Human Molecular Genetics Advance Access originally published online on March 28, 2006
Human Molecular Genetics 2006 15(9):1387-1400; doi:10.1093/hmg/ddl062
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Branching and nucleokinesis defects in migrating interneurons derived from doublecortin knockout mice
1Département de Génétique et Développement and 2Homologous Recombination Laboratory, Institut Cochin, F-75014 Paris, France, 3INSERM U567, Paris, France, 4CNRS UMR 8104, Paris, France, 5Université Paris 5, Faculté de Médecine René Descartes, UM 3, 75014 Paris, France and 6U616 INSERM, Hôpital Pitié-Salpêtrière, 47, Bld de l'Hôpital, 75651 Paris Cédex 13, France
* To whom correspondence should be addressed at: Institut Cochin, Faculté de Médecine Cochin Port Royal, 24 rue du Faubourg Saint Jacques, 75014 Paris, France. Tel: +33 144412429; Fax: +33 144412421; Email: francis{at}cochin.inserm.fr
Received January 4, 2006; Accepted March 10, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Type I lissencephaly results from mutations in the doublecortin (DCX) and LIS1 genes. We generated Dcx knockout mice to further understand the pathophysiological mechanisms associated with this cortical malformation. Dcx is expressed in migrating interneurons in developing human and mouse brains. Video microscopy analyses of such tangentially migrating neuron populations derived from the medial ganglionic eminence show defects in migratory dynamics. Specifically, the formation and division of growth cones, leading to the production of new branches, are more frequent in knockout cells, although branches are less stable. Dcx-deficient cells thus migrate in a disorganized manner, extending and retracting short branches and making less long-distant movements of the nucleus. Despite these differences, migratory speeds and distances remain similar to wild-type cells. These novel data thus highlight a role for Dcx, a microtubule-associated protein enriched at the leading edge in the branching and nucleokinesis of migrating interneurons.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Malformations of cortical development are responsible for a large proportion of cases of mental retardation and epilepsy in children (1
). Type I lissencephaly is one such disorder characterized by a smooth brain surface (agyria) and a severely disorganized cortex. This consists of four jumbled layers of neurons, instead of the six highly organized layers present in a well-formed brain (2). Molecular studies have revealed that the LIS1 and doublecortin (DCX) genes are mutated in a proportion of cases for this disorder (3–5). Nevertheless, the primary pathophysiological mechanisms underlying type I lissencephaly still remain to be elucidated.
The correct lamination of the cortex is dependent on highly regulated neuronal migration. Radial migration (6), principally used by excitatory pyramidal neuron precursors, occurs in an orientation perpendicular to the ventricular surface. Waves of radially migrating neurons thus migrate from the ventricular zone where they are produced to the superficial regions of the cortex, where they progressively form the cortical layers. Tangential modes of migration are also important for cortical development. Thus, inhibitory neuron (interneuron) precursors in rodents are known to be generated in an extra-cortical structure, the medial ganglionic eminence (MGE) and to migrate tangentially into the neocortex in specific migratory streams (7–9). Arriving in the cortex, such cells then use radial migration in order to integrate into the different cortical layers. The exact migratory substrates for tangential migration are still unknown (9). Type I lissencephaly is believed to be a neuronal migration disorder, involving a disorganization of both pyramidal neurons and interneurons (10,11). Interestingly, in human, a proportion of interneurons are produced in the cortical ventricular zone and presumably use mainly radial modes of migration to reach their final destination (12). Massive defects in radial migration are thus probable in type I lissencephaly, although it is unclear if abnormal tangential migration also contributes to this phenotype.
Lis1 and Dcx knockout (KO) mice have been produced and remarkably, in each case the lamination of the mouse neocortex was found to be relatively normal (13–16). Lis1 heterozygotes, equivalent thus to the human genotypic state giving rise to severe lissencephaly, show apparently only minor abnormalities of radial migration in the cortex by bromodeoxyuridine (BrdU) labelling studies. Similar abnormalities in tangential migration were also demonstrated later in Lis1 (+/–) embryonic brains in vivo and in vitro, by immunohistochemistry, and lipophilic dye labelling of migrating cells in slice cultures (17). Although tangential migration was not assessed in Dcx KO mice (15), analyses of radial migration showed no major abnormalities in this model. However, RNAi studies have also been performed for Dcx in rat and mouse (18,19). In these studies, defects were observed in radially migrating neurons, but interestingly, this phenotype was more dramatic in the rat than in the mouse. These data suggest that Dcx's neuronal function may have become increasingly important with increased brain size.
Dcx, expressed both in radially and tangentially oriented neurons in mouse brain (20,21), shows a predominant expression in tangentially oriented neurons in human fetal brain (10). Thus DCX seems likely to be particularly important for tangential migration in human. In both mouse and human, DCX/Dcx is also strongly expressed in growing axons and dendrites. Interestingly, an enrichment of Dcx has been shown at the extremities of growing processes in differentiating neurons and certain migrating neurons in culture (20,22,23), suggesting that this protein is required for process growth. In fitting with this, Dcx has been shown to nucleate and stabilize microtubules, important functions in this region of the cell (20,21,24). Its subcellular localization and function may, however, be regulated by phosphorylation (22,23,25) and thus Dcx may have various functions at different subcellular localizations (26).
In order to further elucidate Dcx function during migration, we focused on tangentially migrating interneuron precursors derived from Dcx KO mice. The migration dynamics of wild-type (WT) cells originating from the MGE have been well characterized previously (27–29). Such cells normally extend a leading process tipped with a growth cone in the direction of migration. Nuclear translocation into the leading process and retraction of a trailing process at the cell rear allow the cell to advance. Using slice cultures, we observed that Dcx KO cells show more branching compared with WT cells. Using an in vitro migration model recently used to characterize the dynamics of nuclear movement and its correlation with branch production in MGE cells (30), we were able to determine by video microscopy experiments that a higher proportion of KO cells have a more branched neuritic arbour compared with WT and that nuclear translocation is also perturbed. Individual KO MGE cells thus migrate in a less organized manner, dynamically extending and retracting short branches and making less long-distant nuclear jumps, although speeds of migration and migratory streams in brain sections seem unchanged. The rostral migratory stream (RMS), however, containing a different population of tangentially migrating interneuron precursors, is abnormal in KO brain sections. Combined, these novel data confirm a role for Dcx in tangential migration and provide insights into the function of Dcx in migrating cells.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Generation of Dcx KO mice
Mice carrying a tri-lox P floxed Dcx allele with lox P sites inserted upstream and downstream of Dcx exon 3 and a PGK-neo selection gene (Fig. 1A) were generated following standard procedures and crossed with Meu-Cre 40 transgenic mice expressing Cre ubiquitously early in development (31
). The partial or complete action of the Cre recombinase led to the generation of KO mice with a deletion of both Dcx exon 3 and the selection gene; and mice deleted for the selection gene but conserving a floxed exon 3. KO mice were successively crossed with either C57BL/6 or Sv129Pas mice.
|
Deletion of exon 3 in mutant KO mice was confirmed by Southern and western blotting (Fig. 1B and C) and immunohistochemistry (Fig. 1D). No Dcx protein was observed using antibodies directed at either the N or C terminus of the protein. Mutant hemizygote males (due to the fact that the Dcx gene is localized on the X chromosome) and homozygote females were born in the correct ratios, ruling out that a complete absence of Dcx is lethal in mouse. Both male hemizygote and female homozygote mice were fertile, thus differing from the Dcx mouse model described by Corbo et al. (15
), where hemizygote males were variably fertile.
Dcx KO mice have no obvious radial migration abnormalities
Histological and immunohistochemical experiments showed no major defects in pyramidal cell generation and migration in the neocortices of KO mouse embryos on both genetic backgrounds. Cortical pyramidal cell neurogenesis occurs between E11 and E17 in the mouse. We, therefore, examined sagittal and coronal sections from embryos at E13.5, E14.5, E16.5, E17.5, E18.5 and newborn mouse brains. In each case, the thickness of the neocortex and the different developmental compartments appeared similar in KO and WT embryos or pups from the same litter. We tested a variety of cortical markers (reelin, microtubule-associated protein 2, nestin, chondroitin sulphate proteoglycan, ßIII tubulin and aristaless-related homeobox gene) to assess the different zones (preplate, marginal zone, cortical plate, intermediate zone and ventricular zone) and the radial organization of the developing cortex, without observing major differences in the KO (Fig. 2A–C and data not shown). Thus, the proliferation, organization and differentiation of future pyramidal neurons in the neocortex appear largely normal in the mutant mice.
|
We also performed injections of BrdU during the time period of cortical neurogenesis and migration. BrdU is incorporated into dividing cells and their progeny and allows an assessment of migrating cells generated at different timepoints. Injections were performed at E11.5, E14.5, E15.5 and E17.5, and in each case embryos were sacrificed either 30 min or 48 h later, or in postnatal stages. These experiments were thus designed to assess cell proliferation (30 min), short-term (48 h) or long-term (postnatal) neuronal migration. Observations of BrdU-positive cells showed no obvious defects in cell proliferation in the neocortex and no inversion of the neocortical layers. Specifically, we observed that migrating neurons born at E14.5 were able to cross through the previously formed neuronal layers in the cortical plate similar to WT neurons (Fig. 2D). Quantification of the numbers of BrdU-labelled cells in each zone of the developing cortex showed no significant differences between mutant and WT cortex (data not shown), similar to the previously described Dcx KO model (15
).
Increased branching of interneuron precursors in slice cultures
The expression of DCX in tangentially oriented neurons in the developing mouse and human brain (10,20,22) led us to question whether migration abnormalities exist in these cells in Dcx KO mice. During mouse embryogenesis, tangentially migrating interneuron precursors are mainly generated between E12.5 and E14.5 in the MGE far away from the neocortex (32,33), and as such, are amenable to detailed studies of migration because of their localized production and the long distances they travel before reaching their final destination. At late stages of corticogenesis, Dcx-positive tangentially migrating neurons have been previously observed in the subventricular zone (SVZ), consistent with the expression of this protein in cells most likely derived from the MGE (20). Nevertheless, the role of Dcx during the migration of such cells has not previously been analysed.
We set out therefore to analyse such tangentially migrating cells derived from the MGE using well-characterized, organotypic slice cultures, containing both the MGE and the cortex (32,33). Three main tangential migratory routes have been recognized (9), a superficial route through the marginal zone, a second route in the intermediate zone and a third deeper route in the SVZ. Slice cultures are useful for assessing these pathways, as well as migration distances, and for observing cell orientation and general migratory morphology. Coronal slices were thus prepared from E13.5 and E14.5 KO and WT embryonic brains and cells were labelled by the injection of 5-chloro-methyl-fluorescein diacetate (CMFDA) dye into the MGE (Fig. 3A). Slices were cultured for 48 h and fixed before analysis by confocal microscopy and Metamorph software, with the investigator blind to the genotypes. No major differences were initially detected, neither in the distance migrated by MGE cells, nor in their distribution in the major migratory streams, nor in the overall length of cell processes (Fig. 3B, WT, 70.10±2.25 µm, n=16, KO, 70.01±3.23 µm, n=11). However, in all three experiments performed, the branching of cells was more apparent in KO slices compared with WT, as demonstrated in Fig. 3C, suggesting potential branching abnormalities. Indeed, it is known that during migration, MGE cells produce branches at their leading edge and show a saltatory progression of their nucleus in these branches (27–29,34). Flat co-cultures have shown moreover that leading growth cone divisions create paired branches that lengthen and pause prior to the selection of one branch to become the new leading process and the second branch for retraction (30,35). Thus, the regulation of branching is an important part of MGE cell migration. Dcx is in fact a microtubule-associated protein known to stabilize microtubules (20,21,24), the effect of which might therefore normally restrain a growing neuronal process from forming superfluous side branches. Thus, a potential branching phenotype in the absence of Dcx may indeed be in fitting with its proposed cellular functions and we thus decided to analyse this further.
|
Quantification of the branching abnormalities using video microscopic analyses
The neuritic arbour being difficult to analyse accurately in the complex 3D environment of slice preparations, we thus decided to continue our analyses in comparable co-culture models where the MGE cells are permitted to migrate on a flat surface. Indeed, it has previously been shown that the major characteristic features of MGE cell migration do not differ in such a model compared with their migration in slice cultures (30
). In the co-culture system, green fluorescent protein (GFP)-expressing KO and WT MGE explants were deposited onto monolayers of GFP-negative dissociated cortical cells derived from WT mice. No differences were observed between Dcx–/– female homozygote mutant neurons and those derived from DcxY/– males in these experiments, although male hemizygote explants were generally used. Cultures were incubated 12–24 h prior to video microscopy, generating films of migrating MGE cells for analysis. Cultures were then also fixed and stained with anti-GFP, in order to amplify the fluorescent signal and to analyse in detail the complexity of the neuritic arbours of a larger number of fixed, migrating MGE cells. Video microscopic analyses of the co-cultures also revealed branching abnormalities in KO MGE cells (Fig. 4A and B, Supplementary Material, Movies S1 and S2, WT and KO, respectively). Four independent video microscopy experiments, involving in total four WT and eight KO embryos, were performed giving reproducible results. Compared with WT cells, which develop a long branched leading process, KO cells were observed to extend shorter branches and to produce new branches at a higher frequency. As branching is dependent on the division or formation of a growth cone, we first analysed growth cone dynamics from the video films. Both growth cone splitting at the leading edge and the formation of new growth cones at other positions along the length of the neuritic processes appeared more frequent in KO cells (Fig. 4C, WT, mean of 1 leading growth cone split per hour, versus KO mean of 1.3 splits per hour; and WT mean of 0.3 versus KO mean of 0.6 new growth cones per hour at other positions). In total, five WT cells were recorded for 299±122 min; versus eight mutant MGE cells, recorded for 247±124 min for this analysis. Thus, the increased branching observed is reflected in the increased formation of new leading edge and side growth cones. Newly formed branches even in WT cells are known to be dynamic, first elongating, then either becoming the new leading process, or retracting and disappearing. Observations of the newly formed branches from the video films showed that growth cones of KO and WT cells apparently exhibit the same migration speed and new branches appear to grow at the same rate in both cases. However, an analysis of the lifetime of new branches, prior to growth cone collapse and retraction, showed that in general they exist for a shorter time in KO than in WT cells (Fig. 4D, WT mean lifetime of 43 min, n=26 branches analysed; KO mean lifetime of 34 min, n=50 branches analysed). The overall result is an apparently increased number of KO MGE cells with numerous short and unstable branches compared with WT, which give rise therefore to a more complex neuritic arbour.
|
The video microscopy observations were confirmed by quantifying MGE cell complexity in fixed and immunostained KO and WT co-cultures. Cells (WT, n=104; Dcx KO, n=109) were thus classed and counted according to whether they exhibited primary, secondary, tertiary, quaternary or quinary terminal processes (Fig. 5A). There were less KO cells with only primary and secondary terminal processes compared with WT (Fig. 5B, WT, 17.3% have primary and 47.1% secondary terminal processes; versus KO, 11.9% have primary and 19.3% secondary terminal processes) but more KO cells had tertiary, quaternary and quinary terminal processes (Fig. 5B, WT, 27.9% with tertiary, 6.7% with quaternary and 1% with quinary terminal processes; versus KO, 41.3% with tertiary, 18.3% quaternary and 8.3% with quinary terminal processes). Nevertheless, comparing WT and KO cells with a particular complexity showed no differences in the actual number of terminal processes, e.g. individual WT and KO cells with terminal tertiary processes were generally asymmetrically branched as shown in Fig. 5A, with only two terminal tertiary processes (data not shown). Thus, it is the number of highly branched cells that differs in the KO and not the branching complexity of an individual cell per se. In agreement with observations of the video films, the average length of a particular type of process (primary, secondary, tertiary, etc.) in the fixed cultures was as much as 20% less in KO cells compared with WT cells (Fig. 5C; for WT cells: trailing process, 61.3±5.7 µm, n=81; primary, 68.6±3.9 µm, n=122; secondary, 50.1±2.3 µm, n=174; tertiary, 38.4±2.4 µm, n=83; quaternary, 28.7±4.9 µm, n=16; quinary, 31.5±22.2 µm, n=2; for KO cells: trailing process, 39.5±4.6 µm, n=64; primary, 53.3±3.3 µm, n=128; secondary 38.0±1.6 µm, n=204; tertiary, 25.3±1.2 µm, n=181; quaternary, 18.9±1.6 µm, n=79; quinary; 13.4±1.3 µm, n=23). Thus our dynamic and static analyses reveal that KO MGE cells exhibit increased growth cone splitting and formation, leading to a higher frequency of new leading edge and side branches, which have however a shorter lifetime and a shorter length than WT MGE cell branches. These results thus strongly support a role for Dcx, a microtubule-associated protein, in growth cone dynamics and neurite stability in migrating MGE cells.
|
0.05, ** at PKO MGE cells also have swelling and nucleokinesis abnormalities
In interneuron precursors, the movement of the nucleus and branch production are tightly coupled (30
). Thus, the saltatory movement of the nucleus is coupled with growth and branching of the leading process, and nuclear jumps occur into a swelling, newly formed within the leading process, after branching has occurred. Correspondingly, analysis of video microscopy films of Dcx KO MGE cells showed that, in addition to abnormal branching, nuclear movement is also affected (Fig. 6, Supplementary Material, Movies S3 and S4, WT and KO, respectively). During migration of WT neurons, a distinct and round swelling that contains the microtubule-organizing centre (MTOC) and Golgi apparatus periodically forms and migrates towards the leading edge at some distance away from the cell soma, and then receives the translocating nucleus and becomes itself the new soma. This has been shown not only for MGE-derived cells but also for SVZ interneuron precursors migrating to the olfactory bulb (36). In KO neurons, a variety of swelling types were observed even in the same cell, ranging from an apparently normally functioning swelling, to a less distinct swelling that does not completely separate from the cell soma, or to a swelling observed at an appreciable distance from the nucleus for abnormally long periods of time (up to 2 h, compare A1 and A2 with B1 and B2 of Fig. 6). Additionally, the size of the swelling was often larger in Dcx KO than in WT cells (Fig. 6A1 and B1). As a consequence of these changes, the frequency of swelling formation is significantly decreased in KO cells (Fig. 6C, right histogram, WT, mean of 2.7 swellings per hour versus KO, mean of 1.3 swellings per hour), whereas the lifetime of swellings is strongly increased (Fig. 6D, e.g. 46% of WT swellings existed between 1–7 min, compared with 16% of KO swellings, whereas 3% of WT swellings existed between 32–50 min compared with 23% of KO swellings, WT, n=67 analysed; KO, n=51 analysed).
|
Unlike WT cells in which the swelling gradually moves towards the leading edge away from the nucleus, in KO cells it can move both forward and backward in the leading neurite (Fig. 6B1, frames 1:00 and 1:12). Such oscillations of the MTOC-containing swelling thus suggest that KO cells are less polarized. Indeed, observations of the video films also showed more KO cells performing complete polarity reversals, and such changes in direction, by selection of the trailing process as the new leading process, are rare in WT cells.
Surprisingly, the changes observed in swelling dynamics do not affect the nuclear migration speed, which is similar in both WT and KO cells (WT, 100.93±4.7 µm/h, n=18; KO, 100.92±4.8 µm/h, n=18). In contrast, the dynamics of nuclear movement often differ between KO and WT cells. For example, as shown for the cells in Fig. 6A and B, long nuclear displacements (jumps superior to 9.5 µm, shown by an asterisk in the upper parts of graphs in Fig. 6A2 and B2) were three times less frequent in the KO cell studied, whereas shorter nuclear displacements (jumps inferior to 9.5 µm, shown by a bar in the upper parts of graphs in Fig. 6A2 and B2) were as frequent in KO as in WT (Fig. 6C). For WT, five cells were analysed for 305±50 min each, and thus 1525 min in total. For Dcx KO, eight cells were analysed for 318±70 min each, and thus 2548 min in total. Observations of the video films also showed that nuclei of KO cells more often advance by short gliding movements than WT cells, instead of using more distinct jumps. Gliding nuclei were observed to move short distances within an oval-shaped soma, not forming an independent swelling. Thus, branching defects in KO MGE cells are coupled with nucleokinesis defects, the latter being characterized by shorter-distance nuclear displacements and abnormal swelling dynamics. This disrupted movement of organelles in KO MGE cells nevertheless does not change overall nuclear migration speed, and as shown in slice cultures, KO cells are able to reach the cortex despite the migration abnormalities.
Dcx KO mice show no major differences in MGE-derived interneuron distribution in vivo, but have an abnormal RMS
Such tangential migration abnormalities detected in vitro might be expected to have consequences on the distribution of MGE-derived interneurons in vivo. In order to look for such interneuron abnormalities in Dcx KO mice, we performed immunohistochemistry experiments testing interneuron markers. Using anti-
-aminobutyric acid (GABA), anti-parvalbumin, anti-calretinin and anti-ARX antibodies, we were unable to detect major abnormalities in the distribution or number of these interneuron populations. Specifically, analyses of the distribution of GABA-positive cells in brain sections at E12.5, E13.5, E14.5 (Supplementary Material, Fig. S1), PN1 and in the adult showed no major differences between KO and WT. Also the distribution of ARX at E13.5, E16.5 and PN1 did not differ (data not shown). On the other hand, quantitative analyses of the calretinin subpopulation of interneurons in adult brain sections showed subtle differences between a KO and WT animal both in the cortex and in the hippocampus (KO sections had on average 88 and 79% the number of cells in WT sections, respectively; Supplementary Material, Figs S2 and S3 and Tables S1 and S2). This result is interesting because it might suggest slowed or aberrant migration of calretinin-positive cells from the ganglionic eminences. Further studies in the future are, however, required to test this interneuron subpopulation in a larger number of KO animals, in order to support this result and to rule out inter-animal variability.
In addition, we analysed polysialic acid–neural cell adhesion molecule (PSA–NCAM)-positive interneuron precursors, which migrate in the RMS towards the olfactory bulb in the adult (37) and have been shown to strongly express Dcx (21,22 and our unpublished results). Such cells show similar migratory dynamics to MGE cells, in the respect that they have a long leading process and that swellings form distal to the nucleus, prior to nuclear displacement (36). Nevertheless, migratory mechanisms and substrates are still likely to be different from MGE cells, as RMS cells form homotypic interactions and migrate in chains, in close proximity to glial cell tunnels (38). Our analyses suggest that Dcx KO mice have subtle RMS abnormalities (Fig. 7). Sagittal sections from two KO animals were analysed in comparison with one WT animal. PSA-NCAM immunohistochemistry was performed using immunoperoxidase detection. PSA-NCAM-positive interneuronal precursors were observed in the RMS, stemming from the anterior horn of the lateral ventricle to the olfactory bulb in both the WT and KO animals. However, in the two KO animals, the RMS appeared thicker than that in WT. Indeed, the WT RMS appears as a thin compact channel, whereas the KO RMS appears more diffuse, covering a larger area. KO cells nevertheless still seem to be organized in chains, as in WT. Interestingly, examination of the olfactory bulb in Dcx KO animals shows no obvious macroscopic or microscopic abnormalities (Supplementary Material, Fig. S4). Thus, it appears that a slightly disorganized migratory stream in the adult is nevertheless compatible with the correct insertion of new cells in the olfactory bulb during adult life. It will now be interesting to confirm this result in a larger number of animals and to analyse the RMS at early postnatal stages in order to assess whether a similar disorganization occurs, or to see if this is restricted to the adult. In summary, subtle migration abnormalities identified in a second type of interneuron precursor in Dcx KO mice strongly support our in vitro findings of aberrant MGE cell migration.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Human lissencephaly is associated with a severe disorganization of the cortex involving an abnormal positioning of both pyramidal neurons and interneurons. Previous rodent models targeting the Dcx gene have indicated either an absence of neocortical abnormalities in a mouse KO model (15
), or, unexpectedly, abnormalities affecting radially migrating neocortical neurons in an RNAi model (18,19). The mice we describe here more closely resemble the previously reported KO model than the RNAi model, as radial migration of pyramidal cell precursors in our KO mice appears largely normal. A possible explanation for the difference in phenotype between Dcx KO models and RNAi-injected mice could be the temporally discrete disruption, or partial inhibition of Dcx in the RNAi model. Thus Dcx's function may not be compensated by other proteins, as may occur in the Dcx KO. Analysis, however, of MGE cells in slice and explant cultures in our KO model reveal subtle abnormalities in the migration dynamics of future interneurons. In particular, migrating KO MGE cells show both increased branching as well as nucleokinesis defects. Our data also show potential migration abnormalities in PSA-NCAM-positive interneuron precursors migrating in the RMS in Dcx KO mice. These combined data support a role for Dcx in tangential migration dynamics. It is possible that similar or more severe abnormalities also exist in human fetal interneuron precursors in type I lissencephaly, with perhaps also more widespread consequences in primate than in rodent brain, but this is difficult to test directly. In mouse, the subtle abnormalities identified provide a further clue to the function of Dcx during tangential migration.
We investigated the behaviour of interneuron precursors migrating from MGE explants onto a monolayer of dissociated WT cortical cells. The complete morphology of individual GFP-positive cells is easily discernible using this co-culture system (30). In live co-cultures, we observed that KO cells more frequently formed and retracted both leading edge and side branches. Specifically, the absence of Dcx leads to a higher frequency of growth cone formation and a decrease in branch stability. As a result, the large majority of KO cells showed a more complex neuritic arbour than WT cells and quantifications in fixed co-cultures showed that KO cells had significantly shorter processes. These results are thus coherent with our initial observations of increased branching in KO slice cultures. Dcx has been shown to be enriched at the extremities of neurites proximal to the growth cone in dissociated cortical neurons (20,39). A similar localization has been shown for Dcx at the extremities of leading processes in migrating interneuron precursors destined for the olfactory bulb (22). In particular, the latter study showed that Dcx was present in nascent secondary processes in these cells when they exist. These combined data suggest a major function for Dcx in growing processes. Interestingly, Dcx-RNAi experiments in radially migrating neurons showed that Dcx-inhibited cells took on a multipolar phenotype (18), with multiple fine processes stemming from the cell soma (40). However, KO MGE cells observed here did not resemble multipolar cells. Instead, they generally exhibited only two principal processes (leading and trailing) stemming from the cell soma and were observed to frequently change direction, even to the extent of selecting the process on the opposing side of the nucleus. One interpretation of these data is that Dcx plays a role in maintaining and stabilizing leading process identity.
Indeed, Dcx's function as an microtubule-associated protein (MAP) (20,21,24), polymerizing and stabilizing microtubules, is quite consistent with a role in process formation and maintenance (41). Thus it is possible, in the absence of Dcx, that microtubules are less stable, with new growth cones more easily formed at ectopic localizations along the neurite shaft. Indeed, local application of nocodazole, a microtubule-depolymerizing agent, has been shown to encourage membrane addition at ectopic sites in differentiating neurons (42). Thus, side branches could perhaps form more easily in the absence of Dcx. We can also imagine that processes with less stabilized microtubules have a transient nature and retract more frequently in these cells. Thus, our data showing a higher frequency of newly formed growth cones, leading to the generation of more numerous, but less stable branches, is in fitting with the loss of an MAP, normally enriched in growing processes. Further support for the importance of microtubule dynamics in controlling the branching of neurites has also been provided by an analysis of dorsal root ganglia (DRG) neurons derived from map1b mouse mutants (43). Cultures of mutant adult DRG neurons showed significantly higher terminal and side branching and these defects were correlated with reduced amounts of acetylated tubulin, known normally to be associated with highly stable microtubule populations. Aside from its microtubule-related functions, Dcx is also potentially involved in vesicle trafficking (39), regulation of cell adhesion and protein stability (44,45), and microtubule-actin crosstalk (46,47). Disruption of each of these processes could also potentially influence leading edge and branching dynamics (41). Thus, further studies are required to investigate each of these mechanisms in Dcx KO cells, in order to fully explain the branching phenotype.
In WT co-cultures, polarity reversal, a rare event involving the change in position of the MTOC and the establishment of a leading process at the rear of the cell, is observed before an MGE cell reverses its direction of migration. As discussed here, Dcx KO cells reverse their polarity more frequently than WT cells. It is possible that this is a consequence of the perturbed nuclear movements and abnormal dynamics of the MTOC-containing swelling, where swelling movements are not so tightly correlated with nuclear movements. KO swellings thus often showed retrograde movements towards the nucleus, rarely observed in WT MGE cells. KO nuclei also perform shorter amplitude jumps than WT cells and complement these with less distinct gliding movements. Interestingly, an involvement of Dcx, together with Lis1, in nuclear–centrosome coupling has previously been suggested, by Dcx overexpression in radially migrating cerebellar Lis1+/– granule neurons (26). Process growth and branching, normally tightly coupled to nucleokinesis in MGE cells, were not however assessed in this study, making it difficult to completely correlate these results with our study. Dcx could thus have a direct role in nuclear–centrosomal mechanisms in MGE cells or alternatively, the nucleokinesis defects could be secondary to leading edge abnormalities. Such defects could also be present in tangentially migrating PSA-NCAM-positive interneuron precursors, although this remains to be thoroughly investigated.
Surprisingly, changes in nucleokinesis dynamics in MGE cells did not affect the mean nuclear speed of migration. Indeed, it has recently been shown that the mechanism of nucleokinesis is heavily dependent upon myosin II function at the rear of the cell (30,36). Thus, inactivation of Dcx, a microtubule-associated protein, most probably has little effect on the major acto-myosin forces required to move the nucleus. Coherent with a normal nuclear speed in co-cultures, we observed that MGE-derived KO cells in slice cultures appeared to migrate equivalent distances to WT cells and GABA-labelling of embryonic brain sections showed similar results between KO and WT. Thus, a large proportion of KO MGE cells are apparently still able to migrate relatively efficiently in vivo. In fitting with these data, we were unable to detect major abnormalities in the distribution or number of several different populations of cortical interneurons detected immunohistochemically in adult KO brain sections. Subtle differences in the number of calretinin-positive cells in the cortex and hippocampus, however, warrant further investigation of this subpopulation. It also remains possible that other subpopulations, not yet examined, will be affected, or that some subpopulations will show differences in distribution at specific times during their development (48). Other interneuron markers such as calbindin, neuropeptide Y, somatostatin and glutamate decarboxylase 67 and 65 should also therefore be tested to complete this analysis. It will also be important to examine the final morphology of each of the different interneuron subtypes, e.g. by Golgi stainings, in order to search for potential differences.
PSA-NCAM-positive interneuron precursors are derived from the anterior SVZ and continue to be produced and migrate during adulthood (37). Cell morphology during migration is similar to MGE cells (36) and these cells have previously been shown to express Dcx at the extremities of their leading processes (22). The abnormalities we identified in Dcx KO animals are similar to those observed in neural cell adhesion (NCAM) KO mice (49). However, these NCAM KO mice also show a severe reduction in the size of their olfactory bulbs, which is not observed in Dcx KO mice. It is likely, therefore, that closer examination of the RMS will reveal differences between these KO models. Interestingly, Dcx has also been suggested to play a role in adhesion (44,45) and therefore perturbed interactions between neurons and their substrates during migration may explain this phenotype. Such abnormalities may thus also fall into the category of subtle tangential migration abnormalities, which have no major lasting effect on the target structure. It will, however, be interesting in the future to test Dcx KO mice for olfactory deficits and to visualize the migratory dynamics of Dcx KO SVZ-derived cells in vitro.
The existence of only subtle interneuron abnormalities in mouse brain in vivo might in fact be expected, when more globally comparing mouse Dcx KO and human type I lissencephaly phenotypes. Indeed, for radially migrating pyramidal cell precursors, there is an enormous difference between the severe disorganization observed in type I lissencephaly compared with the lack of major abnormalities in the radial organization of the mouse KO cortex. Comparing then interneuron abnormalities in cases of Miller Dieker syndrome, which result from reduced amounts of LIS1, calretinin-positive interneurons in the fetal neocortex showed a 10-fold reduction compared with controls (11). Similarly, in cases of DCX-mutated lissencephaly, clusters of calretinin-positive cells have been detected ectopically underneath the white matter, with a corresponding deficit of interneurons observed in the disorganized cortical plate (Gundela Meyer, personal communication). Thus, like radial migration abnormalities, severe interneuron abnormalities in human are also not identically replicated in the mouse KO. It is clear then that the in vitro models we have used, combined with video microscopy, have been extremely useful in helping us identify subtle abnormalities in Dcx KO interneuron precursors and have provided further insight into the function of Dcx in migrating cells. Importantly, these data provide us with hints towards understanding the potential pathophysiological mechanisms involved in type I lissencephaly and further emphasize that perturbations of the interneuron network are very likely to directly contribute to the epilepsy observed in these cases.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Mice
Homozygote, hemizygote and heterozygote Dcx mutant mice were maintained on C57BL/6 and Sv129Pas backgrounds. Mice were genotyped at postnatal day 10 (P10) or at embryonic stages by either Southern blotting or polymerase chain reaction (PCR) following standard methods (50
). The KO mice studied here were generally DcxY/– males on a defined background, generated in most cases with WT littermate controls by crossing heterozygote females with pure C57BL/6 or Sv129Pas males (Charles River, France). Such crosses were preferred because of the ease of generating WT littermate controls (50% of the litter), with heterozygotes (25%) and KO animals (25%) in the same litter.
For video microscopy experiments Dcx KO mice were crossed with ubiquitously GFP-expressing mice (51), on the C57Bl/6 background. WT and Dcx KO GFP-expressing embryos resulted from crosses between Dcx+/– heterozygous females and GFP-expressing WT or DcxY/– males or GFP-expressing Dcx+/– females and WT or DcxY/– males. Pregnant females were killed by cervical dislocation and the embryos dissected in cold Leibovitz medium (L15, Invitrogen, France). No differences were observed between Dcx–/– female homozygote mutant neurons and those derived from DcxY/– males. Experiments involving mice were performed by authorized investigators adhering to French and European guidelines (approved by the French Ministry of Agriculture and Fisheries and following the European Community Council Directive for the care and use of laboratory animals), and experiments and animal care were followed by regional veterinary offices (Directions des Services Vétérinaires Departementaux).
Inactivation of Dcx by homologous recombination
A conditional construct was generated involving the Cre–lox P site-specific recombination system. An 11 kb XbaI–ScaI fragment containing Dcx exon 3 was cloned in pBR322 EcoRI–EcoRV. The EcoRV site in the vector was modified by the addition of an SpeI linker and this site was later ligated to an XbaI site, thus destroying both sites. The ScaI site in the Dcx locus was modified to include EcoRI and HindIII sites to allow the cloning at the other extremity. A lox P site containing a BamHI restriction site was inserted in the BsaI site upstream of exon 3. A floxed PGK-neo selection gene was inserted into the XbaI site downstream of this exon. This construction was cut with HindIII to release the vector and the linearized insert was electroporated in embryonic stem (ES) cells (origin Sv129Pas) following standard methods. ES cell clones were tested by Southern blotting using external 5' and 3' probes as indicated in Fig. 1. Mice carrying a floxed Dcx allele (with lox P sites flanking exon 3) in addition to a floxed selection gene were crossed with Cre transgenic mice expressing the Cre enzyme early in development [Meu-Cre-40, corresponding to mosaic early embryonic ubiquitous (Meu) Cre mice, line 40 in 31]. This cross-generated KO mice (having deleted Dcx exon 3 and the selection gene) and mice deleted for the selection gene but still carrying the floxed exon 3, which would allow a later invalidation of Dcx in a spatially and temporally controlled manner.
Western blots and immunodetections
To confirm the deletion of exon 3 and the absence of Dcx protein in homozygote and hemizygote mutant mice, protein extracts were prepared from E14.5 embryos (where E0.5 is the day of detection of the vaginal plug), and analysed by sodium dodecyl sulphate–polyacrylamide gel electrophoresis and western blotting following standard procedures (50). Antibodies directed against Dcx [Nterm (20), 1:5000, C18, Santa Cruz, CA, USA, 1:500] and 
-tubulin (Sigma-Aldrich, St Louis, MO, USA, 1:10000) were used for immunodetection.
BrdU labellings and immunohistochemistry
Neuronal progenitor cells of embryos at different stages of development were labelled in vivo by intraperitoneal injections of timed-pregnant females with BrdU (Sigma-Aldrich, 150 µg/g body weight). Females were sacrificed and embryos recovered at different times after injection ranging from 30 min to several days. Alternatively, newborn mice were sacrificed and the brains recovered for analysis. Brains were either frozen directly in isopentane or first fixed in 4% (w/v) paraformaldehyde (PFA) and cryoprotected in 10% (w/v) sucrose in phosphate buffer (pH 7.4) prior to freezing. Ten micrometres serial sections were obtained using a Leica CM3050S cryostat. Immunodetections were performed with anti-BrdU antibody (BD Biosciences Pharmingen, CA, USA, 1:25) using an ARK Peroxidase kit (DAKO, CA, USA) and sections were counter-coloured using Hemalun Mayer (Merck, Germany) followed by ammonium acetate (10 mM) treatment. For experiments involving adult animals, these were perfused with 4% PFA or 4% PFA and 2% glutaraldehyde, followed by cryoprotection in 30% sucrose of the brain. Cryostat sections or freezing microtome sections (40 µm, Leitz 1703 Kryomat) were also immunostained for the following cortical markers using immunofluorescence or immunoperoxidase detection: anti-Reelin G10 (1:50, kindly provided by A. Goffinet, University of Louvain Medical School, Brussels, Belgium) (52); anti-PSA-NCAM (1:500, mouse, kindly provided by P. Durbec, IBDM, Marseilles, France), anti-calretinin (1:10000, rabbit, Swant Laboratories, Switzerland), anti-parvalbumin (1:10000, rabbit, Swant Laboratories, Switzerland), anti-CSPG, clone CS-56 (1:50, Sigma-Aldrich); anti-ARX (1:1000) (53); anti-ß III tubulin (1:500, mouse, BabCo, Richmond, CA, USA); anti-nestin, clone Rat-401 (BD Biosciences Pharmingen, 1:40); and anti-MAP2 (1:300, mouse, Sigma-Aldrich). Freezing microtome sections were mounted on gelatinized slides. For fluorescently labelled sections, nuclei were stained with 4'-6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) present in the mounting solution (Mowiol, Calbiochem, La Jolla, CA, USA) at a concentration of 150 µg/ml.
After 2 days in vitro or after the video microscopy, MGE cultures on cover-slips were similarly fixed in 4% (w/v) PFA/10% (w/v) sucrose in phosphate buffer (pH 7.4). After permeabilization, cultures were incubated overnight with primary anti-GFP (1/1000; Molecular Probes, Eugene, OR, USA and Roche Diagnostics, IN, USA). Labelled cultures were analysed using an epifluorescence microscope (DMRD, Leica, Germany) equipped with a Coolsnap camera (Photometrics, CA, USA) and analysed with Neuron J (NIH Image, MD, USA) software.
Slice cultures, injection of fluorescent markers
Brain slices were prepared from E13.5 to E14.5 WT and KO C57Bl/6 and Sv129Pas mouse embryos. No differences were observed between the different backgrounds. Brains were dissected in cold Leibovitz L-15 medium (Invitrogen, France) supplemented with penicillin and streptomycin (PS, 50 UI, Invitrogen, France) and embedded in 3% (w/v) low melting point agarose (type VII, Sigma-Aldrich). Three hundred micrometres thick coronal sections corresponding to the anterior half of the cerebral hemispheres were cut with a tissue slicer (Campden Instruments Ltd, UK) and deposited on Millicell membranes (Biopore semi-permeable culture membranes, Millipore, France) in DMEM F-12 (Invitrogen, France), glucose 2,4 g/l, L-glutamine 100 µM, 1x N2 (Invitrogen, France), 1x B27 (Invitrogen, France), 5% (v/v) fetal calf serum (Invitrogen, France), PS 50 UI (54). In order to label the population of migrating interneurons generated in the ventral telencephalon, fragments of bamboo (on average 500 µm long, 200 µm wide and less than 100 µm thick) soaked in CMFDA (C-2925, Molecular Probes, 10 mM in dimethyl sulphoxide) were inserted in the MGE of each slice. Slices were then transferred to a humidified incubator at 37°C with 5% CO2 and cultured for 48 h prior to 4% PFA fixation. Fixed slices were analysed using an epifluorescence microscope (Leica) equipped with a Coolsnap camera (Photometrics) and a Leica TCS confocal microscope.
Co-cultures of MGE cells on a carpet of cortical neurons
MGE explants from E12.5 or E13.5 WT and Dcx KO GFP-expressing embryos were deposited on WT-dissociated cortical cells. Dissociated cortical cells were prepared from E13.5 or E14.5 embryos and cultures were performed as described in Bellion et al. (30).
Time-lapse video microscopy
MGE explants were cultured on a carpet of cortical neurons in Petri dishes equipped with glass cover-slips and the culture medium was replaced with HEPES (N-(2-hydroxyethyl) piperazine-N'-(2-ethanesulfonic acid)) buffered minimal essential media (MEM) which was phenol red free (Sigma-Aldrich), containing NaHCO3 4 mM, HEPES 20 mM, L-glutamine 2 mM, glucose 33 mM, pyruvate 1 mM, penicillin/streptomycin 10 UI/ml, supplemented with N2 and B27 (Invitrogen, France). Co-cultures were placed in a humidified and thermo-regulated chamber maintained at 37°C on a stage of an inverted epifluorescence microscope (Axiovert 200, Carl Zeiss, Germany), equipped with a monochromater light, a cooled CCD camera (ORCA, Hamamatsu Photonics, Japan), a motorized stage, a piezo z objective and Plan-apochromat 20x/0.75 NA and Plan-neofluar 40x/0,75 NA objectives (Carl Zeiss).
Images of living GFP-expressing migrating interneurons were acquired during 4–16 h at a frequency of one image each minute, each 3 min or each 5 min. Exposure time was 700 or 1000 ms. Precautions were taken to image WT and KO cultures which had been in culture for similar periods of time. Acquisition and illumination devices were driven by Metamorph software (Universal Imaging, West Chester, PA, USA). Sequences of 8- or 16-bit images were analysed using Metamorph, Image J and Neuron J (NIH Image) software.
For all experiments described here, Dcx KO tangentially migrating neurons were analysed on a carpet of WT cortical cells. It would also be possible to test Dcx KO tangentially migrating neurons on a carpet of Dcx KO cortical cells. With such experiments, it should thus be possible to dissect cell autonomous from non-cell autonomous mechanisms.
Analysis of migrating interneurons in vitro
The Metamorph tracking function was used to follow the movements of the nuclei, the MTOC-containing swellings and the growth cones in video films. The nuclear migration speed was measured in a certain time window during the imaging period, to avoid biases due to perturbation of the culture prior to the imaging period and deterioration of the cells towards the end of the imaging period.
Appropriate indices were developed to compare the morphology of migrating neurons from KO versus WT GFP-immunostained co-cultures using Neuron J (NIH Image). These analyses were as follows: (1) the number of cells of different complexities was divided by the total number of cells analysed to generate a percentage (Fig. 5B), (2) the length of process branches (defined as primary, secondary, tertiary, etc.) emerging from a given interneuron, whatever the complexity of the neuron, was traced and measured. Average lengths for each branch type were then calculated (Fig. 5C) and (3) in order to test for branching asymmetry, the numbers of each type of terminal branch were counted and divided by the number of analysed cells in each class. For statistical comparisons between WT and KO conditions, a Shapiro Wilk test was used to test all data for Gaussian distribution, a Student's t-test was thus used (bilateral, =5%) for data following a normal distribution, presented in Figs 3B and 6C (right side of the graph) and for the nuclear migration speed data. For the data presented in Fig. 5C, initially a Kruskal–Wallis test was used on the pooled data, followed by individual Mann–Whitney tests for the different categories. This test was also used for the data presented in Figs 4D and 6C (left side of the graph).
|
SUPPLEMENTARY MATERIAL |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Supplementary Material is available at HMG Online.
|
ACKNOWLEDGEMENTS |
|---|
This work was supported in part by grants from INSERM, CNRS, the European Commission (No. QLG3-CT-2000-00158), the French Ministère de la Recherche (ACI-1066G and fellowships to C.K. and J.-P.B.), the Fondation Jerôme Lejeune (to C.M.) and the Fédération pour la Recherche sur le Cerveau (C.M. and F.F., 2004, 2005). The contribution of the Région Ile-de-France to the Institut Cochin animal care facility is also acknowledged. The authors are very grateful to Gundela Meyer for communicating unpublished data, to Masuru Okabe for providing the act-EGFP green mice, to Egbert Welker for helpful discussions, to André Goffinet for providing anti-reelin antibodies, to Pascale Durbec for providing anti-PSA NCAM antibodies, to Charles Hébert for aid with statistical analyses, to Evelyne Souil for help with immunohistochemistry experiments and to Emma Cheesman and Marie-Claude Vinet for technical assistance. We thank other members of J. Chelly's Laboratory for their contribution to this study.
Conflict of Interest statement. None declared.
|
FOOTNOTES |
|---|
These authors contributed equally. 
|
REFERENCES |
|---|
|
TOPABSTRACT |
|---|