AGL61 Interacts With AGL80 and Is Required for Central Cell Development in Arabidopsis

The central cell of the female gametophyte plays a role in pollen tube guidance and in regulating the initiation of endosperm development. Following fertilization, the central cell gives rise to the seed’s endosperm, which nourishes the developing embryo within the seed. The molecular mechanisms controlling specification and differentiation of the central cell are poorly understood. We identified AGL61 in a screen for transcription factor genes expressed in the female gametophyte. AGL61 encodes a Type I MADS domain protein, which likely functions as a transcription factor. Consistent with this, an AGL61-GFP fusion protein is localized to the nucleus. In the context of the ovule and seed, AGL61 is expressed exclusively in the central cell and early endosperm. agl61 female gametophytes are affected in the central cell specifically. The morphological defects include an overall reduction in size of the central cell and a reduced or absent central cell vacuole. When fertilized with wild-type pollen, agl61 central cells fail to give rise to endosperm. In addition, synergid- and antipodal-expressed genes are ectopically expressed in agl61 central cells. The expression pattern and mutant phenotype of AGL61 are similar to those of AGL80 , suggesting that AGL61 may function as a heterodimer with AGL80 within the central cell; consistent with this, AGL61 and AGL80 interact in yeast. Together, these data suggest that AGL61 functions as a transcription factor and controls the expression of downstream genes during central cell development. , which encodes a Type I MADS domain protein. We show that AGL61 is expressed exclusively in the central cell and endosperm during ovule and seed development, that agl61 mutants have central cell defects similar to those of agl80 , and that AGL61 interacts with AGL80 in yeast. Together, these results suggest that an AGL61-AGL80 heterodimer functions in the central cell to control the expression of downstream genes that are critical for central Growth occurs only when cells contain both AGL80-BD and AGL61-AD (row 1) or both AGL61-BD and AGL80-AD (row 2). Cells containing AGL80-BD only (row 3), AGL80-AD only (row 4), AGL61-BD only (row 5), AGL61-AD only (row 6), or neither AGL61 nor AGL80 (row 7) do not grow.


INTRODUCTION
The central cell of the female gametophyte is critical for several steps of the angiosperm fertilization process. During the late stages of pollen tube growth, a pollen tube grows along the carpel's placental surface, onto the ovule's funiculus, and finally into the ovule's micropyle to reach the female gametophyte. Soon after entering the female gametophyte, the pollen tube releases its two sperm cells to effect double fertilization of the egg cell and central cell, which give rise to the seed's embryo and endosperm, respectively. Endosperm is an important component of the seed because it provides nutrients and other factors to the embryo during seed development and/or to the developing seedling following germination (reviewed in Drews and Yadegari, 2002;Yadegari and Drews, 2004).  (Luo et al., 1999), MEDEA (MEA) (Grossniklaus et al., 1998;Kiyosue et al., 1999;Luo et al., 1999), MULTICOPY SUPPRESSOR OF IRA1 (MSI1) (Kohler et al., 2003;Guitton et al., 2004), and SWINGER (SWN) (Wang et al., 2006). which the FIS proteins form a complex that represses genes involved in endosperm development within the central cell (reviewed in Curtis and Grossniklaus, 2008).

The central cell forms during megagametogenesis. Most species including
Arabidopsis and cereals undergo the Polygonum pattern of megagametogenesis.
During Polygonum-type megagametogenesis, a one-nucleate megaspore undergoes two rounds of mitosis, producing a four-nucleate cell. During a third round of mitosis, phragmoplasts and cell plates form between nuclei, initiating the cellularization process.
Ultimately, the nuclei become completely surrounded by cell walls, resulting in formation of a seven-celled female gametophyte consisting of one central cell, one egg cell, two synergid cells, and three antipodal cells. The central cell inherits two nuclei, the polar nuclei. In Arabidopsis and many other species, the polar nuclei fuse to form the diploid central cell nucleus (secondary nucleus) (reviewed in Willemse and van Went, 1984;Huang and Russell, 1992;Yadegari and Drews, 2004).
Little is known about the regulatory processes controlling central cell development and few transcriptional regulators functioning in this cell have been identified. Those identified include the FIS genes discussed above, as well as AGL80 (Portereiko et al., 2006) and DEMETER (DME) (Choi et al., 2002). DME encodes a DNA glycosylase required for the activation of FIS2, FWA, and MEA expression in the central cell and endosperm (Choi et al., 2002;Jullien et al., 2006 6 cell and endosperm development.

AGL61 is Expressed in the Central Cell
We performed a screen to identify MADS box genes expressed in the female gametophyte. We harvested ovaries from male sterility1 (ms1) (Thorlby et al., 1997;Wilson et al., 2001;Ito and Shinozaki, 2002) and determinant infertile1 (dif1) (Bai et al., 1999;Bhatt et al., 1999;Cai et al., 2003), extracted RNA, and used real-time RT-PCR to assay the expression of genes within this gene family. ms1 ovules are normal and dif1 ovules lack female gametophytes (Steffen et al., 2007); thus, genes exhibiting reduced expression in dif1 ovaries relative to ms1 ovaries are likely to be expressed in the female gametophyte.
These assays identified a gene, AGL61, exhibiting reduced expression in dif1 ovaries relative to wild-type ovaries. The structure of AGL61 is summarized in Figure 1 and the real-time RT-PCR data are provided in Figure 2A.
To determine which cells within the female gametophyte express AGL61, we generated and analyzed transgenic Arabidopsis plants containing a protein-fusion construct, AGL61-GFP, comprising the AGL61 promoter and the entire AGL61 coding region fused with a green fluorescent protein (GFP) coding sequence. Figures 3A to 3C show AGL61-GFP expression during female gametophyte development (female gametophyte stages are described in Christensen et al., 1997). AGL61-GFP expression was first detected in the two polar nuclei just before fusion (late stage FG5; Figure 3A).

AGL61-GFP expression was not detected at earlier developmental stages. Expression
in the central cell continued through stage FG6 ( Figure 3B) and into the mature stage (stage FG7; Figure 3C). During all of these stages, the AGL61-GFP fusion protein was localized to the nucleus, consistent with a predicted function in transcriptional regulation.

7
To determine whether AGL61 is also expressed in developing seeds, we analyzed AGL61-GFP expression at 12 -48 hours after pollination. During this period, AGL61-GFP expression was detected exclusively in the endosperm ( Figure 3D). During endosperm development, AGL61-GFP expression was strongest immediately after fertilization, diminished gradually at progressively older stages, and was not detected after the 8-nucleate stage (stage IV) of endosperm development (endosperm stages are described in Boisnard-Lorig et al., 2001). In reciprocal crosses with plants homozygous for the AGL61-GFP construct and wild type, expression was detected only when the reporter construct was present in the female parent.
We also analyzed expression of an AGL61 promoter-fusion construct, To determine whether AGL61 is expressed elsewhere in the plant, we performed real-time RT-PCR with RNA from various organs. The results from these assays are shown in Figure 2B. Consistent with expression of AGL61-GFP and ProAGL61:GFP in the female gametophyte, strong AGL61 expression was detected in ovaries. In addition, weak expression was detected in siliques, which correlates with limited AGL61-GFP and ProAGL61:GFP expression during seed development, and in stems and anthers.
Expression was not detected by real-time RT-PCR in roots, leaves, and young flowers ( Figure 2B).
In summary, during ovule and seed development, AGL61 is expressed exclusively in the central cell and endosperm, from late stage FG5 (just after central cell cellularization and before the polar nuclei fuse) of female gametophyte development to stage IV (8-nucleate stage) or V (16-nucleate stage) of endosperm development.
Elsewhere in the plant, AGL61 expression is extremely low or is not detected. 8

Mutations in AGL61
To determine whether mutations in AGL61 affect the female gametophyte, we analyzed lines containing T-DNA insertions in this gene. We analyzed two T-DNA alleles, agl61-1 (SALK_009008) and agl61-2 (GABI-Kat 642H10), which were obtained from the Arabidopsis SIGnAL (Alonso et al., 2003) and GABI-Kat (Rosso et al., 2003) collections, respectively. The T-DNA insertion sites in these mutants are shown in Figure 1A.
To determine whether the agl61 mutations affect the female gametophyte, we crossed heterozygous mutant plants as females with wild-type males and scored the number of AGL61/AGL61 and agl61/AGL61 progeny. Table 1 shows that both mutations exhibited reduced transmission through the female gametophyte, indicating that they affect the female gametophyte.
To determine whether the agl61 mutations also affect the male gametophyte, we crossed heterozygous mutant plants as male parents with wild-type females and scored the number AGL61/AGL61 and agl61/AGL61 progeny. With both alleles, homozygous wild-type and heterozygous progeny were present in approximately equal proportions (Table 1), indicating that these mutations do not affect the male gametophyte. Table 1 shows that the agl61 mutations transmit through the female gametophyte at low frequency. Based on the observed gametophytic transmission frequencies (Table   1), homozygous mutants should be present at a frequency of 1.1% -2.5%. However, homozygous mutants were not identified in >800 plants screened for each allele. These results along with the AGL61 expression pattern suggest that the agl61 mutations affect seed development.

Molecular Complementation of the agl61-1 Mutation
To confirm that the female gametophyte defect is due to disruption of AGL61, we introduced a wild-type copy of this gene into the agl61-1 mutant. We identified plants heterozygous for the agl61-1 allele and hemizygous for the rescue construct; these plants contained 25% aborted seeds, as compared to 50% aborted seeds for agl61-1 plants lacking the rescue construct. In the subsequent generation, we identified plants 9 heterozygous for the agl61-1 allele and homozygous for the rescue construct; these plants had full seed set. Together, these data indicate that disruption of the AGL61 gene is responsible for the female gametophyte defect in agl61-1 mutants.
We first analyzed female gametophytes at the terminal developmental stage  Figure   4C).
To determine whether agl61-1 female gametophytes are affected at earlier developmental stages, we analyzed female gametophytes (n = 59) within stage 12c flowers, which contain embryo sacs at stages FG4 to FG6 (Christensen et al., 1997). In flowers at this stage, abnormal female gametophytes were not observed, suggesting that agl61-1 female gametophytes do not exhibit defects at these earlier stages.
To characterize endosperm derived from fertilization of agl61 central cells, we pollinated agl61-1/AGL61 flowers with wild-type pollen, waited 24 hours, and fixed seed tissue for confocal analysis. In the siliques resulting from this cross, ~50% (51/95) of the seeds were normal and ~50% (44/95) were abnormal, suggesting that that the abnormal seeds resulted from fertilization of agl61-1 embryo sacs. In wild-type seeds at 24 hours after pollination, one of the synergid cells is degenerated, the embryo is a single-celled zygote and the endosperm typically consists of four to eight nuclei ( Figure 4D). In most (84%, 37/44) of the abnormal seeds, the embryo sac chamber was filled with highly autofluorescent material ( Figure 4E). A minority (16%, 7/44) of abnormal seeds had a few endosperm nuclei at abnormal positions ( Figure 4F) and a zygote-like structure ( Figure 4G).
To further characterize the defects in agl61-1, we used fluorescence microscopy to analyze development of GFP-marked central cells and endosperm. We analyzed plants heterozygous for the agl61-1 mutation and hemizygous for ProAGL61:GFP, which is expressed in agl61-1 central cells and endosperm (discussed below). In mature female gametophytes (stage FG7), defective central cells were readily apparent. Of the female gametophytes expressing GFP, ~50% (31/63) contained abnormal central cells that resembled those described above: the central cell vacuole was reduced in size or absent and the overall size of the central cell was dramatically reduced ( Figure 3G). At 24 hours after pollination with wild-type pollen, ~50% (51/108) of the seeds were defective and most of these had no endosperm ( Figure 3H).
In summary, agl61 female gametophytes are defective in central cell development. agl61 central cells are reduced in size and have collapsed vacuoles, but appear to be viable, based on expression of a central cell marker. Fertilization of agl61 female gametophytes with wild-type sperm leads to aberrant endosperm development and eventually seed abortion.

agl61 Central Cells Express Synergid and Antipodal Markers
The CLSM analysis discussed above suggests that the egg cell, synergid cells, and antipodal cells are not affected in agl61 female gametophytes. To investigate this issue further, we analyzed expression of markers for these cell types in agl61 embryo sacs.
We analyzed expression of ProDD1:GFP, which is expressed exclusively in the antipodal cells ( Figure 3I), and ProDD3:GFP, which is expressed strongly in the synergid cells and weakly in the egg cell and central cell ( Figure 3K) (Steffen et al., 2007).
In agl61-1 female gametophytes, ProDD1:GFP was expressed in the antipodal cells ( Figure 3J) and ProDD3:GFP was expressed strongly in the synergid cells and weakly in the egg cell ( Figure 3L). These results suggest that the antipodal, synergid, and egg cells are normal in agl61-1 embryo sacs. However, in contrast to wild type, ProDD1:GFP was also expressed in the central cell of agl61-1 embryo sacs ( Figure 3J).
Similarly, ProDD3:GFP, which was expressed weakly in wild-type central cells ( Figure   3K), was expressed strongly in agl61-1 central cells ( Figure 3L). These data indicate that AGL61 is required for suppression of DD1 and DD3 expression in the central cell and that an additional aspect of the agl61 phenotype is mis-expression of antipodal-and synergid-expressed genes.

agl61 Female Gametophyte Attract Pollen Tubes
Analysis of the ccg mutant suggests that the central cell is required for pollen tube guidance by the female gametophyte (Chen et al., 2007). However, the CLSM analysis of developing seeds discussed above suggests that agl61-1 female gametophytes attract pollen tubes and become fertilized. To confirm these results, we analyzed pollen tube growth to agl61 female gametophytes. We observed pollen tubes using pollen from (115/119) of seeds contained a pollen tube in the micropyle and a GFP bolus in the embryo sac, indicating that agl61-1 female gametophytes can attract pollen tubes.
To verify these observations, we performed a similar analysis with central cells expressing ProAGL61:GFP, which allowed us to directly observe mutant embryo sacs (discussed above). At 24 hours after pollination with ProLAT52:GFP pollen, 100% (35/35) of agl61-1 female gametophytes had a pollen tube in its micropyle and a GFP bolus in the embryo sac ( Figures 3N). Together, these data indicate that agl61 female gametophytes are not defective in pollen tube guidance.

AGL61 is Not Autoregulated
Autoregulation is a common feature of MADS box genes (de Folter and Angenent, 2006 Furthermore, the intensity of the GFP signal was approximately equal in wild-type and agl61-1 female gametophytes. Together, these data suggest that AGL61 does not regulate its own expression.

AGL61 Interacts With AGL80
The phenotype of agl61 female gametophytes resembles that of agl80 female gametophytes and the two genes are expressed in a similar pattern (Portereiko et al., 2006), suggesting that AGL61 may interact with AGL80 in the central cell. 13 (AGL61-AD) and full-length AGL80 fused with these domains (AGL80-BD and AGL80-AD). Figure 5 shows that AGL61-BD and AGL61-AD interacted with AGL80-AD and AGL80-BD, respectively, to stimulate transcription of the HIS3 and ADE2 reporter genes. By contrast, control cells containing constructs paired with empty vectors did not activate transcription of the reporter genes. These data indicate that AGL61 interacts with AGL80 in yeast. seed development, suggesting that other Type I genes may also function during these developmental stages.

AGL61 is Required for Central Cell Development
During ovule development, AGL61 is expressed exclusively in the central cell ( Figures   3A-3C). This expression pattern is consistent with the phenotype of agl61 mutants. An additional aspect of the agl61 central cell phenotype is ectopic expression of synergid-and antipodal-expressed genes ( Figures 3J and 3L). These observations indicate that AGL61 is required to suppress the expression of genes in the central cell.
Of two genes tested, both are mis-expressed, suggesting that additional genes are misexpressed in agl61 central cells. Despite the strong morphological defects in the central cell, agl61 female gametophytes are able to attract pollen tubes ( Figures 3N). This is also true of agl80 female gametophytes (Portereiko et al., 2006). These observations are in contrast to those of the ccg mutant, which has subtle or no defects in the central cell but is defective in pollen tube guidance (Chen et al., 2007). These results suggest that the agl61 mutation does not affect CCG expression and production of the central cell factors required for pollen tube guidance.

MADS-domain proteins generally function as homodimers and/or as heterodimers with
other MADS-domain proteins (de Folter and Angenent, 2006). Consistent with this, we have shown that AGL61 interacts with AGL80 in yeast. In a recent study, an interactome map of the Arabidopsis MADS-domain proteins was generated (de Folter et al., 2005). In this study, the AGL61-AGL80 interaction was not reported. In progress are experiments to verify that AGL61 and AGL80 interact in vivo.
Our results suggest that an AGL61-AGL80 heterodimer functions in the central

Plant Transformation
T-DNA constructs were introduced into Agrobacterium strain LBA4404 by electroporation. Arabidopsis plants (ecotype Columbia) were transformed using a modified floral dip procedure (Clough and Bent, 1998). Transformed progeny were selected by germinating surface-sterilized T1 seeds on growth medium containing antibiotics. Resistant seedlings were transplanted to soil after 10 days of growth. For plant-wide real-time RT-PCR, we carried out the experiments and analysis as described in Steffen et al. (2007). Tissue was harvested from plants and placed immediately into liquid nitrogen. Ovaries were harvested from ms1 and dif1 at flower stages 12c (Christensen et al., 1997) and13 (Smyth et al., 1990). Floral cluster tissue includes the inflorescence meristem and flowers at stages 1-10 (Smyth et al., 1990).

Real-Time RT-PCR
Silique tissue includes siliques at 1-2 days after pollination. Leaf tissue includes leaves of sizes 5-12 mm. Roots were harvested from seedlings at 11 days after germination.
We calculated relative expression levels as follows. We first normalized AGL61 transcript levels relative to a standard (ACTIN2) using the formula ∆C T = C T (AGL61) -C T (ACTIN2). We next calculated an average ∆C T value for each tissue. ms1 pistil tissue with the highest relative expression (lowest ∆C T value), was used as the standard for comparison of expression levels. We then calculated relative expression levels using the equation 2 -(average ∆CT (tissue) -average ∆CT (ms1 pistil) .
These constructs were introduced into Arabidopsis plants as described above and transformed plants were selected by germinating T1 seeds on growth medium containing 30 µg/ml kanamycin. The expression patterns reported in the Results are derived from the analysis of at least ten transgenic lines.

Analysis of GFP Expression Patterns
For analysis of mature female gametophytes, we emasculated flowers at stage 12c (Christensen et al., 1997), waited 24 hours, and removed the flowers from the plants.
We then removed the sepals, petals, and stamens, and dissected off the carpel walls using a 30-gauge syringe needle. For analysis of earlier developmental stages, we directly dissected the ovules from stage 12c flowers. For analysis of developing seeds, we emasculated flowers at stage 12c, waited 24 hours, pollinated with self-pollen, waited 12-48 hours, and then dissected the tissue as described above. In all cases, the dissected ovules/seeds were mounted on microscope slides in 10 mM phosphate buffer (pH 7.0) for microscopic analysis. GFP expression patterns were analyzed using a Zeiss Axioplan microscope. GFP was excited using a UV lamp and was detected using a 38 HE EGFP filter set. Images were captured using an AXIOCAM MRM REV2 camera with the AxioVision software package version 4.5 (Zeiss). The T-DNA in agl61-1 is inserted 81 nucleotides upstream of the transcriptional start site, which is 130 nucleotides upstream of the predicted start codon, and is associated with a 17-nucleotide deletion (nucleotides -71 to -65 relative to the transcriptional start site deleted). The T-DNA in agl61-2 is inserted 48 nucleotides downstream of the transcriptional start site, which is immediately upstream of the predicted start codon, and is associated with a 235-nucleotide insertion of unknown origin.

Segregation Analysis
For self-cross analysis, heterozygous plants were allowed to self-pollinate and progeny seed was collected. For reciprocal cross analysis, heterozygous plants were crossed with wild-type plants as outlined in Table 1. In both cases, the progeny F1 seed was germinated on growth medium containing no antibiotics and progeny seedlings were genotyped and scored using PCR. Plants segregating the agl61-1 allele were genotyped using primers LBa1, AGL61-1LP, and AGL61-1RP (see above). Plants segregating the agl61-2 allele were genotyped using primers TDNA1, AGL61-1LP, and AGL61-2RP (see above). Heterozygous plants, identified by PCR were used in the segregation analysis described below.  22 using primers LBa1, AGL61-1RP, pCAMLacZR, and AGL61ATG800R (see above).
These plants were screened for siliques containing full seed set. Plants with full seed set putatively were homozygous for the rescue construct; to verify this, we collected seed from these plants and scored progeny seedlings for the presence of the rescue construct by PCR using primers pCAMLacZR, and AGL61ATG800R (see above).

Yeast Two-Hybrid Analysis
We used the Clontech Matchmaker GAL4 Two-Hybrid System 3 for the yeast two-hybrid analysis. The AGL80 and AGL61 open reading frames (without introns) were fused to the GAL4 activation domain and GAL4 DNA-binding domain in pGAD-T7 and pGBK-T7.
Yeast strain AH109 was cotransformed with combinations of pGAD-T7 and pGBK-T7 constructs (AGL80 plus AGL61 or controls containing one or both empty vectors) and selected on synthetic dropout (SD) medium lacking leucine and tryptophan (SD-LW).
Co-transformants were then assayed for interaction and activation of the histidine and adenine reporter genes on SD medium lacking leucine, tryptophan, histidine and adenine (SD-LWHA). For this, fresh colonies were grown in SD-LW at 30°C overnight to an OD of 1-2, the cells were pelleted and resuspended in 0.5 M sorbitol to an OD of 0.5, and 3 µl of each cell suspension was spotted on SD-LWHA plates using a multi-channel pipetor and grown at 30°C for 2-3 days. In this analysis, the second ATG (at position +49 relative to the transcriptional start site) was used as the start codon.

Analysis of Expression of Promoter:Reporter Constructs in agl61 Female
Gametophytes agl61-1/AGL61 plants were crossed as males with plants homozygous for the promoter:reporter constructs. To identify F1 plants containing the agl61-1 T-DNA allele, PCR was performed with primers LBa1 and AGL61-1RP (see above). F1 seed was plated on growth medium containing 30 µg/ml kanamycin to identify seedlings    Female gametophyte and endosperm stages are described in Christensen et al. (1997) and Boisnard-Lorig et al. (2001), respectively.  All panels are CLSM images. In these images, cytoplasm is gray, vacuoles are black, and nucleoli are white.   -a χ 2 values are not significantly different at a threshold of p = 0.01 from those expected under the hypothesis of a female gametophyte-lethal phenotype (i.e., 1:1:0 segregation). b χ 2 values are not significantly different at a threshold of p = 0.01 from those expected under the hypothesis of wild-type male gametophyte transmission (i.e., 1:1 segregation).