Michele L. Lemons

Intact aggrecan and chondroitin sulfate-depleted aggrecan core glycoprotein inhibit axon growth in the adult rat spinal cord

Aggrecan is a chondroitin sulfate (CS)/keratan sulfate (KS)-substituted proteoglycan (PG) abundant in cartilage which is also present within the mammalian embryonic, adult, and injured adult central nervous system (CNS). Although its role within the CNS is not clear, cell culture studies show that when substituted with CS, aggrecan inhibits neurite extension. To better understand the inhibitory effect of aggrecan on injured adult axons in vivo, we developed a model to independently test intact aggrecan and CS-depleted aggrecan core glycoprotein. Acute rat spinal cord hemisection cavities were filled with a growth-promoting matrix, Matrigel, and severed dorsal rootlets were placed into this matrix. This created an assay in which axons readily grew. The extent of ingrowth in this baseline assay was compared to the ingrowth in Matrigel loaded with intact aggrecan or the purified core glycoprotein of aggrecan. Our results show that both intact aggrecan and equivalent concentrations of the core glycoprotein component significantly inhibit axonal growth in this model system. These results confirm that aggrecan can inhibit the growth of adult axons in vivo and suggest that the inhibitory effects of aggrecan may be mediated, at least in part, by structures located on the core glycoprotein in the absence of the bulk of the CS chains.

Integrin signaling is integral to regeneration.

The inability of the adult injured mammalian spinal cord to successfully regenerate is not well understood. Studies suggest that both extrinsic and intrinsic factors contribute to regeneration failure. In this review, we focus on intrinsic factors that impact regeneration, in particular integrin receptors and their downstream signaling pathways. We discuss studies that address the impact of integrins and integrin signaling pathways on growth cone guidance and motility and how lessons learned from these studies apply to spinal cord regeneration in vivo.

Extracellular matrix in spinal cord regeneration: getting beyond attraction and inhibition.

INTRODUCTION The expression of matrix molecules in the CNS [1,2] and the role of matrix in development [3–5] and regeneration [6,7] have recently been the subject of excellent reviews. It is evident that matrix proteins, particularly proteoglycans, make major contributions to regenerative failure in the adult CNS. Although a number of growth inhibiting molecules are expressed in CNS myelin and on glia, recent data suggests that these molecules are not sufficient to prevent the regeneration of adult neurons. Adult neurons transplanted into intact [8] or degenerating [9] adult white matter tracts are able to robustly regenerate. Regeneration of adult neurons invariably fails when extending axons encounter the glial scar associated with the injury site, suggesting that molecules expressed in the scar play a critical role in adult regeneration failure. Following injury, a large number of extracellular matrix (ECM) proteins that are known to influence the regeneration of neurons in culture are up-regulated or altered in regions of scarring (Table 1). Many of these same molecules are up-regulated following injury in peripheral nerve, where regeneration occurs [10]. Historically, the thinking in the field of regeneration research has been dominated by the relatively simple view that specific matrix molecules either promote or inhibit the growth of neurons. Therapeutic approaches have been directed towards increasing the expression or the efficacy of the former and minimizing the negative impact of the latter. There are reasons to believe, however, that this view does not fully reflect the complexity of the problem. The relationship between growth cone motility and matrix adhesion is not simply a matter of more being better (see Fig. 1). In addition, the response of neurons to specific molecules no longer appears to be a fixed property of the molecule itself, but rather a complex interaction between the molecule, the precise molecular composition of the environment, the particular type of neuron interacting with that environment and the recent history of the neuron. Here we will briefly discuss the role of cell adhesion in growth-cone migration. We will then consider molecules typically thought to be either ‘growth-promoting’ or ‘growthinhibiting’, and review the evidence that suggests this molecular dualism is an inadequate description of how neurons interact with matrix molecules in development and regeneration. MOTILITYANDMATRIXADHESION Most matrix proteins that are considered to be growthpromoting also promote cell adhesion. Indeed, cell adhesion is thought to be a required first step for migration. The ability of adhesive matrix proteins to both promote and prevent the migration of non-neuronal cells has been well studied (Fig. 1). Theoretical calculations and empirical data indicate that non-neuronal cells will only migrate over a narrow range of matrix concentrations where cells are adhered strongly enough to generate traction, without being so strongly adhered that they are unable to change position [11,12]. Surprisingly, this well-established relationship between motility and adhesion has not been consistently applied to the study of growth-cone migration. In some situations, embryonic neurons appear to be a rare exception to the rule that migration will only occur within a narrow window of matrix concentrations. Growth cones of embryonic neurons efficiently migrate on concentrations of the ECM molecule laminin that vary over several orders of magnitude [13–15]. The ability of neurons to migrate over a wide range of laminin concentrations is due to a very unusual regulation of neuronal receptors for laminin [15]. Laminin receptors are down-regulated in response to high laminin concentrations, thereby reducing neuronal adhesion and allowing an intermediate level of attachment to be maintained over a wide range of absolute ligand concentrations. Laminin-receptor levels in embryonic neurons are modulated by desensitization, i.e., receptor is removed from the surface in response to high ligand levels by endocytosis. Thus, embryonic neurons, like all cells, migrate only at intermediate levels of attachment, but (at least on laminin substrata) embryonic neurons are able to dynamically adjust their levels of attachment by adjusting expression of receptors. While embryonic neurons are able to compensate for a wide range of laminin concentrations, the response of growth cones to other matrix molecules that utilize different receptor systems is unknown. There is evidence that the receptor molecule L1 (also known as Ng-CAM) participates in endocytotic recycling in growth cones [16–18], but whether the absolute levels of L1 compensate for availability of ligand to maintain a constant level of L1-mediated attachment is unknown. Nothing is known about the regulation of other receptors that promote neuronal adhesion to components of scar matrix. Moreover, very little is

Combined integrin activation and intracellular cAMP cause Rho GTPase dependent growth cone collapse on laminin-1.

Cyclic nucleotides regulate the response of both developing and regenerating growth cones to a wide range of guidance molecules through poorly understood mechanisms. It is not clear how cAMP levels are regulated or how they translate into altered growth cone behavior. Here, we show that intracellular cAMP levels are influenced by substrata and integrin receptors. We also show that growth cones require a substratum-specific balance between cAMP levels, integrin function and Rho GTPases to maintain motility and prevent collapse. Embryonic chick dorsal root ganglion neurons plated on different concentrations of laminin extend growth cones at similar speeds, yet have distinct levels of integrin expression, integrin activation and intracellular cAMP levels. Either increasing cAMP signaling or activating integrins enhances the rate of growth cone motility, but only on substrata where these two factors are endogenously low (i.e. low concentrations of laminin). Surprisingly, combining these two positive manipulations induces growth cone collapse and retraction on laminin but not on fibronectin. Collapse and retraction on laminin are Rho and Rac1 GTPase dependent and are associated with internalization of integrins, the primary receptors responsible for adhesion. These observations define a novel pathway through which cAMP influences growth cone motility and establish a link between integrins, cAMP and Rho GTPases in growth cones.

Adaptation of Sensory Neurons to Hyalectin and Decorin Proteoglycans

Proteoglycans are abundantly expressed in the pathways of developing and regenerating neurons, yet the responses of neurons to specific proteoglycans are not well characterized. We have shown previously that one chondroitin sulfate proteoglycan (CSPG), aggrecan, is potently inhibitory to sensory axon extension in short-term assays and that over time, embryonic neurons adapt to aggrecan-mediated inhibition through the transcriptional upregulation of integrin expression (Condic et al., 1999). Here, we have compared the response of embryonic sensory neurons to structurally distinct CSPGs that belong to either the hyalectin (or lectican) family of large, aggregating proteoglycans or the decorin (or small leucine-rich proteoglycan) family of smaller proteoglycans. Both of these structurally diverse proteoglycan families are expressed in developing embryos and inhibit outgrowth of embryonic sensory neurons in short-term cultures. These results document a previously uncharacterized inhibitory function for the decorin-family proteoglycan biglycan. Interestingly, embryonic neurons adapt to these diverse proteoglycans over time. Adaptation is associated with upregulation of select integrin α subunits in a proteoglycan-specific manner. Overexpression of specific integrin α subunits improves neuronal regeneration on some but not all decorin-family CSPGs, suggesting that neurons adapt to inhibition mediated by closely related proteoglycans using distinct mechanisms. Our findings indicate that CSPGs with diverse core proteins and distinct numbers of chondroitin sulfate substitution sites mediate a similar response in sensory neurons, suggesting that increased integrin expression may be an effective means of promoting axonal regeneration in the presence of diverse inhibitory proteoglycan species in vivo.

Integrins and cAMP mediate netrin-induced growth cone collapse.

Growth cones integrate a remarkably complex concert of chemical cues to guide axons to their appropriate destinations. Recent work suggests that integrins contribute to axon guidance by interacting with a wide range of extracellular molecules including axon guidance molecules, by mechanisms that are not fully understood. Here, we describe an interaction between integrins and netrin-1 in growth cones that contributes to growth cone collapse. Our data show that netrin-1 causes growth cone collapse in a substratum-specific manner and is integrin-dependent. Netrin-1 causes collapse of cultured chick dorsal root ganglion (DRG) growth cones extending on high levels of laminin-1 (LN) but not growth cones extending on low levels of LN or on fibronectin. Blocking integrin function significantly decreases netrin-induced growth cone collapse on high LN. Netrin-1 and integrins interact on growth cones; netrin-1 causes integrin activation, a conformational shift to a high ligand-affinity state. Netrin-1 directly binds to integrin α3 and α6 peptides, further suggesting a netrin-integrin interaction. Interestingly, our data reveal that netrin-1 increases growth cone levels of cAMP in a substratum-specific manner and that netrin-induced growth cone collapse requires increased cAMP in combination with integrin activation. Manipulations that either decrease cAMP levels or integrin activation block netrin-induced collapse. These results imply a common mechanism for growth cone collapse and novel interactions between integrins, netrin-1 and cAMP that contribute to growth cone guidance.

Isolation and culture of dissociated sensory neurons from chick embryos.

Neurons are multifaceted cells that carry information essential for a variety of functions including sensation, motor movement, learning, and memory. Studying neurons in vivo can be challenging due to their complexity, their varied and dynamic environments, and technical limitations. For these reasons, studying neurons in vitro can prove beneficial to unravel the complex mysteries of neurons. The well-defined nature of cell culture models provides detailed control over environmental conditions and variables. Here we describe how to isolate, dissociate, and culture primary neurons from chick embryos. This technique is rapid, inexpensive, and generates robustly growing sensory neurons. The procedure consistently produces cultures that are highly enriched for neurons and has very few non-neuronal cells (less than 5%). Primary neurons do not adhere well to untreated glass or tissue culture plastic, therefore detailed procedures to create two distinct, well-defined laminin-containing substrata for neuronal plating are described. Cultured neurons are highly amenable to multiple cellular and molecular techniques, including co-immunoprecipitation, live cell imagining, RNAi, and immunocytochemistry. Procedures for double immunocytochemistry on these cultured neurons have been optimized and described here.

Neurexin directs partner-specific synaptic connectivity in C. elegans

In neural circuits, individual neurons often make projections onto multiple postsynaptic partners. Here, we investigate molecular mechanisms by which these divergent connections are generated, using dyadic synapses in C. elegans as a model. We report that C. elegans nrx-1/neurexin directs divergent connectivity through differential actions at synapses with partnering neurons and muscles. We show that cholinergic outputs onto neurons are, unexpectedly, located at previously undefined spine-like protrusions from GABAergic dendrites. Both these spine-like features and cholinergic receptor clustering are strikingly disrupted in the absence of nrx-1. Excitatory transmission onto GABAergic neurons, but not neuromuscular transmission, is also disrupted. Our data indicate that NRX-1 located at presynaptic sites specifically directs postsynaptic development in GABAergic neurons. Our findings provide evidence that individual neurons can direct differential patterns of connectivity with their post-synaptic partners through partner-specific utilization of synaptic organizers, offering a novel view into molecular control of divergent connectivity.