The behaviour of microtubules in chromosomal spindle fibres irradiated

singly or doubly with ultraviolet light

PAULA WILSON and ARTHUR FORER

Biology Department, York University, 4700 Keele Street, North York, Ontario, Canada M3J 1P3

Summary

Areas of reduced birefringence (ARBs) produced byultraviolet microbeam irradiation are areas of de-polymerized microtubules. ARBs probably movepoleward either by microtubule subunit addition atthe kinetochore and loss at the pole, or by micro-tubule subunit addition at one edge of the ARB andloss from the other edge. In this paper we have usedtwo approaches to try to distinguish between thesetwo models. First, we determined whether theedges of the ARB move at the same rate; if ARBmotion is due solely to addition at the kinetochoreand loss at the pole, with the ARB edges unable toexchange subunits, then the two edges of each ARBshould move at the same rate. On the other hand, ifthe exchange is at the ARB edges, then, from datafrom microtubules in vitro, the poleward edgeshould move much faster than the kinetochore-

ward edge. We found that the two edges of the ARBmove at the same rate about half the time, but halfthe time they do not. Second, we studied the behav-iour of two ARBs on a single fibre. If ARB motion isdue solely to subunit addition at the kinetochoreand loss at the pole, then the two ARBs must movepoleward together. We found that after two ARBsare formed on a single fibre the region between theARBs is unstable and rapidly depolymerizes. Theseresults do not fit either model and suggest thatinfluences of kinetochores and poles or other factorsneed to be considered that are not duplicated inexperiments on microtubules in vitro.

Key words: mitosis, microtubules, spindles, ultravioletmicrobeam irradiations, kinetochores.

Introduction

We have studied spindle fibre dynamics using an ultra-violet microbeam. While much is known about thedynamics of spindle microtubules (Salmon et al. 1984;Saxton et al. 1984; Mitchison et al. 1986; Geuens et al.1989), there still is debate about many aspects of howspindle microtubules polymerize prior to anaphase anddepolymerize during anaphase (e.g. see Czaban andForer, 1985a,b; Geuens et al. 1989; Wilson and Forer,1988).

We have used an ultraviolet (u.v.) microbeam toirradiate single chromosomal spindle fibres in livingcrane-fly spermatocytes. Forer (1965) originally reportedthat irradiations of fibres produce areas of reducedbirefringence (ARBs) that both in metaphase and ana-phase move poleward and disappear at the poles; thoughthe molecular changes responsible for forming ARBswere not known at the time, ARB motions were thoughtto be due to treadmilling of microtubule subunits fromkinetochores to poles (e.g. see Inou6 and Ritter, 1975;Margolis et al. 1978). Now that we know that ARBs areindeed regions in which microtubules are absent (Wilsonand Forer, 1988; Hughes et al. 1988) we can use the edgesof ARBs as markers of free ends of microtubules, and by

Journal of Cell Science 94, 625-634 (1989)Printed in Great Britain © The Company of Biologists Limited 1989

studying ARB motion get information about microtubuledynamics.

As discussed previously (Wilson and Forer, 1988),there are two possible models of ARB motion: polewardmovement of an ARB could be due to microtubulesubunits entering microtubules at kinetochores and leav-ing at poles, with the edges of the ARB being inactive, a'treadmilling' model; alternatively, poleward movementof an ARB could be due to microtubule subunits addingat the kinetochore-ward edge of the ARB and leavingfrom the poleward edge of the ARB, with the kinetochoreand pole ends of the fibre being inactive, an 'edge' model.We have tried to distinguish between the two models byfollowing the two edges of the moving ARB, and byforming two ARBs on a single chromosomal spindlefibre.

In considering the two edges of a single ARB, thetreadmilling model predicts that both edges would movepoleward at exactly the same rate. The edge model, onthe other hand, predicts (from data from microtubule-associated protein (MAP)-free microtubules in vitro[Walker et al. 1989; Hotani and Horio, 1988]) that thepoleward edges of the ARBs (the [+] ends of the severedmicrotubules) would move poleward (depolymerize)much faster than the kinetochore-ward edges ([—] ends)

625

polymerize. The two edges of an ARB were reported tomove poleward at the same rate (Forer, 1965), but,because the poleward side was often not measured or themeasurements comprised only four to five points ongraphs of distance versus time (see Figs in Forer, 1965,1966), the data from which this conclusion was drawnwere not sufficiently accurate to rule out alternatives.Thus, careful measurements of the speeds of the twoedges are warranted to test the two models.

In considering two ARBs on the same spindle fibre, thetreadmilling model predicts that the two ARBs wouldmove poleward at exactly the same rate (since allexchanges of subunits take place at pole and kineto-chore). The edge model, on the other hand, predicts(again from experiments on microtubules in vitro) thatthe two poleward edges of the ARBs (the [+] ends)would move poleward faster than the kinetochore-wardedges ([—] ends), and therefore that as the two ARBseach move poleward the unirradiated portion between thetwo ARBs should shorten from its kinetochore-wardedge.

As described below, the two edges of single ARBs donot always move at the same rate, thus ruling outtreadmilling as the sole cause of ARB motion. In ad-dition, the microtubules between the two ARBs on onefibre are unstable and rapidly depolymerize.

Materials and methods

Animal care and cell preparations

Crane flies {Nephmtoma suturalis) were reared as described byForer (1982). Living spermatocytes in meiosis were preparedfrom the testes of fourth instar larvae, as described previously(Wilson and Forer, 1988).

Microscopy and u.v. microbeaniCrane-fly spermatocytes were irradiated in metaphase or ana-phase of meiosis I. They were observed before, during and afterirradiation using polarization microscopy, and, occasionally, byclosing the condenser and increasing the compensator angle, aprocedure which yielded a phase-contrast-like image of the cell(we call this 'pseudo-phase'). The details of the irradiation andobservation system have been described elsewhere (Wilson andForer, 1987, 1988; Hughes et at. 1988). Irradiations andobservations of cells were carried out with a strain-free X32ultrafluar lens with a NA of 0.4 (Carl Zeiss). Irradiations wereusually with monochromatic light of wavelength 260 nm, be-cause this wavelength is most efficient at forming ARBs incrane-fly spermatocytes and least efficient at disrupting chromo-some motion (Sillers and Forer, 1983). The size and shape ofthe irradiated area varied, being either a circle with a diameterof approximately 3 j.im, a slit of width 1.8 fim or a slit of width1.3 jum. The behaviour of the ARB was independent of theparticular irradiating aperture used. The average dose incidenton the cell per irradiation was 0.5xl0~7J f<m~ (« = 1S).

Fluorescence microscope observations were made with aNikon inverted Diaphot-TMD microscope equipped for epi-fluorescence and a Nikon X100 Phase fluor-DL objective lens(NA, 1.30).

Plotting ARB motionWe measured the positions of ARB edges from video images

using an Oculus 200 Board (Coreco, Montreal) to digitize theimages, with software programmed in the laboratory. Measure-ments of single positions from light microscope images arelimited by the resolution limit of the light microscope and wereabout 0.8 ,um for our lens conditions. Given this limitation, ourtactic to get accurate estimates of velocities was to take manymeasurements of position at short time intervals; this shouldresult in a distribution curve which has a peak value at the realvalue of the position in question. By then analysing the positionversus time data statistically we should get accurate estimates ofwhether the two edges of the ARBs move at the same rate. Inthese experiments we generally image-averaged five frames(over 1 or 2 s) and used the averaged images to measurepositions. When necessary, to increase the clarity of the ARBedges (e.g. Fig. 1 E,F), we used the computer to enhance theimages by subtracting background intensities and by spreadingout the remaining intensities over the entire gray scale, and/orto enhance the edges. We analysed ARB positions every 5-10sof real time except when the cell position was being shifted orthe cell was observed using 'pseudo-phase'. The computerstored the user-specified pixel (X,Y) positions of poles, ARBedges, and kinetochores for each time point (i.e. each averagedimage) and then converted them into fim distances from which,using commercially available graphics and statistics softwarepackages, we produced graphs of distance versus time, and didstatistical analyses.

ImmunofluorescenceIrradiated cells were fixed in formaldehyde, extracted in coldacetone, and then processed for indirect immunofluorescencewith rabbit antibodies raised against sea-urchin tubulin, fol-lowed by fluorescein-conjugated anti-rabbit antibodies, as de-scribed by Wilson and Forer (1988). Although formaldehydefixation resulted in higher background staining than othermethods that were tried (e.g., cold methanol fixation or lysisprior to fixation), we used it because the general cell mor-phology was superior to that of cells fixed by other methods;this made finding a single cell from among a large population ofsimilar cells on a coverslip much easier. In addition, we felt thatthe preservation of more labile structures, such as depolymeriz-ing irradiated fibres, would be better than with other fixatives.

Recording imagesLiving cells were recorded using a Panasonic video camera andvideocassette recorder. Immunofluorescence images wererecorded using an RCA SIT camera and Samsung videocassetterecorder. Photographs were taken from the monitor with aNikon F5 35 mm camera using PanF film.

Results

Single irradiations

General observations. In crane-fly spermatocytes,ARBs on chromosomal spindle fibres always move pole-ward (Fig. 1), in general behaving as described for ARBscreated by heterochromatic u.v. light (Forer, 1965). Forexample, as an ARB approaches the pole it sometimesincreases in birefringence (e.g. Fig. 1), probably becausethe ARB is moving into an area where there are other,unirradiated microtubules.

Analysis of ARB motion. ARB motion was analysed in17 cells chosen for optical clarity and the absence ofinclusions in the path of the fibre which might haveobscured the ARB edges. In 14 cells single fibres were

626 P. Wilson and A. Forer

Fig. 1. Illustration of ARB motion. The ARB, formed on a chromosomal spindle fibre in a metaphase cell (using light ofwavelength 260nm), moved poleward and was lost at the pole. Times are in (minutesiseconds) with respect to the end ofirradiation. A. A 'pseudo-phase' image, before irradiation: —01: 16. Bar, lOjUm. B. Polarization image, before irradiation:-00 : 16. C-F. After irradiation. Small white bars mark the approximate positions of the ARB. C, 00:02; D, 02:21; E, 05:37;and F, 08:13. G,H. After the ARB reached the pole. G, 12:28. H. A 'pseudophase' image: 11:57.

irradiated and in three cells two fibres were irradiated(and plotted separately), making a total of 20 ARBs.

Each ARB edge seemed to move poleward at a constantvelocity, as determined by visual inspection of thedistance versus time graphs. Linear regression analysisconfirmed that the velocity was constant since the corre-lation coefficients were close to 1 (70% were 3=0.90 and90% were 3=0.85) and since in the graphs of X valuesversus residuals, the residuals were equally distributedabove and below zero (Figs 2 and 3).

We analysed some of the ARBs more than once, todetermine how reproducible our values were. We re-analysed three ARBs in two cells, for a total of threegraphs for each edge of the three ARBs. The graphsobtained from re-analysing the cells were similar to theoriginal graphs (Table 1), and there was no statisticaldifference in the slopes of the different plots of the sameARB edge, with two exceptions (see the velocities for the

poleward edges for cell 126 in Table 1). Even for theexceptions the differences are not large and the generaltrends remain the same. Most importantly, with regard tothe relationship between the two sides of any one ARB, inall examples the relationship between the velocities of thetwo edges of the ARB are the same - i.e., the polewardedge moved faster than the kinetochore-ward edge for allof the plots (P<0.02). Therefore, with confidence thatthe velocities determined from our plots were accurateand reproducible, we compared the motion of the twoedges of each ARB.

The velocities of the edges of the ARBs varied con-siderably, ranging from 0.35 to 2.07/immin"1 (Table 2).The variations in the velocities appear to be independentof stage (anaphase versus metaphase), wavelength, anddose of ultraviolet light, and we were unable to find apattern to explain them.

The kinetochore-ward and poleward edges of the same

u.v. irradiation of spindle microtubules 627

Time (min)

Fig. 2. Distance versus time graph, regression lines andresidual plots tor ARB motion in the cell in Fig. 1. A. Thepositions of the poleward edge of the ARB (A), thekinetochore-ward edge of the ARB (A), the autosomal pairassociated with the irradiated fibre (O,B) and the pole in theunirradiated half-spindle (V) are plotted with respect to thepole in the irradiated half-spindle. Time 0 is the time atwhich analysis was begun. B. The points and linearregression lines (calculated from least mean squares analysis)for the poleward edge (bottom line) and the kinetochore-wardedge (top line) of the ARB. For the line representing theARB edge closest to the kinetochore, the last five points (ingraph A) were omitted from this analysis because polewardmotion seemed to have stopped, though this could be anartifact caused by the ARB being very difficult to identify asit gets close to the pole, even after image enhancement. Thecorrelation coefficient for both lines is 0.98. C,D. Theresiduals for the kinetochore-ward (C) and poleward (D)edges of the ARB are plotted versus time: the points are closeto 0 and evenly distributed about 0.

10 12

*•'«**.

10 12

• * . . - % .

Time (min)

Fig. 3. Distance versus time graphs of an ARB in cell 126.A. The points and linear regression lines (calculated fromleast mean squares analysis) for the poleward edge (bottomline) and the kinetochore-ward edge (top line) of the ARB.B,C. The residuals for the kinetochore-ward (B) andpoleward (C) edges of the ARB are plotted versus time: thepoints are close to 0 and evenly distributed about 0.

ARB did not always move with the same velocities. Ineight cases the poleward edge moved faster than thekinetochore-ward edge, in 10 cases the two were equal,and in two cases the kinetochore-ward edge moved fasterthan the poleward edge, as determined by (-test compari-sons of the slopes of the regression lines, with P=0.02.(In the two cases where the kinetochore-ward edge grewfaster than the poleward edge, the differences in veloci-ties are not large and the poleward edges of the ARBswere very difficult to see, even with image enhancement,so we are not certain that they represent true differencesin velocities.) When the poleward edge moved faster thanthe kinetochore-ward edge, the ratio of velocities rangedfrom 1.7 to 3.0 (average of 2.2). Whether or not the twoedges of the ARB moved at different rates appeared to beindependent of stage and wavelength (Table 2).

In graphs of ARB motion for cells which enteredanaphase while the ARB was moving, or in which thechromosome stopped after irradiation and then started

628 P. Wilson and A. Forer

Table 1. Re-analysis of ARB motion in two cells. ThreeARBs in two cells were re-analysed to produce three

graphs for each ARB

Cell(ARB)

133

126(ARB1)

(ARB2)

Pole edgevelocity*

(/(inmin"1)

1.59"1.35°1.28°

.12°

.12"

.12"

.37"

.37"

.52"

Kinetochoreedge velocityf

(;(m min~ )

0.65"0.63°0.51°

0.62"0.77b

0.71°'b

0.540.72°0.78°

Not equalNot equalNot equal

Not equalNot equalNot equal

Not equalNot equalNot equal

* Velocity of the ARB edge closest to the pole (the slope of theregression line).

t Velocity of the ARB edge closest to the kinetochore.% Results of Student's /-test, testing the hypothesis that the pole-

edge (\'P) and the kinetochore-edge (VKT) velocities are equal;P=0.02.

"|b Within each group of three velocities, representing the resultsfrom plotting the same edge three different times, those valuesfollowed by the same superscript letter (a or b) are statistically notdifferent; P=0.02.

moving again, we could not detect (by eye) any change inARB velocity as the chromosomes began to move.

Double irradiationsTwo ARBs were obtained on single chromosomal fibresin 18 metaphase or anaphase cells, leaving an unirradiatedbirefringent portion of the fibre (the 'midsection') in

Table 2. Summary of results from single irradiations

Cell

93691

131134

373839

126

128

17272935

127129133

Stage

MetMetMetMetMet

Met/anaMet/anaMet/anaMet/ana

Met/ana

AnaAnaAnaAnaAnaAnaAna

Wave-length(nm)

280265260260260

265265265260

260

260260265265260260260

Kinetochore-edge velocity(((mmin"1)

0.530.690.560.370.350.550.560.810.980.620.541.940.671.000.561.580.671.040.570.65

Pole-edgevelocity

(/immin""1)

0.500.690.580.61*0.81*1.010.99*0.820.951.12*1.37*1.701.98*1.680.47*1.19*0.792.07*0.751.59*

Met, metaphase.Ana, anaphase.Met/ana, cell entered anaphase after irradiation but before the

ARB reached the pole.* Denotes pairs of velocities that were significantly different at

f=0.02.

Ill

Fig. 4. Illustration of the three 'classes' of microtubulescreated after a double irradiation, (i) Microtubules attachedat the kinetochore at their plus ends, with their minus endsfree, (ii) Microtubules attached at the pole at their minusends, with their plus ends free, (iii) Microtubules free at bothends. P, pole.

between. In 11 cells the first ARB was closer to thekinetochore and in seven cells the first ARB was closer tothe pole. The average time between the end of the firstirradiation and the beginning of the second was 21 s.

The two ARBs on a single chromosomal fibre createdthree different groups of sheared microtubules, rep-resented diagrammatically in Fig. 4: (i) those attached tothe kinetochore with newly exposed free minus ends, (ii)those attached to the pole with newly exposed free plusends, and (iii) those forming the midsection, in whichboth ends were free. The first two groups are comparableto the microtubules that remain when a single ARB isproduced on a fibre, but the microtubules in the midsec-tion are novel.

In the majority of cases (12 out of 18), the midsectionlost birefringence until it disappeared; it took approxi-mately one minute to disappear when the first ARBformed was that closer to the kinetochore, and approxi-mately 2 min when the first ARB formed was closer to thepole. A typical result is illustrated in Fig. 5. In the othercases (6 out of 12), the midsection birefringence de-creased, but the birefringent fibre attached to the kineto-chore grew fast enough to eventually grow into the fadingmidsection and past it, so that the ARB closer to thekinetochore appeared to 'fill in' while the poleward ARBmoved normally (e.g., Fig. 6). The distinction betweenthese two results was not always clear - intermediateswere sometimes seen. Thus the birefringent midsectioncould be lost, creating one large ARB which movedpoleward, or the ARB closest to the kinetochore couldrecover birefringence as the midsection faded, leaving thepoleward ARB to move poleward, or there was someintermediate between the two.

The rates of depolymerization of the midsection weredetermined by plotting the positions of the ARB edges(with respect to the pole) versus time. The velocities ofloss of birefringence from the ends of the midsections,estimated by drawing best-fit lines (as judged by eye)through the points representing the edges of the midsec-tions and then finding the slopes of those lines, were inthe same range as those found for the movement of singleARBs (Table 2), and had a similar degree of variation. Atypical distance versus time plot is shown in Fig. 7. Bothfrom the graphs and from observation of the videorecordings, it appears that the birefringence was some-

u.v. irradiation of spindle microtubules 629

Fig. 5. Double irradiation. A chromosomal spindle fibre in a metaphase cell was irradiated twice with 260 nm light to producetwo ARBs. The birefringent portion between the two ARBs depolymerized rapidly, creating one large ARB, which movedpoleward as the chromosomes moved poleward during anaphase. A. A pseudo-phase image, before irradiation. Time (inminutes:seconds) with respect to the end of the second irradiation: —01:41. Bar, 10/im. B-G, Polarization micrographs of thecell. Small bars are at the approximate positions of the ARB edges. B, -00:50; C, -00:04; D, 00: 16; E, 01:49; F, 14:30 andG, 26:19. H. A 'pseudophase' image: 34:57.

times lost from the kinetochore-ward edge towards thepole, sometimes from the poleward edge towards thekinetochore (as in Fig. 7), sometimes from both edges, orsometimes from the entire length of the fibre (Table 3).The particular mode of loss of birefringence was indepen-dent of the stage of division (metaphase or anaphase). Onthe other hand, the direction in which the midsectiondisappeared seems to be correlated with the order inwhich the ARBs were formed (i.e. near the kinetochorefirst or near the pole first; Table 3).

We confirmed by immunofluorescence with antibodiesspecific to tubulin that the loss of birefringence in themidsection was due to the depolymerization of micro-tubules. In cells in which the birefringent midsection haddisappeared prior to fixation there was no fluorescenceabove background in that area (Fig. 8). Thus the loss ofbirefringence in the midsection is apparently due to thedepolymerization of microtubules.

Unlike the edges of the midsection, the other edges ofthe ARBs - the edge closest to the kinetochore and theedge closest to the pole - behaved for the most part assingle ARBs: the kinetochore side grew and the poleward

side retreated at rates similar to those observed when oneARB is on a fibre. One important difference, however,was that the kinetochore side of the kinetochore-wardARB did not always move poleward immediately, butrather it was sometimes delayed for a short time (e.g., seeFig. 9). The kinetochore-ward edge of the ARB wasjudged as 'lagging' in recovery when the plot of distanceversus time did not change position for one minute ormore. Lags ranging from 1 to 5 min were seen in 10 cellsin which there were two ARBs per fibre (56 %) and weremore common in cells in which the kinetochore side wasirradiated first. When there was one ARB per fibre, onthe other hand, in only two cases (10%) were possiblelags observed, both of which were less than 2 min. Thus,the presence of two ARBs rather than one ARB on a fibresomehow alters the behaviour of the ends of the micro-tubules that remain attached to the kinetochore. It isrelevant to note that the behaviour of the midsectionseemed to be independent of whether or not there was alag in the regrowth of the kinetochore edge of the fibre.

Lastly, chromosome motion, though often delayed,was not permanently stopped, despite the loss of micro-

630 P. Wilson and A. Forer

Fig. 6. Double irradiation in which the ARB closest to the kinetochore 'filled in'. A-H. Polarization micrographs. A. Beforeirradiation. Bar, 10 Jim. Time (in minutes:seconds) with respect to the end of the second irradiation: —01: 19. B. After firstirradiation: -00 :43 . White bracket marks the approximate position of the ARB. C, 00:06; D, 00:55 and E 02: 14. The fibrebetween the kinetochore and second ARB is growing into the second ARB at the same time that birefringence increases in thearea, resulting in a decrease in the size and visibility of the ARB. The first (poleward) ARB remains clear, however. F, 02:42;and G, 10:25. The second ARB is no longer visible; the first (poleward) ARB is moving poleward. H, 26:25. The cell hasrecovered and anaphase has begun.

tubules in the two ARBs and in the midsection (e.g., seeFig. 5). Consequently the loss of most of the micro-tubules in a chromosomal fibre does not necessarilyprevent chromosome motion.

Discussion

Microtubide dynamicsOur results rule out the 'treadmilling' hypothesis as thesole cause of ARB motion, because the two ARB edges donot always move poleward at the same rate. They alsorule out the 'edge' hypothesis, or a combination of the twohypotheses, primarily because of the results of the doubleirradiations. That is, while the motion of the two edges ofsingle ARBs might be explained as treadmilling com-bined with variable depolymerization from the polewardedge of the ARB, or might be explained by variable ratesof exchange at the two edges of an ARB, neitherhypothesis, separately or combined, accounts for theinstability of the midsection microtubules after double

2 3 4Time (min)

Fig. 7. Distance versus time graph for ARB motion in thecell in Fig. 5. The pole of the unirradiated half-spindle (O)and the positions of the edges of the first ( •> • ) and second(A,A) ARBs were plotted with respect to the pole in theirradiated half-spindle. In this cell, the midsection (between• and A) disappeared from the poleward edge (A) towardthe kinetochore. The fibre then grew poleward. Time 0 iswhen the analysis was begun.

u.v. irradiation of spindle microtubules 631

Table 3. Summary of results of double irradiations

Cell Stage

Wave-length(nm)

FirstARB

Behaviourof

midsection*

Direction ofmidsection

loss

Time todisappear

(s)

636768

10111581

1171181211237172

120116598286

122

MetMetMetMetMetMetMetMetMetMetMet/anaMet/anaMet/anaMet/anaAnaAnaAnaAna

265265265260260260260260260260261261260260265260260260

KTKTKTKTKTPPPPP

KTKT

PKTKTK TKTP

Loss and fillLostLostLoss and fillLostLostFillLostLostLostLostLostLostLostLostLoss and fillFillLost

GeneralP^KTP-*KTGeneralpGeneral

KT-*PKT-+PKT-»Pi

GeneralKT-*PP^KTBoth

KT-*P

4§

123

135

14250S3

Met, metaphase.Ana, anaphase. Met/ana, cell entered anaphase after irradiation and before ARB reached the pole.KT, kinetochore.P, pole.* When two ARBs were formed on the same fibre, the midsection either disappeared (labelled 'lost') or persisted until met by the fibre

attached to the chromosome (labelled 'fill'), as described in the text, or some intermediate of the two phenomena was observed.

irradiations. Rather, the results after double irradiationsstrongly suggest that we must consider the influences ofkinetochore and pole. This is because microtubulesbetween the two ARBs on a single fibre act differentlyfrom those attached at the pole or at the kinetochore.When the microtubules at the kinetochore-ward edge ofan ARB are directly attached to a kinetochore, themicrotubules are stable and the edge moves poleward;when the microtubules at that same edge are not attachedto the kinetochore, on the other hand, but are rather inthe midsection between two ARBs, the microtubules are

not stable: they depolymerize rather than move pole-ward. A similar argument applies to the poleward ARBedge, and thus the double irradiation experiments impli-cate both kinetochores and poles as influencing thebehaviour of a distant end of a microtubule (at the edge ofthe ARB).

A second line of evidence perhaps implicating thekinetochore is that the kinetochore-ward edge of an ARBbehaves differently when the ARB is the only one on thefibre from the way it does when there is an additionalARB poleward from it. When there is one ARB on the

Fig. 8. Immunofluorescence image of a cell with two ARBs, fixed and stained with anti-tubulin antibodies after the birefringentportion between the ARBs disappeared. A. Polarization micrograph, before irradiation; bar, 10 jum. B. Polarization image justafter formation of 2 ARBs (white brackets). C. Polarization image 79s after final irradiation and prior to fixation. At this timethere is no birefringent midsection but only a single large ARB (marked by the white lines). D. Fluorescence image showingantitubulin antibody-staining of the same cell. There is no antibody staining where the midsection was, indicating that themicrotubules have depolymerized.

632 P. Wilson and A. Forer

10Time (min)

Fig. 9. Distance versus time graph for a double irradiation,illustrating a delay in the poleward motion of the ARB edgeclosest to the kinetochore. The positions of the pole in theunirradiated half-spindle ( • ) , the autosomal pair associatedwith the irradiated fibre ( • ,O) , the first ARB ( • , • ) and thesecond ARB ( V , A ) were plotted with respect to the pole ofthe irradiated half-spindle (x axis). The edge of the first ARBclosest to the kinetochore (D) began to move poleward, butstopped during or after the second irradiation and appearedto remain stationary for approximately 5 min, after which itmoved poleward.

fibre, the kinetochore-ward edge of the ARB alwaysmoves poleward immediately after being formed; whenthere is an additional ARB on the fibre, on the otherhand, the kinetochore-ward edge of the ARB nearer tothe kinetochore often 'lags' before it moves poleward.Since the conditions at the edge of the ARB should be thesame in both cases, these data suggest that there is somemanner of control by the kinetochore, either directlythrough control of addition of microtubule subunits atthe kinetochore or indirectly through control of themicroenvironment.

A third line of evidence implicating the kinetochoreand pole ends of the kinetochore microtubules is fromstaining these cells with antibodies against acetylatedtubulin: at least some of the kinetochore microtubules areacetylated, and the pattern of staining along the fibre(Wilson and Forer, 1989) suggests that the kinetochoreand pole ends of the kinetochore microtubules aredifferent from the remaining regions.

We conclude, then, that the results after doubleirradiations seem not to be able to be interpreted with ourpresent knowledge of microtubule dynamics based onmicrotubule polymerization in vitro. We suggest thatadditional influences, such as those of kinetochore andpole, need to be understood before we can understandmicrotubule dynamics in chromosomal spindle fibres invivo.

It is relevant to note that much of our argument isbased on ultrastructural and immunofluorescence evi-dence that the ARB is devoid of microtubules (Wilsonand Forer, 1988). While from those data we could notrigorously rule out the possibility that microtubule conti-nuity in the ARB was lost during fixation, the depolym-erization of the midsection (between the two ARBs onone spindle fibre) could result only if free microtubule

ends were generated at the edge of the ARB. Thus thesefindings from the present paper rule out the possibilitythat the absence of microtubules in the ARB is an artefactof fixation.

Microtubule stabilizatiotiSome have suggested that kinetochore microtubules arestable because they are capped at both the pole and thekinetochore ends (Kirschner and Mitchison, 1986;Mitchison et al. 1986; Cassimeris et al. 1987). Our datashow that this cannot be so, but rather that kinetochoremicrotubules must be stabilized along their lengths,otherwise they would depolymerize in a catastrophicmanner similar to that exhibited by MAP-free micro-tubules in vitro (e.g. see Hotani and Horio, 1988) and anARB would rapidly fill in. Further, since ARBs oftenmove as a unit, this shows that the kinetochore fibre isfairly stable and cohesive. This conclusion is differentfrom the view presented by Cassimeris et al. (1988),according to which microtubule-kinetochore attach-ments are transient and kinetochore microtubules quitelabile.

We gratefully acknowledge the excellent work of James Kelly,who wrote the software for the distance versus time analyses.We also thank Jim Hodges for his help with maintaining themicrobeam system, Mark DeBoer for help with analysis of thedata, Julia Swedak for comments on the manuscript and DrRobert Keates for helpful discussions and comments on themanuscript. This work was supported by grants from theNatural Sciences and Engineering Research Council of Canada.

References

CASSIMERIS, L., INOUE, S. AND SALMON, E. D. (1988). Microtubuledynamics in the chromosomal spindle fiber: analysis byfluorescence and high-resolution polarization microscopy. CellMotil. Cytoskel. 10, 185-196.

CASSIMERIS, L. U., WALKER, R. A., PRYER, N. K. AND SALMON, E.D. (1987). Dynamic instability of microtubules. BioEssavs 7,149-154.

CZABAN, B. B. AND FORER, A. (1985a). The kinetic polarities ofspindle microtubules in vivo, in crane-fly spermatocytes. I.Kinetochore microtubules that re-form after treatment withcolcemid.J. Cell Sci. 79, 1-37.

CZABAN, B. B. AND FORER, A. (19856). The kinetic polarities ofspindle microtubules in vivo, in crane-fly spermatocytes. II.Kinetochore microtubules in non-treated spindles. J. Cell Sci. 79,39-65.

FORER, A. (1965). Local reduction of spindle fiber birefringence inliving Nephrotoma suturalis (Loew) spermatocytes induced byultraviolet microbeam irradiation. J. Cell Biol. 25, (No. 1, Pt. 2)95-117.

FORER, A. (1966). Characterization of the mitotic traction system,and evidence that birefringent spindle fibers neither produce nortransmit force for chromosome movement. Chromosoma 19, 44-98.

FORER, A. (1982). Crane fly spermatocytes and spermatids: a systemfor studying cytoskeletal components. In Methods in Cell Biology(ed. L. Wilson), vol. 25, pp. 227-252. New York: Academic Press.

GEUENS, G., HILL, A. M., LEVILLIERS, N., ADOUTTE, A. AND

DEBRABANDER, M. (1989). Microtubule dynamics investigated bymicroinjection of Paramecium axonemal tubulin: lack of nucleationbut proximal assembly of microtubules at the kinetochore duringprometaphase.J. Cell Biol. 108, 939-953.

HOTANI, H. AND HORIO, T. (1988). Dynamics of microtubulesvisualized by darkfield microscopy: treadmilling and dynamicinstability. Cell Motil. Cytoskel. 10, 229-236.

u.v. irradiation of spindle microtubules 633

HUGHES, K. D., FORER, A., WILSON, P. J. AND LEGGIADRO, C.

(1988). Ultraviolet microbeam irradiation of microtubules in vitro:the action spectrum for local depolymerization of marginal bandmicrotubules in vitro matches that for reducing birefringence ofchromosomal spindle fibres in vivo. J. Cell Sci. 91, 469-478.

INOU£, S. AND RITTER, H. (1975). Dynamics of mitotic spindleorganization and function. In Molecules and Cell Movement (eds.S. Inou£ and R. E. Stephens), pp.3-30. New York: Raven Press.

KIRSCHNER, M. AND MITCHISON, T. (1986). Beyond self-assembly:from microtubules to morphogenesis. Cell 45, 329-342.

MARGOLIS, R. L., WILSON, L. AND KIEFER, B. 1. (1978). Mitotic

mechanism based on intrinsic microtubule behaviour. Nature,Lond. 272, 450-452.

MITCHISON, T., EVANS, L., SCHULZE, E. AND KIRSCHNER, M.

(1986). Sites of microtubule assembly and disassembly in themitotic spindle. Cell 45, 515-527.

SALMON, E. D., LESLIE, R. J., SAXTON, W. M., KAROW, M. L. AND

MCINTOSH, J. R. (1984). Spindle microtubule dynamics in seaurchin embryos: analysis using a fluorescein-labelled tubulin andmeasurements of fluorescence redistribution after laserphotobleaching. J. Cell Biol. 99, 2165-2174.

SAXTON, W. M., STEMPLE, D. L., LESLIE, R. J., SALMON, E. D.,

ZAVORTINK, M. AND MCINTOSH, J. R. (1984). Tubulin dynamicsin cultured mammalian cells. J Cell Biol. 99, 2175-2186.

SILLERS, P. J. AND FORER, A. (1983). Action spectrum for changes inspindle fibre birefringence after ultraviolet microbeam irradiationsof single chromosomal spindle fibres in crane-fly spermatocytes. J.Cell Sci. 62, 1-25.

WALKER, R. A., INOU£, S. AND SALMON, E. D. (1989). Asymmetric

behavior of severed microtubule ends after ultraviolet-microbeamirradiation of individual microtubules in vitm.J. Cell Biol. 108,931-937.

WILSON, P. AND FORER, A. (1987). Irradiations of rabbit myofibrilswith an ultraviolet microbeam. I. Effects of ultraviolet light on themyofibril components necessary for contraction. Biochem. Cell Biol.65, 363-375.

WILSON, P. AND FORER, A. (1988). Ultraviolet microbeam irradiationof chromosomal spindle fibres shears microtubules and permitsstudy of the new free ends in vivo.J. Cell Sci. 91, 455-468.

WILSON, P. J. AND FORER, A. (1989). Acetylated alpha-tubulin inspermatogenic cells of the crane fly Nephrotoma siituralis:kinetochore microtubules are selectively acetylated. Cell Motil.Cytoskel. (in press.)

(Received 17 May 1989 — Accepted in revised formI September 1989)

634 P. Wilson and A. Forer