PIF transcription factors link a neighbor threat cue to accelerated reproduction in Arabidopsis

Changes in light quality indicative of competition for this essential resource influence plant growth and developmental transitions; however, little is known about neighbor proximity-induced acceleration of reproduction. Phytochrome B (phyB) senses light cues from plant competitors, ultimately leading to the expression of the floral inducers FLOWERING LOCUS T (FT) and TWIN SISTER of FT (TSF). Here we show that PHYTOCHROME INTERACTING FACTORs 4, 5 and 7 (PIF4, PIF5 and PIF7) mediate neighbor proximity-induced flowering, with PIF7 playing a prominent role. These transcriptional regulators act directly downstream of phyB to promote expression of FT and TSF. Neighbor proximity enhances PIF accumulation towards the end of the day, coinciding with enhanced floral inducer expression. We present evidence supporting direct PIF-regulated TSF expression. The relevance of our findings is illustrated by the prior identification of FT, TSF and PIF4 as loci underlying flowering time regulation in natural conditions.

P lants depend on sunlight to fuel photosynthesis. Therefore, growing with potentially reduced light availability, as encountered in dense plant communities, constitutes a threat for plant growth and development. Plants perceive potential competitors because of the reflected far-red (FR) light from neighbors, resulting in reduced red (R)/FR ratio (low R/FR), which leads to the conversion of active phytochrome (Pfr) photoreceptors to their inactive Pr form 1 .
In shade-intolerant plants, such as Arabidopsis thaliana, neighbor detection triggers organ elongation to outgrow competitors and precocious flowering 1 . Among the five phytochromes (phys) present in Arabidopsis, phytochrome B (phyB) is the major regulator of shade avoidance 1 . In normal light conditions, photoactive phyB interacts with a class of basic helix-loop-helix (bHLH) transcription factors, the PHYTOCHROME-INTERACTING FACTORS (PIFs), leading to their phosphorylation and degradation via 26S proteasome or inactivation through less understood mechanisms 2,3 . In contrast, in low R/FR, PIF phosphorylation and degradation is prevented due to the reduction in photoactive phyB enabling PIF binding to Gboxes and PBE-boxes in the promoters of many shade-regulated genes [3][4][5][6][7][8][9][10] . We have started to understand how these transcriptional events lead to shade-induced growth responses 1 ; however, we know very little about developmental transitions triggered by neighbor proximity.
Accelerated flowering is one of the most dramatic responses in the presence of neighboring plants, particularly for annual plants where it represents a unique and irreversible event. In Arabidopsis, accelerated flowering in low R/FR results from the transcriptional induction of FLOWERING LOCUS T (FT) and its close homolog TWIN SISTER OF FT (TSF) in the phloem companion cells of the leaf vasculature [11][12][13][14] . FT acts as a long-distance florigen signal being transported to the shoot apical meristem (SAM) where it forms a flowering activation complex (FAC) after interacting with the basic leucine zipper (bZIP) transcription factor FD and the 14-3-3 proteins [15][16][17] . At the SAM the FAC induces flowering by activating the expression of flowering time genes, such as the MADS-box SUPPRESSOR OF OVER-EXPRESSION OF CO 1 (SOC1) and APETALA1 (AP1) 16,17 .
Despite their importance for shade-induced flowering, the mechanism linking low R/FR ratio perception by phys to enhanced FT and TSF expression in the vasculature remains poorly understood. In Arabidopsis, low R/FR promotes floral transition in a photoperiod-dependent manner 18 in agreement with the attenuated low R/FR response of the photoperiodic mutant constans (co) 11,18 . PHYTOCHROME AND FLOWER-ING TIME 1 (PFT1) was initially proposed to control flowering in response to simulated shade 13 , but was later shown to respond normally to continuous low R/FR 18,19 . Here we show that a subset of PIF transcription factors function genetically downstream of phys to regulate flowering time through FT and TSF expression in response to low R/FR.

Results
PIFs control low R/FR-induced flowering downstream of phyB. PIFs play major roles in neighbor detection growth responses downstream of phyB 4,20,21 . Enhanced PIF expression induces precocious flowering through FT and TSF in the phloem [22][23][24] . Moreover, plants with impaired HFR1 function, a repressor of PIF activity 25 , display increased FT expression in response to low R/FR 26 . Therefore, we hypothesized that PIFs might control flowering time in response to low R/FR. To test this, we scored the flowering transition of PIF loss-of-function mutants under simulated neighbor proximity conditions ( Supplementary Fig. 1a, b; hereafter referred to as low R/FR, while high R/FR refers to standard condition) and found that PIF7 plays a prominent function to accelerate floral transition under low R/FR (Fig. 1a-c; Supplementary Fig. 2a, b). In addition, mutations in PIF4 and PIF5 further enhanced the pif7 late flowering phenotype, indicating that these genes also contribute to the response (Fig. 1a-c;  Supplementary Figs. 2a, b and 3a-c). Importantly, while pif4 pif5 pif7 flowered slightly later than the wild type in high R/FR, the pif4 pif5 pif7 phenotype was much enhanced in low R/FR (significant interaction between genotype and condition both in terms of days to flowering (p = 0.0007) and leaf number (p = 1.52e − 05)) showing the requirement for those three PIF particularly for accelerated flowering in low R/FR ( Fig. 1c; Supplementary Fig. 2b). Moreover, while pif7, pif3 pif4 pif5, and pif4 pif5 pif7 mutants flowered slightly later than the wild type in high R/ FR, in low R/FR late flowering was specifically observed in pif7 and pif4 pif5 pif7 mutants (Fig. 1a-c; Supplementary Fig. 3a-c). Next, we checked whether PIFs mediate precocious flowering of the constitutive shade-avoidance mutant phyB. Consistent with our data in low R/FR, mutations in PIF4, PIF5, and PIF7 were required to fully suppress early flowering in phyB in high R/FR in inductive (long days (LDs)) and non-inductive (short days (SDs)) photoperiods (Fig. 1d Fig. 5a-c), we scored phyB pif4 pif5 pif7 flowering in low R/FR. In this condition, phyB pif4 pif5 pif7 flowers later than phyB, at the same time as pif4 pif5 pif7 ( Supplementary Fig. 5a-c), further supporting the role of PIF4, PIF5, and PIF7 in the phyB pathway. We conclude that PIF4, PIF5, and PIF7 act genetically downstream of phyB, and possibly other phys, to control low R/FR-induced flowering.
PIFs control FT and TSF expression in response to low R/FR. Because flowering in low R/FR depends on both the growth condition and the genetic background 11,18,27 , we tested the flowering response of ft, tsf, ft tsf, and co. In our conditions ft and tsf single mutants responded strongly to low R/FR 11,29 (Supplementary Fig. 6a-d), whereas ft tsf double mutants presented a reduced low R/FR response, similar to co (Supplementary Fig.  6a-d), suggesting that FT and TSF together are needed to promote flowering in low R/FR (significant interaction between genotype (ft vs. ft tsf) and condition both in terms of days to flowering (p = 8.7e − 05) and leaf number (p < 2e − 16)) (Supplementary Fig. 6d).
To compare the role of FT and TSF in the constitutive shadeavoidance mutant phyB with plants growing in low R/FR, we scored the flowering phenotype of phyB ft, phyB tsf, and phyB ft tsf. While phyB tsf double mutants flowered as phyB, both ft and ft tsf abolished phyB early flowering in standard conditions ( Supplementary Fig. 6e). Together, these findings indicate that the phyB mutant does not phenocopy plants growing in low R/FR in terms of flowering, likely because other phys also act in low R/ FR. This is consistent with acceleration of flowering by low R/FR in the phyB mutant 27 (Supplementary Fig. 5a-c) and the extreme early flowering of ft when combined with higher-order phytochrome mutants 30,31 .
Next, we determined whether PIFs contribute to FT and TSF transcriptional regulation in low R/FR. Transcriptome data 32 showed that FT messenger RNA (mRNA) levels increased in cotyledons within 90 min after transfer to low R/FR, while such a rapid induction was not observed for TSF ( Supplementary Fig. 7a, b). In contrast, this early FT up-regulation was absent in the pif4 pif5 pif7 mutant ( Supplementary Fig. 7c). We therefore monitored FT and TSF expression in the wild type and pif4 pif5 pif7 for several days after transfer from high to low R/FR at ZT16, as FT and TSF expression peak at dusk 18,33 . FT and TSF expression were similar in pif4 pif5 pif7 and wild-type plants in high R/FR. In contrast, FT and TSF up-regulation by low R/FR were strongly impaired in pif4 pif5 pif7 (Fig. 2a, b). Consistent with the importance of PIF-dependent TSF up-regulation, ft pif4 pif5 pif7 quadruple mutants flowered later compared to ft and similar to ft tsf under low R/FR ( Supplementary Fig. 8a-d). Moreover, in low R/FR TSF expression was reduced in ft pif4 pif5 pif7 compared to the ft single mutant ( Supplementary Fig. 8e). Low R/FR-mediated induction of FT and TSF were reduced in pif7 single mutant and further reduced in pif4 pif5 pif7 (Fig. 2c,  d), correlating with the flowering time defects of those mutants ( Fig. 1b, c; Supplementary Fig. 2). Similarly, the constitutively higher expression of FT and TSF in phyB was gradually reduced in higher-order pif mutants ( Supplementary Fig. 9a, b). Collectively, these results show that PIF4, PIF5, and PIF7 contribute to low R/FR-mediated floral transition, as well as FT and TSF expression.
In contrast to FT and TSF, CO mRNA expression was only marginally increased by low R/FR light treatments in both wild type and the pif4 pif5 pif7 mutant ( Supplementary Fig. 10). In order to determine whether PIF4, PIF5, and PIF7 control low R/ FR-induced flowering exclusively through the CO-FT/TSF pathway, we compared flowering of co and co pif4 pif5 pif7 mutants.
High R/FR Low R/FR Interestingly, the quadruple mutant flowered later than co both in high and low R/FR, indicating that the PIFs also control flowering through a CO-independent pathway (Supplementary Fig. 11a-c). Moreover, similar to ft pif4 pif5 pif7 and ft tsf, the quadruple co pif4 pif5 pif7 responded to low R/FR, suggesting that independent flowering pathways mediate low R/FR-induced flowering (Supplementary Figs. 8a-c and 11a-c). Collectively, our results identify PIFs as important mediators of FT-and TSF-induced early flowering in response to low R/FR and suggest that PIFs also control flowering through a CO-FT/TSF-independent pathway.
PIF protein accumulation correlates with FT-TSF expression.
To better understand how PIFs control FT and TSF expression, we investigated their temporal and spatial expression pattern. Consistent with the vascular expression of FT and TSF during floral transition 11,12,34 (Fig. 2e), promoter-GUS (β-glucuronidase) fusions showed broad PIF4, PIF5, and PIF7 expression, including the leaf vasculature, in seedlings (Fig. 2e). This is consistent with tissue-specific expression analysis of PIF4, PIF5, and PIF7 35,36 , indicating that PIF4, PIF5, PIF7, FT, and TSF are expressed in the vasculature. FT mRNA expression in the wild type displayed two strong peaks in response to low R/FR, the first early in the light period and the highest peak around dusk, and a third smaller peak during the night (Fig. 3a) 18 . In contrast, there was no induction of FT expression by low R/FR in pif4 pif5 pif7 (Fig. 3a).
The TSF diel expression pattern and its regulation by low R/FR and the PIFs were very similar to FT (Fig. 3a, b). In contrast to FT and TSF, PIF4, PIF5, and PIF7 expression showed one strong peak in the morning as previously observed for PIF4 and PIF5 37 ( Fig. 3c-e). Low R/FR ratio led to a slower decline from peak levels that were particularly obvious for PIF4 and PIF5, but not significant for the lower amplitude cycling PIF7 gene ( Fig. 3c-e). This expression pattern is consistent with a previous study, which showed that low R/FR slows down the circadian clock 38 . Given that phyB inactivation under low R/FR stabilizes PIF4 and PIF5     4 , we decided to also investigate PIF4, PIF5, and PIF7 protein accumulation. Using genomic HA-tagged lines driven by their own promoters ( Supplementary Fig. 12) 39 , we observed diel protein oscillation of HA-tagged PIF4, PIF5, and PIF7 matching mRNA levels (Fig. 3f-h; Supplementary Fig. 13). Moreover, as reported previously 20 , upon transfer to low R/FR we also observed a change in the ratio of slower and faster migrating PIF7 isoforms likely corresponding to different phosphorylation forms ( Supplementary Fig. 14). In addition, we noted a transient increase in total PIF7 levels in low R/FR (Supplementary Fig. 14). Interestingly, PIF4 and PIF5 proteins accumulated to higher levels in low R/FR specifically toward the end of the day, correlating with FT and TSF expression (Fig. 3a, b, f, g; Supplementary  Fig. 13). There was also a tendency for enhanced end-of-day accumulation of PIF7 in low R/FR, but it was less pronounced compared to PIF4 and PIF5, as also observed for the RNA expression patterns (Fig. 3f-h; Supplementary Fig. 13). However, PIF7 nuclear import is induced by low R/FR 3 , indicating that a different mode of low R/FR regulation may be involved for PIF7. Multiple levels of PIF regulation by low R/FR is further suggested by the PIF-dependent expression of FT and TSF in low R/FR early in the day, a time at which no significant difference in PIF4, PIF5, or PIF7 accumulation was observed (Fig. 3a, b, f-h).
PIF proteins directly control TSF expression. PIF4 and PIF5 preferentially bind to G-boxes (CACGTG) and PBE-boxes (CATGTG) 8,9 . Because PIF7 plays a central role in low R/FRinduced flowering and little is known about its DNA-binding preference, we tested its DNA-binding specificity using proteinbinding microarrays. In agreement with recent DNA affinity purification sequencing (DAP-seq) data 40 , and similar to other PIFs 8,9 , we found that PIF7 binds with high affinity to G-boxes (Fig. 4a). Moreover, as observed for other PIFs, among E-boxes it showed the highest affinity for the PBE-box (Fig. 4a).
Interestingly, the analysis of previously published chromatin immunoprecipitation-sequencing (ChIP-seq) data 10 revealed a high-confidence PIF4 binding peak containing 1 G-box and 1 PBE-box overlapping with a highly conserved region located 1.5 kb downstream of FT stop codon ( Supplementary Fig. 15) 41 . Because TSF was shown to integrate environmental signals to influence flowering time in natural conditions 42 and TSF is clearly important for low R/FR-induced flowering in our conditions ( Supplementary Fig. 6a-d), we focused our analysis on PIF regulation of TSF expression. We identified no G-boxes, but identified two PBE-boxes in the TSF promoter located 990 and 437 bases upstream of the ATG (Fig. 4b). The analysis of ChIP-seq data 10 revealed a high-confidence PIF4 peak overlapping with the first PBE-box (−437) (Fig. 4b). To test whether these PBE-boxes are biologically relevant for PIF-mediated TSF expression, we fused its promoter (wild-type and PBE mutants) with luciferase and performed transient transactivation assays in Nicotiana benthamiana. Consistent with PIFs directly regulating TSF expression, PIF4 and PIF7 led to TSF expression (Fig. 4c, d). Importantly, TSF expression was almost completely abolished by a single-nucleotide mutation in the PBE-box present in the PIF4 ChIP peak (−437) 10 or by mutations in both PBE-boxes and one CAAGTG sequence (Fig. 4c, d). Taken together, our data suggest that PIF4 and PIF7 directly bind to PBE-boxes at TSF promoter to induce its expression.

Discussion
Neighbor detection leads to precocious flowering that is controlled via the phyB photoreceptor and FT and TSF genes in Arabidopsis 11,13,14,27,29 . In contrast to well-characterized flowering time pathways, such as photoperiod, vernalization, and gibberellic acid, how sensing neighbor proximity cues are linked to expression of "florigen" genes remains poorly understood. Our experiments identify PIF4, PIF5, and PIF7 as transcription factors acting downstream of phyB to induce flowering in response to neighbor proximity through the floral inducers FT and TSF (Figs. 1, 2; Supplementary Figs. 2-5, 8, and 9). On the other hand, both our data and a recent study analyzing flowering and FT expression profiles in more natural conditions reveal a limited role for the PIFs in high R/FR 43 (Figs. 1, 2; Supplementary  Figs. 2-4). Therefore, our data reveal an important role for PIFs in regulating flowering particularly in response to low R/FR and identify PIF7 as a novel positive regulator of flowering transition.
The perception and response to low R/FR and elevated temperature share sensing and signaling components, including phyB and PIF4 [44][45][46] . However, while the thermal induction of flowering almost completely depends on FT and is mostly independent of CO 47 , low R/FR induction of flowering requires CO (and LDs) and depends both on FT and TSF 11,18,29 (Supplementary Fig. 6). Under non-inductive warm SDs, a key flowering-inducing function was initially attributed to PIF4 in directly mediating FT expression 22 . However, both the underlying mechanism and the importance of PIF4 in thermal-induced flowering via FT regulation have been questioned 23,24,48,49 . Here, we show that PIF7 plays the most prominent role in low R/FR-induced flowering with additional roles played by PIF4 and PIF5 ( ). Finally, we present evidence for direct regulation of TSF expression by PIF7 and PIF4, providing a mechanistic basis for low R/FRinduced flowering (Fig. 4).
Our genetic data also indicate that additional mechanisms contribute to the regulation of flowering in low R/FR independently of FT-TSF and PIF proteins. Indeed, similar to co and ft tsf ( Supplementary Fig. 6) 11,18,29 , co pif4 pif5 pif7 and ft pif4 pif5 pif7 are still responsive to low R/FR ( Supplementary Figs. 8 and 11). CO and FT are major flowering time genes expressed in companion cells of leaves and subsequently FT (and possibly TSF) is transported to the SAM to induce flowering 17,34 . Although it could be speculated that low R/FR directly regulate phys and flowering at the SAM, this seems unlikely because it is normally shaded by emerging leaves. Alternatively, other mobile flowering signals, such as gibberellic acid 50,51 , could be produced in leaves downstream of phys 52,53 and transported to the SAM to induce flowering 53-57 . Indeed, GA metabolism genes are induced by FR light 52 and the GA-responsive flowering time gene LEAFY (LFY) is up-regulated in the phyB mutant background 53 . Alternatively, low R/FR could directly regulate the miR156-SPL aging pathway 58 independently of GA. PIF proteins directly regulate the expression of MIR156 genes and, consequently, several SPL transcription factors in response to end-of-day FR treatments 59 . However, it should be noted that while miR156-SPL age pathway function mostly at the SAM, the tissue-specific expression of PIFs in this tissue have a modest effect on flowering in Arabidopsis 23 . Although we currently do not understand all the pathways induced by low R/FR to accelerate flowering, we identify the phyB-PIF-FT/TSF regulon as one important mechanism (Fig. 4e).
The phyB mutant is often used as a genetic mimic of low R/FRgrown plants. However, when it comes to the regulation of flowering, we and others observe interesting similarities but also differences between both situations. Early flowering in low R/FR and in phyB requires the activity of several PIFs ( Fig. 1; Supplementary Figs. 4 and 5). In contrast, while FT completely accounts for early flowering in phyB (Supplementary Fig. 6e) both in LD or SD photoperiods, low R/FR triggered flowering requires both FT and TSF and long photoperiods 11,18,29,31 (Supplementary Fig. 6c). The requirement for inductive photoperiods to trigger low R/FRinduced flowering is also consistent with the reduced low R/FR response of co and gi mutants 18 . An interesting possibility is that the recently reported interaction of CO with PIF4 24 , or with other PIFs, could contribute to low R/FR-induced flowering in inductive photoperiods. Indeed, low R/FR leads to increased levels of CO 12 , PIF4, and PIF5 4,11 ( Fig. 3; Supplementary Figs. 13 and 14). Moreover, the timing of low R/FR-induced PIF levels is noteworthy as the effect is biggest at the end of the day when CO protein accumulates and coincides with enhanced FT and TSF expression (Fig. 3). Increased levels of PIF4 and PIF5 proteins at the end of the day could be due to both transcriptional and posttranscriptional effects. Indeed, it was previously shown that low R/FR slows down the circadian clock, which may explain the slower decline of PIF4 and PIF5 expression towards the end of the day (Fig. 3c, d) 38 . Therefore, both the higher PIF transcript levels late in the day and enhanced stability of PIF4 and PIF5 protein in low R/FR 4 may contribute to higher PIF levels coinciding with the late FT and TSF expression peaks in low R/FR (Fig. 3). Moreover, we note that low R/FR also leads to higher FT and TSF expression shortly after dawn, a phenomenon that was recently also observed in natural growth conditions (Fig. 3a, b) 43 . Given that at this time of the day we did not observe significant changes in PIF abundance in low vs. high R/FR, we propose that other mechanisms must be operating in shade to control PIF activity. These may include a control of subcellular localization that was reported for PIF7 3 but also direct inhibition of PIF activity by light-activated phyB, which is more abundant in high R/FR 46,[60][61][62] . In summary, the coincidence of high PIF and CO activity towards the end of the day could explain accelerated flowering triggered by neighbor proximity in inductive photoperiods.
In conclusion, multiple mechanisms may contribute to the regulation of PIF protein abundance and activity in low R/FR, but more research is required to fully understand this regulation. However, our data are consistent with the coordinated regulation of FT and TSF expression by PIFs and CO in response to low R/FR 24 (Figs. 2, 3). These transcriptional events likely operate in the leaf vasculature where CO is known to regulate FT and TSF expression 12,34 , and the PIFs are also expressed (Fig. 2e) 11 . The mechanism we uncovered here is likely to be significant in natural environments as "florigen" genes FT and TSF, as well as PIF4, were identified as genes underlying regulation of flowering time and shade-avoidance response in nature 42,63-65 . Our study and further deciphering the mechanism(s) regulating neighbor proximity-induced reproduction may also be relevant to increase yields on increasingly restricted agricultural land.    Table 2). Data corresponding to PIF4 and PIF5 were previously described 8 and shown here for comparison. Box plot represents the distribution of enrichment scores (E-scores) for G-and PBEboxes, as well as for other E-boxes indicated. E-score is a rank-based, non-parametric measure of binding affinity that ranges between −0.5 and +0.5 and is calculated from the median intensity of all the oligonucleotide probes that contain a given 8-mer motif 78 . Boxes represent quartiles 25-75%, the black line within represents the median of the distribution (quartile 50%) and dots denote outliers of the distribution. b Representation of PIF4 chromatin immunoprecipitation-sequencing (ChIP-seq) reads mapped to the TSF locus 10 . Gray box represents high-confidence PIF4 binding peak at TSF promoter and nucleotide sequence represents the wild-type (WT) and All experiments were performed using pif7-1 allele, except those performed using pif7-2/pPIF7::PIF7-3HA-tPIF7. Mutant lines were confirmed by genotyping using oligonucleotides listed in Supplementary Table 1.
RNA isolation and quantitative RT-PCR. Total RNA was extracted using QIA-GEN Plant RNeasy kit and cDNA synthesis was carried out using 1 μg RNA after an on-column DNAseI digestion (RNAse-Free DNase Set, Qiagen). Reverse transcription (RT) was performed using Superscript II Reverse Transcriptase (Invitrogen, Life Technologies) with random oligonucleotides. Quantitative real-time PCR (qPCR) was performed in three biological replicates with three technical replicates for each sample using the QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems). Expression data was normalized against UBC and YSL8 reference genes and relative expression analyses estimated Biogazelle qBase software. Gene-specific oligonucleotides used for qPCR reactions and respective efficiencies are listed in Supplementary Table 1. GUS staining. For GUS staining reactions, samples were fixed with ice-cold 90% acetone for 30 min followed by washing twice with 50 mM NaPO 4 buffer (pH 7.2). Samples were incubated with staining solution (50 mM NaPO 4 , 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, 0.1% Triton X-100 and 2 mM Xgluc) overnight at 37°C in dark. De-staining was performed with ethanol series at room temperature until samples cleared.
Dual-luciferase transient expression in Nicotiana benthamiana. Agrobacterium tumefaciens strain GV3301 was transformed separately with effector and reporter constructs and plated on solid YEP medium supplemented with appropriate antibiotics. After inoculating 30 mL of YEP medium with 1 mL of overnight precultures, cells were incubated at 30°C with vigorous agitation for 16 h. Nicotiana benthamiana leaves were infiltrated with a mixture of effector (final optical density at wavelength (OD 600 ) = 0.3) and/or reporter (final OD 600 = 0.1) cells along with P19 silencing suppressor (final OD 600 = 0.05) using a needleless syringe and infiltrated plants incubated in plant growth room (LD, high R/FR ratio, at 22-23°C ). Three days after incubation, two leaf discs were harvested in liquid nitrogen at zeitgeber (ZT) 8-10 for each biological replicate. After tissue lysis, dual-luciferase assay was performed using Dual-Luciferase Reporter Assay System (Promega). In short, after adding 300 μL of 1× Passive Lysis Buffer, samples were briefly vortexed for 10-15 s and spinned for 15 min at 14,000 × g at 4°C. Ten microliters of supernatant was mixed with 40 μL luciferase substrate and chemiluminescence was measured using GloMax 96 Microplate Luminometer (Promega). The second chemiluminescence measurement was performed after adding 25 μL of 1× Stop&Glo reagent. Relative luciferase activity corresponds to the average firefly/ renilla ratio from four independent infiltrations.
Protein-binding microarray. Protein-binding microarray was performed using Escherichia coli cultures expressing recombinant MBP_PIF7-bHLH proteins (Supplementary Table 2), synthesis of double-stranded microarray, and immunological detection of DNA-protein 77 . In summary, microarrays were scanned in DNA Microarray Scanner (Agilent Technologies) at 2 μm resolution and quantified with Feature Extraction 9.0 software (Agilent Technologies). Normalization of probe intensities and calculation of enrichment scores (E-scores) of all the possible 8-mers were carried out with the PBM Analysis Suite 78 . Perl scripts were modified to adapt them to different custom microarray dimensions and Feature Extraction input files 79 . Data of MBP_PIF4-bHLH and MBP-PIF5-bHLH have been previously published and are shown for comparison 8 .