Type I lissencephaly is a cortical malformation disorder characterized by disorganized cortical layers and gyral abnormalities and associated with severe cognitive impairment and epilepsy. The exact pathophysiological mechanisms underlying the epilepsy and mental retardation in this and related disorders remain unknown. Two genes, LIS1 and doublecortin, have both been shown to be mutated in a large proportion of cases of type I lissencephaly and a milder allelic disorder, subcortical laminar heterotopia (SCLH). Studying the protein products of these genes and the biochemical pathways in which they belong is likely to yield important information concerning both normal and abnormal cortical development. The relationships between the LIS1 and Doublecortin proteins are not yet well defined, but both are believed to play a critical role in cortical neuronal migration. Lis1 is expressed from very early development in the mouse and in both proliferating cells and post-mitotic neurons of the cortex. This protein is likely to have multiple functions since it is a subunit of the enzyme platelet-activating factor acetylhydrolase, which degrades platelet activating factor, and has also been shown to be involved in microtubule dynamics, potentially influencing nuclear migration through its interaction with the dynein motor protein complex. Doublecortin on the other hand is exclusively expressed in post-mitotic neurons and is developmentally regulated. In young developing neurons Doublecortin has a specific subcellular localization at the ends of neuritic and leading processes. This localization, combined with our previous data showing that it is a microtubule-associated protein and that it interacts with adapter complexes involved in vesicle trafficking, suggests a role in the growth of neuronal processes, downstream of directional or guidance signals. The observations summarized here favor the suggestion that whereas LIS1 may play a role in nuclear migration, Doublecortin is instead restricted to functions at the leading edge of the cell.
Type I lissencephaly is a severe cortical malformation disorder characterized by a massive disorganization of neurons throughout the cortex, which results in an absence (agyria) or diminuition (pachygyria) of gyri and sulci and hence a smooth brain surface (Aicardi et al., 1991; Harding, 1996). The affected cortex is in fact four layered with a superficial subpial molecular layer, below this a disorganized outer pyramidal cell layer, then a cell-sparse zone and, finally, a dense inner cell layer, suggesting that many neurons have not completed migration or not migrated at all. Subcortical laminar heterotopia (SCLH) is less severe and is characterized by a relatively organized cortex, although a heterotopic band of neurons is present in the white matter, once again suggestive of incomplete migration. At the clinical level SCLH is associated with epilepsy and sometimes mild mental retardation, but is generally a lot less severe than lissencephaly. Cloning of the LIS1 and doublecortin genes (Reiner et al., 1993; des Portes et al., 1998; Gleeson et al., 1998) and the search for mutations in affected individuals has aided the classification of these conditions and further elaborated the spectrum of phenotypes which are possible. Doublecortin and LIS1 mutations cause very similar clinical and radiological features, although Doublecortin mutations give rise to a more severely affected frontal cortex and LIS1 mutations to more severely affected parietal and occipital cortices (Gleeson et al., 2000). In addition to lissencephaly and SCLH, mutated individuals have now been identified with so-called cryptogenic epilepsy and extremely subtle cortical malformations (des Portes et al., 2002) and also some mutated individuals with no clinical symptoms at all (Demelas et al., 2001). In some cases, only minor changes in protein function might explain the phenotypes observed, in other cases genetic background may be a contributing factor. It is clear that these genetic studies, combined with functional analyses of the Doublecortin and LIS1 proteins, are likely to provide important information concerning the causes of epilepsy and mental retardation in affected individuals.
The strikingly similar spectrum of phenotypes in humans with either LIS1 or Doublecortin mutations suggests that these two proteins may function in the same biochemical pathway. Indeed, one report suggests that Doublecortin and LIS1 can exist in the same protein complex (Caspi et al., 2000). Detailed studies are, however, now necessary to further investigate the respective roles and relationship between these two proteins in the developing neuron. LIS1 was originally identified as a subunit of the intracellular brain form of platelet-activating factor acetylhydrolase (PAFAH) (Reiner et al., 1993; Hattori et al., 1994), but was later shown to also play a role in regulating microtubule dynamics (Sapir et al., 1997). The connection between these two seemingly separate functions has not yet been clearly established. A number of recent studies have, however, shed some light on the microtubule-associated role of LIS1, emerging with a clearer hypothesis of its significance in proliferating cells and developing neurons (Gupta et al., 2002). Several clues to the function of LIS1 were obtained from genetic studies of nuclear distribution mutants in Aspergillus nidulans (Morris, 2000). Through these studies the homolog of LIS1, NudF, was shown to be associated in the same biochemical pathway with the heavy chain of cytoplasmic dynein (NudA) and with the NudE and NudC proteins, originally of unknown function (Aumais et al., 2001). More recent studies have confirmed a physical interaction between LIS1 and dynein (Faulkner et al., 2000; Liu et al., 2000; Smith et al., 2000), which functions as a minus end-directed microtubule-associated motor protein involved in vesicle and organelle transport (Hirokawa, 1998). Indeed, LIS1 has been shown to regulate the function and distribution of dynein along microtubules (Gupta et al., 2002). The evolutionary conservation of these proteins in Aspergillus, and later studies, have led to the proposal that the LIS1–dynein protein complex exerts forces on the microtubules surrounding the nucleus in migrating neurons, eventually pulling this and other somal organelles into the leading process (Gupta et al., 2002). CLIP-170, which interacts with LIS1 and possibly also dynein, may also be implicated in these mechanisms (Coquelle et al., 2002). This nuclear migration hypothesis for the role of LIS1 is supported by the demonstration of a direct interaction between LIS1 and the mammalian NudE homologs, NUDEL and mNUDE, which co-localize at the centrosome and interact with dynein (Feng et al., 2000; Niethammer et al., 2000; Sasaki et al., 2000). Interestingly, LIS1 and NUDEL co-localize predominantly at the centrosome in proliferating cells, but partially redistribute to neuronal processes during differentiation, in association with dynein (Sasaki et al., 2000). It has thus also been proposed that the LIS1/NUDEL/dynein complex influences the transport of plus end-directed microtubules to the periphery of the migrating neuron, hence aiding the extension of the leading process (Gupta et al., 2002). It is also possible, based on the previously described neuronal role of dynein (Hirokawa, 1998), that LIS1 participates in the retrograde transport of vesicles and organelles, such as endosomes, from neuronal processes towards the nucleus. This data is supported by the axonal transport defects identified in Lis1 Drosophila mutants (Liu et al., 2000). These combined data suggest that LIS1 plays a role in nuclear migration in proliferating and/or migrating cortical neurons, but suggest further dynein-related functions in migrating and/or differentiating neurons.
We and others have previously shown that Doublecortin is also a microtubule-associated protein (MAP) (Francis et al., 1999; Gleeson et al., 1999; Horesh et al., 1999), although as we describe here it has a number of features that distinguish it from LIS1. Firstly, Doublecortin has a much more restricted expression pattern showing no expression in proliferating cells, being limited to post-mitotic immature neuronal cells, including tangentially migrating neurons in the embryonic cortex and adult rostral migratory stream and differentiating cortical plate neurons. Secondly, in dissociated cells it has a compartmentalized subcellular localization, concentrated in the extremities of growing neurites, and does not associate with microtubules surrounding the nucleus. Thirdly, at the molecular level the predicted secondary structure of Doublecortin suggests a novel microtubule-binding domain, conserved in other brain-specific proteins of unknown function, but not found in LIS1. Fourthly, as we show here, its microtubule-associated function is likely to differ from that of the LIS1/NUDEL/dynein complex. Intriguingly, our data suggest that Doublecortin is nevertheless involved in vesicle trafficking since it interacts with the AP1 and AP2 adapter complexes (Friocourt et al., 2001) implicated in clathrin-mediated transport to the endosomes and lysosomes and endocytosis, respectively. Our data thus suggest that Doublecortin is specifically required at the tips of growing neuronal processes, where it may play a role in the addition of membrane and/or the regulation of certain receptors or adhesion molecules implicated in cellular and possibly axonal guidance.
Microtubule-associated Doublecortin is Restricted to Neuronal Processes
We have previously shown a localization of microtubule-associated Doublecortin in growing neurites in primary cultures of mouse embryonic neurons (Francis et al., 1999). Our previous double labeling experiments showed that Doublecortin staining is most intense at the extremities of neurites and continues into the proximal regions of growth cones, but is not present in the actin-rich tips. Interestingly, a similar localization of Doublecortin has been observed in tangentially migrating neurons in explant cultures derived from the anterior sub-ventricular zone (B. Schaar and S. McConnell, personal communication). In this case Doublecortin staining was also observed at the extremities of leading processes, which thus suggests similarities in the function of this protein in differentiating cortical neurons and migrating neurons.
The subcellular localizations of LIS1 and NUDEL have been studied in detail in differentiating cortical neurons (Niethammer et al., 2000; Sasaki et al., 2000). These experiments show a change in the distribution of NUDEL and possibly LIS1 following morphological changes as the neuron matures. In primary cortical cultures, NUDEL was predominantly localized in the soma to the region of the centrosome after 1 day in culture and was not present in the neurites, but after 3 days in culture it was also concentrated in the neurites and present in growth cones (Sasaki et al., 2000). In order to assess the behavior of Doublecortin in comparison with these proteins we first examined its localization in very young cultures. Primary cultures were prepared from fetal mouse cerebral hemispheres at 15 days of gestation as previously described (Berwald-Netter et al., 1981) and immunocytochemical staining was performed using anti-Dbcn pep (1:300) (Francis et al., 1999), to detect Doublecortin and anti-α-tubulin (1:1000) (Amersham no. 356). This experiment showed that after only 16 h in culture, the predominant localization of Doublecortin is in the growing processes (Fig. 1A,B), with little if any specific labeling observed in the soma. Indeed, no labeling of the microtubules surrounding the nucleus was observed. Double labelings with a centrosomal marker, γ-tubulin, similarly showed that Doublecortin does not accumulate at the centrosome (data not shown). Doublecortin staining therefore differs from that of NUDEL since it is distributed within the very earliest growing processes in immature neurons in culture. We next examined the distribution of Doublecortin in comparison with LIS1 in 4 day old cultures. Once again the strongest Doublecortin staining was at the extremities of neurites, whereas LIS1, although present in the neurites, showed its strongest expression in the soma (data not shown). These combined data suggest that Doublecortin may play a role which differs from the function of the NUDEL/LIS1 complex in differentiating neurons.
Doublecortin Associates with Microtubules in an ATP-independent Manner
Both LIS1 and NUDEL have been shown to form a complex on microtubules with the motor protein cytoplasmic dynein. Motor proteins can be characterized by their dissociation from microtubules in the presence of ATP and their association in the presence of ADP. Sasaki et al. (Sasaki et al., 2000) showed, using microtubule sedimentation assays, that both LIS1 and NUDEL could be dissociated from microtubules in the motor protein fraction in the presence of ATP. Our data suggest that Doublecortin has a different relationship with microtubules. Firstly, we have previously shown that a direct interaction occurs between Doublecortin and microtubules, not requiring other MAPs (Francis et al., 1999). Here, the microtubule association of Doublecortin was further tested in the presence of ATP and ADP. No effect of these nucleotides was observed on the interaction of either purified Doublecortin with taxol-stabilized microtubules (Fig. 1C) or native Doublecortin sedimented with MAP-associated microtubules (data not shown). GTP and GDP, affecting the binding of other motor proteins such as kinesin, similarly produced no changes in the Doublecortin–microtubule complex. These studies strongly suggest a different type of microtubule association than that exhibited by the NUDEL/ LIS1/dynein proteins.
We have previously shown, nevertheless, an association between Doublecortin and other proteins involved in vesicle trafficking. Using a two-hybrid screen and in vitro and in vivo interaction studies, we identified the μ1 subunit of the AP1 adapter complex as an interacting partner of Doublecortin (Friocourt et al., 2001). Interestingly, the μ2 subunit of a highly similar complex, AP2, known to be involved in clathrinmediated endocytosis, also shows an interaction with Doublecortin (Friocourt et al., 2001). The neuronal function of the AP1 adapter complex has not yet been examined in detail, although this complex has been shown in non-neuronal cells to be involved in clathrin-dependent protein sorting from the trans-Golgi network to the endosomes and lysosomes (Robinson and Bonifacino, 2001). In dissociated neurons in culture we found that AP1 subunits are localized both in association with the trans-Golgi network in the soma, as expected, but also at the extremities of growing neurites, possibly associated with transported vesicles (Friocourt et al., 2001). Thus, although our preliminary data suggest a direct association between Doublecortin and microtubules not requiring dynein, we do however observe an association with certain adapter complexes known to be implicated in transport mechanisms. The relationship between these adapter complexes and dynein now needs to be closely examined.
The Subcellular Localization in Astrocytes of a Protein Homologous to Doublecortin
We also examined the subcellular localization of the Doublecortin-like protein DCLK (Berke et al., 1998; Vreugdenhil et al., 1999; Burgess and Reiner, 2000; Lin et al., 2000), a protein showing strong homology to Doublecortin. Certain isoforms of DCLK show a remarkably similar expression pattern to Doublecortin during embryonic development (Burgess and Reiner, 2000; Lin et al., 2000), including expression in migrating cortical neurons (Mizuguchi et al., 1999). In addition, a number of DCLK isoforms contain a microtubule-binding domain highly homologous to that of Doublecortin, suggesting similar and perhaps connected functions (DCLK1A and 1C, Fig. 2A). Indeed it is possible that, unlike in human, a functional redundancy of these two proteins in mouse can explain the apparently normal cortex observed in doublecortin knockout mice (Corbo et al., 2002). Clearly, additional studies are required to further investigate the functions of DCLK. We describe here a new embryonic isoform of DCLK (DCLK1C, Fig. 2A), which is the most closely related protein to Doublecortin identified to date. Similar to Doublecortin, this protein is 363 amino acids in length (compared to 366 amino acids for Doublecortin), is a phosphoprotein (Fig. 2B) and has an almost identical C-terminus, unlike the other embryonic isoforms of DCLK. Interestingly, in addition to being expressed in developing mouse cortical neurons, this isoform is also expressed in astroglial cells in culture, hence showing a wider expression pattern than Doublecortin.
In order to examine the subcellular localization of this DCLK isoform most closely resembling Doublecortin (DCLK1C, Fig. 2A), we focused on mouse astrocyte cultures since our western blot analysis showed that only this isoform of DCLK was present in these cells. Immunofluorescence studies showed that anti-DCL antibodies labeled some but not all astrocytes in culture and, in particular, astrocytic processes were strongly stained (Fig. 3A–F). These processes are known to be particularly dynamic in vivo, with their extremities found abundantly in perisynaptic regions where they can modulate synaptic function and glial–neuronal communication (Chvatal and Sykova, 2000). Doublecortin itself is not expressed in astrocytes in culture, ensuring that the process staining we observed here is specific for the DCLK protein. Although this protein seems also to partially co-localize with microtubules throughout the soma of these cells, a much stronger, tapered staining pattern is observed towards the ends of the processes. This staining pattern might thus suggest a function of this Doublecortin homolog in the plasticity of the highly dynamic fine processes of astrocytes, which are involved in perisynaptic function (Derouiche and Frotscher, 2001). DCLK in astrocyte processes may thus play a role similar to that of Doublecortin in neuronal processes. This specific process localization, conserved between Doublecortin family members, most probably points to an important function for these proteins in this particular region of the cell.
The migration of neurons from proliferative zones to their final positions in the developing cortex and their subsequent differentiation to form synaptic connections are critical steps for the development of normal cognitive abilities. Severe cortical malformation disorders, showing signs of incomplete or incorrect neuronal migration, are almost always associated with cognitive impairment and epilepsy. In these cases, it is not yet clear whether neuronal mispositioning is responsible for aberrant connectivity or if perturbed axo- and dendritogenesis is an additional primary problem. Molecular studies are increasingly being applied to help answer these questions and to contribute to our understanding of the mechanisms involved in normal brain development. We have focused here on the functions of two molecules, Doublecortin and LIS1, each shown to be mutated in lissencephaly and SCLH, hence normally playing a critical role in human corticogenesis. As yet the relationship between these two proteins remains unclear, although as we highlight here, both have been shown to be associated with microtubules and with proteins involved in vesicle trafficking. An important question to now answer is whether these protein complexes show any overlapping functions during neuronal migration or differentiation.
As a cortical cell begins to express neuronal markers and takes on a neuronal phenotype, it changes its morphology, extending processes to become first a migrating cell and later a differentiated neuron with axons and dendrites. Although different modes of migration exist (Nadarajah and Parnevelas, 2002), process extension is fundamental to each method. Microtubules are known to be important constituents of such processes (Baas, 1999), since they provide an architectural support and allow the transport of cargo to the extremities of the cell. Microtubule components have been extensively studied in differentiating neurons and some populations of migrating neurons, in which they are uniformly oriented in the axon and leading process (Rakic et al., 1996), whereas in dendrites and trailing processes they exist in both orientations. Motor MAPs such as dynein play a critical role in axon and dendrite growth (and probably also leading process growth), by the transport of microtubules of a particular orientation into the processes. As the process lengthens, the microtubules become longer and are stabilized in all regions except the extremities, which remain more dynamic in order to accommodate changes in direction in response to navigational cues and to allow further growth of the process. Less stable microtubule arrays invade the growth cone in the direction of future growth. Our data showing the presence of microtubule-associated Doublecortin at the extremities of growing processes, even the earliest processes grown by neurons in culture, suggests an association of Doublecortin with less stable microtubules. In addition, it is likely that the regulation of Doublecortin by phosphorylation contributes to its compartmentalization and function in this region of the cell. A similar localization in migrating neurons (B. Schaar and S. McConnell, unpublished results) suggests a similar function of Doublecortin in migrating cells, thus influencing a mechanism in common between them. This mechanism seems likely to be the actual growth or remodeling of the neuronal process itself, in response to local environmental cues interpreted by the actin-rich tip. Defects in the ability of a cell to efficiently extend processes may be a feasible explanation for the cause of a neuronal migration disorder such as lissencephaly, characterized by the aberrant positioning of neurons.
The extension of neuronal processes depends not only on the transport of microtubules but equally importantly on the addition of membrane itself. Previous studies have shown that membrane addition in growing axons can only occur in regions containing less stable microtubules (Zakharenko and Popov, 1998), i.e. close to the growth cone. It is possible that membrane addition in leading processes occurs similarly close to the actin-rich tip, although this has not yet been formally demonstrated. The interaction between Doublecortin and adapter complexes involved in vesicle trafficking might indicate a function for Doublecortin in membrane and cargo addition to these regions. Indeed, a co-localization of Doublecortin and the AP1 complex has been demonstrated at the extremities of growing processes in both differentiating neurons in culture (Friocourt et al., 2001) and in migrating neurons (B. Schaar and S. McConnell, unpublished results), which seems to support this hypothesis. Similar mechanisms involving the DCLK protein at the extremities of astroglial processes are also suggested by its interaction with the AP1 and AP2 complexes (Friocourt et al., 2001). These adapter complexes, which are known to attach to membrane vesicles containing receptors and adhesion components, interact with motor MAPs allowing their transport along microtubule tracks, as demonstrated for KIF13A (Nakagawa et al., 2000). The transport in mature neurons is bi-directional with certain motor proteins, such as the KIFs, being involved in anterograde transport (Foletti et al., 1999) and others, such as dynein, implicated in retrograde transport. Although a direct interaction between dynein and the AP1/AP2 complex has not yet been demonstrated it seems likely that it could occur, similar to the situation with KIF13A. Although this may provide a possible link between LIS1/dynein and Doublecortin/AP1/AP2, there is as yet no evidence to suggest that Doublecortin is involved in retrograde transport. Indeed, its presence at the extremities of growing processes in developing neurons might instead suggest a potential role in the detachment of vesicles from microtubules, to allow membrane addition and, hence, process extension (Fig. 4). Interestingly, further links to membrane proteins are suggested by the demonstration of an interaction between Doublecortin and the L1 cell adhesion molecule neurofascin (Kizhatil et al., 2002). In this study Doublecortin was shown to interact with the phosphorylated form of the cytoplasmic domain of neurofascin, a protein showing a similar expression pattern to Doublecortin in the developing cortex and in migrating neurons of the rostral migratory stream. This confirmed interaction of Doublecortin with a membrane protein supports its potential role in dissociating membrane protein vesicles from the adapter complexes, since the presence of Doublecortin could perturb the interaction between the membrane protein and the adapter complex (Fig. 4). By similar mechanisms, the presence of Doublecortin could potentially also regulate the endocytosis of certain membrane proteins. These hypotheses remain to be tested, in addition to the identification of the critical receptors and adhesion molecules which are sorted to the extremities of migrating and differentiating neurons. In addition to neurofascin, several candidates in migrating neurons are the Reelin receptors VLDLR and APOER2 and certain integrins (Gupta et al., 2002).
Although Doublecortin and LIS1 may both play a role in the growth of leading processes, Doublecortin through its implied role in vesicle trafficking and LIS1 by influencing the transport of microtubules and/or the retrograde transport of cargo, it is equally likely that LIS1 instead plays its essential role in the movement of the nucleus (Morris, 2000; Aumais et al., 2001; Gupta et al., 2002). Doublecortin, on the other hand, does not seem to be as highly conserved as LIS1 and has not been identified as a member of the biochemical pathway regulating nuclear migration characterized in Aspergillus nidulans. Thus it is possible that Doublecortin and LIS1 are members of separate biochemical pathways which evolved at different times, influencing neuronal migration (and perhaps also differentiation) through different mechanisms. Alternatively, LIS1 may play a role in nuclear migration only in proliferating cells, and process growth similar to Doublecortin in migrating and differentiating neurons. A further dissection of LIS1 function in these cells is required to answer these questions. One thing remains clear, that although Doublecortin and LIS1 have quite different profiles of expression, in mouse as well as human fetal brain (Meyer et al., 2002), mutations in either of these proteins result in a severe almost identical human phenotype. Clearly, further studies in both mouse and man are required to explain this and related pathophysiologies.
We thank Philippe Denoulet and Yohoved Berwald-Netter for their contributions to this work. We are particularly grateful to Bruce Schaar and Susan McConnell for sharing unpublished data. This work was supported in part by grants from the Institut National de la Santé et de la Recherche Médicale (INSERM), the Centre National de la Recherche Scientifique (CNRS), the European Commission (no. QLG3-CT-2000-00158), the Human Frontier Science Program (RG0283/1999-B), the Ministère de la Recherche (ACI 1A066G) and the Federation pour la Recherche sur le Cerveau.
- platelet activating factor
- congenital abnormality
- ectopic tissue
- dynein atpase
- intellectual disability
- microtubule-associated proteins
- cognitive impairment
- molecular motor proteins
- biochemical pathways
- neuron migration