Lpd is a member of the MIG-10/RIAM/Lpd (MRL) protein family that links signaling from Ras superfamily proteins and phosphoinositides to actin dynamics and cell adhesion during polarization and migration 4–7
. The C. elegans
MRL ortholog, MIG-10, has important roles in neuronal migration, polarization, and axon guidance 8–11
. As Lpd is expressed uniformly throughout the cortex during embryonic development (Supplementary Information, Fig. S1a,b
), we hypothesized that Lpd may function in multiple aspects of cortical development. We used in utero
electroporation to transfect cortical progenitor cells and manipulate Lpd function. E14.5 mouse embryos were co-electroporated with plasmids encoding mCherry (to mark electroporated cells) and a short hairpin RNA (shRNA) vector that knocked down Lpd expression efficiently (Supplementary Information, Fig. S1c
). Analysis of transfected neurons at E18.5 revealed that cells expressing Lpd shRNA accumulated aberrantly within the SVZ and IZ () compared to control cells expressing a random sequence shRNA construct, suggesting that Lpd depletion impairs migration of neurons to the superficial cortical layers.
Lpd silencing impairs neuronal positioning
To ensure the specificity of the cell positioning phenotype for Lpd, two additional Lpd shRNA vectors targeting different regions of Lpd (Data not shown) were electroporated and found to yield similar phenotypes. A rescue assay in which an RNAi resistant Lpd cDNA (by virtue of silent mutations) was co-electroporated with the Lpd shRNA restored normal neuronal positioning, confirmed that the phenotype was Lpd-dependent and not from off-target effects ( and S1d
To determine whether the positioning phenotype resulted from an intrinsic defect in migrating neurons, we expressed shRNA-resistant Lpd selectively in post-mitotic neurons within the developing cortex using the NeuroD1 promoter 12, 13
(pNeuroD1-Lpd) and found that it rescued the Lpd knockdown positioning phenotype (). In a separate test for cell autonomy of the phenotype, embryos were electroporated sequentially: at E13.5 with Lpd shRNA and Venus marker and again at E14.5 with an mCherry marker alone. As expected, the Venus expressing cells exhibited the positioning defect while the mCherry expressing cells were positioned normally (). The morphology of radial glia (by Nestin staining) in transfected regions appeared normal (Supplementary Information Fig. S2
), indicating that the Lpd phenotype was likely independent of defects in radial glial processes. In addition, levels of activated Caspase-3, an apoptotic marker, were unaffected by Lpd depletion, indicating that increased cell death did not contribute to the phenotype (Supplementary Information, Fig. S3
The Lpd shRNA-mediated positioning defect could result from reduced speed of radial migration and/or failure of neurons to exit the SVZ/lower IZ. By time-lapse imaging of organotypic cortical slice cultures prepared from electroporated mouse embryos we found that the small fraction of Lpd shRNA-expressing bipolar neurons that reached the IZ and CP exhibited radial migration rates (9.1 ± 0.75 um/hr, mean±SEM; n = 44) similar to control (7.7 ± 0.44 um/hr; n = 73) (p> 0.05; Student’s t-test) and to rates observed previously for radial bipolar cells (Fig. S4a
. Therefore, Lpd depletion does not affect the migration speed of bipolar neurons that reach the upper IZ and CP. Interestingly, many Lpd-depleted neurons had failed to reach their terminal destination in the CP even when postnatal pups were examined ( and Supplemental Information Fig. S4b
). Therefore, the Lpd depletion phenotype arises primarily from impaired exit of bipolar neurons from the SVZ/lower IZ rather than a delay in radial migration.
We next analyzed the morphology of Lpd knockdown cells aberrantly positioned within the SVZ and lower IZ. Normally, cortical pyramidal neurons transition from multipolar to bipolar morphology in the SVZ/IZ before entering the cortical plate via radial migration15
; defects in the stereotypical shift to bipolar morphology might perturb radial migration. Interestingly, Lpd knockdown increased
the frequency of bipolar cell morphology compared to control (Lpd shRNA: 82.8% ± 3.0%; n = 3 embryos; Control shRNA: 48.85% ± 4.2%; n = 3 embryos; electroporation at E14.5, analysis at E18.5) (). Surprisingly, many bipolar cells expressing Lpd shRNA were oriented tangentially as judged by the angle of their leading process with respect to the pial surface () and appeared parallel to axon fiber tracts in the IZ/SVZ (). Quantification of the average angle of the leading process of these cells and adjacent radial glia fibers confirmed this abnormal tangential orientation (). Many Lpd depleted/tangentially oriented bipolar pyramidal neurons were observed at a significant distance from electroporated regions (); similar phenotypes were observed after electroporation at both early and late stages of cortical development. Therefore, we hypothesized that Lpd inactivation caused pyramidal cortical neurons to migrate tangentially instead of radially along glia. We imaged organotypic cortical slice cultures prepared from E13.5 embryos (electroporated at E11.5) by time-lapse microscopy and observed that all Lpd-depleted bipolar tangential neurons were perpendicular to the radial glial processes and migrated with an average velocity of 10.5 ± 3.4 μm/h (mean±SEM, n = 37) (, Supplemental Videos 1a and 1b
); tangential movement was not observed in control experiments. Interestingly, the average speed of tangentially migrating Lpd-depleted neurons more closely resembled the rates of radially migrating pyramidal neurons than tangentially migrating interneurons 14, 15
Suppression of Lpd increases the number of tangentially oriented bipolar pyramidal neurons in the IZ/SVZ
Lpd depleted bipolar pyramidal neurons migrate tangentially within the IZ/SVZ but do not exhibit a change in cell fate
To determine whether Lpd depletion induced the tangential migration phenotype by a partial or complete change in the fate of pyramidal neurons to interneurons, which normally migrate tangentially through the IZ/SVZ, we immunostained Lpd knockdown neurons for markers that distinguish between those neuronal types. Tangentially oriented Lpd knockdown cells expressed Cux1 (), but not GABA (), indicating that they remained cortical pyramidal neurons and had not converted into tangentially migrating interneurons. Tangentially migrating interneurons within the SVZ must express CXCR4 chemokine receptor 16, 17
to prevent their premature entry into the cortical plate and subsequent radial migration. We reasoned that upregulation of CXCR4 in Lpd knockdown neurons might cause their tangential migration phenotype, however, the tangentially oriented Lpd-knockdown neurons lacked detectable CXCR4(). Together these data indicate that bipolar pyramidal cortical neurons, which normally utilize a radial gliophilic pathway, migrate tangentially in the IZ/SVZ in the absence of Lpd without altering their fate or by expressing CXCR4.
To determine whether long-range cues might influence orientation of tangentially migrating Lpd-depleted cells, we scored the direction of their leading processes () and observed cells that moved both medial to lateral (from the cortex towards the ganglionic eminence) and lateral to medial (from the cortex towards the midline). Although a slight directional bias was observed, we conclude that tangential migration of these neurons likely occurs independently of specific lateral or medial diffusible long-range signals.
We next asked whether tangential migration of Lpd depleted neurons could potentially arise from decreased preference for glial contact, increased preference for migration on or among axons in the axon-rich SVZ/IZ, or both. The vast majority of tangentially oriented cells had a leading process that appeared to be closely apposed to axon fiber tracts (90% ± 0.17; n = 3 embryos; Fig. S5a
). Using an in vitro
culture system, we observed an increased propensity of Lpd depleted cortical neurons to adhere to cortical axon bundles compared to control (Fig. S5b
), raising the possibility that Lpd inactivation caused pyramidal neurons to select axon fiber tracts for tangential migration instead of their normal substrate, radial glia. However, these observations alone are not sufficient to conclude unequivocally that Lpd knockdown neurons migrate tangentially using axons as their substrate.
How does Lpd expression ensure radial glia-guided migration rather than tangential migration? Lpd depletion in cultured cells or in Drosophila
decreases the ratio of filamentous (F) to monomeric (G) actin 5, 18
. We observed that Lpd depletion in cortical neurons resulted in a reduction in F-actin levels (). One way cells adjust to changes in F:G actin ratios involves altered activity of serum response factor (SRF), which together with its co-activator MAL, regulates expression of many cytoskeleton- and adhesion-related genes; when the F:G-actin ratio falls below a critical threshold, G-actin monomer binds to MAL and inhibits its ability to activate SRF 19, 20
. We used an in vivo
SRF reporter assay to examine the effects of Lpd knockdown in the cortex and observed a 4-fold reduction in SRF-responsive gene expression compared to control (). We hypothesized that reduced SRF activity might mediate part or all of the Lpd-knockdown phenotype. Consistent with this idea, shRNA-mediated SRF knockdown resulted in the appearance of tangentially oriented bipolar pyramidal neurons, similar to those observed after Lpd depletion ( and Supplemental Information S6a
). SRF depletion, however, caused an additional neurogenesis defect. Since effective SRF knockdown required expression of the shRNA in neuronal progenitors, it was impossible to evaluate SRF depletion in neurons without also perturbing neurogenesis. To circumvent this problem, we used two well established methods to inhibit SRF selectively in post-mitotic cells: expression of an SRF dominant negative (DN-SRFΔC) that lacks the C-terminal activation domain, or expression of a non-polymerizable β-actin mutant (R62D) that binds to MAL and inhibits its ability to activate SRF 21, 22
. As judged by the in vivo
reporter assay, SRF activity was reduced significantly by neuronal-specific expression of DN-SRFΔC or R62D (). Strikingly, electroporation of either DN-SRFΔC or R62D resulted in tangentially oriented bipolar neurons in the SVZ and lower IZ (, S6b
), a phenotype virtually identical to that of Lpd-depleted neurons. Therefore, similar to Lpd depletion, reduction in SRF activity in post-mitotic neurons results in a shift to tangential orientation.
Lpd affects the orientation of cortical bipolar cells through an SRF/MAL-dependent pathway
To test whether reduced SRF activity was essential for the Lpd knockdown phenotype, we utilized an established approach to bypass the ability of excess G-actin (present in Lpd-depleted cells) to inhibit SRF activity23
. To prevent G-actin from binding to-, and inhibiting the SRF co-activator MAL, we expressed the actin-binding domain within MAL (RPEL) fused to a heterologous nuclear import signal (RPEL-NLS), which acts as a G-actin “sponge”. Expression of RPEL-NLS is known to specifically and effectively restore SRF activity in the presence of low F:G actin levels by preventing excess G-actin from inhibiting MAL-dependent SRF activation23
(). Co-electroporation of RPEL-NLS with the Lpd shRNA rescued both cell positioning () and bipolar tangential cell phenotypes () typical of Lpd depletion. However, co-electroporation of an actin-binding defective RPEL construct, “RPEL(*)-NLS”, with the Lpd shRNA resulted in a phenotype indistinguishable from Lpd knockdown alone. Therefore, Lpd depletion induces the bipolar tangential migration phenotype as a consequence of increased G-actin levels that inhibit MAL activation of SRF (S6c,d). These three independent approaches to perturb G-actin-dependent regulation of SRF via its co-factor MAL are all consistent with the hypothesis that tangential orientation and migration of Lpd-depleted bipolar neurons arise as a consequence of reduced SRF activity.
Expression of MAL G-actin binding motifs (RPEL) rescues the Lpd knockdown orientation and positioning defects of bipolar pyramidal neurons
Altogether, these experiments indicate that Lpd depletion perturbs an SRF/MAL-dependent pathway that ensures that pyramidal neurons migrate along radial glia rather than tangentially. Since the Lpd knockdown neuronal migration phenotype was rescued by uncoupling the ability of excess G-actin to block MAL-dependent SRF activation, Lpd depletion has no other obvious effect on cortical neuronal migration independent of G-actin inhibition of MAL-dependent SRF activation. Therefore, Lpd function may not be required directly for radial migration of pyramidal neurons. Instead, the migration phenotypes described here appear to result solely from the inability of neurons to maintain proper F:G-actin ratios in the absence of Lpd, leading to attenuation of SRF activity. Given the ability of R62D G-actin expression to phenocopy Lpd-depletion, it is likely that perturbation of other regulatory molecules that reduce F:G actin ratios and shut down MAL-dependent SRF activation might also induce a similar phenotype. Thus, Lpd inactivation exposed an unexpected SRF/MAL dependent function in cortical development.
This study provides the first evidence that pyramidal neurons have the intrinsic capacity to switch to tangential migration without changing fate or upregulating CXCR4. Although previous studies have identified many molecules important for radial migration of pyramidal neurons, it remains unclear why these neurons migrate radially instead of tangentially. To our knowledge, this is the first example of a regulatory pathway that instructs pyramidal neurons to migrate along radial glia rather than taking a tangential pathway. This change in behavior of pyramidal neurons may be due to a defect in their ability to choose to migrate along glia. Connexins play a role in regulating adhesion of migrating neurons to radial glia fibers 24,25
, however, silencing of connexins impairs radial migration without causing tangential migration of pyramidal neurons 25
, suggesting that defective neuron:glia adhesion alone is insufficient to cause neurons to switch to a tangential pathway. Alternatively, attenuation of SRF activity may cause pyramidal neurons to respond to a migration cue present in the IZ/SVZ or to exhibit increased adhesion to a non-radial glial substrate in the IZ/SVZ that causes tangential migration parallel to axon fiber tracts, raising the possibility that they migrate along axons. Consistent with this possibility, Lpd-depleted neurons were in close proximity to axons in vivo
and exhibited increased adhesion to axons in vitro
. Whether Lpd-depletion causes pyramidal neurons to contact axons directly is a question for future work. Furthermore, it would be intriguing to determine if Lpd/SRF function affects different types of migration that follow bundled axons such as tumors that invade by spreading along axon fiber tracts.
The vertebrate cortex is a rapidly evolving structure that depends on the appropriate migration and allocation of pyramidal neurons, 80% of all cortical neurons. The present work has identified a mechanism controlling a fundamentally important aspect of pyramidal neuronal migration. Our finding that SRF-dependent transcription drives pyramidal neurons to migrate radially suggests that changes in the expression of SRF targets are sufficient to change migration from radial-glial dependent to tangential; identification of relevant SRF transcriptional targets will provide valuable insight into how pyramidal neurons choose their migration pathway.