The phytohormone abscisic acid (ABA) inhibits the germination of many seeds, including Arabidopsis, but the mechanism for this is not known. In cereals, ABA inhibits the expression of genes involved in storage reserve mobilization. We have found that in Arabidopsis ABA decreases transcription from the promoters of marker genes for β-oxidation and the glyoxylate cycle, essential pathways for the conversion of storage lipid (triacylglycerol) into sucrose. Thirty per cent of stored lipid is broken down over 6 days following imbibition of ABA-treated seed. Sucrose levels in ABA-treated seeds, rather than decreasing as under normal growth conditions, actually double during the 3 days following imbibition. This sucrose is derived from triacylglycerol as demonstrated by two mutants disrupted in the conversion of triacylglycerol into sucrose, kat2 and icl1, which do not accumulate sucrose in the presence of ABA. We conclude that the ABA block on germination is not a consequence of inhibition of storage lipid mobilization. Two independent programmes appear to operate, one that is blocked by ABA, governing developmental growth resulting in germination; and a second that governs storage lipid mobilization which is largely ABA-independent.
Seed germination and storage reserve mobilization are highly controlled processes regulated by environmental factors such as light and temperature, and by development (Bewley and Black, 1994). Following desiccation seeds enter a period of dormancy during which germination will not occur following imbibition, even though environmental conditions may be favourable. When endogenous changes release the seed from dormancy, germination begins with imbibition and is concluded by the emergence of the radicle from the seed coat (Bewley, 1997). Germination is shortly followed by the mobilization of food reserves from the seed storage organs or endosperm, providing essential energy to fuel growth until the seedling becomes photoautotrophic.
Seed storage reserves consist of complex molecules of carbohydrate, protein or lipid (Bewley and Black, 1994). During early post-germinative growth, storage reserves are broken down into soluble molecules such as sucrose which can be transported around the developing seedling. In oilseeds the major storage reserve is lipid, found in the form of triacylglycerol (TAG) (Bewley and Black, 1994). TAG is initially cleaved by lipases, releasing fatty acids that are subsequently broken down by the enzymes of β-oxidation and the glyoxylate cycle (Beevers, 1961; Kornberg and Beevers, 1957). The expression of the genes encoding these enzymes follows a strictly temporal pattern, peaking 1–2 days after imbibition and then falling to low or undetectable levels (Comai et al., 1989; Eastmond and Graham, 2000; Eastmond et al., 2000a; Germain et al., 2001; Graham et al., 1990). Although closely related temporally, germination and storage reserve mobilization are distinct events. While the products of reserve mobilization are required to fuel post-germinative growth, analysis of the β-oxidation mutant ketoacyl CoA thiolase-2 (kat2) (Germain et al., 2001), which is allelic to the ped1 mutant (Hayashi et al., 1998), has shown that fatty acid breakdown is not a requirement for germination. kat2 seeds germinate and then stall in development, but can be rescued by an exogenous carbohydrate source such as sucrose (Germain et al., 2001).
In addition to their role during germination, phytohormones are known to control the mobilization of storage reserves in cereals (Fincher, 1989; Lovegrove and Hooley, 2000). It has been shown that ABA and GA regulate the expression of genes encoding the enzymes required for storage protein and carbohydrate mobilization in rice, wheat and barley, and potential signal transduction components have been identified (Appleford and Lenton, 1997; Cercos et al., 1999; Watanabe et al., 1991). Promoter analysis of the barley cysteine proteinase gene EPB-1 has revealed cis-elements for both ABA- and GA-mediated gene expression. The expression of a GA-Myb transcription factor, which interacts with a GA-response complex in the promoter of EPB-1, is induced by GA and repressed by ABA (Cercos et al., 1999). There is also evidence for the involvement of an ABA-inducible protein kinase (PKABA1) in the inhibition of GA-mediated induction of gene expression in barley and wheat aleurone (Gomez-Cadenas et al., 1999). Overexpression of PKABA1, but not of other protein kinases, blocks the expression of α-amylase– and cysteine proteinase–GUS reporter constructs in aleurone cells (Gomez-Cadenas et al., 1999).
We are interested in understanding the regulatory mechanisms that control the expression of genes involved in TAG breakdown during early post-germinative growth in Arabidopsis. It is possible that the phytohormonal control of reserve mobilization seen in cereals is conserved in dicots, and there is some evidence to support this. Marriott and Northcote (1977) demonstrated that exogenous application of 10 µm ABA inhibits the activity of the glyoxylate cycle enzyme isocitrate lyase (ICL) in castor bean endosperm, an effect that is reversible by treatment with GA. In addition, it has been shown that exogenous ABA treatment results in reduced ICL activity following imbibition in Brassica rapa (Finkelstein and Lynch, 2000b).
Exogenous ABA application inhibits the germination of seeds of many species. In Arabidopsis, it has been observed that sugars or peptone can rescue ABA-mediated inhibition of germination (Garciarrubio et al., 1997). This effect is apparently specific to germination, as the negative effects of ABA on other aspects of seedling growth, such as the development of the first true leaves, are not rescued by exogenous sugars or peptone (Finkelstein and Lynch, 2000b). This observation led to the hypothesis that ABA inhibits germination by preventing the mobilization of storage reserves, thus blocking the supply of energy and nutrients to the developing seedling. The provision of an alternative food supply would allow the seedling to germinate, although the endogenous reserves could not be utilized. This hypothesis suggests that the regulatory mechanisms controlling storage reserve mobilization are conserved between monocots and dicots.
With the aim of furthering our understanding of the regulation of storage reserve breakdown in Arabidopsis, we have investigated the effects of ABA on germination and reserve mobilization. We disprove the hypothesis that ABA inhibits the germination of Arabidopsis seeds by preventing storage reserve mobilization. Indeed there is active lipid mobilization in the presence of ABA and the product of this process, namely sucrose, actually accumulates in the ABA-treated seeds.
Expression of β-oxidation and glyoxylate cycle genes is inhibited in ABA-treated seeds
It has previously been reported that Arabidopsis seeds prevented from germinating by ABA treatment show complete inhibition of protein breakdown (Garciarrubio et al., 1997). To determine the effects of ABA treatment on storage lipid mobilization, the expression of key genes of β-oxidation and the glyoxylate cycle was investigated in the presence and absence of ABA.
A promoter-trapped GUS reporter line for the medium-chain acyl CoA oxidase gene (ACX3) (Eastmond et al., 2000a) was used to examine the effect of ABA on β-oxidation gene expression (Figure 1a,b). For the first 2 days following imbibition GUS activity in ABA-treated seeds does increase, but only to levels that are half that of seeds germinating in the absence of ABA. GUS activity does not increase further in ABA-treated seeds following day 2. Histochemical staining of ACX3::GUS seeds (Figure 1b) confirms that reporter gene expression is induced in non-germinated seeds in the presence and absence of ABA at day 1, but does not increase after day 2 in non-germinating ABA-treated seeds.
Seeds from a transgenic malate synthase (MS):: luciferase reporter line were used to determine the effect of ABA treatment on this key glyoxylate cycle marker gene. This reporter line was constructed using 969 bp of the MS promoter from −3 bp relative to the ATG start site. The typical expression pattern of luciferase in these lines was consistent with that observed by Northern analysis of the Arabidopsis MS gene (Eastmond and Graham, 2001) which is present as a single copy in the Arabidopsis genome. In the absence of ABA, expression of the luciferase reporter gene is strongest 1 day after imbibition and is barely detectable by day 3 (Figure 1c). In the presence of ABA, the peak of reporter gene expression also occurs at day 1, but luciferase activity is reduced to one-third of that in seedlings in the absence of ABA. In contrast to seeds germinating in the absence of ABA, ABA-treated seeds continue to express low but significant levels of luciferase for at least 12 days after imbibition (data not shown).
The reporter gene studies demonstrate that ABA treatment does inhibit the expression of genes encoding enzymes of β-oxidation and the glyoxylate cycle; however, significant levels of expression still occur.
Seeds treated with ABA are able to break down storage lipids
Having observed that the expression of genes required for the breakdown of storage lipid is inhibited in ABA-treated seeds, fatty acid levels were measured by gas chromatography analysis to determine whether TAG breakdown occurs in the presence of ABA. Levels of eicosenoic acid (20 : 1), a marker for TAG (Lemieux et al., 1990), were compared in wild-type (Col0) seeds treated with and without ABA, 6 days after imbibition (Figure 2). In addition, 20 : 1 levels were measured in seeds of the thiolase mutant kat2 (WS background), and in wild-type (WS) seeds, 5 days after imbibition (Figure 2). kat2, which is disrupted in a gene encoding the β-oxidation ketoacyl CoA thiolase enzyme, exhibits limited germination (radicle emergence) and seedling establishment occurs only in the presence of exogenous sugar (Germain et al., 2001).
Under normal conditions, the majority of storage lipid in wild-type seeds is broken down in the first few days after imbibition, with <5% of dry seed 20 : 1 levels remaining after 6 days. Wild-type seeds treated with 10 µm ABA are completely blocked in germination, yet show a 30% reduction in 20 : 1 levels 6 days after imbibition. This is consistent with a reduction in the expression of genes involved in lipid breakdown, as determined by the reporter gene analysis. In contrast, kat2, which does germinate but arrests after radicle emergence (Germain et al., 2001), shows no significant reduction in 20 : 1 levels 5 days after imbibition.
Seeds treated with ABA accumulate sucrose
Inhibition of TAG breakdown as a result of ABA treatment may cause a reduction in energy and nutrient supplies to the embryo sufficient to prevent germination. To address this question, sugar levels were measured in germinating seedlings as an indicator of metabolic status. Sucrose and glucose levels were measured over the 3 days following imbibition in Col0 and WS seeds treated with and without ABA (Figure 3). Immediately after imbibition, seeds contain around 150–230 ng sucrose (open bars), which decreases steadily in the days following imbibition in the absence of ABA (left panels). At the same time, glucose levels (black bars), which are less than 50 ng per seed in imbibed seeds, increase to around 220 ng per seedling by day 3. In contrast, seeds treated with ABA (right panels) accumulate sucrose following imbibition, with levels reaching around 350–450 ng per seed by day 3. Glucose levels in ABA-treated seeds remain low but do not decrease below levels found in day 0 imbibed seeds in the presence or absence of ABA.
Accumulated sucrose is a product of TAG breakdown
Sucrose accumulated in imbibed seeds may result from the interconversion of sugars present within the seed. However, sucrose may also accumulate as a result of the mobilization of seed storage reserves. To determine whether sucrose accumulating in ABA-treated seeds is derived from TAG breakdown, sugar levels were measured in icl1 and kat2 seeds treated with and without ABA (Figure 3). Both these mutants are unable to convert carbon stored in TAG into sucrose (Eastmond et al., 2000b; Germain et al., 2001). As previously mentioned, kat2 is disrupted in β-oxidation. The icl1 mutant is disrupted in the glyoxylate cycle, which is an essential part of the gluconeogenic conversion of acetyl CoA derived from fatty acid β-oxidation into sucrose in germinating oilseeds. However, unlike kat2, icl1 is able to germinate and establish in the absence of exogenous sugar under long-day conditions (Eastmond et al., 2000b). Both icl1 and kat2 seeds are blocked from germinating by ABA treatment.
Similarly to the wild-type seeds, sucrose levels in the two mutants decrease following imbibition in the absence of ABA. By day 3, 25% of kat2 seeds have germinated (radicle emergence) and half the available sucrose present at day 0 is remaining (Figure 3, left panel). Ninety eight per cent of icl1 seeds have germinated and established by day 3, and sucrose is undetectable (Figure 3, left panel). Germination is completely blocked in ABA-treated kat2 and icl1 seeds. In contrast to wild-type seeds, icl1 and kat2 fail to accumulate sucrose in the presence of ABA, and in fact sucrose levels in both mutants decrease following imbibition in the presence of ABA (Figure 3, right panel). However, sucrose levels decrease more slowly in icl1 seeds in the presence of ABA than in its absence, indicating a reduction in sucrose utilization in the presence of ABA. These results demonstrate that accumulated sucrose is derived from TAG breakdown in ABA-treated seeds.
Sucrose accumulation is reduced in abi mutants treated with ABA
The ABA-insensitive mutants abi3-1 and abi4-103 germinate and establish in the presence of 10 µm ABA (Finkelstein, 1994; Koornneef et al., 1984). However, growth of the mutants is slower in the presence than in the absence of ABA, and the percentage of seeds that germinate is reduced. Three days after imbibition in the presence of ABA, 22% of abi3-1 and 37% of abi4-103 seeds have germinated. The physical appearance of representative Col0, Ler, abi3-1 and abi4-103 seeds and seedlings 3 days after imbibition is shown in Figure 4(a) (in the absence of ABA) and in Figure 4(b) (in the presence of ABA). Sugar assays were used to investigate the effect of ABA on sugar levels in abi3-1 and abi4-103 seedlings. Sugar levels were measured in mutant and wild-type seedlings 3 days after imbibition in the absence (Figure 4a) and presence (Figure 4b) of ABA. Separate samples containing only germinated or non-germinated seeds were taken for abi3-1 in the absence of ABA, and abi3-1 and abi4-103 in the presence of ABA. In the absence of ABA, glucose levels in both mutants resemble those in wild-type seeds. Sucrose levels are increased in abi3-1 and abi4-103 seedlings. Both germinated and non-germinated samples of abi3-1 and abi4-103 have increased levels of sucrose in the presence of ABA. In the case of abi3-1, higher sucrose levels were detected in germinated seedlings than in non-germinated seeds. However, sucrose levels in abi3-1 and abi4-103 samples are significantly lower than those in wild-type seeds treated with ABA.
Our experiments demonstrate that Arabidopsis seeds are metabolically active in the presence of ABA, that the signals initiating reserve mobilization are present, and that the breakdown of storage reserves and accumulation of sucrose occurs, although germination does not.
Expression analysis of the ACX3::GUS and MS::luciferase reporter lines demonstrates that genes encoding enzymes of β-oxidation and the glyoxylate cycle are expressed in non-germinating ABA-treated seeds. The timing of the induction of reporter gene expression is not affected by ABA treatment; however, there is a 50–65% reduction in the peak of expression reached 1–2 days after imbibition. Continued monitoring of the MS::luciferase seeds shows that the MS promoter continues to be active at low levels for at least 12 days after imbibition in non-germinated ABA-treated seeds. Consistent with this, measurement of 20 : 1, as an indicator of TAG, reveals that a reduced level of TAG breakdown does occur in non-germinating ABA-treated seeds. These results therefore indicate that the machinery necessary for the mobilization of storage lipid to sugar is present and operating, albeit at reduced levels, in ABA-treated seeds.
Further evidence that storage lipid mobilization in Arabidopsis is not blocked by ABA was sought from measurements of endogenous sugar levels in seeds and seedlings. Arabidopsis seeds contain soluble sugars in the form of glucose, fructose, sucrose and oligosaccharides such as raffinose and stachyose. These sugars are used to provide energy for germination, before the breakdown of storage reserves commences as part of early post-germinative growth. Interconversion of these sugars may take place in non-germinated seeds, for example during drought stress. We predicted that seeds blocked from germination and post-germinative growth by a reduction in the supply of energy and nutrients to the embryo would have low levels of endogenous sugars, having utilized those soluble sugars present within the seed. This hypothesis was supported by the observation that exogenous sugars can rescue both the ABA block on germination and early post-germinative growth (Garciarrubio et al., 1997), and mutants blocked in β-oxidation or the glyoxylate cycle. However, we found that ABA-treated seeds accumulate sucrose to levels approximately double those found in day 0 imbibed seeds, and glucose levels do not fall below those immediately after imbibition, indicating that the inhibition of growth is not due to low soluble sugar levels. Furthermore, icl1 and kat2 do not accumulate sucrose in the presence of ABA, indicating that the majority of sucrose is derived from storage lipid, not from the interconversion of soluble sugars within the seed. Although it can be seen from the icl1 and kat2 mutants that sucrose utilization still occurs in the presence of ABA, measurements from icl1 indicate that sucrose utilization is significantly reduced by ABA treatment.
Sugar levels were measured in the abi mutant seeds in the presence and absence of ABA to gain an insight into a possible relationship between inhibition of growth and the accumulation of sucrose in ABA-treated seeds. abi3-1 and abi4-103 seeds are able to germinate and establish in the presence of ABA, although they are slower to do so than seeds in the absence of ABA. Samples containing only germinated or non-germinated seeds were used for sugar analysis. In the absence of ABA, sucrose levels 3 days after imbibition are increased compared to wild-type seeds, which could in part be due to the fact that sucrose levels in dry seeds of abi3-1 and abi4-103 are also higher than their respective wild types (data not shown). In the presence of ABA, sucrose levels in both germinated and non-germinated abi3-1 and abi4-103 samples are significantly lower than in non-germinated Col0 and Ler samples. However, the observation that sucrose levels were as high or, in the case of abi3-1, higher in samples of germinated seedlings compared to non-germinated seeds suggests that sucrose accumulation is not dependent on the extent of seedling growth in the presence of ABA. Rather, it appears that sucrose accumulates in the presence of ABA as a result of altered sucrose metabolism. However, we were unable to detect any alteration in the activity or expression of genes encoding the key sucrose metabolism enzymes sucrose phosphate synthase, sucrose synthase and invertase (data not shown).
There is evidence to suggest that ABA may inhibit germination by more than one pathway. In Brassica napus, rapid water uptake after approximately 12 h imbibition brings about the expansion of embryonic cells leading to germination (Schopfer and Plachy, 1984). Water uptake is controlled by cell-wall loosening resulting from a decrease in the minimum turgor pressure required for cell expansion, and an increase in the extensibility of the walls (Schopfer and Plachy, 1985). ABA treatment inhibits the changes in the factors affecting cell-wall loosening, thereby preventing cell expansion (Schopfer and Plachy, 1985). In addition, evidence exists in Arabidopsis that ABA inhibits the cell cycle. Wang et al. (1997) identified a cyclin-dependent protein kinase inhibitor (ICK1) which inhibits the activity of kinases involved in the cell cycle. Further investigation revealed that ICK1 is induced in response to ABA treatment in 12-day-old seedlings (Wang et al., 1998), suggesting that ABA treatment may block post-germinative growth by inhibiting cell division.
The involvement of ABA in many different plant responses means that exogenous ABA treatment may have multiple effects after imbibition, and the mode of inhibition of germination and post-germinative growth by ABA remains unclear. The isolation of mutants in ABI3, ABI4 and ABI5 from germination-based screens demonstrates that the transcription factors encoded by these genes play an important role in the regulation of germination. However, the fact that ABA treatment during and following imbibition does not block storage reserve mobilization indicates that ABI3, ABI4 and ABI5 do not play a dominant role in the regulation of this process.
The question still remains as to why sugars and peptone can rescue the ABA block on germination. As ABA-treated seeds accumulate high levels of sucrose from the breakdown of storage lipid, it is unlikely the explanation is as simple as the provision of energy. One possible explanation is that the embryo perceives endogenous and exogenous sugar differently, and that the two have very different effects on the signals underlying germination. Sugar concentration is also an important factor. While concentrations of 14 mm sucrose rescue germination of ABA-treated seeds (Garciarrubio et al., 1997), much higher concentrations of 330 mm sucrose actually inhibit germination and seedling growth (Laby et al., 2000).
We have provided evidence for the existence of two distinct programmes operating after seed imbibition, one driving germination and the other driving reserve mobilization during early post-germinative growth. The observations that the kat2 mutant can germinate in the absence of lipid breakdown (Germain et al., 2001), and that ABA-treated seeds are capable of breaking down lipid storage reserves, demonstrate that each of these programmes can progress in isolation. However, both programmes are required for successful germination and seedling establishment. ABA-treated seeds in which germination has been blocked, but reserve mobilization still occurs, will provide a useful system for unravelling the regulatory mechanisms underlying storage reserve breakdown in Arabidopsis.
Seeds of the Arabidopsis thaliana ecotypes Columbia (Col0), Landsberg erecta (Ler) and Wassilewskija (Ws) were obtained from the Nottingham Arabidopsis Stock Centre. The abi3-1 mutant seeds were obtained from Dr Maarten Koornneef (Wageningen University, the Netherlands). The abi4-103 mutant seeds were obtained from Dr Susan Gibson (Rice University, TX, USA). The kat2 (Germain et al., 2001) and icl1 (Eastmond et al., 2000b) mutant seeds and the ACX3::GUS reporter line (Eastmond et al., 2000a) had previously either been isolated or characterized in the Graham laboratory. Seeds were sown onto plates containing half-strength MS medium (Murashige and Skoog, 1962) with 0.8% (w/v) agar and additions as described. Seeds were imbibed for 4 days in the dark at 4°C, then transferred to 16 h light and 8 h dark at 21°C.
Total fatty acids and eicosenoic acid levels were determined by gas chromatography of fatty acid methyl ester derivatives (Larson and Graham, 2001).
GUS activity was assayed by measuring the conversion of 4-methyl umbelliferyl glucuronide to 4-methyl umbelliferone according to Jefferson (1987). Histochemical detection of GUS activity was carried out according to Jefferson (1987). Samples were stained overnight.
Seedlings were sprayed with 5 mm luciferin, 0.01% Triton X-100 2 h before and immediately prior to imaging. Timed integrations of 100 sec were carried out for each sample using a photon-counting camera and HRPCS-2 Camera Control Unit (Photek, St Leonards on Sea, Sussex, UK).
Sugars were extracted by boiling in 80% ethanol for a total of 3 h. Glucose, fructose and sucrose were determined by a sequential series of assays linked to NADPH production (Stitt et al., 1989).
Isolation of the MS promoter and construction of MS::LUC plants
A 969 bp fragment of 5′ sequence from −3 bp relative to the ATG start site was amplified by PCR. The amplified fragment included an NcoI restriction site at the 3′ end to produce a translatable fusion with the luciferase gene following cloning. Clones from three independent PCRs were sequenced and verified against the genomic sequence. The fragment was cloned as a HindIII–NcoI fragment into pGRT10 (pUC18 into which the luciferase gene had previously been inserted) to produce pMS-GRT. The MS-luciferase cassette was excised with HindIII and SacI and inserted into HindIII–SacI-digested pBI101. To produce a promoterless control, luciferase was excised from pGRT10 with HindIII and SacI and ligated with HindIII–SacI-digested pBI101. Transformation of Arabidopsis was carried out according to Clough and Bent (1998). Expression of luciferase was examined in T2 seeds from three independent transformants and was consistent with Northern analysis of MS gene expression.
We wish to thank The Gatsby Charitable Foundation for provision of a studentship to Sarah Pritchard. Wayne Charlton was funded by the Biotechnology and Biological Sciences Research Council via the Genome Analysis of Agriculturally Important Traits grant 24/GAT/09140 to A.B.