Control of Microtubule Nucleation and Stability in Madin-Darby Canine Kidney Cells: The Occurrence of Noncentrosomal, Stable Detyrosinated Microtubules Marie-Hdlbne Br~, T h o m a s E. Kreis, a n d Eric Karsenti European Molecular Biology Laboratory, D-6900 Heidelberg, Federal Republic of Germany Centrosomal and noncentrosomal micrombules regrew in MDCK ceils with similar kinetics after release from complete disassembly by high concentrations of nocodazole (33 ~tM). During regrowth, centrosomal microtubules became resistant to 1.6 I~M nocodazole before the noncentrosomal ones, although the latter eventually predominate. We suggest that in MDCK cells, microtubules grow and shrink as proposed by the dynamic instability model but the presence of factors prevents them from complete depolymerization. This creates seeds for reelongation that compete with nucleation off the centrosome. By using specific antibodies, we have shown that the abundant subset of nocodazole-resistant microtubules in MDCK ceils contained detyrosinated a-tubulin (glu tubulin). On the other hand, the first microtubnles to regrow after nocodazole removal conrained only tymsinated tubulin. Glu-tubulin became detectable only after 30 min of micmtubule regrowth. This strongly supports the hypothesis that a-tubulin detymsination occurs primarily on "long lived" microtubules and is not the cause of the stabilization process. This is also supported by the increased amount of glu-tubulin that we found in taxol-treated cells. EVERAL lines of evidence support the view that microtubules are involved in cytoplasmic organization and orientation of intracellulartransport. Together with micmtubule-organlzing centers (MTOCs) 1 they are thought to determine the axis of polarity in many ceils (8, 24, 32). Although it is known that micmtubules are polymers in a dynamic equilibrium with a pool of soluble subunlts (17), the intrinsic dynamic characteristics of pure micmmbules have been studied only recently in vitro and called dynamic instability (16, 20, 25, 26). Microinjection of labeled tubulin or microtubule-associated proteins (MAPs) into living cells has also shown that most cytoplasmic microtubules in fibroblast MAP, microtubule-associatedprotein; MTOC, microtubulc-organizing center; tyr-tubulin,tyrosinateda-tubulin. cells turn over with a half-lifeof '~I0 min (30, 31, 34). A smaller fraction of the microtubules is very stable with a half-lifeof several hours (34). The dynamic instabilityof microtubules has severalpotentiallyimportant implications for understanding how differentsubclasses of micmmbule networks are generated during the cell cycle and cell differentiation.Kirschner and Mitchison propose that during morphogenetic processes (polarizationor any sort of cell shape remodeling) microtubules, although globally dynamic,become more stablein a restrictedarea of the cell,thereby providing a basis for cytoplasmic asymmetry (20). This selectivestabilizationhypothesis predicts that: (a) microtubules with differentstabilitycan coexist in a given cell; (b) microtubule-stabilizing factors could be distributed nonhomogeneously in the cytoplasm; and (c) external signals 9 The Rockefeller University Press, 0021-9525/87109/1283/14 $2.00 The Journal of Cell Biology, Volume 105, September 1987 1283-1296 1283 1. Abbreviations used in this paper: glu-tubulin, detymsinated a-tubulin; Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Abstract. The micmtubnle-nucleating activity of centrosomes was analyzed in fibroblastic (Vero) and in epithelial cells (PtK2, Madin-Darby canine kidney [MDCK]) by double-immunofluorescence labeling with anti-centrosome and antitubnlin antibodies. Most of the micmtubules emanated from the centrosomes in Vero ceils, whereas the micmtubule network of MDCK cells appeared to be noncentrosome nucleated and randomly organized. The pattern of micmtubule organization in PtK2 cells was intermediate to the patterns observed in the typical fibmblastic and epithelial cells. The two centriole cylinders were tightly associated and located close to the nucleus in Veto and PtK2 cells. In MDCK cells, however, they were clearly separated and electron microscopy revealed that they nucleated only a few microtubules. The stability of centmsomal and noncentrosomal micmtubules was examined by treatment of these different cell lines with various concentrations of nocodazole. 1.6 gM nocodazole induced an almost complete depolymerization of micmtubules in Vero ceils; some centrosome nucleated micmtubnles remained in PtK2 cells, while many noncentrosomal micmtubules resisted that treatment in MDCK ceils. Materials and Methods Cells Madin-Darby canine kidney epithelial cells (MDCK) strain II were grown in Eagle's minimal essential medium with Eagle's salts supplemented with 10 mM Hepes pH 7.3, 1% L-gintamine, 5% FCS, penicillin (110U/mt), and streptomycin (100 l~g/ml). African green monkey kidney cells (Vero) and Potoroo kidney epithelial ceils (PtK2) were grown as previously described (22). The cells were seeded on glass coverslips and incubated in a humidified atmosphere equilibrated with 5% CO2 in air at 37~ Drug 1~eatments Cells were seeded at a density that normally allows conflncncy between day 2 and 3 (36). All hhe experiments described in this paper were carried out on subconfluent MDCK cells taken 24 h after seeding. Preliminary experiments indicated that the nocodazole concentration required for intermediate de.polymerization in MDCK cells was 1.6 I~M, To induce an almost total depolymerization, a concentration of 33 ttM was required for at least 3 h. Nocodazole (Sigma Chemical GmbH, Deisenhofen, Federal Republic of Germany) was kept as a stock solution in DMSO at -20"C (33 mM) and diluted in culture medium at the appropriate concentration just before use. Experimental protocols are described in the figure legends. Cells were treated at 37~ with taxol at 5 ~tM using a stock solution in DMSO (10 mM) kept at -20~ (Taxot was supplied by Dr. M. Suffness, Natural Product Branch, Division of Cancer Treatment, National Cancer Institute, Betbesda, MD.) Experimental protocols are described in the figure legends. Antibodies Human anti-centrosome antibodies (505t) are a gift of "1".Mitchison and M. Kirschner (26). lmmunofluorescence All the results presented here have been obtained on cells preextracted for 10 s in microtubule-stabilizing medium (80 mM K-Pipes pH 6.8, 5 mM EGTA, 1 mM MgCI2, 0.5% Triton X-100) and fixed in methanol at -20~ for 5 rain as previously described (19). In control experiments, cells were fixed directly in methanol at -20*(2. These and other fixation procedures involving the use of glutaraldehyde combined with extraction buffer gave the same results. ~ t r a c t i o n is useful because the backgrmmd due to free tubulin subunits is elinfinated, making the images sharper. Immunofluorescence was carried out as previously described (19) including 0.1% Triton X-100 in the PBS washes. Coverslips were covered with 20 ktl rabbit anti-glu-tubalin (1:10 dilution) for 10 min. 20 ttl of a monoclonal anti-tyrtubulin (1:400) were added for a further 10 min incubation, Directly mixing the two antibodies gave similar results. After three washes (5 rain each), the coverslips were covered for 10rain with 20 ~tl of fluorescein-labeled goat anti-rabbit and Texas Red-labeled goat anti-mouse antibodies (1:I00 dilution; Dianova GmBH, Hamburg, FRG). After further washes, the coverslips were cleaned quickly in ethanol and mounted in Mowiot. In some experiments, cells were double stained with a monocloual mouse anti-t3-tubulin 0:500 dilution) and a human anti-centrosome antibody (1:100). In this case, secondary antibodies were a goat anti-mouse labeled with fluorescein (1:50) and a goat anti-human labeled with Texas Red (1:100; Dianova GmBH). All antibodies were diluted in PBS containing 0.1% Triton X-100, 3% BSA, and 0.1% sodium azide. lmmunogold Labelingfor Electron Microscopy Ceils grown on plastic coverslips were briefly washed in PBS, preextracted with the stabilizing medium, and fixed with 0.3 % glutaraldehyde in the same medium at 37~ for 10 rain. Free aldehyde groups were fitrated by sodium borohydride (1 nag/nil in PBS) for 7 rain. After washing in PBS, the coverslips were covered with a polyclonal rabbit anti-tubulin antibody for 30 rain and washed five times (15 rain each). The coverslips were then incubated for 30 rain with gold-labeled protein A (8 nm size). After an overnight wash in PBS containing 0.1% Triton X-100, the cells were postfixed in 1% glutaraldehyde in 80 mM K-Pipes pH 6.8, 5 mM EG'I'A, and 1 mM MgCl2. The specimens were then processed for electron microscopy and embedded in Epon. Serial thick (0.I5-0.25 ltm) and thin sections were cut and observed in a Philips 400 electron microscope. Semiquantitative Analysis of Tyrosinatedand Detyrosinated Tubulin by Immunoblotting Cells were seeded on 3-cm petri dishes so as to reach confluency 24 h later when they were treated with nocodazole or taxol as described in the figure legends. At the desired time, the cells were lysed directly in hot Laemmli's sample buffer (0.5 nil) (23a). Care w~s taken to recover the whole cell layer from each dish. Samples were further treated at 950C for 5 min, souicated for 3 rain in a water bath sonicator, spun 5 rain in a microfuge (Eppendorf; Brinkmann Instruments Inc., Westbury, NY), divided into aliquots of 50 I~l each, and kept frozen at -70~ until use. 5 I~l of each sample were applied on 10% Laemmli's polyacrylamide minigels, or 50 [tl on normal 10% polyacrylamide gels. After transfer onto nitrocellulose according to Towbin et al. (39), tyrosinated and detyrosinated tubulins were detected with the monoclonal anti-tyr-antibody (1:6,000) and the polyclonal anti-glu antibody (1:1,000). PeroxidaseAabeled rabbit anti-mouse (1:t,000) and goat anti-rabbit (1:t,000) from Tag., Inc., (Burlingame, CA) were used as secondary antibodies. Diaminobenzidine (5 mg/ml in.50 mM Tris-HCt pH 7.6, 0.01% H:O2) was used as the substratc for the peroxidase reaction. The reaction was carefully controlled and allowed to proceed for the same time for a given set of bints (1-3 min). The reaction was arrested by a quick wash in distilled water followed by a wash in 5 % TCA. Affinity-purified rabbit anti-glu tubulin antibodies was obtained by immunizing rabbits with synthetic peptide containing the 12 carboxy-terminal amino acid of ct-tubulin without the last tyrosine. The monoclonal anti-tyr tubulin was obtained by immunizing mice with the carboxy-terminal 13 amino acid of et-tubulin (including the last tyrosine residue). The preparation and complete characterization of these antibodies wilt be described elsewhere (22a). Each antibody was preabsorbed with the other peptide before use in order to eliminate any possible cross-reaction. The monoclonal anti-[3-tubulin was from Amersham Buchler GmbH, Braunschweig, FRG. After trypsinization, l-d-old MDCK cells were washed twice with 50 mM NaCI, 50 mM "IVIs-HCLpH 7,2, and 1 mM EDTA at 4~ An aliquot of cells was resuspended directly in hot Laemmli's sample buffer at 2 x 106 cells/ml and boiled for 5 rain. Another aliquot was resuspended, at 4~ in 50 mM K-Pipes pH 6.9, I mM EG"rA, and 0,t% Triton at 4 x 10* ceUs/ml. This aliquot was then treated for 20 rain at 3"/~ with 0.25 ttg/ml The Journal of Celt Biology, Volume 105, 1987 1284 Carboxypeptidase A Treatment of Cell Lysates Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 could influence the regional distribution of microtubulestabilizing factors. Another important level of control over microtubule organization concerns the mechanism of polymer nucleation in vivo. It is currently assumed that radial arrays of microtubules are nucleated off the centrosome by the pericentriolar material (2-4, 11, 12, 32). However, the pericentriolar material sometimes dissociates from the centrioles and other microtubule patterns are then generated (37). Moreover, spontaneous nucleation could occur in the cytoplasm if the tubulin concentration were high enough or if microtubulestabilizing factors were present (7, 19). The proportion and dynamics of centrosome-nucleated microtubules may therefore be affected by the number and activity of other potential MTOCs, active tubulin concentration, and microtubulestabilizing factors. In this paper we compare the microtubule patterns in fibroblasts (Vero cells), epithelial-like (PtK2), and epithelial cells (MDCK) as well as the relative stability of microtubules in these cell lines. The relationship between microtubule resistance to nocodazole, age, and a-tubulin detyrosination has also been investigated. We show that stable noncentrosomal microtubules exist in MDCK cells. We discuss the mechanism of generation and stabilization of these microtubules, and we present experimental evidence supporting the general hypothesis ttiat ct-tubulin is detyrosinated when associated with long-lived microtubules (13, 14, 21). of carboxypeptidaseA (phenylmethylsulfonylfluoridetreated; ServaCo., Heidelberg,FRG)preparedas previouslydescribedO8). The reactionwas stoppedby addingan equalvolumeof hot samplebufferto the reactionmixture and furtherboiling for 5 min. 105-eellequivalents(50 Ial)wereloaded on 10% polyacrylamideLaemmli'sgels and analyzedby immunoblotting. Results Centrosomes Are Split and Not Associated with the Nucleus in Subconfluent MDCK Cells In most fibroblastic cells, the two centriole cylinders are tightly associated and bound to the nucleus (1, 2, 10, 27). By immunofluorescence microscopy, using scleroderma human anti-centrosome serum (42), they appeared as one bright dot or two smaller dots tightly associated with each other both in Vero and PtK2 ceils (Fig. 1 a). In subconfluent MDCK cells they were separated, sometimes by distances as large as 10 txm (Fig. 1 b). No specific association with the nucleus was found by electron microscopy (Fig. 2 a). The aspect of the microtubule network in MDCK, Veto, and PtK2 cells stained with a monoclonal anti-13-tubulin antibody is shown in Fig. 3. The microtubule network was clearly centered on the centrosome in most Vero cells. This was less obvious in PtK2 cells. In MDCK cells, the microtubule network was randomly organized. There was no obvious correlation between the position of the split centrosome and the microtubule pattern. The cells shown in Fig. 3 have been double labeled with the centrosome antibody (data not shown). The positions of centrosomes are shown by arrows. We have observed serial thick sections of these cells by electron microscopy after immunogold staining of the microtubules. Between 0 and 15 microtubules were found to radiate from the pericentriolar area in 12 pairs of centrioles observed. Fig. 2 b shows one of these sections in which only 2-3 microtubules were found to radiate from the centrioles. Most of the Noncentrosomal Microtubules Are Stable To determine the relative stability of the noncentrosomal microtubules in MDCK cells, we treated the cells with a concentration of nocodazole (1.6 IxM) known to depolymerize almost all microtubules in a few minutes in I.~9 cells (19). This actually induced an almost complete depolymerization in Vero cells (Fig. 3) fixed 30 min after addition of the drug. Only occasional, long and curly stable microtubules remained, usually deriving from the centrosomes. In PtK2 cells, many more similar microtubules were left, all of them emanating from the centrosome area (Fig. 3). In MDCK cells, a large number of extremely tortuous microtubules persisted 30 min after addition of the drug (Fig. 3). Most of these microtubules apparently did not radiate from the centrosomes (Fig. 3, arrows). Upon prolonged incubation in the same drug concentration, the stable microtubules found in all three cell lines progressively disappeared. Short microtubules started to grow and elongate off the centrosomes between 1 and 4 h after addition of this concentration of the drug. The strongest regrowth was observed in PtK2 cells. At higher nocodazole concentration (33 ~tM, not shown), the kinetics of depolymerization was essentially the same al- Br6 et al. StableNoncentrosomalMicrotubules Figure 1. Centrosome splitting in subconfluent MDCK cells. (a) Centrosomes detected by the human anti-centrosome antibody (diluted 1:100)are visible as one large dot or two closely associated smaller dots in Vero cells. The brighter dots correspond to a centrosome in early prophase (arrow). (b) Each MDCK cell contains two individual structures decorated by the same antibody and sometimes separated by a large distance. Bar, 5 Ixm. though fewer microtubules remained. Microtubule regrowth from centrosomes was never observed after 4 h of treatment. These results show that by adding nocodazole at an intermediate concentration it is possible to induce a shift from a mainly noncentrosomal pattern to an exclusively centrosomal pattern of microtubules in MDCK cells. A large fraction of the rnicrotubules in MDCK cells does disappear after nocodazole addition. Therefore, the population of stable microtubules coexist with more labile microtubules. The ratio of stable versus unstable microtubules seems to be high in MDCK, intermediate in PtK2, and very low in Vero cells. Inversely, the ratio of centrosomal versus noneentrosomal microtubules seems to be low in MDCK, intermediate in PtK2, and very high in Vero cells. Most Stable Microtubules Contain Detyrosinated Tubulin It has been recently reported that a specific class of long sinuous microtubules is enriched in the posttranslationally modified form of a-tubulin, the detyrosinated a-tubulin (glutubulin) (15). A monoclonal anti-tyrosinated tubulin (antityr) and an affinity-purified rabbit anti-detyrosinated tubulin (anti-glu) have been prepared. These antibodies have been 1285 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Centrosomes of MDCK Cells Are Not Preferred Sites of Nucleation cells grown on plastic were immunostained for tubulin as described in Materials and Methods and thick sections screened for centrioles. (a) Two centrioles (arrows)'~3 l~m apart from each other, not associated with the nucleus, are visible. (b) Higher magnification of the centfioles shown in a. Among all the microtubulesdecorated by gold particles, only a few actually emanate from the centrioles (arrowheads) Bars, 1 lma. raised against the synthetic carboxy-terminal peptides of a-tubulin and found to be highly specific for each form of a-tubulin (22a). The microtubule pattern revealed by the anti-tyr antibody (tyr-microtubules) in MDCK cells was similar to the pattern obtained after staining with the monoclonal anti-I~-tubulin (Fig. 4). The anti-glu antibodies stained a distinct subset of long, wavy microtubtdes as previously reported for TC7 cells (glu-microtubules; Fig. 4, b and d) (15). However, MDCK cells appeared to contain many more of these microtubules. Most of the microtubules remaining 30 min after addition of nocodazole (1.6 I~M) were decorated by the glu-antibodies, although to a variable extent (Fig. 4 d). These microtubules were also decorated by the tyr-antibody but very few microtubnles were stained only by the tyrantibody (Fig. 4 c). We will call "glu" any microtubule that is clearly stained by the glu-antibody. This does not mean that it is not detected by the tyr-antibody, inasmuch as some of them being more strongly labeled by this antibody than by the glu-antibody. The number of glu-microtubules slowly decreased with time of incubation in the drug. The short centrosomal microtubules regrowing slowly between 1 and 4 h were first exclusively stained by the anti-tyr-antibody and became stained by the glu-antibody at 4 h, in the vicinity of At the periphery of MDCK cells we occasionally found glumicrotubules with ends exclusively stained by the tyrantibody. Unfortunately, microtubule density hampered our attempts to determine what happened at the other end of these microtubules. It was easy, however, to observe microtubules double stained with tyr- and glu-antibodies over their entire length in cells treated with nocodazole for 1 and 2 h since only a few microtubules remained (Fig. 5). 1 h after addition of nocodazole (1.6 ~tM), very few glu-microtubules contained ends stained only by the tyr-antibody (not shown). However, at 2 h, when tyr-microtubules start to appear around the centrosomes, virtually all the remaining noncentrosomal glu-microtubules had only one end exclusively stained by the tyr-antibody (Fig. 5, a and b). This staining was often much brighter than over the rest of the microtubules and was rarely observed on the glu-microtubules of cells treated with 33 txM nocodazole (Fig. 5, c and d). This concentration prevented the regrowth of tyr-microtubules The Journal of Cell Biology, Volume 105, 1987 1286 the centrosome (Fig. 4 f ) . At this time point, occasional, long glu-microtubules were found associated with the centrosomes. These were probably "old microtubules". One End of Noncentrosomal Glu-Microtubules Can Be Exclusively Composed of 7l~r-Tubulin Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure2. Thick section of MDCK cells observed by electron microscopy after immunogoldstahtingof microtubules. SubconfluentMDCK Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure 3. Microtubule pattern and stability in MDCK, Vero, and PtK2 cells. MDCK (/eft), Veto (center), and PtK (right) cells were stained by double immunofluoreseencefor microtubules and centrosomes as described in Materials and Methods. A mouse monoelonal anti-I[~tubulin was used for mierombule staining. The centrosome staining is not shown, but the position of each centrosome is indicated when not obvious (arrows). Cells were fixed before or after various times of incubation in 1.6 IxM nocodazole. Numbers (left) indicate the time in minutes of incubation of each cell line in nocodazole. Br~ et al. Stable Noncentrosomal Microtabules 1287 MDCK cells were fixed before (a and b), 30 rain (c and d), and 4 h (e andf) after incubation in 1.6 ~tM nocodazole. Double immunostainingwas performed with anti-tyr- (a, c, and e) and anti-glu(b, d, and f ) antibodies. Secondary antibodies were labeled with rhodamine (a, c, and e) and fluoreseein (b, d, and f ) (Bar, 5 txM). from the centrosomes during the treatment. In this case, however, different staining intensities were occasionally observed with the glu-antibodies along the length of a given microtubule (Fig. 5 d). It seems that newly formed microtubules or microtubule fragments contain mainly tyrosinated ct-tubulin. Microtubule Network Reassembly Precedes Tubulin Detyrosination after Nocodazole Removal antibodies (Fig. 6). Microtubules detected by the tyrantibody elongated, and at 30 min an apparently normal microtubule pattern was reformed. It was only then that some mierotubnles became reactive to the glu-antibodies, and only in a fraction of the cell population. A glumicrotubnle pattern comparable to what was found in the normal population was observed at 1 h after nocodazole removal. The ratio of noncentrosomal to centrosomal microtubules increases with time after nocodazole removal. To investigate further the relationship between microtubule polymerization and detyrosination of ct-tubulin in vivo, we examined the kinetics of tyr- and glu-microtubule regrowth after nocodazole reversal (Fig. 6). After 4 h of incubation in 33 ttM nocodazole, most microtubules were depolymerized. Only a few, long centrosomal microtubules containing glutubulin remained (Fig. 6). Centrioles could be identified with the glu-antibodies, which seemed to strongly stain the centrioles (Fig. 6). Interestingly, the tyr-antibody stained numerous dots, scattered in the cytoplasm, which were not stained by the glu-antibodies. This was always found and does not seem to be an artifact of fixation. Indeed, these dots were also detected by the anti-13-tubulin in MDCK cells but not in other cells (Fig. 3). 3 min after nocodazole removal, many short microtubules were detected by the tyr-antibody, growing as small asters from the centrosomes as small foci or as short individual segments scattered throughout the cytoplasm (Fig. 6). With the exception of long (probably old) microtubules, none of the short (newly formed) microtubules were stained with the glu- The low abundance of centrosomal microtubules at steady state in MDCK cells is very surprising in view of the large amount of microtubules regrowing from centrosomes at early time points after nocodazole removal. This means that noncentrosomal microtubules eventually predominate in this cell line. This is different than what has been found in fibroblastic cells where centrosomal microtubules predominate. To investigate the potential correlation between the appearance of noncentrosomal microtubules and polymer stabilization, we incubated cells for 4 h in 33 ~tM nocodazole, then removed the drug, and at various times during microtubule regrowth we further added nocodazole at 1.6 or 33 ttM for 30 min. The cells were then fixed and double stained with the anti-glu and anti-tyr antibodies (Fig. 7). Only tyrmicrotubules are shown. Long centrosomal and short non- The Journalof Cell Biology,Volume105, 1987 1288 Kinetics of Centrosomal and Noncentrosomal Microtubule Stabilization during Regrowth after Nocodazole Removal Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure4. Stable microtubules in MDCK cells contain detyrosinated r centrosomal microtubules were found to be resistant to 1.6 lxM nocodazole 15 rain after release from the first drug treatment (Fig. 7 a). They were not stained by the glu-antibodies (not shown). Long noncentrosomal mierotubules became resistant to this drug concentration only after 30 min of regrowth. They were predominant by 1 h (Fig. 7, b and c). This coincides with the first appearance of rnicrotubules strongly stained by the glu-antibodies (Fig. 6). The use of 1.6 ~M nocodazole for 30 min as a test for the presence of relatively stable micrombules is fully valid since this treatment induces a complete depolymerization in fibroblasts. However, we were concerned that the %table" centrosomal microtubules observed after 15 min of regrowth arose by elongation during the further 30 min of treatment in 1.6 ltM nocodazole. To eliminate this possibility, we repeated the kinetics of polymer stabilization during regrowth by using a high concentration of nocodazole (33 ~tM) (Fig. 7, d-h). The result was striking: many centrosomal microtubules were stabilized immediately after being nucleated (as soon as 3 min; Fig. 7 d). Moreover, the length of stabilized centrosomal microtubules increased with time of regrowth while their number seemed to drop. Stable, short noncentrosomal microtubules arose between 8 and 15 min. Long stable noncentrosomal micrombules were much less abundant at 30 and 60 rain than in cells treated with 1.6 IxM nocodazole (Fig. 7, b, c, g, and h). We conclude from these experiments that: (a) Some centrosomal microtubules are stabilized immediately after being nucleated. (b) The number of stable centrosomal microtubules drops and their length increases with time of regrowth. Br6 et al. StableNoncentrosomalMicrotubutes (c) Noncentrosomal micrombules are stabilized later than centrosomal microtubules. (d) Many of the noncentrosomal micrombules, which are stable in 1.6 I~M, are unstable in 33 I~M nocodazole. (e) Microtubule stabilization is reversible since the pattern of stable microtubules evolves with time of regrowth. ( f ) Tubulin detyrosination occurs after partial microtubule stabilization. Microtubule Stabilization by Taxol Promotes Tubulin Detyrosination The results reported above strongly suggest that detyrosination of a-mbulin occurs primarily on stabilized polymer in vivo. To strengthen this correlation, we examined the microtubule staining with the glu-antibodies in cells treated with taxol, a drug known to reduce micrombule dynamics (Fig. 8). There was a complex rearrangement of the microtubule network over the first 2 h of incubation in taxol. No obvious increase in the number of glu-microtubules could be detected during the first 30 rain of incubation. By 2 h, however, while the microtubule pattern seemed to be relatively stable, most microtubules appeared to be stained by the glu-antibodies. Amount of Glu-Tubulin Is Decreased in Nocodazole and Increased in Taxol-treated Cells The previous set of experiments clearly shows that newly formed microtubules are mostly composed of tyrosinated tubulin. On the other hand, stable microtubules or taxolstabilized microtubules are clearly enriched in glu-tubulin. This strongly suggests that most of the free tubulin is tyrosi- 1289 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure5. In MDCK cells glu-microtubule ends are exclusivelystained by the tyr-antibody during depolymerizationin 1.6 IxM nocodazole. 2 h after the addition of noeodazole at 1.6 lxM (a and b) and 1 h aP~eraddition of noeodazole at 33 IxM (c and d) MDCK cells were fixed and double stained with anti-tyr- (a and c) and anti-glu- (b and d) antibodies. (Arrows)Mierotubule ends exclusivelystained by the anti-tyrantibody. (Large arrowheads) Centrosomes. (Small arrowheads) Ends weakly labeled by the anti-glu-antibody in a microtubule that is strongly stained. The large spots produced by the anti-tyr-antibody around centrosomes are in fact numerous, short microtubules not resolved on the picture. Bar, 5 $tm. The Journal of Cell Biology, Volume 105, 1987 1290 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure 6. Kinetics of tyr- and glu-micmtubule regrowth in MDCK cells after nocodazole removal MDCK ceils were treated for 4 h with 33 gM nocodazole. The drug was removed, the ceils were fixed at the time (in minutes) indicated on the left of the figure, and double stained for tyr- (left) and glu- (right) a-tubulin. (Arrows) Centrosomes; (arrowheads) noncentrosomal microtubule loci. Bar, 5 Ixm. Discussion Centrosome and Centrioles in Subconfluent MDCK Cells The term centrosome in mammalian cells usually refers to the complex formed by the two centriole cylinders and the pericentriolar material that nucleates microtubules (2). We have found that in subconfluent MDCK cells, the two cylinders are largely independent from each other and from the nucleus. A detailed study is underway. This unusual condition of the centrosome has been found in other situations (32, 33, 35). In MDCK cells, the two cylinders migrate up to the apical and central part of the cell at a relatively constant distance from each other during cell polarization (Simons, K., M. H. Br6, and E. Karsenti, unpublished data). This provides us with a unique model to study the control of centrioles and pericentriolar material location during cell Br6 et al. Stable Noncentrosomal Microtubules differenfiaton and polarization. Several examples suggest, indeed, that centrioles are relocated or eliminated during cell differentiation (2, 8, 9, 32, 37, 40, 41, 45). Origin and Persistance of Noncentrosomal Microtubules in MDCK Cells The microtubule pattern in MDCK cells appears to be much more disorganized than in Vero and PtK2 cells, where most microtubules originate at the centrosome. One could argue that this is due to the fact that in MDCK cells the two centrioles are separated. Indeed, by being nucleated in two different locations, the microtubules could form a network that would appear less organized. However, in the electron microscope, the centrioles were found to nucleate only a few microtubules. Therefore, in MDCK cells, many microtubules are generated by a mechanism other than nucleation by the centrosome. One possibility would be that the microtubule-nucleating material that is associated with the centrioles in fibroblasts is more dispersed in MDCK cells. The balance between centrosomal and noncentrosomal microtubules could also be controlled by modulating the dynamics of microtubule assembly and disassembly. Mitchison and Kirschner have shown that microtubule asters formed on purified centrosomes in vitro are very dynamic (25). Microtubules grow and then start to depolymerize catastrophically with a probability that increases with decreasing concentrations of tubulin. The nucleation sites left unoccupied on the centrosome can then nucleate new rnicrotubules below the steady state tubulin concentration. This behavior in vitro may explain why centrosomal microtubules predominate in some cell lines. Indeed, during microtubule re.growth after removal of depolymerizing drugs like nocodazole, both centrosomal and noncentrosomal microtubules appear in the cytoplasm of these cells. However, the noncentrosomal microtubules eventually disappear and at equilibrium a centrosome-nucleated aster of microtubules predominates. In the context of the dynamic instability model, this behavior is easy to explain: Upon removal of nocodazole, the free tubulin concentration in the cell is very high. Spontaneous nucleation will be frequent and compete efficiently with centrosome nucleation. However, with time, more polymer assembles and since the cell is a closed system the free tubulin concentration drops. A steady state situation is reached where the tubulin concentration is too low to allow spontane- 1291 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 nated while detyrosination occurs on the polymer. If this were the case, most of the cellular tubulin in nocodazoletreated ceils should be tyrosinated (since it is soluble) while in taxol-treated cells, tubulin should be enriched in the gluform. To test this possibility, we carried out immunoblots with the anti-glu and anti-tyr-antibodies on total extracts of cells after nocodazole and taxol treatments (Fig. 9). The specificity of the antibodies for tyrosinated and detyrosinated tubulin has been checked on immunoblots made with total cell extracts before and after treatment with carboxypeptidase A (Fig. 9 a). The reactivity of the tyr-antibody was decreased (Fig. 9 a, lane 2) and the reactivity of the gluantibodies dramatically increased (lane 4) after action of carboxypeptidase A. Cells incubated for 4 h in 33 txM nocodazole and lysed directly in sample buffer contained a very low amount of glu-tubulin. Glu-tubulin became detectable between 15 and 30 rain after nocodazole removal. The amount of tyr-tubulin remained constant throughout this period, demonstrating that the lack of detection of glutubulin after nocodazole treatment is not due to a different amount of total tubulin present in these samples (Fig. 9 b). In cells treated with taxol, the amount of glu-tubulin increased progressively with the time of treatment. Here, the amount of tyr-tubulin remained constant or dropped slightly at longer times of incubation in taxol (Fig. 9 c). ous nucleation and therefore only centrosome-nucleated microtubules remain (25). This situation implies that the centrosomal aster is a very dynamic structure and one would expect it to disappear quickly in response to nocodazole. This is what was found for L929 cells and we report a similar situation for Vero cells. In PtK2 cells, many centrosomal microtubules are apparently stable since they are resistant to a similar treatment. In MDCK cells, the situation is different. Most stable microtubules (which are numerous) appear to be noncentrosomai. In a way microtubules behave naturally in these cells as they do in other cells (PtK2, 3T3, human fibroblasts) that have been treated with taxol. De Brabander et al. (6) have shown that taxol induces spontaneous microtubule assembly in the cytoplasm of interphase cells without spatial relation to the centrosome. Moreover, the preexisting centrosomal microtubules gradually disappear. This effect was explained by the properties of taxol that decrease the critical concentration for tubulin polymerization and increase microtubule stability. Therefore, it seems that MDCK cells contain factors capable of stabilizing microtubules as taxol does. These may be specific MAPs. The exact molecular mechanism by which such factors would function is still unclear. Brain MAPs, for example, bind quickly and tightly to microtubules The Journal of Cell Biology, Volume 105, 1987 1292 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure 7. Kinetics of microtubule stabilization aRer nocodazole removal. MDCK cells were treated for 4 h with 33 gM nocodazole. The drug-containing medium was then removed and the cells were quickly washed three times and incubated with fresh, warm medium. After 15, 30, and 60 min of regrowth, nocodazole (1.6 ~tM) was further added for 30 min, and the cells fixed (a-c). After 3, 8, 15, 30, and 60 rain of regrowth, nocodazole (33 ~tM) was further added and the cells fixed (d-h). Double immunofluorescence with anti-tyr- and anti-gluantibodies was carried out. Only the tyr staining is shown. The positions of centrosomes are indicated (arrows). Bar, 5 ~tm. in vitro, leading to a rapid and almost irreversible stabilization of microtubules (18). The presence of such molecules in saturating amounts would not explain the progressive predominance of noncentrosomal microtubules in MDCK cells during regrowth after nocodazole removal. However, Job et al. (18) discuss the possibility that MAPs present in nonsaturating amounts could bind in discrete sites along microtubules because of their fast association rate. They argue that microtubule stability subclasses could be generated in this way. Moreover, they suggest that a given microtubule could have different levels of stability along its length depending on the amount of MAPS bound locally. The presence of limiting concentrations of MAPS in MDCK cells could explain the progressive predominance of noncentrosomal microtubules according to the following mechanism: Initially after nocodazole removal, numerous short microtubules are nucleated in the cytoplasm. Both on free and centrosome-nucleated microtubules, very short segments are stabilized by the binding of MAPs. The existence of such stable segments along noncentrosomal microtubules would prevent them from complete depolymerization when the steady state free tubulin concentration is approached. Such segments could then act as seeds for polymer elongation below the free tubulin concentration required for spontaneous nucleation. If many stable seeds exist, they will compete with the nucleation activity of the centrosome. We find that stable centrosomal microtubules arise early and before the noncentrosomal ones during regrowth. This Br6 et al. Stable Noncentrosomal Microtubules probably occurs because they are capped at one end by the centrosome. This indeed increases the probability of stabilization of a visible microtubule stretch. Interestingly, during regrowth, the length of stable centrosomal microtubules increases in parallel with microtubule elongation. While this occurs the number of stable centrosomal microtubules drops. This behavior fits well with an initial rapid and reversible stabilization due to soluble factors, followed by their random redistribution. Short, stable noncentrosomal microtubules become visible after 8-15 min of regrowth; this occurs when they happen to bind stabilizing factors in two different points. In other words, a stable microtubule could be composed of a long nonstabilized region flanked by two short stabilized segments. The probability for this to occur increases with the length of the microtubule and with its life span. This mechanism would produce microtubules with segments of different ages as previously suggested (18) and with dynamic extremities. Several observations argue in favor of this idea: (a) We found micrombules strongly stained by the glu-antibodies in their middle part with a weaker staining on each side. Since we assume that the extent of r detyrosination is representative of the microtubule age, this suggests that the microtubule ends are not as "old" as the rest of the microtubule. (b) After 1 h of mierotubule depolymerization in the presence of 1.6 lxM nocodazole, tyrmicrotubules regrow slowly from the centrosomes. In the same time, short microtubule segments exclusively stained by the tyr-antibody appear at one end of the remaining noncen- 1293 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 Figure 8. Stabilization of micrombules by taxol increases the proportion of detyrosinated microtubules. MDCK ceils were treated with taxol (5 ~tM) for 30 min (a and b) and 2 h (c and d). After fixation the cells were processed for double immunofluorescencewith the anti-tyr- (a and c) and anti-glu- (b and d) antibodies. Bar, 5 ltm. Microtubule Stability and a-Tubulin Detyrosination trosomal glu-microtubules. These are probably newly polymerized segments. Since most noncentrosomal microtubules have disappeared by 4 h, we must conclude that during depolymerization in the presence of 1.6 BM nocodazole they grow and shrink (the overall shrinking rate being higher than the growth rate). The tyr-containing end might be the active end (+ end) in these microtubules. It has been shown recently that cells contain two classes of microtubules that can be distinguished by antibodies directed against the tyrosinated and detyrosinated forms of ct-tubulin (15). The most abundant class of microtubules contains mainly tyrosinated vt-tubulin. A less abundant class in many cells is composed of long sinuous microtubules containing considerable amounts of detyrosinated a-tubulin. As recently shown, there seems to be a continuum in the ratio of tyrosinated and detyrosinated ct-tubulin in all microtubules (13). It has been suggested that the microtubules strongly stained by the anti-glu antibodies are long lived, stable microtubules (13, 21). In the present paper we have accumulated evidence that this is indeed the case. In fact, it is likely that the turnover of the caboxy-terminal tyrosine of a-tubulin is tightly coupled to microtubule dynamics as proposed by Thompson (38). In vivo as in vitro (28, 29), the tubulin tyrosin ligase seems to function only on the dimer. In vitro the tubulin carboxypeptidase appears to function preferentially on the polymer (23). If in vivo the carboxypeptidase functions on the dimer, we must assume that this reaction is much slower than the tyrosination reaction since after microtubule depolymerization by nocodazole we detect almost no detyrosinated u-tubulin on immunoblots of total cell lysates. Moreover, three lines of evidence suggest that tubulin detyrosination occurs only on long-lived stabilized microtubules: (a) After nocodazole release, the first microtubules to regrow are tyrosinated. (b) The first detyrosinated microtubules appear after 30 rain of regrowth while stable microtubules already exist 3 min after nocodazole removal. This time lag between microtubule polymerization and the appearance of the first detyrosinated microtubules is in agreement with results obtained in vitro (5). (c) Stabilization of microtubules with taxol leads to an increase in the number of detyrosinated microtubules and also to an increased amount of total detyrosinated tubulin. This shows that anti-glu antibodies are good markers of long-lived microtubules and raises intriguing questions as to how the tubulin carboxypeptidase functions. In any case, The Journal of Cell Biology, Volume 105, 1987 1294 Figure 9. Immunoblot analysis of tyrosinated and detyrosinated tubulin in MDCK cell lysates. (a) Cell lysates were prepared and treated by carboxypeptidase A as described in Materials and Methods. The proteins were run on a 10% Laemmli gel, transferred to nitrocellulose, and stained with the anti-tyr (lanes 1 and 2) or anti-glu (lanes 3 and 4) antibodies. Carboxypeptidase-treated sampies are in lanes 2 and 4. (b) Cells were treated with 33 gM nocodazole for 4 h, lysed in hot gel sample buffer at the indicated times (in minutes) after nocodazole removal, and loaded on one 10% Laemmli minigel in duplicate, n, cells not treated by nocodazole. After transfer, the nitrocellulose sheet was split in two parts: onehalf was stained with the anti-tyr antibody (tyr) and the other half with the anti-glu antibody (glu). (c) Ceils were treated with 5 IxM taxol, lysed in hot gel sample buffer at the indicated times (in minutes), and loaded on one 10% Laemmli minigel in duplicate. After transfer, the nitrocellulose sheet was split in two parts: onehalf was stained with the tyr-antibody (tyr) and the other with the glu-antibody (glu). In all cases, the secondary antibody was labeled with peroxidase. See Materials and Methods for details. Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 The regrowth of small centrosomal asters from centrosomes after 4 h of incubation in 1.6 BM nocodazole shows that some active tubulin remains under such conditions. The absence of noncentrosomal microtubules in the same conditions suggests that the mechanism of microtubule stabilization is different from the mechanism of nucleation by the centrosome. The centrosome probably caps irreversibly the "minus" ends of microtubules, while stabilizing factors bind along the microtubule wall. Therefore, noncentrosomal microtubules having two ends free would be more sensitive to very low tubulin concentrations. This interpretation is also supported by the observation that stable centrosomal microtubules resist equally well to 1.6 and 33 BM nocodazole, while many noncentrosomal microtubules do not (Fig. 7). Although the model discussed above for the generation of noncentrosomal microtubules fits well with our experimental observations, other explanations could be proposed: Noncentrosomal microtubules could be stabilized by local interactions with insoluble structures such as intermediate filaments or membrane sites for example (41). Some stable noncentrosomal microtubules could also arise by falling off the centrosome (43). We thank Shamsa Faruki and Heike Wilhelm for expert technical assistance. We thank Gareth Griffith for efficient help with the electron microscopy and critical reading of the manuscript, Kai Simons and Jean Davoust for stimulating discussions and suggestions during the preparation of this manuscript. We thank Alan Summerfield for his photographic work, and Annie Steiner and Anne Walter for typing the manuscript. Received for publication 18 February 1987, and in revised form 14 April 1987. References 1. Bornens, M. 1977. Is the centriole bound to the nuclear membrarre? Nature (Lond.). 270:80-82. 2. Bornens, M., and E. Karsenti. 1984. The centrosome. In Membrane Structure and Function. Vol. 6. E. E. Bittar, editor. L Wiley & Sons, Inc., New York. 99-171. 3. Brinldey, B. R., S. M. Cox, D. A. Pepper, L. Wible, S. L. Brenner, and R. L. Pardue. 1981. Tubulin assembly sites and the organization of cytoplasmic microtubules in cultured mammalian cells. J. Cell Biol. 90:554562. 4. Brinkley, B. R., G. M. Fuller, and D. P. Highfield. 1975. Cytoplasmic microtubules in normal and transformed cells in culture: analysis by tubulin antibody immunofluorescence. Proc. Natl. Acad. Sci. USA. 72:49814985. 5. Deanin, G. G., S. F. Preston, R. K. Hanson, and M. W. Gordon. 1980. On the mechanism of turnover of the carboxyterminal tyrosine of the alpha-chain of tuhulin. Fur. J. Biochem. 109:207-216. 6. De Brabander, M., G. Geuens, R. Nuydens, R. Willebrords, and J. De Mey. 1981. Taxot induces the assembly of free microtubules in living cells and blocks the organizing capacity of the centrosomes and kinetochores. Proc. Natl. Acad. Sci. USA. 78:5608-5612. Br~ et ai. Stable Noncentrosomal Microtubules 7. De Brabander, M., G. Geuens, R. Nuydens, R. Willebrords, and J. De Mey. 1982. Microtubule stability and assembly in living cells: the influence of metabolic inhibitors, taxol and pH. Cold Spring Harbor Syrup. Quant, Biol, 46:227-240. 8. Dustin, P. 1978. Micrombules. Springer Verlag, Berlin/Heidelberg/New York. 9. Dylewsky, D. P., and T. W. Keenan. 1984. Centrioles in the mammary epithelium of the rat. J. Cell Sci. 72:185-193, 10. Fais, D., E. S. Nadezhdina, and Y. S. Sohentsov. 1984. Evidence for the nueleus-centriole association in living cells obtained by ultracentrifugation. Eur. J. Cell Biol. 33:190-196. 11. Frankel, F. R. 1976. Organization and energy-dependent growth of microtubttles. Proc. Natl. Acad. Sci. USA. 73:2798-2802. 12. Fuller, G. M,, B. R. Brinidey, and L M. Boughter. 1975. Immunofluorescence of mitotic spindles by using mono-specific antibodies against bovine brain tubulin. Science (Wash. DC). 187:948-950. 13. Geuens, G., G. G. Gundersen, R. Nuydens, F. Carnelissen, J. C. Bulinski, and M. de Brabander. 1986. Ultrastrucmral eolocaiizafion of tyrosinated and detyrosinated tt-tubulin in interphase and mitotic cells. J. Cell Biol. 103:1883-1895. 14. Gundersen, G. G., and J. C. Bulinski. 1986. Micrombule arrays in differentiated cells contain elevated levels of a post-translationally modified form of tubulin. Euro. J. Cell Biol. 42:288-294. 15. Gundersen, G. G., M. H. Kalnoski, and L C. Bulinski. 1984. Distinct populations of microtubules: Tyrosinated and non tyrosinated alphatubulin are distributed differently in vivo. Cell. 38:779-789. 16. Horio, T., and H. Hatani. 1986. Visualization of the dynamic instability of individual microtubules by dark-field microscopy. Nature (Lond.). 321:605 -607. 17. Inout, S. 1981. Cell division and the mitotic spindle. J. Cell Biol. 91(3, Pt. 2):131s-147s. 18. Job, D., M. Pabion, and R. L. Margolis. 1985. Generation of mierotuhule stability subclasses by micrombule associated proteins: implications for the "microtubule instability~ model. J. Cell Biol. 101:1680-1689. 19. Karseati, E., S. Kobayashi, T. Mitchison, and M. Kirschner. 1984. Role of the centrosome in organizing the interphase microtubale array: the properties of cytoplasts containing or lacking centrosomes. J. Cell Biol. 98:1763-1776. 20. Kirschner, M., and T. Michison. 1986. Beyond self-assembly: from microtubules to morphogenesis. Cell. 45:329-342. 21. Kirschner, M., and E. Schulze. 1986. Morphogenesis and the control of microtubule dynamics in cells. J. Cell Sci. 5(Suppl.):293-310. 22. Kreis, T. E. 1986. Microinjected antibodies against the cytoplasmic domain of vesicular stomatitis virus glycoprotein block its transport to the cell surface. EMBO (Eur. Mol. Biol. Organ.) J. 5:931-941. 22a. Kreis, T. E. 1987. Microtubules containing detyrosinated tuhulin are less dynamic. EMBO (Eur. MoL Biol. Organ.) J. 6:2597-2606. 23. Kumar, N., and M. Flavin. 1981. Preferential action of a brain detyrosinolating carboxypeptidase on polymarized tubalin. J. Cell Biol. 256: 7678-7686. 23a. Laemmli, U. K. 1970. Cleavage of the structural proteins during the assembly of the head of bacteriophage T4. Nature (Lond.). 277:680-685. 24. McIntosh, J. R. 1983. The centrosome as an organizer of the cytoskeleton. Mod. Cell Biol. 2:115-142. 25. Mitchison, T., and M. Kirschner. 1984. Microtubule assembly nucleated by isolated centrosomes. Nature (Lond.). 312:232-237. 26. Mitchison, T., and M. Kirschner. 1984. Dynamic instability of microtubale growth. Nature (Lond.). 312:237-242. 27. Nadezhdina, E. S., D. Fais, and Y. S. Chentsov. 1979. On the association of centrioles with the interphase nucleus. Fur. J. Cell Biol. 19:109-115. 28. Raybin, D., and M. Flavin. 1975. An enzyme tyrosylating ct-tubulin and its role in microtubule assembly. Biochem. Biophys. Res. Commun. 65:1088-1095. 29. Raybin, D., and M. Flavin. 1977. Modification of tubulin by tyrosylation in cells and extracts and its effects on assembly in vitro. Z Cell Biol. 73: 492-504. 30. Saxton, W. M., D. L. Stemple, R. J. Leslie, E. D. Salmon, M. Zavortink, and J. R. Mclntosh. 1984. Tubulin dynamics in cultured mammalian cells. J. Cell Biol. 99:2175-2186. 31. Scherson, T., T. E. Kreis, J. Schlessinger, U. Z. Littauer, G. Borisy, and B. Geiger. 1984. Dynamic interactions of fluorescently labelled microtubale associated proteins in living cells. J. Cell Biol. 99:425--434. 32. Sehliwa, M. 1986. Microtubules.lnTheCytoskeletonanlntroductory Survey. M. Alfert, W. Beermann, L. Goldstein, and K. R. Porter, editors. Springer-Vedag, Wien/New York. 47-81. 33. Schliwa, M., K. B. Pryzwansky, and U. Euteneuer. 1982. Centrosome splitting in neutrophils: An unusual phenomenon related to cell activation and motility. Cell. 31:705-717. 34. Schulze, E., and M. Kirschner. 1986. Microtubule dynamics in interphase cells. J. Cell Biol. 102:1020-1031. 35. Sherline, P., and R. Mascardo. 1982. Epidermal growth factor induced centrosomal separation: mechanism and relationship to mitogenesis. Z Cell BioL 95:316-332. 36. Simons, K., and S. D. Fuller. 1985. Cell surface polarity in epithelia. Annu. Rev. Cell Biol. 1:243-288. 1295 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 our results demonstrate that detyrosination is not the cause of reversible microtubule stabilization. Rather, micrombule stabilization appears to be a prerequisite to tubulin detyrosination, at least in MDCK cells. This may mean that the tubulin carboxypeptidase functions slowly. If a micrombule is short lived, the proportion of detyrosinated subunits may remain too low to be visualized by immunofluorescence. The amount of detyrosinated mbulin in a given micrombule will depend on its age. This may explain why nocodazole-resistant (stable) microtubules contain variable amounts of detyrosinated tubulin (Figs. 4 and 5). Indeed, the age of these stable microtubules could vary broadly. The sort of microtubule stabilization we term "reversible" in this discussion could be better named "short term: It is possible that detyrosination plays a direct or indirect role in the mechanisms involved in "long term" microtubule stabilization. The respective functions of dynamic and stable microtubules are unknown. Our results would imply that in an interphase cell at steady state many dynamic microtubules are centrosome nucleated. Since the centrosome could be located in various places in the cell, it is tempting to speculate that together with its dynamic aster of microtubules, it constitutes a functional entity that can concentrate or keep concentrated specific cytoplasmic organelles in a given area of the cell. Indeed, by sending out new microtubules continuously and at random the centrosome can probe constantly the cytoplasmic space. The stable noncentrosomal microtubules could fulfill other transport or structural functions. It would be interesting to look for such microtubules in various cell systems undergoing specific morphogenetic rearrangements. They might play a scaffolding role as it seems to be the case for microtubules during myogenesis (44). The function of ~t-tubulin detyrosination is still unclear. The specific coupling that seems to exist between microtubule stabilization and detyrosination strongly suggests that this posttranslational modification plays an important functional role. Cell BioL 41:279-289. 42. Tuffanelli, D. L., F. McKeon, D. K. Kleinsmith, T. K. Burnham, and M. Kirschner. 1983. Anti-centrosome and anti-centriole antibodies in the scleroderma spectrum. Arch. Dermatol. 119:560-566. 43. Vorobjev, I. A., and Y. S. Chentsov. t983. The dynamics of reconstitution of microtubules around the cell center after cooling. Eur. J. Celt Biol. 30:149-153. 44. Warren, R. H. 1974. Microtubular organization in elongating myogenic cells. J. Cell Biol. 63:550-566. 45. Zeligs, J. D. 1979. Association of centrioles with clusters of apical vesicles in mitotic thyroid epithelial cells. Are centrioles involved in directing secretion? Cell Tissue Res. 201:11-21. The Journal of Cell Biology, Volume 105, 1987 1296 Downloaded from http://rupress.org/jcb/article-pdf/105/3/1283/1460326/1283.pdf by guest on 23 October 2023 37. Tassin, A. M., B. Maro, and M. Bornens. 1985. Fate of microtubule organizing centers during in vitro myoganesis. J. Cell Biol. 100:35-46. 38~ Thompson, W. C. 1982. The cyclic tyrosination/detyrosination of alphatubulin. Methods Cell Biol. 24:236-255. 39. Towbin, H., T. Staebelin, and J. Gordon. t979. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. $ci. USA. 76:4350-4354. 40. Tucker, J. B. 1984. Spatial organization of microtubule-organizing centers and microtubules. J. Cell Biol. 99(1, Pt. 2):55s-62s. 41. Tucker, J. B., M. J. Milner, D. A. Currie, J. W. Muir, D. A. Forrest, and M. J. Spencer. 1986. Centrosomai microtubule organizing centers and a switch in the control of protofilamant number for cell surfaceassociated microtubules during Drosophila wing morphoganesis. Eur. J.