The discovery that FUS3 misexpression outside embryogenesis can influence vegetative leaf identity suggests that potential FUS3-dependent downstream effectors can be identified through whole-genome microarray analysis 11 . We constructed an inducible misexpression system by transforming the fus3-3 (fus3) mutant with a FUS3-glucocorticoid receptor (GR) translational fusion under the control of the epidermal specific AtML1 promoter (fus3 AtML1:FUS3-GR) 11 . Transgenic seeds were germinated and grown in minimal medium (Murashige and Skoog, MS) for 5 days, then plantlets were transferred to various concentrations of dexamethasone (DEX) to determine the minimal amount of activation needed to influence leaf identity. All concentrations of DEX tested had an effect on vegetative leaf shape; increasing concentrations of DEX resulted in a leaf with an increased paddle-like shape and a progressively shorter petiole (Figure 1A). At higher than 0.5 μM DEX concentrations, the appearance of trichomes on the upper or adaxial surface of the leaves was completely inhibited (data not shown). On the basis of these conditions, 1.0 μM DEX was chosen as the FUS3-activating condition.

Putative FUS3 targets involved in vegetative phase transitions were identified by examining the transcriptome of seedlings that transiently activate FUS3 using the AtML1:FUS3-GR DEX-inducible system 11 . To do this, transgenic seeds from AtML1:FUS3-GR were germinated and grown for 5 days in MS. After this time, half of the seedlings were transferred to MS media supplemented with 1.0 μM DEX (+DEX), and the other half were transferred to MS media supplemented with dimethyl sulfoxide (DMSO) (-DEX or control) for 2 or 4 days (Figure 1B). Putative FUS3 targets were chosen on the basis of the following criteria. Genes that increased (red bars) or decreased (blue bars) at least twofold in expression after two days of FUS3 activation (Figure 1B; +DEX) compared to the control (Figure 1B; -DEX) were considered to represent candidate FUS3 targets involved in phase transition. Utilizing this regime and the average of two replicate experiments, 19 genes increased (Additional file 1, Table S1) and 34 genes decreased (Additional file 2, Table S2) in expression (at least twofold) as the direct result of 2-day FUS3 activation (2 days +DEX) compared to the control (2 days -DEX). To independently verify the expression changes of the 2-day DEX induction experiment, a temporal replicate experiment was performed in which FUS3 activation was extended by transferring 5-day-old seedlings onto DEX for 4 days (4 days +DEX). All but 1 of the 53 genes selected by the 2-day experiment exhibited similar expression ratios in the 4-day experiment, demonstrating that activation of FUS3 results in changes in expression levels of a relatively small but reproducible gene set.

Gene ontology (GO) was utilized to categorize the activated and repressed genes by function (Figure 1C). Thirty-eight percent of genes upregulated by FUS3 activation appear to be involved in enzymatic activity, whereas a number of repressed genes are annotated as having roles in hormone synthesis or action. For example, the CYP707A3 gene that encodes the cytochrome p450 monooxygenase that catabolizes ABA is repressed more than twofold by FUS3 activation (Additional file 2, Table S2) 16 . This observation, in conjunction with the lack of FUS3-dependent induction of known ABA biosynthetic genes, suggests the increased ABA levels observed previously in FUS3 misexpression lines are due to decreased ABA catabolism 11 . A second class of hormone-related genes that are repressed by FUS3 activation at all time points and in both replicates are either involved in ethylene biosynthesis (ACS6, ACC synthase 6) or ethylene response (ERF1, ETHYLENE RESPONSE FACTOR1; ERF104, ETHYLENE RESPONSE FACTOR104; ESE3/EREBP, ETHYLENE AND SALT INDUCIBLE3; EDF4, ETHYLENE RESPONSE DNA BINDING FACTOR4). ERFs and EREBPs are transcription factors which contain an AP2 DNA binding domain, whereas EDF (or RAV) transcription factors contain both an AP2 and a B3 DNA-binding domain. Indeed, transcription factors constitute a dominant category (17%) of FUS3-repressed genes (Additional file 2, Table S2).

Potential downstream targets of FUS3 would be expected to contain the RY promoter element motif CATGCA, to which the B3 domain of FUS3 binds 17 18 . In fact, the RY sequence was found to be statistically enriched in both the up- and downregulated gene set compared to a randomized sample (p-value of 1.77 × 10^-4 and 6.06 × 10^-3 for 1, 000 bp upregulated and downregulated genes, respectively; Additional files 1 and 2, Tables S1 and S2). Together, these data strongly suggest that our experimental conditions identified a small gene set that is responsive to FUS3. Interestingly, this set includes only a few genes that typically mark seed maturation and late embryogenesis, such as seed storage proteins and late embryogenesis abundant proteins (LEA).

A closer inspection of our microarray data yielded additional ethylene-induced transcription factors (ERF2, EDF2 and EDF1) whose expression levels were repressed by FUS3 (Additional file 2, Table S2). These genes did not quite meet the stringent criteria of exhibiting a twofold decrease following both 2- and 4-day periods of FUS3 activation. Of the 14 AP2/EREBP/ERF and RAV/EDF genes that are known to be induced by ethylene exposure, six genes (ERF1, ERF2, ERF104, EDF2, EDF4 and EDF1) were dampened by FUS3 activation, suggesting that FUS3 plays a role in reducing the expression of genes involved in ethylene action 19 . This was verified by RT-PCR performed on ACS6, ERF1, EDF4 and EDF2 at 2 and 4 days of DEX induction (Figure 1D).

FUS3-mediated repression of genes involved in ethylene synthesis and action predict that fus3 loss-of-function mutants would exhibit an inverse expression pattern of these genes. Indeed, expression of ACS6, EDF2 and EDF4 genes were consistently higher in germinating fus3 versus wild type at all time points surveyed after imbibition, whereas ERF1 expression increased after 24 hours (Figure 2A). These results again support a negative role for FUS3 in regulating the expression of these ethylene responsive genes during germination. The presence of the RY element in the promoters of this gene subset may mean FUS3 directly downregulates their expression. Alternatively, FUS3 may act through ethylene synthesis or signaling to modulate their expression. To test these possibilities, we repeated the RT-PCR analysis of ACS6, ERF1, EDF4 and EDF2 on 2-day old wild-type and fus3 seedlings germinated in the presence of either an ethylene biosynthesis inhibitor, aminoethoxyvinylglycine (AVG), or an ethylene signaling inhibitor, silver ions (AgNO₃). In both cases, the inhibitors partially suppressed the increased levels of the gene transcripts in fus3 seedlings compared to untreated controls (Figure 2B). Furthermore, a key transcriptional regulator of ethylene signaling, EIN3 20 21 , showed increased protein stability in the emerging root cells of fus3 compared to wild type, and this stability was eliminated by the addition of silver ions (Figure 2C). Although these results do not exclude a direct role for FUS3 on ethylene-responsive gene transcription, it does suggest that at least part of the aberrant gene expression observed in fus3 mutants does require functional ethylene synthesis or signaling.

To further probe the connection of FUS3 with ethylene action during germination, we studied the growth of dark-grown fus3 loss-of-function seedlings. When wild-type seedlings are germinated in the dark in the presence of ethylene, they show an exaggerated apical hook, a shortening of the hypocotyl and reduced root growth 22 . Termed the "triple response, " this developmental assay has been extremely useful in genetically dissecting the role of various genes involved in ethylene synthesis or signaling in Arabidopsis 23 . Dark-germinated fus3 plantlets grown in the absence of ethylene exhibited shorter hypocotyl growth and hooked cotyledon development (Figure 2D), which is consistent with their ectopic ethylene gene expression. Moreover, these phenotypes were alleviated by the introduction of a mutation (ein2-1) that reduces ethylene signaling into the fus3 genetic background (Figure 2D). To further differentiate ethylene synthesis and response, we repeated the experiment in the presence of the ethylene biosynthesis inhibitor AVG. In contrast to the ein2 mutation, AVG did not restore the hypocotyl length of fus3, which suggests that the ectopic ethylene responses observed in dark-grown fus3 seedlings were not due to increased ethylene synthesis (Figure 2E).

The heightened ethylene responses observed in fus3 after germination suggest that loss of this gene function might have other, uncharacterized effects on vegetative development. Because FUS3 is a regulator of embryonic leaf identity, we decided to examine the vegetative leaf identity and phase transitions of the fus3 loss-of-function mutant more closely. From the emergence of the first leaf after germination through to flowering, each wild-type leaf adopts a unique identity based on size and shape, and this graded growth variation is often referred to as the "leaf heteroblastic series" 23 . Although rosette leaves always produce trichomes on the adaxial side of the wild-type leaf, trichomes on the abaxial side begin to appear on leaves only at later nodes (leaf 5 in wild type). Leaves that produce only adaxial trichomes are considered juvenile, whereas later leaves that show trichomes on both sides are regarded as adult 24 . Also, the ratio of leaf blade-to-petiole lengths is generally lower in early juvenile leaves and becomes higher as the plant transitions to adult leaves and flowering 25 .

Comparisons of leaf profiles between the fus3 mutant and wild-type plants showed that the first two leaves from the mutant plant were more similar in shape and size to the third and fourth leaves of the wild type, and that this shift continued throughout the heteroblastic series (Figure 3A). This precocious shift was reflected in the blade-to-petiole length ratio of each individual leaf (Figure 3B) and in the leaf trichome distribution (Figure 3C). Abaxial trichomes frequently (63%) appeared on leaf 3 in fus3 mutant plants, approximately two leaves earlier than the wild type.

There are two explanations for the vegetative phenotypes of the fus3 mutant. Possibly the altered cotyledon development during embryogenesis in the mutant advances the progression of vegetative leaf identities after germination. Alternatively, FUS3 may function in the developing vegetative meristem. We tested these possibilities directly by measuring FUS3 transcript levels using quantitative RT-PCR (qRT-PCR). In wild-type seeds, FUS3 transcript is detected 6 hours postimbibition and declines over time, as previously shown 26 ; it is still detectable 5 days after germination (Figure 4A). To further clarify the source tissue of postembryonic FUS3 expression, a transgenic line containing a sensitive FUS3:GUS transcriptional fusion was germinated and sampled over time for β-glucuronidase (GUS)-dependent blue histochemical staining 27 . The FUS3:GUS line used in these experiments was previously confirmed to reliably report embryonic FUS3 expression patterns based on FUS3 in situ hybridization 27 . Whole-mount preparations of seedlings showed blue staining in emerging leaf primordia from 2- to 5-day-old seedlings (Figure 4B). A similar expression pattern was found using a FUS3:GFP reporter previously described 26 (data not shown).

To functionally determine the effect of FUS3 from embryonic and vegetative tissues on the precocious vegetative phase transition, we constructed transgenic plants in which FUS3 could be activated either in the embryo or in vegetative tissue. fus3 was transformed with a FUS3-GR translational fusion construct that was under the control of the native FUS3 promoter (fus3 FUS3:FUS3-GR). Seeds produced from these lines in the absence of DEX application were phenotypically indistinguishable from fus3 loss-of-function mutants, with approximately 95% of the seed being desiccation-intolerant after 6 weeks of storage (data not shown). Of the small number of seeds that did survive desiccation, all produced ectopic trichomes on their cotyledons, as expected. Thus, in the absence of DEX application, the FUS3-GR fusion protein cannot rescue any of the known embryonic fus3 phenotypes. By contrast, in two independent experiments, approximately 50% of the seed produced by transgenic plants sprayed with 30 μM DEX during flowering showed desiccation tolerance after 6 weeks of storage. Of these, approximately 85% lacked ectopic trichomes on the cotyledons. The large reduction of fus3 seed phenotypes indicates that DEX application during flowering is sufficient to rescue mutant embryos.

From the transgenic seeds that were rescued, seven independent plants were randomly selected and grown to maturity in the absence of DEX, and leaf profiles were performed. Leaf profiles of all these lines still displayed accelerated heteroblastic development similar to that observed in fus3 leaf profiles (Figure 3A). Furthermore, the lines showed the same advanced vegetative phase transition as fus3 with respect to both individual leaf blade-to-petiole ratios (Figure 3B). The abaxial trichomes, however, did show a partial reversion to a wild-type profile (Figure 3C). Together, these results suggest that specifically removing embryonic FUS3 has a limited influence on the development of the first two juvenile leaves of Arabidopsis, but not on later leaves.

The inability of embryonic FUS3 to fully rescue the precocious vegetative phase transitions of fus3 suggests that this gene may function outside seed development. We tested this possibility by using the fus3 FUS3:FUS3-GR transgenic lines. Unlike in the previous experiment, however, we allowed transgenic lines to produce seed in the absence of DEX application and then germinated the seed in the presence of 10 μM DEX for 2 days. This short DEX application did not influence the heteroblastic phase change in wild-type plants as measured by either the blade-to-petiole transition or the appearance of abaxial trichomes in comparison to untreated plants (Additional file 3, Figure S1). By contrast, exposure of germinating fus3 FUS3:FUS3-GR transgenic seeds to DEX for 2 days resulted in a partial (62.5%) or full (8.3%) shift of leaf heteroblasty back toward a wild-type profile (Figure 4C). This rescue was partially reflected in trichome appearance (Figure 4D) and in full-leaf measurements; in wild type, the largest leaf was leaf 7, reaching a size of 3.9 ± 0.2 cm (n = 24). Similarly, the largest leaf in fus3 FUS3:FUS3-GR plants treated with DEX was leaf 7, which reached a size of 3.6 ± 0.3 cm (n = 23). In contrast, leaf 6 was the largest leaf in fus3 FUS3:FUS3-GR untreated plants (3.0 ± 0.3 cm; n = 23). In conclusion, a short pulse of FUS3 during the first 2 days of postembryonic growth can influence leaf identity in juvenile leaves.

Loss-of-function fus3 mutants show advanced vegetative phase transition, increased expression of ethylene-regulated genes and phenotypes that are characteristic of increased ethylene signaling. To test if these phenotypes are related, a mutation conferring ethylene insensitivity was introduced into the fus3 background. The introduction of the ein2 mutation did suppress defects in vegetative phase change compared to the fus3 single mutant as measured by leaf profiles, blade-to-petiole ratios and abaxial trichome appearance (Figures 5A to 5C). These results suggest that increased ethylene signaling does contribute to the advanced vegetative phase transition phenotype observed in fus3.

To further refine the ethylene contribution, we germinated fus3 seeds on 100 μM AgNO₃ for 2 days and transferred the resulting seedlings to soil. Again, the fus3 phase transition phenotypes in the presence of AgNO₃ were similar to those seen in the fus3 ein2 double-mutant (Figures 5C to 5E). Because in this experiment fus3 embryonic development occurred in the absence of the ethylene signaling inhibitor AgNO₃, any alterations in fus3 vegetative phase transition were due to the transient inhibition of ethylene signaling during germination. In conclusion, the inhibition of ethylene signaling during early germination is able to partially suppress the premature vegetative phase transitions observed in the fus3 mutant.

Four lines of evidence suggest that FUS3 negatively regulates ethylene action. First, gain-of-function activation of FUS3 dampens the expression of a collection of ethylene-responsive genes, whereas a loss-of-function fus3 allele shows the opposite effect. Second, the ectopic expression of these ethylene-responsive genes in light-grown seedlings is dampened by the addition of ethylene synthesis or signaling inhibitors. Third, loss-of-function fus3 seedlings show common ethylene-response phenotypes in the dark which are dependent on ethylene signaling and also have increased stability of the key positive regulator of ethylene signaling, EIN3. Finally, the precocious vegetative phase variation observed in fus3 mutants is suppressed if ethylene signaling is inhibited genetically or pharmacologically.

Inhibition of ethylene biosynthesis does not rescue fus3 phenotypes in the dark, suggesting that alteration of ethylene signaling in fus3 is not merely a consequence of increased ethylene levels and requires functional ethylene signaling. However, genetic manipulation of FUS3 did influence the expression of ACS6, encoding for an enzyme involved in ethylene biosynthesis. The altered expression of this gene might reflect a feedback response to changes in ethylene signaling. On this note, only a subset of ethylene signaling genes was regulated by FUS3, which suggests that the relationship between FUS3 and ethylene is complex. Many of the ethylene-responsive genes regulated by FUS3 contain RY elements in their promoter, which suggests that this transcription factor may bind these promoters directly. A direct regulation of gene transcription may explain why the addition of ethylene inhibitors did not fully alleviate the ectopic expression of the ethylene-responsive genes. Nevertheless, the negative effect of FUS3 on the regulation of ethylene signaling adds this hormone to ABA and GA as those that are dependent on FUS3 during germination 11 .

Coordination of the synthesis and signaling of various hormones is important in regulating overall plant growth and development, and many examples of ethylene, ABA and GA interactions have been reported 28 29 30 31 32 33 . In one of the best-studied cases, initial submergence of deepwater rice plants resulted in the accumulation of ethylene, which in turn enhanced GA sensitivity of the internodes to promote rapid growth 29 . The addition of ABA had the opposite effect by decreasing the sensitivity of internode tissues to GA 32 . With respect to leaf identity, interactions between ethylene and ABA have also been implicated in semiaquatic plants that exhibit heterophylly 34 . ABA induces terrestrial leaf development and ethylene stimulates the formation of submerged leaves in part by reducing ABA levels 30 . In Arabidopsis, where genetic analysis can be applied, loss of ethylene response does increase ABA levels in leaves, and, conversely, Arabidopsis mutants deficient in ABA synthesis show increased ethylene production 35 36 37 38 . The antagonistic and seemingly common relationships between ABA and GA and between ABA and ethylene imply that FUS3 acts as a control point of these multiple hormone pathways.

Suppressing ethylene signaling in fus3 mutants partially rescued the defects in juvenile leaf identities and the accelerated vegetative phase variation of this mutant, but did not rescue the embryonic leaf identity phenotypes. Perhaps this is not surprising, because there are major cellular differences beyond size and shape between embryonic cotyledons and juvenile and adult leaves. For example, genetic programs involved in storage reserves and desiccation appear to be embryo-specific. Related to this is that controlled FUS3 activation outside seed development was not sufficient to upregulate most of the standard seed-specific marker genes such as storage proteins, suggesting that other regulators are also required. The absence of seed-specific marker induction by the controlled activation of FUS3 in seedlings is similar to results obtained by Kagaya et al. 39 , who also utilized a transgene-based activation system. Under their study conditions, vegetatively activated FUS3 could induce ectopic seed storage gene expression only with the addition of ABA. These observations are in contrast to constitutive FUS3 misexpression experiments, which result in storage reserve accumulation in vegetative tissues. Genetically reducing ABA levels in plants that constitutively express FUS3 also attenuates the ability of FUS3 to activate embryonic programs ectopically, consistent with an ABA requirement for full FUS3 function 11 . It appears that there is a limited time during germination during which tissues are more sensitized to FUS3 action. The presence of the transcription factor ABI5 has been posited to define a 60-hour developmental window during germination during which ABA can still arrest the growth of the emerging embryo and induce late-embryogenesis gene expression 40 . Whether this 60-hour checkpoint also defines a window in which FUS3 can reactivate late embryogenesis programming needs to be determined, but our expression studies place postembryonic FUS3 within this developmental window.

Transitions from one leaf identity to the next depend on the coordination of at least two independent processes: the timing of leaf initiation and a program that determines the duration of a developmental phase 3 . In the latter case, the genes that define the juvenile boundary may repress the expression of adult leaf-promoting genes above a certain threshold. One prediction of this model is that these juvenile boundary regulators will decrease in activity over time to allow adult phase change to occur 41 . FUS3 transcripts are detected after germination, and although they decrease, they can still be detected 5 days postgermination at a time when at least four to five leaf primordia have formed 42 . Because leaf 5 is considered to be the first to show adult characteristics, the decreased temporal domain of FUS3 expression correlates well with decreasing juvenile leaf identity. A simple model would suggest that the presence of FUS3 in leaf primordia contributes to juvenile leaf identity and also that as FUS3 activity drops, the transition to adult leaf phases can occur.

Although we have shown a vegetative role for FUS3, embryonic expression of FUS3 also affects the leaf identity of the first two juvenile leaves. This is similar to studies involving LEC1, which suggests that the identity of the embryonic foliar organs influences the identity of subsequent vegetative organs despite their different developmental origins 15 . Unlike fus3 mutants, however, aberrant vegetative phase transitions in lec1 plants are corrected in later leaves 15 . The lack of a detectable LEC1 signal after germination, as shown in public microarrays 43 , may explain the difference between these mutants.

In contrast to animals, plant hormones can function in many different tissues and at various times during development, resulting in various developmental outcomes. One of the mechanisms by which FUS3 prolongs the juvenile phase appears to involve the downregulation of ethylene action, which suggests that this hormone is important in promoting adult-phase transitions (Figure 6). Morphometric analysis of leaf growth of etr1 mutants have shown that this ethylene-insensitive plant produce shorter, broader leaves, which are more akin to a juvenile leaf identity 37 . Consistent with this, the vegetative-to-floral transition in a collection of ethylene-insensitive mutants is delayed compared to the wild type 44 . In agreement with this, ML1:FUS3-GFP plants that constitutively express FUS3 postembryonically also show delayed flowering 11 26 . Using ethylene to quicken vegetative phase transitions is consistent with the role of this hormone in promoting germination 36 38 . Perhaps the short vegetative pulse of FUS3 serves to dampen the action of this hormone after it has stimulated germination so that the leaf identities do not advance too quickly.

fus3-3 (fus3) 7 and all other strains used in this study were derived from a Columbia (Col) genetic background. For all experiments, plants were grown at 20°C under constant light. Seeds from all genotypes were generated from parent plants grown under identical conditions and stored under the same conditions for the same length of time. All transformations were performed as previously described 45 . DEX (Sigma-Aldrich, St Louis, MO, USA) was dissolved in DMSO. We performed leaf profiles according to Telfer et al. 24 , for which 10 to 15 seedlings were germinated and grown in individual 4-inch round pots for approximately 3.5 weeks. To determine if DEX had any off-target effects on leaf heteroblasty, we performed blade-to-petiole measurements and trichome appearance assays on 10 wild-type plants and found that the chemical did not influence either of these markers of phase transition (Additional file 3, Figure S1). For the ein2 fus3 double-mutant construction, ein2-1 homozygotes were identified by screening for an altered triple-response phenotype in a segregating F2 population. Positive ein2-1 plants were grown and assayed for fus3 allele using a cleaved amplified polymorphic sequence marker that marks the mutant polymorphism 11 . For the pharmacological studies, 100 μM of the inhibitor AgNO₃ or 10 μM of the inhibitor AVG were added to MS plates.

The EIN3 cDNA was amplified and carboxy-terminally fused to GFP under the control of the 35S promoter in the pEGAD vector using the primers EIN3-BamHI forward 5'-ATA GGA TCC ATG ATG TTT AAT GAG ATG GGA AAT G-3'; EIN3-BamHI reverse 5'-ATA GGA TCC GAA CCA TAT GGA TAC ATC TTG C-3'; and the BamHI enzyme 46 to generate 35S:GFP-EIN3. The GFP-EIN3 translational fusion reporter fully rescues ein3 loss-of-function ethylene-dependent phenotypes (data not shown).

Thirty PCR cycles were performed with 250 ng of total RNA to amplify ethylene-related genes identified through microarrays using the following primers: ERF1 forward 5'-GTA TCC TCA ACG ACG CCT TTC AC-3'; ERF1 reverse 5'-CTT CAC CGT CAA TCC CTT ATC C-3'; EDF4/RAV1 forward 5'-GTA CAG GTT CCA TCT GTG AAA CC-3'; EDF4/RAV1 reverse 5'-CTC GTC TTC GTC CAT CTT CAC GTC-3'; EDF2/RAV2 forward 5'-GCA TAG ACG AGA TAA GTT CCT CCA C-3'; EDF2/RAV2 reverse 5'-GTT CTT GAA GTT GAC GAC GGC GTC G-3'; ACS6 forward 5'-ACA TCC AGA AGC TTC GAT TTG TAC-3'; and ACS6 reverse 5'-CTC GTC TAC AAA CTC TTC ATC GGA-3'. These primers were also used to check for expression of genes involved in ethylene synthesis and signaling in dry wild-type (Col) and fus3 immature seeds. ACTIN7 was amplified using the following primers: ACT7 forward 5'-GGT GAG GAT ATT CAG CCA CTT GTC TG-3' and ACT7 reverse 5'-TGT GAG ATC CCG ACC CGC AAG ATC-3' and was used as the control. Seeds were imbibed as described for 12, 24 and 48 hours.

RNA extraction and qRT-PCR were performed as previously described 26 . Sequences of gene-specific primers were designed to amplify 190-bp products. These included the following: ACT7₁₉₀ forward 5'-TCA CAG AGG CAC CTC TTA ACC-3'; ACT7₁₉₀ reverse, 5'-CCC TCG TAG ATT GGC ACA G-3'; FUS3₁₉₀ forward, 5'-TGT GAA TGC TCA TGG TCT GC-3'; and FUS3₁₉₀ reverse, 5'-GGA GGA GAA GAT CGT TAA CCA C-3'.

Detection of GUS activity and sectioning was performed as described by Donnelly et al. 47 . Leaf tissues were cleared in 70% ethanol and 8:2:1 chloral hydrate:glycerol:water, then mounted in the same solution on microscope slides.

Wild-type Columbia seeds containing the ML1:FUS3:GR construct 11 were imbibed for 5 days at 4°C on filter paper placed on MS plates. To minimize the effects of biological variation, we followed a pooling strategy according to that described by Zhu et al. 48 in which approximately 100 individual plants per replicate experiment were harvested and pooled before RNA extraction and hybridization. The seedlings were then transferred to 20°C for 5 days. Shoot tissues from seedlings were collected for RNA isolation (day 5). The remaining seedlings on the filter paper were transferred to either MS plates consisting of 1 μM DEX or the same concentration of DMSO. Two and four days after the transfer, shoot tissue was collected from the seedlings. The above protocol was repeated to obtain a second biological replica.

The standard Affymetrix labeling and hybridization protocols were used on an ATH1 Affymetrix GeneChip microarray encompassing 22, 814 probe sets (Affymetrix Inc, Santa Clara, CA, USA). Data were globally normalized using the MAS5.0 global normalization algorithm with a target value of 500. The resulting data were filtered to eliminate MAS5.0 "marginal" and "absent" calls (Present requires both sets of replicates for treatment and control samples). In addition, the r values for the correlation of the expression values between the two replicate set of experiments was greater than 0.95. The complete data set has been submitted to the EMBL-EBI ArrayExpress database under the accession number E-MEXP-3465 in compliance with MIAME standards.

The average expression value for each gene was calculated for two replicate samples at each time point. The Cluster and TreeView software programs were used for cluster analysis 49 . The data set was filtered to identify genes that exhibited more than a twofold increase or decrease in expression level 50 . Cluster analysis was performed using average linkage hierarchical clustering with the centered Pearson correlation coefficient as the similarity metric. RY motif (CATGCA) and p-values were obtained using the "Motif Analysis" tool on TAIR http://www.arabidopsis.org/tools/bulk/motiffinder/index.jsp.

The FUS3:FUS3-GR construct was made by replacing the ML1 promoter in the ML1:FUS3-GR construct with the FUS3 promoter from the FUS3:FUS3-GFP construct previously described 11 . For FUS3:FUS3-GR activation during seed development, DEX was diluted to 30 μM in dH₂O containing 0.05% Triton X-100 and sprayed on flowering plants twice weekly until the plants had senesced.

To study EIN3 stability in a fus3 background, we crossed fus3 with an ein3 35S:GFP:EIN3 transgenic line. The stability of the EIN3 protein in the fus3 background was compared to non-fus3 mutant plants from the same F1 plant. At least 20 seedlings were observed for each genotype. A Nikon inverted fluorescence microscope equipped with a Nikon water immersion objective (Nikon Instruments, Inc, Melville, NY, USA) and a Bio-Rad Radiance 2000 confocal head (Bio-Rad Laboratories, Hercules, CA, USA) was used to detect GFP fluorescence. The same confocal settings were used in all experiments.