Zea mays Meiotic Spindle Ultrastructure Reveals Kinetochore-Microtubule Interface and Embedded Membrane Components

Abstract

Introduction

Spindles are microtubules-based machines whose primary function is to accurately segregate chromosomes in both mitotic and meiotic cell division. The structure of spindles is critical for their function; errors in morphology or attachment to chromosomes lead to aneuploidy, potentially resulting in disease, infertility, and lethality. Electron microscopy studies have yielded fine-detail spindle ultrastructures in many plant and animal species, but no studies have investigated the spindle of Zea mays, a critical crop, and cytogenetic model system.

Methods

Here we use electron tomography (ET), reconstruction, and modeling to obtain three-dimensional, nanometer-resolution of the Z. mays meiotic spindle. Structures such as microtubules, kinetochores, vesicles, membrane channels, and nuclear envelope were modeled through a partial spindle reconstruction, and confirmed using immunostaining and live fluorescence microscopy.

Results

ET revealed that maize spindles contain 8–18 kinetochore microtubules (kMTs) per kinetochore, which are approximately 776 nm in diameter and 316 nm in depth. Small ∼37 nm vesicles were identified, as well as larger (∼5 µm long, 800 nm wide) membrane structures with channels that allow spindle microtubules to pass through. These membrane channels stain positively for the ER-marker protein disulfide isomerase. Imaging of prophase meiotic cells revealed a cross-hatch microtubule arrangement in the perinuclear ring on the external surface of the nuclear envelope, which also contained type II nuclear grooves with transnuclear microtubules passing from the nucleus to the cytoplasm.

Conclusions

Z. mays meiotic spindles are similar to animal counterparts with a comparable number of kMTs and pre-spindle transnuclear microtubules but also plant-specific features such as Golgi-derived vesicles to assist cell plate formation, internal ER membrane channels, and a perinuclear microtubule ring that aids spindle assembly. Maize kinetochores have an electron-diffuse ball in cup morphology that is comparable in size to Drosophila kinetochores and larger than mammalian kinetochores.

Introduction

Spindles are the critical machinery of cell division, segregating chromosomes into the resulting cells. In mitosis, spindles segregate the identical sister chromatids of a replicated chromosome. Meiosis is a more complex process, requiring two rounds of division, in which spindles first separate homologous chromosomes in meiosis I, then sister chromatids in meiosis II. All spindles consist of microtubules, polarized hollow tubes composed of α and β-tubulin heterodimers that dynamically organize into a bipolar structure [1]. Three categories of microtubules provide spindle structure and functionality. Kinetochore microtubules (kMTs), often referred to as k-fibers, reach outward from the spindle poles and attach to chromosomes via the kinetochore, a proteinaceous bridge structure that assembles on the centromere [2]. Interpolar microtubules interact to form the core of the spindle and provide structural integrity, while astral microtubules nucleate away from the spindle to position it within the context of the cell [3]. Together, these microtubules coordinate the critical function of accurately segregating chromosomes. Errors in spindle structure or stability can result in aneuploidy, an incorrect number of chromosomes associated with disease and organismal lethality [4].

Given the critical role of the spindle, many studies have sought to characterize its structure and function through microscopy. Fluorescence microscopy has facilitated studies in live mitotic cells, revealing the dynamics of spindle assembly, chromosome attachment and segregation, and spindle disassembly in animals such as Caenorhabditis elegans [5], Drosophila [6], Xenopus [7], mice [8], and humans [9]. Similar studies have characterized plant mitotic spindles including Arabidopsis [10], the moss Physcomitrella patens [11], cultured tobacco [12], and maize [13]. Live imaging of meiotic spindles has proven more challenging, but studies in mice [14, 15], humans [16], Xenopus [17, 18], Drosophila [19], Arabidopsis [20], and maize [21, 22] have revealed commonalities and differences in overall spindle structure, dynamics, and regulation.

While light microscopy has elucidated many spindle features in both plants and animals, spatial resolution is limited by the wavelength of light. Electron microscopy provides resolution to nanometer scales and can reveal fine-scale ultrastructures not visible by light microscopy [23]. Electron microscopy studies in the budding yeast Saccharomyces cerevisiae have revealed spindle features such as the spindle pole [24], number, type, and orientation of individual microtubules [24–26] and the plasticity of microtubule number based on the number of chromosome attachments [27, 28] Studies in P. patens and tobacco have revealed the axial degree of spindle orientation in asymmetric cell divisions [29]. Organization of microtubules, including their proximity and cross-bridging structure has been shown to provide mechanical integrity to Chlamydomonas and mammalian culture cell spindles [30]. Unlike smaller spindles in budding yeast where individual microtubules reach from the pole to spindle midzone [24, 25], EM studies in larger spindles such as C. elegans [31], Haemanthus [32], Xenopus [33], Chlamydomonas [30], humans [9, 34], and other mammalian cells [30] have shown that interdigitating microtubules create a network of microtubules that produce force to segregate chromosomes.

Microtubules are classified by their function within the spindle. Astral microtubules reach away from the poles of the spindle and assist with spindle positioning [35]. Interpolar microtubules reach inward toward the spindle midzone and provide structural integrity [3]. Spindle force models have theorized that kinesins cross-link interpolar microtubules to generate outward molecular force [27, 36] as deletion or overexpression of these kinesins alters spindle length [37–39]. These bundles of microtubules can support the overall growth and lengthening of each other through reduced catastrophe and increased rescue [9, 31, 40]. kMTs attach to chromosomes via the kinetochore and are responsible for directing chromosome segregation [41]. EM studies of the microtubule-kinetochore interface have highlighted the complexity of chromosome attachments to the spindle. Budding yeast have only a single microtubule attaching per kinetochore [24, 25], a number that does not vary even when its chromosomes are combined into massive mega-chromosomes requiring greater drag force [28]. All other studied eukaryotes have multiple kMTs within a k-fiber attaching to a kinetochore, ranging from 2 to 4 kMTs in the fission yeast Schizosaccharomyces pombe [42], to 15–25 in mammalian species [43, 44], to ∼100 kMTs in Haemanthus [32]. The number and attachment of kMTs to the kinetochore is critical for the spindle checkpoint, a surveillance mechanism that delays the metaphase to anaphase transition until all chromosomes are properly oriented on the spindle [45]. Attachment status and spatial orientation of kMTs to generate tension are the main activators of the checkpoint, and thus, this interface is critical for ensuring accurate chromosome segregation [45, 46]. Advances in specimen preparation including high pressure freezing and freeze substitution result in the excellent preservation of cell fine structure. Electron tomography (ET) of these samples allows for three-dimensional imaging [47] and fine-detail information on transient biological structures, including high resolution imaging of spindles and their interface with chromosomes in humans [34, 48], C. elegans [31], O. tauri [49], S. cerevisiae [27, 49], Chlamydomonas, and mammalian culture cells [30].

Z. mays (maize) is the most widely produced food crop, with an annual worldwide yield nearly double that of the next crop, wheat [50]. In addition to its agronomic importance, maize is also a model system in cytogenetics and chromosome biology, notably used by Barbara McClintock in her Nobel winning discoveries on chromosome structure, meiotic recombination, and transposable elements [51]. Despite the importance of maize, its 3D spindle structure has not been investigated using ET. A single study investigated maize kinetochores using traditional electron microscopy [52], but the spindle ultrastructure and microtubule-kinetochore interface remains uncharacterized. We recently characterized the Z. mays meiotic spindle via live fluorescence microscopy, and discovered the assembly process to be both error-prone and rapid compared to animal systems [21]. The brief meiotic metaphase timespan (15 min) could explain the lack of characterization by electron microscopy. Here, we used ET to collect three-dimensional information of the spatial arrangement of meiotic spindle microtubules in well-preserved samples. Together, our observations provide a comprehensive characterization of the Z. mays meiotic spindle ultrastructure, a structure critical for the accurate segregation of chromosomes in this important crop and cytogenetic model system.

Methods

Maize Lines and Genotyping

Inbred maize (Z. mays ssp. mays) line B73 was used for electron microscopy and immunolocalization experiments. A transgenic maize line previously used in living imaging studies [21, 22] was used for live fluorescence microscopy, which contains an N-terminal fusion of β-tubulin with CFP. The fluorescent transgenic line was generated by the laboratories of Anne Sylvester (University of Wyoming, Laramie, WY, USA) and David Jackson (Cold Spring Harbor Laboratory, NY, USA) [53]. Transgenic plants were genotyped using a CTAB DNA extraction protocol [54] and primers that spanned the CFP-β-TUB1 transgene (forward, anneal in CFP: 5′-GGA​GTA​CAA​CTA​CAT​CAG​CCA​CAA​CGT​C; reverse, anneal in β-TUB1: 5′-CCG​GAC​TGA​CCG​AAG​ACG​AAG​TTG​T). All chemicals and reagents were purchased from Sigma Aldrich unless otherwise noted.

Sample Preparation and Electron Microscopy

Anthers were dissected from the immature tassels of maize inbred B73 and staged for meiosis as previously described [55]. Anthers containing meiocytes in meiosis I and II were placed in aluminum planchettes with 15% (w/v) dextran in water and frozen using a Balzer’s HPM 010 High Pressure Freezing Machine and freeze substituted as previously described [52]. Specifically, samples were stored in substitution fluid at −80°C for 4 days then transferred to −20°C for 3 h, 4°C for 2 h and brought to room temperature for 30 min. Samples were then rinsed three times in anhydrous acetone, and infiltrated with Araldite/Embed 812 resin and flat embedded between two glass slides [52]. After polymerization, samples were observed using light microscopy to identify meiotic anthers.

Regions containing meiotic anthers were excised and remounted on blank Epon blocks. Serial, 300 nm thick sections were cut using a Leica Ultracut-E microtome and collected onto formvar-coated slot grids. The sections were post stained using 2% aqueous uranyl acetate followed by Reynolds lead citrate. Gold particles (15 nm) were then affixed to both surfaces of the section to serve as alignment markers. Sections were imaged in a Tecnai F30 microscope (ThermoFisher Scientific, Waltham, MA, USA) and cells containing meiotic spindles identified. Tilt series from serial sections were collected using the SerialEM program [56] using a Gatan One View camera (Gatan, Inc. Pleasanton, CA, USA) at a pixel size of 1.5 nm. In some cases, montaged tilt series were collected to increase the area of the spindle imaged. Tomographic volumes from the aligned tilt series were computed using the IMOD software package [57, 58]. Serial sections were stitched together to analyze volumes greater than 300 nm; the number of sections stitched depended on structure being analyzed.

3D Tomographic Reconstruction and Modeling

Tomograms were displayed, modeled, and analyzed using the 3dmod program within the IMOD software package [57, 58]. Membranes, microtubules, chromosomes, and vesicles were modeled in the tomographic volume as previously described [59–61]. The slicer tool was used to display and tilt around all three axes, allowing the identification and tracking of microtubules and mapping the interconnected membrane channels. kMTs were identified by their characteristic flared plus end terminating near the kinetochore [62]. Chromosome regions and membrane surfaces were modeled using the contour tool in each tomographic and reconstructing the three-dimensional shape by computing meshes to define the surface of each object [57]. Small vesicles were modeled as spheres and diameter measurements were calculated using the Imodinfo program [61]. Length and width measurements of defined objects were also calculated using Imodinfo.

Immunolocalization

Anthers were dissected and staged for meiosis as described above for electron microscopy [55]. Immunolocalization was performed as previously described [22]. Anthers were fixed for 120 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) and washed three times with 1× PBS with 5 min of shaking incubation between each wash. All subsequent washes described below were similarly performed. Meiocytes were extruded from fixed anthers and adhered to poly-L-lysine coverslips by centrifugation at 200 g for 1 min, then permeabilized for 60 min in solution of 1% Triton X-100, 1 mm EDTA, 1× PBS, then washed. Meiocytes were then blocked in 10% goat serum for 120 min, washed, and incubated overnight in a humid chamber with a monoclonal antibody to ɑ-tubulin (host: mouse, Sigma Aldrich T5168) and a polyclonal antibody to protein disulfide isomerase (PDI) (host: rabbit, Sigma Aldrich P7496). After overnight incubation, cells were washed, blocked again with 10% goat serum for 120 min, washed, incubated with secondary antibodies: ɑ-mouse rhodamine conjugated (Rockland Immunochemicals, 610-1004) and ɑ-rabbit FITC conjugated (Rockland Immunochemicals, 311-1204) for 150 min, washed, and mounted onto slides with ProLong Gold with DAPI (Thermo Fisher Scientific).

Fluorescence Microscopy

Meiocytes were examined using a Zeiss AxioObserver 7 Basic Marianas inverted epifluorescence microscope with an X-Cite 120LED Boost System, a Plan-Apochromat 63×/1.4NA oil objective, and a Prime 95B Black Illuminated Scientific CMOS camera. For immunolocalization, fixed cells were imaged for DAPI (500 ms exposure), FITC (300 ms exposure), and rhodamine (100 ms exposure) with a multi-plane Z stack across a 20-µm range and 0.5-µm step size resulting in 40 stacked images. For live-cell imaging, meiocytes from CFP-tubulin transgenic lines were extruded from anthers into live-cell imaging medium, pH 5.8–5.9, staged for meiosis I and II, loaded onto poly-L-lysine coated coverslips and sealed onto slides as previously described [55]. Images were collected on the system described above every 3–7 min in three dimensions using a 20-µm Z-range and 1-µm steps size with 50 ms exposure for CFP with 2 × 2 binning. All images were analyzed using SlideBook software (3i: Intelligent Imaging Innovations). For live imaging, 8 meiosis II cells were imaged in prophase before nuclear envelope breakdown and 12 cells were imaged in metaphase II with assembled spindles.

Results

We previously imaged the live assembly and disassembly of the Z. mays meiotic spindle using fluorescence microscopy [21, 22]. While this approach provides insight into the real time dynamics of spindle processes, it lacks resolution to characterize the microtubule-kinetochore interface, morphology of kinetochore within its chromatin context, and the structural arrangement of individual microtubules. In this study, we used ET to create 3D reconstructions of the maize meiosis II spindle. Male gametogenesis in plants occurs in anthers, tube-like structures that hold hundreds of meiotically dividing cells called meiocytes [63]. Prepared anthers were dissected to identify meiocytes; Figure 1a and d show lengthwise cross-sections of anthers with box-like tapetal cells that form the outer tube of the anther (yellow arrows) and the hollow inner cavity that houses the larger, rounded meiocytes (red arrows). Meiocytes were assessed for their meiotic stage, and cells were chosen that had fully formed metaphase spindles (Fig. 1a, d, red arrows). Due to the difficult nature of Z. mays anther preparation [52] and the short metaphase time [22], only 2 cells were identified with chromosomes aligned on a fully assembled metaphase spindle (Cell A: Fig. 1a; Cell B: Fig. 1d). Our previous analysis of live spindle dynamics revealed that maize meiosis I and meiosis II cells spend only 45 min in active spindle assembly (15 min in metaphase) in the plant life cycle [22] compared to the 3–8-h range seen in animal meiosis [14, 16, 64]. This limited temporal window provides a possible explanation for the minimal studies on maize meiosis [52] and lack of studies characterizing the spindle. Cell A (Fig. 1a) and Cell B (Fig. 1d) were both in meiosis II (identifiable by its characteristic half-sphere morphology) and used for further characterization of their spindle structures.

Fig. 1.

Z. mays meiotic spindle imaged using electron microscopy. a Low magnification view of a Z. mays anther containing several meiotic cells (half-moon shaped) contained within it. Red arrow indicates the meiotic cell further magnified in b, c (referred to as “Cell A”). The white dashed box indicates the region enlarged in b. Yellow arrows indicate tapetal cells that line the anther. b Low magnification overview of 300 nm section used for tilt series collection with visible half meiotic metaphase spindle. The white box indicates the region enlarged in c. Within the dashed box, one mass of darkly stained chromosomes is visible and a network of microtubules. c Tomographic slice displaying condensed meiotic chromosome masses (red arrow) and microtubules of the spindle (examples indicated with white arrows). d Low magnification view of a Z. mays anther with several meiotic cells (half-moon shaped) contained within it. Red arrow indicates the meiotic cell further magnified in b, c (referred to as “Cell B”). e Low magnification overview of 300 nm section used for tilt series collection with visible midzone of meiotic metaphase spindle. The white box indicates the region enlarged in c. Within the dashed box, two masses of darkly stained chromosomes are visible and a network of microtubules. f Tomographic slice displaying condensed meiotic chromosome masses (red arrows) and microtubules of the spindle (examples indicated with white arrows). Scale bars: 1 µm.

Meiotic Spindle Structure and Kinetochore-Microtubule Interface

Meiotic cells with metaphase spindles and aligned chromosomes were identified in prepared anthers (red arrows, Fig. 1a: Cell A and Fig. 1d: Cell B). Meiotic cells were identified through low magnification light microscopy; several meiocytes can be seen inside the anthers in Figures 1a and d, but only the cells indicated with red arrows have metaphase spindles. White boxes indicate the regions magnified and shown in Figure 1b, e. Chromatin is electron dense; thus, dark, condensed chromosome masses can be seen in both Figure 1b and e. These chromosomes are aligned on spindles which can be seen as a fiber-like network. The regions indicated in white dashed boxes are further magnified in Figure 1c and f. Tomographic slices are shown in Figure 1c and f, with electron-dense chromosomes indicated with red arrows and representative microtubules indicated with white arrows. Figure 1f shows magnification over the spindle midzone with microtubules associating with both sides of the chromosome mass, whereas Figure 1c is a magnification of a half-spindle. ET was performed using multiple tomographic slices, allowing three-dimensional image acquisition and reconstruction of a partial spindle. Subcellular structures including microtubules, chromosomes, vesicles, and other membrane surfaces were modeled using the IMOD program as previously described [59–61].

Microtubules are shown in green, overlaid on a tomographic slice (Fig. 2a, c) or displayed alone in three-dimensional space (Fig. 2b, d). kMTs were identified by their morphology and location; specifically, kMTs terminate in a cluster near the kinetochore with their plus ends identifiable by their flared morphology as previously described [62]. kMTs are shown in pink and their ends are indicated with a light blue sphere in Figure 2. Clusters of microtubules that attach to a single kinetochore are called k-fibers [65], and the number of microtubules in a k-fiber varies greatly across species [66]. Three-dimensional modeling of the kinetochore-microtubule interface through multiple serial sections allowed us to quantify the number of kMTs in maize k-fibers. Six kinetochores were identified (white boxes, Fig. 2b, d) and the number of kMTs were counted (Fig. 2d). There was an average of 10.8 ± 3.8 kMTs per kinetochore and a median of 9.5 kMTs. There was substantial variation across kinetochores, with a high of 18 kMTs identified on one kinetochore and several kinetochores with 8 kMTs. Also visible in Figure 2 are vesicles (white spheres) that will be discussed in below and analyzed in Figure 4.

Fig. 2.

Three-dimensional modeling of maize meiotic spindles. a Cell A: microtubules of the spindle were modeled in three-dimensional space through multiple tomographic slices. Microtubules (green), kinetochore microtubules (kMT, pink), and their associated plus ends (blue spheres) are visible overlaying the tomographic slice. In all images, microtubules that appear short or truncated are microtubules that exit the volume of partial spindle reconstruction. White spheres represent vesicles that will be analyzed in Figure 4. The black box represents a kinetochore further analyzed in Figure 3a. b Cell A: 3D model of microtubules in the meiotic spindle with microtubules (green), kMTs (purple), and their plus ends (blue spheres). One kinetochore is visible (Kinetochore 1, labeled KT1 and white box), identified by the cluster of kMTs that terminate at the chromosome interface. c Cell B: microtubules of the spindle were modeled in three-dimensional space through multiple tomographic slices. Microtubules (green), kMTs (pink), and their associated plus ends (blue spheres) are visible overlaying the tomographic slice. The black boxes represent kinetochores further analyzed in Figure 3b and c. d Cell B: 3D model of microtubules in the meiotic spindle with microtubules (green), kMTs (purple), and their plus ends (blue spheres). Five kinetochores are visible (white boxes, labeled KT 2–6), identified by the cluster of kMTs that terminate at the chromosome interface. e Quantification on kMTs at each kinetochore; Kinetochore 1 was identified in Cell A (a, b); Kinetochores 2–6 were identified in Cell B (c, d). Scale bars: 1 µm.

Fig. 4.

Small vesicles within the meiotic spindle. a 3D model of Cell A spindle containing chromosome mass (yellow), microtubules (green), kMTs (purple), and small vesicles (white). Blue contour lines represent membrane channels that will be analyzed in Figure 5. b 3D model of Cell B spindle containing chromosome mass (yellow), microtubules (green), kMTs (purple), and small vesicles (white). c Histogram of vesicle diameter, binned in 5 nm increments. Scale bars: 1 µm.

Kinetochores morphology varies across the kingdoms. Animal kinetochores appear as a disk-like structure with variable morphology depending on sample preparation [43, 67]. Conventional sample preparation using glutaraldehyde fixation yielded a trilaminar structure with two electron-dense domains separated by a translucent middle layer [43, 68]. More modern sample preparation using high pressure freezing and freeze substitution showed animal kinetochores with two domains, an inner electron-dense region with a light-staining outer fibrous corona domain [34, 69]. Rather than a disk-like structure, plants have a “ball in a cup” morphology with no clearly defined kinetochore visible due to its electron diffuse nature [52, 70]. Plant kinetochores are thus defined by the electron-dense chromatin “cup” that holds the ball-like kinetochore [52, 70, 71]. Previous electron microscopy studies have identified the chromatin cup, but due to the lack of three-dimensional imaging, the dimensions have not been quantified. Here, the three-dimensional chromatin surface was modeled using the IMOD contour tool and the associated computation of a surface mesh as previously described [57]. The chromatin surface is displayed in yellow in Figure 3. The length across each of the six chromatin cup was measured, as well as the depth of the cup (white dashed lines are representative of the dimensions measured) (Fig. 3a–f). There is variability in the size of the chromatin cup, but the average length was 776 ± 168 nm and average depth was 316 ± 145 nm. One of chromatin cups (Cup 2, Fig. 3b) showed a different morphology with two small cups with a chromatin bulge between them. For this cup, we measured across the total length and averaged the depth of the two cups (Fig. 3b). Interestingly, the largest chromatin cup (Cup 3) also had the greatest number of kMTs (Kinetochore 3, 18 kMTs); thus, there may be a correlation between kinetochore size (as defined by the chromatin cup dimensions) and attached microtubules.

Fig. 3.

Three-dimensional analysis of maize “ball in a cup” kinetochore morphology. The chromatin cup was analyzed by modeling the 3D surface of the chromosome mass (yellow contours) and measuring length and depth of the cup (white dashed lines). White spheres represent vesicles that will be analyzed in Figure 4. a Chromatin cup associated with Kinetochore 1, Cell A. b Chromatin cup associated with Kinetochore 2, Cell B. c Chromatin cup associated with Kinetochore 3, Cell B. d Chromatin cup associated with Kinetochore 4, Cell B. e Chromatin cup associated with Kinetochore 5, Cell B. f Chromatin cup associated with Kinetochore 6, Cell B. g Quantification of chromatin cup dimensions: length across the cup and depth into the chromatin measured in nanometers, including average length and depth. Scale bars: 200 nm.

Vesicles and Membrane Channels within the Spindle

The fine resolution of electron microscopy allows for identification of new features. While modeling the microtubules and chromatin surfaces of the maize meiotic spindle, we identified several membrane-based structures: small vesicles (Fig. 4) and larger membrane channels (Fig. 5) scattered throughout the partial spindle volume. In both Cell A (Fig. 4a) and Cell B (Fig. 4b), the modeled partial spindle volumes have over 200 small, spherical vesicles (white spheres, Fig. 4a, b). Diameter of all vesicles was calculated using Imodinfo [61]. The average diameter of vesicles is 37 ± 9 nm (Fig. 4c), which is similar in size to vesicles that have been identified in post-anaphase stage Arabidopsis mitotic spindles and during cell plate formation [61]. These late-stage mitotic vesicles were approximately 55 nm in diameter and shown to be derivatives of the Golgi apparatus [61]. The metaphase meiotic vesicles observed here were smaller than previously observed in Arabidopsis and BY-2 tobacco cells [61, 72]. The size difference could be due to the temporal difference in vesicle size as studies have shown that vesicles fuse post-anaphase, thus growing in size during phragmoplast establishment, whereas vesicles here were measured in metaphase [61, 73].

Fig. 5.

Membrane channels within the meiotic spindle. a Tomographic slice displaying a membrane-bound channel (green) within the meiotic spindle. Chromosome mass is outlined in yellow. b 3D model of a portion of a membrane channel (green) with neighboring microtubules (pink). c Membrane channels (blue) were identified throughout the meiotic spindle. Three channels (numbers indicated) were measured for length and width. d Table of membrane channel measurements. Scale bars: 100 nm.

In addition to the small vesicles dispersed throughout the spindle, ET revealed larger membrane structures embedded in the spindle (Fig. 5). While modeling chromatin surfaces (yellow) and microtubules (purple), membrane surfaces (green) were identified within the spindle (Fig. 5a, b); when modeled through three dimensions using multiple tomographic slices, the membranes were shown to be enclosed structures (volume shaded light green) (Fig. 5a). These “membrane channels” generally run parallel to microtubules (example in Fig. 5b). Some membrane channels were small and rounded, and others were long continuous structures. Three of the long, continuous membrane channels were modeled (Fig. 5c); their interior dimensions were measured at multiple locations along their longitudinal and latitudinal axes and averaged. The average length of the three membrane channels is 4.93 μm and average width is 0.79 μm (Fig. 5d), substantially larger than the vesicles with 37 nm diameters (Fig. 4c).

To confirm these channels were not an artifact of electron microscopy preparation, we investigated their presence in live cells using CFP-tubulin transgenic maize line, as previously described [2, 5]. Spindle images were acquired in the three-dimensional space, and cross-sectional slices of spindles were analyzed. Figure 6a and b are representational images of meiotic cells; the cell volume can be seen due to diffuse, unincorporated CFP-tubulin monomers in the cytoplasm. Figure 6a is a cross-sectional view of two spindles while Figure 6b is a 90° rotational view of the same spindles, where the half-spindle form is apparent with condensed poles. Figure 6c is a magnified image of the spindle cross-section; analyzed images revealed fluorescent bundles of microtubules interspersed with negative, non-fluorescent space. The negative space could be empty, contain a few microtubules, or be a space occupied with other structures, such as membrane channels. Using thresholding to define the object boundaries as previously described [1], we measured the diameter of the cross-sectional negative space and found it to be 1.39 ± 0.6 µm, suggesting these spaces could contain the membrane channels (0.79 μm width) (Fig. 6g).

Fig. 6.

Meiotic spindle and membrane channels imaged by fluorescence microscopy. a Live meiotic cells in metaphase II; CFP-tubulin (β-TUB1) was used to label microtubules (green). The cell volume can be seen by the diffuse green of unincorporated tubulin. A cross-sectional view of the two meiosis II spindles can be seen in a with microtubule bundles and negative, non-labeled spaces. b View of the meiotic spindles in panel A rotated 90° to show the length of the half spindle. c Enlarged cross-sectional view of a meiotic spindle with microtubule bundles (green) and empty spaces. Scale bar: 5 µm. d Immunostained meiosis I cell labeled microtubules (red), chromosomes (blue), and membranes labeled with an ɑ-PDI antibody (green). e Enlarged 3D volume view of the spindle in panel E that shows PDI membranes embedded within the spindle. f Immunostained meiosis II cell labeled microtubules (red), chromosomes (blue), and PDI-labeled membranes (green). g Histogram of membrane channel diameter (binned in 0.25 μm increments) in PDI-immunostained cells (gray) and cross-sectional live cells (black). Scale bars and grid lines: 10 µm (unless otherwise noted).

Membrane channels have been previously identified embedded in mitotic barley spindles [74], and these structures were shown to be derived from the endoplasmic reticulum (ER) [74, 75]. ER membrane channels have been shown to act as both a physical anchor of the spindle, providing structure and load-bearing force [76, 77] as well as calcium reservoirs that release via inositol trisphophate (IP₃) receptors to regulate spindle dynamics [78, 79]. Given the similarity in size, morphology, and spindle location, we suspected our membrane channels were also ER-derived structures. PDI is a protein used for ER identification in animals and plants [80, 81]. Using primary antibodies to PDI, immunofluorescence microscopy revealed the presence of PDI-positive membranes throughout the cytoplasm and embedded within the spindle itself, from the poles to the midzone in both meiosis I (Fig. 6d) and meiosis II (Fig. 6f) spindles. Using 3D volume viewer (Fig. 6e), we confirmed that PDI-positive membranes were embedded within the spindle, rather than on the outer surface (Fig. 6e). We measured the diameter of the PDI-positive membranes and found them to be 0.86 ± 0.3 µm (Fig. 6g). Comparing the diameters of the PDI-positive membrane channels and the cross-sectional empty spaces within the spindle reveals greater variability and a broader distribution in the size (Fig. 6g). This variability in the live-image cross-sectional space is expected given this negative space was measured in live cells that were actively assembling and organizing the spindle. Regardless of the variance, the cross-sectional spaces (average diameter 1.39 ± 0.6 µm) are statistically larger than the PDI-membrane channel (average diameter 0.86 ± 0.3 µm) (p < 0.001, t test) which suggests that these membranes could be located within the spaces. The diameter of the PDI-membrane channels (0.86 ± 0.3 µm) are nearly the same as the EM measured membrane channels (0.79 μm), suggesting that these membranes are ER derived.

Microtubule Organization on the Nuclear Envelope

Before meiotic spindle assembly initiates in plants, microtubules are organized on the surface of the nuclear envelope in a “perinuclear ring” [20, 82, 83]. Live imaging using CFP-tubulin labeled microtubules shows this ring (top cell), which collapses into the nuclear space (bottom cell) upon nuclear envelope breakdown (Fig. 7a, n = 8 cells). Condensed chromosomes localize near the edge of the nuclear periphery (Fig. 7b, n = 8 cells) before nuclear envelope breakdown and establishment of the spindle. Live-imaging resolution is limited, making it difficult to characterize the organization and location of the microtubules relative to the nuclear envelope. Using ET, we imaged meiotic cells before nuclear envelope breakdown (Fig. 7c, d). Three-dimensional modeling revealed localization of chromosomes to the nuclear periphery as seen in live imaging. It also revealed a mesh network of microtubules on the external surface of nuclear envelope organized in a cross-hatch pattern rather than parallel arrays (white box in Fig. 7d, e). Modeling also revealed the extensive presence of microtubules inside of the nucleus (Fig. 7c–e) rather than exclusively on the external surface. Surprisingly, we observed several microtubules that passed directly through the nuclear envelope membrane (white arrows in Fig. 7d, e). These membrane-spanning microtubules are labeled f, g, h in Figure 7d and e, and close-up images of these microtubules are shown in the associated panels (Fig. 7f, g, h). These microtubules do not pass through nuclear pore complexes (orange spheres in Fig. 7c–h) despite their high density in the nuclear envelope; rather they pass directly through the membrane itself. In Fig. 7f and h, the microtubules appear to pass cleanly through the membrane, not through a nuclear pore complex, and with no distortion or torquing of the membrane. The membrane-spanning microtubules in Figure 7g are present in nuclear grooves (blue arrows in Fig. 7d, e, g). Previous studies have suggested that cytoplasmic microtubules can create grooves and invaginations on the nuclear envelope surface, specifically type II grooves which are distinguishable by their deformation of both the inner and outer nuclear membrane [84]. These grooves and the associated microtubules alter chromatin dynamics and organization via the LINC complex [84–86].

Fig. 7.

Nuclear organization in prophase meiotic cells. a Live meiosis II cells with microtubules (green) labeled with CFP-tubulin (β-TUB1). Top cell has an intact nuclear envelope with a network of microtubules; the bottom cell is experiencing nuclear envelope breakdown with microtubules collapsing into nuclear space. b Live meiosis I cell with microtubule network (green) surrounding nuclear envelope with chromosomes (pink) within the nucleus. c Tomographic slice overlaid with modeled chromosome masses (yellow), microtubules (green), nuclear envelope (red), and nuclear pore complexes (orange). d 3D model view of chromosomes (yellow), microtubules (green), nuclear envelope (red), and nuclear pore complexes (orange). White arrows indicate regions where microtubules pass through the nuclear envelope and the blue arrow indicates a nuclear groove. Arrows are labeled f–h, and enlarged views of these microtubles are shown in f–h. White bracket indicates the mesh network of microtubules near the surface of the nuclear envelope. e Enlarged region from e, rotated in 3-dimensional space to highlight the microtubules that pass through the nuclear envelope (white arrows) and the microtubule network (white box). Arrows are labeled f–h, and enlarged views of these microtubles are shown in f–h. f View of membrane-spanning microtubules enlarged from d, e; white arrows indicate two microtubules that pass through the nuclear envelope. g View of membrane-spanning microtubules enlarged from d and e; white arrow indicates a microtubule that pass through the nuclear envelope, and two microtubules sit within a nuclear groove (blue arrow). h View of membrane-spanning microtubules enlarged from d and e; white arrows indicate two microtubules that pass through the nuclear envelope. Scale bars: 10 µm in a, b; 1 µm in c–e; 500 nm in f–h.

Discussion

Spindle structure is critical for its function in the accurate segregation of chromosomes in cell division. The structure of the spindle has been characterized using the fine resolution provided by electron microscopy in protists [32, 87], mitotic [31, 34, 44, 88, 89], and meiotic [90–92] animal cells, as well as mitotic plant cells [75, 93–95], but very few studies have investigated plant meiotic spindles [96, 97]. In this study, we have characterized the ultrastructure of the Z. mays meiotic spindle, including the kinetochore-microtubule interface and membranes structures within the spindle, as well features of the perinuclear microtubules in meiotic prophase.

In all eukaryotic species, spindle microtubules attach to kinetochores to facilitate their segregation in anaphase, but the number of microtubules that attach per kinetochore varies. In the simplest eukaryotic system, the budding yeast S. cerevisiae, a single microtubule attaches to a kinetochore [25]. Fission yeast S. pombe is marginally more complex, with 2–4 kMTs per kinetochore [42]. Metazoans are significantly more complex, with 20 kMTs per human kinetochore [43], 25 kMTs per kinetochore in PTK cells [44], 21–67 kMTs per kinetochore in different grasshopper species [98], and 4–11 kMTs per kinetochore in Drosophila [99]. In plants, the number of kMTs has only been documenting in mitotically dividing cells of the lily Haemanthus with 74–109 kMTs per kinetochore [32]. In our investigation of Z. mays meiotic spindles, we discovered variability in the number of microtubules per kinetochore, with 8–18 kMTs per kinetochore (Fig. 2). Variability in kMT number is common in electron microscopy studies [100] as it represents a snapshot in time during a dynamic process of microtubule attachment. The variability is due to tension on the kinetochore, as previous studies have shown that the number of attached kMTs fluctuates more than 50% during chromosome orientation on the spindle [100]. Our studies showed an average of 10.8 kMTs per maize kinetochore, and also a correlation between kMT number and kinetochore size (Fig. 2e, Kinetochore 3 and Fig. 3c, Cup 3).

Kinetochore size has been characterized in a variety of species. Budding yeast kinetochores are approximately 60–140 nm [101–103]. Electron microscopy of metazoan kinetochores has revealed a disc-like structure approximately 250–560 nm in diameter and 240–340 nm in thickness including its fibrous corona [65, 68, 99, 104, 105]. The plant kinetochore has proven more difficult to characterize due to its lack of electron density, and previous studies have characterized the structure as a “ball in a cup” structure [52, 70]. Using 3D tomography, we were able to measure the dimensions of the chromatin “cup,” providing an indirect approximation of kinetochore size. The chromatin cup is approximately 776 nm in diameter and 316 nm deep (Fig. 3), which is comparable to Drosophila kinetochores (560 nm diameter, 340 nm thick) [99, 104] and larger than mammalian kinetochores (250 nm diameter, 230 nm thick [68, 105, 106].

Investigation of the maize meiotic spindle also revealed the presence of several membrane-bound structures including small vesicles (Fig. 4) and large membrane channels (Fig. 5). The median size and distribution of vesicles were similar to ET studies in Arabidopsis that identified these structures as derived of the Golgi apparatus [61]. In these studies, Golgi vesicles were identified in late anaphase and telophase localizing near the phragmoplast; the presence of these vesicles has not previously been identified in earlier meiotic plant spindles but given the fragmentation and dispersal of the Golgi in the plant meiotic cytoplasm [107], the presence of Golgi vesicles in the metaphase spindle may support the equal distribution. We observed the presence of membrane channels within the maize meiotic spindle (Fig. 5). The diameter of these membrane channels measured by ET corresponded to cross-sectional negative spaces observed within live CFP-tubulin meiotic spindles (Fig. 6a–c). Given previous studies in barley mitotic spindles that showed similar membrane channels derived from the ER [75] and known roles of ER-membrane channels in spindle assembly and regulation [76, 78, 79], we hypothesized that our membrane channels were also ER in origin. Immunostaining with ER-specific protein PDI [80, 81] revealed the presence of PDI membranes that were the same diameter as the membranes measured by tomography (both averaging ∼0.8 µm) (Fig. 6d–g).

Live microscopy and ET images were also obtained in meiotic prophase before spindle assembly. Previous studies have identified a perinuclear ring of microtubules on the nuclear envelope [20, 22, 82, 95]. With high resolution tomography, we find that this ring consists of a cross-hatch lattice structure on the external surface of the nuclear membrane (Fig. 7e) and also internal microtubule structures, including microtubules that pass through the nuclear envelope but not through nuclear pores (Fig. 7e–h). Evidence for these “transnuclear” microtubules has been documented in mammalian cells, resulting from incomplete mitotic spindle breakdown [108, 109]. We observed some, but not all, transnuclear microtubules within type II grooves on the nuclear enveloped, which are invaginations of both the inner and outer membrane [84]. Type II grooves are created by microtubules and other filaments poking the nuclear membrane [85, 110] as well as condensation of meiotic chromatin tethered to the inner membrane [110, 111]. In our live imaging (Fig. 7b) and ET (Fig. 7c, d), we similarly observed chromatin localization near the nuclear periphery near sites of invagination, and in close proximity to microtubules [85, 110]. Internal, external, and transnuclear microtubules all play critical roles in transducing cytoplasmic forces to the nucleus via the LINC protein complex, a network of proteins including the SUN and KASH transmembrane proteins [84]. Transduction of microtubule force helps facilitate nuclear positioning [110], chromatin organization and genomic stability [86, 112], homology search and DNA repair [113, 114], and bouquet formation and synapsis during meiosis [115, 116]. The LINC complex proteins have been identified in maize [117] and our studies confirm the presence of transnuclear microtubules and type II nuclear grooves, suggesting these force transduction pathways are similarly utilized in Z. mays meiosis. Understanding the structure and function of the Z. mays spindle is critical given its importance as both a cytogenetic model organism and an agricultural crop. Increasing our understanding of plant meiotic structures and processes is critical for improving breeding programs [118, 119] and new strategies for increased food production [120, 121].

Acknowledgments

The authors thank R. Kelly Dawe, University of Georgia, for his support in experimental design and discussion. We thank Beth Richardson at the Electron Microscopy Lab, University of Georgia, for her help with the Z. mays meiocyte sample preparation. The authors also thank the Colorado University Boulder Electron Microscopy Core Facility for support with imaging and data analysis. The authors also thank Hamilton College Biology Department and the Dean of Faculty for their support through the Summer Science Research Fellowship funded by the Stephen Harper Kirner Chair.

Statement of Ethics

An ethics statement was not required for this study type since no human or animal subjects were used.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This research was funded by the National Science Foundation, Grant Nos. MCB-1925546 and IOS-1841696. The funder had no role in the design, data collection, data analysis, and reporting of this study.

Author Contributions

N.J.N. conceived and designed experiments. S.M., J.K.C., N.A., E.T.O., and N.J.N. performed the experiments. S.M., J.K.C., B.P., D.M.O., J.K., E.T.O., and N.J.N. analyzed the data. S.M., J.K.C., B.P., D.M.O., J.K., N.A., E.T.O., and N.J.N. drafted and revised the manuscript.

Funding Statement

This research was funded by the National Science Foundation, Grant Nos. MCB-1925546 and IOS-1841696. The funder had no role in the design, data collection, data analysis, and reporting of this study.

Data Availability Statement

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.