|Home | About | Journals | Submit | Contact Us | Français|
Heterozygous LIS1 mutations and males with loss of the X-linked DCX result in lissencephaly, a neuronal migration defect. LIS1 regulates nuclear translocation and mitotic division of neural progenitor cells, while the role of DCX in cortical development remains poorly understood. Here, we uncovered novel neuronal migration and proliferation defects in the Dcx mutant embryonic brains. Although cortical organization was fairly well preserved, Dcxko/Y neurons displayed defective migration velocities similar to Lis1+/ko neurons when characterized by time-lapse video-microscopy of embryonic cortical slices. Dcxko/Y migrating neurons displayed novel multidirectional movements with abnormal morphology and increased branching. Surprisingly, Dcxko/Y radial glial cells displayed spindle orientation abnormalities similar to Lis1+/ko cells that in turn lead to moderate proliferation defects both in vivo and in vitro. We found functional genetic interaction of the two genes, with the combined effects of Lis1 haploinsufficiency and Dcx knockout leading to more severe neuronal migration and proliferation phenotypes in the Lis1+/ko; Dcxko/Y male double mutant compared to the single mutants, resulting in cortical disorganization and depletion of the progenitor pool. Thus, we provide definitive evidence for a critical role for Dcx in neuronal migration and neurogenesis, as well as for the in vivo genetic interaction of the two genes most commonly involved in human neuronal migration defects.
Proper development of the mammalian cerebral cortex relies on the integrated control of neurogenesis and neuronal migration. Proliferation of neuronal progenitor cells during early stages of brain development is critical to expand the progenitor pool at the ventricular surface and later mitotic divisions result in the generation of post-mitotic neural precursors, which then migrate to the cortical plate (Gupta et al., 2002; Götz et al., 2005). Defective neurogenesis or neuronal migration leads to brain malformations, and are often associated with different forms of mental retardation or cognitive disabilities and severe epilepsy Guerrini et al., 2008). For example, classical lissencephaly (or “smooth brain”) is due to a reduced number or absence of gyri and sulci of the cortical surface, resulting in severe mental retardation, seizures and early death (Kato and Dobyns, 2003). Mutations in two genes, LIS1 (Reiner et al., 1993; Lo Nigro et al., 1997) and DCX (Gleeson et al., 1998; des Portes et al., 1998), are responsible for most cases of classical lissencephaly (for review see references: Dobyns et al., 1993; Kato and Dobyns, 2003).
The study of murine Lis1 by genetic knockout or knockdown by RNAi revealed important roles in both neuronal migration and proliferation of neuronal progenitors. Reduction of LIS1 protein leads to neuronal migration defects in a dose dependent manner (Hirotsune et al., 1998; Gambello et al., 2003). The translocation of the nucleus during migration towards the tip of the leading process is generated through dynein motor activity that is ultimately regulated by LIS1 via its binding to phosphorylated NDEL1 (Sasaki et al., 2000; Niethammer et al., 2000) and perhaps NDE1 (Feng et al., 2004), which binds to 14-3-3ε for protection against phosphatase activity (Toyo-oka et al., 2003). The LIS1-NDEL1 complex directly associates with cytoplasmic dynein heavy and light chains to promote dynein motor function in a positive fashion (for review see references: Gupta et al., 2002; Wynshaw-Boris A, 2007; Vallee et al., 2006). The LIS1/dynein complex at the cell periphery also regulates the orientation of the spindle of dividing neuronal precursors at the ventricular surface, and reduction of LIS1 protein levels lead to asymmetric positioning of the spindle of dividing cells at the ventricular surface causing cell death of neuroepithelial stem cells and depletion of radial glial progenitor cells (Yingling et al., 2008).
In contrast to LIS1 function, less is known about the role of DCX during brain development. Similar to LIS1, DCX is a microtubule binding protein that has a strong bundling activity that promotes microtubule polymerization and stability during neuronal migration (Gleeson et al., 1999). The germ line ablation of the Dcx gene in mouse resulted in disorganization of the hippocampus but did not lead to abnormalities in radial movements nor cerebral cortex disorganization (Corbo et al., 2002; Kappeler et al., 2006), although Dcx knock-down by RNAi resulted in neuronal migration defects in rat (Bai et al., 2003). Nuclear translocation defects have been reported in Dcx mouse mutants for tangential movements related to the migration of interneurons in the rostral migratory stream (RMS) (Koizumi et al., 2006) or in the cortex (Kappeler et al., 2006). These migration defects resulted from frequent stalling of migrating cells that were unable to maintain proper bipolar morphology during movement, resulting in disorganized migration secondary to increased branching and persistence of multipolar morphology. Similar results were seen after Dcx knock-down in tangentially migrating neurons from the ganglionic eminence to the cerebral cortex (Friocourt et al., 2007). Genetic redundancy might partially explain the discrepancy existing between the cortical malformations affecting human patients with DCX mutations and the mild or no phenotypic effects of the mutated gene in Dcx mouse knockouts, since there are two other Dcx mutants in mammals: doublecortin-like (DCL) and Dcx-like kinase (DCLK) (Koizumi et al., 2006b; Shu et al., 2006). Additionally, it is unknown whether Dcx expression and function is restricted to postmitotic neurons and whether it plays any role in neurogenesis. DCL is highly expressed during early stages of neocortical development and DCLK does play a role during neurogenesis (Shu et al., 2006; Vreugdenhil et al., 2007). DCLK ensures the correct transition from prometaphase to metaphase in dividing cells by regulating the structural formation of the bipolar spindles (Shu et al., 2006). Whether Dcx plays a role during neurogenesis remains unknown.
We analyzed the combined effects of Lis1 and Dcx mutations on mouse cortical development, to understand if the contemporaneous ablation of Lis1 and Dcx leads to a more severe phenotype than single mutants in both neuronal migration and neurogenesis. In addition to finding more severe defects in double mutants, we surprisingly found neuronal migration and proliferation defects in the Dcx knockout brain. Our study provides strong evidence for a functional genetic interaction of the two genes during brain development as well as for neuronal migration and proliferation defects in the cortex of the Dcx single knockout.
Dcx+/ko female mice (Corbo et al., 2002) were mated to Lis1+/ko male mice (Hirotsune et al., 1998) to produce Lis1+/ko, Dcx+/ko female and Lis1+/ko ;Dcxko/Y male mice in a mixed 129SvJ × NIH Black Swiss background. The thy-1-YFP mice [strain name: B6.Cg-TgN(Thy1-YFP-H)2Jrs] were obtained from Jackson Laboratory (Bar Harbor, ME). Mice were housed in standard cages in Association for Assessment and Accreditation of Laboratory Animal Care (ALAC) approved facilities at UCSD and UCSF. All experiments followed the guidelines of University's animal care and use committee. Mice were maintained on a 12:12 light/dark cycle at 22 °C. All testing occurred during the light portion of the cycle.
All analyses were performed on experimental animals with littermate-matched controls. Brains were fixed in 4% paraformaldehyde and cryo-protected in 30% sucrose for freezing or treated with 70% ethanol for paraffin embedding. At least three brains were analyzed for each experiment and matched sections were stained. Sections were roughly matched by counting the number of coronal sections starting at the rostral-most edge of the brain, and then the sections were more closely matched by anatomical landmarks. Cresyl-violet/Nissl staining was done according to standard protocols. Cortical thickness was measured by marking a line around the whole cortex of the brain sections (n≥25 per genotype) followed by quantification of the total area with ImageJ (http://rsbweb.nih.gov/ij/).
Immunohistochemical staining with guinea pig anti-testis-1 antibody provided cortical layer V-specific staining in the postnatal cortex. Paraffin sections were dewaxed in xylenes, rehydrated, and reacted with primary antibody overnight at 4°C. Primary antibody was diluted 1:1000 in blocking buffer. Secondary antibody (biotinylated goat anti-guinea pig) was diluted 1:200 in blocking buffer and incubated for 2 h at RT. The Vector ABC detection system (Vector Laboratories, Burlingame, CA) was used for development with DAB plus metal enhancement (Sigma) as the chromogen. Cortical layer markers Cux1 (rabbit, Santa Cruz biotechnologies), Foxp1 and Foxp2 (rabbit, Abcam) were diluted 1:200 in blocking agent (PBS, 0.2% Triton X-100, 5% Normal Goat Serum) at incubated overnight. Nestin (mouse, Chemicon), Pericentrin (rabbit, Covance), TuJ (mouse, Sigma), Dcx (rabbit, kind gift of Dr. J.G. Gleeson) and PH3 (rabbit, Sigma) were used at 1:200 dilution. Secondary antibodies were diluted 1:500 and incubated for 1 hour. For BrdU (mouse, Sigma), Ki-67 (rat, DakoCytomation) and Tbr2 (rabbit, kind gift of Dr. R.F. Hevner) double and triple staining, pregnant dams were injected 30 minutes prior embryos collection with 100 µg BrdU per g of mouse weight. Brain sections were pre-treated with antigen-retrieval (10 mM Sodium Citrate pH6) and denaturing solutions (2N HCl, 0.5% Triton-X). Primary antibodies were used together at 1:100 dilution (BrdU and Ki-67) and 1:2000 for Tbr2, while secondary antibodies at 1:500 dilution. Quantification of the single-, double-and triple-stained cells was done manually by counting all the cells present in the images taken from the dorsolateral cortex (n≥20 per genotype for PH3; n≥30 per genotype for the double staining; n≥20 for the triple staining). Acquisition was done at the level of the dorsolateral cortex using the Nikon Spectral C1 confocal microscope with 60× objective (UCSF Nikon Center, Parnassus).
E9.5 and E14.5 brains sections were prepared at 12µM thickness and counter-stained with DAPI in order to see telophase mitotic cells at the ventricular surface. Captured images were analyzed with ImagePro v.6 to determine the angles of the spindle orientation by drawing a line parallel to the ventricular surface and a line parallel to the axis of the dividing cells (Fig. 5A; n>150 cells per genotype) and average angle values were calculated from these measurements (Yingling et al., 2008).
These experiments were performed as previously described (Nadarajah et al., 2001; Youn et al., 2009) and three brains for each genotype were analyzed. Briefly, E14.5 brains were collected and sectioned coronally with a tissue slicer (MX-TX, Siskiyou Inc). Cortical slices were transferred to culture medium with 1% N2 supplement in F12/MEM medium containing 10µg/mL Oregon green (488 BAPTA-1, Invitrogen). Slices were incubated for 1 hour at 37°C in 5% CO2. Individual slices were washed and then placed in a 60 mm culture dish and covered with neutralized collagen (Rat tail, BD biosciences), which was allowed to solidify for 2–3 hours. Oregon green-stained cells were imaged using a 2-photon confocal microscope (Olympus FV300) with a 20× dipping objective. The same media with the addition of HEPES was used during the recording. Time-lapse capture was performed every 7 minutes using the FLUOVIEW program and slices were imaged for 6–8 hours in a temperature controlled chamber. The files generated were merged and projected movies were analyzed using ImagePro v.6. The quantification of the angle changes was done using the same software (n>25 for WT; n>80 for Dcx+/ko; n>100 for Dcxko/Y). The number of branches was quantified in at least 50 moving cells per genotype. All movies have been scaled and calibrated according to the acquisition parameters before the tracking analysis. All cells that displayed a migration of at least 3 cell-body distance was quantified, while the remaining cells were considered stationary (n>200 for WT, n>100 for single mutants; n=10 for double mutant). The percentage of moving cells was estimated within three arbitrary areas from each time-lapse video-microscopy movie at the level of the IZ (Youn et al., 2009).
Primary embryonic cortical cultures were established from E14.5 dissociated brains and cultured in order to generate neurospheres using a modification (Youn et al., 2009) of previously described methods (Reynolds et al., 1992; Rietze et al., 2006; Ishii et al., 2008). In brief, papain-dissociated brains were passed through a 70µm nylon cell-strainer (BD Falcon) and media containing cells homogenized until single cells were visible under light microscope. EGF and bFGF (10ng/mL) were added to the media (DMEM/F12 with N2 supplement) to stimulate the growth of neurospheres in suspension. No significant differences among the genotypes were seen in the number of neurospheres generated at P0. These neurospheres were collected and trypsinized to obtain P1 single cells. 5000 P1 single cells were seeded per each well (12 well-plates) and cultured for 13 days (13DIV). Usually floating neurospheres appeared after 3–5 DIV. For the MTS assay (CellTiter 96 AQueus One solution cell proliferation kit, Promega; Learish et al., 2000; Bantubungi et al., 2008), 5000 P1 single cells were seeded in triplicates in a 96 well plate. MTS solution was added to the media and after 2 hours MTS reduction was recorded at 490nm using an automated microplate reader (Biotek synergy 2).
To study whether Lis1 and Dcx genetically interact during mouse brain development, we generated double mutant Lis1+/ko ;Dcxko/Y mice by mating single mutant Lis1+/ko males with Dcx+/ko females. Due to the lethality of most male Lis1+/ko ;Dcxko/Y mouse mutants shortly after birth, we analyzed cresyl violet-stained sections of adult (P21) female wild-type (WT), Lis1+/ko, Lis1+/ko;Dcx+/ko mice. Both Lis1+/ko and Lis1+/ko;Dcx+/ko mutants displayed an increased volume of the ventricles, disorganized layering and decreased size of the hippocampus with disorganization and loss of other brain areas such as the caudate and putamen (Fig. 1A). Double mutant Lis1+/ko;Dcx+/ko females displayed more severe defects than single Lis1 mutants, especially in the hippocampus (Fig. 1A). Analysis of WT, Lis1+/ko, Lis1+/ko;Dcx+/ko mice containing the thy1-YFP transgene that is expressed in a subset of cerebral cortical layer V pyramidal cells, hippocampal pyramidal cells and cells of the fascia dentate confirmed this graded severity of disruption (Fig. 1A). A few Lis1+/ko;Dcxko/Y males survived to P21. These mice displayed more severe disorganization of the hippocampal structure, as demonstrated by cresyl violet staining and Testis-1 immuno-chemistry (Fig. 1B). The organization of the cerebellum of double mutant males was also severely disrupted to a greater extent than in female Dcx+/ko and Lis1+/ko;Dcx+/ko mice (Fig. 1B).
We examined all genotypes in males prenatally at two developmental stages (P0 and E13.5). At P0 there was an increasing severity of the hippocampal disorganization across the mutants. In agreement with previous reports (Corbo et al., 2002), the hippocampal pyramidal layer of Dcxko/Y males displayed moderate splitting of the CA3 region, while in Lis1+/ko males the splitting extended into the CA1 region (Fig. 1C). The hippocampus of Lis1+/ko;Dcxko/Y males was much smaller with severe disintegration and layering of the pyramidal layer. At E13.5 major structural abnormalities were not detected, but there was a reduction in the cortical thickness of Lis1+/ko and Lis1+/ko;Dcxko/Y brains. There was no significant difference between the Dcxko/Y male single mutant compared to WT, but there were 9–13% (P<0.001) and 15% (P<0.001) differences in the cortical thickness in Lis1+/ko;Dcxko/Y males compared to the single mutants and WT males respectively (Fig. 1C). There was also a slight significant difference between the Lis1+/ko and the Dcxko/Y single mutants (P<0.01). The overall severe phenotype of Lis1+/ko;Dcxko/Y and the Lis1+/ko ; Dcx+/ko mice compared to the single mutants strongly suggested a synergistic effect of the two genes during brain development.
We next investigated neuronal migration ex-vivo using organotypic slice cultures from E14.5 male WT, Dcxko/Y or Lis1+/ko single mutants and Lis1+/ko ;Dcxko/Y double mutant brains (Fig. 2A). Most cells in WT slices migrated away from the starting position (dotted lines) and only a few were stationary (Fig. 2A, blue outlines). In the Lis1+/ko or Dcxko/Y single mutants and Lis1+/ko;Dcxko/Y double mutant male slices there was a decreasing proportion of moving cells (Fig. 2A and Supplementary Movies 1–4). In WT slices 73% of the cells moved away from the starting position. In the Dcxko/Y and Lis1+/ko single mutant slices only 37% and 18% of cells moved, respectively, while cells in Lis1+/ko;Dcxko/Y slices displayed a very severe phenotype with very few moving cells (Fig. 2B). The migration of moving cells was tracked at each time point across the cortex independently of the direction of the movement. WT cells migrated with a speed of 55±2.35 µM/h, while Dcxko/Y and Lis1+/ko displayed slower migration speeds of 38±2.42 and 36±4.29 µM/h respectively (Fig. 2C). Lis1+/ko;Dcxko/Y double mutant cells were mostly stationary and did not move from their initial position, displaying a rounded shape without forming processes (Fig. 2C; Supplementary Movies 1–4). When migration velocities for moving and non-moving cells were combined the difference in the average velocity across the four genotypes was markedly different (Fig. 2D). Of note, the speed of moving cells in the Dcxko/Y and Lis1+/ko single mutants were approximately the same, suggesting that a major difference in the cortical outcomes of these mutants may be due to the proportion of moving cells. We found similar differences for the average velocity, maximal velocity, maximal acceleration and total distance traveled when analyzing migrating cells that were migrating in radial, retrograde and tangential directions (Supplementary Fig. 1). There was also no evidence for a bimodal distribution of migrating cells in female Dcx+/ko cortical slices (Supplementary Fig. 2) as would be expected based on the phenotype of heterozygous mutations in females, although this measurement may not be sensitive enough to detect such differences.
The vast majority of WT neurons migrated exclusively in radial, retrograde and tangential directions Surprisingly, we noticed that many cells in both Dcx+/ko and Dcxko/Y genotypes displayed from a few to several changes of direction during their migration (Fig. 3A,B). These cells moved with unpredictable patterns, such as L-shape, C-shape, S-shape, or zig-zag patterns (Fig. 3A,B; Supplementary Movie 5) or even more complex patterns (Fig. 3A, pink track in Dcxko/Y slice, see enlarged inset). Multidirectional movements were at least three times more frequent in Dcx+/ko and Dcxko/Y neural progenitor cells than in WT or Lis1+/ko progenitor cells (Fig. 3C). Most of these multidirectional-migrating cells (70% of WT and 60% of Dcx+/ko cells) changed direction once, and only a small proportion changed direction 2, 3 or 4 times. By contrast, the majority of Dcxko/Y cells changed direction at least twice, and a number of cells changed direction 3 or 4 times with a few cells changing direction 7 or 8 times (Fig. 3D). Such high numbers of directional changes were never seen in WT or Dcx+/ko mutant. This peculiar migration phenotype was accompanied by abnormal morphology of the Dcxko/Y cells, with these migrating neuronal precursors displaying a typical neurite swelling detached from the cell body and an increased number of branches (Fig. 3E,F), as previously described in similar studies (Kappeler et al., 2006; Koizumi et al., 2006).
To determine whether these neuronal migration defects were associated with anatomical disorganization in the mutants, we examined laminar organization using layer-specific markers on P0 cortical sections. Cux1 (layers II–III) and Foxp1 (layers III–V) were similar to WT in Dcxko/Y and Lis1+/ko single mutants, but a dramatic reduction of Cux1 staining was evident in Lis1+/ko;Dcxko/Y double mutants with very few or no cells present in the outermost layers (Fig. 4, yellow arrowheads). Foxp1 positive cells were mildly abnormal in the Dcxko/Y and Lis1+/ko single mutants, but in the Lis1+/ko;Dcxko/Y double mutant, Foxp1 positive cells were shifted towards the surface. Foxp2 cells (layers V–VI) were in similar places in the lower layers of the cortex in WT and single mutants although there was some reduction of these cells in the Lis1+/ko mutant. By contrast, in the Lis1+/ko;Dcxko/Y double mutant, Foxp2 positive cells were near the pial surface and substantially overlapped with Foxp1 positive cells (Fig. 4, yellow arrows). It appears that Foxp1 and Foxp2 positive cells are mixed, similar to the mixing of cortical layers that we previously showed in Lis1hc/ko brains (Hirotsune et al., 1998; Gambello et al., 2003).
We recently demonstrated that Lis1 controls neuroepithelial expansion and/or radial glial neurogenesis by promoting the capture of microtubules at the cell cortex and controlling the mitotic spindle orientation of neuronal progenitors at the ventricular surface (Yingling et al., 2008). We tested whether haploinsufficiency for Lis1 combined with the absence of Dcx would lead to a more severe spindle orientation phenotype at E9.5 and E14.5 (Fig. 5A). During neuro-epithelial expansion (E9.5), when Dcx is not expressed there were no significant changes in the spindle orientation for Dcxko/Y and Lis1+/ko;Dcxko/Y male mutants compared to WT and Lis1+/ko respectively, although as expected Lis1 haploinsufficiency resulted in a more randomized spindle orientation compared to WT (Fig. 5B). During radial glial neurogenesis (E14.5) we surprisingly found that the average angle of spindle orientation in the Dcxko/Y brains was randomized to a similar degree as Lis1+/ko brains (Fig. 5B), and the Lis1+/ko;Dcxko/Y double mutant brains displayed more randomized spindle orientation compared to the single mutants. We found dividing cells at E14.5 that co-expressed DCX in the SVZ/IZ (Fig. 5C; Brown et al., 2003). Low levels of DCX have been previously reported in cells isolated from the VZ/SVZ of P2 mouse brains, while at E14 20% of same population co-expressed the precursor marker nestin (Walker et al., 2007). Apical-basal polarity was preserved in all mutants, indicated by the correct localization of atypical PKC (Fig. 5A), beta-catenin (data not shown) and the presence of the cadherin hole at the apical membrane (data not shown; Kosodo et al., 2004). Although aPKC was normally localized in the mutants, analysis of the projected Z-stack images revealed a broader expression pattern with a thickened distribution at the ventricular surface, and the distribution of the centrosomes was less organized (Supplementary Fig. 3). Nestin immunostaining displayed a mild reduction of labeled fibers at the VZ/SVZ in the single Lis1+/ko mutants and a normal pattern in Dcxko/Y mutants. Nestin staining in the Lis1+/ko;Dcxko/Y double mutant was significantly reduced at the VZ/SVZ with less organized and more fragmented fibers (Fig. 5D). Thus, the abnormal distribution of atypical PKC likely resulted from a general cellular disorganization.
During radial glial neurogenesis, the positioning of the spindle determines the fate of the two dividing daughter cells. A higher number of asymmetric divisions during neurogenesis should commit more cells to the neuronal lineage and lead to a depletion of the progenitor cells. We investigated both cell proliferation and cell-cycle exit in all genotypes by injecting BrdU 30 min prior to collection of the brains and performing double staining with BrdU and Ki-67 antibodies. In WT brains, 41% of cells were BrdU-positive, while in mutants this was reduced to 39% and 33.7% in the Dcxko/Y and Lis1+/ko single genotypes respectively and 27.2%in the Lis1+/ko;Dcxko/Y double mutant brains (Fig. 6A,B). In WT brains, 17% of dividing cells exited the cell-cycle, while in Dcxko/Y and Lis1+/ko single mutants, 21.4% and 26.8% of dividing cells exited the cell-cycle respectively, and in Lis1+/ko;Dcxko/Y double mutants, 44.5% of dividing cells exited the cell-cycle (Fig. 6A,C). We further investigated whether cell cycle exit was increased in intermediate and/or radial glial progenitors using Tbr2 as third marker. In WT, Dcxko/Y and Lis1+/ko single mutants, cell cycle exit occurred exclusively in the radial glial population, while in the Lis1+/ko;Dcxko/Y double mutants both populations were affected, although with different degree of severity (Fig. 6A,D). We quantified the number of ventricular and abventricular mitoses by staining with phospho-histone H3 antibody and found that the double mutant Lis1+/ko;Dcxko/Y brains had a higher number of abventricular mitoses compared to the other genotypes (Fig. 6A,E). Staining with the neuronal marker TuJ confirmed the presence of more postmitotic neurons in the Lis1+/ko;Dcxko/Y genotype compared to WT, while there were no major differences in the single mutants (Fig 6A). We determined whether the increase in cell-cycle exit was accompanied by an increase in apoptosis. Although more apoptotic cells were found in the Lis1+/ko single mutants compared to WT or Dcxko/Y single mutants (0.81%±0.3 compared to 0.22% ±0.15 or 0.32%±0.11 respectively), no significant differences were seen in the Lis1+/ko;Dcxko/Y double mutants compared to the Lis1+/ko single mutants (0.8%±0.25). These data suggest that there was depletion of the progenitor pool during neurogenesis in the Lis1+/ko;Dcxko/Y double mutant, perhaps due to more randomized mitotic divisions of progenitor cells at the VZ.
We tested whether there were proliferation defects in these mutants in vitro using neurosphere cultures. We counted the number of neurospheres generated at day 5 in vitro (5DIV) after establishment of primary cortical cultures up to 13DIV. We collected data using 9, 5, 3, and 2 plate-replicates for WT, Dcxko/Y, Lis1+/ko and Lis1+/ko;Dcxko/Y cultures, respectively (Fig. 6F). Although there were differences in the number of neurospheres at each time point, the only significant differences were at 11 and 13 DIV between Dcxko/Y and Lis1+/ko;Dcxko/Y and at 13DIV between Lis1+/ko and Lis1+/ko; Dcxko/Y probably due to the intra-individual variability among replicates and the inability of Lis1+/ko and Lis1+/ko;Dcxko/Y cells to reproducibly generate neurospheres in suspension. Therefore we used a second method to asses proliferation and viability based on the absorption of the soluble version of MTT (MTS). Trypsinized P0 neurospheres were plated as single cells and cultured for 6 days under the same conditions used to grow neurospheres and proliferation was measured using an automated plate reader (Day 0 was considered 12 hours after plating). The single Dcxko/Y and Lis1+/ko mutants displayed a moderate but significant decrease in proliferation compared to WT starting at 2DIV (Fig. 6G), while cell viability of the Lis1+/ko;Dcxko/Y double mutant instead was reduced at day 0 and proliferation of these cells severely was reduced starting at 1DIV. In summary, this in vivo and in vitro data strongly suggest a critical role for Dcx in proliferation, resulting in depletion of the progenitor pool that may be related to the differences seen in the positioning of the mitotic spindle of radial precursor cells at the VZ.
We initiated this study to investigate the genetic interaction between Lis1 and Dcx in vivo, and we demonstrated here that the genetic interaction of the two genes results in more severe defects in both neuronal migration and the proliferation of neuronal precursor during brain development. Moreover these findings led us to examine the Dcx single mutant, where we found that neuronal migration and proliferation are also defective, although cortical organization in the Dcx knockout mouse is fairly well preserved at the histological level (Corbo et al., 2002; Kappeler et al., 2006). These studies demonstrate an essential role for Dcx in neuronal migration and, surprisingly, during neurogenesis.
Based on the neonatal lethality of Dcxko/Y males, it is not surprising that Lis1+/ko;Dcxko/Y double mutant also died shortly after birth. Therefore, we analyzed the morphology of adult Lis1+/ko;Dcx+/ko female mutant brains, which displayed severe disruption of the hippocampus, enlargement of the ventricles and loss of other brain areas compared to the milder disorganization of the hippocampus of the Lis1+/ko and Dcxko/Y mice. During embryogenesis, we confirmed the severe disorganization of the hippocampus in the Lis1+/ko;Dcxko/Y male mutants. At E13.5 we detected a reduction in the cortical area in the double male mutant compared to the WT and single mutant genotypes. Subsequent analysis of the P0 brains with layer-specific markers suggested that both neurogenesis and neuronal migration defects affected the normal development of the Lis1+/ko;Dcxko/Y male cortex. Since neurogenesis was found to be severely defective in the double mutant with a conspicuous increase of cells exiting cell cycle, this shift towards more neurogenic divisions may trigger a depletion of the progenitor pool and may explain the reduced size the cortex during early stages and the mislocalization or absence of neurons in the most outer cortical layers at later times.
As previously reported, heterozygous mutation of Lis1 in mice is sufficient to affect neuronal migration (Hirotsune et al., 1998; Gambello et al., 2003). However, the inactivation of the Dcx gene in males did not lead to any major cortical phenotype, although these mice did display moderate defects of the hippocampus (Corbo et al., 2002). Previous studies suggested that there is a cross-talk between the two proteins regarding the binding and stabilization of microtubules in the developing cortex (Caspi et al., 2000). In addition, the migration defects of Lis1+/ko neuronal cells were rescued by Dcx overexpression (Tanaka et al., 2004). By contrast, RNA interference of Dcx in migrating neurons resulted in migration defects, but neither additive or synergistic effects on migration were found when knockdown of both Lis1 and Dcx genes were combined (Bai et al., 2003).
Here we demonstrated that the concomitant genetic deficiency of the Lis1 and Dcx gene products resulted in the inability of most of the neurons to migrate in the cortex and the migration ability of the few remaining neurons was severely reduced. Moreover, we detected a similar decreased average velocity in Lis1 heterozygotes and Dcxko/Y males, supporting the notion that both genes are essential for neuronal migration. The maximal velocity and acceleration of neurons were not severely impaired in the single mutants compared to the WT neurons, suggesting that some cells maintain the mechanical ability to effect short efforts in migration, which likely depends on the availability of the proteins responsible to promote nuclear translocation. The concomitant reduction of the two proteins resulted in a dramatic loss of all aspects of migration, including maximal velocity and acceleration.
In our initial description of the Dcx knockout mouse, we found no major abnormalities in cortical lamination, although migration of cortical neurons was not directly investigated (Corbo et al., 2002). Radial migration was found to be affected in rat neocortex after Dcx RNA interference (Bai et al., 2003), and knockdown of Dcx significantly slowed the migration of rat interneurons from the ganglionic eminence to the cerebral cortex (Friocourt et al., 2007).
Our observations shed some light on the unexplained discrepancy between the phenotypes of the human patients and the normal cortex so far reported for Dcx knockout mice, suggesting that the difference in phenotype may be due to the way the tissue develops and perhaps the extended migratory path needed in human cortex to reach the cortical plate rather than a change in Dcx requirement or redundancy.
We detected an overall reduced velocity of migration and an atypical migration behavior consisting of frequent changes of direction in the cortex. Defective neuronal migration in the Dcx mouse knockouts have been reported only for tangentially migrating interneurons derived from the medial ganglionic eminence or in the RMS (Kappeler et al., 2006; Koizumi et al., 2006). In these studies, the migration of Dcx mutant neurons was disorganized with reduced long-distance movements of the nucleus, and migration was characterized by long stuttering-like pauses resulting in a delayed migration that was independent of direction or responsiveness to Slit chemorepulsion. In both studies, Dcx mutant neurons displayed atypical multipolar morphology with increased number of branches. Our studies definitely demonstrate that Dcx is essential for radial, tangential and VZ-directed migration in the cortex.
DCX is on the X chromosome in humans and mice. Heterogyzote females with DCX mutations display a “double cortex” phenotype due to random X inactivation. Neurons with inactivation of the X chromosome with the normal DCX allele are completely missing DCX protein, and remain as a band of heterotopic neurons, while neurons that inactivate the X chromosome with the mutated DCX gene have normal levels of DCX protein and migrate normally (des Portes et al., 1998; Gleeson et al., 1998; Kato and Dobyns, 2003). In rodents, genetic studies have demonstrated a cell autonomous mechanism for Dcx effects on neuronal migration (Koizumi et al., 2006; Kerjan et al., 2009), while RNA interference studies have shown that partial non-cell autonomous effects may arise likely due to the cooperative property of radially migrating cells in the neocortex (Bai et al., 2003; Ramos et al., 2006;). In our study the reduced velocity of Dcx deficient cells did not demonstrate the clear cell autonomous mechanism for migration in female heterozygotes, since the distribution of both WT and Dcx-deficient populations in the Dcx+/ko female genotype was not bimodal, but rather uniform and in between the WT and Dcxko/Y speed of migration. Similarly the quantification of the multidirectional phenotype and the number of direction changes of the Dcx+/ko and Dcxko/Y genotypes may be interpreted as the effects of both cell- and non cell-autonomous mechanisms. These findings are in agreement with Bai et al. (2003) that demonstrated dual effects on radial migration of cortical rat neurons after inactivation of the DCX protein function. However, an alternative possibility is that there was not an adequate differential velocity between WT and Dcx mutant cells to observe a bimodal distribution, as would be expected based on the phenotype in human female heterozygous (Matsumoto et al., 2001). We have previously shown cell autonomous effects for Lis1 in neuronal migration defects in vitro (Hirotsune et al., 1998; Gambello et al., 2003). Subsequently, using embryonic brain slice transplant cultures of tangentially migrating interneurons, mixing wild type and mutant substrates and migrating cells, we clearly demonstrated that the main effects of Lis1 during tangential migration are cell autonomous, although small non-cell autonomous effects were found to be due to alterations of the migration substrate (McManus et al., 2004). Thus, it is likely that there are major cell autonomous and some additional non-cell autonomous effects when either Lis1 or Dcx is mutated.
A role for Dcx in neurogenesis and proliferation has not previously been reported, and DCX is commonly used as a marker for postmitotic neurons (des Portes et al., 1998; Gleeson et al., 1999). Here, we provide several lines of evidence that Dcx does in fact play a role in neurogenesis and proliferation of neuronal progenitors. First, the average spindle orientation of Dcxko/Y and Lis1+/ko;Dcxko/Y radial glial progenitors was shifted towards more asymmetric divisions compared to WT and Lis1+/ko progenitors, respectively. Second, this change correlated with moderate but significant differences in the number of cells positive for BrdU and in cell cycle exit. Third, cell cycle exit occurred significantly in the radial glial population in both single and double mutants, with Lis1+/ko and Dcxko/Y single mutants showing no change in the intermediate progenitor population; this strongly correlates with the phenotypic differences in the spindle orientation. Fourth, there was an increase in abventricular mitoses in Dcxko/Y and Lis1+/ko;Dcxko/Y embryos compared with WT and Lis1+/ko or Dcxko/Y embryos, respectively. Finally, in vitro studies using embryonic cortical cultures directly support a role for Dcx in proliferation. Importantly, there was no effect of Dcx mutation on spindle orientation in neuroepithelial stem cells at E9.5, when Dcx is not expressed. Additionally, the differences found in the number of BrdU positive cells, the excessive percentage of cells exiting the cell cycle and the reduction in number of later born Cux1 positive neurons correlated with the change in the average spindle orientation of the Lis1+/ko;Dcxko/Y mutant. Taken together, these data indicated that, like Lis1, Dcx plays a crucial role in neuronal migration and a significant role in maintaining the correct balance between neurogenic and stem cell-like mitotic divisions. Loss of Dcx eventually contributes to the depletion of the progenitor pool during cortical development.
Supporting a role for Dcx in proliferation, a sub-population of cells expressing Dcx at low levels was detected that proliferates in a stem cell-like fashion both at E14.5 and P2 (embryonic day 14.5 and postnatal day 2; Walker et al., 2007). A role for Dcx in cell proliferation might be expected based on the study of mutation of its ortholog zyg-8 in C. elegans that demonstrated defective mitotic spindle positioning and microtubule assembly (Gönczy et al., 2001). Other Dcx homologues are clearly involved in neurogenesis; Dclk is highly expressed in regions of active neurogenesis in the cortex and regulate the formation of bipolar mitotic spindles and the proper transition from prometaphase to metaphase during mitosis (Shu et al., 2006). Dcl, a splice variant of Dclk, is expressed in radial glial and neural progenitors and regulates mitotic spindle stability, precursor proliferation and integrity of the radial fiber network (Vreugdenhil et al., 2007). As we demonstrate, Dcx itself plays a significant role in neurogenesis as well.
Our study provides definitive evidence, from both single and double mutants, that Dcx plays important roles in neuronal migration and neurogenesis (Fig. 7), and strongly suggests that functional relationships of these two proteins are critically important in regulating specific cellular functions required for migration and neurogenesis. DCX binds ubiquitously to microtubules in non-neuronal cells and its activity is essential for their bundling and stabilization (Gleeson et al., 1999). Perhaps this function is impaired in Dcx mutants resulting in abnormal branching and change of direction during migration. We have recently uncovered an important role of Lis1 in mitotic divisions of neuroepithelial and radial glial precursor cells at the VZ (Yingling et al., 2008), required for proper attachment of microtubules at the cell cortex. Perhaps DCX plays an important role with LIS1 in cortical dynein capture, although the precise mechanisms have yet to be defined. These findings suggest that the molecular interaction of the LIS1 and DCX is important in both in neuronal migration and neurogenesis. In addition, there is a cortical role of DCX in nuclear translocation and positioning of the mitotic spindle in radial glial mitotic division.
Supplementary Figure 1 Quantification of the migration differences in each of the genotypes based on the direction of movements. A) Radial and retrograde (or VZ-directed) migration was similar for each genotype while, as expected, tangential migration was faster than radial or retrograde movements. Neuronal migration defects were increasingly more severe in each of the mutants, and there were only a few slowly migrating cells in the Lis1+/ko;Dcxko/Y double mutants. Lis1 and Dcx single mutants displayed similar speeds of migration for radial and retrograde migration. B–C) Quantification of maximal velocity and acceleration in each of the genotypes revealed that Lis1 and Dcx single mutant neurons mostly maintained the capacity for a short burst in migration in any direction, which was insufficient to maintain migration over a long run leading to slower average speed. The velocity of migration in Lis1+/ko;Dcxko/Y double mutant was severely affected in all three directions. D) Differences in the distance travelled in each of the genotypes for each direction. Although Lis1 and Dcx single mutants displayed similar migration velocity, Dcxko/Y neurons covered greater distances compared to Lis1+/ko mutants. This was likely due to the difference in number of migrating cells. Error bars indicate S.D. *P<0.05; **P<0.001; ***P<0.0001 by two-tailed T-test distribution.
Supplementary Figure 2 Distribution of velocity of migration in cells from Dcx mutants. We compared the distribution of the velocity of cells from WT, Dcx+/ko and Dcxko/Y cortical slices to determine if mutation of Dcx affected neuronal migration in a cell-autonomous fashion. The Dcx+/ko distribution did not display a bimodal distribution (WT and Dcxko/Y populations).
Supplementary Figure 3 Localization of apical markers indicate disorganization at the VZ. The projection of Z-stack images displayed a broader expression pattern of atypical PKC in Dcxko/Y, Lis1+/ko and Lis1+/ko;Dcxko/Y mutants compared to WT and Dcx+/ko mutants. Accordingly, the distribution of the centrosome at the VZ seems to be less organized. These findings indicate a degree of disorganization of the ventricular surface. Scale bar 10µm.
Supplementary Movie 1 Slice culture imaging of a E14.5 WT brain. The brain slice is oriented with the VZ at the bottom and the pial surface at the top. Most of the labeled cells migrate away from the starting position and migrating neurons move with an appreciable speed.
Supplementary Movie 2 Slice culture imaging of a E14.5 Dcxko/Y brain. The brain slice is oriented with the VZ at the bottom and the pial surface at the top. Compared to WT fewer cells are able to migrate and display slower migration movements.
Supplementary Movie 3 Slice culture imaging of a E14.5 Lis1+/ko brain. The brain slice is oriented with the VZ at the bottom and the pial surface at the top. Compared to WT fewer cells are able to migrate and display slower migration movements.
Supplementary Movie 4 Slice culture imaging of a E14.5 Lis1+/ko;Dcxko/Y brain. The brain slice is oriented with the VZ at the bottom and the pial surface at the top. Neuronal migration is severely affected with nearly all cells were unable to move from the starting position. The majority of cells were unable to generate processes and to initiate the migration movement.
Supplementary Movie 5 Slice culture imaging of a E14.5 Dcxko/Y brain. The brain slice is oriented with the VZ at the bottom and the pial surface at the top. The multidirectional phenotype of a Dcxko/Y migrating neuron that displays a zig-zag movement and increased branching is evident.
This work was supported in part by the NIH (NS041310 and HD047380 to A.W-B.). We thank Lana Bogdanova and Donna Holland for animal handling, and Jeehee Hong and Patricia LaPorte for technical assistance. We especially thank Brendan Brinkman for technical expertise and support at the UCSD Neuroscience Microscopy Shared Facility (P30 NS047101).