The etr1-2 mutation in Arabidopsis thaliana affects the abscisic acid, auxin, cytokinin and gibberellin metabolic pathways during maintenance of seed dormancy, moist-chilling and germination


(fax 604 291 3496; e-mail


In Arabidopsis thaliana, the etr1-2 mutation confers dominant ethylene insensitivity and results in a greater proportion of mature seeds that exhibit dormancy compared with mature seeds of the wild-type. We investigated the impact of the etr1-2 mutation on other plant hormones by analyzing the profiles of four classes of plant hormones and their metabolites by HPLC-ESI/MS/MS in mature seeds of wild-type and etr1-2 plants. Hormone metabolites were analyzed in seeds imbibed immediately under germination conditions, in seeds subjected to a 7-day moist-chilling (stratification) period, and during germination/early post-germinative growth. Higher than wild-type levels of abscisic acid (ABA) appeared to contribute, at least in part, to the greater incidence of dormancy in mature seeds of etr1-2. The lower levels of abscisic acid glucose ester (ABA-GE) in etr1-2 seeds compared with wild-type seeds under germination conditions (with and without moist-chilling treatments) suggest that reduced metabolism of ABA to ABA-GE likely contributed to the accumulation of ABA during germination in the mutant. The mutant seeds exhibited generally higher auxin levels and a large build-up of indole-3-aspartate when placed in germination conditions following moist-chilling. The mutant manifested increased levels of cytokinin glucosides through zeatin-O-glucosylation (Z-O-Glu). The resulting increase in Z-O-Glu was the largest and most consistent change associated with the ETR1 gene mutation. There were more gibberellins (GA) and at higher concentrations in the mutant than in wild-type. Our results suggest that ethylene signaling modulates the metabolism of all the other plant hormone pathways in seeds. Additionally, the hormone profiles of etr1-2 seed during germination suggest a requirement for higher than wild-type levels of GA to promote germination in the absence of a functional ethylene signaling pathway.


Ethylene sensing in Arabidopsis thaliana is mediated by the five members of the ethylene receptor family, ETR1, ETR2, ERS1, ERS2 and EIN4 (reviewed in Bleecker and Kende, 2000; Hall et al., 2000; Hua and Meyerowitz, 1998; Sakai et al., 1998; Schaller and Keiber, 2002). Structurally, these five receptors show the greatest similarity at their N-terminal transmembrane domains where ethylene binding is believed to occur (Gamble et al., 1998, 2002). A majority of the isolated mutations in the N-terminal transmembrane domains of the five receptors confer dominant ethylene insensitivity due to an inability of the receptor to bind ethylene or a failure of the receptor to propagate the signal to downstream effectors after ethylene perception has occurred (Hall et al., 1999; Schaller and Keiber, 2002). The latter appears to be the case with the etr1-2 mutation (Hall et al., 1999; Schaller and Bleecker, 1995).

Several components of the ethylene signal transduction pathway have been identified and, through the use of double mutants, their order of action has been determined (Bleecker and Kende, 2000; Sakai et al., 1998; reviewed in Schaller and Keiber, 2002). There is strong evidence that the receptors function as negative regulators of ethylene responses in the absence of ethylene (Hua and Meyerowitz, 1998). According to the now generally accepted model, in the absence of ethylene, the receptors are normally locked in a signaling-active mode and repress ethylene responses (Clark et al., 1998; Hall et al., 1999; Hua and Meyerowitz, 1998; Schaller and Bleecker, 1995). Upon ethylene binding, the receptors switch into a signaling-inactive state, which alleviates the repression on ethylene signal transduction and allows normal ethylene responses (Gamble et al., 2002). Thus, any receptor mutation that disrupts ethylene binding results in dominant ethylene insensitivity because the receptors are maintained in a signaling-active state. The observation that single knockout mutations in four of the five members of the ethylene receptor family have no impact on ethylene signal transduction led to the suggestion that these receptors are functionally redundant (Gamble et al., 2002; Hua and Meyerowitz, 1998). As suggested by Schaller and Bleecker (1995), the presence of mutant receptors that are locked in a signaling-active state due to a failure to bind ethylene would result in suppression of ethylene responses, even if the other wild-type members of the receptor family were sensing ethylene.

Genetic and biochemical analyses of ethylene-response mutants in Arabidopsis have provided invaluable information on the different components involved in ethylene signaling (Novikova et al., 2000). The etr1-2 mutation in Arabidopsis confers dominant ethylene-insensitivity and results in a larger proportion of mature seeds that exhibit primary dormancy – defined as the failure of mature seeds to germinate under conditions that are conducive to germination (Bewley, 1997) – compared with wild-type seeds (Beaudoin et al., 2000). Conversely, mutants with defects in abscisic acid (ABA) signaling pathways often produce seeds that are characterized by reduced dormancy, which is generally accompanied by a disruption of seed maturation and precocious expression of germinative/post-germinative genes. ABA originating from the developing embryo is required for dormancy inception during development of Arabidopsis seeds (Karssen et al., 1983; Leon-Kloosterziel et al., 1996). Furthermore, de novo synthesis of ABA occurs during imbibition of dormant mature seeds and contributes to dormancy maintenance (reviewed in Kermode, 2004). There are antagonistic effects of ethylene and ABA on dormancy and germination; thus, ethylene may promote germination by interfering with the inception and/or maintenance of seed dormancy by ABA (Ghassemian et al., 2000). For example, the greater proportion of dormant seeds in a population of mature seeds of ethylene-insensitive mutants (e.g. etr1), as compared with those of the wild-type, may be a consequence of their increased sensitivity to ABA (Beaudoin et al., 2000). Likewise, increased ABA biosynthesis in the ein2 mutant indicates that ethylene sensitivity and ABA biosynthesis are coupled in Arabidopsis (Ghassemian et al., 2000). There are varying opinions about the developmental stage during which ethylene [and other hormones such as gibberellin A (GA)] regulate dormancy and some suggest that their role during dormancy inception is minimal, and that their major action is during imbibition to terminate dormancy and/or initiate germination (Ghassemian et al., 2000). To add to the complexity, there are antagonistic interactions between ABA and GA, ethylene and brassinosteroid signaling pathways in the transition from dormancy to germination (Bentsink and Koornneef, 2002; reviewed in Finkelstein and Rock, 2002; Finkelstein et al., 2002). There is extensive cross-talk among the different hormone signaling pathways (Beaudoin et al., 2000; Gazzarina and McCourt, 2001; Ghassemian et al., 2000; Pitts et al., 1998), and between hormone response pathways and sugar signaling (Finkelstein et al., 2002). Thus, a key question concerns the nature and extent of overlap and cross-talk in hormone response networks in plants. How do plant cells integrate and coordinate their responses to developmental cues that involve hormone-dependent events? Do interactions occur at the level of signaling and/or through effects on hormone metabolism?

Some of these questions can be addressed by utilizing hormone-response mutants to investigate changes in hormone metabolism associated with a transition from one physiological state to another. Such a targeted approach should provide invaluable insight into the interactions of seemingly unrelated plant hormone pathways.

The studies described here provide information about the regulation of dormancy and germination by comparing hormone metabolism in dormant seed, during moist-chilling as well as during germination. In a previous study comparing hormone metabolism in germinating and thermodormant lettuce seed (Chiwocha et al., 2003), we showed that accumulation of abscisic acid glucose ester (ABA-GE) was pronounced in germinating seed. In addition, metabolism in imbibed thermodormant seeds showed unexpected fluctuations. The most dramatic fluctuations occurred after 7 days, at which point there were large, transient accumulations of dihydrophaseic acid (DPA) and indole-3-acetic acid (IAA).

To further our understanding about the regulation of the dormancy-to-germination transition, we focus on analyses of hormone metabolism in mature Arabidopsis seeds of wild-type and etr1-2 plants during dormancy termination and germination.


Germination profiles of wild-type and etr1-2 seeds of Arabidopsis

Freshly harvested wild-type (WT) and etr1-2 seeds were immediately imbibed under germination conditions for 10 days (warm-imbibed) in order to determine the degree of dormancy of the mature seeds. The population of wild-type seeds exhibited a lower percentage of dormancy than the population of mutant seeds and 88% germinated in the absence of a previous moist-chilling treatment in contrast to only 20% of etr1-2 seeds (Figure 1). Wild-type and etr1-2 seeds that had been moist-chilled for 7 days prior to their incubation in germination conditions exhibited 100% germination after 2.5 days (Figure 1). Thus, as previously reported, a larger proportion of mature etr1 seeds are dormant as compared with wild-type seeds; however, following moist-chilling, the germination capacities of the two populations are equivalent (Beaudoin et al., 2000).

Figure 1.

Germination profiles of wild-type (WT) and etr1-2 seeds. Germination percentages were scored in seeds that were placed directly under germination conditions without a moist-chilling period (warm-imbibed) or after a 7-day moist-chilling period.

Plant hormone and metabolite profiles of WT and etr1-2 seeds

Changes in hormones and hormone metabolites were analyzed in mature wild-type and etr1-2 seeds that had been imbibed immediately under germination conditions without a previous moist-chilling treatment, in mature seeds that had been subjected to a dormancy-breaking moist-chilling period, and during germination/early post-germinative growth of seeds. It should be noted that during the 2-day imbibition of wild-type seeds under warm conditions (no moist-chilling), only 27% of seeds germinated, so that the profiles obtained from these seeds derive predominantly from un-germinated seed. In contrast, following moist-chilling and subsequent exposure to warm (germination) conditions for the same period, the population of seeds is primarily comprised of germinated seeds. Importantly, the populations of etr1-2 and wild-type seeds germinate to an equivalent extent following a moist-chilling pre-treatment, although the rate of germination is slightly slower in the former (Figure 1, 7-day moist-chilling). The major differences in the hormone profiles of wild-type and mutant seeds are summarized below.

ABA and ABA metabolite profiles of WT and etr1-2 seeds

The ABA content of freshly harvested, mature dry seeds of the Columbia wild-type was 13 ng g−1 DW (Figure 2a). In contrast, ABA levels in mature dry seeds of the etr1-2 mutant were ≈10-fold higher, with a concentration of 120 ng g−1 DW (Figure 2b). ABA levels declined in warm-imbibed seeds to 3 and 17 ng g−1 DW within 6 h in the wild-type and etr1-2, respectively, and remained close to these levels thereafter (Figure 2a,b). In warm-imbibed wild-type seeds, the concentrations of DPA, 7′-hydroxy ABA (7′OH-ABA) and ABA-GE were all greater than endogenous ABA concentrations between 6 and 48 h although phaseic acid (PA) was not detected (Figure 2a). In contrast, DPA, ABA-GE and 7′OH-ABA were present at lower concentrations in warm-imbibed etr1-2 seeds. The biggest difference was observed in ABA-GE concentrations, which were around 10 ng g−1 DW at all time points in wild-type, but barely detected in etr1-2. Most interestingly, the decline in ABA concentration in imbibed etr1-2 seeds was not accompanied by a corresponding increase in any of the ABA catabolites monitored (Figure 2b); indeed, the concentrations of DPA and 7′OH-ABA decreased.

Figure 2.

ABA and ABA metabolite profiles of wild-type seeds (a) and etr1-2 seeds (b) following their incubation under germination conditions without a 7-day moist-chilling period.

During moist-chilling, the ABA levels in wild-type seeds declined (d1–d4), to a similar extent as that found in mature seeds imbibed under germination conditions without a moist-chilling treatment (Figures 2a and 3a). The concentration of ABA-GE increased markedly over the first day of moist-chilling, while other ABA metabolites (DPA and 7′OH-ABA) remained at steady-state levels throughout the moist-chilling period or declined slightly (Figure 3a). In etr1-2 seeds, there was a dramatic decrease in ABA concentration after 1 day of moist-chilling (Figure 3b) (as in warm-imbibed seed; Figure 2b), after which ABA levels remained relatively constant and at lower levels than in warm-imbibed seeds (Figures 2b and 3b). ABA-GE levels in dry and in 1-day moist-chilled etr1-2 seeds were much lower than in wild-type seeds, but increased at later times of moist-chilling, attaining a level similar to that of wild-type seeds after the 7-day period (Figure 3b).

Figure 3.

ABA and ABA metabolite profiles of wild-type (a) and etr1-2 (b) seeds during moist-chilling for 7 days at 3°C.

Following a 7-day moist-chilling period, germination percentages of wild-type versus etr1-2 seeds are similar after 36–48 h in germination conditions; however, during the first 24 h they differ considerably because of a slower rate of germination of mutant seeds (Figure 1). Despite this, ABA and ABA metabolites remained essentially constant in wild-type and etr1-2 seeds for at least the first 18 h in germination conditions (Figure 4). Thereafter, when most seeds of wild-type and etr1-2 plants completed germination, there were significant increases in the ABA and DPA contents of etr1-2 seeds (50 and 30 ng g−1 DW, respectively), which were not observed in the wild-type seeds. Despite the similar concentrations of ABA-GE in wild-type and etr1-2 seeds after the 7-day moist-chilling period, only wild-type seeds showed a marked accumulation of this metabolite during germination (Figures 3 and 4).

Figure 4.

ABA and ABA metabolite profiles during germination of wild-type (a) and etr1-2 seeds (b) initially exposed to a 7-day moist-chilling period prior to incubation under germination conditions.

Auxin profiles of WT and etr1-2 seeds

The auxin IAA and its metabolite indole-3-aspartate (IAAsp) were present at higher concentrations in mature dry seeds of the etr1-2 mutant compared with the seeds of wild-type plants (Figure 5). Upon imbibition, the IAA concentration increased moderately in wild-type seeds and markedly in etr1-2 seeds over the first 6 h; a decline occurred thereafter in seeds of both genotypes. IAAsp conjugate levels declined during imbibition in etr-1 seeds, but not in wild-type seeds. IAA and IAAsp were similar in seeds of both lines at d2 (Figure 5).

Figure 5.

Auxin profiles of wild-type seeds (a) and etr1-2 seeds (b) following their incubation under germination condition without a 7-day moist-chilling period.

The same general trend of changes in IAA and IAAsp concentrations that occurred when the dry wild-type and etr1-2 seeds were immediately imbibed under germination conditions for 2 days (Figure 5) also occurred during moist-chilling of these seeds, with the exception of an increase in IAA in wild-type seeds during the later stages (d4–d6) of moist-chilling (Figure 6).

Figure 6.

Auxin profiles of wild-type (a) and etr1-2 seeds (b) during moist-chilling for 7 days at 3°C.

IAA and IAAsp levels changed very little during germination of wild-type seeds (Figure 7a). The most striking difference between the genotypes was an enormous accumulation of IAAsp conjugate in etr1-2 seeds which increased ≈10-fold in the first 6 h of germination and remained consistently higher than IAA levels thereafter (Figure 7b).

Figure 7.

Auxin profiles during germination of wild-type (a) and etr1-2 (b) seeds initially exposed to a 7-day moist-chilling period prior to incubation under germination conditions.

Cytokinin profiles of WT and etr1-2 seeds

Mature dry seeds of wild-type plants contained significant amounts of the cytokinin isopentenyladenine (2iP) and of the two glycosylated forms of zeatin (the ribosyl- and O-glucosylated zeatins) (16, 9 and 7 ng g−1 DW, respectively; Figure 8a); the concentrations of both zeatin riboside (ZR) and zeatin-O-glucoside (Z-O-Glu) increased during imbibition (Figure 8a). In contrast to imbibed wild-type seeds, where ZR was the predominant cytokinin, the zeatin conjugate, Z-O-Glu, was the major cytokinin of etr1-2 seeds (Figure 8). Zeatin and isopentenyladenosine (IPA) were detected only in dry etr1-2 seeds; 2iP was also present in etr1-2 seeds, albeit at lower amounts than in wild-type seeds (Figure 8).

Figure 8.

Cytokinin profiles of wild-type seeds (a) and etr1-2 seeds (b) following their incubation under germination conditions without a 7-day moist-chilling period.

During moist-chilling, there were moderate accumulations of ZR, and Z-O-Glu in wild-type seeds over the first 4 days (Figure 9a). Z-O-Glu, the predominant cytokinin of etr1-2 seeds, increased during moist-chilling, as in the warm-imbibed seeds, but to a greater extent, reaching a maximum concentration of 110 ng g−1 DW after 7 days (Figure 9b).

Figure 9.

Cytokinin profiles of wild-type (a) and etr1-2 seeds (b) during moist-chilling at 3°C for 7 days.

Upon transfer of moist-chilled seeds to germination conditions, the concentrations of most of the predominant cytokinins and cytokinin conjugates in wild-type and etr1-2 seeds changed very little over the first 18 h (an exception was Z-O-Glu in etr1-2 seeds) (Figure 10a,b). Thereafter, when most seeds of wild-type and etr1-2 plants completed germination, there were significant increases in the endogenous levels of ZR, IPA and Z-O-Glu in wild-type seeds; in etr1-2 seeds, Z-O-Glu increased, and to a greater extent than in warm-imbibed and moist-chilled seeds (Figure 10). IPA was undetectable or virtually undetectable prior to the completion of germination of wild-type seeds (i.e. in directly imbibed and moist-chilled seeds; Figures 8a and 9a and during early incubation of previously moist-chilled seeds under germination conditions; Figure 10a). Thus, the cytokinin profiles of etr1 and wild-type seeds differ considerably.

Figure 10.

Cytokinin profiles during germination of wild-type (a) and etr1-2 seeds (b) initially exposed to a 7-day moist-chilling period prior to incubation under germination conditions.

Gibberellin profiles of WT and etr1-2 seeds

GA4 was the only GA detected in mature, dry wild-type seeds, in which it was present at a concentration of 25 ng g−1 DW. During imbibition, GA4 levels decreased approximately threefold by 48 h (Figure 11a). In contrast, dry seeds of the etr1-2 mutant contained GA1, GA4 and GA7, all at relatively high concentrations compared with the wild-type seeds (Figure 11b). All of the GAs abundant in dry etr1-2 seeds decreased during imbibition (particularly over the first 6 h); however, the GA4 and GA7 levels maintained in imbibed etr1 seeds were substantially higher than those of imbibed wild-type seeds (Figure 11).

Figure 11.

GA profiles of wild-type seeds (a) and etr1-2 seeds (b) following their incubation under germination conditions without a 7-day moist-chilling period.

During moist-chilling, the GA profiles of wild-type and etr1-2 seeds showed similar trends to those observed during warm-imbibition (Figures 11 and 12). However, moist-chilling of etr1-2 seeds led to greater decreases in the endogenous levels of active GAs than directly imbibing the seeds at germination temperatures.

Figure 12.

GA profiles of wild-type (a) and etr1-2 seeds (b) during moist-chilling at 3°C for 7 days.

GA1 accumulated in wild-type seeds over the first 18 h of germination (in contrast to warm-imbibed and moist-chilled seeds), but declined thereafter at a time when GA4 increased slightly; the amounts of GAs were generally maintained in germinating and germinated wild-type seeds at approximately 7–12 ng g−1 DW (Figure 13a). In contrast, etr1-2 seeds contained higher than wild-type levels of GA4 (as in warm-imbibed and moist-chilled seeds), even though the levels were relatively constant over the 48 h germination period (Figure 13b). Furthermore, mutant seeds also exhibited an accumulation of GA1 to higher levels than in wild-type seeds and, in contrast to wild-type, the levels of this compound continued to increase at 2 days (Figure 13a,b).

Figure 13.

GA profiles during germination of wild-type (a) and etr1-2 seeds (b) initially exposed to a 7-day moist-chilling period prior to incubation under germination conditions.


The analyses employed here provide a partial view of how a mutation in the ethylene signal transduction pathway can affect the levels of other major hormones and their metabolites in seeds. It should be noted that a difference in the amount of a metabolite when comparing two genotypes or treatments can arise from altered biosynthesis and/or altered catabolism and that static analyses cannot distinguish these possibilities without additional information. In addition, although the analyses are fairly extensive, they are not fully comprehensive (many metabolites are not measured), and therefore it is not possible to account for the metabolic fate of all compounds. Nevertheless, close examination of changes in the amounts of specific compounds that belong to the same hormone pathway can in some cases allow us to intimate the causes for such changes. Where possible, we have attempted to make the most reasonable interpretations of the data in this way.

ABA metabolism

It is tempting to attribute the higher percentage of dormancy in mature seeds of etr1-2 plants to higher than wild-type ABA levels. Clearly the etr1-2 mutation leads to hyper-accumulation of ABA during seed development, resulting in elevated levels in dry seed (Figure 2b). When mature etr1-2 seeds were placed in germination conditions with no previous moist-chilling period, a treatment which does not elicit high germination, ABA levels rapidly decreased but remained at significantly higher levels than those present in wild-type seeds. At the end of the dormancy-breaking, moist-chilling treatment, ABA levels remained higher in etr1-2 than in wild-type (by approximately twofold; Figure 3). This difference persisted after the seeds were transferred to germination conditions and may have contributed to the slower rate of germination of etr1-2 seeds (3% germination of stratified etr1-2 seeds after 18 h, versus 30% germination in stratified wild-type; Figure 1).

However, the data are not entirely consistent with an explanation of dormancy based solely on ABA concentration. Following dormancy termination, ABA levels in etr1-2 seeds at 18 h in germination conditions (approximately 10 ng g−1 DW; Figure 4b) were almost identical to ABA levels in non-germinating (warm-imbibed) etr1-2 seeds at 2 days (Figure 2b). Similarly, ABA levels were actually lower in etr1-2 seeds after 1 day of moist-chilling than in wild-type seeds (Figure 3), although the full 7-day moist-chilling period is required for complete dormancy-breakage of the former. Thus it may be more accurate to state that differences in dormancy status of mature wild-type seeds versus etr1-2 seeds are determined in part by the considerable differences in ABA turnover within these seeds upon imbibition. Moreover, any flux within the ABA pathway needs to be considered within the context of differences in ABA sensitivity (both as a consequence of the mutation as well as a consequence of treatment/developmental status of the seed).

The ABA sensitivity of seeds (including those of the etr1-2 genotype) must change during moist-chilling/germination; in non-dormant etr1-2 seeds most of the seeds complete germination at a time when endogenous ABA is accumulating. Mature seeds of the etr1-1 mutant and other ethylene insensitive mutants exhibit an increased sensitivity to exogenous ABA as far as the inhibition of germination is concerned (Ghassemian et al., 2000). Furthermore, the sensitivity of seeds to endogenous ABA is suggested to decrease during imbibition in response to ethylene and GA signaling (Beaudoin et al., 2000). Therefore, in the absence of moist-chilling (i.e. in warm-imbibed seeds), the pronounced dormancy of etr1-2 seeds as compared with wild-type seeds may be due to a combination of higher than wild-type levels of ABA in the seeds (Figures 2–4) and to an increased sensitivity of etr1-2 seeds to endogenous ABA due to defective ethylene signaling (Ghassemian et al., 2000). However, the extent to which the responses of etr1 seeds to exogenous ABA reflect sensitivity to the endogenous hormone is not known. In addition, defective ethylene signaling likely contributes to increased dormancy by other (ABA-independent) mechanisms.

ABA catabolism in Arabidopsis seeds occurred through the 8′- and 7′-hydroxylation pathways to DPA and 7′OH-ABA, respectively, as well as by conjugation to ABA-GE. Of these, the conjugation pathway, manifested by an accumulation of ABA-GE, appeared to be most closely connected to the germination of wild-type seeds. This is in contrast to fruits of Brassica species in which the 8′-pathway appears to be the predominant catabolic route leading to large accumulations of DPA (Zhou et al., 2003) and unlike lettuce seed in which 7′-hydroxylation appears to be insignificant (Chiwocha et al., 2003). However, a new pathway for ABA catabolism via hydroxylation at the 9′-position was recently reported (Zhou et al., 2004). In the present study, neo PA, the closed form of 9′OH-ABA, was not measured because the appropriate deuterium-labeled standard for quantification had not yet been synthesized.

In the etr1-2 mutant, although high levels of ABA were present in the dormant mature seeds, the dramatic decrease in ABA levels at 6 h in seeds placed in germination conditions with no previous moist-chilling treatment was not accompanied by increases in the ABA metabolites quantified in this study (Figure 2b). It is possible that the decrease was due to ABA leaching into the surrounding medium or, as suggested above, via metabolism through the 9′-hydroxylation pathway. It is also likely that ABA is metabolized through known pathways but metabolites are being further converted. This must be the case for ABA-GE, for example, where levels rise and then fall during moist-chilling of wild-type seeds.

The conjugation of ABA to ABA-GE appeared to be significant as far as germination capacity is concerned and was clearly perturbed in etr1-2 seed. Moist-chilling of wild-type seed led to a greater increase of ABA-GE as compared with direct imbibition of seeds (Figures 2 and 3) and the amount of this metabolite increased dramatically to more than 150 ng g−1 DW during germination (Figure 4). Similarly, in lettuce seed ABA-GE appeared during germination (to about 50 ng g−1 DW) but was at very low levels in imbibed thermodormant (non-germinating) seed (Chiwocha et al., 2003). As amounts of ABA and of other ABA catabolites change little when ABA-GE levels increase, it seems likely that ABA metabolism is channeled specifically into conjugate formation during dormancy-breakage (moist-chilling) and during germination (although the possibility that ABA-GE is released from a subsequent metabolite of the conjugate cannot be eliminated). It is unclear why this pathway would be required for germination/growth. ABA-GE is generally considered to be an irreversibly inactivated conjugate of ABA but recent studies have suggested that ABA-GE can serve as a transported form of hydrolyzable ABA (Sauter et al., 2002). Therefore, if this is the case, it is possible that germination (or normal growth) may require ABA to be distributed in the growing tissues in an inactive form from which it can be rapidly released in the event of adverse environmental conditions. A certain concentration of endogenous ABA is required for normal growth. The ABA-deficient Arabidopsis mutants aba1-1, aba1-3, and aba1-4 are stunted in their growth due to a reduced capacity to transpire and establish turgor. When these mutants are sprayed with exogenous ABA, cell expansion and normal growth are restored (Finkelstein and Rock, 2002).

Loss of ethylene signaling in the etr1-2 mutant reduces the accumulation of ABA-GE relative to wild-type in all three treatments. Therefore, it is likely that reduction in this catabolic pathway is a significant contributor to elevated ABA levels in etr1-2. During moist-chilling of etr1-2, there is a delayed and less pronounced accumulation of ABA-GE relative to that in wild-type seeds, although the amounts of this metabolite are comparable at the end of the 7-day moist-chilling period. During germination there is a significant increase in ABA-GE, although to much lower levels than in wild-type. However, the amount of ABA-GE in germinating etr1-2 seeds is high compared with warm-imbibed (largely non-germinating) etr1-2 seeds. Therefore, it appears that even in etr1-2 there is an association of ABA-GE with germination.

Altered ABA metabolism during germination (as suggested above) was supported by a striking difference in the ABA metabolites that accumulated in germinating etr1-2 seeds compared with wild-type seeds. Notably, this included: (1) a reduced (although still significant) accumulation of ABA-GE in germinating etr1-2 seeds (Figure 4); (2) a substantial increase in ABA (to about 50 ng g−1 DW) which occurred exclusively in germinating etr1-2 seeds and (3) an increase in 8′-hydroxylation (possibly a relatively minor pathway of ABA metabolism in germinating wild-type seeds) during germination of etr1-2 seeds leading to an accumulation of DPA (to about 30 ng g−1 DW). This suggests that there was a loss of capacity of the ABA-GE pathway to metabolize de novo synthesized ABA in etr1-2 seed, resulting in ABA accumulation and partial compensation by increased metabolism through the 8′-hydroxylation pathway to DPA.

Auxin metabolism

Most plant tissues contain the majority (approximately 90%) of their IAA in conjugated forms, with only minute amounts of free IAA (Cohen and Bandurski, 1982). In the present study, we confined our analyses to monitoring IAA and one conjugate, IAAsp. Mature dry seeds, imbibed immediately or subjected to moist-chilling, exhibited an early and transient peak of IAA that was far more pronounced in etr1-2 seeds than in wild-type seeds (Figures 5 and 6), which would appear to suggest that the increase is unrelated to germination and may be the result of changes during early imbibition and/or to changes in seed environment. However, transfer of moist-chilled seeds to germination conditions was also accompanied by a rapid and transient peak of IAA (Figure 7). Indirect evidence indicating the involvement of auxins in the germination process comes from a recent study by Ogawa et al. (2003). Their microarray analysis of germinating Arabidopsis seeds revealed the upregulation of a number of auxin biosynthesis genes and genes encoding auxin carrier proteins in response to exogenous GA4 application (Ogawa et al., 2003).

Defective ethylene signaling appears to alter the balance between biosynthesis and catabolism of IAA in favor of catabolism; one of the most striking effects of the etr1-2 mutation was manifested in a marked accumulation of IAAsp in etr1-2 seeds during germination, which reached a maximum of almost 500 ng g−1 DW and remained high at a time when most seeds completed germination (Figure 7). The conjugates of IAA appear to be involved in a variety of hormonally related processes such as: transport of IAA within the plant, the storage and subsequent reuse of IAA, protection of IAA from enzymatic degradation, as components of a homeostatic mechanism for the control of IAA levels and as an entry route into the subsequent catabolism of IAA (reviewed in Bartel, 1997). Certain conjugates, such as IAAsp, can be intermediates in IAA destruction as a precursor to the irreversible catabolic oxidation of IAA (Normanly, 1997); therefore accumulation of IAAsp suggests that degradation of IAA is favored during germination. IAAsp levels exhibit a further increase in etr1-2 seeds (e.g. at d2 in germination conditions) suggesting that further metabolism of IAAsp is rate limiting during germination/early post-germinative growth of etr1-2 seeds. The wild-type ETR1 gene therefore appears to generally suppress IAA accumulation, at least in part by increasing conjugation to IAAsp and favoring further catabolism of the conjugate. However, other interpretations cannot be ruled out. For example, other conjugates and catabolites are present in Arabidopsis seeds that we did not measure (Tam et al., 2000). Ethylene signaling might determine the metabolic route in favor of IAAsp formation, but not necessarily alter the amount of IAA that is synthesized.

Cytokinin metabolism

The cytokinin profiles of etr1-2 and wild-type seeds differed considerably. In wild-type seed, ZR was the most abundant cytokinin except in dry seed (in which 2iP was highest) and in seeds that were completing germination (Figure 10a). In the latter case, a striking accumulation of Z-O-Glu was apparent (increasing from <10 ng g−1 DW at 18 h to almost 150 ng g−1 DW at 48 h) together with substantial increases in ZR and IPA. These results suggest that an increase in cytokinin metabolism is associated with the completion of germination and the initiation of post-germinative growth. O-glucosylated cytokinins are resistant to the cleavage of the N6-side chain by cytokinin oxidases; these compounds are considered to be stable storage forms that play an important role in balancing cytokinin levels (Haberer and Kieber, 2002). Z-O-Glu is a stored form of inactive cytokinin that can be re-mobilized to the active aglycone by β-glucosidase action (Mok and Mok, 2001). Therefore, the accumulation of Z-O-Glu suggests a requirement for production of an inactive form of zeatin that can be re-mobilized for post-germination growth.

In all etr1-2 samples Z-O-Glu was the predominant cytokinin, which was present in large quantities (amounts ranging from 50 to 200 ng g−1 DW). Amounts of other, biologically active cytokinins were lower than in wild-type. These observations suggest that defective ethylene signaling resulted in increased zeatin conjugation with a consequent depletion of the biologically active cytokinins (2iP and ZR) normally detected in seed. This was particularly striking during germination when a general increase in cytokinins occurred in wild-type. Therefore, one of the effects of ethylene in wild-type seeds may be to suppress cytokinin metabolism through zeatin O-glucosylation allowing greater accumulation of biologically active hormones. Communication between ethylene and cytokinin signaling through a phosphorelay signaling pathway has been suggested (Hutchison and Kieber, 2002).

Gibberellin metabolism

Although 126 GAs have been identified in higher plants, fungi and bacteria, only a small number are biologically active (e.g. GA1, GA3, GA4 and GA7) (Hedden and Phillips, 2000; Olszewski et al., 2002). Many of the other GAs are biosynthetic intermediates or catabolites of bioactive GAs. Only GA4 was detected in dry seed of wild-type and its levels decreased after imbibition (Figures 11a and 12a). After moist-chilling, seeds that were competent to germinate transiently produced GA1 in addition to GA4 (Figure 13a). There is a well-known association of GAs with seed germination and the effects of GAs are antagonistic to those of ABA (Bentsink and Koornneef, 2002; reviewed in Finkelstein and Rock, 2002; Finkelstein et al., 2002). However, there was no significant net increase in GAs during moist-chilling or germination. The amounts of GAs were generally maintained in germinating and germinated wild-type seeds at approximately 7–12 ng g−1 DW. Surprisingly, dormant etr1-2 seed contained higher amounts of GA4 in addition to GA1 and GA7. All of the GAs decreased upon imbibition in warm-imbibed or moist-chilled seeds (Figures 11b and 12b). Germination and early post-germinative growth of etr1-2 seeds were accompanied by a marked accumulation of GA1; the accumulation of GA1 was greater in magnitude and was extended over a longer period of time, perhaps as a consequence of the slower rate of germination of etr1-2 seeds as compared with wild-type seeds. The production of GA1 during germination may be enabled by changes elicited during moist-chilling. Such is the case in hazel seeds (Corylus avellana) – where chilling has a marked effect on the capacity for GA biosynthesis. GA production (substantial increases in GA1 and GA9) does not take place during moist-chilling itself, but only when the chilled seed subsequently experiences higher (germination) temperatures (Williams et al., 1974). The generally elevated level of GAs in etr1-2 seed suggests that ethylene normally acts to suppress GA production. There is previous evidence for an inter-dependence of GA and ethylene metabolism, for example, Ogawa et al. (2003) reported the upregulation of ethylene biosynthesis genes, as well as the ERS1 ethylene receptor gene, by exogenous GA4 in imbibed Arabidopsis seeds. Therefore, a feedback regulatory loop may exist in which GA stimulates ethylene production, and ethylene represses GA production.

A hormone balance theory has been invoked in which ABA and GA act antagonistically to control both dormancy breakage and germination (Karssen, 1995; Karssen and Lacka, 1986 and references therein). Generally GAs are viewed as important for the promotion and maintenance of germination, while ABA controls seed developmental events including the inception of dormancy (Bewley, 1997). However, GA also appears to act as an antagonist to ABA function during seed development (White et al., 2000). When an ABA-deficient mutant of maize (vp5) is manipulated either genetically or via biosynthesis inhibitors to induce GA-deficiency during early seed development, vivipary is suppressed in developing kernels and the seeds acquire desiccation tolerance and storage longevity. In the presence of ABA, then, de novo GA biosynthesis may be necessary to promote germination (see Bentsink and Koornneef, 2002).

Although the precise mechanism by which GA counteracts the inhibitory effect of ABA on germination is unclear, there is some evidence that it may, in part, involve GA downregulating ABRE-containing genes in Arabidopsis (Ogawa et al., 2003). Thus, because etr1-2 seeds had higher levels of ABA than wild-type seeds, it is conceivable that the mutant may attempt to compensate by accumulating higher than wild-type levels of endogenous GA – both during seed development and during the completion of germination. The biggest differences between mutant and wild-type seeds were after imbibition of dry seed (Figures 11 and 12) and early GA-regulated events taking place after imbibition of dry seed may be important for facilitating germination. There is some evidence consistent with this in the literature. For example, high levels of GA3 and GA20 in non-imbibed seeds of Onopordum nervosum may be involved in seed dormancy breakage (Fernández et al., 2002).


Given that the etr1-2 mutation confers ethylene insensitivity, we reasoned that any differences in the metabolic profiles of wild-type seeds and etr1-2 mutant seeds after they had been exposed to the same conditions would be due to the mutation. The results of the present study indicate that the ethylene signaling pathway is involved in modulating metabolism through other plant hormone pathways during seed maturation and germination/early post-germinative growth in Arabidopsis.

A model summarizing the impact of the etr1-2 mutation on plant hormone metabolism during the dormancy-to-germination transition is provided in Figure 14. The results indicate that defective ethylene signaling has unexpected and sometimes dramatic effects on metabolism in other hormonal pathways.

Figure 14.

Schematic representation of the impact of the etr1-2 mutation during germination in Arabidopsis thaliana. Plant hormones and hormone metabolites that increased in the etr1-2 mutant are in blue. Compounds that were present at higher levels in wild-type seeds are in brown. Compounds that were not detected or that did not show a significant change in concentration are in black. Underlined compounds are precursors, which were not quantified in this study. Note: Only hormones and hormonal metabolites that were investigated in this study are shown in the auxin, abscisic acid, gibberellin and cytokinin pathways. In addition, although not shown in the above model, it should be pointed out that alternative biosynthetic pathways for cytokinins (Åstot et al., 2000) and the auxin IAA (Normanly et al., 1995) do exist in Arabidopsis.

With respect to ABA metabolism, the etr1-2 mutation resulted in elevated ABA levels both during seed maturation and during germination. Reduced ABA catabolism via conjugation to ABA-GE may have contributed to the increased dormancy characteristic of mature etr1 seeds, because the product of this pathway was most noticeably absent under conditions that were ineffective in terminating dormancy (i.e. when mutant seeds were imbibed at germination temperatures with no preceding moist-chilling period). During germination, the etr1-2 seeds compensated for their reduced turnover of ABA via conjugation by increasing the oxidation (8′-hydroxylation) pathway, but presumably this was insufficient to prevent the accumulation of ABA. Although highly speculative, it could be argued that the production of relatively large quantities of hormonally inactive ABA-GE during germination of Arabidopsis seeds (which also occurs in lettuce; Chiwocha et al., 2003) may supply young seedlings with a reserve of ABA that can be mobilized if necessary.

The mutant exhibited generally higher auxin levels with a large build-up of IAAsp during germination. Increased cytokinin conjugation through Z-O-Glu was the largest and most consistent change manifested by the mutant and resulted in a decrease in the levels of biologically active cytokinins. Synthesis of more gibberellins, which were present at higher concentrations, was a further manifestation of the mutant and may have occurred to compensate for the elevated ABA levels in these seeds. The resulting restoration of a more normal GA/ABA balance may have facilitated the successful germination of the moist-chilled mutant seeds upon their transfer to germination conditions.

Hormone response mutants are an invaluable resource for investigating the nature and degree of overlap and cross-talk in plant hormone response networks. From this study, we can conclude that interpretations of genomic and proteomic results (in this and any other hormone response mutants) must take into account the effects of the mutation on all hormone pathways. We anticipate that such an integrated approach will prove to be invaluable in furthering our understanding of hormone action in plants.

Experimental procedures

Plant material and germination testing

Arabidopsis thaliana ecotype Columbia WT and etr1-2 seeds (kindly supplied by Peter McCourt, University of Toronto) were germinated and the plants grown under 125–150 μmol m−2 sec−1 constant light at 22°C with 70% relative humidity. Mature seeds were harvested and stored at 4°C when approximately 90% of the plants had senesced. Germination testing and collection of material for analysis was initiated shortly after harvest. To determine germination capacity, wild-type and mutant seeds were either directly imbibed under germination conditions without any dormancy-breaking treatment (warm-imbibed) or exposed to 7 days of moist-chilling at 3°C to break dormancy prior to their transfer to germination conditions. Germination percentages were scored every 12 h for the first 4 days and daily thereafter for up to 10 days using three replicates of 50 seeds each. Germination conditions were 25°C with 16 h light at approximately 75 μmol m−2 sec−1.

Sampling of material for hormone and hormone metabolite analyses

For each time-point analyzed, triplicate samples of ≈50 mg of seed each (the exact weight was recorded) were imbibed in Petri plates lined with Whatman's No. 1 filter paper which had been moistened with 2 ml double-distilled sterile water. Samples were collected from warm-imbibed seeds at 0, 6, 18 and 48 h after incubation under germination conditions. During the dormancy-breaking treatment, seeds were collected after 1, 4 and 7 days of moist-chilling. Seeds were also collected during germination/growth at 6, 18 and 48 h, after previously being subjected to moist-chilling for 7 days. The samples were immediately frozen in liquid N2, stored at −80°C and later lyophilized for 48 h.

Extraction of plant hormones and metabolites

The extraction procedure was as previously described in Chiwocha et al. (2003). Briefly, lyophilized samples were ground, a mixture of deuterium-labeled internal standards added to triplicate samples of seeds for each time point, and extraction performed using acidic isopropanol. Particulate matter and other unwanted components were removed by solid phase extraction (SPE) with C18 SepPak cartridges (Waters, Mississauga, ON, Canada).

Analysis of endogenous plant hormones and metabolites by HPLC-ESI/MS/MS

The procedure used for quantification of endogenous auxins (IAA and IAAsp), abscisic acid and metabolites (ABA, PA, DPA, 7′OH-ABA and ABA-GE), gibberellins (GA1, GA3, GA4 and GA7) and cytokinins (2iP, IPA, Z, ZR) is as described in Chiwocha et al. (2003), except that three additional cytokinins (CK), namely dihydrozeatin (DHZ), dihydrozeatin riboside (DHZR) and Z-O-Glu, were incorporated in the method. The deuterium-labeled internal standards for these CK were d3-DHZ, d3-DHZR and d5-Z-O-Glu, respectively (Olchemim Ltd, Olomouc, Czech Republic).

HPLC-MS/MS of plant extracts was performed on a 100 mm × 2.1 mm, 4 µm Genesis C18 HPLC column (Jones Chromatography, Hengoed, UK) by gradient elution with acetic acid (pH 3.3) in, aqueous acetonitrile of increasing organic content. The retention times of the three additional cytokinins and their respective internal standards under the HPLC conditions employed were: 9.33 min for DHZ and d3-DHZ; 11.17 min for DHZR and d3-DHZR; and 9.04 min for Z-O-Glu and d5-Z-O-Glu. The precursor- to product-ion transitions used to monitor and quantify the three CK in the positive ionization mode were as follows: DHZ (m/z 222 > 136), d3-DHZ (m/z 225 > 136), DHZR (m/z 354 > 222), d3-DHZR (m/z 357 > 225), Z-O-Glu (m/z 382 > 220), and d5-Z-O-Glu (387 > 225). Multiple reaction monitoring (MRM) calibration curves were generated from standard solutions using the ratio of the chromatographic peak area for each analyte to that of the corresponding internal standard. For the three additional CK, 20 ng of each deuterated internal standard was added to the extraction solvent along with a mixture of the deuterium-labeled internal standards for the other compounds prior to sample extraction (Chiwocha et al., 2003). Thus, the endogenous levels of 18 plant hormones and metabolites could be analyzed simultaneously in each sample. Quality control (QC) samples containing equal amounts (20 ng) of each analyte and internal standard were also prepared and analyzed along with the tissue samples (Ross et al., 2004). The mean QC analyte concentration was 23.8 ± 2.0 ng/sample, with relative errors of <15% for Z-O-Glu, DHZR, IAA, IAAsp, GA1, PA, DPA, IPA and 2iP and <30% for DHZ, ABA, GA3, GA4, GA7, ABA-GE, 7′OH-ABA, Z and ZR.


We would like to extend our thanks to Peter McCourt (University of Toronto) for providing the wild-type (Columbia) and etr1 seeds and Allan Feurtado for his help with the growth of plants to obtain seed. We would like to acknowledge the hard work of the following chemists at PBI for synthesizing all the standards for compounds in the ABA pathway: Ken Nelson, Marek Galka, Irina Zaharia, Vera Cekic and Yuanzhu Gai. We would also like to thank Dick Pharis (University of Calgary) for helpful discussions of the GA results. This research was supported by a Protein Engineering Network of Centres of Excellence (PENCE) grant awarded to Allison R. Kermode (Simon Fraser University), Suzanne R. Abrams, Adrian J. Cutler, Andrew R.S. Ross (Plant Biotechnology Institute), Sharon Regan (Queens University) and Peter McCourt (University of Toronto).