A missed exit: Reelin sets in motion Dab1 polyubiquitination to put the break on neuronal migration
In 1951, Douglas Scott Falconer first described the reeler spontaneous mutant mouse (Falconer 1951). In those mice, cortical neurons are generated normally but migrate abnormally, resulting in an inversion of the cortical laminar organization, with later-born neurons remaining in the deeper layers of the cortex. Forty-four years later, D’Arcangelo et al. (1995) identified the causative gene Rln and the encoded protein Reelin. Reelin pathway mutants, and particularly mice with mutations of its intracellular effector Dab1 (Disabled-1), probably represent the most-studied phenotype of altered neuronal migration. The observations that mutations of Dab1 phenocopy the Rln mutation, and that Dab1 is up-regulated in reeler mice suggested the existence of a negative feedback loop in Reelin signaling via the regulation of Dab1 levels. In the previous issue of Genes & Development, Feng et al. (2007) presented the first coherent model to explain the mechanism of Reelin-mediated Dab1 down-regulation. They proposed that Reelin both activates and down-regulates its effector Dab1, first by inducing its phosphorylation, which then causes its targeting by an E3 ubiquitin ligase complex (EBC) containing SOCS family proteins and Cullin5 (Cul5). The impairment of this down-regulation mechanism in vivo leads to a unique phenotype in the cortex where neurons migrate past their target layer.
The known side of Reelin/Dab1 signaling
Reelin is a large matrix-associated transmembrane glycoprotein expressed by specific neuronal populations throughout the developing CNS. Reelin has been shown to homodimerize, at least in vitro, and to bind to and cluster two receptors from the LDL (low-density lipoprotein) family: VLDLR (very low-density lipoprotein receptor) and ApoER2 (apolipoprotein E receptor type2) (D’Arcangelo et al. 1999; Hiesberger et al. 1999; Trommsdorff et al. 1999; Kubo et al. 2002; Strasser et al. 2004). The binding of Reelin to its receptors induces the tyrosine phosphorylation of the cytoplasmic adaptor protein Dab1, which interacts with the conserved motif NPxY in the cytoplasmic tails of VLDLR and ApoER2 (Trommsdorff et al. 1998; Howell et al. 1999).
In addition, Dab1 binds to and is phophorylated by nonreceptor tyrosine kinases from the Src family (SFK): Src and Fyn (Arnaud et al. 2003a; Kuo et al. 2005). Noticeably, the mouse Dab1 was originally identified in a yeast two-hybrid screen for Src-interacting proteins (Howell et al. 1997). Src and Fyn are activated (phosphorylated) through the Reelin pathway, and thus together, Dab1 tyrosine phosphorylation and SFK phosphorylation are two interdependent mechanisms that are part of a positive feedback loop (Arnaud et al. 2003b; Bock and Herz 2003).
In the developing cortex, Reelin is specifically expressed by horizontally orientated Cajal-Retzius neurons located in the preplate before its splitting and then in the outermost layer of the cortex (the marginal zone of the developing brain, becoming layer I of the mature brain) (Fig. 1; D’Arcangelo et al. 1995; Ogawa et al. 1995). On the other hand, ApoER2, VLDLR, and Dab1 are expressed by both cortical neurons and radial glia cells supporting their migration (Forster et al. 2002; Frotscher et al. 2003; Hartfuss et al. 2003; Luque et al. 2003). Strikingly, mutations of both VLDLR and ApoER2, of Dab1 and particularly point mutations at five Dab1 tyrosine phosphorylation sites (Dab15F mutation of Tyr185, Tyr198, Tyr200, Tyr220, and Tyr232), or mutation of both SFK Src and Fyn result in a phenocopy of the reeler phenotype, with a failure to split the preplate as well as inverting cortical layering (Howell et al. 1997, 2000; Sheldon et al. 1997; Benhayon et al. 2003; Kuo et al. 2005), validating that they participate in a common signaling pathway regulating cortical neuronal migration.
Proposed model of Dab1 down-regulation during cortical neuron migration. During cortical development, neurons (green) migrate along radial glia processes (light gray) from the ventricular zone (VZ) (where they are generated) to their final position within the cortical plate (CP). Later-born neurons (progressively darker green shade) bypass previously deposited neuronal (i.e., earlier-born) neurons (lighter green), thus organizing the cortex in an “inside-out” manner. Reelin is secreted by Cajal-Retzius cells (red) located at early developmental stages (i) within the preplate (PP) before it splits into the subplate (SP) and the marginal zone (MZ), and at later stages (ii) within the marginal zone. Both neurons and radial glia cells express Reelin receptors VLDLR and ApoER2 (orange) and are responsive to Reelin. Reelin signaling displays a negative feedback regulation by which Reelin limits its own action by down-regulating the level of its effector Dab1. Feng et al. (2007) demonstrated that reelin-induced phosphorylation of Dab1 on Tyr185 and Tyr198 cause the polyubiquitination of Dab1 (Ub) by E3 ubiquitin ligases containing SOCS proteins and Cul5. Polyubiquitinated Dab1 is then degraded by the proteasome. Correct cortical development requires a strictly regulated balance of Reelin signaling and its negative feedback. During neuronal migration through the deeper layers of the cortex (a), Dab1 level is normally low (light yellow), potentially due to its distance to the Reelin source at this time, or to its down-regulation induced by Reelin encountered at an early stage of development (i). (b) The uninhibited Dab1 signaling (bright yellow) may be required for the neuron to progress through previously deposited neuronal layers. (c) In turn, Dab1 down-regulation may be necessary to stop neuron movement and avoid “overmigration.” (SVZ) Subventricular zone; (IZ) intermediate zone.
Down-regulating Dab1
Desensitization is a key feature of most signaling pathways as it allows a system to reset and thus regain sensitivity to repeated stimulation. Many studies demonstrated that the Reelin pathway displays such a negative feedback mechanism. It has been shown that the disruption of Reelin signaling (by genetic disruption of Reelin, VLDLR, ApoER2, or Fyn and Src) leads to the accumulation of Dab1 protein in vivo, suggesting that Reelin limits its action in responsive neurons by eventually down-regulating Dab1 levels (Sheldon et al. 1997; Rice et al. 1998; Trommsdorff et al. 1999; Arnaud et al. 2003b; Bock and Herz 2003; Kuo et al. 2005). Interestingly, this Dab1 down-regulation happens post-transcriptionaly, as Dab1 mRNA expression is unchanged in the deficient cells (Rice et al. 1998). Several recent advances provide further understanding of the mechanism of this post-transcriptional down-regulation of Dab1. Arnaud et al. (2003a) and Bock et al. (2004) previously showed in primary neuronal cultures that Dab1 is down-regulated through polyubiquitination and targeting to the proteasome in the presence of Reelin, and that this modification requires tyrosine phosphorylation of Dab1. Moreover, inhibition of Dab1 tyrosine phosphorylation and inhibition of SFK activation blocks Dab1 down-regulation (Arnaud et al. 2003b; Bock et al. 2004; Kuo et al. 2005). Feng et al. (2007) identified two critical Dab1 tyrosine phosphorylation sites (primary Tyr198, and also Tyr185), both at YQxI sequences, essential for the Dab1 degradation in response to Reelin, in an embryonic cortical neuron culture system. Tyr198 had been formerly predicted as a potential Reelin-induced Dab1 phosphorylation site by a Web-based PESTfind algorithm; however, the same method failed to identify Tyr185, but instead predicted the importance of Tyr220 (Bock et al. 2004). Among the potential ubiquitin ligase complexes targeting Dab1 in the presence of Reelin, the E3 ligase Cbl seemed a good candidate to mediate Dab1 degradation. Indeed, Cbl is abundantly expressed in cortical neurons and contains a phospho-tyrosine recognition domain comprising an EF-hand and a SH2-like structure through which it binds phospho-tyrosine residues, followed by a hydrophobic residue at pY + 4 (fourth residue after the phospho-tyrosine). These are similar to the motifs found at phosphorylation sites Tyr185 and Tyr198 of Dab1 (Lupher et al. 1997; Meng et al. 1999). Previous studies showed that Cbl can ubiquitinate Dab1, at least in transfected COS7 cells (Suetsugu et al. 2004). However, Arnaud et al. (2003a) failed to detect complexed Cbl and Dab1 by immunoprecipitation, or tyrosine-phosphorylated Cbl in Reelin-stimulated primary neuron cultures, which are required for its activation as an E3 ligases (Levkowitz et al. 1999). Additionally, targeted deletion of Cbl in mice failed to induce neurodevelopmental defects reminiscent of the reeler phenotype (Thien and Langdon 2001).
Feng et al. (2007) identifed another type of EBC potentially regulating Dab1 in vivo. They also used nonneuronal cells lines such as COS7, in which Dab1, when coexpressed with Fyn and Src, is known to be constitutively tyrosine phophorylated but not degraded. This suggests that a component of the degradation machinery targeting phospho-tyrosine Dab1 is absent or limiting in this in vitro system, which thus makes it particularly adapted for a candidate approach. Feng et al. (2007) tested several E3 ubiquitin ligases in this nonneuronal cell system and show that the SOCS1–3 ligases can bind to Dab1, and induce Tyr185 or Tyr198 phophorylated Dab1 degradation, in a Fyn-dependent manner. Interestingly, they also tested Cbl in this assay and show that it fails to induce Dab1 degradation, further invalidating the Cbl implication hypothesis. Because many SOCS proteins are expressed during brain development (among them SOCS1–3) and might be functionally redundant, Feng et al. (2007) chose to study the consequence of the inactivation of another component of the SOCS-containing EBC on Reelin-induced degradation of Dab1: the cullin Cul5 (Petroski and Deshaies 2005). Cul5 recently has been shown and confirmed in the study by Feng et al. (2007) to be expressed in mouse cortical neurons (Lein et al. 2007), but its role was so far completely unknown. Feng et al. (2007) showed that in cultured cortical neurons, Cul5 binds to Dab1, and that Cul5 knockdown specifically protects Dab1 from Reelin-induced degradation. Noticeably, Dab1 interaction with ubiquitin ligases may also have other biological significance. Dab1 has been shown to interact with the E3 ubiquitin ligase Siah-1A in yeast two-hybrid assays and coimmunoprecipitation experiments in 293T cells. But in that case, Dab1–Siah1 binding induces the inhibition of Siah-1A ubiquitinating activity (Park et al. 2003). Thus, one might wonder if Dab1 could additionally regulate EBC containing SOCS and Cul5 as well, adding a step to the complexity of this Reelin feedback loop.
Overmigration: when Dab1 down-regulation fails to occur
Feng et al. (2007) showed for the first time the in vivo consequences of a failure of Dab1 down-regulation on neuronal migration. Bock et al. (2004) showed that blocking proteasome activity disturbs the proper organization of a cultured embryonic cortical plate, suggesting very indirectly that Dab1 degradation is important for its role in regulating neuronal migration. In this paper, Feng et al. (2007) observed that Cul5 knockdown in the cortex leads to an increased Dab1 level in vivo and to a unique migration defect. Cul5 knockdown neurons are located more superficially within the cortical plate relative to control neurons, and partly intrude into the marginal zone. This is the first description of such a phenotype of “overmigration” that clearly differs from the similarly named phenotype of POMGnT1 (O-mannose β31,2-N-acetyl- glucosaminyl-transferase 1)-deficient mice, in which cortical neurons overmigrate past the neural boundary through breaches in the pia matter caused by overgrown radial glial endfeet (Hu et al. 2007). Interestingly, Cul5 knockdown neurons remain at the surface of the cortex, instead of being bypassed by younger neurons. This defect is partially rescued in vivo by the coelectroporation of Dab1 short hairpin RNA (shRNA). Furthermore Cul5 shRNA stops Dab1 degradation in vitro, validating that the overmigration is due to the inhibition of Dab1 degradation.
In the developing cortex, neurons migrate along the processes of their parental radial glia from the ventricular zone to the more superficial cortical layer and stop when they reach the marginal zone. Later-born neurons migrate through previously deposited neurons in cortical plate, leading to an inside-out lamination of the cortex (Fig. 1). The reeler phenotype displayed by Rln-deficient mice as well as mice deficient for VLDLR and ApoER2, Dab1, and Src and Fyn mutant show a rough inversion of the normal inside-out pattern of cortical migration and an excess of neurons in the normally cell-sparse marginal zone (Caviness and Sidman 1973; Howell et al. 1997, 2000; Sheldon et al. 1997; Benhayon et al. 2003; Kuo et al. 2005). The commonly accepted explanation for this phenotype is that the newly arrived neurons fail to penetrate the preplate and split it appropriately into the marginal zone and subplate, and cannot bypass previously deposited neurons and accumulate in an outside-in manner. Dab1 seems necessary to achieve the bypassing step in mosaic Dab1 wt; Dab1−/− embryos, mutant neurons lie below wild-type neurons (Hammond et al. 2001), and BrdU pulse-labeling experiments showed that Dab1 knockdown neurons accumulate in deep cortical plate and fail to pass their earlier-born siblings. Furthermore, neurons showing an abnormally high level of Dab1 migrate presumably faster through the cortical plate and stick to the marginal zone, preventing the passage of their siblings (Feng et al. 2007).
Overall, those results strongly suggest that Dab1 is necessary for neurons to cross previously deposited neuronal layers, and in turn, Dab1 should be down-regulated in neurons to properly terminate their migration and allow their siblings to bypass them. According to this hypothesis, a precise regulation of Dab1 levels should be required to control the precise location of the migration arrest. A threshold response to a gradient of a signaling molecule may be an efficient way to control a timely stop. However, maintaining a Reelin gradient within the cortical plate may be difficult as the depth and the layered organization of the cortical plate are continuously changing. A transcriptional mode of regulation implies latency of protein synthesis, which is slow in the case of Dab1, and of the protein half-life (12 h for Dab1) (Arnaud et al. 2003a), and thus is probably not an appropriate way to control a timely switch mechanism. Controlling expression level by regulating the degradation of a constantly synthesized protein, in a continuously responding neuron can constitute a very efficient method, in contrast. In migrating cortical neurons, Dab1 is constantly down-regulated (as implied by the fact that in reeler mice, the Dab1 level is increased throughout the depth of the cortex) (Arnaud et al. 2003a) via the Cul5/SOCS-containing E3 ligase complex. Dab1 regulation thus relies on the EBC activity, which may be rapidly controlled by protein complex assembly, to rapidly decrease Dab1 levels and definitively terminate neuronal migration.
Although knockdown results are very informative, essential complementary data would come from the analysis of the brain phenotype of Cul5-deficient mice. The inhibition of ubiquitination in cortical slices by Bock and Herz (2003) may provide some clues. Those slices showed massive disorganization of the layering; however, complementary cell tracing experiments, necessary to fully understand this phenotype, were not performed. In humans, cortical migration disorders caused by Reelin signaling deficiency lead to classical lissencephaly (Hong et al. 2000; Boycott et al. 2005; Chang et al. 2007; Zaki et al. 2007), a condition characterized by a paucity of cortical gyration, leading to severe epilepsy and mental retardation. Impairment of the down-regulation component of Reelin signaling is thus predicted to cause strong cortical developmental defects as well. A precise description of the phenotype of the Cul5-deficient mice cortex may pinpoint human brain abnormalities linked to this mechanism.
Acknowledgments
G.K. is supported by the European Molecular Biology Organization (EMBO).
Footnotes
-
↵1 Corresponding author.
↵1 E-MAIL jogleeson{at}ucsd.edu; FAX (858) 534-1437.
-
Article is online at http://www.genesdev.org/cgi/doi/10.1101/gad.1622907
- Copyright © 2007, Cold Spring Harbor Laboratory Press
