Research ArticlesBOTANY

RETINOBLASTOMA RELATED1 mediates germline entry in Arabidopsis

See allHide authors and affiliations

Science  28 Apr 2017:
Vol. 356, Issue 6336, eaaf6532
DOI: 10.1126/science.aaf6532

Germ cells on demand

Unlike animals, plants do not set aside a germline. Instead, germ cells are developed on demand from somatic lineages. Zhao et al. examined the regulatory pathways that manage the transition from somatic to germ cell development in the small plant Arabidopsis (see the Perspective by Vielle-Calzada). The transcription factor WUSCHEL (WUS) was needed early on for development of ovules. Soon after, a trio of inhibitors that work through a cyclin-dependent kinase allowed a transcriptional repressor to down-regulate WUS. This opened the door to meiosis, while restricting the number of reproductive units per seed to one.

Science, this issue p. eaaf6532; see also p. 378

Structured Abstract

INTRODUCTION

Seeds of flowering plants have evolved to typically carry only a single embryo next to a nourishing tissue, the endosperm that functions analogously to the human placenta. Both structures are formed from a single female gametophyte (embryo sac) that develops from one meiotic product, the functional megaspore, in most sexually reproducing plants. As in many other organisms, including humans, all but one female meiotic product die to assure the development of only one reproductive unit per future seed.

RATIONALE

In contrast to humans and animals, plants do not set aside a specialized cell lineage (germline) that produces meiocytes in early embryogenesis. Instead, the germline of plants is established de novo from somatic cells in floral reproductive organs. Several genes have been identified that control the formation of ovules, which harbor the meiocytes [megaspore mother cells (MMCs)]. These include the homeodomain transcription factor WUSCHEL (WUS), a key regulator of stem cell fate in plants that is essential for the formation of the integuments from which the seed coat is derived. Moreover, WUS is also involved in the specification of MMCs. However, it is not clear how somatic cells that divide mitotically switch to a meiotic cell division program.

RESULTS

Following up expression data of young ovules primordia and MMCs, we reveal a regulatory cascade that controls the entry into meiosis, starting with a group of redundantly acting cyclin-dependent kinase (CDK) inhibitors of the KIP-RELATED PROTEIN (KRP) class. KRPs function by restricting CDKA;1–dependent inactivation of the Arabidopsis Retinoblastoma homolog RBR1. In rbr1 and krp triple mutants, designated meiocytes undergo several mitotic divisions, resulting in the formation of supernumerary meiocytes that give rise to multiple reproductive units per future seed. Live observation revealed that these multiple units can successfully attract a pollen tube and can be fertilized. However, subsequent seed development is blocked, resulting in semisterility of the mutant plants. One of the functions of RBR1 is the direct repression of the stem cell factor WUSCHEL (WUS), which ectopically accumulates in meiocytes of triple krp and rbr1 mutants. Depleting WUS in rbr1 mutants restored the formation of only a single meiocyte. However, ectopic expression of WUS by itself is not sufficient to induce mitotic divisions instead of meiosis, revealing that RBR1 is a central hub of meiocyte differentiation.

CONCLUSION

There is a delicate balance between WUS activation important for ovule primordia formation—including the development of the integuments, as well as a role in specifying the MMC itself—and its inactivation by RBR1 soon afterward to allow entry into meiosis. Different components of the Rb control pathway have been associated with germ cell fate initiation in animals; for example, mutants in the CDK inhibitor dacapo in Drosophila fail to enter meiosis. Similarly, down-regulation of Cdk2-cyclin E, a well-known regulator of Rb, is important for Caenorhabditis elegans germline development. This raises the intriguing question of whether Rb family proteins play a conserved role in germline entry in multicellular organisms.

Formation of multiple MMCs in Arabidopsis.

The left column shows the wild type, the middle column shows krp4 krp6 krp7 triple mutants, and the right column shows rbr1-2. A designated MMC undergoes a mitotic instead of a meiotic division, leading to the production of multiple MMCs in triple krp and rbr1 mutants (top row) instead of a single MMC, as found in the wild type. MMC fate is highlighted by means of KNU-YFP expression (second row). Multiple MMCs give rise to multiple gametophytes in ovules (third row and highlighted in different shades of blue in the bottom row).

Abstract

To produce seeds, flowering plants need to specify somatic cells to undergo meiosis. Here, we reveal a regulatory cascade that controls the entry into meiosis starting with a group of redundantly acting cyclin-dependent kinase (CDK) inhibitors of the KIP-RELATED PROTEIN (KRP) class. KRPs function by restricting CDKA;1–dependent inactivation of the Arabidopsis Retinoblastoma homolog RBR1. In rbr1 and krp triple mutants, designated meiocytes undergo several mitotic divisions, resulting in the formation of supernumerary meiocytes that give rise to multiple reproductive units per future seed. One function of RBR1 is the direct repression of the stem cell factor WUSCHEL (WUS), which ectopically accumulates in meiocytes of triple krp and rbr1 mutants. Depleting WUS in rbr1 mutants restored the formation of only a single meiocyte.

Seeds of flowering plants have evolved to typically carry only a single embryo next to a nourishing tissue, the endosperm. Both structures are formed from a single female gametophyte (embryo sac) that develops from one meiotic product, the functional megaspore (FM), in most sexually reproducing plants. As in many other organisms, including humans, all but one female meiotic product die to assure the development of only one reproductive unit per future seed (1). In contrast to humans and animals, however, plants do not set aside a specialized cell lineage (germline) that produces meiocytes in early embryogenesis. Instead, the germline of plants is established de novo from somatic cells in floral reproductive organs (2, 3).

Several genes have been identified that control the formation of ovules, which harbor the meiocytes [megaspore mother cells (MMCs)] (4). These include the homeodomain transcription factor WUSCHEL (WUS), a key regulator of stem cell fate in plants, that is essential for the formation of the integuments from which the seed coat is derived (5). Moreover, WUS is also involved in the specification of MMCs (6). Here, we ask how somatic cells that divide mitotically switch to a meiotic cell-division program.

Triple krp mutants produce multiple embryo sacs

In order to get insights into the molecular mechanism of this important transition, we analyzed expression data of young ovules (7), indicating that the cyclin-dependent kinase (CDK) inhibitor KIP-RELATED PROTEIN 7 (KRP7) was enriched in MMCs (fig. S1A). Using a previously generated translational reporter line (8), we corroborated that KRP7 is expressed, albeit very weakly, in MMCs (Fig. 1, A to C). However, we did not find any obvious alteration of ovule development in krp7 mutants. Because KRP6 and KRP7 are closely related members of the seven-member-containing KRP family in Arabidopsis (9, 10), and KRP6 is also expressed in MMCs (11), confirmed here with reporter gene analysis and in situ mRNA hybridization (fig. S1, B, C, F, and G), we created a krp6 krp7 double mutant. Because this double mutant did not have an obvious mutant phenotype either, we generated triple mutants combining all remaining krp mutants with krp6 krp7: krp1 krp6 krp7, krp2 krp6 krp7, krp3 krp6 krp7, krp4 krp6 krp7, and krp5 krp6 krp7. All of these triple mutants displayed a similar mutant phenotype, with reduced fertility as expected for mutants affecting MMC formation/function: undeveloped and eventually aborting ovules in maturing siliques (Fig. 2, A, B, and R). Abortion was especially pronounced in krp4 krp6 krp7, with ~30% degenerating ovules (n = 536 ovules). Similar to KRP6 and KRP7, we found by means of in situ mRNA hybridization that KRP4 was weakly expressed in different cells in ovule primordia, including MMCs (fig. S1, D and E). Thus, although these three KRP genes are not exclusively expressed in MMCs, they show a specific overlap in this cell type. We verified that the triple-mutant phenotype results from the redundant function of these KRPs by examining the double mutants krp4 krp6 and krp4 krp7, which did not show ovule abortion (Fig. 2R).

Fig. 1 MMC formation in krp4 krp6 krp7 and rbr1-2 mutants.

(A) Confocal laser scanning micrograph of an ovule primordium of a plant expressing a KRP7-GFP reporter construct showing GFP accumulation in the female meiocyte (MMC, arrow), in the wild type. (B) Bright field image of (A). (C) Overlay of (A) and (B). (D to F) Confocal laser scanning micrographs of MMCs. (D) A single MMC (arrow) is typically found in wild-type ovules. (E) krp4 krp6 krp7 homozygous triple mutants develop supernumerary MMCs (arrows). (F) Supernumerary MMCs (arrows) in homozygous rbr1-2 mutants. (G to I) Confocal laser scanning micrograph of plants expressing a KNU-YFP reporter construct marking MMC fate. (G) Wild-type plant with a single MMC. (H) Homozygous krp4 krp6 krp7 mutants with two MMCs. (I) Homozygous rbr1-2 mutant with four MMCs. (J and K) Confocal laser scanning micrograph of plants expressing a CYCB1;2-GFP reporter construct. (J) CYCB1;2 is not expressed in MMCs in the wild type. (K) CYCB1;2 becomes expressed in one of two MMCs (arrows) in krp4 krp6 krp7 mutants, indicating a mitotic instead of a meiotic cell division. (L) Confocal laser scanning micrograph overlaid with a bright field picture of a krp4 krp6 krp7 mutant expressing a GFP-KRP7 fusion construct under the KNU promoter, which largely rescues the triple krp mutant phenotype. A single MMC (arrow) is formed instead of multiple MMCs in the triple mutant. (M) Differential interference contrast micrograph of a rbr1-2 wus-7/WUS plant in which only a single MMC is formed (arrow). (N) Flow cytometry analyses of flowers from wild type (WT), krp4 krp6 krp7 triple mutants (467), cdka;1 PROCDKA;1:CDKA;1T14D;Y15E (DE), rbr1-2 krp4 krp6 krp7 DE (467DE), and rbr1-2 DE. A lower 2C/4C ratio than in the wild type indicates shorter and a higher 2C/4C ratio indicates longer G1 and S phases; error bars indicate SD; and asterisks indicate statistically significant differences based on a Mann-Whitney test between krp4 krp6 krp7 and krp4 krp6 krp7 DE (P = 0.0286) as well as between rbr1-2 and rbr1-2 DE (P = 0.0286). (O) Kinase assays with plant extracts from wild type (WT) and krp4 krp6 krp7 (krp467) mutant plants, demonstrating slightly higher CKS-associated CDKA;1 kinase activity against Histone H1 as a generic substrate. (Top) Affinity purification of CDKA;1 demonstrating equal amounts of kinase protein. (Middle) Kinase activity shown by autoradiography. (Bottom) Coomassie Brilliant Blue staining showing equal loading of the substrate histone H1. (P) Quantification of the number of MMCs in the indicated genotypes. Error bars indicate SD; asterisks indicate statistically significant difference based on a Mann-Whitney test between krp4 krp6 krp7 and krp4 krp6 krp7 DE (P = 0.0286) as well as between rbr1-2 versus rbr1-2 wus-7/WUS (P = 0.0286) and rbr1-2 versus rbr1-2 wus-101/WUS (P = 0.0286); no significant difference could be found between rbr1-2 and rbr1-2 DE. Scale bars, (A) to (L) 10 μm; (M) 50 μm.

Fig. 2 Embryo sac development in krp4 krp6 krp7 and rbr1-2 mutants.

(A) Wild-type siliques with developing plumb seeds. (B) krp4 krp6 krp7 triple mutants have many undeveloped and early aborting ovules (arrows). (C) Ovule abortion in krp4 krp6 krp7 triple mutants can be nearly completely rescued through expression of GFP-KRP7 driven by the KNU promoter. (D) Ovule abortion in krp4 krp6 krp7 triple mutants can be largely rescued through expression of GFP-KRP7 under the control of the DMC1 promoter. Some ovules are still aborting (arrow). (E to G) Confocal laser scanning micrographs of a mature embryo sac in the wild type and krp4 krp6 krp7 triple mutants. (E) Mature ovule of wild-type plants containing a single embryo sac with one central cell, one egg cell, and two synergids harboring large, medium-sized, and small nuclei, respectively (only one synergid nucleus is visible in the optical section shown). [(F) and (G)] Two consecutive sections of a krp4 krp6 krp7 triple-mutant ovule showing three embryo sacs (movie S5). (H) Diagram of the ovule shown in (F) and (G), with presumably one embryo sac at stage FG5 and two at stage FG6. Nuclei visible in the optical section shown in (F) are colored in blue, nuclei visible in the optical section shown in (G) are colored in white, and nuclei outside the focal plane are colored in red (only visible in movie S5); different blue tones indicate different embryo sacs. (I to P) Confocal laser scanning micrographs of wild-type and krp4 krp6 krp7 embryo sacs expressing central and egg cell fate markers. (I) Only a single cell expresses the central cell marker DD19 in the wild type. (J) Two cells express the central cell marker DD19 in krp4 krp6 krp7 triple mutants. (K) Only one cell expresses the egg cell marker DD45 in wild-type embryo sacs. (L) Two cells express the egg cell marker DD45 in krp4 krp6 krp7 triple mutants. [(M) to (P)] Overlay of the GFP signal in (I) to (L) with autofluorescence in red, revealing the outlines of the ovules. (Q) Quantification of supernumerary embryo sacs (emb sac) in wild-type, krp double-mutant (46, 47, and 67), and krp triple-mutant (167, 267, 367, 467, and 567) ovules. Error bars indicate SD. (R) Quantification of ovule abortion rate in plants with the indicated genotype. K7 indicates PROKNU:KRP7, KG7 indicates PROKNU:GFP:KRP7, D7 indicates PRODMC1:GFP:KRP7, and DG7 indicates PROKNU:GFP:KRP7. All PROKNU and PRODMC1 constructs are in krp4 krp6 krp7 triple-mutant background; error bars indicate SD. Scale bars, (E) to (G) and (M) to (P) 50 μm.

In wild-type plants, the mature embryo sac comprises three cell types that exhibit specific nuclear morphologies: The central cell contains a large and decondensed nucleus, the egg cell nucleus is smaller, and the two synergids have the smallest nuclei (Fig. 2E; fig. S2, A to G; and movies S1 to S3) (12). The krp4 krp6 krp7 triple mutant displayed not just a single embryo sac per ovule but several groups of nuclei, each with one large and several small ones, which is suggestive of multiple embryo sacs per ovule (Fig. 2, F to H). Using reporter lines for egg and central cell fate in the triple-mutant background highlighted two or more egg and central cells per ovule (Fig. 2, I to P). Morphological studies confirmed that these groups of nuclei are separated from each other, indeed representing multiple distinct embryo sacs (fig. S2N and movie S6).

To address their functionality, we asked whether multiple embryo sacs can attract a pollen tube and get fertilized. To this end, we followed by live-imaging double fertilization of krp triple-mutant ovules in a semi–in vitro system using a male parent in which vegetative and sperm nuclei of pollen are labeled with histone H2B-tdTomato (fig. S3A and movies S7 and S8). In addition, we traced pollen tube growth in mutant ovules with Congo red and Aniline blue staining (fig. S3, B to M). Pollen tube attraction was equally successful (~80%, n = 527 ovules) in ovules with two or multiple embryo sacs (fig. S3, A, F, and G, and movies S7 to S8). Because pollen tube attraction depends on functional synergids, our live observation experiments indicated that next to central and egg cells, synergids are also correctly differentiated, at least with respect to pollen tube attraction, in the supernumerary embryo sacs.

However, the egg/central cell and the sperm cells failed to undergo plasmogamy in ~20% of the cases (n = 527 ovules) (fig. S3, E and H). Arabidopsis ovules will attract a second pollen tube by the persistent synergid if gamete fusion fails after reception of a first pollen tube (13, 14). Indeed, we found that 14.2% ± 4.4 (mean ± SD, n = 599 ovules) of the ovules in krp4 krp6 krp7 were targeted by a second pollen tube (fig. S3, I to M). Despite frequent attraction of multiple pollen tubes, we never observed seeds with two embryos or one embryo next to an egg cell (n = 269 seeds) (fig. S4). Taken together, these data show that the developmental arrest of supernumerary gametophytes is due to unknown reasons at the level of plasmogamy and a failure of the fertilized egg/central cell to proliferate.

To trace the origin of the multiple embryo sacs, we first determined the transmission rate of the krp7 mutant allele in reciprocal crosses with plants homozygous for krp4 and krp6 but heterozygous for krp7 (krp4 krp 6 krp7/KRP7). We did not find a significant segregation ratio distortion of the krp7 mutant allele [transmission efficiency through the male (TEM) = 92%, n = 532 seedlings, and through the female (TEF) = 96%, n = 642 seedlings], indicating that the development of multiple embryo sacs in krp4 krp6 krp7 was not due to a gametophytic effect. In accordance with these genetic data, we traced the formation of supernumerary embryo sacs back to the appearance of additional female MMCs and/or supernumerary female megaspores, suggesting a sporophytic defect in krp4 krp6 krp7.

The position of these cells—which could also appear side by side rather then on top of each other, as it is the case for meiotic products—along with their size and their appearance in young ovule primordia, as judged by the not yet extended integuments, suggested that these cells are MMC rather than megaspores (Fig. 1, D and E; fig. S2, H to N; and movies S4 to S6). Furthermore, we found by means of RNA in situ hybridization that the RNA DEAD-box helicase MNEME (MEM), whose expression is enriched in MMCs, is present in these cells (figs. S1A and S5, A to C) (7). In addition, a promoter reporter line for transcription factor KNUCKLES (KNU), which is expressed in MMCs but not the surrounding cells (15), was active in each of these large cells in krp4 krp6 krp7 (Fig. 1, G and H).

The appearance of these cells was dependent on KRPs because the expression of KRP7 or a green fluoresent protein (GFP)–KRP7 fusion under the control of the KNU promoter (PROKNU:KRP7 and PROKNU:GFP:KRP7) largely rescued the krp triple-mutant phenotype, resulting in the formation of a single MMC/functional FM in ovule primordia and restoring plant fertility (Figs. 1, L and P, and 2, C and R). Similar results were obtained when the meiotic DISRUPTED MEIOTIC cDNA1 (DMC1) promoter was used to drive expression of KRP7 or a GFP-KRP7 fusion in krp4 krp6 krp7 MMCs (PRODMC1:KRP7, PRODMC1:GFP:KRP7) (Figs. 1P and 2, D and R, and fig. S1, H and I).

However, because we could not rule out that MEM transcript is carried over into FMs—and similarly, yellow fluorescent protein (YFP) driven from the KNU promoter could also be inherited into early stages of FMs—we wanted to find additional evidence that these large cells are MMCs and not megaspores. Multiple megaspores in one ovule primordium could result from a failure of selecting a single FM after meiosis. In the wild type, the three micropylar-most meiotic products die (fig. S6, A to C), and indeed, KRPs have been shown to lead to cell death when strongly overexpressed (16). However, the observation that often only two or three MMCs (instead of four expected meiotic products) were found in krp triple mutants without any signs of cell death indicates that supernumerary MMCs are formed before the selection of the FM (Fig. 1, E and H, and fig. S2H). Furthermore, we found dead cells on top of two or more surviving megaspores when analyzing later stages of ovule development in the triple krp mutant, indicating that megaspore selection occurs normally, although we cannot fully rule out that always all but one formed megaspore will die in krp triple mutants (figs. S2I and S6, D, E, and F). The selection of the FM can take place at different positions within the ovule primordium of krp triple mutants, suggesting that the selection mechanism in the wild type also involves communication between the meiotic products and does not only depend on the position of the meiotic products with respect to the micropylar and chalazal orientation in the ovule primoridium. Last, the presence of meiotic chromosome configurations in these large cells indicated that these cells are indeed MMCs and that two or more MMCs enter meiosis in parallel (n = 143 ovules) (Fig. 1P and fig. S5, D and E).

These findings, together with our above expression analyses, suggested that KRP4, KRP6, and KRP7 function in restricting meiosis to a single MMC. The defect to enter meiosis was not limited to female meiosis because supernumerary meiocytes were also formed in anthers of the krp4 krp6 krp7 triple mutant (fig. S7, A to D). Viability of the male meiotic products, the male spores that develop into pollen grains, appeared not to be affected (fig. S7, F and G).

Production of multiple MMCs depends on RBR1

Supernumerary MMCs could be formed by two fundamentally different mechanisms. First, KRPs may function in preventing cells adjacent to an existing MMC from adopting germline fate. Such a mechanism has been proposed for the additional germline cells in mutants disrupting ARGONAUTE 9 (AGO9), one of several members encoding the catalytic component of an RNA-induced silencing complex in Arabidopsis (17). Given that KRPs have the potential to move between cells (16), such a non–cell-autonomous function would even be consistent with the expression of KRPs in the designated MMC.

Alternatively, an already specified MMC might undergo one or more mitotic divisions instead of a meiotic division. To discriminate between the latter two scenarios, we capitalized on the observation that a group of B-type cyclins, the B1 class, is expressed in mitotically but not meiotically dividing cells in Arabidopsis (18). Analysis of a CYCB1;2-GFP reporter line in the krp4 krp6 krp7 triple-mutant background showed strong activity in a few MMCs, indicating that at least some, presumably all, of the supernumerary MMCs are formed by a failure of the designated MMC to switch from a mitotic to a meiotic mode of division (Fig. 1, J and K).

Because CDK inhibitors in animals can control cell differentiation independently of cell-cycle regulation (19), we next asked whether the mutant phenotype of krp4 krp6 krp7 was due to a failure to inhibit CDKs or a noncanonical role. Kinase assays and flow cytometry with cell extracts from flower buds of krp4 krp6 krp7 mutants demonstrated slightly higher CDK activities and an increase of G2-phase cells in the triple krp mutant when compared with the wild type (Fig. 1, N and O), which is consistent with the notion that KRPs restrict the entry into S phase in Arabidopsis (10). CDKA;1 is the major cell-cycle kinase in Arabidopsis and the presumed major target of KRP6 and KRP7 (9, 20, 21). CDKA;1 was earlier found to be expressed in MMCs (fig. S1A) (9), and to test whether the KRP proteins and CDKA;1 are in the same regulatory pathway during MMC development, we combined a hypomorphic mutant of CDKA;1 [a weak loss-of-function rescue construct in a homozygous mutant background (22)] with the krp4 krp6 krp7 triple mutant. The resulting quadruple mutant krp4 krp6 krp7 cdka;1 containing the transgene PRO:CDKA;1:CDKA;1DE (short: krp4 krp6 krp7 DE) had more cells in G1 phase than did krp4 krp6 krp7. The frequency of ovules with supernumerary MMCs was reduced by 50% (n = 128 ovules; Mann-Whitney test, P = 0.0286), indicating that the mutant phenotype is largely due to the failure of KRPs to inhibit CDKA;1 (Fig. 1, N and P).

Because the plant Retinoblastoma (Rb) homolog RETINOBLASTOMA-RELATED1 (RBR1) is a target of CDKA;1 (13), we asked whether RBR1 is also involved in MMC specification. RBR1 is required to block proliferation of gametic nuclei and cells within the mature embryo sac (23) and for meiotic recombination (24). Expression analyses revealed that RBR1 is present in all or nearly all cells of ovule primordia, including the MMC, which is consistent with its predominant regulation at the posttranslational level (fig. S1, A, J, and K). Using a homozygous temperature-sensitive rbr1 mutant (rbr1-2) grown at its restrictive temperature (13), we found an increase in the number of meiocytes in both female and male reproductive organs (Fig. 1F; figs. S2, O to U, and S7, B, E, and H; and movies S9 to 13). Supernumerary MMCs in rbr1-2 were also marked by KNU expression; there were often many more than four KNU-positive cells providing additional evidence that these cells are MMCs (Fig. 1I and movie S14). An MMC fate of these cells was further supported by their morphology and position in the young ovule primordia next to meiotic chromosome configuration observed in these cells (fig. S5F). Similar to the triple krp mutants, we found evidence for megaspore selection by cell death in later stages of ovule development (fig. S6, G, H, and I). Detailed morphological analyses showed that homozygous rbr1-2 mutants grown at the restrictive temperature had indeed both the previously described gametophytic and the here-revealed sporophytic defects; we observed multiple, separated embryo sacs in which the gametic cells over-proliferated (fig. S2U and movie S13). An increase in meiocytes was also observed in pollen (fig. S7, B, E, and H).

RBR1 is wired to the cell-proliferation machinery in Arabidopsis via the repression of F-BOX PROTEIN-LIKE17 (FBL17) that, in turn, mediates the SKIP-CULLIN-F-BOX (SCF)–dependent degradation of KRPs in Arabidopsis (8, 9, 25). If the meiocycte defects in rbr1-2 mutants are caused by a deregulated cell proliferation program, a reduction of CDKA;1 activity should rescue the rbr1-2 phenotype. However, when we analyzed the triple-mutant rbr1-2 cdka;1 PRO:CDKA;1:CDKA;1DE (short: rbr1 DE), we did not find a reduction in multiple MMC formation (Fig. 1P). This observation indicates that the failure to inactivate RBR1 through repression of CDKA;1 by KRPs primarily results in a differentiation failure rather than a proliferation defect.

RBR1 represses WUS to allow entry into meiosis

Rb is well known from studies in animals to control cell differentiation, including stem cell behavior (26, 27). Similarly, RBR1 is involved in differentiation processes in Arabidopsis apart from cell-cycle control yet is often associated with the regulation of cell proliferation (28). In Caenorhabditis elegans, entry into meiosis has been associated with the loss of stem cell fate (29). However, none of the developmental regulators found in stem cell differentiation in C. elegans, such as Notch, are present in Arabidopsis. Therefore, we searched for possible plant stem cell factors that could be targeted by RBR1. WUS promoter activity is down-regulated in MMCs, whereas the surrounding cells continue to express WUS (Fig. 3A and fig. S1A) (14). However, previous studies have shown that WUS protein can move in the shoot meristem (30), raising the question of whether a transcriptional regulation of WUS is relevant for WUS protein levels in the MMC. In contrast to the shoot meristem, we found that the translational fusion of WUS with GFP, which was mobile in the shoot, was only present in the cells surrounding MMCs (Fig. 3B). This result was corroborated with immunolocalization experiments showing that WUS protein was only detectable in the epidermal cell layer adjacent to, but never within, the MMC (Fig. 3C).

Fig. 3 RBR1 controls WUS expression.

(A) Confocal laser scanning micrograph of an ovule expressing a PROWUS:NLS-vYFP:3′WUS marker revealing WUS transcription in the epidermal cells of ovule primordia but not in the MMC of wild-type plants. (B) Confocal laser scanning micrograph of an ovule producing a genomic translational WUS protein reporter that shows a similar accumulation pattern in the wild type as the promoter marker shown in (A). (C to E) Light micrographs of WUS protein revealed by means of immunodetection. (C) WUS protein accumulates in the surrounding epidermal cells, but not in the MMCs in the wild type. [(D) and (E)] WUS protein ectopically accumulates in the multiple MMCs of krp4 krp6 krp7 and rbr1-2, respectively. (F) Diagram of genomic region of WUS. Solid boxes indicate exons; white boxes indicates intron; and W1, W2, and W3 represent the fragments used for ChIP in (G). (G) In ChIP experiments with RBR1, a fragment located about 200 bp upstream of WUS transcription start site is enriched while further upstream, and downstream-located fragments could not be amplified. The well-known RBR1 target PROLIFERATING CELL NUCLEAR ANTIGEN 1 (PCNA1) served as positive control (13). The heterochromatin region RB32.5 and two other genes, which are also expressed in epidermis of ovule primorida (At2G43150 and ARGONAUTE 5), were used as negative controls (13, 14). Error bars indicate SD; asterisks indicate statistically significant differences as revealed with Tukey’s multiple comparisons test between W1 IP versus mock (P < 0,0001) and PCNA1 IP versus mock (P < 0,0001). (H) Model of MMC specification. Developmental input (dev input) induces the up-regulation of KRPs in the progenitor MMC. Through the inhibition of CDKA;1, the inactivation of RBR1 is relieved. RBR1 represses cell proliferation through one branch that involves the degradation of KRPs (9). A second regulatory branch leads to the inactivation of WUS expression and allows differentiation of the MMC as shown here. Next to WUS, RBR1 represses additional yet unknown genes (X), which promote self-renewal and inhibit entry into meiosis. Direct regulatory interactions are indicated with a solid line, and indirect and/or currently unknown relationships are indicated with a dashed line. Scale bars, (A) and (B) 10 μm; (C) to (E) 20 μm.

Given that RBR1 functions as a transcriptional repressor and that previous studies have shown that RBR1 does also bind to promoters of target genes that are not directly involved in cell-cycle regulation (28), we scanned the WUS promoter by means of chromatin immunoprecipitation (ChIP) for RBR1 binding sites. We could detect that RBR1 specifically bound to a region of ~200 base pairs (bp) upstream of the WUS transcriptional start site, indicating that WUS is directly controlled by RBR1 (Fig. 3, F and G). RBR1 appeared to be necessary for WUS regulation because we observed by means of immunolocalization that WUS protein ectopically accumulates in the MMCs of rbr1-2 and krp4 krp6 krp7 triple mutants (Fig. 3, D and E).

To test the biological importance of this ectopic WUS expression, we introgressed a hypomorphic wus allele, wus-7, into rbr1-2 mutants. The rbr1-2 wus-7 homozygous double mutant showed a wus phenotype with a centrally located stamen instead of carpels that harbor the ovules. However, we observed a partial yet significant rescue of the mutant phenotype in plants that were heterozygous for wus-7 and homozygous for rbr1-2 (n = 900 ovules; Mann Whitney test, P = 0,0286) (Fig. 1, M and P). MMC formation in heterozygous wus-7 mutants itself is not affected (Fig. 1P). The observation that the inactivation of one WUS copy could partially revert the multiple MMC phenotype indicates a strong WUS-dosage sensitivity of rbr1-2 mutants and supports a direct interaction between RBR1 and WUS. To corroborate this finding, we combined rbr1-2 with another wus allele (wus-101). Similar to the wus-7 combination, the wus mutant phenotype was epistatic in homozygous double-mutant plants. Although weaker than in combination with wus-7, formation of supernumerary MMCs was significantly reduced in homozygous rbr1-2 mutants when wus-101 was heterozygous (n = 161 ovules; Mann Whitney test, P = 0,0286) (Fig. 1P). Thus, we conclude that down-regulation of WUS is key for entry into meiosis in Arabidopsis (Fig. 3H). Next, we asked whether WUS expression is also sufficient to maintain a mitotic fate in designated MMCs. To this end, we generated a gene fusion in which the open reading frame of WUS is fused to both the red fluorescent protein (RFP) (mCherry) and the Glucocorticoid receptor (GR); this fusion is then expressed from the ubiquitin promoter. After induction of WUS activity by applying the synthetic glucocorticoid Dexamethasone to immature flower buds, we found that WUS is relocated from the cytoplasm to the nucleus in MMCs (fig. S8, A to F). However, no multiple MMC phenotype was observed. In addition, we exchanged the ubiquitin promoter with the KNU and the DMC1 promoter to drive mCherry:GR:WUS expression more specifically in the MMC. However, after induction with Dexamethasone, we did not see the formation of multiple MMCs (PROKNU:mCherry:GR:WUS, n = 168 ovules analyzed, percent multiple MMC 11.3; PRODMC1:mCherry:GR:WUS, n = 122 ovules analyzed, percent multiple MMC 13.1) (Fig. 1P). Thus, we conclude that RBR1 is the central gatekeeper of meiocyte differentiation and the regulatory pathway emanating from RBR1 branches off downstream of it. One important function of RBR1 is the down-regulation of WUS as a key factor for entry into meiosis in Arabidopsis. WUS repression is necessary for MMC formation (Fig. 3H).

Conclusions

Here, we have shown that there is a delicate balance between WUS activation important for ovule primordia formation—including the development of the integuments, as well as a role in specifying the MMC itself (5, 6)—and its inactivation by RBR1 soon afterward to allow entry into meiosis (Fig. 3H). However, the observation that krp triple-mutant MMCs eventually entered meiosis indicates that the RBR1-based repression of WUS is backed up by other control pathways that either overrule WUS activity and/or cause a drop in WUS protein levels. Conversely, we found that WUS misexpression in designated MMC is not sufficient to induce mitotic instead of meiotic divisions. Thus, additional factors are required to facilitate the entry into meiosis downstream of RBR1.

The repression by RBR1 combines cell proliferation activity with differentiation steps. This also suggests that CDKA;1 activity is initially very low in meiocytes to prevent the premature inactivation of RBR1. Premeiotic S phases are usually very long in both animals and plants (31, 32), a feature that has not been functionally understood but would be consistent with a low CDK activity in S phase.

In animals, different components of the Rb control pathway have been associated with germ cell fate initiation; for example, mutants in the CDK inhibitor dacapo in Drosophila fail to enter meiosis (33). Similarly, down-regulation of Cdk2-cyclin E, a well-known regulator of Rb, is important for C. elegans germline development (3436). This raises the intriguing question of whether Rb family proteins play a conserved role in germline entry in multicellular organisms.

Materials and methods

Plant material and growth conditions

Arabidopsis thaliana (L.) Heynh. plants were all derived from the Columbia (Col-0) accession except for the hypomorphic wus-7 mutant allele which is in the Landsberg erecta (Ler) background. KRP T-DNA insertion lines krp1 (At2g23430, Salk_100189), krp2 (At3g50630, Salk_068815), krp3 (At5g48820, WsDSLox49707H), krp4 (At2g32710, Sail_248_B06), krp5 (At3g24810, Salk_053533), krp6 (At3g19150, Sail_548_B03), krp7 (At1g49620, GK_841D12), and the temperature sensitive rbr1-2 mutant line (At3g12280, Salk_002946), have been described previously (9, 20, 37). The KNU and WUS promoter reporter lines, the RBR1 reporter line and the hypomorphic cdka;1 mutant DE (cdka;1−/− PRO:CDKA;1:CDKA;1DE) were also described previously (15, 22, 38). The hypomorphic wus-7 allele was isolated from EMS-mutagenized populations as described (39). The amorphic wus-101 allele is a T-DNA insertion line obtained from the Gabi-Kat collection (Gabi Kat-870H12; NASC ID: N349353) (40). The central cell (DD19) and egg cell marker (DD45) lines were described by Steffen and colleagues (2007) (41). All genotypes were determined by polymerase chain reaction (PCR) with the primers indicated in Table S1. All seeds were surface sterilized with chloride gas, sown on 0.8% Phytoagar plates (half-strength Murashige and Skoog (MS) salts, 1% sucrose), and grown under day-neutral conditions (12 hours light at 21°C, and 12 hours dark at 18°C). After germination, plants were transferred to soil and grown under long-day conditions (16 hours day/8 hours night regime at 22°C/18°C). The temperature-sensitive homozygous rbr1-2 mutant together with the comparative material were grown at constant 17°C under long- day conditions. For all crosses, flowers of the female parent were emasculated 2 days before anthesis and hand-pollinated 2 days later.

Expression constructs

To generate the PROKNU:KRP7 construct, a 2,001-bp genomic fragment containing the putative KNU promoter was amplified with the primers pKNU-F1 and pKNU-R1. The full-length KRP7 cDNA was amplified with primers KRP7-F1 and KRP7- R1. A pENTR2B vector was linearized by PCR with the primers pENTR2B-F1 and pENTR2B-R1. These three PCR products were assembled by SLiCE (42), followed by LR recombination reaction with the destination vector pGWB501 (43). To generate the PROKNU:mEGFP:KRP7 construct, a monomeric enhanced GFP (mEGFP) fragment was amplified with the primers mEGFP-F and mEGFP-R. A pENTR2B vector with PROKNU:KRP7 fragment was linearized by PCR with the primers pENTR2B-F2 and pENTR2B-R2. These two PCR products were assembled by SLiCE, followed by LR recombination reaction with the destination vector pGWB501. To generate the PROKNU:WUS:GR construct, the full-length WUS cDNA fused with GR fragment was amplified from p35S:WUS:GR plasmid (44) with the primers WUS-GR-F and WUS-GR-R. A pENTR2B vector with PROKNU fragment was linearized by PCR with the primers pENTR2B-F3 and pENTR2B-R3. These two PCR products were assembled by SLiCE, followed by LR recombination reaction with the destination vector pGWB501. To generate the PROKNU:mCherry:GR:WUS construct, the putative KNU promoter was amplified with the primers pKNU-F2 and pKNU-R2. The pJF359 vector with mCherry:GR:WUS fragment was linearized by PCR with the primers pJF359-F1 and pJF359-R1. These two PCR products were assembled by SLiCE to create the PROKNU:mCherry:GR:WUS fragment. Next, the PROKNU:mCherry:GR:WUS fragment was amplified with the primers attB1-F1 and attB2-R, and was inserted into the entry vector pDONR221 by BP recombination reaction and then into the destination vector pGWB501 by LR recombination reaction. To generate the PRODMC1:KRP7 construct, a 3,254-bp genomic fragment containing the putative DMC1 promoter was amplified with the primers pDMC1-F1 and pDMC1-R1. A pENTR2B vector with a KRP7 fragment was linearized by PCR with the primers pENTR2B-F4 and pENTR2B-R4. These two PCR products were assembled by SLiCE, followed by LR recombination reaction with the destination vector pGWB501. To generate the PRODMC1:mEGFP:KRP7 construct, a 3,254-bp genomic fragment containing the putative DMC1 promoter was amplified with the primers pDMC1-F1 and pDMC1-R2. A pENTR2B vector with a mEGFP-KRP7 fragment was linearized by PCR with the primers pENTR2B-F5 and pENTR2B-R4. These two PCR products were assembled by SLiCE, followed by LR recombination reaction with the destination vector pGWB501. For the PROUBQ10:GR:WUS construct, plasmid pJF359 (pUBQ10:mCherry-GR-linker-WUS:TrbcS) was created by a Gateway LR reaction of the pGreenIIS-based (45, 46) destination vector pFK273 providing the regulatory sequences as well as a BastaTM (glufosinate-ammonium) resistance cassette and the pENTR1A-based entry vector pJF355 contributing the coding sequence. A respective A. thaliana line was established by floral dip transformation of Col-0 wild- type plants followed by selection of T1 plants on soil soaked with 20 mg/L BastaTM (Bayer CropScience Deutschland GmbH, Langenfeld, Germany) water. T2 seeds from individual T1 plants were then sown out on half-strength MS-plates supplemented with 10 mg/L glufosinate-ammonium (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany), 0.5% v/v ethanol from the solvent and with or without 25 μM dexamethasone, respectively. Line #14 displayed uniform strong growth inhibition upon ectopic WUS induction by dexamethason and normal viability without. The T2 seedlings of line #14 also showed a roughly 3:1 segregation ratio of the BastaTM resistance indicating a single transgene insertion and the descendants of two individual homozygous T2 plants were used for all further experiments. To generate the PRODMC1:WUS:GR construct, the putative DMC1 promoter was amplified with the primers pDMC1-F1 and pDMC1-R3. A pENTR2B vector with WUS:GR fragment was linearized by PCR with the primers pENTR2B-F6 and pENTR2B-R4. These two PCR products were assembled by SLiCE, followed by LR recombination reaction with the destination vector pGWB501. To generate the PRODMC1:mCherry:GR:WUS construct, the putative DMC1 promoter was amplified with the primers pDMC1-F2 and pDMC1-R4. A pJF359 vector with mCherry- GR-WUS fragment was linearized by PCR with the primers pJF359-F2 and pJF359-R2. These two PCR products were assembled by SLiCE to create the PRODMC1:mCherry:GR:WUS fragment. Next, the PRODMC1:mCherry:GR:WUS fragment was amplified with the primers attB1-F2 and attB2-R, and was inserted into the entry vector pDONR221 by BP recombination reaction and then into the destination vector pGWB501 by LR recombination reaction.

Microscopy

For fluorescence microscopy analyses, ovules at different developmental stages were dissected from pistils and mounted on microscope slides with tap water. GFP fluorescence of the ovules was analyzed by confocal microscopes (Zeiss 710 and Zeiss 780), using a BP 530-600 filter. Differential Interference Contrast (DIC) microscopy was performed on an Axioimager (Zeiss). For DIC microscopy, mature ovules and developing seeds were prepared from siliques just before and 3 days after pollination (DAP), respectively, mounted on microscope slides in a clearing solution (8:2:1 chloral hydrate:distilled water:glycerol), and kept overnight at 4°C for analysis. For meiotic figure analysis of MMCs, flower buds at the right developmental stage were fixed in Carnoy’s fixation (3:1 absolute ethanol:glacial acetic acid) overnight at 4°C. The pistils (sepal, petal, and stamen were removed from the flower buds) were dehydrated in an ethanol series from 70% to 100% ethanol and cleared by methyl benzoate, ovules were dissected from pistils and mounted on microscope slides, containing a drop of DAPI (4',6-diamidino-2-phenylindole) solution mixed with VECTASHIELD mounting medium according to the manufacturer’s instructions (Vector Laboratories, Inc. Burlingame, CA94010), and observed under the Zeiss confocal microscope.

Confocal laser scanning microscopy of megaspore mother cells and embryo sacs

Stage 10–11 flowers (47) were dissected on double-sided tape. The placentas bearing the developing ovule primordia were excised and placed immediately in fixative (4% (v/v) glutaraldehyde, 12.5 mM cacodylate buffer pH 6.8, modified from Christensen et al. (1997) (48). After two hours the fixative was removed and the placentas were mounted onto microscope slides in a clearing solution (8:2:1 chloral hydrate:distilled water:glycerol). During mounting the two placentral strands were separated and excess tissue was removed. Ovule primordia were observed under an AxioObserver microscope coupled to an LSM710 scanner (Zeiss). To view autofluorescence, excitation was set at 488 nm and fluorescence was detected in the 500 to 600 nm range. Pictures were collected with a 63x/1.2 W Corr C-Apochromat objective at a zoom factor of 2.

Flow cytometry analysis

Flower buds from different genotypes were chopped in 200μl of nuclei extraction buffer, and supplemented with 800μl of staining buffer (Cystain UV Precise P kit, Partec). The suspension was filtered through a 30μm mesh (Green filter, Partec) and analyzed with a Cyflow MB flow cytometer (Partec). Three biological replicates were performed with minimum three measurements for a single biological replicate of each genotype.

Chromatin immunoprecipitation

The ChIP experiments were performed as previously described (9, 49). The inflorescence of PRORBR1:RBR1:mRFP was used as ChIP material.

Immunostaining of megaspore mother cells

The immunostaining procedure was previously described (50).

Kinase assay

Kinase assays were performed as previously described (51). p9CKShs1-associated proteins purified from 300 μg total proteins extracted from inflorescences of wild-type and krp4 krp6 krp7 mutant plants were processed for kinase assays with Histone H10 (NEB) as a substrate. CDKA;1 bound to p9CKShs1-beads was detected by protein blotting with an anti-PSTAIR antibody (Sigma).

Expression analyses

In situ hybridization and capturing of pictures was performed as previously described (7). Briefly, in situ hybridization was performed on 8 μm thin sections of inflorescences or buds from wild-type and krp4 krp6 krp7 mutant plants using KRP4, KRP6, and MEM (7) antisense and sense probes. Heatmaps were based on log2- transformed mean expression values generated by dCHIP as previously described (52).

Pollen mother cell observation and quantification

Inflorescences were collected from plants grown under greenhouse conditions, all opened flowers and largest closed flower buds were removed, fixed overnight in FAA (3,2% Formaldehyde, 5% Acetic acid, 50% EtOH), washed in PBS (2 × 30min), and dehydrated in 50% and 70% EtOH (both 2 × 30min), and 96% EtOH until complete bleaching, followed by staining overnight with a few drops of 1% eosin in 96% EtOH (4°C). The tissue was then washed (3 × 40min) in 100% EtOH and transferred to histoclear through a graded series of EtOH/histoclear of 3v/v; v/v; v/3v (15min each), followed by 3 × 30min 100% histoclear, and thereafter by histoclear/paraplast (v/v) for 2 hours at 60°C. This was followed by 48 hours of washes in paraplast (3 changes a day), followed by embedding and sectioning in 15μm sections. Flowers for male meiocyte (pollen mother cells [PMC]) counts (5 inflorescences per genotype, one flower per inflorescence) were selected by first identifying the flower that showed the latest meiotic stage (either tetrad stage PMCs or separating microspores). Subsequently, from among all smaller flowers in the inflorescence, the largest flower was chosen, which typically had progressed furthest in PMC development. PMCs were counted in three consecutive transverse cross sections from the middle of each of the four pollen sacs. Per genotype, pollen was counted in all four pollen sacs of the four largest anthers in five flowers per genotype, i.e., 5 flowers, 20 anthers, 40 inner and 40 outer pollen sacs per genotype.

Analysis of fertilization in the krp4 krp6 krp7 triple mutant

The krp4 krp6 krp7 triple mutant carrying the PRODD19:DD19:GFP gene was pollinated with pollen from the PRORPS5A:H2B:tdTomato nuclear marker line (53). After 14 HAP, ovules were dissected from the pistils into half-strength Murashige & Skoog’s medium (5% sucrose, adjusted pH to 5.7 with 1 M KOH), and observed by a confocal laser scanning system CSU-X1 (Yokogawa) mounted on an IX-81 microscope (Olympus) at 488 nm and 561 nm excitations. Then, samples were stained on the stage of the microscope with 0.4% Congo red for a few minutes and washed out by half-strength Murashige & Skoog’s medium, and pollen tubes were observed at 561 nm excitation. Semi-in vitro fertilization assay was modified from a previously described protocol by Kawashima et al. (2014) (54). A thin layer of pollen tube growth medium, developed in Sheila McCormick’s laboratory (0.01% boric acid, 5 mM CaCl2, 5 mM KCl, 1 mM MgSO, 10% sucrose pH7.5, 1.5% Nusieve GTG agarose) (55), was prepared in a well of a glass bottom dish (D141410; Matsunami Glass IND., LTD., Japan). Stigmas with styles were removed from emasculated wild-type pistils by a 27-gauge needle (Terumo, Japan) and placed on the medium, then, ~100 pollen grains from plants containing the PRORPS5A:H2B:tdTomato construct were attached on each stigma. After 3 hours incubation at 22°C in the dark, ovules from the krp4 krp6 krp7 triple mutant carrying the PRODD19:DD19:GFP were placed in front of growing pollen tubes on the medium, and incubated for further 2 hours. Time-lapse images were obtained by the CSU-X1 confocal laser scanning system.

Analysis of pollen tube guidance by the aniline blue staining

Pistils were harvested two DAP and subjected to aniline blue staining as described previously (53). Samples were analyzed by a BX51 fluorescence microscope (Olympus) equipped with the U-MWU2 fluorescence mirror unit and a cooled CCD camera DP71 (Olympus).

Supplementary Materials

www.sciencemag.org/content/356/6336/eaaf6532/suppl/DC1

Materials and Methods

Figs. S1 to S8

Table S1

Movies S1 to S14

References and Notes

  1. Acknowledgments: The authors thank M. Heese and P. Bommert for critically reading the manuscript. We thank the Arabidopsis Biological Resource Center (ABRC) and the European Arabidopsis Stock Centre (NASC) for providing seeds of T-DNA lines used in this report. This work was supported by an European Molecular Biology Organization Long-term fellowship (E.W.), an Innovation by Science and Technology (121110) Ph.D. fellowship (M.V.D), the EU Innovative Training Networks grant “COMREC” to M.A.P. and A.S., grants from the “Staatssekretariat für Bildung und Forschung“ in the framework of COST Action FA0903 (to A.S. and U.G.) and the Swiss National Science Foundation (to U.G.), the European Research Council Starting Grant “StemCellAdapt” (282139) (to J.U.L.), a grant from the Deutsche Forschungsgemeinschaft (La606/6, La606/13) (to T.L.), an EU-cofunded INTERREG IV Upper Rhine Project 692 A17 (to A.S. and T.L.), a Fonds Wetenschappelijk Onderzoek (G033711N) research grant (to M.K.N.), a grant from the L’Agence Nationale de la Recherche (ARABIDOSTART) (to A.S.), the European Research Council Starting Grant “Seeds for life” (to A.S.), and core support of the University of Hamburg (to A.S.). The supplementary materials contain additional data.

Stay Connected to Science

Navigate This Article