- © 2019 American Society of Plant Biologists. All Rights Reserved.
Abstract
During meiosis, the stepwise release of sister chromatid cohesion is crucial for the equal distribution of genetic material to daughter cells, enabling generation of fertile gametophytes. However, the molecular mechanism that protects centromeric cohesion from release at meiosis I is unclear in Arabidopsis (Arabidopsis thaliana). Here, we report that the protein phosphatase 2A regulatory subunits B'α and B'β participate in the control of sister chromatid separation. The double mutant b'αβ exhibited severe male and female sterility, caused by the lack of a nucleus or presence of an abnormal nucleus in mature microspores and embryo sacs. 4′,6-Diamidino-2-phenylindole staining revealed unequal amounts of DNA in the mononuclear microspores. Transverse sections of the anthers revealed unevenly sized tetrads with or without a nucleus, suggesting a defect in meiocyte meiosis. An analysis of chromosome spreads showed that the sister chromatids separated prematurely at anaphase I in b'αβ. Immunoblotting showed that AtRECOMBINATION DEFECTIVE8 (AtREC8), a key member of the cohesin complex, was hyperphosphorylated in b'αβ anthers and pistils during meiosis but hypophosphorylated in the wild type. Furthermore, yeast two-hybrid and bimolecular fluorescence complementation assays showed that B'α and B'β interact specifically with AtREC8, AtSHUGOSHIN1 (AtSGO1), AtSGO2, and PATRONUS1. Given that B'α was reported to localize to the centromere in meiotic cells, we propose that protein phosphatase 2A B'α and B'β are recruited by AtSGO1/2 and PATRONUS1 to dephosphorylate AtREC8 at the site of centromere cohesion to shield it from cleavage until anaphase II, contributing to the balanced separation of sister chromatids at meiosis.
Meiosis generates haploid male and female gametophytes, which are essential for sexual reproduction in diploid eukaryotes. Through two consecutive rounds of chromosome division following one round of DNA duplication, the genetic material in a mother cell is distributed to four daughter cells, with a halved number of chromosomes in each (Page and Hawley, 2003; Petronczki et al., 2003); this reduction is a prerequisite for producing healthy progeny. To guarantee the balanced separation of homologous chromosomes at meiosis I and sister chromatids at meiosis II, the cohesion that holds two chromatids together must be released in a stepwise fashion (Page and Hawley, 2003; Petronczki et al., 2003); specifically, chromosome arm cohesion is dissolved at anaphase I, preserved at the kinetochore until metaphase II, and then completely dissociated at the onset of anaphase II (Rieder and Cole, 1999; Uhlmann, 2001; Ishiguro and Watanabe, 2007). Premature dissolution of centromeric sister chromatid cohesion can cause aneuploidy and eventually result in tumorigenesis, birth defects, or sterility (Holland and Cleveland, 2009).
Among eukaryotes, three conserved protein subunits act as the core components of the meiotic cohesin complex that ensures the cohesion of sister chromatids. In yeast, two structural maintenance of chromosome (SMC) proteins, SMC1 and SMC3, and one α-kleisin, RECOMBINATION DEFECTIVE8 (REC8), form a ring-like structure, with REC8 acting to close the ring (Michaelis et al., 1997; Marston, 2014). In mammals, the structural proteins SMC3 and SMC1(β; Garcia-Cruz et al., 2010) create a ring, whereas the meiosis-specific protein REC8 (Bannister et al., 2004; Golubovskaya et al., 2006) or its homologs RAD21L (Lee and Hirano, 2011) and RAD21/SCC1 (Xu et al., 2004) close the ring. Similarly, in Arabidopsis (Arabidopsis thaliana), AtSMC1, AtSMC3 (Liu Cm et al., 2002; Lam et al., 2005), and AtREC8/SYNAPTIC1 (SYN1)/DETERMINATE INFERTILE (DIF1; Bai et al., 1999; Bhatt et al., 1999; Cai et al., 2003; Chelysheva et al., 2005) were found to form a cohesin complex, whereas the structure of the cohesin has not been determined.
In addition, eukaryotes have evolved an elaborate molecular machine, consisting of cohesin-associated proteins, to control the loading and release of cohesion during meiosis. The release of cohesion depends on the phosphorylation status of REC8. Phosphorylation (Brar et al., 2006; Katis et al., 2010; Attner et al., 2013) triggers REC8 cleavage by separase (Katis et al., 2010), leading to ring opening and cohesion loss, whereas dephosphorylation by protein phosphatases shields REC8 from cleavage. The cohesin-associated proteins that protect centromeric cohesion from removal are conserved among eukaryotes. In yeast, the centromeric cohesion-localized protein Shugoshin/MEI-S332 (SGO) is responsible for recruiting a Ser/Thr protein phosphatase 2A (PP2A) B' regulatory subunit, ROX three suppressor 1, which mediates REC8 dephosphorylation and maintains centromeric cohesion until metaphase II (Gregan et al., 2005; Kitajima et al., 2006; Riedel et al., 2006; Clift et al., 2009). In mammals, a meiosis-specific protein, Shugoshin-2 (Llano et al., 2008), recruits PP2A containing a B' subunit (Mailhes et al., 2003; Xu et al., 2009) to dephosphorylate REC8, RAD21L, or RAD21. In Arabidopsis, AtSGO1 mainly protects centromeric cohesion at anaphase I, whereas AtSGO2 plays a minor role (Cromer et al., 2013; Zamariola et al., 2014b). They likely fulfill this role by recruiting a specific but currently unknown protein phosphatase to target AtREC8. Furthermore, a putative anaphase promoting complex/cyclosome (APC/C) inhibitor, PATRONUS (PANS1), protects centromere cohesion during interkinesis (Cromer et al., 2013; Zamariola et al., 2014b). However, whether it recruits a phosphatase to centromeres is unclear.
The PP2A holoenzyme is a heterotrimer consisting of a scaffolding A subunit, regulatory B subunit, and catalytic C subunit. The B subunit determines the substrate specificity and subcellular localization of the PP2A holoenzyme (Shi, 2009; Slupe et al., 2011). Among the three families of B subunits in Arabidopsis, the B' subfamily has nine members. They all contain a conserved B56 domain, like their yeast and mammalian orthologs. Some B' subunits function in development and responses to environmental stimuli. B'γ is involved in the regulation of flowering time (Heidari et al., 2013), whereas B'η and B'ξ play roles in innate immune responses (Segonzac et al., 2014). All B' subunits except B'ε reportedly participate in the regulation of brassinosteroid (BR) signaling (Tang et al., 2011; Wang et al., 2016). Whether and which B' subunits mediate AtREC8 dephosphorylation and protect centromeric cohesion during meiosis are unknown.
Here, we report that B'α and B'β are required to dephosphorylate AtREC8 and protect centromeric cohesion from removal at meiosis I. A lack of B'α and B'β caused premature sister chromatid separation, resulting in male and female sterility. The interaction of B'α and B'β with AtSGO1, AtSGO2, and PANS1 may mean that they are recruited by these cohesion-associated proteins.
RESULTS
The pp2ab'αβ Double Mutant Displays Male and Female Sterility
The pp2ab'αβ (b'αβ) double mutant segregated from the pp2ab'α−/−β+/− parent does not show a typical BR-deficient phenotype in vegetative growth (Supplemental Fig. S1A), implying that the effect of B'α and B'β on BR signaling is limited or probably because of the presence of a redundant gene. However, B'α and B'β are essential in reproductive growth in Arabidopsis. The b'αβ double mutant displays severe sterility and reduced seed setting (Supplemental Fig. S1, A and B; Jonassen et al., 2011). In comparison, as in wild-type Columbia (Col-0), normal seed setting occurred in the single mutants b'α (SALK_149059, transfer-DNA [T-DNA] in first exon) and b'β (SALK_103167, T-DNA in intron) and in two complementary lines, proα:α-YFP (yellow fluorescence protein)/b'αβ and proβ:β-YFP/b'αβ, in which exogenous B'β replaced endogenous B'β (Supplemental Fig. S1, C–F). These results indicate the functional redundancy of B'α and B'β.
We next performed a reciprocal cross between Col-0 and b'αβ plants to look for evidence of male or female sterility. Pollinating Col-0 pistils with pollen of b'αβ in excess improved the seed setting rate greatly, but the number of seeds was still less than that following Col-0 self-pollination (Fig. 1, A, B, and E). In comparison, pollinating b'αβ pistils with Col-0 pollen resulted in limited seed setting, similar to b'αβ self-pollination (Fig. 1, C–E). The results indicate partial sterility in b'αβ male gametophytes and severe sterility in b'αβ female gametophytes.
Reciprocal crosses reveal male and female sterility in b'αβ double mutant plants. A–D, Seed setting and silique size from the indicated crossed plants. The silique images in (A–D) were digitally made into a composite for comparison. Bar = 2 mm. E, Statistical analysis of the seed number per silique from four types of crossed plants. The average number is at the top of the column. Error bars show the means ± sd (sd), n = 30.
The transmission efficiency of b'αβ males and females was tested to determine whether gametophyte or sporophyte defects cause the observed sterility. As shown in Supplemental Fig. S2, a−β− males and females segregated from a+/−β−/− or a−/−β+/− parents produced nearly the same number of progeny as a+β− and a−β+ did, meaning that in heterozygous mother tissue, a−β−microspores and functional megaspores were able to develop into mature pollen grains and embryo sacs. Therefore, the male and female sterility observed in a-β- plants was because of a sporophyte defect.
The Pollen Grains and Embryo Sac of the b'αβ Mutant Exhibited Nuclear Defects and Low Viability
Pollen defects were examined by I2-KI and Alexander’s staining assays. When compared with the well-stained, uniform pollen grains observed in Col-0 (Fig. 2, A and C), b'αβ pollen grains were unequal in size and unevenly stained by I2-KI (Fig. 2B) and poorly stained with Alexander’s stain (Fig. 2D), indicating reduced starch accumulation and low viability. Staining with 4′,6-diamidino-2-phenylindole (DAPI) revealed three nuclei in all Col-0 pollen grains, including one large vegetative nucleus and two small sperm nuclei (Fig. 2, E and F); however, many b'αβ pollen grains lacked DAPI staining regardless of whether their shape and size were normal or abnormal (Fig. 2, G and H), indicating the absence of a nucleus and pollen degeneration probably caused by abnormal chromosomes.
Tests for pollen viability and a nucleus in the b'αβ double mutant. A–D, Pollen viability assays in Col-0 and b'αβ. A and B, I2-KI staining. C and D, Alexander’s staining. Arrows indicate sterile pollen grains. E–H, Pollen nucleus assay by DAPI staining. Circle, small pollen; triangle, pollen without a nucleus; star, pollen with a nucleus. Bars = 100 µm in (A–D) and 20 µm in (E–H).
An examination of mature embryo sacs from the b'αβ mutant revealed similar nuclear defects to those observed in the pollen grains. In contrast with the five-nucleus and four-celled (two synergids, one egg, and one central cell) mature embryo sac in the wild type before fertilization (Fig. 3A), most b'αβ embryo sacs were empty without a nucleus or cell (Fig. 3, B and E). A few embryo sacs contained one or two nuclei (Fig. 3, C and E), but only two or three embryo sacs per silique contained normal nuclei or cells (Fig. 3, D and E).
The b'αβ embryo sac is empty or has an abnormal nucleus. A–D, Three representative embryo sacs from b'αβ mutants (B–D), showing nuclear defects compared with wild-type Col-0 embryo sacs (A) by differential interference contrast (DIC) microscopy. E, Statistical analysis of the embryo sacs with normal and abnormal nuclei shown in (A–D). Error bars show the means ± sd; n > 30 ovules from 10 siliques. F–M, The egg cell marker DD45:GFP in Col-0 and b'αβ embryo sacs. The fluorescent signal (marked by an arrow) indicates an egg cell nucleus. N–S, The central cell marker DD36:GFP in Col-0 and b'αβ embryo sacs. The fluorescent signal (marked by an arrow) indicates a central cell nucleus. The merged images in (G), (I), (K), (M), (O), (Q), and (S) indicate the position of the embryo sac in the ovule. Ccn, central cell nucleus; Ec, egg cell; EcL, egg cell-like; Sc, synergid cell; ScL, synergid cell-like. Bars = 20 μm in (A–D) and 10 μm in (F–S).
The nuclear defects in b'αβ embryo sacs were confirmed using the egg cell marker DD45:GFP and central cell marker DD36:GFP (Steffen et al., 2007). The DD45:GFP signal was concentrated in the egg cell in the wild type (Fig. 3, F and G, arrows); in contrast, it was absent in most b'αβ embryo sacs (Fig. 3, H and I), and abnormally smeared (Fig. 3, J and K, arrows) or normally concentrated (Fig. 3, L and M, arrows) in the egg cells of a small number of b'αβ embryo sacs. Similarly, the DD36:GFP signal was concentrated in the central cell nucleus (Fig. 3, N and O, arrow) and endosperm nucleus 24 h after pollination in the wild type (Supplemental Fig. S3, A and B, arrowheads) but absent in most b'αβ embryo sacs before (Fig. 3, P and Q) and after pollination (Supplemental Fig. S3, C and D). Only a few b'αβ embryo sacs exhibited a GFP signal (Fig. 3, R and S, arrows), which was smeared in the endosperm at 24 h after pollination (Supplemental Fig. S3, E and F, arrowheads).
These data suggest that nucleus formation in the male and female gametophytes was impaired in the b'αβ mutant, resulting in male and female sterility.
The Microspores and Tetrads Produced in b'αβ Anthers Are Abnormal
To determine when the nuclear defects in the gametophytes occurred, mononuclear microspores at the onset of their release from tetrads in early anther stage 8 were examined. In the wild type, the microspores were spherical and contained one nucleus with concentrated DAPI staining in the center (Fig. 4, A and B). In comparison, the b'αβ microspores were diverse in size (Fig. 4, C and D), with some that were larger but most that were smaller than those of the wild type (Fig. 4E). Most b'αβ microspores contained an abnormal nucleus with inconsistent DAPI staining (e.g. peripheral, smeared, split, very little, or intense fluorescence). None of the small microspores exhibited nuclear DNA stained by DAPI; thus, we conclude that they had no nuclei. Only a few of the normal-sized microspores contained a nucleus, with normal DAPI staining like that in the wild type (Fig. 4F). The irregular nuclear morphology we observed suggests that unequal amounts of DNA were delivered to the haploid microspores before their release from tetrads in b'αβ.
Comparison of Col-0 and b'αβ mononuclear microspores. A–D, DAPI staining of mononuclear microspores at the onset of their release from tetrads at stage (St) 8. E, The diameters of the mononuclear microspores shown in Col-0 (B) and b'αβ (D); n = 60 from three individual anthers. Error bars show the means ± sd. F, Statistical analysis of a microspore with a normal nucleus and a microspore with an abnormal nucleus in b'αβ and Col-0. To the right are the standards for normal, various abnormal, and anuclear microspores; n = 150 microspores from three individual anthers. Error bars show the means ± sd. Bars = 20 μm in (A–D) and 10 μm in (F).
Next, tetrad and microspore formation in b'αβ anthers at stages 5–8 was investigated using transverse sections. Up to stage 5, no obvious differences between the Col-0 and b'αβ anthers were found. The microspore mother cells, with a large nucleolus in the nucleus, were surrounded by tapetal cells in b'αβ anthers, as in the wild type (Fig. 5, A1 and A2). At stage 6, the nucleolus almost disappeared, marking the entry of microspore mother cells into meiosis. The wild-type meiocyte was large and enclosed by a thick cell wall. In comparison, the b'αβ meiocyte looked smaller and was enclosed by a thinner cell wall (Fig. 5, B1 and B2). At stage 7, the wild-type microspores in tetrads were uniform and enclosed by a callose wall (Fig. 5, C1); however, in b'αβ, the four microspores in the tetrads were unequal in size (Fig. 5, C2) or, in severely affected anthers, exhibited shrinkage (Fig. 5, C3). An inspection of the tetrads by transmission electron microscopy (TEM) revealed that the uniformly sized microspores in wild-type tetrads each contained a large nucleus with a nucleolus in the center (Fig. 5, C4). In contrast, the four microspores in the b'αβ tetrads were unevenly sized; the smallest one was anuclear and shrunken, whereas the largest one contained an abnormal nucleus with no clear boundary (Fig. 5, C5). At stage 8, some of the microspores released from the b'αβ tetrads displayed a normal morphology, similar to that in Col-0 (Fig. 5, D1 and D2), whereas other microspores collapsed (Fig. 5, D3).
Transverse sections of Col-0 and b'αβ anthers at stages 5–8. A1–D1, Lobe of a Col-0 anther at stages (St) 5–8. A2–D2, Lobe of a b'αβ anther at stages 5–8. C3, A severely affected lobe from a b'αβ anther at the tetrad stage by light microscopy. C4, Lobe of a Col-0 anther at the tetrad stage by TEM. C5, A severely affected lobe from a b'αβ anther at the tetrad stage by TEM. D3, A severely affected lobe from a b'αβ anther at stage 8. The ratios at the top right (C1–C3 and D1–D3) indicate the number of representative lobes over the total number of observed lobes. E, epidermis; En, endothecium; ML, middle layer; MMC, microspore mother cells; MC, meiotic cell; Msp, microspores; N, nucleus; Pe, primary exine; T, tapetum; Tds, tetrads. Bars = 20 μm in (A1–D3) and 2 μm in (C4–C5).
These findings suggest that the observed nuclear defects occurred during meiosis.
Unbalanced Separation of Sister Chromatids during Meiosis I Results in Aneuploid Gametophytes in the b'αβ Mutant
Diploid meiocytes generate four haploid gametes through two rounds of cell division, meiosis I and II, in which equal numbers of chromosomes are separated to four poles to establish nuclei in the resulting tetrads. The observed lack of a nucleus in the tetrads and abnormal DAPI staining in the microspores suggested impaired chromosome delivery. Therefore, we examined chromosome reassortment during meiosis in male meiocytes. In the b'αβ mutant, DAPI-stained chromosome spreads displayed a similar morphology to that of the wild type until metaphase I. Gradual chromosome condensation was detected at leptotene (Supplemental Fig. S4, A and B), zygotene (Supplemental Fig. S4, C and D), and diakinesis (Fig. 6, A and B) of prophase I, and five bivalents oriented at the metaphase I plate (Fig. 6, C and D) were noted in wild-type and b'αβ mutant plants. At anaphase I, five pairs of homologous chromosomes separated to two opposite poles, and the sister chromatids in each chromosome were linked together in the wild type (Fig. 6E). However, eight or more chromosome spreads appeared at early anaphase I in b'αβ (Fig. 6F), indicating sister chromatid disassociation. In the wild type, the sister chromatids stayed linked together at telophase I or interkinesis (Fig. 6G) up through metaphase II, during which more condensed chromosomes were aligned on the metaphase II plate (Fig. 6I). At anaphase II, the sister chromatids separated to four poles (Fig. 6K), with five chromosomes in each at telophase II (Fig. 6M). In the b'αβ mutant, the 10 chromatids entered into the nucleus at interkinesis (Fig. 6H) and condensed at metaphase II, but they were unable to align at the metaphase II plate (Fig. 6J). The random separation of chromatids at anaphase II led to unbalanced chromosome reassortment in four poles (Fig. 6L) and generated numerous aneuploid nuclei at telophase II (Fig. 6N) and, consequently, abnormal nuclei in the tetrads. A statistical analysis showed that in contrast with the 97% normal haploid poles observed in the wild type, only about 31% of the poles in b'αβ obtained five chromosomes, which had a chance to develop into normal male gametes. The other 69% of the poles were aneuploid, which usually led to premature abortion of the gametes (Fig. 6,O and P).
Male meiosis in Col-0 and b'αβ. A–N, Chromosome spreads from wild-type (Col-0; A, C, E, G, I, K, and M) and b'αβ (B, D, F, H, J, L, and N) plants were prepared and observed after DAPI staining. A and B, diakinesis in prophase I; C and D, metaphase I; E and F, late anaphase I; G and H, interkinesis; I and J, metaphase II; K and L, anaphase II; and M and N, telophase II. Bar = 5 µm. O, The percentage of poles containing five sister chromatids in Col-0 and b'αβ meiotic cells counted according to (K) and (L). P, The sister chromatid number in each of the four poles counted for 30 b'αβ meiotic cells according to (L).
B'α and B'β Are Expressed in Meiocytes and Localize to the Nucleus and Chromosomes
The protein expression of B'α and B'β in anthers and ovules was investigated using proα:α-YFP/b'αβ and proβ:β-YFP/b'αβ plants to confirm their function in chromosome behavior during meiosis. B'α-YFP and B'β-YFP were both highly expressed in anther meiocytes undergoing meiosis at stage 6 (Supplemental Fig. S5, A and B, arrows). DAPI staining revealed that B'α-YFP and B'β-YFP were mainly localized to the chromosomes at leptotene, prophase I (Supplemental Fig. S5, C–H). However, we were not able to identify the exact location of B'α-YFP and B'β-YFP by observing chromosomes in living cells. Yuan et al. (2018) recently demonstrated the centromeric location of B'α-GFP in male meiocytes using immunodetection. Together, these results support the redundant role of the two proteins in maintaining the association of sister chromatids during meiosis I. In addition, B'α-YFP and B'β-YFP signals were observed in the nucleus of the megaspore mother cell (Supplemental Fig. S6, A–F, arrows). Because the b'αβ megaspore mother cells had a wild-type–like nucleus before meiosis (Supplemental Fig. S7, A and B) but exhibited either an absent or abnormal nucleus in b'αβ embryo sacs after meiosis (Fig. 3, B and C), we presume that B'α and B'β both play roles in female meiosis, probably via the same mechanism.
AtREC8 Is a Target of B'α- and B'β-PP2A and Is Hyperphosphorylated in b'αβ Anthers and Pistils
In yeast, the phosphorylation status of REC8 determines whether the centromeric cohesion that links sister chromatids together is open or closed during meiosis I (Brar et al., 2006; Katis et al., 2010). Currently, it is unknown whether AtREC8/SYN1/DIF1 phosphorylation would cause centromeric cohesion release in Arabidopsis, although AtREC8 has been shown to carry out a function similar to that of yeast REC8 during meiosis (Bai et al., 1999; Bhatt et al., 1999; Cai et al., 2003; Chelysheva et al., 2005). Therefore, we detected the phosphorylation status of AtREC8 in b'αβ pistils and anthers at stages 6–8. Western blotting using anti-β antibodies showed that B'β was not detectable in b'αβ pistils and anthers, even when the amount of protein loaded was increased and the exposure time was extended. As a positive control, B'β was highly expressed in wild-type anthers (Fig. 7A). This result indicates that no B'β protein was present in the b'αβ double mutant. Next, the levels of AtREC8 in b'αβ and wild-type plants were compared. Western blotting using anti-AtREC8 antibodies showed that AtREC8 appeared mainly as a 70-kD band in the wild type but as 70–100-kD bands in the b'αβ mutant (Fig. 7B). Treatment of the protein extract with increasing amounts of calf intestinal alkaline phosphatase (CIP) resulted in a gradual downward band shift from 100 to 70 kD (Fig. 7C), indicating that AtREC8 was hyperphosphorylated in b'αβ pistils and anthers.
B'α and B'β affect and interact with AtREC8. A, B'β protein expression in Col-0 and b'αβ anthers and pistils at stages 5–7 by immunoblotting (IB) using anti-B'β polyclonal antibodies (middle panel, 10-s exposure). Top, Ponceau S staining of the blot shown as a loading reference. The loading of b'αβ in lanes 2 and 3 is two and five times, respectively, that of the loading of Col-0 in lane 1. Bottom, 10-min exposure. B and C, An immunoblot assay for AtREC8 using anti-AtREC8 antibody revealed the accumulation of 70-kD AtREC8 in Col-0 and higher molecular weight AtREC8 in b'αβ (B), as well as a downward shift of AtREC8 from 100 to 70 kD after the treatment of protein extracts prepared from b‘αβ anthers and pistils in stages 5–7 with increasing amounts of CIP (C). Bottom, Ponceau S staining of the blot shown as a loading reference. D, A yeast two-hybrid assay showing the interaction of PP2AB'α and B'β, but not B'ζ, with meiosis- and mitosis-specific α-kleisin AtRCE8 and AtSCC1. Growth on “−H” or 3-AT” medium indicates an interaction. AD, activation domain fusion vector; BD, DNA-binding domain fusion vector; EV, empty AD vector; +H, His present; −H, His absent ;3-AT, His absent while 3-aminotriazole(3-AT) present. E to G, A BiFC assay showing that PP2A B'α (E) and B'β (F), but not B'ζ (G), interacted with AtREC8 and AtSCC1 in the nucleus of tobacco leaf epidermal cells. The dot-like YFP signal indicates the interaction between the two expressed proteins. DAPI staining marks the nucleus. Bar = 5 µm.
A yeast two-hybrid assay revealed that B'α and B'β interacted with AtREC8 and AtSCC1/SYN4 (meiosis- and mitosis-specific cohesion complex ring closers), respectively (Schubert et al., 2009). In contrast, B'ζ interacted only with AtSCC1, and the strength of the interaction was low (Fig. 7D). The other B' subunits, including ε, θ, and γ, did not interact with any of the cohesion proteins that were tested (Supplemental Fig. S8A). A bimolecular fluorescence complementation (BiFC) assay further demonstrated that B'α and B'β interacted with AtREC8 and AtSCC1 in the nucleus, producing signals that appeared as concentrated dots (Fig. 7, E and F). However, B'ζ did not interact with AtREC8 or AtSCC1 in our BiFC experiment (Fig. 7G), similar to the negative control (Supplemental Fig. S8, B–D).
The above results suggest that PP2A B'α and B'β interact directly with and mediate the dephosphorylation of AtREC8.
B'α and B'β Interact with the Cohesion-Associated Proteins SGO and PANS1
The cohesion-associated protein SGO has been found in yeast and mammals to be in charge of recruiting PP2A to centromeres to dephosphorylate REC8 (Kitajima et al., 2006; Riedel et al., 2006; Clift et al., 2009; Kateneva and Higgins, 2009; Xu et al., 2009). We found that B'α and B'β, but not B'ζ, interacted with AtSGO1 and AtSGO2 in a yeast two-hybrid assay (Fig. 8A) and in a BiFC assay (Fig. 8, B and C). The interactions occurred in the nucleus and resulted in dot-like signals, similar to the interaction of AtREC8 with B'α and B'β (Fig. 7, E and F). Based on this fact, we hypothesize that AtSGO1 and AtSGO2 are capable of recruiting B'α- and B'β-PP2A to sites of centromeric cohesion in Arabidopsis, similar to their orthologs in yeast and mammals. Furthermore, B'α and B'β were found to interact with PANS1, a putative APC/C inhibitor that protects against centromeric cohesion release during interkinesis (Cromer et al., 2013; Zamariola et al., 2014b; Singh et al., 2015), and with its homolog, PANS2 (Fig. 8, A–C). The other B' subunits (ζ, ε, θ, and γ) did not interact with any of these cohesion-associated proteins in our yeast two-hybrid assays (Supplemental Fig. S9).
B'α and B'β interacted with the protectors of centromeric cohesion. A, The interaction of PP2A B'α and B'β, but not B'ζ, with protectors of centromeric cohesion in a yeast two-hybrid assay. Growth on “−H” or "3-AT” medium indicates an interaction. AD, activation domain fusionvector; BD, DNA-binding domain fusionvector; +H, His present; −H, His absent; 3-AT, His absent while 3-aminotriazole(3-AT) present. B to D, PP2A B'α (B) and PP2A B'β (C), but not B'ζ (D), interacted with AtSGO1, AtSGO2, PANS1, and PANS2 in the nucleus of tobacco leaf epidermal cells in a BiFC assay. The dot-like YFP signal indicates the interaction between the two expressed proteins. DAPI staining marks the nucleus. Bar = 5 µm.
These results may mean that B'α-PP2A and B'β-PP2A are specifically recruited by AtSGO1 and AtSGO2, and perhaps by PANS1, to sites of centromeric cohesion to mediate AtREC8 dephosphorylation, enabling the protection of those sites during meiosis.
DISCUSSION
PP2A B'α and B'β Redundantly Control Male and Female Meiosis and Are Crucial for Sexual Reproduction in Arabidopsis
The equal distribution of sister chromatids to daughter cells during meiosis is a prerequisite for producing fertile gametophytes and achieving success during sexual reproduction in plants (Zamariola et al., 2014a). The results of this study show that PP2A subunits B'α and B'β participate in the control of chromatid separation. A lack of B'α and B'β (Fig. 7A; Supplemental Fig. S1F) caused the dissociation of sister chromatids before meiosis II (Fig. 6F). The random distribution of chromatids among the four poles at meiosis II resulted in aneuploid tetrads (Fig. 6, L and N). As expected, most b'αβ microspores with an irregular nucleus or without a nucleus were aborted prematurely (Figs. 2 and 4). About 30% of the poles that obtained five chromatids by chance during meiosis II (Fig. 6, O and P) survived and developed fertile pollen. Therefore, as the number of fertile pollen grains increased (e.g. when an excessive amount of b'αβ pollen was used to pollinate wild-type pistils), the seed setting rate increased (Fig. 1, B and E).
Based on the localization of B'α and B'β to chromosomes in female meiocytes (Supplemental Fig. S6, A–F), and given that a lack of B'α and B'β (Fig. 7A; Supplemental Fig. S1F) resulted in similar nuclear defects in the embryo sac (Fig. 3) to that in microspores (Figs. 2 and 4) after meiosis, we presume that B'α and B'β may control chromatid dissociation during female meiosis. Nevertheless, unlike anthers, which produce large numbers of male meiocytes that in turn each produce four microspores (McCormick, 2004), every ovule contains just one megaspore mother cell that generates four megaspores by meiosis, but three degenerate. Thus, the chalazal-most one with a random genotype survives and develops into a functional megaspore and subsequently a mature embryo sac with seven cells and eight nuclei (Yadegari and Drews, 2004). If the probability of correct chromosome distribution is 30%, as determined for male meiosis (Fig. 6O), the probability of fertile ovules harboring a functional megaspore with the correct ploidy is 1/4×30%, meaning that no matter how many viable pollen grains were used in pollination, no more than four ovules per silique (1/4×30%×50) would be able to set seeds. This putative number is in agreement with the observed seed setting rate per silique in the b'αβ mutant (Fig. 1, C and E).
B'α and B'β Dephosphorylate AtREC8 Specifically to Protect Centromeric Cohesion from Removal until Metaphase II
After DNA duplication, two replicated sister chromatids of each chromosome are held together by a ring-like protein complex of cohesins along the full length of the chromosome arms during prophase I (Haering et al., 2008). Unlike the loss of cohesins from chromosomes at the onset of mitotic anaphase, cohesins are released in a stepwise manner during meiosis, ensuring equal separation of the chromatids to four poles (Page and Hawley, 2003). The precise control of REC8 cleavage is crucial for the stepwise release of chromosome cohesion. In anaphase I, REC8 is cleaved by separase at chromosome arms, but it remains at centromeric regions to protect chromatid cohesion from removal until meiosis II (Brar et al., 2006; Marston, 2014). REC8 dephosphorylation promotes the ability of centromeric cohesins to resist separase activity (Kitajima et al., 2006; Riedel et al., 2006).
In Arabidopsis, an ortholog, AtREC8/SYN1, has been reported to play a role in the stepwise release of cohesion (Bai et al., 1999; Bhatt et al., 1999; Cai et al., 2003; Chelysheva et al., 2005). Here, we demonstrated that B'α- and B'β-PP2A holoenzymes catalyze AtREC8 dephosphorylation during meiosis. B'α and B'β interacted with AtREC8 (Fig. 7, D–F). A lack of B'α and B'β resulted in highly phosphorylated AtREC8 in pistils and anthers (Fig. 7, B and C). In contrast with dephosphorylated AtREC8, which was present at a low level in the wild type, the phosphorylated AtREC8 in b'αβ was highly accumulated. Because AtREC8 was found to disappear at anaphase I in b'αβ (Yuan et al., 2018), we presume that the phosphorylated AtREC8 came from meiocytes before anaphase I. An alternative explanation is that AtREC8 is involved in some other aspect of this process, such as chromosome condensation or ensuring the monopolar orientation of the kinetochore (Bai et al., 1999; Bhatt et al., 1999; Cai et al., 2003; Chelysheva et al., 2005).
In eukaryotes, a centromere-localized cohesin-associated protein, SGO, is required to recruit distinct PP2A to dephosphorylate REC8 at the centromeric region (Kateneva and Higgins, 2009; Xu et al., 2009). Usually, B' subunits are recruited to sites of centromeric cohesion, such as ROX three suppressor 1 in yeast (Kitajima et al., 2006; Riedel et al., 2006) and Wdb and Wrd in Drosophila (Pinto and Orr-Weaver, 2017). In Arabidopsis, two SGO-type proteins, AtSGO1 and AtSGO2, protect centromeric cohesion (Cromer et al., 2013; Zamariola et al., 2013, 2014b). Whether they act as recruiters of protein phosphatases has not been shown. Here, we obtained evidence showing that B'α and B'β interact with AtSGO1 and AtSGO2 (Fig. 8, A–C). Yuan et al. (2018) demonstrated that B'α localized to centromeric regions. Furthermore, the Atsgo1 Atsgo2 double mutant exhibited a similar phenotype to the B'αβ double mutant, including the premature disassociation of sister chromatids at anaphase I (Fig. 6F; Cromer et al., 2013; Zamariola et al., 2014b).Together, these data imply that AtSGO1 and AtSGO2 carry out a B'α- and B'β-PP2A phosphatase relay.
PANS1 is a novel protein identified in Arabidopsis that is required to protect centromeric cohesion during interkinesis (Cromer et al., 2013; Zamariola et al., 2014b). A specific interaction was detected between PANS1 and B'α or B'β, but not B'ζ, probably meaning that PANS1 serves as a second recruiter during interkinesis to carry B'α- and B'β-PP2A to sites of centromeric cohesion or to help stabilize the SGO1-PP2A complex (in addition to its function as a regulator and target of APC/C).
Because REC8 is evolutionarily conserved among eukaryotes, the B regulatory subunit, which determines the substrate specificity of PP2A, should also be conserved to mediate REC8 dephosphorylation. Therefore, B’ subunits are always selected to regulate centromeric cohesion release during meiosis. Additionally, unlike the male- or female-specific molecules that regulate the distinct development of male or female gametophytes (Borg et al., 2009; Yang et al., 2010), the molecules described above function in both male and female meiosis by a conserved mechanism.
The Functions of B'α and B'β in Sporophytic Tissue Remain to be Elucidated
A transmission efficiency test indicated sporophyte-dependent defects in b'αβ male and female gametophytes (Supplemental Fig. S2). Consistently, B'α and B'β also expressed in the tapetum of anthers and the nucellus of ovules (Supplemental Figs. S5 and S6), and we were unable to exclude the function of B'α and B'β in surrounding sporophytic tissues that usually affect germ cell development and maturation (Liu and Fan, 2013). Whether a lack of B'α or B'β in sporophytic tissue affects microspore and megaspore development should be elucidated only after recovering meiotic defects using B'α or B'β driven by a meiosis-specific promoter in the future.
Interestingly, our yeast two-hybrid and BiFC assays revealed that B'α and B'β not only interacted with AtREC8, but also interacted with AtSCC1/SYN4, a mitosis-specific α-kleisin protecting centromeric cohesion (Fig. 7, D–F), implying that in Arabidopsis, B'α and B'β mediate AtSCC1 dephosphorylation during mitosis as well. Furthermore, B'α and B'β interacted with PANS2, which has been reported to function in mitosis (Cromer et al., 2013), suggesting that B'α- and B'β-PP2A are recruited by PANS2. This is supported by the fact that the signals showing the interaction of AtSCC1 with B'α and B'β were similar to those of PANS2 with B'α and B'β: They all occurred in the nucleus as fluorescent dots (Fig. 7, E and F; Fig. 8, B and C). The normal mitosis and healthy vegetative growth, but abnormal meiosis and aborted male and female gametophytes, observed in the b'αβ mutant support the idea that another B’ subunit can replace B'α and B'β functionally in mitosis but not in meiosis.
In conclusion, B'α- and B'β-PP2A may be carried to the centromere by more than one recruiter in Arabidopsis.
MATERIALS AND METHODS
Plant Materials and Growth Conditions
Arabidopsis (Arabidopsis thaliana) plants (ecotype Col-0) were used in all experiments, and tobacco (Nicotiana benthamiana) plants were used in the BiFC assay. The plants were grown under long-day conditions (16 h of light [80 μmol m−2 s−1 white light]/8 h of dark at 22°C).
Genotyping of the T-DNA Insertion Mutants and Complemented Plants
The pp2ab'α T-DNA insertion line was identified by PCR using three primers. One primer (Left Border b1.3) annealed to the T-DNA insertion site, and a pair of primers (α-F/α-R) were designed to amplify the full-length complementary DNA spanning the insertion site (http://signal.salk.edu/). Identification of the pp2ab'β T-DNA insertion line was done according to the method for pp2ab'α using the primer pair β-F/β-R and Left Border b1.3. The complementary plants were identified by amplifying exogenous B'a-YFP or B'β-YFP with the primer pair β-F/YFP-N-R or α-F/YFP-N-R; the background was confirmed by genotyping the T-DNA insertion mutants. The sequences of the primers used are listed in Supplemental Table S1.
Vector Construction and Transformation
To create the proα:α-YFP and proβ:β-YFP expression vectors, the coding sequences of B'α and B'β without the stop codons were amplified, respectively, from bud complementary DNA, inserted into the entry vector pENTR/SD/D-TOPO (Invitrogen), and then introduced into a p101 destination vector harboring the B'α or B'β promoter by attL and attR (LR) recombination. The constructs were sequenced before being introduced into Arabidopsis Col-0 via Agrobacterium GV3101-mediated transformation. The plants used for complementation testing were identified from the progeny of crosses using proα:α-YFP or proβ:β-YFP plants as the female parent and b'αβ as the male parent.
Immunoblotting
Total protein was extracted from anthers and pistils at the meiosis stage using 2×SDS loading buffer. Endogenous B'β and exogenous B'β-YFP were probed using polyclonal anti-B'β antibodies raised against the N terminus (amino acids 1–109) of B'β; the antibody specifically recognizes B'β but not B'α (Zhou et al., 2012). The AtREC8 was probed using anti-AtREC8 antibody (Cai et al., 2003).
For the AtREC8 phosphorylation status assay, total protein was extracted from anthers at stage 6 using the extraction buffer (20 mm HEPES, 40 mm KCl, 1 mm EDTA, and 1% (v/v) Triton X-100, pH 7.5). Next, 50 μL of extract was mixed with 0, 1, or 2 μL of CIP (1 mg/mL; M0290; New England Biolabs) and incubated at 37°C for 40 min. Then, 5 μL of protein extract was separated by SDS-PAGE, and the blot was probed using anti-AtREC8 antibody. Horseradish peroxidase-conjugated goat antirabbit IgG (γ-chain-specific; Sigma-Aldrich) was used as the secondary antibody. The signal was developed using SuperSignal West Femto Chemiluminescence Reagent (Thermo Fisher Scientific).
Pollen Viability Assays
Intact anthers before dehiscence were stained with Alexander’s solution (Alexander, 1969) to detect pollen viability or crushed in 150 µL of 1%(w/v) I2-KI and incubated for 15–20 min at room temperature to detect starch granules in the pollen grains. After being washed with the same buffer, the anthers and pollen grains were observed by microscopy (Zeiss Imager M2).
DAPI Staining and Fluorescence Microscopy
Anthers at stage 8 or 12 were collected and crushed with a fine syringe needle to release microspores or mature pollen grains in a drop of water on the slide. The newly released microspores were stained with 1 μg/mL of DAPI and observed immediately by fluorescence microscopy (Zeiss Imager M2).
For the B'α-YFP and B'β-YFP expression assays, ovules and anthers from the complemented plants were used. Ovules and male meiocytes, released from anthers at the meiosis stage, were stained with 1 μg/ml of DAPI and observed by confocal laser scanning microscopy (Zeiss LSM-710). Fluorescent signals were captured at 526–573 nm (emission) for YFP with 488-nm excitation and 420–490 nm (emission) for DAPI with 340-nm excitation.
Chromosome Spread Preparation
Buds at different stages of meiosis were collected and fixed in Carnoy’s fluid (50% [v/v] ethanol, 5% [v/v] acetic acid, and 3.7% [v/v] formaldehyde) for 4 h at room temperature. After three washes with 10 mm citrate buffer (10 min each), buds harboring yellow anthers were removed and the remaining buds (with white anthers) were treated with an enzyme solution containing 30 mg/mL of pectolyase (Y-008; Seishin Corp.), 15 mg/mL of cellulose (131126-01; Sigma-Aldrich), and 15 mg/mL of β-glucuronidase (SLBV4208; Sigma-Aldrich) in 10 mm sodium citrate (pH 4.5) for 80 min at 37°C to digest the cell wall. After digestion, the anthers were broken with a fine syringe needle, mixed with 60% (v/v) acetic acid at 45°C twice (30 s each time), and then fixed with prechilled Carnoy’s fluid. The broken anthers were stained with 1 μg/mL of DAPI and examined by fluorescence microscopy (Zeiss Imager M2).
Histological Analysis
Inflorescences were fixed in Carnoy’s fluid and then embedded in Spurr’s resin (Structure Probe, Inc.). Microsections (1 μm thick) were stained in 0.05% (w/v) toluidine blue (Sigma-Aldrich) and photographed using a microscope (Zeiss Imager M2).
TEM
Arabidopsis floral buds were fixed in 2% (w/v) paraformaldehyde and 2% (v/v) glutaraldehyde in phosphate-buffered saline (pH 7.4) at 4°C overnight and then postfixed in 1% (w/v) OSO4 for 4 h. After dehydration in an ethanol series, the buds were embedded in Spurr’s resin. Ultrathin sections (∼80 nm) were produced and stained with 2% (w/v) saturated uranyl acetate in 70% (v/v) ethanol before observation under a transmission electron microscope (Hitachi H-7650).
Yeast Two-Hybrid Assays
The coding sequences of AtREC8, AtSCC1, SGO1, SGO2, PANS1, and PANS2 were amplified by PCR using the primer pairs AtREC8-TOPO-F/AtREC8-TOPO-R, AtSCC1-TOPO-F/AtSCC1-TOPO-R, AtSGO1-TOPO-F/SGO1-TOPO-R, AtSGO2-TOPO-F/SGO2-TOPO-R, PANS1-TOPO-F/PANS1-TOPO-R, and PANS2-TOPO-F/PANS2-TOPO-R (Supplemental Table S1). The products were inserted into the entry vector pENTR/SD/D-TOPO (Invitrogen) to create pENTR-AtREC8, pENTR-AtSCC1, pENTR-SGO1, pENTR-SGO2, pENTR-PANS1, and pENTR-PANS2. They were then cloned into pGADT7 by LR recombination to produce prey vectors. The bait vectors, including BD-B'α, BD-B'β, BD-B'ζ, BD-B'θ, BD-B'ε, and BD-B′γ, were generated previously (Tang et al., 2011) and cotransformed with each of the prey vectors AD-AtREC8, AD-AtSCC1, AD-SGO1, AD-SGO2, AD-PANS1, and AD-PANS2 into yeast strain AH109. Transformants that grew on SD/-Leu-Trp-His or SD/-Leu-Trp-His+3AT plates indicated an interaction.
BiFC Assays
pENTR-AtREC8, pENTR-AtSCC1, pENTR-SGO1, pENTR-SGO2, pENTR-PANS1, and pENTR-PANS2 were cloned into pX-cCFP (C-terminal cyan fluorescence protein) by LR recombination to produce AtREC8-cCFP, AtSCC1-cCFP, SGO1-cCFP, SGO2-cCFP, PANS1-cCFP, and PANS2-cCFP; pENTR-B'α, pENTR-B'β, and pENTR-B'ζ were cloned into pX-nYFP by LR recombination to produce B'α-nYFP, B'β-nYFP, and B'ζ-nYFP. All of the constructs were sequenced before transformation into Agrobacterium tumefaciens GV3101. After coinjection of the cCFP- and nYFP-fusion vectors for 36–48 h, tobacco (N. benthamiana) leaves were injected with 1 μg/ml of DAPI and then visualized under a confocal microscope (FV3000; Olympus Corp.) in two channels (YFP and DAPI).
Accession Numbers
The sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: PP2AB'α (At5g03470), PP2AB'β (At3g09880), PP2AB'ζ (At3g21650), PP2AB'θ (At1g13460), PP2AB'ε (At3g54930), PP2AB′γ (At4g15415), PANS1 (At3g4190), PANS2 (At5g12360), AtSGO1 (At3g10440), AtSGO2 (At5g04320), AtREC8/SYN1/DIF1 (At5g05490), and AtSCC1/SYN4 (At5g16270).
Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. B'α and B'β are required for fertility in Arabidopsis.
Supplemental Figure S2. Transmission efficiency test for a−/−β−/− male and female gametophytes.
Supplemental Figure S3. Endosperm nuclei marked by DD36:GFP in Col-0 and b'αβ mutant plants at 24 HAP.
Supplemental Figure S4. Comparison of chromosome spreads between Col-0 and b'αβ during male meiosis I by DAPI staining.
Supplemental Figure S5. B'α-YFP and B'β-YFP expression in male meiotic cells and their location in the chromosome.
Supplemental Figure S6. B'α-YFP and B'β-YFP expression in megaspore mother cells.
Supplemental Figure S7. Comparison of megaspore mother cells (arrow) from Col-0 and b'αβ by DIC microscopy.
Supplemental Figure S8. Other B' subunits did not interact with the cohesion proteins or cohesion-associated components in yeast two-hybrid assays.
Supplemental Figure S9. Negative control for the BiFC assay.
Supplemental Table S1. The primers used in this study.
Acknowledgments
We thank Cathrine Lillo at the Centre for Organelle Research (Stavanger, Norway) for kindly providing the b'αβ seeds; Wei-Cai Yang at Institute of Genetics and Developmental Biology, Chinese Academy of Sciences (Beijing, China) for providing the marker lines DD45:GFP and DD36:GFP seeds; Zhi-Yong Wang at the Carnegie Institution for Science (Washington, DC, USA) for providing the proα:α-YFP and proβ:β-YFP seeds; Raphael Mercier at the Institut Jean-Pierre Bourgin (Versailles, France) for kindly providing the anti-AtREC8 antibody; and Jia-lin Zhou at Hebei Normal University (Hebei, P.R. China) for producing the polyclonal anti-B'β antibody. We also thank Yu-Hong Hu of the Instrumental Analysis Center at Hebei Normal University (Hebei, P.R. China) for her assistance with the TEM.
Footnotes
↵1 This work was supported financially by the National Key Program on the Development of Basic Research in China (2013CB126900) and the National Science Foundation of China (31170264). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
↵2 These authors contributed equally to this work.
↵3 Present address: Institute of Biophysics, Chinese Academy of Science, Chaoyang District, Beijing 100101, People’s Republic of China.
- Received October 22, 2018.
- Accepted January 20, 2019.
- Published January 31, 2019.