TOP1α regulates FLOWERING LOCUS C expression by coupling histone modification and transcription machinery

Chromatin structure is a crucial determinant in the regulation of DNA replication and transcription. Opening the DNA helical structure in highly compact chromatin to recruit the transcription initiation complex and establishing the active chromatin state are two main steps of RNA polymerase II (RNAPII)-mediated transcription in all living cells. Topoisomerase 1 (TOP1) catalyzes the transient cleavages of one strand of DNA and participates in the relaxation of the supercoiled structure of transcription templates (Champoux, 2001). In plants, TOPOISOMERASE 1α (TOP1α), which encodes a type IB topoisomerase, was first isolated in Arabidopsis thaliana by its homology with yeast and human TOP1, and its ability to complement the phenotype of yeast top1 mutants (Kieber et al., 1992). Mutations in this gene were later found to exhibit varied defects in several aspects of plant development, including primordia initiation and phyllotaxis (Laufs et al., 1998; Takahashi et al., 2002), and stem cell homeostasis in the shoot and floral meristems (Graf et al., 2010; Liu et al., 2014). Recently, TOP1α has been shown to play an important role in controlling flowering time by the direct activation of FLOWERING LOCUS C (FLC) and its homologs (Gong et al., 2017). It has been widely proved that histone methylations, such as H3K4me3 (He et al., 2004; Pien et al., 2008; Shafiq et al., 2014; Tamada et al., 2009) and H3K36me3 (Kim et al., 2005; Xu et al., 2008; Zhao et al., 2005), are required for generating the permissive chromatin state to maintain high-level transcriptions of FLC. However, how TOP1α modulates DNA topology to activate transcription and establish a permissive histone modification facilitating FLC transcription is largely unknown. Here, we show that TOP1α directly binds the H3K36me methyltransferase SET DOMAIN GROUP 8 (SDG8) and recruits RNAPII to facilitate FLC transcription elongation. Accordingly, in top1α and sdg8 mutants, transcripts of FLC were dramatically decreased in the early stages of transcription elongation. Our study reveals that TOP1α, together with the H3K36me methyltransferase and RNAPII, controls the main steps of early transcription of FLC.

The top1α mutant exhibits an early flowering phenotype in both long-day and short-day conditions by directly repressing the expression of FLC (Fig. S1) (Gong et al., 2017). Moreover, FLC showed significant epistasis effects in the top1α mutant, in which early flowering phenotypes were completely suppressed by the overexpression of FLC (Gong et al., 2017). To genetically test the interaction between FLC and TOP1α, we crossed the top1α-1 mutant (hereafter top1α) with flc-3 in the Col background and observed that the top1α flc-3 double mutant shows early flowering phenotypes similar to top1α or flc-3 single mutants based on mean total rosette leaf number or flowering time (Fig. 1A,B). To confirm this observation, we also analyzed the flowering phenotypes of the top1α flc-3 double mutant in the FRIGIDA (FRI) background, which contains high levels of FLC expression at the vegetative stage (Johanson et al., 2000; Michaels and Amasino, 2001). The top1α or flc-3 single mutants in the FRI background flowered significantly earlier than wild-type FRI plants (Fig. 1A,B). Likewise, we also observed the same early flowering phenotypes in top1α flc-3 double mutants in the FRI background (Fig. 1A,B). In agreement with these observations, FLC expression levels were observed to be downregulated in 8-day-old seedlings of these three mutants in both backgrounds (Fig. 1C,D).

FLC transcripts are more abundant in young tissues of shoot tips (Michaels and Amasino, 1999; Sheldon et al., 2002; Sung and Amasino, 2004). To test whether TOP1α proteins are colocalized with FLC transcripts in Arabidopsis, we transformed the TOP1α::TOP1α-GFP plasmid into top1α mutants and observed the full complementation of top1α early flowering phenotypes (Fig. S2). As predicted, we observed that both the mRNA and protein of TOP1α mainly accumulated in the shoot apical meristem and young leaves (Fig. S3), which was similar to FLC.

Histone modifications, such as H3K4me and H3K36me, are major regulators of chromatin structure at the FLC locus and are essential for maintaining high levels of FLC transcription (He et al., 2004; Kim et al., 2005; Pien et al., 2008; Shafiq et al., 2014; Tamada et al., 2009; Xu et al., 2008; Zhao et al., 2005). Thus, we further investigated whether TOP1α-mediated changes in DNA topological structure affect H3K4 and H3K36 methylations at the FLC locus. By performing chromatin immunoprecipitation (ChIP) assays, we observed that H3K4me3 enrichment in the top1α mutant was not significantly different compared with that of the wild-type plant at the FLC locus (Fig. 1E). However, levels of H3K36me2 and H3K36me3 were dramatically reduced in top1α mutants, especially in the 500-1500 bp region downstream of the transcription start sites (TSSs), rather than evenly reduced in the entire gene body (Fig. 1F,G), suggesting that TOP1α is involved in regulating di- and tri-methylation levels of H3K36 at the FLC locus in the early phase of transcription elongation.

To shed light on the mechanism underlying the ability of TOP1α to regulate H3K36 methylation, we tested whether TOP1α directly recruits SDG8 methyltransferase, which deposits H3K36me2 and H3K36me3 at the FLC locus. As the mRNA level of SDG8 was not affected in the top1α mutant (Fig. S4A), we performed yeast two-hybrid assays to examine whether TOP1α (full-length) interacts with the SDG8 fragment from amino acids 335-569 and indeed observed an interaction between TOP1α and SDG8 in yeast cells (Fig. 2A). To test their physical interaction in vitro, we performed pull-down assays. Because it is difficult to express the soluble protein of full-length TOP1α in Escherichia coli cells, we divided TOP1α into three truncated fragments, TOP1α-1 (amino acids 1-370), TOP1α-2 (370-582) and TOP1α-3 (510-933), and fused the fragments with His-tag (Fig. 2B). We mixed each of these fragments with purified recombinant GST-SDG8 protein in vitro and could only detect GST-SDG8 in the presence of His-TOP1α-1 after the His-trapped pull-down, but not in the presence of His-TOP1α-2 or His-TOP1α-3 (Fig. 2C), suggesting that SDG8 directly interacts with the N-terminus of TOP1α. To further confirm the interaction between SDG8 and TOP1α-1 in vivo, we performed bimolecular fluorescence complementation (BiFC) experiments in tobacco leaves and observed an interaction between N-terminal enhanced yellow fluorescent protein (eYFP)-fused TOP1α-1 and C-terminal eYFP-fused SDG8 (Fig. 2D). However, the TOP1α-2 fragment did not exhibit any interaction with SDG8 in plants (Fig. 2E). These data suggest that the N-terminus of TOP1α directly interacts with SDG8 in plants.

To test the interaction between TOP1α and SDG8 genetically, we analyzed flowering phenotypes of top1α, sdg8 and top1α sdg8 mutants. Consistent with previous observations (Gong et al., 2017; Zhao et al., 2005), top1α and sdg8 mutants showed early flowering phenotypes (Fig. 2F,G). We observed the same early flowering phenotypes in top1α sdg8 mutants based on the total rosette leaf number and flowering time (Fig. 2F,G). Moreover, FLC expression levels were downregulated in 8-day-old seedlings of top1α sdg8 mutants, as well as the corresponding single mutants (Fig. 2H), suggesting that TOP1α and SDG8 act in the same genetic pathway to control flowering time.

Early studies in animal cells have demonstrated that TOP1 proteins are deposited at TSSs, where they interact with the phosphorylated carboxyl terminal domain (CTD) of the largest subunit of RNAPII to facilitate transcription (Baranello et al., 2016; Carty and Greenleaf, 2002; Kouzine et al., 2013; Teves and Henikoff, 2014; Wu et al., 2010). In Arabidopsis, knocking out TOP1α functions causes a significant reduction of RNAPII enrichment at the transcription start site of FLC (Gong et al., 2017), whereas the mRNA (Fig. S4B) and protein (Gong et al., 2017) levels are not affected in the top1α mutant. To elucidate the molecular mechanism of TOP1α promotion of RNAPII recruitment in plants, we tested whether TOP1α can physically interact with the CTD of RNAPII. By yeast two-hybrid and co-immunoprecipitation (Co-IP) assays, we observed the interaction between CTD and full-length TOP1α in yeast (Fig. 3A) and plant cells (Fig. 3B). As we showed above, SDG8 directly interacted with TOP1α at the N-terminus of TOP1α-1 (Fig. 2C). To explore further the binding fragment of TOP1α with CTD, the same three truncated forms of TOP1α were used for pull-down assays. We could not detect direct binding of GST-CTD with TOP1α-1, TOP1α-2 or TOP1α-3 (Fig. 3C).

Given that TOP1α directly bonds with SDG8, we then investigated whether TOP1α could recruit RNAPII via SDG8 to form a complex. To this end, we first confirmed the direct interaction between CTD and SDG8 by using yeast two-hybrid (Fig. 3A), pull-down (Fig. 3D) and BiFC assays (Fig. 3F). Then we performed TOP1α pull-down assays with GST-CTD in the presence of GST-SDG8, and observed two immunoprecipitated bands representing SDG8 and CTD (Fig. 3E), suggesting that TOP1α interacts with RNAPII indirectly via SDG8. We further performed BiFC-based fluorescence resonance energy transfer (FRET) to examine the interaction of three proteins at the same time in plants. We co-transformed 35S::CFP-TOP1α-1, 35S::nYFP-CTD and 35S::cYFP-SDG8 into tobacco leaves, and observed energy transfer from CFP to YFP (Fig. S5), suggesting that TOP1α, CTD and SDG8 are in the same complex.

RNAPII enrichment in the top1α mutant was found to be significantly reduced, including at the first intron where the major regulatory region for epigenetic modifications is located (Ausín et al., 2004; Bastow et al., 2004; He et al., 2004, 2003; Sheldon et al., 2002; Sung and Amasino, 2004), indicating a role for TOP1α in transcription beyond its recruitment of RNAPII, although it is unknown if and how TOP1α activity is involved in the transcription elongation of FLC. To test this hypothesis functionally, we measured the transcriptional levels of FLC across the entire locus by designing a series of primers from the TSS to the last exon. We observed, surprisingly, that in the top1α mutant, transcripts of FLC were not reduced evenly across the entire locus but rather a dramatic decrease of over 80% at the beginning of transcription elongation, 300-500 bp downstream from the TSS, was observed with relatively stable expression levels thereafter (Fig. 4A). This non-uniform transcription pattern of FLC in the top1α mutant strongly suggests that TOP1α participates in the early phase of transcription elongation. As TOP1α directly interacts with SDG8, which deposited H3K36 methylation in the FLC, we then asked whether downregulation of H3K36 methylation might also cause the same reduction of FLC at the early stages of elongation. In the sdg8 mutant, we observed very similar transcription patterns along the FLC locus, with transcripts decreasing by 90% in the same region as observed for top1α (Fig. 4B), demonstrating that H3K36 methylation is functionally linked with early transcription elongations. Consistent with this observation, we found a major enrichment peak of H3K36me3 in the FLC locus, which was located at the beginning of the first intron (region 5, approximately 500 bp away from the TSS) (Fig. 4C). The di-methylated forms of H3K36 were mainly deposited in the middle of the first intron 1500 bp away from the TSS (Fig. 4D). In the top1α mutant, the H3K36me3 deposition in FLC was mainly reduced in the region of its highly enriched peak (Fig. 1G). From the TSS to this region we observed a rapid decrease of FLC transcripts in top1α (Fig. 4A) and sdg8 (Fig. 4B) mutants. Given that TOP1α interacts with SDG8 and CTD, we conclude that TOP1α regulates early phases of transcription elongation in FLC by recruiting H3K36 methyltransferase and RNAPII.

The early phases of transcription are coupled with recruitment of the transcription initiation complex and movement of the transcription machinery along the gene bodies. Our data are not consistent with the observation in animals that TOP1α directly interacts with RNAPII to form the transcription complex. In contrast, in plants TOP1α recruits RNAPII via SDG8 to promote early transcription elongation by maintaining the open chromatin state via H3K36 methylation (Fig. 4E), suggesting that TOP1α-mediated topological relaxation of the DNA template and recruitment of permissive histone modifications are essential for the early transcription elongation of FLC.

TOP1 in animals has been reported to interact with RNAPII to regulate transcription pauses (Baranello et al., 2016). We found that TOP1α in plants interacts with SDG8 to recruit RNAPII at the N-terminus. The N-terminal domain of TOP1α is a plant-specific fragment that is not conserved in animals (Fig. S6), and its function is not known. Our data indicate that the recruitment of histone modifications to regulate transcription might be a unique feature of plant type IB topoisomerases. Interestingly, in plants, TOP1α is also required for silencing Polycomb group (PcG) target genes by increasing the repressive mark of H3K27me3 at the promoter region (Liu et al., 2014), or transposable elements by RNAPV-dependent RNA-directed DNA methylation (RdDM) and H3K9me2 (Dinh et al., 2014). Therefore, it remains to be determined whether TOP1α directly recruits a variety of histone methyltransferases and different types of RNA polymerases to either positively or negatively regulate gene expression at different transcriptional stages, and how these activities are specified during plant development.

All plants were in the Columbia-0 (Col-0) background except for FRI (CS6209), which was obtained from Arabidopsis Biological Resource Center (ABRC). The top1α mutant (top1α-1) was kindly provided by Prof. Taku Takahashi (Okayama University, Japan). The sdg8 mutant (SALK_036941) and flc-3 were kindly provided by Prof. Yong Ding (University of Science and Technology of China, China). The FRIflc-3 mutant was kindly provided by Prof. Ya-Long Guo (Institute of Botany, Chinese Academy of Sciences, China). All mutants of FRItop1α, FRItop1αflc-3, top1αsdg8 and top1αflc-3 were obtained by crossing the above mutants. All seeds were sterilized by applying 70% ethanol and 0.5% Tween 20 for 10 min, followed by two washes with 95% ethanol and air drying. Plants were grown at 21°C under long-day conditions (16 h of light and 8 h of darkness). All materials for quantitative RT-PCR and ChIP-PCR were 8-day-old seedlings grown on 1/2 Murashige and Skoog (MS) medium containing 0.8% agar and 0.5% sucrose under long-day conditions.

Seedlings were collected and immediately transferred to liquid nitrogen. The Tripure Isolation Reagent (Roche, 94002420) was used to isolate total RNA from plant samples. The PrimeScript RT Reagent Kit (TaKaRa, RR047A) was used for cDNA synthesis. Primers used for qRT-PCR were designed to amplify products that were 100-300 bp in length; gene-specific primer sequences are listed in Table S1. Quantitative PCR was performed with the Thermo PIKO REAL96 Real-Time PCR system using the GoTaq qPCR Master Mix (Promega) with the following conditions: 95°C for 5 min; 40 cycles of 95°C for 10 s, 57°C for 30 s and 72°C for 30 s, followed by 72°C for 10 min. TUBULIN was used to normalize mRNA levels.

The primer pairs used for quantitative RT-PCR (Fig. 4A,B) were: 2, −269 bp upstream of ATG; 3, −200 bp upstream of ATG; 4, 71 bp downstream of ATG; 5, 201 bp downstream of ATG; 6, 688 bp downstream of ATG; 7, 1306 bp downstream of ATG; 8, 2562 bp downstream of ATG; 9, 5506 bp downstream of ATG.

Chromatin immunoprecipitation (ChIP) experiments were performed as previously described (Leibfried et al., 2005). Total chromatin was extracted from 8-day-old seedlings and the Diagenode Bioruptor UCD-200 was used for sonication (30 s on, 30 s off, medium, 40 min duration; sonication buffer: 10 mM Na₃PO₄, 100 mM NaCl, 0.5% sarkosyl, 10 mM EDTA, 1 mM PMSF, protease inhibitor, 1 tablet per 10 ml, pH 7). Chromatin was immunoprecipitated with anti-GFP (Abcam, ab290), anti-H3K4me3 (Abcam, ab8050), anti-H3K36me2 (Abcam, ab9049), anti-H3K36me3 (Abcam, ab9050) and anti-CTD (Abcam, ab1791) antibodies used at 1:200. Quantitative real-time PCR was conducted to measure the amounts of immunoprecipitated fragments of genes of interest on the Thermo PIKO REAL96 Real-Time PCR system using a GoTaq qPCR Master Mix (Promega), and each ChIP sample was quantified in triplicate. The primer pairs used for ChIP-qPCR in Fig. 1E,F and Fig. 4C,D were: 1, −662 bp upstream of ATG (+1); 5, 201 bp downstream of ATG; 6, 688 bp downstream of ATG; 7, 1306 bp downstream of ATG; 8, 2562 bp downstream of ATG; and 10, 5730 bp downstream of ATG. The TSS was located −269 bp upstream of ATG (+1). The primer sequences used for ChIP are listed in Table S1.

The yeast two-hybrid assay was performed according to the standard protocol of Clontech (Clontech, user manual 630489). Saccharomyces cerevisiae strain AH109 was co-transformed with the bait and prey constructs of pGBKT7-SDG8 (amino acids 335-569) and pGADT7-TOP1α (full length); pGBKT7-TOP1α (full length) and pGADT7-CTD; and pGBKT7-SDG8 and pGADT7-TOP1α. Vectors without coding region insertions were used as negative controls. The growth of yeast cells on synthetic defined (SD) medium lacking Trp, Leu, His and adenine was used to detect the interaction.

Total protein was purified from E. coil (Rosetta) that were transformed by the following constructs: 6×His-MBP-TOP1α-1 (from amino acids 1-370), 6×His-MBP-TOP1α-2 (370-582), 6×His-MBP-TOP1α-3 (510-933), 6×His-MBP-CTD, GST-SDG8 (335-569) and GST-CTD.

Glutathione-Sepharose beads (GE Healthcare; Lot: 10236606) were used for the purification of GST-tagged proteins by washing solution [50 mM Tris-HCl, 200 mM NaCl, 1 mM reduced glutathione, 0.2% (v/v) Triton X-100, pH 8.0] and elution solution (50 mM Tris-HCl, 150 mM NaCl, 20 mM reduced glutathione, pH 8.0). Ni-Sepharose beads (GE Healthcare; Lot: 10233021) were used for the purification of His-tagged proteins by washing solution (50 mM Tris-HCl, 200 mM NaCl, 10 mM imidazole, pH 8.0) and elution solution (50 mM Tris-HCl, 150 mM NaCl, 250 mM imidazole, pH 8.0).

Beads were incubated with bait proteins at 4°C for 1 h with 5% skim milk powder in the pull-down solution (20 mM Tris-HCl pH 8.0, 200 mM NaCl, 1 mM EDTA pH 8.0, 0.25% NP-40 and 25 ng/μl PMSF) and then beads were washed several times with the same pull-down solution. Beads were then incubated with 5 µg of soluble protein in 600 μl pull-down solution for 3 h at 4°C. Mock controls included extracts prepared from either the His-Tag or GST-tag vectors. The supernatant was collected as input. The beads were washed five to eight times with pull-down solution, separated on an SDS-PAGE gel, and detected by anti-GST antibody (GenScript, A00866-100, Lot: 13D000626).

Eight-day-old seedlings of Col-0 and pTOP1α:: TOP1α-GFP/top1α were used for Co-IP assays. One gram of seedlings was harvested and ground in liquid nitrogen, mixed with 2 ml protein extraction buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1% (v/v) Triton X-100, 10% (v/v) glycerin, 1 mM PMSF (added fresh) and 1 mM Protease Inhibitor Cocktail (Okayama University, Japan; added fresh). After 3 h on ice and centrifugation at 12,000 g, 100 μl supernatant (total protein) was used as the input, and 30 μl protein A beads (Invitrogen, 1002D) were added to the total protein and incubated on ice for 1 h. To detect protein interactions, 2 μg anti-GFP antibody (Abcam, ab290) was added to the supernatant and incubated overnight. After adding 50 μl protein A beads and incubating for 3 h, beads were washed three times with protein extraction buffer before being separated by SDS-PAGE, and analyzed by western blot using an anti-CTD antibody (Abcam, ab1791 1:1000).

Nicotiana benthamiana were grown in growth chambers under long-day conditions at 21°C. Agrobacterium was transformed with relevant binary vectors of 35S::nYFP-TOP1α-1, 35S::nYFP-TOP1α-2, 35S::nYFP-CTD, 35S::cYFP-CTD and 35S::cYFP-SDG8. A 1:1 mixture of two different bacteria containing plasmids of interest was used to infiltrate the abaxial leaves of 3- to 4-week-old N. benthamiana. Epidermal cells of N. benthamiana were examined 48-72 h after infiltration with a spectral confocal laser-scanning microscope (Leica, LSM710).

Agrobacterium was transformed with relevant binary vectors of 35S::CFP-TOP1α-1, 35S::nYFP-CTD and 35S::cYFP-SDG8. A 1:1:1 mixture of three different bacteria containing plasmids of interest was used to infiltrate the abaxial leaves of 3- to 4-week-old N. benthamiana. Epidermal cells of N. benthamiana were examined 72 h after infiltration with a spectral confocal laser-scanning microscope (Leica, LSM710). The excitation wavelength used to excite the donor molecule (CFP) was 448 nm from an argon ion laser, and 514 nm was used to acquire the acceptor (YFP) image.

The pTOP1α:: TOP1α-GFP was used to transform top1α mutants by the floral-dip method (Clough and Bent, 1998). Those rescued transgenic plants that carried a single insertion in the genome were used for further analysis.

The template of TOP1α was amplified by PCR with specific primers containing T7 and T3 promoter sequences. RNA probes were synthesized by T7/T3 polymerase and labeled with digoxin-UTP. Then, in situ hybridization was performed according to standard protocols (Andersen et al., 2008; Weigel and Jürgens, 2002).

Eight-day-old seedlings of pTOP1α:: TOP1α-GFP/top1α rescue plants were embedded in 6% low melting agarose (Promega, V2111), and sectioned with a vibratome (Leica, VT 1200S) at 100 μm thickness. The sections were collected and immediately imaged with a confocal microscope (Leica, LSM710).

We thank Prof. Taku Takahashi, Prof. Yong Ding and Prof. Ya-Long Guo for sharing mutant seeds.