Genome-wide RNA-seq analysis indicates that the DAG1 transcription factor promotes hypocotyl elongation acting on ABA, ethylene and auxin signaling

Hypocotyl elongation is influenced by light and hormones, but the molecular mechanisms underlying this process are not yet fully elucidated. We had previously suggested that the Arabidopsis DOF transcription factor DAG1 may be a negative component of the mechanism of light-mediated inhibition of hypocotyl elongation, as light-grown dag1 knock-out mutant seedlings show significant shorter hypocotyls than the wild type. By using high-throughput RNA-seq, we compared the transcriptome profile of dag1 and wild type hypocotyls and seedlings. We identified more than 250 genes differentially expressed in dag1 hypocotyls, and their analysis suggests that DAG1 is involved in the promotion of hypocotyl elongation through the control of ABA, ethylene and auxin signaling. Consistently, ChIP-qPCR results show that DAG1 directly binds to the promoters of WRKY18 encoding a transcription factor involved in ABA signaling, of the ethylene- induced gene ETHYLENE RESPONSE FACTOR (ERF2), and of the SMALL AUXIN UP RNA 67 (SAUR67), an auxin-responding gene encoding a protein promoting hypocotyl cell expansion.

pathways, respectively mediated by PHYTOCHROME INTERACTING FACTOR3 (PIF3) and ETHYLENE RESPONSE FACTOR 1 (ERF1). Once the seedling perceives the light, ethylene signaling is switched off by phyB-mediated EIN3 degradation 7 . Abscissic acid (ABA) is known as a growth-inhibitory hormone, although some reports described a stimulatory effect of this hormone in maize, wheat, rice and Arabidopsis [8][9][10] . Recently, it has been proposed that the stimulatory or inhibitory response to exogenous ABA depend on both doses of, and tissue sensitivity to this hormone 11 . Hayashi and collaborators 12 have shown that in dark-grown Arabidopsis seedlings ABA reduces the phosphorylation levels and consequently the activity of the plasma membrane H+-ATPase that triggers growth of the hypocotyl.
Although hypocotyl elongation has been extensively described, the molecular mechanisms underlying the hormonal regulation of this process are not yet fully elucidated.
The Arabidopsis Dof transcription factor Dof AFFECTING GERMINATION 1 (DAG1) is a repressor of light-mediated seed germination acting downstream of the master repressor PIF1 13,14 ; accordingly, dag1 knock-out mutant seeds require less GAs and lower red light fluence rates than wild type seeds to germinate 15,16 .
More recently, we demonstrated that DAG1 plays a key role in the control of the developmental switch between seed dormancy and germination 17 , acting on ABA and GA levels to establish (and maintain) seed dormancy and repress germination. DAG1 negatively controls the ABA catabolic gene CYP707A2 and the GA biosynthetic gene GA3ox1 through direct binding to their promoters. Consistently, in dag1 mutant seeds the ABA level is reduced while the GA level is increased compared to the wild type 17 . We had also shown that light-grown dag1 mutant seedlings have hypocotyls significantly shorter than the wild type, suggesting that DAG1 is a negative regulator in the light-mediated inhibition of hypocotyl elongation 14 .
Here, we investigated the role of DAG1 in the light-mediated inhibition of hypocotyl elongation by analyzing the transcriptome profile of 4 days-old dag1 and wild type hypocotyls and whole seedlings by means of high-throughput RNA-sequencing.

Results
Inactivation of DAG1 reduces hypocotyl cell elongation. We have previously shown that dag1 mutant seedlings grown under continuous red light have significantly shorter hypocotyls compared to wild type 14 . To further corroborate this result we measured hypocotyl length of an Arabidopsis line overexpressing the DAG1-HA chimeric protein in a dag1 mutant background (dag1DAG1-HA) 14 . Five days-old dag1DAG1-HA seedlings grown under red light showed hypocotyls of the same length of wild type ones, suggesting that the chimeric protein DAG1-HA is functional and complements the hypocotyl phenotype of the dag1 mutant (Fig. 1a).
Daily measurements of hypocotyl length for five days under red light revealed that at two days dag1 hypocotyls were slightly longer than wild type, possibly due to their faster germination rate 15 . At three days, hypocotyl length of mutant and wild type seedlings were comparable; at four and five days dag1 hypocotyls were significantly shorter than wild type ones (Fig. 1b).
Most of the hypocotyl cells derive from the embryo, and hypocotyl growth is mainly due to longitudinal expansion 18 . To assess whether the dag1 short-hypocotyl phenotype was due to a reduced number of cells or to decreased cell elongation, the number of hypocotyl epidermal cells was counted in four days-old dag1, dag-1DAG1-HA and wild type seedlings grown under red light. This analysis revealed that dag1, dag1DAG1-HA and wild type hypocotyls do not show a significantly different number of epidermal cells (Fig. 1c). However, while dag1DAG1-HA cells are of the same size of wild type ones, dag1 epidermal cells are significantly shorter (Fig. 1d).
Inactivation of DAG1 affects several classes of (hormone-related) genes in hypocotyls. To elucidate the role of DAG1 in the control of hypocotyl growth, we performed RNA-seq analysis of 4 days-old dag1 and wild type hypocotyls and whole seedlings grown under continuous red light. Three biological replicates of each sample were sequenced using the Illumina Hi-seq platform. For each sample, more then 90% of reads successfully mapped to unique regions of the Arabidopsis genome (TAIR10) (Supplementary Table S1). To evaluate reproducibility among biological replicates, we performed a correlation analysis on normalized gene expression values (CPM, counts per million, see Methods). High positive correlation (Spearman's correlation coefficient >0.95) was observed between the replicates of each sample (Supplementary Figs S1-S4). Clustering analysis of expression values led to a clear separation of samples according to tissue of origin (hypocotyls vs whole seedlings) as well as to sample condition (dag1 vs WT) ( Supplementary Fig. S5a). However, the latter separation is less apparent in the case of hypocotyls, probably due to the higher gene expression variability observed in dag1 hypocotyls, as evidenced by Principal Component Analysis ( Supplementary Fig. S5b).
Two comparison groups were constructed and differential expression analysis performed within each group: 1) dag1 vs wild type hypocotyls (hp dag1/WT) and 2) dag1 vs wild type whole seedlings (ws dag1/WT). The overall results of the differential expression analysis are shown in Fig. 2a, where all genes showing a significant expression change (False Discovery Rate, FDR < 0.05) in at least one comparison group are reported, for a total of 388 DE genes (Additional File 1).
Results of the functional enrichment analysis of DE up-regulated genes are shown in Fig. 3, where representative enriched Gene Ontology (GO) terms are reported for the Biological Process and Molecular Function categories according to REVIGO. Among the most significantly enriched processes we found different representative terms related to plant response to organic substances ("response to chitin", "response to carbohydrate", "response to organic substance"), stimuli ("response to stimulus", "response to endogenous stimulus") and immune response (Fig. 3a). A further inspection of these clusters revealed a subgroup of phytohormones-mediated responses, i.e. "response to hormone", "hormone-mediate signalling pathway", "response to auxin", clustering together in the Hypocotyl growth of dag1 and wild type seedlings. Hypocotyl length was measured every day up to five days, using IMAGEJ software. Stratified seeds were induced to germinate under white light for 24 h, then grown for 5 days under continuous monochromatic red light (40 μmolm −2 s −1 ). Three independent biological replicates were performed with SD values (n > 30). Significant differences were determined using two-way ANOVA followed by Tukey post-hoc test; significantly different groups are indicated by the letters. (c) Epidermal cell number of dag1 (black bar), dag1DAG1-HA (grey bars) and wild type (white bar) hypocotyls of four days-old seedlings grown on horizontal plates under continuous red light (40 μmolm −2 s −1 ). For each sample, the number of cells in an epidermal cell file without stomata was counted. The values are the mean of three biological replicates, presented with SD values. Significant differences were analyzed by t-test. (d) Epidermal cells of wild type, dag1, and dag1DAG1-HA hypocotyls (top to bottom) of four days-old seedlings. The picture is referred to the third cell of the hypocotyl from the apex. Seedlings were grown as in (c). broader category "response to carbohydrate". Instead, "response to abscisic acid", "ethylene biosynthetic process" as well as "response to ethylene" form separate clusters and are also significantly enriched (Additional File 2). Notably, ABA, ethylene and auxin are known to be involved in hypocotyl elongation 12,19,20 . Other significantly enriched processes are "respiratory burst" and "immune system process", possibly suggesting the involvement of DAG1 in the response to environmental stress (Fig. 3a). Finally, as shown in Fig. 3b, enrichment analysis in the Molecular Function category revealed the following significantly enriched GO terms: "transcription factor activity" and "transcription regulator activity" (Fig. 3b; Additional File 2). The analysis of the down-regulated genes (32 out of 257 DE genes) revealed two enriched functional categories related to hormone response, i.e. "response to hormone" and "response to endogenous stimulus", however the results are not statistically significant after correction for multiple testing (Additional File 2).
The analysis on whole seedlings (ws dag1/WT) identified 149 DE genes, most of them down-regulated in the dag1 mutant (105 out of 149) (   are "response to salicylic acid stimulus", "defence response" as well as those related to immune response processes, again suggesting a possible involvement of DAG1 in the response to environmental stress (Additional File 2). No significant enrichment was observed in the Molecular Function category, while "extracellular region" resulted to be significantly enriched in the Cellular Component category. Finally, functional enrichment analysis of up-regulated genes revealed significantly enriched biological processes related to plant response to external/ extracellular stimuli, nutrient levels, starvation and chemical homeostasis. In summary, the results of our differential expression analysis between dag1 and wild type are consistent with the role of DAG1 as a transcription factor and suggest its involvement in the control of hormone-related genes in hypocotyls as well as in response to stress in the whole seedling. Moreover, a differential effect is observed on gene expression levels between hypocotyls and whole seedlings, resulting in a more prominent alteration of biological processes in hypocotyls.
Hormone-related genes. The RNA-seq data suggest that DAG1 is involved in the control of ABA, ethylene and auxin-related genes in hypocotyls. We validated the expression of a number of these hormone-related genes by RT-qPCR on dag1 and wild type hypocotyls.
Of the ABA-related genes, we analysed the expression of RAB GTPASE HOMOLOG B18 (RAB18), a stress-responsive gene involved in ABA and drought response 21,22 , ABA REPRESSOR1 (ABR1) encoding an APETALA2 (AP2) domain transcription factor known as a repressor of ABA 23 and the ABA-responsive WRKY40 transcription factor encoding gene 24 . Among the genes enriching the "response to ethylene" biological process, we validated the expression of five ETHYLENE RESPONSE FACTORS (ERF)-encoding genes (ERF2, ERF5, ERF11, ERF105, ERF109, Additional File 2). This large family of transcription factors includes proteins with very diverse functions, involved in ethylene, ABA and gibberellins signaling 25,26 .
Four SAUR genes -auxin-induced genes highly expressed in hypocotyls 2 where they promote hypocotyl elongation -namely SAUR50, SAUR63, SAUR65, and SAUR67 were among the DE genes belonging to the "response to auxin stimulus" enriched process (Additional File 2).
The results of this RT-qPCR analysis confirmed the RNA-seq data: the transcript levels of the ABA-and ethylene-related genes were higher and those of the SAUR genes were lower, respectively, in dag1 than in wild type hypocotyls (Fig. 4a-c), thus supporting the suggestion that the transcription factor DAG1 is involved in the hormonal regulation -specifically ABA, ethylene and auxin -of hypocotyl elongation.
WRKY genes. Among the DE genes in dag1 hypocotyls, we also found the WRKY transcription factor family, encoding key regulators of many plant processes including the responses to abiotic stresses and to ABA, and seed dormancy/germination 27 . Since DAG1 plays a pivotal role in establishing seed dormancy and repressing seed germination by modulating both ABA and GA level, we set to validate these results by analysing the expression of WRKY6, WRKY18, WRKY28, WRKY33, WRKY46 and WRKY70 in dag1 and wild type hypocotyls by RT-qPCR. This analysis confirmed that inactivation of DAG1 results in a significantly increased expression of the WRKY encoding genes (Fig. 5). DAG1 directly binds the promoters of ERF, SAUR and WRKY genes. DAG1 is known to bind, as all the Dof transcription factors, the CTTT sequence on the promoter of target genes 14,17,28 . Therefore, the validated set of DE genes was analysed for the presence of the CTTT binding site (BS) in their promoter by means of Promomer 29 . Significant matches were found for the ERF2, SAUR67 and WRKY18 promoters, suggesting that they could be direct targets of DAG1.
We performed ChIP assays, using dag1DAG1-HA plants 14,17,30 . Protein-DNA complexes were precipitated with anti-HA antibodies, or without antibodies as a negative control. As additional negative control, we performed the same assays on untransformed dag1 seedlings. Three regions of the ERF2, SAUR67 and WRKY18 promoters were amplified by qPCR (Fig. 6a-c).
As for SAUR67 and WRKY18, the relative amount of promoter fragments b and c (9, 5 and 4, 7 Dof BS in SAUR67 and WRKY18, respectively) precipitated by DAG1-HA were significantly higher than the negative controls, whereas the enrichment of precipitated promoter fragment a (1 and 0 Dof BS in SAUR67 and WRKY18, respectively) was very low in DAG1-HA and in the negative controls (Fig. 6b,c).
Of the three ERF2 promoter fragments (with 4, 1 and 2 Dof BS, respectively), amplification of fragments a and b was the most efficient, compared to the negative controls (Fig. 6a). Although only a single Dof site is present in fragment b, it is located within an optimal sequence context -it has been reported that the sequences flanking the Dof BS may influence DNA binding of the Dof proteins 31 .
This analysis indicates that DAG1 directly binds to the ERF2, SAUR67 and WRKY18 promoters in seedlings. Upstream regions (500 bp) of the 257 DE genes found in the hp dag1/WT comparison group were used as input for the MEME discriminative motif discovery tool 32 to identify patterns shared among, and/or repeated within these sequences.
Two enriched motifs were identified in some of the promoter sequences, i.e. CACGTG (E-value = 4.6e −013 ) and CTCTCTCT (E-value = 5.5e −004 ) ( Supplementary Fig. S6). The former corresponds to a known bHLH-binding motif, i.e. the G-box that is bound by the PIF proteins, a family of transcription factors involved in light-mediated developmental processes 33,34 . Interestingly, DAG1 has been shown to be positively regulated by PIF1, the master repressor of seed germination 13,14 . PIF1, as well as PIF3, PIF4, PIF5 and PIF7, is also involved in the promotion of hypocotyl elongation 35,36 . Among the genes carrying the G-box motif in their promoters we identified seven LEA genes, five ERF genes and four SAUR genes, which we have previously validated by RT-qPCR. Interestingly, ERF2 and SAUR67 show both the PIF and the Dof binding sites (Additional File 3). ABA response is altered in dag1 seedlings. The work described in the previous paragraphs suggests that DAG1 promotes hypocotyl cell elongation and represses the expression of ABA-responsive genes in hypocotyls. Thus, we wondered whether DAG1 promotes hypocotyl elongation by repressing ABA-mediated inhibition. To substantiate this hypothesis, we measured hypocotyl length in dag1 and wild type seedling grown under red light in the presence of increasing ABA concentrations (0, 1, 10, 100, 150 µM). As shown in Fig. 7, while at 0 and 1 µM ABA dag1 hypocotyls are significantly shorter than the wild type, the difference becomes not significant at higher concentrations -thus ABA compensates the lack of DAG1 activity -suggesting that indeed the short-hypocotyl phenotype of dag1 is amenable to the inhibition of ABA-responsive genes (Figs 7 and S7).

Analysis of overlapping DE genes between whole seedlings and hypocotyls comparison groups.
Our transcriptome analysis revealed an overlapping set of differentially expressed genes showing opposite expression change between the two comparison groups (Fig. 8a).
More specifically, the gene encoding the microRNA167D (miR167D), is down-regulated in hp dag1/WT and up-regulated in the ws dag1/WT group; the reads corresponding to miR167D map on a region corresponding to the stem-loop structure containing miRNA and complementary miRNA sequences (pre-miRNA) ( Supplementary Fig. S8), suggesting that lack of DAG1 affects miR167D at the transcriptional and/or processing level. Viceversa, 7 DE genes are up-regulated in hp dag1/WT and down-regulated in ws dag1/WT. The two seed-specific genes Late Embryogenesis Abundant Protein (LEA) gene M17 and Cruciferin 1 (CRA1) were among these genes (Fig. 8a). By means of RT-qPCR on dag1 and wild type hypocotyls, we validated the expression of these genes.
Since DAG1 has been previously shown to play a pivotal role during seed development 17,37 , we validated also the expression of other six LEA genes up-regulated in the hp dag1/WT group (Additional File 1; Supplementary Table S2). The expression analysis corroborated the RNA-seq data as in dag1 mutant hypocotyls the expression of  Table S3. Relative expression levels were normalized with UBQ10 (At4g05320) reference gene. The values of relative expression levels are means of three biological replicates, presented with SD values. Significant differences were analyzed by t-test (*P ≤ 0,05). Scientific  these genes was sharply increased compared to wild type, thus suggesting that DAG1 is required to repress these seed-specific genes during hypocotyl development (Fig. 8b).

Discussion
We have previously demonstrated that DAG1 is involved in the repression of light-mediated inhibition of hypocotyl elongation 14 . Here, we show that DAG1 promotes hypocotyl cell expansion; based on the transcriptome analysis of dag1 hypocotyls, we suggest it does so through the control of ABA, ethylene and auxin signaling. Indeed, the functional enrichment analysis of the DE up-regulated genes of the dag1/WT hypocotyl comparison, revealed that "response to abscisic acid", "response to ethylene" and "response to auxin" are among the significantly enriched processes.
ABA, a growth-limiting hormone, has been shown to suppress hypocotyl elongation in etiolated squash hypocotyl segments 38 , as well as in etiolated Arabidopsis seedlings 12 . In addition, it has been recently demonstrated that ABA-responsive genes are repressed in shade avoidance-driven hypocotyl elongation 39 .
Among the up-regulated genes of the dag1/WT hypocotyl comparison group falling in the "response to abscisic acid" biological process and whose expression we validated by RT-qPCR were RAB18 and ABR1, encoding a GTPase and a transcription factor respectively, involved in ABA signaling [21][22][23] . Our results suggest that DAG1 is required to repress the expression of these genes in hypocotyls, consistent with the dag1 short-hypocotyl phenotype. In addition, our RNA-seq analysis identified seven WRKY transcription factors-encoding genes that are up-regulated in dag1 hypocotyls. The WRKY transcription factors are known as key components of ABA signaling: WRKY18, and the two closely related WRKY60 and WRKY40 proteins, have been shown to cooperate in plant response to biotic stress with both overlapping and distinct functions 40,41 . As for the response to abiotic stress, WRKY18 and WRKY60 have been shown to positively control the response to salt and osmotic stress, as well as ABA sensitivity; this function is counteracted by WRKY40 42 . It has been proposed a complex molecular model where these three WRKY factors may alternatively cooperate or play antagonistic roles to control the expression of the ABA INSENSITIVE 4 and 5 (ABI4 and ABI5) genes 43 . More recently, it has been demonstrated that both WRKY18 and WRKY40 are localised in nuclear bodies (NBs), discrete structures where the photoreceptor phyB, mainly in its active form Pfr 44 , co-localise, and physically interact, with PIF3 and PIF4 24 . Here, we provide evidence that DAG1 negatively controls the expression of both WRKY18 and WRKY40, and that it directly binds the WRKY18 promoter. Recently, it has been shown that WRKY6 is repressed during seed germination and early seedling development, and is induced by exogenous ABA and involved in ABA signaling in these developmental processes 45 . We show that WRKY6 is up-regulated in dag1 hypocotyls, suggesting that DAG1 represses the expression of WRKY6 to promote hypocotyl elongation by negatively controlling ABA signaling. Consistently, we show here that the difference between dag1 and wild type hypocotyl length becomes not significant at higher concentration of ABA, indicating that ABA compensates the lack of DAG1 activity in the dag1 mutant.
Ethylene has opposite effects on hypocotyl elongation depending on light conditions 19,46 , promoting elongation in the light while suppressing it in the dark 4,47 . In promoting hypocotyl elongation, ethylene functions through PIF proteins positively controlling PIF3 transcription in the light 6,48 . Among the genes up-regulated in dag1 hypocotyls are also seven members of the large ERF family encoding transcription factors with very diverse functions including ethylene and ABA signaling 24,25 . Among the ERF genes up-regulated in dag1 hypocotyls, ERF11 is the best characterised and represses the ACS2/5 ethylene biosynthetic genes 49 . Interestingly, ABA induces ERF11 thus repressing ethylene production.
We therefore suggest that DAG1 negatively controls ABA signaling, thus repressing ERF11 and increasing ethylene biosynthesis, ultimately promoting hypocotyl growth in the light.
Ethylene affects hypocotyl growth via different hormones including auxin, as inhibition of auxin transport, biosynthesis or perception suppresses ethylene-promoted hypocotyl elongation 50 . Consistently, our genome-wide analysis also indicates that four auxin-responsive SAUR genes -SAUR50, SAUR63, SAUR65 and SAUR67 -are down-regulated in dag1 hypocotyls; three of these, SAUR50, SAUR63 and SAUR67 are induced by ethylene 50 . In Arabidopsis there are 79 SAUR genes, originally identified because of their rapid induction by auxin 51 . More recently, an organ-specific genome-wide analysis characterised 32 SAUR genes whose expression is light-repressed in hypocotyls 52 suggesting a positive role of these genes in hypocotyl elongation. Four of these SAUR genes are the ones down-regulated in dag1.  1, 10, 100 and 150 µM). Hypocotyl length was measured after 5 days. Three independent biological replicates were performed, with SD values (n > 30). Significant differences were determined using two-way ANOVA followed by Tukey post-hoc test; significantly different groups are indicated by the letters. It was also shown that SAUR50 and SAUR65 are positively controlled by PIF3 and PIF4, which directly bind their promoters 52 . Consistently, the SAUR genes downregulated in dag1 are among the DE genes in hypocotyls of the dag1/WT comparison whose promoters are significantly enriched in the PIF-binding G-box motif.
Among these SAUR genes, the promoter of SAUR67 also contains a multiplicity (26) of DOF binding sites: by ChIP analysis we show here that DAG1 directly binds to this promoter. This suggests a cooperative action of DAG1 with PIF proteins in the regulation of SAUR67 and in promoting hypocotyl elongation.
It has been shown that PIF4 cooperates with the AUXIN RESPONSE FACTORS (ARFs) ARF6 and ARF8 in promoting hypocotyl elongation 53 . The ARF6 and ARF8 transcripts are targets of the MIR167 family of microR-NAs 54 . In Arabidopsis, there are four MIR167 precursor genes (MIR167A-D). Expression profiling of these miR-NAs showed that MIR167D, which has been proposed to be a pseudogene 54 , is primarily expressed in hypocotyls and cotyledons 55 . We show here that the expression of MIR167D is affected by inactivation of DAG1 in opposite ways in hypocotyls and whole seedlings. This suggest that MIR167D, rather than a pseudogene, may be a tissue-specific miRNA involved in hypocotyl elongation.
Finally, we have previously shown that DAG1 is also a key component of the molecular network controlling the seed-to-seedling transition in Arabidopsis 17 . Consistently, quite a high number of DE genes in the hp dag1/ WT comparison group are related to seed-specific functions; in particular, seven LEA genes are up-regulated in dag1 hypocotyls. LEA proteins were first described to accumulate during plant seed dehydration at late stages of embryogenesis 56 . Indeed, the LEA genes that are up-regulated in dag1 hypocotyls are mainly expressed in seeds 57 , suggesting that DAG1 is required to repress the expression of these genes during seed-to-seedling transition.
In conclusion, the resources resulting from our genome-wide analysis substantiate the role of DAG1 in promoting hypocotyl elongation, and provide clues pointing to a role of this transcription factor in regulating and coupling ABA, ethylene and auxin signaling in this developmental process.

Methods
Plant material and growth conditions. All Arabidopsis thaliana lines used in this work were grown in a growth chamber at 22 °C with 16/8-h day/night cycles and light intensity of 300 μmol/m −2 s −1 as previously described 15 . dag1 is the allele described in Papi et al. 15 , in Ws-4 ecotype, dag1DAG1-HA is the transgenic line described in Gabriele et al. 14 . Seeds were surface sterilized and plated on MS agar (halfstrength MS, 0.8% agar, pH 5.7) and stratified at 4 °C for 3 days in the dark.

Phenotypic analysis.
For hypocotyl elongation analysis, all seeds were surface sterilized, sown on MS agar and stratified. After germination was induced by white light for 24 h, plates were incubated for 5 days under continuous red light (40 μmolm −2 s −1 ). Hypocotyl length was measured every day up to five days, or at the 5th day. For ABA experiments, seeds were sown on MS agar with one layer of filter paper 595 (Schleicher & Schull, Dassel, Germany), and 48 h after stratification, seedlings were transferred to plates containing different ABA concentrations (0, 1, 10, 100 and 150 µM). Hypocotyl length was measured after 5 days. Seedlings were scanned and the hypocotyl length was measured using IMAGEJ software. For cell number measurement, four days-old wild type, dag1 and dag1DAG1-HA seedlings, grown on horizontal plates under continuous red light (40 μmolm −2 s −1 ), were fixed in a ethanol/acetic acid mixture (6:1). The samples were cleared in 100% ethanol, incubated for 30 minutes at room temperature. Than the 100% ethanol has been removed and replaced with 70% ethanol, and incubated for 30 minutes. After removing ethanol, samples were placed in a chloral hydrate/glycerol/water mixture (8:1:2, g:ml:ml). Seedlings were mounted in chloral hydrate mixture and images of the samples were taken with a Nikon coolpix 990 camera mounted on Zeiss Axioskop 2 plus microscope equipped with DIC optics. For each sample, the number of cells in an epidermal cell file without stomata was counted. All experiments were performed with three biological replicates, each with three technical replicates.

RNA-seq.
Sterilized seeds were sown on MS agar and stratified. After germination was induced by white light for 24 h, plates were incubated for four days under monochromatic continuous red light (40 μmolm −2 s −1 ). Seedlings were collected and frozen in liquid nitrogen in the dark. For hypocotyls about 1000 seedlings grown in this condition have been dissected, then hypocotyls were frozen in liquid nitrogen in the dark. Three biological replicates for both whole seedlings and hypocotyls were processed. Total RNA was extracted and purified as reported below. Any contaminating genomic DNA was removed using on column DNAse digestion. RNA quality was verified on agarose gel and with Agilent Bioanalyzer 2100 (RNA Integrity Number (RIN) >8).
RNA-seq data processing and detection of differentially expressed genes. RNA-seq reads were mapped to the A.thaliana Tair10 genome assembly using STAR2 58 with default parameters. The gene and transcript annotation from the Ensembl Plant database (http://plants.ensembl.org) was provided during the alignment step. After filtering for uniquely mapped reads, gene-level read counts were obtained using the HTSeq-count algorithm 59 and then processed using the edgeR package 60 . For each sample, raw gene counts were first converted into CPM (counts per million) and those having CPM < 1 in at least three samples (the minimum number of samples in a group) were filtered out in order to filter lowly expressed genes. Gene expression levels and fold-changes were estimated after TMM (Trimmed-Mean of M values) normalization 61 . Both common (all genes in all samples) and separate (tag-wise) dispersion parameters were estimated using the Cox-Reid model and integrated into a Negative Binomial generalized linear model (NB-GLM). Statistical significance of differential expression was assessed using a GLM-likelihood ratio test and the 'Benjamini-Hochberg' correction for multiple testing. A FDR adjusted p-value of 0.05 was set as threshold to define differentially expressed genes.
Gene ontology analysis. Functional  A total of 1 μl of the diluted cDNA was used, along with the specific primers, listed in Table S3 (Supplementary  Table S3). Relative expression levels were normalized with UBQ10 (At4g05320) reference gene.
Chromatin Immunoprecipitation (ChIP) assay. ChIP  The immunoprecipitation was performed using HA-probe antibody (Y-11, sc-805 Santa Cruz), or without antibodies as negative control, overnight at 4 °C. After reverse cross-linking, the enriched DNA levels were quantified by qPCR using specific primer sets (Supplementary Table S3). The Fold enrichment of a specific region was calculated respect to the negative control without antibody. The values are the average of two biological replicates presented with SD values. Significant fold enrichments were analyzed by t-test (*P ≤ 0,05). Statistical analysis. Two-way ANOVA followed by Tukey's HSD posthoc test were used for pairwise multiple comparison (Figs 1 and 7). For hypocotyl elongation assays, statistical significance is indicated by the use of different letters. The values of gene expression analysis and ChIP assays are the mean of three biological replicates presented with SD values. Significant differences were analyzed by t-test (*P ≤ 0,05; **P ≤ 0,01).