Kinetochores can be thought of as having three major functions in chromosome segregation: (a) moving plateward at prometaphase; (b) participating in spindle checkpoint control; and (c) moving poleward at anaphase. Normally, kinetochores cooperate with opposed sister kinetochores (mitosis, meiosis II) or paired homologous kinetochores (meiosis I) to carry out these functions. Here we exploit three- and four-dimensional light microscopy and the maize meiotic mutant absence of first division 1 (afd1) to investigate the properties of single kinetochores. As an outcome of premature sister kinetochore separation in afd1 meiocytes, all of the chromosomes at meiosis II carry single kinetochores. Approximately 60% of the single kinetochore chromosomes align at the spindle equator during prometaphase/metaphase II, whereas acentric fragments, also generated by afd1, fail to align at the equator. Immunocytochemistry suggests that the plateward movement occurs in part because the single kinetochores separate into half kinetochore units. Single kinetochores stain positive for spindle checkpoint proteins during prometaphase, but lose their staining as tension is applied to the half kinetochores. At anaphase, ∼6% of the kinetochores develop stable interactions with microtubules (kinetochore fibers) from both spindle poles. Our data indicate that maize meiotic kinetochores are plastic, redundant structures that can carry out each of their major functions in duplicate.
The congression of chromosomes to the metaphase plate and subsequent poleward movement at anaphase are complex processes that occur with remarkable accuracy during cell division. An important organelle in chromosome movement is the kinetochore, a protein complex that associates with centromeric DNA (for reviews see Rieder and Salmon 1998; Maney et al. 1999). Through the interaction with spindle microtubules, kinetochore proteins have direct roles in propelling chromosomes toward the equatorial plane at prometaphase (e.g., Schaar et al. 1997; Wood et al. 1997), and subsequently away to opposite spindle poles at anaphase (Nicklas 1989). In addition, a handful of kinetochore proteins participate in the spindle checkpoint pathway, which ensures that chromosomes align correctly at the metaphase plate before anaphase begins (for reviews see Rudner and Murray 1996; Skibbens and Hieter 1998; Amon 1999). Even a single unaligned chromosome can activate the spindle checkpoint and prohibit anaphase onset (Li and Nicklas 1995; Rieder et al. 1995). It has been proposed that tension registered at the kinetochore is either directly or indirectly involved in the spindle checkpoint, at least in meiosis (Li and Nicklas 1995, Li and Nicklas 1997; Nicklas 1997; Yu et al. 1999).
Normally, chromosomes possess either two sister kinetochores (mitosis/meiosis II) or the paired kinetochores from homologous chromosomes (meiosis I). A widely held view is that kinetochore pairs are required to ensure that sister/homologous chromosomes segregate to opposite poles. The natural polarity of opposed kinetochores matches the bipolarity of the spindle, allowing the chromosomes to adopt a stable position at the spindle midzone (Rieder and Salmon 1994, Rieder and Salmon 1998; Nicklas 1997). However, the importance of paired kinetochores in chromosome congression was questioned by Khodjakov et al. 1997, who used laser ablation to experimentally remove a kinetochore from each of 50 mammalian mitotic chromosomes. The remaining single kinetochores were sufficient to generate the congression of 38% of the chromosomes analyzed. Electron microscopy of three cells revealed that the single kinetochores were distorted and attached to microtubules from both poles. These, and similar data involving detached kinetochore fragments (Zinkowski et al. 1991; Christy et al. 1995; Wise and Brinkley 1997), suggest that mitotic mammalian kinetochores are composed of subunits that can interact with microtubules independently (Khodjakov et al. 1997). In contrast, single kinetochore chromosomes failed to align at the spindle equator when the same technique was applied to African blood lily (Haemanthus) endosperm cells (Khodjakov et al. 1996), suggesting that single kinetochores and/or their interactions with the spindle differ among species or cell types.
Here, we extend the analysis of single kinetochores to maize meiotic cells. For a source of material, we exploit the phenotype of the maize meiotic mutant absence of first division 1 (afd1) (Golubovskaya and Mashnenkov 1975), which, as a result of premature sister kinetochore separation at meiosis I, produces cells at meiosis II that contain a complete set of single kinetochore chromosomes. By analyzing these single kinetochore chromosomes in detail, we demonstrate that they can align with ∼60% accuracy at metaphase II by interacting with kinetochore fibers from opposite spindle poles. During alignment, the single kinetochores appear to divide into halves that are capable of functioning independently. The connections established by half kinetochores are stable enough to dissociate/dephosphorylate two well-studied spindle checkpoint proteins. Finally, in anaphase, considerable poleward force was generated by the half kinetochores, stretching and nearly separating the kinetochores into two parts.
Materials and Methods
The original stocks carrying the recessive afd1 mutation were provided by Inna Golubovskaya (N.I. Vavilov Institute of Plant Industry Research, St. Petersburg, Russia). This strain was crossed once to the inbred line KYS, and a single resulting Afd1/afd1 plant was self-crossed to generate all of the material used here. Homozygous afd1/afd1 plants were identified cytologically in microsporocytes.
Meiocytes (Yu et al. 1999) or anthers (Yu et al. 1997) from both wild-type and mutant plants were fixed and processed as described previously (Yu et al. 1999). For the analysis of mitosis in afd1 plants, the tips of prop roots were excised, fixed, and sectioned on a cryostat (Yu et al. 1999). The maize centromere protein (CENPC) antibodies, maize MAD2 antibodies, 3F3/2 mAb (a gift from Gary Gorbsky, University of Virginia, Charlottesville; Gorbsky and Ricketts 1993), and mAb against α-tubulin (a gift from David Asai, Purdue University, West Lafayette, IN; Asai et al. 1982) were used as described previously (Yu et al. 1999). The CENPC and MAD2 antibodies were detected by rhodamine-conjugated goat anti–rabbit secondary antibodies, and the 3F3/2 and α-tubulin mAbs were detected by FITC-conjugated goat anti–mouse secondary antibodies (secondary antibodies were purchased from Jackson ImmunoResearch Laboratories). In double labeling studies, primary antibodies were incubated simultaneously. Chromosomal DNA was stained with diamino phenylindole (DAPI) at 0.1 μg/ml.
In Situ Hybridization
For in situ hybridization, a maize centromeric satellite tandem repeat called CentC (Ananiev et al. 1998) was PCR amplified from genomic DNA derived from the inbred line W23 (primers were 5′-GATTGGGCATGTTCGTTGTG and 5′-CACTACTTTAGGTCCAAAAC). Two clones of the ∼155-bp PCR product were sequenced to verify their identity as CentC. Gel-purified PCR products were labeled with fluorescently tagged dUTP and used as probes for in situ hybridization as described previously (Yu et al. 1997), except that the denaturing temperature was reduced to 90°C. In experiments where CENPC and CentC were both labeled, immunolocalization of CENPC was performed first, followed by in situ hybridization.
Microscopy and Data Analysis
Except where specifically noted in the text, all data were collected using a DeltaVision SA3.1 three-dimensional (3D) light microscope workstation as described previously (Yu et al. 1997). The data were processed by constrained iterative deconvolution. For the analysis of meiosis in living cells, meiocytes were cultured in a synthetic culture medium supplemented with the vital DNA stain Syto12 (Yu et al. 1997). Cells regularly survive in this medium for >6 h. Time lapse 3D (4D) data were collected at intervals from 1 to 30 min depending on the experiment.
To estimate the frequency of single kinetochore chromosome alignment in afd1 cells, we first determined that a rectangle with a width of 2 μm encompassed all the kinetochores in four wild-type metaphase II cells. Based on this estimate, a rectangle with a width of 2 μm was applied to the equator of the metaphase II spindles in six afd1 cells (see Fig. 5). The placement of the rectangle in afd1 cells was necessarily subjective, but in each case it was positioned roughly at the equator of the spindle and at right angles to the spindle axis. If a kinetochore was located within the rectangle, it was counted as aligned at the metaphase plate.
To evaluate the effect of tension on the dephosphorylation of the 3F3/2 antigen at the kinetochore, 3F3/2 staining was first normalized for kinetochore size by dividing it by the intensity of CENPC staining. This was done for all the kinetochores in two afd1 prometaphase II cells that did not overlap with another kinetochore or with the background 3F3/2 staining. A square composed of 10 × 10 pixels (pixel size, 0.1103 μm) was used to cover the kinetochore. The gray level intensity of the CENPC and 3F3/2 staining within the square was obtained from three contiguous sections (section thickness, 0.25 μm), averaged, and subtracted from the background intensity. Kinetochore edges were identified as the position half way from the tip to the base of a one-dimensional plot profile drawn over the kinetochore. The longest axis of the kinetochore was used as the length, except when it was spherical, and the diameter was used. To analyze the relationship between staining intensity and kinetochore length, we tested linear, log linear, and power models using maximum coeefficient of determination (R2) as our optimality criterion (using SAS statistical analysis software at the University of Georgia Research Computing Resource Facility).
Sister Kinetochores Separate Prematurely during Meiosis I in afd1 Meiocytes
We have recently identified and characterized a maize homologue of CENPC, a constitutive kinetochore protein (Dawe et al. 1999). Anti-CENPC antibodies effectively label each of the 20 chromosomes of a diploid maize cell at all stages of the cell cycle. As a first step in our study, we used affinity-purified anti-CENPC antibodies to confirm the phenotypic description of afd1 given by Golubovskaya and colleagues (Golubovskaya and Mashnenkov 1975; Golubovskaya et al. 1992). Fig. 1 illustrates a comparison of kinetochores from wild-type (left) and sibling afd1 (right) plants at various stages of meiosis I. A complete description of kinetochore morphology in wild-type cells can also be found in our previous report (Dawe et al. 1999). The earliest detectable prophase stage in afd1 plants is a diplotene-like stage, which in wild-type cells is typified by partially condensed and desynapsed chromosomes. All four (homologous and sister) kinetochores are usually associated at this stage in wild-type cells (Fig. 1 A, and data not shown), such that only 10 CENPC-positive spots are usually observed. Consistent with the assertion that minimal chromosome pairing occurs in afd1 meiocytes (Golubovskaya and Mashnenkov 1975; Golubovskaya 1989), 20 CENPC-positive spots were generally observed at the diplotene-like stage of mutant cells (Fig. 1 B, and data not shown).
After diplotene and prometaphase I, the sister kinetochores in wild-type cells stay conjoined but separate slightly, revealing a doublet structure (Dawe et al. 1999). Approximately ten doublet kinetochores are the norm for each half spindle at metaphase I (Fig. 1 C), anaphase I (Fig. 1 E), and telophase I (Fig. 1 G). However, in afd1 cells, ∼20 single kinetochores were observed in each half spindle at all 3 stages (Fig. 1D, Fig. F, and Fig. H; only the stereo pair in Fig. 1 D shows all the kinetochores). These data confirm the conclusion, made by Golubovskaya and Mashnenkov 1975, that sister kinetochores separate prematurely in afd1 plants to generate single kinetochore chromosomes before meiosis II. Our data also support the data of Chan and Cande 1998, who demonstrated that meiosis I spindle formation is essentially unaltered by the afd1 mutation (Fig. 1). Finally, we observed a low frequency of small chromosome fragments in afd1 plants that lacked visible CENPC staining (Fig. 1 F, discussed below).
To investigate the mitotic phenotype of the afd1 mutation, we extended our studies to somatic cells from afd1 plants. Data acquired from the cells in prop roots (aerial roots extending from the base of the stem) indicate that mitosis in mutant plants is essentially the same as was documented for normal maize mitosis (Yu et al. 1999). As shown in Fig. 2, the sister kinetochores can be distinguished from each other at the earliest stages of mitotic prophase (though they are often still connected; Fig. 2 A), and a complete separation of sister kinetochores occurs as early as prometaphase (not shown). Sister kinetochores then orient (Fig. 2 B) and segregate (Fig. 2 C) to opposite spindle poles. Acentric chromosome fragments were not observed in any of 16 anaphase/telophase cells from 2 mutant plants. Meiosis I and mitosis in the afd1 mutant can be distinguished from each other by several criteria (compare Fig. 1 and Fig. 2). The distinct differences in the timing of kinetochore separation, chromosome condensation patterns, and spindle morphology (mitotic spindles can be seen in Yu et al. 1999) suggest that the afd1 mutation does not substitute meiosis I with a mitotic division.
3D Analysis of Meiosis II in the afd1 Mutant
To determine whether the single kinetochore chromosomes generated by the afd1 mutation align at the metaphase II plate, we first employed 3D time lapse (4D) microscopy. As described previously (Yu et al. 1997), live meiocytes were extruded into a culture medium and stained with the vital DNA stain Syto12. A total of 32 cells from afd1 plants was observed undergoing meiosis II. For 12 of the cells, data collection began before metaphase II and included all or part of prometaphase. As shown in Fig. 3A–D, a clearly identifiable metaphase plate was formed. Once at the metaphase plate (Fig. 3 D), the single kinetochore chromosomes oscillated back and forth in a manner similar to wild-type cells (Fig. 3, A–D; Yu et al. 1997). The full prophase–metaphase II alignment process was observed in two cells, where prometaphase lasted ∼1 h longer (a total of ∼150 min) than expected for a wild-type meiocyte (∼90 min; Yu et al. 1997).
In an additional 6 wild-type cells and 20 cells from sibling afd1 plants, the earliest stages recorded were metaphase or early anaphase II. In each cell where the start of anaphase was documented, all of the chromosomes appeared to begin poleward movement together. The orderly chromosome segregation characteristic of a wild-type cell is shown in Fig. 4. In mutant cells, however, normal chromosome segregation was not observed. Instead, the chromosomes demonstrated erratic behavior typified by irregular rates of movement and frequent changes in direction. The rates of chromosome movement for individual chromosomes varied from <0.4 to 1.4 μm/min (compared with a consistent ∼0.78 μm/min in wild-type plants, Fig. 3; Yu et al. 1999). Anaphase II in mutant cells was typically three to four times longer than is characteristic for wild-type cells; the cell shown in Fig. 3 remained in anaphase for ∼100 min before the distinct nuclei structures characteristic of telophase were observed (not shown, but see Fig. 5 H).
The chromosome fragments generated during meiosis I were observed in living afd1 cells as small Syto12-stained structures. One such fragment was observed in the cell illustrated in Fig. 3 (arrows). The fragment moved slowly towards a spindle pole at 0.22 μm/min during prometaphase and remained suspended in the spindle throughout metaphase II. The same fragment moved rapidly poleward during mid-anaphase II at 0.71 μm/min (Fig. 3, E–H).
A Majority of Single Kinetochore Chromosomes, but Not Acentric Fragments, Aligns at the Spindle Midzone
By treating fixed cells with anti-CENPC antibodies, we were able to view the position and morphology of the single kinetochores during meiosis II. These data are shown in Fig. 5, with control meiocytes from wild-type plants (left) and meiocytes from sibling afd1 plants (right). The spindle in afd1 meiocytes was usually irregular in shape (Fig. 5 D), though a basic bipolar structure was always observed. As in living cells, a majority of the single kinetochores chromosomes appeared to align at the spindle equator in metaphase II (Fig. 5 D). We did not observe any examples of kinetochore-carrying chromosomes located at the spindle poles. Using the thickness of the metaphase II plate in wild-type cells as a standard (see Materials and Methods), we estimated from a sample of six afd1 cells that 60 ± 16% of the single kinetochore chromosomes congressed to the spindle equator (Fig. 5 D; note rectangles in C and D). Those chromosomes that aligned at the plate frequently took on a stretched appearance (Fig. 5 D, arrow and inset), whereas those that failed to align at the plate usually appeared spherical (Fig. 5 D, arrowhead).
Among 72 prometaphase–metaphase II afd1 cells that were analyzed in detail (from 6 plants), 32 possessed at least 1 acentric chromosome fragment, i.e., a small DAPI-stained body that lacked detectable CENPC staining. In contrast, a survey of 922 wild-type cells at the same stages revealed no visible fragments (these data were obtained by standard 2D microscopy from 3 wild-type siblings of mutant plants). The localization of acentric fragments in the spindle can be used to assess the direction of the forces prevailing on chromosome arms. Among the acentric fragments scored in afd1 plants, 4 were located in the vicinity of the spindle midzone, whereas a majority of 28 (88%) were located in a polar region. Acentric fragments that were intermingled among the chromosomes at the midzone would have been difficult to detect, so it is possible that we have overestimated the proportion of fragments that migrated to a pole. At a minimum, however, the data show that while no kinetochore carrying chromosomes was found near the spindle poles, a substantial number of acentric fragments was. The simplest interpretation is that kinetochores, not motile forces associated with chromosome arms, are responsible for the alignment of single kinetochore chromosome at the spindle midzone.
Single Kinetochore Chromosome Alignment Occurs by Tension-sensitive Interactions with the Spindle
In prior studies, we described the localization of two spindle checkpoint proteins on maize mitotic and meiotic kinetochores (Yu et al. 1999). MAD2 is a widely conserved checkpoint protein that binds specifically to unaligned kinetochores (for review see Amon 1999). The 3F3/2 antigen, also a presumed checkpoint protein, is a phosphoepitope that is sensitive to tension applied at the kinetochore (Nicklas et al. 1995; Nicklas 1997). Antibodies to both proteins recognize an outer domain of the kinetochore in wild-type maize meiocytes (Yu et al. 1999). Similarly, single kinetochore chromosomes at prometaphase II stain brightly with both the 3F3/2 antibody and the MAD2 antibody. These data are shown in Fig. 6. The single kinetochores progressively lost 3F3/2 and MAD2 staining as the cells proceeded through prometaphase (Fig. 6A, Fig. C, and Fig. D) and by anaphase II the staining was no longer detectable at kinetochores (Fig. 6 B). The 3F3/2 antibody also stains nonkinetochore sites (arrows in Fig. 6A,Fig. 2 and B2; Yu et al. 1999), which can be distinguished from kinetochore staining by double labeling with either the CENPC or MAD2 antibodies.
We previously demonstrated in wild-type cells that MAD2 and 3F3/2 show nearly identical staining patterns both spatially and temporally (Yu et al. 1999). To test whether MAD2 and 3F3/2 are also colocalized on single kinetochores, we analyzed 20 double-labeled cells ranging from early to late prometaphase II. Each of the cells was fixed at a stage when the checkpoint proteins were detectable on some but not all of the kinetochores present (the average number of MAD2-stained kinetochores was 12.4). We found that 98.8% of the kinetochores that were MAD2-positive were also labeled with the 3F3/2 antibody (247/250 single kinetochores). These data lend strong support to the conclusion that 3F3/2 dephosphorylation and MAD2 dissociation are coincident (Yu et al. 1999), and indicate that 3F3/2 staining at kinetochores is a reliable marker for the presence of MAD2.
Double labeling for the 3F3/2 antigen and CENPC (Fig. 6 A) and for the 3F3/2 antigen and MAD2 (Fig. 6C and Fig. D) indicated that the 3F3/2 antigen lies outside the CENPC domain but in the same domain as MAD2. Since nearly identical results were obtained in wild-type cells (Yu et al. 1999), our analysis suggests that the premature disjunction caused by afd1 does not disrupt the basic composition and organization of the kinetochores. Interestingly, a bipolar staining pattern was clearly observed on several single kinetochores, with the 3F3/2 staining occupying opposite ends of stretched kinetochores (Fig. 6 A3). These data indicate that single kinetochores have the capacity to divide into half kinetochore units, with each end of the elongated kinetochore interacting independently with a pole.
The results of our previous work suggest that the loss of MAD2 and 3F3/2 staining during meiosis is correlated with the level of tension applied to the kinetochores (Yu et al. 1999). Because chromatin is elastic (Waters et al. 1996), we were able to use the distance between paired homologous or sister kinetochores to estimate the tension between the kinetochores and their associated kinetochore fibers (Yu et al. 1999). We used a similar assay on the single kinetochore chromosomes in afd1 meiocytes. At mid-prometaphase II, the single kinetochores can be observed in a variety of states of alignment. Some show no evidence of a bipolar interaction with the spindle, others are stretched, indicating a bipolar interaction with the spindle, and others lie in between these two extremes. We chose two such mid-prometaphase II cells, and for each scorable kinetochore determined the 3F3/2 staining intensity (using CENPC staining to normalize for kinetochore volume) and the diameter/length of the single kinetochore. As shown in Fig. 7, there was a strong negative correlation between the two. These data indicate that the loss of 3F3/2 staining on stretched kinetochores is a result of tension applied to the single kinetochore, not an inherent limitation on the duration of the spindle checkpoint (discussed below).
Poleward Movement at Anaphase Causes Stretching of Single Kinetochores
Anaphase II in afd1 meiocytes can be identified both by the absence of staining for checkpoint proteins (Fig. 6 B) and by the degree of kinetochore stretching that occurs during this stage. While kinetochore stretching during prometaphase and metaphase II was mild (Fig. 5 D and Fig. 6A, Fig. C, and Fig. D), in anaphase II it was frequently extreme (Fig. 5 F and Fig. 6 B). An average of 24% of the single kinetochores at anaphase II were stretched into cylindrical shapes (n = 347 kinetochores in 18 cells), and in several cases the long axes exceeded 5 times the diameter of a normal anaphase II kinetochore (Fig. 5 F).
In wild-type cells, the kinetochores appear to interact with microtubules in a tangential way early in prometaphase and then display distinct end-on interactions with kinetochores in late prometaphase through anaphase (data not shown, and Fig. 1C and Fig. E). To analyze the interaction of single kinetochores with microtubules, we analyzed a set of 153 chromosomes in 8 anaphase II cells. Of 36 stretched single kinetochores, 27 showed primarily tangential interactions with microtubules, while 9 others showed the distinct end-on interactions characteristic of wild-type cells. As shown in Fig. 8B–E, when end-on interactions were apparent, there were two kinetochore fibers from opposite poles interacting with the ends of the stretched kinetochore. It is likely that the microtubules that appeared to interact tangentially with stretched kinetochores (Fig. 8 A) also extended away from the kinetochores in both directions, but other possibilities cannot be excluded.
The observation that a single kinetochore can be stretched between two spindle poles suggests that the force may occasionally divide the kinetochore into two parts. That this is indeed the case was demonstrated by a combination of CENPC immunolocalization and in situ hybridization using an ∼155-bp centromeric DNA repeat called CentC (Ananiev et al. 1998). At prophase II, CentC colocalized well with the CENPC-stained kinetochores (Fig. 9 A), although CentC staining varied considerably from chromosome to chromosome (as documented previously; Ananiev et al. 1998). However, on several anaphase chromosomes (eight chromosomes in five anaphase II cells), the kinetochores had clearly separated into two units that were joined by a thin thread of kinetochore material (Fig. 9 B). In each case, the two kinetochore units were attached by centromeric DNA, ruling out the possibility that these were rare examples of normal (i.e., two-chromatid) chromosomes that may have segregated correctly in meiosis I (in fact, we have no evidence that such chromosomes are ever present at meiosis II in the afd1 mutant). We made an effort to quantify the frequency with which centromere/kinetochores were actually broken during anaphase II, i.e., misdivided (Darlington 1937), by counting the number of CENPC spots in 10 telophase II cells from afd1 plants. Misdivision would be expected to increase the number of kinetochores to a value significantly greater than the expected number of 20. The average kinetochore number in the 10 afd1 cells was 19.78 ± 1.20, which was not significantly different from the kinetochore number in 8 wild-type telophase II cells (19.80 ± 0.84).
Here we demonstrate that meiotic kinetochores, which move to one spindle pole with high fidelity under normal circumstances, will regularly interact with two poles if they are denied a sister kinetochore. The evidence suggests that single kinetochore alignment involves the same basic processes employed during normal chromosome alignment, and that in anaphase, a significant fraction of the single kinetochores divides into half kinetochore units that interact independently with the spindle. In discussing our results, we first evaluate the afd1 phenotype in relation to other published data, and follow with our interpretations of how the single kinetochore chromosomes align, interact with spindle checkpoint proteins, and finally segregate in anaphase.
afd1 Causes a Defect in Sister Chromatid Cohesion at Meiosis I
Sister kinetochores are normally conjoined in meiosis I and then disjoin to move to opposite spindle poles in meiosis II. However, as described by Golubovskaya and colleagues, these events are significantly altered by the afd1 mutation (Golubovskaya and Mashnenkov 1975; Golubovskaya and Khristolyubova 1985; Golubovskaya 1989; Golubovskaya et al. 1992). Plants homozygous for afd1 appear to skip the early prophase stages of meiosis I and then to prematurely segregate the sister kinetochores to opposite poles. The single kinetochore chromosomes released during meiosis I are then carried through to meiosis II, where they form a haphazard metaphase plate and then segregate randomly at anaphase II (Golubovskaya and Mashnenkov 1975; Golubovskaya et al. 1992; Chan and Cande 1998). Using antibodies to the recently identified maize CENPC protein (Dawe et al. 1999), we were able to confirm this phenotype and demonstrate that the equational segregation of sister kinetochores at meiosis I is close to 100% (Fig. 1).
To explain the afd1 phenotype, Golubovskaya and colleagues proposed that the mutation causes a substitution of meiosis I with mitosis-like division, i.e., that meiosis I is absent in afd1 plants (Golubovskaya and Mashnenkov 1975; Golubovskaya and Khristolyubova 1985; Golubovskaya 1989). However, our immunocytochemical analysis suggests that the phenotype is more complex than this. Although some prophase I stages cannot be identified (roughly leptotene to pachytene), sister kinetochore separation in afd1 meiocytes starts after nuclear envelope breakdown (Fig. 1A and Fig. B) as is characteristic of normal meiosis I cells (Dawe et al. 1999). This contrasts with mitosis (Fig. 2 A) or normal meiosis II (Dawe et al. 1999), where kinetochore separation is readily apparent at prophase. Further, the chromatin condensation patterns and spindle morphology of the first division (Fig. 1B and Fig. D) more closely resemble meiosis I in wild-type cells (Fig. 1A and Fig. C) than mitosis (Fig. 2; Yu et al. 1999). We also show here that a low level of chromosome fragmentation occurs during meiosis I in afd1 plants and that many of these fragments lack kinetochores (Fig. 1 H).
The available data suggest that Afd1 encodes a cohesin such as budding yeast rec8p, a meiosis-specific rad21-like protein (Klein et al. 1999). Mutations in REC8 cause premature separation of sister chromatids and inhibit reciprocal recombination. The chromosome fragmentation caused by afd1 may be evidence of aborted recombination events, since meiotic recombination involves the formation of double-strand breaks (Roeder 1997; Zickler and Kleckner 1999). REC8/RAD21-like genes are widely conserved in eukaryotes (Watanabe and Nurse 1999), and mutants of a meiosis-specific RAD21-like gene called SYN1/DIF1 have recently been described in Arabidopsis (Bai et al. 1999; Bhatt et al. 1999). Consistent with a homology between the two genes, there are clear similarities between the afd1 phenotype and the Arabidopsis syn1/dif1 phenotype (Bai et al. 1999; Bhatt et al. 1999).
Recent data from both Saccharomyces cerevisiae and Schizosaccharomyces pombe suggest that cohesin is not required to maintain the integrity of the mitotic kinetochore (Tomoyuki et al. 1999; Takahashi et al. 2000). These data are consistent with our own data, showing that at least three kinetochore proteins (CENPC, MAD2, and the 3F3/2 antigen) show the stage-specific (Fig. 6) and subkinetochore localization (Fig. 6A,Fig. 3 and C3) typical of wild-type kinetochores (Yu et al. 1999). Although we cannot rule out the possibility that afd1 has subtle effects on the morphology or makeup of the kinetochore, the available data suggest that the sister kinetochores are separated but otherwise undisturbed as they enter meiosis II.
Single Meiotic Kinetochores Regularly Align at the Metaphase Plate
Using both living and fixed specimens, we demonstrate that ∼60% of single kinetochore chromosomes align at the metaphase II plate (Fig. 3 and Fig. 5), whereas ∼88% of acentric fragments move in the opposite direction towards a pole (Fig. 3). These data support the observations of Khodjakov et al. 1997, who, with data obtained from a different class of cell division (mitosis) and a distantly related species (rat kangaroo), showed that single kinetochores are capable of autoaligning at metaphase. Our data are also in agreement with earlier data from the same group showing that acentric fragments in Haemanthus mitotic cells move poleward during chromosome alignment (Khodjakov et al. 1996).
There are also notable differences between our data and those published previously. Perhaps the most significant is the demonstration by Khodjakov et al. 1996 that Haemanthus single kinetochore chromosomes fail to align at the spindle midzone. An explanation for this may lie in the fact that meiotic kinetochores display morphological plasticity in the course of their normal function, first associating closely with a sister kinetochore in meiosis I and then subsequently changing their behavior to dissociate from the sister in meiosis II. The inherent capacity for remodeling may make the meiotic kinetochore especially susceptible to aberrant alignment during meiosis II. Another notable difference is that acentric chromosome fragments move plateward during prometaphase in mammalian mitotic cells (Rieder and Salmon 1994), not poleward as in Haemanthus mitosis or maize meiosis. This difference is probably related to the function of centrosomes, which organize the spindle poles in animal mitotic cells, but which are absent in all higher plant cells and the meiotic cells of some animals (Smirnova and Bajer 1992; Rieder et al. 1993). The active polymerization of microtubules outward from centrosomes is thought to be part of the force that drives chromosome fragments plateward (Rieder and Salmon 1994). In cells that lack centrosomes, the dominant force affecting the movement of acentric fragments may be the poleward flux of tubulin monomers within the spindle (Sawin and Mitchison 1994).
Single Kinetochores Interact Normally with Spindle Checkpoint Proteins and Demonstrate Anaphase Motility
In normal cells, sister/homologous kinetochores orient towards opposite spindle poles and generate tension with attached kinetochore fibers (Nicklas 1997). The kinetochores move farther away from each other as increasing tension is applied during prometaphase, such that there is a rough correlation between tension and kinetochore–kinetochore distance (Waters et al. 1996, Waters et al. 1998). MAD2 and 3F3/2 staining decrease as kinetochore–kinetochore distance increases during maize meiosis, suggesting that the release/dephosphorylation of these proteins is tension sensitive (Yu et al. 1999). Here we demonstrate the same effect on single kinetochore chromosomes. Single kinetochores showed variable degrees of stretching during prometaphase (Fig. 7 A), and stretching was negatively correlated with 3F3/2 staining intensity (Fig. 7 B). The double staining required for these experiments revealed that many of the single kinetochores had divided into half kinetochore units. A distinct bipolar staining pattern was observed, where 3F3/2 staining was fully divided and localized at the poles of the elongated kinetochores (Fig. 6 A3).
It is known that tension at the kinetochore makes kinetochore fibers more stable and/or promotes microtubule bundling within the fiber (Ault and Nicklas 1989; Nicklas and Ward 1994). Therefore, it is likely that once bipolar connections are established by the single kinetochore, the resulting tension stabilizes the interaction and serves to further separate and define the half kinetochore units. For the numerous kinetochores that did not appear to adopt a bipolar interaction with the spindle (e.g., those lying outside the rectangle in Fig. 5 D), the loss of MAD2 and 3F3/2 staining may be the result of a normal “timing out” of the checkpoint. In the presence of microtubule-destabilizing drugs that activate the spindle checkpoint in budding yeast, the cell cycle is delayed for a period of ∼6 h, after which the cell cycle proceeds regardless of the state of the chromosomes (Hoyt et al. 1991). Similarly, Li and Nicklas 1995 demonstrated a 5–6-h delay when a single unaligned chromosome was present during insect meiosis. In our study of the afd1 mutant, the full prophase–metaphase II period was documented in only two cells. Prometaphase II extended for ∼1 h longer than expected in these cells, suggesting that the meiosis II spindle checkpoint times out relatively quickly in maize.
Anaphase II in the afd1 mutant is marked by two events: the loss of staining for checkpoint proteins (Fig. 6 B), and a significant increase in the stretching of many single kinetochores (up to five times their normal diameter; Fig. 5 F). In many cases, the kinetochore stretching appeared to be caused by prometaphase-like tangential interactions with microtubules (Fig. 8 A), whereas in others, there were clearly identifiable kinetochore fibers interacting with each end of the stretched kinetochores (Fig. 8 B). The type of end-on interactions shown in Fig. 8 B occurred on 25% of the stretched kinetochores and 6% of all single kinetochores. These data support the idea that single kinetochores can divide into half kinetochore units, and indicate that each unit can undergo poleward motility.
A large body of literature suggests that the centromeric DNA, which underlies the kinetochore and may be entwined within it (for review see Choo 1997), has a redundant structure. The bulk of the cytological evidence for redundancy comes from experiments where centromeres are split by a process known as centromere misdivision (Darlington 1937). One method for detecting misdivision is to observe the behavior of univalents at meiosis I (unpaired chromosomes) or single kinetochore chromosomes at meiosis II. Such chromosomes often appear to break at the centromeres, in some species, such as wheat, at frequencies close to 40% (Sears 1952). More recently, the molecular analysis of centromeres in Drosophila and Arabidopsis has revealed highly reiterated sequence elements (Sun et al. 1997; Copenhaver et al. 1999). Maize centromeres appear to have a similar structure. Several abundant sequence repeats have been identified at maize centromeres (Kaszas and Birchler 1996; Ananiev et al. 1998), and the centromere of the B chromosome can be reduced by misdivision to ∼10% of its natural size and still retain function (Kaszas and Birchler 1996).
Other evidence suggests that the centromere–kinetochore complex has a visibly redundant external structure. Lima-de-Faria 1958 reviewed this literature, which in some cases provides convincing descriptive evidence for half kinetochore units on single kinetochore chromosomes (i.e., at anaphase II; see Figures 45 and 46 of Lima-de-Faria 1958). More recently, Mole-Bajer et al. 1990 used kinetochore-specific human (calcinosis, Raynaud phenomenon, esophageal dismotility, sclerodactyly, telangiectasia [CREST]) autoantisera to show that, under enhanced immunogold detection conditions, the mitotic kinetochores in Haemanthus appear to be composed of two (and sometimes four) distinct units. Similarly, when mammalian mitotic kinetochores are artificially stretched, a repetitive staining pattern is observed (Zinkowski et al. 1991). We show here using functional assays and specific antibodies to a kinetochore outer domain that the maize meiotic kinetochore is divisible into two parts. We consider it unlikely that kinetochores are especially prone or otherwise limited to being divided into two parts. As discussed above, it is probably the interaction with the spindle and the stabilizing properties of kinetochore–microtubule attachment that is responsible for the twofold redundancy we have observed.
Östergren 1947 suggested that the half kinetochore units that were sometimes visible in the light microscope might independently orient towards different spindle poles and cause misdivision (see also Lima-de-Faria 1956, for additional references and further discussion). Our data support this model for centromere misdivision. However, in maize the high frequency of single kinetochore alignment (∼60%) and stretching at anaphase (∼23%) does not result in a comparable level of centromere breakage. In our data set of 10 telophase II cells, we found no evidence for centromere–kinetochore misdivision. Therefore, the data suggest that while maize half kinetochore units frequently orient towards opposite spindle poles, this biorientation is usually resolved by one half of the kinetochore releasing its attachment. The frequency with which a bipolar kinetochore orientation results in centric misdivision is likely to be a species-specific parameter that depends on the size and sequence of the centromere as well as the molecular makeup of the kinetochore.
We thank Carolyn Lawrence and Evelyn Hiatt for help with statistical analysis.
This work was supported by a grant from the National Science Foundation (9513556) to R.K. Dawe.
Abbreviations used in this paper: 3D, three-dimensional; afd1, absence of first division 1; CENPC, centromere protein C; CentC, maize centromeric tandem repeat.