|Home | About | Journals | Submit | Contact Us | Français|
Kinetochore capture and transport by spindle microtubules plays a crucial role in high-fidelity chromosome segregation, although its detailed mechanism has remained elusive. It has been difficult to observe individual kinetochore–microtubule interactions because multiple kinetochores are captured by microtubules during a short period within a small space. We have developed a method to visualize individual kinetochore–microtubule interactions in Saccharomyces cerevisiae, by isolating one of the kinetochores from others through regulation of the activity of a centromere. We detail this technique, which we call ‘centromere reactivation system’, for dissection of the process of kinetochore capture and transport on mitotic spindle. Kinetochores are initially captured by the side of microtubules extending from a spindle pole, and subsequently transported poleward along them, which is an evolutionarily conserved process from yeast to vertebrate cells. Our system, in combination with amenable yeast genetics, has proved useful to elucidate the molecular mechanisms of kinetochore–microtubule interactions. We discuss practical considerations for applying our system to live cell imaging using fluorescence microscopy.
Proper sister chromatid segregation to opposite poles of the cell during mitosis is crucial for the maintenance of genetic integrity in eukaryotic cells. For high-fidelity chromosome segregation, sister kinetochores must be properly captured by spindle microtubules , and then pulled towards opposite spindle poles after the loss of cohesion of sister chromatids at the metaphase–anaphase transition. In animal cells, kinetochores are captured by microtubules in prometaphase after nuclear envelope breakdown [2–4]. Because kinetochore capture and transport could be discerned only in a few cell types, mechanisms of these processes have remained elusive.
The budding yeast, Saccharomyces cerevisiae, is an excellent model organism to study the mechanism for chromosome segregation because of its amenable genetics. In addition, the development of systems to detect chromosome loci by marking them with binding sites for sequence-specific DNA binding proteins [5,6] expanded our knowledge of the mechanism of chromosome segregation by allowing direct visualization of chromosome motion in living cells. In these systems, lac or tet operators are integrated tandemly at chromosome sites of interest, which are bound by the Lac repressor (LacI) or tetracycline repressor (TetR) fused to fluorescent proteins, respectively [5,6]. When centromere movements were visualized, sister centromeres were found to split on the spindle when they bi-orient, while sister chromatids are still being held together by cohesion along their arms [7–10]. However, it is poorly understood how kinetochores interact with microtubules before they bi-orient on the spindle.
Budding yeast undergoes a closed mitosis; the nuclear envelope remains intact throughout the cell cycle. Microtubule organizing centers, called spindle pole bodies (SPBs), are embedded in the nuclear envelope [11,12], and kinetochores are tethered to SPBs by microtubules during most of the cell cycle, including G1 and M phases [12–15]. Nonetheless, several lines of data have suggested that centromeres are released from, and recaptured by microtubules during a brief period in S phase, due to kinetochore disassembly and reassembly upon centromere DNA replication [15,16]. However, in normal S phase, it is difficult to analyze individual kinetochore–microtubule interactions in detail, because all kinetochores interact with microtubules in the vicinity of a spindle pole where microtubules frequently overlap with each other.
To analyze the initial kinetochore–microtubule interaction in budding yeast, we have developed two assay systems, both of which are based on live cell microscopy imaging. The first assay involves delaying the assembly of a kinetochore on a particular centromere . Budding yeast kinetochores assemble on short (~130 bp) centromere sequences , and attach to a single microtubule [19,20]. It was reported that centromere function can be inactivated by activating transcription from an adjacently-inserted promoter, which inhibits kinetochore assembly [21,22]. We inserted the GAL1-10 promoter in the vicinity of CEN3 , to conditionally inactivate and activate the centromere by turning on and off transcription in the presence of galactose and glucose, respectively. This procedure prevents the centromere from associating with the mitotic spindle. While cells are arrested in metaphase, we reactivate the centromere, which leads to kinetochore assembly and interaction with microtubules extending from a spindle pole. In this system, which we have called ‘centromere reactivation system’, we found that kinetochores were captured by the lateral surface of a single microtubule extending from a spindle pole, and were subsequently transported poleward along the microtubule [17,23]. This process of initial kinetochore capture and subsequent transport along microtubule is evolutionarily conserved from yeast to vertebrate cells [2–4,17].
In the second assay, we observe kinetochore–microtubule interactions without artificial regulation of kinetochore assembly or without a cell cycle arrest . Upon centromere DNA replication, kinetochores are disassembled, which causes centromeres to detach from microtubules and move away from a spindle pole. Subsequently (1–2 min later) kinetochores are reassembled and captured by microtubules (kinetochore–microtubule interaction in normal S phase).
The first assay allows observation of the kinetochore–microtubule interactions with higher spatial resolution because the centromere moves away from a spindle pole and other centromeres for longer distances. On the other hand, the second assay enables analyses in a more physiologic condition. We can gain both advantages by comparing outcomes from both assays to analyze the mechanisms underlying the initial kinetochore–microtubule interaction.
In this article, we mainly focus on the centromere reactivation system as a method to dissect the mechanisms of kinetochore–microtubule interaction in budding yeast. We first introduce the technical details of the method and then show the applications of this system to elucidate various aspects of kinetochore–microtubule interactions. We also describe the method to observe kinetochore–microtubule interaction in normal S phase. We will not explain how to analyze chromosome dynamics after sister kinetochore bi-orientation is established or after cells enter anaphase, which has been described elsewhere [7–10]. Finally, we summarize the current view on kinetochore–microtubule interactions during the cell cycle in budding yeast.
We replaced CEN3 on chromosome III with CEN3 under control of the GAL1-10 promoter , to conditionally inactivate and activate the centromere by turning on and off transcription in the presence of galactose and glucose, respectively. For live cell imaging by fluorescence microscopy, we labeled the CEN3–adjacent sequence and α-tubulin (TUB1) with fluorescent proteins to visualize CEN3 and microtubules, respectively . We inactivated the CEN3 and simultaneously arrested cells in metaphase by depleting Cdc20, which is required for sister chromatid separation and for anaphase onset . In this situation, CEN3 localizes away from the spindle and is well separated from all other centromeres attached to the spindle. Then, we reactivated CEN3 by turning off the adjacent GAL1-10 promoter while cells were still in metaphase, and followed the behavior of CEN3.
The basic yeast strain we used for centromere reactivation system (T3531 ) has the following genotype with the W303 strain background: MATa, cdc20Pmet3-CDC20TRP1, cen3Pgal-CEN3-tetOsURA3, leu2TetR-GFPLEU2, trp1YFP-TUB1TRP1. Each construct in T3531 was obtained as follows.
First, Pgal-CEN3 , CYC1 transcription terminator (350 bp amplified by PCR) and 112 tandem copies of tetOs  (spanning 5.6 kb) were cloned into YIplac211 (National Centre for Biotechnology Information X75462) in the above order; second, the left and right CEN3-flanking regions (about 1 kb; not containing CEN3 itself) were amplified by PCR and cloned next to the GAL1-10 promoter in the above plasmid (at the opposite side from CEN3) with the opposite orientation (joining their 3′ ends together) (pT389); third, pT389 was cut between two para-CEN3 DNA fragments and used for transformation of yeast cells (Fig. 1).
For microscopy, cells are mounted either on microscope slides or on glass-bottom dishes depending on the purpose. Cells on microscope slides become somewhat flattened under the cover slip, enabling most of the cells to be kept within a single microscope field and on the same focal plane. On the other hand, cells sometimes change their location during imaging when the mounting medium dries up. In addition, this method may negatively affect cell viability, especially when images are collected over an extended-time period. Mounting cells on glass-bottom dishes has the opposite attributes; cells within a single microscope field are not usually on the same focal plane, but their viability is maintained for longer. When observing temperature-sensitive cells at their restrictive temperature, the objective lens and glass-bottom dishes need to be heated.
As the signal intensity from fluorescent proteins expressed in yeast is very weak mainly due to small number of the molecules in this small organism, a wide-field microscope which can collect more signal is advantageous over a confocal microscope. Three-dimensional live cell imaging followed by deconvolution makes the signal clearer and brighter.
Images are taken using an inverted microscope (Olympus IX-71 in Deltavision microscope (Applied Precision)) with a 100 × 1.4 numerical aperture optical lens, a cooled CCD camera (Photometrics CoolSNAP HQ) and softWoRx software (Applied Precision). In typical experiments, time-lapse images are collected every 10–15 s for 20–40 min with 5–9 (0.5–0.7 μm apart) z-sections at 23 °C (ambient temperature) or 35 °C (for temperature-sensitive cells). To distinguish GFP and YFP signals in time-lapse fluorescence microscopy, the JP3 filter set (Chroma) is used. In typical experiments, exposure time to excitation light is 0.1 s for both GFP/JP3 and YFP/JP3 channels with appropriate neutral density filters, but this depends on the signal intensity of target samples.
Acquired images are deconvoluted and projected with maximum intensity signals to two-dimensional images with the softWoRx software.
Using the centromere reactivation system, we found that the kinetochore is captured by the side of microtubules, often a single microtubule, and transported in a manner to that shown in animal cells . Cells with unattached CEN3s positioned at distance from spindles in elongated nuclei are suitable for analysis of kinetochore capture by long microtubules (Figs. 2 and 4A, 0 s). CEN3s are captured by microtubules within 20 min after addition of glucose in most of the cells. CEN3s are usually captured by the side of the microtubules (Fig. 4A, 40 s), and subsequently transported along the microtubules (Fig. 4A, 100–590 s). Shortly after CEN3s reach a spindle pole, the majority of CEN3-GFP signals split indicating that sister CEN3s bi-orient on the metaphase spindle (Fig. 4A, 620 s).
In the centromere reactivation system, we can divide the kinetochore capture process into several steps  (Fig. 2). First, microtubules extend from spindle poles. Next, kinetochores are captured by the side of a microtubule. Then, kinetochores are transported along the microtubule towards a spindle pole. Finally, after becoming bi-oriented the sister kinetochores separate, and are pulled towards opposite spindle poles. With our centromere reactivation system, we combined mutants that are known to cause chromosome missegregation. By observing these strains for defects in kinetochore capture, we could identify molecules involved in each step of the process, described in Ref. .
One advantage of our centromere reactivation system is that we can clearly see individual microtubules in elongated nuclei. Under normal circumstances microtubules appear only as part of a spindle bundle. This affords the opportunity to analyze the dynamic behaviors of nuclear microtubules, as has been done for cytoplasmic microtubules [27,28]. To follow microtubule dynamics in detail, we need to shorten the time interval for image capture in live cells (3–5 s). This can be achieved by reducing the number of z-sections (three sections per time point) and capturing both CEN3-GFP and YFP-tubulin signal simultaneously using the YFP channel of the JP4 filter set (Chroma). It is possible to distinguish nuclear microtubules from cytoplasmic microtubules by labeling the nuclear rim with Nic96 (a component of the nuclear pore complex) fused with YFP . Nuclear microtubules can also be identified by the presence of Dam1  or absence of Kip2  fused with three tandem copies of GFP along the microtubules (see below).
We could identify molecules involved in the dynamics of nuclear microtubules, and found that capture of kinetochores by microtubules facilitates microtubule rescues (conversion from shrinkage to growth) . We also found two mechanisms of microtubule-dependent poleward kinetochore transport . First, kinetochores slide along the microtubule lateral surface (lateral sliding). Second, kinetochores are tethered at the microtubule distal ends and pulled poleward as microtubules shrink (end-on pulling).
In normal conditions, yeast kinetochores cluster together on the spindle, and it is difficult to discriminate one kinetochore from the others. Although we can assume that certain molecules localize on kinetochores by observing their colocalization with the cluster of centromere signals, it is difficult to analyze localization of certain molecules on individual kinetochores. As the centromere reactivation system enables us to isolate one of the sister kinetochore pairs from others, we can reliably judge the localization of molecules on individual kinetochores. For this purpose we use the centromere reactivation system strains containing TetR-3CFP (Tet repressors fused with three tandem copies of cyan fluorescent protein (CFP))  and CFP-TUB1 instead of TetR-GFP and YFP-TUB1, respectively [17,23]. Molecules of interest are tagged with three tandem repeats of GFP at their genomic loci [32,33]. If the tagged molecules localize to kinetochores, GFP signal is seen on CEN3 after centromere reactivation in the YFP channel of the JP4 filter set, while CEN3 and tubulin are seen in the CFP/JP4 channel. As an example, the kinetochore localization of Stu2, a microtubule plus end-tracking protein (+TIPs) , is shown in Fig. 4B.
The centromere reactivation system is also useful to investigate the localization of molecules of interest on nuclear microtubules using strains as described above. By analyzing the dynamics of a molecule along nuclear microtubules during their polymerization and depolymerization, we can characterize the behavior of the molecules on nuclear microtubules [17,23]. In Fig. 4B, Stu2 is seen at the plus end of growing microtubule, consistent with its property as a +TIP member .
Recent indirect evidence has suggested that kinetochores might be transiently disassembled during S phase, causing centromeres to detach from microtubules [16,17,35]. Such centromere motion has been overlooked in the past, probably because it happens for such a short time period. By setting a very short time interval for image acquisition we have visualized this process .
To study centromere behavior with time-lapse microscopy, we marked CEN5 and CEN15 by the adjacent insertion of a tet or lac operator array [7,10], respectively. These arrays were bound by Tet repressors fused with three tandem copies of CFP (TetR-3CFP) , and by LacI with a single copy of GFP (GFP-LacI) ; thus CEN5 and CEN15 were visualized as small CFP and GFP dots, respectively. Microtubules were also visualized by the expression of α-tubulin (TUB1) fused with YFP (YFP-TUB1).
To follow the trajectory of CENs in detail, CFP/GFP and YFP images are collected every 7.5 s for 8 min with 5 (0.7 μm apart) z-sections (see Section 2.4.1); note that using the JP3 filter set, CFP/GFP and YFP are visualized separately, and the two CENs can be distinguished because CEN15-GFP shows higher intensity than CEN5-CFP. To avoid false judgements, detachment of GFP- and CFP-labeled CENs from YFP-labeled microtubules is scored only when CEN signals do not overlap with microtubule signals for two or more consecutive time points. Detachment is also scored when CEN-spindle pole distance is 700 nm or larger. Moreover, detachment is not scored if the CEN moved only along the z axis, as resolution along the z axis is not as good as on the x–y plane.
Both CEN5 and CEN15 stayed in the vicinity of a spindle pole (<0.5 μm from the center of the pole) during G1 phase (Fig. 4C, 0 s). However, just before bud emergence, which corresponds to early S phase, both CENs detached from microtubules and moved away from a spindle pole (Fig. 4C, 187.5 s shows CEN5 detachment); note that SPBs have not yet separated and cells have a single spindle pole in S phase . While CENs are detached (for 1–2 min), their distance from the spindle is on average 1.0 μm. Centromere detachment from microtubules was found to be dependent on its DNA replication, which results in disassembly of kinetochores. Kinetochores are reassembled soon afterward (Ref. , see Fig. 6), and recaptured by microtubules (Fig. 4C, 240 s shows CEN5 capture). Kinetochores are subsequently transported poleward by microtubules to the vicinity of a spindle pole, where they stay thereafter (Fig. 4C, 240–337.5 s show CEN5 transport). Kinetochores are transported in two ways, lateral sliding and end-on pulling, which are distinguished by observing localization of the Dam1 complex component Ask1 on CENs, in a similar manner as seen in the centromere reactivation system .
The current view on kinetochore–microtubule interactions in budding yeast during the cell cycle is as follows [24,37]:
In the open mitosis of metazoan cells, there is a large temporal gap (G2 phase) between DNA replication and the initial kinetochore–microtubule interaction, because spindle poles must wait for nuclear envelope breakdown in order to organize microtubules for kinetochore capture . By contrast, in the closed mitosis of budding yeast, spindle poles are connected to kinetochores via microtubules throughout most of the cell cycle [12–14], becoming detached only for a short period in S phase. Considering that centromeres are recaptured by microtubules already during S phase, we propose that in budding yeast, (1) there is no G2 phase (and no prophase) and (2) S phase and M phase (prometaphase) significantly overlap (Fig. 5).
In spite of such difference, the mechanisms underlying kinetochore–microtubule interactions are remarkably similar in early mitosis (prometaphase and metaphase) [37,41,42] between budding yeast and metazoan cells. In both, centromeres are initially captured by the lateral surface of microtubules and transported poleward along microtubules by minus-end directed motors; subsequently the Aurora B/Ipl1 kinase facilitates sister kinetochore bi-orientation, which is monitored by a conserved mechanism of the spindle checkpoint; both bi-orientation and spindle-checkpoint mechanisms respond to tension applied on kinetochores, for which the conserved cohesin complex is required.
Kinetochore capture and bi-orientation are fundamental cellular events during mitosis. We therefore expect that many aspects of the underlying mechanisms will be conserved from yeast to vertebrates, even if modifications were added during the evolution process. This is proving to be the case as is recently reported (reviewed in Ref. ). Recent advances in fluorescence microscopy have made yeast cells, which are much smaller than animal cells, available for detailed analysis of kinetochore–microtubule interactions. Further investigation in budding yeast is expected to uncover more regulatory mechanisms of kinetochore capture and transport, which are relevant to mitosis in all eukaryotic cells.
We thank L. Clayton and other members of TUT laboratory for discussions and reading the manuscript; C. Allan and S. Swift for technical help for microscopy/computing; R. Ciosk, F. Uhlmann, K. Nasmyth, K. Bloom, E. Schiebel and the Yeast Resource Centre for reagents. We learned the use of Concanavalin A for yeast-cell immobilization, at the web site of Y. Hiraoka lab. This work was supported by Cancer Research UK, the Wellcome Trust, Human Frontier Science Program, Lister Research Institute Prize and Association for International Cancer Research.