Neuron, Vol. 28, 697–711, December, 2000, Copyright 2000 by Cell Press

NUDEL Is a Novel Cdk5 Substratethat Associates with LIS1and Cytoplasmic Dynein

ments into growing axons (Schnapp and Reese, 1989;Dillman et al., 1996; Phillis et al., 1996; Abe et al., 1997;Waterman-Storer et al., 1997; Ahmad et al., 1998; Allenet al., 1999; Murphey et al., 1999). We found that overex-pression of LIS1 in nonneuronal cells increased retro-

Martin Niethammer,*§ Deanna S. Smith,*§

Ramses Ayala,*§ Junmin Peng,* Jane Ko,*Ming-Sum Lee,* Maria Morabito,*and Li-Huei Tsai*†§

*Department of Pathology and†Howard Hughes Medical Institute grade movement of cytoplasmic dynein and promoted

the peripheral accumulation of microtubules, reflectingHarvard Medical School200 Longwood Avenue the acquisition of neuron-like dynein behaviors. LIS1

deficiency produced the opposite phenotype. Our find-Boston, Massachussetts 02115ings led us to propose that LIS1 abundance in neuronsfacilitates specific dynein functions that play a role inneuronal migration and axon growth.Summary

An Aspergillus nidulans homolog of LIS1, NudF, wasidentified in a screen for mutants affecting nuclear distri-Disruption of one allele of the LIS1 gene causes abution during hyphal growth (Xiang et al., 1995). Othersevere developmental brain abnormality, type I lis-Aspergillus nuclear distribution mutants encode sub-sencephaly. In Aspergillus nidulans, the LIS1 homolog,units of cytoplasmic dynein and dynactin, a dynein regu-NUDF, and cytoplasmic dynein are genetically linkedlatory complex (Xiang et al., 1994, 1999; Beckwith et al.,and regulate nuclear movements during hyphal growth.1998). Recently, NUDF was shown to interact with theRecently, we demonstrated that mammalian LIS1 reg-coiled-coil domain of another protein, NUDE (Efimovulates dynein functions. Here we characterize NUDEL,and Morris, 2000), which is closely related to RO11 ina novel LIS1-interacting protein with sequence homol-the fungus Neurospora crassa (Minke et al., 1999). RO11ogy to gene products also implicated in nuclear distri-interacts with dynein genetically, but its role in nuclearbution in fungi. Like LIS1, NUDEL is robustly expresseddistribution has not been worked out in any detail.in brain, enriched at centrosomes and neuronal growth

Cytoplasmic dynein and LIS1 homologs are also im-cones, and interacts with cytoplasmic dynein. Further-portant for nuclear positioning in Drosophila (Robinsonmore, NUDEL is a substrate of Cdk5, a kinase knownet al., 1999; Swan et al., 1999), and both have beento be critical during neuronal migration. Inhibition ofimplicated in regulation of neuroblast proliferation, neu-Cdk5 modifies NUDEL distribution in neurons and af-ritic branching and axon transport (Liu et al., 2000). Thus,fects neuritic morphology. Our findings point to cross-observations from several organisms suggest that LIS1talk between two prominent pathways that regulatefunctions with cytoplasmic dynein in the regulation of aneuronal migration.number of critical cellular events.

In the developing cerebral cortex, proper neuronalIntroductionmigration also depends on the activity of a serine/threo-nine kinase, Cdk5 (Kwon and Tsai, 2000). Mice lackingIn recent years, molecular pathways regulating neuronalCdk5, or a regulatory partner, p35, display cortical lami-migration have begun to be elucidated (Walsh and Goffi-nation defects (Oshima et al., 1996; Chae et al., 1997;net, 2000). One such pathway involves LIS1. In humans,Gilmore et al., 1998; Kwon and Tsai, 1998). Cdk5 hasLIS1 haploinsufficiency causes a severe developmentalbeen shown to regulate actin cytoskeletal dynamics (Ni-brain disorder, type 1 lissencephaly (see Reiner andkolic et al., 1998; Humbert et al., 2000) and neuronalSapir, 1998). LIS1 knockout in mice is embryonic lethal,adhesion (Kwon et al., 2000). We now report a link be-while heterozygotes display neuronal migration defectstween Cdk5 and the LIS1/dynein pathway. In a screen(Hirotsune et al., 1998).for LIS1-interacting proteins, we isolated a novel geneLIS1 is a WD repeat protein that associates with micro-that displays sequence similarity to Aspergillus NUDE.tubules and reduces the frequency of microtubule catas-NUDEL (for NUDE-like) associates with both LIS1 andtrophe in vitro (Reiner et al., 1993; Sapir et al., 1997,cytoplasmic dynein. Intriguingly, NUDEL is an in vivo1999). LIS1 has also been described as a noncatalyticsubstrate of Cdk5, and phosphorylation of NUDEL bysubunit of platelet-activating factor acetylhydrolase 1bCdk5 affects its distribution in neurons. Thus, NUDEL(PAFAH1B) and may regulate activity or localization ofis a point at which cross-talk occurs between Cdk5 andthis enzyme (Manya et al., 1999). Recently, we demon-LIS1 pathways.strated that LIS1 was enriched in neurons compared to

other cell types and showed that LIS1 interacts withthe microtubule motor cytoplasmic dynein (Smith et al., Results2000). Cytoplasmic dynein, in addition to its mitoticroles, is responsible for retrograde axonal transport in Two-Hybrid Screen with Lis1

Three yeast two-hybrid screens of a human third trimes-neurons, as well as the transport of microtubule seg-ter brain cDNA library were performed using differentfragments of the Lis1 gene. Using the N-terminal 101† To whom correspondence should be addressed (e-mail: li-huei_tsai@amino acids yielded no positives, despite screeninghms.harvard.edu).

§ These authors contributed equally to this work. 2.0 3 107 clones. The C-terminal 309 amino acids, en-

Neuron698

Figure 1. The LIS1-Interacting Protein NUDEL Is a Homolog of Aspergillus NUDE

(A) The predicted amino acid sequence of human NUDEL, isolated in a two-hybrid screen for LIS1-interacting proteins. Three upstream, in-frame stop codons are indicated in bold. A coiled-coil domain, as predicted by the Lupas algorithm (Lupas et al., 1991) is underlined. Fiveconsensus phosphorylation sites for cyclin-dependent kinases are boxed.(B) Alignment of human and mouse NUDEL, human NudE, X. laevis mp43, and fungal nuclear distribution proteins NUDE and RO11, both ofwhich extend beyond the region shown.(C) Table comparing percent identity between NUDEL, mp43, NUDE, and RO11.(D) Schematic of LIS1 interactions found by yeast two-hybrid analysis. An amino-terminal coiled-coil domain of LIS1 did not interact with anyof the constructs (2), and none of the LIS1 constructs interacted with the empty vector pPC86. The WD domain of LIS1 bound to PAFAHsubunits (1), while full-length LIS1 also interacted with NUDEL.

NUDEL Links CDK5 and LIS1/Dynein Pathways699

Figure 2. NUDEL Protein Is Specifically Ex-pressed in Brain and Testes

(A) (Upper panels) Immunoblot analysis ofadult mouse tissues. NUDEL and LIS1 aremost abundant in brain and testes. A vinculinantibody was used to compare protein load-ing. H, heart; L, liver; Br, brain; K, kidney, M,skeletal muscle; T, testes. (Lower panels)NUDEL expression in brain is developmen-tally regulated. Embryonic day 11 (E11), em-bryonic day 17 (E17), postnatal day 5 (P5),and adult (Ad).(B) Immunoblotting of adult rat brain fractionsseparated by differential centrifugation andextraction. H, whole brain; P1, nuclear pelletand debris; S2, cytosol plus light membranes;P2, crude synaptosomal fraction; S3, cyto-solic fraction; P3, light membranes (Golgi,ER); LS1, synaptosomal cytosol; LP1, synap-tosomal membrane fraction; LP2, synaptic-vesicle enriched fraction; LS2, LS1 minusLP2. PSDI is a P2 fraction extracted with Tri-ton X-100 once or twice (PSD II) or with TritonX-100 followed by sarkosyl (PSD III). PSD-95is a postsynaptic density marker.(C and D) NUDEL immunoperoxidase stainingof E16 mouse brain sections. NUDEL is en-riched in the intermediate zone of the devel-oping cerebral cortex (D). (C) shows a preab-sorbed antibody control. LV, lateral ventrical;VZ, ventricular zone; IZ, intermediate zone;CP, cortical plate; MZ, marginal zone. Magni-fication: 203.

(E–L) NUDEL staining in adult mouse brain. NUDEL is prominent in the cortex and striatum (I), in neuropil of CA3 and CA1 regions of thehippocampus (J), in thalamic and hypothalamic structures (K), and in the molecular layer of the cerebellum (L). Controls with no primaryantibody show no staining (E–H). bg, basal ganglia; cc, corpus callosum; dg, dentate gyrus; gl, granular cell layer; h, hypothalamus; ml,molecular cell layer; ot, optic tract; t, thalamus. Magnification: 43.

compassing the seven LIS1 WD repeats (Reiner et al., stop codons (bold in Figure 1A), and we designated thenext downstream methionine as the putative translation1993) yielded eight positives out of 2.6 3 107 clones, seven

of which corresponded to human platelet-activating fac- initiation site. The sequence surrounding this ATG is ingood agreement with the Kozak consensus. The opentor acetylhydrolase a1 subunit (PAFAH1B a1) (Adachi

et al., 1995), and one to PAFAH1B a2 (Adachi et al., reading frame (ORF) for NUDEL encodes a 345 aminoacid protein with a predicted molecular weight of 38.41997). Both serine esterases have been previously

shown to interact with LIS1 (Hattori et al., 1994a, 1994b, kDa. A significant part of NUDEL (amino acid 19–201)is occupied by a coiled-coil motif (predicted by Lupas1995). Full-length Lis1 yielded 32 positives from 7 3 106

clones. Of these, 19 were distinct clones, some of which algorithm; Lupas et al., 1991), underlined in Figure 1A.NUDEL is an acidic protein (predicted pI 5 5.16 for thewere isolated multiple times. Eight clones (five distinct)

encoded human PAFAH1B a1, and four (three distinct) whole sequence, 4.9 for just the coiled-coil domain).The sequence is rich in serines (10%), many of whichPAFAH1B a2. Twenty clones (11 distinct) encoded a

novel protein, which we termed NUDEL (for NUDE-like). are potential phosphorylation sites for casein kinaseII, PKC, or cdk5. No other motifs are apparent in thesequence.Primary Structure of NUDEL

All isolated NUDEL clones encoded protein sequences Human NUDEL shows homology to the nuclear distri-bution proteins NUDE of Aspergillus nidulans (Efimovof 304–383 amino acids and contained the same down-

stream, in-frame stop codon. An identical human neu- and Morris, 2000) and RO11 of Neurospora crassa(Minke et al., 1999). A mouse protein isolated by Sasakironal precursor EST sequence (GenBank accession

number AA232258) was found that had three upstream et al. (2000 [this issue of Neuron]) shows more extensive

(E) NUDEL antibodies 210 and 211 recognize 42 kDa bands in Cos7 cells transfected with NUDEL. A comigrating band and several slower-migrating species are detected in rat brain lysates. Endogenous Cos7 NUDEL is detected if more protein is loaded (lower panel).(F) Antigen preabsorbed NUDEL antibody (211) no longer recognizes NUDEL in brain lysates.(G) NUDEL antibody 211, but not GST antibody, coimmunoprecipitates adenovirally expressed LIS1 from cells transfected with NUDEL. ALIS1 antibody coimunoprecipitates NUDEL.(H) NUDEL coimmunoprecipitates with LIS1 in adult brain extracts. Arrow indicates IgG heavy chain. A longer exposure of the NUDEL inputis shown to the left.

Neuron700

Figure 3. Localization of NUDEL and LIS1

(A and B) NUDEL antibody 211 labeled Cos7 cells transfected with myc-tagged NUDEL (A). NUDEL antibody preabsorbed with NUDEL-GSTdid not stain similar transfected cultures (B).(C–E) Endogenous LIS1 (red) and transfected, myc-tagged NUDEL (green) colocalize at a point near the nucleus (arrows) in Cos7 cells.(F and G) NUDEL staining (arrows, [K]) resembles the centrosomal localization of g-tubulin (arrows, [G]).(H) Transfected myc-tagged NUDEL (red) and g-tubulin (green) colocalize at the centrosome (arrow).(I and J) NUDEL (green, [I]) and LIS1 (green, [J]) are enriched at the center of the MTOC (arrows) as visualized by a-tubulin staining (red).

NUDEL Links CDK5 and LIS1/Dynein Pathways701

homology. Feng et al. (2000 [this issue of Neuron]) ob- tightly associated with core components of the PSD(Figure 2B, right column of panels).tained a different mouse homolog from an E10.5 whole

mouse library screen. That protein, mNudE, and its hu- NUDEL staining of E18 mouse brain sections (Figure2D) reveals enrichment in the intermediate zone of theman counterpart hNudE, are approximately 50% identi-

cal to NUDEL, with the highest homology in the coiled- developing cerebral cortex. This zone contains migrat-ing neurons and thalamocortical axons, both of whichcoil regions. mNudE is similar to a Xenopus laevis mitotic

phosphoprotein, mp43 (Stukenberg et al., 1997). Figure appear to be labeled with the NUDEL antibody. HighNUDEL signal is also detected in the marginal zone con-1B shows an alignment of these proteins. The fungal

proteins (truncated in Figure 1B) are longer than the taining primarily Cajal-Retzius neurons. Developing mid-brain structures were also positive for NUDEL (data notvertebrate proteins. The percentage identity between

human NUDEL, mp43, NUDE, and RO11 are shown in shown). No staining was observed when preabsorbedantibody was used (Figure 2C). In adult brain (FiguresFigure 1C.2E–2L), NUDEL staining is prominent in the cortex andstriatum, in neuropil in CA3 and CA1 regions of theNUDEL Is a Novel LIS1-Interacting Protein

To further characterize the LIS1-NUDEL interaction, we hippocampus, in thalamic and hypothalamic structures,and in the molecular layer of the cerebellum.cotransformed yeast with the three different LIS1 bait

constructs described above and either empty libraryvector pPC86, PAFAH1B a1, PAFAH1B a2, or NUDEL

NUDEL and LIS1 Colocalize at Centrosomes(Figure 1D). The N-terminal domain of LIS1 was not able

in the Microtubule-Organizing Center (MTOC)to bind to any of the proteins. The WD domain alone

A myc-tagged NUDEL construct (NUDEL-myc) waswas sufficient for binding the PAFAH subunits. Only full-

transfected into Cos7 cells in order to perform coim-length LIS1 was able to interact with NUDEL. This was

munostaining of NUDEL and LIS1. The myc-tagged pro-also observed by Sasaki et al. (2000).

tein was specifically recognized by NUDEL antibodyTwo antibodies against NUDEL, 210 and 211, recog-

(Figure 3A) but not preabsorbed antibody (Figure 3B).nize a band in transfected Cos7 cells that comigrates

Punctate staining of endogenous LIS1 (Figure 3C) andwith bands in untransfected cells and adult rat brain

NUDEL-myc (Figure 3D) colocalized at a centrosome-lysates (Figure 1E). Brain lysate also contains multiple

like structure near the nucleus (Figure 3E, arrow). In-slower-migrating species. Preabsorption of antibody

deed, NUDEL and g-tubulin staining was very similar in211 with GST-NUDEL eliminated detection of NUDEL

these cells (Figures 3F and 3G). Furthermore, endoge-in brain lysate (Figure 1F). LIS1 and NUDEL could be

nous g-tubulin and transfected myc-tagged NUDEL co-coimmunoprecipitated from Cos7 cells transfected with

localized in Cos7 cells (Figure 3H). Both endogenousNUDEL and infected with a LIS1-expressing adenovirus

LIS1 and NUDEL are prominent at the center of the(Figure 1G) and from adult mouse brain lysates (Figure

interphase microtubule array in Cos7 cells further sup-1H). Together, these experiments provide strong evi-

porting a role for these proteins at the centrosome (Fig-dence for a physical interaction between LIS1 and

ures 3I and 3J). NUDEL is prominent at centrosomes inNUDEL.

cultured hippocampal neurons (Figures 3K and 3L), andstrong immunoreactivity was also observed in growth

NUDEL Is Enriched in Brain and Testescones (Figure 3M) as was reported for LIS1 (Smith et

A Western blot of adult mouse tissue shows that NUDELal., 2000).

protein is abundant in brain and testes (Figure 2A, toppanels) and barely detectable in other tissues. In brain,NUDEL is most abundant at postnatal day 5, when exten- LIS1 Levels Regulate NUDEL Distribution

LIS1 overexpression or reduction affects dynein distri-sive migration, neuritic growth, and remodeling occur(Figure 2A, lower panel). NUDEL is less abundant in E17 bution (Smith et al., 2000). Because NUDEL also associ-

ates with LIS1, we examined the distribution of NUDELand in adult brains and is not detectable at E11.Biochemical analysis of adult mouse brain reveals an in similar experiments. Cos7 cells were infected with

AdGFP, an adenovirus carrying the GFP gene, or AdLis1,overlapping fractionation profile for NUDEL and LIS1(Figure 2B). NUDEL is primarily in insoluble fractions carrying both Lis1 and GFP genes (Figure 4). Overex-

pression of LIS1 caused a redistribution of transfectedcontaining membrane and cytoskeletal elements (P2,P3, LP1, LP2) and is particularly enriched in a synapto- NUDEL (red, Figures 4A–4D) and endogenous NUDEL

(green, Figures 4E and 4F) toward the nucleus. As shownsomal membrane fraction (LP1). Cdk5 and its activatingpartner, p35, are enriched in the same fractions as previously, LIS1 overexpression also promotes the pe-

ripheral redistribution of microtubules (red, Figures 4ENUDEL. Both LIS1 and NUDEL were found in fractionsenriched for the postsynaptic density (PSD; for review, and 4F). In contrast to cells overexpressing LIS1, NUDEL

distribution in embryonic fibroblasts obtained fromsee Kennedy, 1993). Both are highest in PSDII (resistantto Triton X-100) but are also found in PSDIII (resistant Lis11/2 mice was more dispersed than in cells from wild-

type littermates (Figures 4G–4J).to sarkosyl), suggesting that pools of these proteins are

(K and L) In mouse embryonic cortical neurons, NUDEL staining (green) is intense at the centrosome (arrows). Neurons were costained withfor a-tubulin (red).(M) NUDEL (green) is prominent in axonal growth cones, where it appears to encircle microtubule bundles (red).Scale bars: 50 mm (A and B), 20 mm (I and J), 10 mm (C–E, K, and L), 5 mm (H). Nuclei in (C)–(L) are stained with bizbenzamide (blue).

Neuron702

Figure 4. Effect of LIS1 on NUDEL Localization

(A–D) Overexpression of LIS1 in Cos7 cells causes tighter perinuclear accumulation of transfected NUDEL (A–D). Cells previously transfectedwith NUDEL (red) were infected with a control adenovirus (AdGFP, [A and B]), or a LIS1-carrying virus (AdLis1, [C and D]). Infected cells aredetected by GFP fluorescence (A and C). (B) and (D) show NUDEL staining alone for better visualization of the redistribution.(E and F) Endogenous NUDEL (green) is also redistributed in LIS1-overexpressing cells. Note the tight perinuclear localization of NUDEL inthe AdLis1 infected cell. As shown previously (Smith et al., 2000), microtubules (red) are redistributed toward the periphery in LIS1-overexpress-ing cells.(G–J) Partial loss of LIS1 causes dispersal of endogenous NUDEL. Mouse embryonic fibroblasts derived from Lis1 heterozygous mice (Lis11/2,[G] and [I]) display a more diffuse NUDEL staining than wild-type cells (Lis11/1, [H] and [J]).Scale bars: 10 mm in all panels.

NUDEL, Like LIS1, Interacts with Cytoplasmic Dynein NUDEL with dyneinIC was highest in the cell body(arrow, Figure 5D, panel i), while p150glued and NUDELIn GST pulldown experiments from brain extracts, GST-

NUDEL, but not other GST fusion proteins or GST colocalization was particularly strong in emerging axo-nal growth cones (arrow, Figure 5D, panel ii). In Cos7alone, interacted with cytoplasmic dynein intermediate

chains (DynIC; Figure 5A). NUDEL-GST also associated cells, conspicuous colocalization of NUDEL with thesetwo proteins was detected at the centrosome (arrows,with dynein heavy chains (HC), intermediate chains (IC),

and light intermediate chains (LIC) in a purified prep- Figure 5D, panels iii and iv).In NUDEL-overexpressing cells, dynein (red) andaration of cytoplasmic dynein (Paschal et al., 1991)

(Figure 5B). 20% of each pulldown was analyzed; so by NUDEL (green) distributed to comet-shaped structuresnear the cell periphery (Figures 6A–6C). LIS1 overex-comparison with the 10% input lane, at least half the

dynein was pulled down, strongly suggesting a direct pression did not stimulate this redistribution and did notcolocalize with dynein at the cell periphery (Figure 6D).interaction.

Coimmunoprecipitation experiments confirmed the Together, our data provide strong evidence for an inter-action between NUDEL and dynein and suggest thatinteraction between NUDEL and dyneinIC in postnatal

day 5 (P5) rat brain extracts (Figure 5C). NUDEL was NUDEL and LIS1 may differ in the role they play in dy-nein-mediated events.able to coimmunoprecipitate both LIS1 and dyneinIC,

and LIS1 coimmunoprecipitated p150glued, dyneinIC,and NUDEL. Dynein heavy chain (DHC) coimmunopre- NUDEL Is a Phosphoprotein in Brain

and a Substrate of Cdk5cipitated NUDEL, LIS1, dyneinIC, and p150glued. Littlep150glued was brought down by the NUDEL antibody Incubation of brain lysates at 378C for 2 hr resulted in

a decrease in the slower-migrating NUDEL isoforms and(see Discussion).Dynein IC and p150glued showed partial colocaliza- a concomitant increase in the faster-migrating species,

indicating a shift in mobility. Addition of calf intestinaltion with NUDEL in cultured embryonic hippocampalneurons (Figure 5D, panels i and ii). The overlap of phosphatase (CIP) increased the kinetics of the shift, so

NUDEL Links CDK5 and LIS1/Dynein Pathways703

Figure 5. NUDEL Interacts with Cytoplasmic Dynein

(A) GST-NUDEL pulled down LIS1 and dynein intermediate chains(DynIC) from adult rat brain lysate. GST fused to PAFAH1B subunits(a1, a2) pulled down only LIS1, while GST alone pulled down neither Figure 6. Effect of NUDEL Overexpression on DyneinIC LocalizationLIS1 or DynIC. (A–C) Overexpressed NUDEL (A) and endogenous dyneinIC (B) colo-(B) GST-NUDEL, but not GST alone, pulls down components of calize (C) in comet-shaped structures at the cell periphery in Cos7cytoplasmic dynein from a purified calf brain preparation (10% in- cells (arrows). The large box in (C) shows an enlargement of theput). By silver staining (right), it can be seen that dynein heavy chains region in the smaller box. Nuclei are stained with bizbenzamide(HC), intermediate chains (IC) and light intermediate chains (LIC) (blue).bound to GST-NUDEL. The identity of HC and IC were confirmed (D). A cell transfected with LIS1 (green) and costained for DyneinICby immunoblotting as shown in the two lanes on the left. The proteins (red). LIS1 and DynIC show very little colocalization at the cell pe-originally bound to beads were removed with 1.0% SDS sample riphery, as previously reported (Smith et al., 2000).buffer (right). Scale bars: 10 mm (A, B, and C), 20 mm (D).(C) Coimmunoprecipitation from P5 rat brain lysates. Antibodiesagainst NUDEL, LIS1, and dynein heavy chain (DHC) specificallycoimmunoprecipitated dynein intermediate chains (DynIC). Al- that it was complete by 30 min. This suggests that thethough DHC and LIS1 coimmunoprecipitated the dynactin subunit, multiple bands in brain lysates represent phosphory-p150glued, this protein was not present at substantial levels in

lated species.NUDEL immunoprecipitates.As NUDEL contains Cdk5 consensus sites (Figure 1A)(D) Hippocampal neurons were stained with antibodies against

NUDEL and DynIC (i) or NUDEL and p150glued (ii) and analyzed by and weakly associates with Cdk5 in a GST pulldowndeconvolution microscopy. Nuclei are stained with bizbenzamide assay (data not shown), we tested whether NUDEL could(blue). Endogenous NUDEL (green) and DynIC (red) colocalized in be directly phosphorylated by Cdk5. Both histidine-cell bodies (arrow, [i]), whereas NUDEL and p150glued also colocal- tagged (H6) NUDEL and GST-tagged NUDEL were phos-ized in the axonal growth cone (arrow, [ii]). In Cos7 cells, endoge-

phorylated by purified p35/Cdk5 kinase (Figure 7B) asnous NUDEL (green) strongly colocalized with DynIC (red, [iii]) anddetected by the incorporation of [32P]gATP. GST alonep150glued (red, [iv]) at the centrosome (arrows). Scale bars: 10 mm.was not phosphorylated.

Cdk5 activity caused a NUDEL mobility shift in trans-fected Cos7 cells. NUDEL transfected alone was de-

Neuron704

Figure 7. NUDEL Is a Phosphoprotein in Brain and a Substrate of Cdk5

(A) Incubation of brain lysates at 378C produces a gradual downward shift of slower migrating forms of NUDEL over 120 min. Addition of calfintestinal phosphatase CIP increases the kinetics of the reaction so that by 30 min all NUDEL has shifted to the lower band.(B) Histidine-tagged (H6) NUDEL and GST-NUDEL, but not GST, incorporate [32P]gATP when incubated with recombinant p35/Cdk5 in vitro.The asterisk indicates autophosphorylated p35. H1, histone H1; NS, no exogenous substrate added.(C) Cotransfection of NUDEL (N) with Cdk5 and p25 in Cos7 cells results in a shift in mobility of NUDEL when compared to control cells orcells transfected with p25 or cdk5 alone. p25 is also phosphorylated by Cdk5 (middle panel).(D) Phosphorylated NUDEL in Cos7 cells is not identical to that seen in rat brain extract (H) and fractions P2 and LP1 (see Figure 2).(E) P0 mouse brain lysates from p352/2 mice and p352/2/p392/2 mice (no detectable Cdk5 activity; J. K. et al.,unpublished data) show less ofthe slower-migrating form of NUDEL than wild-type controls.(F) NUDEL mutants with individual or combinatory serine/threonine to alanine mutations in the five Cdk5 consensus sites were cotransfectedinto Cos7 cells alone (left) or with p25/Cdk5 (right). The Cdk5-induced NUDEL mobility shift of wild-type protein was altered in several mutants,particularly the triple mutant, 123, which showed no detectable mobility shift.(G and H) Embryonic hippocampal neurons costained for NUDEL (green) and cdk5 (red). Nuclei are stained with bizbenzamide (blue). Anaxonal growth cone (arrow in [G]; enlarged in [H]) is enriched in both proteins. Scale bar: 10 mm.

tected as a single band on Western blots. However, (Figure 7C). When Cdk5 or p25 were transfected individ-ually (no active kinase), cotransfected NUDEL was notwhen cotransfected with Cdk5 and a potent activator

p25 (Patrick et al., 1999), NUDEL ran as multiple bands shifted (Figure 7C). The phosphorylated forms of NUDEL

NUDEL Links CDK5 and LIS1/Dynein Pathways705

in Cos7 cells do not exactly correlate with the triplet we tate staining (Figure 8D). We also overexpressed a domi-nant-negative mutant of Cdk5 fused to green fluorescentsee in brain fractions (Figure 7D), which could reflect

the actions of other kinases. protein (Cdk5-GFP; Figure 8E). NUDEL staining wasmore punctate in dnk5-expressing neurons (Figure 8E,Analysis of NUDEL in P0 brain extract from mice lack-

ing two known activators of Cdk5, p35, and p39, pro- ii and iv) compared to neurons expressing the wild-typekinase (Figure 8E, panels i and iii). A similar differencevides evidence that it is likely to be a physiological sub-

strate of Cdk5. The p35/p39 double knockout brain was observed in neurons transfected with 123-myccompared to those transfected with wild-type NUDEL-lysates show no detectable Cdk5 kinase activity (J. K.

et al., unpublished data). Slower-migrating species of myc (data not shown). The redistribution of NUDEL fol-lowing inhibition of kinase activity may indicate thatNUDEL in both p352/2 brain lysates and p352/2/p392/2

lysates were less abundant (Figure 7E) but still present underphosphorylated NUDEL is being recruited to ormaintained in more discrete foci.in these lysates, suggesting again that NUDEL may be

phosphorylated by other kinases as well as Cdk5.NUDEL contains five Cdk5 consensus phosphoryla- Discussion

tion sites (Figure 1A). To determine which of these sitesis phosphorylated by Cdk5, we generated point muta- Signaling pathways that control neuronal migration aretions (S/T→A) for each of the five sites (mutants 1–5; rapidly being uncovered by analysis of human geneticFigure 7F), as well as a triple mutant of the first three diseases and mouse mutations. Three main pathwayssites (“123”). When these were cotransfected into Cos7 have been described (Walsh and Goffinet, 2000). Thecells along with Cdk5/p25, mutants 1, 2, and 3 had dis- first involves signaling through Reelin, a large ECM-liketinctly different banding patterns than wild-type NUDEL, molecule responsible for the Reeler phenotype in mice.indicating these sites can be phosphorylated by Cdk5. Reelin triggers tyrosine phosphorylation of a cyto-The 123 mutant showed no mobility shift at all. plasmic adaptor protein, Dab1, through binding to a

Additional evidence supporting an enzyme-substrate variety of transmembrane receptors (Rice and Curran,relationship between Cdk5 and NUDEL comes from 1999; Dulabon et al., 2000; Gilmore and Herrup, 2000).coimmunostaining in embryonic cortical neurons. Sub- In another pathway, Cdk5 kinase activity modulates racstantial colocalization of NUDEL and Cdk5 was ob- signaling/actin cytoskeletal dynamics and also influ-served in cell bodies and in axonal growth cones (Fig- ences N-cadherin/b-catenin adhesion complexes (Ho-ures 7G and 7H). The soma of the lower neuron in Figure mayouni and Curran, 2000; Kwon and Tsai, 2000; Kwon7G stains more intensely for NUDEL and had not yet et al., 2000). LIS1 and PAF are components of a thirdproduced an obvious axon. NUDEL is likely to be prefer- pathway. Our analysis of a novel protein, NUDEL, nowentially transported into newly emerging axons. provides evidence for a link between the Cdk5 and LIS1

pathways and implicates both in the regulation of amicrotubule motor, cytoplasmic dynein.Cdk5 Activity Affects NUDEL Distribution

and Axonal MorphologyNUDEL Is Biochemically and GeneticallyTo explore the consequence of Cdk5 phosphorylationLinked to LIS1of NUDEL, we performed experiments designed to blockWe isolated NUDEL, a novel LIS1-interacting protein inthis event in embryonic cortical or hippocampal culturesa yeast two-hybrid screen. Coimmunoprecipitation and(Figure 8). We first compared transfection of wild-type,GST pulldown experiments confirm the interaction inmyc-tagged NUDEL, and the 123 phosphorylation mu-Cos7 cells and brain lysates. NUDEL displays 21.7%tant, also myc-tagged (123-myc), into cultured corticalsequence identity with NUDE, one of many proteins in-neurons. Double-labeled cells stained for myc (red) andvolved in the control of nuclear distribution in A. nidu-NUDEL (green) appear yellow (Figures 8A and 8B). Thelans. NUDE closely resembles the RO11 protein, whichmost apparent phenotype is the appearance of brightlyis implicated in a similar process in N. crassa (Minke etstained swellings along neuritic processes in 123-myc-al., 1999). nudE was originally isolated as a multicopytransfected cultures (Figure 8A, panels i and ii). Treatingsuppressor of a mutation in nudF, a LIS1 homolog. Morethese cultures with the potent Cdk5 inhibitor, roscovitinerecently, NUDE was shown to interact with NUDF (andalso has this effect in the presence of wild-type or mu-with a fragment of human LIS1) through its coiled-coiltant transfected NUDEL (Figure 8A, panels iii and iv).domain (Efimov and Morris, 2000). Double nudE/nudFRepresentative neurites with varicosities are shown atmutants are similar to nudF single mutants, indicatinghigher magnification in Figure 8B. The axonal swellingsthat they function in the same pathway in fungi. Twoalso contained dynein IC, and growth cones at the oftenother groups also found nudE homologs in yeast two-bulbous ends of these processes were more brightlyhybrid screens for LIS1-interacting proteins. Sasaki etstained for both NUDEL and dynein IC than growthal. (2000) describe a protein identical to NUDEL. Fengcones in control cells transfected with wild-type NUDELet al. (2000) found a distinct gene encoding mNudE, a(Figure 8C).protein with strong homology to NUDEL. Thus, at leastTo examine the effect of phosphorylation on endoge-two members of the NUDE family exist in mammals, andnous NUDEL localization, untransfected hippocampalthese may have specific functions.neurons were treated with roscovitine. NUDEL immuno-

fluorescence in both DMSO-treated control cells androscovitine-treated cells showed the normal enrichment NUDEL Expression in Brain

Immunoblotting of adult mouse tissues revealed theat centrosomes (Figure 8D) but was less diffuse in ros-covitine treated-cells with a correlated increase in punc- highest levels of NUDEL in brain and testes. In E18

Neuron706

Figure 8. Cdk5 Affects NUDEL Distribution and Neuritic Morphology in Cultured Neurons

Overexpression of the mutant NUDEL, 123-myc, or treatment with a Cdk5 inhibitor produces neuritic swellings.(A) Embryonic cortical cultures were transfected with wild-type, myc-tagged NUDEL (NUDEL-myc, [i and iii]), or 123-myc (ii and iv). 1 hr priorto fixation, cultures were exposed to 20 mM roscovitine (iii and iv) or DMSO (i and ii). Cells were processed for myc (green) and NUDEL (red)immunofluorescence. Nuclei are stained with bizbenzamide (blue). Note the axonal swellings in (ii), (iii), and (iv).(B) Representative neurites with varicosities are shown at higher magnification.

NUDEL Links CDK5 and LIS1/Dynein Pathways707

cortex, NUDEL is most pronounced in the intermediate cytoplasmic dynein. Interestingly, overexpression ofLIS1 did not stimulate this behavior, so although LIS1zone, a region containing migrating neurons and fibers.

Interestingly, mNudE shows the opposite pattern of ex- and NUDEL interact, they may play different roles atdifferent subcellular locations.pression at the same age, with highest level of expres-

sion in the ventricular zone and cortical plate (see Feng The effect of NUDEL overexpression on dynein local-ization and NUDEL enrichment at axon endings andet al., 2000). In the adult brain, Nudel staining is strongest

in neuropil and weaker in axon tracts, suggesting that centrosomes suggests that it may play a role at microtu-bule ends. This is interesting in light of a recent reportits primary function may be related to events occurring

at synaptic locations, such as loading and unloading of that in Aspergillus, NUDA (dynein heavy chain) fusionproteins were observed moving as comet-like structurescargo. This is supported by our finding that Nudel is

prominent in LP1 (a synaptic membrane fraction) and the at a rate similar to the rate of cytoplasmic microtubulepolymerization toward the hyphal tip, suggesting thatpostsynaptic density isolated from adult brains. mNudE

mRNA decreases between E13 and birth in mouse brains the motor moved with the polymerizing ends of microtu-bules (Xiang et al., 2000). Similar behavior was described(Feng et. al, 2000). Taken together, our findings suggest

Nudel and mNudE may play spatially and temporally for NUDE and NUDF (Efimov and Morris, 2000). Inter-estingly, we have observed transfected, myc-taggeddistinct roles.NUDEL capping comet-shaped patches that stainedpositively for endogenous NUDEL, as though the exoge-NUDEL Is Prominent at Centrosomesnous protein was being added to an existing frameworkWe find that NUDEL is prominent at centrosomes. Sa-(data not shown).saki et al. (2000) report similar findings, and Feng et al.

Other dynein-interacting proteins, EB1 and p150-(2000) demonstrated a direct interaction between nudEglued, have been described as capping microtubule plusand several well-characterized centrosomal proteins.ends (Berrueta et al., 1999; Lee et al., 2000; Mimori-Centrosomes are microtubule-organizing centers, and,Kiyosue et al., 2000; Tirnauer and Bierer, 2000). In yeast,although these structures are often associated with mi-an EB1 homolog, BIM1, is thought to capture microtu-tosis, there is a proposed role for centrosomes in nucleo-bules at the cortical cytoskeleton and is critical for spin-kinesis. Nucleokinesis is the extension of an anteriordle polarity (Bloom, 2000). NUDEL may be involved inprocess into which the nucleus translocates (Morris,regulating a similar process.2000). This event may be mediated by dynein motors

anchored at peripheral locations that reel in the nucleusby exerting force on microtubules attached to the NUDEL Is a Phosphoproteincentrosome (Karki and Holzbaur, 1999). Dynein is also In brain extracts, NUDEL appears as three or four bands,present at centrosomes, and, indeed, a dynein light in- and phosphatase treatment indicates that these repre-termediate chain interacts directly with pericentrin. Dis- sent different phosphospecies of NUDEL. In Aspergillus,ruption of dynein/dynactin complexes reportedly affects four distinct species of NUDE can be detected, possiblyMTOC behavior at centrosomes (Gonczy et al., 1999; representing phosphorylated forms of NUDE (EfimovPurohit et al., 1999; Quintyne et al., 1999; Robinson et and Morris, 2000). The fact that the NUDEL homolog inal., 1999). NUDEL and LIS1 are thus poised to regulate Xenopus, mp43, was identified as a mitotic phosphoepi-aspects of this dynein behavior in nucleokinesis. tope-containing protein (Stukenberg et al., 1997) sug-

gests that NUDE and its isoforms are extensively phos-phorylated in evolutionarily diverse organisms. NUDELNUDEL and LIS1 Function in a Dynein Pathway

We found that NUDEL, like LIS1, interacts with cyto- has five consensus Cdk phosphorylation motifs. It isphosphorylated by Cdk5 in vitro and in cotransfectedplasmic dynein motors. NUDEL, however, did not inter-

act robustly with p150glued, a component of a dynein cells. Mutagenesis studies suggest that among the fiveCdk5 sites, the first three are likely to be phosphorylatedregulatory complex. It is possible that NUDEL only rec-

ognizes motor complexes that are not associated with by Cdk5. We believe that NUDEL is an authentic physio-logical substrate of Cdk5, as NUDEL phosphorylation isp150glued, possibly stationary complexes. Further ex-

periments will be required to address this question. reduced in brain lysates exhibiting no detectable Cdk5activity derived from mutant mice. Furthermore, inhibi-NUDEL influences dynein behavior, in that overex-

pression causes a redistribution of endogenous dy- tion of Cdk5 in cortical neurons by expression of a domi-nant negative mutant of Cdk5 resulted in reducedneinIC into discrete comet-shaped structures at the cell

periphery. A few of these were observed in nontrans- NUDEL phosphorylation (data not shown).It is clear that other kinases will play a role in NUDELfected cells, suggesting that the redistribution repre-

sents an increase of a normal cellular process involving regulation. First, two isoforms of NUDEL are present in

(C) Growth cones of neurons transfected with NUDEL-myc or 123-myc were double-labeled with NUDEL and dyneinIC antibodies. Note thedynein-rich swellings in both the axon shaft and growth cone of the neuron overexpressing 123-myc compared to NUDEL-myc.(D) Cortical neurons were treated with DMSO (left panel) or 10 mM roscovitine (right panel) and immunostained for endogenous NUDEL. Anarea near the centrosome is enlarged (inset) to emphasize the brighter, more punctate NUDEL staining in the roscovitine-treated cells.(E) A dominant-negative Cdk5 GFP fusion protein (dnk5-GFP) causes a punctate distribution of NUDEL similar to that observed in roscovitine-treated cells. Dnk5-GFP or Cdk5-GFP were overexpressed in neurons using a recombinant herpes virus. (i) and (ii) show both NUDELimmunofluorescence (red) and GFP (green). In (iii) and (iv), only NUDEL staining is shown. Dnk5-GFP.Scale bars: 50 mm (A), 5 mm (B and C), 10 mm (D and E).

Neuron708

brain extracts from mice in which no kinase activity can of vesicles and membranous organelles. Furthermore,there is evidence in yeast that dynein itself can be teth-be detected. In addition, a different band shift of NUDEL

is present in testes and brain, suggesting differential ered to the cortical cytoskeleton. Finally, a dynein lightchain has been found to be associated with Myosin-V, anphosphorylation in these tissues. Finally, we have pre-

liminary evidence to suggest that Cdk5 phosphorylation actin motor protein, and to proteins at the postsynapticdensity (see Naisbitt et al., 2000, and references therein).of NUDEL may regulate its phosphorylation by other

kinases (data not shown). Thus, Cdk5-dynein cross-talk has the potential to regu-late transport between microtubule and actin networks.

Cdk5 has also been linked to the nonreceptor tyrosineCdk5 Regulates NUDEL Subcellular Distributionkinase c-abl pathway via a novel adaptor molecule, Ca-and Neuritic Morphologybles (Zukerberg et al., 2000). This, together with theIn cultured neurons, inhibition of Cdk5 activity of thesimilarity of inversion of cortical layering between reeler/123 mutant stimulates a modest but detectable redistri-scrambler phenotypes and Cdk5-deficient mice, sug-bution of NUDEL into discrete foci. The three Cdk5 phos-gests an intimate relationship between Cdk5 and reelin/phorylation sites on NUDEL reside in the dynein-bindingdisabled pathways. Results presented in this study thusregion (see Sasaki et al., 2000) suggesting that phos-suggest cross-talk occurs between the three knownphorylation could regulate NUDEL binding to the dynein/pathways regulating neuronal migration.dynactin complex. Alternatively, phosphorylation could

regulate dynein function through NUDEL. Indeed, phar-Experimental Proceduresmacological inhibition of Cdk5 activity with roscovitine

or transfection of the 123 mutant into cortical neuronsYeast Two-Hybrid Systemproduced a neuritic swelling phenotype. Similar varicos- Yeast two-hybrid screening was performed using a method pre-

ities have been reported in Drosophila mutants with dis- viously described by Vidal et al. (1996a, 1996b) and a human fetalruption of proteins known to be important for axonal third trimester brain library (Vector pPC86, ProQuest library, Life

Technologies). The baits used consisted of the N-terminal 101 aminotransport (Goldstein and Yang, 2000). Indeed disruptionacids, the C-terminal 309 amino acids (residues 102–410), or theof LIS1 or DHC in individual Drosophila neurons pro-entire mouse LIS1, obtained by PCR using specific primers andduced axonal swellings (Liu et al, 2000). Thus, the effectsubcloned in frame with the Gal4 DNA binding site into vectorof inhibiting NUDEL phosphorylation may be to inhibit pPC97. Sequence alignments were created using DNAstar (DNAstar,

dynein-based transport. Therefore, normal phosphory- Inc., Madison, WI). GenBank accession numbers for the human,lation of NUDEL could either stimulate activation of dy- Xenopus, Aspergillus, and Neurospora sequences are AY004871,

U95097, AF085679, and AF015560, respectively.nein or block NUDEL inhibition of dynein activity. In thelatter case, NUDEL and LIS1 would play antagonistic

Antibodiesroles with respect to dynein activity. Future experimentsTwo rabbit antisera raised against a C-terminal H6 fusion protein ofare necessary to distinguish between these two possi-NUDEL showed similar properties. In most experiments, antibodybilities.211 was used. As a control for specificity toward NUDEL, antibody

Axonal transport is critical for the establishment of 211 was incubated with a 2-fold excess of GST-NUDEL cross-linkedaxodendritic polarity. Interestingly, lissencephalic hu- to glutathione sepharose. A rabbit antibody to LIS1 was produced

as described in Smith et al. (2000). Mouse monoclonal a-tubulin andman brains (Takada et al., 1988) and cortices of p352/2

dyneinIC intermediate chain antibodies and an IgG fraction of rabbitmice (Chae et al., 1997) contain neurons with reversedantiserum against g-tubulin were purchased from Sigma, Inc. Aor aberrant axodendritic polarity. Thus, it is interestingrabbit polyclonal GST antibody and a rabbit polyclonal Cdk5 anti-to speculate on a role for Cdk5/NUDEL/LIS1 in regulat-body (Santa Cruz, Inc.) were used in immunoprecipitations. A mouse

ing this process. monoclonal p150glued antibody was purchased from TransductionLaboratories. A mouse monoclonal Cdk5 antibody, DC17, was pro-duced in this laboratory (Tsai et al., 1994) and used for WesternNUDEL Links Two Pathways Criticalblots.for Neuronal Migration

Motor proteins perform an amazing variety of tasks andExpression Constructs and Western Blottingare implicated in all forms of cellular motility. The speci-Full-length NUDEL was amplified from the yeast two-hybrid clonesficity of kinesin function is thought to lie in the huge and subcloned into vector pcDNA3 (Invitrogen). NUDEL point mu-

diversity of kinesin molecules themselves. Specificity of tants (S/T→A) were generated via PCR using complementary prim-dynein motor function, on the other hand, is thought to ers containing the mutants and subcloned into pcDNA3. All con-

structs were confirmed by sequencing. The expression constructsbe regulated by proteins that interact with the core motorfor Cdk5, p35, and p25 have been previously described (Patrick etcomplex. We recently showed that overexpression ofal., 1999). For immunoblotting, cells were harvested directly intoLIS1 stimulated a change in the size and location ofsample buffer after washing with PBS.dynein complexes (Smith et al., 2000). And here we show

that NUDEL changes dynein localization. If NUDEL con-Fusion Proteins and Pulldowns

fers functional specificity on dynein, then regulation of GST fusion proteins were produced by subcloning full-length humanNUDEL association with the dynein complex becomes NUDEL, and mouse PAFAH subunits a1 and a2 (Albrecht et al.,

1996) were subcloned into vector pGEX-4T-2 (Amersham Phar-a critical point. Cdk5 may play a role in altering NUDELmacia, Uppsala, Sweden). The GST fusion proteins were cross-association with dynein complexes.linked to Glutathione Sepharose 4B resin using 20 mM dimethylpi-The Cdk5 pathway has been implicated in the modula-melimidate. Rat brains were dounce homogenized in 150 mM NaCl,tion of the actin cytoskeleton. Although dynein is a mi-20 mM HEPES (pH 7.4), 1 mM EDTA, 1% Triton X-100. Following

crotubule motor, the dynein regulatory complex, dynac- 1000 3 g centrifugation, the Triton X-100 concentration was in-tin, contains several actin like molecules that are thought creased to 2% for 1 hr, and lysates were centrifuged at 100,000 3

g. For each pulldown, the supernatant was precleared with GSTto regulate association with the cortical cytoskeleton

NUDEL Links CDK5 and LIS1/Dynein Pathways709

protein then incubated with sepharose beads cross-linked with 1 imaging system (Applied Precision, Seattle, WA). Other images werecaptured using 603 or 1003 oil-immersion objectives with a Nikonmg of either GST or the GST fusion proteins for 2 hr at 48C.inverted microscope attached to a SenSys air-cooled CCD camera.The images were colorized and processed using the ImagePro PlusImmunoprecipitationsSystem from Micro Video Instruments (Avon, MA).Immunoprecipitations were performed from extracts of P5 and adult

rat or mouse brains homogenized in one of two buffers: (buffer A)150 mM NaCl, 50 mM Tris (pH 7.4), 5 mM EDTA, 0.1% NP-40, 1 mg/ Cell Culturesml aprotinin, 1 mg/ml pepstatin, 10 mg/ml leupeptin, 1 mM PMSF, Mouse embryo fibroblasts from Lis1 heterozygous mice and wild-25 mM NaF, 10 mM Na3VO4, 1 mM DTT; (buffer B) 100 mM NaCl, 50 type littermates and Cos7 were grown on poly-D-lysine coated cov-mM Tris (pH 7.4), 0.5% NP-40, 1 mg/ml aprotinin, 1 mg/ml pepstatin, erslips in DMEM 110% fetal calf serum. For neuronal cultures, dis-1 mM PMSF, 25 mM NaF, 10 mM Na3VO4. Cos7 cells were lysed in sociated cells from E18 rat brain cortices were grown in DMEMbuffer A (0.5% NP-40) for 1 hr at 48C, followed by centrifugation at (Biowhittaker) supplemented with 1% L-glutamine, 10% fetal calf13,000 3 g for 30 min. Primary antibodies were added to a final serum, and 1% penicillin/streptomycin. Pure hippocampal neuronsconcentration of 4 mg/ml and incubated for 1 hr, 48C. Complexes were grown in MEM supplemented with N2 additives (GIBCO-BRL).were isolated using protein A sepharose or IGM sepharose. Cells were transfected using calcium phosphate 3 days after plating.

16 hr later, cells were fixed in 4% paraformaldehyde for 20 min at378C or ice-cold methanol for 5 min, or both, then permeabilized inBrain Subcellular Fractionations and Postsynaptic0.5% Triton X-100 in phosphate-buffered saline (PBS) for 5 min.Density (PSD) Purification

Rat brain fractions were prepared as previously described (Huttneret al., 1983). Briefly, rat brain was homogenized in 0.32 M sucrose Kinase Assaybuffer (fraction H) and centrifuged at 800 3 g to pellet nuclei and 1 mg of GST, 5 mg GST-tagged NUDEL, and 5 mg of histidine-taggedlarge cellular debris (P1). The supernatant (S1) was subjected to NUDEL were incubated for 20 min at room temperature in 10 ml10,000 3 g centrifugation to yield the crude synaptosomal fraction kinase buffer (KB-30 mM HEPES [pH 7.0], 10 mM MgCl2, 5 mM(P2). The supernatant (S2) was centrifuged for 2 hr, at 165,000 3 g, MnCl2, 1 mM DTT). Similarly, 2 ml of purified p35/CDK5 producedresulting in cytosol (S3) and light membrane and Golgi (P3) fractions. in Bacculovirus was diluted in 30 ml with KB and incubated at roomThe synaptosomal fraction P2 was lysed hypo-osmotically and spun temperature for 20 min. Proteins and kinase were then mixed withfor 20 min at 25,000 3 g to obtain the synaptosomal membrane 0.5 ml of [32P]ATP in a final reaction volume of 50 ml. The reactionfraction LP1. The supernatant, LS1, was centrifuged for 2 hr at was incubated at room temperature for 1 hr, then stopped with 30165,000 3 g, resulting in a synaptic vesicle enriched fraction (LP2) ml of 23 sample buffer. Samples were run on a 9% SDS polyacryl-and a supernatant (LS2). Purified PSD fractions were prepared from amide gel.rat brain lysate and subjected to detergent extraction as previouslydescribed (Carlin et al., 1980; Cho et al., 1992). CIP Assay

100 mg of rat brain lysate in 50 ml 10 mM Tris [pH 8.8], 1 mg/mlPreparation of Cytoplasmic Dynein pepstatin, 10 mg/ml leupeptin, 1 mM PMSF was incubated with 1Cytoplasmic dynein was purified from calf brain essentially as de- ml of CIP (Calf Intestinal Phosphatase) at 378C for 30 min, 1 hr, orscribed in Paschal et al. (1991). Dynein was purified by microtubule 2 hr. Two samples lacking CIP were incubated at 378C for 30 minpelleting in the presence of ATP and amppnp. Kinesins were first or 2 hr. Reactions were stopped with sample buffer.eluted with GTP, and then dynein was eluted separately with ATPand further purified by sucrose fractionation. The proteins shown Acknowledgmentsin Figure 5B are from pooled fractions and have been shown to havemotor activity in a microtubule gliding assay (J. Tirnauer, personal We thank Urs Albrecht for PAFAH1B subunit constructs and Byroncommunication). Taylor for help with yeast-two hybrid screen. We also thank Rachel

Neve for packaging herpes viruses and Michael Gambello and An-Mouse Tissue Preparation thony Wynshaw-Boris for generously providing fibroblasts derivedAdult mouse tissues were homogenized in 150 mM NaCl, 50 mM from Lis11/2 mouse embryos. Ying Zhou assisted with adenovirusTris 7.4, 1 mM EDTA, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, 1 production, and Jennifer Tirnauer and Robert Palazzo provided puri-mg/ml pepstatin, 10 mg/ml leupeptin, 1 mM PMSF, 25 mM NaF, 10 fied dynein. We also thank Janet Volker and Paul Greer for criticalmM Na3VO4. Adult brain and embryonic day 18 (E18) brain sections review of the manuscript and A. Wynshaw-Boris and C. Walsh forwere used for NUDEL immunoperoxidase studies. Embryos were exchanging manuscripts prior to publication. L.-H. T. is an Assistantdrop-fixed in 4% paraformaldehyde, mounted in paraffin, and sec- Investigator of the Howard Hughes Medical Institute, a Rita Allentioned. Antigen retrieval of consecutive sections was performed by Foundation Scholar, and a recipient of an Esther A. and Josephmicrowaving (high power) two times for 4 min in antigen retrieval Klingenstein Fund. This project was partially supported by NIHbuffer citrate buffer (Vector labs). Tissues were treated with 0.5% grants to L.-H. T.TX-100 for 10 min, then processed for NUDEL immunoperoxidasereactivity using a Vectastain Elite kit (Vector). Adult brains were Received August 11, 2000; revised November 9, 2000.obtained following perfusion of mice with 4% paraformaldehyde.20 mM cryosections were processed for immunoperoxidase histo- Referenceschemistry.

Abe, T.K., Tanaka, H., Iwanaga, T., Odani, S., and Kuwano, R. (1997).Production and Use of Recombinant Adenoviruses The presence of the 50-kDa subunit of dynactin complex in theRecombinant adenoviruses were produced using the AdEasy sys- nerve growth cone. Biochem. Biophys. Res. Commun. 233, 295–299.tem (generously provided by the laboratory of Bert Vogelstein at

Adachi, H., Tsujimoto, M., Hattori, M., Arai, H., and Inoue, K. (1995).Johns Hopkins Oncology Center, Baltimore, MD). Infections werecDNA cloning of human cytosolic platelet-activating factor acetylhy-carried out as follows. Cells that had ceased to proliferate bydrolase gamma-subunit and its mRNA expression in human tissue.allowing them to reach confluence for 2 days were trypsinized andBiochem. Biophys. Res. Commun. 214, 180–187.diluted 1/10–1/25 prior to replating on polylysine-coated coverslipsAdachi, T., Aoki, J., Manya, H., Asou, H., Arai, H., and Inoue, K.for immunofluorescence. Cells were exposed 3–6 hr later to adeno-(1997). PAF analogues capable of inhibiting PAF acetylhydrolasevirus at a multiplicity of infection between 25–100 infectious parti-activity suppress migration of isolated rat cerebellar granule cells.cles/cell.Neurosci. Lett. 235, 133–136.

Ahmad, F.J., Echeverri, C.J., Vallee, R.B., and Baas, P.W. (1998).Microscopy and Imaging TechniquesImages were captured using a 1003 oil-immersion objective on a Cytoplasmic dynein and dynactin are required for the transport of

microtubules into the axon. J. Cell Biol. 140, 391–401.Nikon inverted microscope linked to a DeltaVision deconvolution

Neuron710

Albrecht, U., Abu-Issa, R., Ratz, B., Hattori, M., Aoki, J., Arai, H., Graded reduction of Pafah1b1 (Lis1) activity results in neuronal mi-gration defects and early embryonic lethality. Nat. Genet. 19,Inoue, K., and Eichele, G. (1996). Platelet-activating factor acetylhy-

drolase expression and activity suggest a link between neuronal 333–339.migration and platelet-activating factor. Dev. Biol. 180, 579–593. Homayouni, R., and Curran, T. (2000). Cdk5 gets into sticky situa-

tions. Curr. Biol. 10, R331–R334.Allen, M.J., Shan, X., Caruccio, P., Froggett, S.J., Moffat, K.G., andMurphey, R.K. (1999). Targeted expression of truncated glued dis- Humbert, S., Dhavan, R., and Tsai, L. (2000). p39 activates cdk5 inrupts giant fiber synapse formation in drosophila. J. Neurosci. 19, neurons, and is associated with the actin cytoskeleton. J. Cell. Sci.9374–9384. 113, 975–983.Beckwith, S.M., Roghi, C.H., Liu, B., and Ronald Morris, N. (1998). Huttner, W.B., Schiebler, W., Greengard, P., and De Camilli, P. (1983).The “8-kD” cytoplasmic dynein light chain is required for nuclear Synapsin I (protein I), a nerve terminal-specific phosphoprotein. III.migration and for dynein heavy chain localization in Aspergillus nidu- Its association with synaptic vesicles studied in a highly purifiedlans. J. Cell Biol. 143, 1239–1247. synaptic vesicle preparation. J. Cell Biol. 96, 1374–1388.Berrueta, L., Tirnauer, J.S., Schuyler, S.C., Pellman, D., and Bierer, Karki, S., and Holzbaur, E.L. (1999). Cytoplasmic dynein and dynac-B.E. (1999). The APC-associated protein EB1 associates with com- tin in cell division and intracellular transport. Curr. Opin. Cell. Biol.ponents of the dynactin complex and cytoplasmic dynein intermedi- 11, 45–53.ate chain. Curr. Biol. 9, 425–428. Kennedy, M.B. (1993). The postsynaptic density. Curr. Opin. Neuro-Bloom, K. (2000). It’s a kar9ochore to capture microtubules. Nat. biol. 3, 732–737.Cell Biol. 2, E96–E98. Kwon, Y.T., and Tsai, L.H. (1998). A novel disruption of corticalCarlin, R., Grab, D., Cohen, R., and Siekevitz, P. (1980). Isolation development in p35(-/-) mice distinct from reeler. J. Comp. Neurol.and characterization of postsynaptic densities from various brain 395, 510–522.regions: enrichment of different types of postsynaptic densities. J. Kwon, Y.T., and Tsai, L.H. (2000). The role of the p35/cdk5 kinaseCell Biol. 86, 831–845. in cortical development. Results Probl. Cell Differ. 30, 241–253.Chae, T., Kwon, Y.T., Bronson, R., Dikkes, P., Li, E., and Tsai, L.H. Kwon, Y.T., Gupta, A., Zhou, Y., Nikolic, M., and Tsai, L.H. (2000).(1997). Mice lacking p35, a neuronal specific activator of Cdk5, Regulation of N-cadherin-mediated adhesion by the p35-Cdk5 ki-display cortical lamination defects, seizures, and adult lethality. Neu- nase. Curr. Biol. 10, 363–372.ron 18, 29–42.

Lee, L., Tirnauer, J.S., Li, J., Schuyler, S.C., Liu, J.Y., and Pellman,Cho, K.-O., Hunt, C.A., and Kennedy, M.B. (1992). The rat brain D. (2000). Positioning of the mitotic spindle by a cortical-microtubulepostsynaptic density fraction contains a homolog of the Drosophila capture mechanism. Science 287, 2260–2262.discs-large tumor suppressor protein. Neuron 9, 929–942.

Liu, Z., Steward, R., and Luo, L. (2000). Drosophila Lis1 is required forDillman, J.F., 3rd, Dabney, L.P., Karki, S., Paschal, B.M., Holzbaur, neuroblast proliferation, dendritic elaboration and axonal transport.E.L., and Pfister, K.K. (1996). Functional analysis of dynactin and Nat. Cell Biol. 2, 776–783.cytoplasmic dynein in slow axonal transport. J. Neurosci. 16, 6742–

Lupas, A., Van Dyke, M., and Stock, J. (1991). Predicting coiled coils6752.

from protein sequences. Science 252, 1162–1164.Dulabon, L., Olson, E.C., Taglienti, M.G., Eisenhuth, S., McGrath,

Manya, H., Aoki, J., Kato, H., Ishii, J., Hino, S., Arai, H., and Inoue,B., Walsh, C.A., Kreidberg, J.A., and Anton, E.S. (2000). Reelin binds

K. (1999). Biochemical characterization of various catalytic com-a3b1 integrin and inhibits neuronal migration. Neuron 27, 33–44.

plexes of the brain platelet-activating factor acetylhydrolase. J. Biol.Efimov, V.P., and Morris, N.R. (2000). The Lis1-related NUDF protein Chem. 274, 31827–31832.of Aspergillus nidulans interacts with the coiled-coil domain of the Mimori-Kiyosue, Y., Shiina, N., and Tsukita, S. (2000). The dynamicNUDE/RO11 protein. J. Cell Biol. 150, 681–688. behavior of the APC-binding protein EB1 on the distal ends of micro-Feng, Y., Olson, E.C., Stukenberg, P.T., Flanagan, L.A., Kirschner, tubules. Curr. Biol. 10, 865–868.M.W., and Walsh, C.A. (2000). Interactions between LIS1 and Minke, P.F., Lee, I.H., Tinsley, J.H., Bruno, K.S., and Plamann, M.mNudE, a central component of the centrosome, are required for (1999). Neurospora crassa ro-10 and ro-11 genes encode novelCNS lamination. Neuron, this issue, 28, 665–679. proteins required for nuclear distribution. Mol. Microbiol. 32, 1065–Gilmore, E.C., and Herrup, K. (2000). Cortical development: receiving 1076.reelin. Curr. Biol. 10, R162–R166. Morris, N.R. (2000). Nuclear migration. From fungi to the mammalianGilmore, E.C., Ohshima, T., Goffinet, A.M., Kulkarni, A.B., and brain. J. Cell Biol. 148, 1097–1102.Herrup, K. (1998). Cyclin-dependent kinase 5-deficient mice demon- Murphey, R.K., Caruccio, P.C., Getzinger, M., Westgate, P.J., andstrate novel developmental arrest in cerebral cortex. J. Neurosci. Phillis, R.W. (1999). Dynein-dynactin function and sensory axon18, 6370–6377. growth during Drosophila metamorphosis: a role for retrograde mo-Goldstein, L.S., and Yang, Z. (2000). Microtubule-based transport tors. Dev. Biol. 209, 86–97.systems in neurons: the roles of kinesins and dyneins. Annu. Rev. Naisbitt, S., Valtschanoff, J., Allison, D.W., Sala, C., Kim, E., Craig,Neurosci. 23, 39–71. A.M., Weinberg, R.J., and Sheng, M. (2000). Interaction of the post-Gonczy, P., Pichler, S., Kirkham, M., and Hyman, A.A. (1999). Cyto- synaptic density-95/guanylate kinase domain-associated proteinplasmic dynein is required for distinct aspects of MTOC positioning, complex with a light chain of myosin-V and dynein. J. Neurosci. 20,including centrosome separation, in the one cell stage Caenorhab- 4524–4534.ditis elegans embryo. J. Cell Biol. 147, 135–150. Nikolic, M., Chou, M.M., Lu, W., Mayer, B.J., and Tsai, L.H. (1998).Hattori, M., Adachi, H., Tsujimoto, M., Arai, H., and Inoue, K. (1994a). The p35/Cdk5 kinase is a neuron-specific Rac effector that inhibitsThe catalytic subunit of bovine brain platelet-activating factor ace- Pak1 activity. Nature 395, 194–198.tylhydroalse is a novel type of serine esterase. J. Biol. Chem. 269, Oshima, T., Ward, J. M., Huh, C.-G., Longenecker, G., Veeranna,23150–23155. Pant, H.C., Brady, R.O., Martin, L.J., and Kulkarni, A.B. (1996). Tar-Hattori, M., Adachi, H., Tsujimoto, M., Arai, H., and Inoue, K. (1994b). geted disruption of the cyclin-dependent kinase 5 gene results inMiller-Dieker lissencephaly gene encodes a subunit of brain platelet- abnormal corticogenesis, neuronal pathology and perinatal death.activating factor acetylhydrolase. Nature 370, 216–218. Proc. Natl. Acad. Sci. USA 93, 11173–11178.

Hattori, M., Adachi, H., Aoki, J., Tjusimoto, M., Arai, H., and Inoue, Paschal, B.M., Shpetner, H.S., and Vallee, R.B. (1991). PurificationK. (1995). Cloning and expression of a cDNA encoding the b-subunit of brain cytoplasmic dynein and characterization of its in vitro prop-of bovine brain platelet-activating factor acetylhydrolase. J. Biol. erties. Methods Enzymol. 196, 181–191.Chem. 270, 31345–31352. Patrick, G.N., Zukerberg, L., Nikolic, M., de la Monte, S., Dikkes, P.,

and Tsai, L.-H. (1999). Conversion of p35 to p25 deregulates cdk5Hirotsune, S., Fleck, M.W., Gambello, M.J., Bix, G.J., Chen, A., Clark,G.D., Ledbetter, D.H., McBain, C.J., and Wynshaw-Boris, A. (1998). activity and promotes neurodegeneration. Nature 402, 615–622.

NUDEL Links CDK5 and LIS1/Dynein Pathways711

Phillis, R., Statton, D., Caruccio, P., and Murphey, R.K. (1996). Muta- Weiss, D.G., Langford, G.M., and Holzbaur, E.L. (1997). The interac-tion between cytoplasmic dynein and dynactin is required for fasttions in the 8 kDa dynein light chain gene disrupt sensory axon

projections in the Drosophila imaginal CNS. Development 122, 2955– axonal transport. Proc. Natl. Acad. Sci. USA 94, 12180–12185.2963. Xiang, X., Beckwith, S.M., and Morris, N.R. (1994). Cytoplasmic dy-

nein is involved in nuclear migration in Aspergillus nidulans. Proc.Purohit, A., Tynan, S.H., Vallee, R., and Doxsey, S.J. (1999). DirectNatl. Acad. Sci. USA 91, 2100–2104.interaction of pericentrin with cytoplasmic dynein light intermediate

chain contributes to mitotic spindle organization. J. Cell Biol. 147, Xiang, X., Osmani, A.H., Osmani, S.A., Xin, M., and Morris, N.R.481–492. (1995). NudF, a nuclear migration gene in Aspergillus nidulans, is

similar to the human LIS-1 gene required for neuronal migration.Quintyne, N.J., Gill, S.R., Eckley, D.M., Crego, C.L., Compton, D.A.,Mol. Biol. Cell 6, 297–310.and Schroer, T.A. (1999). Dynactin is required for microtubule an-

choring at centrosomes. J. Cell Biol. 147, 321–334. Xiang, X., Zuo, W., Efimov, V.P., and Morris, N.R. (1999). Isolationof a new set of aspergillus nidulans mutants defective in nuclearReiner, O., and Sapir, T. (1998). Abnormal cortical development;migration. Curr. Genet. 35, 626–630.towards elucidation of the LIS1 gene product function. Int. J. Mol.

Med. 1, 849–853. Xiang, X., Han, G., Winkelmann, D.A., Zuo, W., and Morris, N.R.(2000). Dynamics of cytoplasmic dynein in living cells and the effectReiner, O., Carrozzo, R., Shen, Y., Wehnert, M., Faustinella, F., Do-of a mutation in the dynactin complex actin-related protein arp1.byns, W.B., Caskey, C.T., and Ledbetter, D.H. (1993). Isolation of aCurr. Biol. 10, 603–606.Miller-Dieker lissencephaly gene containing G protein b-subunit-

like repeats. Nature 364, 717–721. Zukerberg, L.R., Patrick, G.N., Nikolic, M., Humbert, S., Wu, C.L.,Lanier, L.M., Gertler, F.B., Vidal, M., Van Etten, R.A., and Tsai, L.H.Rice, D.S., and Curran, T. (1999). Mutant mice with scrambled brains:(2000). Cables links Cdk5 and c-Abl and facilitates Cdk5 tyrosineunderstanding the signaling pathways that control cell positioningphosphorylation, kinase upregulation, and neurite outgrowth. Neu-in the CNS. Genes Dev. 13, 2758–2773.ron 26, 633–646.Robinson, J.T., Wojcik, E.J., Sanders, M.A., McGrail, M., and Hays,

T.S. (1999). Cytoplasmic dynein is required for the nuclear attach-GenBank Accession Numbersment and migration of centrosomes during mitosis in Drosophila.

J. Cell Biol. 146, 597–608.The nucleotide sequence for human NUDEL has been submitted to

Sapir, T., Elbaum, M., and Reiner, O. (1997). Reduction of microtu-GenBank under the accession number AY004871.

bule catastrophe events by LIS1, platelet- activating factor acetylhy-drolase subunit. EMBO J. 16, 6977–6984.

Sapir, T., Cahana, A., Seger, R., Nekhai, S., and Reiner, O. (1999).LIS1 is a microtubule-associated phosphoprotein. Eur. J. Biochem.265, 181–188.

Sasaki, S., Shionoya, A., Ishida, M., Gambello, M.J., Yingling, J.,Wynshaw-Boris, A., and Hirotsune, S. (2000). A LIS1/NUDEL/cyto-plasmic dynein heavy chain complex in the developing and adultnervous system. Neuron, this issue, 28, 681–696.

Schnapp, B.J., and Reese, T.S. (1989). Dynein is the motor for retro-grade axonal transport of organelles. Proc. Natl. Acad. Sci. USA 86,1548–1552.

Smith, D.S., Niethammer, M., Ayala, R., Zhou, Y., Gambello, M.J.,Wynshaw-Boris, A., and Tsai, L.-H. (2000). Regulation of cytoplasmicdynein behavior and microtubule organization by mammalian Lis1.Nat. Cell Biol. 2, 767–775.

Stukenberg, P.T., Lustig, K.D., McGarry, T.J., King, R.W., Kuang,J., and Kirschner, M.W. (1997). Systematic identification of mitoticphosphoproteins. Curr. Biol. 7, 338–348.

Swan, A., Nguyen, T., and Suter, B. (1999). Drosophila lissencephaly-1functions with Bic-D and dynein in oocyte determination and nuclearpositioning. Nat. Cell Biol. 1, 444–449.

Takada, K., Becker, L.E., and Chan, F. (1988). Aberrant dendriticdevelopment in the human agyric cortex: a quantitative and qualita-tive Golgi study of two cases. Clin. Neuropathol. 7, 111–119.

Tirnauer, J.S., and Bierer, B.E. (2000). EB1 proteins regulate microtu-bule dynamics, cell polarity, and chromosome stability. J. Cell Biol.149, 761–766.

Tsai, L.H., Delalle, I., Caviness, V.S., Jr., Chae, T., and Harlow, E.(1994). p35 is a neural-specific regulatory subunit of cyclin-depen-dent kinase 5. Nature 371, 419–423.

Vidal, M., Braun, P., Chen, E., Boeke, J.D., and Harlow, E. (1996a).Genetic characterization of a mammalian protein-protein interactiondomain by using a yeast reverse two-hybrid system. Proc. Natl.Acad. Sci. USA 93, 10321–10326.

Vidal, M., Brachmann, R.K., Fattaey, A., Harlow, E., and Boeke,J.D. (1996b). Reverse two-hybrid and one-hybrid systems to detectdissociation of protein-protein and DNA-protein interaction. Proc.Natl. Acad. Sci. USA 93, 10315–10320.

Walsh, C.A., and Goffinet, A.M. (2000). Potential mechanisms ofmutations that affect neuronal migration in man and mouse. Curr.Opin. Genet. Dev. 10, 270–274.

Waterman-Storer, C.M., Karki, S.B., Kuznetsov, S.A., Tabb, J.S.,