Main

The six-layered neocortex is formed by sequential waves of postmitotic neurons that migrate radially from the ventricular zone (VZ) to the pial surface. The first wave establishes a subpial layer termed the preplate zone (PPZ), which the second neuronal wave splits into a superficial marginal zone (MZ) and a deeper subplate (SP), forming a cortical plate (CP) in-between. Thereafter, successive waves of neurons expand the CP by migrating through a cell-sparse intermediate zone (IZ) that separates the SP from the VZ, and they migrate past their neuronal predecessors to arrive directly below the MZ. Hence, neocortical layers are generated in an 'inside-out' manner, with early- and later-migrating neurons forming the deep and superficial CP layers, respectively1.

Radial migration in the neocortex occurs primarily by two distinct migration modes, locomotion or translocation. Neurons moving by locomotion2,3,4 migrate along radial glial fibers2,3,4,5,6, and their cell body and leading edge move in unison to maintain constant length. Neurons moving by translocation, originally called nuclear translocation7,8,9,10,11,12,13,14,15,16 but in the neocortex more recently referred to as somal translocation17, propel their cell somata toward a leading edge that is stably attached to the pial surface and MZ, which effectively shortens neuronal length. Imaging of the embryonic neocortex17 shows that locomotion and translocation peak at different developmental times: early on, around PPZ-splitting, neurons move radially by translocation, whereas later on, when inside-out layering is underway, they first locomote across the IZ into the CP17,18 before translocating to the top CP layer. To date, several signaling pathways have been identified to crucially regulate neocortical layering, including Reelin/VLDLR/ApoER2/Dab1, Cdk5/p35/p39, Lis1, Dlx and filamin-1 (refs. 19–25). Although these pathways affect the early and late phases of neocortical development distinctly, little is known about how they relate to the different migration modes.

p35 is a regulatory activator of Cdk5. In mice deficient for p35 (refs. 26,27) or Cdk5 (refs. 28–30), the PPZ splits properly, but subsequent waves of migrating neurons fail to move past their predecessors, resulting in a CP with inverted layering. Despite the ample molecular evidence that implicates p35/Cdk5 in regulating neuronal function31,32,33,34,35, it remains unclear how Cdk5 signaling specifically influences the cellular physiology of neuronal migration behavior. To address this issue in the neocortex, we performed comparative time-lapse imaging on acute cortical slices of p35 null and wild-type mice.

Here we report that p35−/− neurons migrate by 'branched migration'. This distinct mode of radial migration, characterized by extensive branching, is cell-autonomous and noticeably associated with altered glial guidance. Considering that wild-type neurons locomote during inside-out layering, these results suggest that proper glia-guided migration is important for this process and requires p35/Cdk5 function. Furthermore, as branched migration was not observed in defective Reelin signaling, which also shows layer inversion, p35/Cdk5 and Reelin signaling seem to contribute distinctly to inside-out layering.

Results

Branched migration occurs early in p35 deficiency

p35−/− neurons presumably migrate normally at embryonic day 13 (E13), when the PPZ splits27. To assess whether this assumption is true, we imaged acute E13 cortical slices of wild-type and p35 null mice. In both genotypes, the dorsal aspect of the anterior neocortex was imaged, as the inverted layering phenotype of the p35−/− neocortex is manifested in a graded manner, decreasing from anterior to posterior and from dorsomedial to ventrolateral26.

Consistent with a previous report17, we found that 90% of E13 wild-type neurons (27 of 30 neurons selected randomly from six cortical slices) migrated in the radial direction by somal translocation, which occurred in an undisrupted, straight trajectory (Fig. 1a). By contrast, such migration behavior was not seen in the E13 p35−/− neocortex. Rather, p35−/− neurons showed extensive branching morphology (Fig. 1b and Supplementary Fig. 1a online), which was confirmed by silver impregnations of p35−/− cortical sections (data not shown) and green fluorescent protein (GFP) labeling of migrating neurons in utero (Supplementary Fig. 1b online and Figs. 3d and 4b). The extensive branching made p35-deficient neurons move in a manner that we refer to as branched migration, occurring with a frequency of 83% (25 of 30 radially migrating neurons selected randomly from six cortical slices). Neurons moved by propelling their cell somata along the leading process toward a stably fixed branch point (Fig. 1b and Supplementary Video 1 online).

Figure 1: Migration behavior at E13.
figure 1

(a) Somal translocation of a wild-type neuron. The leading process (blue asterisk) is attached to the upper preplate zone (PPZ) at all times, allowing the cell soma (yellow arrowhead) to translocate from halfway within the cortex into the PPZ, effectively shortening neuronal length. (b) Branched migration of a p35−/− neuron. The leading process is branched, with the leading edges (asterisks) not attached to the PPZ. The cell soma (arrowhead) propels towards the stably positioned branch point (arrow). Significant radial advancement occurs when the cell soma moves through two successive branch points (II–III). Right (III-B), 2× magnification shows branching morphology. (c) Quantification of the zig-zag trajectory of radially migrating p35−/− neurons. Depicted are percentages of wild-type (WT) and p35−/− neurons that change the direction of migration by an angle larger than 30° per hour. Whereas WT neurons adhere to a strictly radial path, p35−/− neurons change direction during radial migration, reflecting their zig-zag trajectory (n, number of neurons). (d) Splitting the p35−/− PPZ involves transition from branched migration to somal translocation. Top, after exiting the VZ by branch migration, with the branches contacting the developing subplate (I,II), the cell soma then enters the PPZ by somal translocation, moving along an unbranched process (III–V) from a previous branch point to the developing marginal zone. Right (II-B), 2× magnification shows branching. Yellow arrowhead, cell soma; white arrow, branch point; white arrowhead (I,II) and blue arrow (II-B), branches; arrowhead (III–V), unbranched leading process; asterisk, leading edge. Bottom, tracings emphasize branch points and branches of the leading process.

Full size image
Figure 3: Relationship between migrating neurons and radial glia.
figure 3

(a) Normal migration behavior of clonal descendents. Left, schematic drawing of a wild-type neuronal clone. Right, normal clonal migration behavior in the posterior p35−/− neocortex. Note the switch to glia-independent migration near the pia. (b) Possible relationships between p35−/− neurons and radial glia during branched migration in the radial direction. (c) Branched neurons in the anterior p35−/− neocortex do not associate with their mother glia. The distance of locomoting wild-type (WT) neurons and branched p35−/− neurons from their mother glia is plotted against their distance from the ventricular surface. Whereas WT neurons maintain close contact with their mother glia (top), p35−/− neurons detach early on and stay detached throughout their radial journey (bottom). (d) Representative examples of three migrating neurons in the anterior p35−/− neocortex with branched leading processes (I–III). All three cell bodies are not associated with the mother glia (cell body in III is in a different z-plane than the radial glia, data not shown). Magnifications of I and III emphasize that individual branches are separated from the mother glia fiber. Magnification of II provides a better view of the branching. Large arrow, cell soma; small arrow, leading process branch; arrowhead, radial glia fiber; CP, cortical plate; IZ, intermediate zone; MZ, marginal zone; SP, subplate; VZ, ventricular zone; n = number of cells from eight different embryos.

Full size image
Figure 4: Cell-autonomous functions of p35.
figure 4

(a) Electroporation into the anterior dorsal neocortex of a wild-type mouse. Unbranched migrating neurons are seen at various radial positions (arrowheads indicate cells that are magnified). (b,c) Branched migration is cell-autonomous. Labeled neurons express (b) GFP-alone or (c) GFP and p35 in the p35-deficient neocortex. Left, overview of a representative field within the anterior dorsal neocortex. Labeled neurons (a–c) are seen in various focal planes with different positions along the radial axis. Right, higher magnifications of individual focal planes. (d) Quantification of the rescue of branching. From experiments like in b and c, the leading process morphology of randomly selected p35−/− neurons was assessed for control (GFP alone) and rescue (GFP + p35) groups. Percentages of unbranched neurons are presented for each group (highly significant difference: P < 0.0001; two-tailed, unpaired student t-test). (e) Radially migrating neurons expressing p35 protein in the anterior p35−/− neocortex move farther toward the cortical plate. The position in relation to the radial axis of the neocortex was assessed for 220 knockout neurons (GFP alone) and 113 rescued neurons (GFP and p35) in six neocortical p35−/− sections. Percentage of labeled neurons is graphed against neocortical regions. Data shown as mean ± s.e.m. In the wild-type neocortex, 91% of neurons were located in the cortical plate under the same experimental conditions (data not shown). CP, cortical plate; IZ, intermediate zone; arrowhead, cell soma; arrow, branch point; asterisk, leading edge; n = number of cells.

Full size image

As the branch points of the migrating p35−/− neurons were found at various locations below the PPZ, radial movement seemed to occur in a disrupted multi-step fashion, distinct from the defined one-step fashion seen in the wild-type neocortex. Indeed, when p35−/− neurons were followed individually over longer periods of time, they traversed much of the neocortex by advancing their cell somata from one branch point to the next (Fig. 1b). Two features of this 'branch-to-branch' migration were striking. First, branching was very dynamic, indicating that the branches change their length and direction very rapidly, until they are finally locked into position (see Supplementary Fig. 2 online). Second, although branched migration allowed the p35−/− neurons to move in an overall radial direction, the trajectory of the actual movement was more of a 'zig-zag' quality (Fig. 1c and Supplementary Fig. 3 online).

In agreement with proper PPZ splitting, p35-deficient neurons were seen to migrate all the way to the developing MZ. However, branched migration was not used for the entire radial journey, but rather until the PPZ was reached. To split the PPZ, p35−/− neurons moved along unbranched processes that were attached to the developing MZ (Fig. 1d), making the final ascent appear indistinguishable from somal translocation (compare Fig. 1a and 1d right). Together, the results indicate that normal layering in the E13 p35−/− neocortex is not associated with a normal pattern of radial migration. Instead, neurons cross most of the neocortex by branched migration before splitting the PPZ by somal translocation.

Branched migration replaces locomotion in p35 deficiency

We next conducted our imaging analysis at E15, when inverted layering is evident in the p35−/− neocortex, and migrating neurons actively contribute to the progression of this defect. We reasoned that imaging at E15 should capture defects in migration behavior that are crucial to the phenotype of inverted layering.

Consistent with previous reports18 that locomotion accounts for 60–80% of total radial migration at E15, we found that 66% of E15 wild-type neurons used locomotion as their mode of radial migration (20 of 30 radially moving neurons selected at random from ten cortical slices). In the lower part of the neocortex, locomotion was present at 86% (18 of 21 neurons randomly chosen from ten cortical slices), emphasizing that it is the principal mode of radial migration across the IZ at later developmental stages17 (Fig. 2a and Supplementary Video 2 online). Hence, our imaging procedure recapitulates not only the migration patterns but also the predominance of locomotion previously observed in the E15 wildtype neocortex.

Figure 2: Migration behavior at E15.
figure 2

(a) E15 wild-type neurons locomote in tandem. Left, a neuron locomotes radially, maintaining a constant neuronal length, as cell soma (arrowhead) and leading edge (arrow) move coordinately. After some time, a second neuron (green arrowhead) appears, which locomotes along the radial path of the first neuron (right), with its leading edge (green arrow) tucked behind the cell soma of the first one. (b) Branched migration in the E15 p35−/− neocortex. Two neurons (white and blue arrows) migrate by branched migration across the lower IZ, to a location where wild-type neurons would usually locomote. In both cases, cell somata (arrowheads) approach a branch point (arrows) stably positioned far below the CP. The leading edges of the process branches (asterisks) are not attached to the marginal zone. (c) Branch-to-branch migration. Top, time-lapse sequence showing a neuron in the IZ. After dynamic changes of the leading branches (compare position of asterisks in I and II), the cell soma (arrowhead) moves through two consecutive branch points (arrows in II–V), within a change of 35° in the radial trajectory. Bottom, camera lucida–like drawings outline the morphology of the neuronal branches. CP, cortical plate; IZ, intermediate zone; VZ, ventricular zone.

Full size image

The pattern of radial migration in the p35-deficient neocortex was markedly different. Neurons did not move by locomotion, but by branched migration to effectively traverse the IZ in the radial direction (Fig. 2b and Supplementary Video 3 online). Indeed, branched migration was the principal mode of radial migration at E15, contributing with 93% (24 of 26 cells) to the overall radial migration, with the residual 7% (2 of 26 cells) remaining inconclusive. When followed over longer time periods, individual neurons showed a migration behavior very much like that observed at E13: considerable dynamic branching (Fig. 2c) and 'branch-to-branch' migration in a zig-zag pattern (Fig. 2c and Supplementary Fig. 4 online). Furthermore, the branches of the leading processes never reached the pial surface, as branch points were distributed almost exclusively within the IZ (95%), with the remainder positioned in the lower CP (5%).

Branched migration associates with altered glial guidance

Two observations from the imaging study suggest that branched migration might be a glia-independent migration. First, glia-guided locomoting cells always possessed an unbranched leading process (Fig. 2a and Supplementary Video 2 online). Second, unlike the unwavering radial path of locomoting cells (Fig. 2a), branched migration had a zig-zag trajectory (Figs. 1c and 2c and Supplementary Fig. 3 online). For mechanistic evidence of this idea, we made use of the finding that a wild-type radial glia cell (RGC) gives rise to neurons that migrate along their mother glial process36 (Fig. 3a, left). We reasoned that investigating the migration behavior of clonal descendents in the p35−/− neocortex should disclose a possible loss of normal neuronal-glial interaction associated with branched migration. p35−/− neurons could either use one of their branches to move along glia or, alternatively, move independently of glia altogether (Fig. 3b).

As retrovirus selectively targets precursor cells36,37,38, we labeled individual neuronal clones by introducing a GFP-carrying retrovirus in utero into E12 p35−/− embryos, and then we assessed the clonal relationship at E15 by GFP-immunostaining. In the posterior neocortex (Fig. 3a, right), which is barely affected by the p35 deficiency26,27, the clonal relationship closely resembled that of the wild-type neocortex36. As with locomoting wild-type cells (Fig. 3c, top), clonal descendents of the posterior p35−/− neocortex traversed the IZ in the radial direction along the glial processes of their respective mother cells (RGCs), demonstrating elongated cell somata and unbranched leading processes. Only near the pia did neurons switch to translocation, with their leading processes occasionally arborized to extensively contact the MZ17 (Fig. 3a, right).

The anterior neocortex, the primary area affected by the p35 deficiency, showed an entirely different picture. Migrating p35−/− neurons showed extensive branching across the IZ, and these neurons were not associated with the glial processes of their mother cells (Fig. 3c, bottom). The average distance between their cell bodies and the mother glia was 20.6 μm (3.5 cell diameters), with the branches not aligned or attached to the glial process (Fig. 3d). Because we detected branched neurons as early as in the subventricular zone (Fig. 3c), p35−/− neurons appeared to detach very early from their mother cell to migrate as branched cells in the radial direction. In sum, these in vivo results suggest that branched neurons have a compromised neuronal-glial interaction, supporting the deductions made from our imaging studies.

Branched migration is cell-autonomous in p35 deficiency

To investigate whether branching is cell-autonomous, we electroporated p35 into the p35-deficient neocortex and assessed whether branching could be rescued in those neurons that expressed p35. In electroporation, plasmid DNA is injected in utero into the ventricle of embryonic brains and taken up by ventricle-lining cells after exposure to an electrical field39 (Fig. 4a). By using plasmid DNA, we avoided potential toxic effects associated with virus-based methods.

We electroporated p35−/− mice at E13, either with GFP alone or with GFP and p35, and assessed neuronal morphology by GFP-immunostaining at E17. Using an IRES-based plasmid allowed us to detect p35 expression through the GFP label. Just as seen in the imaging experiment, p35−/− neurons showed extensive branching of their leading process when expressing GFP alone (Fig. 4b). Notably, however, when GFP and p35 were applied together, GFP-labeled neurons showed unbranched leading processes (Fig. 4c), which made them look like locomoting wildtype neurons. This rescue occurred independently of where the migrating neuron was positioned along the radial axis of the neocortex (Fig. 4c) and was highly significant (93%; Fig. 4d). Thus, we conclude that branching is a cell-autonomous feature of the migrating neuron.

We further observed that p35-expressing neurons were positionally shifted toward the pial surface in comparison to GFP-expressing neurons alone (Fig. 4e). This shift was most pronounced within the IZ, where the majority of p35-expressing neurons were concentrated in the upper IZ. In contrast, many more p35-deficient neurons were located in the lower IZ. Within the CP, the proportion of neurons was similar between the two groups, reflecting the fact that early-born neurons successfully split the PPZ and that rescue of cell positioning was partial. Considering that embryos were killed 4 d after electroporation, p35-expressing neurons may have eventually settled within the CP, given sufficient time. Hence, cell positioning in the p35−/− neocortex may also have a cell-autonomous component.

Branched migration is rare in the scrambler mouse

Finally, we investigated whether branched migration is restricted to p35 deficiency, given that numerous other mouse mutants harbor distinct neocortical layering defects. We were particularly interested in mice with a defective Reelin signaling pathway19,20,21,22, which includes the extracellular matrix protein Reelin, the Reelin receptors ApoER2 and VLDLR, and Dab1, an adaptor protein that binds the Reelin receptors intracellularly40,41,42,43,44,45,46. In mice deficient for Reelin, Dab1, or both ApoER2 and VLDLR, neocortical layering is approximately inverted, as it is in the p35 deficiency23,24,25.

To evaluate migration behavior resulting from defective Reelin signaling, we imaged the scrambler neocortex, which is Dab1-deficient45,46. We imaged at E15 when, just like in p35 deficiency, layer inversion is very pronounced. In contrast to p35−/− neurons, most Dab1-defective neurons did not show any branching morphology and moved radially by locomotion (Fig. 5a), in a proportion indistinguishable from that observed in the wild-type neocortex of identical strain background (Fig. 5b). To confirm that the locomoting neurons seen in our video analysis were attached to radial glia, we conducted a clonal analysis in the E15 scrambler neocortex, using the same conditions as in the p35−/− neocortex. Migrating scrambler neurons indeed moved along the glial process of their mother cell (Fig. 5c), indicating that branched migration is rarely detected in the scrambler mouse at later stages of neocortical development. Therefore, branched migration seems to reflect a defect in p35/Cdk5 but not Reelin signaling.

Figure 5: Migration behavior of scrambler neurons.
figure 5

(a) Branched migration is not evident in the E15 scrambler neocortex. The representative neuron crosses the intermediate zone (IZ) by locomotion: unlike p35−/− neurons, scrambler neurons have unbranched leading processes. Arrowhead, cell soma; asterisk, leading edge. (b) Scrambler and wild-type (WT) neocortices show similar proportions of unbranched migrating neurons at E15. Migrating WT and scrambler neurons were randomly selected from anterior neocortical sections 2 d after infection with a GFP-carrying retrovirus. For each category, three different embryos were assessed and results plotted as percentage of unbranched neurons (n = number of cells). (c) Scrambler neurons use glial guides in the E15 neocortex. The representative neuron shown (box) is attached with its cell soma and unbranched leading process (yellow arrowheads) to the glial process of its mother cell (blue arrowheads).

Full size image

Discussion

Somal translocation and locomotion peak during PPZ-splitting and inside-out migration, respectively17. In p35-null mice, PPZ-splitting is preserved, whereas the CP is inverted26,27. This raises the possibility that p35−/− neurons can translocate but not locomote, with the disruption of locomotion contributing to inverted layering. However, we found that in p35 deficiency, both migration modes are affected, as neurons move by branched migration, a distinct migration mode. Here we discuss the precise nature of this defect and the insight it offers into understanding neocortical layering.

Branched migration prevails in the p35−/− neocortex

The alteration of translocation and locomotion in p35 deficiency shows that both modes of radial migration, although distinct in their phenotype and migratory behavior, are not regulated by distinct signaling pathways. Rather, they share common signaling components that may regulate them differentially. Moreover, as branched migration typically accompanied PPZ-splitting in p35−/− mice, abnormal migration behavior does not necessarily affect neocortical structure. At E15, however, the abundance of branched migration (Fig. 2b,c) does suggest that altered migration behavior may relate to inverted layering in p35 deficiency. As branched migration replaces locomotion at that time, compromised locomotion is tied to defective inside-out layering, with branched migration inadequately compensating for this compromise.

Branch-to-branch migration (Figs. 1b and 2c) seems to be a mechanism for maximizing the advancement of p35−/− neurons in the radial direction. The dynamic nature of branching (Fig. 2c and Supplementary Video 3 online) supports this idea, suggesting that migrating neurons explore their environment to proceed as far as possible. Without branch-to-branch migration, the lower part of the p35−/− neocortex might be filled with permanently and randomly arrested neurons that fail to split the PPZ or form a layered CP. However, advancement by this migration may also come at a cost, as its zig-zag trajectory may contribute to the delayed net movement seen in p35 deficiency26,27.

At E15, we also observed branched migration in radially migrating wild-type neurons, but only in 7% of cases. This concurs with recent evidence that links wild-type branched migration to radially migra-ting interneurons47, which are known to migrate independently of Cdk5 activity30. We confirmed by calbindin staining that branched neurons in the p35 null mice are principally not interneurons (data not shown). If branched migration of p35−/− projection neurons is analogous to that of wild-type interneurons, projection neurons may possess the basic machinery for radial migration, but may require p35/Cdk5 to translocate or locomote at the appropriate developmental stage. p35/Cdk5 may actively suppress branching, indicating that establishing an unbranched process for translocation or locomotion may have an inhibitory component.

Branched migration, abnormal glial guidance and cell autonomy

Our in vivo data showing that branched neurons do not use their mother glia (Fig. 3) offers a cellular mechanism for inverted layering in p35 deficiency: the failure to use glial guides may account for the inability to migrate past neuronal predecessors. Although branched neurons could use non-mother glial guides, this possibility seems less likely. First, wild-type neurons principally use their mother glia to migrate in the radial direction36 (our retroviral studies). Second, in the absence of any known functional or structural differences between neighboring glia, branched neurons are unlikely to selectively use non-mother glial guides, when unable to move along mother glia. Hence, our clonal analysis suggests that inside-out layering may depend on proper neuronal-glial interaction, proposing an explanation for why the loss of glia-guided locomotion cannot be entirely compensated for by branched migration and why locomotion emerges with inside-out migration in normal development. In support of this idea, positionally shifted neurons in our rescue experiment (Fig. 4e) resembled locomoting neurons.

Cell-autonomy of branching (Fig. 4c,d) indicates that branching results from a p35 deficiency in migrating neurons and not from compromised neuronal-neuronal or neuronal-glial interactions requiring Cdk5 activity in surrounding neurons. Further, it shows that branching is caused by a p35 deficiency that manifests itself at the point of development studied and not by a secondary effect arising from a p35 deficiency exerted earlier in development. The fact that the rescue effect was not absolute suggests that some p35-electroporated neurons were radially migrating interneurons or were not expressing p35 protein at the appropriate level and time.

Distinct Cdk5 and Reelin signaling in inside-out layering

As E15 scrambler neurons, unlike p35−/− neurons, move by locomotion (Fig. 5), glia-guided migration is necessary but not sufficient for inside-out layering. The mechanisms of inverted layering are different in the p35/Cdk5 and Dab1 deficiencies, suggesting that p35/Cdk5 and Reelin functions are arranged in parallel, rather than serial, signaling pathways to regulate distinct aspects of inside-out layering. Migrating neurons deficient in Reelin signaling appear to be unable to detach from radial glia48. Curiously, this stands at the opposite spectrum of our result in p35 deficiency, but at the same time has in common with it a principal compromise in proper neuronal-glial interaction. Our cellular results fit well with the increasing molecular evidence that supports a parallel relationship between p35/Cdk5 and Reelin signaling49,50.

Reconciling migration behavior and structural development

Why is branched migration compatible with PPZ-splitting, but not inside-out migration? PPZ-splitting requires a proper response of migrating neurons to Reelin, which is secreted from the developing MZ to presumably act locally20,21,23. It seems plausible that p35−/− neurons retain proper responsiveness to Reelin, after reaching its vicinity by branched migration. This notion fits with a parallel arrangement of the Cdk5 and Reelin signaling pathways49,50 (Fig. 5) and with our result that p35−/− neurons enter the PPZ by somal translocation (Fig. 1e). Hence, PPZ-splitting may occur regardless of how neurons reach Reelin, as long as they respond to it, with somal translocation being part of this response.

Conversely, inside-out migration appears to be incompatible with branched migration. When the neocortex greatly expands and glia-guided migration prevails, compromised guidance along the mother glia may prevent migration past neuronal predecessors, terminating radial migration at significant distance from Reelin protein. Substrates for branched migration may reside on predecessor neurons or on axonal fiber tracts that aberrantly extend through the p35/Cdk5-compromised neocortex27,29, with sequential attachment to them perhaps promoting branch-to-branch migration. However, the inability to reach Reelin eventually prevents proper positioning underneath the MZ.

Methods

Imaging of acute brain slices.

p35 null mice were compared to wild-type mice of equivalent strain background26. Studies were done in accordance with the guidelines of the Animal Studies Committee at Harvard Medical School and the National Institutes of Health. Acute brain slices were prepared as described previously17. To visualize neuronal migration, slices were incubated for 2 h in DMEM/F12 media, containing 10 μg/ml Oregon Green BAPTA-1 488 AM (Molecular Probes) and 0.003% pluronic acid (Molecular Probes) transferred onto poly-D-lysine- and collagen-coated glass coverslips (22 mm2; Corning) and overlaid with additional collagen (Vitrogen from Cohesion) for immobil-ization. After incubation for 1 h without medium in a humidified chamber (5% CO2, 37 °C) and then for 2 h in DMEM/F12 medium with 1× N2-Supplement (Gibco), coverslips were fitted into a temperature-controlled (37 °C) chamber clamped onto the microscope stage. Slices were maintained by continuous perfusion (40 ml/h) with oxygenated DMEM/F12 medium (pH 7.4), enriched with 1× N2-Supplement and 20 mM Hepes.

Labeled cells were viewed through a 20× water-immersion objective (NA 0.75) of a Nikon inverted microscope (488 nm excitation and 522/535 nm emission filters). Images were collected with a cooled CH-350 CCD camera (Roper Scientific) every 10 min as image stacks of 4–6 z-steps (5 μm), with the most superficial image being 40 μm below the slice cut surface. After individual z-sections were deconvoluted (Delta Vision software from Applied Precision), z-sections of each time point were projected onto each other and fused to a movie (NIH Image software).

Clonal analysis.

Pregnant p35 null or scrambler mice were anesthetized with avertin at E12, their uterine horns exposed, and 2 μl of virus solution injected into the lateral ventricle of each embryo with a glass micropipette. After development for 3 more days in utero, embryonic brains were dissected and subjected to GFP immunostaining.

A replication-incompetent retrovirus was used for the in utero injection, which expresses enhanced GFP (EGFP) from the vector pMX-GFP (gift from T. Kitamura, University of Tokyo). Retrovirus was concentrated by centrifugation at 100,000g for 3 h, and precipitated virus particles were diluted in OPTI-MEM at a concentration of 1 × 108 c.f.u./ml. Prior to injection, virus solution was complemented with 200 mg/ml polybrene and 0.01% Fast Green solution.

p35 rescue experiment.

p35 was electroporated into the developing embryo as described previously39. The plasmid backbone used was pCAGGIG (gift from T. Matsuda, Harvard Medical School). This EGFP expression vector is driven by the CAG promoter and directs EGFP expression through the internal ribosomal entry site (IRES). To obtain distal expression, GFP was targeted to the membrane by adding to its N terminus the amino acid sequence MLCCMRRTKQVEEKNDEDQKI of the GAP-43 protein. We then used this modified CAG-IRES-GAP-EGFP construct to express GFP alone, and the CAG-p35-IRES-GAP-EGFP construct to express GFP together with p35. Plasmid DNA was injected at a concentration of 5 μg/μl in phosphate-buffered saline (PBS).

Tissue preparation and GFP immunohistochemistry.

Dissected brains were fixed overnight at 4 °C in 4% paraformaldehyde in PBS, equilibrated for 12–16 h at 4 °C in 30% sucrose in PBS, embedded in O.C.T compound (Sakura Finetek USA Inc.) and frozen in liquid nitrogen. Brains were then cut coronally into 40-μm thick sections with a Leica CM3050 cryostat (Leica Microsystems) and mounted onto Superfrost glass slides (Fisher Scientific).

For GFP immunohistochemistry, cortical sections were blocked for 45 min with 10% goat serum in PBS containing 0.1% Triton X-100 (Sigma) and incubated overnight at 4 °C with a polyclonal anti-GFP antibody (Molecular Probes) diluted at 1:2,000 in blocking solution. The secondary antibody used was a FITC anti-rabbit antibody (ICN/Cappel) incubated for 2 h at room temperature at 1:400 in blocking solution. Fluorescent images were obtained with a CH-350 CCD camera (Roper Scientific) followed by image deconvolution (Delta Vision).

Note: Supplementary information is available on the Nature Neuroscience website.