Process of Dispersion and Fragmentation of Golgi Complexby

Process of Dispersion and Fragmentation of Golgi Complexby Microtubule Bundles Formed in Taxol Treated HeLa Cells Harumi Hoshino, Atsushi Tamaki, and Tatsuo Yagura* Laboratory of Life Science, Department of Chemistry, Faculty of Science, Kwansei Gakuin University, 1-1- 155, Uegahara, Nishinomiya-shi, Hyogo-ken 652, Japan

Keywords: Golgi complex/microtubule/taxol/immunoelectronmicroscopy

ABSTRACT.By means of a monoclonal antibody (mAbG3A5)against Golgi membrane glycoprotein, we have visualized the relative position of cytoplasmic polymerized microtubule bundles to Golgi stack cisternae in taxol treated HeLa cells, and found extensive fragmentation of the Golgi stack cisternae brought about by microtubule bundles. Within a 1 h period of taxol treatment, polymerization of cytoplasmic microtubules increased rapidly to form microtubule bundles, while the Golgi complex dispersed slightly along with the polymerized microtubule bundles. After 2 to 3 h taxol treatment the dispersal of the Golgi complex from the microtubule-organizing center (MTOC)to cytoplasmic periphery rapidly progressed in the direction which the microtubules ran. At early dispersal, the microtubule bundles were oriented apart from the stretched end of the Golgi stack cisternae and exhibited little direct contact with the Golgi stack cisternae membrane. The Golgi stack cisternae then began to wind around the microtubule bundles, followed by the beginning of fragmentation of the Golgi stack cisternae. At this step, some of the microtubules seemed to attach to a part of the Golgi stack cisternae. After prolonged exposure of cells to taxol (25 h) the microtubule bundles were highly developed throughout the cells and most of the Golgi fragments were trapped at their termini. In these cells, extensive fragmentation of Golgi stack cisternae occurred, resulting in small Golgi vesicles bound to the microtubule bundle.

The Golgi complex is composed of several stacks ar- membraneand the reassembly of fragment vesicles into ranged in a circumscribed juxtanuclear region to which Golgi stacks. The transport of newly synthesized pro- the microtubule-organizing center (MTOC) localizes teins through the Golgi cisternae is maintained in cells (19). The association of the Golgi complex and microtu- in which microtubules are depolymerized and the Golgi bules has been demonstrated often in several systems (5, stacks are dispersed (6). These observations have sug- 14, 15), with microtubules being instrumental in the gested that someunknownmechanismis at work to orderly membrane traffic through Golgi elements (12). maintain the integrity, location and function of the The candidate proteins for mediating the interaction be- Golgi complex in addition to or independent of the func- tween the Golgi complex and microtubules have also tion of microtubules. been identified, and have been shownto be associated Wemade a highly specific monoclonal antibody with the cytoplasmic face of Golgi membranes (1, 3). (mAbG3A5) to Golgi membrane glycoprotein (pi38), On the basis of these and several other observations (9, which we used previously to reveal immunocytochemi- 19), microtubules are thought to play a role in maintain- cally the ultrastructure of Golgi vesicles in mitotic cells ing the integrity and location of the Golgi complex in in- (2, 24). In this study, we present a detailed immunocyto- terphase cells. However,although microtubules run in chemical analysis of the mechanismfor the microtubule the vicinity of the Golgi complex (23), ultrastructural bundle mediated intracellular disposition of elements of studies have not revealed the binding of the Golgi cister- the Golgi complex in taxol treated cells using the mAb nae to microtubules (18, 23), such as that shown for mi- G3A5antibody. Taxol, which stabilizes microtubule as- crotubules and endoplasmic reticulum in taxol treated sembly (17), causes the Golgi complex to move toward cells (ll, 17). Recently, Veit et al (21) have shown the minus end of the microtubule bundle (15). Weexam- the microtubule-independent breakdown of the Golgi ined the dispersal event of the Golgi complex through stabilization of microtubule assembly and pursued the fate of the Golgi stack cisternal membrane in taxol * To whomcorrespondence should be addressed. Tel: 81-798-54-6406, Fax: 81-798-51-0914 treated HeLa cells by immunoelectron microscopy us- e-mail: [email protected] ing the mAbG3A5antibody. Our results extend those Abbreviation: MTOC;microtubule-organizing center. of previous studies by showing that the dispersal of the 325 H. Hoshino et al.

Golgi complex is initiated later than the formation of microtubule bundles and that prolonged treatment of taxol breaks the Golgi stack cisternae into small vesicles that bind to the microtubule bundle. Moreover, we showthat microtubule bundles run in the vicinity of the Golgi stack cisternae but that they have only slight contact with the Golgi stack cisternae during early taxol treatment. The Golgi stack cisternae, however, are being stretched by microtubule bundles, suggesting some unknownstructure mediatingthese twostructures.

MATERIALS AND METHODS Reagents. Taxol was obtained from WakoChemical Co.. Stock solution of taxol was prepared at 10 mMin dimethyl sul- foxide and stored in small aliquots at -40°C. Cell Culture. HeLa cells were grown in Eagle's minimum essential medium supplemented with 5%fetal calf serum. In experiments using taxol, the concentration of dimethyl sulfox- ide added along with the drug as a solvent to the culture was 0. 1%(final concentration). Cells were cultivated with the drug at 37°C before preparation as indicated. Antibodies. Hybridoma cell line (mAbG3A5) that se- creted monoclonalantibody against pi38 of humanorigin was prepared as described previously (24) and purified by chro- matography on columns of affi-Gel protein A. Rabbit antibody to tubulin was obtained from ICN Biomedicals, Inc.; FITC-conjugated rabbit anti-mouse IgG antibody was from DAKOA/S; Rhodamine-conjugated donkey anti-rabbit IgG antibody was from CHEMICONInternational Inc. Immunofluorescencemicroscopy and immunoperoxidase microscopy. The preparations were madeby the methods described previously (2). Specimens were observed and photo- graphed with an Olympusfluorescence microscope for immunofluorescence microscopy, and ultrathin sections were examined with a JEOL JEM-100S electron microscope. Western immunoblotting of tubulins. Quantification of microtubule polymer in taxol treated HeLa cells was carried out principally by the method described by Liao et al. (10). HeLa cells were extracted with PEM(100 mMPipes, 10 mM EGTAand 2 mMMgCl2, pH 7.0) containing 0.5 mg/ml saponin for 1 minute after drug treatment, harvested with a scraper, washed 2 times by centrifugation, and then prepared for SDS-PAGE.Blots were performed as described previous-

Fig. 1. Patterns of distribution of the Golgi complex in taxol treated cells. The distribution of the Golgi complex in cells treated with 10 ^Mtaxol was examined with indirect immunofluorescence microscopy after labeling with mAbG3A5antibody. The distribution of the Golgi complex was classified according to their fluorescence patterns: (A) pattern 1, no change, similar to those observed in untreated control cells; (B) pattern 2, a portion of the Golgi stack cisternae are being pulled out and fragmenting; (C) pattern 3, entire Golgi stack cisternae are extended and fragmented; (D) pattern 4, fragmented Golgi Fig. 1. stack cisternae are dispersed to cell periphery. The bar indicates 10 ftm.

326 Dispersion and Fragmentation of Golgi by Taxol ly (8) using anti-tubulin antibody and alkaline phosphatase conjugated-second antibody.

RESULTS S80 /

Microtubule polymerization precedes dispersal of \560- / Golgi complex during taxol treatment. Treatment of f\ cells with taxol for 3 h results both in polymerization of !O / / microtubule bundles in the cell periphery and dispersal 2 40- / of the Golgi complex (16). In HeLacells treated with taxol we could classify several patterns of distribution I£20-/ / of the Golgi complex within the cytoplasm. Fig. 1 sum- marizes the patterns revealed by immunofluorescence 1 / procedure using mAbG3A5antibody and, in Fig. 2, the 0« . , å , . change of patterns over time during taxol treatment. 0 2 4 6 Pattern 1 is the same as the Golgi complex in non- Incubation Time After Addition of Taxol (h) treated control cells, and the Golgi complex in pattern 2 is not dispersed from its MTOCregion but a part of it is Fig. 3. Formation of microtubule bundles in cells during taxol treat- being pulled towards the outer cytoplasm. The total ment. Cells exhibiting apparent microtubule bundles in their cyto- number of cells exhibiting patterns 3 and 4, in which the plasm were counted during the course of taxol treatment (10 fiM) us- Golgi complexes are obviously dispersed in the cyto- ing immunofluorescence microscopy. plasm from the MTOC,gradually increased with time tion of tubulin over time during taxol treatment. First, anti-tubulinwe employed antibody.immunofluorescentA plot of themicroscopyratio of theusingcellsan antibodywith microtubulevs. time bundlesperiod ofvisualizedtaxol treatmentwith anti-tubulinis shown in period of taxol treatment. The ratio of cells exhibiting 70%Fig. of3.theMicrotubulecells at 1bundlesh after weretaxol observedaddition, inwithmorealmostthan withall oftaxol.the cellsNext,exhibitingthe changebundlesin theafteramountlongerof treatmentpolymer- pattern 4, in which the Golgi complex is entirely dis- ofizedcellsmicrotubulesand Westernwasblottingexaminedprocedureby saponinusing anti-tubu-treatment persed to the cell periphery, the same feature described by Sandoval et al. (16) in A549 cells, reached a maxi- mumat 2 h taxol treatment and increased little there- after. As shown in Fig. 2, the dispersion of the Golgi complex began 2 to 3 h after taxol addition in most cells and resulted in the distribution which falls under the cat- egory of patterns 3 and 4. This unexpectedly slow dispersal of the Golgi complex upon taxol treatment made us examinethe degree of bundle formation or polymeriza-

1 2 3 4 5 6

Fig. 4. Change in polymerized microtubules detected by Western 0. 0 1 2 3 4 5 6 immunoblotting with anti-tubulin antibody. Cells were pretreated Incubation Time After Addition of Taxol (h) with saponin and processed for SDSgel electrophoresis. Tubulin blot- ted to the membranewas detected with anti-tubulin antibody. Lane 1 , Fig. 2. Change in the ratio of cells exhibiting the Golgi complex of control cells treated with neither taxol nor saponin weresonicated, each pattern during taxol treatment. The pattern of Golgi distribution thus representing total tubulin containing free non-polymerized tubu- was determined in cells prepared for immunofluorescencemicroscopy lin molecules; lane 2, cells with no taxol treatment but pretreated with using mAbG3A5antibody at the time indicated during taxol treat- saponin; lane 3, cells with 1 h treatment of taxol (10 fiM); lanes 4, 5 ment (10 fjiM). Classification of each type was according to that and 6 are 2 h, 3 h and 5 h treatments with taxol, respectively. Arrow shown in Fig. 1. head indicates tubulin of 55 kDa.

327 H. Hoshino et al. lin antibody. The results are shown in Fig. 4. Free non- gation (lane 2). A small amount of polymerized tubulin polymerized cytoplasmic tubulin existing in cell cyto- was present initially in cells before taxol treatment (lane plasm (lane 1) was completely extracted from the cell 2), but polymerized tubulin increased rapidly in cells pellet after saponin treatment and washings by centrifu- treated with taxol and reached a maximumwithin 3 h of treatment (lanes 3-6). The results shown in Figs. 3 and 4 indicate that the polymerization of tubulin and formation of microtubule bundle are parallel events. These results also show, if compared with those shownin Fig. 2, the apparent retardation of dispersal of the Golgi complex from the time of formation of microtubule bundles, especially at the 1 h period of taxol treatment. To determine if the Golgi complex associates with the microtubule bundles, double-label immunofluorescence microscopy with the mAbG3A5antibody and the antibody recognizing a-tubulin was performed as shown in Fig. 5. At 1 h incubation, the Golgi complex remained localized near the MTOCand did not associate with the running microtubule bundles in most cells (Fig. 5A). At this time the microtubule bundles were formed thor- oughly in the cytoplasm (Fig. 5B, arrows), in accord- ance with the results shown in Fig. 3. Next at 2 h incubation period, the Golgi complex began to be dispersed in the direction of running microtubule bundles (Figs. 5C and D). The cells exposed to prolonged incubation with taxol (25 h) showed quite different images; microtubules were highly polymerized and formed several wide bundles, with the Golgi complexes gathered at one end

Fig. 5. Double-immunofluorescence micrographs of the Golgi complex and microtubule bundle in taxol treated cells. The Golgi complex and microtubule bundle of HeLacells were double stained with mAbG3A5antibody (FITC fluorescence; A, C, E) and anti-tubulin antibody (rhodamine fluorescence; B, D, F) after 10 j«M taxol treatment Fig. 6. Immunoelectron microscopic appearance of Golgi stack for 1 h (A, B), 2h (C, D) or 25h (E, F). The distribution of the pri- cisternae stained with mAbG3A5antibody. Golgi stack cisternae mary antibodies was observed with species-specific FITC- and rhoda- of HeLacells were examined by immunoelectron microscopy as de- mine-conjugated secondary antibodies. Arrows in B and D indicate scribed under Materials and Methods. Sections were examined with the microtubule bundles. The bar indicates 10 //m. no further heavy metal staining. The bar indicates 0.1 /im.

328 Dispersion and Fragmentation of Golgi by Taxol of these wide microtubule bundles (Figs. 5E and F). eral region as seen in a 3 h taxol treated cell. A part of Dispersing Golgi complex does not contact directly the Golgi stack structure seems to be beginning to break with microtubule bundles during early period of taxol down. In addition, although the pi38 was present in the treatment. With immunofluorescence microscopy we Golgi cisterna membraneregion as well as observed in have often observed a part of the Golgi stack cisternae non-treated control cell (see Fig. 6), the distribution of being pulled along with running microtubule bundles. the pi38 became uneven in dispersed Golgi stack cister- These images suggest that the microtubule bundles offer nae, suggesting someunknownalteration occurring in a motive force for the dispersion of the Golgi complex. the fine structure of Golgi cisterna membrane.Microtu- To know the character of this force, an immunoelec- bule bundles were observed in the vicinity of such Golgi tron microscopic study using mAbG3A5antibody was stack cisternae running parallel to the long axis of Golgi done. As shown in Fig. 6, the pl38, labeled with stack cisternae. However, wegot no obvious evidence mAbG3A5antibody, is distributed evenly in the Golgi of direct contact between the Golgi stack cisternae and stack cisterna membranebut is not detected in the Golgi microtubule bundles upon survey of a number of dis- stack cisterna lumen. Thus we expected to be able to persed Golgi complexes although their positions were at pursue the fate of Golgi complex stack cisternae and ex- closer proximity than those seen in stretched Golgi amine their detailed interaction with the microtubule stack cisterna. In another example of a 3 h taxol treated bundles during taxol treatment. cell (Fig. 8B), a portion of the Golgi stack cisternae obvi- Fig. 7 shows an example of stretched Golgi stack cis- ously wound itself around the microtubule bundle (indi- terna visualized by immunoelectron microscopy of a 3 h cated by star mark) and in an other part microtubules taxol treated cell. A few microtubule bundles are seen were seen running in close lateral apposition to the running in the region surrounding one end of stretched Golgi stack (arrows). The Golgi stack cisternae were Golgi stack cisterna (see positions marked by the gear in brought closer to the microtubule bundles in this state. A and stars on the enlarged photograph of B). How- This type of image, in whichthe cisternae are wound ever, unexpectedly, on such preparations examined we around the microtubule bundles, was frequently ob- observed little direct contact between them. Next, we ex- served in 3 h taxol treated cells. amined the relationship between dispersed Golgi stack Prolonged taxol treatmen t causes extensive fragmen- cisternae and microtubule bundles. Fig. 8Ashows an ex- tation of Golgi cisternae to vesicles and results in con- ample of the Golgi complex dispersed to the cell periph- tact between fragmented vesicles and microtubule bun-

Fig. 7. Stretched Golgi cisterna of taxol treated HeLa cell. HeLacells were treated with 10 //M taxol and prepared for immunoelectron microscopy as for Fig. 6. OneGolgi cisternae is stretched from Golgi stack in A. The enlarged image (B) of the area marked with a gear in A shows several microtubule bundles in the vicinity of one end of the stretched Golgi cisternae (indicated by stars). Small arrows in B indicate that these microtubules run in lateral apposition to the Golgi cisterna. The bars indicate 1 ftm (A) and 0.1 [im (B).

329 H. Hoshino et al.

Fig. 8. Positional correlation between the Golgi stack cisternae and microtubule bundles. HeLa cells were treated with 10 //M taxol for 3 h and then processed for immunoelectron microscopy as with Fig. 6. Most of the Golgi complex was dispersed into the cytoplasm from a juxtanuclear position. (A): The Golgi complexat the cell's peripheral region; the star indicates the cross section of one large microtubule bundle, and the arrows indicate small microtubule bundles running in lateral apposition to the long axis of the Golgi stack (small arrows, cross sections; large arrows, longitudinal sections). (B): The Golgi complex whosestack cisternae wind around the microtubule bundle; the star indicates the cross section of microtubule bundle and the arrows indicate the microtubule bundle running in lateral apposition to the Golgi cisternae (longitudinal sections). In both micrographs the localization of pi38 labeled with mAbG3A5antibody in the cisterna membraneappear uneven and a part of stack cisternae have begun to fragment. The bars indicate 1 ^m.

330 Dispersion and Fragmentation of Golgi by Taxol

Fig. 9. Fragmented Golgi cisternae attached to the microtubule bundle. Immunoelectron micrographs of the fragmented Golgi cisternae of a HeLa cell treated for 25 h in 10 //M taxol are shown. Sections were post-stained with uranyl acetate and lead citrate. In A, a part of the Golgi stack (indicated by white arrows) migrated to one end of the microtubule bundle. In B, the Golgi cisternae were fragmented into numerous vesicles and decorated over the microtubule bundle. Arrows and arrowheads indicate the microtubule bundles and Golgi vesicles, respectively. The bars indicate 0.1 /mi.

331 H. Hoshino et al. dies. As shown in Figs. 5E and F, in the 25h taxol of the Golgi complex from the polymerization of micro- treated cells the dispersal of the Golgi complex pro- tubule or growth of microtubule bundle in the taxol ceeded extensively and the Golgi complex accumulated treated cells. As to account for this time lag, it has been at one end of the wide microtubule bundles. In these reported that the pre-existing microtubules around the cells, the Golgi complex seems to be in closer contact to MTOCare not stable and gradually disappear in taxol the microtubule bundles in comparisonto the cells from treated cells although massive assembly of cytoplasmic earlier periods of taxol treatment. Wewanted to exam- free tubulins occurs (4). This positional effect of taxol is ine the contact of these two structures in 25 h taxol explained by the low assembly threshold for polymeriza- treated preparation in more detail, thus immunoelec- tion of microtubule in the MTOC.In taxol treated cells, tron micrographs of these cells were made which the formation of multiple asters was observed during showedseveral phases of interaction betweenthe two the mitotic phase although most cells did not contain structures. In Fig. 9Aa part of the Golgi stack is seen a centrosome suggesting the formation of secondary trapped at one end of a microtubule bundle (white ar- MTOCs(4). Taking these matters into consideration, rows). In the same image, small vesicles labeled with the time lag observed in the present study mayrepresent mAbG3A5antibody appear bound to the outer walls of the time necessary for the disappearance or split out of microtubules (arrowheads). In another image the Golgi the MTOC.After the disappearance or split out of the stack structure was no longer seen and instead many MTOC,the mechanism maintaining the Golgi complex mAbG3A5labeled small vesicles decorate the surface of structure might be disassembled, causing the Golgi com- the microtubules (Fig. 9B, arrowheads). plex to be guided by microtubule bundles to disperse throughout the cytoplasm. DISCUSSION It is of much interest that the microtubule bundles did not directly bind to the Golgi stack cisternae during Illustrated in Fig. 10 is a model for the fragmentation the initial phase of Golgi dispersal, while the cisternae and association of the Golgi stack cisternae with a mi- were apparently stretched in the direction the microtu- crotubule bundle during taxol treatment which inte- bules wererunning. This fact suggests the existence of grates the principal findings of the collective data pre- an unknownmediator between the two structures. Evi- sented in this study. In the first step, a part of the Golgi dence from the existing literature suggests that at least a stack is stretched by the microtubule bundle without part of the function and maintenance of the structure of any direct contact. The elongated Golgi stack cisternae the Golgi complex operates independently of microtu- next winds around the microtubule bundle as the stack bules (13, 18, 20, 21, 22). Based on results obtained by is being fragmented. After prolonged taxol treatment a detailed kinetic analysis of the Golgi complex fragmen- part of the Golgi stack cisternae are shifted to one end tation under reduced temperature, metabolic inhibition of the microtubule bundle and further fragmented into and ionic perturbation, Turner and Tartakoff (20) indi- vesicles which dot the surface of the microtubule bun- cated that the processes of microtubule depolymeriza- dle. The Golgi vesicles seem to bind to the microtubule tion and Golgi complex fragmentation are separate. at this stage. On the other hand, Tassin et al. (18) observed only a Wedetected an apparent time lag before dispersion minimal effect on Golgi complex localization by microtubule depolymerization initiated by nocodazole in myo- tubes. Fromtheses studies, wecan speculate on the existence of "stacking templates" (20) of non-microtubule material by which the Golgi complex organization is pri- marily maintained. This type of unknownstacking tem- plate mayalso mediate some kind of structural link between the Golgi complex and microtubules. Golgi stack cisternae may be stretched in taxol treated cells by microtubule bundles through a putative "stacking structure". Onecandidate for such a structure maybe the cross-bridge structure shownat the gap betweencyto- plasmic organelles and the microtubules in Allogromia laticollaris (7). However, we did not find such a structure in this study. The Golgi stack cisternae dispersed by tubulin bundles were finally fragmented and seemed to bind directly Fig. 10. Schematic model for the fragmentation and association of to the microtubules. This observation suggests that the the Golgi stack cisternae with microtubule bundle. stacking structure was gradually destroyed during the 332 Dispersion and Fragmentation of Golgi by Taxol

fragmentation of Golgi stack cisternae. In this context, 78: 5608-5612. it is interesting that the distribution of pi38 in the Golgi 5. Ho, W.C., Allan, V.J., van Meer, G., Berger, E.G., and Kreis, T.E. 1989. Reclustering of scattered Golgi elements oc- cisternae membranebecameuneven upontaxol treat- curs along microtubules. Eur. J. Cell Biol., 48: 250-263. ment when examined by immunoelectron microscopy. 6. Iida, H. and Shibata, Y. 1991. Functional Golgi units in mi- This lack of uniformity seemedto occur soon after the crotubule-disrupted cultured atrial myocytes. /. Histochem. beginning of dispersal of the Golgi stack cisternae while Cytochem., 39: 1349-1355. their stack structure still remained intact (see Fig. 8). 7. Kachar, B., Bridgman, P.C., andReese, T.S. 1987. Dynam- This change in the distribution of pi38 in the Golgi ic shape changes of cytoplasmic organelles translocating along membranestructure might be a presage of the break- microtubules. /. Cell Biol, 105: 1267-1271. 8. Kozu, T., Seno, T., and Yagura, T. 1986. Activity levels of downof Golgi stack cisternae to vesicles which occur- mouse DNApolymerase a, free from primase activity in syn- red in cells after prolonged taxol treatment; the mem- chronized cells, and a comparison of their catalytic properties. brane units that remain pi38 are thought to be cut out Eur. J. Biochem., 157: 251-259. of the vesicles attached to the microtubule bundle be- 9. Kreis, T.E. 1990. Role ofmicrotubules in the organization of cause such a mosaic distribution of pi38 protein is no the Golgi apparatus. Cell Motil. and Cytoskelet. , 15: 67-70. longer observed in the vesicles (see Fig. 9B). Alterations 10. Liao, G., Nagasaki, T., and Gundersen, G.G. 1995. Low in the Golgi stack cisterna fine structure might affect the concentrations of nocodazole interfere with fibroblast locomo- functional relations among the Golgi stack, stacking tion without significantly affecting microtubule level: Implica- tions for the role of dynamic microtubules in cell locomotion. /. structure, and microtubules. Thus, we conjecture that Cell ScL, 108: 3473-3483. the fragmenting cisternae mayescape from the control ll. Masurovsky, E.B., Peterson, E.R., Crain, S.M., and of the stacking structure, bind to a microtubule, and Horwitz, S.B. 1981. Microtubule arrays in taxol-treated then be transported to the minus ends of microtubule mouse217: dorsal392-398. root ganglion-spinal cord cultures. Brain Res., bundles seen in cells treated with taxol for a prolonged 12. Pastan, I.H. andWillingham, M.C. 1981. Journeytothecen- time period. ter of the cell: role of the receptosome. Science, 214: 504-509. Based on the experiments using taxol our results 13. Rabouille, C, Misteli, T., Watson, R., and Warren, G. showing the detailed process of dispersion and fragmen- 1995. Reassembly of Golgi stacks from mitotic Golgi frag- tation of Golgi stack cisternae are of particular interest ments in a cell-free system. /. Cell Biol., 129: 605-618. whenconsidering the mechanismfor maintaining the 14. Robbins, E. and Gonatas, N.K. 1964. Histochemical and Golgi stack structure in the MTOC.Moreover, the role structural studies on HeLacell cultures exposed to spindle inhib- itors with special reference to the interphase cell. /. Histochem. of pi38 in the Golgi stack cisternae will be a target for Cytochem., 12: 704-711. further study on the mechanismthat wesuggest here. 15. Rogalski, A.A. and Singer, S.J. 1984. Association of ele- Experiments are presently underwayto clone the gene ments of the Golgi apparatus with microtubules. /. Cell Biol., for pi38, in which connection we have thus far deter- 99: 1092-1100. mined about 70%of its nucleotide sequence. At present 16. Sandoval, I.V., Bonifacino, J.S., Klausner, R.D., Henkart, no homology has been found with sequences reported M., and Wehland, J. 1984. Role ofmicrotubules inthe organ- elsewhere. ization and localization of the Golgi apparatus. /. Cell Biol. , 99: 113s-118s. Acknowledgments.Part of this work was performed under the 17. Schiff, P.B. and Horwitz, S.B. 1980. Taxol stabilizes micro- Frontier Research Project in Telecommunications supported by Minis- tubules in mouse fibroblast cells. Proc. Natl. Acad. Sci. USA, try of Posts and Telecommunications and the Support Center for Ad- 77: 1561-1565. vanced Telecommunications Technology Research. 18. Tassin, A.M., Paintrand, M., Berger, E.G., and Bornens, M. 1985. The Golgi apparatus remains associated with microtubule organizing centers during myogenesis. /. Cell Biol. , 101: REFEREN CES 630-638. 19. Thyberg, J. and Moskalewski, S. 1985. Microtubules and Allan, V.J. and Kreis, T.E. 1986. A microtubule-binding pro- the organization of the Golgi complex. Exp. Cell Res., 159: 1- tein associated with membranes of the Golgi apparatus. /. Cell 16. Biol., 103: 2229-2239. 10. Turner, J.R. and Tartakoff, A.M. 1989. The response of Asada, S. and Yagura, T. 1995. Golgi membrane vesicles in the Golgi complex to microtubule alterations: The roles of meta- HeLa mitotic cells are identified with monoclonal antibody bolic energy and membranetraffic in Golgi complex organiza- made against Golgi cisternal membraneprotein pi38. Cell tion. /. Cell Biol., 109: 2081-2088. Struct. Func , 20: 445-453. ll. Veit, B., Yucel, J.K., and Malhotra, V. 1993. Microtubule Bloom, G.S. and Brashear, T.A. 1989. A novel 58-kDa pro- independent vesiculation of Golgi membranes and the reassem- tein associates with the Golgi apparatus and microtubules. /. bly of vesicles into Golgi stacks. /. Cell Biol., 122: 1197-1206. Biol. Chem., 264: 16083-16092. 12. Wehland, J. and Sandoval, I. 1983. Cells injected with gua- DeBrabander, M., Geuens, G., Nuydens, R., Willebrords, nosine 5'-[a,/3-methylene]triphosphate, an a,/3-nonhydrolyzable R., and De May, J. 1981. Taxol induces the assembly of free analog of GTP, show anomalous patterns of tubulin polymeriza- microtubules in living cells and blocks the organizing capacity of tion affecting cell translocation, intracellular movement, and the the centrosomes and kinetochores. Proc. Natl. Acad. Sci. USA, organization of Golgi elements. Proc. Natl. Acad. Sci. USA,

333 H. Hoshino et al.

80: 1938-1941. 24. Yamauchi, T., Higashiura, M., and Yagura, T. 1992. Novel 23. Wehland, J., Henkart, H., Klausner, R., and Sandoval, monoclonal antibody mAbG3A5recognizes 138-kDa glycopro- I.V. 1983. Role of microtubules in the distribution of the tein localized on the Golgi membrane. Cell Struct. Func, 17: Golgi apparatus: Effect of taxol and microinjected anti-a-tubu- 213-222. lin antibodies. Proc. Natl. Acad. Sci. USA, 80: 4286-4290. (Received for publication, February 15, 1997 and in revised form, March 24, 1997)

334