The APETALA2–MYBL2 module represses proanthocyanidin biosynthesis by affecting formation of the MBW complex in seeds of Arabidopsis thaliana

Abstract

Proanthocyanidins (PAs) are the second most abundant plant phenolic natural products. PA biosynthesis is regulated by the well-documented MYB/bHLH/WD40 (MBW) complex, but how this complex itself is regulated remains ill defined. Here, in situ hybridization and β-glucuronidase staining show that APETALA2 (AP2), a well-defined regulator of flower and seed development, is strongly expressed in the seed coat endothelium, where PAs accumulate. AP2 negatively regulates PA content and expression levels of key PA pathway genes. AP2 activates MYBL2 transcription and interacts with MYBL2, a key suppressor of the PA pathway. AP2 exerts its function by directly binding to the AT-rich motifs near the promoter region of MYBL2. Molecular and biochemical analyses revealed that AP2 forms AP2–MYBL2–TT8/EGL3 complexes, disrupting the MBW complex and thereby repressing expression of ANR, TT12, TT19, and AHA10. Genetic analyses revealed that AP2 functions upstream of MYBL2, TT2, and TT8 in PA regulation. Our work reveals a new role of AP2 as a key regulator of PA biosynthesis in Arabidopsis. Overall, this study sheds new light on the comprehensive regulation network of PA biosynthesis as well as the dual regulatory roles of AP2 in seed development and accumulation of major secondary metabolites in Arabidopsis.

AP2 represses proanthocyanidin biosynthesis by directly binding to an AT-rich motif near the promoter region of MYBL2 and by interacting with MYBL2 to form AP2–MYBL2–TT8/EGL3 complexes, disrupting the MBW complex and thereby repressing expression of ANR, TT12, TT19, and AHA10. AP2 is therefore a key regulator of PA biosynthesis in Arabidopsis.

Introduction

Proanthocyanidins (PAs), also known as condensed tannins, are the second most abundant plant phenolic polymers after lignins, and they are present in the leaves, bark, fruits, and seeds of many plants (Dixon et al., 2005; Prior and Gu, 2005). PAs provide protection against predation and give flavor and astringency to beverages such as wine, teas, and fruit juices. PAs enhance soil nitrogen retention, prevent pasture bloat, and improve nitrogen nutrition in ruminant livestock, and they have many beneficial effects on human health (Lees, 1992; Dixon et al., 2005; Joanisse et al., 2009). Significant effort has therefore been focused on characterizing the PA biosynthetic pathway in a range of plant species over the last several decades.

In Arabidopsis, PAs are synthesized in the innermost cell layer of the testa (chalaza, micropyle, and endothelium), conferring to the mature seed its characteristic brown color (Debeaujon et al., 2003; Pourcel et al., 2005; Lepiniec et al., 2006). Biosynthesis of PAs begins in the micropylar region around 1–2 days after fertilization, and deposition progresses in the endothelium toward the chalaza until around 5–6 days after fertilization (Debeaujon et al., 2003; Lepiniec et al., 2006). The genetics of the PA subpathway, as well as the common flavonoid pathway, have been well studied in Arabidopsis, and more than 20 genes have been shown to participate in the early pathway (e.g., CHS, CHI, F3H, and F3'H), late pathway (e.g., DFR, ANS, and ANR), and modification, transport, and oxidation (e.g., TT12, TT19, AHA10, and TT10) (Xu et al., 2014; Supplemental Figure 1).

The MBW ternary complex, consisting of MYB (TT2 and MYB5), bHLH (TT8, GL3, and EGL3) and WD repeat (TTG1) proteins (Baudry et al., 2004; Lepiniec et al., 2006; Xu et al., 2014), is the classical regulatory complex in the PA pathway. The TT2–TT8–TTG1 complex plays a leading role, and three other MBW complexes (MYB5–TT8–TTG1, TT2–GL3–TTG1, and TT2–GL3–TTG1) are also involved in PA biosynthesis in a tissue-specific manner (Baudry et al., 2004; Xu et al., 2014). The functions of the MBW partners in regulation of PAs and flavonoids are well conserved in plants (Xu et al., 2015). In the MBW complex, MYB proteins bind to the MYB core elements and AC-rich elements of late biosynthetic genes in the PA subpathway, whereas bHLH proteins bind to their E/G-box motifs (Koshino-Kimura et al., 2005; Ryu et al., 2005; Ishida et al., 2007; Kang et al., 2009; Song et al., 2011; Xu et al., 2014; Li et al., 2022). TTG1 stabilizes the interaction between TT2 and TT8, which is crucial for the transcriptional activity of the MBW complex in planta (Baudry et al., 2004; Thevenin et al., 2012). Several other regulators also fine-tune the transcriptional activity of MBW complexes by interacting with MYBs or bHLHs to organize or disrupt complex formation (Dubos et al., 2008; Matsui et al., 2008; Appelhagen et al., 2011; Gonzalez et al., 2016). For instance, the TT1 WIP zinc-finger protein, the TTG2 WRKY transcription factor, and the TT16 MADS protein interact with components of the MBW complex to positively regulate PA biosynthesis (Nesi et al., 2002; Sagasser et al., 2002; Appelhagen et al., 2011; Gonzalez et al., 2016; Xu et al., 2017). By contrast, a few other regulators negatively regulate biosynthesis of PAs and other flavonoids; for example, the MADS transcription factor SEEDSTICK inhibits PA accumulation by modifying the chromatin state of key regulatory genes (e.g., TT8, EGL3, and TT16) and ANR (Mizzotti et al., 2014). In particular, the R3-MYB protein MYBL2 negatively regulates anthocyanin and PA biosynthesis by interacting with bHLHs (e.g., TT8, GL3, and EGL3) to inhibit formation of the MBW complex (Sawa, 2002; Zimmermann et al., 2004; Dubos et al., 2008; Matsui et al., 2008). However, bioengineering of PAs based on these regulatory mechanisms of the PA pathway in Arabidopsis and other plant species has not been successful to date (Nesi et al., 2001; Sharma and Dixon, 2005; Verdier et al., 2012; Li et al., 2016), mainly owing to our limited understanding of PA regulation. Although several components involved in regulation of the MBW complex have been identified and characterized, additional regulators still remain to be discovered in order to advance our understanding of PA biosynthesis.

Here, we show that APETALA2 (AP2) disrupts formation of the MBW complex to repress PA biosynthesis. Overexpression of AP2 led to a lighter seed color owing to lower PA content in Arabidopsis seeds, implying that AP2 represses PA biosynthesis. AP2 binds to the promoter of MYBL2 to activate its expression. AP2 also interacts with MYBL2 to repress formation of the MBW complex. Genetic analyses showed that AP2 acts upstream of MYBL2 and the MBW complex in PA regulation. We conclude that AP2 acts as an activator of MYBL2 to repress PA biosynthesis by disrupting formation of the MBW complex.

Results

AP2 represses PA biosynthesis

In addition to its better-known roles in floral organ identity (Bowman et al., 1989; Kunst et al., 1989) and flowering time (Yant et al., 2010), AP2 also has essential roles in seed development (Jofuku et al., 2005; Ohto et al., 2005) and differentiation of the outer layers of the seed coat (Western et al., 2001). Our previous study showed that the Arabidopsis ap2-6 mutant displays a darker seed color than the wild type (Jiang et al., 2013), indicating that the ap2 mutant may accumulate more PAs in the seed coat. To investigate the effect of AP2 on PA accumulation, seeds of the ap2-5 and ap2-6 mutants were stained with p-dimethylaminocinnamaldehyde (DMACA), a dye that is commonly used for PAs can react with PAs to form a blue to purple stain, and the ap2-6 mutant exhibited a darker blue color with a higher PA content than the wild type (Figure 1A and 1B and Supplemental Figure 2).

Figure 1.

AP2 represses PA accumulation and is expressed in the seed coat endothelium.

(A) Seed phenotypes of the wild type (Col-0), ap2 mutants (ap2-5 and ap2-6), and AP2 overexpression lines (AP2m3 OE, AP2m3 Myc OE, and AP2 OE). Scale bars, 0.5 mm.

(B) Relative PA level in mature seeds of Col-0, ap2 mutants, and AP2 overexpression lines. Levels of each type of PA in Col-0 were set to 1.0. Data are presented as mean ± SD, Student’s t-test (n = 3, ∗P < 0.05, ∗∗P < 0.01).

(C–E)In situ hybridization using an AP2 antisense probe (C), sense control (D), and no-probe control (E) in 4-day-old seeds of Col-0. Scale bars, 50 μm.

(F–H) Histochemical analysis of GUS activity generated by proAP2::GUS in 3-day-old siliques (F), seeds without sectioning (G), and seeds with sectioning (H). Scale bars, 200 μm in (F) and 50 μm in (G) and (H).

To further examine the effect of AP2 on PA accumulation, we generated various AP2 overexpression lines (Supplemental Figure 3A and 3B), including AP2 overexpression plants (AP2 OE1-3), AP2m3 overexpression plants (AP2m3 OE1-3), and AP2m3 Myc overexpression plants (AP2m3 Myc OE1-2). AP2 OE indicates overexpression of the native AP2 gene; because AP2 is a target gene of miR172, the transcript level of AP2 is elevated in AP2 OE, but AP2 protein abundance is unchanged. AP2m3 OE1-3 indicates overexpression of a mutant AP2 gene whose mutations disrupt miR172 base pairing (Supplemental Figure 3A and 3B) without changing the AP2 amino acid sequence (Chen, 2004). Both AP2 transcript level and AP2 protein abundance are elevated in AP2m3 OE1-3, owing to the failure of miR172 to bind to AP2m3. AP2m3 Myc OE1-2 indicates AP2m3 with an added Myc tag to facilitate coimmunoprecipitation (coIP) assays; like AP2m3 OE1-3, it has elevated AP2 transcript levels and AP2 protein levels.

Seeds of the AP2m3 and AP2m3 Myc lines exhibited a lighter color and significantly lower PA content than the wild type (Figure 1A and 1B). However, seeds of AP2 OE1-3 showed a phenotype similar to that of the wild type (Figure 1A and 1B). These results indicate that an increase in AP2 protein leads to decreased PA accumulation in the Arabidopsis seed coat.

To evaluate whether flavonoid compounds other than PAs are also affected by AP2, we analyzed anthocyanins and flavonols in seedlings or seeds of these lines, but neither anthocyanins nor flavonols were affected in AP2 knockout or overexpression lines (Supplemental Figure 4), implying that AP2 specifically represses PA accumulation in seeds.

AP2-like miR172-target clade members and miR172 do not affect PA accumulation

AP2-like miR172-target clade members (e.g., TOE1, TOE2, TOE3, SMZ, and SNZ) have been reported to redundantly regulate flowering time (Yant et al., 2010). To determine whether these clade members also affect PA accumulation, we analyzed their expression in the seed coat during various seed developmental stages. AP2 showed the highest expression, and TOE2 was expressed at a higher level than the remaining four genes (Supplemental Figure 5A), suggesting that TOE2 might affect biosynthesis of PAs that accumulate specifically in the seeds. Phylogenetic analysis showed that AP2 was clustered in the same clade with TOE1 and TOE3 (Supplemental Figure 5B). Their similar expression patterns and close phylogenetic relationships suggested that TOE1, TOE2, and TOE3 might be involved in PA accumulation. Thus, we also generated transgenic plants overexpressing TOE1m3, TOE2m3, and TOE3m3 (Supplemental Figure 3); all harbored a truly silent mutation in the miR172 binding site to ensure elevated transcript levels and protein levels. However, we did not observe a lighter seed color in transgenic seeds overexpressing TOE1m3, TOE2m3, or TOE3m3 (Supplemental Figure 6), and their seeds exhibited the same or a similar blue-purple color as wild-type seeds when stained with DMACA (Supplemental Figure 7A). The PA content of these lines was similar to that of the wild type, as determined by quantitative analyses (Supplemental Figure 7B), suggesting that TOE1, TOE2, and TOE3 do not affect PA accumulation.

We next asked whether miR172 itself was involved in PA accumulation. Seed color of the transgenic plants overexpressing miR172a was similar to that of the wild type before and after staining with DMACA (Supplemental Figures 3C, 6, and 7), and their PA contents were similar to that of the wild type (Supplemental Figure 7B), indicating that miR172 does not affect PA accumulation. Taken together, our results clearly indicate that neither AP2-like miR172-target clade members nor miR172 affect PA accumulation.

AP2 is expressed in the seed coat endothelium

Because our results indicated a novel function of AP2 in seeds, where PAs accumulate, we analyzed AP2 expression in early seed development by in situ hybridization. We detected a strong signal above background in the endothelium, which is the innermost cell layer of the inner integument (Figure 1C–1E). Some signal above background was also detected in the outer layers of the seed coat and the embryo (Figure 1C–1E), consistent with previous reports that AP2 is a regulator of outer integument differentiation during seed coat development (Western et al., 2001) and that AP2 is expressed throughout the embryo from the globular stage onward (Wurschum et al., 2006). We next cloned a 5.4-kb fragment upstream of the AP2 cDNA as the AP2 promoter (Zhao et al., 2007), fused it to the coding region of β-glucuronidase (GUS), and introduced it into wild-type Arabidopsis. This promoter fragment drove GUS expression in the seed coat of 3-day-old seeds (Figure 1F–1H), consistent with the expression pattern of AP2 in the in situ hybridization assay. Thus, AP2 is strongly expressed in the seed coat endothelium, where PAs accumulate, suggesting a direct regulatory role of AP2 in PA accumulation.

AP2 regulates expression of late PA pathway genes and transcriptional regulators

To investigate whether transcript abundance of PA pathway genes is affected by AP2, we examined the expression profiles of 14 biosynthetic genes and 11 transcriptional regulators involved in PA accumulation using available online databases. The expression levels of most genes in the developing Arabidopsis seed coat were relatively high at the pre-globular and/or globular developmental stages, decreased dramatically at the heart and/or linear cotyledon stages, and decreased further to very low levels at the mature green stage. However, AP2 was constantly expressed at a high level throughout seed development (Supplemental Figures 5A and 8).

We then measured the relative transcript abundance of all known PA pathway genes in developing seeds (4 days post pollination) by qPCR. In the AP2m3 overexpression line, relative transcript levels of late pathway genes (e.g., DFR, ANS, ANR, TT12, TT19, and AHA10) decreased significantly compared with the wild type. Expression of ANR, TT12, TT19, and AHA10 increased in the ap2-6 mutant compared with the wild type (Figure 2A), indicating that AP2 negatively regulates expression of these genes. Among the 11 transcriptional regulatory genes analyzed, TT2 was significantly inhibited by AP2, whereas MYBL2 was activated by AP2 (Figure 2B). These results indicated that AP2 represses PA accumulation by specifically regulating the expression of late pathway genes (e.g., DFR, ANS, ANR, TT12, TT19, and AHA10) and specific transcriptional regulators (TT2 and MYBL2).

Figure 2.

AP2 represses PA accumulation by regulating the expression of late biosynthetic genes and transcriptional regulator genes.

(A and B) Relative transcript levels of 14 structural genes (A) and 11 transcriptional regulator genes (B) in 4-day-old siliques of Col-0, ap2-6 mutant, and AP2m3 OE lines. The transcript level of each gene in Col-0 was set to 1.0. Data are presented as mean ± SD, Student’s t-test (n = 3, ∗P < 0.05, ∗∗P < 0.01).

AP2 activates MYBL2 transcription by binding to its promoter

To explore how AP2 regulates the PA pathway genes, we performed an in vitro transcription assay in Arabidopsis leaf protoplasts to determine the effects of AP2m3 co-expression on transactivation of a reporter gene (firefly luciferase) driven by the promoters of the PA pathway genes. Surprisingly, we found that AP2m3 activated expression of the reporter gene driven by the MYBL2 promoter by 3.5-fold but did not activate expression of the other 24 genes (Figure 3A). This result suggested that AP2 positively and uniquely regulates the expression of MYBL2 directly, consistent with its expression changes in the ap2 mutant and AP2m3 OE lines (Figure 2B). We also examined the expression level of MYBL2 during seed development and found that MYBL2 expression was reduced in the ap2-6 mutant compared with the wild type during seed development (Supplemental Figure 9), indicating that AP2 activates MYBL2 expression.

Figure 3.

AP2 activates the expression of MYBL2 by binding to AT-rich motifs.

(A)AP2m3 was co-expressed with a reporter gene driven by the promoters of PA pathway genes in Arabidopsis leaf protoplasts. Induction of the reporter gene was measured by assaying firefly luciferase activity. Luciferase activity of the control without AP2m3 was set to 1.0.

(B) The position of five AT-rich motifs (TTTGTT or AACAAA) (M1, −751 to −756; M2, −555 to −560; M3, −256 to −261; M4, −233 to −238; and M5, +4 to +9) located in/near the promoter, coding sequence, and 3′ UTR of MYBL2. Black triangles indicate AT-rich motifs. Bars, 100 bp.

(C)AP2m3 was co-expressed with a reporter gene driven by the native promoter or mutated promoters of MYBL2 in Arabidopsis leaf protoplasts (m indicates that the native AT-rich motif was mutated to CCCCCC). Luciferase activity in the control without AP2m3 was set to 1.0.

(D) Chromatin immunoprecipitation (ChIP) assay of the MYBL2 promoter. One transgenic line (AP2m3 Myc OE-1) overexpressing AP2m3 Myc and Myc antibody were used for the ChIP assay, with the wild type (Col-0) as a negative control. PP2A and UBC30 were included as controls. Three independent experiments were carried out with similar results. Data are presented as mean ± SD, Student’s t-test (n = 3, ∗P < 0.05, ∗∗P < 0.01).

A previous study showed that AP2 directly regulates AGAMOUS (AG), a C-class floral organ identity gene, by binding to the AT-rich motif (TTTGTT or AACAAA) in young flowers (Dinh et al., 2012). Thus, AP2 may directly activate MYBL2 by binding to its AT-rich motif(s). We identified five AT-rich motifs (M1–M5) within/near the promoter region of MYBL2 (Figure 3B) and mutated the native AT-rich motifs (m1–m5 and different combinations) to CCCCCC to examine their effects on MYBL2 expression. AP2 failed to activate MYBL2 expression in in vitro transcription assays whenever M5 (m5) was mutated (Figure 3C). By contrast, mutation of motifs M1–M4 (m1–m4) in different combinations had no significant effect on MYBL2 expression (Figure 3C). These results indicate that AP2 binds to the M5 motif to activate MYBL2 expression.

We next performed a chromatin IP (ChIP) assay to test whether AP2 could bind to the promoter of MYBL2 in vivo. A transgenic Arabidopsis line overexpressing AP2m3 Myc was used for ChIP analysis, along with the wild type as a negative control. AP2m3 Myc bound to the M5F/R region containing the AT-rich M5 motif but not to other regions (Figure 3D), consistent with the mutation analysis of the AT-rich motifs (Figure 3C). These results indicate that AP2 specifically binds to the AT-rich M5 motif of MYBL2.

AP2 interacts with MYBL2 in vivo

Yeast two-hybrid assays were used to determine whether any of the 11 transcriptional regulators of the PA pathway interact with AP2. Yeast cells co-transformed with AP2-BD and MYBL2-AD grew well on selection medium, and those with AP2-BD and TT2-AD and with AP2-BD and MYB5-AD grew slowly (Supplemental Figure 10). Yeast cells co-transformed with AP2-BD and eight other transcriptional regulators did not grow at all (Supplemental Figure 10). These results indicated that AP2 can physically interact with MYBL2, TT2, and MYB5 in yeast. The in vivo interactions between AP2 and MYBL2, AP2 and TT2, and AP2 and MYB5 were examined using bimolecular fluorescence complementation (BiFC) and coIP assays. Co-expression of AP2-N terminal yellow fluorescence protein with MYBL2-CYFP in leaves of Nicotiana benthamiana produced intense yellow fluorescent protein (YFP) fluorescence, whereas no YFP signal was observed when AP2-N terminal yellow fluorescence protein was co-expressed with TT2-CYFP or MYB5-CYFP (Supplemental Figure 11), confirming that AP2 interacts only with MYBL2 in vivo. Consistent with this finding, FLAG-tagged MYBL2 was immunoprecipitated with an antibody against c-Myc when c-Myc-tagged AP2m3 was co-expressed with FLAG-tagged MYBL2 in N. benthamiana leaves (Figure 4A). Taken together, these results demonstrate that AP2 physically interacts with MYBL2 in vivo.

Figure 4.

AP2 interacts with MYBL2, and MYBL2 interacts with TT8, GL3, and EGL3.

(A–D) CoIP analysis of AP2 interaction with MYBL2, and MYBL2 interaction with TT8, GL3, and EGL3.

(A) AP2m3-Myc and MYBL2-FLAG were detected with anti-Myc and anti-FLAG antibodies, respectively.

(B–D) TT8-HA (B), GL3-HA (C), and EGL3-HA (D) and MYBL2-FLAG were detected with anti-HA and anti-FLAG antibodies, respectively.

(E and F) Yeast two-hybrid assays between different truncations of MYBL2 and AP2 or TT8. Left: vectors containing proteins fused with BD or AD. Center: yeast cells containing BD and AD were diluted with sterile demineralized water at ratios of 1:10, 1:100, and 1:1000 on SD/−Trp/−Leu medium at 30°C for 3 days. Right: the above diluted yeast cells grown on SD/−Trp/−Leu/−His/−Ade medium with 2 mM 3-amino-1,2,4-triazole at 30°C for 7 days.

(G) Protein interaction analysis of different truncations of MYBL2 and AP2 or TT8 in leaves of N. benthamiana with a nuclear localization marker (nuclear localization signal-mCherry).

AP2 represses transcriptional activity of the MBW complex by interacting with MYBL2

Previous studies have used yeast two-hybrid assays to show that MYBL2 negatively regulates anthocyanin and PA biosynthesis by interacting with TT8, GL3, and EGL3, thereby inhibiting the transcriptional activity of the MBW complex (Sawa, 2002; Zimmermann et al., 2004; Dubos et al., 2008; Matsui et al., 2008). In this study, we performed BiFC and coIP assays to further confirm the direct interactions of MYBL2 with TT8, GL3, and EGL3 (Figure 4B–4D and 4G and Supplemental Figure 12). Because MYBL2 also interacts with AP2, as shown above, we hypothesized that MYBL2 could interact simultaneously with AP2 and the TT8/GL3/EGL3 complex. To test this hypothesis, we generated different truncated forms of MYBL2 and tested their interactions with AP2 in yeast two-hybrid assays (Supplemental Figure 13A). AP2 bound to C-terminal truncations of MYBL2 (Figure 4E; Supplemental Figure 13A), and TT8 bound to N-terminal truncations of MYBL2 (Figure 4F), as further confirmed by BiFC assays (Figure 4G). Yeast two-hybrid assays also showed that GL3 bound to C-terminal truncations of MYBL2 (Supplemental Figure 13C) and EGL3 to N-terminal truncations of MYBL2 (Supplemental Figure 13D). These results indicate that MYBL2 interacts simultaneously with AP2 and TT8/EGL3 at the C terminus and N terminus, respectively. Additional coIP assays showed that c-Myc-tagged AP2m3 was immunoprecipitated with an antibody against hemagglutinin (HA) when c-Myc-tagged AP2m3 and FLAG-tagged MYBL2 were co-expressed with HA-tagged TT8/HA-tagged EGL3 (Figure 5A and 5B), whereas c-Myc-tagged AP2m3 was not immunoprecipitated when they were co-expressed with FLAG-tagged MYBL2 and HA-tagged GL3 (Figure 5C). These results indicate that MYBL2 bridges AP2 and TT8/EGL3 to form the AP2–MYBL2–TT8 or AP2–MYBL2–EGL3 complex.

Figure 5.

AP2 interacts with MYBL2 and TT8 or EGL3 to form the AP2–MYBL2–TT8/EGL3 complex, thus repressing transcriptional activity of the MBW complex.

(A–C) CoIP analysis of AP2 interaction with MYBL2, TT8, GL3, and EGL3. AP2m3-Myc, MYBL2-FLAG, and TT8-HA (A), EGL3-HA (B), and GL3-HA (C) were detected with anti-Myc, anti-FLAG, and anti-HA antibodies, respectively.

(D) AP2 represses transcriptional activity of the MBW complex by interacting with MYBL2. AP2 enhances MYBL2-mediated transcriptional repression activity of the MBW complex, as demonstrated by analysis of proANR activity. Compared with that of protoplasts co-transfected with proANR:luciferase, TT2, TT8, TTG1, and MYBL2, luciferase activity of protoplasts co-transfected with proANR:luciferase, TT2, TT8, TTG1, MYBL2, and AP2m3 was reduced. Values that differ at P < 0.05 are labeled with different letters according to one-way ANOVA with post hoc Tukey’s Honestly Significant Difference.

MYBL2 has been shown to inhibit ANR expression by repressing the MBW complex (TT2–TT8–TTG1 complex) (Thevenin et al., 2012). We hypothesized that AP2 inhibits the transcription of the MBW complex by interacting with MYBL2. To test this hypothesis, a transient transactivation system was used to test the effect of simultaneous and constitutive expression of AP2 and MYBL2 on expression of proANR:luciferase in the presence or absence of the TT2–TT8–TTG1 complex. It was clear that the presence of the TT2–TT8–TTG1 complex resulted in higher activation activity than the presence of one or two genes (Figure 5D). Quantification of luciferase activity showed that the presence of MYBL2 significantly decreased luciferase activity of the TT2–TT8–TTG1 complex. The presence of AP2m3 with the TT2–TT8–TTG1 complex did not change its activity compared with the TT2–TT8–TTG1 complex alone. By contrast, the presence of both MYBL2 and AP2m3 with the TT2–TT8–TTG1 complex led to even lower activity than that observed for the TT2–TT8–TTG1 complex with MYBL2 (Figure 5D). This result suggests that MYBL2 is required for AP2-repressed transcriptional activity of the TT2–TT8–TTG1 complex for activation of ANR. Taken together, these results demonstrate that AP2 represses the transcriptional activity of the TT2–TT8–TTG1 complex by interacting with MYBL2.

AP2 functions upstream of MYBL2 and the MBW complex in PA biosynthesis

The results above showed that AP2 plays key roles in regulating PA biosynthesis via MYBL2 and the MBW complex. We next investigated their genetic interactions using overexpression and mutant lines. Overexpression of AP2m3 in the mybl2 mutants resulted in seed color and PA content similar to those of the mybl2 mutants (Figure 6A, 6B, and 6D). Overexpression of MYBL2 in the ap2-6 mutant led to seed color and PA content similar to those of the MYBL2 overexpression plants (Figure 6A, 6B, and 6D), suggesting that AP2 acts upstream of MYBL2. Additional genetic studies showed that the double mutants ap2-6/tt2 and ap2-6/tt8 exhibited the typical pale yellow seed color of tt2 and tt8 mutants, consistent with their PA content (Figure 6C and 6E), suggesting that AP2 acts upstream of the MBW complex. AP2 thus functions genetically upstream of MYBL2 and the MBW complex to regulate PA biosynthesis.

Figure 6.

AP2 acts upstream of MYBL2 and the MBW complex, and AP2 repression of PA biosynthesis is dependent on its downstream gene MYBL2.

(A) Seed phenotypes of Col-0, ap2-6, mybl2-1, mybl2-2, AP2m3 OE, MYBL2 OE, and their double mutants. Scale bars, 0.4 mm.

(B) Mature seeds were stained with DMACA for 5 days. Scale bars, 0.4 mm.

(C) Seed phenotypes of tt2, tt8, and their double mutants with ap2-6. Scale bars, 0.4 mm.

(D and E) Relative PA levels in mature seeds. Levels of each type of PA in Col-0 were set to 1.0. Data are presented as mean ± SD. Values that differ at P < 0.05 are labeled with different letters according to one-way ANOVA with post hoc Tukey’s Honestly Significant Difference.

Discussion

Arabidopsis AP2 is a multifunctional gene expressed in the sepal, petal, stamen, and carpel and is associated with flower organ identity (Bowman et al., 1989, 1991; Kunst et al., 1989; Drews et al., 1991; Jofuku et al., 1994; Wollmann et al., 2010; Krogan et al., 2012). AP2 also functions in stem cell maintenance in the shoot apical meristem (Wurschum et al., 2006), control of floral stem cells (Aukerman and Sakai, 2003; Chen, 2004; Yumul et al., 2013; Liu et al., 2014; Huang et al., 2017), and regulation of flowering time (Mathieu et al., 2009; Yant et al., 2010; O'Maoileidigh et al., 2021). In addition, AP2 controls fruit and seed development (Jofuku et al., 1994; Ripoll et al., 2011) and mediates seed size by affecting embryo, endosperm, and seed coat development (Jofuku et al., 2005; Ohto et al., 2005; 2009). AP2 is essential for differentiation of the outer layers of the seed coat (Western et al., 2001), and the ap2-6 mutant exhibits a darker seed color (Jiang et al., 2013). The seed color of Arabidopsis is mainly contributed by PAs; therefore, it is important to clarify whether AP2 also regulates PA biosynthesis in the seed coat.

We found that the ap2-6 mutant with a much darker seed color accumulates more PA than the wild type (Figure 1A and 1B and Supplemental Figure 2). To further determine the effect of AP2 on PA accumulation, we generated AP2 overexpression lines with higher AP2 RNA abundance than the wild type (Supplemental Figure 3B). However, overexpression of AP2 resulted in a brown seed color and a PA content similar to those of the wild type (Figure 1A and 1B). A previous study showed that AP2 RNA abundance was elevated in AP2 overexpression plants but that AP2 protein abundance was not increased compared with the wild type owing to the presence of miR172, a translational repressor that inhibits production of AP2 protein at the translational level (Chen, 2004). Therefore, although AP2 RNA abundance is elevated in the AP2 OE lines, AP2 protein abundance may not be elevated, and PA accumulation may not be affected. To eliminate the effect of miR172 on AP2, we generated AP2m3 overexpression plants that harbor a truly silent mutation in the miR172 binding sites (Chen, 2004). As expected, the AP2m3 overexpression plants had a pale yellow seed color and a much lower PA content (Figure 1A and 1B). Results from in situ hybridization and GUS staining showed that AP2 was strongly expressed in the seed coat endothelium, where PAs accumulate (Figure 1C–1H).

The AP2-like miR172-target clade members play a redundant role in flowering time, in part through tissue-specific expression of particular members (Yant et al., 2010). We determined whether they also play a redundant role in PA accumulation. Their transcript levels in the seed coat and phylogenetic relationships (Supplemental Figure 5) suggested that TOE1, TOE2, and TOE3 might be involved in PA accumulation. However, on the basis of genetic analyses in Arabidopsis, the AP2-like miR172-target clade members are not involved in PA accumulation like AP2, although they play a redundant role in flowering time (Supplemental Figures 3, 6, and 7). This result indicates that the miR172-target clade members have overlapping and independent roles during the same or different developmental processes.

AP2 functions not only as a direct repressor but also as a transcriptional activator in the regulation of SOC1, AG, and AGL15 during flower development (Yant et al., 2010; Dinh et al., 2012; Krogan et al., 2012). This is also the case for its regulation of PA biosynthesis. On the one hand, AP2 negatively regulates expression of late PA biosynthetic genes and the regulatory gene TT2. On the other hand, AP2 activates expression of MYBL2 (Figure 2A and 2B), and AP2 binds to the AT-rich motif of MYBL2 to directly activate its transcription (Figure 3). Our results suggest that AP2 acts as a transcriptional activator to directly promote expression of MYBL2, which in turn disrupts formation of MBW complexes to repress PA accumulation.

Interestingly, besides regulating MYBL2 at the transcriptional level, AP2 can also interact with MYBL2 at the protein level (Figure 4A, 4E, and 4G and Supplemental Figures 10 and 11). AP2 interacts with TOE3 to regulate Arabidopsis floral patterning (Jung et al., 2014), and AP2 represses the expression of its target genes AG, AP3, PI, and SEP3 by physically recruiting the co-repressor TOPLESS and the histone deacetylase HDA19 (Krogan et al., 2012). Previous studies have shown that MYBL2 also interacts with TT8, GL3, and EGL3 in yeast and possibly inhibits formation of the MBW complex in planta (Sawa, 2002; Zimmermann et al., 2004; Dubos et al., 2008; Matsui et al., 2008). Until now, no evidence has shown direct protein interactions between MYBL2 and bHLHs (TT8, GL3, and EGL3) in planta. Our present studies confirmed direct in vivo interactions between MYBL2 and TT8/GL3/EGL3 (Figure 4C, 4D, 4F, and 4G and Supplemental Figure 12), showing that MYBL2 interacts simultaneously with AP2 and TT8/GL3/EGL3 at different protein termini (Figure 4E and 4F and Supplemental Figure 13). CoIP assays further confirmed that MYBL2 can interact with AP2 and TT8/EGL3 simultaneously in planta (Figure 5A and 5B).

Previous studies have shown that the MBW complex transcriptionally activates late biosynthetic genes to regulate PA biosynthesis (Xu et al., 2014). In addition, when MYBL2 is co-expressed with the TT2–TT8–TTG1 complex, the transcript level of ANR is significantly repressed owing to the interaction of MYBL2 with TT8, which disrupts formation of the TT2–TT8–TTG1 complex (Thevenin et al., 2012). Using transient expression assays, we provided new evidence that AP2 represses ANR expression by interacting with MYBL2 to repress the transcriptional activity of the MBW (TT2–TT8–TTG1) complex (Figure 5D). Because the MBW complex activates expression of late biosynthetic genes (e.g., ANR, TT12, and TT19) in a similar manner (Xu et al., 2014), we reason that expression of the other late biosynthetic genes is also repressed similarly by AP2.

Taken together, our results suggest that AP2 regulates the PA pathway at two levels. At the transcript level, AP2 activates expression of MYBL2 by binding to an AT-rich motif, disrupting the formation of the MBW complex by directly interacting with TT8, GL3, or EGL3. At the protein level, AP2 interacts with MYBL2 to promote formation of the AP2–MYBL2–TT8/EGL3 complex, which disrupts formation of the MBW complex and in turn represses expression of PA biosynthetic genes for PA accumulation (Figure 7). In addition, the possibility that AP2 regulates PA biosynthesis via TT2 (Figure 7), which was significantly inhibited by AP2 at the transcriptional level (Figure 2B), cannot be excluded.

Figure 7.

A proposed working model for regulation of PA biosynthesis by AP2 through MYBL2.

AP2 regulates the PA pathway at two levels. At the transcript level, AP2 directly activates expression of MYBL2 by binding to an AT-rich element in the MYBL2 promoter, disrupting formation of the MBW complex by directly interacting with TT8, GL3, or EGL3. At the protein level, AP2 interacts with MYBL2 to promote formation of the AP2–MYBL2–TT8 or AP2–MYBL2–EGL3 complex, which disrupts formation of the MBW complex and in turn represses expression of late biosynthetic gene genes (e.g., ANR), finally inhibiting PA accumulation.

The dual roles played by AP2 in modulating MYBL2 represent a more efficient regulation of PA biosynthesis. Specifically, AP2 activates MYBL2 transcription to produce more MYBL2 protein, which further increases the level of the AP2–MYBL2 complex, more efficiently promoting formation of the AP2–MYBL2–TT8/EGL3 complex and inhibiting formation of the MBW complex.

Several regulatory proteins involved in mediating the accumulation of anthocyanins/PAs either disrupt or promote formation of the MBW complex. For example, TT1 interacts with TT2 and acts synergistically with the MBW complex to promote PA biosynthesis (Sagasser et al., 2002; Appelhagen et al., 2011), and TTG2 promotes PA accumulation by interacting with TTG1 (Gonzalez et al., 2016). Although these proteins interact with members of the MBW complex, they do not disrupt formation of the MBW complex. However, MYBL2 represses PA and anthocyanin accumulation by interacting with TT8/GL3/EGL3, which also disrupts formation of the MBW complex (Dubos et al., 2008; Matsui et al., 2008). CAPRICE, another R3-MYB transcription factor, regulates anthocyanin accumulation by interacting with PAP1 or PAP2, which disrupts formation of the MBW complex (Zhu et al., 2009). SPL9 negatively regulates PA biosynthesis by interacting with PAP1, which inhibits formation of the MBW complex (Gou et al., 2011). The JASMONATE-ZIM DOMAIN proteins in the jasmonate signaling pathway interact with bHLH (TT8, GL3, and EGL3) and MYB (PAP1), thereby disrupting formation of the MBW complex and repressing anthocyanin accumulation (Qi et al., 2011). These proteins interact with members of the MBW complex, disrupting its formation. In this study, AP2 was shown to repress PA biosynthesis through protein interactions that disrupt formation of the MBW complex, but unlike CAPRICE, SPL9, or JASMONATE-ZIM DOMAIN proteins, AP2 did not directly interact with members of the MBW complex. Instead, AP2 was shown to interact with an intermediary, MYBL2, to form the AP2–MYBL2–TT8/EGL3 complex, which disrupts formation of the MBW complex. By contrast, DELLA proteins in the gibberellic acid signaling pathway interact with MYBL2, leading to release of bHLHs (TT8, GL3, or EGL3) and subsequently to formation of the MBW complex (Xie et al., 2016). Although AP2 and DELLA proteins all interact with MYBL2, they regulate formation of the MBW complex in an opposite manner. Whether AP2 is associated with DELLA proteins and whether AP2 orthologs possess the same functions in PA biosynthesis in other plant species are interesting topics for further exploration. Nevertheless, the present study provides new insights into the function of the AP2–MYBL2–MBW cascade in the regulation of PA biosynthesis.

Methods

Plant materials and growth conditions

The ap2-5 (CS6239) and ap2-6 (CS6240) mutants used in the present study have been described previously (Jiang et al., 2013). The Arabidopsis transfer DNA insertion mutant lines mybl2-1 (SALK_107780) and mybl2-2 (SALK_126807) were obtained from the Arabidopsis Biological Resource Center (The Ohio State University, Columbus, OH, USA). Two MYBL2 overexpression lines (MYBL2 OE H3 and MYBL2 OE H6) have been described previously (Dubos et al., 2008). All mutants and transgenic Arabidopsis plants were in the Columbia-0 background.

For DMACA staining and analyses of PA, flavonol, and anthocyanin content, Arabidopsis plants were grown as described previously (Jiang et al., 2015) with minor modifications. For measurement of leaf flavonol content, the plants were grown in a tissue culture room at 22°C under constant light (100 μmol photons m⁻² s⁻¹) for 6 days.

Plasmid construction and Arabidopsis transformation

For generation of transgenic Arabidopsis overexpressing AP2, the AP2 open reading frame (ORF) was amplified and cloned into pENTR/SD/D-TOPO (Invitrogen, Carlsbad, CA, USA). For generation of transgenic Arabidopsis overexpressing AP2m3, TOE1m3, TOE2m3, or TOE3m3, their nucleotide sequences were mutated to the miR172-resistant version without changing their encoded amino acid sequences and then cloned into pENTR/SD/D-TOPO. For generation of transgenic lines overexpressing miR172a, a previously described fragment of miR172a (Schwab et al., 2005) was cloned into pENTR/SD/D-TOPO. All entry vectors were confirmed by sequencing and then recombined into the Gateway plant transformation vector pB2GW7; pENTR-AP2m3 was also recombined into the Gateway plant transformation vector CTAPi-GW-3×Myc.

Constructs used for plant transformation were delivered into the Agrobacterium tumefaciens strain GV3101 by the freeze–thaw method. Transgenic plants were generated by the floral dip method (Clough and Bent, 1998), and seeds were selected on Murashige and Skoog medium supplemented with 10 μg/mL glufosinate ammonium.

In situ hybridization and GUS staining

Four-day-old seeds were isolated from siliques of the wild type (Col-0),fixed in formaldehyde:glacial acetic acid:ethanol solution (1:1:18, by volume), and placed under a vacuum on ice 4 times for 30 min each time. Samples were then incubated at 4°C overnight. After dehydration in an ethanol series, 100% ethanol was replaced by xylene, and the sample was embedded in Paraplast Plus (Sigma). Paraplast sections (8-μm thick) were applied to Poly-Prep slides (Sigma) and incubated at 42°C for 2 days before hybridization. A digoxigenin-labeled antibody was used for in situ hybridization, and immunological detection was performed as described previously (Kouchi and Hata, 1993).

For GUS staining, the 5.4-kb fragment upstream of the AP2 cDNA was cloned as the AP2 promoter as described previously (Zhao et al., 2007). The promoter fragment was fused to the GUS reporter gene and then introduced into wild-type Arabidopsis. Three-day-old seeds were isolated from the transgenic lines, and GUS staining was carried out as described in our previous study (Jiang et al., 2015). After dehydration in an ethanol series, 100% ethanol was replaced with xylene, and the sample was embedded in Paraplast Plus (Sigma). Paraplast sections (8-μm thick) were applied to Poly-Prep slides (Sigma).

Gene expression analysis by qPCR

Total RNA was extracted with the Eastep Super Total RNA Extraction Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. The first-strand cDNA was synthesized using Trans Script One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China). Quantitative RT–PCR was carried out using 2× RealStar Green Fast Mixture (GeneStar, Shanghai, China) and an ABI 7500 Real-Time Detection System (Applied Biosystems, Foster City, CA, USA). The UBQ10 gene was used as a housekeeping gene. The primer sequences used in this study are listed in Supplemental Table 1. The reactions were carried out as follows: 94°C for 30 s, followed by 40 cycles of 94°C for 5 s and 60°C for 34 s. Data were calculated from biological triplicates with technical triplicates.

Analyses of PAs, flavonols, and anthocyanins

Analyses of PAs, flavonols, and anthocyanins were performed as described previously (Jiang et al., 2015) with minor modifications. For analysis of PAs, 10 μg of finely powdered dry seeds was extracted three times with 0.5 ml extraction buffer (70% acetone with 0.5% acetic acid). For analysis of flavonols, a fine powder of 10 μg dry weight was extracted with 500 μl 80% methanol overnight; the extract was hydrolyzed with the same amount of 3 N HCl at 70°C for 40 min, and then the same amount of methanol was added. Flavonol aglycones were analyzed in 100-μl samples on an HPLC 1260 instrument (Agilent Technologies, Santa Clara, CA, USA). For analysis of anthocyanins, 20 μg of fine powder was extracted with 500 μl extraction buffer (80% methanol with 0.1% hydrochloric acid). The contents of PAs, flavonols, and anthocyanins in the wild type were assigned a value of 1.0, and their levels in mutant and transgenic lines were compared with those of the wild type in triplicate.

Transient transcription dual-luciferase assays

The ORF of AP2m3 and the promoter regions of 25 PA pathway genes were cloned into pENTR/SD/D-TOPO and confirmed by sequencing. For mutation of the AT-rich motifs (TTTGTT or AACAAA) in the MYBL2 promoter, mutated promoters in which AT-rich motifs M1–M5 were mutated into CCCCCC were cloned into pENTR/D/TOPO and confirmed by sequencing. The resulting pENTR-AP2m3 was subcloned into the modified destination vector p2GW7 by LR reaction to generate p2GW7-AP2m3. The resulting pENTR-promoter vectors containing individual promoters were subcloned into the modified destination vector p2GW7-LUC by LR reaction to generate p2GW7-promoter vectors (Karimi et al., 2002). The ORFs of TT2, TT8, TTG1, and MYBL2 were cloned into p2GW7. Leaves of 4-week-old Arabidopsis seedlings were used for protoplast isolation as described previously (Yoo et al., 2007). The effector construct (p2GW7-gene) driven by the 35S promoter was co-transfected with each individual reporter construct (p2GW7-promoter) into Arabidopsis protoplasts as described previously (Yoo et al., 2007). A reference construct containing the Renilla luciferase gene driven by the 35S promoter was also co-transfected to measure transfection efficiency. Luciferase activities were measured using the dual-luciferase reporter assay system (Promega), and the firefly luciferase activity was calculated by normalizing against the Renilla luciferase activity in each transfection event. Data are presented as mean ± standard deviation of biological triplicates.

Yeast two-hybrid assays

The ORFs and different truncations of AP2 and MYBL2 were amplified and cloned into the gateway destination vector pGBKT7-DEST with a binding domain (Clontech, Mountain View, CA, USA). ORFs of AP2 and TT2, MYB5, TT8, GL3, EGL3, TTG1, TTG2, MYBL2, TT1, TT16, and STK of the PA pathway were cloned into pGADT7-DEST with an activation domain (Clontech). Yeast strain AH109 cells were co-transformed with specific bait and prey constructs using the LiCl–polyethylene glycol method according to the manufacturer’s instructions (Clontech). The transformants were grown on synthetic SD/−Trp/−Leu dropout medium and further selected on SD/−Trp/−Leu/−His/−Ade medium with 3-amino-1,2,4-triazole at 30°C for 7 days.

BiFC assays

The ORFs of AP2m3, TT8, GL3, and EGL3 were cloned into the Pac I and Spe I restriction sites of the pBI-2YN-CAT vector. The ORFs of MYBL2, TT2, MYB5, TT8, GL3, EGL3, and TTG1 and the C-terminal and N-terminal truncations of MYBL2 were cloned into the Pac I and Spe I restriction sites of the pBI-2YC-CAT vector. Agrobacterium cells containing each construct were incubated at 28°C in Luria-Bertani medium with shaking at 200 rpm until the optical density value reached 0.6–0.8. Agrobacterium cells were then centrifuged for 10 min at 4000 rpm and re-suspended in induction medium (10 mM 2-(N-Morpholino) ethane sulfonate buffer [pH 5.6], 10 mM MgCl₂, and 150 μM acetosyringone), then co-infiltrated into epidermal cells of young leaves of 4-week-old N. benthamiana. YFP fluorescence was detected 40–48 h after infiltration by laser-scanning confocal microscopy (Leica TCS SP5). Emission was collected at 520–550 nm for YFP and at 600–630 nm for mCherry.

CoIP assays

AP2m3 and MYBL2 were cloned into modified pCAMBIA1302 vectors containing the 3×Myc tag and the 3×FLAG tag, respectively. TT8, GL3, and EGL3 were cloned into a modified pCAMBIA1302 vector containing the 3×HA tag. All primers are listed in Supplementary Table 1. To verify the interaction of AP2 with MYBL2, 4-week-old N. benthamiana leaves were infiltrated with A. tumefaciens GV3101 harboring Pro35S:MYBL2-3×FLAG and Pro35S:AP2m3-3×Myc. To test the interaction of MYBL2 with TT8, GL3, and EGL3, GV3101 cells harboring Pro35S:MYBL2-3×FLAG combined with Pro35S:TT8-3×HA, Pro35S:GL3-3×HA, or Pro35S:EGL3-3×HA, respectively, were infiltrated into N. benthamiana leaves. These leaves were harvested 36–48 h after infiltration. Total proteins were extracted from leaves with coIP buffer (100 mM Tris–HCl, 300 mM NaCl, 2 mM EDTA, 10% glycerol, 1% Triton X-100, and 1% protease inhibitor cocktail [pH 7.5]). An anti-Myc tag antibody or anti-HA tag antibody (Abcam, Cambridge, UK) was added to protein extracts and incubated at 4°C for 3 h. Immuno-complexes captured by the beads were washed three times with coIP buffer.

ChIP assays

Wild-type Arabidopsis and transgenic Arabidopsis overexpressing AP2m3-Myc were grown at 23°C under 16-h light and 8-h dark conditions. Siliques of 3- to 5-day-old plants were cross-linked for 15 min in 1% formaldehyde by vacuum infiltration. The ChIP analysis was performed using an anti-Myc tag antibody (Abcam), and quantitative real-time PCR was performed using SYBR Green reagent (GeneStar) and the ABI 7500 Real-time Detection System (Applied Biosystems). Results are presented as the ratio of the amount of DNA immunoprecipitated from AP2m3-Myc samples to that of the wild type in triplicates. The PP2A and UBC30 genes were used as negative controls.

Accession numbers

Sequence data in this article can be found at The Arabidopsis Information Resource under the following accession numbers: AP2 (At4G36920), TOE1 (At2G28550), TOE2 (At5G60120), TOE3 (At5G67180), SMZ (At3G54990), SNZ (At2G39250), miR172a (At2G28056), UBQ10 (At4G05320), PP2A (At1G13320), UBC30 (At5G56150), CHS (At5G13930), CHI (At3G55120), CHIL (At5G05270), F3H (At3G51240), F3′H (At5G07990), DFR (At5G42800), ANS (At4G22880), ANR (At1G61720), TT12 (At3G59030), TT19 (At5G17220), AHA10 (At1G17260), TT10 (At5G48100), TT2 (At5G35550), MYB5 (At3G13540), TT8 (At4G09820), GL3 (At5G41315), EGL3 (At1G63650), TTG1 (At5G24520), MYBL2 (At1G71030), TTG2 (At2G37260), TT1 (At1G34790), TT15 (At1G43620), TT9 (At3G28430), TT16 (At5G23260), and STK (At4G09960).

Funding

This work was supported by grants from the National Natural Science Foundation of China (31870281 to W.J.) and the Agricultural Science and Technology Innovation Program (ASTIP-IAS10).

Author contributions

W.J. and Y.P. designed the research. W.J., Q.Y., J.L., and X.S. performed the analyses of proanthocyanidins, flavonols, and anthocyanins. W.J., Q.Y., and X.H. performed the transient transcriptional activity assay. J.Z. generated plants overexpressing the AP2-like miR172-target clade genes. W.J. and J.Z. performed the Y2H assay and ChIP assay. W.J. and Q.L. performed gene expression analysis, the BiFC assay, and the coIP assay. W.J. analyzed the data and interpreted the results. W.J. and Y.P. wrote the article.

Published: December 5, 2023