Edited by Maarten Koornneef, Max Planck Institute for Plant Breeding Research, Cologne, Germany, and approved February 24, 2016 (received for review January 13, 2016)
Annual plants adapt to their environments by matching their life cycles, particularly seed germination and flowering, to the appropriate seasons. Although genetic evidence has suggested connections among genes regulating seed dormancy and flowering, specific mechanisms for such coordination are unknown. We report that a gene [DELAY OF GERMINATION1 (DOG1)] involved in determining the depth of seed dormancy, and therefore the seasonal timing of germination, also influences the timing of flowering in Arabidopsis and lettuce. We further show that this gene acts through influencing the production of microRNAs that govern the progression of developmental phase transitions through the plant life cycle, providing a molecular genetic mechanism for the coordinate adaptation of seed dormancy and flowering phenotypes to environmental conditions.
Seed germination and flowering, two critical developmental transitions in plant life cycles, are coordinately regulated by genetic and environmental factors to match plant establishment and reproduction to seasonal cues. The DELAY OF GERMINATION1 (DOG1) gene is involved in regulating seed dormancy in response to temperature and has also been associated genetically with pleiotropic flowering phenotypes across diverse Arabidopsis thaliana accessions and locations. Here we show that DOG1 can regulate seed dormancy and flowering times in lettuce (Lactuca sativa, Ls) and Arabidopsis through an influence on levels of microRNAs (miRNAs) miR156 and miR172. In lettuce, suppression of LsDOG1 expression enabled seed germination at high temperature and promoted early flowering in association with reduced miR156 and increased miR172 levels. In Arabidopsis, higher miR156 levels resulting from overexpression of the MIR156 gene enhanced seed dormancy and delayed flowering. These phenotypic effects, as well as conversion of MIR156 transcripts to miR156, were compromised in DOG1 loss-of-function mutant plants, especially in seeds. Overexpression of MIR172 reduced seed dormancy and promoted early flowering in Arabidopsis, and the effect on flowering required functional DOG1. Transcript levels of several genes associated with miRNA processing were consistently lower in dry seeds of Arabidopsis and lettuce when DOG1 was mutated or its expression was reduced; in contrast, transcript levels of these genes were elevated in a DOG1 gain-of-function mutant. Our results reveal a previously unknown linkage between two critical developmental phase transitions in the plant life cycle through a DOG1–miR156–miR172 interaction.
The life cycles of flowering plants are characterized by distinct phase transitions such as from seed to seedling (germination) or from vegetative to reproductive development (flowering) (1). The timing of germination and flowering both require precise environmental sensing and integrated responses to multiple inputs so that developmental transitions can be accurately matched to seasonal conditions (1⇓–3). Seeds use temperature as a signal of the seasonal and current environmental conditions to determine opportune times to germinate with respect to the potential for seedling survival (2, 4, 5). Similarly, in many plants the transition from vegetative to floral development occurs in response to environmental cues, particularly temperature and day length (6, 7). Ecological and evolutionary studies have found that seed germination and flowering traits within species are coadapted across habitat ranges (8⇓⇓–11). Seed dormancy and germination are regulated primarily by the antagonistic actions of the plant hormones gibberellin (GA; promotive) and abscisic acid (ABA; inhibitory), whose synthesis and action vary in response to environmental signals (12). Recent studies indicate that canonical genes regulating flowering, such as FLOWERING LOCUS T (FT) and FLOWERING LOCUS C (FLC), are also involved in the transition from seed dormancy to germination (13⇓⇓–16), suggesting that seed dormancy and flowering may be coordinately regulated through overlapping molecular pathways.
In Arabidopsis, expression of the DELAY OF GERMINATION1 (DOG1) gene responds to seed maturation temperature and determines the depth of seed dormancy (2, 3, 5, 17⇓⇓–20). Deeper dormancy of freshly harvested seeds is associated with high DOG1 transcript levels, which decrease in after-ripened (nondormant) and germinating seeds (18, 19). Functional analyses have shown that DOG1 and ABA are essential for establishing primary seed dormancy (18, 19, 21), although DOG1 can act independently of ABA to delay germination of less dormant seeds (20). Overexpression of DOG1 also increases the sensitivity of seed germination to inhibition by warm temperatures (5, 20). In addition, genome-wide association and genetic linkage mapping studies revealed that the DOG1 locus was associated with flowering phenotypes across diverse accessions and locations (22, 23). However, the mechanism of DOG1 action on seed dormancy or flowering remains unknown.
Lettuce (Lactuca sativa L.) is a major leafy vegetable derived from a progenitor species (Lactuca serriola L.) having a winter annual lifecycle, normally flowering in spring after overwintering as a compact rosette. Lettuce seeds display little primary dormancy, but germination is strongly inhibited by warm temperatures during imbibition, a type of relative dormancy termed thermoinhibition (24, 25), resulting in reduced crop establishment during warm seasons (26). Previously, we genetically mapped and functionally characterized the role of L. sativa 9-cis-EPOXYCAROTENOID DIOXYGENASE4 (LsNCED4), a gene encoding a key regulated enzyme in ABA biosynthesis, in lettuce seed thermoinhibition (24, 25, 27). Expression of this gene is induced in lettuce seeds by imbibition at high temperatures, maintaining high ABA content and preventing germination. A similar role was demonstrated for the homologous gene Arabidopsis thaliana NCED9 (AtNCED9) (28). The mechanism by which these genes are regulated by high temperature remains unknown (29).
Here, we report that suppression of DOG1 expression by RNAi in lettuce or mutation of DOG1 in Arabidopsis markedly increased the maximum temperature for seed germination. In addition, DOG1-RNAi lettuce plants flowered much earlier than control plants. Further analyses in lettuce and Arabidopsis showed that these phenotypes are associated with reduced conversion of primary MIR156 transcript into the microRNA (miRNA) miR156. Effects of overexpression of MIR156 on flowering and seed dormancy were greatly compromised in loss-of-function dog1-3 mutants of Arabidopsis. The expression of several miRNA processing genes was altered in seeds of dog1-3 mutants or DOG1-RNAi lines, possibly associated with the reduced conversion of MIR156 to miR156. In contrast, a gain-of-function dog1-5 mutation that enhanced seed dormancy resulted in increased expression of miRNA processing genes. As have been described previously for flowering (30, 31) and seed and seedling development (32, 33), follow-on effects of modification of miR156 levels on miR172 abundance, or direct overexpression of MIR172, also affected seed dormancy. Our work provides a potential mechanism for DOG1 action and demonstrates a previously unknown linkage between seed dormancy and flowering phenotypes through a DOG1–miR156–miR172 interaction.
To explore whether lettuce homologs of AtDOG1 are involved in regulation of seed thermoinhibition, cDNAs with homology to DOG1 were isolated from four genotypes of lettuce having differing germination thermotolerance. Seeds of cv. Salinas (L. sativa; termed Sal) and PI261653 (Lactuca saligna; termed Saligna) were unable to germinate well at 32 °C, whereas seeds of two thermotolerant genotypes, accession US96UC23 (L. serriola) and PI251246 (L. sativa; termed PI), germinated fully at this temperature (SI Appendix, Fig. S1A). Sequences of LsDOG1 cDNAs from the four genotypes were highly homologous and shared ∼50% overall protein amino acid sequence similarity with Arabidopsis DOG1 (54.7% similarity within DOG1 domain regions; SI Appendix, Fig. S2 and Table S1). LsDOG1 is highly expressed in mature lettuce seeds, whereas its transcript level is much lower in leaves and roots (SI Appendix, Fig. S1C). Its expression in seeds decreased as seed maturation temperature increased (SI Appendix, Fig. S1D), as is also the case for AtDOG1 expression in Arabidopsis seeds (17, 18).
Suppression of LsDOG1 expression in Sal lettuce through RNAi caused 56% and 73% reductions in transcript abundance in two independent lines (Fig. 1A). Seeds of these DOG1-RNAi lines were able to germinate fully at 32 °C, whereas seeds of the segregated nontransgenic control line (i.e., Sal) were completely inhibited (Fig. 1 B and C). In contrast, ectopic expression of LsDOG1 under the native Sal DOG1 promoter (PDOG1:LsDOG1) in the thermotolerant PI lettuce line resulted in increased DOG1 transcripts in mature dry seeds and enhanced seed thermoinhibition at 30 °C (Fig. 1 D–F). A functional homolog of DOG1 is therefore present in lettuce and is involved in regulating germination thermoinhibition.
Regulation by LsDOG1 of lettuce seed thermoinhibition. (A) Relative LsDOG1 mRNA levels in dry seeds of Sal lettuce and its LsDOG1-RNAi lines, 21–5G and 21–1I. (B and C) Seed germination of Sal, 21–5G, and 21–1I lines at 20 and 32 °C. (D) Relative LsDOG1 mRNA levels in dry seeds of PI lettuce and its LsDOG1 over-expression lines, 33–10 and 33–04. (E and F) Seed germination of PI, 33–10, and 33–04 lines at 20 and 30 °C. In D and F, seeds at 20 °C were imbibed for 16 h, whereas seeds at 32 and 30 °C were imbibed for 30 h. Error bars represent SE (n = 3).
Functional DOG1 action in Arabidopsis promotes seed dormancy and extends after-ripening periods required for dormancy alleviation (18, 19), whereas fresh seeds of loss-of-function dog1 mutants can fully germinate without after-ripening or cold stratification (17, 19). To test whether DOG1 might be involved in regulating germination thermoinhibition in Arabidopsis, we imbibed seeds of the Columbia (Col) dog1-3 mutant (Salk_000867) at 32 °C. Mutant seeds germinated more than 80% whereas WT seeds germinated less than 20% at that temperature (SI Appendix, Fig. S1B). Molecular complementation showed that coding regions of LsDOG1 genes from thermosensitive and thermotolerant lettuce genotypes were able to rescue the thermoinhibition phenotype when expressed in Arabidopsis dog1-3 plants (SI Appendix, Fig. S1B), indicating that these homologs are all functional and that their role in regulation of seed thermoinhibition is conserved. The presence of functional DOG1 alleles in US96UC23 and PI251246 is consistent with lack of association of this locus with quantitative trait loci for high-temperature germination in these genotypes (27, 34).
Previous work in lettuce and Arabidopsis showed that NCED genes encoding a key enzyme in the ABA biosynthetic pathway are involved in thermoinhibition of germination (24, 28). We tested the interactive effects of DOG1 and ABA on seed thermoinhibition by overexpressing LsDOG1 in an atnced6-1 atnced9-1 double mutant that displays strong germination thermotolerance as a result of reduced ABA biosynthesis (24, 28). Overexpression of LsDOG1 restored a thermosensitive phenotype to atnced6-1atnced9-1 seeds at 32 °C (SI Appendix, Fig. S1E), indicating that DOG1 and ABA can induce thermoinhibition via parallel pathways. We also generated a dog1-3 × nced9-1 double mutant to further examine the interactive effect of both genes on thermoinhibition. At a higher temperature (35 °C), seeds of the single mutants of dog1-3 and nced9-1 and of the double mutant nced6-1 nced9-1 were fully inhibited from germinating, whereas dog1-3 nced9-1 seeds germinated 95% (SI Appendix, Fig. S1F). This additional effect of dog1-3 in the presence of ABA biosynthesis mutants is in contrast to reports that ABA biosynthesis and functional DOG1 are required for induction of primary dormancy (18, 19). However, consistent with previous reports (17⇓⇓–20), overexpression of LsDOG1 under the CaMV35S promoter in Arabidopsis Col plants resulted in deep seed dormancy that could not be alleviated by the ABA biosynthesis inhibitor fluridone, but could be alleviated by GA (SI Appendix, Fig. S3). Thus, DOG1 and ABA can regulate seed primary dormancy and thermoinhibition, and may act in concert or independently depending on the conditions.
Suppression of DOG1 expression in Sal lettuce caused early flowering (Fig. 2 A–C), as occurred also in wheat (35). Plants from seeds of seven independent transgenic DOG1-RNAi lines that germinated at 32 °C and from segregated control (nontransgenic) seeds were grown in the greenhouse under relatively noninductive day-length conditions for flowering in fall 2013. Salinas DOG1-RNAi lines flowered in ∼120–150 d, whereas control lines did not flower until >200 d (Fig. 2B). Similar results were observed in two DOG1-RNAi lines that were grown in spring 2014, although these plants all flowered earlier because of lengthening days (Fig. 2C).
Regulation of flowering time in lettuce by DOG1. (A) Two independent early-flowering lettuce lines (DOG1-RNAi) and the control Sal line (CTL) are shown at 120 d after seeding. (B) Times to flowering of F2 plants of seven independent transgenic lines and a segregated control line (CTL). Plants were sown in September 2013 and were genotyped for the presence of the transgene (n ≥ 20 plants per line). (C) Times to flowering of two homozygous DOG1-RNAi lines and a control line that were sown in late February 2014 (n ≥ 20 plants per line). (D) Relative levels of miR156 and miR172 and mRNA levels of SPL3, SPL4, and SPL9 in apical meristems of 6-wk-old control (CTL) and DOG1-RNAi (21-5G, 21–1I) lettuce plants; levels were first normalized by LsUBQ10 and are shown relative to those in control for each gene or miRNA. Error bars represent SE (n = 3).
To assess which flowering regulatory pathways were affected by suppression of LsDOG1 expression, transcript levels of key genes in the different pathways were investigated (SI Appendix, Fig. S4). Consistent with early flowering, transcripts of the floral promoter FT were almost 25-fold higher in young DOG1-RNAi lettuce leaves than in similar leaves of the control line (SI Appendix, Fig. S4). Expression patterns of other flowering-related genes were not consistent with effects on the photoperiod-dependent, GA, vernalization, or autonomous flowering pathways (SI Appendix, Supplementary Text and Figs. S4 and S5). However, transcripts of SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPL) 3, 4, and 9 were increased in leaves of the DOG1-RNAi line compared with its control line (SI Appendix, Fig. S4). Furthermore, the relative mRNA levels of SPL3 and 4 were significantly elevated in apical meristems of DOG1-RNAi lines, although SPL9 expression was not affected (Fig. 2D). SPL gene expression is inhibited by DELLA proteins (SI Appendix, Fig. S5), so a decrease in them could result in increased SPL expression (36). However, transcripts of genes encoding two major DELLA proteins (RGL1 and RGA) were unchanged or elevated in DOG1-RNAi leaves (SI Appendix, Fig. S4), making it unlikely that this is the cause of increased SPL expression. SPL transcripts are targeted for degradation by miR156, a regulator of developmental phase transitions in plants, and SPL proteins can induce expression of FT and other genes promoting flowering (30). SPLs also promote production of miR172, a positive regulator of maturation (juvenile to adult) and floral phase transitions (30). Thus, if this mechanism is operative, we would expect early flowering and high SPL transcript levels to be associated with lower miR156 and higher miR172 levels. Consistent with this, miR156 abundance in apical meristems of DOG1-RNAi lines was only one third of that in the control line, whereas miR172 levels were approximately doubled (Fig. 2D). Thus, the early flowering in DOG1-RNAi lettuce plants may be a result of a reduction in miR156 levels, which would result in increased levels of the floral-promoting SPLs and of miR172 (SI Appendix, Fig. S5).
In contrast to early flowering when LsDOG1 expression was suppressed in lettuce, no effects on flowering times were observed in Arabidopsis dog1-3 mutant plants under long-day or short-day conditions (Table 1 and SI Appendix, Table S2). However, when miR156 levels were elevated by overexpression of the gene encoding it, MIR156 (Col-35S:LsMIR156), flowering times were significantly delayed (Fig. 3A), with a majority of plants beginning to flower more than 50 d after germination (DAG) and having more than 36 rosette leaves; in contrast, flowering of homozygous dog1-3 plants also expressing 35S:LsMIR156 was only slightly delayed compared with Col and dog1-3 plants (Table 1 and SI Appendix, Table S2). Further evaluation of 48 dog1-3–35S:LsMIR156 and 72 Col-35S:LsMIR156 independent transgenic lines showed that 87% of the former had flowered within 50 DAG and 95% had less than 30 leaves at flowering, whereas 72% of the latter required more than 50 d to flower and 83% of flowering lines had more than 30 leaves (Table 1 and SI Appendix, Table S2). Although it is possible that sequence homology between the CaMV35S promoter and the T-DNA insertion in dog1-3 could silence the transgene (37), this result was also reproduced in Landsberg erecta (Ler) and its γ-irradiation–induced dog1-1 mutant (19), although the delay in flowering time caused by 35S:LsMIR156 in Ler was not as strong as in Col (Table 1 and SI Appendix, Table S2). As in Arabidopsis, overexpression of LsMIR156 also greatly delayed flowering in Sal lettuce (SI Appendix, Fig. S6). To test whether this phenomenon is specific to lettuce MIR156, we crossed dog1-3 with a Col-35S:AtMIR156 line that exhibits delayed flowering (38). Effects on flowering were essentially identical to those for LsMIR156 (Fig. 3A, Table 1, and SI Appendix, Table S2). The lack of response of dog1-3 plants to overexpression of LsMIR156 or AtMIR156 was evidently caused by reduced conversion of pri-MIR156 transcripts to miR156, as LsMIR156 (or AtMIR156) transcripts were present at equal or greater levels in dog1-3 meristems (Fig. 3B and SI Appendix, Fig. S7F) or leaves (SI Appendix, Fig. S7A) as in Col tissues also expressing LsMIR156 or AtMIR156. The levels of miR156 in dog1-3–35S:LsMIR156 meristems were only 28% or 13% of those in 25-d-old apical meristems or leaves, respectively, of the Col-35S:AtMIR156 plants (Fig. 3B and SI Appendix, Fig. S7B). This was associated with an increase in miR172 and SPL3, 4, 5, and 9 transcripts in dog1-3–35S:LsMIR156 meristems and leaves (Fig. 3B and SI Appendix, Fig. S7 C–E). Similar differences in miR156 and miR172 levels were also observed in Col and dog1-3 meristems in response to overexpression of AtMIR156 (SI Appendix, Fig. S7F).
DOG1 effect on flowering times (days to flowering) of Arabidopsis genotypes
Regulation of flowering time in Arabidopsis by DOG1. (A) Seven-week-old Col, dog1-3, dog1-3–35S:LsMIR156 (dog1-Ls156), dog1-3 × Col-35S:AtMIR156 (dog1×At156), Col-35S:LsMIR156 (Col-Ls156), and Col-35S:AtMIR156 (Col-At156) plants grown in long days. (B) Relative levels of miR156 and miR172 and mRNA levels of LsMIR156, SPL3, SPL4, SPL5, and SPL9 in apical meristems of Col and dog1-3 plants expressing LsMIR156 at 25 DAG; levels were first normalized by AtACT2, and values relative to those in the Col-LsMIR156 meristems for each gene or miRNA are shown. Error bars represent SE (n = 3). (C) Col-35S:LsMIR172 and dog1-3–35S:LsMIR172 plants grown for 65 d in short days.
It was recently reported that dog1-5 (SALK_022748) is a gain-of-function mutant in which the level of DOG1 protein is greatly increased and seed dormancy is enhanced compared with Col WT (39), as we have confirmed here (SI Appendix, Fig. S8). In this case, the mutation in DOG1 did not negate the effect of MIR156 overexpression on delaying flowering, but rather somewhat enhanced it; 93% of independent dog1-5-LsMIR156 transgenic lines required more than 50 DAG to flower compared with 72% of Col-LsMIR156 plants, with similar data for numbers of leaves at flowering (Table 1 and SI Appendix, Table S2). Interestingly, as for dog1-3, we did not observe a significant difference in flowering time for dog1-5 plants in the absence of MIR156 overexpression (Table 1 and SI Appendix, Table S1). Thus, in contrast to lettuce, DOG1 regulation of flowering time in Arabidopsis is evident only when MIR156 expression is enhanced.
ABA has also been reported to be involved in floral transitions (40), raising the question of whether the early flowering observed in lettuce DOG1-RNAi plants and dog1-3 plants overexpressing MIR156 could be caused by the alteration of ABA contents, as mutants in ABA biosynthetic genes have similar seed germination phenotypes as dog1-3 mutant seeds (SI Appendix, Fig. S1 B and E). However, the delayed flowering time phenotypes conferred by 35S:AtMIR156 and 35S:LsMIR156 in Col plants also occurred in an nced9-1 mutant background (Table 1 and SI Appendix, Table S2). Thus, although phenotypic effects of MIR156 overexpression on flowering depend upon functional DOG1, expression of NCED9, which is primarily seed-expressed and required for thermoinhibition (28), is not necessary.
The results reported here show that loss of DOG1 function negatively affects processing of MIR156 to miR156 in lettuce and Arabidopsis and elevates miR172 abundance. As miR172 is known to promote early flowering in Arabidopsis (30), we tested the effect of overexpression of LsMIR172 on flowering in Col WT and dog1-3 mutant plants. To maximize phenotypic differences, we grew 100 (for Col-35S:LsMIR172) and 77 (for dog1-3–35S:LsMIR172) independent transgenic lines under short-day conditions. Under short days, WT Col plants developed more than 40 leaves before flowering and 90% of plants did not flower before 90 d, whereas 85% of Col-35S:MIR172 plants had less than 40 leaves at flowering and 39% started flowering in less than 90 DAG (Fig. 3C, Table 1, and SI Appendix, Table S2). This advancement of flowering as a result of overexpression of MIR172 was much reduced in dog1-3 plants: more than 60% of dog1-3–35S:LsMIR172 plants had more than 40 leaves at flowering, as did the dog1-3 plants, and only 9% of dog1-3:35S:LsMIR172 plants flowered in less than 90 DAG (Fig. 3C, Table 1, and SI Appendix, Table S2). Thus, functional DOG1 plays a role in the processing or action of MIR172 as well as of MIR156 (SI Appendix, Fig. S5).
Suppression of DOG1 expression in lettuce, which improved germination at warm temperatures (Fig. 1 B and C), also resulted in large reductions in miR156 content in dry seeds, with levels only 5–7% of that in control seeds (Fig. 4A). In mature dog1-3 mutant seeds, the miR156 level was slightly lower than that in mature WT Col seeds, but this difference was more significant when LsMIR156 was overexpressed (Fig.4B). We therefore tested whether expression of MIR156 and MIR172 would affect primary dormancy and thermoinhibition of Arabidopsis seed germination in WT and dog1-3 mutant plants. Expression of 35S:LsMIR156 or 35S:AtMIR156 in Col plants reduced germination at 25 °C, likely reflecting increased primary dormancy, and strongly enhanced seed thermoinhibition at warmer temperatures (Fig. 4C). In contrast, overexpressing LsMIR156 in dog1-3 plants did not affect germination at 25, 30, or 32 °C (Fig. 4C). Although seeds of the nced9-1 mutant, like dog1-3 seeds, were thermotolerant, expression of 35S:AtMIR156 in nced9-1 plants had similar effects in suppressing germination as in Col plants (Fig. 4C). This demonstrates that elevation of miR156 can inhibit germination independently of seed ABA biosynthesis and that the effect of MIR156 expression on seed germination inhibition is dependent on functional DOG1. In contrast, overexpression of LsMIR172 promoted germination of fresh (primary dormant) Col seeds, even at elevated temperatures (Fig. 4D). As dog1-3 seeds did not exhibit primary dormancy and germinated fully at all temperatures tested, an additional effect of LsMIR172 overexpression on germination of dog1-3 seeds could not be detected (Fig. 4D). Thus, miR156 and miR172 can influence seed thermoinhibition as well as flowering and other developmental phase transitions (SI Appendix, Fig. S5).
Lettuce seed content of miR156, effects of modification of miR156 or miR172 expression on germination, and effects of dog1 mutations on expression of DICER-related genes. (A) Levels of miR156 in dry seeds of control and three homozygous DOG1-RNAi lettuce lines, 21–6A, 21–5G, and 21–1I. (B) miR156 levels in dry seeds of Col, dog1-3 mutant, Col expressing LsMIR156, and dog1-3 mutants expressing LsMIR156. (C) Effect of expression of 35S:AtMIR156 and 35S:LsMIR156 on germination of seeds of Col, nced9-1, and dog1-3 genotypes of Arabidopsis. All seeds were tested at 3 wk after harvesting. (D) Germination at 25, 30, or 32 °C of Col WT or dog1-3 mutant seeds and of seeds of these genotypes that had been transformed to express 35S:LsMIR172. Seeds were tested 5 d after harvesting when primary dormancy was still present in Col-WT seeds to demonstrate alleviation of dormancy by LsMIR172 overexpression. (E–G) Transcripts of genes associated with miRNA processing or transcription (DCL1, HYL1, TGH, SE, and CDC5) in dry seeds of Col and its T-DNA dog1-3 (loss-of-function) and dog1-5 (gain-of-function) mutants (E), of Ler-CviDOG1 introgression line and its irradiation-induced mutant dog1-1 (F), and of Salinas lettuce (CTL) and its DOG1-RNAi silencing line (21–1I) (G). mRNA levels were first normalized by AtACT2; values relative to those in Col (E), in Ler-CviDOG1 (F), and in Sal (G) are shown. Error bars represent SE (n = 3).
As the conversion of primary MIR156 to miR156 was greatly compromised in seeds of lettuce DOG1-RNAi suppression lines and of Arabidopsis dog1-3 mutants (Fig. 4 A and B), we tested whether transcripts of genes encoding proteins associated with miRNA transcription or processing were altered. After preliminary screening, we focused on five genes: DICER-LIKE 1 (DCL1), HYPONASTIC LEAVES1 (HYL1), the G-patch domain protein TOUGH (TGH), the zinc finger protein SERRATE (SE), and CELL DIVISION CYCLE 5 (CDC5) that binds RNA polymerase II and MIR promoters to positively regulate transcription of MIR-encoding genes (41, 42). Although the specific effects varied somewhat among the Arabidopsis and lettuce genotypes and mutants tested, transcript levels of all of these genes differed from their respective WTs in at least some cases (Fig. 4 E–G). For example, the transcript levels of DCL1, HYL1, and CDC5 in dry dog1-3 mutant seeds were 56%, 55%, and 76%, respectively, of those in dry Col WT seeds, whereas expression of these genes (and of TGH and SE) was elevated compared with Col-WT in seeds of the gain-of-function dog1-5 mutant (Fig. 4E). Consistent reduction in expression of these genes was also observed in dry seeds of the dog1-1 mutant compared with those in Ler-CviDOG1 (Fig. 4F), which contains a strong CviDOG1 allele (19), and transcripts of DCL1, SE, TGH, and CDC5 were also lower in dry DOG1-RNAi lettuce seeds (Fig. 4G). These results indicate that DOG1 could affect miR156 (and possibly miR172) processing in seeds through altering expression of genes encoding miRNA-processing proteins (SI Appendix, Fig. S5).
Seed germination and flowering are two critical developmental-phase transitions in the plant life cycle. Plants sense environmental cues such as temperature to coordinate endogenous genetic and molecular mechanisms that determine the timing of both phase transitions (1, 6, 7, 12, 43, 44). Although its mechanism of action remains unknown, DOG1 has been demonstrated to be a positive regulator for the establishment and maintenance of primary seed dormancy in several species (19, 20, 35). DOG1 is involved in transducing maternal environmental conditions during seed development into effects on seed dormancy and in the loss of dormancy through dry after-ripening or moist chilling (2, 5, 17, 45, 46). ABA is also a strong positive regulator of seed dormancy (12) and promoted DOG1 expression in imbibed Lepidium sativum seeds (47). Arabidopsis seeds may require ABA and functional DOG1 to establish primary dormancy, as reduced dormancy in Arabidopsis seeds was still observed in ABA-deficient aba1 mutants, even when the strong Cvi-DOG1 allele was present (18, 19), whereas seeds of a dog1-2 cyp707a2-1 double mutant that have higher levels of ABA nonetheless exhibited reduced seed dormancy as a result of the loss of DOG1 function (18). Here, we demonstrated that DOG1 is involved in regulation of thermoinhibition of seed germination in lettuce and Arabidopsis, but, unlike the dependence of primary dormancy on ABA and DOG1, DOG1 can act in the absence of enhanced ABA synthesis to impose thermoinhibition, consistent with the relative ineffectiveness of an ABA biosynthesis inhibitor to alleviate effects of overexpression of DOG1 on thermoinhibition (SI Appendix, Fig. S3) (20). Thus, ABA and DOG1 may regulate seed thermoinhibition through parallel but mutually reinforcing pathways that may not be identical to those by which they act on primary dormancy (29). On the contrary, GA can alleviate the effects of overexpression of DOG1 on thermoinhibition (SI Appendix, Fig. S3) (20), as might be expected from the suppression of expression of GA biosynthetic enzymes by high temperatures in Arabidopsis and lettuce (25, 28, 34).
Ecological genetics studies have shown that, under natural seasonal variation, DOG1 can influence seed germination and flowering times (48). Although, initially, the effects of DOG1 on flowering were suggested to be an indirect response to the seasonal timing of germination, further study showed that the alleviation of seed dormancy resulted in an acceleration of flowering that was independent of germination timing (49). Genome-wide association and genetic linkage mapping analyses have implicated the DOG1 locus in the regulation of flowering phenotypes (22, 23), but these results were attributed to the possibility of being caused by closely linked genes, as DOG1 was assumed to be active only in seeds. However, our results demonstrate that DOG1 itself can have effects on seed dormancy and flowering, suggesting a more direct mechanism for the evolutionary coadaptation of life-cycle transitions to match seasonal environmental conditions (16, 44, 49, 50). In addition, previous studies have shown that genes related to flowering are expressed in association with seed development and germination. For example, a genome-wide expression correlation network in Arabidopsis showed that maturity-phase transition regulators such as SE, EMBRYONIC FLOWER1, and EARLY BOLTING IN SHORT DAYS are coexpressed with germination-associated genes and affect seed germination (14). Arabidopsis HISTONE MONOUBIQUITINATION1 and 2 (HUB1 and HUB2) genes regulate flowering and seed germination; seeds of hub1 and hub2 mutants displayed reduced seed dormancy (51), and hub1 seeds also exhibit strong thermotolerance to high temperature at germination (52). The increased germination of hub1 and hub2 seeds could be attributed to the altered expression of several seed dormancy-related genes, including DOG1 and NCED9 (51). Loss of function of HUB1 and HUB2 also caused early flowering phenotypes through repression of FLC by chromatin modification (53). In addition to repressing flowering, FLC also promotes seed germination at low temperature (10 °C) via changing expression of ABA-catabolism (CYP707A2) and GA-biosynthesis (GA20ox1) genes (15). Consistent with these observations, the flowering-promotive protein FT maternally controls seed dormancy through alteration in seed coat tannin content in Arabidopsis (13). Thus, although there have been previous indications that seed dormancy and flowering might share common genetic and molecular mechanisms, our understanding of how such mechanisms might be connected to each other and to environmental cues has been limited.
Our results demonstrate that DOG1 regulates seed germination and flowering time at least in part through an influence on generation and/or action of miR156 and miR172. miRNAs have been implicated previously as being involved separately in seed germination and flowering time (30, 32). Mutants that are defective in DCL1, HYL1, and SE, all of which are required for miRNA biogenesis (41), display abnormal flowering and alteration in ABA sensitivity of seed germination (54, 55). Interestingly, the delayed-flowering phenotype of dcl1-7 mutant could be rescued by expression of a DCL4-dependent MIR839 in which the mature sequence was replaced by the one encoding MIR172, suggesting that the delayed flowering in dcl1-7 plants is a result of the reduction in miR172 (56). Earlier transition from the juvenile to the adult phase was also observed in loss-of-function mutants of SE and HYL1, in association with substantial reduction in miR156, and could be rescued by overexpression of MIR156 (57, 58). Another miR156 target, SPL13, plays an important role in the regulation of the postgerminative switch from the cotyledon stage to the vegetative-leaf stage during seedling growth (59). Mutants in dcl1 exhibit early seed maturation programming, probably caused by increased levels of SPL10 and SPL11 as a consequence of reduced miR156 (60, 61). Our data indicate that DOG1 may affect the conversion of primary MIR156 to miR156 or primary MIR172 to miR172 through an effect on expression of genes involved in miRNA processing (SI Appendix, Fig. S5). Loss of function or gain of function of DOG1 decreases or increases, respectively, expression of DICER-related genes (Fig. 4 E–G), and previous studies found that miR156 and miR172 levels are altered in mutants of DCL1, HYL1, SE, TGH, and CDC5 (42, 56, 58, 62, 63). However, multiple feedback loops are associated with transcription and processing of miRNAs (41), so further study is needed to clarify the specific mechanism by which DOG1 affects miRNA levels. In addition, the strong effects of DOG1 overexpression in enhancing seed dormancy even in the absence of MIR156 overexpression (SI Appendix, Fig. S3) (18, 19) indicate that DOG1 may also act on seed germination through additional mechanisms (SI Appendix, Fig. S5).
miR156 accumulates to high levels during seed development (32, 64) and is abundant in young seedlings and decreases as plants age, in association with the transition to adult phase and flowering (30, 38, 65). This pattern indicates that high miR156 levels are reset during embryogenesis and seed development, as occurs for FLC, a strong floral repressor (7). Winter annual flowering plants require prolonged chilling to epigenetically silence FLC to flower the next spring; to ensure the requirement for vernalization in every generation, it is essential to reactivate FLC gene expression during seed development (66). Consistent with this, expression and maintenance of FLC during late embryogenesis are required for late flowering phenotypes (67, 68). Similarly, DOG1 may play a role in resetting high miR156 levels during seed development (Fig. 4A), and defects in DOG1 function could result in altered postgerminative plant development and floral transition as well as reduced seed dormancy (SI Appendix, Fig. S5) (32, 41, 64, 65). We also note that a primary phenotype of DOG1 action is an extension of dry after-ripening times required for dormancy alleviation (19). A role for miRNAs produced during seed development in maintaining seed dormancy provides a mechanism for the after-ripening effect, in which seeds lose dormancy during dry storage when metabolism is prevented (12). miR156 accumulation during seed development, and its subsequent loss during storage as a result of oxidation, as shown for stored mRNAs (69), would provide a timing mechanism for after-ripening via loss of the inhibitory effect of stored miR156 on germination.
Our results support the conclusion that the transition from dormancy to germination is a life-cycle phase transition comparable to vegetative maturation or flowering, and apparently is influenced by the same miRNA-dependent systems (SI Appendix, Fig. S5) (32). Effects on these systems are also consistent with the hypothesis that DOG1 primarily affects the underlying environmental sensitivity of processes it influences, rather than directly determining final phenotypes (5). As the miR156/SPL/miR172 pathway is involved in modulating the inputs from a diverse array of environmental signals and transducing them into developmental responses (6, 7, 32, 59), interaction of DOG1 with this system would enable it to play an integrating role by matching the environmental sensitivities of seed dormancy and flowering to current or anticipated conditions. A dual role in sensing environmental signals (e.g., temperature) and coordinating developmental-phase transitions in the plant life cycle would explain the repeated identification of DOG1 as a significant locus in ecological genetics studies of flowering phenotypes (22, 23, 48). It would also make evolutionary sense, as seed dormancy and germination timing influence the environment in which subsequent flowering and reproduction occur, and vice versa (16, 44, 50). An integrated mechanism for coordinating these two major life cycle transitions having significant impacts on fitness and survival would be subject to coselection to optimize (or bet-hedge) both (9, 11, 43, 49, 50). Much remains to be done to fully understand these complex interactions between plant life cycles and the environment, but our results demonstrate that DOG1 functions as an important molecular integrator that exerts its effects on developmental phase changes at least in part through miRNA-regulated pathways.
Seeds of lettuce (L. sativa L.) were germinated at room temperature and then transferred to an automatically controlled greenhouse at 23 ± 2 °C (day) and 20 ± 2 °C (night) in long days (14 h light/10 h dark) for seed production unless stated otherwise. Seeds of A. thaliana were stratified at 4 °C for 2–4 d before being transferred to a growth chamber at 21–22 °C under long days (16 h light/8 h dark, 100 or 135 µmol⋅m−2⋅s−1 light intensity) or short days (8 h light/16 h dark, 135 µmol⋅m−2⋅s−1 light intensity) for seed production. Seeds of the γ-irradiation–induced dog1-1 mutant were provided by Leonie Bentsink, Wageningen University, Wageningen, The Netherlands (19), seeds of the T-DNA mutant nced9-1 by Eiji Nambara, University of Toronto, Toronto, and seeds of the T-DNA double-mutant nced6-1 nced9-1 by Annie Marion-Poll, Institut Jean-Pierre Bourgin, Versailles, France. Seeds of Arabidopsis WT (Col and Ler), T-DNA dog1-3 mutant (Salk_000867) (19), T-DNA dog1-5 mutant (SALK_022748) (39), and AtMIR156-overexpression line (CS67849) (30) were purchased from the Arabidopsis Biological Resource Center. The dog1-3 mutant line was selfed and genotyped for three generations before this study. The Arabidopsis T-DNA mutants were genotyped using the primers listed in SI Appendix, Table S3. Details of experimental procedures including isolation of LsDOG1 and vector construction, generation of dog1-3 × 35S:AtMIR156 and nced9–1 × 35S:LsMIR156 lines, seed germination assays, measurements of flowering times, identification of LsMIR156 and LsMIR172, vector construction for overexpression of LsMIR156 and LsMIR172, plant transformation, and mRNA and miRNA analyses are described in SI Appendix, Supplementary Methods. Primers used in this study are listed in SI Appendix, Table S3.
We thank University of California, Davis, student interns Phat Hua, Alex Wai, and Macrae Dec-Hull and visiting Universidade Federal de Viçosa (Brazil) student Danubia Nobre for assistance in maintaining plants. Extramural funding for this work was provided by United States Department of Agriculture (USDA)-National Institute of Food and Agriculture Grant 2008-02509, USDA-Cooperative State Research, Education, and Extension Service Regional Research Project W3168, and Rijk Zwaan B.V., De Lier, The Netherlands. Access to lettuce genomic sequences was provided through the University of California, Davis, Genome Center and supported in part by National Science Foundation Award 0820451 and the Compositae Genome Project. The visit of S.W. to University of California, Davis, was supported by China Scholarship Council (201203250049) and Natural Science Foundation of China (31171867).
Author contributions: H.H. and K.J.B. designed research; H.H. and S.W. performed research; H.H. and K.J.B. analyzed data; and H.H. and K.J.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1600558113/-/DCSupplemental.
