INTRODUCTION

It is well documented that the Golgi apparatus typicallyoccupies a perinuclear position in cells at the centrosomalcomplex. This organizational scheme is highly conserved andis restored after the Golgi is disrupted or repositioned duringcentrosomal migration (Kupfer et al., 1982), cell division(Lucocq et al., 1989; Lucocq and Warren, 1987) and micro-tubule (Mt) depolymerization (Ho et al., 1989). Numerousauthors have described an intimate structural interactionbetween the Golgi and the Mt cytoskeleton (Kronebusch andSinger, 1987; Rogalski and Singer, 1984; Thyberg andMoskalewski, 1985; Turner and Tartakoff, 1989). Thesestudies have predicted that Mts may participate in the mainte-nance of Golgi structure and support vesicular traffickingpathways to and from this organelle. Because of its dynamiccapacity, it is likely that the association of the Golgi with Mtsis mediated, in part, by Mt-dependent motor enzymes. Recentfunctional studies have implicated the Mt-dependent enzyme,cytoplasmic dynein, in the positioning of the Golgi at the cen-trosome (Corthesy-Theulaz et al., 1992). While this function isconsistent with the retrograde activity of a dynein motor, it isnot known if dynein acts alone or with other motor enzymesto support the organization of the Golgi apparatus. Indeed, itis difficult to explain how dynein could support the extension

of tubulovesicular processes of the

trans-Golgi outward fromthe centrosome along Mts, as has been seen by video fluores-cence microscopy (Cooper et al., 1990), or the translocation ofnascent secretory vesicles away from the Golgi. These obser-vations suggest the association of the Golgi with an antero-grade motor such as kinesin.

Kinesin is a Mt-activated ATPase, originally isolated fromneuronal tissues (Brady, 1985; Vale et al., 1985), whichsupports the translocation of vesicular organelles such assecretory vesicles (Ferreira et al., 1992; Leopold et al., 1992;Rothwell et al., 1993; Urrutia et al., 1991), pigment granules(Rodionov et al., 1991) and lysosomes (Hollenbeck andSwanson, 1990) outward from the cell centre to the peripheryin various cell types. Despite detailed morphological and bio-chemical studies of kinesin distribution in neurons (Hirokawaet al., 1991; Leopold et al., 1992) and sea urchin coelomecytes(Henson et al., 1992), there is little evidence for the associa-tion of kinesin with the Golgi. In addition, there is no infor-mation on the distribution of kinesin in epithelial cells. To thisend, we have utilized two well-characterized mono- and poly-clonal kinesin antibodies to conduct a detailed morphologicaland biochemical study on the distribution of this enzyme in apolarized epithelial cell, the rat hepatocyte.

Hepatocytes are highly polarized cells that maintain well-differentiated apical (canalicular) and basolateral (sinusoidal)

2417

The Golgi apparatus is a dynamic membranous structure,which has been observed to alter its location and morphol-ogy during the cell cycle and after microtubule disruption.These dynamics are believed to be supported by a closestructural interaction of the Golgi with the microtubulecytoskeleton and associated motor enzymes. One micro-tubule-dependent motor enzyme, kinesin, has been impli-cated in Golgi movement and function although directevidence supporting this interaction is lacking. In thisstudy, we utilized two well-characterized kinesin antibodiesin conjunction with subcellular fractionation techniques,immunoblot analysis and immunofluorescence microscopyto conduct a detailed study on the association of kinesinwith the Golgi and other membranous organelles in apolarized epithelial cell, the primary rat hepatocyte. Wefound that kinesin represents ~0.3% of total protein in ratliver homogenates, with ~30% membrane-associated andthe remainder in the cytosol. Among membrane fractions,

kinesin was concentrated markedly in Golgi-enrichedfractions, which were prepared using two independenttechniques. Kinesin was also abundant in fractionsenriched in transcytotic carriers and secretory vesicles,with lower levels detected on fractions enriched inendosomes, endoplasmic reticulum, lysosomes and mito-chondria. Immunofluorescence microscopy showed thatkinesin is concentrated on Golgi-like structures in bothprimary cultured hepatocytes and rat hepatocyte-derivedclone 9 cells. Double-label immunofluorescence demon-strated that kinesin staining colocalizes with the Golgimarker,

α-mannosidase II, in both cell types. These resultsprovide compelling evidence showing that kinesin is asso-ciated with the Golgi complex in cells and implicate thismotor enzyme in Golgi structure, function and dynamics.

Key words: liver, microtubule motors, immunofluorescence,transcytotic carriers

SUMMARY

Association of kinesin with the Golgi apparatus in rat hepatocytes

David L. Marks, Janet M. Larkin and Mark A. McNiven*

Center for Basic Research in Digestive Diseases, Mayo Clinic and Mayo Foundation, Rochester, MN 55905, USA

*Author for correspondence

Journal of Cell Science 107, 2417-2426 (1994)Printed in Great Britain © The Company of Biologists Limited 1994

2418

domains via Mt-dependent vesicular trafficking processes(Bartles et al., 1985; Roman and Hubbard, 1983). These cellspossess well-developed synthetic machinery and secrete over30 different proteins into the basolateral sinusoid, while inter-nalizing, transporting and degrading numerous plasma proteins(Donohue et al., 1990; Marks and LaRusso, 1993). In addition,hepatocytes utilize a transcytotic pathway in which nascentproteins destined for the canalicular domain are first trans-ported to the sinusoidal plasmalemma, then vesiculated andtranscytosed to the canalicular membrane (Bartles et al., 1985).Thus, the hepatocyte possesses both conserved and uniquevesicular pathways and provides an exceptional model withwhich to examine the participation of kinesin in organelletransport and Golgi dynamics.

In this study, we have conducted extensive purification andcharacterization of different vesicular organelles from rat liverand, by immunoblot analysis, provide strong evidence thatkinesin is associated with secretory vesicles, transcytoticcarriers and, most dramatically, the Golgi. Immunofluorescentstaining of the Golgi in primary hepatocytes and a hepatocyte-derived cell line (clone 9) with several different kinesin anti-bodies supports this observation. These results suggest thatkinesin is involved in multiple vesicular pathways in the hepa-tocyte and provide novel evidence for an interaction of kinesinwith the Golgi apparatus.

MATERIALS AND METHODS

MaterialsPolyclonal antisera to α-mannosidase II (α-man II) and rab7 were giftsfrom Dr Marilyn Farquhar, University of California, San Diego, andMarino Zerial, EMBL, Heidelberg, respectively. Anti-α-man II mono-clonal ascites fluid was purchased from Berkeley Antibody Co.,Berkeley, CA. Hybridoma supernatant containing SUK-4, a previouslydescribed monoclonal antibody made against sea urchin kinesin heavychain (Ingold et al., 1988), was obtained from the Developmental StudiesHybridoma Bank (Iowa City, IA). Purified rat liver kinesin was preparedusing a modification of a method used to purify kinesin from bovineadrenal medulla (Urrutia and Kachar, 1992). Leupeptin, pepstatin, tosylarginine methyl ester, phenylmethylsulfonyl fluoride, benzamidine andsoybean trypsin inhibitor were from Sigma (St Louis, MO).

Preparation and purification of a polyclonal antibody tokinesin heavy chain (MMR44)Peptides (KKLSGKLYLVDLAGSEKVSKTGAEG and HIPYRD-SKLTRILQESLGGNARTT), based on consensus regions of themicrotubule-binding domain of the kinesin heavy chain conserved insquid, fly and sea urchin (Wright et al., 1991), were synthesized by theMayo Clinic Peptide Core, Rochester, MN. The peptides were conju-gated to keyhole limpet hemocyanin, suspended in Freud’s adjuvent,and injected into male New Zealand White rabbits. Serum sampleswere collected and screened for the ability to recognize kinesin fromrat liver and rat brain on immunoblots. For immunofluorescence, thepolyclonal anti-kinesin antibodies were affinity purified from serumusing the above mentioned peptides immobilized on an Amino-Linkcolumn (Pierce Chemical, Rockford IL). Purified antibodies wereeluted from the columns using the Immunopure buffer system (PierceChemical), and concentrated and equilibrated into Dulbecco’sphosphate buffered saline (DPBS: 8 mM Na2HPO4, 137 mM NaCl,2.7 mM KCl, 1.5 mM KH2PO4, 0.5 mM MgCl2, 0.04% NaN3, pH7.25) using Centricon-10 filters (Amicon Co., Beverly, MA).

Isolation of rat liver subcellular fractionsAll liver fractionation steps were performed at 4°C. First, fractions

enriched in specific organelles were prepared from rat liver bycombining the crude liver fractionation scheme of de Duve et al.(1955) with a method used to subfractionate total liver microsomes(Larkin and Palade, 1991). Rats were anesthetized with pentobarbitaland the livers perfused with buffer A (100 mM Tris-base, 0.25 Msucrose plus multiple protease inhibitors (10 µg ml−1 leupeptin, 10 µgml−1 tosyl arginine methyl ester, 1 µg ml−1 pepstatin, 10 mM benza-midine, 0.1 mg ml−1 soybean trypsin inhibitor, 1 mM phenylmethyl-sulfonyl fluoride) via the portal vein to remove blood. Livers werethen homogenized in two volumes of buffer A by ten passes of aPotter-Elverjeim homogenizer at 2000 rpm. The homogenate was cen-trifuged at 800

g for 10 minutes and the supernatant was saved; thepellet was homogenized twice (single pass of the homogenizer at 2000rpm) in buffer A and centrifuged as above to prepare a nuclear-enriched (N) pellet (de Duve et al., 1955). The postnuclear super-natants were pooled and centrifuged at 36,000 g for 11 minutes; thepellet was twice homogenized (single pass at 2000 rpm) in buffer Aand centrifuged at 36,000 g for 11 minutes to prepare a pellet enrichedin mitochondria and lysosomes (ML). The remaining supernatantswere pooled and centrifuged at 105,000 g for 90 minutes to separatea total microsome pellet (TM fraction) from the final supernatant orcytosol (S). The total microsome pellet was resuspended in buffer Aand fractionated through two sucrose gradient centrifugation steps asdetailed previously (Larkin and Palade, 1991) to prepare fractionsenriched in Golgi membranes (Golgi light (GL) and Golgi heavy(GH)), transcytotic carriers (TC), secretory vesicles (SV) and crudeendoplasmic reticulum (ER).

In addition, we used separate techniques to prepare liver fractionsenriched in specific organelles. Hepatic lysosomes were purified bythe Ca2+ shift method of Yamada et al. (1984). Intact Golgi cisternaewere isolated from rat liver by the method of Hamilton et al. (1991).Hepatic mitochondria were prepared from rat liver by sucrose gradientcentrifugation as described previously (Rickwood et al., 1987).Fractions enriched in late endosomes were prepared as described(Mullock and Luzio, 1992). Each technique was carried out asdescribed except that homogenization buffers contained proteaseinhibitors as listed above.

Total protein contents in liver subcellular fractions were measuredusing the BCA method (Pierce Chemical, Rockford IL) after solubi-lization of fractions in 1% SDS. Enzymatic activities in subcellularfractions of β-N-acetylglucosaminidase (β-NAG), a lysosomal markerenzyme, and malate dehydrogenase (MDH), a mitochondrial enzyme,were measured as previously described (Dupourque and Kun, 1969;LaRusso and Fowler, 1979). Enrichment for α-man II in Golgifractions was determined by quantitative immunoblotting as describedbelow for kinesin, except that polyclonal anti-α-man II antibody(1:1000 dilution) was used as the primary antibody.

SDS-PAGE and immunoblottingKinesin-containing fractions were run on 8.5% acrylamide SDS-PAGE gels with dithiothreitol, using the method of Laemmli (1970)with 20:1 (v/v), acrylamide:bis-acrylamide (Porter et al., 1987).Protein bands in gels were visualized by Coomassie Blue staining. Forimmunoblots, proteins were transferred from gels to polyvinyldifluo-ride by established methods (Towbin et al., 1979). Blots were blockedwith 5% bovine serum albumin, incubated with polyclonal anti-kinesin antibodies, diluted to 1:1000 followed by 1:2000 goat anti-rabbit-alkaline phosphatase (Tago Immunochemicals, BurlingameCA) and then visualized as previously described (Dubreuil et al.,1985). Signals detected on blots were scanned using a UMAX ultra-vision scanner attached to a MacIntosh Quadra 700; data were quan-tified using the Image 1.47 program. Samples of purified rat liverkinesin were run as internal standards on all kinesin blots.

Cell culture and immunofluorescenceClone 9 cells (a rat hepatocyte-derived cell line from the AmericanType Culture Collection, Rockville MD) were grown on glass cover-

D. L. Marks, J. M. Larkin and M. A. McNiven

2419Golgi-associated kinesin

slips in DMEM with 10% FBS. Rat hepatocytes were isolated by themethod of Seglen (1976) and cultured on type I rat tail collagen-coated coverslips in DMEM supplemented with 10% FBS, 0.5 i.u./mlinsulin, 20 ng/ml EGF, 7.5 µg/ml hydrocortisone, 200 µg/ml strepto-mycin, and 200 i.u./ml penicillin for 2-4 hours. Cells were fixed for10 minutes with 2% formaldehyde, permeabilized with Triton X-100(0.1% for 4 minutes for clone 9 cells; 0.2% for 10 minutes for hepa-tocytes), quenched with 0.01 M glycine (3× 5 minutes), washed withDPBS and incubated for 1 hour at room temperature with blockingbuffer (5% normal goat serum, 5% glycerol in DPBS). Cells were thenincubated with primary antibodies overnight at 4°C. Primary anti-bodies used were 1:400 diluted affinity-purified polyclonal anti-kinesin antibody, undiluted SUK4 hybridoma supernatant, 1:1000polyclonal anti-α-man II and 1:1000 monoclonal anti-α-man II. Afterwashing with DPBS, cells were incubated in appropriate secondaryantibodies for 1 hour at room temperature. Secondary antibodies usedwere 1:500 diluted FITC-conjugated goat anti-rabbit, 1:500 FITC-conjugated goat anti-mouse, 1:200 TRITC-conjugated goat anti-rabbit(TAGO Immunochemicals, Burlingame, CA), and 1:200 TRITC-con-jugated goat anti-mouse (Kirkegarde and Perry, Gaithersburg, MD).The cells were then washed with DPBS and mounted on slides inSlowfade (Molecular Probes, Eugene, OR). In double-label immuno-fluorescence experiments, cells were incubated simultaneously withpolyclonal and monoclonal primary antibodies, washed and thenincubated simultaneously with two secondary antibodies conjugatedto different fluorophores. All other procedures were as describedabove for single label immunofluorescence.

RESULTS

Characterization of polyclonal kinesin antibodies To study the distribution of kinesin in hepatocytes, weprepared polyclonal antibodies to synthetic peptides represent-ing two different conserved regions of the kinesin Mt-bindingdomain (KKLSGKLYLVDLAGSEKVSKTGAEG (MMR43and 44) and HIPYRDSKLTRILQESLGGNARTT (MMR48)).Because all three of these antibodies gave identical results byimmunoblotting and immunfluorescence, we only present datafor MMR44. Characterization of these antibodies wasperformed by immunoblotting kinesin-containinghomogenates from rat liver, rat brain and hepatocyte-derivedclone 9 cells. As shown in Fig. 1, MMR44 recognizes a singleband (molecular mass ~120 kDa) in homogenates from wholeliver (L) and clone 9 cells (C9). In brain homogenate (Br),however, MMR44 recognizes three protein bands, including aprominent doublet at ~120 kDa and a single band at ~130 kDa.

The usefulness of MMR44 for quantitative immunoblottingwas established by blotting serial dilutions of purified liverkinesin. Kinesin signals detected on blots were linearly relatedto the amount of kinesin loaded between 50 and 300 ng (datanot shown). In addition, MMR44 immunoprecipitates a single~120 kDa protein band from hepatocyte homogenates that isrecognized by multiple kinesin antibodies via immunoblotting(data not shown). Thus, from these biochemical criteria, we areconfident that MMR44 recognizes, and is specific for, kinesinsfrom different cells and tissues, making it a useful tool for thestudies described below.

Kinesin is associated with Golgi-enriched fractionsfrom rat liverTo determine the intracellular distribution of kinesin in ratliver, we isolated crude subcellular fractions and fractions

L Br C9 K K'

Kinesinheavy chain

Kinesinlight chains

Tubulin

Fig. 1. The polyclonal antibody, MMR44, recognizes kinesin intissue and cell homogenates. Homogenates of liver (L), brain (Br),clone 9 cells (C9), and purified liver kinesin (K) were run on SDS-PAGE gels, transferred to polyvinyldifluoride membranes andprobed with MMR44. For comparison, a Coomassie Blue-stained gelof purified liver kinesin is shown (K′). From left to right, proteinloaded per lane was 50, 50, 15, 0.02 and 5 µg. In liver and clone 9cells, a single kinesin heavy chain band (molecular mass ~120 kDa)is recognized; while in brain, a doublet (molecular mass ~120) andan upper band (molecular mass ~130 kDa) are detected.

Table 1. Relative enrichment of organelle marker enzymesin liver fractions

Fold enrichment*

Golgi Lysosomes Mitochondria(α-man II)† (β-NAG)‡ (MDH)§

Crude fractionsNuclear pellet (N) 0.25 0.31 0.67Lysosomal/mitochondrial pellet (LM) 3.00 4.18 0.81Total microsomes (TM) 0.16 0.19 0.09Cytosol (S) <0.3

| 0.16 1.04

Organelle-enriched fractionsGolgi heavy (GH) 5.70 0.25 0.22Golgi light (GL) 16.75 0.40 0.00Intact Golgi (IG) 67.00 1.05 0.12Transcytotic carriers (TC) 5.30 0.10 0.02Secretory vesicles (SV) 4.80 0.14 0.25Crude endoplasmic reticulum (ER) 0.55 0.24 0.03Mitochondria (M) <0.3| 7.12 4.42Purified lysosomes (L) <0.3| 44.50 0.29Endosomes (E) 2.49 0.36 0.27

*Values are means of replicate measurements from representativepreparations and are expressed as relative specific activity compared tospecific activity in homogenate samples. Numbers in bold indicate fractionswith the greatest enrichment in each enzyme.

†Relative specific enzyme mass of α-man II (α-mannosidase II) wasmeasured by quantitative immunoblotting (see Materials and Methods).

‡Relative specific activity of β-NAG (N-acetyl-β-glucosaminidase) wasmeasured by fluorescence assay using 4-methyl umbelliferyl-2-aceto-amido-2-deoxy-β-D-glucopyranoside as a substrate as previously described(LaRusso and Fowler 1979).

§Relative specific activity of MDH (malate dehydrogenase) measured byfluorescence assay as previously described (Dupourque and Kun, 1969).

|No detectable signal when 20 µg of protein was immunoblotted with an α-man II antibody.

2420

highly enriched in different populations of vesicularorganelles. Each fraction was characterized extensively usingimmunoblot analysis and enzyme assays for standard organellemarker proteins. Mitochondria (M), lysosomes (L), secretoryvesicles (SV), and transcytotic carriers (TC), were isolatedusing previously established procedures (see Materials andMethods), while Golgi was isolated using two independentmethods. Heavy (GH) and light Golgi (GL) fractions were

separated by the fractionation method used by Larkin andPalade (1991), while intact Golgi cisternae (IG) were preparedby the method of Hamilton et al. (1991). Late endosome-enriched fractions, which were isolated by the method ofMullock and Luzio (1992) were identified by density, thepresence of endocytosed 125I-labeled asialofetuin, and anenrichment of the late endosomal marker rab7 (data notshown). Table 1 shows the enrichment of Golgi (α-man II),lysosomal (β-NAG), and mitochondrial (MDH) markerenzymes in different organelle fractions. The Golgi marker, α-man II, was enriched in the Golgi fractions (GH, GL and IG)by 5.7, 16.7 and 67-fold, respectively, relative to homogenate.The activity of the lysosomal enzyme, β-NAG, was enriched44-fold in lysosome fractions (L). Finally, the mitochondrialmarker MDH was enriched exclusively (4-fold) in mitochon-drial fractions (M). Thus, using established techniques, we

D. L. Marks, J. M. Larkin and M. A. McNiven

0

2

4

6

0

20

40

60

80

Kin

esin

(% r

eco

vere

d)

A

B

C

K H ML TMN S

H ML TMN S

ML TMN S

Kin

esin

(µg

mg

−1)

Fig. 2. Distribution of kinesin in crude liver fractions. Hepaticsubcellular fractions were prepared and assayed for kinesin byimmunoblot analysis using MMR44 (see Materials and Methods).(A) A representative immunoblot of crude liver fractions probed withMMR44. K, purified liver kinesin (0.1 µg); H, liver homogenate; N,nuclear pellet; ML, mitochondrial-lysosomal fraction; TM, totalmicrosomes; S, supernatant. Lanes H, N, ML, TM and S were eachloaded with 25 µg protein. (B) Quantification of kinesin distributionas assessed by scanning densitometry of immunoblots using purifiedliver kinesin as an internal standard. Values are expressed as µgkinesin/mg total protein and represent means ± s.e.m. of 10 blots,which included fractions prepared from 8 rat livers. Labels are as inA. (C) Distribution of kinesin in liver fractions expressed as apercentage of total kinesin recovered. Labels are as in A. Thepercentage of total kinesin in each fraction was calculated as follows:(µg kinesin/mg protein in each fraction (shown in B) × total proteinrecovered in each fraction) ÷ (total kinesin recovered in the N, ML,TM and S fractions combined) × 100. Note that ~30% of recoveredkinesin is associated with membranous (N, ML and TM) fractionsand the rest with cytosol (S).

Kinesinheavychain

A

B

GH GL TC SV L M ER E K

0

2

4

6

8

IG GH GL TC SV L M ER E

10

Kin

esin

(µg

mg

−1)

IG

Fig. 3. Kinesin is enriched on the Golgi andvesicle carriers. Liver fractions enriched forspecific organelles were prepared andimmunoblotted using kinesin antibody, MMR44(see Materials and Methods). (A) Arepresentative immunoblot showing kinesindistribution in characterized membrane fractions.Lane K was loaded with 0.1 µg of purified liverkinesin. Other lanes were loaded with 20 µg ofprotein. Abbreviations are defined in Table 1.(B) Quantification of kinesin distribution inorganelle-enriched fractions. Kinesinquantification was performed as in Fig. 2. Valuesare expressed as µg kinesin/mg total protein andrepresent means ± s.e.m. of ≥3 blots.

2421Golgi-associated kinesin

obtained and characterized highly enriched populations ofspecific hepatocellular organelles with purities equal to thosepreviously reported by others (Andersson and Glaumann,1987; Declercq et al., 1984; Yamada et al., 1984).

We then utilized quantitative immunoblotting to assess the

prevalence of kinesin in crude liver fractions and the charac-terized organelle populations described above. Fig. 2A is a rep-resentative immunoblot showing the distribution of kinesin inliver homogenate (H) and in fractions enriched in nuclei (N),mitochondria and lysosomes (ML), cytosol (S), and total

Fig. 4. Immunofluorescence imagesof a hepatocyte-derived cell line,clone 9 cells, stained with twodifferent anti-kinesin antibodies. (a and b) Clone 9 cells stained withthe kinesin monoclonal antibodySUK4. Kinesin is localized to aperinuclear cap situated on one sideof the nucleus and its distribution issimilar to that of the Golgiapparatus (arrows). (c and d) Twoclone 9 cells stained with thekinesin polyclonal antibody,MMR44. Again, a Golgi-likeperinuclear staining (c) is observed(arrows). When the same two cellsare viewed at a different focal plane(d), numerous, putative cytoplasmicvesicles are observed (arrowheads).Bar, 20 µm.

Fig. 5. Localization of kinesin in cultured rat hepatocytesby immunofluorescence. (a and b) Two hepatocytecouplets cultured for three hours were stained with thepolyclonal kinesin antibody, MMR44. The kinesinantibody stains a complex membranous network localizedto the pericanalicular and perinuclear cytoplasm. Arrowsindicate putative canaliculi between the two cells in eachcouplet. (a′and b′) The same couplets viewed with phase-contrast microscopy. Bar, 20 µm.

2422

microsomes (TM), a fraction that includes plasma membrane,Golgi, ER, endosomes and peroxisomes. Kinesin was quanti-fied in these fractions using scanning densitometry and foundto be most enriched in cytosol (5-6 µg/mg) with lower amounts(0.9-2.2 µg/mg) detectable in the N, ML and TM fractions (Fig.2 B). The majority of hepatocellular kinesin is cytosolic, withonly ~30% associated with membranes (N, ML, TM; Fig. 2 C),a distribution that coincides with that previously reported incultured fibroblasts by Hollenbeck (1989). The association ofkinesin with enriched vesicular organelle fractions (Fig. 3) wassurprising and contrasted with the observations of othersstudying kinesin distribution in non-epithelial tissues (Hensonet al., 1992; Leopold et al., 1992). We found kinesin to behighly abundant in Golgi (Fig. 3), particularly in the IG andGH fractions, which contain ~8 µg kinesin/mg protein. Inaddition, there were significant amounts of kinesin associatedwith the TC fractions. Much lower levels (0.5-2.0 µgkinesin/mg protein) of the motor were found in L, M, andendosome (E) fractions. Predictably, secretory vesicles weresignificantly enriched in kinesin.

Immunofluorescence localization of kinesin to theGolgi apparatus in cultured hepatocytesTo support the biochemical studies described above, it was

important to show morphologically that kinesin is present onthe Golgi apparatus in cultured cells. To this end, weperformed immunofluorescence studies utilizing severaldifferent kinesin antibodies. We obtained a monoclonalantibody made to the N-terminal domain of purified sea urchinkinesin heavy chain (Ingold et al., 1988). This antibody hasbeen well characterized in other molecular (Scholey et al.,1989) and morphological (Henson et al., 1992) studies and wasused in conjunction with the library of polyclonal kinesin-peptide antibodies described above. Kinesin antibodies wereapplied to two distinct cultured cell populations: rat hepato-cyte-derived clone 9 cells, which are flat, amenable to fluores-cence microscopy, and secrete several hepatocyte proteins suchas albumin and transferrin (data not shown) but are notpolarized; and primary cultured rat hepatocytes, which arethick and difficult to stain but are differentiated and remainpolarized for several hours after isolation. Thus, both cell typeshave distinct limitations and advantages for this study. Clone9 cells stained with the monoclonal kinesin antibody (SUK4)display a delicate, reticular network localized to one side of thenucleus, which is reminiscent of the Golgi apparatus (Fig. 4aand b), while the polyclonal antibody (MMR44) stains asimilar structure in these cells (Fig. 4c). Interestingly, whencells were viewed at different planes of focus, we observed that

D. L. Marks, J. M. Larkin and M. A. McNiven

Fig. 6. Kinesin is localized to the Golgi apparatus in clone 9 cells. Clone 9 cells were double labeled with antibodies to kinesin and a Golgimarker (α-man II). (a and a′) Cells stained with the monoclonal kinesin antibody, SUK4 (a), and a polyclonal Golgi marker antibody (a′). (band b′) A second field of cells stained with the polyclonal kinesin antibody, MMR44 (b), and a monoclonal Golgi marker antibody (b′). Notethe extensive colocalization between kinesin and the Golgi (arrows). Bar, 20 µm.

2423Golgi-associated kinesin

MMR44 also stained punctate vesicular-like structures and anelaborate network that extends outward from one side of thenucleus toward the peripheral cytoplasm (Fig. 4d). The moreextensive staining patterns seen in cells labeled with MMR44probably reflect the higher titer of MMR44 compared to SUK4hybridoma supernatant. Similar staining patterns in these cellswere observed with the other polyclonal kinesin peptide anti-bodies (MMR43 and MMR48; not shown).

In isolated hepatocytes, MMR44 stained an extensivetubulovesicular membrane network located in the pericanalic-ular and perinuclear regions (Fig. 5a and b) as shown by com-parison with phase-contrast images of the same couplets (Fig.5a′ and b′). Again, similar staining patterns were seen withantibodies MMR43 and MMR48 (not shown). On the basis oftheir appearance, we suspected that these immunofluorescentstaining patterns indicate a localization of kinesin to the Golgi,which would be consistent with our biochemical observations;however, the distribution of the Golgi in polarized hepatocyteshas not been defined. Thus, to explore the possibility thatkinesin antibodies indeed stain the Golgi apparatus in clone 9cells and isolated hepatocytes, we conducted double-labelingexperiments using both mono- and polyclonal kinesin anti-bodies in conjunction with antibodies to the Golgi marker α-man II (Figs 6 and 7). In both cell types, we saw a strong cor-

relation between the staining patterns produced by the kinesinand mannosidase antibodies. Again, the other peptide poly-clonal antibodies to kinesin (MMR43 and MMR48) producedsimilar results (data not shown). These data provide definitivemorphological evidence that kinesin is associated with theGolgi apparatus.

DISCUSSION

In this study, we conducted a detailed examination of the dis-tribution of kinesin in a polarized epithelial cell, the rat hepa-tocyte. Several well-characterized polyclonal antibodies tokinesin (Fig. 1) were utilized for immunoblotting analysis ofcrude cell and tissue fractions (Fig. 2), and highly enrichedorganelle preparations (Fig. 3; Table 1), to identify thevesicular compartments to which kinesin associates. Kinesin ispredominantly a soluble enzyme in liver with only ~30%bound to membranes (Fig. 2) as shown previously in culturedfibroblasts (Hollenbeck, 1989). Substantial amounts of kinesinwere seen in Golgi fractions prepared by two differentmethods, with little kinesin found associated with the endo-plasmic reticulum, mitochondria, lysosome, and endosomefractions (Fig. 3). To confirm these biochemical observations,

Fig. 7. Kinesin is localized to the Golgi apparatus in polarized hepatocyte couplets. Cultured hepatocytes were double labeled with antibodies,kinesin (MMR44), and Golgi (α-man II) antibodies. (a and b) Color images of isolated hepatocytes stained for kinesin (a) and Golgi (b). Notethe striking coincidence of the two staining patterns (arrows). (c) In the hepatocytes, colocalization of the kinesin and Golgi antibodies isdemonstrated in a double exposed image. The yellow color indicates an overlap of kinesin- and Golgi-stained structures. (d) For comparison,the same hepatocytes are shown viewed by phase-contrast microscopy. Bar, 20 µm.

2424

we applied the same kinesin antibodies as well as a widely usedkinesin monoclonal antibody, SUK4 (Ingold et al., 1988), toimmunostain both polarized hepatocyte couplets in culture anda non-polarized hepatocyte-derived cell line (clone 9). Wefound that all the antibodies used produced a strong perinu-clear Golgi-like stain in the clone 9 cells (Fig. 4), while thehigh titer polyclonal peptide antibodies stained similar butmore complex structures in the pericanalicular cytoplasm ofthe hepatocyte couplets (Fig. 5). Double immunofluorescencestaining of cells with antibodies to kinesin and the Golgimarker enzyme, α-man II, showed colocalization of the twoantigens (Figs 6 and 7), confirming the association of kinesinwith the Golgi apparatus. These studies provide thorough andcompelling biochemical and morphological evidence thatkinesin is associated with the Golgi apparatus in hepatocytes.

The immunoblotting studies presented here show that kinesinis also prevalent in hepatocellular fractions that are highlyenriched in secretory vesicles (SVs) and transcytotic carriers.While we expected to find kinesin associated with SVs, basedon previous reports in other cell types (Rothwell et al., 1993;Urrutia et al., 1991), an interaction with TC vesicles was notpredicted because the movement of these vesicle carriers isassumed to be exclusively from the sinusoid to the apicalcanalicular domain. This transport would require the action ofa retrograde motor like cytoplasmic dynein as opposed tokinesin. While contamination from other organelles is possible,this vesicle carrier fraction has been characterized (Larkin andPalade, 1991; and see Table 1) and has more associated kinesinthan any other organelle fraction that we have examined (Fig.3). Thus, contamination from other organelles could notaccount for the levels of kinesin found in this fraction.

The morphological identification of the Golgi apparatus incultured hepatocyte couplets using α-man II antibodies showsa preferential association of this organelle with the canalicu-lus, thus extending previous electron microscopic studies(DePierre et al., 1988; Jones, 1990; Macchiarelli and Motta,1993). In our study, the distribution of the Golgi complex wasviewed in its entirety by immunofluorescence and most of theGolgi was shown to be on either side of the nucleus near eachcanalicular domain (Fig. 7). This organization of the Golgi waspresent only in short-term (2-4 hours) cultured hepatocytes thatstill retained some degree of polarity; in hepatocytes culturedfor 24 hours or longer the Golgi was reorganized into perinu-clear structures typical of the arrangement of the Golgi incultured cells (data not shown). While the organization of Mtsin hepatocytes in situ has not been established, it is likely tocoincide with that observed in a recently established polarizedhepatocyte cell line (Ihrke et al., 1993), which is known to havea centrosomal complex at each canalicular domain (Meads andSchroer, 1994). Thus, we predict that the polarized hepatocytepositions the centrosome and associated Golgi complex at itsapical domains, with Mts and nascent secretory vesiclesextending outward toward the sinusoidal (basolateral)domains. This organizational motif appears to be in contrast tothat seen in other polarized epithelia such as the pancreaticacinar cells (Smith et al., 1984) and the columnar cells of theintestinal brush border (Fath and Burgess, 1993).

Is kinesin associated with the Golgi apparatus inother cells?The results presented here both support and contrast with

earlier observations on the localization of kinesin in cells.Immunolocalization studies have suggested that kinesin isassociated with orthograde moving organelles in neurons(Hirokawa et al., 1991) and with membranous vesicles in seaurchin coelomocytes (Henson et al., 1992), primary rat brainneurons and PtK cells (Pfister et al., 1989). The precise identityof these organelles has remained both elusive and unclear, withsome studies providing evidence for an interaction of kinesinwith mitochondria (Leopold et al., 1992), melanophoregranules (Rodionov et al., 1991), and components of thesecretory pathway such as the endoplasmic reticulum (Daboraand Sheetz, 1988; Toyoshima et al., 1992; Vale and Hotani,1988; Yu et al., 1992) and secretory vesicles (Ferreira et al.,1992; Leopold et al., 1992; Rothwell et al., 1993; Urrutia etal., 1991). In addition, other studies have provided morpho-logical or functional evidence that kinesin associates with ret-rograde moving components of the endocytic pathway such ascoated vesicles (Leopold et al., 1992), endosomes (Henson etal., 1992) and lysosomes (Hollenbeck and Swanson, 1990).Recently, Leopold et al. (1992) conducted a thorough study onthe distribution of kinesin in neuronal cells and tissues usingimmunoblot analysis and immunoelectron microscopy, andreported an association of kinesin with mitochondria and endo-plasmic reticulum at levels comparable to those observed byus. Surprisingly, none of the studies cited above have localizedkinesin to the Golgi apparatus. However, in agreement withour results, two recent studies have reported an association ofkinesin with the Golgi; one in Sertoli cells (Johnson et al.,1994), and the other in bovine corneal endothelial cells andchromaffin cells of the adrenal medulla (Schmitz et al., 1994).

What then can account for the differences between thesestudies on kinesin distribution? Certainly, there are likely to bevariations in results depending on the type and origin of cellsexamined, the antibody reagents used, and the methods bywhich specific organelles are isolated. Indeed, it is possible thatkinesin associates with different organelles in a hepatocyte ascompared to a neuron and that the kinesins utilized by variousorganelle populations in these cells are different as well. It isnow generally accepted that there are many different isoformsof kinesin, which may perform distinct and specific cellularfunctions (Bloom, 1992; Vale, 1992; Vale and Goldstein,1990). For example, there are likely to be more than a dozendifferent kinesin-like proteins in the axon alone (Aizawa et al.,1992), some of which may be used to move synaptic vesicles(Gho et al., 1992; Hall and Hedgecock, 1991). For thesereasons, we conducted this study using a collection of anti-bodies (SUK4, MMR43, MMR44, MMR48) directed toconsensus domains of kinesin. The regions we selected for thegeneration of peptide antibodies include two different, highlyconserved Mt binding sites, which are nearly identical betweenurchin, squid, fly (Wright et al., 1991) and human kinesins(Navone et al., 1992). The monoclonal antibody, SUK4, madeto purified sea urchin kinesin, is directed to similar conservedmotor domains as has been shown by molecular methods(Scholey et al., 1989). Thus, we predict that most kinesinisoforms associated with vesicular organelles in the hepatocytewill be recognized with these reagents using both immunoflu-orescence and immunoblot criteria.

Finally, it is important to note that the purification methodsemployed to isolate enriched populations of vesicularorganelles from cells and tissues are highly variable not only

D. L. Marks, J. M. Larkin and M. A. McNiven

2425Golgi-associated kinesin

in purity but in proteolytic activity as well. For example, Golgiisolated from liver by a method modified by Bloom andBrashear (1989) was reported previously to have little or noassociated kinesin (Leopold et al., 1992). In our hands, Golgifractions prepared from rat liver by two different methodscontained large amounts of kinesin (Hamilton et al., 1991;Larkin and Palade, 1991). We found that the rapid executionof all procedures, the use of six different protease inhibitors,and low temperature during all steps were critical, to reducethe proteolytic degradation of kinesin. Golgi fractions preparedwithout strict adherence to these technical details were foundto contain much less kinesin as assessed by immunoblotting(data not shown). Thus, it is likely that a summation of the bio-logical and technical differences described above are respon-sible for the reported variability of kinesin distribution in cells.

Potential functions for Golgi-associated kinesinWe believe that it is not surprising to find kinesin associatedwith the Golgi apparatus, since there are multiple, motilefunctions performed by this dynamic organelle. First, it hasbeen proposed that kinesin may support the movement or“retrieval” of nascent, protein-containing vesicles from theGolgi back to the endoplasmic reticulum. This model is rea-sonable in light of the classic brefeldin A studies by Lippincott-Schwartz et al. (1989, 1990) that have suggested that theretrieval pathway from Golgi to the ER is Mt-dependent,requiring the participation of an orthograde Mt motor. Second,the positioning of the Golgi at the centrosome is likely toinvolve an antagonistic interaction between kinesin and the ret-rograde motor cytoplasmic dynein, which has been shown to beassociated with the Golgi by functional studies (Corthesy-Theulaz et al., 1992). Indeed, the extensive, anastomosing Golginetworks visualized in many cells are likely to be formed bylong tubular extensions from the trans-Golgi outward along Mts(Cooper et al., 1990). Such elaborate membranous structureswould likely require the added participation of an orthogradeacting motor like kinesin, since dynein acting alone wouldresult in a tightly compacted Golgi mass situated at the centro-some. This proposed outward extension of the trans-Golgi bykinesin is attractive, since it would facilitate vesicle formation,budding, and the subsequent kinesin-mediated transport ofsecretory vesicles to the cell surface. We are currently con-ducting functional studies to define how kinesin supports themovements of the Golgi complex and its associated vesicles.

We thank Barbara Oswald for technical assistance, KennethCamacho for preparation of endosome-enriched fractions and SuePeterson for typing this manuscript. This work was supported by aThompson-Mayo fellowship to D.L.M. and NIH grants DK44650 andAA09227-02 to M.A.M. Preliminary versions of this work werepresented previously in abstract form (Mol. Biol. Cell 1993, 4, 274A;Hepatology 1993, 18, 166A)

REFERENCES

Aizawa, H., Sekine, Y., Takemura, R., Zhang, Z., Nangaku, M. andHirokawa, N. (1992). Kinesin family in murine central nervous system. J.Cell Biol. 119, 1287-1296.

Andersson, G. and Glaumann, H. (1987). Current methods for the isolation ofcoated vesicles, endocytic vesicles, the Golgi complex, lysosomes andautophagic vacuoles. In Lysosomes: Their Role in Protein Breakdown (ed. H.Glaumann and F. J. Ballard), pp. 61-113. New York: Academic Press.

Bartles, J. R., Braiterman, L. T. and Hubbard, A. L. (1985). Endogenousand exogenous domain markers of the rat hepatocyte plasma membrane. J.Cell Biol. 100, 1126-1138.

Bloom, G. S. and Brashear, T. A. (1989). A novel 58-kDa protein associateswith the Golgi apparatus and microtubules. J. Biol. Chem. 264, 16083-16092.

Bloom, G. S. (1992). Motor proteins for cytoplasmic microtubules. Curr. Opin.Cell Biol. 4, 66-73.

Brady, S. T. (1985). A novel brain ATPase with properties expected for the fastaxonal transport motor. Nature 317, 73-75.

Cooper, M. S., Cornell-Bell, A. H., Chernajavsky, A., Dani, J. W. andSmith, S. J. (1990). Tubulovesicular processes emerge from trans-Golgicisternae, extend along microtubules, and interlink adjacent trans-Golgielements into a reticulum. Cell 61, 135-145.

Corthesy-Theulaz, I., Pauloin, A. and Pfeffer, S. R. 1. (1992). Cytoplasmicdynein participates in the centrosomal localization of the Golgi complex. J.Cell Biol. 118, 1333-1345.

Dabora, S. L. and Sheetz, M. P. (1988). The microtubule-dependentformation of a tubulovesicular network with characteristics of the ER fromcultured cell extracts. Cell 54, 27-35.

Declercq, P. E., Haagsman, H. P., Van Veldhoven, P., Debeer, L. J., VanGolde, L. M. G. and Mannaerts, G. P. (1984). Rat liver dihydroxyacetone-phosphate acyltransferases and their contribution to glycerolipid synthesis. J.Biol. Chem. 259, 9064-9075.

de Duve, C., Pressman, B. C., Gianetto, R., Wattiaux, R. and Appelmans,F. (1955). Tissue fractionation studies. 6. Intracellular distribution patternsof enzymes in rat liver tissue. Biochem. J. 60, 604-616.

DePierre, J. W., Andersson, G. and Dallner, G. (1988). Endoplasmicreticulum and Golgi complex. In The Liver, Biology and Pathobiology (ed. I.M. Arias, W. B. Jakoby, H. Popper, D. Schachter and D. A. Shafritz), pp.165-187. New York: Raven Press.

Donohue, T. M. Jr, Jennett, R. B., Tuma, D. J. and Sorrell, M. F. (1990).Synthesis and secretion of plasma proteins by the liver. In Hepatology, aTextbook of Liver Disease (ed. D. Zakim and T. D. Boyer), pp. 124-137.Philadelphia, PA: W. B. Saunders Co.

Dubreuil, R., Byers, T. J., Branton, D., Goldstein, L. S. B. and Kiehart, D.P. (1985). Drosophila spectrin. I. characterization of the purified protein. J.Cell Biol. 105, 2095-2102.

Dupourque, D. and Kun, E. (1969). Cytoplasmic and mitochondrial malatedehydrogenases from beef kidney. Meth. Enzymol. 13, 116-122.

Fath, K. R. and Burgess, D. R. (1993). Golgi-derived vesicles from developingepithelial cells bind actin filaments and possess myosin-I as a cytoplasmicallyoriented peripheral membrane protein. J. Cell Biol. 120, 117-127.

Ferreira, A., Niclas, J., Vale, R. D., Banker, G. and Kosik, K. S. (1992).Suppression of kinesin expression in cultured hippocampal neurons usingantisense oligonucleotides. J. Cell Biol. 117, 595-606.

Gho, M., McDonald, K., Ganetzky, B. and Saxton, W. M. (1992). Effects ofkinesin mutations on neuronal functions. Science 258, 313-316.

Hall, D. H. and Hedgecock, E. M. (1991). Kinesin-related gene unc-104 isrequired for axonal transport of synaptic vesicles in C. elegans. Cell 65, 837-847.

Hamilton, R. L., Moorehouse, A. and Havel, R. J. (1991). Isolation andproperties of nascent lipoproteins from highly purified rat hepatocytic Golgifractions. J. Lipid. Res. 32, 529-543.

Henson, J. H., Nesbitt, D., Wright, B. D. and Scholey, J. M. (1992).Immunolocalization of kinesin in sea urchin coelomocytes. Association ofkinesin with intracellular organelles. J. Cell Sci. 103, 309-320.

Hirokawa, N., Sato-Yoshitake, R., Kobayashi, N., Pfister, K. K., Bloom, G.S. and Brady, S. T. (1991). Kinesin associates with anterogradelytransported membranous organelles in vivo. J. Cell Biol. 114, 295-302.

Ho, W. C., Allen, V. J., van Meer, G., Berger, E. G. and Kreis, T. E. (1989).Reclustering of scattered Golgi elements occurs along microtubules. Eur. J.Cell Biol. 48, 250-263.

Hollenbeck, P. J. (1989). The distribution, abundance and subcellularlocalization of kinesin. J. Cell Biol. 108, 2335.

Hollenbeck, P. J. and Swanson., J. A. (1990). Radial extension ofmacrophage tubular lysosomes supported by kinesin. Nature 346, 864-866.

Ihrke, G., Neufeld, E. B., Meads, T., Shanks, M. R., Cassio, D., Laurent,M., Schroer, T. A., Pagano, R. E. and Hubbard, A. L. (1993). WIF-BCells: an in vitro model for studies of hepatocyte polarity. J. Cell Biol. 123,1761-1775.

Ingold, A. L., Cohn, S. A. and Scholey, J. M. (1988). Inhibition of kinesin-driven motility by monoclonal antibodies to kinesin heavy chains. J. CellBiol. 107, 2657-2667.

2426

Johnson, K. J., Hall, E. S. and Boekelheide, K. (1994). Kinesin localizes tothe trans-Golgi network of epithelial cells. Mol. Biol. Cell 4, 274a.

Jones, A. L. (1990). Anatomy of the normal liver. In Hepatology (D. Zakim andT. D. Boyer), pp. 3-30. Philadelphia, PA: W.B. Saunders Co.

Kronebusch, P. J. and Singer, S. J. (1987). The microtubule-organizingcomplex and the Golgi apparatus are co-localized around the entire nuclearenvelope of interphase cardiac myocytes. J. Cell Sci. 88, 25-34.

Kupfer, A., Louvard, D. and Singer, S. J. (1982). Polarization of the Golgiapparatus and the microtubule-organizing center in cultured fibroblasts at theedge of an experimental wound. Proc. Nat. Acad. Sci. USA 79, 2603-2607.

Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly ofthe head of bacteriophage T4. Nature 227, 680-685.

Larkin, J. M. and Palade, G. E. (1991). Transcytotic vesicular carriers forpolymeric IgA receptors accumulate in rat hepatocytes after bile ductligation. J. Cell Sci. 98, 205-216.

LaRusso, N. F. and Fowler, S. (1979). Coordinate secretion of acid hydrolasesin rat bile: hepatocyte exocytosis of lysosomal protein. J. Clin. Invest. 64,948-954.

Leopold, P. L., DcDowall, A. W., Pfister, K. K., Bloom, G. S. and Brady, S.T. (1992). Association of kinesin with characterized membrane-boundedorganelles. Cell Motil. Cytoskel. 23, 19-33.

Lippincott-Schwartz, J., Yuan, L. C., Bonifacino, J. S. and Klausner, R. D.(1989). Rapid redistribution of Golgi proteins into the ER in cells treated withbrefeldin A: evidence for membrane cycling from Golgi to ER. Cell 56, 801-813.

Lippincott-Schwartz, J., Donaldson, J. G., Schweizer, A., Berger, E. G.,Hauri, H.-P., Yuan, L. C. and Klausner, R. D. (1990). Microtubule-dependent retrograde transport of proteins into the ER in the presence ofbrefeldin A suggests an ER recycling pathway. Cell 60, 821-836.

Lucocq, J. M. and Warren, G. (1987). Fragmentation and partitioning of theGolgi apparatus during mitosis in HeLa cells. EMBO J. 6, 3239-3246.

Lucocq, J. M., Berger, E. G. and Warren, G. (1989). Mitotic Golgi fragments inHeLa cells and their role in the reassembly pathway. J. Cell Biol. 109, 463-474.

Macchiarelli, G. and Motta, P. M. (1993). Ultrastructure of the hepatocyte. InHepatic Transport and Bile Secretion: Physiology and Pathophysiology (ed.N. Tavoloni and P. D. Berk), pp. 103-116. New York: Raven Press.

Marks, D. L. and LaRusso, N. F. (1993). Vesicle-dependent transportpathways in liver cells. In Hepatic Transport and Bile Secretion: Physiologyand Pathophysiology (ed. N. Tavoloni and P. D. Berk), pp. 513-530. NewYork: Raven Press.

Meads, T. and Schroer, T. A. (1994). Localization of cytoplasmic dynein andthe dynactin complex in polarized epithelial cells. Mol. Biol. Cell 4, 402a.

Mullock, B. M. and Luzio, J. P. (1992). Reconstitution of rat liver endosome-lysosome fusion in vitro. Meth. Enzymol. 219, 52-60.

Navone, F., Niclas, J., Hom-Booher, N., Sparks, L., Bernstein, H. D.,McCaffrey, G. and Vale, R. D. (1992). Cloning and expression of a humankinesin heavy chain gene: interaction of the COOH-terminal domain withcytoplasmic microtubules in transfected CV-1 cells. J. Cell Biol. 117, 1263-1275.

Pfister, K. K., Wagner, M. C., Stenoien, D. L., Brady, S. T. and Bloom, G. S.(1989). Monoclonal antibodies to kinesin heavy and light chains stainvesicle-like structures, but not microtubules, in cultured cells. J. Cell Biol.108, 1453-1463.

Porter, M. E., Scholey, J. M., Stemple, D. L., Vigers, G. P. A., Vale, R. D.,Sheetz, M. P. and McIntosh, J. R. (1987). Characterization of themicrotubule movement produced by sea urchin egg kinesin. J. Biol. Chem.262, 2794-2803.

Rickwood, D., Wilson, M. T. and Darley-Usmar, V. M. (1987). Isolation andcharacteristics of intact mitochondria. In Mitochondria: a PracticalApproach (ed. V. M. Darley-Usmar, D. Rickwood and M. T. Wilson), pp.Oxford, England: IRL Press.

Rodionov, V. I., Gyoeva, F. K. and Gelfand, V. I. (1991). Kinesin isresponsible for centrifugal movement of pigment granules in melanophores.Proc. Nat. Acad. Sci. USA 88, 4956-4960.

Rogalski, A. A. and Singer, S. J. (1984). Associations of elements of the Golgiapparatus with microtubules. J. Cell Biol. 99, 1092-1100.

Roman, L. M. and Hubbard, A. L. (1983). A domain-specific marker for thehepatocyte plasma membrane: localization of leucine aminopeptidase to thebile canalicular domain. J. Cell Biol. 96, 1548-1558.

Rothwell, S. W., Deal, C. D., Pinto, J. and Wright, D. G. (1993). Affinitypurification and subcellular localization of kinesin in human neutrophils. J.Leuk. Biol. 53, 372-380.

Schmitz, F., Wallis, K. T., Rho, M., Drenchkahn, D. and Murphy, D. B.(1994). Intracellular distribution of kinesin in chromaffin cells. Eur. J. CellBiol. 63, 77-83.

Scholey, J. M., Heuser, J., Yang, J. T. and Goldstein, L. S. B. (1989).Identification of globular mechanochemical heads of kinesin. Nature 338,355-357.

Seglen, P. O. (1976). Preparation of isolated rat liver cells. Meth. Cell Biol. 13,29-83.

Smith, Z. D. J., D’Eugenio-Gumkowski, F., Yanagisawa, K. and Jamieson,J. D. (1984). Endogenous and monoclonal antibodies to the rat pancreaticacinar cell Golgi complex. J. Cell Biol. 98, 2035-2046.

Thyberg, J. and Moskalewski, S. (1985). Microtubules and the organizationof the Golgi complex. Exp. Cell Res. 159, 1-16.

Towbin, H., Staehelin, T. and Gordon, J. (1979). Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets: procedure andsome applications. Proc. Nat. Acad. Sci. USA 76, 4350-4354.

Toyoshima, I., Yu, H., Steuer, E. R. and Sheetz, M. P. (1992). Kinectin, amajor kinesin binding protein on ER. J. Cell Biol. 118, 1121-1131.

Turner, J. R. and Tartakoff, A. M. (1989). The response of the Golgi complexto microtubule alterations: the roles of metabolic energy and membranetraffic in Golgi complex organization. J. Cell Biol. 109, 2081-2088.

Urrutia, R., McNiven, M. A., Albanesi, J. P., Murphy, D. B. and Kachar,B. (1991). Purified kinesin promotes vesicle motility and induces activesliding between microtubules in vitro. Proc. Nat. Acad. Sci. USA 88, 6701-6705.

Urrutia, R. and Kachar, B. (1992). An improved method for the purificationof kinesin from bovine adrenal medulla. J. Biochem. Biophys. Meth. 24, 63-70.

Vale, R. D., Reese, T. S. and Sheetz, M. P. (1985). Identification of a novelforce-generating protein, kinesin, involved in microtubule-based motility.Cell 42, 39-50.

Vale, R. D. and Hotani, H. (1988). Formation of membrane networks in vitroby kinesin-driven microtubule movement. J. Cell Biol. 107, 2233-2241.

Vale, R. D. and Goldstein, L. S. B. (1990). One motor, many tails: anexpanding repertoire of force-generating enzymes. Cell 60, 883-885.

Vale, R. D. (1992). Microtubule motors: many new models off the assemblyline. Trends Biochem. Sci. 17, 300-304.

Wright, B. D., Henson, J. H., Wedaman, K. P., Willy, P. J., Morand, J. N.and Scholey, J. M. (1991). Subcellular localization and sequence of seaurchin kinesin heavy chain: evidence for its association with membranes inthe mitotic apparatus and interphase cytoplasm. J. Cell Biol. 113, 817-833.

Yamada, H., Hayashi, H. and Natori, Y. (1984). A simple procedure for theisolation of highly purified lysosomes from normal rat liver. J. Biochem. 95,1155-1160.

Yu, H., Toyoshima, I., Steuer, E. R. and Sheetz, M. P. (1992). Kinesin andcytoplasmic dynein binding to brain microsomes. J. Biol. Chem. 267, 20457-20464.

(Received 24 March 1994 - Accepted 9 May 1994)

D. L. Marks, J. M. Larkin and M. A. McNiven