摘要
Semaphorins are multifunctional proteins essential for embryonic development and for various physiological functions in the adult. These proteins interact with members of the plexin and neuropilin families of transmembrane receptors, leading to the activation of a diverse range of intracellular signalling events. A key characteristic of semaphorins is their ability to alter cytoskeletal dynamics, which is important for different cellular processes, including the guidance of neuronal growth cones and neural crest cells, for angiogenesis, and also for tumorigenesis and metastasis. Recent research has increasingly given clear indications of the complex ways in which different cells regulate responsiveness to semaphorins. This diversity is manifested by the presence of a multitude of semaphorin receptor complexes and the regulation of an array of downstream signalling events. Semaphorin-induced cellular responses are not fixed, but can be modified in a context-dependent fashion. Furthermore, it has been suggested that transmembrane semaphorins can participate in reverse signalling, functioning as receptors in specific settings. Despite the wealth of information regarding the biological functions of these molecules, many aspects of the regulation of these functions remain uncertain. However, several molecular features through which semaphorins mediate responses in different cell types are shared by a number of family members, facilitating our understanding of semaphorin signalling. This primer will summarise these common characteristics. Currently, more than 30 semaphorins have been characterised, according to the unifying feature of having a ∼500 amino-acid extracellular semaphorin (Sema) domain. On the basis of different domain compositions of the carboxyl terminus and considerations of the phylogenetic characteristics of these domains, the members of the semaphorin family have been divided into eight classes (Figure 1). Some semaphorins are secreted molecules with no membrane attachment site and therefore act at a distance. Other members contain single-transmembrane-spanning domains or are attached to the cell surface by a glycosylphosphatidylinositol (GPI) anchor and evoke responses in a contact-dependent fashion. Whilst semaphorins function predominantly as ligands, class 1 and class 6 (transmembrane) semaphorins can also participate in reverse signalling. Additional structural features of different members of the semaphorin family include the presence of a single immunoglobulin (Ig)-like domain, a set of thrombospondin type I repeats or a highly charged carboxy-terminal tail. Several semaphorins are subject to proteolytic cleavage, which increases the diverse way in which semaphorin function can be regulated. On one hand, membrane-bound semaphorins can be released from cells producing them through cleavage by matrix metalloproteinases, a mechanism that retains the biological activity of the full-length semaphorin protein. On the other hand, proteolytic cleavage has also been shown to modify the potency of semaphorins. In the light of work that further identified essential functions for heparan and chondroitin sulphate proteoglycans in modulating semaphorin function and cellular behaviour, it appears that the way in which cells respond to semaphorins crucially depends on the composition of the extracellular milieu in which semaphorins are produced and/or presented. The most important receptors for semaphorins are members of a family of single-transmembrane-spanning proteins, termed plexins. Originally identified as cell-surface proteins in the developing Xenopus nervous system, plexins were later discovered to participate in class 3 semaphorin responses by functioning as an essential receptor component and, for the other semaphorin classes, were shown to act as the sole binding and transducing receptor. In the extracellular region, plexin family members encompass a conserved Sema domain, which mediates ligand interactions, and also maintains — at least in the case of members of the PlexinA family — the receptor in an inhibitory conformation. Consistently, loss of the Sema domain renders the receptor constitutively active. In their intracellular region, plexins have an intrinsic GTPase activating (GAP) domain, as well as binding sites for different members of the Ras superfamily of GTPases and for guanine nucleotide exchange factors (GEFs). Docking of intracellular tyrosine kinases to the intracellular domain of plexins has also been reported, which, in turn, regulates plexin phosphorylation following ligand occupation. Other proteins that directly associate with the plexins include those that may connect to cellular adhesion and redox signalling (molecules interacting with CasL; MICAL), those that regulate cyclic nucleotides (Nervy) or those that regulate cellular trafficking events (RanBPM). Similarly, collapse response mediator proteins (CRMPs) and the Src family tyrosine kinase Fyn interact directly with plexins, potentially conveying plexin activation to changes in microtubule dynamics. Plexins also associate with other membrane proteins and these include the receptor tyrosine kinases Off-track (OTK), VEGF receptors, c-Met, Ron and ErbB2. In addition to plexins, the two transmembrane proteins neuropilin-1 and neuropilin-2 function as exclusive binding receptors for the class 3 semaphorins (with the exception of Sema3E). However, binding of Sema3 to neuropilins alone is not sufficient to activate cellular responses, which instead require the formation of ‘holo-receptors’ through association with other proteins at the cell surface. These include PlexinA family members, but also c-Met and VEGFR, as well as cell adhesion molecules of the Ig superfamily (IgCAMs). Of interest in this context is the discovery that soluble IgCAMs can alter cellular responses to semaphorins; neurons that are repelled from a source of Sema3A demonstrate attraction in the presence of a soluble form of the IgCAM L1. Although this phenomenon has so far only been investigated in vitro, soluble forms of IgCAMs have been detected in the developing brain and interfering with their production leads to axon guidance errors. Thus, controlling the expression of soluble IgCAM might turn out be an additional avenue by which semaphorin signals could be modified in vivo. One of the initial intracellular molecular events that occurs following semaphorin binding to plexins involves the receptor's intrinsic GTPase activity, which is itself regulated by small GTPases. It has been shown, for example, that following Sema3A stimulation in neurons the GEF FARP2 dissociates from PlexinA1 and activates the small GTPase Rac. This event leads to the subsequent association of PlexinA1 with another small GTPase, Rnd1, which permits activation of PlexinA1's RasGAP activity. Although this specific situation might reflect signalling responses to Sema3A only, it ssuggests the possibility that plexin activation — and thus semaphorin signalling — might be viewed as a consequence of two sequential events. The first of these is the ability of plexins to associate directly with small GTPases in their activated conformation, which leads to the second common event, the activation of the intrinsic GTPase activity of plexins. This unifying mechanism for plexin function is substantiated by the high degree of conservation of the GTPase–plexin association motif in the intracellular region of all plexin family members, and the fact that several plexin family members have indeed been shown to interact with active GTPases. Similarly, the intrinsic plexin RasGAP activity is found to be present in all plexin family members. Analysis of the structure of PlexinB1 also suggests that the GTPase–plexin interaction induces a conformational change in the plexin extracellular domain, which increases ligand-binding affinity. Thus, through intracellular association with activated GTPases, plexins may be switched to high-affinity binding receptors for semaphorins, an event commonly referred to as ‘inside-out signalling’. To put it simply, cells that maintain high levels of GTPase activity through activation of parallel signalling pathways may turn out to be more responsive to semaphorins. As a consequence of semaphorin engagement with cognate cell-surface receptors, different signal transduction cascades are activated that involve small GTPases, serine/threonine and tyrosine kinases, and various lipid kinases/phosphatases. It is currently unknown whether certain semaphorin receptor combinations regulate specific downstream events. However, it is increasingly apparent that several signalling events are common to the function of a number of semaphorins (Figure 2), such as the regulation of cytoskeletal dynamics and cell adhesion. A key semaphorin-dependent signalling pathway controlling cell migration and linked to semaphorin-induced reorganisation of the actin cytoskeleton involves the actin-binding protein cofilin. Cofilin is regulated by phosphorylation at the carboxyl terminus, an event that inhibits its known actin-binding and actin-severing activity. This phosphorylation is mediated by LIM kinase (LIMK) and is dependent on its upstream regulators Rho-associated kinase (ROCK) and p21-activated kinase (PAK), which take part in a number of semaphorin-mediated responses. For example, Sema3A is known to activate LIMK in neuronal growth cones, and Sema4D signalling has been shown to act upstream of both PAK and ROCK. It is reasonable to expect that cofilin is essential for semaphorin responses, given that it is potentially associated with those responses that rely on rapid changes in the actin cytoskeleton, and this is indeed the case. Examples include the semaphorin-induced growth cone collapse response in neurons, changes in cellular adhesion of dendritic cells, and effects on platelet aggregation. A further avenue for regulating semaphorin responses through cofilin is the cell's translational machinery. It is known that subsets of mRNAs are specifically maintained at sites of increased cytoskeletal dynamics, which is thought to confine the synthesis of their protein products to regions where they are needed most. To date, two independent semaphorin-mediated events have been shown to require localised translation of cofilin: Sema3A-induced growth cone collapse and a morphogenetic programme in the nematode worm. Thus, semaphorins appear to have the ability to activate two divergent signalling systems that converge on the common target cofilin. In the first instance, semaphorins may tune overall levels of cofilin by regulating localised translation, and, at the same time, semaphorin may regulate the severing activity of cofilin by phosphorylation. This dual control offers a mechanism that generates, and potentially amplifies, small changes in signalling at discrete cellular sites. Similar semaphorin signalling relationships seem to occur with RhoA, as this GTPase has been suggested to be controlled by localised translation, at least in neuronal growth cones, and its activity is known to regulate a number of cellular responses evoked by semaphorins. Semaphorin-induced actin remodelling is commonly associated with alterations in the cell's anchorage to the extracellular matrix. Stimulation with semaphorins generally decreases adhesion by reducing integrin activation, but can also increase integrin-dependent adhesion via classical adhesion-associated signalling proteins, such as focal adhesion kinase and MAP kinase. A link between cellular adhesion and the semaphorin response is provided by the small GTPase. R-Ras, an established activator of integrin-mediated adhesion. Following semaphorin stimulation, the intrinsic GTPase activity of plexins targets R-Ras and antagonises integrin-mediated adhesion. FARP2 also suppresses R-Ras activity; however, the additional involvement of this Rho GEF in inhibiting the type-I phosphatidylinositol phosphate kinase PIPKIγ661 provides a further interesting example of divergence of semaphorin signalling pathways. PIPKIγ661 is a lipid kinase targeted to focal adhesions and, through the spatial generation of phosphatidylinositol 4,5-bisphosphate (PIP2) at these sites, promotes interactions between vinculin with actin and talin. Thus far, R-Ras has been portrayed as a key protein that couples semaphorin receptors to intracellular signalling and cell migration. However, its involvement goes beyond controlling semaphorin function through cellular adhesion. In neurons and various cancer cells, semaphorin stimulation results in decreased phosphoinositide 3-kinase (PI3K) signalling and inactivates one of the principle downstream effectors of PI3K, Akt. This process again involves the intrinsic GTPase activity of plexins and R-Ras and also requires PTEN, a phosphoinositide 3-phosphatase widely distributed in mammalian tissues. Conversely, semaphorins have also been shown to activate PI3K signalling, for example, during endothelial cell migration. As well as altering actin polymerisation and cellular adhesion, cell migration relies on the rapid reorganisation of microtubules and, perhaps not unexpectedly, semaphorins have been shown to alter microtubule dynamics, particularly in the context of axon guidance. One prevailing example is the involvement of CRMPs in different semaphorin responses. Semaphorin reduces the activity of glycogen synthase kinase 3 towards the microtubule-binding protein CRMP2, which impairs microtubule polymerisation and affects axonal extension. Additional evidence suggests that semaphorins may regulate microtubule dynamics independently of CRMPs. Our understanding of the functions of semaphorins now extends far beyond their initial characterisation as axon guidance cues, with roles identified in vascular and cardiac development, cancer progression, and the immune system, whilst their important roles in the pathology of various diseases and injury states are becoming increasingly evident. It is now accepted that semaphorins are key regulators of the cytoskeleton and cell adhesion during cell migration, but that they also evoke responses such as cell survival, proliferation and differentiation. Semaphorins stimulate a complex signalling network involving a multitude of receptors and signalling molecules, which allows for a diverse range of outcomes, often in a cell-type-specific manner. Within a given cell type, a particular semaphorin signal can also generate different responses depending on the presence of a variety of modulatory signals, such as cyclic nucleotides, adding a further layer of complexity to the network. Clearly, future questions in this field have to be directed at analysing the significance of semaphorin signalling systems in controlling cellular responses in vivo, which promises to deepen our understanding of the diverse and important functions now attributed to semaphorins.