Anaphase asymmetry and dynamic repositioning of the division plane during maize meiosis

Organisms package their genetic material into chromosomes that must be faithfully separated into two new cells during division. Although this process is crucial in mitosis where mistakes can lead to tumorigenesis (Gordon et al., 2012; Stumpff et al., 2014), mistakes in meiosis result in congenital birth defects and disorders (Nagaoka et al., 2012), and embryo lethality (Hyde and Schust, 2015). Despite being more error-prone than mitosis (Nagaoka et al., 2012), less is known about the meiotic regulation of spindle assembly and chromosome segregation. Female meiosis in mammals (Schatten and Sun, 2009), Caenorhabditis elegans (Sumiyoshi et al., 2002), Drosophila (Casal et al., 1990) and Xenopus (Kalt, 1973) are different from their mitotic and male meiotic divisions. They must assemble spindles, position them within the cell volume and segregate chromosomes, all without the organizational support of centrosomes (Dumont and Desai, 2012). The lack of centrosomes could account for the higher rates of mis-segregation seen during female meiosis compared to male meiosis (Hassold and Hunt, 2001). A study on human oocytes has found that less than 20% of meiosis I spindles are stable. The majority of these spindles cannot focus their poles, form multi-polar spindles or have a high frequency of lagging chromosomes (Holubcová et al., 2015).

Female meiosis in animals is also known for unequal cellular division (Brunet and Verlhac, 2011). In mammals, the meiotic spindle is positioned near the cell periphery to produce one large egg and three small polar bodies (Brunet and Verlhac, 2011). Mistakes in spindle positioning are characteristic of aging and low-quality oocytes that do not produce successful embryos (Brunet and Verlhac, 2011). Meiotic spindles are positioned at the cell cortex by actin microfilaments (Longo and Chen, 1985), their nucleator FORMIN2 (Dumont et al., 2007; Leader et al., 2002) and myosins (Simerly et al., 1998; Weber et al., 2004). A dense actin network interacts directly with both chromosomes (Longo and Chen, 1985; Maro et al., 1986) and microtubules (Azoury et al., 2008) to position the spindle at the cell cortex for asymmetric division. In other asymmetric cell divisions, such as the first zygotic division of C. elegans (Cowan and Hyman, 2004) or Drosophila neuroblast development (Yu et al., 2006), spindles are positioned within the cell volume through cortical pulling forces (Kiyomitsu, 2015). Astral microtubules interact with the minus-end-directed motor dynein, which is asymmetrically anchored on one side of the cell cortex through G-coupled protein receptors (Kotak et al., 2012), and the spindle is pulled into position (Siller and Doe, 2009). Cortical pulling is also used to position spindles symmetrically, but in these divisions dynein localization is non-polarized. Accurate spindle positioning within the cell volume is crucial because the spindle mid-zone has been shown to set the location of the cleavage furrow and, thus, the size and shape of resulting daughter cells (Burgess and Chang, 2005).

Much less is known about spindle positioning and chromosome segregation in plants, where all cell divisions occur in the absence of centrosomes (Schmit, 2002; Wasteneys, 2002; Zhang and Dawe, 2011). Meiosis in plants such as maize have been heavily studied at the level of fixed specimens (Dawe, 1998; Zamariola et al., 2014) but far less studied at the level of live cells (Yu et al., 1997, 1999), leaving gaps in our understanding of spindle and chromosome movements.

In this study, we imaged live-cell division during male maize meiosis I and meiosis II. Maize meiotic cells are easily collected from immature tassels and provide an excellent system in which to study acentrosomal spindles. Early attempts to live-image maize meiosis were limited to observing only chromosome movements with a cell-permeant DNA stain (Yu et al., 1997), but a newly developed fluorescent tubulin fusion (Mohanty et al., 2009) has allowed this first study of live meiotic spindle dynamics. Imaging meiosis I and meiosis II spindles revealed an unexpected and previously undocumented phenomenon – the separation of chromosomes is not consistently symmetric. The observed asymmetry is limited to anaphase A and correlates with the position of the spindle relative to the cell cortex. In cases where the spindle was positioned significantly off-center, the mass of chromosomes furthest from the edge of the cell moved faster and farther, helping to adjust the position of the chromosomes and to correct the initial asymmetry. Additional data show that the phragmoplast, the plant equivalent of the cleavage furrow which determines the plane of division (Otegui et al., 2005), forms midway between the chromosome masses and not at the spindle mid-zone, helping to assure that cell division accurately divides the cell into equal parts.

Chromosome segregation in male maize meiosis was studied by performing live imaging. Microtubules were labeled with cyan fluorescent protein (CFP) fusion with β-tubulin (Mohanty et al., 2009), and chromosomes were labeled with SYTO12, a green nucleic acid stain that penetrates live cells (Yu et al., 1997). Meiotic cells (meiocytes) were extruded from anthers into a culture medium that has been previously demonstrated to support cell growth (De La Peña, 1986; Yu et al., 1997) and were imaged by using fluorescence microscopy. Spindle morphology and chromosome movements were tracked over time, and movies were captured of both meiosis I (Fig. 1A; Movie 1) and meiosis II (Fig. 1B; Movie 2). We confirmed that the CFP–tubulin tag did not affect spindle dynamics by comparisons with unlabeled fixed cells (Fig. S1A,B). All characterization and data analysis presented below is based on 41 live cells.

In both meiosis I and meiosis II, chromosomes align on the spindle metaphase plate, separate into two masses in anaphase and decondense into two new nuclei that are separated by a growing phragmoplast (n=41). Anaphase I lasted an average of 12.7±3.2 min (mean±s.d.), and anaphase II lasted an average of 11.0±3.7 min; this timing was not statistically different (Student's t-test, P=0.36). Anaphase was defined as the period of time from chromosome separation until spindle breakdown. Only one lagging chromosome was observed out of 41 imaged cells (2%) (Fig. 1C); it remained in the center of the spindle and was not pulled to one pole. Maize meiotic chromosome segregation is thus more stable than human female meiosis where these ‘persistent’ lagging chromosomes have been observed in 40% of cells (Holubcová et al., 2015). In 17% of cells, at least one chromosome was observed to be trailing the main mass after initial separation but always caught up to the mass within 5–10 min (Fig. 1C). Because these slow-moving chromosomes have not been previously reported in studies on fixed cells (Yu and Dawe, 2000; Li and Dawe, 2009), we cannot rule out the possibility that culturing and imaging the meiocytes impact chromosome segregation. We also observed astral-like microtubules in some spindles that appeared to contact and curl around the cell cortex (Fig. S2A).

Spindle dynamics are similar in both meiosis I and meiosis II. Metaphase spindle length in meiosis I is 34.2±3.6 μm and 35.3±5.2 μm in meiosis II (mean±s.d.) (Fig. 2A). Eukaryotic chromosomes are separated in anaphase through two mechanisms – movement of chromosomes to the spindle poles (anaphase A) and elongation of the spindle with the poles pushing apart (anaphase B) (Fig. 2B). The usage of these two mechanisms and their degree of contribution to chromosome separation varies from species to species (Maiato and Lince-Faria, 2010). We found that spindles did not elongate from their metaphase length during anaphase (Fig. 2A). In fact, spindles were significantly shorter 20 min after anaphase onset during both meiosis I (24.9±7.0 μm, P<10⁻⁵) and meiosis II (23.7±5.2 μm, P=0.01) (Fig. 2C). The segregation of maize chromosomes appears to be exclusively achieved through movement in anaphase A, with no observed contribution during anaphase B.

Chromosome movement to the poles (anaphase A) exhibits differences between meiosis I and meiosis II. Although spindle length was the same, chromosomes in meiosis II were pulled significantly further apart than in meiosis I. Homologous chromosomes were pulled an average of 5.6±2.1 μm (mean±s.d.) from the metaphase plate to the poles in meiosis I, and sister chromatids were pulled 6.7±0.7 μm (Fig. 2D, P=0.001). If we normalize by spindle length, where being pulled 50% of spindle length means being pulled from the metaphase plate all the way to the poles, meiosis I chromosomes were pulled an average of 32±12% (mean±s.d.), and meiosis II chromosomes were pulled 40±8% (Fig. S3A, P=0.009). Meiosis II chromosomes were pulled closer to the poles of the spindle, both in terms of raw distance and as a percentage of spindle length.

We found that the standard deviation of chromosome segregation distance (anaphase A distance) was larger during meiosis I than during meiosis II (Fig. 2D), suggesting more variation between the distances traveled by each mass of chromosomes. We tracked the movement of the two masses over time and found that chromosomes moved asymmetrically on the spindle in anaphase (Fig. 3A). Fig. 3B shows an example of an asymmetric division where the chromosome masses are outlined in white and tracked in three dimensions from metaphase (T=0 min) through anaphase (T=5–10 min). The asymmetry was quantified by designating one chromosome mass as ‘A’ and determining that it traveled distance D_A, and designating the other mass as ‘B’ and that it traveled distance D_B. The difference in segregation distance is then ΔD_A−B=D_A−D_B (Fig. 3C). A small ΔD_A−B value means that the two masses of chromosomes traveled approximately equal distances toward their respective poles, whereas a large value means the chromosomes segregated asymmetrically.

The asymmetry in chromosome movement was statistically significant. The average total distance traveled by mass A (D_A) was statistically greater than that of mass B (D_B) (Fig. 3D; 6.6±1.9 μm vs 4.9±1.7 μm, P<10⁻⁴; mean±s.d.). Mass A also traveled further than mass B over multiple time points throughout anaphase, beginning 10 min after the metaphase–anaphase transition (P<0.004) (Fig. 3E). The average difference in chromosome segregation distance (ΔD_A−B) was twice as large in meiosis I than in meiosis II (Fig. 3F; 1.8±1.5 μm vs 0.9±0.6 μm, P=0.02), meaning chromosomes segregated more asymmetrically during meiosis I (Fig. 3F). Approximately half of meiosis I cells segregated in a similar manner to meiosis II cells with less than 2-μm difference, whereas the other half traveled quite asymmetrically with a range of 2- to 7-μm difference in chromosome paths (Fig. 3G).

To better compare chromosome movements between cells, D_A and D_B were normalized to total movement (D_A+D_B). In symmetrically segregating cells, D_A and D_B were nearly equal, thus each contributed 50% to chromosome movement. The percentage of total chromosome movement attributed to each chromosome mass was plotted and ordered by symmetry (Fig. 3H). Plotting this data shows that there were not two discrete populations of symmetric and asymmetric cells; instead, we observed a continuum of asymmetry from 50–50% to 80–20% at the extremes. A plot of all cells and their non-normalized chromosome segregation distances, as well as their spindle lengths can be found in Fig. S3B. The asymmetry in chromosome segregation distance (ΔD_A−B) was not correlated with spindle length (Fig. S3C), total distance traveled during anaphase A (Fig. S3D) or the presence of slow-moving chromosomes (Fig. S1C).

The timing of anaphase was not correlated with asymmetry (Fig. S3E), and chromosomes that traveled further did not spend a longer time moving (P=0.4), suggesting that the two masses of chromosomes move at different rates. The data show that in 95% of the observed cells (39/41), the chromosome mass that traveled further (mass A) did so at a faster rate than the chromosome mass that moved the shorter distances (mass B) (P<0.0001, Fisher's exact test). Both the average rate of segregation and the maximum speed was greater for chromosome mass A than that for mass B (Fig. 3I; P=0.003 and P=0.05, respectively; see Fig. S3G,H for plot of all cells). Because cells showed a continuum of symmetry phenotypes rather than two discrete populations (Fig. 3H), we compared the top 25% most asymmetric divisions with the 25% least asymmetric divisions (symmetric divisions). By comparing top and bottom quartiles for the phenotype, we found that in symmetric divisions there was no difference in mass A and mass B speeds at any point during anaphase (Fig. S3F; P>0.7), but in asymmetric divisions mass A was faster than mass B at both 5 and 10 min post-transition to anaphase (Fig. 3J; P≤0.005; see Fig. S3I for plot of all cells). There was also a significant increase in the speed of chromosome mass A from the 5-min time point (0.58 μm min⁻¹) to that at the 10-min time point (0.90 μm min⁻¹, P=0.01). The difference in the initial rate of movement between masses A and B suggests that the chromosome masses could be under different force conditions at metaphase.

As they segregate at different rates, the two masses of chromosomes could be experiencing different forces at the metaphase–anaphase transition. Forces are generated on chromosomes during metaphase as microtubules depolymerize and pull the paired chromosomes toward the poles, but these forces are resisted by chiasma during meiosis I and cohesin during meiosis II. Microtubule polymerization, chromosome elasticity and viscous drag also generate force, which all balance to align chromosomes at metaphase (Dumont and Mitchison, 2009). It is possible that the chromosomes in asymmetrically dividing cells were not aligned in the center of the spindle at the start of anaphase, and thus experienced unequal forces, driving them apart (Fig. 4A). The three-dimensional location of the spindle center was measured and subtracted from the three-dimensional location of the metaphase chromosomes to determine the chromosome–spindle offset (ΔChrom-Spindle). If the asymmetric movement is caused by misalignment of the chromosomes, we would expect the offset distance to be greater in the top quartile of asymmetric divisions, as one mass must trave­l further to approach its pole (D_A>D_B; Fig. 4A). Chromosomes were offset from spindle center by 0.7±0.4 μm (mean±s.d.) in asymmetric divisions and by 0.7±0.3 μm in symmetric divisions (Fig. 4B), with no significant difference in offset. Thus, the asymmetric movement was not due to a failure to align chromosomes in a central position on the spindle.

Positioning of the spindle within the cell volume can define the size of daughter cells, as the site of cell division is often specified by the spindle mid-zone in both animals and plants (Kiyomitsu, 2015). Given that four equally sized products is the optimal outcome in maize male meiosis, it is possible that the asymmetric segregation is a post-metaphase correction mechanism to pull chromosomes into equal volumes when the spindle is not centered in the cell (Fig. 4C). If this were true, we would expect spindles in asymmetric divisions to be more offset from the cell center (ΔSpindle-Cell) than in symmetric divisions. Spindles in asymmetric divisions were indeed more offset (3.3±2.2 μm) than those in symmetric divisions (1.5±0.9 μm, P=0.04) (Fig. 4D).

We also expect that the further-moving chromosome mass (mass A) should travel in the direction of larger cell volume (Fig. 5A). We measured the distance from spindle pole to cell cortex (distances X and Y) and found our expectation to be true. In the top quartile of asymmetric divisions, all cells had greater chromosome movement on the side of the spindle furthest from the cell cortex (distance X), which was significant compared to randomly oriented movement (Fisher's exact test, P=0.01) (Fig. 5B). The difference in distance between the pole and cell wall (X−Y) was greater in asymmetric divisions, both in terms of absolute distance (X−Y, P=0.003) and when normalized for cell size (X−Y/X+Y, P=0.002) (Fig. 5C). In cells in which spindles were offset by more than a standard deviation, asymmetry of chromosome movement was significantly greater than the average of all observed cells (P=0.01; Fig. 5D).

Most convincingly, the extent to which spindles were offset correlated with asymmetric movement such that the greater the spindle was shifted towards one wall, the greater the difference in chromosome movement. The position of the spindle in the cell can explain ∼59% of the observed anaphase asymmetry (R²=0.588; Fig. 5E). Two cells were excluded from this analysis owing to confounding features that could have increased asymmetry in ways that were unrelated to spindle position (red data points in Fig. 5E). One had a persistent lagging chromosome with possible merotelic attachments, which impede chromosome movement towards one pole, and the other had a large spindle morphology defect. Images of these divisions can be found in Fig. S2B,C. No other cells showed significant aberrations in chromosome alignment or spindle shape.

Asymmetry is often dictated by spindle position (Siller and Doe, 2009). In some systems, such as the first zygotic division in C. elegans, the whole spindle is shifted during anaphase to achieve the desired location within the cell volume (Li, 2013). To determine if offset spindles shift toward the cell center, we measured the pole-to-wall distance (X) throughout anaphase (T=0–20 min) in the top quartile of asymmetric divisions. Distance at each time point (X_T) was normalized to the initial pole–wall distance (X₀), such that a value less than 1 indicates a decrease in distance and shifting of the spindle towards cell center (Fig. 5F). The data show that the average pole–wall distance was not significantly different at the end of anaphase than at the beginning. Because the spindles in our assay did not shift (Fig. 5F) or elongate (Fig. 2A,C), the observed asymmetric movement is due solely to unequal movement of chromosomes toward their poles. This represents a new mechanism for positioning chromosomes within a cell volume.

Following chromosome segregation during anaphase, cytokinesis structures form at the spindle mid-zone to split the cell in two (Otegui et al., 2005). These include the cleavage furrow in animals (Gould, 2016) and the phragmoplast in plants (Müller and Jürgens, 2016). Our observations indicated that chromosomes are asymmetrically pulled away from the spindle mid-zone, suggesting that the distance from the phragmoplast to the chromosomes might also be asymmetric (Fig. 6A, option 1). However, anaphase asymmetry (ΔD_A−B) clearly did not correlate with the cytokinesis asymmetry (ΔP_A−B) (Fig. 6B; R²=0.0007) and, instead, the phragmoplast formed at a site that was equidistant from the chromosome masses (Fig. 6A, option 2, 6C). The distance from phragmoplast to chromosome mass strongly correlated with the midpoint distance between chromosomes, calculated by averaging D_A and D_B (Fig. 6C, R²=0.8424). By establishing itself relative to the position of the chromosomes, the phragmoplast might be able to provide a back-up mechanism to segregate chromosomes and ensure that each daughter cell receives at least some genetic material. We observed two instances where the spindle failed to segregate chromosomes, in which the phragmoplast was able to push chromosomes apart into two masses (Fig. 6D).

We investigated the dynamics of chromosome segregation in male maize meiosis; a system that segregates chromosomes with an acentrosomal spindle (Zhang and Dawe, 2011). We found that the two masses of chromosomes did not segregate consistently equal distances (Fig. 3D). Instead, we saw a large variation in anaphase A distance, with masses of chromosomes traveling asymmetric distances on the spindle. In the most extreme case, one mass of chromosomes traveled 80% of the total anaphase A distance, whereas the other mass traveled only 20% (Fig. 3H). The asymmetry observed here is different from the high incidence of lagging chromosomes seen in human (Holubcová et al., 2015) and mouse (Yun et al., 2014) oocytes, as well as in the perturbed meiosis in other species (Dumont et al., 2010). In these previous studies, the two masses of chromosomes moved apart equally, with individual chromosomes lagging behind; here, we saw that all ten of the chromosomes on one side of the spindle frequently moved at different speeds than those of the chromosomes on the other side (Fig. 3I).

To our knowledge, asymmetrical segregation during anaphase A has not been previously described. Previous live-imaging studies of meiosis in human (Holubcová et al., 2015), mouse (Lane et al., 2012), C. elegans (Dumont et al., 2010; Segbert et al., 2003) and Drosophila (Gilliland et al., 2007) showed no evidence of asymmetry, and the previous data on mitosis also suggest that anaphase A is predictably symmetric. There are, however, a few documented examples of unequal chromosome movement during anaphase B. In Drosophila (Kaltschmidt et al., 2000) and C.elegans (Ou et al., 2010) neurogenesis, chromosomes are segregated by a centrally positioned spindle, and in anaphase B, one side of the spindles elongates further than the other, creating a large neuroblast and a small ganglion mother cell (Kaltschmidt et al., 2000; Li, 2013; Ou et al., 2010). In Drosophila, the asymmetry in spindle morphology is created by a larger centrosome that nucleates more microtubules to pull the chromosomes further in one direction (Cai et al., 2003), and in C. elegans, polarization of myosin II squeezes one side of the spindle, allowing the other side to preferentially elongate (Ou et al., 2010). Additionally, in a study on mitotic tobacco culture cells, unequal elongation during anaphase B pulls chromosomes further on one side (Hayashi et al., 2007). Although there is evidence for anaphase B in mitotic maize cells (Duncan and Persidsky, 1958), we have found no evidence for anaphase B in meiosis because the spindle does not elongate (Fig. 2A,C). Our observations indicate that all of the asymmetry occurs during movement during anaphase A (Figs 2E–I and 5F). Maize does not have centrosomes to create an imbalance in microtubule nucleation, and there is no evidence of polarized actin in maize meiosis (Staiger and Cande, 1991), so neither of these established mechanisms can explain the asymmetry seen here.

The function of asymmetric chromosome segregation is typically to move chromosomes into unequal cell volumes to produce daughter cells of different sizes and/or cell fates (Li, 2013). Asymmetry generally arises from the orientation and positioning of the spindle (Siller and Doe, 2009). In mouse meiosis, the spindle is positioned near the cell periphery to create one large egg and three small polar bodies (Brunet and Verlhac, 2011). In the first zygotic division of C. elegans, the spindle is offset from cell center, producing a small posterior cell that gives rise to the germline and muscle, and a larger anterior cell that becomes all other somatic tissue (Gilbert, 2006; Li, 2013). Asymmetry and orientation of the spindle is also used in stem cells to produce one self-renewing and one differentiating daughter (Knoblich, 2008). Maize male meiosis presents the opposite problem, where the spindle is frequently offset from the cell center but the final products of meiosis are highly uniform in size. If the purpose of anaphase asymmetry in maize is to place the new nuclei in equal volumes, it should be correlated with how offset the spindle is within the cell (Fig. 5A). We find that there is indeed such a correlation, as the greater the spindle was shifted towards the edge of the cell, the greater the asymmetry (Fig. 5E). When the spindle was offset to a large extent (the top quartile of asymmetric divisions), chromosomes always moved further on the side of the spindle with greater pole–wall spacing (Fig. 5B). The spindle itself did not move (Fig. 5F), but the rate and distance of chromosome movement within the spindle served to correct for irregularities in spindle position and to help to place chromosomes into opposite hemispheres of the cell.

The phragmoplast forms the new cell wall after cell division in plants. In mitotic cells, the position of both the spindle and phragmoplast is dictated by the pre-prophase band, a microtubule array around the cell periphery (Mineyuki et al., 1991), and its associated proteins, TONNEAU (Azimzadeh et al., 2008; Camilleri et al., 2002; Traas et al., 1987) and TANGLED (Cleary and Smith, 1998; Smith et al., 1996). Meiotic plant cells lack a pre-prophase band (Chan and Cande, 1998), and analysis of fixed specimens suggests that phragmoplasts are formed from the central fibers of the spindle that remain after chromosomes segregation (Otegui and Staehelin, 2000; Shamina et al., 2007; Staehelin and Hepler, 1996). Although, in most cell types, the position of the metaphase plate marks the location of the spindle mid-zone, owing to anaphase asymmetry, these positions often differed in our assay. We found that the phragmoplast appeared halfway between the two chromosome masses after anaphase rather than at the original spindle mid-zone (Fig. 6A–C). Dynamic establishment of the phragmoplast relative to chromosomes allowed cells to correct the division plane if the spindle was improperly positioned (Fig. 5B,E) and provided a back-up mechanism for segregation when the spindle failed to fully separate chromosomes (Fig. 6D).

We found that chromosomes moved farther and faster on the side of the spindle that was most distant from the cell cortex (Fig. 3I,J, Fig. 5A,B). Maize meiotic cells must have a sensing mechanism in order to integrate spindle position with chromosome segregation distances. One possibility is astral microtubules that reach from the spindle poles towards the cortex and relay positional information. Higher plants lack centrosomes (Schmit, 2002; Wasteneys, 2002) and were thought to lack astral microtubules as well (Lloyd and Hussey, 2001; Smirnova and Bajer, 1992), but recent microscopy analyses have revealed astral-like microtubules that reach from the acentrosomal pole towards the cell cortex in Arabidopsis (Chan et al., 2005) and tobacco (Dhonukshe et al., 2005). We observed similar astral connections to the cortex in our time-lapse data (Fig. S1A). In animals, astral microtubules work in concert with cortical dynein to move spindles into position (Siller and Doe, 2009). However, higher plants lack dynein (Lawrence et al., 2001). It is possible that astral-like microtubules act as messengers, relaying positional information to the spindle and modulating microtubule dynamics. Astral-like microtubules could bind to regulatory molecules such as KLP10 or CLASP, which regulate microtubule flux rates in Drosophila and produce a similar asymmetry phenotype with unequal movement of chromosomes towards poles when depleted (Matos et al., 2009). Although very little is known about microtubule flux regulation in plants (Dhonukshe et al., 2006), our data suggest that such regulators could exist and are sensitive to spindle position within the cell. Overall, our findings demonstrate that positional signals are transduced from the cortex to modulate chromosome dynamics as late as anaphase to correct errors in spindle placement.

A maize line (Zea mays ssp. mays) containing CFP fused to the N-terminus of β-tubulin (β-TUB1) was generated by the laboratory of Anne Sylvester (University of Wyoming, WY). Plants were genotyped for the CFP–tubulin transgene using a CTAB DNA extraction protocol (Clarke, 2009) on leaf tissue and primers that annealed within the CFP (5′-GGAGTACAACTACATCAGCCACAACGTC) and tubulin (5′-CCGGACTGACCGAAGACGAAGTTGT) sequences. The maize line containing dv1 was obtained from the Maize Genetics Cooperation Stock Center (University of Illinois) and genotyped as previously described (Higgins et al., 2016). All chemicals and reagents, unless otherwise stated, were purchased from Sigma Aldrich.

Male meiotic cells were harvested from immature tassels as previously described (Yu et al., 1997). Meiocytes were extruded from anthers into live-cell imaging medium, pH 5.8–5.9 (De La Peña, 1986; Yu et al., 1997), that contained a final concentration of 2 μM SYTO12 Green DNA dye (Invitrogen Molecular Probes). Cells were staged for meiosis I and meiosis II, loaded onto poly-L-lysine-coated coverslips (Corning) and sealed onto microscope slides (Fisher Scientific). Cells were imaged on a Zeiss Axio Imager.M1 fluorescence microscope with a 63× Plan-APO Chromat oil objective. Images were collected every 3–7 min in three dimensions using a 20-μm Z-range and 1-μm step size with 50-ms exposure for CFP and 30-ms exposure for SYTO12 and 2×2 binning. Three-dimensional volume renderings of an example meiosis I cell (Movie 3) and meiosis II cell (Movie 4) demonstrate that cells were not compressed between the coverslip and slide during imaging. The sample size was 41 cells, and the asymmetric and symmetric categories were defined as the top and bottom quartile of the asymmetric phenotype. Sample size was sufficient for statistical power in both Student's t-test and Fisher's exact test, used as described in the text.

Images were analyzed using Slidebook software (Intelligent Imaging Innovations, Denver, CO, USA). Cells, spindles, chromosomes and phragmoplasts were identified as objects by thresholding. Cells, labeled by diffuse cytoplasmic CFP–tubulin monomers, were thresholded at approximately 35% above CFP background, spindle and phragmoplast signals at approximately 50% above background, and chromosomes at approximately 5% above FITC background. Object statistics were extracted including center of volume (x, y, z coordinates of the center of the object) and longest chord (distance between the two furthest pixels within the object). Spindle length was measured as the longest chord within the spindle object after constrained iterative deconvolution using a calculated point-spread function. Chromosome movements were calculated as the three-dimensional distance between the center of volume at different time points, and chromosome speed was calculated as this distance divided by time. The total distance traveled in anaphase (D_A or D_B) was calculated as the distance between the initial position on the metaphase plate and the final position near the spindle pole. The asymmetry value (ΔD_A−B) is the difference in these distances: ΔD_A−B=D_A−D_B. Congression of chromosomes on the metaphase plate (ΔChromosome-Spindle) was calculated as the difference between the chromosome center of volume and the spindle center of volume at metaphase. The offset of the spindle from the cell center (ΔSpindle-Cell) was calculated as the difference between the spindle center of volume and the cell center of volume. Cytokinesis ΔP_A−B is the difference in distance from each chromosome mass to the phragmoplast center of volume. The xyz position of the spindle pole was determined by the edge of the thresholded spindle outline (50% above deconvolved background CFP signal) as the furthest point from the center of the spindle, measured using Slidebook ruler function. The distance from pole to cell cortex was measured by determining the shortest xy distance from the pole to the thresholded outline of the cell (35% above non-deconvolved background CFP signal) within the same z plane as the pole using Slidebook ruler function.

Immunolocalization was performed as previously described (Higgins et al., 2016). Anthers from immature tassels were fixed for 60 min in 4% paraformaldehyde-PHEMS buffer (60 mM PIPES, 25 mM Hepes, 10 mM EGTA, 2 mM MgCl₂, 0.35 M Sorbitol, pH 6.8), washed three times in 1× PBS and dissected for meiotic cells. Staged meiocytes were adhered to poly-L-lysine coverslips by centrifugation at 100 g for 1 min, then permeabilized for 1 h in a 1% Triton X-100, 1 mM EDTA, 1× PBS solution. Coverslips with affixed cells were blocked in 10% goat serum for 90 min, incubated with a monoclonal antibody against sea urchin α-tubulin (Asai et al., 1982) at 37°C overnight, blocked again with 10% goat serum and then incubated with a Rhodamine-conjugated AffiniPure Goat Anti-Mouse IgG (H+L) secondary antibody (Jackson ImmunoResearch) for 150 min. Between each step, coverslips were washed three times with 1× PBS solution. Coverslips were mounted with ProLong Gold with DAPI (Thermo Fisher Scientific) and imaged as described above, except there was no binning and exposure times were optimized for each channel and ranged from 0.1–1 s.

We thank Caroline Jackson for genotyping the CFP–tubulin plants and Amy Hodges for genotyping the dv1 plants, as well as Jonathan Gent and other members of the Plant Biology Department for discussion of the data.