|Home | About | Journals | Submit | Contact Us | Français|
The cerebral cortex is a laminated sheet of neurons composed of the arrays of intersecting radial columns1,2,3. During development, excitatory projection neurons originating from the proliferative units at the ventricular surface of the embryonic cerebral vesicles migrate along elongated radial glial fibers4 to form a cellular infrastructure of radial (vertical) ontogenetic columns in the overlaying cortical plate (CP)5. However, a subpopulation of these clonally related neurons also undergoes a short lateral shift and transfers from their parental to the neighboring radial glial fibers6, and intermixes with neurons originating from neighboring proliferative units5,7. This columnar organization serves as the primary information processing unit in the cortex1,8,9. The molecular mechanisms, role and significance of this lateral dispersion for cortical development are not understood. Here we provide the evidence that EphA/ephrin-A signalling-dependent shift in allocation of clonally related neurons is essential for the proper assembly of cortical columns. In contrast to the relatively uniform labelling of the developing CP by various molecular markers and retrograde tracers in wild-type mice, we found alternating labelling of columnar compartments in ephrin-A knockout mice that are caused by impaired lateral dispersion of migrating neurons rather than altered cell production or death. Furthermore, in utero electroporation revealed that lateral dispersion depends on the expression levels of EphAs and ephrin-As during neuronal migration. This heretofore unrecognized mechanism for lateral neuronal dispersion appears to be essential for the proper intermixing of neuronal types in the cortical columns, which, when disrupted, might contribute to neuropsychiatric disorders associated with abnormal columnar organization8,10.
Functional columns in the neocortex emerge developmentally from a proliferative zone protomap, and consist of multiple classes of interconnected neurons1,11,12 defined anatomically and physiologically.. In the present study, we have explored possible roles of A-type Eph receptors and their ligands, ephrin-As, in the integration of neurons in radial columns because they have been implicated in controlling cell positioning in a variety of developmental contexts by either sorting cell types13,14,15, restricting their intermingling16, or regulating their migration17. First, to determine their expression during the period when the majority of postmitotic neurons migrate radially to the CP, sections were incubated with human Fc-tagged ephrin-A and EphA proteins that bind to all EphAs and ephrin-As, respectively. We observed an elaborate but stereotypical binding pattern of both receptors and ligands across the thickness of the neocortex with a peak expression in the intermediate zone (IZ), where they exhibited opposing gradients (Supplementary Fig. 1). In fact, multiple EphA receptors and ephrin-A ligands were detected in the subventricular zone (SVZ) and/or IZ through which neurons originated in the ventricular zone (VZ) migrate to the CP (Supplementary Fig. 1). To examine the roles of EphAs/ephrin-As in neuronal migration in the neocortex, we analyzed the triple knockout (TKO) mouse18 for ephrin-A2, A3, and A5 that account for almost all ephrin-As in the developing neocortex19. To minimize the effects of mistargeted afferent and efferent projections19,20,21, we performed most analyses at postnatal day 0 (P0) or embryonic stages before the establishment of these projections. It has been demonstrated recently that EphA/ephrin-A signaling has a proapoptotic effect in the embryonic cortex without affecting the proliferation or cell cycle progression22. Consistently, proliferation and cell cycle progression were not affected in the TKO cortex (data not shown). In addition, similar to the EphA7 knockout mice22, less than 10% of the 52 embryos showed an exencephalic phenotype at E14.5, which is likely caused by decreased apoptosis of neural progenitors22. These animals were excluded from our analyses. The remaining TKO embryos did not exhibit any difference in the number of apoptotic cells in the cortex, gross brain shape, brain mass, cortical surface size and cell numbers in a cortical hemisphere compared to wild-type embryos (Supplementary Fig.2).
The overall patterning of cortical areas at P0, revealed by in situ hybridization , appeared unaffected in TKO (Supplementary Fig.2, 3). However, more detailed analysis revealed spotted, irregular distribution of cells labeled by these markers, and alternating thickening and thinning of the CP in TKO mice (Supplementary Fig. 3). Analysis of the wild-type (n ≥ 7) and TKO (n ≥ 9) brains showed highly uneven distribution of Ctip2+ neurons in layer 5 along the tangential axis in the TKO, although their radial positioning was unaffected (Fig. 1a). Most domains with an excess of Ctip2+ neurons were found adjacent to domains with reduced Ctip2+ neurons (Fig. 1b, c). Similarly, labeling for Sox5 and Cutl1, markers of neurons in layers 5–6 and 2–4, respectively, showed normal radial but irregular tangential distribution of labeled neurons in the TKO (Fig. 1d, e). Importantly, double labeling with Ctip2 and Cutl1, and Ctip2 and Lmo4 (which is expressed in neurons of layers 2–6), revealed that the irregular tangential distribution of marker-positive and -negative neuronal subtypes has a certain independency among cortical layers (Fig. 1e, f). Labeling of cell nuclei exhibited uneven CP thickness and cell packing among neighboring columnar domains (Fig. 1f, Supplementary Fig. 4). However, their increase and decrease, and those of marker-positive neurons, were not parallel, so that the thicker cortical domains with higher cell density often contained less Lmo4+ neurons in the upper cortical layers (Fig. 1f, compare Lmo4 and nuclei labeling). To gain insight into the functional organization of the TKO neocortex, corticotectal projection neurons (CTPN) and callosal projection neurons (CPN) were retrogradely labeled at P2 and observed at P4 (n = 6 per genotype). Whereas layer distribution of labeled neurons was normal, their numbers were highly variable along the tangential axis in the TKO neocortex (Fig. 1g). Tangential axis-specific disruption of cortical organization was further confirmed at P14 by combining Cutl1 immunohistochemistry with 5-bromodeoxyuridine (BrdU) injection made at E14.5 (Supplementary Fig. 4). In addition, no apparent defect was observed in the organization of the marginal zone and pial surface, Reelin expression in Cajal-Retzius cells, and the organization of radial glial fibers in the TKO neocortex at P0 (Supplementary Fig. 5).
To test the possibility that neurons in TKO mice might be allocated into unusual cortical columns, we examined the positional relationship among the clonally related CP neurons by infecting precursors with a Green Fluorescent Protein (GFP)-retrovirus at E12.5 and analyzing the location of GFP+ neurons at P0. The results show that the width of ontogenetic columns of GFP+ clones in TKO was smaller (Fig. 2a, n = 7) than that in the wild-type neocortex (Fig. 2a, n = 10). Furthermore, in the majority of the TKO CP (n = 5/7), the labeled neurons were aligned in straight columns with one to two cell diameters, which was never encountered in the wild-type CP (Fig. 2b). This straight alignment of neurons was not due to a crowding effect of increased cell packing density (Supplementary Fig. 6). Because lateral movement of radially migrating neurons occurs within the SVZ and IZ during “the multipolar stage”23, we measured the lateral distance between migrating neurons and the parental radial fibers at the top of the IZ at E14.5 in the neocortex infected with GFP retrovirus at E12.5. The distance was significantly less in TKO than that in wild-type (Fig. 2c), indicating that EphA/ephrin-A signaling is required for lateral dispersion/intermingling of neuronal clones during migration. To address the mechanism of lateral dispersion, we performed cell ablation experiments, in which the diphtheria toxin A-chain fragment (DTA) and an Enhanced Yellow Fluorescent Protein (EYFP) reporter plasmids were electroporated into the cortex at E13.5, and the effects on the migrating neurons and upper layer neurons were examined at E15.5 and at P0. As DTA causes inhibition of protein synthesis followed by apoptotic cell death24, we observed death of infected progenitor cells in the VZ and, importantly, also of the migrating neurons in the IZ, before they arrive in the CP (Fig. 2d, see also Methods in detail). With this system, we found that, despite significant numbers of dying cells in the DTA-electroporated domains, surviving NeuroD+ young neurons25 in the lower IZ were distributed continuously along the tangential axis of the cortex in the wild-type neocortex at E15.5, irrespective of the size of electroporated areas (Fig. 2d, e, n = 12). In contrast, most electroporated TKO cortices showed interrupted distribution of NeuroD+ neurons (Fig. 2d, n = 8/9), even in the brains that have very limited electroporated areas (Fig. 2e). Consistent with these observations, Cutl1-labeled upper layers was apparently thinner in the TKO cortex at P0, specifically within the domains of DTA electroporation (Fig. 2f). Finally, we observed no difference in proliferation and neuronal production in the progenitors in the electroporated domains between wild-type and TKO (data not shown). These results demonstrate that the developing neocortex possesses a striking capacity to fill in the vacated space of ablated neurons with neighbouring migrating neurons, but that this capacity is lost in the TKO neocortex.
To further examine the role of EphA/ephrin-A signaling in lateral dispersion of neocortical neurons, we co-electroporated either a control or an EphA7 expression plasmid with an EYFP reporter plasmid at E12.5, and examined the distribution of EYFP labeled neurons at E18.5. In contrast to the random distribution of EYFP+ neurons in control brains (Fig. 3a, n ≥ 20), EphA7 overexpressed neurons formed striking multiple EYFP+ columns (Fig. 3a, n ≥ 20) in all cortical regions examined (Supplementary Fig. 7). Introducing another EphA receptor, EphA4, generated the same phenotype (data not shown). The exaggerated columnar pattern of EYFP+ neurons appears one day after EphA7 electroporation, and continues during embryonic and postnatal stages (Fig. 3a, Supplementary Fig. 7, n = 7 [E13.5], n ≥ 20 [P4]). Non-labeled neurons situated between EYFP+ columns had a similar cell packing density (Fig. 3a, Supplementary Fig. 7 and 8) and the overall cytoarchitecture in the CP was unaffected as previously reported21. Organization of apical dendrites and formation of radial glial fibers also were not affected grossly by EphA7 overexpression (Supplementary Fig. 8). TUNEL and BrdU labeling at E18.5 following EphA7 electroporation showed no difference between EYFP+ and EYFP− domains (Supplementary Fig. 8). Furthermore, the cell cycle index and cell death rate displayed no change in the EphA7 electroporated neocortex (Supplementary Fig. 8). Finally, sequential electroporation with the EphA7 expression plasmid, with the EYFP (first) and DsRed2 (second) marker plasmids, demonstrated that the isolated columns of EphA7 electroporated neurons consist of multiple clones (Supplementary Fig. 7). EphA7 electroporation into the TKO cortex did not form the columnar segregation (Supplementary Fig. 8). Together, these results indicate that EYFP+ columns are not formed by increased production or death of a subset of EYFP+ clones, but rather, by tangential segregation of EYFP+ and EYFP− neurons.
Since EphAs and ephrin-As can serve as both receptors and ligands to mediate forward and reverse signaling26 and both are required for cell sorting13,15, we electroporated ephrin-A2, A5, and the truncated forms of EphA7 and EphA4, together with the EYFP plasmid to determine which signaling is involved in columnar segregation. We found segregation of EYFP+ and EYFP− neurons, but no effect on cell cycle progression and cell death, in the ephrin-A2 and A5 electroporated cortex (Supplementary Fig. 9, and data not shown). However, electroporation with the truncated EphAs, which can induce only reverse signaling as ligands, failed to induce segregation (Supplementary Fig. 9). These results indicate that forward signaling is necessary for tangential segregation, although we cannot rule out completely that reverse signaling might also be involved cooperatively.
To determine the mechanism of columnar segregation during radial migration, we analyzed the E15.5 neocortex electroporated at E12.5. No difference was observed in the random probability of gene transfer among progenitors in the VZ between control (n = 16) and EphA7 electroporated brains (n = 23)(Fig. 3b, c). In both groups, EYFP+ cells in the VZ and CP have radially elongated shapes without any evidence of tangential movement (Fig. 3a, b). In the IZ of all cases, most of the EYFP+ migrating neurons assumed a multipolar shape. They were more randomly distributed than EYFP+ progenitors in the VZ in controls (Fig. 3b, c), whereas EphA7 electroporated migrating neurons exhibited distinct segregation from EYFP− neurons in the IZ, forming EYFP+ clusters (Fig. 3b, c). Live imaging observation confirmed that this segregation is mediated by tangential movement of neurons during the multipolar stage (Supplementary Movie 1, 2). These results indicate that neurons segregate from each other due to different levels of EphA/ephrin-A signaling that they receive during the multipolar phase of migration. Columnar grouping of EphA7-overexpressing neurons of multiclonal origins, through extensive tangential movement, exhibits the opposite phenotype to that observed in TKO, in which tangential dispersion of cortical neurons is impaired and columnar arrays of clonally related neurons are formed (Fig. 2a, b). Taken together, our analyses indicate that EphA/ephrin-A signaling regulates lateral (tangential) neuronal dispersion and intermingling during the multipolar stage of radial migration, and that this mechanism is required to generate cortical columns with appropriate cellular, molecular and anatomical components (see also Supplementary Discussion). Given these results, what is its biological significance? It is well established that progenitor cells have heterogeneous molecular properties, proliferative capacity and lineage commitment 27,28,29. We found that in wild-type, but not in TKO, an adjustment of neuron numbers, and an appropriate intermixing of neuronal subtypes can be achieved by lateral dispersion/intermixing of migrating neurons (Supplementary Fig. 10). These adaptive processes may be essential to achieve the modular function of cortical columns1,8,12. Another possibility is that, given the preferential development of specific microcircuits within radially arrayed ontogenetic clones9, appropriate radial and tangential distribution of clonally related neurons may be essential for proper synaptic development and function of the radial units. The new role of EphA/ephrin-A signaling in assuring the homogeneous and continuous aspects of the neocortex demonstrated here contrasts with its previously known roles in restriction of cell movement and inhibiting cell intermingling during formation of boundaries in other systems13,15,16. Although inappropriate neuronal positioning and abnormal columnar organization has been reported in post-mortem analysis of cortical tissue from subjects with psychiatric disorders, the molecular mechanisms of these abnormalities remain unknown. Our finding provides a novel perspective for investigating typical and atypical development that underlies higher order information processing in normal and pathophysiological states.
Generation/genotyping of ephrin-A TKO was performed as described18. Wild-type C57/Bl6 and CD-1 mice were used as controls. Histology, imunohistochemistry and in situ hybridization. were performed on cryosections or vibratome sections as described previously21. Radial glial cells were labeled by injecting saturated DiI (1,1'-dioctadecyl- 3,3,3',3'-tetramethylindocarbocyanine perchlorate, Invitrogen) solution into the lateral ventricle of fixed P0 brains. CTPN and CPN were retrogradely labeled by injection of FluoroGold (Invitrogen) into the superior colliculus and red fluorescent microspheres (FluoSperes, Invitrogen) into the contralateral parietal associative/visual cortex, respectively. GFP retrovirus (a gift from F.H. Gage) was injected in utero into the lateral ventricle of the embryo at E12.5. The distance between migrating neurons and their parental radial glia at E14.5 was quantified in z-stacked confocal images using LSM Image Bowser. In utero electroporation was performed at E12.5 – E15.5. Each EphA and ephrin-A expression plasmid was injected with pCAG-EYFP or pCAGGS-DsRed2. DTA expression plasmid was injected with pCAG-EYFP. The cortical area of each hemisphere was quantified using ImageJ software. Total numbers of cortical cells were estimated using the Isotropic Fractionator method. The thickness of Cutl1+ cortical layers was measured using ImageJ on coronal sections at the center of electroporated domains and the corresponding region in the contralateral hemisphere, and the ratios were calculated. Pregnant mice (E13.5) were labeled with BrdU for 30 min and 24 h, respectively, and labeling index (proportion of BrdU+ cells among Ki67+ cells) and cell cycle exit index (proportion of Ki67− cells among BrdU+ cells) were calculated. Apoptotic cell death was analyzed by counting TUNEL+ cells in 10,000 µm2 in the neocortex. For quantification of the pixel intensity along the tangential axis of the neocortex, the regions were delimited from medial to lateral in the neocortex, straightened with ImageJ, and the plot profiles were made.
All experiments using animals were in accordance with the protocols approved by Yale University Institutional Animal Care and Use Committee. The Ephrin-A2/A3/A5 triple knockout mouse was generated and genotyped as previously described18,30,31,32. Wild-type mice were from the C57/Bl6 strain (Jackson). Wild-type CD-1 mice (Charles River) were also used in EphA/ephrin-A overexpression analysis. The day on which a vaginal plug was observed was designated as embryonic day 0.5.
In situ hybridization cryosections or whole-mount was performed as described33,34. The probes for EphA5, EphA7, ephrin-A5 and cadherin-8 are described previously21. The probes for Id2, Lmo3 and Lmo4 are as described35 (gifts from J.L.R. Rubenstein). The probes for EphA4 (1014 bp mouse partial clone), ephrni-A2 (mouse full-length clone) and ephrin-A3 (~200 bp rat partial clone) were gifts from L. F. Kromer.
Brains were fixed with 4% paraformaldehyde (PFA) in phosphate buffered saline overnight, and 60–70 µm coronal vibratome slices or 20 µm coronal cryosections were collected. Immunohistochemistry was performed as described previously21. The following primary antibodies were used: polyclonal anti-GFP (also recognize EYFP, 1:3000; Invitrogen, 1:250; Abcam), anti-DsRed (1:5000; BD Biosciences), anti-Cutl1 (1:500; Santa Cruz Biotechnology), anti-Sox5 (1:5000; Aviva Systems Biology), anti-Lmo4 (1:4000; Abcam), anti-Laminin (1:1000; Sigma-Aldrich), anti-NeuroD (1:300; Santa Cruz Biotechnology), and monoclonal anti-Ctip2 (1:700; Abcam), anti-Ki67 (1:100; Neomarker), anti-BrdU (1:100 Beckton and Dickinson), anti-RC2 (1:5; DSHB), anti-Reelin (1:1000; Millipore), anti-MAP2 (1:200; Sigma-Aldrich) antibodies. Sections were nuclear counterstained with TO-PRO3 or 4’-6-diamidino-2-phenylindole (DAPI, Invitrogen) when necessary. Cresyl violet staining was performed following a standard protocol. All images were captured using a confocal LSM 510 NLO system or an Axioplan2 microscope (Carl Zeiss) equipped with epifluorescence.
Ligand and receptor binding histochemistry on were performed as described21 using recombinant human ephrin-A5/Fc and EphA7/Fc chimeric protein (R&D Systems), respectively.
To label radial glial cells and their endfeet, saturated DiI solution (in ethanol) was injected into the lateral ventricle of P0 wild-type and TKO brains fixed with 4% PFA. After 48 hr of incubation at 37 °C, coronal vibratome slices were made at 100 µm thickness and photographed. CTPN and CPN in visual cortex were retrogradely labeled by pressure-injection of Fluorogold (Invitrogen) into the superior colliculus and red fluorescent microspheres (FluoSperes, Invitrogen) into the contralateral parietal associative/visual cortex, respectively, of P2 pups (to minimize the effect of activity-dependent reorganization) . Coronal cryosections were made at P4 and images were captured using an Axioplan2 microscope (Carl Zeiss).
GFP-retrovirus (a gift from F.H. Gage) was produced and injected in utero into E12.5 embryos as previously described36. Brains were collected at E14.5 or P0, fixed in 4% PFA for overnight and thick vibratome sections (70 µm) were made for GFP immunohistochemistry. To quantify the distance between migrating neurons and their parental radial glia, z-stack images for each slice were taken using a confocal LSM 510 NLO system (Carl Zeiss), and the distance was measured using LSM Image Browser (Carl Zeiss). The investigator was blind to the genotype for analysis.
EphA7 expression plasmid (pCAGGS-EphA7-IRES-Venus), control plasmid (pCAGGS-IRES-EGFP), and pCAGGS-DsRed2 were described previously21. pCAG-EYFP and pCAG-DTA were provided by T. Saito37 and K. Kohno38, respectively. Full length cDNA of mouse EphA4 (a gift from N.Y. Ip), ephrin-A2, and ephrin-A5 (gifts from L.F. Kromer) were inserted into pCAGGS-IRES-EGFP as their expression plasmids. A truncated form of EphA4 (EphA4-T) lacking its intracellular domain (at Val597) was designed as previously described39, and EphA7 truncated form (EphA7-T) was similarly designed (truncated at Ile609). cDNA fragments were amplified by Polymerase Chain Reactions using full-length expression plasmids as templates, and the following primers: EphA4-T 5’-AGGAGCAGCGTTGGCACC-3’ (forward) and 5’-CACCTAAGTTCTAACACCTTGATT-3’ (reverse), EphA7-T 5’-CCATGGTTGTTCAAACTCGGTA-3’ (forward), and 5’-AATCTAGGTTTTGGTGCCTGGA-3’ (reverse), and the fragments were inserted into pCAGGS-IRES-EGFP.
In utero electroporation was performed as previously described37 at E12.5 – E15.5. Each EphA and ephrin-A expression plasmid (2 mg/ml) was injected with pCAG-EYFP (0.5 mg/ml) or pCAGGS-DsRed2 (1 mg/ml). In some cases, BrdU (50 mg/kg body weight) was intraperitoneally injected immediately or 24 hrs after surgery for cell proliferation and cell cycle exit analyses as described below. DTA expression plasmid (1 mg/ml) was injected with pCAG-EYFP (2 mg/ml). We observed a massive increase of TUNEL+ cells in the electroporated domains of both wild-type and TKO within 24 hr (n = 3 and 2, respectively, data not shown). This striking increase of dying cells was restricted to the VZ and IZ, but was not present in the CP at either E14.5 (data not shown) or E15.5 (Fig.2d, n = 6 [wild-type] and 4 [TKO]). However, the increase of cell death was no longer evident by P0 (data not shown). Whole brain images were captured using Stemi SV 11 Apo stereomicroscope (Carl Zeiss) equipped with fluorescent source.
Cortical slices were prepared as previously described40 from the E15.5 embryos electroporated at E13.5 (n = 6 each for control- and EphA7-electroporation). The slices were transferred to a RC25 recording chamber (Warner Instruments) maintained at 36°C and perfused with fresh culture media. Time-lapse imaging of EYFP-labeled cells around the VZ-IZ was performed following the previously described procedure41. Time series were collected every 15 min for ~15 h per slice through 20× objective on a confocal LSM 510 NLO system fitted with a motorized stage, and edited with Adobe Premier Pro.
Investigator was blind to the genotype or electroporated gene. The cortical area of each hemisphere was quantified using ImageJ from images of the whole brain taken as dorsal view using Stemi SV 11 Apo stereomicroscope (Carl Zeiss), and averaged for wild-type and TKO. Total brain mass was weighed after fixation in 4% PFA overnight, and averaged for each genotype. The thickness of Cutl1+ cortical layers was measured using ImageJ on coronal sections at the center of electroporated domains and the relative thickness was calculated against the corresponding region in the non-electroporated hemisphere, and averaged for each genotype. A few sections that showed an apparent TKO phenotype of abnormal cortical thickness around the domain of measurement on either hemisphere were excluded from quantification. For proliferation and cell cycle exit analyses, pregnant mice were injected intraperitoneally with BrdU (50 mg/kg body weight) at E13.5, and embryos were collected after 30 min and 24 h, respectively. Brain sections were immunostained for Ki67 and BrdU, and BrdU labeling index (proportion of BrdU+ cells among Ki67+ cells) and cell cycle exit index (proportion of Ki67− cells among BrdU+ cells) were calculated from the counted cell numbers as previously described42. For quantification of apoptotic cell death, TUNEL assay was performed using ApopTag in situ apoptosis detection kit (Millipore) following the manufacture’s protocol, and TUNEL+ cells in 10,000 µm2 was calculated from counted cell numbers throughout the neocortex.
The fluorescent pixel intensities of EphA7/ephrin-A5-Fc binding histochemistry in the IZ (Supplementary Fig. 1), Ctip2 immunohistochemistry in the layer 5 (Fig. 1), and EYFP+ electroporated cells in the VZ and IZ (Fig. 3), along the medial-lateral axis of the neocortex, were quantified using a previously described method43 with modification. Regions for quantification were delimited from medial to lateral in the neocortex, straightened with ImageJ software (‘straighten’ plugin), and the plot profiles of the fluorescent intensity was made as gray values (y) against the tangential distance (x). To present the distribution of EphA/ephrin-A-Fc binding and EYFP+ electroporated cells, the pixel intensity was normalized and expressed as a percentage to the total value measured inside the region of interest. Pixel intensity of the Ctip2 labeling was not normalized between the wild-type and TKO cortex for comparison of their total values. Tangential distance was subgrouped into ~30 bins for better presentation for EphA/ephrin-A-Fc binding and Ctip2 labeling.
Because of the highly anisotropic cerebral cortex of TKO, total numbers of cells in a cortical hemisphere were estimated using the Isotropic Fractionator method as described44 instead of using stereological method. Briefly, a cortical hemisphere was dissected from each wild-type and TKO brain at P0 and a suspension of nuclei was obtained through mechanical dissociation in 40 mM sodium citrate and 1% Triton X-100 using a tissue homogenizer. The homogenate was collected by centrifugation and suspended in PBS containing 1% DAPI, and the nuclear density was determined using a hemocytometer under fluorescence microscope to estimate the total cell number in the original tissue. For each genotype, 6 pups from 3 dams were used. The coefficients of variation (CV) were < 0.1 for both wild-type and TKO, indicating the reliability of the method. The investigator was blind to the genotype for analysis.
We are grateful to Drs. T. Cutforth, D.A. Feldheim, J.G. Flanagan, J. Frisen, F.H. Gage, N.Y. Ip, K. Kohno, L.F. Kromer, C. Redies, J.L.R. Rubenstein and T. Saito for providing materials. We also thank M.R. Sarkisian and A. Bonnin for helpful comments, M. Pappy, J. Bao, C. Anderson and S. Ellis for technical assistance. This work was supported by the NARSAD Young Investigator Award (M.T.), the Kavli Institute for Neuroscience at Yale (P.R.) and the National Institute of Health (P.L. and P.R.)
Supplementary Information accompanies the paper on www.nature.com/nature.
Author ContributionsM.T. initiated the project, conducted experiments, analyzed the data, and wrote the manuscript. K.H-T. conducted experiments, analyzed the data, and helped to write the manuscript. P.L. and P.R. contributed to the interpretation of results and writing of the manuscript.
Reprints and permissions information is available at npg.nature.com/reprintsandpermissions