Low Concentrations of Nocodazole Interfere with Fibroblast Locomotion

Journal of Cell Science 108, 3473-3483 (1995) 3473 Printed in Great Britain © The Company of Biologists Limited 1995 JCS6965

Low concentrations of nocodazole interfere with ﬁbroblast locomotion without signiﬁcantly affecting microtubule level: implications for the role of dynamic microtubules in cell locomotion

Guojuan Liao1, Takayuki Nagasaki1 and Gregg G. Gundersen1,2,* 1Departments of Anatomy & Cell Biology and 2Pathology, Columbia University, New York, NY 10032, USA *Author for correspondence

SUMMARY

The role of microtubules (MTs) in cell locomotion is polymer levels by cell extraction and western blotting uncertain: while MTs are not essential for motility of certain yielded results similar to those obtained by quantification of cells, MTs are necessary for the directed translocation of MT fluorescence. A comparison of the locomotion rate mea- large cells such as fibroblasts, endothelial cells and neuronal surements with the MT level measurements indicated that growth cones. Based on previous studies, we hypothesize that over half of the cell locomotion rate could be blocked by cell locomotion may involve MTs in two possible ways: (1) nocodazole without significantly affecting MT levels in the the rate of cell locomotion is proportional to MT level; or (2) cell; the remaining locomotion rate was reduced propor- cell locomotion is not proportional to MT level but requires tionally to MT levels. These results do not support the notion a critical level of MTs to proceed. To test these hypotheses, that a critical level of MTs is required for cell locomotion we measured the rate of locomotion of NRK fibroblasts and suggest that only a portion (<50%) of the speed of the migrating into an in vitro wound, before and after treatment cells is proportional to MT levels. Rather, by analogy with with different concentrations of nocodazole to generate cells studies of MT antagonists on the mitotic spindle, they with different levels of MTs. Locomotion of cells was suggest a third possibility: that low concentrations of noco- monitored directly using timelapse recording and analyzed dazole interfere with MT dynamics and thus, MT dynamics with an Image-1 image analysis program. Addition of noco- are critical for the maximal speed of cell locomotion. This dazole (у50 nM) resulted in a rapid reduction in locomotion notion was further supported by analogous effects of taxol to a new rate that was maintained for >2 hours. We found and vinblastine on cell locomotion: at concentrations that that addition of as little as 100 nM nocodazole decreased the reportedly cause little change in the level of MTs, taxol and rate of locomotion by more than 60%; and that 300 nM vinblastine also dramatically decreased the rate of locomo- nocodazole completely stopped cell locomotion. Although tion of NRK cells. In summary, our results establish the rela- 100 nM nocodazole decreased locomotion over 60%, we tionship between microtubule levels and locomotion rate and detected no qualitative change in MT distribution by suggest that dynamic MTs are rate-limiting for fibroblast immunofluorescence. Quantitative analysis of MT fluores- locomotion. cence in immunofluorescently stained preparations showed that 100 nM nocodazole had no detectable effect on MT levels and that 300 nM nocodazole only decreased MT levels Key words: cell motility, wound healing, cytoskeleton, video to ~40% of controls. Quantitative analysis of tubulin microscopy, timelapse recording, taxol, vinblastine

INTRODUCTION trophils (Zigmond et al., 1981), MTs do not seem to be immediately necessary for locomotion, since complete breakdown Microtubules (MTs) are major components of the cytoskeleton of MTs does not affect the short term motility of the cells. From that are present in nearly every eukaryotic cell. Previous these studies, it seems unlikely that intact MTs are necessary studies have found that MTs play a role in the directed cell for producing the driving force for motility. Nonetheless, the locomotion of some, but not all cell types. Intact arrays of MTs essential requirement for intact MTs in locomotion of large are clearly necessary for the directed migration of larger cells cells suggests that MTs do play an important role. The corre- such as fibroblasts (Vasiliev et al., 1970; Goldman, 1971; Gail lation between cell size and MT dependence of cell locomo- and Boone, 1971), endothelial cells (Gotlieb et al., 1983), tion (Schliwa and Honer, 1993) suggests that MTs may monocytes (Zakhireh and Malech, 1980), and nerve growth become essential in organizing or contributing to the motile cones (Bamburg et al., 1986). However, in some smaller cells apparatus of the cell as it becomes larger. such as fish keratocytes (Euteneuer and Schliwa, 1986) or neu- Previous studies on how MTs contribute to cell locomotion 3474 G. Liao, T. Nagasaki and G. G. Gundersen were done almost exclusively by MT ‘knock out’ experiments The hypothesis that MTs may be involved in delivering in which complete MT breakdown was induced by high con- membrane components to the leading edge is supported by sub- centrations of MT antagonists. Breakdown of the bulk of the stantial structural data. Yet, despite some effort, there are a MTs results in the loss of polarization of the cell as assessed number of features of this model that have not been demon- by the presence of ruffling around the entire circumference of strated. One concern is that it is not particularly clear that the cell rather than in a localized area (Vasiliev et al., 1970; newly synthesized membrane is necessary for cell locomotion; Goldman, 1971). These results are consistent with the idea that it is possible that membraneous precursors may come from without MTs, the cell is incapable of polarizing its actin recycled or stored sources (Erickson and Trinkaus, 1976). activity and consequently attempts to go in every direction at Also, studies have failed to detect a population of vesicles near once (Vasiliev, 1991). Similar conclusions were drawn from the leading edge that might be candidates for the Golgi-derived experiments using epithelial cells in which the effect of MT insertion vesicles. Other studies designed to measure a antagonists was to cause the misexpression of actin activity (in polarized ‘flow’ of membrane material rearward from the this case microvilli rather than ruffles) on basolateral surfaces leading edge, as would be predicted if components were being of the cell where they are not normally assembled (Achler et inserted selectively at the leading edge, have generally failed al., 1989). At this level, one function of MTs in directed cell to produce evidence that such a flow exists (Holifield et al., locomotion seems to be the establishment and maintenance of 1990; Sheetz et al., 1989; Lee et al., 1990). Finally, additional cell polarity, so that the cell can utilize its actin-based driving ideas need to be added to the polarized membrane insertion force to locomote itself in a unidirectional manner. model to account for the clear effect that MT depolymeriza- The studies described above have explained the role of MTs tion has on polarized actin activity, an activity which is clearly in cell locomotion by emphasizing the depolarizing effect that important for locomotion. depletion of MTs has on actin-based activity. This has been a Is the role of MTs in cell locomotion explained by the idea useful idea for considering the role of MTs in cell locomotion, that MTs serve as tracks for vesicle delivery? Or is this role however, it does not suggest a molecular mechanism by which instead explained by the ability of MTs to establish intracel- MTs contribute to cell motility. It also cannot explain the lular polarity, a function that may require a critical level of ability of MT antagonists to completely block locomotion of MTs? Such questions could be addressed by determining the cells at the edge of a wound (see Gotlieb et al., 1983; also this relationship between MT levels and the rate of cell locomo- study). In such cells, actin activity is naturally constrained to tion. If there is a positive correlation between the MT level the portion of the cell not in contact with surrounding cells. If and the rate of cell locomotion, it would suggest that MTs play MT antagonists redistribute the actin activity that is polarized a direct role in determining the migration rate. Such a role at the front of such cells, it should lead to a reduction in the would be consistent with the notion that MTs serve in the speed of cell locomotion but not a complete inhibition, since transport of rate-limiting materials to the leading edge. An some actin activity would still be found at the front of the cells. alternative hypothesis is that there is a particular level of MTs Indeed, measurements of the effect of MT antagonists on the above which, but not below which, cells can migrate. This membrane ruffling of wound-edge fibroblasts have shown that ‘critical level’ of MTs may be important for the gross organ- ruffling is decreased 4-6 fold over that in controls, but it is not ization and/or maintenance of cellular elements or organelles reduced to zero (Bershadsky et al., 1991). necessary for the locomotion of a cell. For example, this might Other explanations for MT function in cell locomotion include polarizing actin activity (Vasiliev, 1991) or clustering suggest that MTs may polarize the delivery of membrane of the Golgi apparatus so that a functional polarity is main- proteins to the leading edge. This hypothesis is supported by tained (Kreis, 1990; Singer and Kupfer, 1986). numerous studies which have shown that locomoting cells In this study, we have determined the relationship between have a propensity for reorienting their microtubule organizing cellular MT levels and the rate of fibroblast locomotion. To center (MTOC) to a position between the nucleus and the generate cells with varying levels of MTs, we treated wounded leading edge of the cell (Gotlieb et al., 1981, 1983; Kupfer et monolayers of NRK fibroblasts with nocodazole, a MT depoly- al., 1982; Albrecht-Buehler and Bushnell, 1979; Gundersen merizing agent, over a range of concentrations. We then deter- and Bulinski, 1988). Since the Golgi apparatus is localized at mined the rate of cell locomotion and quantified the MT levels or near the MTOC in most cells, the reorientation of the MTOC in these cells. Our results show that there are two components also reorients the Golgi apparatus in the cell (Kupfer et al.. to the inhibition of cell locomotion by nocodazole: at low con- 1982; Nemere et al., 1985). Protein products from the Golgi centrations of nocodazole, ~60% of the cell locomotion can be are envisioned to travel along MT tracks before insertion into inhibited without any detectable change in the level of MTs; the nearest cell border. For migrating cells with an oriented at higher concentrations of nocodazole, the rate of locomotion MTOC, this would result in the preferential delivery of appears to be correlated with MT level. We found no evidence membrane proteins to the leading edge. Evidence for the for a particular critical level of MTs necessary for cell loco- polarized insertion of membrane proteins by a MT-dependent motion. Given the likely effects of low doses of nocodazole on mechanism has been obtained in a series of elegant studies MT dynamics (Wilson et al., 1993; Wilson and Jordan, 1994), using the ts mutant of the G coat protein of VSV virus as a our results suggest that much of the speed of cell locomotion marker for newly delivered membrane glycoproteins is related to the dynamics of MTs, not to the MT level. In (Bergmann et al., 1983; Rogalski and Singer, 1984). Given the further experiments, we show that low concentrations of taxol, preferential orientation of stable, detyrosinated MTs towards which stabilize MTs, also inhibited cell locomotion, support- the leading edge of wound-edge fibroblasts, it is possible that ing the idea that dynamic MTs play an important role in fibro- the delivery of Golgi products could occur on specialized MT blast locomotion. We discuss possible mechanisms by which tracks (Gundersen and Bulinski, 1988; Nagasaki et al., 1992). MTs may contribute to the regulation of cell locomotion rate. Microtubules and cell locomotion 3475

MATERIALS AND METHODS of an Image-1 image analysis program (Universal Imaging Co., West Chester, PA) as previously described (Nagasaki et al., 1994). The Reagents average position of the wound edge was determined from the record- Nocodazole was purchased from Aldrich (Milwaukee, WI); vinblas- ings using images spaced 20 minutes apart. The rate of locomotion tine, saponin, casein and horse plasma-derived serum (PDS) were before and after switching to medium containing drugs or devoid of purchased from Sigma (St Louis, MO); taxol was a generous gift from drugs (for controls), was determined by a linear regression analysis Dr Ven L. Narayanan (National Cancer Institute, Bethesda, MD). using the interval between 60 and 0 minutes before the treatment, and Nocodazole, vinblastine and taxol were prepared as stock solutions in the interval between 0 and 120 minutes after the treatment, respec- DMSO. The final concentration of DMSO in cell culture medium tively. never exceeded 0.1% (v/v), and this concentration of DMSO had no Indirect immunofluorescence effect on NRK cell locomotion or MT levels. Indirect immunofluorescence of fixed cells was performed with a mouse monoclonal antibody (3F3, generously provided by Dr J. Cell culture Lessard, University of Cincinnati) reactive to all β-tubulin isoforms NRK fibroblasts were maintained in DMEM with 10% calf serum as and a rhodamine-conjugated donkey anti-mouse secondary antibody described previously (Nagasaki et al., 1994), and for experiments, (Jackson ImmunoResearch, West Grove, PA) as described (Khawaja cells were seeded on acid-washed glass coverslips (Fisher Scientific, et al., 1988). 25 mm diameter for timelapse recording, 18 mm for immunofluorescence). For preparation of MT polymer fractions, cells were plated Quantification of MT levels by fluorescence microscopy in 35 mm tissue culture dishes. For all experimental manipulations, Immunofluorescently-stained NRK cells were observed with a Nikon cells were incubated for two days to obtain confluent monolayers. Optiphot epifluorescence microscope using a 63× objective lens ‘Wounds’ were then made by scraping narrow strips of cells from the (Zeiss, Planapo Chromat, NA 1.40) and a zoom lens (Nikon 0.9- monolayer with a jeweler’s screw driver. Medium was immediately 2.25×) at 2.25× to further magnify the image sent to the video camera. removed, and the wounded monolayer of cells was rinsed twice with Images of cells at the wound edge were captured with a SIT camera the appropriate fresh medium before proceeding with timelapse (mti-65, Dage-MTI Inc. Precision video, Michigan City, IN) with recording, indirect immunofluorescence, or preparation of MT manual gain and black level controls, adjusted so that background flu- polymer fractions, as described below. For determination of MT orescence was minimal and MTs were the brightest objects in the levels by immunofluorescence, cells were extracted with 37°C PHEM field. Once the gain and black levels of the SIT camera were (60 mM Pipes, 25 mM Hepes, 1 mM EGTA, 2 mM MgCl2, pH 6.95) optimized, they were not changed for a series of measurements. containing 0.5 mg/ml saponin and 5 mg/ml casein (dialyzed against Images were captured and digitized with Image-1 and stored on a PHEM) for 1 minute to remove the monomeric tubulin generated by computer optical drive (Panasonic). Fluorescence intensity was quan- MT depolymerization drugs. One minute was found to be the tified using the ‘area brightness measurement’ function of Image-1. minimum time necessary to remove all the monomeric tubulin. We A rectangular bar (96 pixels × 1 pixel) was overlaid on cells of interest found that inclusion of casein in the extraction buffer reduced the in such a way that the bar was parallel with the leading edge. With background staining when cells were stained as described below. this orientation, virtually all MTs were perpendicular to the long Immediately after extraction, cells were rinsed once in the buffer and dimension of the bar, thus ensuring that individual MTs cross the bar − fixed in 20°C methanol for 5 minutes. Methanol-fixed cells were rather than run along its length. For each measurement, the average rehydrated in Tris-buffered saline for indirect immunofluorescence pixel intensity (API) was recorded for the bar-covered area in the cell, (see below). and the API of the background was measured in a region adjacent to the cell. The background API was subtracted from the API of the cor- Timelapse recording of cell locomotion responding MT measurement. For each nocodazole treatment, the API Timelapse recording was carried out to monitor the effect of MT drugs of the bar-covered area after background substraction was determined on the locomotion of NRK cells. A confluent monolayer of NRK cells for about 50 cells and the mean was calculated. Standard deviation of was wounded with a jeweler’s screw driver, and rinsed several times these measurements fall between 12 and 20% of the mean fluor- with recording medium consisting of DMEM containing 25 mM escence intensity. For each series of experiments, mean MT fluor- HEPES, 100 units/ml penicillin, 100 µg/ml streptomycin, and 10% escence intensity of nocodazole-treated cells was normalized to that PDS, without bicarbonate (pH 7.4). After wounding, the coverslip in untreated control cells. As described in the results, we found that with the wounded monolayer of NRK cells was immediately mounted making measurements in the leading lamella of cells at 60% of the in a Sykes-Moore chamber (volume ~1 ml, Bellco Glass Inc., distance from the nucleus to the leading edge where fluorescence Vineland, NJ) for timelapse recording at 37¡C as previously described intensity was not saturating was most representative of total MT (Nagasaki et al., 1994). Briefly, medium was perfused via two 20 levels. gauge needles inserted through the silicone gasket of the Sykes-Moore chamber. The flow of medium was controlled by a TRIS peristaltic Preparation and quantification of MT polymer levels pump (ISCO Inc. Lincoln, NE) connected to the outlet needle. The MT polymer fractions from control and nocodazole-treated cells were inlet needle was connected to two reservoirs via a three way stopcock. prepared and analyzed by SDS-PAGE and western blotting. A 35 mm MT antagonists were added to one of the reservoirs and 2 hours after dish of confluent NRK cells was used for each sample. The confluent wounding, the flow was switched to the reservoir containing the drug. monolayers of cells were wounded as described above (15 wounds Cells were recorded for 2 hours after addition of the drug and usually were made on each dish). After wounding, cells were incubated at another 2 hours after rinsing out the drug with normal recording 37¡C for 1 hour in DMEM (Gibco) supplemented with penicillin, medium. Phase contrast images using a 20× objective were captured streptomycin, 10 mM Hepes (pH 7.4) and 10% PDS before addition with a video camera (model 70, DAGE-MTI, Michigan City, IN), and of different doses of nocodazole. The incubation with added nocod- recorded using an optical memory disc recorder (Model TQ-2028F, azole was for one hour, and then the cells were rinsed once in PEM Panasonic, Secaucus, NJ) controlled by a desktop computer with a (100 mM Pipes, 10 mM EGTA and 2 mM MgCl2, pH 7.0), extracted program developed in this laboratory (by T. Nagasaki). Images were with PEM containing 0.5 mg/ml saponin for 1 minute, and rinsed recorded at an interval of 2 minutes. twice in PEM buffer (all at 37¡C). After thoroughly removing the last For measurement of the rate of cell locomotion, we determined the rinse, the cells were solubilized in SDS sample buffer containing rate of advance of the wound edge using the ‘track points’ function phenylmethylsulfonyl fluoride. The protein concentration of the MT 3476 G. Liao, T. Nagasaki and G. G. Gundersen polymer fractions was determined with the BCA assay, using BSA as permanent effect on cell locomotion or have any cytotoxic standard (Pierce, Rockford, IL). An equal amount of protein (10 µg) effects on the cells. from each sample was run on SDS-PAGE with tubulin standards and A plot of the distance migrated vs time after wounding is then western-blotted as described (Gundersen et al., 1994). shown in Fig. 2 for another experiment (with 300 nM nocod- β Blots were incubated in monoclonal antibody (3F3) to -tubulin as azole). Such a plot shows two significant aspects of the effect described by Gundersen et al. (1994), except that 1/7,500 alkaline of nocodazole on the locomoting cells. One is that the inhibi- phosphatase-conjugated goat anti-mouse IgG was used as the secondary antibody. After development with 5-bromo-4-chloro-3- tion of locomotion by nocodazole occurred rapidly; within a indolylphosphate and p-nitro-tetrazolium blue, blots were dried in the few minutes of addition of nocodazole, the cells have adopted dark and imaged with a Newvicon video camera (DAGE-mti, 70 a new rate of locomotion. The second important result is that series). After digitization and background subtraction with Image-1, the rate of locomotion was constant before, during, and after the density of the bands was quantified with the ‘area brightness measurement’ function of the Image-1 program. The tubulin standard curve obtained by this method was linear between the range of 25 ng and 200 ng of tubulin, and experimental samples were loaded so that the tubulin levels fell within this range.

RESULTS

Low concentrations of nocodazole decrease the rate of locomotion of NRK ﬁbroblasts We made timelapse recordings to examine the effects of nocodazole on NRK cells migrating into a wound. Before recording, we wounded conﬂuent monolayers of NRK cells (see Materials and Methods) to obtain a uniform population of motile cells. The cells at the wound margin began migration into the wound almost immediately after the wounding and then maintained a constant rate of locomotion. We found that this rate remained constant for several hours (see Nagasaki et al., 1994). Because these wound-edge cells are only capable of migrating in one direction, our study is focused only on the directed translocation activity of cells rather than the random movement of sparse cells, which we found to be intermittent in both duration and direction of movement (unpublished observations). In a series of experiments, we determined the rate of locomotion before and after different concentrations of nocodazole (25 nM to 400 nM) were perfused into cells in a Sykes-Moore chamber. Fig. 1 shows an example of a timelapse recording of cells migrating into the wound before and after treatment with nocodazole (in this case, 400 nM). As can be seen from the timelapse sequence (Fig. 1), addition of 400 nM nocodazole resulted in a complete inhibition of cell locomotion that was maintained throughout the nocodazole treatment (у2 hours). If nocodazole was subsequently removed from the medium, cells resumed their migration at about the same rate as before nocodazole addition, indicating that nocodazole did not have a

Fig. 1. Timelapse recording of NRK fibroblasts migrating into an in vitro wound before and after treatment with nocodazole. Confluent NRK monolayers were wounded and the locomotion of cells into the wound was monitored by timelapse recording. Wounded monolayers were perfused with recording medium for 1 to 2 hours before switching to medium containing nocodazole, and cell locomotion in the presence of nocodazole was monitored for ∼2 hours. At this point, nocodazole was washed out with fresh recording medium and the recording continued for another 2 hours. Shown is a representative recording of NRK cell locomotion before and after treatment with 400 nM nocodazole (only ~1/4 of the field that was recorded is shown). Elapsed time (in minutes) and the time points of the addition (‘+NZ’) and wash-out (‘−NZ’) of nocodazole are indicated at the upper right corner of each panel. Bar, 35 µm. Microtubules and cell locomotion 3477

values for the rate of locomotion in different experiments were 24.7±3.9 µm/hr (n=21). Nocodazole treatment resulted in a dose-dependent decrease in the rate of locomotion (Fig. 3). We detected a significant inhibition of cell locomotion with concentrations of nocodazole as low as 25 nM (74.4±7.2%, n=2) and the inhibition was more than 60% with 100 nM nocodazole (39.9±4.2%, n=3). Cells stopped forward movement completely after the addition of ≥300 nM nocodazole. Since we perfused cells at a higher rate during nocodazole addition, we checked to see if perfusion of medium without nocodazole (mock treatment) affected cell locomotion. Cells after mock treatment showed only a slight decrease in the locomotion rate (93.6±1.9%, n=3; Fig. 3) and this difference was not statisti- cally significant (P=0.34, Student’s t-test). Fig. 2. A trace of the distance migrated vs time after wounding of To determine if the effect of nocodazole on cell locomotion NRK cells before, during, and after treatment with nocodazole (300 was attributable to its action on MTs rather than some other nM, in the experiment presented here). Nocodazole was added at 120 action of nocodazole, we tested the effect of low concentra- minutes after wounding (‘+NZ’) and washed out at 240 minutes after tions of vinblastine on NRK cell locomotion in an identical wounding (‘−NZ’). The average position of the wound edge was manner. Vinblastine is another MT antagonist that has a com- determined from timelapse recordings using images spaced 20 pletely different structure from nocodazole (reviewed by minutes apart (see Materials and Methods). Dustin, 1984). Previous studies have shown that vinblastine induces depolymerization of MTs and inhibition of cell proliferation at concentrations substantially lower than those of the nocodazole treatments. We found that the rate of locomo- nocodazole (Jordan et al., 1992). We found that vinblastine at tion analyzed by linear regression had a correlation coefficient a concentration of 5.5 nM reduced the rate of NRK cell loco- that varied by less than 5% during these intervals. Accordingly, motion by 78% (data not shown). This supports the idea that we used a single rate to represent the cell locomotion over the nocodazole affects cell locomotion through its action on MTs. treatment intervals. Further evidence supporting this conclusion is presented In reporting the effect of nocodazole on cell locomotion, we below. expressed the rate of locomotion as the percentage of the control rate determined in each experiment. This control value Effects of nocodazole on the level of MTs in NRK was the rate of locomotion in the 60 minutes prior to addition cells of nocodazole and was determined for each recording. Control Nocodazole has long been established as a MT antagonist which depolymerizes MTs by blocking MT assembly (Hoebeke et al., 1976). More recent in vitro studies have shown that nocodazole decreases the MT dynamic instability parameters of growth rate and shortening rate, and increases the time MTs spend in pause (neither growing nor shrinking) (Wilson et al., 1993; Wilson and Jordan, 1994). To determine the effect of different concentrations of nocodazole on MT levels, we undertook three different approaches: visual inspection of cells immunofluorescently stained with anti-tubulin antibody; determination of MT levels by quantitative measurement of fluorescence in immunofluorescently-stained preparations; and quantification of tubulin polymer levels by western blot analysis. Incubation of cells with 100 nM nocodazole, which reduced cell locomotion more than 60%, did not result in any obvious qualitative change in the level or distribution of MTs in cells Fig. 3. Locomotion rates of NRK fibroblasts after treatment with at the wound edge. As assessed by immunofluorescence, in different concentrations of nocodazole. Confluent NRK monolayers controls, as well as in cells treated with 100 nM nocodazole were wounded and the locomotion of cells into the wound before and (for 1 hour), MTs appear to fill the cytoplasm and extend after nocodazole treatment was monitored as in Fig. 1. Mock virtually the entire distance to the leading edge of the cell (Fig. treatment of cells was performed in an identical manner except that 4a and b). The close approach of the MTs to the leading edge the medium did not contain nocodazole. The rate of cell locomotion in both controls and 100 nM nocodazole-treated cells is shown before nocodazole or mock treatment was determined and used as for representative cells in Fig. 5. By comparing the MT distri- the 100% level in each experiment. Rates of cell locomotion were determined by analyzing timelapse recordings as described in the butions in the immunofluorescence images (Fig. 5a and c) with Materials and Methods and expressed as percentage of the control the phase images of the leading edges (Fig. 5b and d, respec- rate. Error bars are standard deviations from 2 determinations for 25 tively), it is clear that a number of MTs in both types of cells nM nocodazole, and 3 or more determinations for the rest of the data closely approach the limiting membrane of the cell edge. In points. cells treated with 200 nM or 300 nM nocodazole (Fig. 4c and 3478 G. Liao, T. Nagasaki and G. G. Gundersen d), MTs were still detected but they appeared fewer in number microscopic assay that is based on measuring the fluorescence and were shorter than in controls. Cells treated with ≥200 nM intensity of MTs in cells that have been immunofluorescently nocodazole were also less extended than control cells, which stained with tubulin antibodies (see Materials and Methods). may be related to the presence of shorter MTs. We did not Briefly, images of immunofluorescently stained cells obtained observe obvious variations in MT levels during the course of with a SIT camera were digitized, and the MT level was quan- the nocodazole treatment (0.5 hours to 2 hours) (data not tified using the ‘area brightness measurement’ function of the shown), so throughout our study, we measured MT levels in Image-1 program. As described below, by paying attention to cells after 1 hour nocodazole treatment. the cellular location at which the measurement is made, we At the level of resolution afforded by light microscopy, these have been able to derive quantitative values that reflect the observations suggest that low concentrations of nocodazole levels of MTs in cells. (≤100 nM), which were effective in inhibiting cell locomotion Because of the radial distribution of MTs in cultured cells, (Fig. 3), did not have a significant effect on the distribution of an important consideration in attempting to sample the MT flu- MTs in NRK cells. However, quantitative information con- orescence in stained cells is that MT density is dependent on cerning the level of MTs in cells at the wound edge after such the distance from the centrosome. In the extracted cells that we treatments cannot be obtained from visual inspection of stained used for measurements, the MT array is constrained to essen- MTs. To acquire quantitative data, we have developed a novel tially two-dimensions so the MT density would fall off as the

Fig. 4. Distribution of MTs in motile NRK cells after incubation with different concentrations of nocodazole. Monolayers of NRK cells were wounded and after a recovery period of 1 hour, were further incubated either in the absence or presence of nocodazole for 1 hour. At the end of the incubation, cells were extracted, ﬁxed with methanol and processed for tubulin immunoﬂuorescence staining. (a) Control cells without nocodazole treatment; (b,c,d) cells treated with 100, 200, and 300 nM nocodazole, respectively. Bar, 20 µm. Microtubules and cell locomotion 3479

Fig. 6. Fluorescence intensity vs distance from the nucleus in Fig. 5. Distribution of MTs in the leading edge of motile NRK cells. wound-edge cells immunofluorescently-stained for tubulin. Cells NRK cells were wounded, incubated with or without 100 nM were fixed and stained with an antibody to β-tubulin. Measurements nocodazole, extracted, fixed and immunofluorescently stained for of fluorescence intensity in individual cells were made along a line tubulin as in the legend of Fig. 4. Shown here are the from the nucleus to the leading edge (see text). Shown here are immunofluorescence images of MTs (a and c) and the corresponding individual measurements of 7 cells of different size, shape and phase contrast images of the leading edge (b and d) of representative length. (a) Tubulin fluorescence intensity vs the absolute distance cells. (a and b) Control cells; (c and d) cells treated with 100 nM µ from the nucleus to leading edge. (b) Tubulin fluorescence intensity nocodazole. Bar, 5 m. replotted vs the relative distance from the nucleus to leading edge, with the relative distance of the nucleus edge defined as 0 and the leading edge defined as 100. square of the distance from the centrosome. Accordingly, we made all our measurements of fluorescence intensity with reference to the distance between the centrosome and the relative distance from the nucleus (Fig. 6). As expected for a leading edge. To determine the optimal distance from the cen- radial-based array of MTs, the fluorescence intensity decreased trosome at which to make our measurements, we considered with increased distances from the nucleus. Yet because of dif- two factors: (1) as one moves towards the centrosome, the ference in cell length, the fluorescence intensity decreased at density of MTs increases and at some point, MTs overlap different absolute distances from the nucleus (Fig. 6a). spatially and cannot be resolved by the optical system and/or However, when the data were plotted as fluorescence intensity saturate the detector; and (2) the measurement should be made vs relative distance from the nucleus, despite differences in the as close as possible to the centrosome to include the greatest shape, size and length of cells, the traces fell within a relatively number of MTs, which increases the sensitivity and precision narrow range of values and showed a similar overall shape of the measurement. (Fig. 6b). This shows that the fluorescence intensity of MTs in With these criteria in mind, we first established the rela- a particular region of wound-edge cells was predictably related tionship between MT fluorescence intensity and the distance to the relative distance from the nucleus. This observation was from nucleus. The fluorescence intensity within a 1×96 pixel very important to our analysis: it suggested that by making bar (this bar was 1/3-3/4 the width of the cells at the magnifi- measurements of fluorescence intensity in different cells at the cation used; see Materials and Methods) was recorded at every same relative distance from the nucleus, we would be able to pixel along the shortest line from the nucleus to the leading obtain values that accurately represent the MT levels in the edge (phase was used to detect the leading edge). The fluor- cells. The curve of fluorescence intensity vs relative distance escence intensity was plotted against both the absolute and seemed to saturate at distances ≤50% of the distance from the 3480 G. Liao, T. Nagasaki and G. G. Gundersen

Table 1. MT polymer levels in cells treated with different concentrations of nocodazole Tubulin (% of control, Nocodazole (µM) mean ± s.d.) 0 100 0.1 99.1±10.6 0.2 62.6±10.2 0.3 48.1±16.7 0.4 45.2±2.2 10 (−5.9)±5.1

Quantitation of MT levels in cells after treatment with indicated concentrations of nocodazole was carried out by western blot analysis as described in Materials and Methods. The same amount of protein (10 µg) from each sample was loaded on SDS-polyacrylamide gels. The calculated amount of tubulin in samples was normalized to that in the controls. Data shown is obtained from the average of 2 to 4 separate experiments.

Fig. 7. The fluorescence intensity of MTs in wound-edge NRK cells after treatment with different concentrations of nocodazole. MTs throughout the leading edge for untreated cells, and cells Monolayers of NRK cells were wounded, treated with nocodazole treated with nocodazole. and immunofluorescently stained as in the legend of Fig. 4. To measure MT levels with a separate technique, we Immunofluorescence staining patterns of MTs were captured, performed a western blot analysis of tubulin levels in polymer digitized and the fluorescence intensity of MTs at 60% of the fractions prepared from wounded cells treated with different distance from the nucleus to the leading edge were quantified (see concentrations of nocodazole. To quantify the absolute levels Materials and Methods). For each sample, more than 50 cells were measured and the mean of the recorded fluorescence intensities was of MT polymer in these cells, we compared the density of the calculated and normalized to that in untreated control cells. Results bands in immunoblots to that of different amounts of a purified are compiled from 3 separate experiments. brain tubulin standard loaded on the same blot. Such a quantitative analysis of MT polymer in cells treated with different amounts of nocodazole is summarized in Table 1. This analysis nucleus to the cell edge (Fig. 6b). Accordingly, we made all confirmed quantitatively that the MT polymer level showed no our measurements at a point 60% of the distance to the leading detectable change with 100 nM nocodazole. For concentrations edge. above this, decreased polymer was detected, analogous to the The results of quantitative measurements of MT fluores- fluorescence intensity results obtained microscopically. At 10 cence in wound-edge cells treated with various concentrations µM nocodazole, a concentration that effectively depolymerizes of nocodazole are shown in Fig. 7. The MT fluorescence virtually all MTs in the cells as detected by immunofluores- intensity determined for cells after treatment with ≤100 nM cence (not shown), we measured no MT polymer, confirming nocodazole was indistinguishable from that of control cells that the assay was dependent on the presence of polymeric (P=0.723 for 100 nM nocodazole treated cells, Student’s t- MTs and that complete extraction of the monomeric tubulin test). Cells showed reduced levels of MT fluorescence with was achieved by the protocol we used. A point-by-point com- concentrations of nocodazole at ≥150 nM, and with 200 nM parison of the fluorescence and western blotting assays showed nocodazole, cells showed significantly reduced levels of MTs that for any given nocodazole concentration, the two measure- (P<0.0001). Nonetheless, with 300 nM nocodazole, the level ments yielded values that agreed within ~10% (Fig. 7 and of nocodazole that can completely block cell locomotion (see Table 1). Fig. 3), cells still retained ~40% of their MT fluorescence. With 10 µM nocodazole for 1 hour, a treatment that completely Relationship between the levels of MTs and the rate depolymerizes MTs (see below), the cells had a MT fluores- of NRK cell locomotion cence intensity of 9.9% of that in control cells, indicating that The major goal of this study was to explore the role of MTs in the background fluorescence intensity for this microscopic cell locomotion by determining the relationship between the assay is about 10%. levels of MTs and the speed of NRK cell locomotion. We have The above measurements were made at a set distance from replotted the rates of NRK cell locomotion determined at the nucleus. When we checked the fluorescence intensity vs different concentrations of nocodazole against the correspond- relative distance for the entire length from the nucleus to the ing MT fluorescence intensities determined after similar treat- leading edge for cells treated with nocodazole, we found that ments with nocodazole. As shown in Fig. 8, for low nocoda- an averaged trace for cells treated with 100 nM nocodazole, zole concentrations, the rate of directed locomotion of NRK was virtually indistinguishable from that for untreated cells cells was dramatically decreased (up to 60%) even though little (traces from seven cells were summed and averaged for this MT depolymerization was detected. Yet, further decreases in comparison). For cells treated with 300 nM nocodazole, the the rate of locomotion (at higher nocodazole concentrations) shape of the averaged trace was similar to the control curve, did show a close correlation between MT levels and the but the curve was shifted to the left (indicative of lower MT remaining ~40% of the rate of locomotion (Fig. 8). These levels). Within the limitations of this analysis, these results results suggest that the contribution of MTs to cell locomotion confirm that our measurements made at 60% of the distance can be separated into two components: one that is not from the nucleus to the leading edge, reflected the levels of dependent on MT level and one that is. Microtubules and cell locomotion 3481

Fig. 8. Relationship between the levels of MTs and the rate of Fig. 9. Locomotion rates of NRK fibroblasts after treatment with low locomotion in NRK cells. Locomotion rates of NRK fibroblasts after concentrations of taxol. Confluent NRK monolayers were wounded treatment with different concentrations of nocodazole (0-400 nM) and the locomotion of cells into the wound was monitored by were obtained as in Fig. 3. These locomotion rates were replotted timelapse recording. Wounded monolayers were perfused with against the level of MT fluorescence intensity at the same recording medium for approximately 2 hours before switching to nocodazole concentrations (from Fig. 7). The numbers in parentheses medium containing taxol, and the recording continued for another 2- next to the individual points represent the concentration of 3 hours. Mock treatment of cells was performed in an identical nocodazole (in nM) used to derive the point. manner as the taxol-treated cells, except that the medium did not contain taxol. Wounding, recording, and rates of cell locomotion were determined as in the legend of Fig. 3. Error bars are standard Effects of low concentrations of taxol on NRK cell deviations of means as determined in 2-4 separate experiments. locomotion Our quantitative analysis of the effects of nocodazole on NRK cell locomotion and on the levels of MT polymer demonstrated study, we set out to extend these hypotheses by testing two that low concentrations of nocodazole interfered with the ideas which could explain the involvement of MTs in cell loco- directed locomotion of NRK cells with little or no detectable motion, namely, whether there is a positive correlation between MT depolymerization. One explanation for these results is that the MT level and the rate of cell locomotion, or whether there low concentrations of nocodazole influenced the dynamic is a critical level of MTs required for cell locomotion. To behavior of the MTs. To test this idea, we determined the address these questions, we established an experimental system effects of low concentrations of taxol on the rate of NRK cell in which we could use different concentrations of nocodazole locomotion. Taxol is an antimitotic drug that stabilizes MT by to alter the level of MTs in the cell. With this system, we deter- binding to MT polymer (Schiff et al., 1979) and low concen- mined the relationship between the levels of MTs and the rate trations of taxol have been reported to decrease the dynamic of locomotion in NRK fibroblasts (Fig. 8). Surprisingly, we instability of MTs in vitro (Jordan et al., 1993; Derry et al., observed that with low concentrations of nocodazole, inhibi- 1995). If the dynamics of MTs is involved in the regulation of tion of NRK cell locomotion (60%) occurred before significant directed locomotion of fibroblasts, then low concentrations of changes in MT levels were induced. Moreover, there seemed taxol should interfere with the locomotive behavior of fibro- to be a close correlation between the MT levels and the blasts. In wounded monolayers of NRK cells, 10 nM taxol remaining 40% of the rate of cell locomotion (Fig. 8). These reduced the rate to 65% of the control rate and 50 nM taxol results do not support the notion that there is a critical level of reduced the rate to 18% of the control rate (Fig. 9). These con- MTs required for cell locomotion. In contrast, they suggest that centrations of taxol have been reported to have little effect on MTs may be involved in more than one process related to cell MT polymer level in HeLa cells (Jordan et al., 1993). Taxol at locomotion (one unrelated to MT level and one related to MT 100 nM almost completely blocked forward locomotion (Fig. level). There is reason to think that the inhibition of cell loco- 9). These results, combined with the observation that low con- motion at nocodazole concentrations that do not significantly centrations of nocodazole decreased NRK cell locomotion affect MT polymer levels is due to nocodazole effects on MT without changing MT level, support the idea that dynamic MTs dynamics (see below for more discussion), so these results are rate-limiting for a step or steps in fibroblast locomotion. suggest that MT dynamics are an important element in fibroblast locomotion.

DISCUSSION Measurements of MT levels in cells The level of MTs in cells treated with nocodazole was quanti- Relationship between MT level and cell locomotion tated by two independent approaches: quantitation of immuno- Current models for the function of MTs in directed cell loco- fluorescence images using digital video microscopy and quan- motion have emphasized the role of MTs in the establishment titative western blot analysis. When we analyzed the cells that and maintenance of cell polarity (reviewed by Singer and had been treated with 100 nM nocodazole, both the micro- Kupler, 1986; Vasiliev, 1991). These ideas are based in part scopic method and the biochemical method showed that the on experiments in which the complete breakdown of MTs was MT level in these cells did not change significantly from that induced by high concentrations of MT antagonists. In this in untreated control cells. We cannot, however, rule out the 3482 G. Liao, T. Nagasaki and G. G. Gundersen possibility that there were subtle changes in the MT level in with the nocodazole data. The idea that MT dynamics are cells treated with 100 nM nocodazole, since the error in deter- critical for cell locomotion is not adequately accounted for by mining MT levels by either method is in the range of ±15%. models which propose that MTs serve as tracks for delivering Nonetheless, even if 100 nM nocodazole decreased the MT membrane components to the leading edge or that MTs polarize levels by 15%, the rate of locomotion was decreased by a actin activity (see Introduction). We hypothesize that dynamic greater proportional amount (∼60%). Thus, based on these MTs contribute to cell locomotion by interacting with transient measurements, it is unlikely that the highest speed of NRK structures or assemblies (‘targets’), whose stabilization or mat- fibroblast locomotion is directly related to MT levels. uration is rate-limiting for cell locomotion. Without interaction with MTs, the transient structures are not stabilized or matured Mechanism of inhibition of fibroblast locomotion by and cannot contribute to the advancement of the cell. By low concentrations of nocodazole dampening MT dynamics, nocodazole (and taxol) would There are several mechanisms that could account for the decrease the chance of MTs hitting their targets and would thus inhibitory effect of low concentrations of nocodazole on cell inhibit locomotion. At higher concentrations, nocodazole may locomotion in the absence of detectable effects on MT levels. also limit access of the MTs to their putative targets by decreas- These include: (1) non-specific effects of nocodazole; (2) a sub- ing the length of the MTs. population of MTs sensitive to nocodazole and important for cell While it is hard to know what targets the dynamic MTs might locomotion; (3) subtle changes in MT length; or (4) altered MT interact with without further experiments, three possibilities are dynamics. The first possibility is unlikely since low concentra- worth considering: (1) lamellipodial protrusions that advance tions of vinblastine, another MT depolymerizing agent, also the front of the cell; (2) nascent focal adhesion plaques that inhibited NRK cell locomotion, implying that nocodazole affects provide traction for forward movement; and (3) putative MT- cell locomotion most probably through its action on MTs rather stabilizing sites. Both lamellipodial protrusions and focal than through some unique, unknown effect of nocodazole. This adhesion plaques form transiently at the front of the cell and are idea is also made unlikely by our results with taxol. The second necessary for fibroblast locomotion. The maturation and/or possibility can be envisaged if there were a subset of MTs that maintenance of these structures might be essential for cell loco- was particularly sensitive to nocodazole and was particularly motion and MTs might provide the necessary maturing or sta- important for cell locomotion. However, this mechanism cannot bilizing effect on them. Another possible target for dynamic explain the observation that low concentrations of taxol, an agent MTs are putative factors involved in MT stabilization. with well-characterized MT-stabilizing effects (Schiff et al., According to this idea, it is not the structures that the dynamic 1979), also inhibited locomotion. The possibility that low con- MTs interact with that are important for locomotion, but the centrations of nocodazole decreased MT lengths without causing effect this interaction has on the MTs themselves. Once stable reduction in MT levels is a viable possibility. However, any MTs are formed by interacting with the target, they might serve shortening of MTs would have to be subtle, otherwise we would an important function for locomotion, e.g. as preferred tracks have detected it as an obvious decrease in the approach of MTs for the transport of vesicles to the leading edge. Stable, dety- to the leading edge; in both control and 100 nM nocodazole- rosinated MTs in motile fibroblasts have been shown to be pref- treated cells, we observed MTs to contact the leading edge of erentially oriented towards the direction of locomotion the cell (see Fig. 5). These results suggest that the fourth (Gundersen and Bulinski, 1988; Nagasaki et al., 1992). In other mechanism, i.e. dampened MT dynamics, is the most likely studies, we have found that stable, detyrosinated MTs are the explanation for the decreased locomotion by low concentrations preferred sites of interaction for other cellular organelles such of nocodazole. This mechanism is supported by recent studies as intermediate filaments (G. Gurland and G. G. Gundersen, which have shown that MT drugs such as nocodazole, vinblas- unpublished). While these data support a role for stable MTs in tine and taxol decrease the dynamic instability of MTs in vitro cell locomotion, this idea still needs to be tested directly. (Jordan et al., 1993; Wilson et al., 1993; Wilson and Jordan, Our results suggest that dynamic MTs are necessary for 1994; Toso et al., 1993; Derry et al., 1995) and that vinblastine fibroblast locomotion; nonetheless, we also observed a strong affects MT dynamics in vivo (Iyengar et al., 1993; Tanaka and correlation between the MT level and a portion of the rate of Kirschner, 1995). Moreover, when used at low concentrations in cell locomotion (Fig. 8). This result is consistent with the idea vivo, these MT drugs interfere with the mitotic spindle in HeLa that part of the contribution of MTs to cell locomotion is due cells without affecting the MT polymer level (Jordan et al., 1992, to the transport of rate-limiting materials to the leading edge 1993). Tanaka et al. (1995), showed that growth cone motility during cell locomotion. However, with the higher concentra- was decreased with concentrations of vinblastine that were tions of nocodazole that affect MT polymer levels, it is also confirmed to decrease MT dynamics, although they did not conceivable that MT dynamics were further dampened. Even measure the effect of vinblastine on MT levels. We are currently if no further reduction in individual MT dynamics resulted measuring the effects of low concentrations of nocodazole on from higher nocodazole concentrations, according to our the parameters of dynamic instability of MTs in vivo. ‘transient target’ model, a reduction in MT number would lead to a further reduction in MTs interacting with the transient How can MTs regulate cell locomotion rate? target, resulting in a further decrease in locomotion. Thus, the Our data suggest that the dynamics of MTs could regulate the correlation in MT levels and locomotion rate, which we locomotion of NRK cells. We showed that low concentrations observed for nocodazole concentrations ≥150 nM, is consistent of nocodazole inhibited NRK cell locomotion without signifi- with a role for MTs in delivering vesicles to the leading edge cantly affecting the level of MTs. The inhibition of locomotion or in interacting with transient targets. by low concentrations of another MT-depolymerizing drug In summary, we have provided evidence that MTs play an (vinblastine) and a MT-stabilizing drug (taxol) is consistent important role in determining the rate of fibroblast locomotion. Microtubules and cell locomotion 3483

Our data establish the relationship between MT levels and Jordan, M. A., Thrower, D. and Wilson, L. (1992). Effects of vinblastine, locomotion rate and suggest that MT dynamics is a rate- podophyllotoxin and nocodazole on mitotic spindles. Implications for the limiting factor in the regulation of fibroblast locomotion. These role of microtubule dynamics in mitosis. J. Cell Sci. 102, 401-416. Jordan, M. A., Toso, R. J., Thrower, D. and Wilson, L. (1993). Mechanism data point to the notion that MTs may act to interact with of mitotic block and inhibition of cell proliferation by taxol at low transient targets in the leading edge of motile cells and we are concentrations. Proc. Nat. Acad. Sci. USA 90, 9552-9556. currently examining this idea. Khawaja, S., Gundersen, G. G. and Bulinski, J. C. (1988). Enhanced stability of microtubules enriched in detyrosinated tubulin is not a direct We thank Beth Prairie for contributing to the initial phases of this function of detyrosination level. J. Cell Biol. 106, 141-149. project. G. Liao was supported by a NIH Training Grant (NIAAC Kreis, T. E. (1990). Role of microtubules in the organization of the Golgi apparatus. Cell Motil. Cytoskel. 15, 67-70. 00189). This project was supported by a grant from NIH (GM 42026) Kupfer, A., Louvard, D. and Singer, S. J. (1982). Polarization of the Golgi to G. G. Gundersen. apparatus and the microtubule-organizing center in cultured fibroblasts at the edge of an experimental wound. Proc. Nat. Acad. Sci. USA 79, 2603-7. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of REFERENCES the head of bacteriophage T4. Nature 227, 680-685. Nagasaki, T., Chapin, C. J. and Gundersen, G. G. (1992). Distribution of detyrosinated microtubules in motile NRK fibroblasts is rapidly altered upon Achler, C., Filmer, D., Nerte, C. and Drenckhahn, D. (1989). Role of cell-cell contact: implications for contact inhibition of locomotion. Cell microtubules in polarized delivery of apical membrane proteins to the brush Motil. Cytoskel. 23, 45-60. border of the intestinal epithelium. J. Cell Biol. 109, 179-189. Nagasaki, T., Liao, G. and Gundersen, G. G. (1994). Isolated plasma Albrecht-Buehler, G. and Bushnell, A. (1979). The orientation of centrioles membranes induce the loss of oriented detyrosinated microtubules and other in migrating 3T3 cells. Exp. Cell Res. 120, 111-118. contact inhibition-like responses in migrating NRK cells. J. Cell Sci. 107, Bamburg, J. R., Bray, D. and Chapman, K. (1986). Assembly of 3413-23. microtubules at the tip of growing axons. Nature 321, 788-790. Nemere, I., Kupfer, A. and Singer, S. J. (1985). Reorientation of the Golgi Bergmann, J. E., Kupfer, A. and Singer, S. J. (1983). Membrane insertion at apparatus and the microtubule-organizing center inside macrophages the leading edge of motile fibroblasts. Proc. Nat. Acad. Sci. USA 83, 1367- subjected to a chemotactic gradient. Cell Motil. 5, 17-29. 1371. Rogalski, A. A. and Singer, S. J. (1984). Associations of elements of the Golgi Bershadsky, A. D., Vaisberg, E. A. and Vasiliev, J. M. (1991). Pseudopodia apparatus with microtubules. J. Cell Biol. 99, 1101-09. activity at the active edge of migrating fibroblasts is decreased after drug- Sheetz, M. P., Turney, S., Qian, H. and Elson, E. L. (1989). Nanometre-level induced microtubule depolymerization. Cell Motil. Cytoskel. 19, 152-158. analysis demonstrates that lipid flow does not drive membrane glycoprotein Derry, W. B., Wilson, L. and Jordan, M. A. (1995). Substoichiometric movements. Nature 340, 284-288. binding of taxol suppresses microtubule dynamics. Biochemistry 34, 2203- Schiff, P. B., Fant, J. and Horwitz, S. B. (1979). Promotion of microtubule 2211. assembly in vitro by taxol. Nature 277, 665-667. Dustin, P. (1984). Microtubules. Berlin: Springer-Verlag. Schliwa, M. and Honer, B. (1993). Microtubules, centrosomes and Erickson, C. A. and Trinkaus, J. P. (1976). Microvilli and blebs as sources of intermediate filaments in directed cell movement. Trends Cell Biol. 3, 377- reserve surface membrane during cell spreading. Exp. Cell Res. 99, 375-384. 381. Euteneuer, U. and Schliwa, M. (1986). The function of microtubules in Singer, S. J. and Kupfer, A. (1986). The directed migration of eukaryotic directional cell movement. Ann. NY Acad. Sci. 466, 867-886. cells. Annu. Rev. Cell Biol. 2, 337-365. Gail, M. H. and Boone, C. W. (1971). Effect of colcemid on fibroblast Tanaka, E., Ho, T. and Kirschner, M. W. (1995). The role of microtubule motility. Exp. Cell Res. 65, 221-227. dynamics in growth cone motility and axonal growth. J. Cell Biol. 128, 127- Goldman, R. D. (1971). The role of three cytoplasmic fibers in BHK-21 cell 137. motility. I. Microtubules and the effects of colchicine. J. Cell Biol. 51, 752- Toso, R. J., Jordan, M. A., Farrell, K. W., Matsumoto, B. and Wilson, L. 762. (1993). Kinetic stabilization of microtubule dynamic instability in vitro by Gotlieb, A. I., May, L. M., Subrahmanyan, L. and Kalnins, V. I. (1981). vinblastine. Biochemistry 32, 1285-93. Distribution of microtubule organizing centers in migrating sheets of Vasiliev, J. M., Gelfand, I. M., Domnina, L., Ivanova, O. Y., Komm, S. G. endothelial cells. J. Cell Biol. 91, 589-94. and Olshevskaja, L. V. (1970). Effect of colcemid on the locomotory Gotlieb, A. I., Subrahmanyan, L. and Kalnins, V. I. (1983). Microtubule- behavior of fibroblasts. J. Embryo. Exp. Morph. 24, 625-640. organizing center and cell migration: effect of inhibition of migration and Vasiliev, J. M. (1991). Polarization of pseudopodial activities: cytoskeletal microtubule disruption in endothelial cells. J. Cell Biol. 96, 1266-1272. mechanisms. J. Cell Sci. 98, 1-4. Gundersen, G. G., Kim, I. and Chapin, C. J. (1994). Induction of stable β Wilson, L., Toso, R. J. and Jordan, M. A. (1993). Vinblastine, nocodazole, microtubules in 3T3 fibroblasts by TGF- and serum. J. Cell Sci. 107, 645- and colchicine suppress the dynamic instability of microtubules: 659. implications for the mechanism of antimitotic action. Cell. Pharmacol. Gundersen, G. G. and Bulinski, J. C. (1988). Selective stabilization of (Suppl.) I, S35-S40. microtubules oriented toward the direction of cell migration. Proc. Nat. Wilson, L. and Jordan, M. A. (1994). Pharmacological probes of microtubule Acad. Sci. USA 85, 5946-5950. function. In Microtubules (ed. Hyams and Lloyd), pp. 59-83. Wiley-Liss. Hoebeke, J., Van Nijen, G. and DeBrabender, M. (1976). Interaction of Zakhireh, B. and Malech H. (1980). The effect of colchicine and vinblastine oncodazole (R17934) a new antitumoral drug, with rat brain tubulin. on the chemotactic response of human monocytes. J. Immunol. 125, 2143- Biochem. Biophys. Res. Commun. 69, 319-324 2153. Holifield, B. F., Ishihara, A. and Jacobson, K. (1990). Comparative behavior Zigmond S. H., Levitsky, H. I. and Kreel, B. J. (1981). Cell polarity: an of membrane protein-antibody complex on motile fibroblasts: Implications examination of its behavioral expression and its consequences for for a mechanism of capping. J. Cell Biol. 111, 2499-2512. polymorphonuclear leukocyte chemotaxis. J. Cell Biol. 89, 585-92. Iyengar R., Jordan, M. A., Thrower, D., Wendell, K., Sheldon, E., Wilson, L. and Wadsworth, P. (1993). Low concentrations of vinblastine suppress microtubule dynamic instability in vivo. Mol. Biol. Cell (Suppl.) 4, 50a. (Received 9 March 1995 - Accepted 27 July 1995)