PHOSPHORYLETHANOLAMINE CYTIDYLYLTRANSFERASE 1 (PECT1) regulates phosphatidylethanolamine biosynthesis and controls the phosphatidylethanolamine:phosphatidylcholine ratio in Arabidopsis thaliana. Previous studies have suggested that PECT1 regulates flowering time by modulating the interaction between phosphatidylcholine and FLOWERING LOCUS T (FT), a florigen, in the shoot apical meristem (SAM). Here, we show that knockdown of PECT1 by artificial microRNA in the SAM (pFD::amiR-PECT1) accelerated flowering under inductive and even non-inductive conditions, in which FT transcription is almost absent, and in ft-10 twin sister of ft-1 double mutants under both conditions. Transcriptome analyses suggested that PECT1 affects flowering by regulating SHORT VEGETATIVE PHASE (SVP) and GIBBERELLIN 20 OXIDASE 2 (GA20ox2). SVP misexpression in the SAM suppressed the early flowering of pFD::amiR-PECT1 plants. pFD::amiR-PECT1 plants showed increased gibberellin (GA) levels in the SAM, concomitant with the reduction of REPRESSOR OF GA1-3 levels. Consistent with this, GA treatment had little effect on flowering time of pFD::amiR-PECT1 plants and the GA antagonist paclobutrazol strongly affected flowering in these plants. Together, these results suggest that PECT1 also regulates flowering time through a florigen-independent pathway, modulating SVP expression and thus regulating GA production.
The timing of flowering is precisely regulated by external and internal cues (Andrés and Coupland, 2012). Arabidopsis thaliana is a facultative long-day (LD) plant; the LD photoperiod allows the plant to produce sufficient florigen to trigger flowering (Srikanth and Schmid, 2011). In Arabidopsis, the florigen is encoded by FLOWERING LOCUS T (FT) and its paralogue TWIN SISTER OF FT (TSF) (Kardailsky et al., 1999; Kobayashi et al., 1999; Yamaguchi et al., 2005). FT and TSF transcription is promoted in phloem companion cells in the leaf by CONSTANS (CO), which is a nuclear-localized zinc-finger transcription factor (An et al., 2004; Wenkel et al., 2006). FT and TSF are transported to the shoot apical meristem (SAM), in which they activate the expression of SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1) and APETALA1 (AP1) and initiate floral primordia development (Andrés and Coupland, 2012; Corbesier et al., 2007).
Under non-inductive short-day (SD) conditions, Arabidopsis plants show late flowering (Koornneef et al., 1991). In the light, CO is stabilized by the circadian clock component PSEUDO RESPONSE REGULATOR to promote FT and TSF transcription; in the dark, CO is degraded by the RING motif-containing E3 ligase CONSTITUTIVE PHOTOMORPHOGENIC 1 (Hayama et al., 2017; Jang et al., 2008; Liu et al., 2008). Therefore, under non-inductive conditions, the low amount of CO limits FT and TSF expression to levels that are not sufficient to trigger flowering (Takada and Goto, 2003; Yamaguchi et al., 2005).
Flowering under SD conditions is regulated by the gibberellin (GA)- and age-dependent pathways (Srikanth and Schmid, 2011). In the SAM, GA accelerates flowering and floral development by activating the transcription of downstream targets including LEAFY (LFY), SOC1, FRUITFULL (FUL), SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 3 (SPL3), SPL5 and SPL9 (Blázquez et al., 1998; Galvao et al., 2012; Hyun et al., 2016; Moon et al., 2003; Yu et al., 2012). Furthermore, SPL15 plays an important role downstream of GA in promoting the floral transition under SD conditions through activation of its target genes in cooperation with SOC1 (Hyun et al., 2016). During the vegetative phase, SHORT VEGETATIVE PHASE (SVP) counteracts the effect of GA in promoting flowering (Andrés et al., 2014) by repressing the crucial GA biosynthesis gene GIBBERELLIN 20 OXIDASE 2 (GA20ox2) in the SAM. SVP also represses the expression of SOC1 and SEPALLATA3 (SEP3) to delay flowering (Li et al., 2008; Liu et al., 2009).
FT is a member of the conserved phosphatidylethanolamine binding protein (PEBP) family (Bernier and Jollés, 1984; Serre et al., 1998). FT was suggested to interact with phosphatidylcholine (PC) (Nakamura et al., 2014); PC is generated from phosphoethanolamine, a precursor of phosphatidylethanolamine (PE), and PC levels undergo diurnal oscillations. In Arabidopsis, the rate-limiting enzyme PHOSPHORYLETHANOLAMINE CYTIDYLYLTRANSFERASE 1 (PECT1) modulates the PC:PE ratio (Mizoi et al., 2006). Knocking down PECT1 using artificial microRNA specifically in the SAM (pFD::amiR-PECT1) increases PC levels and thus elevates SOC1 and AP1 expression levels, which eventually accelerates flowering time by PC-FT binding (Mizoi et al., 2006; Nakamura et al., 2014). Therefore, it was suggested that the PC-FT interaction in the SAM is important for initiation of flowering by FT (Nakamura et al., 2014). However, the early flowering of pFD::amiR-PECT1 plants under LD conditions was only partially suppressed in the ft tsf mutant background (Nakamura et al., 2014), suggesting that PECT1 also regulates other flowering pathways.
Here, we report a novel role of PECT1 in the regulation of flowering time. PECT1 regulates flowering through transcriptional regulation of the floral repressor SVP in the SAM. The pFD::amiR-PECT1 transgenic lines showed lower expression levels of SVP and higher expression levels of its downstream targets GA20ox2, SOC1 and SEP3, resulting in early flowering, even under non-inductive conditions such as short photoperiod and low ambient temperature, in which FT transcripts are almost absent or present at very low levels. Our results suggested that, in addition to modulating flowering through its effect on PC, PECT1 also modulates flowering through a florigen-independent pathway via the regulation of SVP and GA20ox2 expression.
RESULTS AND DISCUSSION
SAM-specific knockdown of PECT1 accelerates flowering independent of florigen
To investigate the role of PECT1 in flowering time regulation under different environmental conditions, we measured the flowering time of pFD::amiR-PECT1 plants, in which PECT1 expression in the SAM is targeted by an artificial miRNA (Nakamura et al., 2014), at 23°C (LD and SD) and 16°C (LD). We confirmed significant downregulation of PECT1 in the SAM-enriched samples of pFD::amiR-PECT1 plants, suggesting the lines are suitable for subsequent studies (Fig. S1A).
By counting the total number of leaves at flowering to quantify flowering time, we observed that pFD::amiR-PECT1 plants flowered earlier than wild-type (WT) Col-0 plants under all conditions tested (Fig. 1A; Table S1) (Nakamura et al., 2014). In particular, pFD::amiR-PECT1 plants (lines #2, #4 and #5) flowered earlier than WT under 16°C LD and 23°C SD conditions (Fig. 1A,B), in which the expression levels of FT and TSF were very low (Fig. S2). Under 23°C SD conditions, WT plants flowered with about 50 leaves, whereas pFD::amiR-PECT1 lines flowered with 15-17 leaves (Fig. 1A). The ratios of the number of leaves at flowering at different temperatures (16°C/23°C LD) and under different photoperiods (23°C SD/LD) of pFD::amiR-PECT1 lines #2, #4 and #5 were lower than those of WT plants, indicating that pFD::amiR-PECT1 plants were temperature- and photoperiod-insensitive (Fig. 1C). Consistent with their accelerated flowering, we observed an increase in the mRNA levels of the floral integrator gene SOC1 and the floral meristem identity gene AP1 in pFD::amiR-PECT1 lines under 23°C LD and 23°C SD conditions (Fig. S3).
The finding that the pFD::amiR-PECT1 plants flowered earlier than WT plants under non-inductive conditions raised the possibility that PECT1 also regulates flowering time in a florigen-independent manner. Therefore, we examined the effect of pFD::amiR-PECT1 in ft-10 tsf-1 double mutants, which lack both florigens. Under 23°C LD conditions, pFD::amiR-PECT1 ft-10 tsf-1 plants flowered earlier than ft-10 tsf-1 mutants (31.2±1.5 versus 55.8±4.9 leaves; mean±s.d.) (Fig. 1D,E; Table S1), consistent with a previous report (Nakamura et al., 2014). In particular, pFD::amiR-PECT1 ft-10 tsf-1 plants flowered earlier than ft-10 tsf-1 mutants under 16°C LD (34.0±1.7 versus 60.5±3.0 leaves) and 23°C SD conditions (39.9±4.0 versus 72.0±4.4 leaves), revealing that acceleration of flowering in pFD::amiR-PECT1 plants is also independent of FT and TSF.
RNA-seq analyses identified SVP and GA20ox2 as candidates for the accelerated flowering in pFD::amiR-PECT1 plants
To identify the genes responsible for accelerated flowering in pFD::amiR-PECT1 plants in SD, we compared the transcriptome profiles of the SAM-enriched samples of the 10- and 16-day-old pFD::amiR-PECT1 plants (line #5) grown under 23°C SD conditions (Fig. 2A). RNA-seq analyses showed that the mRNA levels of 652 genes (321 upregulated and 331 downregulated) are affected (≥2-fold and P≤0.01, Fisher's exact test) by knockdown of PECT1 in the SAM at day 10, whereas 426 genes (212 upregulated and 214 downregulated) were affected at day 16 (Fig. 2B; Table S2). Among the differentially expressed genes (DEGs), 74 were commonly upregulated and 70 were commonly downregulated (Fig. 2C).
Functional classification of the DEGs in pFD::amiR-PECT1 plants was performed using Gene Ontology (GO) terms. Fig. S4 lists the top 15 enriched GO-terms of commonly upregulated (Fig. 2D) or downregulated (Fig. 2E) DEGs from both stages. Many commonly upregulated genes were associated with reproductive organ development (shown in blue in Fig. 2D), consistent with the early flowering of pFD::amiR-PECT1 plants under 23°C SD conditions (Fig. S3B). By contrast, many of the commonly downregulated genes in pFD::amiR-PECT1 plants were related to biotic stimulus, immune response and cell death (Fig. 2E).
To identify the genes responsible for early flowering, we cross-referenced the DEGs with known flowering time genes (FLOR-ID) (Bouche et al., 2016). In 10-day-old pFD::amiR-PECT1 plants, the levels of 6 and 7 transcripts increased and decreased, respectively, compared with WT (Fig. 2F), and in 16-day-old pFD::amiR-PECT1 plants, 3 and 6 transcripts increased and decreased, respectively (Fig. 2G). Among the upregulated genes, GA20ox2 and CYCLING DOF FACTOR 4 (CDF4) showed an increase at both stages. However, we excluded CDF4 from further analyses because it is a floral repressor (Fornara et al., 2009) and its increased expression is inconsistent with the early flowering phenotype of pFD::amiR-PECT1 plants. By contrast, among the downregulated genes, only SVP consistently showed a decrease (8.49- and 6.42-fold) at both stages (Fig. 2F,G). Therefore, the early flowering of pFD::amiR-PECT1 plants seen under inductive and non-inductive conditions (Fig. 1A) is probably caused by the downregulation of SVP in the SAM, which indirectly elevated GA20ox2 mRNA levels (Andrés et al., 2014).
PECT1 regulates flowering via SVP and its downstream targets in the SAM
Next, we performed quantitative real-time PCR (qPCR) to confirm the alteration of SVP and GA20ox2 expression. We first prepared different tissues (whole seedlings, leaves and SAM-enriched samples) and validated our sample preparation using SHOOT MERISTEMLESS (STM) as the SAM marker and RUBISCO SMALL SUBUNIT 3B (RBCS3B) as the leaf marker (Fig. S5). Under 23°C SD conditions, the SVP mRNA levels were lower in the SAM-enriched samples of pFD::amiR-PECT1 lines compared with WT (3.0- to 3.7-fold) (Fig. 3A). A similar decrease (2.1- to 3.0-fold) was seen in whole seedlings, but not in the leaf. Consistent with the qPCR data, pSVP::GUSplus expression in the SAM was lower in pFD::amiR-PECT1 plants than WT under 23°C SD conditions (Fig. 3B). Concomitant with the downregulation of SVP, we observed an increase in mRNA levels of GA20ox2 (2.9- to 4.9-fold), SOC1 (1.5- to 1.9-fold) and SEP3 (7.4- to 12.7-fold), the known downstream targets of SVP, in the SAM-enriched samples (Fig. 3C; Fig. S6A,B). Furthermore, the mRNA levels of LFY, FUL and SPL3, GA downstream targets, were also elevated (Fig. S6C,D), suggesting that the GA pathway is activated in pFD::amiR-PECT1 plants.
Importantly, the expression of SVP, GA20ox2, SOC1 and SEP3 was altered in the ft-10 tsf-1 background. The SVP mRNA levels in pFD::amiR-PECT1 ft-10 tsf-1 plants were 3.8-fold lower compared with WT, whereas the GA20ox2, SOC1 and SEP3 mRNA levels were 4.3-, 1.7- and 19.6-fold higher, respectively (Fig. 3D; Fig. S7), consistent with the early flowering of pFD::amiR-PECT1 ft-10 tsf-1 plants (Fig. 1D,E). This suggested that the altered expression of SVP, GA20ox2, SOC1 and SEP3 was independent of florigens. Under 23°C LD conditions, SVP mRNA levels of pFD::amiR-PECT1 plants were also lower compared with WT, concomitant with higher GA20ox2 mRNA levels (Fig. S8).
To further elucidate the role of PECT1, we expressed amiR-PECT1 and PECT1:3×HA using an estrogen-inducible XVE system (Zuo et al., 2000) (Fig. S9A). The estradiol treatment in the pER8::amiR-PECT1 plants reduced PECT1 mRNA levels, whereas it increased PECT1 mRNA levels in pER8::PECT1:3×HA plants (Fig. 3E; Fig. S9B,C). Furthermore, we observed a reduction in SVP mRNA levels in pER8::amiR-PECT1 plants (1.8-fold reduction in the SAM-enriched samples), whereas we observed elevated SVP mRNA levels in pER8::PECT1:3×HA plants (2.8-fold increase in the SAM-enriched samples) after estradiol treatment (Fig. 3F; Fig. S9D), suggesting that the late flowering of pFD::PECT1 plants is also due to the elevated levels of SVP (Nakamura et al., 2014). A weak alteration in GA20ox2 and SEP3 mRNA levels was seen in pER8::amiR-PECT1 and pER8::PECT1:3×HA plants (Fig. S9D,E). These data suggest that PECT1 acts upstream to regulate SVP.
PECT1-mediated flowering is SVP- and GA-dependent
To test the genetic interaction between PECT1 and SVP, we generated pFD::amiR-PECT1 svp-32 mutants and measured their flowering time at 23°C under LD and SD conditions. No additive acceleration was observed in pFD::amiR-PECT1 svp-32 plants under 23°C LD conditions (Fig. 4A; Fig. S10A; Table S1) and very weak acceleration of flowering (8.8% decrease in total leaf number) was observed in pFD::amiR-PECT1 svp-32 mutants only under 23°C SD conditions. The mRNA levels of GA20ox2, SOC1 and SEP3 were similar in pFD::amiR-PECT1 svp-32 plants and svp-32 mutants (Fig. S10B), suggesting that PECT1 and SVP act in the same genetic pathway. Furthermore, expression of SVP:3×HA in the SAM driven by the KNOTTED-LIKE FROM ARABIDOPSIS THALIANA (KNAT1) promoter (Fig. S11A) partially rescued the early flowering of pFD::amiR-PECT1 plants under 23°C SD and LD conditions (Fig. 4B; Fig. S11B,C), consistent with the changes in expression levels of SVP, GA20ox2 and SEP3 in the transgenic lines (Fig. 4C; Fig. S11D).
To confirm whether the alteration of GA20ox2 mRNA levels in pFD::amiR-PECT1 plants affected endogenous GA levels, we performed liquid chromatography-mass spectrometry (LC-MS) to quantify the levels of bioactive GAs (GA3 and GA4) and the precursors (GA5) in the SAM-enriched samples. We observed that the levels of all of the tested GAs [both 13-hydroxylated GAs (GA3 and GA5) and non-13-hydroxylated GA (GA4)] were significantly increased (3.9-fold) in pFD::amiR-PECT1 plants (Fig. 4D), consistent with the increased GA levels of svp-41 mutants (Andrés et al., 2014). Consistent with the result, the levels of REPRESSOR OF GA1-3 (RGA), a DELLA protein (Dill et al., 2001), were reduced in the SAM-enriched samples of pFD::amiR-PECT1 plants, but not in the leaf (Fig. 4E).
To further examine the involvement of GA, we treated the WT and pFD::amiR-PECT1 plants with exogenous GA3 and the GA biosynthesis inhibitor paclobutrazol (PAC) and measured the flowering time under 23°C SD conditions. We confirmed that WT and pFD::amiR-PECT1 plants responded similarly to GA3 and PAC treatments by checking the GA20ox2 mRNA levels (Fig. S12) (Fukazawa et al., 2017). Flowering of pFD::amiR-PECT1 plants was only slightly accelerated (0.6 to 12.6% reduction in leaf number) by GA3 treatment, whereas WT plants showed a 40.3% reduction in leaf number (Fig. 4F; Table S1). However, flowering of pFD::amiR-PECT1 plants was severely delayed upon PAC treatment (64.8 to 70.1% increase in leaf number), whereas WT plants showed a 40.8% increase, suggesting that the accelerated flowering in pFD::amiR-PECT1 plants was associated with increased GA levels (Fig. 4D). Nevertheless, flowering of PAC-treated pFD::amiR-PECT1 plants was still earlier than PAC-treated WT plants, suggesting that GA is not the only factor involved in the regulation of flowering time by PECT1, as SOC1 and SEP3 were also upregulated in pFD::amiR-PECT1 plants (Fig. S6A,B).
In summary, our study identified another novel pathway for flowering time regulation by PECT1. In addition to the mechanism shown in Nakamura et al. (2014), our results revealed that the effect of PECT1 on flowering time is also through the modulation of SVP expression and its downstream targets (GA20ox2, SOC1 and SEP3) (Fig. 4G). The methylation of PE during PE:PC biosynthesis affects histone methylation status, which in turn modulates gene expression in yeasts and human cells (Ye et al., 2017), suggesting that membrane phospholipids and histones are closely linked in cell metabolism; however, further research is required to test this possibility in plants.
MATERIALS AND METHODS
Plant materials, growth conditions, chemical treatments and flowering time measurement
The mutants used in this study are in the Columbia (Col) background. pFD::amiR-PECT1, pFD::amiR-PECT1 ft-10 tsf-1, ft-10 tsf-1, and svp-32 have been described previously (Lee et al., 2007; Nakamura et al., 2014; Yamaguchi et al., 2005). The pFD::amiR-PECT1 lines were kindly provided by Prof. George Coupland (The Max Planck Institute for Plant Breeding Research, Germany). The pFD::amiR-PECT1 ft-10 tsf-1 mutants were kindly provided by Prof. Yuki Nakamura (Institute of Plant and Microbial Biology, Academia Sinica, Taiwan). Plants were grown in soil or on half-strength Murashige and Skoog (MS) media at 16°C or 23°C under LD (16 h:8 h light:dark) and at 23°C under SD (8 h:16 h light:dark) conditions. For chemical treatments, β-estradiol (estradiol) (Sigma-Aldrich), GA3 (Duchefa Biochimie) and PAC (Sigma-Aldrich) stock solutions were prepared in DMSO. Mock (DMSO) and 100 µM estradiol were prepared in water and added on the top of MS medium containing 8-day-old plants, which were incubated under 23°C SD conditions for 2 days. For flowering time measurement, mock (DMSO), 50 µM GA3 and 5 µM PAC were prepared in water and applied by irrigation twice a week until flowering. The total leaf number was counted when the primary inflorescence reached ∼5 cm. Box plots were constructed to represent flowering time distribution (Postma and Goedhart, 2019). The data are summarized as a boxplot, with the box indicating the interquartile range (IQR), the whiskers showing the range of values that are within 1.5× IQR and a horizontal line indicating the median.
To generate pSVP::GUSplus transgenic plants, the promoter region of SVP was cloned into pENTR2B:GUSplus entry vector using In-Fusion cloning method (Clontech). The fused pSVP::GUSplus fragment was then cloned into the binary vector pEarleyGate 304 (Earley et al., 2006) using the Gateway cloning system (Invitrogen). To generate pKNAT1::SVP:3×HA, the coding sequence of SVP was cloned into 3×HA-containing entry vector (Wang et al., 2013). The fused SVP:3×HA fragment was then cloned into modified pJHA212G binary vector containing Nopaline Synthase terminator (NosT) and the KNAT1 promoter (Yoo et al., 2005) using the Gateway cloning system. To generate pER8::amiR-PECT1, pER8::PECT1:3×HA and pER-GUS constructs, the respective sequences were cloned into the entry vector pENTR2B for amiR-PECT1 and GUS, and PECT1 was cloned into an entry vector containing 3×HA (Wang et al., 2013). The resulting chimeric sequences were then cloned into the pER8 vector using the Gateway cloning system. Information on the primers used for amplification is given in Table S3.
Gene expression analyses
For mRNA analyses, total RNA was extracted from plant tissues using Plant RNA Purification Reagent (Invitrogen), and RNA was used as a template for cDNA synthesis using the M-MLV RT, RNaseH− (ElpisBio). The transcript levels were measured by qPCR using a QuantStudio5 qPCR instrument (Applied Biosystems). For normalization, two stably expressed genes, PP2AA3 (AT1G13320) and a SAND family protein (AT2G29390) were used (Hong et al., 2010). For validation assays, sampling was done at zeitgeber time (ZT) 8 for 23°C SD conditions and ZT16 for 23°C LD conditions. All qPCR experiments were carried out in three biological replicates, each with three technical replicates. Primers for qPCR (Table S3) were designed according to previously suggested criteria (Die and Roman, 2012). The relative abundance of transcripts was calculated according to the geNorm using the PCR efficiency and the Ct values (Hong et al., 2010).
For protein analyses, total protein was extracted from the seedlings using ProPrep Protein Extraction Buffer (iNtRON Biotechnologies). Proteins were separated by SDS-PAGE and transferred to Immobilon-P PVDF membranes (Millipore Corporation). Anti-HA high affinity monoclonal antibodies (clone 3F10, Sigma-Aldrich, 1:5000), anti-RGA polyclonal antibodies (AS11 1630, Agrisera, 1:1000) and anti-Actin polyclonal antibodies (AS13 2640, Agrisera, 1:10,000) were used as primary antibodies, with anti-rat (A9037, Sigma-Aldrich, 1:10,000) or anti-rabbit HRP-conjugated secondary antibodies (A6154, Sigma-Aldrich, 1:10,000 for anti-RGA and 1:20,000 for anti-Actin).
RNA-seq and bioinformatics analyses
RNA was isolated from SAM-enriched samples, which were collected from the shoot apical region of at least 20-25 plants, of the WT and pFD::amiR-PECT1 plants (line #5). The libraries were constructed using NEBNext Ultra II Directional RNA-seq kit (New England Biolabs) and sequenced using the Illumina Hiseq X Ten platform. The resulting fastq files (GSE147212) were checked for quality using FastQC and then the reads were aligned to TAIR10 reference genome. DEGs were identified using CLC genomics Workbench v.11 as the genes with at least 2-fold change and a P-value smaller than 0.01. For GO analysis, Plaza 3.0 (Proost et al., 2015) was used.
GUS staining assays
GUS staining was performed as previously described (Jefferson et al., 1987). The pSVP::GUSplus and pSVP::GUSplus pFD::amiR-PECT1 seedlings were grown at 23°C under SD conditions. Plant samples, pre-treated with 90% acetone, were soaked in X-Gluc solution and incubated at 37°C. Chlorophyll was removed in ethanol at 60°C, and samples were visualized using an Axioskop 2 plus microscope (Zeiss).
GA levels were quantified using 10-day-old SAM-enriched samples of pFD::amiR-PECT1 plants grown on MS media under 23°C SD conditions (ZT8). Experiments were performed using three biological replicates. GA extraction and quantification was performed as previously described (Li et al., 2017) without a derivatization procedure, using a 6460 Triple quadrupole mass spectrometer (QQQ) comprised of a 1200 series HPLC system at Korea Basic Science Institute (KBSI). A Chemcobond 5-ODS-H (50×2.1 mm, 5 µm) (ChemcoPak) was used with an oven (40°C). GA3 (Sigma-Aldrich), GA4, GA5 and deuterium isotope-labeled GA4 ([17-2H2]-GA4) (Olchemim) were used as standard. The monitoring conditions for all of the tested GA and the internal standard are listed in Table S4.
We thank Prof. George Coupland (Max Planck Institute for Plant Breeding Research, Germany) and Prof. Yuki Nakamura (Institute of Plant and Microbial Biology, Academia Sinica, Taiwan) for kindly providing research materials. We thank Jae Kwan Kim, Hocheol Hwang and Hyewon Jeong for their technical assistance.
Conceptualization: H.S., J.H.A.; Methodology: H.S.; Software: Z.N.; Formal analysis: H.S., S.J., G.Y.; Investigation: H.S., K.G., J.-Y.J.; Writing - original draft: H.S., J.H.A.; Writing - review & editing: H.S., Z.N., J.H.A.; Supervision: H.S., J.H.A.; Project administration: J.H.A.; Funding acquisition: J.H.A.
This work was supported by a National Research Foundation of Korea grant funded by the Korean government (NRF-2017R1A2B3009624 to J.H.A.) and Samsung Science and Technology Foundation (SSTF-BA1602-12 to J.H.A.).
The raw RNA-sequencing data are available at NCBI under accession number GSE147212.
The authors declare no competing or financial interests.