BIOLOGICAL OVERVIEW

Actin and microtubules (MTs) are tightly coordinated during neuronal growth cone navigation and are dynamically regulated in response to guidance cues; however, little is known about the underlying molecular mechanisms. Drosophila Pod-1 can crosslink both actin and MTs. In cultured S2 cells, Pod-1 colocalizes with lamellar actin and MTs, and overexpression remodels the cytoskeleton to promote dynamic neurite-like actin-dependent projections. Consistent with these observations, Pod-1 localizes to the tips of growing axons, regions where actin and MTs interact, and is especially abundant at navigational choice points. In either the absence or overabundance of Pod-1, growth cone targeting but not outgrowth is disrupted. Taken together, these results reveal novel activities for pod-1 and show that proper levels of Pod-1, an actin/MT crosslinker, must be maintained in the growth cone for correct axon guidance (Rothenberg, 2003).

The growth cone is a specialized structure at the tip of a growing axon that integrates extracellular guidance cues into cytoskeletal changes that underlie axon guidance. At the leading edge, finger-like filopodia consisting of parallel bundles of actin continuously extend and retract, exploring the immediate environment for guidance cues. Between the filopodia, veil-like membranous sheets, called lamellipodia, contain a complex branched network of actin filaments. Farther back from the lamellipodia, the central region of the growth cone contains MTs that continuously explore the peripheral regions and exhibit dynamic instability, often extending and retracting along actin filaments. When a filopodium encounters an attractive cue, it becomes dilated, and invading MTs are captured and stabilized. These polymerizing MTs then help drive the directional extension of the axon (Rothenberg, 2003 and references therein).

Actin and MT structures in growth cones often appear to be crosslinked and depend on each other for structural integrity. Disruption of either actin or MTs in growth cones affects both actin and MTs and can disrupt axon steering. The molecules underlying this relationship remain unknown (Rothenberg, 2003).

Pod-1 was isolated from early C. elegans extracts based on binding to F-actin, though Pod-1 is not required for the overall integrity of the actin cytoskeleton. Pod-1 is a strict maternal effect gene required for all aspects of early embryonic asymmetry and anterior-posterior axis formation (Rappleye, 1999), processes that depend on intact MT and actin networks. Pod-1 contains two tandem coronin repeats (Rappleye, 1999). Coronin was isolated by MT affinity chromatography and has been shown to bind both actin and MTs (Goode, 1999). However, C. elegans pod-1 has to date not been implicated in MT regulation (Rothenberg, 2003).

The Drosophila homolog of Pod-1 crosslinks actin and MTs in vitro. In S2 cells, Drosophila Pod-1 colocalizes extensively with newly assembled actin and often colocalizes with MTs that extend out to the lamellar edge or into filopodia. Depolymerization of actin causes Pod-1 to localize to MTs, whereas depolymerization of MTs has no effect on Pod-1 localization. Overexpression of Pod-1 induces dramatic changes in cell shape: long, neurite-like, actin-rich processes form in an actin-dependent (but not MT-dependent) manner and are subsequently invaded by MTs. Interestingly, at their tips, a subset of these processes localize Enabled, an important cytoskeletal regulator that functions together with several different transmembrane receptors (e.g., Robo and UNC-40/DCC) involved in many axon guidance decisions. In developing neurons, Pod-1 is concentrated in growing neurites, where it is especially enriched at the tips of extending axons often at navigational choice points. The primary defect in embryos completely lacking Pod-1 is aberrant axon targeting. Furthermore, the level of Pod-1 is critical, as postmitotic neuronal overexpression of Ppd1 causes severe defects in axon pathfinding. It is proposed that Pod-1 is an actin/MT crosslinker that functions in cytoskeletal remodeling during axon navigation and may facilitate the flow of guidance information to cytoskeletal networks (Rothenberg, 2003).

In vivo Pod-1 was highly enriched in the developing nervous system where it localizes to the tips of growing neurites and concentrates in axons at navigational choice points. In embryos completely lacking Pod-1, the fidelity of axon targeting is disrupted, and axons exhibit frequent guidance defects -- perhaps due to dysregulation of the growth cone cytoskeleton during turning and/or branching. Interestingly, no general defect was observed in early axon outgrowth, since axons invariably extend a long distance from the cell body. Thus, Pod-1 plays a specialized role in the growth cone. Furthermore, axon targeting requires proper levels of Pod-1, since postmitotic neuronal overexpression of Pod-1 is sufficient to disrupt axon guidance. Together, these results suggest that Pod-1 is an actin/MT crosslinker that coordinates cytoskeletal dynamics to ensure the fidelity of axon targeting (Rothenberg, 2003).

kakapo/shortstop is an actin-MT crosslinker of the plakin family and is required for continued axon extension in Drosophila: embryos homozygous for severe kak/shot alleles cannot project sensory axons more than a short distance from the soma and cannot direct motor axons to their targets. In strong alleles the phenotype is severe, and nearly all axons stop short. Plakins in other systems have been implicated in cell adhesion at sites of mechanical stress; perhaps axons lacking kak/shot have adhesive defects or problems with the 'clutch' mechanism that enables axons to grasp a substrate and extend. Although kak/shot may be required for axon targeting, its requirement for continued axon extension precludes this knowledge (Rothenberg, 2003).

In contrast, embryos lacking all Pod-1 can still extend but not target axons, perhaps because of turning and/or branching difficulties. Thus, Pod-1 and Kak/Shot have distinct functions. Additional data support the idea that Kak/Shot -- but not Pod-1 -- plays a primary role in neurite extension: while Pod-1 and Kak/Shot are both found at the tips of dendrites of lateral chordotonal neurons, kak/shot mutants have difficulty extending these dendrites, whereas embryos devoid of Pod-1 do not (Rothenberg, 2003).

Furthermore, preliminary genetic interaction data suggest that Pod-1 may function in part to transmit guidance signals to the cytoskeleton. For example, several observations were suggestive of a relationship between Pod-1 and Enabled. (1) Axon defects in embryos devoid of Pod-1 resemble defects in embryos mutant for enabled (ena). (2) Pod-1 overexpression recruits Ena to the ends of the neurite-like projections in S2 cells. (3) Extensive colocalization is observed between Pod-1 and Ena in S2 cells as well as in embryos. Therefore genetic interactions between pod-1, ena, and the robo receptor (one of several axon guidance receptors that directly binds to Ena) were tested to ask whether the genes might function together in midline repulsion, a specific axon guidance decision that involves Robo and Ena. Indeed, it was found that while pod-1 zygotic mutants, ena heterozygotes, or robo heterozygotes do not exhibit midline crossing errors, when gene dosages of ena or robo are reduced simultaneously with pod-1, frequent (i.e., in approximately 30% of abdominal segments) midline crossing errors are observed (Rothenberg, 2003).

If Pod-1 functions together with guidance signaling molecules such as Robo and Ena, one possible difference between Pod-1 and Kak/Shot is that Pod-1 may have a more subtle or regulated function in the transmission of guidance information from receptors to the cytoskeleton, rather than a constitutive structural role (Rothenberg, 2003).

In S2 cells, high levels of Pod-1GFP dramatically remodel both the actin and MT cytoskeletal networks to cause the outgrowth of dynamic, actin-rich, actin-dependent processes. Many of these processes localize Ena to their tips and are invaded by MTs, much like the dilated filopodia of axonal growth cones that have encountered chemoattractant cues. Consistent with this, postmitotic neuronal overexpression of Pod-1 in embryos causes defects in axon targeting. However, the axons in these embryos do not appear significantly different from wild-type: even when rapid fixation techniques are employed to preserve filopodia, the same kind of dramatic changes in cell shape observed in cell culture overexpressors are not abserved. However, Drosophila growth cones are extremely small; it remains possible that very careful live imaging may reveal a dynamic difference. Also, while expression levels in the overexpression embryos are high enough to alter axon targeting (presumably by affecting signaling and/or the cytoskeleton in a subtle or regulated way at choice points), the levels achieved may not have been high enough to strongly affect cell shape. In fact, in the overexpression embryos, specific Pod-1 localization to choice points is still observed, suggesting that the machinery that localizes Pod-1 is not saturated in spite of overexpression. Thus, while high levels of overexpression can be achieved in S2 cells to drastically alter cell shape, lower levels of overexpression are sufficient to affect navigating growth cones in embryos (Rothenberg, 2003).

It was surprising that reducing Pod-1 to undetectable levels by RNAi had no apparent effect on cytoskeletal dynamics in S2 cells. Notably, depletion of several other molecules that function as important cytoskeletal regulators in growth cones (such as Ena and Kak/Shot) also has no effect on S2 cell cytoskeletal dynamics, perhaps because S2 cells are nonpolarized nonmotile phagocytes and are therefore different from neurons. Whereas S2 cells can be informative about Pod-1's capabilities to remodel cytoskeletal networks and recruit regulatory components (e.g., Ena) to the tips of cellular processes, it is conceivable that Pod-1 performs these functions in neurons in response to signaling information that S2 cells do not receive (Rothenberg, 2003).

Although embryos devoid of all Pod-1 have frequent axon targeting defects, some axons are still able to reach their targets. Thus, Pod-1 is not absolutely required for axon targeting but instead ensures its fidelity. Perhaps Pod-1 has a regulatory role, or perhaps it functions redundantly with other molecules in growth cones. At least two models for the function of Pod-1 could account for the observed defects. They are not mutually exclusive. (1) Pod-1 may function as part of an 'information scaffold' that links important signaling molecules to the actin and MT networks. As part of an information scaffold, Pod-1 may function as a bridge that physically connects signaling molecules downstream of guidance receptors with actin and MTs. In this way, Pod-1 might facilitate the flow of extracellular guidance information to the cytoskeleton. (2) Pod-1 could also play a structural role by stabilizing cytoskeletal networks or certain cytoskeletal structures in growth cones (Rothenberg, 2003).

Interestingly, Pod-1 contains a +xxPxxP domain in its central region, a class 1K SH3 binding domain (Cesareni, 2002), as well as several other PXXP motifs that may bind to SH3 domain-containing proteins. Perhaps Pod-1 interacts with one or more of the several known SH3 domain-containing signaling proteins that play important roles in many axon guidance decisions (Rothenberg, 2003).

As a cytoskeletal crosslinker, Pod-1 could provide structural support to the growth cone cytoskeleton and thereby enable guidance information to be effectively translated into concerted cytoskeletal changes. Unfortunately, Drosophila growth cones are too small to allow a detailed description of the growth cone cytoskeleton, and it is not possible to detect subtle effects loss of Pod-1 may have on growth cone cytoskeletal networks. However, biochemical experiments show that Pod-1 possesses three distinct biochemical activities: actin bundling, MT crosslinking, and actin/MT crosslinking. Any of these may be important in the growth cone (Rothenberg, 2003).

Studies have shown that actin bundles are key elements in growth cone steering. Stabilization of actin bundles in a subregion of the growth cone anticipates attractive turning; conversely, focal loss of actin bundling induces local growth cone collapse and repulsive turning. Moreover, growth cone filopodia, while not required for continued axon extension, apparently determine the direction of axon growth. Since growth cones devoid of all Pod-1 still have filopodia, Pod-1 may play a role in regulating or modulating actin bundles or filopodia in vivo (Rothenberg, 2003).

Many studies have also illustrated the importance of MTs and MT regulatory proteins in axon guidance. In fact, local stabilization of MTs can induce axon attraction, while local destabilization of MTs can induce repulsion. Since Pod-1 localizes to MTs at the lamellar edge and can crosslink MTs in vitro, it may contribute to the capture and stabilization of MT ends or other aspects of MT regulation in growth cones (Rothenberg, 2003).

Increasing evidence has also emerged demonstrating that actin and MTs are precisely coordinated and highly regulated in growth cones. Polymerizing MTs preferentially extend along actin bundles during attractive growth, and MTs are probably linked to actin bundles as they undergo retrograde flow during filopodial retraction. Pharmacological disruption of actin affects MT organization, and vice versa. Disruption of actin/MT crosslinking would be expected to disrupt axonal steering but not extension. Thus, Pod-1 may be involved in actin/MT crosslinking in the growth cone. This would be consistent with the axon defects in embryos lacking Pod-1, the observations in S2 cells, and the in vitro experiments. In support of this, since actin/MT interactions are known to be required in migrating cells, embryos lacking Pod-1 are found to exhibit PNS abnormalities consistent with cell migration defects. Normally, the lateral chordotonal neurons migrate to become evenly aligned along the dorsal-ventral (D/V) axis; however, in embryos lacking all Pod-1, these neurons vary in their D/V position. Incidentally, this phenotype is also seen in ena mutants (Rothenberg, 2003).

This study has identified Pod-1 as an actin/MT crosslinker that can remodel the cytoskeleton and play an essential role in ensuring the fidelity of axon targeting. Pod-1 is highly conserved across evolution and may be important in neural development in different organisms. Mice and humans each possess a single pod-1. Interestingly, mouse pod-1 is expressed in the developing nervous system, with high levels in the dorsal root ganglia and neural tube. Subsequent work may reveal whether mammals and insects utilize pod-1 in similar ways during neural development (Rothenberg, 2003).

PROTEIN STRUCTURE

Amino Acids - 1074

Structural Domains

Sequence analysis of a full-length cDNA for Pod-1 shows that the predicted protein has 1074 amino acids, is 31% identical and 46% similar to C. elegans Pod-1, and has a nearly identical domain structure and length. Thus, like the worm, the fly has a single copy of pod-1. Notably, both mice and humans also possess a single pod-1 gene. Drosophila Ppd1 contains two tandem repeats of coronin homology, domains that likely mediate F-actin binding. Each of these domains has three WD repeats, a protein-protein binding motif found in a large number of proteins with diverse functions. Between the coronin repeats Pod-1 contains a highly charged stretch of 236 amino acids weakly homologous (19% identical, 44% similar) to the MT binding domain of MAP1B, a MT-associated protein that suppresses MT instability. The C terminus of Pod-1 has no identifiable domain but is highly conserved. Thus, the primary sequence of Pod-1 is consistent with an actin/MT crosslinker (Rothenberg, 2003).

date revised: 5 September 2003

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.