Arabidopsis thaliana plants were grown at 22–24°C under long days (16 h light/8 h dark) in a growth room with cool-white fluorescent light (90–100 μmol m^-2 s^-1) for general growth and seed harvest. Mutants and transgenic lines are described in Supplementary Table S1. For phenotypic analyses, surface-sterilized seeds were plated on Murashige and Skoog (MS, Duchefa, M0222) agar plates (half-strength MS, 0.8% phytoagar, and 0.05% MES, pH 5.7), stratified for 3 days at 4°C in darkness, transferred to white light for 3 h to synchronize seed germination, and grown in different experimental conditions. ACC (Sigma, A3903) was dissolved in water, ethephon (Sigma, C-0143) was dissolved in DMSO, and AgNO₃ (Sigma, S-0139) was dissolved in either water or DMSO depending on the experiment. For molecular experiments, seedlings were grown under darkness or under red light (13 μmol m^-2 s^-1) for the indicated period.

For photobleaching assays, seedlings grown in the dark for the indicated period were transferred to continuous white light (100 μmol⋅m^-2⋅s^-1) for 3 days. Then, bleached seedlings were counted. For protochlorophyllide quantification, 10 seedlings grown in the dark for 4 days were gently agitated in 1 mL of ice-cold 80% acetone for 1 h in the dark at 4°C to extract pigments. The protochlorophyllide level in 100 μl of supernatant was determined using a fluorescence spectrophotometer (Tecan, infinite 200 PRO) with an excitation wavelength of 440 nm, a bandwidth of 4 nm, and an emission wavelength of 600–720 nm.

Ethylene levels were measured in 100 seedlings grown in 14 ml vials containing 10 ml growth medium and 4 ml headspace. The vials were refreshed with hydrocarbon-free air before they were sealed gas-tight and further incubated for 24 h in the dark. The headspace air was retrieved and ethylene was quantified by gas chromatography (Hewlett-Packard, 5890 series II).

Total RNA from plant tissues was isolated using a plant total RNA extraction kit (Sigma). First-strand cDNAs were prepared with 2 μg of total RNA and M-MLV reverse transcriptase (Promega) according to the manufacturer’s instructions. Gene expression levels were determined by qPCR using SYBR green on a CFX Connect machine (Bio-Rad). Gene expression was normalized to PP2A as an internal control. The gene-specific primers used for qPCR are listed in Supplementary Table S2.

Seedlings were harvested and flash-frozen in liquid nitrogen under a dim green light. The seedlings were then ground in liquid nitrogen and homogenized in denaturing buffer (100 mM NaH₂PO₄, 10 mM Tris-HCl, 8 M urea, pH 8.0) by vigorous vortexing. The debris was removed by centrifugation at 20,000 ×g for 10 m at 4°C. For immunoblot analysis, the supernatants were separated on an 8% SDS-polyacrylamide gel. Then, the proteins were transferred to a nitrocellulose membrane (Hybond ECL, Amersham) using transfer buffer (5.8 g l^-1 Tris base, 29 g l^-1 glycine, 20% methanol, and 0.01% SDS). A rabbit polyclonal anti-EIN3 antibody for native EIN3 (Kim et al., 2013), a rabbit polyclonal anti-Myc antibody (Santa Cruz, CA, USA) for PIF4-Myc, a mouse monoclonal anti-FLAG antibody (Sigma, USA) for EIN3-FLAG, and a mouse monoclonal anti-tubulin antibody (Sigma, USA) for the loading control were used for protein detection. All antibodies were diluted in PBS buffer containing 0.05% Tween 20. Blots were washed three times with the same buffer and then incubated with the appropriate secondary antibodies. After washing three times, the horseradish peroxidase activity of the secondary antibodies was detected using an ECL detection kit (AbFRONTIER, Korea).

Plants overexpressing GFP-Myc, PIF1-Myc, PIF3-Myc, PIF4-Myc, PIF5-Myc, or EIN3-FLAG were grown for 4 days under the indicated conditions before cross-linking for 20 m with 1% formaldehyde under vacuum. Chromatin complexes were isolated and sonicated as described with slight modifications (Oh et al., 2009). An anti-Myc monoclonal antibody (mouse, Cell Signaling) or an anti-FLAG polyclonal antibody (rabbit, Sigma), and Protein A agarose/salmon sperm DNA (Millipore) were used for immunoprecipitation. After reverse cross-linking and protein digestion, DNA was purified using the QIAquick PCR Purification Kit (Qiagen) before being used for qPCR.

All microarray analysis was performed with R version 2.15.0. The limma package was used for background correction and intra- and inter-array normalization. Then, lmFit was used to fit a linear model to the data so statistical calculations could be made using ebayes. ChIP-chip/Seq data was mapped to the TAIR10 genome using bowtie, analyzed by CisGenome v2.0, and visualized by IGV v2.3. Gene Set Enrichment Analysis was conducted with the GSEA package v2.08 (Broad Institute, MIT) according to the online user guide. ACC-responsive gene sets were generated based on published microarray data (Nemhauser et al., 2006).

PIF1 (AT2G20180), PIF3 (AT1G09530), PIF4 (AT2G43010), PIF5 (AT3G59060), ETR1 (AT1G66340), ETR2 (AT3G23150), ERS1 (AT2G40940), ERS2 (AT1G04310), EIN4 (AT3G04580), CTR1 (AT5G03730), EIN2 (AT5G03280), EIN5 (AT1G54490), EBF2 (AT5G25350), EIN3 (AT3G20770), EIL1 (AT2G27050), EBP (AT3G16770), ACO1 (AT2G19590), ERF1 (AT3G23240), EXP9 (AT5G02260), FHL (AT5G02200), HB52 (AT5G53980), AHP1 (AT3G21510), HLS1 (AT4G37580), LOG5 (AT4G35190), GRF2 (AT4G37740), CEL1 (AT1G70710), BRG3 (AT3G12920), SBP1 (AT4G14030), GUN4 (AT3G59400), CHLH/GUN5 (AT5G13630), HEMA1 (AT1G58290), PORA (AT5G54190), PORB (AT4G27440), PORC (AT1G03630), PP2A (AT1G13320), EF1ALPHA (AT5G60390).

The pifq mutant and the ethylene-related mutants etr1, ein2, and ein3 show open cotyledons in the dark and excessive photobleaching when transferred to the light (Supplementary Figure S1). This suggests PIF signaling and ethylene signaling interact. We therefore analyzed four published microarray datasets that compare dark-grown pifq mutants to wild type or ethylene-treated dark-grown wild type to non-treated wild type. These four microarray datasets overlap on 378 genes with significantly altered levels of expression (moderated t-statistic, P < 0.05). Of these 378 genes, 196 show an inverse correlation in the pifq mutants versus wild type ethylene-treated seedlings (Figure 1A). This suggests PIFs positively regulate many ethylene-responsive genes. We next used Gene Set Enrichment Analysis (GSEA) to determine whether ethylene-responsive genes are statistically enriched among PIF-regulated genes. For the analysis, we divided all ethylene-responsive genes into those showing up-regulation of the ethylene precursor ACC (ACC up) and those showing down-regulation of ACC (ACC down). We found ACC up-regulated genes tend to be significantly down-regulated in the pifq mutant, while ACC down-regulated genes tend to be significantly up-regulated in the pifq mutant (Figure 1B). We also observed suppression of several well-known ethylene-responsive markers in both 2- and 4-day-old dark-grown pifq seedlings (Figure 1C). Together, these results indicate PIFs and ethylene regulate large numbers of genes significantly in the same direction. This is consistent with the reduced ethylene signaling phenotypes of the pifq mutants.

Since dark-grown PIF5-overexpressing seedlings synthesize far more ethylene than wild type seedlings, we asked whether the suppression of ethylene-responsive genes in pifq mutants is due to low levels of ethylene production. Thus, we measured endogenous ethylene levels in dark-grown pifq mutant seedlings at 24 h intervals (Figure 1D). Surprisingly, we found PIFs do not robustly promote ethylene biosynthesis. From 1 to 2 days after germination, we found the pifq mutants produce roughly half the ethylene wild type seedlings produce. This pattern reverses after day 2 with the pifq mutants producing higher levels of ethylene than wild type (Figure 1D). These results are inconsistent with the hypothesis that reduced ethylene biosynthesis is responsible for the suppression of ethylene-responsive genes in pifq mutants (Figure 1C). We further examined the expression of well-known ethylene-inducible genes (i.e., EBP, ETR2, and ERS2) in the presence of either a saturating level of the ethylene perception inhibitor silver nitrate (AgNO₃) or the ethylene-producing compound ethephon. Ethylene induces and AgNO₃ represses these ethylene-inducible markers in both wild type and pifq mutants (Figure 2A). Interestingly, the expression of these marker genes is lower in the pifq mutants than in wild type seedlings at all doses of AgNO₃ and ethephon (Figure 2A). On the other hand, PIF4- and PIF5-overexpressing lines show increased marker gene expression for the same doses of AgNO₃ (Figure 2B). In these experiments, the AgNO₃ dose (20 μM) was high enough to fully suppress ethylene responses even with simultaneous treatment of excessive ethylene (Figures 2C,D). These results, thus, suggest PIFs promote the expression of ethylene-responsive genes regardless of endogenous ethylene levels.

Reduced expression of ethylene signaling genes may explain the reduced responsiveness of ethylene marker genes in pifq mutants in response to exogenous ethylene. We thus examined whether PIFs regulate the expression of ethylene signaling genes (Supplementary Figure S2A). Although we did find via microarray (Supplementary Figure S2B) and qRT-PCR (Supplementary Figure S2C) that some ethylene signaling genes are significantly suppressed in pifq mutants, the suppressed genes include both positive (EIN2) and negative regulators of ethylene signaling (ETR2 and CTR1). This complicates any prediction of their net effect on ethylene responses. Since ethylene signaling pathway ultimately impinges on EIN3 protein stabilization (Supplementary Figure S2A), we examined whether the ethylene-mediated EIN3 protein stability is affected in the pifq mutant. However, the EIN3 proteins were stabilized by both ACC and ethephon in the pifq mutants, just as it is in wild type (Supplementary Figure S3A). Red light treatment has no effect on ethephon-induced stabilization of EIN3 (Supplementary Figure S3B), nor is EIN3 expression significantly altered in pifq mutants or by red light treatment (Supplementary Figure S3C). These results suggest the transcriptional regulation of ethylene signaling genes by PIFs does not significantly affect ethylene signaling upstream of EIN3. Instead, the altered expression of ethylene responsive genes in pifq mutants may relate to the activity of EIN3 itself.

It is possible EIN3 activity depends on the activity of the PIFs. By comparing the known genome-wide targets of four PIFs (PIF1, PIF3, PIF4, and PIF5) with those of EIN3, we found a significant number of shared targets (584 genes, hypergeometric test, p < 10^-90; Figure 3A). These shared targets account for 11% of the PIF targets and 44% of the EIN3 targets. We next examined pifq and ein3/eil1 microarrays (Shin et al., 2009; Zhong et al., 2009) to determine how PIFs and EIN3 regulate the expression of these shared targets (class c from Figure 3A). Although only 331 of the 584 shared targets were included in the microarray analyses (Figure 3B), we found 51 of 331 are significantly regulated by both PIFs and EIN3/EIL1, mostly in the same direction (Figure 3C). We observed enriched binding of both PIF4 and EIN3 in the promoter regions of their shared targets (Supplementary Figure S4B). Furthermore, many of the PIF4 and EIN3 binding peaks precisely overlap (Figures 3D,E) and most of the binding peaks fall within 200 bp of each other (Supplementary Figure S4C, blue). In other words, these two transcription factors bind closely to one another on their shared target promoters. To exclude the possibility that this proximity of the PIF4 and EIN3 binding peaks is attributable to chance, we selected PIF4 and EIN3 binding peaks from random target promoters instead of from the same promoter and then calculated the distances between each peak (Supplementary Figure S4A). Compared to the random peak distances (Supplementary Figure S4C, orange), we found the actual PIF4-EIN3 peaks (Supplementary Figure S4C, blue) are strongly biased toward shorter inter-peak distances. This supports the binding of PIFs and EIN3 to their shared targets in close proximity.

The proximity of the PIF and EIN3 binding peaks suggests PIFs and EIN3 may enhance one another’s binding to their shared target promoters. Since red light dissociates PIFs from their target promoters and enhances PIF degradation (Park et al., 2004, 2012; Shen et al., 2005; Al-Sady et al., 2006; Oh et al., 2006; Lorrain et al., 2008), it may also inhibit EIN3 binding to promoters it co-targets with PIFs. We therefore performed a ChIP assay with transgenic plants expressing FLAG-tagged EIN3 grown either in the dark or under red light (Figure 4A). After confirming EIN3 protein stability is unaffected by red light (Figure 4B), we found red light does not significantly affect EIN3 binding to four PIF co-targeted promoters (i.e., those of HLS1, GRF2, LOG5, and SOB3) or to two non-binding control promoters (i.e., those of EF-1alpha and FHL; Figure 4A). This suggests EIN3 binding to the target promoters it shares with the PIFs is independent of PIF binding. We next asked whether ethylene signaling enhances PIF binding to co-targeted promoters by performing a ChIP assay with transgenic plants expressing MYC-tagged PIF4 grown in the presence of either AgNO₃ or ACC (Figure 4C). Since ethylene stabilizes EIN3, AgNO₃ should inhibit and ACC should enhance the binding of PIF4 to PIF4/EIN3 co-targeted promoters if EIN3 is required for PIF4 binding. After confirming neither AgNO₃ nor ACC treatment significantly alters PIF4 protein stability (Figure 4D), we found PIF4 binds equally to four co-targeted promoters regardless of the presence of AgNO₃ or ACC (Figure 4C). This suggests the binding of PIF4 to its target promoters is independent of ethylene signaling. Together, these results suggest PIFs and EIN3 independently bind their target promoters.

Phytochrome-interacting factors and EIN3 bound to the same promoters may independently or interdependently regulate the expression of their shared targets. We therefore measured the expression of their shared targets in pifq, ein2, and pifq;ein2 quintuple mutants. EIN3 and EIL protein levels are very low in ein2 mutants because they are constitutively degraded by EBF1 and EBF2 (Guo and Ecker, 2003; Potuschak et al., 2003). Some of the shared target genes (e.g., LOG5 and AT4G37240) are equally repressed in the pifq, ein2, and pifq;ein2 mutants compared to wild type, suggesting PIFs and EIN3 interdependently activate their expression (Figure 5A). The expression levels of other shared target genes (e.g., HLS1 and GRF2) are slightly higher in the pifq mutant than in the ein2 and pifq;ein2 mutants (Figure 5A). Since the rest of the PIFs (e.g., PIF7) remain active in pifq mutants, they may be responsible for the residual HLS1 and GRF2 expression observed in pifq mutants. To remove the residual PIF activities, we treated wild type seedlings with red light because red light robustly suppresses all PIFs (including PIF7) via Pfr phytochrome (Leivar et al., 2008a). The red light treatment of wild type seedlings reduced shared target expression to the levels observed in ein2 mutants and unlike in wild type seedlings, ein2 mutants showed low expression of co-targeted genes regardless of red light treatment (Figure 5B). On the other hand, the expression of shared target genes (HLS1 and LOG5) in ein3;eil1 double mutants was reduced by the red light treatment (Supplementary Figure S5) suggesting that the activity of residual EIN3-like protein (e.g., EIL2) is also dependent on PIF activities. Taken together, our results suggest both PIFs and ethylene signaling are required for high expression of these co-targeted genes in the dark. However, not all the shared target genes are interdependently regulated by PIFs and EIN3 as shown by further decreased expression of other subset of co-targeted genes (BRG3 and SBP1; Figure 5B). These results support the hypothesis that PIFs and EIN3 either interdependently or additively activate the expression of their shared targets.

Hookless1 mediates the ethylene-induced formation of the apical hook. Since HLS1 is one of the genes co-targeted by PIFs and EIN3 and since PIFs and EIN3 are required for the prevention of photobleaching in etiolated seedlings, we asked whether HLS1 is also involved in the prevention of photobleaching. When 4-day-old etiolated wild type seedlings are transferred to white light, their cotyledons turn green. However, the cotyledons of hls1 mutant seedlings fail to turn green and are instead photobleached (Figure 6A). Photobleaching occurs when protochlorophyllide over-accumulates, and etiolated hls1 seedlings consistently accumulate more protochlorophyllide than wild type seedlings (Figure 6B). We therefore measured the expression of a series of chlorophyll biosynthesis genes in the hls1 mutants. Consistent with their photobleaching phenotype, hls1 mutant seedlings show high expression of HEMA1 and CHLH and low expression of PORA and PORB. Together, these results suggest PIFs and EIN3 prevent photo-oxidative damages of etiolated seedlings in the dark to light transition by activating HLS1, which represses HEMA1 and CHLH and activates PORA and PORB.