Doublecortin-like, a microtubule-associated proteinexpressed in radial glia, is crucial for neuronal precursordivision and radial process stability

Erno Vreugdenhil,1 Sharon M. Kolk,3 Karin Boekhoorn,2 Carlos P. Fitzsimons,1 Marcel Schaaf,4 Theo Schouten,1

Angela Sarabdjitsingh,1,2 Rosana Sibug1 and Paul J. Lucassen21Division of Medical Pharmacology, Leiden ⁄Amsterdam Center for Drug Research, Leiden University Medical Center, PO Box 9502,2300 RA Leiden, The Netherlands2Center for Neuroscience, Swammerdam Institute for Life Sciences, University of Amsterdam, The Netherlands3Department of Pharmacology and Anatomy, Rudolf Magnus Institute of Neuroscience, University Medical Center Utrecht, TheNetherlands4Institute for Biology, Clusius Laboratory, Leiden University, The Netherlands

Keywords: cortical development, doublecortin, DCLK, mouse, neuroblast

Abstract

During corticogenesis, progenitors divide within the ventricular zone where they rely on radial process extensions, formed by radialglial cell (RG) scaffolds, along which they migrate to the proper layers of the cerebral cortex. Although the microtubule-associatedproteins doublecortin (DCX) and doublecortin-like kinase (DCLK) are critically involved in dynamic rearrangement of the cytoskeletalmachinery that allow migration, little is known about their role in early corticogenesis. Here we have functionally characterized amouse splice-variant of DCLK, doublecortin-like (DCL), exhibiting 73% amino acid sequence identity with DCX over its entire length.Unlike DCX, DCL is expressed from embryonic day 8 onwards throughout the early neuroepithelium. It is localized in mitotic cells,RGs and radial processes. DCL knockdown using siRNA in vitro induces spindle collapse in dividing neuroblastoma cells, whereasoverexpression results in elongated and asymmetrical mitotic spindles. In vivo knockdown of the DCLK gene by in uteroelectroporation significantly reduced cell numbers in the inner proliferative zones and dramatically disrupted most radial processes.Our data emphasize the unique role of the DCLK gene in mitotic spindle integrity during early neurogenesis. In addition, they indicatecrucial involvement of DCLK in RG proliferation and their radial process stability, a finding that has thus far not been attributed to DCXor DCLK.

Introduction

Generation of the cerebral cortex is a complex and tightly controlledprocess involving a multitude of proliferation-, migration- anddifferentiation-related events. In mice, cortical development starts asearly as embryonic day (E)8 when the neuroepithelium lines theneuronal tube. Following massive proliferation, neuronal progenitorsline up to form the preplate at !E12 (Gupta et al., 2002). From E12onwards a second phase of development can be distinguished in whichthe preplate splits into the marginal zone and subplate in betweenwhich the cortical plate (CP) will form. In this second phase, waves ofneural precursor cells (NPCs) migrate radially, guided by fibre tracts ofthe radial glial cell (RG) scaffold (Schmechel & Rakic, 1979; Guptaet al., 2002; Marin & Rubenstein, 2003; Kriegstein & Noctor, 2004).

Various genes encoding microtubule-associated proteins (MAPs)regulating radial migration have been shown to be crucial for thesecond phase of cortical development. For example, doublecortin(DCX) induces phosphorylation-dependent microtubule stabilizationand is important for neuronal migration, growth of neuronal processes

and correct cortical layering (Francis et al., 1999; Gleeson et al., 1999;Taylor et al., 2000; Bai et al., 2003; Friocourt et al., 2003; Gdalyahuet al., 2004; Schaar et al., 2004; Tanaka et al., 2004). Interestingly,although DCX mutations cause migratory defects associated with theDCX syndrome in humans (des Portes et al., 1998; Gleeson et al.,1998), DCX-null mutations in mice neither disrupt neuronal migrationnor cause DCX syndrome (Corbo et al., 2002). However, in uteroRNA-interference of DCX in rats (Bai et al., 2003; Ramos et al.,2006) does create an animal model of DCX syndrome, suggesting thepresence of related complementary genes.The human and rodent genome encode a gene, doublecortin-like

kinase (DCLK), that has substantial sequence identity to DCX and issubject to extensive alternative splicing (Matsumoto et al., 1999;Burgess & Reiner, 2002; Engels et al., 2004). One of its alternativetranscripts, DCLK-long, encodes a protein of 756 amino acids thatcontains two DCX domains and a functional kinase domain. DCLK-long is expressed from E12 onwards (Omori et al., 1998; Mizuguchiet al., 1999; Lin et al., 2000) and is, like DCX, capable of microtubulepolymerization (Omori et al., 1998; Mizuguchi et al., 1999; Burgess &Reiner, 2000; Lin et al., 2000). Interestingly, the analogue inC. elegans, zyg-8, is involved in mitotic spindle positioning (Gonczyet al., 2001). Consistent with this, recent studies in mice have

Correspondence: Dr E. Vreugdenhil, 1Division of Medical Pharmacology, as above.E-mail: vreugden@lacdr.leidenuniv.nl

Received 6 July 2006, revised 8 November 2006, accepted 27 November 2006

European Journal of Neuroscience, Vol. 25, pp. 635–648, 2007 doi:10.1111/j.1460-9568.2007.05318.x

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd

implicated the DCLK gene in mitotic spindle formation (Shu et al.,2006), migration (Koizumi et al., 2006) and vesicle transport (Deuelet al., 2006). Given this, the DCLK and DCX genes may function inpartially redundant pathways in the second phase of cortical devel-opment (Koizumi et al., 2006).Although the functions of DCX and DCLK have been well-

characterized in the second phase of cortical development, little isknown about their possible involvement in the early phases ofcortical development. Using a variety of different techniques wehave identified and characterized another splice variant of theDCLK gene, doublecortin-like (DCL), encoding a 362-amino acidprotein exhibiting high sequence identity with DCX over its entirelength but missing the kinase domain of DCLK-long. We showspecific expression of DCL, but not of DCX, in the first phase ofcortical development, indicating a unique role for the DCLK gene.In addition, we provide experimental evidence indicating a crucialrole for the DCLK gene in the correct construction of the RGscaffold.

Materials and methods

Cloning of the murine DCL

We have cloned a novel CaMK-related peptide, called CARP(Vreugdenhil et al., 1999). As the predicted C-terminal amino acidsof CARP are highly similar to the C-terminal 17 amino acids of DCXand are located on a single exon (exon 8 in Fig. 1A), we suspected analternative splice product of the DCLK gene which could combine thisexon with upstream-located DCX-like exons. To test this, we havedeveloped an antisense primer 1A: CTGGAATTCTTACACTGAG-TCTCCTGAG (EcoR1 site underlined) corresponding to the stop-codon region of the CARP-specific exon and a sense primer 2S:GCAGGTTCTCACTGACATTACCG corresponding to exon 3 of themurine DCLK gene. As predicted, in 30 cycles of PCR we amplified a457 bp fragment using mouse embryonic cDNA as a template andpolymerase PfuI (Stratagene, Amsterdam, The Netherlands).DNA sequence analysis confirmed that the DNA sequence was

DCLK-specific. Subsequently, a DCL cDNA encoding the completeDCL protein was amplified using CCAGGATCCACCATGTCGTT-CGGCAGAGATATG (BamH1 site underlined) as a sense and 1A asan antisense primer. It was cut with BamH1 and EcoR1 and subclonedin the expression plasmid pcDNA 3.1 (InVitrogen, Groningen, TheNetherlands).

RT-PCR experiments

Total RNA was extracted from mouse heads collected at E8, E10, 12,14, 16 and 18 as well as from the adult mouse brain, and reverse-transcribed into cDNA using reverse transciptase of Gibco-BRL(Breda, The Netherlands). DCL fragments were generated in 30 cyclesusing primers 1A and 2S described above. As a positive control for thecDNA synthesis, actin primers CATCGTCACCAACTGGGACG-ACA (sense) and GCTCGGTGAGGATCTTCATGAGG (antisense)were used. A negative control for each cDNA synthesis was includedby the omission of the reverse transcriptase. To address a possibledevelopmental role of CARP, cDNAwas amplified in 32 cycles usingthe CARP-specific sense primer AGAAGAGGCTCTGGCTCTTGand the DCL- and CARP-specific primer 1A. In order to verify cDNAamplification in the logarithmic phase, different numbers of cycleswere applied. The identity of DCL and CARP RT-PCR fragments wasconfirmed by DNA sequence analysis. Experiments were performed inquadruplicate with similar results.

In situ hybridization

DCL mRNA includes exon 8 (Fig. 1), which is absent in most otherDCLK transcripts except for CARP. Because CARP is expressed atvery low levels during embryonic development (see RT-PCR experi-ments) we have developed a 40-mer antisense oligonucleotide 5¢-TTTGCTGTTAGATGCTTGCTTAGGAAATGGGAAACCTTGA-3¢,complementary to position 280–241 of AF045469 and specific toexon 8. Therefore, the in situ hybridization signal obtained with theseprobes represents DCL and not DCLK-long. As a negative control wehave used.5¢-TTTGATGTTATATGCTTGATTAGGACATGGGACACCT-

GGA-3¢, which contains six mismatches (underlined). Both oligonu-cleotides were end-labelled with a-33P dATP (NEN Life ScienceProducts, Hoofddorp, The Netherlands; 2000 Ci ⁄mmol, 10 mCi ⁄mL)using terminal transferase, according to the manufacturer’s instruc-tions (Roche Molecular Biochemicals, Almere, The Netherlands).In situ hybridization and visualization of the signals were per-formed as described before (Vreugdenhil et al., 2001; Engels et al.,2004).

Antibodies

The anti-DCLK antibody has been described previously (Kruideringet al., 2001). Mouse monoclonal anti-tubulin was obtained fromSigma (Nieuwegein, The Netherlands). Goat polyclonal anti-DCX(C-18) antibody, rhodamine-conjugated secondary antibodies andhorseradish peroxidase-conjugated secondary antibodies were fromSanta Cruz Biotechnology, Inc. or from Chemicon (guinea pig anti-DCX polyclonal antibody).

Cell culture and treatments

All cell culture chemicals were obtained from Life Science Technol-ogies, Inc. (Woerden, The Netherlands) unless otherwise stated. Allcells were maintained at 37 !C, 5% CO2. COS-1 cells were cultured inDulbecco’s modified Eagle’s medium (DMEM), supplemented with100 units ⁄mL penicillin, 100 lg ⁄mL streptomycin and 10% fetalbovine serum. N1E-115 cells were cultured in DMEM without sodiumpyruvate, supplemented with 100 units ⁄mL penicillin, 100 lg ⁄mLstreptomycin, hybridoma (HAT) mix and 10% fetal bovine serum. Fortransient transfection experiments, cells were cultured on plates orcoverslips coated with poly l-lysine.

SiRNA experiments

For siRNA experiments, the mouse neuroblastoma cell-line N1E-115(ATCC number CRL-2263) was used because it was found toendogenously expresses DCL (data not shown). Three differentsynthetic RNA oligos, 5¢-GCTCACTCCTTCGAGCAGGTT-3¢ and5¢-CCTGCTCGAAGGAGTGAGCTT-3¢ (annealed siDCL-1), 5¢-CA-AGAAGACGGCUCACUCCTT-3¢ and 5¢-GGAGUGAGCCGUCU-UCUUGTT-3¢ (annealed siDCL-2) and 5¢-GAAAGCCAAGAAGG-UUCGATT-3¢ and 5¢-TCGAACCUUCUUGGCUUUCTT-3¢ (annealedsiDCL-3), in which the 3¢ thymidines are deoxynucleotides, wereobtained from Eurogentec. For gene silencing, 60 pmol siRNA duplexwas dissolved in 50 lL opti-MEM (Life Technologies, Woerden, TheNetherlands) and mixed by pipetting with 3 lL oligofectamine reagent(Invitrogen, Groningen, The Netherlands), and dissolved in 12 lLopti-MEM. After 20 min of incubation at room temperature, thevolume was increased with 32 lL opti-MEM and the total mixture

636 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

Fig. 1. Genomic organization of DCL and alignment with DCX. (A) Genomic organization of the DCLK gene and the cloning strategy of the DCL cDNA. Onlythe exon–intron structure of the DCL part is indicated, including the recently identified exon 8 encoding the common 3¢-end of CARP as well as DCL (Vreugdenhilet al., 2001). Exons are represented by rectangles and indicated by arabic numbers, while introns are solid lines. The DCL transcript derived from this part of theDCLK gene is indicated below (DCL) the genomic structure. The open reading frame is represented by a rectangle and nontranslated sequences by lines. For thegenomic organization of the DCLK gene and biosynthesis of all DCLK transcripts we refer to Engels et al. (2004). The part of the protein that overlaps with theunique synthetic peptide sequence used to generate the DCLK antibody is indicated in (nonhatched) grey, and involves the combined spliced sequence of exons 7and 8. The synthetic oligonucleotide, used in the in situ hybridization experiments and labelled with 33P, is indicated below the DCL mRNA. It is complementary tothe 3¢-untranslated region that is derived from exon 8. This exon is incorporated into the DCL mRNA but not into that of DCLK-long. The location of the primersused to clone DCL is indicated by arrows. The position of siDCL-1, -2 and -3 used to silence expression of DCL are indicated by bars. (B) Alignment of the mouseDCL protein with DCX. Identical residues are dark grey and conserved substitutions are light grey. The two DCX domains and the subplate-rich domain areindicated by arrows.

DCL directs neuronal precursor division 637

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

(100 lL) was added to the cells (500 lL). After 48 h, gene silencingwas tested by Western blot analysis and immunofluorescence. Theoptical density of DCL bands was measured and the percentage DCLknockdown was calculated from four independent experiments bysetting mock-transfected cells as a 100% reference.

Immunocytochemistry

Cells were transfected as described above. At the indicated times,cultured cells were fixed with 80% aceton in water onto coverslips.Following a washing and block with 5% normal goat serum (Sigma,Nieuwegein, The Netherlands) in PBS, primary antibody incubationwas followed by rhodamine-conjugated secondary antibodies. Nucleiwere stained with 0.2 lg ⁄mL Hoechst 33258 and images wereobtained with an Olympus AX70 fluorescence microscope coupled toa Sony 3CCD colour video camera operated by AnalysisR software(Soft Imaging System Corp.).To obtain information on its in vivo role, DCL protein distribution

was mapped in CD1 mouse embryos of E9, 10 and 11 that weredissected, briefly washed in PBS and then fixed for 4 h in methanol,acetone and water (40 : 40 : 20). They were then stored for 2 weeksin 70% ethanol and embedded in Paraplast Plus (Kendall, TycoHealthcare, Mansfield, MA 02048, USA). Six-micrometre-thicksections were cut. After clearing and rehydration, nonspecific bindingwas reduced using 1% milk powder solution in PBS. The primaryDCL antibody was applied at a 1 : 50 dilution in 0.25% gelatin and0.5% triton X-100 in TBS (Supermix). The secondary antibody wasbiotinylated antirabbit (1 : 200; Amersham Life Sciences,’s-Hert-ogenbosch, The Netherlands) which was amplified two times withavidin–biotin (ABC) Elite (Vector Laboratories, Burlingame, CA,USA) and biotinylated tyramide (1 : 500) with 0.01% peroxide.Chromogen was visualized with diaminobenzidine, and sections werecounterstained with Cresyl Violet.For comparison, DCX protein distribution was studied in adjacent

sections using the C-18 DCX antibody (Santa Cruz Biotechnology,South Cruz, CA, USA) at a 1 : 75 dilution. The same protocol asabove was used except for omission of the blocking step and the use ofa biotinylated antigoat secondary antibody.

Double and triple immunofluorescence

Sections were mounted on Superfrost Plus slides (Menzel-Glaser),dewaxed, hydrated and incubated in Zamboni mix (paraformaldehyde,2%; picric acid, 0.1%; and gluteraldehyde, 0.025%; in 0.1 mphosphate buffer, pH 7.4). To retrieve masked antigen epitopes,microwave pretreatment was performed in 0.1 m citrate buffer,pH 3.0, for 15 min. Primary antibodies were C-18 (used at 1 : 100),DCL (used at 1 : 50) and vimentin (used at a 1 : 150 dilution in BSA,0.1%; Triton X-100, 1%; and donkey normal serum, 3% in TBS.Secondary antibody was conjugated to Alexa Fluor488 (1 : 400,antigoat; Molecular Probes, Oregon, USA), Cy3 (1 : 400, antirabbit;Jackson Immunoresearch) or Alexa Fluor647 (1 : 400, antimouse;Molecular Probes). Sections were mounted in Vectashield containingDAPI or Hoechst and examined using a Zeiss LSM510 confocalmicroscope.

Protein extraction and Western blotting

Mouse tissue and cells were solubilized with lysis buffer (triethanol-amine pH 7.5, 20 mm; NaCl, 140 mm; deoxycelate, 0.05%; sodiumdodecyl sulphate, 0.05%; and Triton X-100, 0.05%), supplemented

with CompleteTM EDTA-free protease inhibitor mixture (RocheMolecular Biochemicals) and centrifuged at 16 000 g for 30 min.Supernatant was collected and the protein concentration was deter-mined using the Pierce method. Equal amounts of protein (25 lg celllysate) were seperated by SDS-PAGE and transferred to immobilon-PPVDF membranes (Millipore). Blots were blocked in Tris-bufferedsaline with 0.2% Tween 20, with 5% milk and incubated with primaryantibodies (anti-DCLK, 1 : 1000; and monoclonal anti-a-tubulinDM1A, 1 : 1000; Sigma–Aldrich, The Netherlands) and horseradishperoxidase-conjugated secondary antibodies. Antibody binding wasdetected by ECL (Amersham Pharmacia Biotech).

In utero electroporation (IUE)

All experiments with embryos, including the ones with IUE, wereapproved by the animal experimental committee of the University ofLeiden. All animal use and care during IUE were in accordance withinstitutional guidelines: University of Utrecht Experimental AnimalCommittee protocol no. 05.08.076. The IUE experiments using DCLsiRNA were performed in C57Bl ⁄ 6JolaHsD mice purchased fromHarlan Laboratories (Horst, The Netherlands). Killing was performedby cervical dislocation. The day of the vaginal plug was consideredE0.5 and the day of birth as postnatal day (P)0.Cells lining the ventricle of the developing cortex were transfected

with DCL siRNA in vivo by means of IUE. Timed pregnantC57Bl ⁄ 6JolaHsD mice were anaesthesized with a cocktail of ketamineand xylazine (100 mg ⁄ kg and 10 mg ⁄ kg, respectively). Followinglaporatomy, plasmid DNA (2 lg ⁄ lL for the control plasmids pCMV-YFP and 2 lg ⁄ lL for pSuper-DCL183 and pSuper-DCL183 mis-match in Tris-buffered 0.02% Fast Green) was injected through theuterine wall into one of the lateral ventricles of E14.5 embryos usingcalibrated pulled glass capillaries (WPI, Sarasota) in a survival surgeryset-up. Double-stranded small hairpin (sh)DNA was subcloned inpSuper (Brummelkamp et al., 2002). For unexplained reasons, double-stranded shDNA sequences based on the previously used siDCL-2 andsiDCL-3 sequences did not result in efficient DCL knockdown whentested in N1E-115 cells. Therefore, novel shDNA sequences weredesigned, subcloned in pSuper and tested in the same cells. The mostefficient construct, pSuper-DCL183 (CCAAGCTAAGATGTCTTTA),gave 86% DCL knockdown (determined from three independentexperiments). As a negative control we used pSuper-DCL mismatch(CCACGCTAAGCTGTCTTTA): < 5% DCL knockdown in N1E-115cells.One hemisphere was used for the mismatch or the DCL siRNA and

the other served as an internal control. A series of five unipolar square-wave current pulses (33 V max) were delivered over the embryo’shead using ‘Tweezertrodes’ (BTX, Harvard Apparatus). Embryoswere placed back into the dam and gestation was allowed to proceed.Embryos were harvested 1 day later and processed for immunocyt-ochemistry. Only embryos showing effective electroporation withinthe target area were included in the analyses.Pictures were taken using either a Zeiss Axioplan epifuorescence

microscope or a Leica TCS NT confocal microscope. Alexa488 wasexcited at 488 nm and Alexa555 at 568 nm, and emissions weredetected at 500–540 and 580–650 nm, respectively. Excitation of thetwo fluorophores was performed sequentially.

Data analysis

At least three embryos per group (control vs. DCL siRNA) wereanalysed per timepoint. Only animals in which the cortex was broadly

638 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

transfected were included in the analysis. Optical fields of 3–7 sectionsper animal, at least 80 lm apart, and at the same neuroanatomical levelin each group, were captured with specialized software. A 10·objective of a Zeiss Axioplan was used and a bin of 0.2 mm wide wasplaced in the centre of the transfected area. The length of eachembryonic zone [proliferative–ventricular zone (PZ), intermediate zone(IZ) and cortical plate (CP)] was measured per bin in each optical field.Yellow fluorescent protein (YFP)-positive cells were quantified perzone in each optical field. Data were normalized to total number ⁄mm2,averaged for each embryo, pooled and tested for significance byone-way anova (a " 5%) and expressed as mean ± SEM.

Results

DCL, a microtubule-associated protein, was abundantlyexpressed during the earliest stages of neocortical developmentand preceded the onset of DCX expression

DNA sequence analysis of a DCL cDNA clone revealed an openreading frame of 362 amino acids with a predicted molecular mass of40 kDa (Fig. 1B) which also showed a 73% amino acid identity (81%similarity) with mouse DCX over the entire length (see Fig. 1). Similarto DCX, DCL is a MAP which can stabilize microtubules (seeSupplementary material, Fig. S1).

To study a possible role for DCL in early neurogenesis, we firstanalysed the spatiotemporal expression pattern of DCL mRNA duringearly embryonic development. RT-PCR with DCL-specific primersclearly showed mRNA expression in embryo heads of E8 and E10,with the highest expression found at E12 and E14 (Fig. 2). Theexpression slowly declined after E14. In contrast to DCL, expressionof CARP, a closely related but different splice variant (see Materialsand methods; also Vreugdenhil et al., 1999) was nearly absent duringembryonic development (Fig. 2) and did not increase in intensityduring embryonic development.

Using in situ hybridization with an oligonucleotide probe recogni-zing the 3¢ UTR of the DCL transcript but not the DCLK-longtranscript (see Materials and methods and Fig. 1A) we observed(Fig. 3A) DCL mRNA expression from E8 onwards along the lengthof the early neuroepithelium, a major site of neurogenesis (Noctoret al., 2001; Haubensak et al., 2004). At E10, when massive precursordivision occurs, substantial DCL expression was found in the earlydiencephalon, telencephalon, mesencephalon and neural tube(Fig. 3B). Consistent with our immunocytochemical data (see below),

DCL expression at E12 (Fig. 3C) was greatly elevated throughout themain neurogenic regions of the developing CNS, relative to E8–10.Mismatch controls yielded no signal (Fig. 3D).We have previously raised an antibody using synthetic CARP, a

splice variant of the DCLK gene, as antigen (Vreugdenhil et al., 1999;Kruidering et al., 2001); Western blot analysis (Fig. 4) of embryonicbrain homogenates probed with this antibody revealed a 40 kDaprotein and no other DCLK-related immunoreactive bands, e.g. the80 kDa DCLK-long (Burgess & Reiner, 2002) or the 80 kDa DCK2protein (Edelman et al., 2005), indicating that this antibody recognizesspecifically DCL during embryonic development. In line with our RT-PCR experiments, CARP protein in embryonic lysates was belowdetection level in our Western blot analysis. DCL protein on blot wasexpressed from E10 onwards with the highest levels of DCL proteinfound at E12 and E14, after which expression levels graduallydeclined. In contrast to DCL, DCX protein expression was notdetected until E12, not even after prolonged exposure (Fig. 4B).We next performed immunocytochemistry to map the spatiotem-

poral distribution of DCL protein at E8–11. DCL was expressed in themain neuroepithelia from E9 onwards, an age at which DCX wasnotably still absent (Fig. 5, compare B with C). At E10 and 11, DCLwas strongly and selectively expressed in the telencephalon, dien-cephalon, dorsal root and sympathetic ganglia whereas non-neuronaltissues were DCL-negative (Fig. 5A). DCL was expressed in manyradial processes originating from the neuroepithelia and extendingradially towards the pial surface. Transverse cross-sections illustrateda differential and only partly overlapping distribution pattern of DCXand DCL. DCX expression commenced at !E11, primarily intangentially orientated processes in the early cortical plate andmarginal zone (Fig. 5H), whereas DCL was prominently expressedthroughout the ependymal zone from E9 onwards (Fig. 5C, D, F, I, Jand K); the ependymal zone was DCX-negative at this age (Fig. 5B).Striking DCL immunoreactivity was found in mitotic cells throughout

Fig. 2. RT-PCR analysis of DCL mRNA expression in the embryonic heads.cDNA, reverse transcribed from RNA from mouse embryonic heads at agesE8–18, was used with DCL-specific primers. Actin primers were used as apositive control for cDNA synthesis and a minus reverse transcriptase (–RT)‘cDNA’ sample was used as a negative control.

Fig. 3. In situ hybridization analysis of DCL mRNA expression in theembryonic brain. In situ hybridization of (A) a transverse brain section at E8and (B and C) sagittal sections at E10 and E12, respectively. (D) A saggitalsection at E12 was hybridized with a mismatch oligonucleotide as negativecontrol. On E8, clear signals were detected in the neuroepithelium (ne). OnE10, a clear signal was detected in the neuroepithelium of the neuronal tube(n.e. o.n.t)–early spinal cord, and in the diencephalon (di) and mesenceophalon(me). On E12 a strong signal was observed in the main neurogenic regionssurrounding the ventricles. No signal was observed in the mismatch control. lv,lateral ventricle; mo, medulla oblongata; mt, metencephalon; mv, mesenche-phalic vesicle; nc, neopallial cortex; ng, neural groove; rh, rhombencephalon;te, telencephalon; tv, telencephalic vesicle; IV v, fourth ventricle. Exposuretime, 14 days. Scale bars, 1 mm.

DCL directs neuronal precursor division 639

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

the VZ, the neocortex and neural tube as well as in blast-like cellsengaged in cell cycle (Fig. 5E, F and K–O). In mitotic cells, DCL wasseen in close association with centrosome-like structures (Fig. 5M)and also observed between the separated chromosomes (Fig. 5K, L, Nand O) as well as in the periphery of a mitotic cell. Although we do notaddress this in detail here, this pattern suggests that DCL is at least notselective for either the kinetochore (middle arrow in Fig. 5K) or thepolar microtubule population (arrowhead on the right in Fig. 5J).To examine the possibility that DCL is a neuronal precursor marker

and to address its possible overlap with DCX, we performed tripleimmunofluorescent labelling for DCL, DCX and the intermediatefilament protein vimentin which identifies neurogenic radial glia (seeFig. 6; also Noctor et al., 2001; Gotz et al., 2002; Malatesta et al.,

2003; Anthony et al., 2004). Triple labelling revealed a differentialdistribution pattern of DCL and DCX in early development with anearly complete overlap of DCL expression with vimentin-positivecells at E12 (Fig. 6G–J) mainly in the VZ, but also in individualmitotic cells (Fig. 6K). Vimentin+, DCL+ and DCX+ triple-labelledcells were only seen in the IZ (white cells in Fig. 6K), consistent withthe presence of migratory neuronal precursors undergoing secondarydivisions in these regions (Kriegstein & Noctor, 2004). In contrast,DCX did not overlap with vimentin in other regions of the neocortexor spinal cord (Fig. 6K and N). Many long DCL+ radial processeswere devoid of DCX signal (Fig. 6L and M). DCL+ mitotic cells in theVZ displayed vertical as well as horizontal cleavage plane orientations(Fig. 6A, B and D). Taken together, these findings show that DCLmRNA and protein expression were abundant in the main neurogenicregions during early cortical development, where they occurred inradial processes and mitotic cells. The spatiotemporal distribution ofDCL was different from that of DCX and the onset of expression ofDCL preceded that of DCX.

Suppression of DCL function by siRNA affectedmicrotubule architecture and disrupted the mitoticspindle in neuroblastoma cells

Our immunohistochemical analysis suggested the presence of DCL individing neuronal precursors where it seemed to colocalize withmitotic spindles. To study the subcellular location of DCL in moredetail and to address its possible function we first screened a numberof different cell lines for endogenous DCL expression and found thatDCL was expressed endogenously only in several neuroblastoma celllines but not in any other cell line, including PC12 cells (data notshown), suggesting specific DCL expression in cells with a neuroblast-like phenotype.To address the subcellular localization of DCL we performed

immunocytochemistry on N1E-115 cells, a mouse neuroblastoma cellline expressing high levels of DCL as indicated by Western blotanalysis (data not shown). In line with our in vivo observations, strongDCL immunoreactivity was observed in all dividing N1E-115 cellsduring metaphase or early anaphase (see Fig. 7B and E). a-Tubulincolabelling (Fig. 7C) and DNA staining with Hoechst (Fig. 7A)confirmed that DCL was expressed in association with the mitoticspindles, suggesting a role for DCL in mitotic spindle formation.

Fig. 4. Western blot analysis of DCL (A) and DCX (B) in embryonic lysatesobtained at E8–18. Note that the 40 kDa DCL protein was expressed at E10,before any other bands were visible. DCX protein expression was observedfrom E12 onwards. Western blot analysis was performed with uniquemembranes for DCX and DCL using the same batch of embryonic lysates.

Fig. 5. Immunocytochemical analysis of DCL and DCX protein expression in the embryonic brain. (A) At E11, DCL protein (sagittal section) was foundthroughout the telencephalon and diencephalon (left and right, respectively). DCL was found in the outer layers close to the pia as well as in the inner VZ(arrowheads; see also below). Non-neuronal tissue such as the mandibular component of the first branchial arch (M) was devoid of any signal. IV, fourth ventricle.(B and C) Adjacent transverse (coronal) sections from the early neuroepithelium at E9 immunostained for (B) DCX and (C) DCL. DCX was absent at this age(arrowheads in B) whereas DCL was abundantly expressed in the inner ependymal (upper two arrows in C) and outer early preplate–marginal zone region (lowerarrow pointing leftward in C). (D) Sagittal section of the neuroepithelium of the neural tube at E11, showing abundant expression at the luminal border(arrowheads) while, in the developing neuronal tissue, various mitotic cells expressed DCL (arrows). (E) Detail of a DCL+ mitotic cell in the neocortical VZ.Arrowhead points to the chromosomes orientated in the midline cleavage plane. (F) DCL expression in the neuroepithelium and VZ (arrowhead on the left) of theearly telencephalon at E10. Many DCL+ processes extended radially towards the pial surface (arrowheads in upper left corner). In the preplate (arrowhead on theright), DCL was also found in tangentially orientated processes. DCL+ mitotic cells in the IZ are also visible (arrows). (G) Immunostaining for the intermediatefilament protein vimentin, which identifies neurogenic radial glia, shows immunopositive cellbodies (arrow) in the lumenal surface (asterisks) and radial processesrunning through the VZ and SVZ (E11). (H and I) Transverse cross-sections illustrating the differential yet partly overlapping distribution of DCX and DCL.(H) DCX was expressed at E11 in the upper preplate and CP–marginal zone (arrow). (I) In contrast, DCL was already strongly expressed at E9 in the VZ–ependymal layer (arrowhead to the left), with DCL+ radial processes extending towards the IZ (arrowhead on the right). The ventricular layer (asterisks in H) wasdevoid of DCX but not of DCL signal. (J) Detail of the VZ at E9 showing DCL expression in processes radially extending into the SVZ and IZ (middle arrowhead).A mitotic cell lacking DCL expression between the chromosomes (right arrowhead) is also visible. (K) Detail of the VZ showing DCL+ dividing cells, which appearto be in telophase (left) and anaphase (middle), with a vertical cleavage plane. The cell on the left is probably in telophase or may represent two cells in interphase.A DCL+ cell on the right is visible with DCL expression between the chromosomes, with a horizontal cleavage orientation (arrowheads). (L) DCL immuno-reactivity in the VZ at E11, in prophase and telophase cells (arrowheads) and in a blast-like precursor cell in metaphase or anaphase (arrow). (M) Two DCL+ mitoticcells in the neuroepithelium displaying expression between chromosomes (upper arrow) as well as in structures resembling centrosomes (lower arrows). (N and O)DCL immunoreactivity in dividing cells in anaphase II or telophase II (N) and in metaphase or anaphase I, with the chromosomes clearly visible (arrow in O), andin a radially extending process (arrowheads). Scale bars, 150 lm (A), 25 lm (B and C), 70 lm (D), 5 lm (E), 15 lm (F and G), 12 lm (J), 8 lm (K), 10 lm(H, I and L), 1.5 lm (M), 1 lm (N and O).

640 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

Functional aspects of DCL were addressed using three differentsiRNA molecules against DCL (see Fig. 1A). Western blot analysisindicated that siDCL-1 failed to knock down DCL protein (Fig. 7E,lane 1), which might relate to the lack of TT dinucleotides at the 3¢-endin this antisense strand. SiDCL-1 was subsequently used as a negativecontrol for possible off-target effects. Compared to nontreated andsiDCL-1, transfection with siDCL-2 and siDCL-3 induced a knock-down of, respectively, 80% and > 90% (Fig. 7A, lanes 2 and 3).

Our immunohistochemical data suggested a role for DCL inproliferative division of neural precursors. Therefore, we have studieddividing N1E-115 cells after DCL knockdown by siRNA technology.Strong DCL immunoreactivity was observed in all dividing N1E-115cells during metaphase or early anaphase (see Fig. 7B and F).a-Tubulin colabelling confirmed that DCL was expressed in associ-ation with the mitotic spindles and near the centrosome, suggesting arole for DCL in mitotic spindle formation. Indeed, DCL knockdown

DCL directs neuronal precursor division 641

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

by siDCL-2 and siDCL-3 caused a dramatic collapse of the mitoticspindle (Fig. 7I–N) in ! 40% of all dividing cells (10 out of 24 cells inmetaphase in two independent experiments) for siDCL-2 and in alldividing cells (30 out of 30, in two independent experiments)transfected with siDCL-3. Inefficient knockdown by siDCL-1 leftDCL colocalization as well as the mitotic spindle unchanged (Fig. 7Fand G). Also, integrity of the centrosomes was inspected by c-tubulinlabelling, which was not affected when DCL was knocked down (seeSupplementary material, Fig. S2). Clearly, DCL is involved in mitoticspindle stability in neuroblastoma cells. Combined with the in vivodistribution of DCL, our siRNA experiments suggest a role for DCL inneuronal precursor division.

DCL overexpression induced elongated and aberrant mitoticspindlesNext, we tested the effect of DCL overexpression on mitotic spindleformation in dividing COS-1 cells, a cell line that does not expressDCL endogenously. Consistent with our earlier observations, DCLcolabelled mitotic spindles in transfected COS-1 cells (see Fig. 8).DCL gain-of-function in these cells induced two different phenotypes;the first, observed in 20% (n " 126) of the dividing COS-1 cellsanalysed, was characterized by DCL colabelling with a-tubulin. Thisphenotype is similar to the endogeneous pattern in dividing N1E-115cells (Fig. 8D–F). The second phenotype was present in 80%(n " 603) of the cells in M phase and was characterized by

Fig. 6. Confocal double-immunofluorescence analysis of DCL and DCX protein expression in the embryonic brain. (A–D) DCL expression in (D; E10) the VZand in mitotic cells at the ventricular surface (arrows in D). DCL appears to be associated with both (A, arrow) kinetochore microtubules and(B) polar microtubules (cytoplasmic staining surrounding the chromatin). Blue stain, Hoechst-identified chromatin. (Inset C) Mitotic cells are DCX negative,indicating a function of DCL different from DCX at these ages. (E and F) At E11, many DCL+ precursor cells (arrows) were found in the higher regions of the SVZ–IZ. These blast cells resemble those shown in Fig. 5D, and were DCX-negative; DCX+ processes are in close apposition to, but not overlapping with, the mitoticDCL+ blasts (inset E). (G–J) Triple labelling of the neuroepithelial layer of the VZ of an E13 mouse revealed expression of DCL in vimentin+ RG and neuralprecursors (merged image in J). (H, arrow) A mitotic cell orientated close to the ventricular surface is visible. (K) Triple labelling (E11) showing DCL expressionin vimentin+ RG cells (purple) throughout the neuroepithelium and VZ and in neuroblasts in the SVZ–IZ. This distribution pattern is distinctly different from DCX(green), with weak expression in a.o. tangentially orientated migratory fibres in the outer layers of the VZ–SVZ and more robust expression in the early preplate–primitive plexiform zone (PPZ) at this age. Only in the second proliferative layer between (S)VZ and PPZ, where migrating precursors can undergo M-phase again(Kriegstein & Noctor, 2004), are vimentin+, DCX+ and DCL+ triple-labelled (white) cells found (arrows). Precursors in the VZ–lower SVZ are only double-positivefor vimentin and DCL (purple cells). (L) DCX and DCL colabelling of the spinal cord (E11) reveals DCL expression at upper and lower border regions (arrows)with DCL+, DCX– (see M for a higer magnification) long RG-like fibres (arrowheads) extending radially (*), amid extensive (nonoverlapping) DCX expression.(M) Detail of L. (N) Unlike DCL, DCX expression was not expressed in vimentin+ RG (two lower arrows) in the neural tube at E11. Only very rarely wascolabelling seen, indicating neuronal migration under guidance of RGs. Scale bars, 5 lm (A, B and E), 12 lm (D and G–J), 20 lm (F and M), 16 lm (K), 60 lm(L), 18 lm (N).

642 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

abnormally orientated (Fig. 8B) elongated asymmetrical monopolarmitotic spindles. This phenotype was also characterized by an aberrantsegregation of the chromosomes (blue stain in Fig. 8C). The elongatedasymmetrical spindles of cells overexpressing DCL were clearlydifferent from the stereotypical bipolar spindles in control cells (‘ref’in inset in Fig. 8B). Clearly, the overexpression of foreign proteins wasartificial and interpretation of these results should therefore be judgedcarefully. However, they do show that DCL can regulate mitoticspindle stability and length, consistent with another recent study (Shuet al., 2006) and with the described polymerization and stabilization

effects of DCX domains on microtubules (Francis et al., 1999; forreview see Couillard-Despres et al., 2001).

RNA interference in utero demonstrated an in vivo role for DCLin radial process stability and progenitor division

To study the in vivo role of the DCLK gene in RG proliferation and instability of the radial processes, we performed IUE using RNAinterference in cells lining the ventricle of the developing cortex.Plasmids encoding effective short-hairpin RNA molecules were

Fig. 7. DCL knockdown led to deformation of mitotic spindles. (A–D) DCL colocalized with mitotic spindles in N115 cells. Confocal analysis of mitotic N115cells showed that DNA staining with DAPI (A) colocalized with both (B) DCL and (C) a-tubulin. (D) Merged picture. (E) Effectiveness of RNA interference inknocking down DCL. Western blot analysis of DCL expression in N1E-115 cells with (1–3) and without (4) siRNA treatment, performed in duplicate. Three differentsiRNA molecules targeting DCL were used: siDCL-1 (lanes 1), siDCL-2 (lanes 2) and siDCL-3 (lanes 3). SiDCL-2 and -3 induced an effective knockdown (80%and 90%, respectively) while siDCL-1 failed to do so, and was subsequently used a control for the siRNA procedure. As a reference, the same membrane wasre-stained with a-tubulin. (F–N) Confocal analysis of mitotic neuroblastoma cells treated with DCL siRNA. (F–H) Transfection with siDCL-1 (F) failed to induceDCL knockdown and (G) left the mitotic spindle intact. Effective DCL knockdown by (I–K) siDCL-2 or (L–N) siDCL-3 not only (I and L) caused allimmunoreactive DCL signal to disappear, but also (J and M) induced collapse and deformation of the mitotic spindles. Scale bars, 10 lm.

DCL directs neuronal precursor division 643

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

selected for DCL knockdown by transfection of neuroblastoma cellsand monitoring the endogenous DCL protein levels by Western blotanalysis. As negative controls, we used mismatch short hairpin siRNAmolecules and empty vector. The most effective plasmid, pSuper-DCL183 (> 80% knockdown, data not shown; see also Fig. 7) and itsmismatch control (pSuper-DCL183m, no knockdown compared withempty plasmid), were used for IUE. To visualize transfected cells,plasmids were cotransfected with plasmids encoding YFP. We appliedIUE at E14.5: for the control plasmid group, 92% of the embryossurvived and 70% of these survivors expressed YFP; for the DCLi-electroporated group, 84% survived and 73% of these expressed YFP.Embryos transfected with empty vector or mismatch controls at

E14.5 and collected 24 h later revealed normal ongoing corticaldevelopment with large numbers of YFP-expressing, almond-shapedelongated progenitors, typical of radial migrating neurons, present inthe proliferative zone [subventricular zone (SVZ)–VZ; asterisks inFig. 9A and D]. Radial fibres (Fig. 9H) spanning all embryoniccortical zones with tight endfeet at the pial surface were present(arrows in Fig. 9A and G).One day after transfection with pSuper-DCL183, cell numbers in

the PZ were strongly reduced and, only occasionally, YFP+ cellsaccumulated in the upper region of the SVZ–VZ (Fig. 9B and I).Structural organization of the dividing precursors was aberrant andmany of the remaining cells had an enlarged multipolar morphologyand lacked, or only had a very short, leading and ⁄ or trailing process(Fig. 9E and inset in E). Quantification revealed a significant

reduction in the number of YFP+ cells between the control and DCLsiRNA group (Fig. 9F) in the PZ (control, 3752.3 ± 400.9; pSuper-DCL183, 2180.7 ± 115.64; P < 0.05). Cortical thickness of the PZ,IZ and CP did not differ between the groups (data not shown).When expressed as a percentage of all cells in PZ and IZ, nodifferences were observed between numbers of cells (data notshown). This lack of difference indicates no changes in migration orpositional mismatch, implying that all cells were affected in asimilar manner and their position had not changed as a result of theknockdown. Thus, the decrease in number of cells in the PZ mayreflect a direct effect of the DCL knockdown on the progenitors atthe PZ–IZ boundary. Furthermore, a profound reduction in theextent and number of individual radial processes of DCLi-transfected cells was apparent, particularly in the IZ and CP wheremany shortened and twisted radial processes (Fig. 9B and I–L) wereobserved. Also, very few endfeet had reached the pial surface ascompared to empty vector or mismatch controls (Fig. 9B and I).Individual processes in the IZ were often distorted, twisted andsometimes obliquely orientated (Fig. 9J–L). The straight longprocesses observed in both empty vector and mismatch controls(Fig. 9H) were never seen in embryos electroporated with pSuper-DCL183, indicating a crucial role for DCL in the stabilization ofmicrotubules as part of the cytoskeletal organization inherent tonormal radial glia fibre formation.To address possible compensatory changes in DCX expression

after DCL knockdown, additional immunostainings for DCX

Fig. 8. DCL overexpression in dividing COS-1 cells. (A–C) A normally dividing COS-1 cell stained with a-tubulin is shown as reference (‘ref’ in inset). (A)Overexpression of DCL leads to (B) elongation, often unilateral, of the mitotic spindle microtubules. The mitotic spindle length, indicated by arrows, of transfectedcells was enhanced compared to nontransfected cells (ref; reference length). DNA was stained with DAPI. (D–I) DCL overexpression in COS-1 cells during celldivision revealed two phenotypes: one (D and F) very similar to wild-type COS-1 cells, in which DCL (D) largely colocalized with a-tubulin (E). Similarly to thedividing N1E-115 cells (see Fig. 7), DCL was also associated with kinetochore microtubuli and overlapped with mitotic spindles (a-tubulin). (G–I) The otherpredominant phenotype revealed a profound elongation and monopolar orientation of the mitotic spindles. DCL (A, D and G); a-tubulin (B, E and H); and mergedimages (C, F and I). Scale bar, 10 lm.

644 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

(Chemicon antibody) failed to reveal any difference in the DCXdistribution pattern between electroporated (Fig. 9C) and control(not shown) tissue, nor was any DCX expression induced in thesiRNA targeted VZ zone (Fig. 9C), indicating that expression of

DCX was not affected by DCL knockdown. Together, these datademonstrate a critical role for DCL not only in mitotic spindleformation and proliferation but also in the stability of the radial fibrenetwork.

Fig. 9. In utero electroporation of DCLi reduced cell number in the SVZ–VZ and severely disturbed radial fibre organization. (A) Mismatch control plasmidspCMV-YFP delivered into E14.5 embryonic brain labelled many neural progenitors in SVZ–VZ and an extensive network of radial processes spanning the entirethickness of the cortex with a dense rim of pial endfeet (arrow, detail in G). Control plasmids yielded identical results. Asterisk indicates the VZ. Note theresemblance to the radial DCL+ network depicted in Figs 5F, I and J and 6K. (B) Knockdown of DCL after pSuper-DCL183 delivery induced a profoundreduction in SVZ–VZ precursor number and an almost complete ablation of the radial processes, with small cells in the IZ and only a few pial endfeet remaining.(C) The pattern of DCX (red) immunoreactivity was unaffected after pSuper-DCL183 delivery (GFP-positive cells). Note the same distribution in the left orright part of the picture. Also, no DCX immunoreactivity was induced in the targeted VZ region. (D) After control plasmid delivery, the SVZ–VZ showednumerous dividing (arrows) and migrating almond-shaped elongated precursors (inset), with straight radial fibres (arrowhead and H). (E) After DCL knockdown,a strong reduction and aberrant organization of the remaining cells (arrows, inset) was seen; they have shortened and nonradially orientated fibre elements(arrowheads). (F) Numbers of electroporated cells in the proliferative zone (PZ; VZ and SVZ together) were significantly reduced. In the CP, no differences werefound (not shown). (G) Details of radial processes (arrow) in the CP and their endfeet (arrowhead) at the pial surface after control plasmid delivery. Knockdownof DCL caused very few, if any, endfeet to reach the pia (see arrow in B). (H) Detail of straight elongated radial fibres in the IZ after mismatch control vectordelivery. (I) DCL knockdown reduced progenitor number, with the remaining ones displaying a disorganized and often multipolar cellular shape. See alsoE. (J) Detail of the IZ after DCL knockdown (left is SVZ–VZ, right is CP), showing many aberrantly orientated, shortened, curly and twisted processes andsmall cellular elements. (K and L) Details of these aberrant and curly fibres (arrows). Scale bars, 20 lm (A–C and I), 15 lm (D and E), 10 lm (G, H, J and K),5 lm (L).

DCL directs neuronal precursor division 645

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

Discussion

We have shown that the DCLK splice variant DCL is a predominantMAP during early corticogenesis. DCL regulates mitotic spindlestability, precursor proliferation and integrity of the radial fibrenetwork, which will form the early radial scaffold, enabling neuronalmigration during later developmental stages.Our results provide an important addition to the current knowledge

on DCLK, as they specifically address the role of DCL during theearly stages of cortical development. Microtubule stabilization of theearly pial-orientated processes allows somal or nuclear translocation, aprocess by which neuroblasts translocate their somata along radiallyorientated processes. The processes are typically restricted to E10–13,the period when DCL is abundantly expressed. During this period,DCX is either absent or expressed in a pattern that does not overlapwith DCL, suggesting an important and exclusive role of the DCLKgene during early corticogenesis.Aside from the expression in radial fibres, DCL is selectively found

in RGs and neural progenitors. Recent data has shown that RGs are thesource of most, if not all, neurons generated during corticaldevelopment, with neuronal progeny being derived from clonallyrelated RGs (Malatesta et al., 2000, 2003; Alvarez-Buylla et al., 2001;Noctor et al., 2001; Gotz et al., 2002; Rakic, 2003; Anthony et al.,2004). The selective expression of DCL in vimentin-positive neuro-genic RGs in mitotic cells and in neuroblastoma cell lines is consistentwith the concept that DCL is a RG- and neuronal precursor-specificMAP. Indeed, microtubules but not microfilaments have already beenshown to be critically involved in the polarized morphology of RGs(Li et al., 2003). This common origin, or centralization of neurogen-esis and radial glia, raises the interesting possibility that one cell typeearly in development is critical for division, initial neuronal productionand radial process extension and hence for the proper development ofcortical architecture.Our results indicate an important role for DCL in control of the

length and stability of mitotic spindles of dividing NPCs in earlycortical development. Numerous MAPs have been shown to beassociated and to affect the form of mitotic spindles (for review seeGadde & Heald, 2004). Many MAPs, like, e.g., the kinesins, areubiquitously expressed and share a common evolutionarily conservedrole in cell division (for review see Hirokawa & Takemura, 2005).Interestingly, the MAP LIS1 has been shown to affect spindleformation and to be implicated in the developmental regulation ofNPCs (Faulkner et al., 2000). Similarly, mutations in DCX lead tomalformation of the developing cortex and mutated DCX causesabnormal spindle orientation while impairing mitotic progressionin vitro (Couillard-Despres et al., 2004). In contrast to LIS1, our datashow a differential spatiotemporal expression pattern for DCL andDCX particularly in the early developmental period, suggesting aunique role for DCL in NPC proliferation.We have shown DCL to regulate mitotic spindle stability in a

neuroblastoma cell line. Recent findings by us and others (Shu et al.,2006) on the role of DCLK in mitotic spindle formation furthersubstantiate a crucial role for spindle stabilization in neuronalproliferation and differentiation. The stability and length of themitotic spindle had already been shown to affect the fate of daughtercells in D. melanogaster; the generation of daughter cells withdifferent sizes and phenotypes depends on the asymmetric formationof the spindles during anaphase (Kaltschmidt et al., 2000). Similarly,in C. elegans, the orthologue of the mammalian DCLK gene zyg-8 isinvolved in asymmetric division of the one-cell stage in embryos.Also, zyg-8 is associated with mitotic spindles and promotesmicrotubule assembly during anaphase (Gonczy et al., 2001).

Mutations in zyg-8 disrupt division of the one-cell stage of C. elegansembryos, suggesting evolutionarily conserved functions for themammalian DCLK and zyg-8 with regards to spindle formation.Similar to a recent study by Shu et al. (2006), we report that DCLK

proteins are expressed in neurogenic embryonic brain areas. Our effectof siRNA-mediated knockdown on spindle disruption in N1E-115 cellsresembles their effects found after DCLK knockdown in HEK293 cells.Furthermore, Shu et al. (2006) have shown cell cycle exit in vitro andpromotion of neuronal identity of neuronal progenitors in vivo afterDCLK knockdown, which is in accordance with the strong reduction inRGs after DCL knockdown in utero in our study. In addition to thesimilarities, we report several aspects that are novel and unique in ourstudy: (i) we present the functional characterization of DCL, a specificsplice variant of the DCLK gene; (ii) we show that, in contrast to DCXand DCLK-long, which are expressed from E12 and 13 onwards, DCLis expressed from E8 onwards; (iii) DCL is expressed in differentregions and cell types than DCX between E9 and E12; and (iv) weprovide evidence strongly suggesting that DCLK is involved in RGproliferation and radial fibre stability.Neurogenic radial glia can generate neurons directly through

asymmetric cell division or indirectly through generation of interme-diate progenitors or transit amplifying cells that migrate to the SVZ–IZ, where they divide symmetrically to produce two daughter neurons.Hence, the vimentin+, DCX+ and DCL+ triple-stained cells in theSVZ–IZ (see Fig. 6) may represent such secondary neurogenicdivisions of migratory intermediate progenitors (Haydar et al., 2003;Haubensak et al., 2004; Kriegstein & Noctor, 2004; Noctor et al.,2004) and illustrate possible interactions between DCL and DCX inthese layers (Deuel et al., 2006; Koizumi et al., 2006). In the innerVZ, DCL but no DCX expression occurs in mitotic cells. Afterinspection, all types of mitotic cells, irrespective of whether they hadvertical, oblique or horizontal cleavage plane orientations, were foundto be DCL+. Given the possible prediction of the mode of divisionbased on cleavage plane orientation (Kriegstein & Noctor, 2004), thereappears to be at least no strong preference of DCL for a specificorientation or mode of division.The nearly complete depletion of the radial fibre network after

in utero knockdown demonstrates a critical role of DCLK in theregulation of radial fibre stability. The siRNA construct used in ourstudy targets the DCX domain of DCLK. As this domain is present inboth DCLK-long and DCL, it is difficult to attribute the RG phenotypeobserved in our study solely to a specific transcript encoded by theDCLK gene. However, DCLK-long has been reported to be expressedfrom E13 onwards (Lin et al., 2000; Shu et al., 2006) whereas theonset of DCL expression is at E9. Also, no changes in radial fibreintegrity were reported in the in utero knockdown study of Shu et al.(2006) targeting DCLK. A possible explanation may lie in the siRNAconstruct used in that study that targets specifically the full-lengthDCLK-long transcript but leaves the DCL transcript and protein intact(see supplementary data of Shu et al., 2006). Thus, the difference inphenotypes between our present study and that of Shu et al. (2006)may pinpoint DCL as a selective protein necessary for radial fibreformation (Koizumi et al., 2006; Shu et al., 2006). Other possibleexplanations include the precise time-point of analysis (E17 vs. E14 inour study) as well as the methodology used (in utero lentiviral deliveryvs. in utero electroporation of a large cortical swatch in our study).Given the early prominent role for DCLK, one would expect a

severe or even lethal phenotype in DCLK mutants. However, loss ofthe full-length DCLK, the DCLK DCX-like isoform or the kinasedomain containing isoforms of DCLK, results in a preserved neocor-tical lamination (Deuel et al., 2006; Koizumi et al., 2006). Similardiscrepancies between acute siRNA-mediated knockdown and germ

646 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

line knockout have been described for the DCX gene. These resultedin a severe and very mild phenotype, respectively (Corbo et al., 2002;Bai et al., 2003), and suggest that other genes could have compensatedfor the loss of DCX function in DCX knockout mice. Indeed, onlydouble DCLK–DCX-null mice display disrupted cortical lamination(Koizumi et al., 2006; Shu et al., 2006).

Recent evidence (Ramos et al., 2006) further indicates that thesiRNA effect for DCX is rat-specific, as knockdown of DCX in micefailed to induce subcortical band heterotopia. This bears considerablerelevance to the relatively mild phenotypes in DCLK mouse mutants(Deuel et al., 2006; Koizumi et al., 2006), which could relate togenetic redundancy with the recently-described DCK2 gene (Edelmanet al., 2005). The levels of DCK2 were indeed not altered in DCLKhomozygous null mouse brains (Koizumi et al., 2006). Despite theexpected close functional overlap between DCL, DCX, DCK2 andDCLK variants, little compensation occurred in view of the strongphenotype presently observed after in utero knockdown of DCL. Thelack of a compensational response in DCX following DCL knock-down (Fig. 9C) further underscores the important role of DCL in radialstability during embryonic development.

Genes associated with the centrosome and spindle formation arekey elements in mitosis regulation. Mutations in the genes formicrocephalin and abnormal spindle in microcephaly (ASPM) areassociated with microcephaly, a human condition characterized by asmall cortex (Woods, 2004). As DCLK regulates mitotic spindleformation in early neuronal progenitors, mutations in this gene arelikely to affect the precursor pool, as confirmed by the present in uteroknockdown of DCL. DCL thus seems of potential interest for studiesaimed at understanding genetic disorders related to altered brain size.

In conclusion, during cortical development, two related proteins,DCX and DCL, are expressed and both promote microtubule stability.In contrast to recent studies on the DCLK gene, DCX and DCL differin their early spatiotemporal expression pattern and function.Although their functions at later stages of corticogenesis appearsynergistic and largely overlapping, DCX is mainly involved in thestabilization of microtubules during neuronal migration after E12.DCL expression, in contrast, starts earlier, i.e. !E8, and appears to beinvolved in stabilization of radial processes and in mitosis ofneurogenic precursors through spindle stabilization. This earlydevelopmental expression, in contrast to other DCLK splice variants,is of particular general relevance given the extent of the progenitorpool expansion and neuronal production during this period. Interfer-ence with these processes may ultimately affect cortical number andbrain size. We propose that DCL is crucial for precursor proliferationand stability of the early radial glial scaffold and expect that it mayprove to be of great interest for our understanding of earlyneurogenesis and of disorders associated with alterations in initialneuronal production, such as microcephaly.

Supplementary material

The following supplementary material may be found on www.blackwell-synergy.comFig. S1. DCL is a MAP and stabilizes microtubules.Fig. S2. DCL knock-down does not affect centrosome structure.

AcknowledgementsWe thank Drs Nicole Datson, Lee Fradkin, Ron de Kloet, Jasprien Noordermeerand Onno Meijer for helpful comments and discussions. We thank ProfessorDr A.F.M. Moorman and Ms. C. De Gier–De Vries (Academic MedicalCentre, Amsterdam) for their assistance with embryonic tissue processing,

Dr I. Huitinga (Netherlands Institute for Brain Research, Amsterdam) forprovision of the biotinylated tyramide, and Trudy van Kempen, Ans Tijssen(LACDR), Suharti Maslam, Hendrik Kommerie, Bart Engels and Karin dePunder (IN, SILS, UvA) for excellent technical assistance. Dr A. Campbell(SILS-CNS UvA) is acknowledged for his corrections to the English language.P.J.L. is supported by the Dutch Brain Foundation and the VolkswagenStiftung. P.J.L. and K.B. are supported by the Internationale StichtingAlzheimer Onderzoek (ISAO). S.M.K. is supported by grants from theNetherlands Organization for Scientific Research and the Dutch BrainFoundation (grant 12F04(2).67 to M.P. Smidt).

AbbreviationsCARP, CaMK-related peptide; CP, cortical plate; DCL, doublecortin-like;DCLK, doublecortin-like kinase; DCX, doublecortin; E, embryonic day; IUE,in utero electroporation; IZ, intermediate zone; MAP, microtubule-associatedprotein; NPC, neural precursor cells; P, postnatal day; PZ, proliferative zone;RG, radial glial cell; siRNA, small interference RNA; SVZ, subventricularzone; VZ, ventricular zone; YFP, yellow fluorescent protein.

ReferencesAlvarez-Buylla, A., Garcia-Verdugo, J.M. & Tramontin, A.D. (2001) A unified

hypothesis on the lineage of neural stem cells. Nat. Rev. Neurosci., 2, 287–293.

Anthony, T.E., Klein, C., Fishell, G. & Heintz, N. (2004) Radial glia serve asneuronal progenitors in all regions of the central nervous system. Neuron, 41,881–890.

Bai, J., Ramos, R.L., Ackman, J.B., Thomas, A.M., Lee, R.V. & LoTurco, J.J.(2003) RNAi reveals doublecortin is required for radial migration in ratneocortex. Nat. Neurosci., 6, 1277–1283.

Brummelkamp, T.R., Bernards, R. & Agami, R. (2002) A system for stableexpression of short interfering RNAs in mammalian cells. Science, 296, 550–553.

Burgess, H.A. & Reiner, O. (2000) Doublecortin-like kinase is associatedwith microtubules in neuronal growth cones. Mol. Cell Neurosci., 16, 529–541.

Burgess, H.A. & Reiner, O. (2002) Alternative splice variants of doublecortin-like kinase are differentially expressed and have different kinase activities.J. Biol. Chem., 277, 17696–17705.

Corbo, J.C., Deuel, T.A., Long, J.M., LaPorte, P., Tsai, E., Wynshaw-Boris, A.& Walsh, C.A. (2002) Doublecortin is required in mice for lamination of thehippocampus but not the neocortex. J. Neurosci., 22, 7548–7557.

Couillard-Despres, S., Uvanik, G., Ploetz, S., Karl, C., Koch, H., Winkler, J. &Aigner, L. (2004) Mitotic impairment by doublecortin is diminished bydoublecortin mutations found in patients. Neurogenetics, 5, 83–93.

Couillard-Despres, S., Winkler, J., Uvanik, G. & Aigner, L. (2001) Molecularmechanisms of neuronal migration disorders, quo vadis. Curr. Mol. Med., 1,677–688.

Deuel, T.A., Liu, J.S., Corbo, J.C., Yoo, S.Y., Rorke-Adams, L.B. &Walsh, C.A. (2006) Genetic interactions between doublecortin anddoublecortin-like kinase in neuronal migration and axon outgrowth. Neuron,49, 41–53.

Edelman, A.M., Kim, W.Y., Higgins, D., Goldstein, E.G., Oberdoerster, M. &Sigurdson, W. (2005) Doublecortin kinase-2, a novel doublecortin-relatedprotein kinase associated with terminal segments of axons and dendrites.J. Biol. Chem., 280, 8531–8543.

Engels, B.M., Schouten, T.G., van Dullemen, J., Gosens, I. & Vreugdenhil, E.(2004) Functional differences between two DCLK splice variants. Brain Res.Mol. Brain Res., 120, 103–114.

Faulkner, N.E., Dujardin, D.L., Tai, C.Y., Vaughan, K.T., O’Connell, C.B.,Wang, Y. & Vallee, R.B. (2000) A Role for the lissencephaly gene LIS1 inmitosis and cytoplasmic dynein function. Nat. Cell Biol., 2, 784–791.

Francis, F., Koulakoff, A., Boucher, D., Chafey, P., Schaar, B., Vinet, M.C.,Friocourt, G., McDonnell, N., Reiner, O., Kahn, A., McConnell, S.K.,Berwald-Netter, Y., Denoulet, P. & Chelly, J. (1999) Doublecortin is adevelopmentally regulated, microtubule-associated protein expressed inmigrating and differentiating neurons. Neuron, 23, 247–256.

Friocourt, G., Koulakoff, A., Chafey, P., Boucher, D., Fauchereau, F., Chelly, J.& Francis, F. (2003) Doublecortin functions at the extremities of growingneuronal processes. Cereb. Cortex, 13, 620–626.

Gadde, S. & Heald, R. (2004) Mechanisms and molecules of the mitoticspindle. Curr. Biol., 14, R797–R805.

DCL directs neuronal precursor division 647

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648

Gdalyahu, A., Ghosh, I., Levy, T., Sapir, T., Sapoznik, S., Fishler, Y., Azoulai, D.& Reiner, O. (2004) DCX, a new mediator of the JNK pathway. EMBO J., 23,823–832.

Gleeson, J.G., Allen, K.M., Fox, J.W., Lamperti, E.D., Berkovic, S., Scheffer, I.,Cooper, E.C., Dobyns, W.B., Minnerath, S.R., Ross, M.E. & Walsh, C.A.(1998) Doublecortin, a brain-specific gene mutated in human X-linkedlissencephaly and double cortex syndrome, encodes a putative signalingprotein. Cell, 92, 63–72.

Gleeson, J.G., Lin, P.T., Flanagan, L.A. & Walsh, C.A. (1999) Doublecortin is amicrotubule-associated protein and is expressed widely by migratingneurons. Neuron, 23, 257–271.

Gonczy, P., Bellanger, J.M., Kirkham, M., Pozniakowski, A., Baumer, K.,Phillips, J.B. & Hyman, A.A. (2001) zyg-8, a gene required for spindlepositioning in C. elegans, encodes a doublecortin-related kinase thatpromotes microtubule assembly. Dev. Cell, 1, 363–375.

Gotz, M., Hartfuss, E. & Malatesta, P. (2002) Radial glial cells as neuronalprecursors: a new perspective on the correlation of morphology and lineagerestriction in the developing cerebral cortex of mice. Brain Res. Bull., 57,777–788.

Gupta, A., Tsai, L.H. & Wynshaw-Boris, A. (2002) Life is a journey: a geneticlook at neocortical development. Nat. Rev. Genet, 3, 342–355.

Haubensak, W., Attardo, A., Denk, W. & Huttner, W.B. (2004) Neurons arise inthe basal neuroepithelium of the early mammalian telencephalon: a majorsite of neurogenesis. Proc. Natl Acad. Sci. USA, 101, 3196–3201.

Haydar, T.F., Ang, E. Jr & Rakic, P. (2003) Mitotic spindle rotation and modeof cell division in the developing telencephalon. Proc. Natl Acad. Sci. USA,100, 2890–2895.

Hirokawa, N. & Takemura, R. (2005) Molecular motors and mechanisms ofdirectional transport in neurons. Nat. Rev. Neurosci., 6, 201–214.

Kaltschmidt, J.A., Davidson, C.M., Brown, N.H. & Brand, A.H. (2000)Rotation and asymmetry of the mitotic spindle direct asymmetric celldivision in the developing central nervous system. Nat. Cell Biol., 2, 7–12.

Koizumi, H., Tanaka, T. & Gleeson, J.G. (2006) Doublecortin-like kinasefunctions with doublecortin to mediate fiber tract decussation and neuronalmigration. Neuron, 49, 55–66.

Kriegstein, A.R. & Noctor, S.C. (2004) Patterns of neuronal migration in theembryonic cortex. Trends Neurosci., 27, 392–399.

Kruidering, M., Schouten, T., Evan, G.I. & Vreugdenhil, E. (2001) Caspase-mediated cleavage of the Ca2+ ⁄ calmodulin-dependent protein kinase-likekinase facilitates neuronal apoptosis. J. Biol. Chem., 276, 38417–38425.

Li, H., Berlin, Y., Hart, R.P. & Grumet, M. (2003) Microtubules are critical forradial glial morphology: possible regulation by MAPs and MARKs. Glia,44, 37–46.

Lin, P.T., Gleeson, J.G., Corbo, J.C., Flanagan, L. & Walsh, C.A. (2000)DCAMKL1 encodes a protein kinase with homology to doublecortin thatregulates microtubule polymerization. J. Neurosci., 20, 9152–9161.

Malatesta, P., Hack, M.A., Hartfuss, E., Kettenmann, H., Klinkert, W.,Kirchhoff, F. & Gotz, M. (2003) Neuronal or glial progeny: regionaldifferences in radial glia fate. Neuron, 37, 751–764.

Malatesta, P., Hartfuss, E. & Gotz, M. (2000) Isolation of radial glial cells byfluorescent-activated cell sorting reveals a neuronal lineage. Development,127, 5253–5263.

Marin, O. & Rubenstein, J.L. (2003) Cell migration in the forebrain. Annu. Rev.Neurosci., 26, 441–483.

Matsumoto, N., Pilz, D.T. & Ledbetter, D.H. (1999) Genomic structure,chromosomal mapping, and expression pattern of human DCAMKL1(KIAA0369), a homologue of DCX (XLIS). Genomics, 56, 179–183.

Mizuguchi, M., Qin, J., Yamada, M., Ikeda, K. & Takashima, S. (1999) Highexpression of doublecortin and KIAA0369 protein in fetal brain suggeststheir specific role in neuronal migration. Am. J. Pathol., 155, 1713–1721.

Noctor, S.C., Flint, A.C., Weissman, T.A., Dammerman, R.S. & Kriegstein,A.R. (2001) Neurons derived from radial glial cells establish radial units inneocortex. Nature, 409, 714–720.

Noctor, S.C., Martinez-Cerdeno, V., Ivic, L. & Kriegstein, A.R. (2004) Corticalneurons arise in symmetric and asymmetric division zones and migratethrough specific phases. Nat. Neurosci., 7, 136–144.

Omori, Y., Suzuki, M., Ozaki, K., Harada, Y., Nakamura, Y., Takahashi, E. &Fujiwara, T. (1998) Expression and chromosomal localization of KIAA0369,a putative kinase structurally related to Doublecortin. J. Hum. Genet, 43,169–177.

des Portes, V., Pinard, J.M., Billuart, P., Vinet, M.C., Koulakoff, A., Carrie, A.,Gelot, A., Dupuis, E., Motte, J., Berwald-Netter, Y., Catala, M., Kahn, A.,Beldjord, C. & Chelly, J. (1998) A novel CNS gene required for neuronalmigration and involved in X-linked subcortical laminar heterotopia andlissencephaly syndrome. Cell, 92, 51–61.

Rakic, P. (2003) Developmental and evolutionary adaptations of cortical radialglia. Cereb. Cortex, 13, 541–549.

Ramos, R.L., Bai, J. & Loturco, J.J. (2006) Heterotopia formation in rat but notmouse neocortex after RNA interference knockdown of DCX. Cereb.Cortex, 16, 1323–1331.

Schaar, B.T., Kinoshita, K. & McConnell, S.K. (2004) Doublecortinmicrotubule affinity is regulated by a balance of kinase and phosphataseactivity at the leading edge of migrating neurons. Neuron, 41, 203–213.

Schmechel, D.E. & Rakic, P. (1979) Arrested proliferation of radial glial cellsduring midgestation in rhesus monkey. Nature, 277, 303–305.

Shu, T., Tseng, H.C., Sapir, T., Stern, P., Zhou, Y., Sanada, K., Fischer, A.,Coquelle, F.M., Reiner, O. & Tsai, L.H. (2006) Doublecortin-like kinasecontrols neurogenesis by regulating mitotic spindles and M phaseprogression. Neuron, 49, 25–39.

Tanaka, T., Serneo, F.F., Tseng, H.C., Kulkarni, A.B., Tsai, L.H. & Gleeson,J.G. (2004) Cdk5 phosphorylation of doublecortin ser297 regulates its effecton neuronal migration. Neuron, 41, 215–227.

Taylor, K.R., Holzer, A.K., Bazan, J.F., Walsh, C.A. & Gleeson, J.G. (2000)Patient mutations in doublecortin define a repeated tubulin-binding domain.J. Biol. Chem., 275, 34442–34450.

Vreugdenhil, E., Datson, N., Engels, B., de Jong, J., van Koningsbruggen, S.,Schaaf, M. & de Kloet, E.R. (1999) Kainate-elicited seizures induce mRNAencoding a CaMK-related peptide: a putative modulator of kinase activity inrat hippocampus. J. Neurobiol., 39, 41–50.

Vreugdenhil, E., Engels, B., Middelburg, R., van Koningsbruggen, S., Knol, J.,Veldhuisen, B. & de Kloet, E.R. (2001) Multiple transcripts generated by theDCAMKL gene are expressed in the rat hippocampus. Brain Res. Mol. BrainRes., 94, 67–74.

Woods, C.G. (2004) Humanmicrocephaly.Curr. Opin. Neurobiol., 14, 112–117.

648 E. Vreugdenhil et al.

ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 25, 635–648