Disabled-1 (Dab1) is an essential intracellular protein in the Reelin pathway. It has a nuclear localization signal (NLS; hereafter referred to as “NLS1”) and 2 nuclear export signals, and shuttles between the nucleus and the cytoplasm. In this study, we found that Dab1 has an additional unidentified NLS, and that the Dab1 NLS1 mutant could translocate to the nucleus in an unconventional ATP/temperature-dependent and cytoplasmic factor/RanGTP gradient-independent manner. Additional mutations in the NLS1 mutant revealed that K67 and K69 are important for the nuclear transport. Furthermore, an excess of the intracellular domain of the Reelin receptors inhibited the nuclear translocation of Dab1. An in utero electroporation study showed that a large amount of Dab1 in the cytoplasm in migrating neurons inhibited the migration, and that forced transport of Dab1 into the nucleus attenuated this inhibitory effect. In addition, rescue experiments using yotari, an autosomal recessive mutant of dab1, revealed that cells expressing Dab1 NLS1 mutant tend to distribute at more superficial positions than those expressing wild-type Dab1. Taken together, these findings suggest that Dab1 has at least 2 NLSs, and that the regulation of the subcellular localization of Dab1 is important for the proper migration of excitatory neurons.
Excitatory neurons of the cerebral cortex are produced in the proliferative zone and migrate radially toward the brain surface (Angevine and Sidman 1961; Berry and Rogers 1965; Rakic 1972; Nadarajah et al. 2001; Tabata and Nakajima 2003; Ayala et al. 2007; Tabata et al. 2012; Evsyukova et al. 2013), passing through early-born neurons, stopping their migration beneath the marginal zone (MZ) to form a cortical plate (CP), where they begin to differentiate the dendrites (Kawauchi et al. 2010; Franco et al. 2011; Jossin and Cooper 2011; Sekine et al. 2012, 2014; Cooper 2013; Gil-Sanz et al. 2013). Therefore, CP neurons eventually reside in a so-called inside-out sequence, with superficial positions populated by late-born neurons and deep positions populated by early-born neurons (Caviness 1982). The secreted glycoprotein Reelin has critically important roles in these processes, because the disruption of this gene causes a roughly inverted positioning of the neurons (Falconer 1951; Caviness 1982; Goffinet 1983; Bar et al. 1995; D'Arcangelo et al. 1995; Ogawa et al. 1995; Nakajima et al. 1997; D'Arcangelo 2005; Honda et al. 2011) and aberrant dendrite formation (Niu et al. 2004; Kohno et al. 2015). Reelin signal transmission to the cytoplasm is mediated by apolipoprotein E receptor 2 (ApoER2) or very low-density lipoprotein receptor (VLDLR; D'Arcangelo et al. 1999; Hiesberger et al. 1999; Trommsdorff et al. 1999; Hirota et al. 2015). The cytoplasmic protein Disabled-1 (Dab1) then transmits the signal to downstream pathways (Rice et al. 1998; Howell et al. 1999). As dab1 mutations cause a phenocopy of the reeler mutation (reelin mutant mice; Howell et al. 1997; Sheldon et al. 1997; Ware et al. 1997; Yoneshima et al. 1997; Kojima et al. 2000), an understanding of the physiological functions of Dab1 is important (Olson et al. 2006; Matsuki et al. 2008; Franco et al. 2011; Sekine et al. 2011, 2012). Previously, we showed that Dab1 is a nucleocytoplasmic shuttling protein with a nuclear localization signal (NLS) and 2 nuclear export signals (NESs; Honda and Nakajima 2006), suggesting that Dab1 may have a function in the nucleus and/or that nucleocytoplasmic shuttling itself might play some role in Reelin signaling.
In eukaryotes, some proteins must be actively transported across the nuclear pore complex (NPC) for signal transduction into the nucleus (Raices and D'Angelo 2012), with the protein's NLS binding either to soluble transporters or directly to proteins within the NPC itself (Xu and Massague 2004). The most frequently identified NLS is the “classical” NLS (Kalderon et al. 1984), and the soluble nuclear transporters Importin α and β (Gorlich and Kutay 1999; Cook et al. 2007) are mainly responsible for mediating nuclear transport (Gorlich and Kutay 1999). In contrast, a smaller but significant number of proteins can be translocated into the nucleus without the involvement of cytosolic soluble transporters (Xu and Massague 2004).
In this study, we found that the Dab1 NLS1 mutant could translocate into the nucleus in a soluble factor-independent manner, suggesting that Dab1 has at least one other unidentified NLS. An overexpression study of Dab1 in migrating neurons showed that an increase in Dab1 protein in the cytoplasm inhibited neuronal migration, whereas forced transport into the nucleus reduced this inhibitory effect. Migration rescue experiments using yotari mice revealed that NLS1 mutation of Dab1 caused changes in the final positioning of rescued neurons. Taken together, these results suggest that the active nuclear transport of Dab1 via NLS1 and an unidentified NLS may control the amount of Dab1 in the cytoplasm, thereby regulating the proper migration of excitatory neurons.
Materials and Methods
Pregnant ICR mice were purchased from Japan SLC (Hamamatsu, Japan). The day on which a vaginal plug was detected was designated as embryonic day 0 (E0). All the animal experiments were performed under the control of the Keio University Institutional Animal Care and Use Committee, in accordance with the Institutional Guidelines on Animal Experimentation at Keio University, the Japanese Government Law Concerning the Protection and Control of Animals, and the Japanese Government Notification of Feeding and Safekeeping of Animals.
All plasmid vectors used in this study are described in Supplementary Materials and Methods.
Leptomycin B Treatment and Examination of Subcellular Localization
For the examination of the subcellular localization of various non-tagged or EGFP-tagged Dab1 in the presence or absence of leptomycin B (LMB; Sigma-Aldrich, St Louis, MO, USA), Neuro2a cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum on a 35-mm dish, and plasmid DNA [(1) 1.25 μg of Dab1-EGFP or Dab1 expression vector shown in Figs 1, 4, 5A,C, and 7B, and Supplementary Figs 1–3 or (2) 0.1 μg of Dab1-EGFP expression vector and 1 μg of GST-ApoER2/VLDLR ICD-HA expression vector, or 3.5 μg of pCAGGS-GST-HA shown in Fig. 8B–J] were transfected with Superfect transfection reagent (Qiagen, Valencia, CA, USA). As for pCAGGS-GST-HA, the amount of transfected plasmid DNA was increased to adjust the protein expression level to those of the other GST/HA-tagged proteins. Approximately 36 h after transfection, LMB was added to the medium at 10 ng/mL and was incubated for 1 or 12 h. As a vehicle control, the same volume of methanol was added instead of the LMB. For the detection of non-tagged Dab1 protein, the cells were incubated with rabbit anti-Dab1 antibody (Millipore) at room temperature (RT) for 1 h, and anti-rabbit IgG Alexa Fluor 488 (Life Technologies, Grand Island, NY, USA) at RT for 1 h. To analyze the pattern of the subcellular localization of Dab1 proteins, we classified the subcellular localization pattern into 3 categories: C, the fluorescent signal in the cytoplasm was more intense than that in the nucleoplasm; N, the fluorescent signal in the nucleoplasm was more intense than that in the cytoplasm; or C = N, the fluorescent signals in the cytoplasm and the nucleus were approximately equal.
One microgram of each plasmid DNA was transfected into HEK293T cells on 35-mm dishes using Superfect DNA transfection reagent (Qiagen). After 2 days, the cells were collected, lysed using a cell lysis buffer (50 mM Tris–HCl [pH 7.4], 150 mM NaCl, 1% Nonidet P-40 [NP-40], 5 mM MgCl2, 50 mM NaF, 1 mM Na3VO4, and 2 mM dithiothreitol [DTT]) containing a protease inhibitor mixture (Roche Applied Science, Mannheim, Germany), and then clarified using centrifugation. To immunoprecipitate the EGFP-tagged protein, the lysate was incubated with a rabbit anti-GFP antibody (MBL, Nagoya, Japan) at 4 °C for 1 h on a rotator, then protein G agarose (Thermo Fisher Scientific, Waltham, MA, USA) was added to the reaction and rotated for an additional 1 h. The beads were then washed 3 times with the cell lysis buffer. The immunoprecipitated complex was dissolved using SDS sample buffer, separated using SDS–PAGE, and electroblotted onto a membrane. To detect the precipitated ApoER2-HA, the membrane was reacted with a mouse anti-HA antibody (Covance, Princeton, NJ, USA), followed by a reaction with horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). The bands were visualized using a BM chemiluminescence western blotting substrate (Roche Applied Science), and the chemiluminescence was detected using an ImageQuant LAS 4000 Mini (GE Healthcare Life Sciences, Buckinghamshire, UK). The same membrane was then stripped with stripping buffer (Thermo Scientific), reacted with rabbit anti-GFP antibody (MBL), incubated with HRP-conjugated anti-rabbit IgG (Jackson ImmunoResearch Laboratories), and visualized. To detect the expression level of ApoER2-HA and the EGFP-tagged proteins, 1% volume of the input lysate was analyzed using western blotting, as described above.
One microgram of each plasmid DNA, except for pCAGGS-GST-HA, was transfected into HEK293T cells (Fig. 8A). For pCAGGS-GST-HA, 3.5 μg of the plasmid DNA was transfected to adjust the relative protein expression level to that of the other proteins. Two days after the transfection, the cells were lysed using a cell lysis buffer (50 mM Tris–HCl, pH 7.4, 1% NP-40, 150 mM NaCl, 2 mM DTT, and 0.4% skim milk) containing a protease inhibitor mixture (Roche Applied Science) and cleared using centrifugation; the supernatants were then collected. To pull down the GST-fusion proteins and their interacting proteins, the lysate was reacted with glutathione Sepharose 4B beads (GE Healthcare Life Sciences) at 4°C for 1 h. After 3 washes using cell lysis buffer, the binding proteins and 1% volume of the input lysate were dissolved in SDS sample buffer, then analyzed as described above.
Plasmid Transfection into Cerebral Cortical Neurons and Primary Cultures
To observe the subcellular localization of Dab1-EGFP-fusion proteins in primary neurons, exo utero electroporation was performed and primary cultures were prepared as described previously (Honda and Nakajima 2006).
Purification of Proteins
GST-tagged and His-tagged proteins were obtained as previously described (Honda and Nakajima 2006). The purified proteins were then dialyzed against complete transport buffer (CTB: 20 mM HEPES–NaOH [pH 7.4], 110 mM potassium acetate, 2 mM magnesium acetate, 0.5 mM EGTA, 2 mM DTT, 1 μg/mL of aprotinin, 1 μg/mL of pepstatin, and 1 μg/mL of leupeptin) at 4 °C through a microdialyzer, then stored at 4 °C until use. As to RanQ69L, the purified protein was dialyzed against 0.1 mM GTP-containing CTB.
Nuclear Import Assay
The nuclear import assay was performed as described previously (Kehlenbach and Paschal 2006), with some modifications. HeLa S3 cells were plated on a 35-mm dish containing glass coverslips at 5 × 105 cells/dish and were cultured for approximately 2 days. The coverslips were then transferred to a 35-mm dish containing ice-cold CTB and placed on ice. The cells were washed 3 times with ice-cold CTB and were then incubated with 40 μg/mL of digitonin-containing CTB for 5 min on ice. The cells were washed 3 times with ice-cold CTB. The coverslips were transferred to a 35-mm dish containing prewarmed CTB (30 °C) and were incubated at 30 °C on a heat block for 15 min; the cells were then washed 5 times with prewarmed CTB (30 °C), and the coverslips were transferred to a 35-mm dish containing ice-cold CTB on ice. The coverslips were then transferred to a Parafilm-lined, humidified, 10-cm Petri dish, and 25 μL of import reaction mix consisting of import substrate protein (∼0.2 μM), an energy-generating system (1 mM ATP, 1 mM GTP, 5 mM creatinine phosphate, and 20 units/mL of creatinine phosphokinase), 50% (v/v) rabbit reticulocyte lysate, 2 mM DTT, 5 mg/mL of bovine serum albumin (BSA), 20 mM HEPES–NaOH (pH 7.4), 110 mM potassium acetate, 2 mM magnesium acetate, and 0.5 mM EGTA was added to each coverslip. The coverslips were then incubated at 30 °C in a water bath or at 0 °C on ice for 20 min. To test for ATP-dependency, the energy-generating system was omitted and 100 units/mL of apyrase (Sigma-Aldrich) was added to the import reaction mix. Cytosol dependency was tested by exchanging the rabbit reticulocyte lysate for distilled water. After 20 min of incubation, the import reaction mix was absorbed using a Kimwipe and the cells were washed in ice-cold CTB 3 times; the cells were then fixed using 4% paraformaldehyde (PFA) in a 0.1-M sodium phosphate buffer (pH 7.4; 4% PFA) at RT for 5 min. The cells were washed using phosphate-buffered saline (PBS) and stained using 4′,6-diamidino-2-phenylindole (DAPI, Life Technologies). For wheat germ agglutinin (WGA, Sigma-Aldrich) and the RanQ69L-GTP assay, digitonin-permeabilized cells were washed 5 times and incubated with CTB for the control, 2 μg/μL of WGA, or approximately 2 μM of RanQ69L supplemented with 2-mM GTP at 30 °C for 15 min. Then, the buffers were removed, and the cells were incubated with the import reaction mixture, 2 μg/μL of WGA, or approximately 2 μM of RanQ69L-GTP containing the import reaction mixture at 30 °C on a water bath for 20 min.
In Utero Electroporation
In utero electroporation was performed using E14.5 ICR (Fig. 9 and Supplementary Fig. 6) or yotari (Fig. 10 and Supplementary Fig. 4) mouse embryos, as described previously (Tabata and Nakajima 2001, 2008). Briefly, pregnant mice were deeply anesthetized with pentobarbital sodium, and the uterus was exposed. For the Dab1 overexpression experiment (Fig. 9), plasmid DNA solutions containing (1) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pCAGGS1, (2) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pCAGGS-Dab1-HA, or (3) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pCAGGS-Dab1NES1/2mt-HA were prepared. For the yotari-rescue experiment (Fig. 10 and SupplementaryFig. 4), plasmid DNA solutions containing (1) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pTα1, (2) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pTα1-Dab1-HA, (3) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pTα1-Dab1NLS1mt-HA, or (4) 2.5 mg/mL of pCAGGS-EGFP and 2.5 mg/mL of pTα1-Dab1NLS1 + Cmt-HA were prepared. For the Dab1 and Notch intracellular domain (NICD) overexpression experiment (see Supplementary Fig. 6), plasmid DNA solutions containing (1) 2.5 mg/mL of pCAGGS-EGFP, 2.5 mg/mL of pCAGGS-Dab1-HA, and 2.5 mg/mL of pTα1, or (2) 2.5 mg/mL of pCAGGS-EGFP, 2.5 mg/mL of pCAGGS-Dab1-HA, and 2.5 mg/mL of pTα1-Myc-NICD were prepared. After addition of 0.1% Fast green at 1/10 volume, the plasmid DNA solutions were injected into the lateral ventricle using a glass capillary. Immediately after the injection, 50-ms, 34-V electronic pulses were applied 4 times at 950-ms intervals using an electroporator (NEPA21; NEPA Gene, Chiba, Japan) with a forceps-type electrode (CUY650P5; NEPA Gene). After electroporation, the uterus was placed back into the abdominal cavity, and the embryos were allowed to develop normally.
Preparation of Sections and Immunohistochemistry
On E18.5 or postnatal day 0 (P0), mice were anesthetized and transcardially fixed with 4% PFA. The brains were then removed and post-fixed at 4 °C overnight. The fixed brains were washed with PBS, soaked in 10, 20, and 30% sucrose-containing PBS in a stepwise fashion, and then embedded in OCT compound (Sakura Finetek, Tokyo, Japan). The brains were sectioned coronally at a thickness of 20 μm using a cryostat (Leica Microsystems, Germany). The sections were washed 3 times in PBS containing 0.1% Tween 20 (PBStw) at RT for 5 min each time, then treated with 5% BSA-containing PBStw (5% BSA/PBStw) for 5 min at RT. For the detection of the CAG-driven HA-tagged Dab1 (Fig. 9), the sections were incubated with rat anti-HA antibody (3F10; Roche Applied Science) at RT for 1 h. After 3 washes, the sections were incubated with anti-rat IgG Alexa Fluor 594 (Life Technologies). As for the detection of the Tα1-driven Dab1-HA proteins or endogenous Dab1 (Fig. 10 and Supplementary Fig. 4), the sections were treated with rabbit anti-Dab1 antibody (Millipore) at RT for 1 h, biotynylated anti-rabbit IgG (Vector Lab or Jackson ImmunoResearch Laboratories) at RT for 1 h, and Dylight 594 Streptavidin (Jackson ImmunoResearch Laboratories) at RT for 30 min. To detect the Myc-NICD (see Supplementary Fig. 6), the sections were treated with mouse anti-Myc-antibody (9E10; Santa Cruz Biotechnology, Santa Cruz, CA, USA) at RT for 1h, and anti-mouse IgG Alexa Fluor 594 (Life Technologies) at RT for 1 h.
Measurement of Relative Bin Positions of EGFP-Labeled Cells
To analyze the positioning of the neurons quantitatively, single focal plane images were taken around the caudal visual cortex level (Fig. 9) or anterior commissure level (Fig. 10) using an FV1000 laser scanning confocal microscope (Olympus, Tokyo, Japan). We then measured the distance from the ventricular surface to the EGFP-labeled neurons (D1, see Figs 9F and 10I) and the distance from the ventricular surface to the border between the CP and the MZ (D2 shown in Fig. 9F) or top of the CP (D2 shown in Fig. 10I) through the EGFP-labeled cells using the ImageJ software, and calculated the relative bin position by dividing D1 by D2 and multiplying by 10. The relative bin positions were defined as follows: 0 ≤ bin 1 < 1, 1 ≤ bin 2 < 2, 2 ≤ bin 3 < 3, 3 ≤ bin 4 < 4, 4 ≤ bin 5 < 5, 5 ≤ bin 6 < 6, 6 ≤ bin 7 < 7, 8 ≤ bin 8 < 8, 8 ≤ bin 9 < 9, 9 ≤ bin 10 < 10. For the statistical analysis, more than 100 EGFP-labeled cells obtained from 3 independent experiments were measured, and the mean value for each bin of paired samples was analyzed by one-way ANOVA with Tukey's post hoc test using SPSS Statistics for Mac, Version 22 (IBM, Armonk, NY, USA).
Dab1 with Mutations in the NLS1 Sequence Can Translocate into the Nucleus
In our previous study, we reported that Dab1 is a nucleocytoplasmic shuttling protein containing a bipartite NLS sequence (NLS1) and 2 NESs (NES1 and NES2), and that the transport of Dab1 from the nucleoplasm to the cytoplasm is mainly mediated by chromosomal region maintenance 1 (CRM1)/Exportin1 (Honda and Nakajima 2006). The microinjection of a fusion protein GST/EGFP-tagged Dab1 NLS1 sequence (RKKGQDRSEATLIKRFK) (GST-Dab1NLS1-EGFP) into Neuro2a cells showed the nuclear transport of the protein into the nucleus, whereas the introduction of mutations into this NLS1 sequence significantly attenuated the nuclear transport activity, indicating that Dab1 NLS1 is functional and sufficient for the translocation of other unrelated proteins into the nucleus (Honda and Nakajima 2006). Since Dab1 does not contain any other typical NLS sequences like those of SV40 T-antigen (Kalderon et al. 1984) or nucleoplasmin (Robbins et al. 1991), we originally speculated that the NLS1 is the only NLS for the Dab1 to enter the nucleus. However, we found that, when the Dab1 NLS1-mutant-expressing cells were treated with LMB, an inhibitor of CRM1-dependent nuclear export (Wolff et al. 1997), or methanol (as a vehicle control) for 1 h (see Supplementary Fig. 1A,B) or 12 h (Fig. 1A,B), the mutant Dab1 proteins could translocate into and accumulate in the nucleus in both cases. In addition, a non-tagged version of the Dab1 NLS1 mutant also showed a similar nuclear accumulation pattern (Fig. 1C,D). These results suggest that Dab1 is transported into the nucleus via an unidentified NLS, in addition to the previously identified NLS1.
Dab1 NLS1 Mutant Is Transported into the Nucleus via a Temperature-Dependent/ATP-Dependent and a Cytosol-Independent/RanGTP Gradient-Independent Manner
To characterize the nuclear translocation property of the Dab1 NLS1 sequence and the full-length Dab1 NLS1 mutant, a digitonin-permeabilized cell assay (Adam et al. 1990; Kehlenbach and Paschal 2006) was performed using Hela S3 cells (Fig. 2). Digitonin is a detergent and can selectively permeabilize the plasma membrane without affecting the nuclear envelope. Although digitonin permeabilization induces the release of cytosolic soluble factors from the cytoplasmic compartment, nuclear import can be reconstituted through the addition of exogenous factors. Using this assay system, the properties of nuclear import can be classified into at least 2 groups: A cytosolic soluble factor-dependent nuclear import pathway and a soluble factor-independent pathway (Xu and Massague 2004). To examine the nuclear transport properties of Dab1 wild-type NLS1 and the full-length Dab1 NLS1 mutant in permeabilized Hela S3 cells, a GST-Dab1NLS1-EGFP-fusion protein and an EGFP/His-tagged full-length Dab1 NLS1-mutant fusion protein (Dab1NLS1mt-EGFP-His6) were prepared (Fig. 2A). For comparison, the wild-type version of the full-length Dab1-fusion protein (Dab1-EGFP-His6) was also prepared. In addition, as a positive control for Importin-dependent nuclear import, a GST/EGFP-tagged SV40 NLS-fusion protein (GST-SV40NLS-EGFP) was prepared. When digitonin-permeabilized Hela S3 cells were incubated with GST-SV40NLS-EGFP, the fluorescent signal was observed mainly in the nucleus (Fig. 2B,B′). Although a microinjection experiment of GST-Dab1NLS1-EGFP into Neuro2a cells showed the nuclear transport of the protein into the nucleus (Honda and Nakajima 2006), we were unable to observe the nuclear translocation of this protein in this assay using HeLa S3 cells (Fig. 2D,D′). Incubation with Dab1-EGFP-His6 or Dab1NLS1mt-EGFP-His6 resulted in a diffuse distribution in both the nucleus and the cytoplasm (Fig. 2K,K′,O,O′). When the cells were incubated on ice or in the absence of ATP, the nuclear translocations of GST-SV40NLS-EGFP and both Dab1-EGFP-fusion proteins were diminished (Fig. 2F, F′, J, J′, N, N′, H, H′, L, L′, P,P′). In the absence of cytoplasm, the nuclear translocation of GST-SV40NLS-EGFP was not observed (Fig. 2I,I′), whereas both full-length Dab1-fusion proteins, Dab1-EGFP-His6 and Dab1NLS1mt-EGFP-His6, were translocated into the nucleus (Fig. 2M,M′,Q,Q′).
Next, we examined whether the nuclear translocation of the full-length Dab1-fusion proteins occurs through the NPC, rather than by simple diffusion, using WGA. WGA is an inhibitor of NPC-mediated nuclear translocation that does not affect passive diffusion (Finlay et al. 1987; Yoneda et al. 1987; Wolff et al. 1988). As the above experiment showed that Dab1-EGFP-fusion proteins do not require cytosolic factors, in this experiment, nuclear import assays of Dab1-EGFP-fusion proteins were performed without cytosolic soluble factors. As a result, the addition of WGA significantly blocked the nuclear translocation of all the EGFP-fusion proteins, indicating that nuclear translocation is mediated through the NPC (Fig. 3B,B′,E,E′,H,H′). Of note, it appeared that wild-type Dab1 was trapped in the cytoplasm in the WGA-treated cells (Fig. 3E,E′), which was not observed for GST-SV40NLS-EGFP (Fig. 3B,B′) or Dab1NLS1mt-EGFP-fusion protein (Fig. 3H,H′; see also “Discussion”).
Furthermore, we examined whether the nuclear translocation of the Dab1-fusion proteins requires a GTP-bound Ran (RanGTP) gradient. For the Importin-dependent nuclear translocation of classical NLS-containing proteins, the GTPase family member Ran is essentially required (Gorlich and Kutay 1999), since RanGTP is abundant in the nucleus and promotes the dissociation of an NLS-containing cargo-Importin α/β ternary complex in the nucleus, providing the directionality of the transport system. To disturb the RanGTP gradient, we pretreated cells with RanQ69L-GTP. RanQ69L is a GTPase activity-deficient mutant of Ran and thus exists predominantly as a GTP-bound form (Bischoff et al. 1994). Similar to the above experiment, cytosolic soluble factors were removed for the Dab1-EGFP-fusion protein import assay. Consequently, RanQ69L-GTP treatment efficiently blocked the nuclear translocation of the SV40NLS-fusion protein (Fig. 3C,C′), but not that of the Dab1-fusion proteins (Fig. 3F,F′,I,I′).
Altogether, these results suggest that Dab1 has at least 2 independent nuclear transport pathways: one is the N-terminal NLS1-dependent pathway, and the other is the unidentified NLS-dependent pathway, which depends on ATP and temperature but not on the RanGTP gradient and cytoplasmic soluble factors.
Mutations at Both NLS1 and K67/K69 Diminish the Nuclear Translocation of Dab1
We next examined which part of Dab1 is important for the nuclear translocation of the Dab1 NLS1-mutant protein. As mentioned before, Dab1 does not have any obvious classical NLS motif except for the NLS1, suggesting that Dab1 might have a non-classical NLS. As non-classical NLSs reportedly have a tendency to possess relatively high amounts of lysine and arginine residues (Nguyen Ba et al. 2009), we hypothesized that the unidentified NLS(s) of Dab1 might also have such a tendency. We therefore searched for basic amino acid clusters in the Dab1 primary sequence and replaced them with neutral alanine residues in various patterns (Fig. 4A,B). All the Dab1-mutant proteins were expressed in Neuro2a cells as a fusion protein with EGFP, and we observed the subcellular localization patterns in the presence or absence of LMB for 1 h (see Supplementary Fig. 2) or 12 h (Fig. 4). As a result, when mutations were introduced into the basic amino acid cluster C (K67 and K69) or E (K165 and R169), the nuclear accumulation of the Dab1-mutant protein was significantly inhibited at both 1 and 12 h (see Supplementary Fig. 2C,D and Fig. 4C,D, NLS1 + ABCFH, NLS1 + ABCDFH, and NLS1 + ABEFGH, marked by asterisks) of incubation.
Next, we examined whether the simultaneous mutations in both NLS1 and basic amino acid cluster C or E are sufficient to inhibit Dab1 nuclear translocation. Mutations in both NLS1 and basic amino acid cluster C, but not E, inhibited the nuclear translocation of the Dab1-EGFP-fusion protein (Fig. 5A,B). On the other hand, when only the basic amino acid cluster C was mutated, the nuclear translocation ability of the Dab1 mutant was not affected (Fig. 5A,B). These results were confirmed by using non-tagged Dab1 NLS1 + C mutant and Dab1 C mutant (Fig. 5C,D). In addition, we obtained similar results using the Dab1-EGFP-fusion proteins in the HEK293T cells (data not shown) and primary neurons (Fig. 5E,F). These results indicate that the basic amino acid cluster C contains amino acids that are essential for the nuclear import of the Dab1 NLS1 mutant, raising the possibility that the basic amino acid cluster C is part of an unidentified NLS and that 2 independent nuclear import pathways exist for Dab1.
Simultaneous Mutations in the NLS1 and the Basic Amino Acid Cluster C of Dab1 Abolish its Ability to Bind to the Reelin Receptor
To examine whether the Dab1 mutants can bind to Reelin receptors, we co-transfected Dab1-EGFP and HA-tagged ApoER2 expression vector into HEK293T cells. Two days after the transfection, Dab1-EGFP was immunoprecipitated and immunoblotted using an antibody for HA. As a result, ApoER2-HA was coimmunoprecipitated with wild-type Dab1, the Dab1 NLS1 mutant, and the Dab1 C mutant. However, the Dab1 NLS1 + C mutant did not bind to ApoER2-HA (Fig. 6), suggesting that the Dab1 structure required for receptor binding is abolished by the double mutations in NLS1 and cluster C. These results raised the possibility that the binding of Dab1 to ApoER2 might be required for the cluster C-dependent (unidentified NLS-dependent) nuclear translocation of Dab1.
If binding between Dab1 and ApoER2 itself is required for the nuclear transport activity of the unidentified NLS, mutations in the critical amino acid residues of Dab1 for binding to the Reelin receptor would also inhibit the nuclear translocation of the NLS1 mutant of Dab1. As the previous study showed that S114 and F158 of Dab1 are essential amino acids for Dab1 binding to the NPxY motif of ApoER2 (Yun et al. 2003), we next constructed Dab1 S114T and Dab1 S114T/F158V mutants (Dab1S114Tmt-EGFP and Dab1S114T/F158Vmt-EGFP). In addition, we introduced an additional mutation into the NLS1 sequence of these mutants (Dab1NLS1 + S114Tmt-EGFP and Dab1NLS1 + S114T/F158Vmt-EGFP). Consistent with a previous report (Yun et al. 2003), these Dab1 mutants lost their ability to bind to ApoER2 (Fig. 7A). However, when they were treated with LMB for 1 h (see Supplementary Fig. 3) or 12 h (Fig. 7B,C), all the mutants had accumulated in the nucleus. These results show that the inhibition of binding between Dab1 and ApoER2 itself does not affect the function of the unidentified Dab1 NLS, suggesting that the failure of the NLS1 + C mutant to undergo nuclear translocation was not caused simply by the defect in receptor binding.
ApoER2 ICD and VLDLR ICD Inhibit the Nuclear Translocation of Dab1
Several reports have shown that ApoER2 ICD or VLDLR ICD is cleaved and released into the cytoplasm (May et al. 2003; Hoe and Rebeck 2005; Wagner et al. 2013). As for ApoER2, Reelin stimulation significantly increased its ICD production (Hoe and Rebeck 2005; Hoe et al. 2006). Considering that a soluble cytoplasmic fragment of the C-cadherin ICD is known to inhibit the nuclear translocation of β-catenin (Suh and Gumbiner 2003), the cytoplasmic ICD of the Reelin receptors might also affect the subcellular localization (nucleocytoplasmic shuttling) of Dab1.
To address this issue, we constructed GST/HA-tagged ApoER2 or VLDLR ICD (GST-ApoER2 ICD-HA, GST-VLDLR ICD-HA) expression vectors and also prepared a GST-HA expression vector as a negative control. First, we examined whether GST-ApoER2 ICD-HA and GST-VLDLR ICD-HA bind with Dab1-EGFP-fusion proteins using HEK293T cells by a GST pull down assay (Fig. 8A). As a result, both ICD proteins were able to bind with wild-type Dab1 and the Dab1 NLS1 mutant. However, the binding between the Dab1 C mutant and the receptor tails was significantly reduced, suggesting the importance of the extracellular domain of the receptors for their stable binding to Dab1 (Fig. 6, C mutant).
Next, we co-transfected the Dab1-EGFP expression vectors and the GST/HA-tagged protein expression vectors into Neuro2a cells, and the cells were treated with methanol or LMB for 12 h (Fig. 8B–K). As a result, the nuclear translocation of both Dab1-EGFP and Dab1NLS1mt-EGFP was significantly inhibited by the co-expression of GST-ApoER2 ICD-HA or GST-VLDLR ICD-HA (Fig. 8C,D,F,G,K), whereas the inhibition of the nuclear translocation of the Dab1 C mutant was not observed (Fig. 8I,J,K). While it is not clear whether a single or both NLS function(s) were inhibited in the Dab1-EGFP, at least for the case of Dab1NLS1mt-EGFP, it is plausible that the unidentified NLS-dependent nuclear transport pathway was inhibited by the ICDs. Therefore, an excess amount of ICDs of the Reelin receptors appeared to have the ability to inhibit at least the unidentified NLS function, suggesting that proper turnover of the receptor ICDs might be important for appropriate nuclear translocation of Dab1.
Excessive Amount of Dab1 in the Cytoplasm, But Not in the Nucleoplasm, Inhibits Radial Neuronal Migration in the Cerebral Cortex
So far, many reports have shown that Dab1 is endogenously involved in the neuronal migration (Olson et al. 2006; Franco et al. 2011; Sekine et al. 2011, 2012) and dendrite formation (Niu et al. 2004; Olson et al. 2006; Kubo et al. 2010) of excitatory neurons. However, no information is available regarding how Dab1 nucleocytoplasmic shuttling is involved in these events, and the subcellular localization of endogenous Dab1 is not known due to technical difficulties (see Supplementary Fig. 4). To explore the physiological role of the nucleocytoplasmic shuttling of Dab1, wild-type Dab1-HA and Dab1NES1/2mt-HA, which contain mutations in each of the 2 NESs, were overexpressed in cerebral cortical neurons using in utero electroporation at E14.5, and the effect on their radial migration was investigated at P0 (Fig. 9). Immunohistochemical staining for HA revealed that the Dab1-HA-fusion protein was almost exclusively localized in the cytoplasm (Fig. 9B,D–D″), whereas Dab1NES1/2mt-HA was localized mainly in the nucleus (Fig. 9C,E–E″). When an empty vector was co-expressed with an EGFP expression vector, most EGFP-positive cells migrated to the top of the CP, as shown by the relative bin position 10 (Fig. 9A,F,G). However, the overexpression of Dab1-HA severely impaired the neuronal migration (Fig. 9B,G). Most of the GFP-positive cells were stalled between the intermediate zone and the middle region of the CP. On the other hand, the overexpression of Dab1NES1/2mt-HA did not clearly affect the neuronal migration (Fig. 9C,G). These results indicated that an excessive amount of cytoplasmic Dab1 inhibits radial neuronal migration, and that forced nuclear transport can attenuate its inhibitory effect on migration.
Overexpression of Wild-Type Dab1 and NLS1-Mutant Dab1 Result in Different Migration Properties of Neurons in the Most Superficial Part of the Yotari Cortex
To further examine the physiological function of nuclear translocation of Dab1, we next evaluated the rescue ability of Dab1 for neuronal migration in the cerebral cortex of a dab1-mutant mouse, yotari. Previous studies have shown that the defective migration in the dab1-mutant cortex (Howell et al. 1997; Hammond et al. 2001; Tabata and Nakajima 2002) was rescued by exogenously expressed Dab1 (Sanada et al. 2004; Morimura and Ogawa 2009; Simo et al. 2010). Thus, we examined the differences in the migration rescue effect among the Dab1 mutants using this assay system. At E14.5, empty, HA-tagged wild-type Dab1 (Dab1-HA), Dab1 NLS1 mutant (Dab1NLS1mt-HA), or Dab1 NLS1 + C mutant (Dab1NLS1 + Cmt-HA) expression vectors were co-introduced with the GFP expression vector into the yotari cortex, and the brains were fixed at E18.5. In the absence of Dab1, GFP-positive cells were distributed between the intermediate zone and middle part of the cortex (Fig. 10A,E,E′, and “gray bar” in J), and never migrated into the superficial part of the cortex, as shown by the absence of any gray bars in the relative bin positions 9 and 10 (Fig. 10J). In contrast, in the presence of exogenously expressed wild-type Dab1 (Fig. 10B,F–F″, “black bar” in J), a part of the GFP-positive cells could migrate to the more superficial part of the cortex as indicated by the black bar in the relative bin positions 9 and 10 (Fig. 10J). As to the Dab1 NLS1 mutant (Fig. 10C,G–G″, “blue bar” in J), GFP-positive cells were more abundantly observed in the most superficial part of the cortex, indicated by the relative bin position 10, than the wild-type Dab1-expressing cells (Fig. 10J). When Dab1 NLS1 + C mutant was expressed (Fig. 10D,H–H″, “magenta bar” in J), none of the GFP-positive cells were observed in the superficial part of the cortex shown in relative bin positions 9 and 10 (Fig. 10J), similar to the observation in the empty vector-transfected control experiments (Fig. 10A). These results show that overexpression of wild-type Dab1 and Dab1 NLS1 mutant caused different migration properties of the neurons in the yotari mice.
Subcellular Localization of NICD Is Not Influenced by the Presence of Wild-Type or Mutant Dab1, and Neuronally Expressed NICD Does Not Affect the Migration Defect Caused by Dab1 Overexpression
Recently, it has been reported that Dab1 forms a complex with NICD in mouse brains (Hashimoto-Torii et al. 2008) and in human neural progenitor cells (Keilani and Sugaya 2008); Notch1 and Dab1 colocalize in the dentate gyrus (Sibbe et al. 2009), and exogenously expressed NICD can rescue the neuronal migration failure in the reeler cerebral cortex (Hashimoto-Torii et al. 2008). Thus, it is possible that Dab1 mutants alter NICD localization. To verify this possibility, we examined whether the various Dab1 NLS or NES mutants can alter the subcellular localization of NICD. We transfected Neuro2a cells with NICD and Dab1 expression vectors either alone or in combination (see Supplementary Fig. 5). Examination by immunocytochemical staining for NICD showed that the subcellular localization of NICD was not influenced by the presence of any Dab1 proteins.
We next examined whether migration failure caused by overexpression of wild-type Dab1 could be suppressed by the co-expression of NICD, because it is possible that the interaction between Dab1 and NICD in vivo might be stronger than that in vitro and could affect the subcellular localization of Dab1. As Notch signaling is known to have critically important function in regulating neuronal progenitor differentiation (Yoon and Gaiano 2005), we used the Tα1 promoter to preferentially express the NICD in post-mitotic neurons (Gloster et al. 1994, 1999; Hashimoto-Torii et al. 2008). Then, the wild-type Dab1 expression vector was co-introduced into the cerebral cortex with the Tα1 empty vector or pTα1-HA-NICD vector at E14.5, and the mice were fixed at P0 (see Supplementary Fig. 6). While the overexpression of NICD could not be easily detected by immunohistochemistry, it was readily detectable in a pTα1-HA-NICD-transfected HEK293T cell by western blot analysis (data not shown), and NICD overexpression did not affect the neuronal migration failure caused by overexpression of wild-type Dab1 (see Supplementary Fig. 6B–B″).
In this study, we showed that the Dab1 NLS1 sequence and the full-length Dab1 NLS1 mutant have different properties for nuclear import in digitonin-permeabilized HeLa S3 cells, and that Dab1 has at least 2 different NLSs. We previously showed that the Dab1 NLS1 sequence alone could confer the ability to translocate into the nucleus to other unrelated proteins using a microinjection experiment with a Dab1 NLS1-fusion protein (GST-Dab1NLS1-EGFP) and Neuro2a cells (Honda and Nakajima 2006). However, in this study, we could not observe the nuclear translocation of the Dab1 NLS1-fusion protein using a nuclear import assay with HeLa S3 cells. This difference might have been caused by 4 major differences in assay conditions, including the cell type, cytoplasmic soluble factors, assay time, and assay temperature. The nuclear transporter for the Dab1 NLS1 existing in Neuro2a cells might not have been included in the rabbit reticulocyte lysate and HeLa S3 cells. The other possibility is that the nuclear import mediated by the Dab1 NLS1 may take a longer time to show its effect on the accumulation of the protein in the nucleus. This is because the HeLa S3 cells in the nuclear import assay and the Neuro2a cells in the previous microinjection study were incubated for 20 min at 30 °C and 2 h at 37 °C, respectively, before the observation of the subcellular localization of the protein. These experimental differences might have caused the different results for the nuclear import of GST-Dab1NLS1-EGFP.
Nuclear import assays showed that the full-length Dab1 NLS1 mutant can translocate into the nucleus in a temperature-dependent/ATP-dependent and a soluble factor-independent/RanGTP gradient-independent manner. This unconventional type of nuclear transport is very rare, and only a few studies have reported such mechanisms with regard to soluble factor-independency and ATP-dependency, such as those for hnRNP K (Michael et al. 1997) and HIV-1 Tat (Efthymiadis et al. 1998). The KHS domain (YDRRGRPGDRYDGMVGFSADETWDSAIDTWSPSEWQMAY) of hnRNP K (Michael et al. 1997) and the NLS sequence (GRKKRRQRRRAP) of the HIV-1 Tat (Efthymiadis et al. 1998) protein have been shown to confer a nuclear translocation ability to these respective proteins. However, no sequence similarity was found between these sequences and Dab1. Additionally, we found that Dab1-EGFP-His6 may be trapped in the cytoplasm in the WGA-treated cells (Fig. 3E,E′); this phenomenon was not observed in the case of GST-SV40NLS-EGFP and Dab1NLS1mt-EGFP-His6 (Fig. 3B,B′,H,H′). A similar tendency to that observed with Dab1-EGFP-His6 was also observed when the cells were treated with RanQ69L-GTP (Fig. 3F,F′), or even with a control CTB solution (Fig. 3D,D′), although the distribution pattern differed: relatively uniform distribution in the WGA-treated cells, but patchy distribution in the control CTB- or RanQ69L-GTP-treated cells. As the other proteins were clearly washed out from the cytoplasm, these results might suggest that wild-type Dab1, but not Dab1 NLS1 mutant, binds to some cytoplasmic structure under this experimental condition, and WGA might affect its distribution by an unknown mechanism.
Substitutional mutation analyses have shown that the nuclear translocation of the Dab1 NLS1 mutant can be diminished by additional mutations in the basic amino acid cluster C (K67 and K69). We first hypothesized that the basic amino acid-rich region may be used as an NLS and found that the basic amino acid cluster C was critically important for the nuclear translocation of Dab1. At present, however, we cannot exclude the possibility that the basic amino acid cluster C may not be part of the unidentified NLS, since the mutations in the Dab1 NLS1 sequence and the basic amino acid cluster C concurrently caused the inhibition of Dab1 nuclear translocation and the impairment of Dab1 binding to ApoER2. There are 2 possible interpretations for this observation. One possibility is that when the NLS1 sequence was mutated, the amino acids of the basic amino acid cluster C might have become critical amino acids for the binding to ApoER2 ICD; thus, the double mutation might have caused a simultaneous loss of function. The other possibility is that the simultaneous mutations might have caused a conformational change in Dab1, resulting in the concomitant impairment of nuclear transport and Reelin receptor binding. A plausible interpretation might be the latter explanation, since the ApoER2-binding residues of Dab1 suggested by a crystal structure analysis are not involved in either the NLS1 sequence or the basic amino acid cluster C (Stolt et al. 2003).
To examine the physiological significance of the nucleocytoplasmic shuttling of Dab1, we carried out in vivo experiments. Overexpression of Dab1 and Dab1 mutants using a CAG promoter in the wild-type cortex revealed that the cells overexpressing wild-type Dab1 stopped migrating in the deep part of the CP, whereas those overexpressing Dab1 NES1/2 mutant migrated to the top of the CP, similar to the observation in the control. Although the amount of Dab1 protein in an overexpressed situation is thought to be much higher than the endogenous level and we do not know how much cytoplasmic Dab1 is required to affect neuronal migration, these results might suggest that the active nuclear translocation of Dab1 is one of the regulatory mechanisms responsible for its function, similar to the situations for p42/p44 MAPKs (Volmat et al. 2001) or Net1 (Schmidt and Hall 2002).
The migration rescue experiment using the yotari mouse revealed that the cells expressing the Dab1 NLS1 mutant were distributed more superficially than those expressing wild-type Dab1. Dab1-deficient cells in the yotari mouse never migrated to the most superficial part of the cortex as indicated by the relative bin positions 9 and 10 (gray bar in Fig. 10J), and cells overexpressing wild-type Dab1 containing a large amount of Dab1 in the cytoplasm, but not in the nucleus, showed severe migration defect, resulting in a scarcity of GFP+ cells in bins 9 and 10 (“black bar” in Fig. 9G). Thus, it is speculated that the amount of cytoplasmic Dab1 protein in the migrating neurons has to be properly regulated during neuronal migration. Therefore, wild-type Dab1-expressing neurons in bins 9 and 10 in the yotari mouse (black bar in Fig. 10J) might contain Dab1 protein in the cytoplasm in sufficient and proper amounts for their migration. If this is the case, the presumed increase in the cytoplasmic expression of Dab1 protein or change in the balance of the Dab1 proteins between the cytoplasm and the nucleus caused by Dab1 NLS1 mutation might affect the efficiency of neuronal migration, and eventually cause a difference in the migration pattern of the Dab1-rescued neurons. However, at present, it is difficult to determine the amounts of Dab1 in the cytoplasm and nucleoplasm (see Supplementary Fig. 4), and which nuclear import pathway is used for Dab1 transport in vivo. As to the Dab1 NLS1 + C mutant, because it has already been reported that the loss of the Reelin receptor-binding ability of Dab1 by substitutional mutations in S114 and F158 causes loss of the ability of Dab1 to rescue the migration defect in yotari mice (Morimura and Ogawa 2009), it is plausible to think that one of the reasons for the lack of the migration-rescuing ability of the Dab1 NLS1 + C mutant was the inhibition of binding to the Reelin receptor. Thus, it is not clear whether the overall inhibition of the nuclear import of Dab1 might have any influence on the neuronal migration. Collectively, although further investigation is required, it is possible that the NLS1-dependent nuclear transport pathway of Dab1 might regulate the neuronal migration through modulating the amount of Dab1 protein in the cytoplasm.
In conclusion, we have shown that the Dab1 NLS1 mutant can translocate into the nucleus via an unconventional temperature-dependent/ATP-dependent and cytoplasmic soluble factor-independent/RanGTP gradient-independent manner, suggesting that Dab1 has 2 different nuclear translocation pathways. Since many proteins with critical functions in the nucleus seem to have evolved to possess multiple nuclear translocation pathways (Wagstaff and Jans 2009), Dab1 might also have evolved to maintain alternative pathways to ensure the nuclear translocation of Dab1. Further studies are needed to clarify the physiological meaning of the nuclear translocation of Dab1.
This work was supported by Grants-in-Aid for Scientific Research (Grant Numbers: 19700321, 21700383, 24700357, 15K06746, 15H01586, 15H02355, and 22111004) of Japan Society for the Promotion of Science, the Strategic Research Program for Brain Sciences (“Understanding of molecular and environmental bases for brain health”) of the Ministry of Education, Culture, Sports, Science and Technology of Japan, the Keio Gijuku Fukuzawa Memorial Fund for the Advancement of Education and Research, Research Grants for Life Sciences and Medicine of Keio University Medical Science Fund, the Keio Gijuku Academic Development Funds, Terumo Life Science Foundation, Takeda Science Foundation, and Life Science Foundation of Japan.
We thank Dr Toshifumi Morimura (Shiga University of Medical Science, Shiga, Japan) for providing the pSRα-FLAG-Dab1 (S114T) and pSRα-FLAG-Dab1 (S114T/F158V), Dr Junichi Miyazaki (Osaka University, Osaka, Japan) for the pCAGGS, Dr Jonathan A. Cooper (Fred Hutchinson Cancer Research Center, Seattle, WA, USA) for the pBS-mDab555, Dr Johannes Nimpf (Medical University of Vienna, Vienna, Austria) for the pMSCVpuro-mmApoER2(−), Dr Tokuo Yamamoto (Tohoku University, Sendai, Japan) for the pRC/CMV-mouse ApoER2, and Dr Dirk Gorlich (ZMBH, University of Heidelberg, Heidelberg, Germany) for the pQE32-RanQ69L, Dr Freda Miller (University of Toronto, Toronto, Canada) for p253, and Dr Yukiko Gotoh (The University of Tokyo, Tokyo, Japan) for the pEFBOS-FCDN. Conflict of Interest: None declared.