We previously showed that FHY3 and FAR1 proteins physically interact with both EIN3 and PIF5 transcription factors (Liu et al., 2017, 2020), while both EIN3 and PIF5 were reported to up-regulate the expression of ORE1, a key NAC transcription factor promoting leaf senescence (Sakuraba et al., 2014; Qiu et al., 2015). Thus, we hypothesized that FHY3 and FAR1 may regulate leaf senescence through modulating the functionality of the EIN3-ORE1 and PIF5-ORE1 transcriptional modules. To test our hypothesis, we first compared the leaf phenotype of the fhy3 far1 double mutant and wild type plant (Col-0) grown under long-day (16 h light/8 h darkness) conditions. The result showed that the fhy3 far1 plants indeed showed an obvious early leaf senescence phenotype, with lower chlorophyll contents and higher expression levels of several well-known senescence-induced genes (SEN4, SAG12, SAG13 and SAG29) in the fourth leaves of 30 and 36 day-old plants compared with those in the same-aged wild type plants (Supplementary Figures 1A–D). Consistent with this, the leaves of FHY3 overexpressors (FHY3OE) senesced later than the wild type plants (Supplementary Figure 2). Since dark treatment is known to induce rapid and synchronous leaf senescence and is adopted to simulate natural senescence (Weaver et al., 1998; Weaver and Amasino, 2001), thus we exposed the third and fourth rosette leaves detached from 4-week-old wild type (Col-0) and fhy3 far1 mutant plants to darkness. We found that the detached leaves from fhy3 far1 mutants showed significantly faster senescence than the wild type, with significantly lower chlorophyll content (Figures 1A,B), consistent with their effects on age-dependent senescence. The expression levels of SEN4 and SAG12 were also much higher in the fhy3 far1 mutant than those in wild type plants (Figure 1C). Moreover, with 4-day darkness treatment (4 DDI), 32-day-old fhy3-11 and fhy3 far1 plants exhibited early leaf senescence when compared with the same-aged wild type plants, while the FHY3OE plants showed late leaf senescence (Supplementary Figure 2). Collectively, these results indicate that FHY3 and FAR1 negatively regulate leaf senescence with or without darkness induction.

It is well known that ethylene promotes leaf senescence, and two closely related transcription factors, EIN3 and EIL1, are essential for ethylene signaling (Chao et al., 1997). It has been shown that EIN3 and EIL1 can directly activate the expression of several senescence-associated genes, such as ORE1, NAP, and WRKY75 (Qiu et al., 2015; Guo et al., 2017; Zhang et al., 2017), to promote leaf senescence. We previously showed that FHY3 and FAR1 can directly interact with EIN3 and EIL1 in vivo and in vitro to coordinately regulate the phosphate starvation response in Arabidopsis (Liu et al., 2017). Thus, we examined the available databases of EIN3 and FHY3 target genes (Ouyang et al., 2011; Chang et al., 2013; Wang et al., 2016) and identified several senescence associated genes among the co-regulated genes by FHY3 and EIN3, including WRKY75, ORE1, NAP, SAG20, and SAG21. qRT-PCR analysis verified that the expression patterns of these genes indeed changed in the fhy3 far1 mutant compared with the wild type (Supplementary Figure 3).

To investigate the genetic interaction between FHY3/FAR1 and EIN3/EIL1 in regulating leaf senescence, we constructed the fhy3 ein3 eil1 triple mutant and compared its leaf senescence phenotype with that of ein3 eil1, and fhy3-11. We found that after 4 days in darkness, the detached leaves of the fhy3 ein3 eil1 mutant, but not fhy3-11, maintained green just like the ein3 eil1 parental mutant (Figure 2A). In addition, their distinct leaf yellowing phenotypes of the detached 4-day-old leaves were consistent with the results of a quantitative assay of chlorophyll contents in these mutants (Figure 2B). We also observed a similar senescence pattern for naturally senescence plants with or without darkness treatment (Supplementary Figure 4). Moreover, expression of the senescence associated genes SAG12, ORE1, and WRKY75 at the indicated leaf ages were also in consistent with the senescence phenotypes of the dark-treated leaves (Figures 2C–E). Taking together, these results indicate that FHY3 and FAR1 act upstream of EIN3/EIL1 to regulate leaf senescence.

Previous studies have shown that ORE1, a key regulator of leaf senescence, is a direct target gene of EIN3 (Qiu et al., 2015). To elucidate the genetic relationship between ORE1 and FHY3/FAR1 in controlling leaf senescence, we generated ore1 single mutant using CRISPR/Cas9 technology (Supplementary Figures 5A,B) and constructed ore1 fhy3 far1 triple mutant. The fourth detached leaves from two independent ore1 single mutants, ore1-3 and ore1-4, showed later leaf senescence with darkness treatment than those from wild type plants (Supplementary Figure 5C). The detached leaves of the fhy3 far1 mutant became yellow after dark treatment for 4 days, whereas the detached leaves of ore1 and ore1 fhy3 far1 remained green (Figure 3A and Supplementary Figure 5C). The distinct leaf yellowing phenotypes were consistent with the results of chlorophyll content assay (Figure 3B). Similarly, the 32-day-old plants with or without darkness induction also showed that the ore1 fhy3 far1 triple mutant exhibited a slower senescence phenotype than the fhy3 far1 double mutant (Supplementary Figure 5D). Consistently, the expression of SAG12 and SEN4 was induced much slower and lower in the ore1 fhy3 far1 triple mutant compared with the fhy3 far1 double mutant, similar to the ore1 single mutant (Figures 3C,D). These observations indicate that ORE1 acts downstream of FHY3 and FAR1 in regulating leaf senescence.

Considering that WRKY75 is another direct target of EIN3 (Chang et al., 2013), we also constructed wrky75 fhy3 far1 triple mutant and observed its senescence phenotype. The leaf senescence phenotype and chlorophyll content showed that the mutation of WRKY75 partially rescued the early senescence phenotype of fhy3 far1 (Supplementary Figures 6A,B). Consistent with this, the expression of SAG12 and SEN4 was much higher in the 30 and 36-day wrky75 fhy3 far1 triple mutant compared with those in the same age wrky75 or wild type, and was close to that in the fhy3 far1 double mutant (Supplementary Figures 6C,D), suggesting that WRKY75 may also play a minor role in FHY3/FAR1-regulated leaf senescence.

We previously reported that FHY3 and FAR1 can directly interact with the DNA binding domain of EIN3/EIL1 proteins (Liu et al., 2017), we thus speculated whether FHY3 and FAR1 can regulate the function of EIN3/EIL1 during leaf senescence. Results from a dual-luciferase reporter (DLR) system with a 3.5-kb ORE1 promoter sequence in Nicotiana benthamiana leaves (Figures 4A,B) showed that EIN3 alone promoted the transcription of ORE1, whereas FHY3 alone seemed have no obvious effect on ORE1 transcription (Figures 4A,B). However, when FHY3 was co-expressed with EIN3 in N. benthamiana leaves, the induction of ORE1 expression by EIN3 was significantly repressed (Figures 4A,B), suggesting that FHY3 represses the transcriptional activation activity of EIN3 on ORE1. We next examined whether the physical interaction between FHY3 and EIN3 may affect the DNA binding activity of EIN3. We first used yeast one-hybrid assay to test the effect of FHY3 protein on binding of EIN3 to the EIN3 binding site (EBS, 5′-ATGAACCT-3′, 5x EBS was used here) in the ORE1 promoter. The results showed that there was no obvious binding between the FHY3 protein and ORE1 promoter or the 5xEBS fragment (Figure 4C). However, we found that when FHY3 protein was added (construct AD-FHY3), the ability of EIN3 binding to the ORE1 promoter or the 5xEBS fragment was drastically decreased (Figure 4C). To confirm this, we detected the direct binding activity of EIN3 by electrophoretic mobility shift assay (EMSA). Since the N-terminal fragment of EIN3 (EIN3N, amino acids 141–352) contains the DNA binding domain and has been verified to possess the binding ability (Li et al., 2013), thus we produced the EIN3N proteins and tested its binding activity to the biotin-labeled 60-bp ORE1 promoter fragment (containing the EBS, designed as Biotin-ORE1 EBS). We found that when FHY3 protein was added, the ability of EIN3 binding to the ORE1 promoter was drastically decreased (Figure 4D). Since no binding of FHY3 to the ORE1 promoter was detected (Figure 4C), these results suggest that FHY3 likely regulates ORE1 expression via the FHY3-EIN3 interaction rather than through direct binding to the ORE1 promoter. To verify this, we generated transgenic plants expressing HA-tagged EIN3 (EIN3-HA) in the wild type (EIN3-HA/Col-0) and fhy3 far1 double mutant backgrounds (EIN3-HA/fhy3 far1). Chromatin immunoprecipitation combined with quantitative PCR (ChIP-qPCR) showed more enrichment of the ORE1 promoter fragment containing the EBS (Solano et al., 1998) in the EIN3-HA/fhy3 far1 seedlings, compared to the EIN3-HA/Col-0 seedlings (Figure 4E). These results indicate that FHY3 represses the DNA binding activity of EIN3 to the ORE1 promoter.

Previous studies have shown PIF4 and PIF5 promote leaf senescence by activating the expression of EIN3, ORE1, ABI5, and EEL (Sakuraba et al., 2014). Our previous work showed that FHY3 and FAR1 interact with PIF5 both in vivo and in vitro, and that the bHLH domain of PIF5, which is necessary for DNA binding, is responsible for the interaction with FHY3 (Liu et al., 2020). We thus speculated that FHY3 and FAR1 may also regulate PIF5 function in inducing leaf senescence. To test this possibility, we constructed fhy3 pif5 double mutant and phenotypic assay showed that, compared with wild type, pif5-3 leaves displayed a little delayed senescence and the fhy3 pif5 double mutant exhibited an intermediate phenotype between the fhy3-11 and pif5-3 single mutants after dark treatment. Consistent with this, chlorophyll content in the fhy3 pif5 double mutant was lower than the pif5-3 single mutant but higher than the fhy3-11 single mutant (Figures 5A,B). Moreover, the expression patterns of ORE1, SAG12, and SEN4 in leaves of these mutants were consistent with their senescence phenotypes (Figures 5C–E). These results suggest that FHY3 acts antagonistically with PIF5 in regulating leaf senescence. In support of this, we found that overexpression of FHY3 partially repressed the early senescence phenotype of the PIF5OE plants (Supplementary Figures 7A,B).

We next tested whether FHY3 can repress PIF5’s DNA binding activity to the ORE1 promoter. Our yeast one-hybrid assay and EMSA experiments showed that PIF5 could directly bind to the ORE1 promoter fragment (Figures 5F,G), which is consistent with previous report by Sakuraba et al. (2014), and the addition of FHY3 proteins repressed or interfered with the binding activity of PIF5 (Figure 5F). ChIP-qPCR and DLR assays also verified that PIF5 could directly bind to the ORE1 promoter and induce expression of ORE1 in vivo, while addition of FHY3 repressed PIF5-induced expression of ORE1 (Figures 5H,I), suggesting that the interaction between FHY3 and PIF5 represses PIF5 binding to the ORE1 promoter.

We previously showed that FHY3 protein directly interacts with both EIN3 and PIF5 (Liu et al., 2017, 2020), thus we wondered whether FHY3, EIN3, and PIF5 can form a tri-protein complex. To test this hypothesis, we conducted yeast three-hybrid experiment and luciferase complementation imaging (LCI) assay in tobacco. Yeast three-hybrid result showed that the interaction between EIN3 and PIF5 could hardly be detected in yeast, but addition of FHY3 obviously increased the interaction between EIN3 and PIF5 (Figure 6A), suggesting that FHY3 may bridge the interaction between EIN3 and PIF5. Similarly, the LCI results showed that the interaction between EIN3 and PIF5 was very weak, but when FHY3 was co-expressed, the luciferase activity was strongly induced (Figure 6B).

To further investigate the genetic relationship between PIF5 and EIN3/EIL1, we generated the PIF5OE/ein3 eil1 mutant combination by genetic crosses. Leaf phenotyping, chlorophyll content measurement and downstream senescence-associated genes (ORE1 and SAG12) expression assay all revealed that mutation of EIN3 and EIL1 partially rescued the early senescence phenotype of PIF5OE (Figures 6C,D). We also generated the EIN3OE/pifq mutant. Phenotypic assay showed that its leaves senesced earlier than pifq but later than EIN3OE (Figure 6E). The expression levels of their downstream targets ORE1 and SAG12 in the EIN3OE/pifq plants were also intermediate between their parents EIN3OE and pifq mutants (Figure 6F). These results suggest that PIF5 likely works in parallel with EIN3/EIL1, but downstream of FHY3.

To investigate how the expression and protein levels of FHY3 and FAR1 were regulated during the aging process, we analyzed the expression of FHY3 and FAR1 at the four indicated age in wild type (Col-0) leaves. qRT-PCR analysis showed that the expression of FHY3 and FAR1 was up-regulated in leaves of about 24-day-old and then down-regulated afterward (Figure 7A). To examine developmental regulation of the FHY3 protein level, we generated transgenic plants expressing GFP-FHY3 fusion protein driven by the 35S promoter (GFP-FHY3/Col-0). The functionality of the over-expressed GFP-FHY3 fusion protein was verified by phenotyping under mimicked shade conditions (Supplementary Figure 8). Total protein was extracted from the fourth leaf at the indicated leaf ages and we found that the levels of GFP-FHY3 fusion protein rapidly decreased in leaves older than 24 days (Figure 7B). This result was verified in wild type plants using anti-FHY3 antibodies (Supplementary Figure 9). By contrast, most of the senescence-associated genes (including ORE1) were sharply up-regulated in leaves older than 24 days, as previously shown (Figure 7C and Supplementary Figures 1D,E). These results indicate that FHY3 possibly functions at the early stage to repress leaf senescence.

Next, we examined the changes of FHY3 protein levels during dark treatment. The GFP-FHY3 transgenic seedlings were grown under continuous white light for 7 days and then transferred to darkness for the indicated times. Western blot analysis showed that the FHY3 protein levels decreased quickly upon dark treatment (Figure 7D). By contrast, we observed a rapid up-regulation of ORE1 gene expression in dark-treated seedlings (Figure 7E).

The Arabidopsis thaliana ecotype Columbia (Col-0) is the parent line for all mutants and transgenic plants used in this study. Transgenic lines in different genetic backgrounds and multiple mutants are constructed by genetic crosses. fhy3-11 (SALK_002711) and far1-4 (SALK_031652) were obtained from the ABRC, ein3 eil1 (Alonso et al., 2003), pif5-3, pifq (Zhong et al., 2012) were described previously. FHY3 overexpressors (FHY3OE) and FAR1 overexpressors (FAR1OE) have been described in Ma et al. (2017) and Liu et al. (2019), respectively. The wrky75 mutant has been described in Guo et al. (2017). Double or triple mutants were generated by genetic crosses.

Arabidopsis seeds were surface-sterilized and plated on Murashige and Skoog (MS) medium (4.4 g/L MS salts, 1% [w/v] sucrose, pH 5.8, and 8 g/L agar). After stratification at 4°C for 3 days, the seedlings were transferred to soil and grown at 22°C under long-day conditions (16-h light/8-h dark). The white light source was provided by LED (PAR = 100 μmol m^–2 s^–1).

For JG-EIN3, JG-PIF5, and JG-FHY3 constructs, the individual full-length coding sequences of EIN3, PIF5, and FHY3 were ligated to the vector pB42AD and designed as JG-EIN3, JG-PIF5, and JG-FHY3, respectively. For AD-FHY3 construct, the coding sequences of FHY3 was ligated to the pEG202 vector (Clontech) and designed as AD-FHY3. To create pORE1:LacZ, the ORE1 promoter was amplified from genomic DNA and inserted into pLacZ2μ vector (Lin et al., 2007) digested with EcoRI and XhoI. To construct p5x EBS:LacZ, 5 repeats of the ORE1 promoter fragment containing the EIN3 binding site (5′-aatatactttacaaggttcatgcatgcatacattgttttc-3′) was amplified and inserted into SalI digested pLacZi2μ vector (Lin et al., 2007). The five tandem repeats of EBS (5x EBS) were designed in the primer pairs P03 and P04. Two subfragments, 5x EBS-1 (amplified with primer pair P03) and 5xEBS-2 (amplified with primer pair P04), together with pLacZi2μ vector (digested with EcoRI/XhoI) were incubated in 2x Gibson Assembly Master Mix (New England Biolabs) to generate the construct p5xEBS:LacZ for yeast one hybrid.

For yeast three-hybrid assay, the coding sequence of EIN3 was cloned from cDNA into EcoRI-digested pGADT7 vector to generate the AD-EIN3 construct. PIF5 coding sequence was amplified from cDNA and cloned into EcoRI-digested pBridge vector to generate the pBridge-PIF5 construct. Then FHY3 coding sequence was amplified from cDNA and inserted into BglII-digested pBridge-PIF5 to generate the pBridge-PIF5-FHY3 construct.

To generate GST-FHY3 (aa 186-TAA), the FHY3 fragment (aa 186-TAA) was amplified from cDNA and inserted into EcoRI digested pGEX-5x-1.

Plasmids of the 35S promoter-driven effectors for dual Luc reporter system were described previously (Liu et al., 2017, 2019). To generate pORE1:Luc, a 3.5-kb genomic promoter sequence upstream of the coding region of ORE1 was amplified, and inserted into SalI digested pGreenII-0800 vector (Hellens et al., 2005).

The oligonucleotide primers for the constructs above are summarized in Supplementary Table 1. The constructs were verified by DNA sequencing analysis.

FHY3OE (35S:FLAG-FHY3-HA), PIF5OE (35S:PIF5-HA) were lab stock (Liu et al., 2017; Xie et al., 2017). FHY3OE PIF5OE transgenic plants were generated by crossing FHY3OE and PIF5OE. PIF5OE/ein3 eil1 transgenic plants were generated by crossing PIF5OE and ein3 eil1. GFP-FHY3 (35S:GFP-FHY3) transgenic plants were obtained by cloning FHY3 coding sequences into the pEGAD vector and transforming the construct into Col-0 background. Homozygotes were characterized by hygromycin resistance in the T₃ population. EIN3-HA (35S:EIN3-HA) transgenic plants were obtained by cloning EIN3 coding sequences into the pCAMBIA1307 vector and transforming the construct into Col-0 background. Homozygotes were characterized by hygromycin resistance in the T₃ population. EIN3HA/fhy3 far1 transgenic plants were obtained by crossing EIN3-HA transgenic plants with fhy3 far1, and the homozygotes were characterized by PCR-based genotyping of the F₂ population.

Chlorophyll contents were measured in the third and fourth leaves using a SPAD Chlorophyll Meter (SPAD-502 Plus, Konica Minolta). Each leaf was evenly divided into 5–6 spots, and one measurement was taken per spot. The average value of the 5–6 measurements (SPAD Unit) represents a single data point and one biological replicate. Six individual leaves of each genotype are measured, and three biological replicates were performed.

Total RNA was extracted from the fourth leaf of the indicated leaf ages using Trizol reagent (Invitrogen). Reverse transcription was performed using reverse transcriptase (Tiangen). cDNA was diluted 1:10 and subjected to quantitative PCR using SuperReal PreMix Plus (Tiangen) and a 7500 Real Time PCR System (Applied Biosystems, United States) cycler according to the manufacturer’s manual. The level of ACT2 transcript was adopted as an internal control. The oligonucleotide primers for Real-time PCR are summarized in Supplementary Table 1.

To detect the binding of EIN3, FHY3 or PIF5 proteins to the ORE1 promoter, plasmids of indicated JG-fusion proteins (such as JG-FHY3, JG-EIN3 or JG-PIF5) were cotransformed with the indicated ORE1 promoter reporter plasmids into the yeast strain EGY48. Transformants grown on the SD/-Trp/-Ura medium (Clontech, United States) were transferred to the selection medium containing raffinose, galactose, and 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (Amresco, United States) for blue color development. To test the effect of FHY3 on the binding of PIF5 or EIN3 to ORE1 promoter, AD-FHY3 and JG-EIN3 or JG-PIF5 were cotransformed with the indicated ORE1 promoter reporter plasmids into the yeast strain EGY48. Transformants grown on the SD/-Ura/-Trp/-His medium (Clontech, United States) were transferred to the selection medium for blue color development.

Vectors were cotransfected into the AH109 yeast strain according to the manufacturer’s protocol (Clontech, United States). Yeast were grown on selection plate (SD/-Trp/-Leu) for 3–4 days and then transferred to selection plate (SD/-Trp/-Leu/-Met/-His). Positive interactions were recognized by growth on the SD/-Trp/-Leu/-Met/-His plate.

Biotin-labeled/unlabeled or mutant ORE1 and EIN3 promoter oligonucleotide probes were listed in Supplementary Table 1. MBP-EIN3N and GST-PIF5 bHLH vectors were constructed as described previously (Liu et al., 2017; Xie et al., 2017). GST, GST-PIF5 bHLH, GST-FHY3 (aa 186-TAA), MBP, and MBP-EIN3N fusion proteins were expressed in the Escherichia coli strain BL21. The recombinant proteins were purified using either GST-agarose or amylose resin affinity chromatography. EMSA was performed using a LightShift Chemiluminescent EMSA kit (Pierce, United States) according to the manufacturer’s instructions. Briefly, synthetic DNA oligonucleotide probes labeled with biotin were incubated with the indicated recombinant proteins in the presence or absence of excess amounts of unlabeled competitors for 10 min at room temperature. The DNA-protein complexes were separated on 6% native polyacrylamide gels. To analyze FHY3 protein function, 1, 2, and 4 μg of GST-FHY3 were used.

Chromatin immunoprecipitation was performed as described previously (Saleh et al., 2008). Briefly, 2 g of leaf tissues from 4-week-old Col-0 (used as a negative control, set to a value of 1), EIN3-HA, EIN3-HA/fhy3 far1 were collected and fixed in 1% formaldehyde for 20 min under a vacuum, followed by neutralization using 0.125 M glycine for additional 5 min. The leaves were then washed for three times with water followed by chromatin isolation. Anti-HA antibodies were added to the sonicated chromatin followed by incubation overnight to precipitate the bound DNA fragments. After salmon sperm-sheared DNA/protein A agarose beads, the bound DNA was eluted and amplified with primers corresponding to sequences in the ORE1 promoter. Each experiment was performed three times using different pools of seedlings. The oligonucleotide primers for qPCR are summarized in Supplementary Table 1.

Transient expression in N. benthamiana was performed as described previously (Sparkes et al., 2006). A. tumefaciens strain GV3101 carrying the reporter plasmid (pORE1:LacZ or p5x EBS:LacZ) and effector plasmids were cultured in liquid Luria-Bertani medium overnight. The dense cultures were incubated into fresh medium by 1:100 dilution and incubated for 6–8 h. The bacteria were then pelleted at 4,000 rpm for 15 min, and resuspended in an infiltration buffer (5 g/L glucose, 10 mM MgCl₂, 10 mM MES-KOH, pH 5.7; adding 150 μM acetosyringone right before use) to and OD600 of 0.6. The resuspended agrobacteria containing different constructs were mixed equally and then infiltrated into tobacco leaves using 1 mL syringes without needles. Plants were incubated for 2 or 3 days. Firefly luciferase and Renilla luciferase activities were assayed as described previously (Li et al., 2010).

The firefly LCI assays were performed using N. benthamiana leaves. Plasmids for LCI were described previously (Liu et al., 2017, 2019). Both the nLUC- (N-terminal luciferase) and cLUC- (C-terminal luciferase) fusion constructs with or without pSPYNE-FHY3 (empty vector pSPYNE as control) were co-infiltrated into N. benthamiana leaves via A. tumefaciens-mediated co-infiltration. The infiltrated plants were incubated for 2 or 3 days before examining using the Night SHADE LB 985 Plant Imaging System (Berthold, German).

Sequences of all genes analyzed in this work are available at TAIR under the flowing AGI codes: ORE1 (AT5G39610), FHY3 (AT3G22170), FAR1 (AT4G15090), PIF5 (AT3G59060), NAP (AT1G69490), WRKY75 (AT5G13080), SEN4 (AT4G30270), EIN3 (AT3G20770), SAG12 (AT5G45890), SAG13 (AT2G29350), SAG21 (AT4G02380), SAG20 (AT3G10985), SAG29 (AT5G13170), UBQ10 (AT4G05320), ACT2 (AT3G18780).