We report in this paper that activation of Ral by integrin signaling increases neurite branching through pathways involving the exocyst complex, PLD activation, and PKC-mediated phosphorylation of GAP-43. To date, there have been only few studies on the function of Ral in neurons. Inhibition of Ral has been reported to enhance NGF-mediated neurite outgrowth in PC12 cells and, although this seems contradictory to the results reported here, it is likely that this may have had more to do with inhibition of the differentiation program than with cytoskeletal changes (Goi et al., 1999
). A previous study in synaptosomes isolated from transgenic mice expressing dominant-negative RalA implicated Ral in the regulation of neurosecretion (Polzin et al., 2002
). Interestingly, these authors showed that dominant-negative Ral suppressed both phorbol ester– and calcium-induced phosphorylation of known PKC substrates, such as MARCKS and SNAP-25. The animals did not display any gross morphological defects in the brain, but the use of the neuronal-specific enolase promoter, which becomes active in mature neurons only at the time of synaptogenesis, may have prevented the discovery of potential defects in neuronal morphology during development. Cortical neurons expressing dominant-negative Ral maintain a polarized morphology with an axonlike process and a few shorter neurites, suggesting that Ral activity is dispensable for the establishment of neuronal polarity. Indeed, we still observe neurite growth in both cortical and SCG neurons expressing dominant-negative Ral, indicating that this GTPase is not necessary for neurite extension per se. Ral activation may instead be associated with local signaling events required during development and leading to branching, such as axon pathfinding, formation of terminal arborizations, and lesion-induced neurite sprouting.
We show that both RalA and -B promote phosphorylation of GAP-43 but only RalA acts through the exocyst complex. These results indicate that the two Ral isoforms may promote branching by different mechanisms; in support of this, expression of Rlf-CAAX, a Ral-GEF able to activate both endogenous RalA and -B, produced the most striking effect on branching. In addition, knocking down a single Ral isoform by RNAi produced a significant decrease in branching, indicating that RalA and -B may play distinct roles in regulating branch number. RalA has been reported to bind to the exocyst complex more effectively than active RalB both in vitro and in cultured cells (Shipitsin and Feig, 2004
). Even though the two Ral isoforms are 85% identical, they differ in their COOH-terminal domain, which is responsible for their distinct cellular localization, as shown in MDCK cells (Shipitsin and Feig, 2004
). It is possible that the different localization of Ral isoforms could differentially modulate their interaction with various effectors.
Disrupting the interaction of RalA with the exocyst impaired the ability of active RalA to promote branching, implying that the exocyst plays an important role in branching. Previous papers have implicated the exocyst, a complex consisting of at least eight subunits, in cytoskeletal regulation. Interaction between RalA and the exocyst complex has been shown to mediate Cdc42-dependent filopodia formation in fibroblasts (Sugihara et al., 2002
). PC12 cells expressing dominant-negative sec10 and ventral cord neurons from sec5
mutant Drosophila melanogaster
larvae fail to extend neurites (Vega and Hsu, 2001
; Murthy et al., 2003
). Interestingly, sec5
larvae also show growth arrest of neuromuscular junctions and impaired motor neuron arborizations on muscles, suggesting a role for the exocyst complex in regulating the normal development of terminal innervation sites (Murthy et al., 2003
). Neurite branching is characterized by the splaying apart and fragmentation of microtubules and by filopodial activity (Gallo and Letourneau, 2004
), events that we observe upon Ral activation. Interestingly, local destabilization of microtubules along axons can promote plasma membrane addition, and the exocyst complex has been shown to inhibit tubulin polymerization in vitro (Wang et al., 2004
). In normal rat kidney cells, overexpression of the exocyst subunit exo70 results in disruption of the microtubule network and increased plasma membrane addition in the form of long and thin plasma membrane protrusions (Wang et al., 2004
). Thus, recruitment of the exocyst complex by active Ral could promote branching through a dual mechanism involving the coordinated regulation of microtubule dynamics and membrane delivery.
Our data indicate that the association between Ral and PLD plays a crucial role in promoting branching, particularly for RalB. Although Ral has no effect on the activity of PLD in vitro, activation of PLD by v-Src, for example, is Ral dependent (Jiang et al., 1995
; Luo et al., 1998
). Moreover, interaction of Ral with PLD synergistically enhances Arf6-dependent PLD activation downstream of Ras (Xu et al., 2003
), and Arf6 has been implicated in the regulation of both axonal and dendritic branching (Hernandez-Deviez et al., 2004
) and in the efficient recycling of β1 integrin (Brown et al., 2001
). In addition, GTP-bound Arf6 can promote membrane recycling toward specialized plasma membrane regions by directly interacting with the exocyst subunit Sec10 (Prigent et al., 2003
). Arf6 activity has also been consistently associated with the formation of actin-based plasma membrane protrusions (Prigent et al., 2003
). Although the existence of a signaling pathway downstream of integrins involving Ral and Arf6 remains to be demonstrated, these intriguing links further point to Ral as an attractive candidate molecule likely to coordinate cytoskeletal reorganization and polarized endocytic recycling in response to extracellular matrix stimuli.
PLD generates phosphatidic acid, a multifunctional lipid that can be further metabolized to other important signaling lipids, such as DAG, which is a PKC activator, and lysophosphatidic acid. Several in vitro studies have suggested that PLD plays a role in neurite outgrowth. PLD expression and activity increase during neuronal differentiation of PC12 cells and the immortalized mouse hippocampal stem cell line HiB5 (Min et al., 2001
; Sung et al., 2001
). In addition, inhibiting PLD activity impairs L1-stimulated neurite outgrowth in cerebellar granule neurons (Watanabe et al., 2004
). Interestingly, PLD1 increasingly associates with PKC-α and -βII during NGF-induced neuronal differentiation of PC12 cells (Min et al., 2001
), suggesting that PLD and PKC activities are correlated to induce differentiation. Active Ral might then serve a dual function by recruiting PLD to the membrane and enhancing its activity, leading to PKC activation and subsequent phosphorylation of substrates such as GAP-43. In agreement with this, in models of experimentally induced sprouting such as kainate-induced outgrowth of hippocampal mossy fibers, both PLD and GAP-43 are strongly reexpressed (Zhang et al., 2004
). Sprouting responses are usually triggered after injury by the release of diffusible growth factors and the up-regulation of several extracellular matrix molecules, among them laminins, which facilitate adhesion and outgrowth of the sprouting processes. Ral could therefore play an important role in coupling extracellular matrix signals to the cytoskeleton via GAP-43 and to vesicle trafficking via the exocyst complex, ultimately leading to branching.
We show that Ral regulates branching through phosphorylation of GAP-43. GAP-43 is generally considered an intrinsic determinant for neurite outgrowth and plasticity. It is maximally expressed during nervous system development and reinduced in injured and regenerating neural tissues, and it serves to potentiate growth cone and nerve terminal responses to local growth and guidance signals. Its expression is not an absolute requirement for neurite outgrowth either in vitro or in vivo (Aigner and Caroni, 1995
; Strittmatter et al., 1995
). However, depletion experiments in cultured neurons indicate that GAP-43 greatly promotes adhesion and persistent growth cone spreading and branching (Aigner and Caroni, 1995
). Membrane anchoring of GAP-43 through palmitoylation plays a crucial role for its morphogenic effects, possibly promoting its recruitment to specialized membrane microdomains enriched in sphingolipids. Incorporation of a lipidated motif such as the one present in GAP-43 into lipid rafts may create local changes in membrane tension and the extension of filopodial structures (Gauthier-Campbell et al., 2004
). We found that overexpressed active Ral (particularly RalB) colocalizes in patches along neurites with endogenous GAP-43 (unpublished data). These patches might represent functional adhesion sites, as GAP-43 has been shown to associate with areas of membrane that are tightly bound to the substrate and belong to the membrane cytoskeletal fraction (Meiri and Gordon-Weeks, 1990
). Extracellular matrix–induced localization of integrins in lipid rafts can help generate and maintain a specific signaling environment (Decker et al., 2004
). Activation of Ral by integrin stimulation along neurites might help recruit important branch-promoting effectors, such as the exocyst complex and PLD, to such signaling microdomains.
We found that Ral activation increases phosphorylation of endogenous GAP-43 in cortical neurons. Interestingly, PKC-mediated phosphorylation of GAP-43 correlates with increased stabilization of the interaction between growth cones and substrate and, conversely, unphosphorylated GAP-43 is always found in retracting or collapsing areas (Dent and Meiri, 1992
). In addition, NGF-treated PC12 cells lacking endogenous GAP-43 and expressing nonphosphorylatable GAP-43 display abnormalities in long-term adhesion to laminin substrates, surface expression of integrins, and cell morphology, with marked membrane blebbing and varicosities (Meiri et al., 1996
). Phosphorylation of GAP-43 by PKC therefore seems to maintain a functional coupling between the plasma membrane and the underlying cytoskeleton. Indeed, phosphorylated GAP-43 can stabilize long actin filaments in vitro (He et al., 1997
) and can enhance the accumulation of phosphatidylinositol-4,5-bisphosphate–rich plasmalemmal patches, which are possible sites of signal-induced actin assembly (Laux et al., 2000
). Maintenance of phosphorylated GAP-43 induced by local Ral activation might therefore contribute to the stabilization of signaling pathways promoting actin polymerization and, ultimately, branching.
GAP-43 knockout mice display pathfinding defects and perturbed development of cortical maps (Strittmatter et al., 1995
; Maier et al., 1999
), whereas mice overexpressing GAP-43 show striking spontaneous and lesion-induced nerve sprouting, with highly potentiated terminal arborization during innervation (Aigner et al., 1995
). Interestingly, several extracellular matrix molecules are up-regulated after injury to either the peripheral or the central nervous system, among them both laminins and collagens (Ivins et al., 2000
). In this context, and given that exposure to laminin activates Ral, it is tempting to speculate that this GTPase could act as a mediator between integrin signaling and GAP-43 to potentiate lesion-induced sprouting. Integrins are critical determinants of basic morphogenesis for many aspects of embryogenesis, including nervous system development. Indeed, they participate in a variety of processes, such as brain lamination, neuroblast migration, and axon navigation and synaptogenesis (Milner and Campbell, 2002
). Our data point to Ral as an important molecular link between integrin-dependent signaling and cytoskeletal reorganization in neurons. Further investigation should help clarify the involvement of Ral in axonal pathfinding and synapse formation, events relying on both integrin signaling and branch/filopodia formation.