The LPA-LPA4 axis is required for establishment of bipolar morphology and radial migration of newborn cortical neurons

Development of the mammalian cerebral cortex is achieved by the intervention of a series of cellular events including neurogenesis, neuronal migration and neuronal maturation. Neuronal migration is a fundamental process for establishing the laminated structure of the cerebral cortex (Ayala et al., 2007; Cooper, 2014; Evsyukova et al., 2013; Kriegstein and Noctor, 2004; Marín et al., 2010; Tsai and Gleeson, 2005). In the developing neocortex, neurons are born from neural progenitor cells/intermediate progenitor cells residing at the ventricular zone/subventricular zone (VZ/SVZ). Newborn neurons initially show a multipolar morphology in the SVZ and lower IZ (LoTurco and Bai, 2006; Marín et al., 2010; Noctor et al., 2004; Tabata and Nakajima, 2003). Subsequently, they extend a ventricle-directed thin tailing process (axon), reorient their centrosomes and Golgi toward the pia in accordance with their radial polarization, and form a single pia-directed thick process (leading process) to adopt a bipolar shape. Neurons with bipolar shape in the lower IZ thereafter exit the IZ and move radially toward the pial surface. These processes of radial polarization and morphological transformation most likely depend on the coordinated actions of extracellular factors and intracellular signaling. To date, only the extracellular glycoprotein reelin and guidance cue semaphorin 3A have been shown to promote the multipolar-to-bipolar transition of newborn neurons through induction of radial polarization (Cooper, 2014; Evsyukova et al., 2013; Marín et al., 2010). However, the extracellular signals and downstream pathways that remodel the cytoskeleton to control morphological transformation of newborn neurons remain largely unknown.

Lysophosphatidic acid (LPA) is a bioactive lipid molecule composed structurally of a phosphate, a glycerol and a fatty acid (Tokumura, 1995). The extracellular actions of LPA are mediated by at least 6 subtypes of G-protein-coupled receptors referred to as LPA1 to LPA6 (Yung et al., 2015). LPA1, LPA2 and LPA4 are known to express in murine embryonic brains (Choi et al., 2010; Yung et al., 2015). Both LPA1 and LPA2 are reported to play a role in neural progenitor cells (Choi et al., 2010; Fukushima et al., 2007; Kingsbury et al., 2003). On the other hand, the roles of LPA4 in the developing brain remain unclear, as mice deficient for LPA4 show normal gross brain anatomy but this may be due to compensatory effects (Lee et al., 2008; Sumida et al., 2010).

In the present study, we find that LPA4 is expressed in migratory neurons of the mouse developing neocortex. Depletion of LPA4 in neurons profoundly impairs neuronal migration and accumulates multipolar neurons in the lower IZ. Using neuronal culture system, we discovered that LPA4-depleted neurons display normal apical localization of the centrosome/Golgi, but fail to form the pia-directed process. Therefore, they remain in a multipolar state, instead of adopting a bipolar morphology. Conversely, LPA treatment leads to precocious formation of the pia-directed process through LPA4, and this morphological change is accompanied with aberrant neuronal migration. Furthermore, LPA4 depletion is coupled with altered remodeling of the actin cytoskeleton, as well as with destabilization of the F-actin-binding protein filamin A (FlnA). Finally, FlnA overexpression rescues defects in remodeling of the actin cytoskeleton, formation of the pia-directed process and neuronal migration of LPA4-depleted neurons.

We first localized LPA4 in the mouse developing neocortex. As shown in Fig. 1A, LPA4 immunoreactivity was prominent at embryonic day (E) 14 and E17, from mid to late corticogenesis. However, no immunoreactivity was detected at E11, whereas reduced signal was detected at postnatal day (P) 1 (Fig. 1A). At E14 and E17, LPA4 immunoreactivity was enriched in the cortical plate (CP) and intermediate zone (IZ), where migratory neurons are abundant (Fig. 1A,B). In addition, relatively weaker LPA4 immunoreactivity was detected in the ventricular zone (VZ) where neural progenitor cells reside. Within the IZ and CP, LPA4 immunoreactivity showed remarkable overlapping expression with the neuronal marker Tuj1. As almost no immunoreactivity was observed in brain sections immunostained using LPA4 antibody pre-absorbed with its immunogen (Fig. S1A), these observations together indicate that LPA4 is expressed in migratory/maturating neurons and to a lesser extent in neural progenitor cells. In addition, in cultured cortical neurons prepared from the E14 neocortex, LPA4 immunoreactivity was detected throughout cell soma and enriched in tips of neurites (Fig. 1C). Specificity of the immunostaining signals was confirmed by the diminished LPA4 immunoreactivity in primary neurons depleted of the LPA4 protein by RNAi (Fig. S1D,E; see below), further supporting the neuronal expression of LPA4.

To characterize the functions of LPA4 in migratory neurons in the developing neocortex, we used DNA-based RNAi approach to acutely knock down the expression of LPA4. For this, we generated plasmids expressing two different short hairpin RNAs (shRNA) against LPA4 (LPA4 shRNA#1 and shRNA#2). These shRNA constructs efficiently silenced LPA4 overexpressed in HEK cells (Fig. S1B). In addition, the shRNA constructs diminished endogenous levels of LPA4 in NIH3T3 cells and primary neurons (Fig. S1C-E). We then electroporated the LPA4 shRNA constructs together with the GFP-expression construct into E14 neocortices to knockdown LPA4 in migratory neurons. In E17 neocortices electroporated with the control shRNA at E14, the majority of the GFP-labeled neurons migrated into the CP and a smaller population of the cells were detected in the IZ. On the other hand, in LPA4 shRNA-introduced neocortices, most of the GFP-labeled cells were located at the interface between the VZ and IZ, and a small population of the cells resided into the CP (Fig. 2A,B). Noticeably, control GFP-positive cells around the lower IZ showed a typical bipolar morphology or a unipolar morphology with a pia-directed process but no detectable tailing process (Fig. 2C,D). On the other hand, most of the LPA4 shRNA-introduced neurons in the lower IZ had multiple short thin processes, or displayed no detectable processes (Fig. 2C,D). The defects in neuronal morphology and positioning of LPA4 shRNA#1-introduced neocortices are almost completely rescued by expression of LPA4 with two silent mutations within the LPA4 shRNA#1 target sequence (LPA4-res; Fig. S1B). These observations not only confirm the specificity of LPA4 shRNA but also indicate that LPA4-depleted neurons display impaired multipolar-to-bipolar transition. As LPA4-depleted neurons are still mis-positioned at the subcortical region at P4 (Fig. 2E,F), we reason that LPA4 knockdown arrests (rather than delays) neuronal migration.

Of note, as in control neurons, almost all misplaced LPA4-depleted neurons were positive for Cux1, a marker of cells destined for layer 2-4 neurons (Nieto et al., 2004) (Fig. 2G). This observation suggests that laminar fate was not altered upon LPA4 depletion and consequently that alteration of laminar fate does not account for the mis-positioning of LPA4-depleted neurons. Nonetheless, immunofluorescent intensities of Cux1 in LPA4-depleted neurons were significantly lower than that in control GFP-labeled neurons (Fig. 2G,H). As Cux1 immunoreactivity is known to increase with neuronal maturation (Nieto et al., 2004), this observation suggests that maturation of LPA4-depleted neurons was likely delayed. In addition, mis-positioning of LPA4-depleted neurons was unlikely caused by disruption of the radial glial scaffold, as indicated by the normal radial fibers reaching the pia in LPA4 shRNA-introduced neocortices (Fig. S2). Neuronal differentiation of progenitor cells was also unaffected in LPA4-electroporated neocortices: the populations of GFP-labeled cells positive for Pax6 (a neural progenitor marker) and Tbr2 (an intermediate progenitor marker) were unchanged in E15 neocortices electroporated with LPA4 shRNA at E13 (Fig. S3). Finally, apoptosis was not enhanced following LPA4 shRNA electroporation (Fig. S4). In summary, these results suggest that LPA4 is required for neurons to adopt a bipolar shape and to undergo radial migration.

To further examine the effects of LPA4 knockdown on neuronal morphology, we used the slice overlay culture system (Polleux et al., 2000) in which dissociated GFP (or mCherry)-labeled cortical neurons are placed on neonatal cortical slices. In this particular system, cortical neurons prepared from E16 neocortices electroporated with GFP at E14 initially display multipolar morphology with short thin processes after ∼24 h in culture (Fig. 3A). After ∼2 days in culture, these neurons adopt a bipolar shape with a pia-directed thick process that is negative for the dendritic marker MAP2, and a ventricle-directed thin axon (Fig. 3A). The morphological transformation is reminiscent of that of migratory neurons in vivo. This system allows us to perform biochemical and molecular manipulations on growing neurons in an environment resembling the in vivo situation. By 5 days in culture, neurons on slices eventually differentiate into pyramidal neurons characterized by a single MAP2-positive pia-directed process, several basal processes and a ventricle-directed long axon (Fig. 3A; Polleux et al., 2000). Using this culture system, we first examined the effect of LPA4 knockdown on cell morphology. For this, E14 neocortices were electroporated with either control shRNA/mCherry or LPA4 shRNA/GFP into dorsal VZ progenitors, which eventually give rise to pyramidal neurons, and cortical cells were prepared at E16. The control shRNA/mCherry- and LPA4 shRNA/GFP-introduced neurons were then mixed and subjected to the slice overlay culture (Fig. 3A). After 2 days in culture, most of the control shRNA-introduced cells displayed a bipolar morphology characterized by a pia-directed thick process and a ventricle-directed axon (Fig. 3B,C). In contrast, electroporation of LPA4 shRNA plasmids resulted in a smaller population of GFP-labeled cells with bipolar morphology. Instead, most of the LPA4 shRNA-introduced cells displayed a multipolar morphology characterized by multiple thin processes emerging from the cell body (Fig. 3B,C). The defect in absence of pia-oriented process in LPA4 shRNA#1-treated neurons was alleviated with expression of LPA4-res (Fig. 3D). Of note, most LPA4 knocked down neurons extended a single ventricle-directed axon-like process (96.3+0.47% in control sh versus 95.5+1.13% in LPA4 sh#1), although the length of the process was shorter than that of control neurons (Fig. S5).

The secreted factors Sema3A and reelin have been shown to influence to the radial polarization of multipolar neurons in the developing cortex by reorienting the centrosome/Golgi apparatus (Chen et al., 2008; Jossin and Cooper, 2011; Shelly et al., 2011). We sought to determine the orientation of the centrosomes and Golgi in multipolar neurons depleted of LPA4. As shown in Fig. 4, LPA4 depletion by shRNA did not affect formation of a ventricle-directed axon-like process nor did it alter the normal apical localization of the centrosome (Fig. 4A,B). The normal apical positioning of the centrosome/Golgi in LPA4-depleted multipolar neurons was also observed in vivo (Fig. 4C-F). Together, these results suggest that LPA4-depleted neurons normally orient the centrosomes and Golgi but fail to form the pia-directed process.

Given that LPA4 affects the multipolar-to-bipolar transition, we next assessed the effects of LPA ligands on the establishment of bipolar morphology. Using slice overlay culture system, we found that LPA supplementation (36 h) increases the fraction of neurons with bipolar morphology in a dose-dependent manner (Fig. 5A). Importantly, the increase was totally abolished upon LPA4 knockdown (Fig. 5A), thereby indicating that the effect of LPA treatment on the pia-directed process formation is mediated by LPA4. To further support the notion that LPA is responsible for establishment of bipolar morphology, we investigated the effects of perturbing LPA synthesis. LPA is mainly produced via two pathways (Aoki et al., 2008): by hydrolysis of phosphatidic acids by phospholipase A1 (PLA1) and A2 (PLA2); or via hydrolysis of lysophospholipids by the secreted enzyme lysophospholipase D (autotaxin; ATX). In the latter pathway, lysophospholipids are produced by phospholipase A1 (PLA1) and A2 (PLA2)-dependent hydrolysis of phospholipids predominantly present within cell membranes (Aoki et al., 2008). We first applied methyl arachidonyl fluorophosphonate (MAFP), an inhibitor for PLA1 and PLA2 (Higgs and Glomset, 1996; Lucas and Dennis, 2005), in a slice overlay culture system. The addition of MAFP (48 h) dramatically reduced the fraction of neurons with a pia-directed thick process (Fig. 5B). Noticeably, as co-treatment of MAFP with LPA almost completely reversed the impaired morphology of neurons observed with MAFP application, we reasoned that MAFP-induced impairment of the pia-directed process formation was attributable to inhibition of LPA synthesis. Next, we performed experiments in the presence of the potent ATX blockers HA155 and PF8380 (Albers et al., 2010; Gierse et al., 2010). We found that these inhibitors significantly reduced the proportion of neurons with bipolar morphology (Fig. 5C). In summary, interfering with LPA-LPA4 pathway impairs formation of the pia-directed process in neurons, whereas LPA treatment promotes its formation.

Given that LPA supplementation and inhibition of LPA synthesis affect morphological transformation of cortical neurons in the slice overlay culture system, we next determined whether these manipulations can alter migration of cortical neurons. For this purpose, neocortical slices were prepared from E16 neocortices electroporated with GFP at E14 and cultured in the presence of LPA and/or inhibitors for LPA synthesis for 36 h. When compared with the distribution of the GFP-labeled cells in control-treated GFP-electroporated slices, a significantly larger fraction of the GFP cells was found at the upper part of the neocortices treated with LPA (Fig. 6A). In addition, more GFP-positive cells in the IZ exhibited the pia-directed thick process in the presence of LPA (Fig. 6B). On the other hand, MAFP treatment resulted in a smaller fraction of GFP cells positioned at the upper part of the neocortex as well as in a smaller fraction of GFP cells with the pia-directed thick process. Importantly, the mis-positioning and attenuated morphological transformation of these neurons were reversed with LPA (Fig. 6A,B). Similar to MAFP, PF8380 treatment of cortical slices resulted in impaired neuronal migration and morphology, and co-treatment with MAFP and PF8380 worsened the defects in neuronal positioning and morphology (Fig. 6C,D). Taken together, these results reveal that manipulations of LPA signaling have a significant impact on radial migration and neuronal morphology. They are consistent with our findings of altered multipolar-to-bipolar transformation and arrested neuronal migration in LPA4-depleted neocortices.

To further confirm the distinct role of LPA-LPA4 signaling in neuronal migration, we treated LPA4 shRNA-electroporated cortical slices with LPA or MAFP, and examined their effect on neuronal positioning. We found that neither LPA nor MAFP had significant effect on the cell positioning in LPA4-depleted cells (Fig. S6). This result indicates that LPA4 is the primary receptor for LPA signaling during neuronal migration in the developing neocortex.

Based on the finding that the LPA-LPA4 axis contributes to neuronal morphogenesis, one can suspect that LPA signaling affects the cytoskeleton in newborn neurons. As remodeling of the actin filaments is central to cell morphogenesis, we sought to characterize the spatiotemporal formation of filamentous actin (F-actin) in neurons using the slice overlay culture system. To visualize F-actin, neurons co-transfected with GFP and the Lifeact peptide (an F-actin binding peptide; Riedl et al., 2008) tagged with DsRed were subjected to the slice overlay culture. As shown in Fig. 7, after 18 h in culture, almost all GFP-introduced neurons showed multipolar shape with intense F-actin signals (Lifeact-DsRed signals) distributed throughout the cell soma and in individual processes (Fig. 7A). After 48 h in culture, about one half of the neurons with multipolar shape showed F-actin signals distributed throughout the cell soma and in proximal parts of individual processes, similar to those at 18 h in culture. In contrast, the remaining half of the cells with multipolar shape displayed enrichment of F-actin signals at the apical side of the cell soma with reduced or no signal in individual processes (Fig. 7A). By contrast, almost all neurons with bipolar morphology showed F-actin signals accumulating at the apical compartment of the cell soma (at the base of the pia-directed process) (Fig. 7A). These observations suggest that in neurons with multipolar shape, F-actin is initially present throughout the cell soma and in individual processes. As differentiation proceeds, F-actin becomes enriched at the apical side beneath the to-be-extended pia-directed process. In support of the theory, time-lapse imaging also showed changes in distribution of Lifeact-DsRed from cell soma/neuritis to the apical compartment, prior to pia-directed process extension (Fig. S7).

Importantly, when LPA4-depleted neurons were cultured in the slice overlay culture system, a larger population of the cells showed widespread distribution of F-actin (14.7±1.7% in control sh versus 41.1±2.4% in LPA4 sh#1, P<0.001 by a two-tailed Student's t-test) (Fig. 7B,C) with no clear apical side enrichment of F-actin. These observations reveal that LPA4 depletion is coupled with altered remodeling of the actin cytoskeleton in neurons at multipolar stage.

The F-actin-binding protein filamin A (FlnA) organizes the actin cytoskeleton and plays a key role in radial migration of neurons in the developing neocortex (Nagano et al., 2004; Razinia et al., 2012; Stossel et al., 2001). We found that FlnA signals were enriched at the apical side of the cell soma in multipolar cells that displayed F-actin enrichment (Fig. S8). Such enrichment of FlnA was not observed in multipolar cells with wide distribution of F-actin (Fig. S8). In addition, in cells with bipolar morphology, FlnA was distributed in the pia-directed process (Fig. S8). These results imply the involvement of FlnA in apical F-actin enrichment and pia-directed process formation. We observed a decrease in FlnA protein levels in cultured neurons prepared from E16 neocortices electroporated with LPA4 shRNA at E14, as assessed by immunostaining (Fig. 8A,B). We then sought to examine the potential effect of LPA4 knockdown on the stability of FlnA in NIH3T3 cells expressing HA-tagged FlnA. When compared with control, shRNA-introduced cells showing no significant decrease in FlnA levels; a large (∼50%) decrease in FlnA levels was observed under LPA4-depleted condition and in the presence of CHX, a protein synthesis inhibitor (Fig. 8C,D). These results suggest that LPA4 signaling leads to stabilization of FlnA. In addition, in HEK293 cells, we found that FlnA associates with LPA4 (Fig. S9), suggesting that LPA4 localizes at close proximity to FLA through direct or indirect interaction; this may be important for stabilization of FlnA by LPA4.

We next investigated the potential link between FlnA- and LPA4-mediated pia-directed process formation of newborn neurons. We found that expression of a dominant-negative form of FlnA lacking the actin-binding domain (ΔABD-FlnA) produced fewer GFP-labeled cells with pia-directed process and apical enrichment of F-actin in the slice overlay culture system (Fig. 8E, Fig. S10), as did LPA4 depletion. In addition, the LPA-induced increase in the fraction of neurons with bipolar morphology was totally abolished upon ΔABD-FlnA expression (Fig. 8E). Furthermore, defects in actin reorganization and pia-directed process formation observed with LPA4 depletion were significantly rescued with expression of FinA (Fig. 8F). Importantly, expression of FlnA almost completely reversed the impaired neuronal positioning and morphology in LPA4 shRNA-depleted neocortices (Fig. 9). These findings indicate that pia-directed process malformation caused by LPA4 depletion is intimately coupled with the decreased levels of FlnA and impaired reorganization of the actin cytoskeleton.

In the present study, we identify the LPA-LPA4 axis as an extrinsic signaling pathway required for the formation of the pia-directed process and hence for the establishment of bipolar morphology of cortical newborn neurons and for their radial migration. A previous report states that LPA inhibits migration of neurons (Fukushima et al., 2002). In the study, the authors showed that almost no neurons emerge from small pieces of cultured cortical explants in the presence of exogenous LPA. They also showed, in whole-brain culture, a marked difference in the distribution of β-tub III/Tuj1-positive cells following LPA application. However, as acknowledged by the authors, their observations cannot exclude the possibility that exogenous LPA alters neuronal processes other than migration per se (such as differentiation, survival, etc.) in the cortical cultures, thereby keeping neurons in the explant and changing the number of neurons produced and/or alive in the whole-brain culture (Fukushima et al., 2002; Kingsbury et al., 2003). In the present study, we directly examined the effects of LPA4 depletion, exogenous LPA and LPA synthesis inhibitors on neuronal migration in cortical slices. We found that migration of newborn neurons is promoted by exogenous LPA and retarded by LPA4 depletion and by LPA-synthesis inhibitors. Of note, maturation of LPA4-depleted neurons appears to be delayed, as indicated by the decreased levels of Cux1 (Fig. 2G,H). This is possibly due to the inappropriate environment surrounding the mis-positioned LPA4-depleted neurons. Alternatively, interference of LPA4 signaling may lead to cell-intrinsic defects in maturation programs. In these scenarios, altered migration of LPA4-depleted neurons may be partly attributable to delayed neuronal maturation.

In the developing brain, LPA acts as a repellent factor for axons. In dispersed cortical neuron and retinal cultures, LPA causes axonal retraction (Fukushima et al., 2002; Campbell and Holt, 2003). In addition, LPA guides thalamocortical axons to proper cortical targets in brain slices (Cheng et al., 2016). In the present study, we uncovered a novel role for LPA signaling in the formation of the pia-directed process during the multipolar-to-bipolar transition. Thus, LPA mediates multiple context-dependent neuronal functions in the developing brain.

It is of note that mice with genetic deletion of Lpa4 (Lpar4 - Mouse Genome Informatics) show no overt phenotype during CNS development (Lee et al., 2008; Sumida et al., 2010), although the possibility of compensatory effects induced by genetic knockout cannot be excluded. This is reminiscent to the situation of Lpa1/Lpa2 double-knockout mice that show mostly normal neocortical development (Contos et al., 2002), despite their potential involvement in progenitor biology. On the other hand, in utero electroporation of shRNA that allows the acute downregulation of genes, circumvents the compensatory mechanisms that may be activated in knockout animals [e.g. doublecortin (Bai et al., 2003), α-chimaerin (Ip et al., 2012) and β-amyloid precursor protein (Young-Pearse et al., 2007)]. This technique, combined with a RNAi approach enabled us to uncover novel function for LPA and LPA4 signaling in neuronal transformation and radial migration.

Sema3A and reelin have been reported to contribute to multipolar-to-bipolar transition of newborn neurons in the developing neocortex. Importantly, secreted from the marginal zone, Sema3A and reelin function as polarizing cues for migratory neurons in vivo. Interfering with Sema3A and reelin signaling pathway disrupts the normal apical positioning of centrosome/Golgi in multipolar neurons. This defect is accompanied with the production of both misoriented bipolar neurons and multipolar neurons (Chen et al., 2008; Jossin and Cooper, 2011; Shelly et al., 2011). In contrast, disruption of LPA-LPA4 signaling did not affect positioning of the centrosome/Golgi. Thus, the action of the LPA-LPA4 axis on the multipolar-to-bipolar transition is distinct and unique: it occurs after the positioning of the centrosome and Golgi, and affects specifically the pia-directed process. Thus, during the multipolar-to-bipolar transition, multipolar newborn cortical neurons reorient their centrosome/Golgi via extracellular polarizing factors such as Sema3A/reelin. Via LPA-LPA4 signaling that acts as a later morphogenic cue, they then form the pia-directed process and acquire the bipolar morphology.

Dynamic remodeling of the actin network occurs during the multipolar-to-bipolar transition (Fig. S7). Importantly, normal apical enrichment of F-actin within differentiating multipolar neurons is prevented by LPA4 depletion (Fig. 7). Instead, F-actin remained all over the cell. Furthermore, we identified the F-actin-binding protein FlnA as a key effector of the LPA-LPA4 cascade: FlnA is destabilized in LPA4-depleted cells and overexpression of FlnA rescues the pia-directed process formation as well as cortical radial migration and neuronal positioning. Remodeling of the cytoskeleton via FlnA upon LPA/LPA4 pathway activation is rather specific. Indeed, RhoA GTPase, which also impacts the actin network, was not activated in cortical neurons in which actin rearrangement occurs (Fig. S11). On the other hand, stabilization of microtubules, as assessed by levels of Ac-tub, has been reported to control multipolar-to-bipolar morphological transition of cortical neurons (Ip et al., 2012; Wu et al., 2012). In our hands, depletion of LPA4 did not affect the levels of Ac-tub in cortical neurons (Fig. S12). This result suggests that LPA-LPA4 signaling is unlikely to direct multipolar-to-bipolar morphological transition through microtubule stabilization and/or tubulin acetylation. Nevertheless, LPA4 could modulate microtubule dynamics via other mechanisms to induce this transition.

FlnA has been reported to contribute to VZ surface lining by rearranging the actin cytoskeleton and interacting with various extracellular matrix proteins (Carabalona et al., 2012; Feng et al., 2006; Sarkisian et al., 2006). Despite the expression of LPA4 in both neural progenitors and migrating neurons, LPA4-depletion by in utero electroporation (i.e. LPA4 depletion in progenitor cells) resulted in no disturbance of the VZ surface. This result suggests that the LPA-LPA4 pathway is unlikely to contribute to regulation of FlnA levels in progenitors. Thus, the FlnA stabilization mechanism reported herein is neuron specific. Further studies examining how LPA4 regulates FlnA level will lead to a better understanding of the stage-specific regulation/function of FlnA.

LPA is mainly generated through pathways involving PLA1/PLA2 and ATX (Aoki et al., 2008). We found that application of both ATX inhibitors (HA155 and PF8380) and a PLA1/2 inhibitor (MAFP) resulted in impaired formation of pia-directed process and altered neuronal migration. These observations indicate that PLA1/PLA2 and ATX are both involved in production of extracellular LPA to affect neuronal morphogenesis. Of note, depletion of ATX has little effect on neuronal migration (Greenman et al., 2015). As ATX is a secreted enzyme, ATX depletion could be rescued by ATX produced by neighboring cortical cells and/or a non-cortical source. In the present study, we used bath application of PF8380 to inhibit the ATX activity in the entire cultured cortex, thereby dissecting out roles of the ATX-LPA pathway in neuronal migration. Importantly, ATX is expressed in the CP and subplate (Greenman et al., 2015; Cheng et al., 2016), whereas PLA is present in cortical neurons (Kishimoto et al., 1999; Ong et al., 2010). In addition, LPA is present in the IZ and CP with enrichment around the CP and subplate in the neocortex at E16 (Cheng et al., 2016). LPA can also be synthesized and secreted by cortical neurons in culture (Fukushima et al., 2000). Considering these studies and our data, we provide a working model in which multipolar neurons in the SVZ and lower IZ are exposed to LPA produced in the SP/CP; upon LPA action on LPA4, these newborn neurons adopt a bipolar morphology and initiate radial migration from the IZ.

In conclusion, our study provides novel insights into the morphogenesis of newborn neocortical neurons and their migratory environment and ability. The identification of the LPA-LPA4-FlnA axis carries the hope of deciphering novel mechanisms underlying environmental regulation of neuronal morphogenesis and neuronal migration, which are central processes in corticogenesis.

ICR mice were purchased from SLC (Hamamatsu, Japan) and were housed under a 12 h light-12 h dark cycle with ad libitum access to food and water. Both male and female mice were employed without distinction in all the experiments. All animal experiments were conducted in accordance with guidelines set by The University of Tokyo and approved (permit number 21-01) by the Committee on Animal Care and Use of the Graduate School of Science in The University of Tokyo.

pCAGIG plasmid (expressing GFP under the control of the CAG promoter), GFP-ΔABD-FlnA-expressing plasmid, HA-FlnA-expressing plasmid and pBS/U6 plasmid were kind gifts from Drs Takahiko Matsuda (Kyoto University, Japan), Makoto Sato (Osaka University, Japan), Thomas Stossel (Harvard Medical School, Boston, MA, USA) and Yang Shi (Harvard Medical School, Boston, MA, USA), respectively. For generating a plasmid expressing LPA4, mCherry and DsRed2-CentrinII under the control of CAG promotor, the full-length open reading frame of each gene was amplified by PCR with Pfu Turbo polymerase (Stratagene) and subcloned into the pCAGEN plasmid (a kind gift from Dr Takahiko Matsuda). Plasmids encoding silent mutation of LPA4 (LPA4-res) were generated using a QuikChange mutagenesis technique (Agilent Technologies) with primers 5′-catct atggg agcat gctgt ttctc acctg catca gtg-3′ (forward) and 5′-cactg atgca ggtga gaaac agcat gctcc catag atg-3′ (reverse). For HA-tagged LPA4, an oligonucleotide-encoding HA epitope sequence was fused to the 5′ end of the full-length LPA4 coding sequence. Plasmid expressing Lifeact-DsRed (Riedl et al., 2008) was generated by inserting the fragment encoding MGVADLIKKFESISKEE-DsRed into the pCAGEN plasmid. Plasmids encoding LPA4 shRNA were generated by inserting the annealed oligonucleotides into the pBS/U6 plasmid (Sui et al., 2002). The target sequences for RNAi are as follows: for LPA4 shRNA#1, 5′-gggag caugc ucuuc cucac c-3′; for LPA4 shRNA#2, 5′-gggac ugcgu uccuc accaa c-3′.

Antibody against LPA4 was generated by Sigma Genosis as described below. Synthetic peptides corresponding to the amino acids Thr³⁴¹-Gln³⁵⁶ of LPA4 were conjugated to keyhole limpet hemocyanin and were injected into rabbits. LPA4 antibody was affinity-purified from the serum by a column coupled with immunogen peptides.

The following antibodies were used for immunostaining. Rabbit anti-LPA4 (1:100), rat anti-GFP (1:2000; 04434-34, Nakalai Tesque), rabbit anti-GFP (1:2000; A11122, Invitrogen), mouse anti-MAP2 (1:5000; ab11268, Abcam), mouse anti-Tuj1 (1:3000; MMS-435P, Covance), mouse anti-nestin (1:1000, 556309, BD Biosciences), rabbit anti-BLBP (1;1000, AB9558, Chemicon), rabbit anti-Pax6 (1:1000; PRB-278P, Covance), rabbit anti-Tbr2 (1:500; ab23345, Abcam), rabbit anti-Cux1 (1:100; sc-13024, Santa Cruz Biotechnology), mouse anti-GM130 (1:300, 610822, BD Biosciences), mouse anti-FlnA (1:100, ab80837, Abcam), mouse anti-acetylated tubulin (1:500, T6793, Sigma-Aldrich) and rabbit anti-cleaved caspase 3 (1:500; #9661, Cell Signaling Technology).

The following antibodies were used for immunoblotting and immunoprecipitation. Rabbit anti-GFP (1:2000; A11122, Invitrogen), rabbit anti-HA (1:1000, 561, MBL), mouse anti-FLAG (1:1000; F1804, Sigma-Aldrich) and mouse anti-β-actin (1:50,000; A1978, Sigma-Aldrich).

DNA solution in PBS containing 0.01% Fast Green was injected into the lateral ventricle of mouse embryos. Thereafter, electroporation (five 50 ms square pulse with 950 ms intervals; Nepa gene, CUY21-EDIT) was carried out with forceps-type electrodes (CUY650P5; Nepa gene) to introduce plasmids into neural progenitor cells in the VZ of the developing neocortex. Electroporation voltage used was 42V. Final concentrations of the plasmids used are as follows. Plasmids expressing GFP (pCAGIG: 2-5 μg/μl), mCherry (2 μg/μl), DsRed2-CentrinII (0.5 μg/μl), LPA4 with silent mutation (5 μg/μl), Lifeact-DsRed (0.5 μg/μl), LPA4 shRNA (1-2 μg/μl), HA-FlnA (8 μg/μl) and GFP-ΔABD-FlnA (5 μg/μl).

Brains were fixed with 4% paraformaldehyde in PBS for 30 min (for embryonic brains) or 2 h (for postnatal brains) at room temperature and cryoprotected in 30% sucrose in PBS overnight at 4°C. Thereafter, the brains were embedded in a solution of a 2:1 mixture of 30% sucrose/PBS and OCT compound (Sakura, Tokyo, Japan), frozen by liquid nitrogen and stored at −80°C until use. Thick cryosections (20 μm) were made. Brain sections were washed with PBS, incubated with blocking solution [3% (w/v) BSA, 5% (v/v) FBS and 0.2% (w/v) Triton X-100 in PBS] and then incubated with primary antibodies overnight at 4°C. The sections were then incubated with Alexa488/Cy3/Cy5-conjugated secondary antibodies overnight at 4°C and mounted in a Prolong Gold mounting solution (Invitrogen). Nuclei were visualized with TO-PRO-3 (Thermo Fisher Scientific). Images were obtained with a 63× objective (Plan-Apochromat; Zeiss) on Zeiss LSM 5 confocal microscope.

HEK293 and NIH3T3 cells were purchased from the European Collection of Authenticated Cell Cultures (ECACC). We did not authenticate the cell line in our lab. The lot was authenticated by ECACC before shipment. We did not test for mycoplasma contamination in our lab. The lot was tested at ECACC before shipment. HEK293 or NIH3T3 cells maintained in 10% FBS/DMEM were transiently transfected using Lipofectamin 3000 (Invitrogen) according to the manufacturer's instructions. Transfected cells were then cultured in 10% FBS/DMEM for 48 h. For detection of LPA4, the cells were serum starved overnight before the harvest. For inhibition of protein synthesis, cycloheximide (final concentration, 100 μM; Nacalai Tesque) was added to the culture medium. The cells were then subjected to immunoblotting, immunoprecipitation or immunocytochemistry.

For neocortical neuronal culture, E14 embryos were electroporated with plasmids encoding GFP, mCherry, DsRed2-CentrinII, LPA4 shRNA and LPA4 with silence mutation. Forty-eight hours after electroporation, neocortical cells were prepared and cultured as described earlier (Asada et al., 2007). Cultured cortical neurons were fixed at 1 day in vitro (DIV; the plating day is defined as 0 DIV) or 2 DIV with 4% paraformaldehyde in PBS for 30 min at 37°C, permeabilized with 0.5% Triton X-100 in PBS for 5 min, blocked with 3% BSA/0.2% Triton X-100 in PBS and incubated with primary antibodies in the blocking solution at 4°C overnight. The coverslips were then incubated with Alexa488/Cy3/Cy5-conjugated secondary antibodies for 2 h at room temperature and mounted in a Prolong Gold mounting solution (Invitrogen). F-actin was visualized using acti-stain phalloidin (Cytoskeleton). Fluorescent images were obtained using Zeiss LSM5 confocal microscope.

The harvested cells were lysed in lysis buffer [20 mM Tris-HCl, 1% (vol/vol) Triton X-100, 10% (vol/vol) glycerol, 137 mM NaCl, 1 mM dithiothreitol, 2 mM EDTA, 50 mM NaF, 1 mM Na₃VO₄ and protease inhibitor cocktail (Complete, EDTA-free; Roche); pH 8.0 at 4°C]. After a 30 min incubation on ice, the cell lysates were centrifuged for 10 min at 20,000 g, and the resultant supernatant was collected. The cell extracts were then incubated with 1.0 μg of a precipitating antibody at 4°C overnight, followed by incubation with 15 μl of protein G-Sepharose at 4°C for 3 h. The beads were washed three times with lysis buffer and subjected to immunoblotting.

E16 neocortices were incubated in 1 ml HBSS (Invitrogen) containing papain (20 units; Roche), L-cystein (0.32 mg/ml) and 20 mM HEPES-NaOH (pH 7.4) for 30 min. The neocortices were rinsed twice with HBSS and dispersed with a fire-polished Pasteur pipet in Neurobasal media (Invitrogen) containing B27 (Invitrogen), glutamine and PenStrep. Neocortical cultures were treated at DIV1 with 1 µM LPA. At DIV1, almost all of the neocortical cells (>98%) were Tuj1-positive neurons. For F-actin staining, neurons were fixed with 4% paraformaldehyde/PBS for 20 min at 37°C. Neurons were then rinsed with PBS, followed by treatment with 0.5% Triton X-100/PBS for 5 min at room temperature. Thereafter, F-actin was visualized with acti-stain phalloidin (Cytoskeleton). RhoA activity was measured with a RhoA activation assay kit (Abcam) using neocortical cells prepared as above. NIH3T3 cells were serum starved in DMEM supplemented with 0.1% FBS for 24 h, prior to LPA treatment.

Acute brain slices (300 μm) of postnatal day 1 or 2 mouse brains were prepared as described previously (Asada and Sanada, 2010). Neocortical regions were dissected out from the slice and transferred onto a transparent porous membrane (1.0 μm pore size, BD Falcon; precoated with poly-D-lysine and laminin) in a 35 mm well containing culture media [F-12/DMEM supplemented with N2 and 5% horse serum (Invitrogen)]. Neocortical neuronal culture was prepared from E16 mouse brains as described above, and plating on top of cortical slices that were settled for 2 h after preparation. Three hours after the cell plating, culture membranes were transferred to the culture media without serum. For LPA/drug treatment, LPA (final concentration, 200 nM or 1 μM: Cayman Chemical), HA155 (final concentration, 1 μM: Sigma-Aldrich), PF8380 (final concentration, 1 μM: Sigma-Aldrich) and methyl arachidonyl fluorophosphates (MAFP) (final concentration, 10 μM: Cayman Chemical) were added to the media. Thirty-six to 48 h after the cell plating, neuronal cultures were fixed with 4% paraformaldehyde in PBS for 30 min at 37°C, then subjected to immunostaining.

For time-lapse imaging of Lifeact-DsRed, neuronal cultures were first treated with MAFP and PF8380 for 37 h to maintain and to synchronize them at the multipolar stage. Thereafter, LPA was supplemented to induce pia-directed process formation. Images of GFP and DsRed were then taken every 1-6 h for about 24 h. The experiments were designed to minimalize phototoxicity but allowed us to clearly image the multipolar-to-bipolar transition. At each time-point, the Petri dishes with slice overlay culture were transferred from the CO₂ incubator to an Axio observer upright microscope (Carl Zeiss Microimaging), and images of GFP-expressing cells were collected with an AxioCam cooled CCD camera (Carl Zeiss Microimaging).

Plasmid expressing GFP was electroporated into E14 neocortices. Forty hours later, acute brain slices of electroporated brains were prepared using the method described above. Slices were cultured on a transparent porous membrane in a six-well culture plate containing culture medium (F-12/DMEM supplemented with N2 and B27) for 36 h. The reagents used were LPA (final concentration, 1 μM), PF8380 (final concentration, 1 μM) and MAFP (final concentration, 10 μM). Cultured slices were fixed in 4% paraformaldehyde and then subjected to immunostaining.

All bar graphs were plotted as mean±s.e.m. Direct comparisons were made using two-tailed Student's or Welch's t-test. One-way ANOVA followed by Bonferroni's multiple comparison test was used for experiments for three or more datasets. Repeat measures two-way ANOVAs were performed for comparisons of neuronal positioning of cortical slice cultures. These statistical analyses were performed using Microsoft Excel. The significance level was set at P<0.05 for all tests. No statistical methods were used to pre-determine sample sizes but our sample sizes are similar to those generally employed in the field. The experiments were randomized, and data collection and analyses were performed blind to the experimental condition.

We thank Drs Takahiko Matsuda, Makoto Sato, Thomas Stossel and Yang Shi for plasmids.