Reelin and cofilin cooperate during the migration of cortical neurons: a quantitative morphological analysis

Neuronal migration is a fundamental process in brain development resulting in the formation of neuronal layers. In the mammalian cerebral cortex, pyramidal neurons born in the ventricular zone (VZ) migrate radially towards the marginal zone (MZ) along the processes of radial glial cells and form the six-layered cortex (Rakic, 1971, 1972; Nadarajah et al., 2001, 2003; Cooper, 2008). In the dentate gyrus of the hippocampus, the granule cells born in the hilus also migrate towards the MZ and form a densely packed granular layer (Förster et al., 2002, 2006a,b; Frotscher et al., 2003; Weiss et al., 2003; Zhao et al., 2004, 2006). In the cerebellum, Purkinje cells migrate along Bergmann glial fibers and form the Purkinje plate beneath the transient external granule cell layer (Yuasa et al., 1991). In all these different brain regions, radial neuronal migration is controlled by reelin, an extracellular matrix protein synthesized by cells located in the target direction of the migrating neurons, suggesting a role for reelin in directed migration. Cajal-Retzius cells in the MZ are a major source of reelin in the cortex and hippocampus (D'Arcangelo et al., 1995; Frotscher, 1997, 1998; Alcántara et al., 1998; Tissir and Goffinet, 2003); external granule cells synthesize reelin in the cerebellum (Schiffmann et al., 1997). In reeler mutant mice deficient in reelin (Reln), cortical lamination is disrupted (Falconer, 1951; Kubo and Nakajima, 2003; Tissir and Goffinet, 2003; Hack et al., 2007), granule cells in the hippocampus are loosely distributed throughout the dentate gyrus (Stanfield and Cowan, 1979; Förster et al., 2002, 2006a,b; Zhao et al., 2004, 2006) and Purkinje cells in the cerebellum are ectopically positioned (Yuasa et al., 1993).

Reelin binds to two lipoprotein receptors, apolipoprotein-E receptor 2 (ApoER2 also known as Lrp8) and very low-density lipoprotein receptor (VLDLR; Trommsdorff et al., 1999; D'Arcangelo et al., 1999; Hiesberger et al., 1999) and induces the phosphorylation of the adaptor protein disabled 1 (Dab1; Sheldon et al., 1997; Howell et al., 1999; Benhayon et al., 2003). ApoER2 and Vldlr double-knockout mice and Dab1 knockout mice show similar neuronal migration defects as seen in reeler mutants (Rakic and Caviness, 1995; Trommsdorff et al., 1999; Walsh and Goffinet, 2000; Drakew et al., 2002).

Despite some insight into the reelin signaling cascade, it is still poorly understood how reelin regulates neuronal migration. Since the neurons always migrate towards the reelin-containing zones, reelin has been proposed to act as a chemoattractive factor for radially migrating neurons (Caffrey et al., 2014). Conversely, reelin has been suggested to be a stop or detachment signal because the MZ, which is enriched in reelin, is almost cell-free in wild-type mice, but is invaded by numerous neurons in the reeler mutant (Frotscher, 1998; Hack et al., 2007; Chai et al., 2009; Zhao and Frotscher, 2010; Hirota et al., 2015). By binding to VLDLR or integrin receptors on the leading processes, reelin in the MZ seems to arrest migrating neurons (Anton et al., 1999; Dulabon et al., 2000; Sanada et al., 2004; Schmid et al., 2005; Hack et al., 2007; Chai et al., 2009; Sekine et al., 2012; Hirota et al., 2015).

Neuronal migration is a coordinated movement comprising extension of the leading process, translocation of the nucleus and retraction of the trailing process. These different processes are all associated with rearrangements of the cytoskeleton (Bamburg, 1999; Pollard and Borisy, 2003; Jovceva et al., 2007). Thus, in order to understand migration, we need to know how reelin signaling regulates cytoskeletal dynamics. We have shown previously that reelin signaling enhances the activity of LIM kinase 1 (LIMK1; Chai et al., 2009), which phosphorylates the actin-binding protein cofilin (non-muscle cofilin 1 or CFL1; Arber et al., 1998; Yang et al., 1998). Actin-depolymerizing factor (ADF) and cofilin sever actin filaments (F-actin) and thereby generate new filament barbed ends (Lappalainen and Drubin, 1997; Ichetovkin et al., 2000; Andrianantoandro and Pollard, 2006), which are available for new rounds of elongation, and monomeric actin, which can be reused to build new filamentous actin structures. Thus, cofilin participates in the dynamic reorganization of the actin cytoskeleton and defines the direction of cell migration (Ghosh et al., 2004). Phosphorylation of cofilin at serine 3 inhibits all cofilin-actin interactions, blocks actin dynamics and subsequent process extension (Nagaoka et al., 1996; Zebda et al., 2000; Huang et al., 2006; Bravo-Cordero et al., 2013). Reelin-induced cofilin phosphorylation in the leading processes of migrating neurons stabilizes their cytoskeleton, anchors them to the marginal zone and promotes somal translocation (Chai et al., 2009; Frotscher, 2010; Förster et al., 2010; Franco et al., 2011; Jossin and Cooper, 2011). Thus, reelin-induced phosphorylation of cofilin appears to play a pivotal role in the proper positioning of migrating neurons. Both reelin and cofilin are required because conditional cofilin knockout mice show severe migration defects of late generated cortical neurons destined to superficial layers of the cerebral cortex (Bellenchi et al., 2007).

In the present study, we aimed to better understand the interplay between reelin and cofilin during the migration of cortical neurons by interfering with reelin-induced cofilin phosphorylation at serine 3. We generated different cofilin constructs, a nonphosphorylatable form of cofilin and a pseudophosphorylated form, cloned in-frame into the vector pCAG-GFP. We used timed in utero electroporation (IUE) and real-time microscopy to monitor the migratory behavior of the transfected cells.

We hypothesized that IUE with a nonphosphorylatable form of cofilin (cofilin^S3A) would result in a phenotype similar to that of reeler because cofilin phosphorylation is significantly reduced in the reeler mutant (Chai et al., 2009). We expected a contrasting phenotype when transfecting embryonic cortical neurons with a pseudo-phosphorylated form of cofilin (cofilin^S3E). Control neurons transfected with pCAG-GFP alone were compared with neurons transfected with cofilin^S3A, cofilin^S3E, cofilin^WT and with neurons in reeler embryos transfected with pCAG-GFP.

On E16.5, only minor differences in the migratory behavior were observed between neurons transfected with the different constructs 2 days previously (Fig. 1). Thus, the majority of all transfected cells were still found in the intermediate zone (IZ), subventricular zone (SVZ) and ventricular zone (VZ), respectively. At this stage, many cells in the inner portion of the IZ appeared multipolar (Hatanaka and Yamauchi, 2013), whereas those neurons that had already reached the outer IZ or the cortical plate (CP), such as many pCAG-GFP-transfected control cells (Fig. 1A), showed a long leading process oriented towards the marginal zone (MZ). We found that ∼15% of the control cells had already reached the cortical plate (Fig. 2A,B) compared with 25% in embryos transfected with cofilin^WT (Fig. 1B, Fig. 2A,B). By contrast, in cofilin^S3A-transfected slices, in slices transfected with cofilin^S3E and in reeler slices most cells were clustered in the IZ, SVZ and VZ, and very few cells were found in the CP (Fig. 1C-E, Fig. 2).

When studying the transfected neurons in the upper IZ at high-power magnification, we observed some morphological differences between the mutant cells. When compared with control cells (Fig. 3A), cofilin^WT-transfected neurons gave rise to very long and thin leading processes. Occasionally several processes originated from the cell body or proximal leading process (Fig. 3B). Cofilin^S3A cells were characterized by many short processes originating from the cell body (Fig. 3C), whereas cofilin^S3E-transfected neurons had long, varicose leading processes and several short processes, which gave rise to filopodia-like protrusions (Fig. 3D). Surprisingly, reeler neurons displayed a normal bipolar morphology that was indistinguishable from control cells (Fig. 3E). Although the lengths of the leading processes were not significantly different between the various mutant cells (except for cofilin^S3E-transfected neurons, Fig. 4A), the number of supernumerary processes originating from the cell body was found increased in neurons transfected with the different cofilin constructs (Fig. 4B).

Next, we performed live imaging and measured the migratory speed and directionality of migrating neurons using Imaris software. Statistical analysis of migratory speed indicated no significant differences between GFP-transfected control cells and cells transfected with the different cofilin constructs (n=50 cells for each experimental condition; Fig. 4C). Reeler neurons also showed a normal migratory speed when compared with control cells. However, the direction of migration varied to some extent. While cofilin^WT and cofilin^S3A cells did not show obvious abnormalities, some cofilin^S3E and reeler neurons were observed that migrated back towards the VZ (Fig. 4D), as described before in a study of neuronal trajectories in reeler mutants (Britto et al., 2011). Taken together, there were some morphological abnormalities and minor migration defects as early as 2 days after IUE on E14.5.

Three days after IUE, the majority of GFP-positive cells in controls were scattered over the CP and IZ, and only very few cells were still seen in the SVZ and VZ, respectively (Fig. 5A). The cells that had reached the CP had not yet formed a distinct cell layer (Tabata and Nakajima, 2008). When dividing the CP into an upper, middle and lower portion, almost equal numbers of neurons were found in these subzones of CP, together amounting to more than 60% of all GFP-labeled neurons in controls (Fig. 6A-D). Also in cofilin^WT-transfected embryos, numerous cells had invaded the CP (∼45%), but more cells than in controls were found in the SVZ/VZ (Fig. 5B, Fig. 6A-F). In cofilin^S3A-transfected sections, the majority of labeled neurons were clustered in the IZ and SVZ, and only very few cells had entered the CP (Fig. 5C, Fig. 6A-F). In cofilin^S3E-transfected slices, many labeled neurons were still seen in the IZ and SVZ, but ∼30% of them had migrated into the CP (Fig. 5D, Fig. 6A-F). By contrast, in sections from reeler mutants, virtually all GFP-labeled neurons were seen in the deep layers of the cortex, with many of them still present in the VZ/SVZ (Fig. 5E, Fig. 6A-F). We hypothesized that the migration defect in reeler is at least partially due to insufficient phosphorylation of cofilin. In fact, when we transfected reeler embryos with the pseudophosphorylated form of cofilin (cofilin^S3E) on E14.5 and studied the brains on E17.5, we observed a partial rescue of the reeler phenotype (Fig. 5F, Fig. 6A-F). A similar partial rescue was observed when reeler embryos were transfected with LIM kinase 1 (Limk1), which is known to phosphorylate cofilin (Fig. 5G, Fig. 6A-F).

Next, we again studied the structural characteristics and migratory behavior of neurons transfected with the different constructs (Fig. 7, Movies 1-5). Control cells revealed their characteristic bipolar shape and moved smoothly by nuclear translocation (∼85 µm in 200 min; Fig. 7A, Movie 1). By contrast, many neurons transfected with cofilin^WT showed very little forward movement during this time period (Fig. 7B, Movie 2). Moreover, cofilin^WT-transfected cells were observed to form supernumerary processes (Fig. 7B, red arrowhead) and cofilin-actin rods (Ono et al., 1996; Bernstein and Bamburg, 2003; Bernstein et al., 2006). Occasionally, we noticed that portions of the soma were squeezed in the leading process before the rest of the soma hooked up during the process of nuclear translocation (Fig. 7B). However, this was similarly observed in all other types of transfected neurons. Individual neurons of embryos transfected with cofilin^S3A and cofilin^S3E were regularly found to change their shape over time but without clear forward movement of the soma (Fig. 7C,D, Movies 3 and 4). The leading processes of cofilin^S3A neurons often gave rise to branches (Fig. 7C). Cofilin^S3E neurons and reeler cells, but also cofilin^S3A neurons, were often oriented towards the ventricular zone (Fig. 7D,E, Movies 4 and 5). Many reeler neurons transfected with cofilin^S3E or Limk1 showed a normal migratory behavior (Fig. 7F,G, Movie 6).

These observations were confirmed by quantitative analyses (Fig. 8). We reported recently that reelin-induced branching of the leading processes anchors them to the marginal zone (Chai et al., 2015). Therefore, we again measured the lengths of leading processes in the different mutant cells and quantified the percentage of them that had reached the marginal zone. The data revealed that the leading processes of cofilin^WT-transfected neurons were particularly long and many of them had reached the MZ similar to those of control cells when compared with the leading processes of all other mutant cells (Fig. 8A,B).

When compared with control cells, the average migratory speed of neurons transfected with the different cofilin constructs as well as of pCAG-GFP-transfected reeler neurons was significantly reduced (Fig. 8C). Moreover, the percentage of neurons migrating towards VZ was significantly increased in reeler neurons (Britto et al., 2011) and cofilin^S3A- and cofilin^S3E-transfected cells (Fig. 8D), consistent with the reduced number of leading processes of these cells reaching the MZ (Fig. 8B). Of note, both migratory speed and migration direction towards the MZ were partially rescued by transfecting reeler neurons with cofilin^S3E or Limk1 (Fig. 8C,D).

In summary, 3 days after IUE, the majority of migrating neurons transfected with nonphosphorylatable or pseudophosphorylated cofilin mutants were unable to perceive and convey instructive positional signals from the marginal zone, which together with structural abnormalities resulted in reduced migratory speed and altered directionality of the migratory process. This migration defect resembled that of reeler mutants in which the reelin signal from the MZ is absent. A remarkable rescue of the reeler phenotype was observed when reeler neurons were transfected with Limk1 or pseudophosphorylated cofilin (cofilin^S3E).

Next, we studied the transfected animals after birth. At this time point, most neurons have terminated their migration and reached their destinations in the CP. Control cells formed a compact cell layer subjacent to the MZ (Fig. 9A). More than 80% of all labeled cells in controls were found in the upper cortical plate (layer II/III of the cortex; Fig. 10). The IZ and SVZ/VZ, contained only very few neurons that were still migrating. In tissue sections of cofilin^WT-transfected mice, there were minor differences to controls; these neurons were also capable of forming a cell layer near the marginal zone. However, significantly fewer cells compared with controls (about 50%) were found in the upper portion of the CP (Fig. 9B, Fig. 10). The vast majority of GFP-positive neurons in animals transfected with cofilin^S3A (Fig. 9C, Fig. 10) or cofilin^S3E (Fig. 9D, Fig. 10) were observed in the deep layers of the cortex. In reeler tissue sections, GFP-positive cells were scattered all over the cortex with many cells showing an inverted orientation towards the VZ. Very few cells had invaded the upper layers of the CP subjacent to the MZ (Fig. 9E, Fig. 10).

Since at that stage most neurons transfected at E14.5 have terminated their migration – having arrived at normal or abnormal locations – we refrained from measuring migratory speed and documenting migratory behavior but analyzed their morphological characteristics. For this purpose, we focused on neurons that were still migrating in each experimental group that were behind the main cohort of cells and were located at the border between the upper IZ and the CP.

Such ‘late’ neurons in control slices still showed the characteristic asymmetric shape of migrating cells (Fig. S1A,F). In cofilin^WT-transfected cells, many small branches originated from the cell bodies and leading processes (Fig. S1B,F). Cofilin^S3A cells displayed an asymmetric bipolar shape with short, branched leading processes. Some neurons exhibited two or more processes directed towards the MZ (Fig. S1C,F). Cofilin^S3E neurons had lost the characteristic asymmetric polarity of migrating neurons. The two processes originating from the opposite poles of the cell body appeared equally short and thick, and no typical leading or trailing processes could be identified (Fig. S1D,F). GFP-labeled reeler neurons displayed a variety of different shapes with leading processes oriented in all directions, often towards the white matter (Fig. S1E,F).

Taken together, 5 days after IUE at E14.5, the migration defects observed after shorter survival times have become more obvious because GFP-labeled neurons have now terminated their migration and, in controls, formed a cell layer in the upper portion of the cortical plate. Although cofilin^WT-transfected neurons were largely similar to controls, the majority of cofilin^S3A and cofilin^S3E cells were unable to migrate to the upper cortical plate, pointing to a central role of phosphorylatable cofilin in the migratory process. Numerous reeler neurons invaded the cortical plate, but very few of them reached its upper portion.

Directed neuronal migration is a process that requires the coordinated reorganization of the actin cytoskeleton. Changes in cell shape during migration are associated with cytoskeletal remodeling; however, stability of the leading process is needed for nuclear translocation to take place (Nadarajah et al., 2001; Nadarajah and Parnavelas, 2002; Miyata and Ogawa, 2007; Cooper, 2013). Hence, neuronal migration is characterized by well-coordinated, consecutive periods of cytoskeletal stability and reorganization of the leading process. Actin-depolymerizing molecules such as ADF and cofilin play important roles because they control the turnover and stability, respectively, of actin filaments. Forced expression of cofilin^WT, but in particular transfection with a nonphosphorylatable form of cofilin (cofilin^S3A) or a pseudo-phosphorylated form (cofilin^S3E), resulted in alterations of this fine-tuned balance, in structural abnormalities of the migrating neurons and in migration defects. To some extent, the observations in these cofilin mutant cells resembled those in reeler neurons, probably because reelin is crucially involved in cofilin phosphorylation (Chai et al., 2009). As an example, cofilin^S3A and cofilin^S3E cells, as well as reeler neurons, showed backward migration towards the VZ (Britto et al., 2011; present study), suggesting that a reelin gradient from the MZ is pivotal for the stabilization of the leading process to the MZ by cofilin phosphorylation and thus for directed migration to the CP. Both cofilin and reelin are involved in cell proliferation (Bellenchi et al., 2007; Zhao et al., 2007). In particular, cofilin plays an important role in cell cycle exit (Bellenchi et al., 2007). Thus, the results presented in the present study have to be interpreted with the caveat that IUE with cofilin mutants might also have affected progenitor cell proliferation.

Our present results have shown that overexpression of cofilin^WT leads to the formation of supernumerary processes resulting in a loss of the normal asymmetric bipolarity of migrating cells entering the CP (Miyata et al., 2001; Noctor et al., 2004,, 2007; Kriegstein and Noctor, 2004; Tsai and Gleeson, 2005; Ayala et al., 2007). These structural abnormalities most likely contribute to the migration defects seen 3 and 5 days following IUE. Overexpression of ADF and cofilin increases cytoskeletal remodeling and neurite outgrowth (Meberg and Bamburg, 2000; Endo et al., 2003; Flynn et al., 2012), including the formation of rods within the cells (Ono et al., 1996; Bernstein and Bamburg, 2003; Bernstein et al., 2006). We found that many neurons overexpressing cofilin^WT formed a relatively long leading process and cofilin-actin rods; they largely reached their appropriate destinations in the CP, but layer formation was delayed, consistent with a reduction in migratory speed. These results are in line with studies reporting that overexpression of ADF and cofilin inhibits motility and invasiveness of different types of cancer cells (Lee et al., 2005; Yap et al., 2005). Whether the overexpressed cofilin induced elongation of processes or initiated new branches is much dependent on its relative concentration to F-actin (Andrianantoandro and Pollard, 2006) and its cooperation with the Arp2/3 protein complex. The Arp2/3 complex has been shown to bind to the site of a pre-existing ‘mother’ filament, which then induces the lateral outgrowth of a new filament, leading to a branched F-actin network (Volkmann et al., 2001; DesMarais et al., 2004; Egile et al., 2005; Rouiller et al., 2008). Cofilin supports Arp2/3 complex activity by severing existing capped, i.e. elongation-blocked actin filaments, thus generating free fast-growing barbed ends that will then elongate and form new binding sites for the active Arp2/3 complex (DesMarais et al., 2004), supporting the notion that the synergistic interaction between cofilin and the Arp2/3 complex might have been responsible for the branching and process extension observed in cofilin^WT-transfected neurons.

There are other proteins involved in the organization of the cytoskeleton. Pacary et al. (2011) have previously shown that co-transfection of a non-phosphorylatable form of cofilin (cofilin^S3A) together with shRNA for the proneural transcription factors Rnd2 and Rnd3 rescued the migration defect of Rnd2-silenced neurons, but not that of Rnd3-silenced cells, pointing to a significant role of Rnd proteins in the reorganization of the actin cytoskeleton.

Although essentially two modes of neuronal migration have been described, i.e. somal translocation and locomotion along guiding radial glial fibers (Rakic, 1971, 1972; Nadarajah et al., 2001; Nadarajah and Parnavelas, 2002; Kriegstein and Noctor, 2004; Cooper, 2008, 2013), translocation of the nucleus into the leading process is common to both forms. Soma translocation requires that the leading process is under tension (Miyata and Ogawa, 2007; Cooper, 2013) and that the soma is pulled forward by a myosin II-dependent flow of actin filaments towards the tip of the leading process (He et al., 2010). Movement of the nucleus involves SUN-domain proteins connecting microtubule-based motor proteins with the nuclear membrane (Zhang et al., 2009). We have shown here that the majority of neurons expressing a form of cofilin that cannot be phosphorylated do not populate the CP. As soma translocation and directed migration necessitates a leading process with a stabilized actin cytoskeleton (Chai et al., 2009), overexpression of cofilin^S3A might result in the depolymerization of the actin filaments necessary for the functionality of the leading process, loss of directed migration and eventually even backward migration.

Cofilin^S3A is likely to produce many free barbed F-actin ends, which might initiate the branching activity of the Arp2/3 complex (Svitkina and Borisy, 1999; Pantaloni et al., 2000; Volkmann et al., 2001; Amann and Pollard, 2001; Ichetovkin et al., 2002). Ongoing reorganization of the actin cytoskeleton in S3A neurons might not only lead to the formation of supernumerary processes but might also result in remodeling of microtubules as well as altered localization of the centrosome (Solecki et al., 2009; Sakakibara et al., 2014), which determines neuronal polarity (de Anda et al., 2005). As a consequence, the cells will change migration direction. We hypothesize that exuberant process growth and insufficient stabilization of the actin cytoskeleton are crucially involved in the inability of cofilin^S3A-transfected cells to migrate properly towards the cortical surface.

The migration defect of cofilin^S3A-transfected neurons is reminiscent of that in reeler mutants. Reelin, when concentrated in the MZ of the developing cortex, was previously found to phosphorylate cofilin by activating LIM kinase 1 (Chai et al., 2009). Thus, the leading processes of migrating neurons are stabilized as they approach the MZ, which supports directed migration towards the surface of the cortex. Reeler neurons and cofilin^S3A-transfected neurons have in common that deficient stabilization of the leading processes prevents them from becoming anchored to the MZ, resulting in aberrant (non-directed) migration including backward migration towards the VZ (Britto et al., 2011). The inverted course of many pyramidal cell apical dendrites, the former leading processes, in the adult reeler cortex reflects this mal-orientation of cortical neurons during the developmental period (Terashima et al., 1985, 1992; Frotscher et al., 2009). Mal-orientation of migrating neurons in the reeler cortex eventually results in the malformation of cortical layers as visualized by using layer-specific markers (Hack et al., 2007; Dekimoto et al., 2010).

We regard it as an important result exemplifying the cooperation of reelin and cofilin in the migration of cortical neurons that transfection of neurons in reeler embryos with Limk1 or pseudophosphorylated cofilin (cofilin^S3E) partially rescued the reeler phenotype. Future studies will aim at downregulating cyclin-dependent kinase 5 (Cdk5), because suppression of Cdk5 was found to increase cofilin phosphorylation in cortical neurons (Kawauchi et al., 2006).

During development, reelin is also detectable in the germinal zone and in layer V neurons of the cortical plate (Alcántara et al., 1998; Schiffmann et al., 1997; Jossin et al., 2007; Chai et al., 2009). It has been suggested that reelin activates Rap1, which, in turn, upregulates N-cadherin through Rab-GTPase-dependent endocytic pathways (Kawauchi et al., 2010; Jossin and Cooper, 2011; Franco et al., 2011). N-cadherin is needed for the orientation of migrating multipolar neurons in the intermediate zone and their transformation to a bipolar shape (Kadowaki et al., 2007; Shikanai et al., 2011; Jossin and Cooper, 2011). We hypothesize that reelin in the germinal zone and in layer V enables late-generated neurons to bypass their predecessors and to migrate to their destination in superficial cortical layers (Pinto-Lord et al., 1982; Nadarajah et al., 2001, 2003).

A thick leading process pointing towards the marginal zone and a thin trailing process, the future axon, originating from the opposite pole of the cell body, are characteristic features of a migrating neuron. Controlled actin turnover is likely to be involved in this asymmetric cell differentiation, but is altered in neurons transfected with cofilin^S3E. In these cells, the leading and the trailing processes are equally thick and poorly differentiated. Cofilin^S3E mimics phosphorylated cofilin (Moriyama et al., 1996) and thus inhibits the supply of G-actin, which is essential for actin polymerization. In conditional cofilin knockout mice, layer II and III of the cerebral cortex are missing, implying migration defects of the neurons destined to these layers (Bellenchi et al., 2007). We conclude that the severing activity of cofilin and remodeling of the actin cytoskeleton, respectively, are required for the differentiation of asymmetric cell polarity, process extension and motility of migrating neurons.

Our results point to an important role of cofilin and its coordinated regulation by reelin-dependent phosphorylation during the directed migration of cortical neurons. Further studies will be necessary to specify the involvement of the reelin receptors ApoER2 and VLDLR, which were found previously to have specific functional roles (Zhao et al., 2006; Hack et al., 2007; Zhao and Frotscher, 2010) and distinct spatiotemporal expression patterns (Hirota et al., 2015).

Pregnant wild-type mice (C57/BI/6J; n=50) and heterozygous reeler mutants (C57/BI/6J-reln; n=12) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Animals were bred in the Experimental Animal Center of the University Medical Center Hamburg-Eppendorf. All animals were maintained in accordance with the institutional guidelines of the University of Hamburg (license no. 48/13). Care was taken to minimize suffering of the animals during surgical procedures. The day on which the vaginal plug was detected was designated as embryonic day 0.5 (E0.5). The first neonatal day was considered to be postnatal day 0 (P0). Reeler genotypes were confirmed by PCR analysis of genomic DNA and immunostaining for reelin (Deller et al., 1999; Chai et al., 2009).

Cofilin^WT was generated by PCR from a mouse Cfl1 cDNA library (GenBank accession no. NM_007687) using the forward primer: 5′-CCGGAATTCGCCACCATGGCCTCTGGTGTGGCTGTC-3′ and reverse primer: 5′-GCGGGATCCCCCAAAGGCTTGCCCTCCAGG-3′. Two cofilin mutants were generated that mimic either the dephosphorylated (constitutively active) or phosphorylated (dominant negative) form by changing Ser3 to alanine (cofilin^S3A) or aspartate (cofilin^S3E), respectively. Wild-type Limk1 was generated by PCR from a mouse Limk1 cDNA library (GenBank accession no. NM-010717) using the forward primer: 5′-GTCGACGCCACCATGAGGTTGACGCTACTTTGTTGCA-3′ and reverse primer: 5′-AGATCTGTCAGGGACCTCGGGG-3′. The resulting fragments were cloned in-frame into the vector pCAG-GFP, a mammalian expression vector driven by the chicken actin promoter (Addgene, plasmid 11150; Niwa et al., 1991). The constructs were further tested by sequence analysis and enzyme restriction. The resulting plasmids were then purified using an Endo-free maxi prep kit from Qiagen.

IUE was performed as described previously (Saito and Nakatsuji, 2001; Tabata and Nakajima, 2001). Briefly, timed pregnant mice were deeply anesthetized with isoflurane (Abbott, Wiesbaden, Germany) at E14.5 (wild-type mice, n=50; reeler mice, n=12) and then firmly fixed on a heating mat. After dehairing and disinfecting the abdomen with iodine tincture, a 3 cm midline laparotomy was performed and the uterus was exposed. Plasmids were dissolved in phosphate-buffered saline (PBS; pH 7.4) at a concentration of 1 µg/µl. Fast Green solution (0.1%) was added to the plasmid solution in a ratio of 1:10 to monitor the injection. Approximately 1 µl of the plasmid solution was injected into one of the lateral ventricles with a glass micropipette made from a microcapillary tube (GB100TF-10; Science Products, Hofheim, Germany), which was connected to an aspirator tube (Sigma). The heads of embryos in the uterus were placed between the 7 mm platinum tweezers electrodes (Model 520; Harvard Apparatus). Electronic pulses (30 V, 50 ms) were given five times at intervals of 950 ms with an ECM830 BTX square wave electroporator (ECM 830 BTX; Harvard Apparatus). The uterine horns were placed back in their original location. The abdominal wall and skin were stitched with surgical sutures, and the embryos were allowed to continue development. Plasmid pCAG-GFP was injected into the lateral ventricles of wild-type embryos as a control. Alternatively, the different n-cofilin constructs (cofilin^WT, cofilin^S3A and cofilin^S3E) were injected. Reeler mutants were injected either with pCAG-GFP alone, with cofilin^S3E or Limk1. After electroporation, the uterine horns were placed back in their original location. Dams were sacrificed and the brains of the embryos used for the preparation of slice cultures (see below). Alternatively, 2, 3 or 5 days after IUE, brains were fixed in 4% paraformaldehyde (PFA) and sectioned transversally into 50-μm-thick sections.

Two or three days after electroporation, pregnant mice at E16.5 (wild-type mice, n=16; heterozygous reeler mice, n=2) and E17.5 (wild-type mice, n=24; heterozygous reeler mice, n=4), respectively, were decapitated under hypothermic anesthesia. The embryos at E16.5 (for each cofilin construct and control plasmid, n=24; for reeler, n=5) and at E17.5 (for each cofilin construct and control plasmid, n=36; for reeler mice, n=14) were collected and rapidly placed in ice-cold Hank's balanced salt solution (HBSS; Invitrogen). The brains were dissected from the skull and checked under a fluorescence microscope.

GFP-positive cerebral cortices were then dissected and sectioned (300 μm) perpendicular to the longitudinal axis of the cerebral cortex using a McIlwain tissue chopper. The slices were placed onto culture inserts (Millipore) and transferred to 6-well plates with 1 ml/well nutrition medium containing 25% heat-inactivated horse serum, 25% Hank's balanced salt solution, 50% minimal essential medium and 2 mM glutamine (pH 7.2; Invitrogen), and incubated in 5% CO₂ at 37°C for at least 3 h. After recovery, the culture inserts were placed in Petri dishes (30 mm diameter) with a glass bottom containing fresh medium and then transferred for live imaging.

For live imaging of slice cultures, an Improvision confocal spinning-disc microscope (Zeiss) and 20× air immersion objective were used to acquire z-series of cortical slices at 5 µm intervals through a tissue depth of about 20 µm. The z-series were then visualized as single optical scans with concurrent orthogonal views using Volocity6 software (Perkin Elmer). The time interval was 10 min. Duration of imaging was up to 15 h. This microscope was equipped with a chamber for the control of temperature, humidity and CO₂. Results of time-lapse imaging of the slice cultures were analyzed using Volocity6 software, and the average speed of neuronal cell migration was measured by using Imaris software. Measurements from 50 GFP-expressing cells were collected from three videos of each different plasmid transfection and stage (2 and 3 days after IUE). Only neurons whose migration could be tracked for a substantial distance were included. In these cells, the average speed of movement and the direction of migration (towards the marginal zone or ventricular zone) were measured in the x- and y-axes.

Brains were prepared for histological analysis 2, 3 and 5 days after IUE. Of the embryos transfected with the different cofilin constructs, at least three per different construct and stage and three GFP-transfected reeler brains were used. Three reeler mice each were used for the rescue experiments with cofilin^S3E and Limk1, respectively, 3 days after IUE on E14.5. Two and three days after IUE, pregnant wild-type mice and heterozygous reeler mutants were anesthetized with isoflurane under hypothermic anesthesia and the embryos dissected in ice-cold HBSS. Five days after IUE, corresponding to the day of birth, newborn mice were sacrificed by cervical dislocation under hypothermia and the brains dissected and rapidly placed in HBSS. Brains were viewed under a fluorescence microscope to verify transfection. GFP-positive hemispheres were fixed in 4% PFA overnight at 4°C. After washing in 0.1 M PBS at room temperature (RT) for several hours, brains were embedded in 5% agar and cut transversally into 50-µm-thick slices on a LeicaVT 1000S vibratome (Leica Microsystems) and the sections were placed in a 24-well plate containing 0.1 M PBS, counterstained with propidium iodide (PI; Sigma), and mounted in Mowiol on glass slides. Sections were photographed using a confocal laser-scanning microscope (Leica TCS SP5) and 20× air or 63× oil immersion objectives. z-series of brain sections at 0.5 μm intervals through a tissue depth of 9 μm were acquired and visualized as single optical scans with concurrent orthogonal views using Leica LAS software (Leica).

In fixed slices, all GFP-positive neurons in the different zones of the cortex were counted. Two days after IUE, cells were counted in CP, IZ and SVZ/VZ (n=3 sections from three different mice). Three days after IUE, when the cortex had increased in thickness, the cortical plate was subdivided into three portions: an upper CP (UCP), middle CP (MCP) and lower CP (LCP) (n=3 sections from three different mice). Five days after IUE, cells were counted in the UCP, IZ and SVZ/VZ (n=3 slices from three different mice). Cell counts were performed using ImageJ (NIH) software.

In 50-100 neurons from each transfected group of cells, the lengths of the leading processes 2 days and 3 days after IUE were measured using iTEM software (Zeiss). In addition, 2 days after IUE, the percentage of migrating neurons giving rise to more than two leading processes and 3 days after IUE the percentage of GFP-positive neurons with leading processes reaching the marginal zone was determined.

Results were documented using Excel software (Microsoft) and presented as means±s.e.m. Differences between groups were tested for statistical significance (one-way ANOVA with Tukey's multiple comparison test, *P<0.05; **P<0.01; ***P<0.001).

We thank Dr Froylan Calderon de Anda for supplying the ECM830 BTX electroporator.