Atypical E2F activity restrains APC/C^CCS52A2 function obligatory for endocycle onset

The atypical E2F transcription factor E2Fe/DEL1 of Arabidopsis has been shown to control the endoreduplication level in roots, hypocotyls, and leaves (22). Compared with control plants, the knockout plants E2Fe/DEL1^KO (del1–1) displayed enhanced endoreduplication, whereas in the E2Fe/DEL1^OE overexpressing lines, the DNA ploidy level was reduced. Using β-glucuronidase (GUS) reporter line, E2Fe/DEL1 transcription was detected only in dividing cells (Fig. 1), confirming previous in situ mRNA hybridization data (22). Therefore, we postulated that E2Fe/DEL1 operates as a repressor of the endocycle onset, preventing dividing cells from exiting the cell cycle prematurely. To test this hypothesis, we compared the timing of the endocycle onset in developing leaves of del1–1 and control plants. Cells from the first leaf pair divide up to 9–11 days after sowing, after which they gradually exit the cell cycle and start to endoreduplicate (22, 24, 25). Correspondingly, 8-day-old leaves of both del1–1 and control lines had an equally low endoreduplication index (EI), that is, the mean number of endoreduplication cycles per nucleus (Fig. 2A). In contrast to wild-type leaves, the EI of E2Fe/DEL1-deficient leaves increased significantly (P < 0.005) and reproducibly between day 8 and day 10 because of a rise in the cell population with an 8C DNA content. The difference in EI between wild-type and del1–1 leaves was maintained on day 12, when wild-type leaves also began to endoreduplicate, and endured as the leaves matured. Thus, in the absence of a functional E2Fe/DEL1 protein, cells entered the endocycle more quickly. Endocycle onset and cell cycle exit are linked events. Correspondingly, del1–1 leaf cells exited the cell cycle program prematurely [supporting information (SI) Fig. S1], eventually resulting in a reduced total cell number (16,489 ± 1322 vs. 18,666 ± 741 epidermal cells in del1–1 and wild-type plants, respectively; P < 0.05). These findings were supported by analysis at the cellular level (Fig. 2B), showing a greater number of abaxial epidermal leaf cells with a high DNA content in 12-day-old del1–1 leaves than in control leaves at the same age.

To examine how E2Fe/DEL1 represses mitotic exit, we compared the transcriptome of wild-type plants with that of plants that either overexpressed or were deficient in E2Fe/DEL1 (see SI Text). We identified 10 genes with expression profiles that positively correlated with the endoreduplication phenotype (Fig. 2C; Table S1). Among these, the CCS52A2 gene displayed the most significant changes in expression level. These changes in CCS52A2 expression in response to altered E2Fe/DEL1 levels were confirmed by quantitative reverse-transcription polymerase chain reaction (RT-PCR) analysis (Fig. S2). Although the Arabidopsis genome encodes three atypical E2F proteins, CCS52A2 transcripts were specifically modulated by E2Fe/DEL1, because the expression levels remained the same in E2Fd/DEL2 knockout (del2–1) and E2Ff/DEL3 knockdown (del3–1) plants (Fig. 2D).

The CCS52A2 gene encodes a putative activator of the APC/C and is related to the Drosophila FZR and mammalian CDH1 proteins. The CCS52A genes of Medicago sativa and Medicago truncatula have been shown to control the onset of endoreduplication during nodule development, and likewise, the CDH1 and FZR genes regulate the mitosis-to-endocycle transition in human and Drosophila cells, respectively (12–14, 26). Arabidopsis has two CCS52A/FZR/CDH1-related genes, designated CCS52A1 and CCS52A2 (27). In analogy to their leguminous and nonplant counterparts, both CCS52A1 and CCS52A2 were found to control the endocycle, as indicated by the low EI of mature leaves of the knockout lines CCS52A1^KO and CCS52A2^KO (Fig. 3A; Fig. S3).

The endoreduplication phenotypes of CCS52A1^KO and CCS52A2^KO plants suggest that both CCS52A1 and CCS52A2 may be direct target genes of E2Fe/DEL1. But only CCS52A2 transcript levels were altered in del1–1 plants (Fig. 3B). Correspondingly, ChIP assays with a specific anti-E2Fe/DEL1 antibody demonstrated that E2Fe/DEL1 associated only with the CCS52A2 promoter (Fig. 3C). Locus scanning revealed that the E2Fe/DEL1-binding site within the CCS52A2 promoter coincided with the position of a putative E2F cis-acting element located just upstream of the ORF of the CCS52A2 gene (Fig. 3D; Fig. S4A). Proof that this site is required for E2Fe/DEL1 binding was provided by introducing either the endogenous CCS52A2 promoter or an identical promoter construct with a mutated E2F cis-acting element into plants. The wild-type promoter fragment, but not its mutant variant, could be immunoprecipitated by the anti-E2Fe/DEL1 antibody (Fig. 3E), even in an E2Fe/DEL1^OE background, implying that E2Fe/DEL1 binding requires a functional E2F regulatory sequence.

The changes in CCS52A2 transcript abundance in the E2Fe/DEL1 transgenic lines and the direct association of E2Fe/DEL1 with the CCS52A2 promoter indicates that E2Fe/DEL1 might control the temporal expression of CCS52A2. Therefore, the CCS52A2 transcript levels were analyzed in wild-type and del1–1 plants during leaf development. E2Fe/DEL1 mRNA levels were abundant mainly in the early stages of leaf development. In contrast, CCS52A2 transcripts accumulated and reached a maximum at day 10, the time of cell cycle exit and onset of endoreduplication (Fig. 4A). Because of its low expression level before day 10, CCS52A2 might be repressed by E2Fe/DEL1 during the dividing phase of leaf development. This hypothesis was confirmed in del1–1 plants, in which CCS52A2 expression levels were clearly higher during the early leaf growth stages than those of control plants (Fig. 4B). No significant changes in CCS52A1 expression levels were observed (Fig. 4C), again indicating that E2Fe/DEL1 controls CCS52A2 expression only.

To determine whether the increase in CCS52A2 transcript levels in young leaves of del1–1 plants could account for the premature onset of endoreduplication, DNA ploidy changes were analyzed in plants overexpressing CCS52A2. CCS52A2^OE leaves entered the endoreduplication cycle earlier than control leaves, as indicated by the increased number of cells with a high DNA ploidy level (Fig. 4D), in agreement with the anticipated role of CCS52A2 as an activator of the endoreduplication program.

Our data suggest that E2Fe/DEL1 levels determine the timing of cell cycle exit and onset of endoreduplication by controlling when APC/C^CCS52A2 is active. To understand mechanistically how decreasing E2Fe/DEL1 levels can account for the division-to-endoreduplication transition, we mathematically modeled the cell cycle phase–dependent expression pattern of CCS52A2 during leaf development. In a synchronized cell culture, E2Fe/DEL1 and CCS52A2 display complementary transcription profiles, with a predominance of CCS52A2 expression during the G₁ and S phases (Fig. S5B) (28). The CCS52A2 expression profile corresponded with the anticipated function of its gene product in preventing premature accumulation of mitotic cyclins in interphase cells but allowing their accumulation during the late S and G₂ phases, allowing the M phase to proceed (29). E2Fe/DEL1 expression levels peaked during G₂, similar to what has been observed for its mammalian counterparts E2F7 and E2F8 (30, 31). Because cell division cannot be synchronized experimentally in a developing leaf and endoreduplication cannot be triggered in Arabidopsis cell cultures, we combined leaf and cell culture expression data mathematically (see SI Appendix, Modeling). This mathematical modeling permitted an in silico visualization of the cell cycle phase–dependent relationship between E2Fe/DEL1 and CCS52A2 in a developmental context. The simulation revealed that decreasing E2Fe/DEL1 levels during leaf maturation triggered a preferential increase in CCS52A2 transcripts during the late S-G₂ and M phases (Fig. 4E; Movie S1). These data suggest that E2Fe/DEL1 controls the cell cycle phase–dependent CCS52A2 transcription profile in a developmentally dependent manner.

In analogy to CCS52A2, a consensus E2F cis-acting element was detected in the promoter region of the human CDH1 gene (Fig. S4B). To investigate whether the observed control of the APC/C activator genes by atypical E2F proteins might be conserved among plants and metazoans, we evaluated the binding of the human E2F7 transcription factor to the CDH1 promoter by ChIP analysis in human bone tissue–derived osteosarcoma cells (U2OS). As a positive control, we tested the association of E2F7 with the E2F1 gene, recently demonstrated to be an E2F7 target gene (20, 21). No E2F1 or CDH1 promoter DNA was precipitated with a nonspecific antibody, but both promoters could be detected in the E2F7 immunoprecipitates (Fig. 5).

Plants were grown at 22°C and a 16-h photoperiod (65 μE m⁻²s⁻¹) on agar-solidified culture medium (0.5 × Murashige and Skoog medium, 0.5 g/liter of 2-(N-morpholino)ethanesulfonic acid [MES], 10 g/liter of sucrose, and 0.8% plant tissue culture agar). The del1–1 and del3–1 alleles have been described previously (22, 23); del2–1, ccs52a1–1, ccs52a2–1, and ccs52a2–2 are the SALK T-DNA insertion lines 093190, 083656, 001978, and 073708, respectively. The SAIL T-DNA insertion line 797-F01 represents ccs52a1–2. All lines were provided by the Signal Insertion Mutant Library (http://signal.salk.edu). Primers for genotyping are listed in Table S2.

The intergenic region containing the CCS52A2 (At4g11920) promoter was isolated by PCR (for primers used, see Table S2) and cloned into the Gateway pKm43GW vector (39). The resulting plasmid was used to mutate the E2F-binding site with a site-directed plasmid mutagenesis (for primers, see Table S2). The coding region of CCS52A2 was amplified by PCR (for primers, see Table S2) and cloned into the Gateway pDONR221 vector by attB × attP recombination and subsequently recombined into the pH2GW7 vector by attL × attR recombination. All vectors were used to transform Arabidopsis thaliana (L.) Heyhn plants by the flower-dip method (40). Transgenic homozygous plants containing only one T-DNA were obtained on a selective medium.

Briefly, young seedlings were incubated in 80% acetone for 30 min. After the material had been washed in phosphate buffer, it was immersed in the enzymatic reaction mixture (1 mg/ml of 5-bromo-4-chromo-3-indolyl β-d-glucuronide, 2 mM ferricyanide, and 0.5 mM ferrocyanide in 100 mM phosphate buffer [pH 7.4]). The reaction was run at 37°C in the dark for 4 h. The material was cleared in ethanol and examined under a light microscope.

Flow cytometry and densitometry (7) were used to create the DNA ploidy maps. The EI was calculated (3) from the number of nuclei of each represented ploidy level multiplied by the number of endoreduplication cycles necessary to reach the corresponding ploidy level. The sum of the resulting products was divided by the total number of nuclei.

A GST-tagged fusion protein containing the last 100 aa of the E2Fe/DEL1 protein was produced in Escherichia coli BL21-CodonPlus(DE3)-RIL according to standard methods. Polyacrylamide gel slices containing this recombinant protein were injected into rabbits to produce polyclonal anti-E2Fe/DEL1 antiserum.

Aphidicolin block/release was done as described previously (41).