Proc. Natl. Acad. Sci. USAVol. 94, pp. 9654–9659, September 1997Cell Biology

Identification and classification of 16 new kinesin superfamily(KIF) proteins in mouse genome

TERUNAGA NAKAGAWA, YOSUKE TANAKA, EIJI MATSUOKA, SATORU KONDO, YASUSHI OKADA, YASUKO NODA,YOSHIMITSU KANAI, AND NOBUTAKA HIROKAWA*Department of Cell Biology and Anatomy, Graduate School of Medicine, University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113, Japan

Communicated by Setsuro Ebashi, National Institute for Physiological Sciences, Okazaki, Japan, June 30, 1997 (received for review March 25,1997)

ABSTRACT KIF (kinesin superfamily) proteins are mi-crotubule-dependent molecular motors that play importantroles in intracellular transport and cell division. The extent towhich KIFs are involved in various transporting phenomena,as well as their regulation mechanism, are unknown. Theidentification of 16 new KIFs in this report doubles theexisting number of KIFs known in the mouse. Conservednucleotide sequences in the motor domain were amplified byPCR using cDNAs of mouse nervous tissue, kidney, and smallintestine as templates. The new KIFs were studied with respectto their expression patterns in different tissues, chromosomallocation, and molecular evolution. Our results suggest that (i)there is no apparent tendency among related subclasses ofKIFs of cosegregation in chromosomal mapping, and (ii)according to their tissue distribution patterns, KIFs can bedivided into two classes–i.e., ubiquitous and specific tissue-dominant. Further characterization of KIFs may elucidateunknown fundamental phenomena underlying intracellulartransport. Finally, we propose a straightforward nomencla-ture system for the members of the mouse kinesin superfamily.

The dynamic aspect of intracellular organelle transport andcell division involve proper functioning of microtubule-dependent molecular motors such as kinesin and dyneinsuperfamily proteins. Since the biochemical identification (1)and cDNA cloning (2) of kinesin heavy chain, cDNAs ofkinesin superfamily genes were genetically identified fromvarious species (3–6). The deduced amino acid sequences ofthese members revealed the presence of a motor domain ofhigh homology, and a more divergent nonmotor domain. Thelatter comprises a large portion of the whole molecule, withamino acid sequences that are characteristic for each members.These findings support the observation by quick-freeze, deep-etch electron microscopy techniques that in nerve axons, manydifferent types of crossbridge structures are observed betweencellular organelles and microtubules (7, 8). Some of thesecrossbridges were later identified to be kinesin (9). These andother findings suggest that the machinery for intracellulartransport is highly sophisticated, and that a large family ofgenes encoding the kinesin superfamily is to be expected.

Molecular biological approaches were instrumental in iden-tifying members of the kinesin superfamily. Several amino acidmotifs in the motor domain, such as IFAYGQT, DLAGSE,and HIPYR, are highly conserved among species. Fragmentsflanked by these sequences can therefore be amplified by PCRusing degenerate oligonucleotide primers. This strategy iden-tified many kinesin superfamily members in various organisms,such as Drosophila (10, 11), mouse (12, 13), Xenopus (14), fish(15), yeast (3), and rat (16). The reason for the diversity of the

kinesin superfamily remains elusive, but it appears that dif-ferent members are assigned specific cellular functions. In themouse there are neuron-specific members such as KIF5A (12),nKHC (17), KIF1A (18), and KIFC2 (13) that are apparentlynot involved in cell division. [Note that Aizawa et al. (12)originally described KIF5. However, further careful studiesrevealed that KIF5 consists of three members that are referredto as KIF5A, -5B, and -5C in this paper. Henceforth, the nameKIF5A is used for the sequence referred to as KIF5 in Aizawaet al.]

For neurons, our data suggest that certain classes of mem-branous organelles are transported separately down the axonby different motors. For example, KIF1A is associated withorganelles that contain synaptotagmin, synaptophysin, andRab3A, but not SV2, syntaxin 1A, or SNAP-25. Motors suchas conventional kinesin or KIF3 were not associated withKIF1A-containing organelle (18). This evidence suggests thatdifferent members of the KIF carry different cargos, eventhough the possibility of functional redundancy still remains.In neurons, at least, motor molecules other than KIF1A mayexist that transport vesicles containing SV2, syntaxin 1A, andSNAP-25. Thus with our present knowledge of KIF we may beunable to obtain the overall view of the complex intracellulartransport system. These and other considerations stronglymotivated us to search for the additional members of the KIF.Initially, based on PCR approach, our group has identifiedseven KIFs—i.e., KIF1A (18), KIF1B (19), KIF2 (20), KIF3A(21), KIF3B (22), KIF4 (23), and KIF5A (ref. 12; see notes inthe previous paragraph) in mouse brain. Three more KIFswere identified by another PCR strategy (13). In the presentstudy, the search for additional KIFs was intensified further byan improved PCR strategy. We identified 16 new members ofKIF, virtually doubling the number of known KIFs in themouse. In this report the partial amino acid sequences of thesenew KIFs are presented with the classification based onmolecular evolutionary analysis and tissue distribution. Wealso present the chromosomal mapping for some of the knownKIFs. Finally, we propose the use of a simplified nomenclatureas for the mouse kinesin superfamily.

MATERIALS AND METHODS

Isolation of mRNA. Mice were killed to obtain the followingmaterial: 0-day-old whole brain, 4-week-old hippocampus,4-week-old olfactory bulb, 4-week-old kidney, and 4-week-oldintestine. Each tissue was dissected from separate mice to

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked ‘‘advertisement’’ inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

© 1997 by The National Academy of Sciences 0027-8424y97y949654-6$2.00y0PNAS is available online at http:yywww.pnas.org.

Data deposition: The sequences reported in this paper have beendeposited in the GenBank database [accession nos. AB001456(KIF1C), AB001433 (KIF3C), AB001434 (KIF6), AB001435 (KIF7),AB001436 (KIF8), AB001437 (KIF9), AB001426 (KIF10), AB001427(KIF11), AB001428 (KIF12), AB001429 (KIF13A), AB001430(KIF13B), AB001431 (KIF14), AB001432 (KIF15), AB001425(KIF16A), AB001423 (KIF16B), AB001424 (KIF17), AB001457(KIFC3)].*To whom reprint requests should be addressed.

9654

maintain freshness. The rest of the purification procedures areperformed as described (24).

Synthesis of First-Strand cDNA. Two hundred units ofSuper Script II (GIBCOyBRL) reverse transcriptase was usedto transcribe 50–500 ng of poly(A)1 RNA per 20 ml reactionby an oligo(dT) primer (45°C for 50 min). Other reactionconditions follow the manufacturer’s protocols.

PCRs to Identify New KIFs. The oligonucleotide primersused in this study are listed in Table 1. The primers weredesigned according to the conserved amino acid sequencesIFAYGQT and VDLAGSE in the motor domain. For allPCRs, AmpliTaq DNA polymerase (Perkin–Elmer) and Ge-neAmp PCR system 9600 (Perkin–Elmer) were used. PrimersIFAYGQT and DLAGSE were used to amplify 0-day-oldmouse whole brain first-strand cDNA; the reaction programwas as follows: 30 cycles at 94°C for 1 min, at 56°C for 1 min,and at 72°C for 1 min. The reaction mixture contained 0.5 mlof cDNA solution (directly used from cDNA synthesis afterheat inactivation), 1 ml of 10 mM dNTP, 5 units of AmpliTaqDNA polymerase, 100 pmol of each primer, 10 mM TriszHCl(pH 8.3) (at 25°C), 50 mM KCl, 1.5 mM MgCl2, and 0.001%(wtyvol) gelatin, in a total volume of 50 ml. Multiple bandsappeared when separated by agarose gel electrophoresis. Formost of the known kinesin superfamilies, the nucleotidesequences between IFAYGQT and VDLAGSE correspond to'500 bp. Thus, bands around 500 bp were cut out and theDNAs were recovered by Gene CleanII (Bio 101). Purifiedfragments were treated with T4 DNA polymerase and sub-cloned into EcoRV site of pBlueScript (Stratagene). Sequenc-ing was done by Applied Biosystems auto sequencer models373 and 377. For the other template cDNAs the remainingprimers were used for amplification. PCR schedule was asfollows: 96°C for 45 sec, 55°C for 4 min, and 72°C for 3 min with6-sec extension per cycle for 35–40 cycles (25). PrimersIFAY-RI, IFAY-BAM, and IFAY-CLA contain EcoRI,BamHI, and ClaI sites, respectively. These sites were used tosubclone the amplified fragments into pBlueScript. Calf intes-tine alkaline phosphatase (Toyobo, Osaka) treatment of thesubcloning vectors was done with less than 1 mg of DNA in 50ml volume. The buffer used for the reaction was 50 mMTriszHCl (pH 9.0), 1 mM MgCl2, 0.1 mM ZnCl2, and 1 mMSpermidine (Sigma). Three units of enzyme was added to startthe reaction (1 h at 37°C). An additional three units of enzymewas added at the middle of the reaction. Competent cells wereprepared with DH5a bacteria as described (26).

Northern Blot Analysis. Total RNA was purified as de-scribed (24) from brain, lung, heart, liver, spleen, kidney,intestine, and testis tissue of 4-week-old mice. Twenty micro-grams of total RNA was loaded on each lane of a 1% agarosegel. The concentration of RNAs was measured by light absor-bance as described (27). Probes were labeled with T7 Quick

prime kit (Pharmacia). Quick Hyb (Stratagene) was used forhybridization according to manufacturer’s protocol. The washschedule was 23 SSCy0.1% SDS at 65°C for 10 min (threetimes). For some probes additional wash in 0.13 SSCy0.1%SDS for 10 min was needed to eliminate the cross-hybridization signal.

Sequence Alignment and Molecular Evolutional Analysis. Aphylogenetic tree of the motor domain sequences of the KIFwas generated using the CLUSTALW program (28) based on theneighbor joining (NJ) method (29). The confidence limits onthe tree were calculated by performing bootstrap resampling1,000 times. The region between IFAYGQT and LAGSE wasused for the alignment. However, for some of the newlyidentified KIFs, only a partial sequence of the correspondingdomain was obtained due to the cloning artefacts. The accu-racy of the tree obtained was confirmed by comparing the treesdeduced by the NJ method and the maximum parsimony (MP)method. MP method was performed according to the de-scribed procedures (30). Sequences other than the newlyidentified KIFs were obtained from the GenBank database.The tree was drawn using the software TREEVIEW (31).

Cytogenetic Mapping. A l EMBL3 library of genomic DNAderived from the mouse embryonic stem cell line J1 wasscreened with cDNA probes containing the motor domains ofKIF1A, -1B, -2, -3A, -3B, -5A, -5B, -5C, and -C2 by standardmethods (27). Overlapping clones of about 20 kb long wereisolated. Clones were proved to contain the correct genes bypartial sequencing. For fluorescence in situ hybridizationmapping standard methods were used (32–34) [see DNABiotech (Ontario, Canada) and Nippon Gene (Sendai, Ja-pan)].

RESULTS

Identification of New KIFs. Fig. 1 shows the alignment of theamino acid sequences of the PCR products of the new KIFsidentified together with the previously identified ones (12, 29).The identification of the new KIFs are summarized in Table 2.

Tissue Distribution Patterns of the Newly Identified KIFs.Some of the new KIFs turned out to be homologues ofpreviously identified and characterized kinesin related pro-teins in other species. Fig. 2 shows the tissue distributionpatterns of the transcripts of some of the new KIFs asdetermined by Northern blot analysis. Twenty micrograms oftotal RNA was loaded per lane unless otherwise specified.KIF5C turned out to be brain specific. The transcript of KIF1C(Fig. 3A) was ubiquitously distributed and 4 bands weredetected, suggesting alternative splicing or multiple polyade-nylaytion sites. Two micrograms of poly(A)1 RNA from4-week-old mouse brain was blotted against KIF1C probe inFig. 3B. Four bands appear clearly in this Northern blot even

Table 1. Degenerate oligonucleotide primers used in this study

Primer Sequence

IFAYGQT 59-AT(TyCyA),TT(TyC),GCI,TA(TyC),GGI,CA(AyG),AC-39DLAGSE 59-(CyT)TC,I(AyC)(AyT),ICC,IGC,IAG,(AyG)TC-39IFAY 59-AT(AyTyC),TT(CyT),(AyG)CI,TA(CyT),GGI,CA(AyG),AC-39IFAY-RI 59-GGG,AAT,TCA,T(AyTyC)T,T(CyT)(AyG),CIT,A(CyT)G,GIC,A(AyG)A,C-39IFAY-BAM 59-CGG,GAT,CCA,T(AyTyC)T,T(CyT)(AyG),CIT,A(CyT)G,GIC,A(AyG)A,C-39IFAY-CLA 59-CCA,TCG,ATA,T(AyTyC)T,T(CyT)(AyG),CIT,A(CyT)G,GIC,A(AyG)A,C-39LAGSE1 59-CTC,(AyG)CT,ICC,IGC,(CyT)A(AyG),(AyG)TC,IA-39LAGSE2 59-CTC,(AyG)CT,ICC,IGC,(AyG)AG,(AyG)TC,IA-3LAGSE3 59-CTC,IGA,ICC,IGC,(CyT)A(AyG),(AyG)TC,IA-39LAGSE4 59-CTC,IGA,ICC,IGC,(AyG)AG,(AyG)TC,IA-39LAGSE5 59-TTC,(AyG)CT,ICC,IGC,(CyT)A(AyG),(AyG)TC,IA-39LAGSE6 59-TTC,(AyG)CT,ICC,IGC,(AyG)AG,(AyG)TC,IA-39LAGSE7 59-TTC,IGA,ICC,IGC,(CyT)A(AyG),(AyG)TC,IA-39LAGSE8 59-TTC,IGA,ICC,IGC,(AyG)AG,(AyG)TC,IA-39

Cell Biology: Nakagawa et al. Proc. Natl. Acad. Sci. USA 94 (1997) 9655

after harsh wash condition. In Fig. 3C developmental changesof the KIF1C transcript is shown by applying 2 mg of brainpoly(A)1 RNA for each developmental stage. Northern blot-ting revealed KIF6 and KIF7 (data not shown) to be testis-specific. KIF8 was a mouse homologue of yeast KIP1 (37).KIF9 was testis-dominant with less signals detected from brain,lung, and kidney. KIF10 was a mouse homologue of CENP-E(38). KIF11 was a mouse homologue of Xenopus Eg5 (39).KIF12 was kidney-dominant. KIF13A transcript was overallubiquitous but dominant in brain, lung, heart, kidney, andtestis. The transcript of KIF13B was distributed ubiquitously.KIF14 transcript was detected in brain and kidney. KIF15transcript was dominantly detected in spleen and testis.KIF16A was overall ubiquitous but dominant in lung, heart,kidney, and testis. KIF16B was lung-dominant.

As seen from the Northern blot results, the source of thecDNA material does not necessarily correlate with transcriptdistribution pattern. For example, most of the testis-dominantKIFs were identified with nervous tissue as a starting material.Our results suggest that we can roughly classify the KIFs bytheir transcript distribution patterns into two classes—i.e.,ubiquitous and tissue-dominant. KIF1C, -13B, and -16A arethe ubiquitously distributed, whereas the others (KIF5C, -6, -9,-12, -13A, -14, -15, and -16B) have the biased tissue distribu-tion.

Although the washing procedure of the Northern blotmembranes were done with care, the possibility of cross-hybridization still remains. However, considering the differ-ence of the molecular weight of each transcript, the cross-hybridization may not be between the different families.

Two Brain-Specific Kinesin Heavy Chains Exist in theMouse. In the previous search, we identified KIF5 as abrain-specific type of conventional kinesin heavy chain (12).Another kinesin heavy chain identified by Kato (35) wassimilar to both KIF5 (12) and to ubiquitous kinesin (36). Thusto avoid confusion in nomenclature, we decided to use thename KIF5A for the sequence referred to as KIF5 in Aizawaet al. (12), KIF5B for the ubiquitous kinesin heavy chain, and

KIF5C for the molecule identified by Kato (35). Using cDNAfragment of the motor domain kindly provided by K. Kato, wefound that this was also specifically expressed in the brain (Fig.2). Thus there are three different kinesin heavy chains inmouse brain.

Phylogenetic Analysis of Known Kinesin Superfamily Pro-teins. To classify the newly identified KIFs, molecular evolu-tionary analysis was performed. Phylogenetic trees were ob-tained by both neighbor joining (Fig. 4A) (29) and maximumparsimony methods (Fig. 4B) (40). Both trees are unrooted. Bycomparing the two trees, some of the branching patterns of theperipheral branches are nearly identical. For example, thebranching patterns of the KIF1, KIF13, and KIF16 families areidentical in both methods. In other cases differences are seen.For example, KIF14 locates in a completely different branch.The reliability of each branch point was evaluated by calcu-lating the bootstrap values.

These results allow us to assess whether a newly identifiedKIF is a homologue of previously identified motors in otherorganisms. For example, KIF11 is highly homologous to Eg5 inXenopus. However, careful interpretation is needed. First, onlythose that are shown to be evolutionarily close by the twotree-generating methods may be considered to be highlyhomologous. Second, even if two genes from different speciesappear evolutionarily close in both trees, there still remains thepossibility that the two are not species homologues but ratherorthologues. One good example is seen in the branchingpatterns for KIF13A, KIF13B, and CELF56E3 (a sequenceobtained from the Caenorhabditis elegans genome project).KIF13A can be a species homologue of CELF56E3 just aslikely as KIF13B. Thus, the information obtained from thephylogeny tree should be interpreted with caution. Full cDNAsequences of each of the molecules are required to finallydetermine whether two molecules are species homologues ornot. In any case, by neglecting the details of molecularevolution we were able to name the newly identified molecules.The numbers were given according to the chronological ordersof identification. However, some of the names were laterchanged when we noticed that the molecule can be consideredto consist a subfamily with another molecule. Thus, thenumbering itself has no more meaning than each molecules aredifferent.

Chromosomal Mapping of the KIFs and KIFCs. To examinethe distribution of the KIF genes on the chromosome, weperformed f luorescence in situ hybridization using thegenomic sequences as probes. The results are shown in Fig. 5.There seems to be little relation between closely related KIFsand their chromosomal localization. For example KIF1A and1B, KIF3A and -3B, KIF5A, -5B, and -5C, respectively, areclosely related members. However, their chromosomal loca-

FIG. 1. Amino acid sequence of the KIF motor domains. Sequences between IFAYGQT and DLAGSE, the highly conserved kinesin motordomain motif, were aligned using the program CLUSTALW. Arrows indicate the positions of the primers. Asterisks are identical amino acids. Dotsare similar amino acids. For some new KIFs only a partial sequence within this domain is shown. This is due to the cloning artefact. The rest ofthe region was either lost by the restriction enzyme digestion or appeared as a hybrid molecule with another KIF which may be caused by lowstringency PCR. Various cDNA templates were obtained from whole brain, hippocampus, olfatory bulb, kidney, and intestine of 4-week-old mice.

Table 2. Summary of searching the new KIFs

Library

No. ofclones

sequenced Identification

0 day whole brain 83 KIF3C, KIF1C, KIF8, KIF114 week hippocampus 50 KIF6, KIF7, KIF9, KIF104 week olfactory bulb 100 KIF16B4 week kidney 200 KIF12, KIF13A, KIF13B, KIF14,

KIF16A, KIF174 week intestine 100 KIF15

9656 Cell Biology: Nakagawa et al. Proc. Natl. Acad. Sci. USA 94 (1997)

tions differ and no apparent segregation seems to exist. Thesedata implies that the evolution of the KIF genes may bedifferent from those genes that are clustered in certain regionsof the genome, such as globin genes.

DISCUSSIONThe improved PCR strategy lead to the identification of 16 newKIFs in the mouse, virtually doubling the number of the known

mouse KIF molecules. The total number of the known KIFs inmouse has now reached more than 30. Although 3 (KIF8,KIF10, and KIF11) are likely to be the homologues of thepreviously characterized genes in other species, still 13 are new.In a PCR with a mixture of degenerate oligonucleotide prim-ers, the outcome of the amplified cDNA fragment is stronglybiased by the annealing efficiency between the template andthe primers. Thus, the probability of discovering the new KIFsdoes not necessarily correlate with the level of the transcriptof that particular KIF in the cDNA template. Our findings doesnot give any information about the actual size of the KIF familyin the mouse genome, and there are reasons to believe theremay be even more unknown KIFs.

What Are the Functions of the New KIFs? When focusing onmicrotubule-dependent intracellular transport, highly differ-entiated cells such as neurons and epithelial cells are goodmodel systems to study. However, can we actually distinguishwhether the new KIFs are for intracellular transport or for celldivision on the basis of the information we have at thismoment? Even if the two molecules from different species areshown to be evolutionarily very close, it does not necessarilymean that the two are the species homologues. This is becausethe phylogenetic analysis performed by us uses only partialamino acid sequences of the motor domain.

From our previous study, KIF1A and KIF1B are highlyhomologous in the region that covers the entire motor domain,but the cargos they carry are completely different. However,there seems to be some relationship between molecular evo-lution and function. For example, the branching pattern in thephylogenetic tree deduced by two different methods matchesfor KIF8, KIP1, CIN8, BimC, Eg5, and Cut7. This is the socalled the BimCyEg5 family (8), and most of the molecules inthis class are concerned in cell division. The branching patternsfor KIF1A, -1B, -1C, unc104, CELF56E3, KIF13A, -13B,KIF16A, and -16B are also same between the two differentanalysis method. If the relationship between molecular evo-lution and function found for the BimCyEg5 family holds truefor this family, the five new KIFs in this KIF1, -13, and -16branch may be involved in intracellular organelle transport.Because KIF12 and KIF14 are put in different branches in the

FIG. 2. Our first approach to elucidate the function of each new KIFs is to examine the tissue distribution. We performed Northern blot analysesfor the new KIFs that have almost no functional implication from the updated database. Twenty micrograms of total RNA from various mousetissues (4 week old) was loaded on each lanes. Probes used are from the PCR amplified and subcloned fragment of the motor domain listed inFig. 1. Controls were taken by probing the membrane with rat b-actin probe. The arrowhead with the number indicates the molecular weight inkilobases. The numbers in the left of each membrane is the name of the molecule (KIF is abbreviated). ACT is the beta actin control blot. Lanes:1, testis; 2, intestine; 3, kidney; 4, spleen; 5, liver; 6, heart; 7, lung; 8, brain.

FIG. 3. Northern blot analysis of KIF1C. Twenty micrograms oftotal RNA from various mouse tissues (4 week old) was loaded on eachlanes. Probes used are from the PCR amplified and subclonedfragment of the motor domain. KIF1C appears as four bands in theNorthern blot. (A) Tissue distribution of KIF1C transcripts. (B) Twomicrograms of mRNA from 4-week-old mouse brain was resolved andblotted likewise. The molecular weights are marked in the right sidein kb. (C) Two micrograms of mRNA was resolved to examine thedevelopmental transcription pattern of KIF1C in brain. Br, brain; Lu,lung; H, heart; Lv, liver, Sp, spleen; K, kidney; Int, intestine; Tes, testis.

Cell Biology: Nakagawa et al. Proc. Natl. Acad. Sci. USA 94 (1997) 9657

two trees, we interpret these are unclassified for the moment.KIF12 is dominantly expressed in kidney and KIF14 in brainand kidney. Taking into account the tissue distribution pat-terns, the evidence is enough to encourage us to furthercharacterize these two molecules too.

What Can Be Said from the Tissue Distribution NorthernBlotting Data? We classified the newly identified KIFs intoubiquitous and tissue-dominant types according to the North-ern blotting data. In general, one might suggest that moleculesthat are ubiquitously expressed in the entire organism mayfunction in some sort of fundamental processes of cells ingeneral. KIF1B, for example, is a ubiquitous molecule andfunctions as a transporter of mitochondria (19). Mitochondriaare ubiquitous organelle, which is consistent with the distri-bution of KIF1B. However, cell division is also a processobserved in all cells except certain highly differentiated cells,such as neurons. In new born animals, when the actual numberof cells in the body is increasing, a certain population of almostevery tissue should be undergoing cell division. But in rela-tively grown up animals cell division events should be limitedto specific area of the organism, such as testes, lymphaticorgans (spleen), and certain epithelial tissues (alimentary duct,dermis, etc.). Our materials for the Northern blot analyseswere obtained from 4-week-old mice. By the criteria above,KIF15 may be involved in cell division. KIF6, -7, and -9 aredominantly transcribed in testes compared with other tissues,but it may be premature to conclude that they are involved incell division. We believe the situation in testes is complicted.In this tissue active spermatogenesis (involving cell division)

take place in a vectorial manner from the basement to theapical lumen (directional intracellular transport). However,the sperm cells thus formed also have active flagella whichinclude a lot of KIFs and dyneins. Thus it is difficult to sayanything at this moment about the function of KIF6, -7, and -9.

On the other hand, the KIFs that are neither testes- norspleen-dominant, such as KIF12, -13A, -14, and -16B, arepotentially interesting as motors for specific intracellular trans-ports.

What Can We Learn from the New KIFs. Much of thestudies aimed toward intracellular transport are done in neu-rons and polarized epithelial cell systems. The involvement ofKIFs in axonal transport, dendritic transport, and intracellulartrafficking in epithelial cells remain elusive. The functionalspecificity of each KIF is believed to reside in the nonmotordomain of the molecule. This belief is based on the evidencethat the amino acid sequences of the nonmotor domain arequite distinct between different members. Initially, the onlytools to further devide the nonmotor domains into differentsubdomains were the prediction of three-dimensional struc-ture by analysis softwares based on the primary amino acidsequence such as the coiled–coil algorithm (41, 42). However,as more and more information on full-length cDNA sequenceaccumulated, subdomains in the nonmotor domains becameclear for some of the new KIFs. One example is the AF6ycnodomain (43). The function of this domain is unknown, but thedomain exists in the region flanking the motor domain towardthe carboxyl terminal in KIF1A and -1B. The question iswhether this domain exists in KIF1C. The full cDNA sequence

FIG. 4. Phylogenetic tree was constructed using the new sequence data of the KIFs listed in Fig. 1. Other sequences were available from thedatabase and added to generate the tree. The amino acid sequence between IFAYGQT and DLAGSE was used to align the sequences. The treeis unrooted. We confirmed the accuracy of the resultant phylogenetic tree by comparing the trees deduced via the neighbor joining (NJ) method(A) and maximum parsimony (MP) method (B). In A, branching that gives bootstrap values higher than 90 are marked. In B, bootstrap values arelisted in the tree. The GenBank accession numbers of the sequence data used are as follows: KIF1A, D29951; KIF1B, D17577; KIF1C, AB001456;KIF2, D12644; KIF3A, D1264; KIF3B, D26077; KIF3C, AB001433; KIF4, D12646; KIF5A, C44259; KIF5B(MMuKHC), L27153;KIF5C(MMnKHC), X61435; KIF6, AB001434; KIF7, AB001435; KIF8, AB001436; KIF9, AB001437; KIF10, AB001426; KIF11, AB001427; KIF12,AB001428; KIF13A, AB001429; KIF13B, AB001430; KIF14, AB001431; KIF15, AB001432; KIF16A, AB001425; KIF16B, AB001423; KIF17,AB001424; KIFC1, D49545; KIFC2, D49544; KIFC3, AB001457; unc104, M58582; CELF56E3, U41536; CHO1, X67155; CHO2, X83576; Kid,D38751; SMY1, M69021; nod, M94188; MCAK, U11790; XKCM1, U36485, U36486; ncd, X57475; KAR3, M31719; Eg5, X54002; Cut7, X57513;BimC, M32075; Cin8, M90522; KIP1, Z11962; CENP-E, Z15005; XKLP1, X82012; Chromokinesin(Chromokine), U18309; NKIN, L47106; unc116,L19120; DmKHC, M24441; HsnKHC, U06698; HsuKHC, X65873; KLP68D, U15974; FLA10, L33697; KRP85(KinII), L16993; KRP95(KinII),U00996.

9658 Cell Biology: Nakagawa et al. Proc. Natl. Acad. Sci. USA 94 (1997)

will give an answer. In other KIFs as well subdomains that areshared within the subfamilies of the new KIFs may be revealed.

Nomenclature of the KIF. The growing number of newlyidentified KIF proteins may lead to some confusion becausedifferent nomenclatures are used in different species. Wepropose to use the following rule: (i) the term KIF is used forthe names of the mouse kinesin superfamily genes; (ii) indi-vidual members of the superfamily in mouse are termed KIFfollowed by a number such as KIF2, KIF4, and so on; (iii) acapital letter is added after the number when there aresubfamilies such as, KIF1A, KIF1B, and so on; and (iv) a ‘‘C’’is added before the number when the molecule has its motordomain in the carboxyl-terminal half, such as KIFC1, KIFC2,and so on. In this way, we think the nomenclature will helppeople both in and outside the field of molecular motors.

We thank Hiroto Okayama for helpful discussion, and MasaeSugaya, Hiromi Sato, and Haruyo Fukuda for their technical andsecretarial assistance. This study was supported by a grant for Centerof Excellence (COE) from the Ministry of Education, Science andCulture to N.H. T.N. is supported by the Japan Society for thePromotion of Science Research Fellowship for Young Scientists.

1. Vale, R. D., Reese, T. S. & Sheetz, M. S. (1985) Cell 42, 39–50.2. Yang, J. T., Saxton, W. M. & Goldstein, L. S. B. (1998) Proc. Natl.

Acad. Sci. USA 85, 1864–1868.3. Meluh, P. B. & Rose, M. D. (1990) Cell 60, 1029–1041.4. Enos, A. P. & Morris, N. R. (1990) Cell 60, 1019–1027.5. Zhang, P., Knowles, B. A., Goldstein, L. S. B. & Hawley, R. S.

(1990) Cell 62, 1053–1062.6. Hagan, I. & Yanagida, M. (1990) Nature (London) 347, 563–566.7. Hirokawa, N. (1982) J. Cell Biol. 94, 129–42.

8. Hirokawa. N. (1996) Trends Cell Biol. 6, 135–141.9. Hirokawa, N., Pfister, K. K., Yorifuji, H., Wagner, M. C., Brady,

S. T. & Bloom, G. S. (1989) Cell 56, 867–878.10. Endow, S. A. & Hatumi, M. (1991) Proc. Natl. Acad. Sci. USA 88,

4424–4427.11. Stewart, R. J., Pesavento, P. A., Woerpel, D. N. & Goldstein,

L. S. B. (1991) Proc. Natl. Acad. Sci. USA 88, 8470–8474.12. Aizawa, H., Sekine, Y., Takemura, R., Zhang, Z., Nangaku, M.

& Hirokawa, N. (1992) J. Cell Biol. 119, 1287–1296.13. Saito, N., Okada, Y., Noda, Y., Kinoshita, Y., Kondo, S. &

Hirokawa, N. (1997) Neuron 18, 425–438.14. Vernos, I., Heasman, J. & Wylie, C. (1993) Dev. Biol. 157,

232–239.15. King-Smith, C., Bost-Usinger, L. & Burnside, B. (1995) Cell

Motil. Cytoskeleton 31, 66–81.16. Sperry, A. O. & Zhao, L. (1996) Mol. Biol. Cell. 7, 298–305.17. Niclas, J., Navone, F., Hom-Booher, N. & Vale, R. (1994) Neuron

12, 1059–1072.18. Okada, Y., Yamazaki, H., Sekine-Aizawa, Y. & Hirokawa, N.

(1995) Cell 81, 769–780.19. Nangaku, M., Sato-Yoshitake, R., Okada, Y., Noda, Y., Take-

mura, R., Yamazaki, H. & Hirokawa, N. (1994) Cell 79, 1209–1220.

20. Noda, Y., Sato-Yoshitake, R., Kondo, S., Nangaku, M. & Hiro-kawa, N. (1995) J. Cell Biol. 129, 157–167.

21. Kondo, S., Sato-Yoshitake, R., Noda, Y., Aizawa, H., Nakata, T.,Matsuura, Y. & Hirokawa, N. (1994) J. Cell Biol. 125, 1095–1107.

22. Yamazaki, Nakata, T., H. Okada, Y. & Hirokawa, N. (1995) J.Cell Biol. 130, 1387–1399.

23. Sekine, Y., Okada, Y., Noda, Y., Kondo, S., Aizawa, H., Take-mura, R. & Hirokawa, N. (1994) J. Cell Biol. 127, 187–201.

24. Okayama, H., Kawaichi, M., Brownstein, M., Lee, F., Yokota, T.& Arai, K. (1987) Methods Enzymol. 154, 3–28.

25. Buck, L. & Axel, R. (1991) Cell 65, 175–187.26. Inoue, H., Nojima, H. & Okayama, H. (1990) Gene 96, 23–28.27. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular

Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press,Plainview, NY), 2nd Ed.

28. Thompson, J. D., Higgins, D. G. & Gibson T. J. (1994) NucleicAcids Res. 22, 4673–4680.

29. Saitou, N. & Nei, M. (1987) Mol. Biol. Evol. 4, 406–425.30. Tanaka, Y., Zhang, Z. & Hirokawa, N. (1995) J. Cell Sci. 108,

1883–1893.31. Page, R. D. M. (1996) Comput. Appl. Biosci. 12, 357–358.32. Heng, H. H. Q., Squire, J. & Tsui, L.-C. (1992) Proc. Natl. Acad.

Sci. USA 89, 9509–9513.33. Feng, G. S., Shen, R., Heng, H. H. Q., Tsui, L.-C., Kazlauskas,

A. & Pawson, T. (1994) Oncogene 9, 1745–1750.34. Heng, H. H. Q. & Tsui, L-C., (1993) Chromosoma 102, 325–332.35. Kato, K. (1991) Eur. J. Neurosci. 2, 704–711.36. Gudkov, A. V., Kazarov, A. R., Thimmapaya, R., Axenobich,

S. A., Mazo, I. A. & Roninson, I. B. (1994) Proc. Natl. Acad. Sci.USA 91, 3744–3748.

37. Roof, D. M., Meluh, P. B. & Rose, M. D. (1992) J. Cell Biol. 118,95–108.

38. Yen, T. J., Compton, D. A., Wise D., Zinkowski, R. P., Brinkley,B. R., Earnshaw, W. C. & Cleveland, D. W. (1991) EMBO J. 10,1245–1254.

39. Le Guellec, R., Paris, J., Couturier, A., Roghi, C. & Philippe, M.(1991) Mol. Cell. Biol. 11, 3395–3398.

40. Eck, R. V. & Dayhoff, M. O. (1966) Atlas of Protein Sequence andStructure (Natl. Biomed. Res. Found., Silver Spring, MD).

41. Lupas, A., Van Dyke, M. & Stock, J. (1991) Science 252,1162–1164.

42. Lupas, A. (1996) Trends Biochem. Sci. 21, 375–382.43. Ponting, C. P. (1995) Trends Biochem. Sci. 20, 265–266.

FIG. 5. Cytogenetic mapping were performed for some of theknown KIFs. (A and B) Examples of the fluorescence in situ hybrid-ization (FISH) mapping (probed for KIF5B). (A) The FISH signals onchromosome (arrow). (B) The same mitotic figure stained with49,6-diamidino-2-phenylindole (DAPI) to identify chromosome 13.(C) Rest of the results are summarized.

Cell Biology: Nakagawa et al. Proc. Natl. Acad. Sci. USA 94 (1997) 9659