Coordinating Synaptic Growth without Being a Nervous Wreck

Neuron, Vol. 41, 489–494, February 19, 2004, Copyright 2004 by Cell Press

Previews

Coordinating Synaptic Growth without Being a Nervous Wreck

The function and regulation of actin-cytoskeletal dynamics during synaptic growth is poorly understood. In this issue of Neuron, Coyle et al. report the identification of nervous wreck (nwk), a synapse-specific adaptor molecule in Drosophila that regulates synaptic growth and morphology via Wasp, a well-characterized mediator of actin dynamics. The wiring of the brain shares an underlying driving force with many other basic cellular processes—actin cytoskeletal dynamics. For example, during axon pathfinding, changes in cytoskeletal structure are intricately coordinated in response to external cues, resulting in directional growth, retraction, turning, and stalling of the growth cone. After decades of study, much is known about this process: many guidance receptors have been identified, and adaptor molecules that link the activation of these transmembrane receptors to the actin cytoskeleton have been found to play a central role in mediating guidance (Dent and Gertler, 2003; Korey and Van Vactor, 2000). While the importance of actin dynamics in axon pathfinding is clear, little is known of its role in the next step of neural development—the formation and growth of synapses. In this issue of Neuron, Coyle and colleagues expand our view of actin at the synapse with the identification of nervous wreck (nwk), a cytoskeletal adaptor molecule localized to synapses that functions in regulating synaptic growth in Drosophila (Coyle et al., 2004). The Drosophila larval neuromuscular junction (NMJ) has become a favorite model system for studying synaptic growth. As the larva grows dramatically in the course of only 4 days, the postsynaptic muscles increase ⵑ100fold in surface area. To maintain a synapse capable of contracting this ever-enlarging muscle, the motoneuron adds many new synaptic boutons filled with neurotransmitter release sites. These new boutons grow from preexisting boutons in a process that resembles yeast budding (Zito et al., 1999). The new boutons can grow from the ends of bouton strings or by intercalation between preexisting boutons. Sometimes a bud will form from a bouton that already has two neighbors, resulting in a branch. This process is highly regulated, so that each motorneuron forms a stereotyped arbor of bouton strings and branches. Defects in this stereotyped morphology can be readily detected, and previously identified mutants have suggested that synaptic morphology is regulated by neuronal activity, cell adhesion, microtubule dynamics, protein turnover, and various signaling cascades (Jin, 2002; Koh et al., 2000). The study by Coyle and colleagues adds regulation of the actin cytoskeleton to this list with the identification of nwk. This new molecule, nervous wreck, is named after the

behavior of the unfortunate flies that have this mutation. While their movement and coordination appear normal at room temperature, upon a shift to 38⬚C they rapidly lose coordination and undergo seizures followed by paralysis. Previously characterized temperature-sensitive paralytic mutations in Drosophila have given great insights into ion channel function and synaptic transmission. The characterization of nwk now extends this analysis to synaptic morphology. nwk mutants have a uniquely aberrant synaptic morphology. These mutant NMJs have more, and smaller, boutons. Their most striking phenotype, however, is the abundance of hyperbranched boutons. At wild-type synapses, almost all branches are bifurcations; at nwk synapses, three or more branches often extend from a single bouton. This phenotype suggests that nwk functions to inhibit synaptic branching. The cloning of nwk revealed a gene product with a very interesting assortment of sequence domains, and database searches identified a small family of homologs in organisms from yeast to humans. This family is defined by a unique “ARNEY” motif within the protein that is shared with no other proteins and a characteristic arrangement of domains. These include an FCH domain (implicated in actin binding), two SH3 domains, and a proline-rich C terminus that contains at least five potential SH3 binding sites. Some of the molecules in this family contain a Rho GTPase activating domain (RhoGAP) instead of the first SH3 domain. Since the Rho GTPase is a direct regulator of the actin cytoskeleton and SH3 and SH3 binding domains are common modes of interaction among molecules in membrane-associated signaling pathways, the members of this NwkARNEY family are likely to function as links between signaling pathways and the actin cytoskeleton. In support of this model, the family includes the srGAPs, which link the guidance receptor Robo for the chemorepellant Slit to the Rho GTPase, and thus to cytoskeletal remodeling (Wong et al., 2001). Two clues from sequence homologies led the authors to the more specific hypothesis that the Nwk-ARNEY family could influence the actin cytoskeleton via Wiscott-Aldrich Syndrome associated Protein (Wasp). First, the yeast Nwk-family member Bzz1/Las7 interacts with the yeast Wasp homolog Las17 via its SH3 domains (Soulard et al., 2002). Second, the first SH3 domain of Nwk most closely resembles the SH3 domain of the adaptor protein DOCK/Nck, which binds to Wasp via its SH3 domains (Buday et al., 2002). Wasp is an extensively characterized mediator of actin cytoskeletal dyanamics. Wasp directly activates the ARP2/3 complex, which nucleates the growth of new actin filaments and new branches of filaments off of existing filaments. By influencing the extent of filament branching, ARP2/3 is a central regulator of filamentous actin (F-actin) structure (Pollard and Borisy, 2003). The authors provide several lines of evidence that Nwk interacts with Wasp in the synaptic bouton to regulate synaptic morphology. First, Nwk localizes to a honeycomb-like network of the presynaptic membrane that

Neuron 490

surrounds active zones (sites of synaptic contact). This staining pattern has been previously described as the periactive zone and is shared with other molecules identified as regulators of synaptic growth (Sone et al., 2000). These include the neural adhesion molecule Fasciclin II and the putative Rho guanine exchange factor still life. Expression of a Nwk-containing transgene in neurons can at least partially rescue all of the nwk mutant phenotypes, proving that Nwk functions presynaptically, as its localization suggests. Second, the Drosophila Wasp protein (Wsp) localizes to presynaptic boutons. Wsp localization overlaps with Nwk although it is more diffuse, more closely resembling that of F-actin within the presynaptic bouton. Third, Nwk can biochemically interact with Wsp via its SH3 domain. Fourth, mutations in wsp phenocopy the nwk mutants; both exhibit an increase in bouton number and the appearance of hyperbranched boutons. Finally, wsp and nwk mutants display dosage-sensitive genetic interactions—double heterozyotes (in which one copy of each gene is mutant, while the other is wild-type), have significantly overbranched synapses when compared to mutants that are heterozygous for either gene alone. This type of genetic interaction is rarely observed for gene products that do not directly influence each other. In addition, when both genes are mutant there is a dramatic enhancement of the hyperbranching phenotype. In the double mutants, boutons with as many as six new branches are frequently observed. Since the double mutant is more impaired than either single mutant, there may be additional synaptic activators of Wasp and likely additional effectors for Nwk. In summary, these findings suggest a model in which Nwk (and other members of the newly delineated NwkARNEY family) functions via Wasp. Presumably, by binding Wasp, Nwk can recruit or activate the ARP2/3 complex and thus stimulate branching of actin filaments and reorganize the F-actin cytoskeleton. Of note, while F-actin has long been suspected to play a role in determining bouton morphology, this has not been previously shown. The identification Nwk provides the strongest link yet between F-actin and bouton structure. The implied function of ARP2/3 complex remains to be formally demonstrated. Likewise, additional factors that influence ARP2/3 or play a role in the elaborate dance of F-actin dynamics might also regulate synapse morphology. However, many of the already-known mediators of F-actin dynamics function in nearly every cell and every stage of development. In contrast, Nwk appears to play a more specialized role at the synapse. It remains to be determined how the branching of F-actin influences synaptic growth and bouton branching. Figure 1 proposes a model based on our understanding of the role of branched F-actin in motile cells and growth cones, where it functions in lamellopodia as part of the force that pushes forward the leading edge. In contrast, the forces that drive the motile filopodia in growth cones sensing their environment favor filament elongation and disfavor filament branching. Transferring this paradigm to the synapse suggests that a bouton enlarges in a manner analogous to the leading edge of translocating lamellopodia, while bouton budding is akin to formation of exploratory filopodia. In this view, Nwk, acting via Wasp and ARP2/3, could function at the synapse to

Figure 1. Model for Nwk and F-actin Branching during Synaptic Growth

stimulate F-actin branching and thus promote the enlargement of an individual bouton while inhibiting its budding. This could account for the phenotype in the nwk mutant of smaller bouton size and excess budding and branching. While these speculations may take some time to address, other pressing questions await. Having identified Nwk as a synaptic molecule that can activate changes in the F-actin cytoskeleten, what acts upstream of Nwk? Presumably Nwk links a membrane associated signal to the cytoskeleton. What is this signal? By analogy to the ARNEY family member srGAP and its interaction with Robo, could Nwk bind a guidance receptor for synaptic growth or morphology? Finally, the existence of Nwk homologs in C. elegans, mice, and humans suggests a conserved funtion for nwk. In support of this hypothesis, the Nwk family member srGAP3 has recently been identified as the culprit in a severe mental retardation syndrome (Endris et al., 2002). Nwk family members are thus excellent candidates to regulate vertebrate synaptic development.

Catherine A. Collins and Aaron DiAntonio Department of Molecular Biology and Pharmacology Washington University School of Medicine St. Louis, Missouri 63110 Selected Reading Buday, L., Wunderlich, L., and Tamas, P. (2002). Cell. Signal. 14, 723–731.

Previews 491

Coyle, I.P., Koh, Y.-H., Lee, W.-C.M., Slind, J., Fergestad, T., Littleton, J.T., and Gantetzky, B. (2004). Neuron 41, this issue, 521–534. Dent, E.W., and Gertler, F.B. (2003). Neuron 40, 209–227. Endris, V., Wogatzky, B., Leimer, U., Bartsch, D., Zatyka, M., Latif, F., Maher, E.R., Tariverdian, G., Kirsch, S., Karch, D., and Rappold, G.A. (2002). Proc. Natl. Acad. Sci. USA 99, 11754–11759. Jin, Y. (2002). Curr. Opin. Neurobiol. 12, 71–79. Koh, Y.H., Gramates, L.S., and Budnik, V. (2000). Microsc. Res. Tech. 49, 14–25. Korey, C.A., and Van Vactor, D. (2000). J. Neurobiol. 44, 184–193. Pollard, T.D., and Borisy, G.G. (2003). Cell 112, 453–465. Sone, M., Suzuki, E., Hoshino, M., Hou, D., Kuromi, H., Fukata, M., Kuroda, S., Kaibuchi, K., Nabeshima, Y., and Hama, C. (2000). Development 127, 4157–4168. Soulard, A., Lechler, T., Spiridonov, V., Shevchenko, A., Li, R., and Winsor, B. (2002). Mol. Cell. Biol. 22, 7889–7906. Wong, K., Ren, X.R., Huang, Y.Z., Xie, Y., Liu, G., Saito, H., Tang, H., Wen, L., Brady-Kalnay, S.M., Mei, L., et al. (2001). Cell 107, 209–221. Zito, K., Parnas, D., Fetter, R.D., Isacoff, E.Y., and Goodman, C.S. (1999). Neuron 22, 719–729.

The Kiss of Death The programmed cell death (PCD) of neurons is generally thought to be cell autonomous and not to require a death signal from other cells. A recent study by MarinTeva et al., in this issue of Neuron, brings this theory into question and suggests that neighboring microglia actively participate in the PCD of Purkinje cells in the cerebellum. The question of whether developmental PCD of neurons is cell autonomous versus mediated by death signals derived from adjacent cells has, until recently, been largely unresolved. Like most cell-physiological processes, the death and scavenging of cells requires flawless orchestration of numerous players. The conductor of this cellular symphony leading to the climactic death and removal of cells is largely unknown. Previous studies suggest that it is the apoptotic cell that initiates phagocytosis; however, recent work presented in this issue of Neuron suggests that microglia may be the impetus for death (Marin-Teva et al., 2004). One might imagine a number of cellular “verses” that must be accurately executed between these two cell populations for complete engulfment of the apoptotic cell to occur. Verses played by the apoptotic cell include “come hither,” triggering the recruitment of phagocytic cells, and “eat me,” indicating the need of the dying cell to be engulfed. The phagocyte responds by stimulating its mobilization to the target area, then voices “eat it” signals causing the phagocyte to tether itself to the apoptotic cell, followed by reorganizing “commands” that alter the cytoskeletal architecture necessary for engulfment, and finally an “activate” song resulting in the release of proinflammatory cytokines, often induced by lysis of the dying cell. Although microglia have long been appreciated as the resident scavenger cells of the nervous system, little

thought has been given to them as active killers of other cells. The laboratory of Michel Mallat has used an innovative strategy to selectively ablate microglia in a slice preparation of the early postnatal cerebellum. This microglial destruction results in the dramatic rescue of Purkinje neurons that otherwise would die within 24 hr in vitro. Furthermore, the authors postulate a mechanism by which Purkinje cells are compromised by the release of superoxide (O2.⫺) following the respiratory burst elicited by the engulfing microglia. This burst, a key feature of most engulfing phagocytes, is the result of the conversion of oxygen to O2.⫺ by activated NADPH oxidase. Additionally, O2.⫺ dismutates, resulting in the formation of H2O2 as well as other potentially toxic molecules. These results raise the question of whether the microglial respiratory burst is orchestrating Purkinje cell death or whether the dying Purkinje cells initiate the microglial respiratory burst. Although further studies are required to definitively answer this question, these authors have made great strides toward understanding this issue. When one attempts to understand the complex schedule of events associated with PCD, past experience tells us that the nematode C. elegans may be a helpful informant, whereas the immune system may play a similar role for understanding the potential involvement of microglia in this process. Using the powerful genetic tools provided by C. elegans, several “engulfment” genes have been identified that are thought to be part of two discrete transduction pathways, which exhibit some functional redundancy. The first group, ced-1, ced-6, and ced-7, encode homologs of a lowdensity lipoprotein receptor-related protein (LRP), the adaptor protein GULP, and a 12 transmembrane ATP binding cassette transporter protein, respectively. The second group, ced-2, ced-5, ced-10, and ced-12, encode homologs of CrkII, DOCK180, Rac, and ELMO. The first set of proteins is thought to function in corpse recognition, whereas the second is postulated to control the cytoskeletal rearrangements necessary for engulfment to occur. In an effort to understand the role of these engulfment genes in programmed cell death, two recent studies investigated the effect of mutating these genes in concert with ced-3, a caspase-3 homolog. When engulfment genes were mutated alone, no alterations in cell number were observed, suggesting that these genes alone cannot actively induce death. However, in weak ced3(op149) mutants, where death is only partially suppressed and a basal level of caspase activation occurs, some cells advanced through the early morphological stages of cell death but then reverted to their normal appearance and survived. These observations suggest that, under reduced caspase activation, cells have the ability to elude death, even after the initiation of an apoptotic program (Hoeppner et al., 2001). Furthermore, these studies support a threshold mechanism by which sufficient caspase-3 activation is required for the cell to undergo a complete death, which includes its engulfment and subsequent digestion. Accordingly, a critical question is what stimulates caspase activation above this theorized threshold? Some might argue that it is the physical process of cell engulfment that is the final trigger for death, as mutants