Proc. Natl. Acad. Sci. USAVol. 75, No. 8, pp. 3863-3866, August 1978Cell Biology

Association of mitochondria with microtubules in cultured cells(intracellular transport)

MICHAEL H. HEGGENESS, MELVIN SIMON, AND S. J. SINGERDepartment of Biology, University of California, San Diego, La Jolla, California 92093

Contributed by S. J. Singer, June 6, 1978

ABSTRACT By indirect immunofluorescence techniques,microtubules and mitochondria were localized in normal ratkidney cells, human WI38 fibroblasts, mouse peritoneal mac-rophages, and a putative smooth muscle rat cell line, in mono-layer culture. The mitochondria were found to be arrangedalong the cytoplasmic microtubules in each cell type. Disruptionof the microtubules with colcemid caused a redistribution ofthe mitochondria in these cells. There was no correlation be-tween the location of the mitochondria and actin-containingfilaments. This evidence suggests that mitochondria are directlyor indirectly associated with microtubules in these cells.

The control mechanisms that determine the location andmovement of intracellular organelles are not well understood.Microtubules have often been implicated in such phenomena,particularly in connection with membrane-bound intracellularorganelles. By now, the literature on this subject is too volu-minous to cite adequately; as examples, see refs. 1-6. In thepresent paper, we confine our attention to possible interactionsbetween microtubules and mitochondria. The most impressiveprevious evidence for the existence of such interactions hascome from electron microscopic studies of neuronal axons (7,8). In these elongated cell processes viewed in cross sections, theproximity of mitochondria to microtubules has been shownstatistically to be much greater than expected from a randomdistribution; furthermore, by heavy positive staining, whatappear to be cross bridges between microtubules and proximalmitochondria have been demonstrated. These studies, however,leave open the question whether these microtubular-mito-chondrial interactions are general to all cells or only to highlyspecialized axonal processes, as well as many other questionsrelating to the detailed mechanisms and possible functions ofsuch interactions. Furthermore, the thin-section electron mi-croscopic method for detecting such interactions is feasible onlywith oriented cell processes such as axons, and, at least intially,other methods are required for the more general case.

During studies from this laboratory on the fluorescencestaining of specific intracellular components, we have obtainedclear evidence at the light microscopic level of resolution foran association between microtubules and mitochondria inseveral unrelated types of cultured cells. This evidence is pre-sented in this paper.

MATERIALS AND METHODSAntitubulin Antibodies. Tubulin was prepared from 12-

day-old chick embryo brain by the disaggregation-reaggre-gation method of Shelanski et al. (9). The preparations that wereused for immunization had a relatively small amount of mi-crotubule-associated protein in addition to microtubule protein.The animals were immunized by a modification of the proce-

dure described by Van Vunakis et al. (10). The protein (1.0 mg)was mixed with an equal weight of methylated bovine serumalbumin and Freund's complete adjuvant and injected in thetoe pads and foot pads of New Zealand white rabbits. After 3weeks the rabbits were injected intramuscularly with tubulin(1.0 mg) in Freund's adjuvant. Three weeks later they wereinjected (1.0 mg of protein) intravenously and bled after an-other 7 days. The intravenous injection was repeated three moretimes until high-titer antibody was obtained.

In order to test the specificity of the antiserum, we preparedtubulin, free of microtubule-associated protein, by DEAE-Sepharose chromatography and by phosphocellulose chroma-tography (11, 12). The serum was absorbed at equivalence withthe purified tubulin. The absorbed serum showed little or noreaction when tested by complement fixation (13) against theinitial tubulin preparation (Fig. 1). The specificity of the an-titubulin reaction was further demonstrated by testing it againsta variety of other proteins. It gave no reaction against purifiedsmooth muscle myosin or chicken'gizzard filamin.

Cells and Cell Staining Methods. W138 and NRK cells weregrown as described (14). The putative smooth muscle rat cellline AIO (15) was obtained from David Shubert of the Salk In-stitute and was grown in Dulbecco's modified minimal essentialmedium supplemented with 10% fetal calf serum. Mouse per-itoneal macrophages were obtained from unstimulated BALB/cmice and maintained in the modified minimal essential me-dium supplemented with 5% fetal calf serum. All cells weremaintained at 370 in a humid atmosphere of 90% air/10%CO2.

Cells grown on coverslips were fixed and made permeableby a 3O-min treatment with 3% formaldehyde followed by a2-min exposure to 0.1% Triton X-100 as has been described (16).They were then stained by indirect immunofluorescence, withrabbit antibodies and rhodamine-conjugated goat anti-rabbit

100-IO-

< 80

z 60

40C.). 20

0.2 0.4 0.6 0.8 1.0pg PROTEIN

FIG. 1. Specificity of antitubulin antiserum. The reaction ofantiserum before and after absorption with microtubular proteinseparated from high molecular weight components was measured bycomplement fixation (13). The upper curve shows the reaction of theantiserum with tubulin (@). There was no reaction with chicken giz-zard filamin (0) and chicken muscle myosin (0). The absorbed an-tiserum showed no reaction with tubulin (X).

3863

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked"advertisement" in accordance with 18 U. S. C. §1734 solely to indicatethis fact.

3864 Cell Biology: Heggeness et al.

IgG. Rabbit anti-chick brain tubulin IgG (0.2 mg/ml) was usedto localize microtubules (16). Rabbit antiserum against beefheart cytochrome c oxidase (dilution 1/20), the gift of EfraimRacker, was used to localize the mitochondria. The specificityof this antibody has been described (17). As a control, normalrabbit serum was found to give only very low backgroundstaining.

Actin was localized by using biotinyl heavy meromyosin andfluorescein-avidin as described (14, 16). The specificity of thisstaining was established by its elimination byMg pyrophosphateor free biotin.

21;.u.1 NL{K celtltaineild for tuilltn. lii NRIK cell StK'vstalbsr cv'tochronwolx1(j:icse to visiialize ithe tnil(chondria. c NRK clisned Jiltaneoily wit 1itantiththa ulifn anti anti ctohro(rco(xidse. N(te ihat ihe ax is (Uf ehlminglottXr-it (ornria isll1ul\d

paral II tI it associated irit ol Ie. A.\ rroiw nd wictessla 1ItI1)cho'ndrion. id) Alti cell stane is 11 4 Mus' rphaoze -tamned

i1 ai(.1 . X7 1)(

RESULTS

When indirect immunofluorescent staining of the specificmitochondrial enzyme cytochrome c oxidase was carried outon NRK cultured fibroblasts, the distribution of the mito-chondria was revealed. There was a fairly dense concentrationaround the nucleus and an almost fibrillar array of mitochon-dria radiating towards the cell periphery (Fig. 2b). The distri-bution of mitochondria was strikingly similar to that observedwhen microtubules (18) were specifically stained by indirectimmunofluorescence (Fig. 2a) on other similar cells. The im-

Proc. Natl. Acad. Sci. USA 75 (1978)

Proc. Nati. Acad. Sci. USA 75 (1978) 3865

FIG. 3. (a) NRK cell stained with anticytochrome c oxidase to visualize the mitochondria after 3.5 hr of exposure to 0.25 fg of Colcemidper ml. Note that the mitochondria are clustered around the nucleus rather than dispersed through the cytoplasm as in Fig. 2b. (b) Nomarskiimage of the cell shown in a. (c) WI-38 cell stained for actin with biotinyl heavy meromyosin and fluorescein-avidin. (d) Same WI-38 cell shownin c stained for cytochrome c oxidase and tubulin using rhodamine antibodies as in Fig. 2. Note that in areas of the cell, such as the lower leftportion, where microtubules are largely absent, the'mitochondria are also absent, but actin is abundantly present. (X700.)

munofluorescent staining of both microtubules and mito-chondria was then carried out simultaneously on the same cellswith the same fluorochrome, rhodamine. This was possiblebecause the labeled mitochondria were easily distinguishedmorphologically from the labeled microtubules. In such fields(Fig. 2c), it was clear that in the peripheral regions of the cellsthe large majority, if not all, of the mitochondria were arrayedalong the microtubules with their long axes generally parallel

to the long axes of the microtubules. Essentially the same resultswere seen with human WI-38 fibroblasts (Fig. 3d), macro-phages (Fig. 2e), and A10 cells (Fig. 2d). If the microtubulesin these cells were first disaggregated by a 3.5-hr treatment withColcemid at 0.25 ,g/ml and then stained for cytochrome coxidase, the mitochondria were found to have retracted fromthe cell periphery to the region around the nucleus. An NRKcell after such treatment is shown in Fig. Sa. Macrophages

Cell Biology: Heggeness et al.

3866 Cell Biology: Heggeness et al.

treated similarly also showed a redistribution and someclumping of mitochondria. However, the drug treatment alsocaused dramatic shape changes in the macrophages, compli-cating the interpretation of this experiment.The possibility that actin-containing filaments might also be

associated with the mitochondria was explored by a triple-staining procedure. The cytochrome c oxidase and tubulinstaining was carried out simultaneously with rhodamine indi-rect immunofluorescence; on the same specimens the actin wasthen specifically stained with biotinyl heavy meromyosin andfluorescein-conjugated avidin.

Macrophages, which lack an extensive network of stress fi-bers, showed diffuse actin staining in all regions of the cell, andno correlation could be discerned between the location of themitochondria and microtubules on the one hand and that of theactin on the other (not shown). In the fibroblasts and the smoothmuscle cells, which contain a distinct array of actin-containingfibers of regular orientation, the distribution of the mitochon-dria was clearly independent of that of the actin (Fig. 3 c andd).

DISCUSSIONThese experiments demonstrate that in several different typesof cultured cells most, if not all, of the mitochondria are asso-ciated with cytoplasmic microtubules. This association is notadventitious because there is no indication of an association ofmitochondria with cytoplasmic actin in these cells. The asso-ciation of mitochondria with microtubules strongly suggests theexistence of some kind of specific linkage, either direct or in-direct, between them; no other reasonable explanation has oc-curred to us. It is hard to imagine, for example, that somenonspecific electrostatic interaction can be responsible for theassociation when so many other intracellular components (suchas actin) are also present in the immediate vicinity of the mi-tochondria and microtubules.

It is possible that the association of mitochondria with mi-crotubules is not direct, but is through a linkage to otherstructures associated with micotubules. For example, in somecases (19) intermediate filaments have been observed to formcylindrical arrays around microtubules in cultured fibroblasts;if the mitochondria were linked to the intermediate filments,they would thus appear to be associated with microtubules inour experiments at the resolution of the light microscope. Theelectron microscopic evidence, however, is that such an asso-ciation of intermediate filaments with microtubules is not aregular ultrastructural feature; microtubules are not alwaysobserved to be so surrounded. We are therefore inclined to theview that our observations of an association of mitochondriawith microtubules is due to a direct linkage between them. Thisview also draws on the electron microscopic observations ofSmith and his colleagues (7, 8) with neuronal axons, which havedemonstrated that cross bridges exist between apposed mito-chondria and microtubules in these cells. Such cross bridges,however, have not been shown to be present in the cultured cellsthat we have studied.The occurrence of this association of microtubules and mi-

tochondria in such diverse cells as fibroblasts, macrophages, andsmooth muscle cells and in neuronal axons makes it likely thatthe phenomenon is of general physiological significance. It maywell be that the intracellular distribution of mitochondria is

regulated by the association with microtubules. The fact thatthe distribution of mitochondria is markedly altered when themicrotubules are disrupted by colcemid treatment is consistentwith this view. Such regulation of the distribution of mito-chondria inside cells is likely to be important in controlling localconcentrations of ATP, divalent cations, and other cell const-itutents involved in mitochondrial metabolism. The observedmicrotubular-mitochondrial association may also be directlyinvolved in the intracellular movements of mitochondria. It hasbeen proposed that such movements involve the successivemaking and breaking of linkages, with the mitochondriamoving along the microtubules as along a track (8).

Associations between microtubules and other membrane-bound intracellular organelles have been proposed (20-22) andtheir possible existence could be explored by the appropriatespecific immunofluorescence techniques as used in thispaper.

This work was supported by U.S. Public Health Service GrantsGM-15971 and CA-22031 to S.J.S., and National Science FoundationGrant PCM 76-17197 to M.S. We thank Dr. Efraim Racker for thegenerous gift of anti-cytochrome c oxidase antibody. We are gratefulto Dr. Roger Carroll and Dr. J. F. Ash for helpful discussions and Dr.L. Scheffler for providing cell culture facilities. S.J.S. is an AmericanCancer Society Research Professor.1. Allen, R. D. (1975) J. Cell Biol. 64, 497-503.2. Tilney, L. G. & Porter, K. R. (1965) Protoplasna 60, 317-344.3. Bickle, D., Tilney, L. G. & Porter, K. R. (1966) Protoplasnw 61,

322-45.4. Murphy, D. B. & Tilney, L. G. (1974) J. Cell Biol. 61, 757-

779.5. Holmes, K. V. & Choppin, P. W. (1968) J. Cell Biol. 39,526-

543.6. Wagner, R. C. & Rosenberg, M. D. (1973) Cytobiologie 17,

20-27.7. Smith, D. S., Jirlfors, U. & Cameron, B. F. (1975) Ann. N.Y. Acad.

Sd. 253,472-506.8. Smith, D. S., JArIfors, U. & Cayer, M. L. (1977) J. Cell Sd. 27,

255-272.9. Shelanski, M. L., Gaskin, F. & Cantor, C. R. (1973) Proc. Nati.

Acad. Sci. USA 70,765-768.10. Van Vunakis, H., Kaplan, J., Lehrer, H. & Levine, L. (1966)

Immunochemistry 3,393-402.11. Murphy, D. B. & Borisy, G. G. (1975) Proc. Natl. Acad. Sd. USA

72,2696-2700.12. Himes, R. H., Burton, P. R., Kersey, R. N. & Pierson, G. (1976)

Proc. Natl. Acad. Sci. USA 73,4397-4399.13. Wasserman, E. & Levine, L. (1961) J. Immunol. 87,290-296.14. Heggeness, M. H. & Ash, J. F. (1977) J. Cell. Biol. 73, 783-

788.15. Kimes, B. W. & Brandt, L. (1976) Exp. Cell Res. 98,349466.16. Heggeness, M. H., Wang, K. & Singer, S. J. (1977) Proc. Natl.

Aazd. Scd. USA 74,3883-3887.17. Eytan, G. D., Carroll, R. C., Schatz, G. & Racker, E. (1975) J. B"d.

Chem. 250,859886.18. Brinkley, B. R., Fuller, G. NC & Highfield, D. P. (1975) Proc. Natl.

Acad. Sci. USA 72,4981-4985.19. Goldman, R. D. & Follett, E. A. C. (1969) Exp. Cell Res. 57,

263-276.20. Olmsted, J. B. & Borisy, G. C. (1973) Annu. Rev. Biochem. 42,

507-540.21. Porter, K. R. (1966) in Principles of Biomolecular Organization,

eds. Woistenholme, G. E. W. & O'Conner, M; J. and A. ChurchillLtd., London), pp. 308-343.

22. Snyder, J. A. & McIntosh, J. R. (1976) Annu. Rev. Biochem. 45,699-720.

Proc. Natl. Acad. Sci. USA 75 (1978)