Nuclear actin cable with multiple branches during yeast meiosis

Actin polymerizes to form filaments/cables for motility, transport, and structural framework in a cell. Recent studies show that actin polymers are present not only in cytoplasm, but also in nuclei of vertebrate cells, and their formation is induced in response to stress. Here, by electron microscopic observation with rapid freezing and high-pressure freezing, we found a unique polymerized form of actin inside of nuclei of budding yeast cells undergoing meiosis. The nuclear actin cable during meiosis consists of several actin filaments with a rectangular lattice arrangement and is associated with multiple branched cables/filaments showing “feather-like” appearance. The cable is immuno-labeled with anti-actin antibody and sensitive to an actin-depolymerizing drug. Like cytoplasmic actin cables, nuclear actin cables are rarely seen in pre-meiotic cells and spores, and are induced during meiotic prophase-I. We speculate that nuclear actin cables play a role in nuclear events during meiotic prophase I.


Introduction
In the cytoplasm, as a cytoskeletal protein, actin polymerizes to form a filament (F-actin) for various cellular functions such as motility, division, phagocytosis, endocytosis, and membrane trafficking 1 . Dynamics of cytoplasmic actin filaments are highly regulated by various factors in different environments. Actin is also present in nuclei 2 . Actin monomer functions as a component of several chromatin-remodeling complexes for transcription 3 .
Recent studies showed that polymerized forms of actin are present in nuclei of various types of vertebrate and invertebrate cells [4][5][6][7][8][9] . About forty-years ago, Fukui and his colleagues identified actin bundles in Dictyostelium and HeLa cells upon the treatment with dimethyl sulfoxide [10][11][12] . In Xenopus oocytes, nuclear actin forms a mesh of filaments, which is involved in the protection of nucleoli from gravity-induced aggregation 13 . In starfish oocytes, actin filaments promote the breakdown of nuclear envelope and, by forming a mesh, the capture of chromosomes by spindles in cell division 14 . In mouse oocytes, actin filaments promote chromosome segregation during meiosis I and II 5 . Somatic mammalian cells induce the formation of actin polymers transiently in a nucleus in response to stress; serum starvation, heat shock, and DNA damage such as DNA double-strand breaks (DSBs). For serum starvation, the F-actin is involved in transcription by helping the activity of a transcriptional cofactor, MRTF (myocardin-related transcription factor) 15,16 . Nuclear F-actin also promotes repair of DSBs in mammalian and fruit fly cells 4,7,17 .
In budding and fission yeasts, actin is present in the cytoplasm as a polymerized form such as rings, patches and cables [18][19][20] as well as less-defined structures called filasome 21 . In budding yeast Saccharomyces cerevisiae, actin cables in cytoplasm play a role in the transport in mitotic budded cells. Previous electron microscopic (EM) analysis of cytoplasm in fixed mitotic cells revealed a linear actin bundle containing multiple actin filaments 22 . Actin cables/bundles and some other actin-filament associated structures are also visualized by using a green fluorescent protein (GFP) fused with an actin-binding protein, Abp140, or by staining with an actin-specific peptide with a fluorescent dye such as phalloidin 23 .
Actin filaments/cables are present also in of the cytoplasm of meiotic yeast cells 24,25 . The actin cables are induced to form during meiotic prophase-I (meiotic G2 phase) and form a network that surrounds the nucleus. It is proposed that meiosis-specific motion of telomeres on the nuclear envelope is mediated by "Piggy-Back" mechanism, in which cytoplasmic actin cables attached to NE drives the motion of telomeres, thus chromosomes through protein ensembles embedded in NE 24,26 . Previous EM observation of a nucleus of chemically fixed cells undergoing meiosis showed synaptonemal complex (SC), a meiosis-specific chromosome structure as well as nuclear microtubules from Spindle pole body (SPB), a yeast centrosome [27][28][29] . However, it still remains less known about ultra-structure such as cytoskeletons including actin cables inside of yeast cells during meiosis.
In this study, we analyzed ultra-structures inside meiotic yeast cells, particularly an actin-associated structure called actin cables, by using freeze-substitution electron microscope 30 . As expected, meiotic prophase-I cells induce the formation of cytoplasmic actin cables. Interestingly, we also detected actin cables inside nuclei in cells undergoing meiosis I, but not in pre-meiotic cells or spores. The structure of nuclear actin cables is similar to that in the cytoplasm of meiotic cells. The cables are immuno-labeled with anti-actin antibody and are sensitive to the treatment with an actin-depolymerizing drug.
Meiosis-specific nuclear actin cables consist of multiple actin filaments with a regular arrangement and show multiple branches. These results indicate that actin cables or actin cable-networks are formed inside nuclei of cells undergoing chromosomal events in prophase-I; e.g. recombination, chromosome motion and the SC formation. The biological implications of nuclear actin cables in meiotic cells are discussed.

Electron microscopic observation of meiotic yeast cells
Previous analysis of actin localization in budding yeast by staining with phalloidin and live-cell imaging using a GFP-fusion of Abp140 revealed that, during prophase of meiosis I, yeast cells contain actin cables in cytoplasm 2,25,26 . To get more detailed ultra-structures of cytoskeletons and its spatial relationship with organelles inside of meiotic yeast cells, we used a transmission electron microscope (TEM). Cells were quickly frozen, and substituted with the fixative and stained with osmium (freeze-substitution method) 30 . Thin sections of cells were observed under TEM (Fig. 1). With the freeze-substitution method, cellular organelles including nucleus, mitochondrion, and vacuoles in cytoplasm filled with dense-stained ribosomes were well preserved (Fig. 1a, c, f, j). The nucleus was surrounded with double-layered nuclear membranes and contained electron-dense regions, which corresponds with nucleolus (Fig. 1b). During meiosis prophase-I, i.e. 4 h after the induction of meiosis by incubating cells with sporulation medium, nuclei contact vacuoles forming nuclear-vacuole junctions (NVJ; Fig. 1f) as shown previously 31 . At 8 h, four pre-spore cells were formed inside of the cells (Fig. 1j).

Previous EM analyses of mitotic cells of both budding and fission yeasts have
shown three distinct actin sub-cellular structures; rings, cables, and patches in addition to filasome and amorphous structure [18][19][20][21] . We could detect a cable structure in cytoplasm during meiotic prophase-I (Fig. 1f, g, i). This cable contains actin since immuno-gold labeling using anti-actin antibody showed that gold particles were seen on the cables in cytoplasm (Fig. 2a). Thus, we called the structure as "actin cable" hereafter. Previous EM analysis detected similar cytoplasmic actin cables in mitotic yeast cells 22 and later immuno-EM confirmed the presence of actin cables 32 . We also detected gold particles on less electron dense areas in cytoplasm, which might correspond to filasomes (see also Fig. 2a'), a novel actin-containing membrane-less sub-cellular structure in cytoplasm originally found in fission yeast 21 . The filasome may differ from actin patches involved in endocytosis. Consistent with previous cytological studies of actin 24,25 , actin cables in cytoplasm are rarely seen in pre-meiotic cells at 0 h (in G1 or G0 phases; Fig. 1a and 3a for quantification) and 2 h (in S-phase, Fig.1c), but are observed in cells at 4 h after the induction of meiosis ( Fig. 1f and 3a). The formation of the cable in cytoplasm peaks at 5 h, which corresponds with late prophase-I and gradually decreases during further progression of meiotic divisions (Fig. 3a). In some cases, cytoplasmic actin cables are located near mitochondria (Fig.1i). Among the sections, we rarely saw the attachment of cytoplasmic actin cables to the nucleus, although a previous cytological study suggested that cytoplasmic actin cables attach to the NE 24 .
In addition to cytoplasmic actin cables, we detected the similar cables inside of nuclei of meiotic cells (Fig. 1c, d, e, f, h). The structure of cables in nuclei is similar to that in cytoplasm (compare Fig. 1g and 1h). This cable is structurally different from microtubules in the nucleus emanating from Spindle Pole body (SPB: see Fig. 6). Importantly, immuno-gold labeling using anti-actin antibody showed gold labels on these cables in nuclei as those in cytoplasm (Fig.   2b), indicating that the nuclear cable contains actin. Thus, we referred a cable containing actin in nuclei to as "nuclear actin cable".

Nuclear actin cables are formed during prophase-I
We checked the presence of nuclear actin cables in different stages of meiosis ( Fig. 3a). As a control, we measured nuclei containing microtubules. The nuclear microtubules are seen at 0 h and sections positive to microtubules are increased slightly during meiotic prophase-I. Nuclear actin cables appear earlier during meiosis than cytoplasmic cables (Fig. 3a). At 0 h before the induction of meiosis in which most of cells are G1, we detected few cables in nuclei ( Fig. 1a; 0/10, 1/10 and 1/34 cells in three independent time courses) as well as in cytoplasm (1/10, 1/22 and 1/34 cells). Few cables were also found in four nuclei of pre-spore cells, suggesting that the cables disassemble by the formation of pre-spore cells (Fig. 1j). Kinetic analysis revealed that nuclear actin cables were seen at 2 h post-induction of meiosis (Fig. 1c,  between two nuclei undergoing anaphase-I (Fig. 4b). This suggests that the actin cables passed into the daughter nuclei during meiosis I division.

Nuclear actin cables form branches
Both actin cables in cytoplasm and nuclei are structurally similar. The cables contain three to ten thin parallel filaments and exhibit multiple branched filaments, which look like "feather" (Fig. 1g, h). Cross-sections of the nuclear cables often reveal a regular rectangular/square arrangement (lattice) of filaments with repeated units of alternate single and double filaments ( Fig. 1e; shown as red dots in Fig.1e'). From a long cable, additional cables/filaments seem to form a branch ( Fig. 1e, g, h). At this resolution some branched filaments look attached directly to filaments in a main cable while the other filaments are not.
The diameter of the thin filament in nuclear actin cables is around 7-8 nm To get more spatial information on nuclear actin cables, we checked serial sections of a nucleus in yeast cells at 4 h in meiosis ( Fig. 6 and 7). To achieve in-depth freezing in samples, we froze cells under a high pressure 36 . With high-pressure freezing specimens suitable for sectioning were obtained. Importantly, we did not see any change of sub-cellular structures including actin cables in yeast cells prepared by rapid freezing or high-pressure freezing (compare Fig. 1 with Fig. 6 and 7). We often detected multiple "feather-like actin structures" in different sections (Fig. 6), indicating that actin cables are an abundant nuclear structure. In some sections, a long cable that spans entire nucleus is observed (Fig. 7-5 and 7-6), and the end of the cable is likely attached to the nuclear envelope ( Fig. 7; arrows in Fig. 7-6).

Nuclear actin cables by chemical fixation
Yeast cells are surrounded with thick cell walls, which impair penetration of staining reagents such as osmic tetroxide. We also stained spheroplasted meiotic yeast cells with osmic acid after fixation with glutaraldehyde (without any freezing, Supplementary Fig. 2) and, in some cases, prepared serial sections for EM observation (Supplementary Fig. 3). With this procedure, the mitochondria are contrasted highly ( Supplementary Fig. 2). Membrane structures such as nuclear membrane were partially deformed, possibly due to hypo-osmotic conditions. Importantly, even under this condition, we could detect actin cables in both the nucleus and the cytoplasm of cells during meiotic prophase-I ( Supplementary Fig. 2d, e, g). Observation of serial sections revealed that actin cables attach to the inner nuclear membrane (Supplementary Fig. 3).

Discussion
In this study, by using TEM with rapid freezing-fixation, we found nuclear actin cables in budding yeast cells undergoing the physiological program of meiosis.
We could also detect actin cables in cytoplasm, which are induced during prophase I 24 . Since we used rapid freezing to preserve structures inside of cells, it is unlikely that the actin cable is an artifact produced by sample preparation, which might be induced by external stress and/or fixation. Moreover, we also detect the actin cables in nuclei fixed with chemicals without freezing (Supplementary Fig. 2 and 3).
Both nuclear and cytoplasmic actin cables consist of multiple parallel filaments. This is consistent with previous EM analyses showing cytoplasmic actin cables in fixed mitotic cells, which are present as a linear cable of multiple filaments 22 . On the other hand, the actin cables formed during meiosis appear to have a unique ultra-structure: a single actin cable exhibits multiple lateral branches of filaments (Fig. 8). Some branched filaments are directly attached to filaments in main cables while the other filaments are not (Fig. 8a). This branch structure is different from a branch of actin filament from a single actin filament, which is seen in cytoplasm in moving edges of various vertebrate cells and is mediated by the Arp2/3 complex. In yeast, branched "short" actin filaments (not cables) are often associated with mitotic actin patches that mediate endocytosis 20 . More recently, it is shown that, in mammalian cells, Arp2/3 controls the formation of nuclear F-actin in response to DSBs 7 . The angle between the Arp2/3-dependent branched filaments is ~70° 35 . On the other hand, angles in the meiotic nuclear actin cables are ~35°. Moreover, meiotic nuclear actin cables do not show a single branch of the filament, but rather showed a branch of the cables with multiple filaments, suggesting a unique branching mechanism for nuclear actin cables in the yeast. Therefore, the actin cables with multiple branches in yeast meiotic nuclei are a unique polymerized form of actin (Fig. 8c).
Nuclear (and cytoplasmic) actin cables seem to contain a unique arrangement of actin filaments, in which alternate pattern of 1 and 2 filaments is observed in a cross section of the cables (Fig. 8a and b). This alternate pattern provides rectangular/square arrangement of filaments in a single cable. The distance of inter-filaments is mainly 10-15 nm. In vitro biochemical reconstitution of actin cables with an actin-bundling protein, fimbrin, showed packing of the filaments with 12 nm inter-filament distance [37][38][39] . Thus, the formation of actin cables in meiotic nuclei might be mediated by a fimbrin-like actin-binding protein.
Budding yeast contain multiple actin-bundling proteins; Sac6/fimbrin, Scp1, Ipg1, Crn1 and Abp140 20 . Among them, fimbrin/Sac6 is a major protein for actin bundling, which plays a critical role in bundling the filaments in cytoplasmic actin cables in mitotic cells. In vitro reconstitution of actin bundles in the presence of human fimbrin shows an arrangement of actin filaments in a hexagonal lattice 39 .
This arrangement is similar to in vivo arrangement in terms of a unit of equilateral triangle. However, in meiotic nuclear actin cables, rather hexagonal, tetragonal lattice is observed in which two sides are required for the self-assembly of filaments in the lattice and the others are for branch formation.
In muscle, actin filaments exhibit tetragonal lattice in the Z-band of adjacent sarcomeres 40,41 . In this case, α-actinin seems to mediate inter-filament interaction. It has been reported that fission yeast, but not budding yeast, contain α-actinin-like protein 42,43 . It is interesting to find an α-actinin-like protein in budding yeast.
The average length of nuclear actin cables is 300-400 nm, which is slightly shorter than the cables which form a contracting ring in fission yeast 18, 44 . Nuclear actin cables are able to elongate up to 1 µm (Fig. 3b and Fig. 7). This long cable consists of a few cables in a linear array, rather than a single cable.
We speculate that some branched actin cables are assembled into a long cable.
Several actin cables are present in a single nucleus of meiotic cells (Fig. 6 and 7), indicating that nuclear actin cables are abundant. Nuclear actin cables are induced from very early meiotic prophase-I such as 2-h post induction of meiosis and are present to at least by meiosis I nuclear division. Actin cables are abundant not only in nuclei, but also in cytoplasm particularly during late prophase-I (Fig. 4a). On the other hand, we rarely see nuclear or cytoplasmic actin cables in a G1 cell (pre-meiotic) or in spores (Fig. 1a, i).
Recent studies in mammalian cultured cells demonstrate the presence of a polymerized form actin in nuclei, which can be detected with a fluorescence probe such as Life-ACT 17 or directly by EM 12 . The nuclear F-actin formation is induced in response to stress such as serum starvation, heat, and DNA damage 8 . The nuclear F-actin is involved in the gene expression, the relocation of chromosomal loci as well as DNA repair 4,7,17 . Like mammalian cells, meiotic actin cable formation in meiotic prophase-I yeast cells may appear in response to DNA damage, since meiotic cells are programed to induce multiple DNA DSBs 45 . However, during meiosis, the formation of nuclear actin cables starts at 2 h post-induction of meiosis, prior to DSB formation, which begins at 3 h, suggesting that nuclear actin cable formation might not be associated with DSB formation during yeast meiosis. Therefore, we need more analysis to know relationship between nuclear actin cables and meiotic DSB formation/repair.

Competing interests
The authors declare no competing interests.
Yeast cell culture and time-course analyses of the events during meiosis and the cell cycle progression were performed as described previously 47,48 .
Briefly, 1 ml of diploid yeast culture in YPAD was cultured in 200 ml of SPS media for 16 h. Cells were collected and, after washing with H 2 O, were re-suspended in 200 ml of SPM media (20 mM KCH 3 COO, 0.02% raffinose) to start meiosis.

Rapid Freezing for transmission electron microscopy
Specimens for freeze-substitution electron microscopy were prepared according to previously described method 49,50 , with slight modifications. Cells were harvested by centrifugation. The cell pellets were sandwiched between two copper disks (3 mm in diameter). Specimens were quickly frozen with liquid propane using a rapid freezing device (KF80; Leica, Vienna, Austria).

High pressure freezing fixation for transmission electron microscopy
The specimens for high-pressure freeze-substitution electron microscopy were prepared according to previously described method 51 with slight modifications.
The pelleted cells were pipetted into aluminum specimen carriers (Leica) and frozen in a high-pressure freezing machine HPM-010 (BAL-TEC, Liechtenstein).
The cells were transferred to 2% OsO 4 in cold absolute acetone. Substitution fixation was carried out at −90°C for over 80 h. After the fixation, the specimens were warmed gradually (at -40°C for 2 h, at -20°C for 2 h, at 4°C for 2 h and at room temperature for 1 h) and washed with absolute acetone and then with exchange to 0.1% uranyl acetate in absolute acetone. After the staining, samples were washed with absolute acetone and rinsed with QY-2. Substitution and embedding were described in above.

EM grid preparation and observation
Specimens for morphological observation were sectioned using a ULTRACUT-S ultramicrotome (Reichert-Nissei, Tokyo, Japan). Ultrathin sections were cut with thickness of 50-60 nm and mounted on copper grids. Specimens for immuno-staining were mounted on nickel grids. Ultrathin sections were stained with 4% uranyl acetate for 12 min in dark at room temperature and citrate mixture (SIGMA-ALDRICH) for 2 min at room temperature, and examined using JEM-1200EXS at 120kV or JEM-1400 transmission electron microscope (JEOL, Tokyo, Japan ) at 100 kV.

Immunoelectron Microscopy and Immunolabelling
For immuno-electron microscopy, chemical fixation samples shown below were etched with 1% H 2 O 2 for 5min at room temperature. Otherwise, cell pellets were sandwiched between two aluminum disks (diameter 3 mm). And specimens were quickly frozen with liquid propane. They were freeze-substituted in cold absolute acetone containing 0.01% OsO 4 . Substitution took place for 48-72 h at -80°C, and were then warmed gradually (at -40°C for 4 h, at -20°C for 2h and at 4°C for 1h) and washed with dehydration ethanol at 4°C. Then the cells were washed twice with cold ethanol and substituted at 4°C and infiltration was done in the LR white and ethanol mixture (10,30,50,70,80, 90, 100% pure resin).
Specimens were embedded in pure resin and polymerized at 50°C for 1-2 days.
Immunolabelling was carried out by previously described method 51

Latrunculin B treatment
After cultured in SPM medium for four hours, cells were treated with 30 µM Latrunculin B (R&D Systems) dissolved in 0.1% DMSO at 30°C for 1 hour. After the post-treatment, cells were collected and fixed by the rapid freezing mentioned later. µM phalloidin (Sigma-Aldrich) for 1 min, these treatments were performed as described previously 18,44 .

Chemical fixation method for electron microscopy
Specimens for TEM were prepared as described previously 21,44,52,53 .

Image processing and measurement
Measurement of diameter of actin cables and distance between actin cables was carried out using ImageJ (NCBI, https://imagej.nih.gov/ij/).

Statistics
Statistical significance for Length of actin cables between that in nucleus and cytoplasm was analyzed using the Mann-Whitney's U-test. The null hypothesis was that there exists no variation of the length between nucleus and cytoplasm.
Statistical significance for existence of actin cables in cells with and without LatB treatment was analyzed using Fisher's exact test. The null hypothesis was that LatB treatment did not affect existence of actin cables in cells. Two-sided P-value was shown.