生物谷報道：神經(jīng)遷移是近年來的研究的熱點,，有關這方面進展十分迅速,，本篇報道綜述了FAK和Cdk5在其中的關鍵作用，使讀者對神經(jīng)遷移的分子機制的進展有一個較全面的認識,。

The molecular mystery of neuronal migration: FAK and Cdk5

Margareta Nikolic

The basic building blocks of a cell are its cytoskeletal proteins, the orderly but dynamic organization of which is essential. How signalling molecules regulate the cytoskeleton in the developing nervous system is still largely unknown. A recent breakthrough sheds light on a pathway involving Cdk5 (cyclin-dependent kinase 5) and FAK (focal adhesion kinase), demonstrating their role in regulating microtubule structure and thus nuclear positioning in radially migrating cortical neurones.

The development of a multicellular organism depends largely on the ability of its cells to move to specific predetermined areas in a coordinated and controlled manner. This is very apparent in the mammalian central nervous system (CNS) particularly in the cerebral cortex. Here, postmitotic neurones undergo long-range directed migration from deep within a proliferative ventricular zone to create the adult cortex, a structurally and functionally complex tissue [1]. Remarkably, neurones that cease to proliferate at later developmental stages migrate past the neurones ‘born’ earlier, thus creating an inverted layering pattern ( Box 1). Errors in these migratory paths can have severe consequences, as seen in several human diseases (lissencephaly type I and II, X-linked double-cortex malformation and periventricular heterotopias) and mouse loss-of-function models (reelin, reelin receptors Vldlr and ApoER2, Dab1, filamin and Cdk5 knockouts) [2 and 3]. Despite our extensive knowledge of the anatomy and cellular structure of the CNS, until recently the molecular mechanisms underlying its development remained a mystery. One of the major breakthroughs has been the discovery that the proline-directed serine/threonine kinase Cdk5 plays an essential role in the developing CNS [4 and 5]. Now, novel and exciting findings from the laboratory of Li-Huei Tsai [6] reveal that in migrating neurones nuclear movement is regulated by a specialized microtubule (MT) structure, the organization of which depends on FAK phosphorylation by Cdk5. The highlights and implications of these data are discussed here. These findings are important because, together with the data of Hoshino and colleagues [7], they demonstrate the benefits of a novel method to study the role of signalling proteins, such as kinase cascades in the developing cortex, and the nature of the questions that can be addressed in the future.

Box 1. Mouse corticogenesis in the presence and absence of Cdk5 activity

The mammalian cerebral cortex forms in a temporally and spatially organized manner. In mice, this begins ten days postfertilization (E10) and proceeds after birth [1]. Neurones that constitute the cerebral cortex arise in the proliferating neural epithelium that lines the inner (ventricular) surface of the neural tube and is termed the ventricular zone (VZ). The first postmitotic neurones of the cortex generate a layer called the preplate (PP), which is subsequently split apart by layer VI neurones, to form the marginal zone (MZ) and subplate (SP) layer (Figure Ia). The mammalian cortex forms in an inside-out manner, that is, the younger the neurone, the further it has to migrate to reach its final destination. In this way, layer V neurones pass through layer VI, layer IV neurones pass through layers VI and V, layer III neurones migrate through layers VI, V and IV, whereas layer II neurones work their way through layers VI, V, IV and III. It appears that the general rule is to ‘stop under the marginal zone’, which is most probably caused by the effects of the reelin protein [3]. Mice engineered to lack Cdk5 (die at birth) or its activator p35 (viable adults) have abnormally structured cortices owing to defects in neuronal migration [4 and 5]. Interestingly, the layer VI neurones in mutant mice appear to migrate correctly and split the PP; however, subsequent layers form under, rather than over, the SP leading to an abnormally organized adult cortex in p35 knockout mice (Figure Ib). Postnatal analysis of cortices in Cdk5 knockout mice is prevented by their perinatal lethality. Currently, it is not clear whether the abnormal axonal tracts seen in p35−/− mice are because of the incorrect positioning of cortical neurones or other cell-autonomous defects. During normal corticogenesis, the method of neuronal migration changes. At early stages, neuronal precursors are seen to contact the ventricular and pial surface (PS), extending across the entire cortical plate (CP) (Figure Ia, cell 1). Differentiation into a neurone is marked by the detachment of these cells from the ventricular surface and the dorsal migration of their soma through somal translocation and nucleokinesis ( Figure Ia, cell 2) [10]. As the cortex thickens, neurones migrate with the aid of radially extending glia (Figure Ia, cells 3 and 5). These locomoting neurones have a motile leading process that maintains a relatively constant distance from the soma. Once the leading process contacts the pial surface, the cells reach their final destination by somal translocation ( Figure Ia, cell 6). IZ (intermediate zone) is the axon-rich area of the cortex, whereas SVZ/SEPZ (subventricular zone/subependymal zone) contains proliferating precursors of the adult cortex. The arrows indicate the direction of migration.

Figure I.

1. Cdk5 phosphorylates FAK in migrating neurones
Cdk5 is a major regulator of cytoskeletal components known to affect neuronal MTs, microfilaments and intermediate filaments during development and neurodegeneration (Box 2) [8]. Animal models with compromised Cdk5 activity (genetically modified to lack Cdk5 or its activating protein, p35) have disrupted radial migration of neurones from the ventricular zone to the cortical plate ( Box 1) [3]. Thus, how does Cdk5 control neuronal migration? The recent findings of Tsai and colleagues [6] have provided part of this answer, because we now know that, in vivo, Cdk5 phosphorylates FAK on S732. Interestingly, in migrating neurones, unlike FAK, S732-phosphorylated FAK does not colocalize with actin microfilaments [9]. Instead, it accumulates in and associates with distinct MT structures that stem from MT-organizing centres (MTOCs), extending around the nucleus with a branched fork-like appearance. Neurones devoid of Cdk5 or overexpressing a FAK mutant that does not phosphorylate (FAK732A) display a malformed MT fork, resulting in a rounded nuclear morphology and consequently migration defects. Therefore, Cdk5-mediated neuronal migration can be attributed, at least in part, to the phosphorylation of FAK.

Box 2. The function and regulation of p35, Cdk5 and FAK

Cdk5 is a proline-directed serine/threonine kinase and a member of the Cdk family [23]. It is highly homologous to Cdk2 and Cdc2; however, it does not promote cell-cycle progression. Kinase-active Cdk5 has only been detected in nonproliferating cells, primarily postmitotic neurones. To function, Cdk5 needs to associate with a regulatory partner at least two of which are known – p35 and p39. In addition to activating Cdk5, p35 and p39 affect the subcellular localization of this kinase. Cdk5 phosphorylates many proteins and affects the organization of all three cytoskeletal elements microfilaments, microtubules (MTs) and intermediate filaments. FAK is a nonreceptor tyrosine kinase that becomes phosphorylated and consequently activated after engagement with integrins [11 and 16]. The generation of multiple phosphotyrosines and phosphoserines is thought to create docking sites for several classes of signalling molecules [23]. FAK is, therefore, better known for its role as an adaptor protein than as a kinase. The most commonly studied FAK is the ubiquitously expressed (FAKo), whereas the nervous system contains several alternatively spliced isoforms, some of which result from the insertion of short peptide sequences into the coding region. In some cases, the splice variants exhibit enhanced FAK kinase activity; however, the functional roles of these variants are largely unknown [11]. Recently, FAK was shown to lie directly downstream of Cdk5 in regulating MT organization and neuronal migration in the developing cortex [6]. These two kinases might have overlapping functional roles in several other cellular processes, as depicted in Figure II. Cdk5 and FAK are both enriched in axonal growth cones and can promote the outgrowth of neurites. FAK regulates focal adhesion turnover, enhancing the migration of different cell types and has been implicated in the development of cancer [16]. Cdk5 regulates N-cadherin-dependent neuronal adhesion. Both kinases can affect several aspects of synapse function including presynaptic (membrane recycling) and postsynaptic (acting as the neuromuscular junction, dopaminergic signalling, regulation of the NMDA receptor function and long-term potentiation) roles [8 and 11]. Cdk5 and FAK can also contribute to neurodegeneration, inducing or preventing cell death and affecting the progression of diseases such as Alzheimer's [8 and 24]. Cdk5 currently has a more defined role in axonal guidance; however, this could be because of the early embryonic lethality of FAK-deficient mice that has prevented these analyses in the mouse model. Together, the functional roles of these two kinases have many overlapping aspects, because they might exist in a single signalling pathway or at least affect different aspects of the same biological processes.

Figure II.

A striking observation is that neurones expressing nonphosphorylatable FAK have not completely lost their ability to move, accumulating in the intermediate zone and subplate [6]. Similar findings were recently reported by the Hoshino laboratory upon the expression of a dominant-negative Cdk5 mutant and were observed in p35 and Cdk5 knockout mice ( Figure 1g) [4, 5 and 7]. It is, therefore, possible that Cdk5 activity and FAK phosphorylation are essential for part of the migratory process that does not include its initiation. During cortical formation, neurones migrate using at least two types of radial movement. Early in development, they reach the cortex by somal translocation. As the cortex thickens, neurones move out of the ventricular zone with the help of radial glia (a process referred to as locomotion) and only utilize somal translocation for the late stages of this migratory process when they are in the vicinity of the pial surface ( Box 1). In translocating neurones, the nucleus moves up a long extending process that is attached to the pial surface, resulting in a rapid and constant motion of the cell. By contrast, locomoting cells utilize a leading process that on average retains a constant distance from the soma and these neurones move in short fast bursts [10]. Xie et al. [6] and Kawauchi et al. [7] have demonstrated that Cdk5 and FAK phosphorylated on S732 are essential for locomotion. It would be interesting to determine exactly at what stage the FAK-dependent MT fork is required and whether it has a role in translocating neurones, as well as locomoting neurones. The examination of Cdk5 or FAK conditional knockouts and a nonphosphorylatable FAK knockin mouse model would aid these studies.

Figure 1. The use of in utero electroporation to demonstrate the role of Cdk5 in neuronal migration. (a) In utero electroporation is a novel technique that involves the injection of DNA into the lateral ventricle of immature embryos, often after 14 days of gestation (E14), from outside the uteri of anaesthetized pregnant mice. The embryos are subsequently subjected to electric pulses delivered using forceps-style electrodes, which transport the injected DNA from the ventricle into the surrounding neuroepithelium. The embryos are placed back into the mother and allowed to develop until the required stage for analysis that could be either embryonic or postnatal [7]. (b–d) The application of this technique ensures the transfection of a subset of cortical neurones (layers II–IV) that are known to ultimately migrate into superficial layers of the cortex, as shown by the expression of enhanced green fluorescent protein (EGFP). (b) One day post electroporation, the cells lie deep in the ventricular zone (VZ); (c) three days later they are migrating through the intermediate zone (IZ); (d) and (e) forming the superficial layers of the cortex seen at birth (P0) and four days later (P4). (f) Radially migrating neurones have a distinct morphology. (g) Neurones expressing dominant-negative Cdk5 (DN-Cdk5) are unable to migrate into the cortex forming a layer of cells termed the underplate (UP). Scale bars represent 200 m (b–e) and 10 m (f). Figure adapted from [7]. CP, cortical plate.

2. Regulation of MTs by FAK
The kinase activity of FAK is believed to be secondary to its chaperoning role in assembling complexes containing other kinases (Src, Fyn and PI3K) and their substrates (Box 2) [11]. Accordingly, the phosphorylation of FAK at S732 does not affect its catalytic activity and yet the MT fork is malformed when FAK phosphorylation is decreased. Why? A logical assumption is that the phosphorylation of FAK by Cdk5 positively or negatively affects its ability to assemble a particular protein complex. On the basis of mutant expression, immunostaining and copurification studies, Xie et al. [6] propose that Cdk5 acts to displace FAK from the MTOCs. The identification and characterization of protein(s) that associate with FAK in a phospho-S732-dependent manner is the most appropriate approach to test this model.

3. Neuronal and non-neuronal FAK
In the adult brain, FAK is regulated by neurotransmitters and controls synaptic plasticity (Box 2). The higher levels of FAK seen in embryonic cortex when neuronal migration and neurite outgrowth predominate suggest a nonsynaptic function for this kinase [6 and 12]. In differentiating neurones, FAK is detected in growth cones and can regulate growth-factor- and integrin-induced neurite outgrowth [9, 13 and 14]. Xie et al. [6] have revealed that FAK also controls neuronal migration by affecting their MTs. In other cell types, FAK enhances motility by increasing focal adhesion turnover, which is also induced by repetitive MT targeting [15 and 16]. Putting all these together, one can speculate on the multiple cytoskeletal roles of FAK; however, are these roles ubiquitous or cell-type specific?

Despite its ubiquitous expression pattern, the kinase activity of Cdk5 has been primarily close family member Cdc2 have identical substrate specificities and can both phosphorylate amphiphysin1 and Pak1 [17 and 18]. It is interesting to find out whether the phosphorylation of FAK on S732 is cell-cycle dependent, particularly because, in proliferating cells, FAK is targeted on multiple serine residues, some of which are preferentially phosphorylated during mitosis [19].

4. New tricks to study neuronal migration
A large part of what is known about the mechanism of neuronal migration comes from the studies of human diseases and animal models, both of which take time to obtain and characterize. To understand the processes that underlie neuronal development, it is essential to utilize a manipulative system where modifications in signalling pathways can be made and the consequences can be analyzed at a cellular level. For many years, researchers have utilized slice and explant cultures to examine the behaviour of neurones in the developing forebrain. These techniques are well suited for the analysis of axonal outgrowth and guidance in vitro. However, the applicability of these methods for the analysis of neuronal migration from the ventricular zone is limited, most probably because the disruptions encountered by the ventricular surface during tissue preparation (e.g. the loss of three-dimensional architecture and short-term viability of the cells). The lack of a good in vitro system has therefore been apparent. In utero electroporation is an excellent method to study the migration of cortical neurones. The technique enables alterations of a single or groups of proteins in individual neurones, the consequence of which can be subsequently followed in an in vivo environment [20 and 21]. Xie et al. [6] have clearly demonstrated the power of this technique revealing the importance of FAK phosphorylation by Cdk5 in migrating neurones. In addition, Hoshino and coworkers recently used this approach to reveal the functional role of the small GTPase Rac1, its activators Tiam1 and Stef, as well as one of its downstream targets JNK kinase, in radially migrating cortical neurones [7]. They demonstrated that the inhibition of Tiam1/Stef, Rac1 or JNK signalling arrests neuronal migration. In dissociated neurones, the inhibition of JNK results in decreased phosphorylation of the MT-associated protein Map1B suggesting a role for JNK in inducing MT instability. Interestingly, Map1B is highly expressed in radially migrating cells and might also be a substrate of Cdk5 [22].

5. Concluding remarks
In the developing cortex, neurones undergo many complex processes involving major changes in shape, mobility and function, which are all regulated by a multitude of proteins that interact and affect the function of one other. To understand simple aspects of the signalling pathways that determine neuronal fate and function, it is essential to be able to analyze the consequences of lost or altered protein activity in an appropriate experimental system. The Tsai laboratory [6] have demonstrated this, contributing several important points to the field of neuroscience, one of which is our better understanding of the role Cdk5 plays in vivo. For nearly a decade, it has been known that Cdk5 activity is required in differentiating neurones; however, how it performs this function at a molecular level remained unknown until now. Xie and colleagues [6] reveal a novel role for FAK and an unknown signalling pathway involving Cdk5 in neuronal migration. They link the function of these two kinases with the control of MT organization, for the first time demonstrating the importance of a particular MT structure for nuclear movement during neuronal migration. Another major contribution of their study is the demonstrated applicability of a powerful technique such as in utero electroporation. It is important to note that, in mice, the loss of Cdk5, FAK or Rac1 expression results in embryonic lethality, which in the case of FAK and Rac1 is too early to examine even the onset of corticogenesis. Electroporation of mutant proteins into individual neurones enables the study of their fate in an environment that is otherwise typical of that experienced by wild-type proteins well past the maximum time allowed by knockout mouse models. In addition to in utero electroporation, recent progress in several techniques, including the use of small inhibitory RNAs, green fluorescent marker proteins and live imaging, has set the scene for future major discoveries in the fields of molecular, developmental and cellular neurobiology. Not only will such combinatorial approaches help to answer further questions regarding the role of Cdk5 but also they can be extended to analyze many other proteins with suspected functions in the developing and adult CNS, as already demonstrated by Hoshino and colleagues [7]. This promises an exciting future for neuroscience.