Bim1p Localizes to the Spindle and the Tips of Cytoplasmic Microtubules
We examined the localization of Bim1p in living cells using a functional GFP-Bim1p fusion protein expressed from the native BIM1
promoter. Bim1p fluorescence was visible in nearly every cell at every stage of the cell cycle (Fig. ). In cells containing a short bipolar spindle, GFP-Bim1p fluoresced in a bright band along the length of the spindle (Fig. , A and H). Following anaphase, fluorescence was maintained along the length of the spindle but was brightest at the poles and in the region of microtubule overlap at the spindle midzone (Fig. , B, C, F, and G). In the cytoplasm, fluorescence usually (80% of cells) appeared as an intense spot or spots in the expected position of cytoplasmic microtubule tips, equivalent to the plus ends (28
) (arrows). This discontinuous localization contrasts the more uniform microtubule localization seen when the protein was overexpressed from the ACT1
), but the tip fluorescence is remarkably similar to the recently reported staining of human EB1 and RP1 (23
). The potential enrichment of Bim1p at microtubule ends has implications for models to explain the role of Bim1p in regulating microtubule dynamics in vivo.
Figure 1 Localization of GFP-Bim1p. Images of asynchronously grown bim1Δ cells expressing GFP-Bim1p from the BIM1 promoter are shown. These images are two-dimensional composites of Z-focal plane series 0.5 μm apart. Arrows indicate microtubule (more ...)
bim1Δ Cells Have Shorter Cytoplasmic Microtubules
Observation of the microtubule cytoskeleton in bim1Δ mutant cells by either indirect immunofluorescence or GFP tubulin fluorescence revealed a cell cycle–dependent change in the length of cytoplasmic microtubules (data not shown and Fig. A). During G1, the cytoplasmic microtubules in exponentially growing cells were shorter in bim1Δ than in BIM1 cells (mean length 1.05 μm vs. 1.59 μm, P < 0.0001). Synchronization in G1 with the mating pheromone α-factor accentuated this effect on microtubule length (mean length 1.52 μm in bim1Δ vs. 2.60 μm in BIM1, P < 0.0001). By contrast, during anaphase the mean length was slightly greater in bim1Δ compared to BIM1 cells (1.56 μm and 1.32 μm, P = 0.03). Cell size did not differ between the strains at any cell cycle stage (data not shown).
Figure 2 Cells lacking BIM1 have shorter cytoplasmic microtubules during G1. (A) Fluorescence images of GFP-Tub1p in BIM1 and bim1Δ cells. Bar, 1 μm. (B) Serial section electron micrographs through the spindle pole bodies of BIM1 and bim1Δ (more ...)
By fluorescence microscopy, microtubules can be reliably measured once their length reaches 0.5 μm. When we performed electron microscopy of bim1Δ mutant cells, we saw additional, even shorter microtubules in the 100–200-nm range that were not present in wild-type controls. Fig. B shows an example of the cytoplasmic outer plaque of the yeast spindle pole body with cytoplasmic microtubules emanating from it. The mean length of these short microtubules was 79 nm. We were also able to visualize the spindle pole bodies and the proximal (minus) microtubule ends by electron tomography, and these both appeared morphologically normal (data not shown). Based on these images, we hypothesized that Bim1p would affect microtubule assembly in a cell cycle–specific manner.
Cytoplasmic Microtubule Dynamics in Wild-Type Cells
Cytoplasmic microtubule dynamics were measured in living cells expressing GFP-Tub1p under the control of the TUB1 promoter. The level of expression of the GFP-Tub1p fusion protein was approximately one-fourth that of untagged α-tubulin (see Fig. , lanes 5 and 6). The GFP-Tub1p fusion protein produced uniform fluorescence along nuclear and cytoplasmic microtubules similar to that seen by indirect immunofluorescence in fixed wild-type cells, and the growth rates of strains containing GFP-Tub1p were indistinguishable from the corresponding nonfluorescent strains (data not shown). Initial observations of the microtubule movements in living cells revealed dynamic changes in the length of microtubules coupled with vigorous pivoting and swiveling of the microtubules about the spindle pole body. We used a measure of microtubule length determined from the two-dimensional projection of the microtubule in a composite of Z-focal plane sections and the path through serial Z sections at every time point (see Materials and Methods). This technique eliminated underestimates of rates based on the length of the two-dimensional projection alone and expanded the analysis to include microtubules that made excursions out of the focal plane.
Figure 5 Steady-state tubulin levels are unaffected by deletion of BIM1. BIM1 and bim1Δ cells were grown asynchronously (lanes 1, 2, 5, and 6) or arrested with α-factor (lanes 3 and 4) and Western blotting was performed on cell lysates. (Top) (more ...)
cells, cytoplasmic microtubules exhibited dynamic instability in vivo as has been described (5
), but the rates were considerably faster in our system than previously reported. There was greater microtubule dynamic instability during G1 than in preanaphase or anaphase, and this was due to increases in the shrinkage rate as well as both transition frequencies. These dynamic changes, discussed in more detail below, are illustrated in the time-lapse sequences of a BIM1
cell in the G1 phase of the cell cycle, shown in Fig. A, and in the life history tables in Fig. A, and summarized in Tables and .
Figure 3 Cytoplasmic microtubule dynamics in living cells expressing GFP-Tub1p. The sequences shown are two-dimensional projections of Z-focal plane series 0.3 μm apart; the interval between images is 8 s. (A) Unbudded (G1) BIM1 cell. The arrow points (more ...)
Figure 4 BIM1 deletion reduces cytoplasmic microtubule dynamics in G1. BIM1 and bim1Δ cells expressing GFP-Tub1p were imaged by time-lapse microscopy at 8-s intervals. Life history plots were constructed from measurements of three-dimensional microtubule (more ...)
Cytoplasmic Microtubule Dynamic Rates and Frequencies
Time Distribution Among Growing, Shrinking, and Pausing
In BIM1 cells, microtubules were most dynamic during the G1 phase of the cell cycle. The shrinkage rate was significantly faster than the growth rate during G1 (shrinkage 3.2 μm/min vs. growth 2.2 μm/min, P = 0.006, Table ). These rates were faster than the shrinkage and growth rates in anaphase, which were 2.2 μm/min and 1.6 μm/min, respectively (the difference in shrinkage rates between G1 and anaphase was statistically significant, P = 0.007). Rates were similar in preanaphase compared to anaphase (preanaphase shrinkage rate 2.7 μm/min and growth rate 1.7 μm/min). In addition to growth and shrinkage rates, the frequency of dynamic transitions was greater in G1 compared to mitosis. In G1, catastrophe and rescue frequencies were 0.008/s and 0.007/s, respectively, compared to the preanaphase catastrophe frequency of 0.004/s and rescue frequency of 0.002/s, and the anaphase catastrophe frequency of 0.007/s and rescue frequency of 0.004/s.
A comparison of time distribution between growing, shrinking, and pausing was also informative (Table ). During G1, when the shrinkage rate was faster, the amount of time spent shrinking was similar to the rest of the cell cycle, but the amount of time spent growing was greater. BIM1
cells spent 51% of their time growing during G1, vs. 38% of their time growing during anaphase. For comparison, we also used the dynamicity parameter (50
), which takes into account the total activity during the microtubule lifetime. Dynamicity was increased 1.6-fold during G1, 44 dimers/s vs. 27 dimers/s during mitosis. Thus, although microtubule length was similar in G1 and mitotic cells, microtubules in G1 cells were significantly more dynamic, with faster shrinkage rates, more frequent catastrophes and rescues, and less time spent pausing. The microtubule lengths predicted from multiplying the mean growth rate with the mean time spent growing correlated well with the lengths measured in static images of BIM1
cells (predicted length 1.7 μm vs. measured length 1.6 μm in G1; predicted length 1.6 μm vs. measured length 1.3 μm in mitosis).
bim1Δ Cells Have Abnormal Cytoplasmic Microtubule Dynamics during G1
One explanation for the reduction in microtubule length and number in the bim1Δ
mutant could be that Bim1p functions as a classical MAP and stabilizes microtubules (6
). However, as alluded to above, dynamic behavior cannot be predicted from the microtubule lengths in static images. In fact, the overall effect of BIM1
deletion was to make microtubules significantly less dynamic than in BIM1
cells, demonstrating that Bim1p promotes dynamic instability.
The most dramatic differences between microtubule dynamics in BIM1 and bim1Δ cells were observed during the G1 phase of the cell cycle. During G1, the shrinkage rate of microtubules in the bim1Δ mutant was 1.8 μm/min, compared to the rate of 3.2 μm/min in BIM1 cells (P = 0.002, Table ). By contrast with BIM1 cells, where shrinkage was significantly faster than growth, the growth rate was equivalent to the shrinkage rate in bim1Δ cells during G1. In preanaphase and anaphase, the growth and shrinkage rates showed a trend, not statistically significant, to slower rates in the bim1Δ mutant compared to BIM1 cells.
Transition frequencies were also lower in the bim1Δ mutant. During G1, the catastrophe frequency in bim1Δ cells was twofold less than in BIM1 (0.004/s, vs. 0.008/s) and the rescue frequency was threefold less than in BIM1 (0.002/s vs. 0.007/s). In preanaphase, the catastrophe and rescue frequencies were more similar between bim1Δ and BIM1 (catastrophe frequencies 0.002/s in bim1Δ vs. 0.004/s in BIM1; rescue frequencies 0.001/s in bim1Δ vs. 0.002/s in BIM1). During anaphase, transition frequencies rose again slightly (catastrophe frequencies 0.005/s for bim1Δ vs. 0.007/s for BIM1; rescue frequencies 0.002/s for bim1Δ vs. 0.004/s for BIM1). During G1, the dynamicity of microtubules in bim1Δ cells, 12 dimers/s, was 3.7-fold lower than the wild-type level of 44 dimers/s. Consequently, microtubules in bim1Δ cells showed greater dynamicity during mitosis than G1. Because our analysis was limited to microtubules in the bim1Δ mutant that were long enough to measure, it might not be representative of the population as a whole; however, based on the electron micrographs which showed very short microtubules in G1 bim1Δ cells, we believe the measured dynamicity is likely to be conservative, rather than to exaggerate the effect of BIM1 deletion. The decreases in dynamic instability are illustrated in Figs. and and summarized in Table .
Microtubules in bim1Δ cells exhibited a marked increase in the amount of time spent in the paused state during G1 (66% vs. 16% in BIM1 cells, Fig. B and Table ). This increase in pausing was entirely accounted for by a decrease in the time spent growing (8% of G1 spent growing by microtubules in bim1Δ cells vs. 51% in BIM1 cells). During preanaphase, microtubules in bim1Δ cells spent 22% of their time pausing and spent 32% of their time growing (compared to 51% of time spent growing during preanaphase in BIM1). Microtubules in bim1Δ cells spent 21% of their time shrinking in G1, compared to 47% in preanaphase. During anaphase, the time distribution in bim1Δ was nearly identical to BIM1 (27% spent pausing, 33% spent growing, and 28% spent shrinking). Thus, microtubules in bim1Δ cells spent less time growing and more time pausing during G1. The rate of shrinkage was statistically different from BIM1, and both catastrophe frequencies and the time distribution were markedly altered. As with BIM1 cells, the microtubule lengths predicted from the dynamics parameters correlated well with the measured lengths (predicted length 1.0 μm vs. measured length 1.1 μm in G1; predicted length 1.5 μm vs. measured length 1.6 μm in mitosis). Taken together, these data support the conclusion that, rather than acting as a microtubule stabilizing factor, Bim1p promotes microtubule dynamics.
To assess whether the Bim1p-specific effect on microtubule dynamics was related to changes in the concentration of α- and β-tubulin, Western blotting for tubulin was performed in cells growing asynchronously or arrested in G1. Fig. shows that α- and β-tubulin levels were equivalent in BIM1 and bim1Δ strains, both during growth and during α-factor–induced G1 arrest, as well as in the BIM1 and bim1Δ strains expressing GFP-Tub1p used for microtubule dynamics measurements.
Bim1p Regulates Nuclear Microtubule Function
The compactness of the yeast mitotic spindle prevented us from visualizing nuclear microtubule dynamics in the bim1Δ mutant. However, two lines of evidence suggest that Bim1p is important for nuclear microtubule function. First, cells with simultaneous deletion of BIM1 and the gene encoding the spindle assembly checkpoint protein Mad1p were inviable, suggesting that BIM1 deletion activates the spindle assembly checkpoint (data not shown). Second, electron micrographs of bim1Δ cells showed aberrant spindle structures not observed in wild-type controls. Fig. shows an example of a budded bim1Δ cell whose spindle pole bodies have duplicated but a bipolar spindle is not present. Instead, cytoplasmic microtubules intersect at right angles between the two poles. Such a structure could represent an intermediate in spindle formation or a collapsed bipolar spindle. This spindle morphology has not been seen in the BIM1 control or in other wild-type cells examined.
Figure 6 Aberrant spindle structures in the bim1Δ mutant. Two spindle pole bodies (SPB1 and SPB2) are seen in the mother cell, yet the nucleus (N) can be seen in the bud. Nuclear microtubule arrays (nMT) emanate from the SPBs but do not form a bipolar (more ...)
BIM1 Is Transcriptionally Regulated
We performed Northern and Western blotting to test whether BIM1
was cell cycle regulated. bim1Δ
cells expressing a functional GFP-Bim1p fusion protein from the BIM1
promoter were released from an α-factor arrest and cells harvested every 10 min through 2 cell cycles. GFP-BIM1
(and untagged BIM1
) mRNA was cell cycle regulated, peaking during G1-S and dropping during mitosis (Fig. A and data not shown), consistent with its observed effect on microtubule dynamics which predominates during G1. This cell cycle regulation of mRNA was similar to the fluctuations in BIM1
mRNA seen by whole genome microarray analysis (8
). Western blotting for the GFP epitope showed a significantly blunted fluctuation in the amount, and no change in the electrophoretic mobility, of Bim1p during the cell cycle (Fig. B). The same result was obtained using Bim1p fused to an HA epitope tag (data not shown). Thus, although BIM1
was transcriptionally regulated, under these conditions the levels of epitope-tagged Bim1p were less affected. The cell cycle effect of Bim1p on microtubule dynamics may therefore be regulated by posttranslational modifications, or by a functional interaction with another protein.
Figure 7 BIM1 is transcriptionally regulated. GFP-BIM1 cells were arrested in G1 and samples were taken at 10-min intervals after release. Cell cycle position was confirmed by cell morphology. (A) Northern blotting was performed using a probe to the BIM1 coding (more ...)
bim1Δ Cells Have a Nuclear Position Defect
The principal function of cytoplasmic microtubules in yeast is to position the nucleus properly during growth and mating (21
). To investigate the consequences of shorter, less dynamic cytoplasmic microtubules caused by BIM1
deletion, we examined nuclear positioning in bim1Δ
cells during vegetative growth. Nuclear position, consisting of both nuclear movement close to the bud neck and alignment of the spindle angle relative to the mother-bud axis, was measured in live cells containing the spindle pole body protein Nuf2p fused to GFP (24
). In the bim1Δ
mutant, the localization of Nuf2p-GFP at the poles was more diffuse than in BIM1
cells (data not shown), but the growth of the bim1Δ
strain containing Nuf2p-GFP was only minimally reduced (doubling time 2 h vs. 2.2 h at 24°C). As shown in Fig. , cells lacking BIM1
displayed a random orientation of the preanaphase spindle relative to the mother-bud axis (the mean spindle orientation angle was 43° in bim1Δ
cells vs. 32° in BIM1
cells) and an increased distance between the nucleus and the bud neck (the mean distance from the proximal spindle pole to the bud neck was 2.1 ± 1.2 μm in bim1Δ
cells vs. 1.0 ± 0.4 μm in BIM1
cells). This nuclear position defect in bim1Δ
cells was also observed by indirect antitubulin immunofluorescence (see Fig. C).
Figure 8 Nuclear position defect of bim1Δ cells. BIM1 and bim1Δ cells expressing the spindle pole marker Nuf2p-GFP were grown asynchronously, and preanaphase cells were photographed for spindle measurements. The spindle angle was calculated as (more ...)
Figure 10 Functional opposition between bim1Δ and kar3Δ. (A) bim1Δ suppresses the temperature-sensitive growth defect of kar3Δ cells. Serial fivefold dilutions of the indicated strains were grown at 24°C and 37°C (more ...)
While spindle position in the bim1Δ
mutant was abnormal, relatively few bim1Δ
cells went on to execute an abnormal anaphase and produce binucleate mother cells. By DAPI staining, 4% of budded binucleate BIM1
cells retained both nuclei in the mother cell, whereas in the bim1Δ
cells 5% of budded binucleate cells retained both nuclei in the mother (n
= 700, data not shown). Why are the consequences of abnormal spindle position so mild in the bim1Δ
background? In bim1Δ
cells with misoriented anaphase spindles, we observed a consistent pattern of spindle correction (n
= 5). Fig. shows a time-lapse series of one such cell, in which a cytoplasmic microtubule enters the bud neck, contacts the lateral cortex of the bud, and appears to pull the elongated spindle through the neck. During the initial part of this correction process, sliding appears to occur without depolymerization. Cytoplasmic dynein and Kip3p are candidate motors for producing this sliding force (5
), and, consistent with this idea, deletions of both of these proteins produce synthetic lethality with bim1Δ
(data not shown).
Figure 9 Correction of the spindle position defect in a bim1Δ cell. bim1Δ cells expressing GFP-Tub1p were photographed by time-lapse microscopy. The images are shown from a representative cell with a misoriented anaphase spindle that was rapidly (more ...)
Spindle Elongation Kinetics Are Mildly Reduced in bim1Δ
Once properly positioned, the yeast spindle elongates through the bud neck. Elongation of the spindle reflects concerted changes in kinetochore, pole to pole, and interdigitating microtubules that undergo simultaneous polymerization, depolymerization, and sliding within the nucleus, as well as pulling forces on the nucleus generated by cytoplasmic microtubules. It follows a biphasic pattern consisting of an initial rapid elongation phase followed by a period of slower continued elongation (24
). The initial phase of spindle elongation was mildly reduced in bim1Δ
cells, at 0.41 μm/min, compared to 0.59 μm/min in BIM1
= 0.07, n
= 7), while the slow phases were equivalent at 0.18 μm/min in bim1Δ
cells and 0.22 μm/min in BIM1
cells. This small decrease in the initial, rapid phase of spindle elongation in bim1Δ
cells suggests a subtle effect of Bim1p on some microtubule populations later in the cell cycle, and it could be due to an effect on either nuclear (pushing) or cytoplasmic (pulling) microtubules.
Bim1p and Kar3p Have Opposing Effects on Microtubules
The effect of Bim1p loss on G1 microtubule dynamics— slowing of the shrinkage rate, reduction in the frequencies of catastrophe and rescue transitions, and increase in the pause time—produced shorter microtubules. We hypothesized that BIM1 deletion might produce a net opposing effect to mutations that increase microtubule length, such as deletion of KAR3. Kar3p protein levels were unchanged by BIM1 deletion (data not shown). We investigated the synthetic phenotype of a bim1Δkar3Δ double mutant and found that deletion of BIM1 suppressed the temperature-sensitive growth defect of the kar3Δ mutant (Fig. A). This suppression correlated with an intermediate microtubule length during G1, the time when these mutations produce the most dramatic effects. Fig. B shows the microtubule morphology in representative fields of single and double mutant cells arrested in G1 with α-factor, by indirect antitubulin immunofluorescence. kar3Δ mutant cells expressing GFP-Tub1p grew poorly relative to kar3Δ cells, so this analysis was not performed in living cells.
As discussed above, bim1Δ
mutant cells are defective in positioning the mitotic spindle. Examination of spindle position in single and double mutant cells by indirect antitubulin immunofluorescence revealed correction of the bim1Δ
spindle position defect by simultaneous deletion of KAR3
(Fig. C). While the Bim1p and Kar3p mechanisms and sites of action (discussed below) may differ, these results demonstrate that the loss of opposing activities can produce normal appearing microtubule structures which correlate with suppression of defects in cell growth and spindle position. This cross-suppression between bim1Δ
did not extend to the microtubule-based process of karyogamy (data not shown). It was relatively specific to bim1Δ
, as other mutations which shortened microtubules, such as deletion of BIK1
), could not rescue KAR3
deletion, and other mutations which lengthened microtubules, such as deletion of DYN1
, could not suppress the bim1Δ
To better understand the mechanism of the bim1Δkar3Δ interaction, we measured microtubule dynamics in the bim1Δkar3Δ double mutant expressing GFP-Tub1p, as done above for the bim1Δ single mutant. Strikingly, microtubule dynamics in the bim1Δkar3Δ double mutant during G1 were like those in the bim1Δ single mutant. During G1, the microtubule shrinkage rate was 2.1 μm/min, the rescue and catastrophe frequencies were 0.004/s and 0.003/s, dynamicity was 17 dimers/s, and the percentage time pausing was 17% (Tables and ). During preanaphase and anaphase, the parameters were similar to those of the wild-type strain, also as observed with the bim1Δ single mutant. These measurements suggest that simultaneous deletion of KAR3 is able to functionally suppress the effects of bim1Δ on microtubule length and spindle orientation without suppressing the effects on microtubule dynamics directly.