Microtubule Dynamics in the Chromosomal Spindle Fiber: Analysis by Fluorescence and High-Resolution Polarization Microscopy

Transcription

1 Cell Motility and the Cytoskeleton 10: (1988) Microtubule Dynamics in the Chromosomal Spindle Fiber: Analysis by Fluorescence and High-Resolution Polarization Microscopy L. Cassimeris, S. Inoue, and E. D. Salmon Department of Biology, University of North Carolina, Chapel Hill (L. C., E. D. S.) and Marine 5iological Laboratory, Woods Hole, Massachusetts (S. 1.) We describe preliminary results from two studies exploring the dynamics of microtubule assembly and organization within chromosomal spindle fibers. In the first study, we microinjected fluorescently labeled tubulin into mitotic PtKl cells and measured fluorescence redistribution after photobleaching (FRAP) to determine the assembly dynamics of the microtubules within the chromosomal fibers in metaphase cells depleted of nonkinetochore microtubules by cooling to C. FRAP measurements showed that the tubulin throughout at least 72% of the microtubules within the chromosomal fibers exchanges with the cellular tubulin pool with a half-time of 77 sec. There was no observable poleward flux of subunits. If the assembly of the kinetochore microtubules is governed by dynamic instability, our results indicate that the half-life of microtubule attachment to the kinetochore is less than several min at C. In the second study, we used high-resolution polarization microscopy to observe microtubule dynamics during mitosis in newt lung epithelial cells. We obtained evidence from 150-nm-thick optical sections that microtubules throughout the spindle laterally associate for several sec into rods composed of a few microtubules. These transient lateral associations between microtubules appeared to produce the clustering of nonkinetochore and kinetochore microtubules into the chromosomal fibers. Our results indicate that the chromosomal fiber is a dynamic structure, because microtubule assembly is transient, lateral interactions between microtubules are transient, and the attachment of the kinetochores to microtubules may also be transient. Key words: mitosis, kinetochore, video microscopy INTRODUCTION Our understanding of the mechanisms of microtubule assembly and chromosome movement have been significantly advanced by approaches using light microscopy. A major advantage of these approaches is that processes can be observed in real time. Two light microscopy approaches have been particularly valuable for visualizing the dynamics of microtubules in mitotic spindle fibers in living cells: fluorescence and polarization microscopy. In this paper, we describe preliminary results from experiments using fluorescence and polarization microscopy to examine the dynamics of assembly and association of microtubules comprising the chromo- somal spindle fibers, primarily during the prometaphasemetaphase stages of mitosis. The bipolar mitotic spindle is formed from overlapping arrays of microtubules extending from opposite spindle poles. Each half-spindle is composed of 100s-1,000s of microtubules, all oriented with their (+), or fast growing ends, distal from the spindle pole. Usually, the majority of microtubules within each half- Received January 5, 1988; accepted January 12, Address reprint requests to L. Cassimeris, Dept. of Biology, University of North Carolina, Chapel Hill, NC Alan R. Liss. Inc.

2 186 Cassimeris et al. spindle are nonkinetochore, polar microtubules. A smaller proportion of microtubules attaches to the kinetochores of the chromosomes; they are termed kinetochore microtubules. For example, each half-spindle in PtKl cells at 37 C contains approximately 1,300 nonkinetochore microtubules and about 275 kinetochore microtubules (25 kinetochore microtubules per chromosome) [McIntosh et al., 1975; Rieder, 1981bl. The spindle fiber that links a kinetochore to a pole has been termed in the literature either a chromosomal fiber or a kinetochore fiber. We will use the term chromosomal fiber to define the fiber, visible by fluorescence or low-magnification polarization microscopy, which runs between the kinetochore and the pole. This term was chosen, rather than kinetochore fiber, to avoid implying that we are specifically referring to kinetochore microtubules. The chromosomal fiber is a complex of kinetochore and nonkinetochore microtubules, but the percentage of kinetochore and nonkinetochore microtubules within the fiber has not been well established. Serial reconstruction of electron micrographs revealed, for the cold-stable chromosomal fiber of PtKl cells, that about 50% of the microtubules extend uninterrupted from the pole to the kinetochore. About 25% extended from the kinetochore but ended before reaching the pole. The remaining microtubules were not associated with the kinetochore [Rieder, b]. Observations of changes in spindle birefringence by polarization microscopy have established that the majority of spindle microtubules exist in a labile, dynamic equilibrium with a cellular pool of tubulin subunits [InouC, 1981al. A variety of agents, such as cold temperatures or high hydrostatic pressure, induce rapid, reversible depolymerization of the nonkinetochore spindle microtubules [InouC, 1981a; Salmon et al., 1984al. Kinetochore microtubules are differentially more stable to depolymerization by these conditions. Fluorescence techniques have also provided valuable information on the rate and pattern of microtubule assembly in living cells, and these results have been reviewed extensively elsewhere [Saxton et al., 1984; McIntosh et al., 1985; Wadsworth and Salmon, 1985; Salmon and Wadsworth, 1986; Cassimeris et al., We have used tubulin labeled with the fluorescent probe dichlorotriazinyl-aminofluorescein (DTAF-tubulin) and measurements of fluorescence redistribution after photobleaching (FRAP) to study steady-state microtubule polymerization dynamics during mitosis. Our studies were initiated in collaboration with the laboratory of Dick McIntosh, and similar approaches have been used to analyze microtubule assembly dynamics by these and other labs. Previous results from FRAP experiments and studies following incorporation of labeled subunits have identified dynamic instability [Mitchison and Kirschner, 1984a,b] as the dominant pathway of microtubule assembly for the majority of spindle microtubules, the nonkinetochore microtubules [Cassimeris et al., FRAP studies have shown that the majority of microtubules within the spindle are very dynamic. The half-time of microtubule turnover within the half-spindle at physiological temperature is approximately 20 sec for sea urchin embryos [Salmon et al., 1984b], approximately sec for mammalian tissue culture cells [Saxton et al., 1984; Wadsworth and Salmon, 1985, 1986a1, and sec for newt lung epithelial cells [Wadsworth and Salmon, 1986al. The turnover of the majority of microtubules is extremely rapid, and fluorescence appears to recover uniformly throughout the bleached region in all cell types analyzed [Wadsworth and Salmon, Studies of the sites of tubulin incorporation into microtubules of animal cells have shown that tubulin adds on to the ends of microtubules distal from the centrosome [Soltys and Borisy, 1985; Schulze and Kirschner, 1986; Mitchison et al., Careful analysis of photometrically measured fluorescence recovery curves by Wadsworth and Salmon [1986a] suggested that there was a subpopulation of spindle microtubules that recovered fluorescence at a slower rate than the majority of the spindle microtubules. This subpopulation of more stable microtubules may represent the kinetochore microtubules. Due to the normally low percentage of kinetochore microtubules within the spindle, it has been much more difficult to specifically analyze the assembly dynamics of the kinetochore microtubules within living cells. Several recent experiments have begun to address this question. These studies have been limited to observations of preparations fixed at time points after microinjection of labeled tubulin subunits (incorporation studies) or after photobleaching. Mitchison et al. [ used electron microscopy to follow the incorporation of biotinylated tubulin into microtubules during prometaphase-metaphase in mitotic BSCl cells. One min after injection, labeled tubulin was only seen at the ends of the kinetochore microtubules proximal to the kinetochore. Approximately 10 min later, the kinetochore microtubules were uniformly labeled all along their length. From this data they postulated a poleward flux of subunits in kinetochore microtubules. A major limitation of this study was the inability to follow kinetochore microtubules for significant distances from the kinetochore in adjacent sections. We have conducted similar experiments [Wise et al., in PtKl cells by following DTAF-tubulin incorporation into chromosomal fibers using fluorescence microscopy and digital image processing. Our results confirm the observation of Mitchison et al. [ that initial incorporation occurs at the kinetochore, and

3 Microtubule Dynamics in the Chromosomal Fiber 187 after 10 min the chromosomal fiber appears fully labeled. However, several min after microinjection, the extent of incorporation of labeled tubulin from the kinetochore toward the pole was heterogeneous among chromosome fibers within the same cell and asymmetric between sister chromosomal fibers. The sites of tubulin dissociation from kinetochore microtubules during metaphase have not been previously resolved [Mitchison et al., 1986; Wadsworth and Salmon, 1986al. During anaphase, tubulin subunits appear to dissociate from the kinetochore microtubules at the kinetochore region as the chromosomes move poleward [Mitchison et al., 1986; Gorbsky et al., In this paper we will present the results of FRAP experiments on PtKl cells in which we have reduced nonkinetochore microtubule density by cooling cells to room temperature. These results show that the assembly of microtubules within the chromosomal fibers is dynamic during the prometaphase-metaphase stage of mitosis and that there does not appear to be an obvious poleward flux of subunits within the fiber. In addition, we will describe observations of microtubule dynamics in spindles of newt lung epithelial cells viewed with high-resolution polarization microscopy. These observations provide evidence that the chromosomal fiber is a dynamic cluster of microtubules produced by transient lateral associations between kinetochore and nonkinetochore microtubules. MATERIALS AND METHODS PtKl Cell Culture and Microinjection PtKl cells were grown in F-10 medium supplemented with 10% fetal bovine serum (Flow Labs) and antibiotics (Sigma), ph 7.3. For FRAP experiments, cells were plated onto glass coverslips and used within 2 days. DTAF-labeled tubulin was prepared as described previously [Leslie et al., 1984; Wadsworth and Salmon, 1986al and pressure microinjected into PtKl cells as described by Wadsworth and Salmon [1986a]. Cells were microinjected at room temperature and then returned to the 37 C incubator for 10 min to allow cells to equilibrate after injection. Coverslips with injected cells were then placed in Rose chambers and the chambers filled with room temperature saline G [Wadsworth and Salmon, 1986al. Newt Lung Cell Culture Primary cultures of newt lung epithelia were grown based on the procedures of Boss [ and Rieder et al. [1986]. Briefly, small pieces of excised lung, approximately 3 mm in diameter, were grown in Rose chambers in L- 15 medium diluted 1 : 1 with distilled water and sup- plemented with 5 mm HEPES, 5% whole egg ultrafiltrate (Gibco), 10% fetal bovine serum (Flow Labs), antibiotics, and antimycotics (Sigma), ph Coverslips with mitotic cells were removed from the Rose chambers, mounted on glass slides, and examined by high-resolution polarization microscopy as described below. FLUORESCENCE MICROSCOPY, PHOTOBLEACH- ING, AND DIGITAL IMAGE PROCESSING The FRAP experiments reported here were performed on an optical bench microscope as described in detail elsewhere [Salmon and Wadsworth, 1986; Wadsworth and Salmon, 1986al. Briefly, local regions of spindles were photobleached by exposure to a 0.1 -sec pulse of focused, 488-nm laser light (100-mW power) from a Spectra Physics model 164-argon laser. Recovery of fluorescence within the bleached region was then measured either by photomultiplier recordings (PMT- FRAP) or from digitized video images (Video-FRAP). A diagram of the microscope system is shown in Figure 1. Photometric measurements of fluorescence recovery were made as described by Wadsworth and Salmon [ 1986al. Cells were observed with a Zeiss 40 X Neofluor lens. Fluorescence within a 2.8-pm-diameter circular spot was photobleached with the laser beam focused at the field diaphragm plane of the epi-illuminator using a spherical lens. Fluorescence was measured with a coaxial beam attenuated 10,000-fold compared with the bleaching intensity. The fluorescence image was projected through a pinhole aperture to a photomultiplier (EM1 9863A). An EM1 C-10 photon counter was used to count the number of photons every sec. An Apple I1 Plus Computer was used to store, plot, and analyze the data. Video images of fluorescence recovery within an approximately 1.8-pm-diameter bar pattern bleach were recorded by methods similar to those described previously [Wadsworth and Salmon, 19861, except that images were electronically enhanced by digital image processing. Fluorescence images were projected to a Dage SIT 66 lowlight-level video camera, and images were acquired with a home-built digital image processor (5 12 X 5 12 pixel resolution). The image processor consisted of Max-Video (Data-Cube, Peabody, Mass) digital image processing boards installed in a Force VME bus computer and driven by a processor. Each final image was obtained by summing 32 frames (1-sec integration time). Images were stored on video tape using a Sony VO 5800 H tape recorder. Micrographs were made by photographing the image from a video monitor. Polarization Microscopy Visualization of chromosomal fiber birefringence and measurement of the birefringent retardation of spin-

4 188 Cassimeris et al. / s 3 -; :z 1 PROCESSOR -! - ' C /+... r/t) + m -1 APPLE COUNTER El 1 Fig. I. Schematic diagram of the microscope system used for fluorescence microscopy, laser photobleaching, electronic imaging, and photomultiplier measurements. The design is a modification of an apparatus described in detail by Salmon and Wadsworth [ and Wadsworth and Salmon [ 1986al. L1, argon ion laser; L2, 100-W Hg illuminator; L3, 100-W Hg illuminator or 60-W tungsten illuminator; C, condenser; OB, objective; EF, fluorescence excitation filters; DM, dichroic mirror; BF, fluorescence blocking filters; AIP, aerial image plane; FL, lens that focuses laser beam at AIP; PMT, photomultiplier; BS, beam splitter; P, pin-hole; SI, S2, S3, S4, and S5, shutters; LLL, low-light-level video camera; Apple, Apple I1 computer; DIGITAL IMAGE PROCESSOR, Max-Video image processor. dle fibers requires sensitive polarization microscopy methods. The birefringent retardation (r) of individual chromosome fibers is of the order of = 0.5 nm, or 1/1,000 of the wavelength (A) of green light. The maximum light intensity produced by a birefringent specimen between crossed polars in the polarization microscope is proportional to the illumination intensity and proportional to sin2 (27rr/h). Thus the light intensity corresponding to individual chromosome fibers is only 4 x lop6 the illumination intensity. Detection and measurement of the birefringence of individual chromo- somal fibers requires that the polarization microscope be designed for high extinction, sensitivity, and resolution. This requires the high-numerical-aperature (NA) condenser and objective lenses, an intense light source, high-quality polarizers, and optical components free of birefringence and corrected, using rectifiers [Inoue, 19861, for the rotation of the plane of polarization that occurs at curved optical surfaces. The highest resolution (about 0.2 pm) is obtained with full aperture illumination of Plan Apochromatic objectives whose NA = Image contrast is very low and difficult to detect with the eye when using optical components well corrected for geometrical aberrations and the full objective aperture illumination required for the highest resolution. Fortunately, the contrast of a weakly birefringent specimen can be significantly enhanced using video detectors and electronic contrast enhancement produced by analog [Allen, et al., 1981; InouC, 1981bl and digital methods [InouC, The custom-built high-extinction polarization microscope designed by InouC [ 1986, Figs , ) was used in combination with electronic contrast enhancement techniques to observe and record the dynamic organization of microtubules in the chromosomal fibers in mitotic newt cells. Cells were viewed at low magnification with a 40 x /NA = 0.75 Nikon Rectified objective and at high magnification with a 100 X /NA = 1.4 Nikon DM series Plan Apochromatic objective. The specimen was illuminated using an NA = 1.35 rectified condenser lens whose aperture was fully and uniformly illuminated through a fiberoptic light scrambler with 546-nm green light. Video images were generated in real time, using a Newvicon camera, and digitally processed by an Image I/AT system (Universal Imaging Corporation, Ithaca, NY). Enhancement of image quality was achieved by the subtraction of a stored, out-of-focus background and a jumping average of four frames. Images were recorded with a :-inch VCR as well as with time-lapse recording with a monochrome, high-resolution optical memory disc recorder (OMDR) (Panasonic TQ-2021FBC) once every 4 sec. We tested the ability of this instrument to detect and resolve microtubules by comparing the contrast and resolution of single microtubules assembled in vitro from purified tubulin with that of axonemes purified from sea urchin flagella [Walker et al., A single 25-nm microtubule was seen as a thin black or white thread depending on the orientation of the microtubule to the transmission direction of the slow axis of the compensator. The width of the image of a microtubule was equivalent to the diffraction limitation of the microscope optics. This width was about 200 nm for the NA = 1.4 objective used in our high-resolution observations (data not shown). A 200-nm-diameter axoneme purified from

5 sea urchin sperm flagella gave rise to a diffraction image only slightly wider than the image generated by a single 25-nm-diameter microtubule. However, the contrast of the axoneme was much greater than the contrast of a single microtubule, corresponding to the difference between the mass of a single microtubule and the mass of the axoneme (which consists of nine outer microtubule doublets, a central microtubule pair and other accessory proteins). The number of microtubules giving rise to the diffraction image could be estimated from the area under an intensity trace across the diffraction image [InouC, RESULTS AND DISCUSSION FRAP Measurements of Microtubule Dynamics in Mitotic PtKl Spindles at C To analyze steady-state microtubule turnover within the chromosomal fibers of living cells, we have taken advantage of the reduced microtubule density present in PtKl spindles cooled to room temperature (23-24 C). PtKl cells will continue through mitosis at this lower temperature, but at a much slower rate [Rieder, 198 la]. Immunofluorescent and electron microscopic observations have shown that the density of nonkinetochore microtubules within the spindle is greatly reduced at this temperature. Chromosomal fibers are clearly visible either by antitubulin immunofluorescence (data not shown) or by DTAF-tubulin fluorescence visualized in microinjected cells (Fig. 2B). We examined steady-state microtubule turnover within the chromosomal fibers of these cells at room temperature during the prometaphase-metaphase stage of mitosis by FRAP analysis. Photometric measurements were used to determine the rate and extent of fluorescence recovery following photobleaching. A typical computer record of fluorescence recovery within a photobleached 2.8-~m-diameter circular spot is shown in Figure 2A. The kinetic constants derived from the recovery curves are summarized in Table I. Fluorescence recovery at room temperature occurred with a half-time of approximately 77 sec and was slower than the sec half-time of recovery reported for metaphase PtKl spindles at 37 C by Saxton et al. [ The slower rate of fluorescence recovery at room temperature is most likely a reflection of both the temperature dependence of the rate of microtubule turnover and the increased proportion of the more stable kinetochore microtubules. The extent of recovery (approximately 70%, Table I) suggests that the great majority of microtubules are still dynamic under these conditions. Fluorescence recovery within a bar bleach pattern was examined to determine whether there was a visible Microtubule Dynamics in the Chromosomal Fiber 189 flux of subunits within the chromosomal fibers. The report by Mitchison et al. [1986] predicts that the bar pattern should move poleward during the recovery phase. As shown in Figure 2B, fluorescence appears to recover uniformly within the bar pattern, with no detectable translocation of the bleached region. Because the bar pattern extends completely across the half-spindle, recovery cannot occur by lateral movement of unbleached microtubules into the bleached region. A major unresolved question is the life-time of a microtubule attachment to the kinetochore. Kinetochore microtubules have been shown to be differentially stable to agents such as cooling, hydrostatic pressure, and colchicine [Rieder, 1981b; Salmon et al., 19761, which promote depolymerization of nonkinetochore microtubules. This differential stability may be due in part to the stabilization of the ( + ) ends of the kinetochore microtubules by their attachment to the kinetochore [Salmon et al., On the other hand, these depolymerization agents could somehow enhance the affinity of the kinetochore for the ends of these microtubules, thus producing nonphysiological stable microtubule attachments. McIntosh and Vigers [ have suggested, from measurements of FRAP for PtKl spindles at 37"C, that tubulin within the kinetochore microtubules exchanges as rapidly as tubulin within the nonkinetochore microtubules. The high density of nonkinetochore microtubules in PtKl spindles at 37 C makes it difficult to measure the dynamics of the kinetochore microtubules by FRAP procedures. At 37"C, the kinetochore microtubules comprise only 20-25% of the total number of microtubules within the half-spindle. Our FRAP results for mitotic PtKl cells depleted of nonkinetochore microtubules indicate that the assembly of microtubules within the chromosomal fibers is dynamic and that there is no observable synchronous poleward flux of subunits within the fibers. The FRAP measurements show that at least 72% of the tubulin within the microtubules of the chromosomal fiber exchanges with tubulin from the cytoplasmic pool with a half-time of 77 sec. If most microtubules in the chromosomal fibers are kinetochore microtubules, our FRAP results support the conclusion of McIntosh and Vigers [ that microtubule attachment at the kinetochore is transient and indicates, to a first approximation, that the half-life of a kinetochore microtubule in a PtKl spindle at 23 C is on the order of 1-2 min. Furthermore, if dynamic instability characterizes the assembly of spindle microtubules, the rate and uniform pattern of FRAP suggcsts that kinctochorc microtubules must continuously bind and release from the kinetochore. This process would occur simultaneously with assembly of new microtubules and capture by kinetochores. Capture of the (+) microtubule ends

6 190 Cassimeris et al. Sec Fig. 2. FRAP of PtKl Cells at C. A: Computer-generated record of PMT-FRAP for a metaphase spindle microinjected with DTAF-tubulin and photobleached as described in Materials and Methods. Photons were counted intermittently in I-sec intervals by opening a manual shutter. Fluorescence measurements, F,,,, were normalized to 100 using the average number of photon counts for five samples preceding the recording of the data. The data is plotted over the first 200 seconds, then during sec (between the first and second vertical bars), and finally between 735 and 765 sec (the second vertical bar to the end). The solid line plotted through the data is an exponential regression line used to calculate the half-time of fluorescence recovery in the microtubule array as described by Wadsworth and Salmon [ 1986aJ. The initial rapid phase of fluorescence recovery is due to the diffusion of tubulin subunits into the bleached region. B: VIDEO-FRAP of a metaphase PtKl cell microinjected with DTAF-tubulin and photobleached at room temperature as described in Materials and Methods. The upper half-spindle was photobleached with a 1.8-pm-wide bar pattern between the chromosomes and the pole (arrowhead). The phase-contrast image of the cell is shown on the left. It was taken 10 sec prior to photobleaching. The fluorescence images during fluorescence recovery after bleaching are shown to the right of the phase-contrast image, and the time after photobleaching is indicated on each frame. Before bleaching, the distribution of fluorescence in the spindle was symmetrical about the metaphase plate. The lower half-spindle tilted slightly out of focus during the recovery period. Bar, 5 pm. by the kinetochore would allow the kinetochore microtubules to persist longer than uncapped microtubules. However, this is only a working hypothesis, since the percentage of kinetochore and nonkinetochore microtubules within the chromosomal fiber is not yet known. Microtubule Dynamics Measured by Polarization Microscopy We have used high-resolution polarization microscopy to examine the organization of microtubules within the chromosomal fibers of mitotic newt lung epithelial

7 Microtubule Dynamics in the Chromosomal Fiber 191 TABLE I. Fluorescence Redistribution After Photobleaching in PtKl Cells at C k" t 1%" %R" No. Metaphase PtK cells ? 33 72? "Determined from plots of In(F, - si,) vs. time after photobleaching (Fig. 2) for fluorescence recovery within the microtubule array, as described in detail elsewhere [Salmon and Wadsworth, 1986; Wadsworth and Salmon, 1986al. K is the first order rate constant, tl/z the half-time of fluorescence recovery, and %R the percent recovery. cells. Figure 3 shows a low-magnification view (40X/NA = 0.75 objective) of chromosomal fibers taken from a time-lapse video recording of a mitotic cell during metaphase through early anaphase. This unusual cell had two additional microtubule organizing centers located near the cell surface along the right side of the cell. Most of the chromosomes attached and oriented between the poles of the central spindle. A clear view of the organization of the birefringent chromosomal fibers was seen for the chromosome that attached to microtubules emanating from the upper pole of the central spindle and the upper microtubule organizing center near the cell surface. The birefringence of the bundle of microtubules in the chromosomal fibers was distinctly greater than the birefringence of the nonkinetochore microtubules surrounding the fibers. Thus the density of microtubules within the fiber is, on average, higher than the average density of microtubules outside the fiber. At metaphase, the birefringent chromosomal fibers appear to have a constant diameter of about 1 pm continuously between the chromosome and the poles. The crosssectional intensity pattern also appears uniform along the length of the chromosomal fibers. This indicates that the average number of microtubules within a cross section of the chromosomal fiber is constant along the length of the fiber. We have not attempted to measure the retardation of the chromosomal fibers in the newt, so an estimate of the number of microtubules clustered within the fiber is unavailable. Note that birefringence extends across the centromere region of the chromosome from one chromosomal fiber to the other. This indicates that a fraction of the microtubules clustered within the chromosomal fibers are nonkinetochore microtubules that extend beyond the kinetochore into the fiber attached to the opposite kinetochore. The birefringence of these microtubules can be seen extending beyond the chromosomes into the interzonal region as the chromosomes separate in anaphase (Fig. 3C). From this low-resolution view of the chromosomal fibers, we can conclude that the chromosomal fiber at metaphase is a bundle of microtubules, about I-pmdiameter, extending between the kinetochore and the Fig. 3. Low-magnification polarization micrographs of a multipolar newt lung epithelial mitotic cell. Frames were selected from a time-lapse video recording. The arrowheads in A denote birefringent sister chromosomal fibers clearly visible because a single chromosome has oriented between the upper spindle pole and a peripheral microtubule organizing center at the upper right. Contrast in the prints was optimized for this fiber. Spindle fibers in the central spindle appear in bright contrast, while the fibers oriented at 90" to the central spindle axis appear in dark contrast. See text for a full description of observations. Time in minutes is given on each frame. Bar in A = 10 pm. pole. The average number of microtubules in a cross section of the fiber appears constant over time and along the length of the fiber. At this level of magnification and resolution, the organization of microtubules within the chromosomal fibers appears to be unchanging or changing slowly during metaphase. At high resolution, the view of the chromosomal fiber was much more dynamic. Figure 4 shows four frames from time-lapse video records of cells viewed with a 100 X /NA = 1.4 objective. The depth of field of these optical sections was 150 nm, and the lateral resolution was about 200 nm. The spindles had a fine fibrillar appearance. In dark contrast, short, thin black rods, whose width was equivalent to the diffraction limit of the microscope, appeared abruptly in view and then abruptly disappeared after a variable length of time (up to several sec) (Fig. 4a and b). Both the length (1-4 pm) and the contrast of the threads were variable. They appeared asynchronously and stochastically throughout the half-spindle region from prometaphase through anaphase. At prometaphase, a birefringent cable, about 250-

8 192 Cassimeris et al. Fig. 4. High-magnification polarization micrographs of newt lung epithelial mitotic cells. Frames were selected from time-lapse video recordings. Images in a, c, and d are from the same cell. The image in b is from a different spindle. These optical sections are approximately 150-nm-thick. a and b: Mid-prometaphase. Birefringent cables, about 250-nm in diameter (large arrows) are seen extending from the kinetochores toward the poles. Within the half-spindle, the short thin black rods obvious on real-time play-back of the time-lapse video records generally appear in these single micrographs as the fine fibrillar texture of the spindle. In b, several rods are visible (small arrows).

9 Microtubule Dynamics in the Chromosomal Fiber 193 Fig. 4. c: Early anaphase. The density of the microtubules in the half-spindle is higher than at prometaphase and the 250-nm-diameter fibers extending Irurri the kinetochores are not obvious. d: Late anaphase. The dynamic lateral associations between microtubules are easiest to visualize at this stage of mitosis because microtubule density in the interzone is reduced. The arrows point to birefringent filaments that may be individual microtubules. The thick, dense black rods are probably overlapped bundles of microtubules of stem-bodies. See text for details. Bar = 10 pm.

10 194 Cassimeris et al. Fig. 5. A schematic drawing of the dynamic lateral associations between spindle microtubules, which is derived from our highresolution polarization micrographs. Lateral associations between kinetochore microtubules from the cable of microtubules attached to the kinetochore. Transient lateral associations between the nonkineto- chore microtubules and the kinetochore microtubules splay apart the kinetochore cable at a site towards the spindle pole. The double arrows indicate that the position of microtubule splaying in the kinetochore cable is dynamic. Rods are clusters of a few microtubules as described in the text. nm-diameter, was seen extending poleward from each kinetochore within the optical section (Fig. 4a and b). These cables were likely to be tight bundles of kinetochore microtubules, since they ended abruptly at the kinetochores. The contrast produced by the birefringence of these cables was similar to the contrast produced by the birefringence of the centrioles at the spindle poles. The average number of kinetochore microtubules for the newt cell, approximately 20 [Rieder, Cassimeris and Salmon, unpublished observations], is similar to the number of microtubules within the centrioles. The width of these kinetochore microtubule cables was similar to the diameter of a newt kinetochore [0.25 km (Rieder and Bajer, 1977)l. By through-focus optical sectioning, several of these fibers were seen to extend all the way to the poles in prometaphase. However, many of these kinetochore cables ended before reaching the pole. The microtubules at the poleward ends of thess cables appeared to splay out. The position of the splayed ends was very dynamic. The length of the kinetochore cable grew and shortened over time, but became progressively shorter as the cell approached anaphase. The position of the splayed end of the kinetochore cable appeared to be very close to the kinetochore at late metaphase and anaphase. Occasionally, the kinetochore cable could be seen extending several microns towards the pole from a kinetochore, but in general it was difficult to detect. The density of microtubules in the spindle regions between the chromosomes and the poles was too high to permit identification of individual microtubules with certainty. In the interzonal region between separating chromosomes in anaphase, the density of microtubules was low. As shown in Figure 4d, the lateral association of the clusters of one or a few microtubules into the larger bundles of microtubules, termed stem-bodies, was clearly visible. We interpret these observations in terms of weak, transient lateral associations between spindle microtubules that are in close proximity to each other, as shown schematically in Figure 5. The transient appearance of the rods could be due in part to single microtubules laterally moving into the optical section. More likely, it is produced by the transient lateral association of a few microtubules. The contrast of the rods was variable but only slightly different from the contrast of single microtubules viewed in solution in vitro. Their contrast was much less than the contrast of either the centrioles, kinetochore fibers, or stem-bodies. The brief duration and short extent of these transient bundles indicates that the energy of lateral association between microtubules is comparable to the thermal kinetic energy of the microtubules. FRAP studies indicate that the average life-time of a tubulin subunit within a nonkinetochore microtubule is on the order of sec in the newt spindle [Wadsworth and Salmon, 1986al. The life-time of the transient bundles we observed was much shorter than this duration, being from 1 to several sec. Thus it appears unlikely that the transient

11 Microtubule Dynamics in the Chromosomal Fiber 195 nature of the dynamic instability of microtubule polymerization regulates the duration of lateral association. Continuous but random transient lateral associations between microtubules throughout the spindle could produce the mechanical cohesiveness of the half-spindle [Inout, 1981al and could contribute to the forces that push the chromosome arms and other large cytoplasmic organelles out of the central spindle in mitotic animal cells [Rieder et al., Transient lateral interactions between microtubules may contribute significantly to the organization of microtubules within the chromosomal fiber, as initially proposed by Bajer and Mole-Bajer [1975]. First, consider situations where the density of nonkinetochore microtubules in the half-spindle is low. Rieder and Bajer [ have shown, using electron microscopy, that depolymerization of the bulk of the nonkinetochore microtubules within the newt spindle, either by cooling or heating, induces the formation of tight bundles of microtubules that extend all the way from the kinetochores to the poles. The dimensions of these bundles correspond to those of the cables we observed in untreated cells at prometaphase. At this stage, we found that the kinetochore cables extended more than 10 pm in length before the microtubules splayed apart in regions of the spindle near the pole. Perhaps, when the density of nonkinetochore microtubules is low, as it appears to be in early prometaphase near the spindle equator, the transient lateral associations between the kinetochore microtubules cooperatively pulls them into a tight bundle (Figure 5). In regions of the spindle near the pole, the density of nonkinetochore microtubules is higher. At this higher density of nonkinetochore microtubules, lateral associations between the kinetochore microtubules and the surrounding nonkinetochore microtubules may splay apart the kinetochore microtubules of the kinetochore cable. As cells approached metaphase, their spindles became progressively shorter. The density of microtubules increased substantially within the half-spindle regions between the chromosomes and the poles as judged by the contrast of the birefringence in the half-spindle (compare Fig. 4a with c), and the lengths of the kinetochore fibers were much shorter or undetectable in comparison with prometaphase. At metaphase, the density of nonkinetochore microtubules throughout the half-spindle may be sufficiently high to splay apart the kinetochore microtubules near the kinetochore. Under these conditions, the nonkinetochore microtubules would interdigitate with the kinetochore microtubules, so the chromosomal fiber would be a dynamic mixture of kinetochore and nonkinetochore microtubules (Fig. 5). Extensive intermingling of kinetochore and nonkinetochore microtubules has been demonstrated for chromosomal fibers in mitotic Hemanthus endosperm cells [Inoue et al., CONCLUSIONS Our observations indicate that the chromosomal fiber is a dynamic structure because microtubule assembly is transient, and lateral interactions between microtubules are also transient. At metaphase, the great majority of microtubules within the chromosomal fiber are undergoing dynamic instability assembly based on the tubulin analog and FRAP studies described above. The half-life of attachment of a microtubule to a kinetochore in metaphase may be less than 1 to 2 min. As one microtubule within a chromosomal fiber rapidly shortens and disappears, a new, elongating microtubule appears to be recruited into the chromosomal fiber by the transient lateral interactions between microtubules. The clustering of growing nonkinetochore microtubules into the chromosomal fiber surrounding the kinetochore microtubules could facilitate the attachment of a growing microtubule to an available attachment site on the kinetochore. Rapid reattachment would produce a constant steady-state number of kinetochore microtubules, even though the life-time of attachment of an individual microtubule at the kinetochore is very brief. ACKNOWLEDGMENTS We are indebted to D. Wise and C. Rieder for their initial observations on the behavior of room temperature. Ptk cells. We thank Brenda Bourns and Susan Whitfield at Chapel Hill and Linda Golden at the Marine Biological Laboratory for their photographic assistance. Thanks to Rich Walker for help with Fig. 1. We are also indebted to Bob Knudsen (Marine Biological Laboratories), Ted Inouk, Dyon Anniballi, and Dan Green (Universal Imaging Corp, Ithaca, NY), for their construction of the apparatus that provided the through-focus, thin-optical sections of our specimens. This study was supported in part by NIH GM to EDS and NSF DCB and NIH R-37-GM to S.I. REFERENCES Allen, R.D., Travis, J.L., Allen, N.S., and Yilmaz, H. (1981): Video-enhanced contrast polarization (AVEC-POL) microscopy: A new method applied to the detection of birefringence in the motile reticulopodial network of Allogromia laticollaris. Cell Motil. 1: Bajer, A., and Mole-Bajer, J. (1975): Lateral movements in the spindle and the mechanism of mitosis. In Inoue, S. and Stephens, R.E. (eds): Molecules and Cell Movement. New York: Raven Press, pp BOSS, J. (1954): Mitosis in cultures of newt tissues. I. A critical study of the methods and materials. Exp. Cell Res. 7: Cassimeris. L.U., Walker, R.A., Pryer. N.K., and Salmon, E.D. (1987): Dynamic instability of microtubules. Bioessays 7:

12 196 Cassimeris et al. Gorbsky, G.J., Sammak, P.J., and Borisy, G.G. (1987): Chromosomes move poleward in anaphase along stationary microtubules that coordinately disassemble from their kinetochore ends. J. Cell Biol. 104:9-18. Inoue, S. (1981a): Cell division and the mitotic spindle. J. Cell Biol. 9 I : 131 s- 147s. Inoue, S. (1981b): Video image processing greatly enhances contrast quality, and speed in polarization microscopy. J. Cell Biol Inoue, S. (1986): Video Microscopy. New York: Plenum Press. Inoue, S. (1988): Imaging of unresolved objects, superresolution, and precision of distance measurements with video microscopy. Methods Cell Biol. 30: In press. Inoue, S., Mole-Bajer, J. and Bajer, A.S. (1985): Three-dimensional distribution of microtubules in Hemanfhus endosperm cells. In DeBrabender, M., and DeMey, J. (eds.): Microtubules and Microtubule Inhibitors. Amsterdam: Elsevier, pp Leslie, R.J., Saxton, W.M., Mitchison, T.J., Neighbors, B., Salmon, E.D., and McIntosh, J.R. (1984): Assembly properties of fluorescein-labeled tubulin in vitro before and after fluorescence bleaching. J. Cell Biol. 99: McIntosh, J.R., and Vigers, G.P.A. (1987): Microtubule dynamics in the mitotic spindle. In G.W. Bailey (ed.): Proceedings of the 45th Annual Meeting of the Electron Microscopy Society of America. San Francisco: San Francisco Press, pp McIntosh, J.R., Cande, W.Z., and Synder, J.A. (1975): Structure and physiology of the mammalian mitotic spindle. In Inoue, S., and Stephens, R.E. (eds.): Molecules and Cell Movement. New York: Raven Press, pp McIntosh, J., Saxton, W.M., Stemple, D.L., Leslie, R.J., and Welsh, M.J. (1985): Dynamics of tubulin and calmodulin in the mammalian mitotic spindle. Ann. NY Acad. Sci. 466: Mitchison, T., and Kirschner, M. (1984a): Microtubule assembly nucleated by isolated centrosomes. Nature 312: Mitchison, T., and Kirschner, M. (1984b): Dynamic instability of microtubule growth. Nature 312: Mitchison, T., Evans, L., Schulze, E., and Kirschner, M. (1986): Sites of microtubule assembly and disassembly in the mitotic spindle. Cell Rieder, C.L. (1981a): Effect of hypothermia (20-25 C) on mitosis in PtKl cells. Cell Biol. Int. Rep Rieder, C.L. (1981b): The structure of the cold-stable kinetochore fiber in metaphase PtKl cells. Chromosoma 84: Rieder, C.L., and Bajer, A.S. (1977): Heat-induced reversible hexagonal packing of spindle microtubules. J. Cell Biol Rieder, C.L., Davison, E.A., Jensen, L.C.W., Cassimeris, L., and Salmon, E.D. (1986): Oscillatory movements of monooriented chromosomes and their position relative to the spindle pole result from the ejection properties of the aster and half-spindle. J. Cell Biol. 103: Salmon, E.D., Goode, D., Maugel, T.K., and Bonar, D.B. (1976): Pressure-induced depolymerization of spindle microtubules. Ill. Differential stability in HeLa cells. J. Cell Biol. 69: Salmon, E.D., and Wadsworth, P. (1986): Fluorescence studies of tubulin and microtubule dynamics in living cells. In Taylor, D.L., Waggoner, A.S., Murphy, R.F., Lanni, F., and Birge, R.R. (eds): Applications of Fluorescence in the Biomedical Sciences. New York: Alan R. Liss, pp Salmon, E.D., McKeel, M., and Hays, T. (1984a): Rapid rate of tubulin dissociation from microtubules in the mitotic spindle in vivo measured by blocking polymerization with colchicine. J. Cell Biol. 99: Salmon, E.D., Leslie, R.J., Saxton, W.M., Karow, M.L., and McIntosh, J.R. (l984b): Spindle microtubule dynamics in sea urchin embryos: Analysis using a fluorescein-labeled tubulin and measurements of fluorescence redistribution after laser photobleaching. J. Cell Biol. 99: Saxton, W.M., Stemple, D.L., Leslie, R.S., Salmon, E.D., Zavortink, M., and McIntosh, J.R. (1984): Tubulin dynamics in cultured mammalian cells. J. Cell Biol. 99: Schulze, E., and Kirschner, M. (1986): Microtubule dynamics in interphase cells. J. Cell Biol. 102: Soltys, B.J., and Borisy, G.G. (1985): Polymerization of tubulin in vivo: Direct evidence for assembly onto microtubule ends and from centrosomes. J. Cell Biol. 100: Wadsworth, P., and Salmon, E.D. (1985): Microtubule dynamics in mitotic spindles of living cells. Ann. NY Acad. Sci. 466: Wadsworth, P., and Salmon, E.D. (1986a): Analysis of the treadmilling model during metaphase of mitosis using fluorescence redistribution after photobleaching. J. Cell Biol. 102: Wadsworth, P.. and Salmon, E.D. (1986b): Preparation and characterization of fluorescent analogs of tubulin. Methods Enzymol Walker, R.A., Pryer, N.K., Cassimeris, L.U., Soboeiro, M. and Salmon, E.D. ( 1986): Axoneme-nucleated MAP-free microtubules exhibit Polarity-Dependent Dynamic Instability: A realtime observation. J. Cell Biol. 103:432a. Wise, D., Cassimeris, L.U., Rieder, C.L., Wadsworth, P., and Salmon, E.D. (1986): Incorporation of tubulin into kinetochore microtubules-relation to chromosome congression. J. Cell Biol. 103:412a.