Mechanisms of neuronal growth cone guidance: an historical perspective

Abstract

At the distal most aspect of motile extending axons and dendrites lies the growth cone, a hand like macro-organelle of membrane bound cytoskeleton, packed with receptors, adhesion molecules, molecular motors and an army of regulatory and signaling proteins. Splayed out along the substratum in-vitro, the growth cone resembles an open hand with bundles of filamentous actin, barbed ends outstretched, as if fingers extending from a central domain of dynamic microtubule plus ends. The growth cone acts first as a sensory platform, analyzing the environment ahead for the presence of guidance cues, secondly as a mechanical dynamo establishing focal contact with the extracellular matrix to drive processive forward outgrowth, and thirdly as a forward bio-chemical command center where signals are interrogated to inform turning, extension, retraction or branching. During his career Paul Letourneau has made major contributions to our understanding of how growth cones respond to their environment. Here we will summarize some of these major advances in their historical context. Letourneau’s contributions have provided insights into cytoskeletal organization, growth cone dynamics, and signaling pathways. His recent work has described some important molecules and molecular mechanisms involved in growth cone turning. Although much remains to be understood about this important and intriguing structure, Letourneau’s contributions have provided us with “growth cone guidance.”

Introduction

It takes the culmination of millions of connections to produce first functional neural circuits which arrayed just so will then facilitate the autonomous existence of the more complex metazoan life forms. During embryonic development the design and assembly of these diverse neural scaffolds is repeatedly executed with great precision despite the often large distances traversed between a neuronal soma and its ultimate axon terminal. The burden of arranging this architecture ultimately lies on a simple and elegant structure, the neuronal growth cone, first described 120 years ago (Ramon y Cajal, 1890). Recent years have seen great advances in building a conceptual model of growth cone dynamics and function. The state of the current model is indebted to the pioneering work of Dr. Paul Letourneau, whose research has much like a growth cone, laid the foundations of many key concepts, steered the course towards our understanding of important signaling pathways and today continues to challenge established paradigms and in this way contribute directly on its refinement.

Paul Letourneau received his doctoral training in the laboratory of Norm Wessells at Stanford University. Following the discovery of actin in neuronal growth cones (Fine and Bray, 1971), the Wessells lab published a pioneering paper demonstrating for the first time the structure of actin filaments in growth cones and the use of cytochalasin B (CB) to disrupt growth cone morphology through effects on actin ultrastructure (Yamada et al., 1971). It was in this environment that Letourneau started his graduate studies. Herein we will explore in brief a sub-set of Letourneau’s manuscripts each developing important elements of growth cone guidance and signaling. This tribute will begin with his work on growth cone adhesion and localization of cytoskeletal structures by electron microscopy and immunocytochemistry and then follow the progression of his work detailing key signaling pathways regulating actin polymerization in response to attractive cues to establish our current mechanistic model of growth cone guidance.

Growth cone adhesion and identification of myosin II in growth cones

In the initial studies wherein growth cones were treated with CB and growth cone collapse ensued, neurite outgrowth was also inhibited (Yamada et al., 1971). The general consensus at that time was that removal of actin prevented neurite elongation, which was envisioned by many as a pulling force that towed the neurite behind it. Indeed there were many papers over the two decades following this report that argued about push and pull forces in neurite extension. During his graduate work, Letourneau studied different effectors of neurite elongation and non-neuronal cell migration including lanthanum ions (La⁺³) and axonal outgrowth in an agar matrix (Strassman et al., 1973; Letourneau and Wessells, 1974). When he started his own lab he continued to work on growth cone interactions with different substrates (Letourneau, 1975a, b). From these studies he came to the conclusions that cell migration and nerve axon elongation differ in mechanism, with respect to both adhesive interactions and the activity of microfilament systems. He also concluded that interactions between growth cone filopodia and the substratum are important determinants in the directions and pathways of axonal elongation (Letourneau, 1975b).

Letourneau then showed that it was growth cone to substratum adhesion that is required for neuronal morphogenesis and neurite elongation (Letourneau, 1979), and he was the first person to visualize myosin II bipolar filaments within growth cones (Letourneau, 1981). These studies also suggested that the effects of CB in inhibiting neurite outgrowth, as observed by Yamada et al in 1971, might have come from inhibition of actin-based adhesion, rather than from the direct modulation of actin assembly processes thought to be involved in growth cone pathfinding. Letourneau was the first to directly demonstrate the separation between neurite outgrowth and growth cone dynamics, but before we get to that story it is worthwhile to touch on his important findings with regard to the cytoskeletal organization of the growth cone.

Cytoskeletal organization in growth cones

Through combinations of fixation methods before or during permeabilization/extraction of neurites and growth cones undergoing active outgrowth, Letourneau gave us an ultrastructural picture of the growth cone cytoskeleton (Letourneau, 1983). He identified the actin filaments through their binding of myosin fragments (HMM and S1) and showed these were sparse within neurites. He found that microtubules, which were well known to extend into the central domain of growth cones, would often penetrate the actin filament-rich peripheral domain, often terminating along bundles of actin filaments (filopodial bundles) at the growth cone margins. This finding helped clarify a controversial area and led to speculation that the microfilament system at the leading edge may have a direct effect on microtubule organization and force transduction during neurite outgrowth and turning.

Neurite outgrowth and pathfinding

In a groundbreaking study, Marsh and Letourneau (1984) used dissociated chick dorsal root ganglion (DRG) neurons that were cultured in the continuous presence of CB on a good adhesive substrate, poly-ornithine. Interestingly these cells, though lacking filopodia, lamellipodia or any semblance of a growth cone, were capable of neurite extension, often elaborating very long and meandering neurites. This behavior was found to be dependent upon a highly adhesive substratum as neurons cultured on untreated plastic showed growth cone collapse without neurite extension as was originally observed by Yamada et al in 1971. Time lapse films demonstrated that in the presence of CB, neurites lacking growth cones extend in a pulsatile fashion resembling the extrusion of material from blunted tips. In addition, saltatory bidirectional motion of intracellular particles was observed in CB treated neurites. Although some small amounts of residual F-actin could be observed using ultrastructural methods, the filaments were very short and clumped. Taken together these findings indicate an organized and dynamic actin network is not required for neurite extension. It also indirectly demonstrates that actin filaments are not required for vesicular transport or delivery of new plasma membrane to growing tips. These studies set the stage for the in vivo experiments out of David Bentley’s lab two years later (Bentley and Toroian-Raymond, 1986), which showed that treatment of grasshopper embryos with CB did not affect neurite elongation but impaired pathfinding during the development of T1 pioneer neurons in vivo.

Over the next two decades Letourneau and collaborators continued to examine the great complexity of the neuronal cytoskeleton and its regulation, particularly the effects of neurotrophic factors and calcium signaling processes. This work will be highlighted in some of the companion articles found in this special edition. To remain focused on actin in growth cones we will next continue with a trio of articles on this subject beginning in 2004.

Growth cone filopodial dynamics and signaling pathways

In a series of three manuscripts, Letourneau’s lab greatly expanded our knowledge of how signaling pathways involving the Rho family of GTPases respond to neurotrophin (NT) binding to coordinate filopodial reorganization in growth cones. In the first of these papers, signaling via neurotrophin receptors and downstream substrates was examined in cultured DRG and embryonic retinal neurons treated with NGF or brain derived neurotrophic factor (BDNF), respectively (Gehler et al., 2004a). Treatment of cultured neurons with NTs produces an increase in growth cone filopodial length and number in a dose-dependent manner. Surprisingly, this response was mediated by the p75 NT receptor (p75^NTR) and not via the high affinity Trk receptors. Interestingly however, neurons from p75^NTR knock-out mice showed enhanced filopodial length beyond what was observed in wild type growth cones treated with NTs leading to the conclusion that the unoccupied p75^NTR has a negative effect on filopodial length that is absent when p75^NTR is not expressed. Letourneau’s group demonstrated NT treatment of wild type neurons reduces RhoA activity; inhibition of RhoA in the absence of NT is sufficient to stimulate filopodial length whereas expression of constitutively active RhoA effectively blocks NT signaling to growth cones.

In the second installment (Gehler et al., 2004b) NT signaling immediately upstream of actin was examined in greater detail. Pharmacological inhibition of the RhoA effector ROCK similar to treatmentwith NT increased filopodial length and number. Co-treatment of cultured neurons with ROCK inhibitor plus NT had no additive effect on filopodial length but did increase filopodial number over inhibitor alone. This result revealed that the NT stimulated increased filament length requires ROCK signaling, while increases in filopodial number may be regulated by an NT independent pathway via ROCK. Looking downstream of ROCK, Gehler et al. (2004b) focused on the actin depolymerizing factor (ADF)/cofilin family of actin dynamizing proteins (Bamburg and Bernstein, 2010). These proteins are inhibited via phosphorylation on serine 3 by the ROCK substrate LIM kinase. Application of NT stimulated ADF/cofilin dephosphorylation (activation) and introduction of the inactive (S3E) ADF/cofilin had a dominant negative effect by blocking the increased filopodial length produced by p75^NTR activation. Chariot vector-mediated introduction of constitutively active (S3A) ADF/cofilin in the absence of NT stimulation increased filopodial length, an effect that was not additive with NT. To determine if depolymerizing or severing activity of ADF/cofilin was required for mediating the change in filopodial length and number, Gehler et al. employed the dominant negative (S3E) and non-severing (KK95,96QQ) mutants. While neither had an effect on filopodial length when administered alone, each could block increased filopodial length following NT treatment. Neither mutant alone decreased basal filopodial number but S3E blocks most and KK95,96QQ bocks all of the increases in filopodial number following NT application. Together these results demonstrate that the severing activity of ADF/cofilin is necessary and sufficient for mediating the effects of NT on filopodial length; ADF/cofilin activity is also required for NT-mediated increased filopodial number.

To examine the pathways underlying this divergence in signaling, Gehler et al. next demonstrated that inhibition of myosin-II activity (a substrate for ROCK activation) also produces increased filopodial length. Co-treatment of neurons with NT and a myosin-II inhibitor had an additive effect on filopodial length indicating that they act through separate pathways. This was additionally confirmed by an additive increase in filopodial length produced by treatment with both ROCK- and myosin-II inhibitors. The additional observations that increased filopodial length resulting from myosin-II inhibition was not blocked by ADF/cofilin S3E and was additive with S3A, suggests that NT signaling to growth cone filopodia via ADF/cofilin occurs independently of the ROCK-myosin-II pathway.

In the third installment of these studies, the Rho family GTPase Cdc42, implicated in mediating effects of positive guidance cues, is shown to modify NT signaling on growth cone actin dynamics (Chen et al., 2005). NT treatment produced an activation of Cdc42; expressing constitutively active Cdc42 in the absence of NT decreases active RhoA, increases active ADF/cofilin, and recapitulates the NT effects of increased filopodial number and length. Incorporation of a dominant negative Cdc42 increased RhoA activity and prevented NT driven ADF/cofilin activation and subsequent increases in filopodial dynamics. The Cdc42-stimulation of filopodial dynamics was prevented by inhibition of ROCK, thus demonstrating that crosstalk between Cdc42 and RhoA occurs upstream of ROCK. The elegant experimental design involving the chariot loading of proteins so that their acute effects can be studied, as well as the careful quantification, illuminated the signaling pathways and regulatory proteins involved from receptor binding to downstream changes in actin dynamics in detail (Figure 1). These studies set the stage for understanding mechanisms of growth cone pathfinding.

Figure 1:

Pathways regulating growth cone filopodial length and number. The unbound p75^NTR receptor signals to RhoA inducing LIMK mediated phosphorylation of ADF/cofilin (pAC) to reduce filopodial length and number. This observation was confirmed in p75^NTR knock out animals where growth cone filopodia resemble those in wild type neurons treated with NT. NT-bound p75^NTR activates Cdc42 which inhibits RhoA to produce the net effect of ADF/cofilin activation, increasing both filopodial length and number. ROCK inhibition can also mediate an increased filopodial length via the inhibition of myosin II phosphorylation (activation). However, myosin II can also be activated through non ROCK-dependent pathways, allowing for additive effects on filopodial length for ROCK and myosin II inhibitors.

Current View of Growth Cone Pathfinding

Letourneau and colleagues have recently followed up on the story of ADF/cofilin regulation and growth cone morphology by carefully examining the processes by which attractive extracellular cues modify ADF/cofilin activity and, thereby, localized actin. dynamics to facilitate turning (Marsick et al., 2010). Two embryonic culture systems were employed in parallel, one utilizing netrin-1 applied to temporal retinal neurons and the other utilizing NGF applied to embryonic DRG neurons. Netrin-1/retinal neurons represent pathfinding in the central nervous system whereas NGF/DRGs are a model system for understanding pathfinding in the peripheral nervous system. The similar results obtained in these studies suggest that the underlying signaling pathways are highly conserved. This work concludes that chick growth cones respond to attractive cues by generating a gradient of elevated ADF/cofilin activity toward the side of the growth cone exposed to the highest concentration of attractive cue. Furthermore, this work made use of chariot vector packaged active ADF/cofilin applied as a gradient from a local source, which caused growth cones to steer toward this source. Turning did not occur toward chariot vector alone or chariot vector packaged control proteins. The turning response required the severing activity of cofilin to generate an increase in F-actin barbed ends to nucleate actin polymerization on the turning side. These findings however seemingly contradict previous work in which Xenopus laevis growth cones, in response to BMP7, turned toward the side of reduced ADF/cofilin activity (Wen et al.,2007). Such contradictory findings may underscore the complexity of actin regulation. Both the total local actin concentration, the ratio of F-actin to free/globular (G)-actin, and the ratio of active cofilin to actin play critical roles in how actin dynamically responds to ADF/cofilin (Bamburg and Bernstein, 2010). This manuscript concludes by examining and documenting differences in the first two of these parameters between growth cones from Xenopus and chick. Chick growth cones contain higher levels of actin than Xenopus growth cones and a higher F- to G-actin ratio, perhaps explaining the need to activate ADF/cofilin in chick and inhibit it in Xenopus to achieve the same balance in severing and assembly dynamics. A stylized cartoon depicts the overall summary of this study as it pertains to chick neuron growth cone turning (Figure 2).

Figure 2:

Growth cone turning in response to an attractive cue. Neurotrophins (NT) increase the number and length of filopodia, decrease the pool of G-actin and increase F-actin barbed-(+) ends on the leading edge of a turning growth cone. A gradient of de-phosphorylated (active) ADF/cofilin (AC) forms, driving actin re-organization at the leading edge. Retrograde actin flow remains stable indicating the increased actin polymerization drives these observed changes. Both a Trk inhibitor (K252a) and CA-LIMK block the effect of NT on F-actin, and the actin depolymerizing agent Cytochalasin-D blocks NT-driven increased (+)-ends.

Whereas the outcome of these studies reveals a greater hidden complexity at the level of regulation of actin dynamics, Letourneau and colleagues have continued to advance our understanding of the important linkages between signaling, cytoskeletal dynamics, adhesion and pathfinding. Although his leadership in the development of this journal has prompted us to recognize his 65^th birthday, it is his sense of humor, collegiality and scientific contributions that are most appreciated by those of us who have had the opportunity to work.with him. We look forward to his continuing insights into the mechanisms of growth cone behavior for many years to come.