http://orcid.org/0000-0001-6255-7670
Figures
Abstract
While widespread transcriptome changes were previously observed with seed dormancy loss, this study specifically characterized transcriptional changes associated with the increased seed dormancy and dormancy loss of the gibberellin (GA) hormone-insensitive sleepy1-2 (sly1-2) mutant. The SLY1 gene encodes the F-box subunit of an SCF E3 ubiquitin ligase needed for GA-triggered proteolysis of DELLA repressors of seed germination. DELLA overaccumulation in sly1-2 seeds leads to increased dormancy that can be rescued without DELLA protein destruction either by overexpression of the GA receptor, GA-INSENSITIVE DWARF1b (GID1b-OE) (74% germination) or by extended dry after-ripening (11 months, 51% germination). After-ripening of sly1 resulted in different transcriptional changes in early versus late Phase II of germination that were consistent with the processes known to occur. Approximately half of the transcriptome changes with after-ripening appear to depend on SLY1-triggered DELLA proteolysis. Given that many of these SLY1/GA-dependent changes are genes involved in protein translation, it appears that GA signaling increases germination capacity in part by activating translation. While sly1-2 after-ripening was associated with transcript-level changes in 4594 genes over two imbibition timepoints, rescue of sly1-2 germination by GID1b-OE was associated with changes in only 23 genes. Thus, a big change in sly1-2 germination phenotype can occur with relatively little change in the global pattern of gene expression during the process of germination. Most GID1b-OE-responsive transcripts showed similar changes with after-ripening in early Phase II of imbibition, but opposite changes with after-ripening by late Phase II. This suggests that GID1b-OE stimulates germination early in imbibition, but may later trigger negative feedback regulation.
Citation: Nelson SK, Steber CM (2017) Transcriptional mechanisms associated with seed dormancy and dormancy loss in the gibberellin-insensitive sly1-2 mutant of Arabidopsis thaliana. PLoS ONE 12(6): e0179143. https://doi.org/10.1371/journal.pone.0179143
Editor: Tzvetanka D. Dinkova, Universidad Nacional Autonoma de Mexico, MEXICO
Received: March 11, 2017; Accepted: May 24, 2017; Published: June 19, 2017
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: Microarray raw data files are available at the ArrayExpress database (http://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-4782.
Funding: This work was supported by National Science Foundation (NSF) Award 0850981 to CMS (https://www.nsf.gov/funding/pgm_summ.jsp?pims_id=501087), and USDA-ARS project 424575 to CMS (https://www.ars.usda.gov/pacific-west-area/pullman-wa/whgq/).
Competing interests: The authors have declared that no competing interests exist.
Introduction
The evolution of seeds that carry plant embryos in a state of arrested growth was critical to the success of land plants and to agriculture [1]. Because seeds carry nutrient reserves for the germinating embryo, they are a convenient food source, comprising 70% of the human diet (reviewed in [2]). Seeds of temperate species can be dormant at maturity, meaning that they are unable to germinate under favorable conditions [3]. Dormancy prevents germination out of season (fall versus spring), and allows species to survive natural disasters as seeds in the soil. Seed dormancy can be lost through a period of dry storage called after-ripening, through a period of moist chilling called cold stratification, or through seed coat scarification. Seed dormancy is associated with high levels of the hormone abscisic acid (ABA) and germination is associated with high levels of the hormone gibberellin (GA) (reviewed in [4]). Increased ABA signaling or decreased GA signaling may promote seed dormancy through transcriptional regulation. Previous studies identified transcripts differentially regulated with dormancy loss through after-ripening [5–10]. This study examined changes in the global pattern of gene expression associated with seed dormancy and dormancy loss in a GA-insensitive mutant that over-accumulates DELLA protein repressors of seed germination. The events that occur during seed germination and GA signaling are reviewed to provide background for the transcriptome analysis.
Seed germination is both an event and a three-phase process (reviewed in [2]). Phase I involves the rapid uptake of water and cellular rehydration, and occurs in both dormant and non-dormant seeds. Phase II processes prepare the seed for germination, including roughly in order: the repair of DNA damage sustained during dehydration and storage, mitochondrial repair and initiation of cellular respiration, the mobilization of stored nutrient reserves, the initiation of transcription and protein translation, and lastly DNA synthesis and cell expansion in the radicle or embryonic root. Phase III begins with germination per se, embryo emergence from the seed, and includes the post-germinative processes of cell division, seedling growth, and the majority of nutrient mobilization. Dormant seeds undergo the Phase II processes of cellular restoration, DNA repair, initiation of respiration, and some RNA transcription and protein translation, but never reach Phase III. Differences in dormant seed Phase II processes likely prevent germination per se. Since dormant seeds will not complete the germination process, we will refer to ungerminated seeds undergoing Phase I and II as “imbibing seeds.” While imbibing in the cold breaks dormancy, Arabidopsis thaliana seeds do not reach Phase III during this cold stratification and eventually enter secondary dormancy [5,11]. Brief cold stratification (i.e. 3–5 d at 4°C) synchronizes Arabidopsis seeds in early Phase II [12,13]. GA stimulates multiple germination processes, including radicle cell elongation, cotyledon expansion, and the production of enzymes that mobilize stored reserves and weaken barrier tissues such as the aleurone and testa (reviewed in [14], [12,15]).
GA stimulates GA responses by negatively regulating DELLA (Asp-Glu-Leu-Leu-Ala) domain repressors of GA signaling (reviewed in [16]). In the GA biosynthesis mutant, ga1-3, DELLA protein levels are high and block GA responses including stem elongation, cell division, the transition to flowering, and seed germination [17,18]. GA treatment of ga1-3 results in rapid DELLA proteolysis followed by rescue of phenotypes including germination [19–21]. The partially overlapping functions of the five Arabidopsis DELLA genes were defined based on differences in the ability of DELLA knockouts to rescue ga1-3 phenotypes (reviewed in [22]). Only the DELLA RGL2 (RGA-LIKE2) knockout rescued ga1-3 germination in the light, but mutations in DELLAs RGA (REPRESSOR OF GA1-3) and GAI (GA INSENSITIVE) were also needed to rescue ga1-3 dark germination [23].
DELLAs positively and negatively regulate transcription by interaction with other transcription factors involved in a wide range of processes [24–27]. While DELLA does not directly bind DNA, DELLA RGA chromatin immunoprecipitation (ChIP)-seq and ChIP-qPCR analysis revealed association with a wide range of promoter elements including cytokinin-regulated, GA-responsive, light-regulated, stress-regulated, and ABA-related genes [24,28]. Examples of DELLA-interactors include: 1) the basic helix-loop-helix (bHLH) transcription factors that regulate cell elongation in response to light, PIF3 (PHYTOCHROME INTERACTING FACTOR3), PIF4, PIL2 (PIF3-LIKE2), and PIF1/PIL5, 2) the regulator of brassinosteroid signaling BZR1 (BRASSINOZALE-RESISTANT1), 3) the negative regulator of jasmonic acid signaling JAZ1 (JASMONATE ZIM DOMAIN1), and 4) the MYB transcription factor GL1 (GLABRA1) and bHLH GL3 involved in trichome initiation (reviewed in [22], [25,29]).
GA triggers DELLA proteolysis via the ubiquitin-proteasome pathway. The GA receptor GID1 (GA-INSENSITIVE DWARF1) undergoes a conformational change upon GA binding, stimulating interaction with DELLA proteins [30–33]. The formation of the GID1-GA-DELLA complex causes DELLA recognition by the SLEEPY1 (SLY1) F-box subunit of an SCF (Skp, Cullin, F-box) E3 ubiquitin ligase [34–38]. The SCFSLY1 complex catalyzes DELLA polyubiquitination, targeting it for degradation via the 26S proteasome. GA-stimulated DELLA destruction lifts DELLA repression of GA responses. Like ga1-3, the triple knockout of the three Arabidopsis GID1 receptors, GID1a, GID1b, and GID1c, causes failure to germinate unless the seed coat is cut [21,35,39,40]. The GA-insensitive sly1 F-box mutants exhibit varying degrees of increased seed dormancy associated with failure to degrade DELLA protein. This study makes use of the sly1-2 allele, a 2-bp deletion resulting in a frameshift and the loss of the last 40 of 151 amino acids, including the proposed DELLA binding site.
DELLA repression in the Arabidopsis sly1 and the rice gid2 F-box mutants can be lifted by non-proteolytic mechanisms [32,41,42]. Although sly1 mutants accumulate more DELLA protein than ga1-3 or gid1a gid1b gid1c triple mutants, they show weaker GA-insensitive phenotypes. This suggests that some GA signaling occurs in sly1 mutants despite high DELLA protein levels [17,35,39,43]. The germination of sly1-2 seeds can be rescued by two mechanisms that do not reduce DELLA protein accumulation, after-ripening and GID1 gene overexpression [21,42]. One to two years of dry after-ripening are needed to relieve sly1-2 dormancy, compared to two weeks of after-ripening in Landsberg erecta wild-type (Ler wt). The sly1-2 mutant allows us to look at after-ripening mechanisms that can function without DELLA destruction. Partial rescue of sly1 germination by GID1a, GID1b, and GID1c overexpression (GID1-OE) was associated with increased GID1-GA-DELLA complex formation, suggesting that GID1 can inactivate DELLA repressors without SLY1-directed DELLA destruction. GID1b-OE rescued sly1-2 seed germination better than GID1a-OE and GID1c-OE, likely because GID1b has higher affinity for GA and DELLA [42,44,45]. The model is that sly1-2 after-ripening leads to increased GA hormone levels and higher GID1b protein levels, thereby inactivating DELLA via GID1-GA-DELLA complex formation [42,46]. If both after-ripening and GID1b-OE rescue germination through increased GID1-GA-DELLA complex formation, then they should result in overlapping changes in the global pattern of transcript abundances. Moreover, genes showing differential regulation with after-ripening and GID1b-OE in sly1-2 may represent the non-proteolytic mechanisms needed to overcome DELLA repression of seed germination.
It is important to define specific pathways governing seed dormancy. This has been difficult given the plethora of expression changes observed in multiple Arabidopsis ecotypes, times in imbibition, and under multiple environmental conditions [5–8,47,48]. This problem is due to the fact that multiple genetic mechanisms cause dormancy in different species, genotypes, and at different times in the germination process. By identifying global patterns of transcript abundance associated with sly1-2 dormancy and dormancy loss, this study teased apart DELLA-dependent dormancy mechanisms, as well as dormancy-release mechanisms that do not require DELLA destruction. Distinct transcriptional changes were observed during early and late Phase II of germination. Interestingly, sly1-2 dormancy loss due to GID1b-overexpression was associated with changes in a very small set of transcripts. Thus, rescue of sly1-2 germination can occur with little change in the transcriptome. Finally, gene ontology analysis revealed that GA and SLY1 are important for increasing abundance of transcripts involved in protein translation with dormancy loss.
Results
Strategy for identifying SLY1-dependent and -independent regulation of seed dormancy
Germination kinetics following cold stratification for 4 days at 4°C were determined for seeds of: a) non-dormant wild-type Ler (WT) after-ripened for 2 weeks, b) dormant sly1-2 after-ripened for 2 weeks (sly1-2(D)), c) non-dormant sly1-2 after-ripened for 19 months (sly1-2(AR)), and d) non-dormant sly1-2 GID1b-overexpressed (sly1-2 GID1b-OE) after-ripened for 2 weeks (Fig 1A). WT showed neither testa rupture nor radicle emergence at 12h of imbibition, but achieved 96% germination within 24 h, indicating lack of seed dormancy. In contrast, the sly1-2(D) sample showed no germination after 8 d of imbibition, indicating failure to lose dormancy with the same 2 weeks (wk) of after-ripening. Germination of sly1-2 was rescued to 74% by day 8 in sly1-2 GID1b-OE seeds and to 51% germination by day 6 by sly1-2(AR). Cold stratification was used to improve the synchrony of germination in sly1-2 GID1b-OE, sly1-2(AR), and Ler wt. It should be noted that cold stratification does not rescue the germination of dormant sly1-2 seeds.
(A) Germination of seeds used for microarray analysis including: 2-week-old Ler wt, 2-week-old sly1-2 (dormant), 19-month-old sly1-2 (after-ripened), and 2-week-old sly1-2 GID1b-OE seeds. (B) For comparisons, the four seed samples are referred to as WT, D, AR, and GID, respectively. Seeds of each line were cold stratified for 4 days at 4°C in the dark and then transferred to the light at 22°C. Seeds were harvested for RNA extraction at the “0h” (blue) timepoint immediately after the cold stratification, and at the “12h” (green) timepoint (cold stratification followed by 12 h in the light at 22°C). (C) Experimental comparisons made in this paper, including comparisons from reanalysis of data from Carrera et al. [6] and Cao et al. [49].
Using these four sets of seeds, an experiment was designed to examine transcriptional mechanisms of seed dormancy and dormancy loss in the sly1-2 mutant. Seeds were sampled at two imbibition timepoints representing early and late Phase II, and subjected to Affymetrix oligonucleotide-based microarray analysis. Cold stratification for 4 d at 4°C in the dark was used to synchronize seeds in Phase II [2]. Seeds at the “0 hour” (0h, early Phase II) timepoint were harvested immediately after cold stratification, while seeds at the “12 hour” (12h, late Phase II) timepoint were imbibed for 12 h in the light at 22°C after cold stratification (Fig 1B). These timepoints allowed examination of transcriptome differences during the early and late germination process, before germination per se.
This experimental design allowed comparisons of dormant and less dormant samples within the current study, and between this study and reanalyzed microarray datasets from previous studies (Fig 1C). D refers to sly1-2(D or Dormant) and AR refers to sly1-2(AR or After-ripened) for all comparisons but for the Ler ARvsD from Carrera et al. [6]. WT refers to Ler wt, while GID refers to sly1-2 GID1b-OE. Our 0h imbibition timepoint matched the seed imbibition conditions used by Cao et al. [49] for WT, ga1-3, and ga1-3 della transcriptomes (where ga1-3 della refers to ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1). The Ler D and AR seeds used in Carrera et al. [6] were sampled at 24 h of imbibition without cold stratification representing late Phase II, allowing comparison with the sly1-2 12h timepoint as, in both cases, germination per se of non-dormant seeds occurred approximately 12 h after sampling.
Transcriptome differences associated with sly1-2 dormancy and after-ripening
The DvsWT comparison revealed transcriptome differences associated with the sly1-2 dormancy phenotype, including 1741 up- and 1818 down-regulated transcripts at the 0h and 3247 up- and 3434 down-regulated transcripts at the 12h imbibition timepoint (S1 Table). Since the sly1-2 mutant over-accumulates DELLA protein, these differentially-regulated genes may be indirectly considered DELLA-regulated genes. DELLA protein over-accumulation occurs in both sly1-2 and ga1-3 mutants, but only sly1-2 germination can be rescued through after-ripening, likely via DELLA-proteolysis-independent GA signaling [21]. To compare the effects of DELLA accumulation in sly1 to DELLA accumulation in ga1-3, we compared 0h DvsWT to ga1-3 vs WT (S1A Fig). If the ga1-3 and sly1-2 germination phenotypes result solely from accumulation of DELLA repressors of seed germination, then these two mutations should result in nearly identical changes in transcript accumulation. Instead, the sly1 DvsWT and ga1-3 vs WT comparisons had a large but incomplete overlap of about 39% of either individual comparison. 60% of the DvsWT transcriptome changes were unique to the sly1-2 comparison, while about 61% of the ga1-3 vs WT changes were unique to the ga1-3 comparison. Transcriptome differences between ga1-3 and sly1-2 may shed light on the role of DELLA-proteolysis-independent GA signaling in sly1 dormancy loss.
Based on transcriptome profiles, the dormancy and dormancy-loss mechanisms of sly1-2 mutants have much in common with wild-type dormancy mechanisms. There was more than a 50% overlap between the WTvsD and Ler ARvsD comparisons in late Phase II (Fig 2A). Thus, many transcripts associated with sly1-2 dormancy are also regulated with dormancy loss through after-ripening of Ler wt. This suggests that the transcripts with altered expression may actually cause the dormancy phenotype of sly1-2. After-ripening of sly1-2 was associated with 1385 transcriptome changes similar to those observed in wild-type Ler (S1B Fig). Moreover, the 86% that overlapped were also differentially abundant in the 12h WTvsD comparison (S1D Fig). These changes occurring with after-ripening of both Ler and sly1-2 do not require DELLA destruction, while those that occur only in wild-type Ler after-ripening likely involve DELLA destruction by SCFSLY1 (S2 Table).
(A) Overlap between genes inversely regulated with the sly1 mutation at 12h with genes differentially-regulated with after-ripening of wild-type Ler at 24 h (12h WTvsD ∩ 24 h Ler ARvsD). (B) Overlap between genes AR-regulated genes in sly1-2 at 0h and at 12h (0h sly1-2 ARvsD ∩ 12h sly1-2 ARvsD). (C) Overlap of genes inversely regulated with the sly1 mutation and with sly1-2 after-ripening at 12h (12h WTvsD ∩ 12h sly1-2 ARvsD).
AR-dependent transcriptome differences in sly1-2 (ARvsD) were identified at early (0h) and late (12h) timepoints in Phase II of the germination process (S1 Table). The two imbibition timepoints were transcriptionally distinct. Although a similar number of differentially-regulated transcripts were observed at 0h and 12h in sly1-2 ARvsD (2599 and 2579, respectively), there was only a 22% overlap between differentially-regulated transcripts at the two imbibition timepoints (Fig 2B). Moreover, analysis of genome-wide expression plots using the adjusted Fisher-Pearson standardized moment coefficient (G1) as a measure of skew indicated that the 12h sly1-2 ARvsD dataset was more highly skewed towards positive changes in expression (G1 = 1.19) (Fig 3B), than the 0h sly1-2 ARvsD dataset (G1 = 0.37, where G1 = 0 indicates a completely symmetrical dataset) (Fig 3A) [50]. Thus, it appears that more transcripts are induced as germination per se approaches.
Genes differentially regulated with sly1-2 after-ripening at the (A) 0h and (B) 12h imbibition timepoints. Genes differentially regulated with GID1b-OE (GIDvsD) at (C) 0h and (D) 12h. Genome-wide expression plots show the skew, magnitude, and chromosomal distribution of differentially-regulated genes. The log2-fold changes (logFCs) in transcript abundances for each comparison were plotted on the y-axis versus a number between 1 and 22,810 corresponding to a gene’s chromosomal location on the x-axis. Genes with significant differential regulation are in blue or green (based on FDR p < 0.05). Positive logFC indicates up- and negative logFC down-regulation. Shaded area mark the ±2 and ±3 logFC to allow comparison of magnitude and skew.
Gene ontology analysis of after-ripening and imbibition time
The gene categories differentially regulated with sly1-2 after-ripening were examined at the two imbibition timepoints by TAGGIT gene ontology analysis using categories defined by relevance to seed germination [6] (see Materials and Methods; S2 Fig). At 0h, sly1-2 after-ripening was differentially enriched for categories associated with early Phase II, including DNA repair, protein degradation/inhibition of protein degradation (involved in mobilization of seed storage proteins), and protein translation (Fig 4A). At 12h, sly1-2 after-ripening was differentially enriched for categories associated with late Phase II and the transition to Phase III, including stored reserve mobilization (beta-oxidation, glycolysis and gluconeogenesis, protein degradation), cell division, cell-wall modification (needed for seedling emergence and growth), and photosynthesis (Fig 4B). The TAGGIT profile at 12h resembled late Phase II of the Ler ARvsD (Fig 4C). Transcriptome changes in 24 h Ler ARvsD accounted for 54% of those in sly1-2 ARvsD at the 12h, but only 28% of those at the 0h timepoint (S1B Fig).
(A) Genes differentially regulated with sly1-2 after-ripening at 0h in early Phase II (sly1-2 ARvsD comparison), and (B) at 12h in late Phase II. (C) Genes differentially regulated with Ler after-ripening in late Phase II (24 h Ler ARvsD). The x-axis value shows the percentage of total up- or down-regulated genes within a dataset.
The TAGGIT category with the most distinct difference between sly1-2 and wild-type was the protein translation category. Protein translation was more highly up-regulated in the 24h Ler ARvsD than at either of the sly1-2 ARvsD timepoints (Fig 4C). Consistent with the notion that SLY1 is needed for translation-associated gene induction, TAGGIT analysis of the DvsWT dataset revealed down-regulation of the translation-associated category in sly1-2 at 0h and 12h (Fig 5A and 5B). Translation-associated genes down-regulated in the DvsWT and up-regulated in the Ler ARvsD comparisons included ribosomal structural proteins, tRNA synthetase genes, and elongation factors, as shown in S3 Table. Thus, SLY1 appears to be needed for the efficient up-regulation of genes involved in protein translation with after-ripening. If so, then we would expect GA to regulate translation-related gene expression. Indeed, our TAGGIT analysis revealed that protein translation is the most strongly GA-upregulated and DELLA-downregulated gene category in early Phase II seeds (S3A, S3B and S3C Fig; [49]).
(A) At 0h in early Phase II. (B) At 12h in late Phase II. The value on the x-axis shows the percentage of either the total up-regulated or total down-regulated genes within a dataset.
Some categories switched from being more up- to more down-regulated (or vice versa) between the 0h and 12h timepoints, suggesting that negative regulation of germination per se early in Phase II may be reversed in late Phase II. For example, the dormancy-related category was more up-regulated, while the germination and protein degradation categories were more down-regulated at 0h than at 12h. The ABA category was differentially enriched at both timepoints, but showed a greater ratio of up- to down-regulation at 0h. Note, however, that the TAGGIT ABA category does not differentiate between positive and negative regulators. There was strong down-regulation in the ethylene category at the 12h timepoint, but little change in the GA-related category. The Ler ARvsD comparison also showed little change in the GA category, suggesting that this was not due to the sly1-2 mutation (Fig 4C; [6]).
After-ripening regulated transcription factor targets
Transcription factors (TFs) possibly controlling after-ripening-regulated transcription were identified in the sly1-2 ARvsD datasets using the Plant GeneSet Enrichment Analysis (PlantGSEA) tool that identifies TFs based on enrichment of their experimentally identified targets in a differentially-regulated gene set (Table 1; [51–53]). There was strong enrichment for E2Fa–DPa (the heterodimeric E2Fa-Drosophila melanogaster a complex) and PIF1/PIL5 targets at both 0h and 12h. E2Fa–DPa may promote germination per se because it is a positive regulator of the cell division cycle, and stimulates gene expression involved in DNA replication, nitrate assimilation, and cell wall biosynthesis [54,55]. PIF1/PIL5 is a light-repressed negative regulator of seed germination, and was enriched in the down-regulated fraction [56,57]. There was also enrichment for targets of AP2 (APETALA2), and light-responsive TFs, PIF3 and HY5 (ELONGATED HYPOCOTYL5) [58]. In addition to floral development, AP2 regulates seed size, seed coat formation, and storage of seed mucilage during development [59]. Interestingly, PIF1/PIL5 and PIF3 are DELLA-interacting proteins, raising the possibility that DELLA-regulated genes are after-ripening-regulated via PIF1 and PIF3.
Zentella et al. [24] identified 18 putative DELLA RGA targets by microarray, and confirmed DELLA localization to 8 of these promoters by DELLA ChIP-qPCR. We examined whether these 18 putative DELLA target genes were differentially regulated by after-ripening, sly1, GA, or DELLA (Table 2). Our reanalysis of the Cao et al. [49] microarray datasets defined GA-regulated transcripts based on the WT vs ga1-3 comparison, and DELLA-regulated transcripts based on ga1-3 vs ga1-3 della (Fig 1C). For convenience, transcripts down-regulated in the presence of GA (lower in WT than ga1-3) are called GA-DOWN, and those up-regulated by DELLA (higher in ga1-3 than in the ga1-3 della) are called DELLA-UP. DELLA-UP transcripts are generally also GA-DOWN [24,49]. Since sly1-2 mutants over-accumulate DELLA, DELLA-UP genes are expected to be up-regulated in sly1-2 (sly1-UP), and vice versa. Indeed, out of the 3213 DELLA-regulated and the 3559 0h sly1-regulated genes, only 72 genes showed opposite DELLA and sly1-regulation. If sly1-2 after-ripening leads to down-regulation of DELLA by a non-proteolytic mechanism, then we would expect DELLA-UP genes to be after-ripening (AR)-DOWN, and vice versa. Of the 18 putative DELLA targets, 8 showed significant differential AR-regulation, and of these 6 AR-DOWN targets were also DELLA-UP and/or sly1-UP at 12h. Moreover, the DELLA target GID1a was AR-DOWN/DELLA-UP at 0h and 12h. The DELLA-UP target XERICO was AR-UP at 0h and AR-DOWN at 12h of imbibition, suggesting that its regulation changes over imbibition time. The MYB TF At3g11280 was DELLA-UP and AR-UP.
Few transcripts are differentially regulated with GID1b overexpression, and a subset of these are DELLA-regulated
Although GID1b-OE rescued sly1-2 germination as well as after-ripening, it was associated with far fewer changes in transcript accumulation, including only 23 transcripts over the 0h and 12h timepoints (Fig 6; S4 Fig; S1 Table). In the GIDvsD comparison, there were 13 differentially-regulated genes at 0h and 14 at 12h. Four genes showed the same differential regulation at both imbibition timepoints, including GID1b, At1g21630, BAG6 (BCL-2-Associated Anthogene6), and At2g46250. Given that it is overexpressed on the 35S promoter, GID1b served as an internal control, showing the strongest up-regulation in GIDvsD at 0h and 12h (logFC > 8, Fig 3C and 3D). Down-regulated genes BAG6 (At2g46240) and At2g46250, may share a promoter as they are oriented head-to-tail on chromosome 2. The putative EF hand domain protein At1g21630 was GID1b-UP at 0h and 12h in both the GIDvsD and sly1-2 GIDvsAR comparisons (logFC > 2.68).
The GIDvsD column provides the GID1b-OE–UP (red) and -DOWN (purple) regulated genes at 0h and 12h. The GA column indicates GA-regulation based on the Ler wt vs ga1-3 comparison and DELLA column indicates DELLA-regulation based on the ga1-3 vs ga1-3 4x della comparison (Cao et al. [49]). The DvsWT column indicates the DELLA/sly1-regulation. Black borders indicate significance based on an FDR of p < 0.05.
In the non-proteolytic GA signaling model, GID1b overexpression rescues germination by down-regulating DELLA repressors via protein-protein interaction [41,42]. Thus, GID1b-OE was expected to mainly cause changes in DELLA-regulated transcript levels. Indeed, only 3 of the 22 GID1b-OE-regulated genes were not found to be GA-, DELLA-, or sly1-regulated (Fig 6; S4 Fig). It was expected that GID1b-OE-UP genes would also be GA-UP/DELLA-DOWN/sly1-DOWN or vice versa. Surprisingly, 15 of the 22 GID1b-OE-regulated gene showed the opposite expression pattern, suggesting that the change with GID1b-OE may not be due to lifting DELLA repression (Fig 6). The EXPL1 gene behaved as expected, showing GID1b-OE-UP and sly1-DOWN regulation at 12h. Six of the GID1b-OE-DOWN genes behaved as expected, showing GA-DOWN/DELLA-UP or sly1-UP regulation. Of these, the putative oxidoreductases At5g07480 (KUOX1) and At2g44800 are MYB-type TF GL1 targets, while putative nuclear transport factor 2 family member At5g01740 is a TF GL3 target based on PlantGSEA. Both GL1 and GL3 are known DELLA-interacting proteins [29], supporting the notion that GID1b-OE may alter expression of these genes by down-regulating DELLA.
Comparison of after-ripening and GID1b-OE differentially regulated transcripts
Since long after-ripening and GID1b-OE both rescue sly1-2 germination without DELLA destruction, we examined whether they were associated with similar transcriptome changes. Many more genes were differentially regulated with after-ripening than with GID1b-OE (Fig 3A, 3B, 3C and 3D). More of the GID1b-OE-regulated transcripts (GIDvsD) showed similar AR-regulation (sly1-2 ARvsD) at the 0h than at the 12h imbibition timepoint. Of the 13 GID1b-OE-regulated genes at 0h, 1 GID1b-OE-UP and 7 GID1b-OE-DOWN genes showed similar AR-regulation (Fig 7). Of the 14 GID1b-OE-regulated genes at 12h, only 1 GID1b-OE-DOWN gene (KUOX1) showed similar AR-regulation (Fig 8). This gene was also down-regulated in 12h WTvsD and in the 24 h Ler ARvsD comparison, suggesting that it is AR-regulated in both wild-type and sly1-2 [6]. Thus, it appears that there is more similarity in early Phase II than as the seeds approach germination per se.
Overlap between sly1-2 AR-regulated genes with GID1b-OE-regulated genes at 0h (0h sly1-2 ARvsD ∩ 0h GIDvsD) is shown in a Venn diagram. The three heatmaps correspond to the three areas of the Venn diagram: the top 50 AR-regulated genes that were not GID1-OE-regulated (light blue), the 5 GID1-OE-regulated genes that were not AR-regulated (orange), and the 8 GID1b-OE-regulated genes that were AR-regulated (brown). Up-regulation is red, down-regulation purple, and significant differences are marked with a black border (FDR of p < 0.05). The ARvsD columns indicate sly1-2 ARvsD and the DELLA/DELLA-regulated columns indicate DELLA-regulation based on the ga1-3 vs ga1-3 4x della comparison from Cao et al. [49] (0h timepoint). Black triangles indicate significant regulation in the expected direction if sly1-2 after-ripening occurs through negative regulation of DELLA.
Overlap of sly1-2 AR-regulated with GID1b-OE-regulated genes at 12h (12h sly1-2 ARvsD ∩ 12h GIDvsD) is shown in a Venn diagram. The Ler ARvsD and Ler columns indicate the 24 h Ler ARvsD comparison from Carrera et al. [6]. The black circle marks GASA4, a 12h GID1b-OE-regulated transcript based on RT-qPCR.
We expected to find agreement between comparisons of lines that germinate to lines that do not germinate in late Phase II, including the 12h GIDvsD, the 12h WTvsD, and both the 12h sly1-2 ARvsD and 24 h Ler ARvsD comparisons (Fig 8). Indeed, all of the 50 most differentially-regulated genes in the 12h sly1-2 ARvsD comparison showed similar differential regulation in the 12h WTvsD, and all but 1 with 24 h Ler ARvsD. In contrast, most of the 12h GID1b-OE-regulated genes did not show similar regulation in WTvsD, or with after-ripening of sly1-2 and Ler. In fact, 7 showed regulation in the opposite direction. For example, HSFA9, GASA2, At2g34740, RAB28, and At5g45690, were GID1b-OE-UP but down-regulated in 12h sly1-2 ARvsD, 12h WTvsD, and 24 h Ler ARvsD. Additionally, GASA4 and AHb1 (Arabidopsis nonsymbiotic Hemoglobin1) were GID1b-OE-DOWN, but up-regulated in 12h WTvsD, and with sly1-2 after-ripening. Thus, many of the 12h GID1b-OE differentially-regulated genes actually showed opposite regulation from other germinating versus non-germinating comparisons. This may suggest a negative feedback response.
Comparison of RT-qPCR and microarray analyses of gene expression
RT-qPCR was conducted to validate microarray results using six genes differentially regulated in multiple comparisons (Fig 9). For comparison, both RT-qPCR and microarray expression were plotted relative to HBT (HOBBIT, At2g20000), a constitutive control gene [60]. There appeared to be a high degree of similarity and a linear correlation between the RT-qPCR results and the microarray results, although some differences were better observed by RT-qPCR than by microarray (S5 Fig; Fig 9). As in Ler ARvsD, GASA2 and HSFA9 were AR-DOWN both in the 12h sly1-2 ARvsD microarray and RT-qPCR analyses. RT-qPCR also confirmed differential regulation of GASA2 and HSFA9 by GID1b-OE at 12h (Fig 9; S6 Fig).
For each gene, relative expression in microarray (blue) and RT-qPCR (green) analyses are shown relative to the same calibrator, set to a height of 1 (100) and indicated by the red dotted-line. RMA-normalized microarray data was analyzed using the ddCT method relative to the same constitutively expressed HBT control gene used for the ddCT analysis of RT-qPCR data. Statistical significance was determined by pairwise t-test with Bonferroni-Holm correction for multiple comparisons (see S6 Fig for p-values). Error bars represent SD.
Based on a previous RT-PCR analysis in sly1-2 seeds imbibed for 24 h following cold stratification, XERICO and GASA4 were expected to be AR-regulated at 12h [21,42]. Indeed, XERICO was significantly AR-DOWN and GASA4 AR-UP in sly1-2 ARvsD at 12h based on microarray and RT-qPCR (Fig 9; S6 Fig). Moreover, GASA4 was significantly AR-UP (logFC = 1.05; p = 1.26 x 10−4) and XERICO significantly AR-DOWN (logFC = -1.32; p = 6.14 x 10−3) in the Ler ARvsD comparison [6]. A previous analysis detected GID1b-OE-down-regulation of XERICO and GASA4 at 24 h of imbibition following cold stratification [42]. Although these genes were not significantly GID1b-OE-regulated in our microarray study based on an FDR adjusted-p-value < 0.05, they both exhibited a downward trend at the 12h timepoint. By RT-qPCR analysis at 12h, XERICO showed a non-significant downward trend, whereas GASA4 was significantly GID1b-OE-DOWN (p = 2.9 x 10−3; Fig 9; S6 Fig). Thus, GASA4 was added to the short list of GID1b-OE-regulated transcripts.
In previous studies, DOG1 (DELAY OF GERMINATION1) expression levels were positively correlated with higher seed dormancy [61,62]. Consistent with this, DOG1 mRNA levels were lower with Ler after-ripening (logFC = -0.91, p = 0.02), and higher in dormant sly1-2 than in WT at 0h (Fig 9; S6 Fig). DOG1 was not down-regulated with sly1-2 after-ripening or with GID1b-OE. RT-qPCR analysis showed that DOG1 mRNA levels were AR-UP in sly1-2 at 0h (logFC = 0.93, p = 0.05). Although DOG1 expression at 12h was very low, a similar expression pattern was observed by RT-qPCR.
The gene with the largest fold increase with after-ripening was SLY1 (logFC: 5.70 (0h), 5.34 (12h); Figs 7 and 8). This was consistent with RT-PCR analysis and previous work (Fig 9; [63]). Note that the 2-bp deletion in the sly1-2 transcript did not interfere with detection by RT-PCR primers or by the 11 SLY1 Affymetrix ATH1 probes. The only gene with a SLY1-like expression pattern was DDB1b, a subunit of the ultraviolet-damaged DNA-binding protein complex (logFC: 3.90 (0h), 3.69 (12h)) [64]. Both SLY1 and DDB1b were more highly expressed in WT and in sly1-2(AR) than in sly1-2(D), suggesting that SLY1 and after-ripening stimulate expression of these genes. Both of these transcripts were DELLA-UP (Fig 7; [49]).
Discussion
The mechanisms governing seed dormancy release remain one of nature’s most interesting mysteries. Genome-wide transcriptional assays in sly1-2 were designed to ask specific questions about DELLA-regulated and DELLA-independent seed dormancy loss. Although sly1-2 cannot degrade DELLA repressors via the ubiquitin-proteasome pathway, non-proteolytic GA signaling allows some rescue of seed dormancy by long after-ripening or by GID1b overexpression [42]. Comparing sly1-2 ARvsD to Ler ARvsD revealed that over half of the transcriptome changes with after-ripening depend upon DELLA destruction (S1B and S1D Fig; Fig 4; [6]). Furthermore, TAGGIT analysis revealed a major loss in abundance of transcripts associated with protein translation in the sly1-2 mutant (Fig 5). The GIDvsD comparison revealed that the rescue of sly1-2 germination can occur with remarkably few changes in gene transcript levels (Figs 3 and 6). Within this small number of changes, after-ripening and GID1b-OE appear to rescue sly1-2 germination by only partly overlapping transcriptional mechanisms (Figs 7 and 8). GID1b-OE differentially-regulated genes overlapped with AR-regulated genes more in early than in late Phase II. Many DELLA-regulated genes were inversely regulated with after-ripening and GID1b-overexpression.
After-ripening resulted in both SLY1 and DELLA-proteolysis dependent and independent transcriptome changes
We know that SLY1 and DELLA destruction play a role in after-ripening because the sly1-2 single gene mutation causes a large change in the imbibing seed transcriptome associated with a drastic shift in the time required for dry after-ripening from 2 weeks to around 2 years. Genes associated with seed dormancy due to DELLA overaccumulation were identified in the DvsWT transcriptome comparison. Over half of these sly1-regulated genes were oppositely regulated with after-ripening of Ler or sly1-2, supporting the role of DELLA in seed dormancy and its release (Fig 2A and 2C; [6]).
Although sly1-2 mutants cannot degrade DELLA protein, GA treatment can cause up-regulation of GA-stimulated gene expression [42]. A similar fraction of late Phase II AR-regulated genes in sly1-2 (35%) or wild-type Ler (33%) were known GA-regulated genes, in spite of the larger total number of transcriptome changes in Ler (S7A and S7B Fig; [49]). Thus, some GA-regulated transcripts do not require SLY1 and associated DELLA destruction to respond to after-ripening.
If after-ripening and GID1b-OE are rescuing sly1-2 seed germination at least in part through DELLA-regulated transcriptional changes, then we might expect to identify known targets of DELLA-interacting or GA-regulated transcription factors in our differentially regulated gene sets. Indeed, enrichment analysis for transcription factor targets within after-ripening-regulated genes in sly1-2 identified: 1) the DELLA-interacting TFs, PIF1/PIL5 and PIF3, 2) the GA- and light-regulated TF, HY5 (Table 1; [58,65,66]). HY5 is active in early Phase II and is a positive regulator of ABA responses during seed germination through ABI5 [67–69]. Among AR-regulated genes, there were also 54 and 127 known targets of the DELLA interacting transcription factors GL1 and GL3, respectively (S4 Table). GL1 and GL3 interact with DELLA to regulate jasmonate (JA) and GA signaling [29]. Given evidence suggesting that JA signaling stimulates dormancy loss and germination in barley and wheat, future work should examine whether GL1 and GL3 mediate crosstalk between GA and JA signaling during seed germination [70,71].
Gene categories associated with after-ripening
Although both dormant and after-ripened seeds enter Phase II, TAGGIT analysis suggests that what prevents dormant seeds from germinating may be AR-regulation of genes associated with early and late Phase II processes at 0h and 12h, respectively (Fig 4A and 4B). Thus, seed dormancy may block germination in part by preventing the accumulation of transcripts involved in Phase II processes. Changes in similar transcript categories were associated with after-ripening of sly1-2 and Ler wt, suggesting that sly1-2 undergoes a “natural” after-ripening process (S1B Fig; [6]).
TAGGIT gene ontology analysis suggested that after-ripening upregulates protein translation, and that this upregulation involves GA signaling. After-ripening resulted in massive up-regulation of protein translation-associated genes in late Phase II in wild-type Ler, but not in sly1-2 (Fig 4). Additionally, comparison of 2 week after-ripened sly1-2 and wild-type showed similarly massive down-regulation of the same category in sly1-2 in both early and late Phase II (Fig 5). Furthermore, our re-analysis of GA- and DELLA-regulated genesets from Cao et al [49] showed that translation-associated genes were the most highly regulated category. This suggests that GA signaling and SLY1 are required for the efficient induction of protein translation-associated genes in germinating seeds. Future work should examine whether the failure of GA biosynthesis and signaling mutant germination is associated with a failure in protein translation. This notion is interesting given that inhibition of protein translation with cyclohexamide fully blocked germination, whereas inhibition of de novo transcription by α-amanitin slowed seed germination [72]. Thus, proteins translated from stored mRNAs or mRNAs transcribed early in seed imbibition (before α-amanitin uptake) appear to be sufficient for germination per se. Investigation of the stored mRNA transcriptomes in dry seeds or of the translatome in early imbibition may shed light on key proteins involved in seed germination.
Reduced sly1 dormancy due to GID1b-OE was associated with few transcriptome changes
Very few transcriptional changes occurred with GID1b-OE in sly1-2 while many changes occurred with sly1-2 after-ripening (Fig 3A, 3B, 3C and 3D). This result indicates that a major change in sly1-2 seed germination potential can occur with few changes in transcript levels. There are three possible explanations: 1) GID1b-OE rescues sly1-2 germination via a non-transcriptional mechanism, 2) GID1b-OE rescue of sly1-2 germination results from transcriptional changes during seed development rather than during seed imbibition, or 3) the 22 GID1b-OE-regulated transcripts are particularly critical regulators of dormancy loss. While future work will need to examine whether loss or gain of GID1b-OE-regulated gene function alters seed germination potential, it is interesting to consider the functional categories of these GID1b-OE-regulated transcripts.
Several of the GID1b-OE-regulated transcripts encode ABA-responsive seed-related regulatory genes (Fig 6). The TF HSFA9, and the LEA (Late Embryogenesis Abundant protein) RAB28 were GID1b-OE-up-regulated. Both play roles in seed desiccation tolerance and are regulated by the seed-specific TF ABI3 of the ABA signaling pathway [73–77]. Consistent with its up-regulation, overexpression of RAB28 increased germination rates under both standard and osmotic stress conditions [78]. HSFA9 is not heat stress-responsive, and is the only one of 21 Arabidopsis HSFs that is solely expressed in seeds starting late in seed development [74]. HSFA9 also regulates seed longevity. Another GID1b-OE-up-regulated transcript, At2g34740, encodes a protein phosphatase type 2C (PP2C). Although not well characterized, At2g34740 was induced by ABA and repressed in the ABA-signaling SNF1-related protein kinase 2 (SnRK2) gene triple mutant, srk2D srk2e srk2i [79]. This suggested a possible role in ABA signaling during seed development or germination [80,81].
Both nitric oxide (NO) and reactive oxygen species (ROS) have been implicated in stimulating seed germination (reviewed in [82]). NO regulates seed oxygen levels during germination, but also acts as a signaling molecule to promote seed germination [83]. NO appears to promote DELLA protein accumulation and represses SLY1 expression in seedlings [84]. The GID1b-OE-down-regulated non-symbiotic hemoglobin AHb1 modulates NO accumulation in seeds and regulates expression of ROS metabolism and ABA signaling genes [85,86]. ROS likely stimulates seed germination through endosperm weakening and promotion of programmed cell death in the aleurone, but ROS accumulation also leads to oxidative damage. In total, 6 of the 15 GID1b-OE-down-regulated genes were implicated in oxidative stress regulation, including GASA4, AHb1, BAG6, CYSa, KUOX1, and the oxidoreductase At2g44800 [86–89]. Oxidation of specific RNAs and protein disulfide bonds may play a functional role in dry after-ripening [9,90–92]. Moreover, germinating seeds need protection from oxidative damage as metabolism is activated (reviewed in [2], [93]).
We also found GA-regulated or signaling genes among the GID1b-OE-regulated transcripts (Fig 6). PTR3 is a GA transporter with homology to peptide transporters, and is likely involved in stress tolerance during germination [94–96]. PTR3 was GID1b-OE- and AR-up-regulated in early Phase II. GA promotes cell elongation by inducing cell wall loosening enzymes such as the GID1b-OE-UP expansin family member, EXPL1 [97–99]. Two members of the GA-regulated GASA gene family, GASA2 and GASA4, were oppositely regulated with GID1b-OE and after-ripening (Fig 9). The fact that GASA4 is sly1-down and AR-upregulated is consistent with the fact that GASA4-overexpression promotes seed germination. Since multiple GASAs block germination, it is possible that GASA2 is sly1-UP and AR-DOWN because it negatively regulates germination [89,100].
One of the fastest GA responses is an increase in cytosolic calcium [101]. The calcium signaling gene BAG6 (a putative calmodulin-binding protein) was GID1b-OE-DOWN, while the calcium-binding EF hand family protein (At1g21630) was one of the most strongly GID1b-OE-UP transcripts (Fig 3). Future work should examine whether GA acts via calcium signaling in the cytosol to stimulate seed germination. This notion is credible because: 1) the rice and Arabidopsis GID1 proteins localize both to the nucleus and to the cytoplasm, and 2) Arabidopsis GID1a was shown to stimulate GA signaling both when altered to localize solely to the nucleus and solely to the cytoplasm [39,102,103].
The idea of a non-transcriptional mechanism of GA signaling is interesting given that gene ontology showed relatively few changes in GA-related transcripts associated with after-ripening of sly1-2, wild-type Ler, and of barley (Fig 4A, 4B and 4C; [6,70]). In the barley aleurone, GA signaling via a calcium-dependent protein kinase in the cytosol occurs independently of nuclear GA signaling [104,105]. The idea that some GID1-mediated GA signaling may not involve direct transcriptional regulation suggests a DELLA-independent mechanism of GA signaling, since DELLA is a nuclear-localized protein believed to function entirely through transcriptional regulation [19,36,106,107]. Future studies should examine if GID1b-OE rescue of sly1 germination requires the ability to physically interact with DELLA protein.
Similarity of GID1b-OE- and AR-regulation in sly1-2 depends on imbibition time
The majority of GID1b-OE-regulated genes showed similar AR-regulation in early Phase II (0h), whereas most of the genes differentially regulated with GID1b-OE showed the opposite regulation with after-ripening in late Phase II (12h) (Figs 7 and 8). This suggests that GID1b mainly contributes to dormancy loss in early Phase II–a crucial time for regulating the transition from quiescent to germinating seed. Some of the changes we are seeing in late Phase II with GID1b-OE may be negative feedback effects rather than the primary transcriptional changes resulting in dormancy loss. The accumulation of GID1b, GID1a, and GID1c proteins increased in response to after-ripening in Arabidopsis [46]. The nine genes showing similar transcriptional changes with GID1b-OE and after-ripening likely represent the subset of AR-regulated genes responding to increased GID1b protein levels with after-ripening (Figs 7 and 8).
If after-ripening and GID1b-OE rescue sly1-2 seed germination by down-regulating DELLA through increased GID1-GA-DELLA complex formation, then we would expect GID1b-OE- and AR-regulated genes to be similarly regulated by GA and inversely by DELLA. While this held true for both sly1-2 and Ler AR-regulated genes, it did not always hold true for GID1b-OE particularly in late Phase II (DELLA-regulation defined by DvsWT or Cao et al. [49]) (Fig 8; [6]). Only 7 of the 22 GID1b-OE differentially-regulated genes were inversely regulated by DELLA (Figs 6, 7 and 8). A large portion of GID1b-OE-regulated genes at 12h were DELLA-regulated in the opposite direction as expected if GID1b-OE-regulation occurs through negative regulation of DELLA (Figs 6 and 8). It was also the case that a larger fraction of AR-regulated genes were inversely DELLA-regulated at the 12h than at the 0h timepoint (Figs 2C, 7 and 8; S1C Fig). Thus, more transcriptional effects from negative regulation of DELLA were apparent later in Phase II.
There are likely many mechanisms that can alleviate seed dormancy
There are many genetic mechanisms contributing to seed dormancy and dormancy release. This study specifically defined DELLA/sly1-dependent and independent aspects of seed dormancy and dormancy release. Previous work also found that there were ABA-dependent and independent transcriptome changes associated with dormancy and after-ripening [8]. Of the genes regulated by after-ripening, 52% are DELLA-regulated and 79% are ABA-regulated (Fig 2B; [6,8]). Clearly, the DELLA- and ABA-regulated aspects of dormancy and after-ripening cannot be mutually exclusive. This is consistent with previous work showing that some GA-regulated genes are also ABA-regulated, and vice versa [14,42]. Moreover, DELLA directly regulates XERICO, a positive regulator of ABA biosynthesis [24]. Transcriptome studies in the highly dormant ecotype Cape Verde Islands have also defined transcriptome changes associated with both primary and secondary seed dormancy [5,7]. The real question is, what are the first proteins or transcripts to initiate the processes needed to germinate, and how is the behavior of these regulators altered by dormancy-breaking processes such as after-ripening? Future work may be able to identify these first events by examining either dry seeds or seeds earlier in imbibition (Phase I).
While some changes in gene expression as seeds transition from early to late Phase II may be involved in activating the germination program, others may prevent germination. There were genes whose transcript levels increased from 0h to 12h of imbibition solely in after-ripened seeds, solely in dormant seeds, and in both dormant and after-ripened sly1-2 seeds (S7C Fig). The fact that there were a substantial number of genes down and up-regulated going from 0h to 12h of imbibition solely in DORMANT seeds, suggests that dormancy is not a passive process where seeds fail to induce genes needed for germination. Instead, it suggests that some genes may actively prevent germination of dormant seeds. Such genes may be preparing the seed to survive re-desiccation or may be inhibiting the activity of genes/proteins needed for germination. The Dekkers et al., [108] study made similar observations and defined dormancy-dependent genes and after-ripening-dependent genes in Cape Verde Islands in different tissues and across multiple imbibition timepoints.
Materials and methods
Plant materials and growth conditions
Arabidopsis thaliana ecotype Landsberg erecta (Ler) wild-type, sly1-2 and sly1-2 GID1b-OE in the Ler background were previously described [34,41]. The GID1b-OE allele is an HA:GID1b translational fusion driven by the strong cauliflower mosaic virus 35S promoter in the sly1-2 background. Arabidopsis used for all experiments were cultivated side-by-side in a Conviron® growth chamber according to McGinnis et al. [35]. After harvest, seeds were stored in open tubes for 2 wk at room temperature and low humidity (≈15–30%) to allow dry after-ripening. Note that 2-wk-old seed means that seeds were stored under dry after-ripening conditions for 2 wk. Thereafter, seeds were stored in closed tubes at -20°C to preserve dormancy until plated. The long after-ripened sly1-2(AR) sample was a separate sly1-2 batch grown in advance under the same conditions and stored at room temperature for 19 mo.
Germination experiments
For the germination screen (Fig 1A), 80 to 100 seeds for each of the four seed batches were sterilized with 70% ethanol and 0.01% SDS for 5 minutes followed by 10% bleach and 0.01% SDS for a further 10 minutes, washed and plated on MS-agar plates containing 0.8% agar, 0.5× MS salts (Sigma-Aldrich), and 5 mM MES (2-(N-morpholino)ethanesulfonic acid), pH 5.5. The germination of the seed batches used for microarray analysis were scored daily for 8 d during incubation in the light at 22°C following cold stratification for 4 d at 4°C. Germination and testa rupture were also scored at the 12 h imbibition timepoint used for microarray analysis.
Total RNA isolation from seeds
For RNA extraction, 20 mg of dry seed per sample was sown on filter paper moistened with sterile 0.5× MS salts buffered with 5mM MES, pH 5.5. Seeds were collected at two timepoints: 1) a 0h timepoint harvested immediately after cold stratification for 4 d at 4°C in the dark, and 2) a 12h timepoint incubated 12h in the light at 22°C following the 4 d cold stratification at 4°C. RNA isolation was performed using a phenol-chloroform extraction developed for tough tissues such as Arabidopsis seeds based on Oñate-Sánchez and Vicente-Carbajosa [109] with an additional chloroform extraction and the use of Phase Lock (5-PRIME) gel tubes to allow complete recovery of the aqueous phase without risk of organic phase contamination. RNA concentration was determined using a NanoDrop ND-2000c spectrophotometer (Thermo Scientific), and RNA quality checked by gel electrophoresis using RNA denatured at 70°C for 5 minutes in a formaldehyde dye mixture and run on a standard 1.1% agarose gel.
Microarray and data analysis
Affymetrix ATH1 chip (22,810 genes) oligonucleotide-based DNA microarray analysis was performed using three biological replicates at the 0h and 12h imbibition timepoints, for each genotype: Ler wt (2 wk old), dormant sly1-2 (2 wk old), after-ripened sly1-2 (19 mo old), and sly1-2 GID1b-OE (2 wk old). For each biological replicate, 2 μg of RNA was submitted to the Molecular Biology and Genomics Core Laboratory at Washington State University for synthesis of biotin-labeled cRNA, hybridization to ATH1 Arabidopsis microarray chips (Affymetrix), and chip scanning (http://crb.wsu.edu/core-laboratories/molecular-biology-and-genomics-core). Data analysis of raw CEL files was performed using the LIMMA package as part of the Bioconductor suite of tools in the R statistical programming environment [110–112]. Microarray raw data files are available at ArrayExpress (http://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-4782 [113]. Background correction and normalization was performed by Robust Multi-array Average (RMA), control probesets were removed, and significance was determined by False Discovery Rate (FDR) with α = 0.05 [114,115]. For genome-wide expression plots, the adjusted Fisher-Pearson standardized moment coefficient (G1) measurement of skew was calculated using the equation: G1 = g1 * sqrt(n(n-1)) / (n-2) [50].
Raw datasets from Cao et al. [49] and Carrera et al. [6] were obtained from ArrayExpress and NASCarrays (http://affymetrix.arabidopsis.info), respectively, and reanalyzed by the same methods to allow fair comparisons to the current study. In the Cao et al. [49] study, seeds were dry after-ripened for approximately 1 mo, then imbibed for 4 d at 4°C before harvesting for RNA isolation. In the Carrera et al. [6] study, the dormant Ler were freshly harvested from yellow siliques, and the after-ripened Ler were stored dry for 2 mo. Note that the freshly harvested Ler were dormant, whereas the 2 week after-ripened Ler wt seeds in Fig 1 were non-dormant and germinated efficiently.
Microarray analysis measures relative steady-state transcript levels, not active gene transcription. Thus, differential regulation could result from differences at the level of gene transcription or at the level of transcript stability. References to differential regulation or expression indicate differential regulation, but do not necessarily indicate differences in gene transcription. When referring to the differential regulation in A relative to B, or AvsB, up in AvsB means up-regulated in A (or down-regulated in B), whereas down in AvsB means down-regulated in A (or up-regulated in B).
In the GIDvsD comparison, very few differentially-regulated genes were identified considering the very different germination phenotypes. To ensure that the very high expression of GID1b was not causing artifactually low significance for other genes as a result of the multiple comparisons adjustment, significance testing was repeated with GID1b removed. This reanalysis of the data did not result in any change to the number of differentially-regulated genes in GIDvsD comparison at either imbibition timepoint. To determine whether differences in signal intensity might account for differences in the number of differentially-regulated genes, we also graphed signal intensities for sly1-2 ARvsD and GIDvsD at both imbibition timepoints (S9 Fig). It was clear that the significant differences were observed over a wide range of signal intensities, allowing confidence that these differences were not artifacts due to low signal intensity.
MicroarrayTools R package
R software tools were written for this study and are available through github in the microarrayTools package (https://github.com/bakuhatsu/microarrayTools). The microarrayTools package includes four functions: getProbeID, venndia, TAGGITontology, and TAGGITplot. getProbeID can be used to convert a gene or list of genes from AGI identifiers (e.g. At4g24210) or short gene names (e.g. SLY1) to ATH1 Affymetrix probe ids. Overlaps between comparisons were identified using the venndia function (Fig 2; S1 and S7 Figs). This function allows comparison of up- and down-regulated genes simultaneously and can be nested to simplify complex comparisons (overlaps of overlaps) (S1D Fig). In addition to plotting, overlapping genes can be returned as lists of up- and down-regulated genes for further analysis. For seed-specific gene ontology classification of differentially-regulated genes, an R pipeline was developed similar to the TAGGIT pipeline originally developed as an Excel macro by Carrera et al. [6]. The TAGGITontology function takes a list of significant genes and outputs a data frame for plotting and and an excel file containing hits. For searching gene descriptions TAGGITontology calls specialized functions written in C++ via the Rcpp package to greatly improve search speeds over pure R code [116]. The TAGGITplot function, which utilizes the ggplot2 package, automates plotting bar charts such as those published in Fig 4, S2 Fig and S3 Fig [117]. An example script is shown in S9 Fig.
TAGGIT gene ontology analysis in R
Unlike standard Gene Ontology (GO), which identifies gene categories based on enrichment within a set of differentially regulated genes, TAGGIT quantifies enrichment for genes in a set of 26 specific categories defined for their involvement in seed dormancy and germination [6]. Genes were categorized a) by using the provided list of search terms for each category and comparing to the current gene descriptions, and b) by comparison to AGI identifiers from the original TAGGIT list (S2 Fig; S5 Table; [6]). Since gene annotation information, including descriptions, was pulled live from the ath1121501.db online database (Bioconductor) it was possible to identify TAGGIT genes that were not present in the original TAGGIT analysis [6,110]. This list of gene descriptions can be updated at any time by re-defining the myAnnot object. Using the TAGGITontology and TAGGITplot functions, it is possible to plot genes enriched for in the up-regulated fraction as red bars, and in the down-regulated fraction as blue bars. For simplicity, a higher degree of enrichment in the up-regulated fraction of a certain category was referred to as “more up-regulation” in this category, while a higher degree in the down-regulated fraction as “more down-regulation” in this category.
Transcription factor target gene enrichment analysis
The web-based PlantGSEA toolkit (http://structuralbiology.cau.edu.cn/PlantGSEA/) with the Transcription Factor Targets (TFT) dataset identified known gene targets of transcription factors within differentially regulated genesets [51–53]. PlantGSEA identifies transcription factor targets based on published results of in vivo DNA-binding techniques such as ChIP coupled with DNA sequencing (ChIP-seq) and ChIP followed by microarray hybridization (ChIP-chip). The TFT dataset includes both targets of transcription factors and of transcriptional regulators that interact with a DNA-binding proteins, both referred to as TFs in this work. TFTs are considered “confirmed” if their function was proven by two or more approaches including in vivo evidence. Significantly enriched targets are presented in Table 1, the “All” category includes both “confirmed” and “unconfirmed” TFTs. Complete tables of TFT hits are available in S4 Table. Enrichment relative to the whole transcriptome was determined using a Fisher statistical test with the Yekutieli (FDR under dependency) correction for multiple testing adjustment with α = 0.06 [118,119].
RT-qPCR analysis for microarray validation
RT-qPCR analysis for microarray validation was performed using gene-specific primers for select differentially regulated genes: GASA2, GASA4, HSFA9, DOG1, XERICO, and SLY1 (Fig 9; S6 Fig). cDNA for RT-qPCR was synthesized from 1 μg of total RNA using the ProScript® M-MuLV First Strand cDNA synthesis (New England Biolabs), and RT-qPCR analysis was conducted using the Lightcycler LightCycler® FastStart DNA Master SYBR Green I kit (Roche). Primers were designed using the QuantPrime online tool (http://www.quantprime.de), with the exception of the previously published HSFA9 and DOG1 primers (S10 Fig; [120–122]). Reaction efficiencies were calculated based on dilution curves, all primer pair efficiencies were within 10% of each other and ±10% of 100% efficiency. Conditions for RT-qPCR were as follows: 10 minutes at 95°C (initial denature), followed by 45 cycles of 10 s at 95°C (denaturation), 5 s at the primer-specific annealing temperature (see S10 Fig), and 10 s at 72°C (extension). Changes in mRNA transcript abundance were quantified for the three biological replicates per gene using the Delta-Delta Ct method [123] and calculated relative to the constitutive HOBBIT (HBT; At2g20000) control [60]. The HBT control was selected based on the Graeber et al. [60] comparison of 12 Arabidopsis control genes in germinating seeds, where it was demonstrated to have the highest expression stability (lowest intersample transcript abundance variation) of the genes tested. Statistical testing was performed by pairwise Student’s t-test with Bonferroni-Holm correction for multiple comparisons with α = 0.07 (S6 Fig; [124]).
Supporting information
S1 Fig. Venn diagrams of differentially-regulated genesets to identify overlaps.
(A) Overlap between genes that are DELLA/sly1-regulated at 0h (based on DvsWT) with those that are oppositely GA-regulated at 0h in Cao et al. [49] (0h DvsWT ∩ 0h ga1-3 vs WT). (B) Overlap between genes differentially regulated with after-ripening of sly1-2 at 12h with those of Ler at 24 h (12h sly1-2 ARvsD ∩ 24 h Ler ARvsD). (C) Overlap between genes inversely regulated by the sly1 mutation at 0h with genes differentially regulated with sly1-2 after-ripening at 0h (0h WTvsD ∩ 0h sly1-2 ARvsD). (D) Overlap of 12h sly1-2 and 24 h Ler after-ripening-regulated genes with those inversely regulated by the sly1 mutation at 12h. The sly1-regulated genes are likely also DELLA-regulated. Throughout this study, up- (red) and down-regulated (blue) genes are based on FDR cutoff of p < 0.05.
https://doi.org/10.1371/journal.pone.0179143.s001
(PDF)
S2 Fig. Diagram of pipeline for TAGGIT gene ontology analysis with the TAGGITontology and TAGGITplot R functions.
https://doi.org/10.1371/journal.pone.0179143.s002
(PDF)
S3 Fig. TAGGIT gene ontology analysis of GA- and DELLA-regulated transcriptome changes.
Using reanalyzed datasets from Cao et al. [49] with timepoint equivalent to 0h in early Phase II (imbibed for 4 d at 4°C). (A) GA-regulated genes determined by the Ler wt vs ga1-3 comparison, (B) DELLA-regulated genes determined by the ga1-3 vs ga1-3 4x della comparison, and (C) total GA-regulated and total DELLA-regulated genesets compared. The value on the x-axis shows the percentage of the total differentially regulated genes within a dataset.
https://doi.org/10.1371/journal.pone.0179143.s003
(PDF)
S4 Fig. Table with logFCs for differentially regulated genes with rescue of sly1-2 germination by GID1b-overexpression.
https://doi.org/10.1371/journal.pone.0179143.s004
(PDF)
S5 Fig. Plots of the correlation between RT-qPCR and microarray data for each comparison at 0h and 12h timepoints.
https://doi.org/10.1371/journal.pone.0179143.s005
(PDF)
S6 Fig. Tables of p-values for pairwise comparisons of RT-qPCR data.
Significant values are indicated in black text. Significance is based on pairwise t-tests with Bonferroni-Holm correction for multiple comparisons with α = 0.07.
https://doi.org/10.1371/journal.pone.0179143.s006
(PDF)
S7 Fig. Venn diagrams of differentially-regulated genesets to identify overlaps.
(A) Overlap between genes differentially regulated with after-ripening of sly1-2 at 12h with those that are GA-regulated based on the Ler wt vs ga1-3 comparison from Cao et al. [49] (12h sly1-2 ARvsD ∩ 0h WT vs ga1-3). (B) Overlap between genes differentially regulated with after-ripening of Ler wt at 24 h (Carrera et al. [6]) with those that are GA-regulated based on the Ler wt vs ga1-3 comparison (24 h Ler ARvsD ∩ 0h WT vs ga1-3). (C) Overlap between genes differentially regulated in sly1-2(D) from 0h to 12h of imbibition with those of sly1-2(AR) from 0h to 12h of imbibition (sly1-2(AR) 12hvs0h ∩ sly1-2(D) 12hvs0h). Significance based on an FDR cutoff of p < 0.05.
https://doi.org/10.1371/journal.pone.0179143.s007
(PDF)
S8 Fig. Plots comparing intensities of microarray data after RMA normalization.
Red circles indicate genes that were significantly differentially regulated (FDR, p < 0.05). Black circles were plotted with transparency of 25% such that darker areas indicate a larger accumulation of dots at the same location. Normalized probe intensities (log2 scale) were compared. A, 0h sly1-2(AR) (y-axis) versus 0h sly1-2(D) (x-axis). B, 12h sly1-2(AR) (y-axis) versus 12h sly1-2(D) (x-axis). C, 0h sly1-2 GID1b-OE (y-axis) versus 0h sly1-2(D) (x-axis). D, 12h sly1-2 GID1b-OE (y-axis) versus 12h sly1-2(D) (x-axis).
https://doi.org/10.1371/journal.pone.0179143.s008
(PDF)
S9 Fig. Example code for analyzing a dataset with TAGGITontology and plotting it with TAGGITplot.
https://doi.org/10.1371/journal.pone.0179143.s009
(PDF)
S10 Fig. Table of primers used for RT-qPCR with primer sequences and annealing temperatures used in this study.
https://doi.org/10.1371/journal.pone.0179143.s010
(PDF)
S1 Table. Complete tables of significant differences 1) in the DvsWT comparison, 2) with sly1-2 after-ripening, and 3) with GID1b-overexpression, for both the 0h and 12h timepoints.
https://doi.org/10.1371/journal.pone.0179143.s011
(XLSX)
S2 Table. Table of 1) DELLA-regulated genes that change with after-ripening, which likely involve DELLA-proteolysis, and 2) DELLA-regulated genes which likely do not involve DELLA-proteolysis.
https://doi.org/10.1371/journal.pone.0179143.s012
(XLSX)
S3 Table. Table of translation associated genes identified by TAGGIT analysis.
Tables represent enrichment in 1) the down-regulated fraction of 0h DvsWT, 2) the downregulated fraction of 12h DvsWT, and 3) up-regulated fraction of 24 h Ler ARvsD (Carrera et al., [6]).
https://doi.org/10.1371/journal.pone.0179143.s013
(XLSX)
S4 Table. Complete table of PlantGSEA enrichment analysis.
Transcription factor targets differentially regulated with sly1-2 after-ripening at 0h and 12h.
https://doi.org/10.1371/journal.pone.0179143.s014
(XLSX)
S5 Table. TAGGITguideAGIs and TAGGITguideSearchTerms tables.
Used by the TAGGITontology R function to classify genes into TAGGIT gene ontology categories based on lists of AGI locus identifiers, and lists of search terms corresponding to each category, respectively.
https://doi.org/10.1371/journal.pone.0179143.s015
(XLSX)
Acknowledgments
We thank the members of the Steber lab, A. Hauvermale, S. Martinez, K. Tuttle, and T. Harris, for helpful suggestions about the research and manuscript. Thanks are also due to M. Neff, H. Hellmann, A. McCubbin, and P. Okubara for helpful comments on the manuscript. This research was funded by National Science Foundation (NSF) Award 0850981 and USDA-ARS project 424575 (to C.M.S.).
Author Contributions
- Conceptualization: SKN CMS.
- Data curation: SKN.
- Formal analysis: SKN.
- Funding acquisition: CMS.
- Investigation: SKN.
- Methodology: SKN CMS.
- Project administration: SKN CMS.
- Resources: SKN CMS.
- Software: SKN.
- Supervision: CMS.
- Validation: SKN.
- Visualization: SKN CMS.
- Writing – original draft: SKN CMS.
- Writing – review & editing: SKN CMS.
References
- 1.
Corner EJH. The life of plants. Cleveland: World Pub. Co; 1964.
- 2.
Bewley JD, Bradford KJ, Hilhorst HWM, Nonogaki H. Seeds: physiology of development, germination and dormancy. New York, NY: Springer; 2013.
- 3. Finch-Savage WE, Leubner-Metzger G. Seed dormancy and the control of germination. New Phytol. 2006;171: 501–523. pmid:16866955
- 4. Finkelstein R, Reeves W, Ariizumi T, Steber CM. Molecular aspects of seed dormancy. Annu Rev Plant Biol. 2008;59: 387–415. pmid:18257711
- 5. Cadman CS, Toorop PE, Hilhorst HW, Finch-Savage WE. Gene expression profiles of Arabidopsis Cvi seeds during dormancy cycling indicate a common underlying dormancy control mechanism. Plant J. 2006;46: 805–822. pmid:16709196
- 6. Carrera E, Holman T, Medhurst A, Peer W, Schmuths H, Footitt S, et al. Gene Expression Profiling Reveals Defined Functions of the ATP-Binding Cassette Transporter COMATOSE Late in Phase II of Germination. Plant Physiol. 2007;143: 1669–1679. pmid:17322332
- 7. Finch-Savage WE, Cadman CSC, Toorop PE, Lynn JR, Hilhorst HWM. Seed dormancy release in Arabidopsis Cvi by dry after-ripening, low temperature, nitrate and light shows common quantitative patterns of gene expression directed by environmentally specific sensing. Plant J. 2007;51: 60–78. pmid:17461781
- 8. Carrera E, Holman T, Medhurst A, Dietrich D, Footitt S, Theodoulou FL, et al. Seed after-ripening is a discrete developmental pathway associated with specific gene networks in Arabidopsis. Plant J. 2008;53: 214–224. pmid:18028281
- 9. Bazin J, Langlade N, Vincourt P, Arribat S, Balzergue S, El-Maarouf-Bouteau H, et al. Targeted mRNA oxidation regulates sunflower seed dormancy alleviation during dry after-ripening. The Plant Cell. 2011;23: 2196–2208. pmid:21642546
- 10. Gao F, Jordan MC, Ayele BT. Transcriptional programs regulating seed dormancy and its release by after-ripening in common wheat (Triticum aestivum L.). Plant Biotechnol J. 2012;10: 465–476. pmid:22292455
- 11. Penfield S, Springthorpe V. Understanding chilling responses in Arabidopsis seeds and their contribution to life history. Philos Trans R Soc Lond, B, Biol Sci. 2012;367: 291–297. pmid:22144391
- 12. Ogawa M, Hanada A, Yamauchi Y, Kuwahara A, Kamiya Y, Yamaguchi S. Gibberellin biosynthesis and response during Arabidopsis seed germination. The Plant Cell. 2003;15: 1591–1604. Available: http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&id=12837949&retmode=ref&cmd=prlinks pmid:12837949
- 13. Yamauchi Y, Ogawa M, Kuwahara A, Hanada A, Kamiya Y, Yamaguchi S. Activation of gibberellin biosynthesis and response pathways by low temperature during imbibition of Arabidopsis thaliana seeds. The Plant Cell. 2004;16: 367–378. pmid:14729916
- 14. Sun T-P, Gubler F. Molecular mechanism of gibberellin signaling in plants. Annu Rev Plant Biol. 2004;55: 197–223. pmid:15377219
- 15. Penfield S, Gilday AD, Halliday KJ, Graham IA. DELLA-mediated cotyledon expansion breaks coat-imposed seed dormancy. Curr Biol. 2006;16: 2366–2370. pmid:17141619
- 16.
Nelson SK, Steber CM. Gibberellin hormone signal perception: down-regulating DELLA repressors of plant growth and development. Annual Plant Reviews, Volume 49. Chichester, UK: John Wiley & Sons, Ltd; 2016. pp. 153–188. https://doi.org/10.1002/9781119210436.ch6
- 17. Koornneef M, van der Veen JH. Induction and analysis of gibberellin sensitive mutants in Arabidopsis thaliana (L.) heynh. Theor Appl Genet. 1980;58: 257–263. pmid:24301503
- 18. Telfer A, Bollman KM, Poethig RS. Phase change and the regulation of trichome distribution in Arabidopsis thaliana. Development. 1997;124: 645–654. pmid:9043079
- 19. Silverstone AL, Jung HS, Dill A, Kawaide H, Kamiya Y, Sun TP. Repressing a repressor: gibberellin-induced rapid reduction of the RGA protein in Arabidopsis. The Plant Cell. 2001;13: 1555–1566. pmid:11449051
- 20. Tyler L, Thomas SG, Hu J, Dill A, Alonso JM, Ecker JR, et al. DELLA proteins and gibberellin-regulated seed germination and floral development in Arabidopsis. Plant Physiol. 2004;135: 1008–1019. pmid:15173565
- 21. Ariizumi T, Steber CM. Seed germination of GA-insensitive sleepy1 mutants does not require RGL2 protein disappearance in Arabidopsis. The Plant Cell. 2007;19: 791–804. pmid:17384169
- 22. Hauvermale AL, Ariizumi T, Steber CM. Gibberellin signaling: a theme and variations on DELLA repression. Plant Physiol. 2012;160: 83–92. pmid:22843665
- 23. Cao D, Hussain A, Cheng H, Peng J. Loss of function of four DELLA genes leads to light- and gibberellin-independent seed germination in Arabidopsis. Planta. 2005;223: 105–113. pmid:16034591
- 24. Zentella R, Zhang Z-L, Park M, Thomas SG, Endo A, Murase K, et al. Global analysis of DELLA direct targets in early gibberellin signaling in Arabidopsis. The Plant Cell. 2007;19: 3037–3057. pmid:17933900
- 25. Bai M-Y, Shang J-X, Oh E, Fan M, Bai Y, Zentella R, et al. Brassinosteroid, gibberellin and phytochrome impinge on a common transcription module in Arabidopsis. Nat Cell Biol. 2012;14: 810–817. pmid:22820377
- 26. Hirano K, Kouketu E, Katoh H, Aya K, Ueguchi-Tanaka M, Matsuoka M. The suppressive function of the rice DELLA protein SLR1 is dependent on its transcriptional activation activity. Plant J. 2012;71: 443–453. pmid:22429711
- 27. Zhiponova MK, Morohashi K, Vanhoutte I, Machemer-Noonan K, Revalska M, Van Montagu M, et al. Helix-loop-helix/basic helix-loop-helix transcription factor network represses cell elongation in Arabidopsis through an apparent incoherent feed-forward loop. 2014;111: 2824–2829. pmid:24505057
- 28. Marín-de la Rosa N, Pfeiffer A, Hill K, Locascio A, Bhalerao RP, Miskolczi P, et al. Genome Wide Binding Site Analysis Reveals Transcriptional Coactivation of Cytokinin-Responsive Genes by DELLA Proteins. Yu H, editor. PLoS Genet. Public Library of Science; 2015;11: e1005337. pmid:26134422
- 29. Qi T, Huang H, Wu D, Yan J, Qi Y, Song S, et al. Arabidopsis DELLA and JAZ Proteins Bind the WD-Repeat/bHLH/MYB Complex to Modulate Gibberellin and Jasmonate Signaling Synergy. The Plant Cell Online. 2014;26: 1118–1133.
- 30. Murase K, Hirano Y, Sun T- P, Hakoshima T. Gibberellin-induced DELLA recognition by the gibberellin receptor GID1. 2008;456: 459–463. pmid:19037309
- 31. Shimada A, Ueguchi-Tanaka M, Nakatsu T, Nakajima M, Naoe Y, Ohmiya H, et al. Structural basis for gibberellin recognition by its receptor GID1. Nature Publishing Group; 2008;456: 520–523. pmid:19037316
- 32. Ueguchi-Tanaka M, Hirano K, Hasegawa Y, Kitano H, Matsuoka M. Release of the repressive activity of rice DELLA protein SLR1 by gibberellin does not require SLR1 degradation in the gid2 mutant. The Plant Cell. 2008;20: 2437–2446. pmid:18827181
- 33. Hirano K, Asano K, Tsuji H, Kawamura M, Mori H, Kitano H, et al. Characterization of the molecular mechanism underlying gibberellin perception complex formation in rice. The Plant Cell. 2010;22: 2680–2696. pmid:20716699
- 34. Steber CM, Cooney SE, McCourt P. Isolation of the GA-response mutant sly1 as a suppressor of ABI1-1 in Arabidopsis thaliana. Genetics. 1998;149: 509–521. pmid:9611170
- 35. McGinnis KM, Thomas SG, Soule JD, Strader LC, Zale JM, Sun T-P, et al. The Arabidopsis SLEEPY1 gene encodes a putative F-box subunit of an SCF E3 ubiquitin ligase. The Plant Cell. 2003;15: 1120–1130. pmid:12724538
- 36. Dill A, Thomas SG, Hu J, Steber CM, Sun T-P. The Arabidopsis F-box protein SLEEPY1 targets gibberellin signaling repressors for gibberellin-induced degradation. The Plant Cell. 2004;16: 1392–1405. pmid:15155881
- 37. Ariizumi T, Lawrence PK, Steber CM. The role of two f-box proteins, SLEEPY1 and SNEEZY, in Arabidopsis gibberellin signaling. Plant Physiol. 2011;155: 765–775. pmid:21163960
- 38. Ariizumi T, Steber CM. Mutations in the F-box gene SNEEZY result in decreased Arabidopsis GA signaling. Plant Signal Behav. 2011;6: 831–833. pmid:21455025
- 39. Willige BC, Ghosh S, Nill C, Zourelidou M, Dohmann EMN, Maier A, et al. The DELLA domain of GA INSENSITIVE mediates the interaction with the GA INSENSITIVE DWARF1A gibberellin receptor of Arabidopsis. The Plant Cell. 2007;19: 1209–1220. pmid:17416730
- 40. Iuchi S, Suzuki H, Kim Y-C, Iuchi A, Kuromori T, Ueguchi-Tanaka M, et al. Multiple loss-of-function of Arabidopsis gibberellin receptor AtGID1s completely shuts down a gibberellin signal. Plant J. 2007;50: 958–966. pmid:17521411
- 41. Ariizumi T, Murase K, Sun T-P, Steber CM. Proteolysis-independent downregulation of DELLA repression in Arabidopsis by the gibberellin receptor GIBBERELLIN INSENSITIVE DWARF1. The Plant Cell. 2008;20: 2447–2459. pmid:18827182
- 42. Ariizumi T, Hauvermale AL, Nelson SK, Hanada A, Yamaguchi S, Steber CM. Lifting della repression of Arabidopsis seed germination by nonproteolytic gibberellin signaling. Plant Physiol. 2013;162: 2125–2139. pmid:23818171
- 43. Griffiths J, Murase K, Rieu I, Zentella R, Zhang Z-L, Powers SJ, et al. Genetic characterization and functional analysis of the GID1 gibberellin receptors in Arabidopsis. The Plant Cell. 2006;18: 3399–3414. pmid:17194763
- 44. Nakajima M, Shimada A, Takashi Y, Kim Y-C, Park S-H, Ueguchi-Tanaka M, et al. Identification and characterization of Arabidopsis gibberellin receptors. Plant J. 2006;46: 880–889. pmid:16709201
- 45. Yamamoto Y, Hirai T, Yamamoto E, Kawamura M, Sato T, Kitano H, et al. A Rice gid1 Suppressor Mutant Reveals That Gibberellin Is Not Always Required for Interaction between Its Receptor, GID1, and DELLA Proteins. The Plant Cell. 2010;22: 3589–3602. pmid:21098733
- 46. Hauvermale AL, Tuttle KM, Takebayashi Y, Seo M, Steber CM. Loss of Arabidopsis thaliana Seed Dormancy is Associated with Increased Accumulation of the GID1 GA Hormone Receptors. Plant Cell Physiol. Oxford University Press; 2015;56: 1773–1785. pmid:26136598
- 47. Nakabayashi K, Okamoto M, Koshiba T, Kamiya Y, Nambara E. Genome-wide profiling of stored mRNA in Arabidopsis thaliana seed germination: epigenetic and genetic regulation of transcription in seed. Plant J. 2005;41: 697–709. pmid:15703057
- 48. Preston J, Tatematsu K, Kanno Y, Hobo T, Kimura M, Jikumaru Y, et al. Temporal expression patterns of hormone metabolism genes during imbibition of Arabidopsis thaliana seeds: a comparative study on dormant and non-dormant accessions. Plant Cell Physiol. 2009;50: 1786–1800. pmid:19713425
- 49. Cao D, Cheng H, Wu W, Soo HM, Peng J. Gibberellin mobilizes distinct DELLA-dependent transcriptomes to regulate seed germination and floral development in Arabidopsis. Plant Physiol. 2006;142: 509–525. pmid:16920880
- 50. Joanes DN, Gill CA. Comparing Measures of Sample Skewness and Kurtosis. Journal of the Royal Statistical Society Series D (The Statistician). Wiley for the Royal Statistical Society; 1998;47: 183–189.
- 51. Yilmaz A, Mejia-Guerra MK, Kurz K, Liang X, Welch L, Grotewold E. AGRIS: the Arabidopsis Gene Regulatory Information Server, an update. Nucleic Acids Res. 2010;39: D1118–D1122. pmid:21059685
- 52. Lai L, Liberzon A, Hennessey J, Jiang G, Qi J, Mesirov JP, et al. AraPath: a knowledgebase for pathway analysis in Arabidopsis. Bioinformatics. Oxford University Press; 2012;28: 2291–2292. pmid:22760305
- 53. Yi X, Du Z, Su Z. PlantGSEA: a gene set enrichment analysis toolkit for plant community. Nucleic Acids Res. 2013;41: W98–103. pmid:23632162
- 54. De Veylder L, Beeckman T, Beemster GTS, de Almeida Engler J, Ormenese S, Maes S, et al. Control of proliferation, endoreduplication and differentiation by the Arabidopsis E2Fa-DPa transcription factor. EMBO J. 2002;21: 1360–1368. pmid:11889041
- 55. Vlieghe K, Vuylsteke M, Florquin K, Rombauts S, Maes S, Ormenese S, et al. Microarray analysis of E2Fa-DPa-overexpressing plants uncovers a cross-talking genetic network between DNA replication and nitrogen assimilation. J Cell Sci. 2003;116: 4249–4259. pmid:12953064
- 56. Oh E, Kim J, Park E, Kim J-I, Kang C, Choi G. PIL5, a phytochrome-interacting basic helix-loop-helix protein, is a key negative regulator of seed germination in Arabidopsis thaliana. The Plant Cell. 2004;16: 3045–3058. pmid:15486102
- 57. Oh E, Yamaguchi S, Kamiya Y, Bae G, Chung W-I, Choi G. Light activates the degradation of PIL5 protein to promote seed germination through gibberellin in Arabidopsis. Plant J. 2006;47: 124–139. pmid:16740147
- 58. Alabadí D, Gallego-Bartolomé J, Orlando L, García-Cárcel L, Rubio V, Martínez C, et al. Gibberellins modulate light signaling pathways to prevent Arabidopsis seedling de-etiolation in darkness. Plant J. 2008;53: 324–335. pmid:18053005
- 59. Jofuku KD, Boer den BG, Van Montagu M, Okamuro JK. Control of Arabidopsis flower and seed development by the homeotic gene APETALA2. The Plant Cell. 1994;6: 1211–1225. pmid:7919989
- 60. Graeber K, Linkies A, Wood ATA, Leubner-Metzger G. A guideline to family-wide comparative state-of-the-art quantitative RT-PCR analysis exemplified with a Brassicaceae cross-species seed germination case study. The Plant Cell. 2011;23: 2045–2063. pmid:21666000
- 61. Bentsink L, Jowett J, Hanhart CJ, Koornneef M. Cloning of DOG1, a quantitative trait locus controlling seed dormancy in Arabidopsis. Proc Natl Acad Sci USA. 2006;103: 17042–17047. pmid:17065317
- 62. Chiang GCK, Bartsch M, Barua D, Nakabayashi K, Debieu M, Kronholm I, et al. DOG1 expression is predicted by the seed-maturation environment and contributes to geographical variation in germination in Arabidopsis thaliana. Mol Ecol. 2011;20: 3336–3349. pmid:21740475
- 63. Lee KP, Piskurewicz U, Turecková V, Strnad M, Lopez-Molina L. A seed coat bedding assay shows that RGL2-dependent release of abscisic acid by the endosperm controls embryo growth in Arabidopsis dormant seeds. 2010;107: 19108–19113. pmid:20956298
- 64. Bernhardt A, Mooney S, Hellmann H. Arabidopsis DDB1a and DDB1b are critical for embryo development. Planta. 2010;232: 555–566. pmid:20499085
- 65. Feng S, Martinez C, Gusmaroli G, Wang Y, Zhou J, Wang F, et al. Coordinated regulation of Arabidopsis thaliana development by light and gibberellins. 2008;451: 475–479. pmid:18216856
- 66. Gallego-Bartolomé J, Minguet EG, Marín JA, Prat S, Blázquez MA, Alabadí D. Transcriptional diversification and functional conservation between DELLA proteins in Arabidopsis. Mol Biol Evol. 2010;27: 1247–1256. pmid:20093430
- 67. Chen H, Xiong L. Role of HY5 in abscisic acid response in seeds and seedlings. Plant Signal Behav. 2008;3: 986–988. pmid:19704428
- 68. Lau OS, Deng XW. Plant hormone signaling lightens up: integrators of light and hormones. Curr Opin Plant Biol. 2010;13: 571–577. pmid:20739215
- 69. Toh S, Mccourt P, Tsuchiya Y. HY5 is involved in strigolactone-dependent seed germination in Arabidopsis. Plant Signal Behav. 2012;7: 556–558. pmid:22516816
- 70. Barrero JM, Talbot MJ, White RG, Jacobsen JV, Gubler F. Anatomical and transcriptomic studies of the coleorhiza reveal the importance of this tissue in regulating dormancy in barley. Plant Physiol. 2009;150: 1006–1021. pmid:19386806
- 71. Jacobsen JV, Barrero JM, Hughes T, Julkowska M, Taylor JM, Xu Q, et al. Roles for blue light, jasmonate and nitric oxide in the regulation of dormancy and germination in wheat grain (Triticum aestivum L.). Planta. 2013;238: 121–138. pmid:23588419
- 72. Rajjou L, Gallardo K, Debeaujon I, Vandekerckhove J, Job C, Job D. The effect of alpha-amanitin on the Arabidopsis seed proteome highlights the distinct roles of stored and neosynthesized mRNAs during germination. Plant Physiol. 2004;134: 1598–1613. pmid:15047896
- 73. Arenas-Mena C, Raynal M, Borrell A, Varoquaux F, Cutanda MC, Stacy RA, et al. Expression and cellular localization of Atrab28 during arabidopsis embryogenesis. Plant Mol Biol. 1999;40: 355–363. pmid:10412913
- 74. Kotak S, Vierling E, Bäumlein H, Koskull-Döring von P. A novel transcriptional cascade regulating expression of heat stress proteins during seed development of Arabidopsis. The Plant Cell. 2007;19: 182–195. pmid:17220197
- 75. Hundertmark M, Hincha DK. LEA (late embryogenesis abundant) proteins and their encoding genes in Arabidopsis thaliana. BMC Genomics. 2008;9: 118. pmid:18318901
- 76. Prieto-Dapena P, Castaño R, Almoguera C, Jordano J. The ectopic overexpression of a seed-specific transcription factor, HaHSFA9, confers tolerance to severe dehydration in vegetative organs. Plant J. 2008;54: 1004–1014. pmid:18315542
- 77. Tejedor-Cano J, Prieto-Dapena P, Almoguera C, Carranco R, Hiratsu K, Ohme-Takagi M, et al. Loss of function of the HSFA9 seed longevity program. Plant Cell Environ. 2010;33: 1408–1417. pmid:20444218
- 78. Borrell A, Cutanda MC, Lumbreras V, Pujal J, Goday A, Culiáñez-Macià FA, et al. Arabidopsis thaliana atrab28: a nuclear targeted protein related to germination and toxic cation tolerance. Plant Mol Biol. 2002;50: 249–259. pmid:12175017
- 79. Nakashima K, Fujita Y, Kanamori N, Katagiri T, Umezawa T, Kidokoro S, et al. Three Arabidopsis SnRK2 protein kinases, SRK2D/SnRK2.2, SRK2E/SnRK2.6/OST1 and SRK2I/SnRK2.3, involved in ABA signaling are essential for the control of seed development and dormancy. Plant Cell Physiol. Oxford University Press; 2009;50: 1345–1363. pmid:19541597
- 80. Winter D, Vinegar B, Nahal H, Ammar R, Wilson GV, Provart NJ. An “Electronic Fluorescent Pictograph” browser for exploring and analyzing large-scale biological data sets. Baxter I, editor. PLoS ONE. 2007;2: e718. pmid:17684564
- 81. Bassel GW, Fung P, Chow T-FF, Foong JA, Provart NJ, Cutler SR. Elucidating the germination transcriptional program using small molecules. Plant Physiol. 2008;147: 143–155. pmid:18359847
- 82. Šírová J, Sedlářová M, Piterková J, Luhová L, Petřivalský M. The role of nitric oxide in the germination of plant seeds and pollen. Plant Sci. 2011;181: 560–572. pmid:21893253
- 83. Borisjuk L, Rolletschek H. The oxygen status of the developing seed. New Phytol. Blackwell Publishing Ltd; 2009;182: 17–30. pmid:19207684
- 84. Lozano-Juste J, León J. Nitric oxide regulates DELLA content and PIF expression to promote photomorphogenesis in Arabidopsis. Plant Physiol. 2011;156: 1410–1423. pmid:21562334
- 85. Abbruzzetti S, Faggiano S, Spyrakis F, Bruno S, Mozzarelli A, Astegno A, et al. Oxygen and nitric oxide rebinding kinetics in nonsymbiotic hemoglobin AHb1 from Arabidopsis thaliana. IUBMB Life. 2011;63: 1094–1100. pmid:22034287
- 86. Thiel J, Rolletschek H, Friedel S, Lunn JE, Nguyen TH, Feil R, et al. Seed-specific elevation of non-symbiotic hemoglobin AtHb1: beneficial effects and underlying molecular networks in Arabidopsis thaliana. BMC Plant Biol. BioMed Central; 2011;11: 48. pmid:21406103
- 87. Zhang X, Liu S, Takano T. Two cysteine proteinase inhibitors from Arabidopsis thaliana, AtCYSa and AtCYSb, increasing the salt, drought, oxidation and cold tolerance. Plant Mol Biol. 2008;68: 131–143. pmid:18523728
- 88. Nishizawa-Yokoi A, Yoshida E, Yabuta Y, Shigeoka S. Analysis of the regulation of target genes by an Arabidopsis heat shock transcription factor, HsfA2. Biosci Biotechnol Biochem. 2009;73: 890–895. pmid:19352026
- 89. Rubinovich L, Weiss D. The Arabidopsis cysteine-rich protein GASA4 promotes GA responses and exhibits redox activity in bacteria and in planta. Plant J. 2010;64: 1018–1027. pmid:21143681
- 90. Job C, Rajjou L, Lovigny Y, Belghazi M, Job D. Patterns of protein oxidation in Arabidopsis seeds and during germination. Plant Physiol. 2005;138: 790–802. pmid:15908592
- 91. Alkhalfioui F, Renard M, Vensel WH, Wong J, Tanaka CK, Hurkman WJ, et al. Thioredoxin-linked proteins are reduced during germination of Medicago truncatula seeds. Plant Physiol. 2007;144: 1559–1579. pmid:17513483
- 92. Oracz K, El-Maarouf-Bouteau H, Farrant JM, Cooper K, Belghazi M, Job C, et al. ROS production and protein oxidation as a novel mechanism for seed dormancy alleviation. Plant J. 2007;50: 452–465. pmid:17376157
- 93. Cheah KS, Osborne DJ. DNA lesions occur with loss of viability in embryos of ageing rye seed. 1978;272: 593–599.
- 94. Karim S, Lundh D, Holmström K-O, Mandal A, Pirhonen M. Structural and functional characterization of AtPTR3, a stress-induced peptide transporter of Arabidopsis. J Mol Model. 2005;11: 226–236. pmid:15889294
- 95. Karim S, Holmström K-O, Mandal A, Dahl P, Hohmann S, Brader G, et al. AtPTR3, a wound-induced peptide transporter needed for defence against virulent bacterial pathogens in Arabidopsis. Planta. 2007;225: 1431–1445. pmid:17143616
- 96. Chiba Y, Shimizu T, Miyakawa S, Kanno Y, Koshiba T, Kamiya Y, et al. Identification of Arabidopsis thaliana NRT1/PTR FAMILY (NPF) proteins capable of transporting plant hormones. Journal of plant research. Springer Japan; 2015;128: 679–686. pmid:25801271
- 97. Cosgrove DJ. Characterization of long-term extension of isolated cell walls from growing cucumber hypocotyls. Planta. 1989;177: 121–130.
- 98. Lee Y, Kende H. Expression of alpha-expansin and expansin-like genes in deepwater rice. Plant Physiol. 2002;130: 1396–1405. pmid:12428004
- 99. Sampedro J, Cosgrove DJ. The expansin superfamily. Genome Biol. 2005;6: 242. pmid:16356276
- 100. Roxrud I, Lid SE, Fletcher JC, Schmidt EDL, Opsahl-Sorteberg H-G. GASA4, one of the 14-member Arabidopsis GASA family of small polypeptides, regulates flowering and seed development. Plant Cell Physiol. 2007;48: 471–483. pmid:17284469
- 101. Bush DS. Effects of gibberellic acid and environmental factors on cytosolic calcium in wheat aleurone cells. Planta. Springer-Verlag; 1996;199: 89–99–99.
- 102. Ueguchi-Tanaka M, Ashikari M, Nakajima M, Itoh H, Katoh E, Kobayashi M, et al. GIBBERELLIN INSENSITIVE DWARF1 encodes a soluble receptor for gibberellin. 2005;437: 693–698. pmid:16193045
- 103. Livne S, Weiss D. Cytosolic Activity of the Gibberellin Receptor GIBBERELLIN INSENSITIVE DWARF1A. Plant Cell Physiol. 2014;55: 1727–1733. pmid:25104541
- 104. Gilroy S, Jones RL. Perception of gibberellin and abscisic acid at the external face of the plasma membrane of barley (Hordeum vulgare L.) aleurone protoplasts. Plant Physiol. 1994;104: 1185–1192. pmid:12232156
- 105. McCubbin AG, Ritchie SM, Swanson SJ, Gilroy S. The calcium-dependent protein kinase HvCDPK1 mediates the gibberellic acid response of the barley aleurone through regulation of vacuolar function. Plant J. 2004;39: 206–218. pmid:15225286
- 106. Gubler F, Chandler PM, White RG, Llewellyn DJ, Jacobsen JV. Gibberellin signaling in barley aleurone cells. Control of SLN1 and GAMYB expression. Plant Physiol. 2002;129: 191–200. pmid:12011350
- 107. Itoh H, Ueguchi-Tanaka M, Sato Y, Ashikari M, Matsuoka M. The gibberellin signaling pathway is regulated by the appearance and disappearance of SLENDER RICE1 in nuclei. The Plant Cell. 2002;14: 57–70. pmid:11826299
- 108. Dekkers BJW, Pearce SP, van Bolderen-Veldkamp RPM, Holdsworth MJ, Bentsink L. Dormant and after-Ripened Arabidopsis thaliana Seeds are Distinguished by Early Transcriptional Differences in the Imbibed State. Front Plant Sci. Frontiers; 2016;7: 479. pmid:27625677
- 109. Oñate-Sánchez L, Vicente-Carbajosa J. DNA-free RNA isolation protocols for Arabidopsis thaliana, including seeds and siliques. BMC Res Notes. 2008;1: 93. pmid:18937828
- 110. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004;5: R80. pmid:15461798
- 111. Smyth G. Limma: linear models for microarray data. Bioinformatics and computational biology solutions using R and bioconductor. 2005;: 397–420.
- 112.
R Core Team. R: A language and environment for statistical computing [Internet]. R Foundation for Statistical Computing. Vienna, Austria; 2016. Available: https://www.R-project.org/
- 113. Kolesnikov N, Hastings E, Keays M, Melnichuk O, Tang YA, Williams E, et al. ArrayExpress update—simplifying data submissions. Nucleic Acids Res. 2015;43: D1113–6. pmid:25361974
- 114. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological). Wiley for the Royal Statistical Society; 1995;57: 289–300.
- 115. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003;4: 249–264. pmid:12925520
- 116. Eddelbuettel D, Francois R. Rcpp: Seamless R and C++ Integration. Journal of Statistical Software. 2011;40: 1–18.
- 117.
Wickham H. ggplot2. Springer Science & Business Media; 2009.
- 118. Fisher RA. On the Interpretation of χ2 from Contingency Tables, and the Calculation of P. Journal of the Royal Statistical Society. Wiley for the Royal Statistical Society; 1922;85: 87–94.
- 119. Benjamini Y, Yekutieli D. The control of the false discovery rate in multiple testing under dependency. The Annals of Statistics. Institute of Mathematical Statistics; 2001;29: 1165–1188.
- 120. Arvidsson S, Kwasniewski M, Riano-Pachon DM, Mueller-Roeber B. QuantPrime—a flexible tool for reliable high-throughput primer design for quantitative PCR. BMC Bioinformatics. BioMed Central Ltd; 2008;9: 465. pmid:18976492
- 121. Mortensen SA, Sønderkær M, Lynggaard C, Grasser M, Nielsen KL, Grasser KD. Reduced expression of the DOG1 gene in Arabidopsis mutant seeds lacking the transcript elongation factor TFIIS. FEBS Lett. 2011;585: 1929–1933. pmid:21569772
- 122. Guan Q, Wen C, Zeng H, Zhu J. A KH Domain-Containing Putative RNA-Binding Protein Is Critical for Heat Stress-Responsive Gene Regulation and Thermotolerance in Arabidopsis. Mol Plant. 2013;6: 386–395. pmid:23087326
- 123. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001;25: 402–408. pmid:11846609
- 124. Holm S. A simple sequentially rejective multiple test procedure. Scandinavian journal of statistics. 1979;6: 65–70.