synergistic nuclear import of neurod1 and its partner transcription factor, e47, via...

14
Research Article Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization Rashid Mehmood a , Noriko Yasuhara a , Souichi Oe b , Masahiro Nagai b , Yoshihiro Yoneda a,b, a Department of Frontier Biosciences, Graduate School of Frontier Biosciences, Osaka University,1-3 Yamada-oka, Suita, Osaka 565-0871, Japan b Department of Biochemistry, Graduate School of Medicine, Osaka University, 2-2 Yamada-oka, Suita, Osaka 565-0871, Japan ARTICLE INFORMATION ABSTRACT Article Chronology: Received 1 July 2008 Revised version received 13 February 2009 Accepted 21 February 2009 Available online 9 March 2009 The transition from undifferentiated pluripotent cells to terminally differentiated neurons is coordinated by a repertoire of transcription factors. NeuroD1 is a type II basic helix loop helix (bHLH) transcription factor that plays critical roles in neuronal differentiation and maintenance in the central nervous system. Its dimerization with E47, a type I bHLH transcription factor, leads to the transcriptional regulation of target genes. Mounting evidence suggests that regulating the localization of transcription factors contributes to the regulation of their activity during development as defects in their localization underlie a variety of developmental disorders. In this study, we attempted to understand the nuclear import mannerisms of NeuroD1 and E47. We found that the nuclear import of NeuroD1 and E47 is energy-dependent and involves the Ran- mediated pathway. Herein, we demonstrate that NeuroD1 and E47 can dimerize inside the cytoplasm before their nuclear import. Moreover, this dimerization promotes nuclear import as the nuclear accumulation of NeuroD1 was enhanced in the presence of E47 in an in vitro nuclear import assay, and NLS-deficient NeuroD1 was successfully imported into the nucleus upon E47 overexpression. NeuroD1 also had a similar effect on the nuclear accumulation of NLS-deficient E47. These findings suggest a novel role for dimerization that may promote, at least partially, the nuclear import of transcription factors allowing them to function efficiently in the nucleus. © 2009 Elsevier Inc. All rights reserved. Keywords: NeuroD1 E47 bHLH Heterodimer Nuclear import Nuclear localization signal Introduction The differentiation and development of early embryos largely depend on the well-ordered control of gene activity determined by a network of transcription factors. Helix loop helix (HLH) transcription factors function as major players in a variety of differentiation programs (for review, see [1, 2]). On the basis of their functional relationships and gene expression patterns, HLH proteins are classified into seven classes [1]. NeuroD1 is a type II bHLH transcription factor with confined expression in certain parts of the brain, pancreas and intestinal endocrinal tissues. NeuroD1 was identified more than a decade ago as a protein that, when overexpressed, can transform endodermal tissue into neurons [3]. Since then, NeuroD1 has been extensively studied and many of its upstream, downstream and interacting proteins have been identified [4]. Moreover, it was shown that mice lacking NeuroD1 die within 5 days of birth due to severe diabetes mellitus marked by high ketone levels in the urine as a result of a loss of insulin producing pancreatic β-cells [5]. However, NeuroD1 knock-out mice were rescued transgenically by introducing myc-tagged NeuroD under the control of an insulin promoter [6], and geneti- cally, by crossing NeuroD null mice with a MATH-2/NEX-1-null genetic background [7]. These rescued mice displayed severe neurological disorders due to the depletion of cerebellar granule EXPERIMENTAL CELL RESEARCH 315 (2009) 1639 1652 Corresponding author. Department of Frontier Biosciences, Graduate School of Frontier Biosciences, Osaka University, 1-3 Yamada-oka, Suita, Osaka 565-0871, Japan. E-mail address: [email protected] (Y. Yoneda). 0014-4827/$ see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.yexcr.2009.02.025 available at www.sciencedirect.com www.elsevier.com/locate/yexcr

Upload: rashid-mehmood

Post on 13-Sep-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te /yexc r

Research Article

Synergistic nuclear import of NeuroD1 and its partnertranscription factor, E47, via heterodimerization

Rashid Mehmooda, Noriko Yasuharaa, Souichi Oeb, Masahiro Nagaib, Yoshihiro Yonedaa,b,⁎aDepartment of Frontier Biosciences, Graduate School of Frontier Biosciences, Osaka University, 1-3 Yamada-oka, Suita, Osaka 565-0871, JapanbDepartment of Biochemistry, Graduate School of Medicine, Osaka University, 2-2 Yamada-oka, Suita, Osaka 565-0871, Japan

A R T I C L E I N F O R M A T I O N

⁎ Corresponding author. Department of FrontieOsaka 565-0871, Japan.

E-mail address: [email protected]

0014-4827/$ – see front matter © 2009 Elseviedoi:10.1016/j.yexcr.2009.02.025

A B S T R A C T

Article Chronology:

Received 1 July 2008Revised version received13 February 2009Accepted 21 February 2009Available online 9 March 2009

The transition from undifferentiated pluripotent cells to terminally differentiated neurons iscoordinated by a repertoire of transcription factors. NeuroD1 is a type II basic helix loop helix(bHLH) transcription factor that plays critical roles in neuronal differentiation and maintenance inthe central nervous system. Its dimerization with E47, a type I bHLH transcription factor, leads tothe transcriptional regulation of target genes. Mounting evidence suggests that regulating thelocalization of transcription factors contributes to the regulation of their activity duringdevelopment as defects in their localization underlie a variety of developmental disorders. Inthis study, we attempted to understand the nuclear import mannerisms of NeuroD1 and E47. Wefound that the nuclear import of NeuroD1 and E47 is energy-dependent and involves the Ran-

mediated pathway. Herein, we demonstrate that NeuroD1 and E47 can dimerize inside thecytoplasm before their nuclear import. Moreover, this dimerization promotes nuclear import as thenuclear accumulation of NeuroD1 was enhanced in the presence of E47 in an in vitro nuclearimport assay, and NLS-deficient NeuroD1 was successfully imported into the nucleus upon E47overexpression. NeuroD1 also had a similar effect on the nuclear accumulation of NLS-deficientE47. These findings suggest a novel role for dimerization that may promote, at least partially, thenuclear import of transcription factors allowing them to function efficiently in the nucleus.

© 2009 Elsevier Inc. All rights reserved.

Keywords:

NeuroD1E47bHLH

HeterodimerNuclear importNuclear localization signal

Introduction

The differentiation and development of early embryos largelydepend on thewell-ordered control of gene activity determined bya network of transcription factors. Helix loop helix (HLH)transcription factors function as major players in a variety ofdifferentiation programs (for review, see [1, 2]). On the basis oftheir functional relationships and gene expression patterns, HLHproteins are classified into seven classes [1]. NeuroD1 is a type IIbHLH transcription factor with confined expression in certain partsof the brain, pancreas and intestinal endocrinal tissues. NeuroD1was identified more than a decade ago as a protein that, when

r Biosciences, Graduate Sc

u.ac.jp (Y. Yoneda).

r Inc. All rights reserved.

overexpressed, can transform endodermal tissue into neurons [3].Since then, NeuroD1 has been extensively studied and many of itsupstream, downstream and interacting proteins have beenidentified [4]. Moreover, it was shown that mice lacking NeuroD1die within 5 days of birth due to severe diabetes mellitus markedby high ketone levels in the urine as a result of a loss of insulinproducing pancreatic β-cells [5]. However, NeuroD1 knock-outmice were rescued transgenically by introducing myc-taggedNeuroD under the control of an insulin promoter [6], and geneti-cally, by crossing NeuroD null mice with a MATH-2/NEX-1-nullgenetic background [7]. These rescued mice displayed severeneurological disorders due to the depletion of cerebellar granule

hool of Frontier Biosciences, Osaka University, 1-3 Yamada-oka, Suita,

Page 2: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1640 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

cells and loss of the dentate gyrus in the hippocampus [6,7], whichmeans that NeuroD1 is crucial in neurogenesis. In general, type IIbHLH transcription factors such as NeuroD1, which are expressedin a tissue-specific manner, interact with type I bHLH E proteins(exemplified by E12/E47), which are expressed ubiquitously, toensure the transcription of target genes [8–10]. The products ofgene E2A (E12/E47) have been identified as NeuroD1-dimerizingpartners in pancreatic cells [11], enteroendocrine cells [12] and,more recently, in neuronal cells [13,14].

Functional molecules communicate continuously between thenucleus and the cytoplasm through the nuclear pore complexes(NPCs) present in the nuclear envelope. Themovement of ions andsmall molecules through NPCs mainly occurs by passive diffusionwhile the transport of macromolecules requires a selectivetransport pathway [15,16]. Most of the proteins that follow aselective nuclear import pathway harbor nuclear localizationsignals (NLSs). In the classical nuclear import pathway, the NLS ofthe cargo molecule is identified by adaptor protein importin α,which forms a heterodimerwith the actual transporter, importin β.The importin α/β-cargo complex then docks to the NPC for thesubsequent translocation step [17–22]. Apart from this classicalnuclear import pathway, numerous cases have been found whereimportinβ alone can interact andmediate the nuclear import of thecargo proteins [23]. In mammals, importin α constitutes 6 familymembers while importin β constitutes at least 20 family membersboth with tissue-specific expression and substrate specificity[23,24]. In addition, it is known that small GTPase Ran plays acritical role as it controls the directionality of the transport process[25,26]. It has been shown that Ran exists mainly in a GDP-boundform in the cytoplasm and in a GTP-bound form in the nucleus. TheGTP-bound form triggers the dissociation of the import complex inthe nucleus [27] where cargo molecules perform their respectivefunctions while the import receptors are recycled back to thecytoplasm for another import cycle.

Emerging evidence suggests that the nuclear import of a varietyof transcription factors is regulated during differentiation anddevelopment [28,29]. Recently Oct6, a critical transcription factorfor neuronal differentiation, was shown to be cytoplasmic in undif-ferentiated ES cells and to translocate into the nucleus uponneuronal differentiation [29]. Furthermore, the study clearly de-monstrated that switching the expressionpattern of the importinαsubtype triggers the differentiation of ES cells to neuronal cells invitro, showing that nuclear import factors play a crucial role in celldifferentiation [29]. Consequently, anomalies in the nucleartransport process are known to correlate to various developmentaldefects, diseases and certain forms of cancer [30–34].

Although it is well established that NeuroD1 and E47 functiontogether in the nucleus as a heterodimer to control gene

Fig. 1 – NeuroD1 is selectively transported into the nucleus. (a–c) Efixed rat primary cultured neurons. (d–f) NeuroD1–EGFPwas transfethe nucleus. (g–n) In vitro nuclear import assay. Digitonin-permeacontaining 0.3 μM import substrate, an ATP regeneration system, anNeuroD1 and GST–NLS–GFP accumulated in the nucleus in the presATP regeneration system and lysate (k, l) or the addition of WGA (mSimilar results were obtained in the microinjection assay (o–r). RecNIH3T3 cells either alone (o) or together with 1 μMWGA (q). The celfollowed by anti-goat Alexa488 and visualized under a microscope.

expression, it is not yet elucidated how both NeuroD1 and E47are transported into the nucleus and whether their interactionmutually affects their nuclear transport. Therefore, in this study,we set out to determine the nuclear import manner of NeuroD1and its interacting partner E47 in a cellular context. We show thatdirect heterodimerization between NeuroD1 and E47 via HLHdomains can take place inside the cytoplasm before their nuclearimport. Furthermore, the basic amino acid-rich domain inNeuroD1 acts as a functional NLS of NeuroD1. NLSs of eitherNeuroD1 or E47 sufficiently carry out the nuclear import of theheterodimer. Thus, the heterodimerization between NeuroD1 andE47 may confer an additional effect on the localization and actionof these partner proteins.

Materials and methods

Cell culture

NIH3T3 and HeLa cells that do not express endogenous NeuroD1,were cultured in Dulbecco's modified MEM (DMEM; Sigma-Aldrich) supplemented with heat-inactivated 10% fetal bovineserum at 37 °C in 10% CO2.

Construction of mammalian expression vectors

Both NeuroD1 and E47 were cloned from mouse brain cDNA.NeuroD1 was cloned between the EcoRI and SmaI sites of a pEGFP-N2 vector (BD Biosciences clonetech). NeuroD1 mutants wereprepared using different primers, albeit with the same restrictionsites of pEGFP-N2. To construct NeuroD1 and mutants with twoGFP molecules, EGFP from pEGFP-C2 was digested with NheI andBglII and cloned upstream of NeuroD1–EGFP. E47 was clonedbetween the HindIII and SacII sites in a pmRFP vector. Similarly, theE47 NLSmutant, which lacks aa159–177 and aa537–557 of E47wasprepared by amplification and was cloned between the HindIII andSacII sites of the same vector. Flag-E47-RFP was constructed in apFlag-CMV2 vector (Sigma) between the HindIII and NotI sites.

Bacterial expression vectors and purifiedrecombinant proteins

NeuroD1 was cloned in a pET21d vector (Novagen) as follows. ThepET21d vector was digested with NcoI and treatedwithMung beannuclease followed by digestion with XhoI. The NeuroD1 upstreamregion was blunted with T4 DNA polymerase treatment while thedownstream regionwas digested with XhoI. Similarly, E47-pET21dwas generated by cloning E47 between the BamHI and XhoI sites of

ndogenously expressed NeuroD1 is localized to the nucleus incted intoNIH3T3 cells and the GFP signal wasmainly detected inblized HeLa cells were incubated with 10 μl reaction mixtured Ehrlich cell lysate. GST–NLS–GFP was used as a control. (g, h)ence of lysate and the ATP regeneration system. (i, j) Lack of an, n) inhibited the nuclear import of NeuroD1 and GST–NLS–GFP.ombinant NeuroD1-His was cytoplasmically microinjected intols were fixed 1 h aftermicroinjection, treatedwith anti-NeuroD1Anti-rat Alexa568 was co-injected as an injection marker (p, r).

Page 3: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1641E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

the pET21d vector. To construct NeuroD1–GST, GST was clonedbetween the XhoI sites in NeuroD1-pET21d. GST–NDNLS–GFP wasconstructed in a PGEX6p1 vector. GFP was from EGFP-N2. NDNLSwas cloned between GST and GFP utilizing EcoRI and SmaI sites.

NeuroD1-pET21d and GST–NDNLS–GFP were transformed intoa BL21 strain of Escherichia coli. The expression of the fusionproteins was induced by 0.1 mM IPTG at 20 °C for 16 h. E47-pET21dwas transformed into another E. coli strain, Rosetta. The expression

Page 4: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

Fig. 2 – A long stretch of basic amino acids contributes to thenuclear localization of NeuroD1. (a) Various deletion mutantsof NeuroD1 were constructed. (b–j) Wild type NeuroD1 and itsmutants fused with EGFP were transfected into NIH3T3 cells.The cells were fixed and observed under a microscope 20 h postTransfection. (c) Mutation outside the basic chain did not affectthe nuclear accumulation of NeuroD1, while various mutationsinside the basic chain did (d–i). (j) EFGP alone was transfectedinto the cells as a control. (k) aa87–aa103 from the basic chainof NeuroD1 when fused between GST and GFP (namedGST–NDNLS–GFP) efficiently translocated the fusion into thenucleus upon microinjection, while GST–GFP fusion lackingthis chain did not (m). Anti-rat Alexa568 was co-injected as aninjection marker (l, n). (o, p) Transfection of wild typeNeuroD1–EGFP and Δ81–103–2EGFP into cultured primaryhippocampal neurons. (q) Cross species comparison of thebasic amino acids chain. (r) Florescence quantification of thetransfected cells. N/C florescence intensity was calculated usingAdobe photoshop 5.5. Transfection of every construct wasperformed at least three times (n=20). Error bars denote thestandard deviation.

1642 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

of fusion protein was induced by 0.1 mM IPTG at 37 °C for 5 h.Bacterial cultures were harvested, treated with liquid Nitrogen andstored at−80 °C overnight. After thawing and sonication, the clearlysates of GST-tagged proteins were incubated with GlutathioneSepharose™ 4B (Amersham biosciences), while the lysates of His-tagged proteins were incubated with Ni-NTA agarose (Qiagen)following the manufacturer's instructions. RanQ69L was purifiedas described previously [35]. The purified proteins were finallydiluted in Transport Buffer (TB: 20 mM HEPES–KOH pH 7.3,110 mM potassium acetate, 5 mM sodium acetate, 2 mMmagnesium acetate, 0.5 mM EGTA, 2 mM DTT, 1 μg/ml each ofaprotinin, pepstatin and leupeptin) using a PD-10 column (GEHealthcare).

In vitro import assay and immunofluorescence

Digitonin-permeabilized HeLa cells were prepared essentially asdescribed previously [36]. Permeabilized cells were incubated atroom temperature for 10 min in TB and washed twice with TB tominimize residual proteins in the cytoplasm. The cells wereincubated at 37 °C for 50 min with 1% BSA in TB, transportsubstrates, Ehrlich cell lysate and an ATP regeneration system in atotal volume of 10 μl per sample. After incubation, the cells werefixed with 4% paraformaldehyde in phosphate-buffered saline(PBS) followed by permeabilization of the nuclear membrane with0.5% Triton X-100 in PBS. Immunostaining of NeuroD1 wasperformed with a primary anti-NeuroD polyclonal antibody(Santa Cruz) and a secondary Alexa488-labeled anti-goat IgGantibody (Molecular Probes). For detection of E47, primary anti-E47 (Santa Cruz) was used followed by Alexa568-labeled, anti-rabbit IgG antibody (Molecular Probes).

Transfection assays

The transfection of NIH3T3 cells was carried out using Effectene(Qiagen) according to the manufacturer's operation manual.NIH3T3 cells were grown on 12-well multidishes (NUNC). Foreach expression vector, 300 ng of plasmid was transfected per welland the cells were incubated for 20 h before observation under aZeiss Axiovert 200Mmicroscope. Cotransfection experiments wereperformed the same way with equal quantities of expressionplasmids. Quantification of images was performed using adobephotoshop 5.5. Quantification was carried out by obtaining thefluorescence intensity ratios of the nucleus and cytoplasm/back-ground. Similarly, ImageJ was used for the quantification ofconfocal images.

Primary neuronal culture and transfection

Primary hippocampal neurons that express endogenous NeuroD1were cultured as described [37]. The hippocampi of embryonicday 18 rats were isolated. Then, following dissociation withtrypsin treatment, cells were plated in Neurobasal media(Invitrogen) containing 2.5 mM L-glutamine (Invitrogen), B-27(Invitrogen, used at 1:50 dilution), and antibiotics/antimycotic(Invitrogen, used at 1:100 dilution) in 12-well dishes onpolyethyleneimine (Sigma, P2636) coated coverslips. Transfectionwas carried out 10–14 days after plating the cells using atransmessenger transfection kit (Qiagen) with 1.6 μg of expressionvectors. Fluorescent images of the paraformaldehyde fixed

neurons were taken using a confocal laser-scanning microscopeattached to an inverted microscope (Axiovert 100M; Carl Zeiss)with a ×40/0.75 Plan-Neofluar.

Endogenous NeuroD1 and E47 in primary hippocampalneurons were immunostained by fixing the cells with 4%paraformaldehyde in PBS supplemented with 4% agarose and2 mM MgCl2 followed by 70% methanol treatment and antibodiesas described previously.

Co-immunoprecipitation and immunoblotting

Cells were cultured on a 6-well dish. Flag-E47-RFP and NeuroD1–2EFGP and its mutants were co-transfected using Effectene(Qiagen) according to the manufacturer's operation manual. Thecells were harvested at 24 h post-transfection and resuspended inTB with 1% Triton X-100 and protease inhibitors. The lysates wereimmunoprecipitated with washed M2 beads for 4 h at 4 °C.Interacting proteins bound to beads were eluted in the samplebuffer (2% SDS, 10% glycerol, 60 mM Tris–HCl (pH 6.8), 5% βmercaptoethanol, and 0.01% bromophenol blue), resolved by 12.5%SDS-polyacrylamide gel electrophoresis and transferred onto a

Page 5: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

Fig. 2 (continued).

1643E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

nitrocellulose transfer membrane (Whatman). The blots were firstincubated in blocking buffer (5% (w/v) non-fat dry milk in PBSplus 0.05% tween 20) overnight at 4 °C. The blots were thenincubated with primary antibody for 1 h at room temperature.Both anti-NeuroD (sc-1084) and anti-E47 (sc-763) were fromSanta Cruz biotechnology and anti-GFP (A11122) was fromInvitrogen Molecular Probes. After extensive washing, the blotswere incubated with horseradish peroxidase-conjugated second-

ary antibodies for 1 h at room temperature. Antigen-antibodycomplexes were visualized using ECL (GE Healthcare).

In vitro binding assay

In vitro binding assay was performed in TB. GST and NeuroD1–GSTwere immobilized on the GST beads. E47-His was added to theimmobilized proteins. The mixture was incubated at 37 °C for

Page 6: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1644 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

90 min to a final volume of 300 μl. Beads were extensively washedwith TB, treated with sample buffer and immunoblotted usinganti-E47 and anti-GST (Santa Cruz biotechnology).

Microinjection

NIH3T3 cells on a 35 mm culture dish were cytoplasmicallymicroinjected using a micromanipulator system as described [38].Samples were prepared in TB and filtered through a 0.22 mm PVDFfilter (Millipore) prior to injection. Alexa568-labeled anti-rat IgGwas used as an injection marker. The cells were fixed with 4%paraformaldehyde in PBS at 1 h post-microinjection, followed bypermeabilization of the nuclear membrane by 0.5% Triton X-100 inPBS and immunostaining of NeuroD1 by primary anti-NeuroDpolyclonal antibody and a secondary Alexa488-labeled anti-goatIgG antibody (Molecular Probes).

Results

NeuroD1 is primarily localized in the nucleus and its nuclearimport is energy-dependent

To elucidate the dynamic behavior of transcription factors forneuronal differentiation, we focused on NeuroD1 and its bindingpartner, E47. To clarify the exact subcellular localization ofNeuroD1, the endogenous expression of NeuroD1 was detectedby immunostaining in rat primary hippocampal neurons where itwas seen to primarily localize in the nucleus (Figs. 1a–c).NeuroD1–EGFP, when transiently expressed in NIH3T3 cells, wasalso localized in the nucleus (Figs. 1d–f). These results show thatNeuroD1 has the ability to migrate into the nucleus.

To investigate the nuclear import fashion of NeuroD1, weperformed in vitro transport assay using digitonin-permeabilizedHeLa cells (Figs. 1g–n). The addition of both Ehrlich ascites tumor

Fig. 3 – E47 is selectively transported into the nucleus. (a–h) Digitonmixture containing 0.3 μM of import substrate, an ATP regeneratiocontrol. Lack of an ATP regeneration system (c, d) or lysate (e, f) orand GST–NLS–GFP (E47-upper panel; GST–NLS–GFP-lower panel).

cell lysate and ATP efficiently mediated the nuclear import ofNeuroD1-His (Fig. 1g). Nuclear import was undetectable in theabsence of either an energy source (Fig. 1i) or cell lysate (Fig. 1k).The import was also inhibited by the addition of wheat germagglutinin (WGA), a well-known inhibitor of selective nuclearimport [39] (Fig. 1m), which shows that NeuroD1-His does notdiffuse into the nucleus passively. GST-SV40NLS-GFP (heretoforereferred to as GST–NLS–GFP) was used as a control. To extendthese findings to an in vivo system, the recombinant NeuroD1-Hiswas cytoplasmically microinjected into living NIH3T3 cells with orwithout the addition of WGA (Figs. 1o–r). The results showed thatthe nuclear localization of the injected NeuroD1-His was inhibitedby co-injectionwithWGA. These results indicate that NeuroD1-Hiscan translocate into the nucleus and that the nuclear import ofNeuroD1-His requires energy and soluble cytosolic factors.

A long stretch of basic amino acids in the basic domain ofNeuroD1 functions as an intrinsic NLS

Most proteins that are targeted to the nucleus via a selectivetransport mechanism harbor an NLS in their primary sequences. Toclarify the sequence(s) responsible for the nuclear import ofNeuroD1, we carried out domain localization analysis using variousdeletion mutants (Fig. 2a). Deletions outside the basic domain(Fig. 2c) did not affect the nuclear accumulation of NeuroD1, whilevarious deletions inside the basic domain variably disrupted thenuclear accumulation of NeuroD1 (Figs. 2d–i). The deletion of theentire stretch of basic amino acids (mutantΔ81–103-EGFP) (Fig. 2i)significantly blocked the nuclear accumulation of NeuroD1. Only aportion of this basic peptide (aa 87–aa 103), when fused betweenGST and GFP (GST–NDNLS–GFP), was sufficient to mediate thenuclear import of theGST–GFP fusionproteinwhen cytoplasmicallymicroinjected (Fig. 2k), whereaswhenGST–GFP lacked this stretch,it was exclusively cytoplasmic (Fig. 2m). To increase the molecularweight of the fusion proteins, two GFP molecules were fused with

in-permeablized HeLa cells were incubated with 10 μl reactionn system, and Ehrlich cell lysate. GST–NLS–GFP was used as athe addition of WGA (g, h) inhibited the nuclear import of E47

Page 7: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1645E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

wild type NeuroD1 and its mutants. NeuroD1–2EGFP was stillconfined to the nucleus while Δ81–103–2EGFP had relatively highcytoplasmic retention. In addition, when the expression vector forΔ81–103–2EGFP was transfected into cultured primary hippocam-pal neurons, therewas clear cytoplasmic retention of the GFP signal(Figs. 2o, p). These results indicate that the basic domaincontributes to the nuclear localization of NeuroD1 and that the81–103 stretch of basic amino acids acts as an intrinsic NLS ofNeuroD1. Furthermore, itwas found that this peptide of basic aminoacids is conserved across species (Fig. 2q).

E47 is imported into the nucleus in an energy-dependent andreceptor-mediated manner

It is well known that NeuroD1 heterodimerizes with E47 and thatthis dimerization is important for the transactivation activity ofNeuroD1 [34]. Therefore, we next focused on the nuclear importproperties of E47 and on the significance of this dimerizationprocess in terms of nuclear accumulation of the heterodimer. Toexplore the import characteristics of E47, the in vitro import assayusing recombinant E47-His was carried out both in the presenceand absence of each Ehrlich cell lysate, ATP and WGA (Figs. 3a–h).Like NeuroD1, E47-His was efficiently imported into the nucleus inthe presence of lysate and ATP (Fig. 3a), and its nuclear importwas abolished in the absence of either ATP or lysate (Figs. 3c, e).The addition of WGA also inhibited the nuclear accumulation ofE47-His (Fig. 3g). These results indicate the involvement of asoluble factor-mediated selective nuclear import mechanism forE47.

Small GTPase Ran is involved in the nuclear import ofNeuroD1 and E47

Small GTPase Ran is involved in the selective nuclear transportmediated by the importin β family transport factors. To investigatethe impact of Ran on the nuclear import of NeuroD1 and E47,RanQ69L-GTP, a dominant-negative form of Ran, which is deficientin GTPase activity and is predominantly in the GTP-bound form

Fig. 4 – Nuclear import pathway of NeuroD1 and E47 is Ran-dependabsence (a, c, e) and presence (b, d, f) of 20 μMRanQ69L-GTP. Cells wMethods section.

[40,41], was used in the in vitro transport assay. As shown in Fig. 4,both E47-His (Fig. 4a) and NeuroD1-His (Fig. 4c) accumulated inthe nucleus in the presence of lysate while the addition ofRanQ69L-GTP hampered their nuclear accumulation (Fig. 4b, d).This result clearly indicates that the nuclear import of NeuroD1and E47 is Ran-dependent and probably requires importin β familymember proteins.

E47 enhances the nuclear accumulation of itsheterodimerizing partner, NeuroD1

To identifywhether the heterodimerization of NeuroD1with E47 isimportant for the transactivation function aswell as for the nuclearimport process, we assessed the impact of E47 on the nuclearimport of NeuroD1. The nuclear import of NeuroD1-His wasexamined in the in vitro transport assay in the presence of E47-His.Surprisingly, we found that the nuclear import efficiency ofNeuroD1 was significantly enhanced by the addition of E47-Histo the import assay mixture, while the import efficiency of E47-Hisappeared to be unchanged even with the addition of NeuroD1-His(Figs. 5a–c). The nuclear accumulation rates of NeuroD1and E47when added to the import mixture alone or together are showngraphically in Figs. 5d and e. These results suggest that E47 playssome role in the nuclear import process of NeuroD1. The impactof E47 on the nuclear accumulation of NeuroD1 was specific.When added to NeuroD1-His or E47-His in the import mixture,GST–NLS–GFP had no affect on the nuclear accumulation of eitherNeuroD1 or E47. Likewise, the nuclear import of GST–NLS–GFP alsoremained unchanged (supplementary Fig. 1).

To gain insight into how E47modulates the nuclear localizationof NeuroD1, RFP-tagged E47 was transfected either alone ortogether with NeuroD1–2EGFP in cultured neuronal cells. Flag-E47-RFP was mainly localized in the nucleus when transfectedalone (Fig. 5f). Upon cotransfection, wild type NeuroD1–2EGFPand flag-E47-RFP were co-localized in the nucleus as a hetero-dimer (Figs. 5g–j). Endogenous NeuroD1 and E47 were also foundto localize mainly in the nuclei of rat primary hippocampalneurons (supplementary Figure 2), though faint cytoplasmic

ent. In vitro import assay was carried out both in theere fixed and immunostained, as described in the Materials and

Page 8: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1646 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

Page 9: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

Fig. 5 – E47 promotes the nuclear accumulation of NeuroD1. (a) Nuclear accumulation of NeuroD1 in the absence and presence of E47in an in vitro nuclear import assay (n=30). (b) Nuclear accumulation of E47 in the absence and presence of NeuroD1 in an in vitronuclear import assay (n=30). (c) Graphical representation of (a) and (b). Error bars indicate Standard Deviation. (d) The nuclearaccumulation rate of NeuroD1 in the absence and presence of E47. (e) The nuclear accumulation rate of E47 in the absence andpresence of NeuroD1. Importmixture was added to the permeablized cells and incubated at 37 °C for 10min, 20min, 30min, 40min,and 50 min. After fixation, E47 and NeuroD1 were immunostained using anti-E47 and anti-NeuroD1, respectively. (f–v) Primarycultures of neurons from rat hippocampi were transfected with the indicated expression plasmids. At 24 h post-transfection, theneuronal cells were observed under a confocal laser-scanning microscope attached to an inverted microscope (Axiovert 100M; CarlZeiss) with a ×40/0.75 Plan-Neofluar. (f) Wild type E47-RFP was localized in the nucleus. (g–j) Wild type NeuroD1 and E47 wereco-localized in the nucleus. (k–n) Wild type E47 redirected the localization of the otherwise cytoplasmic NeuroD1 mutantΔ81–103–2EGFP to the nucleus. (o–r) Δ81–161–2EGFP, a mutant lacking both an NLS and an HLH domain did not change itslocalization even after cotransfection with E47-RFP. (s–v) E47 did not change the localization of 2EGFP alone. The transfectionexperiments were carried out three times independently. The average number of cells transfected and the localization of GFP andRFP signals are shown in the supplementary table and Fig. 6p. (w–y) Interaction of NeuroD1 with E47 through their HLH domains.(w) 2EGFP, NeuroD1–2EFGP and its mutants were transfected into NIH3T3 cells and their expression was detected by westernblotting. (x) 2EGFP, NeuroD1–2EFGP and its mutants were co-transfected with flag-E47-RFP. At 24 h post-transfection, the cells wereharvested and the lysates were immunoprecipitated using M2 agarose (Sigma). The immunoprecipitated proteins wereimmunoblotted with anti-GFP (upper panel) and anti-E47 (lower panel). (y) In vitro binding assay was performed to demonstratedirect interaction of NeuroD1 and E47. NeuroD1–GSTwas immobilized on GST beads followed by the addition of E47. As a control, E47was added to GST bound beads. The samples were immunoblotted with anti E47 (left panel) and anti-GST (right panel).

1647E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

Page 10: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1648 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

signals were detected for E47. However, whether this dimerizationcan take place in the cytoplasm or has any role in the nuclearimport of these proteins remains elusive.

Therefore, we sought to address these issues by co-transfectingvarious mutants of NeuroD1 with flag-E47-RFP. Interestingly,the wild type E47 modulated the nuclear localization of Δ81–103–2EGFP, which is a NeuroD1mutant that lacks the intrinsic NLS

but still has the HLH domain that may interact with E47, and wasretained in the cytoplasm in the absence of E47 (Fig. 5k–n). Toshow that the nuclear localization of the NLS mutant form ofNeuroD1 depends on its interaction with E47 through their HLHdomains, another NeuroD1 mutant, Δ81–161–2EGFP, which lacksboth the NLS and HLH domain, was used. When this mutant wastransfected alone into the neuronal cells, its localization was

Page 11: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1649E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

nucleocytoplasmic, most likely because of its smaller sizecompared with Δ81–103–2EGFP. Cotransfection of flag-E47-RFPand Δ81–161–2EGFP caused no change in the localization of thisNeuroD1 mutant, although the localization of E47 itself wasconfined to the nucleus (Fig. 5o–r) suggesting that the specificityof the interaction between the HLH domains affects their nuclearaccumulation. Similar nuclear redirection by E47 was observedwhen the cotransfection was performed in NIH3T3 (supplemen-tary Figure 3) and Cos-7 cells.

The interaction between NeuroD1 and E47 through their HLHdomains was further confirmed by immunoprecipitation. GFP-tagged wild type NeuroD1 and its mutants were expressed inNIH3T3 cells (Fig. 5w). Flag-E47-RFP was co-transfected withGFP-tagged NeuroD1 and its mutants into NIH3T3 cells andimmunoprecipitation was carried out using M2 beads (Fig. 5x).As shown in Fig. 5w, only NeuroD1 constructs harboring HLHdomains interacted with E47, while the mutant lacking the HLHdomain did not. In addition, to show that NeuroD1 and E47 canbind directly to one another without the presence of DNA, an invitro binding assay was carried out using recombinant NeuroD1–GST and E47-His. As shown in Fig. 5y, the recombinant E47-Hisspecifically bound to the recombinant NeuroD1–GST. Thus, weconcluded that E47 and NeuroD1 directly interact with oneanother via their HLH domains and this interaction has asignificant effect on the nuclear import process. The data alsosuggest that the NLSs of E47 alone are sufficient to mediate thenuclear import of the heterodimer.

NeuroD1 also supports the nuclear import ofNLS-deficient E47

Next, we attempted to determinewhether NeuroD1 plays a similarrole in the nuclear import of E47, i.e., whether NeuroD1 affects thenuclear import of E47. To address this, an NLSs-deleted mutant ofE47, ΔNLS-E47-RFP, was used. This mutant lacked aa159–177 andaa537–557 of E47. E47 reportedly has two NLSs [42] and deletion ofboth confines its localization to the cytoplasm. Consistent with thatreport, we found that deleting both NLSs abolished the nuclearaccumulation of E47 (Figs. 6a–c). Upon cotransfection of ΔNLS-E47-RFP with wild type NeuroD1–2EGFP in cultured neuronalcells, the otherwise exclusively cytoplasmic ΔNLS-E47-RFP shiftedtowards the nucleus (Figs. 6d–g). The NLS-deficient mutant ofNeuroD1 Δ81–103–2EGFP did not alter the localization of ΔNLS-E47-RFP upon cotransfection (Figs. 6h–k). These results indicatethat import-competent NeuroD1 is also able to support the nucleartranslocation of NLS-deficient E47 in neuronal cells. From thesefindings, we concluded that in neuronal cells NeuroD1 and E47mutually co-operate with one another for efficient and timelynuclear localization.

It is noteworthy that in NIH3T3 cells, wild type NeuroD1 couldnot redirect the nuclear accumulation of ΔNLS-E47-RFP as

Fig. 6 –NeuroD1 can alsomediate the nuclear import of E47. Primarythe indicated expression vectors and observed under a microscopeaccumulation when transfected into neuronal cells. (d–g) Cytoplasupon cotransfectionwith wild type NeuroD1–2EGFP. (h–k) The NLSaccumulation of ΔNLS-E47. (l–o) 2EGFP, when cotransfected with Δ(p) N/C florescence of the primary hippocampal neuronal cells tran(summarized in the supplementary table). NeuroD1 and its mutanError bars indicate the standard deviation. (n=10).

efficiently as in the primary neurons. Even the nuclear import ofwild type NeuroD1 was slightly shifted to the cytoplasmiccompartment when cotransfected with ΔNLS-E47-RFP. The E47mutant might act as a dominant negative in this case (supple-mentary Figure 4). This data suggest that they mutually cooperateonly in the primary neurons for nuclear redirection. Fig. 6pgraphically describes the localization of NeuroD1 and E47-derivedmutant proteins in the primary neurons when cotransfected in theindicated combinations.

Discussion

HLH proteins are crucial transcription regulators during a widevariety of developmental programs in invertebrates and verte-brates [1,43]. On the basis of their functional relationships andgene expression patterns, HLH proteins are classified into sevenclasses [1]. In the developing nervous system, class II, V and VIHLH transcription factors have been extensively scrutinizedconcerning their roles in neural programming, while class I HLHproteins like E47 have been merely relegated to the role ofbinding partners to ensure transcriptional activities [44].Mounting evidence suggests that the class II factor NeuroD1heterodimerizes with E47 in pancreatic beta cells [45], endocrinecells [11] and neuronal cells [13] to carry out its transactivatingactivities.

As hypothesized by Heng and Tan [44], E proteins like E47 mayhave additional roles in the developing nervous system aside fromacting merely as heterodimerizing partners in the nucleus, asindicated by recent evidence showing the involvement of E12/47in stabilizing Neurogenin, a proneural bHLH transcription factoracting upstream of NeuroD1 [46]. Furthermore, the role of Eproteins in the cotransport of other dimerizing bHLH transcriptionfactors has been reported in muscle and lymphocyte developmentwhere they interact with and transport MyoD [42] and SCL/tal[47], respectively. However, there have been no correspondingstudies for the developing nervous system. In this study, our dataassign an additional role to E47, i.e. the role of nuclear localizationby modulating the nuclear transport of NeuroD1 via heterodimer-ization. This is the first report describing the involvement of Eproteins in the nuclear localization of partner molecules in thenervous system.

Our data indicate that NeuroD1 and E47 are transported intothe nucleus by importin β family members in a Ran-dependentmanner; however, we did not determine which importin β familymembers are involved. The data also imply that NeuroD1 and E47can bind to one another in the cytoplasm and then be importedinto the nucleus as a heterodimeric complex. Although NeuroD1was sufficient for the nuclear import of ΔNLS-E47-RFP in theprimary neurons, there was only partial nuclear redirection ofΔNLS-E47-RFP in NIH3T3 cells. On the other hand, the nuclear

cultures of neurons from rat hippocampiwere transfectedwith24 h post transfection. (a–c) ΔNLS-E47-RFP lost its nuclearmic localization of ΔNLS-E47-RFP shifted towards the nucleusmutant of NeuroD1, Δ81–103–2EGFP, did not restore the nuclearNLS-E47-RFP, did not change the localization of the RFP signal.sfected with the indicated combinations of expression vectors

ts were GFP-tagged while E47 and ΔNLS-E47 were RFP-tagged.

Page 12: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1650 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

accumulation of NeuroD1 was slightly compromised suggestingthat ΔNLS-E47 might have a dominant-negative effect on NeuroD1(supplementary Figure 4). Therefore, it is likely that multipleoverlapping of nuclear import pathways exists for monomeric orhomodimeric E47, NeuroD1 and of their mutual heterodimer. Suchdifferential binding to specific import receptors by the individualproteins and their mutual heterodimer has already been reportedfor the RXR-VDR heterodimer [48]. This hypothesis is strengthenedby the fact that NLS-deficient E47 is imported efficiently into thenucleus when cotransfected with wild type NeuroD1 in theprimary neurons, but not as efficiently, however, in eitherNIH3T3 or cos-7 cells. This might suggest that the particularimport receptor is specifically, or, perhaps highly, expressed in theprimary neurons. The question of why endogenous proteins do notchange the localization of their NLS-deficient counterparts mightbe explained by tight regulation over the expression of theendogenous proteins. A low expression of endogenous proteinsmay make them incapable of efficiently relocating their over-expressed counterparts. The consequences of heterodimerization-mediated nuclear import remain to be seen, but it is likely that thisphenomenon influences the nuclear accumulation of proteins, and,consequently, their actions. The phenomenon appears to beparticularly important because a small change in the localizationof critical transcription factors may contribute to abnormalphenotypes as exemplified by a point mutation in the TWISTgene that leads to a small change in the localization of TWIST. Thisultimately results in Saethre–Chotzen syndrome [49].

Although the effect of E47 on the nuclear accumulation ofNeuroD1 was quite similar to its effects on MyoD [42], our datashed additional light on the unique aspect of the NeuroD1/E47import process. The role of E47 in the nuclear import of MyoD wasa one-sided process, i.e. E47 redirected the nuclear accumulation ofNLS-deficient MyoD, whereas MyoD lacked the ability to importNLS-deficient E47 [42]. By contrast, we found that both partnerproteins in the NeuroD1/E47 heterodimer assisted one another inthe nuclear import process, although it remains unknown howNeuroD1, but not MyoD, affects the nuclear import process of E47.This differencemay arise from the fact that despite being membersof the same family, NeuroD1 and MyoD have differing modes ofprotein binding and stability as evidenced by the crystal structuresof the E47–NeuroD1 and E47–MyoD helices [50, 51]. Moreover, it islikely that MyoD and NeuroD1 may encounter different regulationin the tissues where each is expressed.

NeuroD1 undergoes extensive modifications with profoundconsequences [4]. O-linked glycosylation of NeuroD1 by O-linkedGlc NAc transferase reportedly regulates the subcellular localiza-tion of NeuroD1 in the mouse insulinoma cell line MIN6 [52]. In asimilar manner, the MEK-ERK-mediated phosphorylation of Neu-roD1 that is stimulated by high glucose levels also alters thesubcellular localization of NeuroD1 [53]. Interestingly, serine 274 isthe common target site for glycosylation and phosphorylation. Tosee the impact of these modifications on the subcellular localiza-tion of NeuroD1, we mutated serine 274 to alanine (non-phosphorylable form) and aspartic acid (which mimics thephosphorylated state) and transfected these mutants into ratprimary hippocampal neurons (supplementary Figure 5). How-ever, both of these mutants remained localized to the nucleussuggesting that these modifications are context dependent [54]and pointing out the existence of multiple pathways that regulateNeuroD1 activities. Moreover, these mutants also retained the

ability to relocate ΔNLS-E47-RFP into the nucleus (supplementaryFigure 5).

A multitude of class II bHLH proteins play roles in the neuraldifferentiation process, and class I bHLH E proteins reportedlyfunction as their heterodimerizing partners to control target geneexpression [2]. Some of these type II transcription factors actupstream of NeuroD1, e.g. E47 heterodimerizes with Neurogenin,to carry out transactivating functions [46,55]. E proteins are knownto be expressed ubiquitously. Therefore, in order for E proteins toenact tissue-specific or developmental stage-specific functions inspecialized cells, it seems that they have to bind to specific partnerproteins. On the other hand, class II bHLH transcription factorsalone are not capable of binding to DNA as homodimers in order tocarry out transcriptional regulation [56]. Thus, it is most likely thatin neurogenesis, one of the roles of ubiquitously expressed E47 is topromote the nuclear import of its binding partner NeuroD1through heterodimerization, which enables NeuroD1 to functionefficiently in the nucleus. Neuronal differentiation is accompaniedby dynamic expression patterns of nucleocytoplasmic transportfactors [29]. Heterodimeric nuclear transport might be a strategyutilized by cells to respond to such developmental cues effectively.It will be interesting to explore whether E47 plays the same role innuclear targeting of all of its partner proteins. Thus, this studyprovides novel insights regarding the question of why NeuroD1forms a heterodimer with the transcriptional co-factor E47 toachieve its function in neurogenesis.

Acknowledgments

We thank Yoneda labmembers for their discussions. This workwassupported by the Japanese Ministry of Education, Culture, Sports,Science and technology, the Japan Society for the Promotion ofScience, the Takeda Science Foundation and the Foundation ofSanyo Broadcasting.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.yexcr.2009.02.025.

R E F E R E N C E S

[1] M.E. Massari, C. Murre, Helix–loop–helix proteins: regulators oftranscription in eucaryotic organisms, Mol. Cell. Biol. 20 (2000)429–440.

[2] R. Kageyama, T. Ohtsuka, J. Hatakeyama, R. Ohsawa, Roles of bHLHgenes in neural stem cell differentiation, Exp. Cell. Res. 306(2005) 343–348.

[3] J.E. Lee, S.M. Hollenberg, L. Snider, D.L. Turner, N. Lipnick,H. Weintraub, Conversion of Xenopus ectoderm into neurons byneuroD, a basic helix–loop–helix protein, Science 268 (1995)836–844.

[4] J.H. Chae, G.H. Stein, J.E. Lee, NeuroD: the predicted and thesurprising, Mol. Cells. 18 (2004) 271–288.

[5] F.J. Naya, H.P. Huang, Y. Qiu, H. Mutoh, F.J. DeMayo, A.B. Leiter, M.J.Tsai, Diabetes, defective pancreatic morphogenesis, and abnormalenteroendocrine differentiation in BETA2/neuroD-deficient mice,Genes Dev. 11 (1997) 2323–2334.

Page 13: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1651E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

[6] T. Miyata, T. Maeda, J.E. Lee, NeuroD is required for differentiationof the granule cells in the cerebellum and hippocampus, GenesDev. 13 (1999) 1647–1652.

[7] M.H. Schwab, A. Bartholomae, B. Heimrich, D. Feldmeyer, S.Druffel-Augustin, S. Goebbels, F.J. Naya, S. Zhao, M. Frotscher, M.J.Tsai, K.A. Nave, Neuronal basic helix–loop–helix proteins (NEXand BETA2/Neuro D) regulate terminal granule cell differentia-tion in the hippocampus, J. Neurosci. 20 (2000) 3714–3724.

[8] C. Murre, P.S. Mccaw, H. Vaessin, M. Caudy, L.Y. Jan, Y.N. Jan, C.V.Cabrera, J.N. Buskin, S.D. Hauschka, A.B. Lassar, H. Weintraub, D.Baltimore, Interactions between heterologous helix–loop–helixproteins generate complexes that bind specifically to a commonDNA sequence, Cell 58 (1989) 537–544.

[9] M.L. Vetter, N.L. Brown, The role of basic helix–loop–helix genesin vertebrate retinogenesis, Semin. Cell. Dev. Biol. 12 (2001)491–498.

[10] S.E. Ross, M.E. Greenberg, C.D. Stiles, Basic helix–loop–helixfactors in cortical development, Neuron 39 (2003) 13–25.

[11] F.J. Naya, C.M. Stellrecht, M.J. Tsai, Tissue-specific regulation of theinsulin gene by a novel basic helix–loop–helix transcriptionfactor, Genes Dev. 9 (1995) 1009–1019.

[12] H. Mutoh, B.P. Fung, F.J. Naya, M.J. Tsai, J. Nishitani, A.B. Leiter, Thebasic helix–loop–helix transcription factor BETA2/NeuroD isexpressed in mammalian enteroendocrine cells and activatessecretin gene expression, Proc. Natl. Acad. Sci. U. S. A. 94 (1997)3560–3564.

[13] M.B. Breslin, M. Zhu, M.S. Lan, NeuroD1/E47 regulates the E-boxelement of a novel zinc finger transcription factor, IA-1, indeveloping nervous system, J. Biol. Chem. 278 (2003)38991–38997.

[14] Y. Liu, M. Encinas, J.X. Comella, M. Aldea, C. Gallego, Basichelix–loop–helix proteins bind to TrkB and p21(Cip1) promoterslinking differentiation and cell cycle arrest in neuroblastoma cells,Mol. Cell Biol. 24 (2004) 2662–2672.

[15] W.M. Bonner, Protein migration into nuclei. I. Frog oocyte nucleiin vivo accumulate microinjected histones, allow entry to smallproteins, and exclude large proteins, J. Cell. Biol. 64 (1975)421–430.

[16] B. Naim, V. Brumfeld, R. Kapon, V. Kiss, R. Nevo, Z. Reich, Passiveand facilitated transport in nuclear pore complexes is largelyuncoupled, J. Biol. Chem. 282 (2007) 3881–3888.

[17] N. Imamoto, T. Shimamoto, T. Takao, T. Tachibana, S. Kose, M.Matsubae, T. Sekimoto, Y. Shimonishi, Y. Yoneda, In vivo evidencefor involvement of a 58 kDa component of nuclear pore-targetingcomplex in nuclear protein import, EMBO J. 14 (1995)3617–3626.

[18] D. Görlich, U. Kutay, Transport between the cell nucleus and thecytoplasm, Annu. Rev. Cell Dev. Biol. 15 (1999) 607–660.

[19] I.G. Macara, Transport into and out of the nucleus, Microbiol. Mol.Biol. Rev. 65 (2001) 570–594.

[20] Y.M. Chook, G. Blobel, Karyopherins and nuclear import,Curr. Opin. Struct. Biol. 11 (2001) 703–715.

[21] E. Conti, E. Izaurralde, Nuclear transport enters the atomic age,Curr. Opin. Cell Biol. 13 (2001) 310–319.

[22] M. Stewart, Molecular mechanism of the nuclear protein importcycle, Nat. Rev. Mol. Cell Biol. 8 (2007) 195–208.

[23] H. Fried, U. Kutay, Nucleocytoplasmic transport: taking aninventory, Cell. Mol. Life Sci. 60 (2003) 1659–1688.

[24] Y. Yoneda, Nucleocytoplasmic protein traffic and its significanceto cell function, Genes Cells 5 (2000) 777–787.

[25] M.S. Moore, G. Blobel, The GTP-binding protein Ran/TC4 isrequired for protein import into the nucleus, Nature 365 (1993)661–663.

[26] F. Melchior, B. Paschal, J. Evans, L. Gerace, Inhibition of nuclearprotein import by nonhydrolyzable analogues of GTP andidentification of the small GTPase Ran/TC4 as an essentialtransport factor, J. Cell Biol. 123 (1993) 1649–1659.

[27] S. Kuersten, M. Ohno, I.W. Mattaj, Nucleocytoplasmic transport:Ran, beta and beyond, Trends Cell Biol. 11 (2001) 497–503.

[28] C. Hogarth, C. Itman, D.A. Jans, K.L. Loveland, Regulatednucleocytoplasmic transport in spermatogenesis: a driver ofcellular differentiation? Bioessays 27 (2005) 1011–1025.

[29] N. Yasuhara, N. Shibazaki, S. Tanaka, M. Nagai, Y. Kamikawa, S. Oe,M. Asally, Y. Kamachi, H. Kondoh, Y. Yoneda, Triggering neuraldifferentiation of ES cells by subtype switching of importin-alpha,Nat. Cell Biol. 9 (2007) 72–79.

[30] I.K. Poon, D.A. Jans, Regulation of nuclear transport: central role indevelopment and transformation? Traffic 6 (2005) 173–186.

[31] R.S. Faustino, T.J. Nelson, A. Terzic, C. Perez-Terzic, Nucleartransport: target for therapy, Clin. Pharmacol. Ther. 81 (2007)880–886.

[32] M. Fabbro, B.R. Henderson, Regulation of tumor suppressors bynuclear-cytoplasmic shuttling, Exp. Cell Res. 282 (2003) 59–69.

[33] B. Li, W. Zhang, G. Chan, A. Jancso-Radek, S. Liu, M.A. Weiss,Human sex reversal due to impaired nuclear localization of SRY. Aclinical correlation, J. Biol. Chem. 276 (2001) 46480–46484.

[34] S. Preiss, A. Argentaro, A. Clayton, A. John, D.A. Jans, T. Ogata, T.Nagai, I. Barroso, A.J. Schafer, V.R. Harley, Compound effects ofpoint mutations causing campomelic dysplasia/autosomal sexreversal upon SOX9 structure, nuclear transport, DNA binding,and transcriptional activation, J. Biol. Chem. 276 (2001)27864–27872.

[35] T. Sekimoto, K. Nakajima, T. Tachibana, T. Hirano, Y. Yoneda,Interferon-gamma-dependent nuclear import of Stat1 ismediated by the GTPase activity of Ran/TC4, J. Biol. Chem. 271(1996) 31017–31020.

[36] S.A. Adam, R.S. Marr, L. Gerace, Nuclear protein import inpermeabilized mammalian cells requires soluble cytoplasmicfactors, J. Cell Biol. 111 (1990) 807–816.

[37] K. Takano, T. Miki, J. Katahira, Y. Yoneda, NXF2 is involved incytoplasmic mRNA dynamics through interactions with motorproteins, Nucleic Acids Res. 35 (2007) 2513–2521.

[38] I. Kotera, T. Sekimoto, Y. Miyamoto, T. Saiwaki, E. Nagoshi, H.Sakagami, H. Kondo, Y. Yoneda, Importin α transports CaMKIV tothe nucleus without utilizing importin β, EMBO J. 24 (2005)942–951.

[39] Y. Yoneda, N. Imamoto-Sonobe, M. Yamaizumi, T. Uchida,Reversible inhibition of protein import into the nucleus by wheatgerm agglutinin injected into cultured cells, Exp. Cell Res. 173(1987) 586–595.

[40] C. Klebe, H. Prinz, A. Wittinghofer, R.S. Goody, The kineticmechanism of Ran–nucleotide exchange catalyzed by RCC1,Biochemistry 34 (1995) 12543–12552.

[41] M. Rexach, G. Blobel, Protein import into nuclei: association anddissociation reactions involving transport substrate, transportfactors, and nucleoporins, Cell 83 (1995) 683–692.

[42] J.M. Lingbeck, J.S. Trausch-Azar, A. Ciechanover, A.L. Schwartz, E12and E47 modulate cellular localization and proteasome-mediateddegradation of MyoD and Id1, Oncogene 24 (2005)6376–6384.

[43] E.N. Olson, W.H. Klein, bHLH factors in muscle development: deadlines and commitments, what to leave in and what to leave out,Genes Dev. 8 (1994) 1–8.

[44] J.I. Heng, S.S. Tan, The role of class I HLH genes in neuraldevelopment—have they been overlooked? Bioessays 25 (2003)709–716.

[45] J.M. Moates, S. Nanda, M.A. Cissell, M.J. Tsai, R. Stein, BETA2activates transcription from the upstream glucokinase genepromoter in islet β-cells and gut endocrine cells, Diabetes 52(2003) 403–408.

[46] J.M. Vosper, C.S. Fiore-Heriche, I. Horan, K. Wilson, H. Wise, A.Philpott, Regulation of neurogenin stability byubiquitin-mediated proteolysis, Biochem. J. 407 (2007)277–284.

[47] A.N. Goldfarb, K. Lewandowska, Nuclear redirection of acytoplasmic helix–loop–helix protein via heterodimerizationwith a nuclear localizing partner, Exp. Cell Res. 214 (1994)481–485.

Page 14: Synergistic nuclear import of NeuroD1 and its partner transcription factor, E47, via heterodimerization

1652 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 5 ( 2 0 0 9 ) 1 6 3 9 – 1 6 5 2

[48] R. Yasmin, R.M. Williams, M. Xu, N. Noy, Nuclear import of theretinoid X receptor, the vitamin D receptor, and their mutualheterodimer, J Biol Chem. 280 (2005) 40152–40160.

[49] N. Funato, S.R. Twigg, N. Higashihori, K. Ohyama, S.A. Wall, A.O.Wilkie, M. Nakamura, Functional analysis of natural mutations intwo TWIST protein motifs, Hum. Mutat. 25 (2005) 550–556.

[50] A. Longo, G.P. Guanga, R.B. Rose, Crystal structure ofE47-NeuroD1/beta2 bHLH domain-DNA complex: heterodimerselectivity andDNA recognition, Biochemistry 47 (2008) 218–229.

[51] H.Wendt, R.M. Thomas, T. Ellenberger, DNA-mediated folding andassembly of MyoD-E47 heterodimers, J. Biol. Chem. 273 (1998)5735–5743.

[52] S.S. Andrali, Q. Qian, S. Özcan, Glucose mediates the translocationof NeuroD1 by O-linked glycosylation, J. Biol. Chem. 282 (2007)15589–15596.

[53] H.V. Petersen, J.N. Jensen, R. Stein, P. Serup, Glucose inducedMAPK signalling influences NeuroD1-mediated activationand nuclear localization, FEBS Lett. 528 (2002)241–245.

[54] C. Dufton, E. Marcora, J.H. Chae, J. McCullough, J. Eby, M. Hausburg,G.H. Stein, S. Khoo, M.H. Cobb, J.E. Lee, Context-dependentregulation of NeuroD activity and protein accumulation, Mol. Cell.Neurosci. 28 (2005) 727–736.

[55] H.P. Huang, M. Liu, H.M. El-Hodiri, K. Chu, M. Jamrich, M.J. Tsai,Regulation of the pancreatic islet-specific gene BETA2 (neuroD)by neurogenin 3, Mol. Cell Biol. 20 (2000) 3292–3307.

[56] M. Peyton, C.M. Stellrecht, F.J. Naya, H.P. Huang, P.J. Samora, M.J.Tsai, BETA3, a novel helix–loop–helix protein, can act as anegative regulator of BETA2 and MyoD-responsive genes, Mol.Cell Biol. 16 (1996) 626–633.