaptotagmin, 1:500; rabbit poly- tion of MTs is also thought to be an early event in the clonal anti-Dap160, 1:200; and rat polyclonal anti-Nervous wreck, pruning of axons in the Drosophila CNS [4]. Although 1:1000 (gift from B. Ganetzky). All secondary antibodies and Cy3- signaling molecules that influence the organization of and Cy5-conjugated anti-HRP were obtained from Jackson Immu- synaptic microtubules have been identified [22, 46, 47], noresearch Laboratories and Molecular Probes, used at a 1:200– the mechanisms that normally stabilize synaptic MTs to 1:1.000 dilution, and applied for 1–2 hr at room temperature (RT). Images were captured with an Axiovert 200 (Zeiss) inverted micro- prevent MT retraction remain to be defined. The spec- scope and a cooled CCD camera. Hardware and software analysis trin skeleton appears to be essential for the normal were driven by Intelligent Imaging Innovations (3I) software. maintenance of a stable presynaptic MT cytoskeleton. In support of this observation, it has been demon- Electrophysiology strated in Drosophila that both α- and β-Spectrin can Third-instar larvae were selected and dissected according to pre- bind to microtubules [48]. It is also possible, however, viously published techniques [17, 32]. Whole-muscle recordings that spectrin directly stabilizes the synaptic actin cy- were performed on muscle 6 in abdominal segment A3 with sharp Ω toskeleton and this in turn influences the stability of microelectrodes (12–16 M ). Recordings were selected for analysis only with resting membrane potentials more hyperpolarized than synaptic MTs [5]. −60 mV and with input resistances greater than 5 MΩ. The average spontaneous miniature excitatory postsynaptic potential (mepsp) Experimental Procedures amplitude was quantified by measuring the amplitude of approxi- mately 100–200 individual spontaneous release events per syn- Fly Stocks apse. The average per-synapse mepsp amplitudes were then Flies were maintained at 25°C on normal food. The following strains averaged for each genotype. Measurement of mepsp amplitudes were used in this study: w1118 (wild-type), elavC155-GAL4 (neuron was semiautomated (Synaptosoft). The average superthreshold- specific), sca-GAL4 (neuron specific), BG57-GAL4 (muscle expres- evoked EPSP amplitude was calculated for each synapse, ensuring sion from mid-first-instar on), da-GAL4 (ubiquitous expression), that both motor axons innervating muscle 6 in segment A3 were UAS-DN-Glued84, and MHC-Shaker-GFP. We experimentally deter- recruited. Quantal content was calculated as the average EPSP mined the combination of neuronal GAL4 drivers and UAS-dsRNA amplitude divided by the average mepsp amplitude. Quantal insertions necessary to eliminate presynaptic α-orβ-Spectrin pro- content was determined for each synapse and then averaged tein. Two copies of UAS-a-spectrin-dsRNA were driven by elavC155- across synapses to generate the average quantal content for each GAL4 to eliminate presynaptic expression of α-Spectrin. A higher genotype. Data for steady-state synaptic transmission were ac- level of b-spectrin dsRNA expression was necessary to eliminate quired in HL3 saline (0.5 mM extracellular calcium). Data were col- protein, and this was achieved with the stronger neuronal GAL4 lected with a Digidata 1200B analog-to-digital board and PClamp driver line scabrous-GAL4 in combination with two copies of UAS- software (Axon Instruments). Data were analyzed offline with Mini- b-spectrin dsRNA. Postsynaptic elimination of α-orβ-Spectrin pro- Analysis software (Synaptosoft). tein was achieved by driving one copy of the UAS-dsRNA constructs with BG57-GAL4. Electron Microscopy Third-instar larvae, mutant and wild-type, were prepared for Generation of UAS-spectrin-dsRNA Constructs electron microscopy as follows: Larvae were filleted and pinned out and Germline Transformation in physiological saline that was then exchanged with 2% glutaral- We used the pWIZ-Vector [13] to generate UAS-a-spectrin-dsRNA dehyde in 0.12 M Na-cacodylate buffer (pH 7.4). The filleted larvae and UAS-b-spectrin-dsRNA constructs. The primers below were were fixed in place for 10 min, then transferred to vials containing used to amplify fragments of the a- and b-spectrin cDNAs that do fresh fixative and fixed for a total of 2 hr with rotation. The larvae Spectrin Stabilizes the Synapse 927

were rinsed3×15minwith 0.12 M Na-cacodylate buffer and post- 11. Featherstone, D.E., Davis, W.S., Dubreuil, R.R., and Broadie, K. fixed with 1% osmium tetroxide in 0.12 M Na-cacodylate buffer (2001). Drosophila alpha- and beta-spectrin mutations disrupt with rotation. After postfixation, the specimens were rinsed 2 × 15 presynaptic neurotransmitter release. J. Neurosci. 21, 4215– min with 0.12 M Na-cacodylate buffer and 2 × 15 min with water, 4224. and then stained en bloc with 1% aqueous uranyl acetate for 1 hr. 12. Hammarlund, M., Davis, W.S., and Jorgensen, E.M. (2000). Mu- The larvae were rinsed with water, dehydrated with ethanol, passed tations in beta-spectrin disrupt axon outgrowth and sarcomere through propylene oxide, and embedded in Eponate 12 resin. All structure. J. Cell Biol. 149, 931–942. processing steps were done at room temperature. Sections were 13. Lee, Y.S., and Carthew, R.W. (2003). Making a better RNAi vec- cut with a Leica Ultracut T microtome, collected on Formvar- tor for Drosophila: Use of intron spacers. Methods 30, 322–329. coated slot grids, and stained with uranyl acetate and Sato's lead. 14. Van Roessel, P., Hayward, N.M., Barros, C.S., and Brand, A.H. Sections were photographed with a JEOL 1200 EX/II TEM operated (2002). Two-color GFP imaging demonstrates cell-autonomy of at 80 kV. Average synaptic-vesicle diameters were measured for GAL4-driven RNA interference in Drosophila. Genesis 34, vesicles within 250 nm of the active zone. The average vesicle di- 170–173. ameter was determined for each active zone, and then measure- 15. Roignant, J.Y., Carre, C., Mugat, B., Szymczak, D., Lepesant, ments per active zone were averaged for each genotype. J.A., and Antoniewski, C. (2003). Absence of transitive and sys- temic pathways allows cell-specific and isoform-specific RNAi in Drosophila. RNA 9, 299–308. Supplemental Data 16. Thomas, G.H., Zarnescu, D.C., Juedes, A.E., Bales, M.A., Lon- Two supplemental figures are available at http://www.current- dergan, A., Korte, C.C., and Kiehart, D.P. (1998). Drosophila be- biology.com/cgi/content/full/15/10/918/DC1/. taHeavy-spectrin is essential for development and contributes to specific cell fates in the eye. Development 125, 2125–2134. 17. Eaton, B.A., Fetter, R.D., and Davis, G.W. (2002). Dynactin is necessary for synapse stabilization. Neuron 34, 729–741. Acknowledgments 18. Guan, B., Hartmann, B., Kho, Y.H., Gorczyca, M., and Budnik, V. (1996). The Drosophila tumor suppressor gene, dlg,isin- We would like to thank Christian Klämbt, in whose laboratory J. volved in structural plasticity at a glutamatergic synapse. Curr. Pielage generated the UAS-a-spectrin dsRNA flies. We thank Jes- Biol. 6, 695–706. sica Tierney for technical assistance. We thank Benjamin Eaton, 19. Thomas, U., Ebitsch, S., Gorczyca, M., Koh, Y.H., Hough, C.D., Heather Heerssen, and Bruno Marie for critical reading of this Woods, D., Gundelfinger, E.D., and Budnik, V. (2000). Synaptic manuscript. We thank Eric Buchner, Vivian Budnik, Ron Dubreuil, targeting and localization of discs-large is a stepwise process and Barry Ganetzky for antibodies as indicated and Richard Car- controlled by different domains of the protein. Curr. Biol. 10, thew for the pWIZ vector. This work was supported by a grant from 1108–1117. the National Institutes of Health to G.W.D. (NS047342). 20. Watts, R.J., Schuldiner, O., Perrino, J., Larsen, C., and Luo, L. (2004). Glia engulf degenerating axons during developmental Received: February 10, 2005 axon pruning. Curr. Biol. 14, 678–684. Revised: March 24, 2005 21. Hummel, T., Krukkert, K., Roos, J., Davis, G., and Klambt, C. Accepted: April 12, 2005 (2000). Drosophila Futsch/22C10 is a MAP1B-like protein re- Published: May 24, 2005 quired for dendritic and axonal development. Neuron 26, 357– 370. 22. Roos, J., Hummel, T., Ng, N., Klambt, C., and Davis, G.W. References (2000). Drosophila Futsch regulates synaptic microtubule or- 1. Bishop, D.L., Misgeld, T., Walsh, M.K., Gan, W.B., and Licht- ganization and is necessary for synaptic growth. Neuron 26, man, J.W. (2004). Axon branch removal at developing synapses 371–382. by axosome shedding. Neuron 44, 651–661. 23. Gunawardena, S., and Goldstein, L.S. (2001). Disruption of axo- 2. Goda, Y., and Davis, G.W. (2003). Mechanisms of synapse as- nal transport and neuronal viability by amyloid precursor pro- sembly and disassembly. Neuron 40, 243–264. tein mutations in Drosophila. Neuron 32, 389–401. 3. Sanes, J.R., and Lichtman, J.W. (1999). Development of the 24. Holleran, E.A., Ligon, L.A., Tokito, M., Stankewich, M.C., Mor- row, J.S., and Holzbaur, E.L. (2001). beta III spectrin binds to vertebrate neuromuscular junction. Annu. Rev. Neurosci. 22, the Arp1 subunit of dynactin. J. Biol. Chem. 276, 36598–36605. 389–442. 25. Muresan, V., Stankewich, M.C., Steffen, W., Morrow, J.S., Holz- 4. Watts, R.J., Hoopfer, E.D., and Luo, L. (2003). Axon pruning baur, E.L., and Schnapp, B.J. (2001). Dynactin-dependent, dy- during Drosophila metamorphosis: Evidence for local degener- nein-driven vesicle transport in the absence of membrane pro- ation and requirement of the ubiquitin-proteasome system. teins: A role for spectrin and acidic phospholipids. Mol. Cell 7, Neuron 38, 871–885. 173–183. 5. Bennett, V., and Baines, A.J. (2001). Spectrin and ankyrin- 26. Kristiansen, L.V., Velasquez, E., Romani, S., Baars, S., Berezin, based pathways: Metazoan inventions for integrating cells into V., Bock, E., Hortsch, M., and Garcia-Alonso, L. (2005). Genetic tissues. Physiol. Rev. 81, 1353–1392. analysis of an overlapping functional requirement for L1- and 6. Baines, A.J., Keating, L., Phillips, G.W., and Scott, C. (2001). NCAM-type proteins during sensory axon guidance in Dro- The postsynaptic spectrin/4.1 membrane protein “accumula- sophila. Mol. Cell. Neurosci. 28, 141–152. tion machine”. Cell. Mol. Biol. Lett. 6, 691–702. 27. Hall, S.G., and Bieber, A.J. (1997). Mutations in the Drosophila 7. Goodman, S.R., Zimmer, W.E., Clark, M.B., Zagon, I.S., Barker, neuroglian cell adhesion molecule affect motor neuron path- J.E., and Bloom, M.L. (1995). Brain spectrin: Of mice and men. finding and peripheral nervous system patterning. J. Neurobiol. Brain Res. Bull. 36, 593–606. 32, 325–340. 8. Phillips, G.R., Huang, J.K., Wang, Y., Tanaka, H., Shapiro, L., 28. Dubreuil, R.R., MacVicar, G., Dissanayake, S., Liu, C., Homer, Zhang, W., Shan, W.S., Arndt, K., Frank, M., Gordon, R.E., et al. D., and Hortsch, M. (1996). Neuroglian-mediated cell adhesion (2001). The presynaptic particle web: Ultrastructure, composi- induces assembly of the membrane skeleton at cell contact tion, dissolution, and reconstitution. Neuron 32, 63–77. sites. J. Cell Biol. 133, 647–655. 9. Dubreuil, R.R., Wang, P., Dahl, S., Lee, J., and Goldstein, L.S. 29. Schuster, C.M., Davis, G.W., Fetter, R.D., and Goodman, C.S. (2000). Drosophila beta spectrin functions independently of al- (1996). Genetic dissection of structural and functional compo- pha spectrin to polarize the Na,K ATPase in epithelial cells. J. nents of synaptic plasticity. I. Fasciclin II controls synaptic sta- Cell Biol. 149, 647–656. bilization and growth. Neuron 17, 641–654. 10. Lee, J.K., Coyne, R.S., Dubreuil, R.R., Goldstein, L.S., and 30. Coyle, I.P., Koh, Y.H., Lee, W.C., Slind, J., Fergestad, T., Little- Branton, D. (1993). Cell shape and interaction defects in alpha- ton, J.T., and Ganetzky, B. (2004). Nervous wreck, an SH3 spectrin mutants of Drosophila melanogaster. J. Cell Biol. 123, adaptor protein that interacts with Wsp, regulates synaptic 1797–1809. growth in Drosophila. Neuron 41, 521–534. Current Biology 928

31. Desai, C.J., Popova, E., and Zinn, K. (1994). A Drosophila receptor tyrosine phosphatase expressed in the embryonic CNS and larval optic lobes is a member of the set of proteins bear- ing the “HRP” carbohydrate epitope. J. Neurosci. 14, 7272– 7283. 32. Marie, B., Sweeney, S.T., Poskanzer, K.E., Roos, J., Kelly, R.B., and Davis, G.W. (2004). Dap160/Intersectin scaffolds the peri- active zone to achieve high-fidelity endocytosis and normal synaptic growth. Neuron 43, 207–219. 33. McCabe, B.D., Marques, G., Haghighi, A.P., Fetter, R.D., Crotty, M.L., Haerry, T.E., Goodman, C.S., and O'Connor, M.B. (2003). The BMP homolog Gbb provides a retrograde signal that regulates synaptic growth at the Drosophila neuromuscular junction. Neuron 39, 241–254. 34. Hafezparast, M., Klocke, R., Ruhrberg, C., Marquardt, A., Ah- mad-Annuar, A., Bowen, S., Lalli, G., Witherden, A.S., Humme- rich, H., Nicholson, S., et al. (2003). Mutations in dynein link motor neuron degeneration to defects in retrograde transport. Science 300, 808–812. 35. LaMonte, B.H., Wallace, K.E., Holloway, B.A., Shelly, S.S., As- cano, J., Tokito, M., Van Winkle, T., Howland, D.S., and Holz- baur, E.L. (2002). Disruption of dynein/dynactin inhibits axonal transport in motor neurons causing late-onset progressive degeneration. Neuron 34, 715–727. 36. Puls, I., Jonnakuty, C., LaMonte, B.H., Holzbaur, E.L., Tokito, M., Mann, E., Floeter, M.K., Bidus, K., Drayna, D., Oh, S.J., et al. (2003). Mutant dynactin in motor neuron disease. Nat. Genet. 33, 455–456. 37. Munch, C., Sedlmeier, R., Meyer, T., Homberg, V., Sperfeld, A.D., Kurt, A., Prudlo, J., Peraus, G., Hanemann, C.O., Stumm, G., et al. (2004). Point mutations of the p150 subunit of dynactin (DCTN1) gene in ALS. Neurology 63, 724–726. 38. Komada, M., and Soriano, P. (2002). [Beta]IV-spectrin regulates sodium channel clustering through ankyrin-G at axon initial segments and nodes of Ranvier. J. Cell Biol. 156, 337–348. 39. Lacas-Gervais, S., Guo, J., Strenzke, N., Scarfone, E., Kolpe, M., Jahkel, M., De Camilli, P., Moser, T., Rasband, M.N., and Solimena, M. (2004). BetaIVSigma1 spectrin stabilizes the nodes of Ranvier and axon initial segments. J. Cell Biol. 166, 983–990. 40. Parkinson, N.J., Olsson, C.L., Hallows, J.L., McKee-Johnson, J., Keogh, B.P., Noben-Trauth, K., Kujawa, S.G., and Tempel, B.L. (2001). Mutant beta-spectrin 4 causes auditory and motor neuropathies in quivering mice. Nat. Genet. 29, 61–65. 41. Yang, D., Lu, H., and Erickson, J.W. (2000). Evidence that pro- cessed small dsRNAs may mediate sequence-specific mRNA degradation during RNAi in Drosophila embryos. Curr. Biol. 10, 1191–1200. 42. Sikorski, A.F., Sangerman, J., Goodman, S.R., and Critz, S.D. (2000). Spectrin (betaSpIIsigma1) is an essential component of synaptic transmission. Brain Res. 852, 161–166. 43. Zimmer, W.E., Zhao, Y., Sikorski, A.F., Critz, S.D., Sangerman, J., Elferink, L.A., Xu, X.S., and Goodman, S.R. (2000). The do- main of brain beta-spectrin responsible for synaptic vesicle association is essential for synaptic transmission. Brain Res. 881, 18–27. 44. Eaton, B.A., and Davis, G.W. (2003). Synapse disassembly. Genes Dev. 17, 2075–2082. 45. Jenkins, S.M., Kizhatil, K., Kramarcy, N.R., Sen, A., Sealock, R., and Bennett, V. (2001). FIGQY phosphorylation defines discrete populations of L1 cell adhesion molecules at sites of cell-cell contact and in migrating neurons. J. Cell Sci. 114, 3823–3835. 46. Ruiz-Canada, C., Ashley, J., Moeckel-Cole, S., Drier, E., Yin, J., and Budnik, V. (2004). New synaptic bouton formation is disrupted by misregulation of microtubule stability in aPKC mutants. Neuron 42, 567–580. 47. Pennetta, G., Hiesinger, P., Fabian-Fine, R., Meinertzhagen, I., and Bellen, H. (2002). Drosophila VAP-33A directs bouton formation at neuromuscular junctions in a dosage-dependent manner. Neuron 35, 291–306. 48. Sisson, J.C., Field, C., Ventura, R., Royou, A., and Sullivan, W. (2000). Lava lamp, a novel peripheral golgi protein, is required for Drosophila melanogaster cellularization. J. Cell Biol. 151, 905–918.