zebrafish mitotic kinesin-like protein 1 (mklp1) functions ... · t8–2, t11–3, and t11–8)...

Zebrafish mitotic kinesin-like protein 1 (Mklp1)functions in embryonic cytokinesis

MING-CHYUAN CHEN,1 YI ZHOU,2 AND H. WILLIAM DETRICH III1

1Department of Biology, Northeastern University, and 2Division of Hematology/Oncology,Children’s Hospital and Howard Hughes Medical Institute, Boston, Massachusetts 02115Received 31 May 2001; accepted in final form 21 November 2001

Chen, Ming-Chyuan, Yi Zhou, and H. William De-trich III. Zebrafish mitotic kinesin-like protein 1 (Mklp1)functions in embryonic cytokinesis. Physiol Genomics 8:51–66, 2002. First published November 27, 2001; 10.1152/physiolgenomics.00042.2001.—To understand the functionsof microtubule motors in vertebrate development, we areinvestigating the kinesin-like proteins (KLPs) of the ze-brafish, Danio rerio. Here we describe the structure, intra-cellular distribution, and function of zebrafish mitotic KLP1(Mklp1). The zebrafish mklp1 gene that encodes this 867-amino acid protein maps to a region of zebrafish linkagegroup 18 that is syntenic with part of human chromosome 15.In zebrafish AB9 fibroblasts and in COS-7 cells, the zebrafishMklp1 protein decorates spindle microtubules at metaphase,redistributes to the spindle midzone during anaphase, andbecomes concentrated in the midbody during telophase andcytokinesis. The motor is detected consistently in interphasenuclei of COS cells and occasionally in those of AB9 cells.Nuclear targeting of Mklp1 is conferred by two basic motifslocated in the COOH terminus of the motor. In cleavingzebrafish embryos, green fluorescent protein (GFP)-taggedMklp1 is found in the nucleus in interphase and associateswith microtubules of the spindle midbody in cytokinesis.One- or two-cell embryos injected with synthetic mRNAsencoding dominant-negative variants of GFP-Mklp1 fre-quently fail to complete cytokinesis during cleavage, result-ing in formation of multinucleated blastomeres. Our resultsindicate that the zebrafish Mklp1 motor performs a criticalfunction that is required for completion of embryonic cytoki-nesis.

embryo; vertebrate

METAZOAN DEVELOPMENT GENERALLY begins with cleavage,the period during which the zygote undergoes rapidcell division. In vertebrates, karyokinesis (division ofthe nucleus) alternates with cytokinesis (division of thecytoplasm) to partition the zygote into a multicellularblastula. The requirement for two mechanochemicalsystems, the mitotic spindle in nuclear division and theactin-myosin contractile ring in cytoplasmic partition-ing, to accomplish cleavage has long been recognized.However, the regulatory systems that ensure the tem-

poral and spatial coordination of mitosis and cytokine-sis in cleaving embryos remain poorly understood.

The mitotic spindle itself plays a major role in con-trolling the timing and the positioning of cleavagefurrow formation (34, 55, 56). Both the spindle polesand the spindle midzone have been shown to regulatedevelopment of the furrow (34, 55, 56), probablythrough interactions with the cell cortex. Recent obser-vations suggest that a “cleavage signal” is deliveredfrom the spindle midzone to the cell cortex to initiatefurrow formation (5). Furthermore, reduction in thenumber of midzone microtubules near the cell cortexinhibits or reverses progression of the cleavage furrow(71). Thus the spindle midzone apparently is requiredboth for the initiation and progression of cytokinesis.

Although the importance of the spindle midzone tocytokinesis is clear, the nature of the midzone cleavagesignal and the molecular mechanism of its interactionwith the cell cortex remain to be determined. Recentstudies suggest that motors of the kinesin-like micro-tubule motor superfamily (more generally, kinesin-likeproteins, or KLPs; see Refs. 4, 28, 45) may be involvedin transport of cytokinetic factors to, or in stabilizationof, the spindle midzone. Several KLPs [CENP-E,MKLP1/CHO1 (hereafter MKLP1), KLP3A] are presentin the spindle midzone during anaphase (48, 72, 74).The distribution of these proteins appears ideal forparticipation in formation of the cleavage furrow, butonly KLP3A has been shown to be required for cytoki-nesis in Drosophila (72), where this function is re-stricted to male meiosis. CENP-E may be involved inthe metaphase-anaphase transition (74) but appar-ently does not function in cytokinesis (60). HumanMKLP1 has been reported to play a role in anaphase Bmovement based on its ability to cross-link antiparallelmicrotubules and to slide them past one another (48).However, two new members of the MKLP1 family,PAV-KLP from Drosophila melanogaster and ZEN-4from Caenorhabditis elegans, have recently beenshown by genetic analysis to be essential for cytokine-sis (1, 54). These motors are found at the spindlemidzone shortly after the onset of anaphase and ap-pear to stabilize the midzone microtubule bundles.Furthermore, PAV-KLP may mobilize mitotic and cy-tokinetic regulators, such as the polo kinase (41, 43,50). Whether MKLP1 performs primarily mitotic orcytokinetic functions in vertebrates remains unclear.

Article published online before print. See web site for date ofpublication (http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: H. W.Detrich III, Dept. of Biology, Northeastern Univ., 414 Mugar Hall,360 Huntington Ave., Boston, MA 02115 (E-mail: [email protected]).

Physiol Genomics 8: 51–66, 2002.First published November 27, 2001; 10.1152/physiolgenomics.00042.2001.

1094-8341/02 $5.00 Copyright © 2002 the American Physiological Society 51

The zebrafish, Danio rerio, is an excellent experi-mental model for genetic and cell-biological analysis ofKLP function in vertebrate development (8, 12, 15, 22,29). Our objectives in this report are to determine thestructure of Mklp1, the zebrafish ortholog of humanMKLP1 and hamster CHO1, to map the chromosomallocus of the zebrafish mklp1 gene, and to assess thefunction of Mklp1 during embryonic cleavage. To thisend, we have cloned a full-length zebrafish mklp1cDNA, mapped the zebrafish mklp1 locus to linkagegroup 18, and employed epifluorescence microscopy tolocate wild-type or green fluorescent protein (GFP)-tagged Mklp1 in cultured cell models and in zebrafishembryos. Using Mklp1-deletion mutants, we have de-termined the nuclear targeting signals of this motorand have shown that wild-type Mklp1 plays an impor-tant role in embryonic cytokinesis. We propose thatsuccessful completion of cytokinesis in vertebrate em-bryos requires the participation of MKLP1 orthologs inthe stabilization of the microtubule midbody, in thetransport of critical molecules to the cleavage furrow,or both.

EXPERIMENTAL PROCEDURES

Fish maintenance and embryo culture. Wild-type ze-brafish, D. rerio, were obtained from EKK Will WaterlifeResources (Gibsonton, FL) and were maintained in 40-lfreshwater aquaria on a 14:10-h light/dark photoperiod at28–29°C (70). Embryos, obtained by mating two males withthree or four females, were collected and placed in egg water(0.03% Instant Ocean synthetic sea salts in deionized water)within the first hour postfertilization. Staging of embryosfollowed the criteria of Kimmel et al. (35).

Cell culture. Zebrafish AB9 cells, a primary fibroblast cellline developed from fin tissue of the AB strain (51), weregrown in sterile 60-mm culture dishes containing DMEMmedium supplemented with 10% FBS. Cells were incubatedin a humid, 5% CO2 environment at 29°C. Under theseconditions, cells split 1:4 doubled in number in �72 h whenfed with fresh culture medium at 3-day intervals.

African green monkey COS-7 cells were cultured at 37°C inDMEM supplemented with 10% FBS in a humid, 5% CO2

incubator. Cells split 1:10 doubled every 24 h when fed freshculture medium at 3-day intervals.

Cloning and sequencing of the zebrafish mklp1 cDNA. Anoligo(dT)-primed zebrafish head kidney cDNA library (66) inthe vector Lambda ZAP Express was screened for clonescontaining potential mlkp1 cDNA inserts by hybridization toa 166-bp, mklp1-related cDNA [clone mp8; obtained as aserendipitous by-product of an RT-PCR based screen forhematopoietic transcription factors (66)]. A total of 30 candi-date mklp1 cDNA isolates were obtained from a screen of�500,000 recombinant phage, and three of these (clonesT8–2, T11–3, and T11–8) were carried through tertiaryplaque purification and in vivo excision to generate subclonesin the plasmid pBK-CMV. Plasmid pmklp1, excised fromphage clone T11–3, contained a full-length cDNA insert of�3.7 kb. The parental cDNA, and nested deletions (26)thereof, were sequenced on both strands by use of thedideoxynucleotide chain termination method (59).

Confirmation that the cDNA insert of pmklp1 encoded thezebrafish ortholog of mammalian MKLP1 was obtained byscanning the deduced protein sequence against the GenBankdatabase using the BLASTP program (National Center for

Biotechnology Information). Pairwise comparisons of theMklp1 protein sequence to those of other KLPs (GenBankaccession numbers accompany text) were performed usingthe algorithm of Needleman and Wunsch (47) as imple-mented by DNASTAR AALIGN (ver. 1.65).

GenBank data deposition. The sequence of the zebrafishmklp1 cDNA has been deposited in the GenBank databaseunder the accession number AF139990.

Radiation hybrid mapping. The zebrafish mklp1 gene wasmapped to the zebrafish genome using the Goodfellow T51panel, which contains 94 radiation hybrid lines (18, 38, 39),and the PCR-based protocol described on the Children’s Hospi-tal Zebrafish Genome Initiative web page (http://zfrhmaps.tch.harvard.edu/ZonRHmapper/). The primers, whichwere derived from the 3�-untranslated region (3�-UTR) of themklp1 cDNA, were 5� CAGGTGAATTTTTCATGGCAACA 3�and 5� AGCCTCTCATCACTGTGTGCAATA 3�. Amplifica-tion was performed for 40 cycles using the following program:1) template denaturation at 94°C for 30 s; 2) primer anneal-ing at 55°C for 30 s; and 3) elongation at 72°C for 60 s. Theexpected size of the fragment amplified using these primerswas 164 bp. The map position of zebrafish mklp1 was calcu-lated on the Goodfellow T51 panel using SAMapper 1.0 (64).

Synteny analysis. Orthologous gene pairs surrounding thezebrafish mklp1 and human MKLP1 loci were identified byreciprocal best-hit BLASTX and BLASTN searches using thecriteria and databases described by others (2, 52, 73). Ze-brafish genes were mapped to the radiation hybrid T51 panelas described previously. The map positions of the correspond-ing human genes were obtained from the Human GenomeProject Working Draft at University of California, SantaCruz (freeze of December 12, 2000) at http://genome.ucsc.edu.

Electrophoresis and immunoblotting of cell and embryoextracts. Lysates of AB9 cells were prepared by homogeniza-tion of whole cells in electrophoresis sample buffer (40).KLP-enriched extracts of zebrafish cleavage-stage embryoswere isolated by the 5�-adenylylimidodiphosphate (AMP-PNP)-dependent microtubule-affinity protocol of Wagner etal. (69). Proteins in these samples were separated by SDS-polyacrylamide gel electrophoresis (40) on 10% gels. Electro-phoretic transfer of proteins from SDS-polyacrylamide gelsto Immobilon-P membranes (Millipore) was performed bythe method of Towbin et al. (67). Proteins were visualizedby transillumination of membranes wetted with 20% meth-anol (method at http://www.millipore.com/analytical/pubdbase.nsf/docs/TP001.html).

To determine whether AB9 cell extracts and microtubulepreparations from cleavage-stage embryos contained Mklp1,Western blots were probed with mouse monoclonal and rab-bit polyclonal antibodies, respectively, that have been shownto be specific for the CHO1 antigen (i.e., MKLP1) of mammals(37, 61). (For immunoblotting protocol, see Pierce Chemicaltechnical document 0636, available at http://www.piercenet.com. Primary antibodies were generously pro-vided by Dr. Ryoko Kuriyama, University of MinnesotaSchool of Medicine.) CHO1 IgM-class monoclonal and IgG-class polyclonal antibodies were used at dilutions of 1:1,000and 1:3,000, respectively, and secondary antibodies [horse-radish peroxidase (HRP)-conjugated goat anti-mouse IgM(Sigma) and HRP-conjugated goat anti-rabbit IgG (Sigma)]were used at dilutions of 1:50,000 and 1:75,000. Antigen-antibody complexes were detected using the SuperSignalWest Pico Chemiluminescent Substrate (Pierce Chemical).

Immunocytochemistry of zebrafish cells and embryos.Freshly confluent cultures of AB9 cells were split 1:2 andplated onto sterile glass coverslips in 60-mm culture dishes.

52 ZEBRAFISH MKLP1 IN EMBRYONIC CYTOKINESIS

Physiol Genomics • VOL 8 • www.physiolgenomics.org

On the following day, the cells were fixed with 100% metha-nol for 4 min at �20°C. After rehydration in blocking solution(2% BSA, 2% normal goat serum, 0.1% Triton X-100, 0.75%glycine, and 0.2% sodium azide in PBS), the fixed cell prep-aration was incubated (2 h, 37°C) with the mouse monoclonalantibodies CHO1 and DM1D (mouse anti-chicken �-tubulin;IgG class; Sigma Chemical) at dilutions of 1:500 and 1:250,respectively. After washing the coverslips with blocking so-lution (three 5-min washes), secondary antibodies [indocar-bocyanine (Cy3)-conjugated goat anti-mouse IgM and fluo-rescein isothiocyanate (FITC)-conjugated goat anti-mouseIgG; Jackson ImmunoResearch Laboratories] were applied tothe fixed cells at 1:200 dilutions (2 h, 37°C). Finally, the cellpreparations were washed three times with blocking solu-tion. Hoechst 33258 (Polysciences) was added to the lastwash (1 �g/ml) to stain nuclei. Coverslips were mounted onglass slides with clear nail polish.

Dechorionated zebrafish embryos were processed for wholeembryo immunostaining by a modification of the procedure ofSolnica-Krezel and Driever (63). Because the epitope recog-nized by the CHO1 antibody is disrupted by aldehyde fixa-tives, embryos were fixed with cold methanol (�20°C), whichyields suboptimal, but acceptable, preservation of microtu-bules. Fixed embryos were incubated with the CHO1 andDM1D primary antibodies on a rocking platform for 16–18 hat 4°C and washed with blocking solution (four washes of 1 heach, room temperature). Finally, the embryos were incu-bated with the Cy3- and FITC-conjugated secondary antibod-ies (16–18 h with rocking, 4°C), washed with blocking solu-tion (4�1 h, room temperature), and then mounted in ananti-fading solution (70% glycerol, 0.2% sodium azide inPBS) on bridge slides.

RT-PCR analysis of zebrafish mklp1 mRNA levels duringembryonic development. The steady-state level of zebrafishmklp1 mRNA at discrete developmental stages was analyzedqualitatively by RT-PCR. To provide a framework for com-parison, the stage-specific transcript levels of zebrafish elon-gation factor 1� (EF1�) and transcription factor Pou2 werealso determined. Details of the protocol, including primerpairs, can be found at http://www.biology.neu.edu/detrich.html.

Whole embryo in situ hybridization. Whole mount in situhybridization using zebrafish mklp1 antisense RNA was per-formed by a modification (66) of the method of Detrich et al.(11). Stained embryos were transferred to 80% glycerol in 1�PBS for microscopic observation. The protocol can be found athttp://www.biology.neu.edu/detrich.html.

Construction of expression plasmids encoding GFP-taggedwild-type and mutant Mklp1s. The expression plasmidspGFP-Mklp1 [encodes GFP (6) fused through an 18-residuespacer (GSKEFGTRRRGIKLKHLS) to the amino terminus ofwild-type Mklp1], pGFP-Mklp1(�N1–275) (deletes the first275 codons of mklp1), and pGFP-Mklp1(�C591–867) (re-moves the last 277 codons of the mklp1 cDNA) were engi-neered from pmklp1 as described previously (7). Four additionaldeletion subclones, pGFP-Mklp1(�721–867), pGFP-Mklp1(�815–818), pGFP-Mklp1(�863–867), and pGFP-Mklp1(�815–818, �863–867), were generated by PCR-mediated mutagenesisusing pfu DNA polymerase (Stratagene). The construct pGFP-Mklp1(T119N), encoding a point mutation (T119N) in Mklp1,was created via primer-based mutagenesis. Full details regard-ing construction of the new deletion mutants and the pointmutant may be found at http://www.biology.neu.edu/detrich.html. As a control, pBK-GFP was constructed by sub-cloning the GFP cDNA, containing 5�-SpeI and 3�-BamHI link-ers (7), between the SpeI and BamHI sites of pBK-CMV. Allconstructs created by PCR were sequenced to verify, in the

regions subjected to amplification, both the intended alterationsand the absence of secondary mutations.

Transfection of COS-7 cells with wild-type and deletionmutants of pGFP-Mklp1 and immunodetection of the fusionproteins. COS-7 cells were transfected with pGFP-Mklp1 (orone of its derivatives) by a modification of the DEAE-dextranprotocol described by Sambrook et al. (58). COS-7 cell cul-tures were split 1:2, and the cells were seeded onto sterileglass coverslips in 60-mm culture dishes as described abovefor zebrafish AB9 cells. Twenty-four hours later, the cellswere incubated with 2-ml DEAE-dextran solution (4 mg/mlDEAE-dextran, 1 mM chloroquine in DMEM) containing�1.5 �g of plasmid for 1–3 h at 37°C. The DEAE-dextransolution was removed, and cells were then shocked by expo-sure to 2 ml of DMSO for 2 min at room temperature. Afterremoval of the DMSO solution, cells were washed with PBS(37°C), fed with 2-ml DMEM plus 10% FBS, and incubatedfor 20–24 h before processing for immunostaining withmouse anti-� tubulin IgG (Sigma) and rabbit anti-GFP IgG(Clontech) primary antibodies (see procedures under Immu-nocytochemistry of zebrafish cells and embryos, above). Thesecondary antibodies were FITC-conjugated goat anti-mouseIgG and Cy3-conjugated goat anti-rabbit IgG (Jackson Im-munoResearch Laboratories).

In vitro synthesis of mRNA and embryo microinjection.pGFP-Mklp1, pGFP-Mklp1(�N1–275), pGFP-Mklp1(T119N),and pBK-GFP were cut with KpnI to generate linear tem-plates for in vitro transcription. 5�-Capped mRNAs weresynthesized from the templates using T3 RNA polymerase(Promega) and T3 Cap-Scribe nucleotide solution (Boeh-ringer-Mannheim Biochemicals). Only those preparationsthat gave a single, nondegraded RNA band of appropriatesize when evaluated by denaturing electrophoresis (58) wereused for microinjection.

Microinjection of synthetic mRNAs into manually dechori-onated embryos was performed at the one-cell stage as de-scribed by Westerfield (70) using a PLI-100 Picoinjector(Medical Systems). Approximately 5 nl of an mRNA solution(or of sterile water as a control) were injected per embryo.Data are reported only for those experiments in which �80%of the water-injected controls developed through the shieldstage. Embryos either were examined alive by confocal epi-fluorescence microscopy to visualize the GFP signal or werefixed (3.7% formaldehyde, 4 h, room temperature) after 4–5 hof development and processed for immunostaining. Cellboundaries in fixed embryos were delineated by incubationwith a rabbit anti-pan-cadherin polyclonal antibody (IgGclass; Sigma Chemical) followed by a Cy3-conjugated goatanti-rabbit IgG secondary antibody (Jackson ImmmunoRe-search). The nuclei of fixed embryos were stained with SY-TOX per the manufacturer’s instructions (Molecular Probes).

Microscopy. Epifluorescence microscopy of AB9 and COS-7cells was performed with an Olympus Model AH-2 micro-scope equipped with 20�, 40�, and 100� objectives and withFITC (excitation wavelengths 380–490 nm, emission �515 nm), Texas Red (excitation 465–550 nm, emission �590 nm), and ultraviolet (excitation wavelengths 330–380nm, emission � 420 nm) filter sets. Photomicrographs re-corded on Kodak Ektachrome 1600 slide film were digitizedat 405 dots/cm by use of a Polaroid Sprint 35 slide scanner.The intracellular distributions of GFP-Mklp1 fusion proteinsin living zebrafish embryos, and of fixed embryos stainedwith anti-pan-cadherin antibody and SYTOX, were examinedby confocal epifluorescence microscopy using a Bio-Radmodel MRC-600 microscope. Embryos hybridized in situ withRNA probes were examined and photographed with a Nikonmodel SMZ-U dissecting microscope.

53ZEBRAFISH MKLP1 IN EMBRYONIC CYTOKINESIS


ron

ron

ron

ron

ron

RESULTS

Primary sequence and structural organization of ze-brafish Mklp1. The 867-residue Mklp1 protein wasfound to be closely related to, but shorter than, two

MKLP1s from mammals, human MKLP1 and CHO1(68 and 71% overall sequence identity, respectively)(Fig. 1). The sequence identity of Mklp1 to theseMKLP1s was greatest in their motor domains (77% toboth human MKLP1 and CHO1 covering residues

Fig. 1. Primary sequence of zebrafish mitotic kinesin-like protein 1 (zMklp1). Sequence alignment of Mklp1with respect to its orthologs in the MKLP1/CHO1family: CgCHO1, MKLP1 from the Chinese hamster,Cricetulus griseus (GenBank accession no. X83575);HsMKLP1, MKLP1 from the human, Homo sapiens(accession no. X67155); PAV-KLP, MKLP1 encoded bythe Drosophila melanogaster gene pavarotti (accessionno. AJ224882); and ZEN-4; MKLP1 encoded by thezen-4 gene of Caenorhabditis elegans (accession no.AF057567). Residues that are identical and/or con-served in three or more of the five proteins are shown inreversed text. Dashes indicate insertions introduced tomaximize sequence identity. The P-loop sequence, GS-GKT (residues 115–119), is underscored and labeled,and two oligopeptides common to most KLPs, SSRSHSand DLAGSE (residues 298–303 and 337–342), areindicated by dashed underscore. Three putative nuclearlocalization signals located near the amino- and car-boxyl termini are underscored and labeled with Romannumerals.



1–350) but was also significant in their presumptivecargo-binding tails (71% to MKLP1, 70% to CHO1covering Mklp1 residues 650–867). However, the ze-brafish motor lacked the long insertion of CHO1 (resi-dues 691–793) and the �100-residue, COOH-terminalextension of the human motor. The similarity of Mklp1to those of invertebrates was considerably lower (39%for D. melanogaster PAV-KLP; 34% for C. elegans ZEN-4). We conclude that Mklp1 is a relatively short mem-

ber of the vertebrate subgroup of MKLP1 family mo-tors.

Sequence motifs shared by Mklp1, human MKLP1,CHO1, and many other KLPs included a P-loop se-quence, GSGKT (residues 115–119), and two oligopep-tides, SSRSHS and DLAGSE (residues 298–303 and337–342) that are likely to be involved in sensing thestatus of bound adenosine nucleotide and propagatingthe conformational changes that generate mechano-

Fig. 1.—Continued



chemical work (36, 68). Furthermore, potential nu-clear-localization signals were present in both the ami-no- and carboxy-terminal regions (Fig. 1, residues8–12, 815–818, and 863–867).

The quaternary structural organization of Mklp1also appeared to be similar to that of its mammaliancounterparts. Prediction of secondary structure usingCOILS (ver. 2.1) revealed that residues 550–640 ofMklp1 have a high propensity for formation of an�-helical coiled coil. Thus Mklp1, like human MKLP1and hamster CHO1, probably forms a parallel coiled-coil homodimer through its stalk, thus positioning twoNH2-terminal motor domains at one end of the elon-gated molecule and the COOH-terminal tails at theother (37).

Mapping of mklp1 to the zebrafish genome and com-parative analysis to the human genome. Using a ze-brafish radiation hybrid panel, we mapped the mklp1gene to within 96 cR of the centromere marker z5479 oflinkage group 18 (Fig. 2). The orthologous human gene,MKLP1 (accession no. X67155), was found on the longarm of chromosome 15. To determine whether thezebrafish and human chromosomal segments encom-passing MKLP1 constitute syntenic elements con-served between the two genomes, we examined theregions on either side of the motor gene. On the telo-meric side of mklp1, we identified three genes, grou-cho2, MAPKK1, and smad6, whose human orthologsalso mapped near the MKLP1 gene. The absolute po-sitions of the four genes of this group appear to havebeen altered by two inversions. No zebrafish/humansyntenies were found on the centromeric side of mklp1,perhaps indicating that the zebrafish gene resides nearan inversion or translocation breakpoint that differen-tiates the two genomes.

Intracellular distribution of Mklp1 in zebrafish AB9cells. The monoclonal CHO1 antibody has been shownpreviously to be specific for an epitope present in the

Fig. 2. Synteny of zebrafish and human chromosomal segments encompass-ing the mklp1/MKLP1 gene. The region of zebrafish linkage group 18 con-taining the MKLP1, groucho2, MAP2K1, and smad6 genes (left bar, betweensimple-sequence length polymorphism markers z13426 and z5479) is com-pared with the segment of human chromosome 15 (right bar, 15q22.31–15q23) that contains the orthologous human genes. Zebrafish and humanorthologs are indicated by “EST/GenBank” or “GenBank/GenBank” designa-tors (e.g., fk08g05.y1 AF035528 or Y12466 A142116, respectively), thenames of the genes are shown below the designators in italic font, and theirloci are indicated by the solid lines and white bars. Physical separationbetween zebrafish genes is shown on the vertical axis in centirays (one cR �60 kb), whereas human genes are bracketed by their base positions (kb).Arrows indicate direction to the respective chromosomal centromeres (cen).EST, expressed sequence tag.

Fig. 3. Detection of zebrafish Mklp1 using anti-CHO1 antibodies. A:whole cell extract of AB9 cells (protein) and immunoblot stained withthe monoclonal CHO1 antibody (mCHO1). B: KLP-enriched micro-tubule fraction obtained from cleavage-stage zebrafish embryos (pro-tein) and immunoblot stained with the polyclonal CHO1 antibody(pCHO1). The molecular masses of protein standards (kDa) areindicated in the middle.



COOH-terminal half of MKLP1/CHO1 family motorsfrom several mammalian cell lines (37, 48, 49). Fig-ure 3A shows that the CHO1 antibody recognizedspecifically a single protein of �99 kDa in whole cellextracts of zebrafish AB9 cells, which is consistentwith the predicted mass of Mklp1 (98,673 Da). Sim-ilarly, a polyclonal CHO1 antibody (37) detected asingle, 99-kDa protein in a KLP-enriched microtu-bule fraction obtained from cleavage-stage zebrafishembryos (Fig. 3B). These observations indicate thatthe CHO1 antibodies cross-react with zebrafishMklp1.

The cell cycle-dependent intracellular distribution ofMklp1 was analyzed by staining zebrafish AB9 cellswith the monoclonal CHO1 antibody. Figure 4 showsthat Mklp1 colocalized with a subset of microtubules ina pattern similar to that described for MKLP1 motorsin other cell systems. The motor epitope first appearedassociated with the microtubules of mitotic spindles atpro-metaphase (Fig. 4C), became concentrated in themicrotubules of the spindle interzone in anaphase (Fig.4D), and remained associated with the spindle mid-body during telophase and cytokinesis (Fig. 4, E andF). In some cultures we detected Mklp1 in the nuclei ofAB9 cells during interphase, but this observation hasnot been consistently reproducible (e.g., Fig. 4, A andB). (Consideration of the significance of this observa-tion is reserved to the DISCUSSION, below.) Staining ofcentrosomes, either in interphase or in mitosis, was notobserved.

Expression and subcellular distribution of Mklp1 intransfected COS-7 cells. To verify the cell cycle-depen-dent behavior of Mklp1 revealed by our immunocyto-

chemical observations, we engineered the expression ofa GFP-Mklp1 fusion protein in COS-7 cells. Figure 5shows that the GFP-tagged motor was readily detectedin the nuclei of cells in interphase (A" and D"), redis-tributed to spindle microtubules at metaphase (B"),migrated to the spindle interzone in anaphase (C"), andfinally became concentrated in the midbody duringtelophase and cytokinesis (D"). Thus the distribution ofMklp1 throughout the cell cycle in this heterologouscell system resembles that reported for other verte-brate MKLPs.

Nuclear targeting of Mklp1. Sequence analysis ofMklp1 revealed potential nuclear localization signalsin both the amino- and carboxy-terminal regions ofthe motor (Fig. 1). To define the sequence motifs thatconfer nuclear targeting of Mklp1 during interphase,we constructed a series of deletion mutants in pGFP-Mklp1 and examined the cell cycle-dependent behav-ior of the mutant motors in COS-7 cells. Removal ofthe amino terminus (residues 1–275) had no effecton nuclear localization during interphase, thus rul-ing out the sequence element therein (KTPRR, resi-dues 8–12) as a functional targeting signal (Fig. 6).In contrast, deletion of large portions of the carboxyterminus (residues 591–867 or 720–867) abrogatednuclear accumulation. Removal of either of theCOOH-terminal sequence elements, 815RKRR818 or863KRRKP867, resulted in an ambiguous, nuclear/cytoplasmic phenotype, whereas their removal incombination eliminated nuclear targeting. There-fore, the two elements apparently constitute a bipar-tite nuclear localization motif (13).

Fig. 4. Distribution of Mklp1 in zebrafish AB9 cells during the cell cycle. Cells at interphase (A and B columns),pro-metaphase (C column), anaphase (D column), telophase (E column), and cytokinesis (F column). Each cell wasexamined by phase-contrast microscopy (no prime mark) and by epifluorescence microscopy to detect DNA(Hoechst 33258 signal, single prime mark), microtubules (FITC signal, double prime), or Mklp1 (Cy3 signal, tripleprime). Immunocytochemistry of proteins and staining of DNA were performed as described under EXPERIMENTALPROCEDURES. Bar in F 5 �m. Cy3, indocarbocyanine; FITC, fluorescein isothiocyanate.



Expression of the mklp1 gene in zebrafish embryos.We investigated the quantity and localization of mklp1mRNA in zebrafish embryos, both preceding and fol-lowing the mid-blastula transition (MBT), by RT-PCRand by whole mount in situ hybridization. mklp1mRNA was abundant in the zygote, cleavage, and early

blastula stages, which indicates that message is ma-ternally supplied (for figure, visit http://www.biology.neu.edu/detrich.html). Following MBT, transcriptlevels declined moderately during the gastrula andsomitic stages and then increased markedly by thehatching stage. By contrast, stage-specific transcript

Fig. 5. Cell cycle-dependent distribution of GFP-Mklp1 in COS-7 cells. Cells at interphase (A), metaphase (B),anaphase (C), and cytokinesis (D). Each cell was examined by phase-contrast microscopy (no prime mark) and byepifluorescence microscopy to detect microtubules (green FITC signal, single prime), or GFP-Mklp1 (red Cy3signal, double prime). Transfection of cells with pGFP-Mklp1 and immunodetection of microtubules and GFP-Mklp1 were performed as described under EXPERIMENTAL PROCEDURES. Arrows in D indicate the midbody. Bar inD" 5 �m. GFP, green fluorescent protein.



levels of zebrafish EF1� and Pou2 were consistent withprior reports (17, 25), with the former abundant afterMBT and the latter abundant up to the mid-somitestage. We conclude that embryonic zebrafish mklp1mRNAs are supplied by transcription of both the ma-ternal and zygotic genomes.

Whole mount in situ hybridization was performed toexamine both the temporal and spatial distribution ofmklp1 mRNAs in zebrafish embryos (data not shown).mklp1 transcripts appeared to be distributed uni-formly in the cells of embryos from the zygote periodthrough gastrulation (0–10 h). During early somito-genesis (10–12 h), mklp1 mRNAs were presentthroughout the embryo, with greater concentrationsapparent in the head and tail regions. At late somito-genesis (18–22 h) and the prim-5 stage (24 h), themessage was concentrated primarily in the embryonicbrain, which was undergoing neuromere formation andthe sculpturing of fore-, mid-, and hindbrain regions.

Location of wild-type Mklp1 in cleaving zebrafishembryos. We evaluated the spatial distribution of theMklp1 motor during early embryogenesis by epifluo-rescence confocal microscopy of embryos stained withCHO1 and �-tubulin antibodies. Figure 7A shows thestaining observed at the cleavage furrow between twoblastomeres at the eight-cell stage, and Fig. 7B pre-sents an interpretative diagram. Mklp1 was found in asubplasmalemmal annulus in close contact with theouter microtubules of the spindle midzone and nascentmidbody. This pattern was observed at all cleavagefurrows in early embryos.

When living zebrafish embryos were injected at theone- or two-cell stage with synthetic mRNAs encodingwild-type GFP-Mklp1, the tagged motor was detectedboth in the nuclei of, and in discrete foci equidistantbetween, cleaving blastomeres (Fig. 8, A and B). Thelatter probably correspond to cytokinetic midbodies.Optical sectioning of the cleavage furrow (arrowhead)between the blastomeres indicated in Fig. 8B (curvedlines) showed that the motor was distributed as a ring

beneath the equatorial cortex (data not shown; forfigure, visit http://www.biology.neu.edu/detrich.html). Some staining of the central midbody was alsoobserved. These results are consistent with physicallinkage of the microtubule midbody to the cleavagefurrow via Mklp1 and suggest that the motor mayparticipate in cytokinesis.

Function of Mklp1 in zebrafish development. To in-vestigate the function of Mklp1 in embryogenesis, weemployed a dominant-negative strategy (27) by ex-pressing the GFP-tagged, amino-terminal mutantsGFP-Mklp1(�N1–275) and GFP-Mklp1(T119N) in pre-blastula embryos. Both mutations should disrupt themotility of the Mklp1 motor, in the former by eliminat-ing most of its motor domain and in the latter byreducing or abolishing its ATPase activity (46). In thecase of the homodimeric Mklp1 motor, expression ofthe dominant-negative derivatives in excess shouldcause the wild-type motor chains to be sequestered infunctionally inactive hybrids. The resulting phenotypiceffects on zebrafish embryogenesis would then suggestthe function of the wild-type motor.

To determine whether the dominant-negative mo-tors were defective in their ability to associate withmicrotubules and to translocate to the cleavage furrow,we transfected COS-7 cells with the correspondingexpression plasmids, after which the fusion proteinswere detected immunocytochemically. Figure 9 showsthat the GFP-Mklp1(�N1–275) fusion protein accumu-lated in the nuclei of cells in interphase (Fig. 9, A andA�) but failed to associate with the microtubule cy-toskeleton or to migrate to the midbody in cells com-pleting cytokinesis (B and B�).1 Similar results wereobtained for the GFP-Mklp1 motor with the alteredATP-binding motif (T119N) (data not shown). One in-

1The absence of a gross phenotypic effect of the mutant Mklp1s onCOS-7 cells probably results from low levels of expression of theexogenous motor during the short posttransfection incubations.

Fig. 6. Nuclear targeting of Mklp1. COS-7cells were transfected with pGFP-Mklp1,and deletions thereof, and the expressedproteins were detected by indirect immuno-cytochemistry (see EXPERIMENTAL PROCE-DURES). Deletion mutants, all tagged at theamino terminus with GFP: Mklp1�N,Mklp1 lacking the first 275 residues;Mklp1�C, Mklp1 lacking the final 277 res-idues; Mklp1(�720–867), Mklp1(�815–818), and Mklp1(�863–867), Mklp1 lackingthe residues indicated; Mklp1(double �),Mklp1 lacking both residues 815–818 and863–867. The subcellular distribution ofwild-type and deletion mutants of GFP-Mklp1 is indicated by N (nuclear), C (cyto-plasmic), or N/C (both nuclear and cytoplas-mic). M, motor domain; S, stalk domain; T,tail domain.



stance of an interphase COS-7 cell with four GFP-positive nuclei was observed for the T119N transfec-tion, which suggests that the dominant-negative motormight be able to cause a defect in cytokinesis. Ratherthan pursue the latter observation in this heterologouscell line, we continued our functional studies of Mklp1in zebrafish embryos.

Zebrafish embryos injected with small quantities(0.5 ng) of mRNAs encoding the dominant-negativemutants of GFP-Mklp1 [GFP-Mklp1(�N1–275), Fig. 8,C and D; GFP-Mklp1(T119N), data not shown] consis-tently demonstrated, during cleavage, brightly stain-ing nuclei but few of the intercellular foci observedwith wild-type GFP-mklp1 mRNA (Fig. 8A, B). Occa-sionally, binucleated blastomeres were observed (Fig.8C). These results indicated the following: 1) full-length fusion proteins were translated from the syn-thetic mRNAs (the functional nuclear localization sig-nals reside at the COOH terminus of Mklp1); 2) bothGFP-Mklp1(�N1–275) and GFP-Mklp1(T119N) weredefective in microtubule-dependent translocation; and3) completion of cytokinesis was inhibited in someblastomeres. However, no impairment of cleavage rateor subsequent morphogenetic movements was appar-ent, presumably because production of the defectivemotor chains in most blastomeres was not sufficient toprevent formation of near-normal levels of wild-typeMklp1 dimers. We interpret the GFP signal detected ata small proportion of cleavage furrows to be the resultof rare movement to the midbody of translocation-

deficient heterodimers (see DISCUSSION) composed of oneendogenous wild-type Mklp1 subunit and one GFP-tagged dominant-negative subunit.

The observation that cytokinesis appeared to be in-hibited in some blastomeres at low doses of injecteddominant-negative GFP-mklp1 mRNA motivated anexamination of its effects at higher quantity. Zebrafishembryos at the one- or two-cell stage were injected withGFP-mklp1(�N1–275) mRNA, or with wild-type GFP-mklp1 mRNA or GFP control mRNAs, at doses of 1.5,3.0, or 4.5 ng (Table 1). Those receiving the lowest doseshowed little, if any, developmental phenotype irre-spective of mRNA type. At the highest dose, cleavagewas disrupted by both dominant-negative and controlmRNAs, most likely due to nonspecific developmentaleffects of such large boluses of mRNA (30). The inter-mediate level of GFP-mklp1(�N1–275) mRNA was,however, quite informative. Approximately 30% of em-bryos (11 of the 37 scored) that received the GFP-mklp1(�N1–275) mRNA failed to develop beyond thefour- to eight-cell stage. These embryos repeatedlyinitiated cytokinesis, but their cleavage furrows re-tracted prior to achieving cell partition. Ten percent ofexperimental embryos (4 of 37) did not initiate epibolicmovement within the interval during which controlsreached 50% epiboly. The remaining 60% (22 of 37scored embryos) appeared to develop normally to thebud stage but contained larger blastomeres duringcleavage and exhibited slower epibolic movement thandid embryos that received the control mRNAs. Some of

Fig. 7. Colocalization of Mklp1 and mi-crotubules at the cleavage furrow. A:merged confocal image of the cleavagefurrow between two blastomeres of aneight-cell embryo showing Mklp1 inred (CHO1 primary antibody, Cy3-con-jugated secondary antibody), microtu-bules in green (anti-chick brain �-tubu-lin primary antibody, FITC-conjugatedsecondary), and the overlap of the an-tigens in yellow. Preservation of themicrotubules was suboptimal due tothe use of methanol fixation to preservethe CHO1 epitope of Mklp1. The em-bryo is oriented with the animal pole tothe left and the vegetal pole/yolk cell(not shown) to the right. Bar 100 �m.B: interpretation of the staining ob-served in A. Midzone microtubules (orsmall bundles of microtubules) are in-dicated by the straight lines that inter-sect to form the left-to-right chevronpattern, and Mklp1 is represented bythe solid circles located at the chevronvertices.



the embryos of this last group developed through thesomite stages but showed poor tissue organization and,at 24 h, did not move in response to mechanical stim-ulation (data not shown). Tissue degeneration com-menced at 28 h in this subset of embryos. By contrast,embryos that received 3.0 ng of the wild-type GFP-mklp1 or GFP mRNAs showed few developmental ab-normalities (Table 1).

The cellular defect resulting from injection of mutantmRNA was investigated by confocal microscopy. Fig-ure 10 compares embryos injected at the one- or two-cell stage with dominant-negative or wild-type GFP-mklp1 mRNAs at the intermediate dose. Blastomeresfrom embryos injected with GFP-mklp1(�N1–275)mRNA contained many nuclei in a common cytoplasm(Fig. 10, A–D), consistent with failure of cytokinesis.By contrast, normal cleavage furrows developed inembryos that received the control mRNAs, and multi-nucleated blastomeres were never observed (Fig. 10, Eand F).

DISCUSSION

Cytoplasmic microtubules and their associated mo-tors participate in many processes that govern devel-

opment of the metazoan embryo, including the mitoticdivisions of cleavage, segregation of the germ plasm,localization of determinants during embryonic axis for-mation, and morphogenetic movements (16, 32, 57, 63,65). Less well appreciated is the active participation ofthe microtubule cytoskeleton in embryonic cytokinesis.In this report, we characterize the zebrafish kinesin-like microtubule motor Mklp1, map the locus of themklp1 gene, and demonstrate that the motor is re-quired for successful completion of cytokinesis in cleav-ing zebrafish embryos.

Classification and structure of Mklp1. Structuralanalysis provides compelling support for assignment ofMklp1 to the MKLP1/CHO1 motor family. The primarysequence of Mklp1 in toto bears a striking resemblanceto those of the CHO1 and MKLP1 motors of Chinesehamster and human, and its predicted domain organi-zation (NH2-terminal motor, central stalk, and COOH-terminal tail) is identical to the mammalian proteins.The strong sequence similarity of the COOH-terminaltail domains of the three motors, a feature generallyrestricted to members of the same KLP family, furthersupports the assignment of Mklp1 to the MKLP1/CHO1 family (also termed the KIF23/CHO1 family;

Fig. 8. Distribution of wild-type anddominant-negative GFP-Mklp1 in liv-ing zebrafish embryos. One- or two-cellembryos were injected with mRNA (0.5ng) encoding either GFP-Mklp1 orGFP-Mklp1(�N1–275), and the GFPsignal of the expressed fusion proteinwas detected by confocal microscopy oflate blastulae. A and B: wild-type GFP-Mklp1. Numerous labeled nuclei andmidbody remnants (examples indi-cated by arrow and arrowhead, respec-tively) are visible (A). In dividing blas-tomeres (B), the fusion protein forms aring (arrowhead) equidistant betweenthe nascent daughter cells (boundariesindicated by curved lines). C and D:GFP-Mklp1(�N1–275). Many labelednuclei are present (C), but midbodystaining is rare (D). The arrows in Cand D show multinucleated blas-tomeres, and the arrowhead in D indi-cates a rare midbody containing GFP-Mklp1(�N1–275). Bars 50 �m.



Ref. 44) of the N-6 class of the KLP superfamily (28).Based on secondary-structure prediction, we suggestthat Mklp1 chains associate to form homodimersvia coiled-coil dimerization of its stalk domain [cf.Kuriyama et al. (37)].

Zebrafish/human synteny near the MKLP1 locus.Our observation that the MKLP1 gene orthologs ofzebrafish and human are contained in larger chromo-somal elements shared by linkage group 18 and chro-mosome 15, respectively, supports prior analyses thatportions of these chromosomes are evolutionarily re-lated (2, 73). Furthermore, the presence of two inver-sions within the four-gene synteny is consistent withthe hypothesis that rearrangements within chromo-somes, rather than translocations between, have

played the greater evolutionary role in the remodelingof vertebrate karyotypes (52). Given the large size ofthe KLP motor superfamily (4, 28, 44, 45), futuresystematic mapping of KLP genes to the genomes ofrepresentative vertebrates should provide a compellingtest of mechanistic models of vertebrate chromosomalevolution.

Distribution of Mklp1 during the cell cycle. In ze-brafish AB9 cells, the mitotic distribution of Mklp1,determined by staining with the CHO1 antibody (49),resembles that reported for CHO1 and human MKLP1in several mammalian cell lines (37, 48, 49), i.e., asso-ciation with spindle microtubules at metaphase, redis-tribution to the spindle interzone in anaphase, andconcentration in the midbody during telophase. During

Fig. 9. Cell cycle-dependent distribution of domi-nant-negative GFP-Mklp1 in COS-7 cells. A and B:cells in interphase and cytokinesis, respectively.During interphase, GFP-Mklp1(�N1–275) waspresent in the nucleus (A and A�). At cytokinesis (Band B�), in contrast, the fusion protein was locatedthroughout the cytoplasm with no significant accu-mulation at the midbody (arrows). Nuclei werestained with Hoechst 33258 (A). GFP-Mklp1(�N1–275) (A� and B�) and microtubules (B) were detectedimmunocytochemically. Bar in A� 5 �m.

Table 1. Phenotypes of embryos injected with dominant-negative and control mklp1 mRNAs

mRNA Dose, ng Embryos Injected Embryos Scored Impaired Cleavage Impaired Epiboly Epiboly Completed

GFP-mklp1(�N) 1.5 43* 35 0 7 28GFP-mklp1 1.5 22 16 0 3 13GFP 1.5 22 17 0 1 16GFP-mklp1(�N) 3.0 50* 37 11 4 22GFP-mklp1 3.0 25 20 0 3 17GFP 3.0 25 21 0 2 19GFP-mklp1(�N) 4.5 40* 34 34 NA NAGFP-mklp1 4.5 20 17 17 NA NAGFP 4.5 20 15 15 NA NA

Embryos were examined for a green fluorescent protein (GFP) signal 2 h after injection of mRNA. Only those showing a detectable signalwere scored for developmental phenotype. Epiboly, the flattening and spreading of the blastoderm over the yolk cell that commences late inthe blastula period and continues through gastrulation (70); GFP-Mklp1(�N), the GFP-Mklp1(�N1-275) fusion protein; NA, data notavailable. *Data pooled from two independent experiments.



interphase, by contrast, we detect the antigen occasion-ally in the nuclei of AB9 cells, in contrast to the con-sistent staining of nuclei in mammalian cells, andlocalization to the centrosome is not observed. Giventhe presence of a functional nuclear localization signalin the COOH terminus of Mklp1 (see below), the failureto detect consistently the motor in the nuclei of AB9cells is puzzling. Perhaps the epitope recognized by theCHO1 antibody is altered or not accessible whenMklp1 is present in nuclei of this cell line. An alterna-tive explanation may be that a second zebrafish para-log of Mklp1, available by virtue of tetraploidization ofthe zebrafish lineage (53) but devoid of the COOH-terminal epitope recognized by the CHO1 antibody,concentrates preferentially, but not exclusively, in thenuclei of AB9 fibroblasts to perform the nuclear func-tions of the single MKLP1/CHO1 motor of mammals.To test the latter hypothesis, additional zebrafish klpcDNAs and expressed sequence tags (ESTs) must bescreened for sequence affinity to the mklp1 gene.

When fused to GFP and expressed in COS-7 cells,Mklp1 demonstrates both nuclear localization duringinterphase and dynamic redistribution within the spin-dle during mitosis [cf. Nislow et al. (48, 49), Kuriyamaet al. (37)]. We therefore used COS-7 cells to map thenuclear localization signals of the Mklp1 motor. Ourresults indicate that nuclear targeting is conferred by abipartite, COOH-terminal motif composed of the se-quence elements RKRR and KRRKP separated by 44amino acid residues. In contrast, human MKLP1 and

PAV-KLP have been reported to contain the putativenuclear targeting motif KTPR near their NH2 termini,but the function of this motif has not been evaluated bydeletion analysis. Our demonstration that the corre-sponding element in Mklp1 (residues 8–11) does notcontribute to targeting of the motor to the interphasenucleus suggests that the nuclear localization se-quences of the human, hamster, and Drosophila or-thologs should be sought in their COOH termini. In-deed, CHO1 and human MKLP1 possess the firstMklp1 element RKRR, and PAV-KLP a variant(RKRP), in regions of high sequence conservation (Fig.1), and CHO1 also contains the second element(KRKK� vs. KRRKP in Mklp1). Furthermore, deletionmapping of large segments of human MKLP1 has im-plicated two overlapping sites, 797PNGSRKR803 and801RKRR804 (the latter corresponding to 815RKRR818 ofMklp1) in nuclear targeting of the motor (10). C. el-egans ZEN-4, which lacks the COOH-terminal regioncontaining the bipartite targeting motif (Fig. 1), asso-ciates with centrosomes in interphase cells, but not,apparently, with the nucleus (54).

In cleaving zebrafish embryos, the intracellular dis-tribution of Mklp1 resembles that observed in culturedcells, with specific localization to interphase nuclei, tocleavage furrows, and to midbody remnants. However,Mklp1, whether wild-type or GFP-tagged fusion, hasnot been detected in the metaphase spindles of cleav-ing embryos, possibly due to either rapid transit of themotor through the spindle to the future site of furrow

Fig. 10. Cytokinetic phenotypes of embryos injected with dominant-negative or control GFP-mklp1 mRNAs. A andB: multiple nuclei in a common cytoplasm. In the most severe phenotype observed, single blastomeres containedmany nuclei. B is a 2� enlargement of part of the embryo shown in A. C and D: less severe phenotype in whichblastomeres contained 2–4 nuclei. Note the clear delineation of the plasma membrane by the pan-cadherinstaining (D). E and F: embryos (8-cell stage) that received wild-type GFP-mklp1 mRNA formed normal cleavagefurrows (arrows). Multinucleated blastomeres were never observed in embryos that received the control mRNAs.Embryos were injected with dominant-negative or control mRNAs at the one- to two-cell stage as described underEXPERIMENTAL PROCEDURES. Bars in A, B, E, and F 50 �m, whereas those in C and D 10 �m.



formation or to a low density of the motor along spindlemicrotubules. At the furrow, Mklp1 forms a ring, ordisk, at the rim of the cytoplasmic bridge linking thenascent daughter cells [cf. C. elegans ZEN-4; Raich etal. (54)]. As mitosis ends and cytokinesis begins, Mklp1condenses to form a compact mass coincident with themidbody. With few exceptions, dominant-negativeGFP-tagged Mklp1 (i.e., the motor domain deletion)fails to appear in midbody remnants between zebrafishblastomeres completing normal cytokinesis (Fig. 8).Since the rapid, processive movement of kinesins alongmicrotubules generally depends on the presence of twofunctional motor domains (3, 20, 24, 75), relatively fewwild-type/dominant-negative hybrid Mklp1 motorswould be expected to accumulate at the midbody.Taken together, our observations of the intracellulardistribution of Mklp1 in the blastomeres of cleavingzebrafish embryos support strongly a role for Mklp1 incytokinesis.

Functional requirement for Mklp1 in cytokinesis, notmitosis. Our observation that Mklp1 is required forcytokinesis in embryonic zebrafish stands in apparentcontradiction to prior reports that suggest that themammalian orthologs function in mitosis. HumanMKLP1 and Chinese hamster CHO1 were originallyproposed to provide the force necessary for separationof the spindle poles during mitosis (37, 48, 49). Recom-binant human MKLP1 has been shown to bundle an-tiparallel microtubules in vitro and to disperse thebundles by ATP-dependent sliding (48), activities rem-iniscent of the antiparallel sliding of polar microtu-bules during anaphase B. Furthermore, perturbationexperiments employing anti-MKLP1 antibody demon-strated mitotic arrest when the antibody was injectedinto cells prior to onset of anaphase (49). Althoughsuggestive, these results are subject to alternative in-terpretations (1, 54) and may not reflect the actualfunction of MKLP1/CHO1 motors in vivo.

Recent genetic analyses of MKLP1/CHO1 orthologsfrom invertebrates (1, 54) demonstrate that this familyof motors is necessary for cytokinesis rather than mi-tosis. Mutations in the Drosophila gene pavarotti (en-coding PAV-KLP) and in the Caenorhabditis genezen-4 (encoding the ZEN-4 KLP) disrupt cytokinesiswithout perturbing anaphase B spindle elongation.Thus, if MKLP1/CHO1 motors are involved in spindleelongation, then this function is redundant and can beperformed by other KLPs.

Our studies of Mklp1 demonstrate clearly thatMKLP1/CHO1 motors are also required for cytokinesisin vertebrates. Expression of dominant-negative mu-tants of Mklp1 in zebrafish embryos produces a cyto-kinetic phenotype, i.e., cleavage arrest and the forma-tion of multinucleated blastomeres in the most extremecases, rather than a mitotic defect. The downstreamconsequences of perturbing Mklp1 function, which maybe direct or indirect, are also severe, with gross tissuedisorganization apparent in those embryos that reachthe late somitic stages (�18–24 h). We propose thatMKLP1/CHO1 motors are required for cytokinesis inall metazoan embryos and may also perform later de-

velopmental functions (e.g., dendritic differentiation;Refs. 62 and 76).

Proposed mechanism of Mklp1 function in cytokine-sis. How do motors of the CHO1/MKLP1 family partic-ipate in the initiation and progression of the cleavagefurrow? At least two general mechanisms, distinct butnot mutually exclusive, can be invoked to explain theircytokinetic function. First, these motors, by virtue oftheir ability to cross-link antiparallel microtubules,may maintain the integrity of the spindle midbody (1,54), hence providing a structural framework for deliv-ery of the “cleavage signal” to the cell cortex (5) or forassembly of the contractile ring (1, 19). Second, thesemotors appear ideally suited to transport molecules ofcritical importance to the cytokinetic process to thenascent cleavage furrow. Particularly important maybe the polo-like kinase, a serine/threonine kinase thatregulates mitotic spindle assembly and, in yeasts, cy-tokinesis (21). Polo-like kinase interacts in vivo withMKLP1/CHO1 and colocalizes with the motor throughmitosis and cytokinesis (42), suggesting that the motortransports the kinase to the spindle midzone where thelatter delivers a signal or signals necessary for furrowformation. Alternatively, MKLP1/CHO1 may delivermaternal stores of membrane via midbody or adjacentmicrotubule bundles to the site of membrane additionat the leading edge of the furrow (9). In zebrafish,transport of cadherin and �-catenin via intracellularmembranes to the furrow surface to mediate blas-tomere cohesion requires microtubules (31) and prob-ably KLPs. Other potential cargos for MKLP1/CHO1motors include components of the contractile ring com-plex (e.g., actin, anillin, septins; Ref 1).

We propose that Mklp1 serves a multifunctional rolein cytokinesis during zebrafish morphogenesis, both bystabilizing midbody microtubules and by transportingstructural components and/or signaling molecules tothe cleavage furrow. Our mechanistic understanding ofthese functions and the identification of molecularcomponents that interact with this motor are likely tobe accelerated by analysis of early acting developmen-tal mutations (e.g., the early arrest mutants; Ref. 33)obtained in recent large-scale mutagenic screens (14,23) in this fish. Given the substantial pool of mRNAsfor Mklp1 (this work) and other KLPs (M.-C. Chen andH. W. Detrich III, unpublished results) that are storedin the zebrafish egg, the design and implementation ofmaternal-effect screens targeted to motor-dependentprocesses should be a particularly promising strategyfor pursuing this goal.

We thank Dr. Leonard I. Zon for the gift of the zebrafish Mklp1-related cDNA mp8, Dr. Ryoko Kuriyama for kindly providing themonoclonal and polyclonal CHO1 antibodies, and Sandra Parker forhelp with the Western blots and the preparation of the figures.

This work was supported by National Science Foundation GrantsOPP-9420712, OPP-9815381, and OPP-0089451 (to H. W. Detrich).

Present address of M.-C. Chen: Academica Sinica, Taipei, Taiwan.

REFERENCES

1. Adams RR, Tavares AAM, Salzberg A, Bellen HJ, andGlover DM. pavarotti encodes a kinesin-like protein required to



organize the central spindle and contractile ring for cytokinesis.Genes Dev 12: 1483–1494, 1998.

2. Barbazuk WB, Korf I, Kadavi C, Heyen J, Tate S, Wun E,Bedell JA, McPherson JD, and Johnson SL. The syntenicrelationship of the zebrafish and human genomes. Genome Res10: 1351–1358, 2000.

3. Berliner E, Young EC, Anderson K, Mahtani HK, andGelles J. Failure of a single-headed kinesin to track parallel tomicrotubule protofilaments. Nature 373: 718–721, 1995.

4. Bloom GS and Endow SA. Motor proteins 1: kinesins. ProtProfile 2: 1109–1171, 1995.

5. Cao LG and Wang YL. Signals from the spindle midzone arerequired for the stimulation of cytokinesis in cultured epithelialcells. Mol Biol Cell 7: 225–232, 1996.

6. Chalfie M, Euskirchen G, Ward W, and Prasher DC. Greenfluorescent protein as a marker for gene expression. Science 263:802–805, 1994.

7. Chen MC and Detrich HW III. Kinesin-like microtubule mo-tors in early development. Methods Cell Biol 59: 227–250, 1999.

8. Currie PD. Zebrafish genetics: mutant cornucopia. Curr Biol 6:1548–1552, 1996.

9. Danilchik MV, Funk WC, Brown EE, and Larkin L. Re-quirement for microtubules in new membrane formation duringcytokinesis in Xenopus embryos. Dev Biol 194: 47–60, 1998.

10. Deavours BE and Walker RA. Nuclear localization of C-terminal domains of the kinesin-like protein MKLP-1. BiochemBiophys Res Commun 260: 605–608, 1999.

11. Detrich HW, III, Kieran MW, Chan FY, Barone LM, Yee K,Rundstadler JA, Pratt S, Ransom D, and Zon LI. Intraem-bryonic hematopoietic cell migration during vertebrate develop-ment. Proc Natl Acad Sci USA 92: 10713–10717, 1995.

12. Detrich HW, III, Westerfield M, and Zon LI. Overview of thezebrafish system. Methods Cell Biol 59: 3–10, 1999.

13. Dingwall C and Laskey R. Nuclear targeting sequences: aconsensus? Trends Biol Sci 16: 478–481, 1991.

14. Driever W, Solnica-Krezel L, Schier AF, Neuhauss SCF,Malicki J, Stemple DL, Stainier DYR, Zwartkruis F, Ab-delilah S, Rangini Z, Belak J, and Boggs C. A genetic screenfor mutations affecting embryogenesis in zebrafish. Development123: 37–46, 1996.

15. Eisen JS. Zebrafish make a big splash. Cell 87: 969–977, 1996.16. Eyal-Giladi H. Establishment of the axis in chordates: facts and

speculations. Development 124: 2285–2296, 1997.17. Gao D, Li Z, Murphy T, and Sauerbier W. Structure and

transcription of the gene for translation elongation factor 2subunit alpha of zebrafish (Danio rerio). Biochim Biophys Acta1350: 1–5, 1997.

18. Geisler R, Rauch GJ, Baier H, van Bebber F, Brobeta L,Dekens MP, Finger K, Fricke C, Gates MA, Geiger H,Geiger-Rudolph S, Gilmour D, Glaser S, Gnugge L, Ha-beck H, Hingst K, Holley S, Keenan J, Kirn A, Knaut H,Lashkari D, Maderspacher F, Martyn U, Neuhauss S, andHaffter P. A radiation hybrid map of the zebrafish genome. NatGenet 23: 86–89, 1999.

19. Giansanti MG, Bonaccorsi S, Williams B, Williams EV,Santolamazza C, Goldberg ML, and Gatti M. Cooperativeinteractions between the central spindle and the contractile ringduring Drosophila cytokinesis. Genes Dev 12: 396–410, 1998.

20. Gilbert SP, Webb MR, Brune M, and Johnson KA. Pathwayof processive ATP hydrolysis by kinesin. Nature 373: 671–676,1995.

21. Glover DM, Okhura H, and Tavares A. Polo kinase: thechoreographer of the mitotic stage? J Cell Biol 135: 1681–1684,1996.

22. Grunwald DJ. A fin-de-siecle achievement: charting new wa-ters in vertebrate biology. Science 274: 1634–1635, 1996.

23. Haffter P, Granato M, Brand M, Mullins MC, Hammer-schmidt M, Kane DC, Odenthal J, van Eeden FJM, JiangYJ, Heisenberg CP, Kelsh RN, Furutani-Seiki M, Vogel-sang E, Beuchle D, Schach U, Fabian C, and Nusslein-Volhard C. The identification of genes with unique and essen-tial functions in the development of the zebrafish, Danio rerio.Development 123: 1–36, 1996.

24. Hancock WO and Howard J. Processivity of the motor proteinkinesin requires two heads. J Cell Biol 140: 1395–1405, 1998.

25. Hauptmann G and Gerster T. Pou-2: a zebrafish gene activeduring cleavage stages and in the early hindbrain. Mech Dev 51:127–138, 1995.

26. Henikoff S. Unidirectional digestion with exonuclease III cre-ates targeted breakpoints for DNA sequencing. Gene 28: 351–359, 1984.

27. Herskowitz I. Functional inactivation of genes by dominantnegative mutations. Nature 329: 219–222, 1987.

28. Hirokawa N. Kinesin and dynein superfamily proteins and themechanism of organelle transport. Science 279: 519–526, 1998.

29. Holder N and McMahon A. Genes from zebrafish screens.Nature 384: 515–516, 1996.

30. Hyatt TM and Ekker SC. Vectors and techniques for ectopicgene expression in zebrafish. Methods Cell Biol 59: 117–126,1999.

31. Jesuthasan S. Furrow-associated microtubule arrays are re-quired for the cohesion of zebrafish blastomeres following cyto-kinesis. J Cell Sci 111: 3695–3703, 1998.

32. Jesuthasan S and Strahle U. Dynamic microtubules and spec-ification of the zebrafish embryonic axis. Curr Biol 7: 31–42,1996.

33. Kane DC, Maischein HM, Brand M, van Eeden FJM, Fu-rutani-Seiki M, Granato M, Haffter P, Hammerschmidt M,Heisenberg CP, Jiang YJ, Kelsh RN, Mullins MC,Odenthal J, Warga RM, and Nusslein-Volhard C. The ze-brafish early arrest mutants. Development 123: 57–66, 1996.

34. Kawamura K. Microdissection studies on the dividing neuro-blast of the grasshopper, with special reference to the mecha-nism of unequal cytokinesis. Exp Cell Res 106: 127–137, 1977.

35. Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, andSchilling TF. Stages of embryonic development of the zebrafish.Dev Dyn 203: 253–310, 1995.

36. Kull FJ, Sablin EP, Lau R, Fletterick RJ, and Vale RD.Crystal structure of the kinesin motor domain reveals a struc-tural similarity to myosin. Nature 380: 550–555, 1996.

37. Kuriyama R, Dragas-Granoic S, Maekawa T, Vassilev A,Khodjakov A, and Kobayashi H. Heterogeneity and microtu-bule interaction of the CHO1 antigen, a mitosis-specific kinesin-like protein. J Cell Sci 107: 3485–3499, 1994.

38. Kwok C, Critcher R, and Schmitt K. Construction and char-acterization of zebrafish whole genome radiation hybrids. Meth-ods Cell Biol 60: 287–302, 1999.

39. Kwok C, Korn RM, Davis ME, Burt DW, Critcher R, Mc-Carthy L, Paw BH, Zon LI, Goodfellow PN, and Schmitt K.Characterization of whole genome radiation hybrid mappingresources for non-mammalian vertebrates. Nucleic Acids Res 26:3562–3566, 1998.

40. Laemmli UK. Cleavage of structural proteins during the assem-bly of the head of bacteriophage T4. Nature 227: 680–685, 1970.

41. Lane HA and Nigg EA. Antibody microinjection reveals anessential role for human polo-like kinase 1 (Plk1) in the func-tional maturation of mitotic centrosomes. J Cell Biol 135: 1701–1713, 1996.

42. Lee KS, Yuan YL, Kuriyama R, and Erickson RL. Plk is anM-phase-specific protein kinase and interacts with a kinesin-likeprotein, CHO1/MKLP1. Mol Cell Biol 15: 7143–7151, 1995.

43. Llamazares S, Moreira A, Tavares A, Girdham C, SpruceBA, Gonzalez C, Karess RE, Glover DM, and Sunkel CE.Polo encodes a protein kinase homolog required for mitosis inDrosophila. Genes Dev 5: 2153–2165, 1991.

44. Miki H, Setou M, Kaneshiro K, and Hirokawa N. All kinesinsuperfamily protein, KIF, genes in mouse and human. Proc NatlAcad Sci USA 98: 7004–7011, 2001.

45. Moore JD and Endow SA. Kinesin proteins: a phylum ofmotors for microtubule-based motility. Bioessays 18: 207–219,1996.

46. Nakata T and Hirokawa N. Point mutation of adenosinetriphosphate-binding motif generated rigor kinesin that selec-tively blocks anterograde lysosome membrane transport. J CellBiol 131: 1039–1053, 1995.



47. Needleman SB and Wunsch CD. A general method applicableto the search for similarities in the amino acid sequence of twoproteins. J Mol Biol 48: 443–453, 1970.

48. Nislow C, Lombillo VA, Kuriyama R, and McIntosh JR. Aplus-end-directed motor enzyme that moves antiparallel micro-tubules in vitro localizes to the interzone of mitotic spindles.Nature 359: 543–547, 1992.

49. Nislow C, Sellito C, Kuriyama R, and McIntosh JR. Amonoclonal antibody to a mitotic microtubule-associated proteinblocks mitotic progression. J Cell Biol 111: 511–522, 1990.

50. Ohkura H, Hagan IM, and Glover DM. The conserved Schizo-saccharomyces pombe kinase, plo1, required to form a bipolarspindle, the actin ring, and septum, can drive septum formationin G1 and G2 cells. Genes Dev 9: 1059–1073, 1995.

51. Paw BH and Zon LI. Primary fibroblast cell culture. MethodsCell Biol 59: 39–43, 1999.

52. Postlethwait JH, Woods IG, Ngo-Hazelett P, Yan YL, KellyPD, Chu F, Huang H, Hill-Force A, and Talbot WS. Ze-brafish comparative genomics and the origins of vertebrate chro-mosomes. Genome Res 10: 1890–1902, 2000.

53. Postlethwait JH, Yan YL, Gates MA, Horne S, Amores A,Brownlie A, Donovan A, Egan ES, Force A, Gong Z, GoutelC, Fritz A, Kelsh R, Knapik E, Liao E, Paw B, Ransom D,Singer A, Thomson M, Abduljabbar TS, Yelick P, Beier D,Joly JS, Larhammar D, and Rosa F. Vertebrate genomeevolution and the zebrafish gene map. Nat Genet 18: 345–349,1998.

54. Raich WB, Moran AN, Rothman JH, and Hardin J. Cytoki-nesis and midzone microtubule organization in Caenorhabditiselegans require the kinesin-like protein ZEN-4. Mol Biol Cell 9:2037–2049, 1998.

55. Rappaport R. Establishment of the mechanism of cytokinesisin animal cells. Int Rev Cytol 105: 245–281, 1986.

56. Rappaport R and Rappaport BN. Establishment of cleavagefurrows by the mitotic spindle. J Exp Zool 189: 189–196, 1974.

57. Robb DL, Heasman J, Raats J, and Wylie C. A kinesin-likeprotein is required for germ plasm aggregation in Xenopus. Cell87: 823–831, 1996.

58. Sambrook J, Fritsch EF, and Maniatis T. Molecular Cloning:A Laboratory Manual. Cold Spring Harbor, NY: Cold SpringHarbor Laboratory Press, 1989.

59. Sanger F, Nicklen S, and Coulson AR. DNA sequencing withchain-terminating inhibitors. Proc Natl Acad Sci USA 74: 5463–5467, 1977.

60. Savoian MS, Earnshaw WC, Khodjakov A, and Rieder CL.Cleavage furrows formed between centrosomes lacking an inter-vening spindle and chromosomes contain microtubule bundles,INCENP, and CHO1 but not CENP-E. Mol Biol Cell 10: 297–311, 1999.

61. Sellito C and Kuriyama R. Distribution of a matrix componentof the midbody during the cell cycle in Chinese hamster ovarycells. J Cell Biol 106: 431–439, 1988.

62. Sharp DJ, Yu W, Ferhat L, Kuriyama R, Rueger DD, andBaas PW. Identification of a microtubule-associated motor pro-

tein essential for dendritic differentiation. J Cell Biol 138: 833–843, 1997.

63. Solnica-Krezel L and Driever W. Microtubule arrays of thezebrafish yolk cell: organization and function during epiboly.Development 120: 2443–2455, 1994.

64. Stewart EA, McKusick KB, Aggarwal A, Bajorek E, BradyS, Chu A, Fang N, Hadley D, Harris M, Hussain S, Lee R,Maratukulam A, O’Connor K, Perkins S, Piercy M, Qin F,Reif T, Sanders C, She X, Sun WL, Tabar P, Voyticky S,Cowles S, Fan JB, Mader C, Quackenbush J, Myers RM,and Cox DR. An STS-based radiation hybrid map of the humangenome. Genome Res 7: 422–433, 1997.

65. Strahle U and Jesuthasan S. Ultraviolet radiation impairsepiboly in zebrafish embryos: evidence for a microtubule-depen-dent mechanism of epiboly. Development 119: 909–919, 1993.

66. Thompson MA, Ransom DG, Pratt SJ, MacLennan H, Ki-eran MW, Detrich HW, III, Vail B, Huber TL, Paw B,Brownlie AJ, Oates AC, Fritz A, Gates MA, Amores A,Bahary N, Talbot WS, Her H, Beier DR, Postlethwait JH,and Zon LI. The cloche and spadetail genes differentially affecthematopoiesis and vasculogenesis. Dev Biol 197: 248–269, 1998.

67. Towbin H, Staehelin T, and Gordon J. Electrophoretic trans-fer of proteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. Proc Natl Acad Sci USA 76:4350–4354, 1979.

68. Vale RD. Switches, latches, and amplifiers: common themes ofG proteins and molecular motors. J Cell Biol 135: 291–302, 1996.

69. Wagner MC, Pfister KK, Brady ST, and Bloom GS. Purifi-cation of kinesin from bovine brain and assay of microtubule-stimulated ATPase activity. Methods Enzymol 196: 157–175,1991.

70. Westerfield M. The Zebrafish Book. Eugene, OR: University ofOregon Press, 1995, p. 1.1–11.78.

71. Wheatley SP and Wang Y. Midzone microtubule bundles arecontinuously required for cytokinesis in cultured epithelial cells.J Cell Biol 135: 981–989, 1996.

72. Williams BC, Riedy MF, Williams EV, Gatti M, and Gold-berg ML. The Drosophila kinesin-like protein KLP3A is a mid-body component required for central spindle assembly and ini-tiation of cytokinesis. J Cell Biol 129: 709–723, 1995.

73. Woods IG, Kelly PD, Chu F, Ngo-Hazelett P, Yan YL,Huang H, Postlethwait JH, and Talbot WS. A comparativemap of the zebrafish genome. Genome Res 10: 1903–1914, 2000.

74. Yen TJ, Compton DA, Wise D, Zinkowski RP, Brinkley BR,Earnshaw WC, and Cleveland DW. CENP-E, a novel humancentromere-associated protein required for progression frommetaphase to anaphase. EMBO J 10: 1245–1254, 1991.

75. Young EC, Mahtani HK, and Gelles J. One-headed kinesinderivatives move by a nonprocessive, low-duty ratio mechanismunlike that of two-headed kinesin. Biochemistry 37: 3467–3479,1998.

76. Yu W, Sharp DJ, Kuriyama R, Mallik P, and Baas PW.Inhibition of a mitotic motor compromises the formation ofdendrite-like processes from neuroblastoma cells. J Cell Biol136: 659–668, 1997.



zebrafish mitotic kinesin-like protein 1 (mklp1) functions ... · t8–2, t11–3, and t11–8)...

Documents