BIOLOGICAL OVERVIEW

Cyclin-dependent kinase 5 (Cdk5) is one of a subfamily of Cdks involved in the control of cell differentiation and morphology rather than cell division. Specifically, Cdk5 and its activating subunit, p35, have been implicated in growth cone motility during axon extension. Both Cdk5 and p35 are expressed in post-mitotic neurons and are localized to growth cones. The Cdk5-p35 complex interacts with the Rac GTPase (see Drosophila Rac1), a protein required for growth cone motility. Studies using cultured neurons have suggested that Cdk5 activity controls the efficiency of neurite extension. Mutant mice lacking p35 exhibit subtle axon-guidance defects, but these mice have severe defects in neuronal migration, making it difficult to define precisely the role of the Cdk5-p35 complex in vivo. Drosophila Cdk5 functions in axon patterning in the Drosophila embryo. Although the data support the idea that Cdk5-p35 is involved in axonogenesis, they do not support the view that Cdk5 simply promotes growth cone motility. Instead, disrupting Cdk5 function causes widespread errors in axon patterning (Connell-Crowley, 2000 and references therein).

The functional consequences of altering Drosophila Cdk5 activity were tested. Cdk5, Cdk5dn and/or Cdk5 activator-like protein (also known as Dp35), the Drosophila counterpart of the vertebrate Cdk5 activating subunit, were ectopically expressed in post-mitotic neurons using the elav-GAL4 driver, and motor nerves innervating lateral and dorsal body wall muscles of stage 16-17 embryos were examined. Altering neuronal Cdk5 activity leads to two important observations. (1) Either increasing or decreasing Cdk5 activity produces errors in axon patterning. (2) There is no simple correlation between Cdk5 activity and axon/dendrite length. When endogenous neuronal Cdk5 is ectopically activated by expressing one copy of Dp35, subtle nerve defects appear in 32%-40% of the embryos. The majority of defects occur in the transverse nerve (TN), with a few defects observed in segmental nerve a (SNa) and the intersegmental nerve (ISN). In contrast to observations in cultured neurons, overextended nerves, stalled nerves, and nerves with errors in pathfinding and target recognition were found. Two copies of Dp35 increases the percentage of embryos with defects, suggesting that the effects of Dp35 are dose dependent. The severity of the phenotype is greatly enhanced by coexpressing Cdk5 with Dp35; 80% of the embryos exhibited nerve defects that were more severe and widespread than when Dp35 alone was expressed. Nevertheless, the spectrum of phenotypes observed (overextended, stalled and misrouted nerves) was similar to that seen when Dp35 alone was expressed (Connell-Crowley, 2000).

Neuronal Cdk5 activity was decreased by expressing dominant negative Cdk5 (Cdk5dn). One copy of Cdk5dn has no effect on axon patterning, but two copies cause 32%-35% of the embryos to exhibit subtle nerve defects. The majority of defects occur in the TN, with some defects in SNa and the ISN. In addition, a similar spectrum of phenotypes as that seen with increased Cdk5 activity is seen. With four copies of Cdk5dn, the percentage of embryos with defects increases to 56%, indicating that the effects of Cdk5dn, like those of Dp35, are dose dependent. Cdk5dn is likely to act, in part, by titrating endogenous Dp35, because coexpression of two copies of Cdk5dn with two copies of Dp35 suppresses the axonal defects produced by two copies of Dp35 alone (31% versus 57%). The phenotypic suppression observed with Cdk5dn/Dp35 coexpression contrasts with the phenotypic enhancement caused by Cdk5/Dp35 coexpression and indicates that the Cdk5/Dp35-induced defects require kinase activity. Additionally, the phenotypes of Cdk5 and Cdk5dn appear to be mutually suppressive because, while the effects of either Cdk5 or Cdk5dn are dose dependent, coexpression of two copies of Cdk5 with two copies of Cdk5dn causes a smaller percentage of embryos to exhibit defects than does four copies of Cdk5dn (26% versus 56) (Connell-Crowley, 2000).

The data presented confirm that Cdk5 is involved in axon development in vivo, consistent with predictions of Cdk5 function from studies using cultured neurons. Analysis of the phenotypes resulting from Cdk5/Dp35 expression in vivo reveal, however, a different picture of Cdk5 function than that observed in cultured neurons. Experiments using cultured neurons suggest a simple correlation between Cdk5 activity and the efficiency of neurite outgrowth. However, in these experiments, it was not possible to examine axon patterning. It is clear that Cdk5 and p35 are not absolutely required for axon growth. Rather, altering Cdk5 activity degrades the accuracy of axon patterning: either increasing or decreasing activity promotes or retards axon growth, and causes errors in pathfinding and target recognition. Perhaps the main function of Cdk5 is to ensure coordination of the many signaling pathways that are active in the growth cone, akin to the 'checkpoint' function of cell-cycle Cdks, and not simply to increase growth cone motility. Such a 'surveillance' mechanism would be consistent with the relatively subtle phenotypes observed when Cdk5 activity is modulated (Connell-Crowley, 2000).

Cdk5, like all Cdks, must associate with a regulatory subunit to become active. Thus, while Cdk5 is expressed ubiquitously during development, the spatial and temporal expression of its regulatory subunit limits where and when Cdk5 can be active. An examination was carried out to see whether a Cdk5 regulatory subunit is present at a time and place consistent with a role in axon outgrowth in the Drosophila embryo (Connell-Crowley, 2000).

Dp35, the activating subunit of Cdk5 is expressed in neurons at a time and place consistent with a role in axon development in vivo. Dp35 is expressed exclusively in the nervous system,with strong expression in the brain,ventral nerve cord and post-mitotic neurons of the peripheral nervous system (PNS). The earliest Dp35 expression was observed at stage 12 in clusters of cells in the central nervous system (CNS) where the neurons that extend the first CNS axons reside. Dp35 expression is not observed in neuroblasts, suggesting that fly Dp35-Cdk5, like mammalian p35-Cdk5, is not involved in cell division. Also examined was the subcellular localization of Dp35-Myc expressed in a small subset of neurons using the 15J2 GAL4 driver. Consistent with the localization of mammalian p35, Dp35-Myc protein expressed in CNS dMP2 neurons is present in the neuron cell body and axon, including the growth cone. Dp35-Myc, expressed in PNS lateral chordotonal neurons, is present in neuron cell bodies, dendrites and axons (Connell-Crowley, 2000).

It is interesting that either increasing or decreasing Cdk5 activity causes qualitatively similar nerve defects. This is reminiscent of the effects produced when activity of the Rac GTPase is increased or decreased. As p35 interacts with Rac, it will be of interest to determine whether the axonal effects of Cdk5/Dp35 in vivo are mediated or regulated by Rac. More generally, the ability to modulate Cdk5 activity in vivo will permit Cdk5 to be placed genetically within the signaling pathways that control growth cone motility and guidance (Connell-Crowley, 2000).

The rest of this essay will consider the role of mammalian Cdk5 in regulating neuronal adhesion and cytoskeletal dynamics (reviewed in Homayouni, 2000). Cdk5 kinase activity is not detected in dividing cells. In fact, the catalytically active form of Cdk5 is present only in differentiated neurons of the brain, where it associates with a neuron-specific 35 kDa regulatory subunit, p35. Cdk5 expression gradually increases during neurogenesis until it reaches maximum levels in the adult brain. Recent evidence, reported by Kwon (2000) suggests that Cdk5 plays a critical role in the regulation of N-cadherin-mediated cell adhesion during cortical development. Insights into the function of Cdk5 and p35 come from studies on mice with targeted disruption of the gene encoding either Cdk5 or p35 (Chae, 1997; Ohshima, 1996). The two mutants were found to exhibit similar abnormalities in the laminar structure of the cerebral cortex. But the cdk5-/- mice have additional defects in other brain structures and died shortly after birth. A small fraction of p35-/- mice die after spontaneous seizures, and greater than 50% died after seizure-inducing treatments that are not fatal in normal mice. The differences between the phenotypes of cdk5-/- and p35-/- mice may be due to the presence of other Cdk5 regulatory subunits in the brain. Thus far, several other Cdk5 regulators have been identified, including p39, Munc18 (p67) and a truncated form of p35, p25. At least in the forebrain, both p35 and Cdk5 play an essential role in the formation of the cortical layers during development (Homayouni, 2000 and references therein).

The mammalian cortex is assembled through a choreographed series of events that ultimately result in the segregation of neurons with similar properties into six layers. In the earliest phase of development, the preplate, composed of Cajal-Retzius and subplate neurons, is formed between the pial surface and the ventricular zone (VZ), where cells are actively dividing. Cells exit the cell cycle in the VZ and migrate radially outward toward the pial surface along glial fibers. The first-born neurons migrate past the subplate, displacing this layer away from the Cajal-Retzius cells in an area known as the marginal zone. Splitting of the preplate requires Reelin, a large extracellular protein secreted by Cajal-Retzius cells. The next wave of post-mitotic neurons migrates along the same fibers, past the subplate and the older neurons in the cortical plate, before inserting beneath the marginal zone. In this manner, the classical inside-out pattern of the neo-cortex arises in which the sequential generation of layers II-VI occurs beneath the marginal zone (Homayouni, 2000).

In the cortices of both p35-/- and cdk5-/- mutant mice,the marginal zone is unaffected and Reelin expression is normal. The preplate splits in p35-/- and cdk5-/- brains, and the first-born population of neurons -- the future layer VI -- appears to migrate correctly past the subplate cells. But the migration of the later-born neurons is impeded, in the mutants, such that the second wave of neurons fails to migrate past the established layer of cortical neurons and, instead, accumulates below the subplate. On the basis of these observations, it has been proposed that p35 and Cdk5 are required for neurons to bypass one another during corticogenesis, although they are not required for splitting of the preplate (Homayouni, 2000).

To form distinct neuronal layers in the developing cortex, one would expect migrating neurons to recognize and preferentially adhere to their cohorts in the respective layers. Indeed, in aggregation assays, early-born cortical neurons selectively associate with one another. A good candidate molecule to mediate this cell-cell adhesion is the Ca 2+-dependent neuronal adhesion molecule N-cadherin (Drosophila homolog: Cadherin-N). N-cadherin is expressed transiently throughout the developing cortical plate, but it persists only in the deepest layers of the postnatal brain. Thus, it is possible that N-cadherin plays a role in maintaining the integrity of the cortical plate at the time newly generated neurons migrate past. This model presents a paradox, however: how can migrating neurons expressing the homophilic protein N-cadherin bypass cortical plate neurons that also express N-cadherin? The recent work of Kwon (2000) on p35-interacting proteins has shed some light on this subject. A novel interaction between p35 and the intracellular regulator of N-cadherin, alpha-catenin was identified. Expression of Cdk5 and p35 in cultured cells decreases the association between alpha-catenin and N-cadherin, resulting in decreased cell adhesion. Conversely, in neuronal aggregation assays, loss of Cdk5 kinase activity by pharmacological inhibition or ablation of p35 resulted in increased cell adhesion. These results provide important insights into the molecular mechanism by which Cdk5 and p35 regulate neuronal migration in the developing cortex (Homayouni, 2000 and references therein).

On the basis of these findings a new molecular model can be put forward to describe cortical development. As neurons exit the cell cycle, they express p35, which activates Cdk5, causing down-regulation of N-cadherin-mediated cell adhesion. This allows the neurons to migrate past their predecessors in the cortical plate, which express large amounts of N-cadherin. In cdk5-/- and p35-/- brains, the migrating neurons are unable to down-regulate N-cadherin, so they cannot bypass the earlier-born cortical plate neurons. This may explain why later-born neurons accumulate beneath the subplate in these mutant mice (Homayouni, 2000 and references therein).

One question remaining is how does a migrating neuron that has reached the marginal zone down-regulate the p35/Cdk5 kinase pathway and reactivate N-cadherin mediated adhesion? Reelin has been proposed to function as a stop signal for migrating neurons. It is possible that Reelin acts as a stop cue by inhibiting the p35/Cdk5 kinase pathway. Thus, Reelin may activate N-cadherin-mediated cell adhesion to terminate migration and initiate homotypic adhesion between the new arrivals and neurons resident in the cortical plate. Recent results by Hiesberger (1999) support this hypothesis. Tau, a substrate of Cdk5, is hyperphosphorylated in reeler mutant brains. One explanation for this observation is that Reelin down-regulates p35/Cdk5 kinase activity under normal conditions. Therefore, it is possible that, in reeler mice, p35/Cdk5 kinase remains active in migrating neurons, resulting in a failure of post-mitotic neurons to adhere to one another in the presumptive marginal zone. This may in part explain the disruption of cortical layers observed in reeler brains (Homayouni, 2000 and references therein).

Increasing evidence points to the fact that Cdk5 is a key regulator of neuronal function. It modulates cell adhesion and cytoskeletal dynamics, processes that are essential during development and in the adult nervous system. Not surprisingly, deregulation of Cdk5 has severe consequences for brain development, causing disruption of cell positioning and perinatal lethality. In the adult nervous system, aberrant activation of Cdk5 by p25 is associated with neurodegeneration. Cdk5 family members play critical roles in the control of cell division and they have attracted attention as a target for anti-cancer therapies. Perhaps Cdk5 plays a similarly critical role in postmitotic, developing and adult neurons. It may also attract attention as a therapeutic target for neurodegenerative diseases (Homayouni, 2000).

GENE STRUCTURE

Transcript length - 1.6 kb

Exons - 4

PROTEIN STRUCTURE

Amino Acids - 294

Structural Domains

The D. melanogaster homolog of mammalian CDK5 has been cloned and its chromosomal location determined. The gene for Cdk5 consists of 4 exons separated by 3 short introns ranging in size from 61-160 bp. Northern blot analysis has revealed a single mRNA of approximately 1.6 kb that is expressed at highest levels in the adult fly. The putative amino acid sequence for Drosophila Cdk5 predicts a protein with a mass of approximately 32 kDa that is 77% identical to its mammalian counter-parts. Drosophila Cdk5 gene is located in polytene chromosomal region 52BC of the right arm of chromosome 2 (Hellmich, 1994).

date revised: 30 May 2000

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.