The splicing factor 1–FLOWERING LOCUS M module spatially regulates temperature-dependent flowering by modulating FLOWERING LOCUS T and LEAFY expression

The AtSF1-FLM module spatially controls temperature-dependent flowering by negatively regulating the expression of FT and LFY in the leaf and shoot apex, respectively. Alternative splicing mediated by various splicing factors is important for the regulation of plant growth and development. Our recent reports have shown that a temperature-dependent interaction between Arabidopsis thaliana splicing factor 1 (AtSF1) and FLOWERING LOCUS M (FLM) pre-mRNA introns controls the differential production of FLM-β transcripts at different temperatures, eventually resulting in temperature-responsive flowering. However, the molecular and genetic interactions between the AtSF1-FLM module and floral activator genes remain unknown. Here, we aimed to identify the interactions among AtSF1, FLM, FLOWERING LOCUS T (FT), and LEAFY (LFY) by performing molecular and genetic analyses. FT and TWIN SISTER OF FT (TSF) expression in atsf1-2 mutants significantly increased in the morning and middle of the night at 16 and 23 °C, respectively, under long-day conditions. In addition, ft mutation suppressed the early flowering of atsf1-2 and atsf1-2 flm-3 mutants and masked the temperature response of atsf1-2 flm-3 mutants, suggesting that FT is a downstream target gene of the AtSF1-FLM module. LFY expression significantly increased in the diurnal samples of atsf1-2 mutants and in the shoot apex regions of atsf1-2 ft-10 mutants at different temperatures. The chromatin immunoprecipitation (ChIP) assay revealed that FLM directly binds to the genomic regions of LFY but not of APETALA1 (AP1). Moreover, lfy mutation suppressed the early flowering of flm-3 mutants, suggesting that LFY is another target of the AtSF1-FLM module. Our results reveal that the AtSF1-FLM module spatially modulates temperature-dependent flowering by regulating FT and LFY expressions.


Introduction
In eukaryotes, splicing is the removal of non-coding intronic regions from between exons in primary transcripts (pre-mRNAs) to form mature messenger RNAs (mRNAs) using a large RNA-protein complex called the spliceosome, consisting of U1, U2, U4/U6, and U5 small nuclear ribonucleoproteins (snRNPs) as the main building blocks (Wahl et al. 2009;Will and Luhrmann 2011). During the splicing process, conserved cis-acting elements, such as the 5′-splicing site (5′-ss), the branch point sequence (BPS), the polypyrimidine tract, and the 3′-splice site (3′-ss) within the intron are required for the binding of various splicing factors. Pre-mRNA splicing begins with the recognition of 5′-ss by base pairing with U1 snRNA via U1 snRNP and the identification Communicated by Youn-Il Park.
Keh Chien Lee and Hee Tae Lee contributed equally to this manuscript.
1 3 of 3′-ss by the binding of splicing factor 1 (SF1) to the BPS followed by the recruitment of U2 snRNP auxiliary factor 65 (U2AF65) and U2AF35 to the polypyrimidine tract and 3′-ss with the help of U2 snRNP. During this process, the domain structures within SF1 and U2AF65 are also important for their binding to the conserved cis-acting elements in 3′-ss. The KH-QUA2 domain at the N-terminus of SF1 interacts with BPS and buries BPS-adenosine to facilitate the formation of the BPS-U2 snRNA helix (Arning et al. 1996;Liu et al. 2001). The U2AF ligand motif at the N-terminal region of SF1 is also required for its interaction with the U2AF homology motif of U2AF65 (Selenko et al. 2003;Wang et al. 2013;Zhang et al. 2013).
Diverse environmental cues, different developmental stages, and abiotic or biotic stresses affect 5′-or 3′-ss selection, thereby producing two or more mRNA isoforms from the same pre-mRNA in a process called alternative splicing (AS) (Chaudhary et al. 2019;Staiger and Brown 2013;Syed et al. 2012). AS is a widespread mechanism that has the potential to greatly increase the complexity of the transcriptome and proteome, with 60-70% of intron-containing genes in higher plants undergoing AS (Chamala et al. 2015;Zhang et al. 2019Zhang et al. , 2017. AS generates non-productive mRNA isoforms containing premature termination codons, which target nonsense-mediated mRNA decay, thereby regulating transcript abundance Kurihara et al. 2009). Moreover, AS leads to the formation of several protein variants with altered amino acid sequences, which control post-transcriptional regulation via protein-protein interactions (Marquez et al. 2015).
AS plays an important role in the regulation of flowering time in response to changes in ambient temperature, suggesting that the AS of pre-mRNAs serves as a 'molecular thermometer' (Capovilla et al. 2015;Deng and Cao 2017). One known example is the temperature-dependent AS of FLOWERING LOCUS M (FLM) pre-mRNA in temperature-responsive flowering. The alternative use of mutually exclusive second or third exons in this pre-mRNA produces two spliced isoforms, FLM-β and FLM-δ at low and high temperatures, respectively (Lee et al. 2013Pose et al. 2013). The resulting proteins FLM-β and FLM-δ interact with SHORT VEGETATIVE PHASE (SVP) to form distinct protein complexes at different temperatures, thereby contributing to the repression and promotion of flowering, respectively. However, further studies have revealed that only the FLM-β protein is functional in the formation of a temperature-dependent repressor complex with SVP (Capovilla et al. 2017;Lutz et al. 2015Lutz et al. , 2017. Furthermore, high ambient temperature causes AS coupled with nonsense-mediated mRNA decay of aberrant FLM transcripts containing premature termination codons (Sureshkumar et al. 2016). These results suggest that a subset of splicing factors affects the balance between functional and non-functional transcript isoforms of FLM to modulate the temperature-dependent flowering time. Several U2 snRNP-related factors, such as AtU2AF65A, AtU2AF65B, and AtSF1, regulate the AS of FLM pre-mRNA (Lee et al. 2020(Lee et al. , 2017Park et al. 2019). The U1 snRNP component RNA BINDING PROTEIN 45d (RBP45d) and the glycine-rich proteins AtGRP7 and AtGRP8 also modulate the temperature-dependent AS of FLM (Chang et al. 2021;Steffen et al. 2019).
We have previously shown that a lesion in Arabidopsis thaliana SF1 (AtSF1) leads to the misregulation of AS of FLM pre-mRNA, and a temperature-dependent interaction between AtSF1 and the introns of FLM pre-mRNA regulates temperature-responsive flowering via the differential production of FLM-β transcripts at varying temperatures (Lee et al. 2020). Although AtSF1 mutation leads to the reduced expression of floral repressors, such as FLM and SVP, explaining the early flowering phenotype of atsf1-2 mutants (Lee et al. 2017(Lee et al. , 2020Park et al. 2019), the molecular and genetic interactions between AtSF1 and floral activator genes involved in the ambient temperature pathway are poorly understood. In this study, we aimed to investigate the expression levels of the floral activator genes FLOWERING LOCUS T (FT), SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1), and TWIN SISTER OF FT (TSF) in atsf1-2 mutants as well as the genetic interactions among AtSF1, FLM, and FT. We hypothesised that the expression of LEAFY (LFY), which is specifically expressed in the shoot apex, will be increased in atsf1-2 mutants and FLM will be specifically bound to LFY genomic regions. Furthermore, we also expect to observe a genetic interaction between FLM and LFY.

Plant materials and growth conditions
All mutants used in this study had a Columbia (Col-0) background, similar to that of wild-type plants. The processes of generating atsf1-2, flm-3, ft-10, and lfy-12 mutants have been previously described (Jang et al. 2014;Lee et al. 2013;Moon et al. 2005;Yoo et al. 2005). The lfy-12 and ft-10 mutants were provided by Dr. Lee (Seoul National University) and Dr. Ahn (Korea University), respectively. Wildtype and mutant plants were grown either in Sunshine Mix 5 (Sungro, Agawam, MA, USA) or in half-strength Murashige and Skoog (MS) medium at 16 and 23 °C under long-day (LD) (16 h light/8 h dark) conditions at a light intensity of 120 μmol m −2 s −1 . Flowering time was measured by scoring the total number of rosette and/or cauline leaves, and the data were presented as box plots (Postma and Goedhart 2019). In the plots, the box indicates the interquartile range (IQR), the whiskers show the range of values within 1.5 × IQR, and the horizontal line indicates the median. The number of plants counted is shown above each genotype in a box plot. To reveal the statistical differences in flowering time, the data were analysed using SPSS version 24 (IBM SPSS Statistics). Under LD conditions, the leaf number ratio (LNR; 16 °C/23 °C) was used as an indicator of temperatureresponsive flowering (Lee et al. 2007(Lee et al. , 2020.

Generation of double mutants and genotyping
Homozygosity of the double mutants (atsf1-2 flm-3, atsf1-2 ft-10, and atsf1-2 lfy-12) was verified by polymerase chain reaction (PCR) genotyping. The genomic DNA isolation was performed as previously described (Weigel and Glazebrook 2002). For genotyping of the atsf1-2 allele, 062177L and LBb1.3 primers were used (Jang et al. 2014). For genotyping of the flm-3 allele, JH4014 and mLBa1 primers were used (Lee et al. 2013). For the ft-10 allele, the JH2295, JH2296, and JH2297 primers were used (Yoo et al. 2005). For the lfy-12 allele, the PCR amplification was done using DW1228 and DW1019 primers, and the mutation was confirmed by sequencing (Moon et al. 2005). The oligonucleotide primers used for genotyping are listed in Table S1.

RNA expression analyses
For RNA expression analysis, total RNA was extracted from wild-type or mutant seedlings using TRIzol (Invitrogen, Carlsbad, CA, USA). For real-time quantitative polymerase chain reaction (RT-qPCR), the samples were harvested at the indicated Zeiteberg time (ZT) points, immediately frozen in liquid nitrogen, and stored at − 80 °C until further use. RNA quality analysis and complementary DNA (cDNA) synthesis were performed as described previously (Lee et al. 2013). RT-qPCR analysis was carried out in 384-well plates using a LightCycler 480 (Roche Applied Science, Penzberg, Germany) or CFX real-time system (Bio-Rad, Hercules, CA, USA). A stably expressed gene (PP2AA3) was used as the reference gene. All RT-qPCR experiments were performed in three biological replicates (independently harvested samples), each with three technical replicates. The relative transcript abundance was determined using the geNorm algorithm and the PCR efficiency and Ct (threshold cycle) values (Lee et al. 2013). The oligonucleotide primers used for RT-qPCR are listed in Table S1.

Chromatin immunoprecipitation (ChIP) analysis
pFLM:gFLM:GFP seedlings (Lee et al. 2013;Pose et al. 2013) grown on MS medium at 16 and 23 °C under LD conditions (ZT16) were cross-linked in 1% formaldehyde on ice using vacuum infiltration. Nuclear extracts were isolated, and an immunoprecipitation assay was conducted as described by (Lee et al. 2013). After shearing chromatin via sonication, approximately 5 µg of rabbit anti-green fluorescent protein (GFP) polyclonal antibody (Santa Cruz Biotechnology, Dallas, TX, USA) was used to immunoprecipitate the genomic DNA fragments. The DNA recovered from immunoprecipitation or 10% input DNA was used for qPCR analysis. ChIP experiments were carried out in three biological replicates (samples independently harvested on different days) with three technical triplicates each (ChIP samples processed on the same day), and the results were presented as a percentage of input (% input) (Livak and Schmittgen 2001). Error bars indicate standard error of the mean (SEM) of three biological replicates. The oligonucleotide primers used for the ChIP analysis are listed in Table S1.

Effect of AtSF1 mutation on FT and TSF expression at different temperatures
In a previous study, we found that atsf1-2 mutants exhibited early flowering phenotypes at different temperatures upon downregulation of the expression of SVP, FLM-β, and TEM-PRANILLO 2 (Lee et al. 2020). As these important floral repressors affect the transcript levels of FT, SOC1, and TSF, all of which are involved in integrating temperature-dependent flowering signals within the ambient temperature pathway (Lee et al. 2013;Pose et al. 2013), we measured the expression of flowering-promoting genes over a 2-day diurnal time course in similar developmental atsf1-2 mutant and wild-type plants grown at 16 and 23 °C under LD conditions. Although the expression of FT and TSF was specifically increased in the middle of the night (ZT20) in 7-and 8-day-old atsf1-2 mutant seedlings grown at 23 °C, FT expression at 16 °C showed a statistically significant increase in the morning (ZT0 and 8, and ZT4 to 12 in 11-and 12-day-old seedlings, respectively), whereas TSF expression increased from night to morning (ZT8 and 20, and ZT0 to 8 in 11-and 12-day-old seedlings, respectively) (Fig. 1a, c). Therefore, phase shifts affected the expression patterns of these two genes. However, SOC1 expression in the atsf1-2 mutants decreased at 16 and 23 °C, when compared to the wild-type plants (Fig. 1b), indicating that SOC1 expression could not explain the early flowering phenotype in atsf1-2 mutants. These results suggest that atsf1 mutation leads to early flowering at different temperatures by partially regulating FT and TSF expression.

Effect of genetic interactions among AtSF1, FLM, and FT on temperature-insensitive flowering of atsf1-2 mutants
Because FT acts as a major floral integrator gene under LD conditions (Yoo et al. 2005), we tested the genetic effect of AtSF1, FLM, and FT on flowering time. We first generated double and triple mutants with different combinations of atsf1-2, flm-3, and ft-10 single mutants and measured their flowering times at 16 and 23 °C under LD conditions. Our analysis showed that wild-type plants flowered with a mean total leaf number (TLN) of 27.8 and 14.1 at 16 and 23 °C, respectively. Moreover, the atsf1-2 and flm-3 mutants flowered significantly earlier than the wild-type plants, whereas the ft-10 mutants flowered later than other single mutants and wild-type plants (Fig. 2a), which was consistent with the previous reports (Lee et al. 2013(Lee et al. , 2020. The atsf1-2 flm-3 mutants flowered with 10.3 and 9.1 leaves at 16 and 23 °C, respectively, which was comparable to flm-3 single mutants at 16 °C, but not at 23 °C (Fig. 2a). The atsf1-2 ft-10 mutants (TLN = 35.3 and 17.0 leaves at 16 and 23 °C, respectively) flowered earlier than ft-10 single mutants (TLN = 51.7 and 32.3 leaves) at different temperatures (Fig. 2a). However, flm-3 ft-10 mutants flowered with a TLN of 44.5 and 32.6 leaves at 16 and 23 °C, respectively, which was comparable to that of ft-10 mutants. Furthermore, the atsf1-2 flm-3 ft-10 Error bars indicate the SEM of three biological replicates. One-way ANOVA followed by Duncan's multiple comparison tests were performed to test the statistical significance (P < 0.05). Letters indicate significant differences determined by Duncan's method mutants flowered earlier than the flm-3 ft-10 mutants (40.1 and 28.2 leaves at 16 and 23 °C, respectively) (Fig. 2a). These results indicate that the flm or ft single mutation and flm ft double mutation did not completely suppress the flowering phenotype of atsf1-2 mutants at different temperatures.
To assess the temperature sensitivity of the various mutants, we calculated their LNRs using TLN values at 16 and 23 °C. An LNR close to 1.0 indicates that different temperatures have minor effects on flowering. The atsf1-2 or flm-3 single mutants and atsf1-2 flm-3 double mutants had LNR values of approximately 1.0, compared to those of wild-type plants (LNR = 2.0) (Fig. 2b). However, the LNRs of the atsf1-2 ft-10, flm-3 ft-10, and atsf1-2 flm-3 ft-10 mutants were 2.1, 1.4, and 1.4, respectively, indicating that both FLM and FT acted downstream of AtSF1 during temperature response. These results suggest that AtSF1 regulates other flowering time genes, in addition to FLM and FT, to modulate flowering time at different temperatures.

Effect of AtSF1 on LFY expression in the shoot apex regions
Since atsf1-2 ft-10 mutants flowered earlier than ft-10 mutants (Fig. 2a) and our previous study have shown that LFY expression was increased at ZT16 in 8-day-old atsf1-2 mutants at 23 °C under LD conditions (Lee et al. 2017), we measured the transcript levels of floral activators LFY and APETALA1 (AP1) over an approximately 2-day diurnal time course in the atsf1-2 and wild-type plants grown to similar developmental stages at 16 and 23 °C under LD conditions. RT-qPCR analysis showed that LFY expression levels were increased throughout the time course in the whole seedlings of atsf1-2 mutants grown at 23 °C, but not in those grown at 16 °C (Fig. 3a). However, the expression of AP1 in atsf1-2 mutants decreased at most ZT points at 23 °C, but increased at 16 °C (Fig. 3b). These results indicate that lesions in AtSF1 affect LFY and AP1 expression significantly in all seedlings at 23 °C only.
It has been reported that AtSF1 is strongly expressed in shoot apex regions (Jang et al. 2014); thus, we also examined the expression levels of LFY and AP1 at ZT16 in the shoot apices of 8-and 12-day-old atsf1-2 mutant and wild-type plants grown at 23 and 16 °C under LD conditions, respectively. LFY expression was significantly increased in atsf1-2 and atsf1-2 ft-10 mutants at the different temperatures compared to that in the wild-type plants (Fig. 3c). In contrast, AP1 expression in atsf1-2 mutants was lower than that in the wild-type plants at 23 and 16 °C (Fig. 3d). Unlike diurnal expression levels of AP1 in the whole seedlings of atsf1-2 mutants, AP1 expression decreased in the shoot apex regions of the atsf1-2 mutants at different temperatures. Furthermore, LFY expression was almost similar in atsf1-2, flm-3, and atsf1-2 flm-3 mutants (Fig. 3e). These results suggest that AtSF1 influences flowering by negatively regulating LFY expression via FLM in the shoot apex.

Direct binding of FLM to the LFY genomic region
Since it has been established that AtSF1 regulates the AS of FLM pre-mRNA (Lee et al. 2020), we examined whether FLM protein is involved in directly regulating LFY expression. FLM is a type of MADS-box transcription factor, which are DNA-binding proteins that recognise the CArG or vCArG motifs in their target genes (Lee et al. 2013;Pose et al. 2013); thus, we performed ChIP experiments on 8-dayold pFLM:gFLM:GFP flm-3 mutant and wild-type plants at 16 and 23 °C under LD conditions (ZT16). In these experiments, two regions (upstream promoter region I and part of the first exon and intron region II) and one region (part of the first exon and intron region I) containing CArG or vCArG motifs of the LFY and AP1 loci, respectively, were explored (Fig. 4). Each region lacking the CArG or vCArG motifs of the LFY and AP1 loci was used as a negative control. The FLM-GFP protein was bound to two regions (I and II) in the LFY locus compared to that in the wild-type plants (Fig. 4a), which was consistent with the FLM ChIP-sequencing data . We observed that the FLM binding to one LFY region I was temperature-dependent, with an approximately 1.8-fold increase in binding at 16 °C compared to that at 23 °C (Fig. 4a). These observations indicate that FLM preferentially binds to the promoter region of LFY at low temperatures. In contrast, FLM binding occurred at both 16 and 23 °C in region II (Fig. 4a), suggesting that FLM can bind to the LFY genomic region in a temperature-independent manner. However, differential binding of FLM to the LFY region (NC), a negative control lacking conserved motifs, was not observed at different temperatures. Moreover, no significant FLM enrichment was observed in region I and negative control of the AP1 genomic region (Fig. 4b).
Based on the increased LFY expression in atsf1-2 mutants (Fig. 3c) and the direct FLM binding to the LFY genomic locus (Fig. 4a), we conclude that LFY is an in vivo target of FLM and AtSF1 may influence LFY expression through FLM.

Genetic interactions between FLM and LFY
To test the genetic effects of FLM and LFY on flowering time, we measured the flowering times of flm-3, lfy-12, and flm-3 lfy-12 mutants at 16 and 23 °C under LD conditions. The flm-3 lfy-12 mutants flowered significantly later (18.8 and 13.4 leaves, respectively) than flm-3 single mutants at 16 and 23 °C (15.0 and 12.0 leaves, respectively) (Fig. 5a). This indicates that the lfy mutation partially suppressed the early flowering phenotype of flm-3 mutants at different temperatures. The LNRs of flm-3 lfy-12 and flm-3 mutants were 1.4 and 1.3, respectively, compared to those of lfy-12 mutants (LNR = 2.3), indicating that the temperature insensitivity of flm-3 lfy-12 mutants was similar to that of flm-3 mutants (Fig. 5b).
In contrast, the LNR of the lfy-12 mutant was 2.3. This result indicates that the temperature sensitivity of lfy-12 mutants was completely suppressed by flm mutation. The results suggest that LFY is another target of the AtSF1-FLM module in addition to FT and that this module regulates flowering by modulating LFY expression. and AP1 (b) in the atsf1-2 mutant at 16 and 23 °C under LD conditions, measured by RT-qPCR (Student's t test, * P < 0.05; ** P < 0.01). Expression levels in wild-type (Col-0) plants at ZT0 on day 7 or 11 at the indicated temperatures were defined as 1.0. Error bars indicate the SEM. c, d Expression levels of LFY (c) and AP1 (d) at ZT16 in the 12-and 8-day-old shoot apices of atsf1-2, ft-10, and atsf1-2 ft-10 mutants at 16 and 23 °C, respectively, under LD conditions. e LFY expression at ZT16 in the 12-and 8-day-old shoot apices of atsf1-2, flm-3, and atsf1-2 flm-3 mutants at 16 and 23 °C, respectively, under LD conditions. Error bars indicate the SEM of three biological replicates. Statistical analysis was performed as described in Fig. 2

Discussion
Our previous reports have revealed that AS variants of FLM transcripts produced by temperature-dependent interactions between AtSF1 and FLM pre-mRNA control the flowering time in response to temperature fluctuations (Lee et al. 2020). However, the molecular and genetic interactions between the AtSF1-FLM module and the important floral integrator genes involved in temperature-dependent flowering remain unknown. In the present study, we showed that the AtSF1-FLM module spatially regulates FT and LFY, which play important roles in regulating temperature-dependent flowering.
We observed that the LNRs of ft-10 soc1-2 tsf-1 mutants was close to 1.0 compared to those of their single or double mutants (LNRs of ft-10, soc1-2, tsf-1, ft-10 soc1-2, and ft-10 tsf-1 = 1.4, 1.9, 1.7, 1.2, and 1.3, respectively), and that FT, TSF, and SOC1 are common downstream genes of SVP and FLM, two main ambient temperature pathway mediators, these factors are known to be the major outputs of this pathway (Lee et al. 2013;Pose et al. 2013). Since FT and Asterisks indicate statistically significant differences (Student's t test, * P < 0.05, ** P < 0.01). ns, not significant. Error bars indicate the SEM of three biological replicates. b ChIP analysis of FLM binding to the AP1 genomic region in 8-day-old seedlings of pFLM:gFLM:GFP and wild-type (Col-0) plants at different temperatures. In a schematic diagram of the AP1 genomic region, a horizontal line (I) indicates the predicted vCArG motifs in the AP1 genomic region (NC, negative control)

Fig. 5
Genetic interaction between FLM and LFY. a Box plot showing flowering time of flm-3, lfy-12 and flm-3 lfy-12 mutants grown at 16 and 23 °C under LD conditions. Note that we selected rosette leaf numbers to measure flowering time, because lfy-12 mutants was epistatic to flm-3 mutants with regard to cauline leaf formation. Num-bers above the x-axis represent n. b Leaf number ratio (LNR) of wildtype (Col-0) and the mutants used in a. Error bars indicate the SEM of three biological replicates. Statistical analysis was performed as described in Fig. 2 SOC1 expression remains unaltered at ZT16 in 8-day-old atsf1-2 mutants grown at 23 °C (Lee et al. 2017) and atsf1 mutation greatly affects the AS of FLM pre-mRNA at different temperatures (Lee et al. 2020), it is important to determine whether the AtSF1-FLM module actually regulates FT, TSF, and SOC1. In this study, we showed that FT and TSF expression in atsf1-2 mutants was significantly increased in the morning and the middle of the night at 16 and 23 °C under LD conditions (Fig. 1). Increased expression levels of FT and TSF at specific ZT points under continuously low and high temperature conditions explain the early flowering phenotype of the atsf1-2 mutants. Moreover, several ambient temperature pathway mutants that show early flowering exhibit higher FT expression in the morning than in the afternoon under natural LD conditions (Song et al. 2018), and minor changes in FT expression are observed in svp mutants only at dawn relative to wild-type plants under day/ night temperature regulation (Kinmonth-Schultz et al. 2016). Furthermore, our genetic interaction data showed that the ft mutation suppressed the early flowering of the atsf1-2 and atsf1-2 flm-3 mutants (Fig. 2). Results of the present study and previous reports indicate that FT is primarily expressed in leaves (Corbesier et al. 2007), making it a potential downstream target of the AtSF1-FLM module in leaves. As phytochrome A (phyA) and EARLY FLOWERING 3 (ELF3) regulate FT expression in the morning and CONSTANS (CO) protein is more stable in the morning under natural LD conditions (Song et al. 2018), further investigation is required to explain the role of AtSF1 in AS of phyA, ELF3, and CO at different temperatures.
Several reports have shown that LFY is a floral activator gene in the shoot apex (Klepikova et al. 2015;Michaels and Amasino 1999). In the present study, we observed that the atsf1-2 ft-10 and atsf1-2 flm-3 ft-10 mutants flowered earlier than the ft-10 mutants at different temperatures (Fig. 2), and LFY expression was significantly upregulated in the shoot apices of atsf1-2, flm-3, atsf1-2 ft-10, and atsf1-2 flm-3 mutants at both 16 and 23 °C (Fig. 3c, e). Therefore, LFY may be another target of the AtSF1-FLM module in shoot apex regions. Consistent with previous reports on the detection of the spatial expression patterns of AtSF1, FLM, and LFY in shoot apex regions (Hempel et al. 1997;Jang et al. 2014;Lee et al. 2013), several lines of evidence found in this study support our hypothesis. First, the diurnal expression patterns of LFY were significantly increased at 23 °C under LD conditions, and specifically increased in the middle of the night at 16 °C (Fig. 3a). Second, LFY expression was significantly upregulated in the shoot apices of the atsf1-2, flm-3, atsf1-2 ft-10, and atsf1-2 flm-3 mutants at different temperatures (Fig. 3c, e). Third, FLM directly bound to the LFY genomic region (Fig. 4a). Finally, flm-3 lfy-12 mutants flowered later than flm-3 mutants at different temperatures, and the temperature insensitivity of the flm-3 lfy-12 mutants was comparable to that of the flm-3 mutants (Fig. 5). However, considering that the increase in LFY expression remained in the atsf1-2 ft-10 mutants similar to the atsf1-2 mutants (Fig. 3c), we could not exclude the possibility that AtSF1 directly affects LFY expression. Furthermore, the preferential binding of FLM to one LFY region I was observed at 16 °C (Fig. 4a). Thus, further genetic interaction studies on AtSF1, FLM, and LFY at different temperatures are required.
Several reports have also revealed that AP1 down-regulates the expression of floral repressors, such as TEM-PRANILLO1 (TEM1) and TEM2, which are known to repress the expression of FT (Castillejo and Pelaz 2008; Kaufmann et al. 2010). In this study, our data showed that diurnal expression patterns of AP1 in the whole seedlings of atsf1-2 mutants was increased and decreased at 16 and 23 °C, respectively (Fig. 3b). However, we observed the reduced expression levels of AP1 expression in the shoot apices of the atsf1-2 mutants (Fig. 3d) and no significant FLM enrichment in AP1 genomic regions at different temperatures (Fig. 4b). Therefore, it is likely that AtSF1 but not the AtSF1-FLM module may directly affect AP1 expression, which has the minor effect of flowering. This notion is further supported by our previous observation that TEM2 expression was dramatically reduced in atsf1-2 mutants (Lee et al. 2020).
In the present study, we identified the molecular and genetic interactions among the AtSF1-FLM module, FT, and LFY in response to ambient temperature. AtSF1 greatly affects the temperature-responsive AS of FLM pre-mRNA (Lee et al. 2020), and the AtSF1-FLM module spatially regulates FT and LFY expression to repress precocious flowering in a temperature-dependent manner.