Most plant seeds are dispersed in a dry, mature state. If these seeds are non-dormant and the environmental conditions are favourable, they will pass through the complex process of germination. In this review, recent progress made with state-of-the-art techniques including genome-wide gene expression analyses that provided deeper insight into the early phase of seed germination, which includes imbibition and the subsequent plateau phase of water uptake in which metabolism is reactivated, is summarized. The physiological state of a seed is determined, at least in part, by the stored mRNAs that are translated upon imbibition. Very early upon imbibition massive transcriptome changes occur, which are regulated by ambient temperature, light conditions, and plant hormones. The hormones abscisic acid and gibberellins play a major role in regulating early seed germination. The early germination phase of Arabidopsis thaliana culminates in testa rupture, which is followed by the late germination phase and endosperm rupture. An integrated view on the early phase of seed germination is provided and it is shown that it is characterized by dynamic biomechanical changes together with very early alterations in transcript, protein, and hormone levels that set the stage for the later events. Early seed germination thereby contributes to seed and seedling performance important for plant establishment in the natural and agricultural ecosystem.
‘… I have had one experiment some little time in progress which will, I think, be interesting, namely, seeds in salt water, immersed in water of 32°–33° […] I have in small bottles out of doors, exposed to variation of temperature, cress, radish, cabbages, lettuces, carrots, and celery, and onion seed—four great families. These, after immersion for exactly one week, have all germinated, which I did not in the least expect (and thought how you would sneer at me); for the water of nearly all, and of the cress especially, smelt very badly, and the cress seed emitted a wonderful quantity of mucus (the ‘Vestiges’ would have expected them to turn into tadpoles), so as to adhere in a mass; but these seeds germinated and grew splendidly. The germination of all (especially cress and lettuces) has been accelerated, except the cabbages, which have come up very irregularly, and a good many, I think, dead. One would have thought, from their native habitat, that the cabbage would have stood well. The Umbelliferae and onions seem to stand the salt well.’ (April 13th, 1855, cited from: Darwin, 1887).
Charles Darwin's interest in seed germination was a focus within his wider interest in plant development. He published several papers about the above findings in the Gardeners’ Chronicle and Agricultural Gazette (including Darwin, 1855a, b, c, d). His interest in seed germination was indeed well founded: seed germination is a crucial process in the seed plant life cycle. It determines when plants enter natural or agricultural ecosystems and is the basis for crop production. This review deals with the early events during this important life cycle transition. Early seed germination is defined here as imbibition plus the early plateau phase of water uptake. It is thus positioned between the dry state of the seed and the late phase of germination. Germination is completed by visible radicle protrusion through the seed covering layers, and followed by seedling establishment (Fig. 1). Late germination has been the focus of seed research for many decades (summarized in recent reviews, e.g. Finch-Savage and Leubner-Metzger, 2006; Holdsworth et al., 2008; Nonogaki et al., 2007; North et al., 2010). It is believed that unravelling the mechanisms underlying germination requires the integration of all of its facets including early events.
Most mature angiosperm seeds consist of an embryo surrounded by covering layers such as the maternal testa (seed coat) and the triploid endosperm. Seeds exhibit species-specific differences in their structure and the composition of their storage compounds (Obroucheva and Antipova, 1997; Linkies et al., 2010). Interestingly, Charles Darwin already worked with some species that would later become model species in seed biology, namely lettuce, on which the red/far red light-induced reversibility of phytochrome effects was discovered (Borthwick et al., 1952), and cress, radish, and cabbages, which are members of the Brassicaceae family for which the first plant genome was sequenced (Arabidopsis thaliana; Koornneef and Meinke, 2010). This review will focus on a range of orthodox eudicot model systems of seed germination. Pea (Pisum sativum, Fabaceae, Fig. 2A) seeds store mainly proteins and starch in the embryo's storage cotyledons; mature pea seeds have no endosperm (Obroucheva and Antipova, 1997; Melkus et al., 2009). The Brassicaceae oil-seeds of Arabidopsis (Fig. 1A) and garden cress (Lepidium sativum, ‘cress’) contain a thin endosperm layer (Haughn and Chaudhury, 2005; Müller et al., 2006), while the oil-seeds of tobacco (Nicotiana tabacum, Solanaceae) contain a thicker endosperm layer (Leubner-Metzger, 2003; Manz et al., 2005). This review will follow the physiological timeline of events during early seed germination, from the dry seed to fully reactivated metabolism. This is supported by corresponding figures and in addition by Supplementary material available at JXB online.
The dry seed stage: moisture content, after-ripening, and the stored transcriptome
Seed maturation and desiccation were recently reviewed (Holdsworth et al., 2008; Angelovici et al., 2010). This discussion will start at the end-point of these reviews: with the biochemical properties of the desiccated mature orthodox seed, which constitutes a desiccation-tolerant state of the sporophyte with typical average water contents of ∼10%. This ‘dry’ seed state is therefore in fact a ‘low-hydrated’ state, and dry seeds are not completely metabolically inert. The physiological state of dry seeds changes during after-ripening (i.e. a prolonged period of dry storage at room temperature of freshly harvested, mature seeds). After-ripening storage is associated with a loss of dormancy, although dormancy release and after-ripening may be separate pathways (Carrera et al., 2008; Holdsworth et al., 2008). After-ripening depends on temperature and seed moisture content. The optimal moisture content for after-ripening is lower for oil-storing compared with starchy seeds, but in general after-ripening takes place at seed moisture contents between 8% and 15% (Probert, 2000; Bazin et al., 2011). Moisture sorption isotherms of seeds (Fig. 1B) show that water molecules are weakly bound at this water content, which means its availability for biochemical reactions is limited. The shape of the moisture sorption isotherm curves is similar for oil-seeds (e.g. tobacco and Arabidopsis; Hay et al., 2003; Manz et al., 2005) and starchy seeds (e.g. pea; Chen, 2003); even though absolute values differ slightly. However, water distribution is inhomogeneous within seeds, and seed tissues differ in their moisture sorption isotherms (see references in Hay et al., 2003; Manz et al., 2005; Wojtyla et al., 2006). 1H-nuclear magnetic resonance (NMR) spectroscopic imaging of tobacco seeds suggests that there are local pockets of higher hydration, in which water may be freely available for biochemical reactions. This may lead to differing biochemical capabilities between seed organs and tissues. Although evidence is still fragmentary, there are several reports that indicate that low-level transcription, post-transcriptional processing, and translation may be possible during seed after-ripening of tobacco (Leubner-Metzger, 2002, 2005), Arabidopsis (Müller et al., 2009), barley (Leymarie et al., 2007), and other species (Holdsworth et al., 2008). Holdsworth et al. (2008) also provide a critical review of this issue which is recommended for further reading. Future experiments to elucidate these highly debated findings are required and could include the use of novel imaging techniques as well as tissue-specific transcriptome and proteome analyses during after-ripening storage.
There are indications that after-ripening includes protein oxidation by reactive oxygen species (ROS). ROS are formed in the dry state as can be shown by the redox state of dry seeds shifting towards a more oxidized cellular environment (Kranner et al., 2006, 2010b). Antioxidant enzyme activity will be limited or impossible in most parts of the seeds due to the lack of available water. The seeds therefore rely on small antioxidant molecules for their protection from oxidative damage, with the glutathione (GSH) system probably playing a major role. ROS oxidize GSH to its dimer GSSG, which accumulates during seed storage (Kranner and Grill, 1993). In addition, the lipophilic antioxidant tocopherol, which protects membranes from lipid peroxidation, is essential for seed longevity and germination characteristics, as was shown in a mutant approach in Arabidopsis (Sattler et al., 2004; Mene-Saffrane et al., 2010). A third major antioxidant, ascorbate, is only present in small amounts in dry seeds and therefore probably plays a minor role in regulating the redox situation in the dry state (Wojtyla et al., 2006; Dowdle et al., 2007).
Oracz et al. (2007, 2009) proposed a causal role for ROS in sunflower embryo dormancy release, and Müller et al. (2009) showed that the Arabidopsis mutant atrbohB which does not after-ripen has an altered pattern of protein oxidation in the dry seeds. In this context, the concept of an oxidative window has been suggested which assumes that oxidative processes in seeds first lead to after-ripening and loss of dormancy, but later tip the scale towards oxidative damage, deterioration, and loss of viability (Bailly, 2004).
The processes that take place in dry seeds and lead to after-ripening or deterioration are an important aspect of seed biology and the topic of active research. As even small changes in overall seed moisture content influence storability and longevity of seeds (Buitink et al., 2000; Finch-Savage and Leubner-Metzger, 2006), understanding these processes is an issue of economic importance and a major concern of seed banks (Nagel and Börner, 2010). New non-invasive imaging approaches (see last section) will be very helpful in elucidating this stage of the plant life cycle.
Dry seeds contain mRNAs stored during maturation, also called long-lived transcripts to indicate that they survived desiccation (Rajjou et al., 2004). Over 10 000 different stored mRNAs have been identified in transcriptome analyses of Arabidopsis (Nakabayashi et al., 2005; Kimura and Nambara, 2010; Okamoto et al., 2010). Similar numbers were found in barley and rice (Howell et al., 2009; Sreenivasulu et al., 2010). In Arabidopsis, abscisic acid (ABA)-responsive elements are over-represented in the promoters of genes whose transcripts are stored (Nakabayashi et al., 2005), in accordance with the major role of ABA during seed maturation (Nambara et al., 2010; Radchuk et al., 2010). So far the published transcriptomes are from whole dry seeds, but it is known that the different seed compartments, for example the endosperm and embryo, accumulate different transcripts during seed development (Le et al., 2010). Kimura and Nambara (2010) showed that major portions of the dry seed transcriptomes of the non-dormant Arabidopsis ecotype Columbia (Col) and the dormant ecotype Cape Verde Island (Cvi) are very similar. The majority of stored mRNAs are of the LEA (‘late embryogenesis abundant’) group or transcripts of storage proteins, supporting the view that the dry seed transcriptome mirrors the process of seed maturation as well as prepares the seed for the following germination. The transcriptomes of the two ecotypes differ in an over-representation of heat shock proteins and ROS-related transcripts in Col and of phosphate and lipid metabolism as well as cytoskeleton-associated transcripts in Cvi. Larger transcriptome differences between these dormant and non-dormant seeds only develop after imbibition. A comparison between the dry seed transcriptomes of near isogenic lines (NILs) representing ‘Delay of Germination’ (DOG) quantitative trait loci (QTLs) of Arabidopsis that differ in after-ripening and/or dormancy suggests that natural variations for these traits are mainly controlled by additive genetic and molecular pathways, rather than epistatic interactions (Bentsink et al., 2010). It will be interesting to see if, as is the case for the comparison of Col and Cvi (Kimura and Nambara, 2010), the differences between the transcriptomes of the NILs become larger during imbibition and if these distinct DOG pathways remain independent. This review will come back to the dry seed transcriptome and its changes upon imbibition when metabolism is explored in a later section.
Physical, morphological, and physiological aspects of imbibition and testa rupture
Seed germination begins when the dry seeds come into contact with water under favourable conditions. It comprises three phases of water uptake. Dry seeds have very low water potentials (Woodstock, 1988; Obroucheva and Antipova, 1997) which cause rapid water influx during phase I (imbibition, Figs 2B, 3). As this process is driven by the matrix potential, it also occurs in dead seeds (Krishnan et al., 2004). During imbibition the seed rapidly swells and changes in size and shape (e.g. Robert et al., 2008; Preston et al., 2009). 1H-NMR image analyses of imbibition with pea, tobacco, and other species demonstrate that there are major entry points for water uptake such as the micropyle, and that the progress of imbibition differs between seed tissues (e.g. Manz et al., 2005; Wojtyla et al., 2006). In Arabidopsis the shape of the imbibed wild-type seed approximates a prolate spheroid during this period. Robert et al. (2008) showed that ethylene mutants differ in seed shape and imbibition behaviour from wild-type seeds. These differences in changes in size and shape of imbibing seeds could in the future be used for large-scale mutant screens, as computational approaches facilitate the high-throughput analysis of image time series (e.g. Joosen et al., 2010).
The Arabidopsis testa contains volcano-shaped cell wall structures on the seed surface known as columellae (Fig. 4A). Upon first contact with water of Arabidopsis and other mucilaginous seeds, mucilage is released quickly from the columellae—the ‘wonderful quantity of mucus’ that Darwin observed on his cress seeds is a further example. Arabidopsis mucilage is composed mainly of rhamnogalacturonan pectins and cellulose arranged in an outer water-soluble layer and an inner layer covalently bound to the testa by cellulose microfibrils (Windsor et al., 2000; Macquet et al., 2007). Possible functions of the mucilage are the adherence to surfaces and animals for seed dispersal (Mummenhoff et al., 2004) and aiding germination in osmotically and saline-stressful environments (Yang et al., 2010) as the mucilage is very hydrophilic and delays water loss.
Initial imbibition is often accompanied by a massive leakage of cellular solutes. Similar phenomena can be observed in resurrection plants and pollen that rapidly return from a dry quiescent state to a fully hydrated state (Hoekstra et al., 1999). While leakage can accelerate germination by lowering inhibitor concentrations within seeds (Matilla et al., 2005), it is also a sign of damage to membranes and cellular compartments caused by fast and/or inhomogeneous rehydration (Powell and Matthews, 1978). In order to deal with the damage imposed during dehydration, storage, and, most significantly, rehydration, seeds activate a number of repair mechanisms during imbibition (Fig. 3). This includes the repair of membranes as well as of proteins in which aspartyl residues were damaged by conversion to isoaspartyl. The latter can be reversed by isoaspartyl methyltransferase (Oge et al., 2008).
Damage to genomic DNA includes progressive loss of telomeric sequences during prolonged dry storage (Boubriak et al., 2007) as well as strand breaks and other types of DNA damage that result from cumulative effects of temperature, moisture, oxygen, and ROS levels (reviewed by Bray and West, 2005). The accumulation of chromosomal damage and/or an inability to repair such damage during the imbibition period appear to be significant factors contributing to loss of seed viability during storage. Seed dehydration during maturation and rehydration during imbibition lead to the appearance of a large number of DNA single-strand breaks in maize seeds, most of which can be attributed to imbibitional damage. This includes the appearance and repair of apurinic/apyrimidinic sites in DNA during early germination. DNA damage, which would obviously be a major obstacle during germination, can be repaired by DNA ligases. DNA ligase expression is activated quickly upon imbibition of Arabidopsis seeds, and high levels of de novo DNA synthesis have been observed in the absence of nuclear DNA replication or cell division, indicating a role in DNA repair. In aged seeds, which have suffered more severe damage during storage, enhanced early DNA synthesis has been observed (Bray and West, 2005). Insertional knock-out mutants of two DNA ligases, AtLIG4 and the plant-specific AtLIG6, consequently showed a delay in germination under optimal conditions which was aggravated under cold stress conditions and in the presence of ROS (Waterworth et al., 2010). Bray and West (2005) state that the seed provides an ideal ‘model’ system for investigating the effects of a variety of endogenous DNA-damaging agents and environmental stresses on genome integrity.
The permeability of the testa, being the part of the seed that comes into contact with the ambient water in most seeds, plays a central role in the rate of water uptake (Chachalis and Smith, 2000; Wojtyla et al., 2006; Koizumi et al., 2008). Brassica seeds with different testa morphology show altered germination characteristics (Zeng et al., 2004; Matilla et al., 2005). Arabidopsis testa mutant seeds with reduced pigmentation are more permeable to tetrazolium salts than the wild type, and the seeds show a lower dormancy and differ in hormone sensitivities during germination (Debeaujon and Koornneef, 2000; North et al., 2010).
Once the rate of water uptake and changes in seed size and shape start to stagnate, germinating seeds move into water uptake phase II, during which the water content remains stable and which can vary widely in duration. In species with a ‘two-step germination’ process such as Arabidopsis, cress, and tobacco (Liu et al., 2005; Manz et al., 2005; Müller et al., 2006), phase II encompasses testa rupture (Figs 1C, 3, 4B). This is followed by phase III water uptake, endosperm rupture, and radicle protrusion; that is, the completion of germination sensu stricto (Fig. 4C). Phase III water uptake continues during the transition to seedling growth (Figs 1C, 3). Endospermless species such as pea and Brassica napus (Brassica) enter phase III after testa rupture (‘one-step germination’, Fig. 2). In non-dormant seeds, exogenous ABA inhibits the transition from water uptake phase II to III and late embryo cell expansion, but does not affect phase I and II and testa rupture (Figs 2B, 3) (e.g. Schopfer and Plachy, 1984; da Silva et al., 2004; Manz et al., 2005; Müller et al., 2006). The presence of endosperm in mature seeds provides an additional target tissue for regulating the completion of germination by ABA and environmental factors. Its visibility as a two-step process appears to be a phylogenetically widespread trait determined by the anatomy of the seed-covering layers (Petruzzelli et al., 2003; Linkies et al., 2010).
Aquaporins, small membrane proteins that can transport water as well as non-polar small molecules, facilitate cell–cell water transport and may contribute to spatial distribution of water within seed tissues during imbibition as well as to the timing of testa rupture in tobacco (Schuurmans et al., 2003; Maurel et al., 2009). Vander Willingen et al. (2006) analysed the involvement of aquaporins in the germination of cold-stratified Arabidopsis seeds and found that tonoplast intrinsic proteins (TIPs) show a germination-related shift of protein accumulation from TIP3 to TIP1 at the later stages of germination and found a similar pattern for RNA transcription levels. In contrast to TIPs, they did not find strong indications for the involvement of plasma membrane intrinsic proteins (PIP) as neither protein nor transcript presence and changes were detected. A different conclusion for the PIP-type aquaporins comes from evidence obtained with knock-down tobacco mutants of the plasma membrane aquaporins PIP1 and PIP2, for which testa rupture was affected differentially (Ernst, 2007): testa rupture occurred earlier in pip2 mutant seeds, while it was delayed in pip1. The time period between testa rupture and the completion of germination was not altered in the mutants compared with the wild type. Taken together, these contrasting results in tobacco and Arabidopsis clearly show that further research in this area is needed.
While Arabidopsis seed coat development including the genes and hormones involved in this process has been studied in detail (Haughn and Chaudhury, 2005), little is known about the changes of the seed coat's mechanical and biochemical properties that ultimately lead to testa rupture. In tobacco, testa rupture starts near the micropylar seed end that covers the radicle and spreads along the ridges on the testa (Leubner-Metzger, 2003). Progression of tobacco testa rupture is facilitated by channel-like structures underlying the ridges, suggesting pre-determined breaking points. Arabidopsis testa rupture (Fig. 4B) also starts at the micropylar seed end, but it is unknown if pre-determined breaking points exist. Future experiments concerning a potential enzymatic weakening, spatial water redistribution in connection with cell elongation, and transcriptomic activity before testa rupture are required as they might shed light on how testa rupture is controlled. Tissue-specific transgenic approaches might help to elucidate the role of the embryo, endosperm, and integument layers in testa rupture.
Embryo cells elongate prior to the completion of seed germination of Arabidopsis, Brassica, Medicago, and other species; cell division is not evident in the embryos of these seeds during germination (Barroco et al., 2005; Gimeno-Gilles et al., 2009; Sliwinska et al., 2009). After the initial swelling is completed, all changes in seed size and shape during germination are caused by cell expansion. Expanding plant cells adjust the extensibility of their cell walls by remodelling the major components of the wall, the cellulose microfibrils and/or the pectin/hemicellulose matrix. Loosening of the wall allows water influx which drives cell expansion and generates cellular turgor pressure (Schopfer, 2006). This led to the model that embryo growth during germination depends primarily on changes in cell wall extensibility. These changes are accompanied by progressing vacuolation during late phase II of water uptake. Embryo and endosperm cells are not fully vacuolated during early phase II, but display many small vacuoles (Bethke et al., 2007). Sliwinska et al. (2009) proposed that Arabidopsis embryo elongation occurs in a distinct and confined elongation zone between the radicle and the lower hypocotyl.
Hormonal and temperature regulation of early gene expression in imbibed seeds
Hormone contents, signalling, and interactions play important roles in determining the physiological state of the seed and in regulating the germination process (Kucera et al., 2005). The endogenous ABA contents of non-dormant and dormant seeds rapidly decline upon imbibition during the early phase of germination (within 6–12 h, Fig. 5; Chiwocha et al., 2005; Nakabayashi et al., 2005; Hermann et al., 2007; Linkies et al., 2009; Preston et al., 2009). However, in dormant and thermoinhibited seeds (i.e. seeds in which high temperatures inhibit germination) this decrease stops, and de novo ABA synthesis in the imbibed state causes higher ABA contents which are required for dormancy maintenance and for inhibiting germination (Nambara et al., 2010). A sufficient decrease in endogenous ABA content during imbibition and early phase II is thus a major prerequisite for the completion of germination. Exogenous treatment of after-ripened Arabidopsis or cress seeds with ABA does not affect the kinetics of testa rupture, but inhibits endosperm weakening and rupture (Fig. 6A and Müller et al., 2006). Nine-cis-epoxycarotenoid dioxygenaes (NCEDs) and ABA 8'-hydroxylases (CYP707As) are the major key regulatory enzymes for ABA biosynthesis and degradation, respectively (Fig. 5A). NCEDs and CYP707As are encoded by multigene families, and their tissue- and environment-specific regulation determines the ABA contents (Seo et al., 2006; Toh et al., 2008). CYP707A2 transcripts are expressed in the radicle upon seed imbibition (Okamoto et al., 2006). Changes in hormone contents during the early germination phase are also evident in Arabidopsis seeds for jasmonic acid, whose content decreases, and indole acetic acid, whose content increases (Preston et al., 2009).
The germination-inhibiting effect of ABA is counteracted by gibberellins (GAs) and by ethylene. The effects of these hormones on the late germination process have been extensively reviewed (Kucera et al., 2005; Holdsworth et al., 2008; Linkies et al., 2009; North et al., 2010) and their interaction with each other and with light has been studied (e.g. Debeaujon and Koornneef, 2000; Seo et al., 2008; Piskurewicz et al., 2009; North et al., 2010). Ethylene has important roles during the late phase of germination and counteracts the ABA inhibition by interfering with ABA signalling, but it does not affect ABA contents (Linkies et al., 2009). In contrast, GAs are important during the early and the late phase of germination and counteract the ABA inhibition. Bioactive GA4 was already present in physiologically relevant amounts in the dry, after-ripened seeds that Ogawa et al. (2003) used for their transcriptome analysis (Fig. 7B), and a further increase in GA4 contents occurs during late germination. GA20 and GA3 oxidases (GA20ox and GA3ox) are the major key regulatory enzymes for GA biosynthesis, while GA2 oxidases mediate GA degradation (Fig. 7A). Transcripts of GA20ox and GA3ox accumulate during early germination. Ogawa et al. (2003) demonstrated that GA biosynthesis localizes to the radicle, hypocotyl, and micropylar endosperm during germination. Due to the rapid ABA degradation, the GA/ABA ratio increases ∼3-fold during early germination and ∼10-fold during late germination (compare Figs 5B and 7B). While for the early germination phase Ogawa et al. (2003) did not find altered ABA contents upon treatment of GA-deficient ga1-3 Arabidopsis seeds with exogenous GA, Yano et al. (2009) found that GA4 contents and GA3ox1 transcript levels were decreased in ABA-overproducing cyp707a2 Arabidopsis seeds. This is most probably due to the increased ABA contents and it therefore seems that ABA can inhibit GA biosynthesis during early germination.
GA20ox and GA3ox are induced by red light and cold stratification (Supplementary Fig. S1 at JXB online; Yamauchi et al., 2004; Kucera et al., 2005). Moist cold stratification of Arabidopsis (i.e. incubation of imbibed seeds at 4 °C in darkness usually for 1–4 d) is often used to break dormancy and promote subsequent germination in the light. Yamauchi et al. (2004) have demonstrated that cold stratification is associated with increased contents of bioactive GAs (Fig. 7C) and by the accumulation of GA20 and GA3 oxidase transcripts (Supplementary Fig. S1B). Furthermore, cold stratification induced a spatial change in GA3ox1 transcript expression; in addition to the radicle it strongly accumulated in the micropylar endosperm (Yamauchi et al., 2004). The seed-specific eFP-browser and the eNorthern tool at www.bar.utoronto.ca visualize transcript expression patterns based on global transcriptome analyses during Arabidopsis seed germination (Winter et al., 2007; Bassel et al., 2008). These tools were used for the purpose of this review to analyse early temporal transcript changes and their regulation by GAs and ABA or upon cold stratification. As discussed later, the transcriptional changes of key metabolic enzymes were analysed during the early phase of germination (0–24 h) of non-dormant, non-stratified Arabidopsis seeds (Nakabayashi et al., 2005; Preston et al., 2009) to test and generate hypotheses for the activation of seed metabolism (Fig. 8). Selected transcript changes of non-stratified and stratified seeds were also compared in order to determine the effects of cold stratification (Yamauchi et al., 2004), and GA and ABA treatments (RIKEN transcriptome sets at www.bar.utoronto.ca). This approach is obviously limited as changes in transcript level do not necessarily correspond to similar changes in protein level and/or enzyme activity. The authors are aware of the descriptive nature of this approach and its limitations, but it is believed that this approach is useful to generate hypotheses that can be tested in subsequent experiments on the physiological, protein, and activity level and by mutant approaches.
Here, this approach is applied to investigate the transcript expression patterns of α-expansins and ABA-related genes in Arabidopsis seeds. α-Expansins are known to be induced in the endosperm upon imbibition (Penfield et al., 2006; Carrera et al., 2008; Linkies et al., 2009). They are a group of proteins proposed to be involved in cell wall remodelling important for cell expansion growth, and they exhibit extensive regulation during early germination (Fig. 6B–D). During the early phase of Arabidopsis seed germination, transcripts of EXPA1, 2, 3, 8, 9, 15, and 20 accumulate 100- to 500-fold from 0 h to 12 h in whole unstratified seeds imbibed in the light (Fig. 6B). This induction upon imbibition is also evident during moist cold stratification (4 °C in the dark), but not if seeds are imbibed at 22 °C in the dark (Fig. 6C). When the 4 °C/22 °C transcript ratios are compared, an ∼30-fold cold induction was evident for EXPA1 and EXPA2 (Fig. 6D). Many α-expansin genes are GA inducible, as shown by the +GA/–GA transcript ratios obtained with GA-deficient ga1-3 mutant seeds; EXPA2 accumulates >200-fold upon GA treatment (Fig. 6D). The cold induction of α-expansin expression could therefore be mediated by GAs as cold stratification is associated with the induction of GA biosynthesis and increased contents of bioactive GAs (Fig. 7C, and Supplementary Fig. S1B at JXB online). Cold stratification is also associated with enhanced GA biosynthesis in the micropylar endosperm (Yamauchi et al., 2004) where α-expansin is localized. Cold stratification also promoted the decline in ABA contents and, associated with this, caused earlier completion of germination (Fig. 5B). In contrast to GAs, ABA treatment did not affect the α-expansin transcript expression (+ABA/–ABA ratios, Fig. 6D). ABA also did not affect the kinetics of testa rupture of after-ripened Arabidopsis seeds (Fig. 6A). Taken together, these temporal, hormonal, and cold-inducible transcript expression patterns of EXPA2 and other α-expansins in the micropylar endosperm are in agreement with the hypothesis that they could have roles in endosperm-mediated processes during early germination that lead to and control testa rupture.
Recent advances in Arabidopsis molecular genetics have revealed the core ABA signalling pathways (Nambara et al., 2010). Group A members of the protein phosphatase 2C (PP2C) family of genes (Supplementary Fig. S1C at JXB online), including ABA-INSENSITIVE1 (ABI1), seem to act as negative regulators of seed germination (Kucera et al., 2005; Nishimura et al., 2007). The PYR1/PYL1/RCAR family of START proteins is a family of ABA receptors and may have a prominent function in seed ABA responsiveness through regulation of PP2C activity in an ABA-dependent manner (Nambara et al., 2010; Nishimura et al., 2010). Targets of the PP2C are members of the SNF1-related protein kinase subfamily 2 (SnRK2) that act as positive regulators of ABA signalling in activating ABRE-binding transcription factors such as ABI5 (Nakashima et al., 2009). The SnRK2s become active when they are de-repressed from their inhibition by PP2Cs. Changes in the phosphorylation status of >50 proteins have been demonstrated in 12-day-old Arabidopsis plants after the addition of 50 μM ABA to the growth medium (Kline et al., 2010). This included an increase in phosphorylation of four SnRK2s after 30 min of treatment. Transcription is important for seed ABA responsiveness and is mediated, at least in part, by the transcription factors ABI5, ABI4, and ABI3 (Holdsworth et al., 2008; Nambara et al., 2010). ABA degradation (Fig. 5B) combined with a decrease in ABA sensitivity, for example by targeted proteolysis of ABI3 and ABI5 via the N-end rule pathway (Holman et al., 2009), promotes seed germination. Cold stratification not only induces a decline in ABA contents, but also affects the transcript expression of several ABA signalling components including PYL6, ABI4, ABI5, and several PP2Cs and SnRK2s (Supplementary Fig. S1C). In contrast to the group A PP2Cs mentioned above that act as negative regulators of ABA signalling, PP2C5 was found to be a positive regulator of ABA signalling (Brock et al., 2010). ATHB20 is a transcription factor involved in ABA sensitivity that is induced in the micropylar endosperm during early germination (Barrero et al., 2010). The MFT (MOTHER OF FT AND TFL1) gene serves as a mediator in response to ABA and GA signals, and regulates seed germination through a negative feedback loop modulating ABA signalling in Arabidopsis (Xi et al., 2010). Interestingly, the ABA-inducible expression of MFT is confined to the embryo elongation zone identified by Sliwinska et al. (2009). Based on these findings and the rapid decline of ABA contents upon imbibition, the seed tissue-specific regulation of ABA signalling is an emerging research field important for early seed germination.
A major decision for seed germination-related experiments is whether or not moist (cold) stratification should be used to release dormancy and achieve fast and uniform germination. The stratification treatment not only releases dormancy, but also promotes germination, and it is often hard to draw a clear line between the two interconnected processes. While a homogenous population is desirable, stratified seeds will already have gone through many processes that are important in early germination, and these will be lost to the subsequent observations. This also implies that early germination is hard to study in deeply dormant seeds such as many conifers which might need multiple months of moist chilling before they are able to germinate (Zeng et al., 2003). The important point is that early germination differs between non-stratified and stratified seeds, and if moist stratification is used it cannot simply be regarded as a technical treatment. It is suggested that if stratification is used, sampling during the stratification period should be included as part of the experimental investigations. In the following sections, the Arabidopsis transcriptome will be used to describe the activation of metabolism during early germination of non-dormant, non-stratified seeds. In addition, how ABA, GAs, and cold stratification affect the transcriptome responses regarding metabolism during early seed germination will be addressed.
Reactivation of metabolism: transcription and translation
During water uptake phases I and II, large metabolic changes take place in seeds which set the course for subsequent radicle protrusion. Metabolism is reactivated with enzymes that were stored in the seed during maturation. This has been shown in proteomic approaches in Arabidopsis, where a large number of enzymes involved in the major metabolic pathways were found in dry seeds and remained stable or even accumulated further during early germination (www.seed-proteome.com; Gallardo et al., 2001; Rajjou et al., 2004; Fu et al., 2005). Proteomic evidence for this includes enzymes from energy production pathways in dry Arabidopsis seeds (Supplementary Table S1 at JXB online): glycolysis [6-phosphofructokinase (PFK), phosphoglycerate kinase (PGK)], gluconeogenesis [PEP carboxykinase (PEPCK)], fermentation [alcohol dehydrogenase (ADH)], pyruvate dehydrogenase (PDH), the tricarboxylic acid (TCA) cycle [succinate dehydrogenase, succinyl-CoA ligase, malate dehydrogenase (MDH)], the glyoxylate cycle (isocitrate lyase), and the amino acid aminotransferases. However, the number of proteins detected by proteome analyses of seeds of Arabidopsis (see above), cress (Müller et al., 2010), sugar beet (Catusse et al., 2008), Medicago (Boudet et al., 2006), barley (Sreenivasulu et al., 2010), and rice (Yang et al., 2007) is limited (<1000) and therefore does not allow a complete comparison with the genome-wide transcriptome analyses. In Supplementary Table S1 the demonstrated presence of these proteins by proteome analysis of dry Arabidopsis seeds has been compared with the dry seed transcriptome. In almost all the cases where proteome analyses demonstrated a protein to be present in dry seeds, the corresponding transcript is abundant in the dry seed transcriptome. For example, the proteins of one cytosolic PGK and one cytosolic MDH were detected by proteome analysis. In each of these cases there are three genes, only one of which shows high transcript abundance in the dry seed. The proteins detected correspond to these highly abundant transcripts. It seems clear, therefore, that many of the abundant dry seed transcripts simply reflect translation during seed maturation. This conclusion is further supported by the fact that in many cases the most abundant transcripts in dry seeds are rapidly degraded upon imbibition, whereas several transcripts with lower abundance in the dry state accumulate during the early germination phase (for examples see the next section).
Upon imbibition, dramatic changes in the transcriptome can be observed after as little as 1–3 h, that is in phase I of water uptake (Howell et al., 2009; Preston et al., 2009; Okamoto et al., 2010). Some of these changes have been shown to be tissue specific (Okamoto et al., 2010). Transcripts displaying strong changes in abundance and following similar expression patterns during early germination have been found to share common cis-acting elements in their gene promoters. 3' Untranslated regions (UTRs) with motifs associated with RNA stability are enriched in transcripts with a strong regulation between 0 h and 3 h in rice (Nakabayashi et al., 2005; Howell et al., 2009; Preston et al., 2009), possibly indicating that many changes are due to degradation of stored mRNAs. It would therefore be particularly interesting to perpetuate earlier work (Marcus and Feeley, 1964; Dure and Waters, 1965; Comai and Harada, 1990) and investigate polysome-associated transcripts in addition to the general transcriptome in order to get a clearer picture of the actual translational activity of the many RNAs found in seeds during early germination.
A radiolabelling approach in Arabidopsis showed that de novo protein synthesis already took place in the first 8 h after imbibition and peaked between 8 h and 16 h (Rajjou et al., 2006), which corresponds to early phase II. All components of the transcriptional machinery are stored in dry seeds and are quickly activated upon imbibition, as has been demonstrated by the fact that the addition of the translation inhibitor cycloheximide does not alter early transcript up-regulation in Arabidopsis (Kimura and Nambara, 2010). Interestingly, cycloheximide disrupted early down-regulation of transcripts. Translation thus seems to be necessary to activate mRNA degradation mechanisms. In addition, ribosomal proteins are among the first to be transcribed in Arabidopsis (Tatematsu et al., 2008) and maize (Beltran-Pena et al., 1995). Translation is required for successful germination, while inhibition of transcription delays, but does not block, the completion of germination of Arabidopsis (Rajjou et al., 2004). Tobacco, on the other hand, can proceed to testa rupture but not to endosperm rupture in the presence of a transcriptional inhibitor (Arcila and Mohapatra, 1992; Leubner-Metzger, 2003). In agreement with a high ABA content during imbibition and early phase II of Arabidopsis seeds (i.e. before the ABA levels decline, Fig. 5B), the first 8 h comprise a short window of time in which ABA-induced genes that belong to the set of seed maturation genes that are typically expressed in an ABA-inducible manner during seed maturation are transcribed and translated (Lopez-Molina et al., 2002; Rajjou et al., 2006). From the work of Rajjou et al. (2004, 2008) it is clear that translation from stored mRNAs can differ depending on the physiological seed state, but which of the stored transcripts are absolutely required for the completion of germination is still an unsolved question.
Reactivation of metabolism: energy production
Minutes after the start of imbibition, a sharp increase in oxygen uptake and carbon dioxide release can be observed (Botha et al., 1992; Bewley, 1997). Gases may be released from colloidal adsorption or simply pushed out from gas-filled spaces by the water rushing in (Cloetens et al., 2006). Oxygen uptake then stagnates or increases only slowly; this lasts until the end of phase II (Botha et al., 1992; Bewley, 1997).
Dry seeds contain only low amounts of ATP, but a rapid production is initiated upon cellular hydration in association with the gas exchange (Botha et al., 1992; Spoelstra et al., 2002; Benamar et al., 2008). It is clear that respiratory pathways operate in imbibed seeds (Fig. 8), but their relative contribution and the substrates for ATP production during germination are still a matter of debate. Most of our knowledge about metabolic pathways at the level of enzyme activities and their products during the early phase of seed germination comes from studies of species with large seeds such as pea (Obroucheva and Antipova, 1997; Macherel et al., 2007; Benamar et al., 2008; Smiri et al., 2009). Most of the pathways presented in Fig. 8 operate in the cytosol. Mitochondria in dry seeds need repair and differentiation before contributing significantly to ATP production by oxidative phosphorylation. However, substrate oxidation of succinate by mitochondria extracted from dry pea seeds is possible (Morohashi and Bewley, 1980; Botha et al., 1992). Mitochondrial enzymes and membranes in starchy seeds such as pea seem to be protected by LEA proteins, and repair of pre-existing mitochondria takes place upon imbibition (Grelet et al., 2005; Tolleter et al., 2010). In contrast, biogenesis of new mitochondria is more important in oil-seeds (Morohashi et al., 1981; Morohashi, 1986). It is known that many key enzyme activities of the TCA cycle accumulate during early germination.
Oxygen-sensitive microsensors have been used to investigate the spatial and temporal oxygen status of germinating seeds of legumes, cereals, sunflower, and oilseed rape. The results point toward a limitation of oxygen uptake by seed-covering layers leading to hypoxia in seeds. In pea seeds, the internal oxygen content dropped during imbibition to anoxic levels while the respiration rate increased continuously (Obroucheva and Antipova, 1997; Benamar et al., 2008; Rolletschek et al., 2009). The ratio between fermentation and aerobic respiration varies between species and during the progression of germination, as does the sensitivity of seeds to oxygen availability (Benamar et al., 2008; Rolletschek et al., 2009). The adenylate energy charge of pea seeds can be increased and the accumulation of fermentation products decreased by providing the imbibing seeds with additional oxygen, which again points to a limiting role for oxygen rather than enzyme availability. Additional oxygen does not, however, increase the final germination percentage of pea, pointing to the fact that the seeds are adapted to germinate under hypoxic conditions. The outcome of Darwin's experiment described in the Introduction, which imposed salt stress and hypoxia during germination, testifies to the amazing ability of seeds to germinate under conditions of oxygen deprivation.
At the level of enzyme activities far less is known about the early germination of the small seeds of Arabidopsis, but the dynamics of the proteome (e.g. Gallardo et al., 2001; Fu et al., 2005) and the transcriptome (e.g. Nakabayashi et al., 2005; Preston et al., 2009) have been investigated. Temporal transcript expression patterns of key metabolic enzymes during the early phase of germination (0–24 h, water phase I and II) of non-dormant, non-stratified Arabidopsis seeds (Nakabayashi et al., 2005; Preston et al., 2009) were therefore used to test and generate hypotheses for the activation of seed metabolism (Fig. 4) and to determine the effects of moist cold stratification (incubation of Col seeds in darkness for 96 h at 4 °C compared with 22 °C; Yamauchi et al., 2004), ABA (Ler seeds at 24 h without and with ABA added), and GA (GA-deficient ga1-3 seeds at 6 h without and with GA added) treatments (RIKEN transcriptome sets at www.bar.utoronto.ca). It should be kept in mind that while the approach with ga1-3 has been very informative, the results may have been influenced by the fact that a mutant was used and might not be completely applicable to the wild type. The dry seed transcriptome for energy metabolism was also compared with its counterpart during early germination (imbibed for 6 h) to highlight the similarities and differences in transcript abundances (Fig. 8, and Supplementary Table S1 at JXB online). It was presumed that an at least 2-fold increase in transcript abundance is an indication for the activation of the encoded enzyme. As mentioned above, this approach is obviously limited as an effect on the transcript level does not necessarily correspond to a similar change on the protein and activity level. Where possible, information available from proteome and mutant work with Arabidopsis is therefore included.
Based on an at least 2-fold increase in transcript levels, the Arabidopsis seed transcriptome during early germination supports the view that glycolysis, fermentation, the TCA cycle and the oxidative pentose phosphate pathway (OPPP) are activated during early germination (Fig. 8, Supplementary Fig. S2 at JXB online). This is in agreement with evidence from enzyme activity measurements (e.g. Bettey and Finch-Savage, 1996; Wakao et al., 2008; Smiri et al., 2009) and proteome analyses (e.g. Gallardo et al., 2001; Fu et al., 2005; Müller et al., 2010). These activations are, however, not simply evident for each gene of a particular pathway, but seem to be a complex combination of stored proteins, stored transcripts, and de novo transcription and translation. Ethanolic fermentation by pyruvate decarboxylase (PDC) and ADH is a good example of this. In the dry seed transcriptome, PDC2 and ADH are the most abundant transcripts of the fermentation pathway, while the abundance of PDC1 and others is low. However, upon imbibition, the PDC2 and ADH transcript levels rapidly decline (3- to 5-fold within 6 h; Fig. 8, and Supplementary Table S1 and Fig. S2). Proteome analysis, however, demonstrated that ADH protein levels remain constant; that is, ADH activity depends on stored ADH protein that accumulated during seed maturation and/or on newly synthesized ADH that replaced stored ADH subjected to protein degradation (Gallardo et al., 2001). The high abundance of ADH transcripts in dry seeds combined with its rapid degradation would then simply be a remnant from seed maturation. On the other hand, de novo translation and ADH protein accumulation have also been described (Fu et al., 2005; Rajjou et al., 2006). ADH transcript expression is induced by cold stratification, but not regulated by GA (ga1-3 seeds ±GA) or ABA (wild-type seeds ±ABA) (Supplementary Fig. S2). While PDC2 is not inducible by these factors, PDC1 transcripts accumulate upon imbibition, cold stratification, and GA treatment (Supplementary Fig. S2). This example demonstrates that activation of energy metabolism during early germination is complex and cannot simply be predicted from the most abundant transcripts in the dry seed transcriptome.
Supplementary Fig. S2 also shows that imbibition itself and cold stratification are the most important factors for up-regulating transcripts of the sugar-related metabolic pathways mentioned above; neither up-regulation by GA nor down-regulation by ABA appears to be of major importance. However, other mechanisms for hormonal regulation exist; for example, ABA induces AtPirin1 (Lapik and Kaufman, 2003), which is known to regulate pyruvate catabolism by inhibiting the PDH complex (Soo et al., 2007). Cold stratification does not cause an increase in the levels of the TCA cycle metabolites citrate, malate, and succinate, but subsequent incubation in the light causes increases in their levels (Angelovici et al., 2011).
In contrast to the pathways for glycolysis, fermentation, the TCA cycle, and the OPPP mentioned above, transcript expression for key genes of gluconeogenesis, sucrose synthesis, and the peroxisomal pathways (fatty acid β-oxidation and the glyoxylate cycle) is not up-regulated during early germination (Fig. 8, and Supplementary Table S1, Fig. S2 at JXB online). Transcripts of the fatty acid β-oxidation enzyme 3-ketoacyl-CoA thiolase (KAT2/PED1) and the glyoxylate cycle enzyme isocitrate lyase (ICL) are among the top 100 most highly expressed transcripts in dry Arabidopsis seeds (Kimura and Nambara, 2010). However, the ICL transcript levels decline 25-fold (6 h/dry) upon imbibition and remain low until the late germination phase for which ICL protein accumulation has been shown (Supplementary Table S1, Fig. S2; Gallardo et al., 2001). Gluconeogenesis, sucrose synthesis, and the peroxisomal pathways are most important for post-germinative seedling establishment (e.g. Penfield et al., 2004; Holdsworth et al., 2008; Holman et al., 2009). Seedling arrest, but not germination phenotypes were evident for single-gene knock-out mutants of these pathways (Penfield et al., 2005). In contrast, recent evidence from double mutants for the peroxisomal pathways (the glyoxylate cycle and fatty acid β-oxidation, e.g. Pinfield-Wells et al., 2005; Pracharoenwattana et al., 2005, 2010) and careful consideration of the physiological seed state combined with distinct germination conditions (distinct media, sucrose addition, cold stratification; Footitt et al., 2006) demonstrates that the peroxisomal β-oxidation determines germination potential. This includes the ABC transporter COMATOSE (CTS, also known as PED3) required for the import of substrates for peroxisomal β-oxidation (Russell et al., 2000), for which a complex interaction with ABA has been proposed (Footitt et al., 2006; Kanai et al., 2010). The recent results of Kanai et al. (2010) suggest that CTS/PED3 promotes seed germination by suppressing the ABA-mediated inhibition of pectin degradation in the seed-covering layers.
Cleavage of triacylglycerol (TAG, seed oil) by TAG lipases at the water–oil interface of oil bodies provides glycerol and fatty acids. The Arabidopsis early germination transcriptome suggests that a glycerol shunt pathway is activated that feeds into glycolysis via dihydroxyacetone phosphate (DHAP) and involves glycerol-3-phosphate dehydrogenase (G3PDc), glycerol kinase (Gycerol-insensitive1/Nonhost1, GLI1/NOH1), and TAG lipases (Fig. 8). GLI1/NOH1 transcript levels increase >20-fold, the corresponding mutants have a germination phenotype (Eastmond, 2004), and in seedlings a metabolic connection to DHAP involving GLI1/NOH1 and G3PDc is known (Chanda et al., 2008). Transcripts of several AtTLG-type TAG lipase genes accumulate dramatically (e.g. AtTGL1 transcripts >300-fold) (Fig. 8, and Supplementary Table S1, Fig. S2 at JXB online). TAG lipase enzyme activity of AtTGL1 has been demonstrated, and the corresponding mutants show delayed germination which could be overcome by sucrose treatment (Körner, 2005). It is therefore hypothesized that glycerol released by AtTLG-type TAG lipases together with a seed glycerol shunt pathway (Fig. 8) could provide energy during early oil-seed germination. In addition, several of the AtTLG-type TAG lipase transcripts accumulate upon cold stratification, AtTGL1 is GA induced, but none of the AtTLG-type TAG lipase transcripts is affected by ABA (Supplementary Fig. S2). Other types of TAG lipase transcripts also accumulate upon imbibition, but none of them as strongly as AtTGL1 and AtTGL8.
Interestingly, fatty acid metabolism is repressed by ABA in the embryo, but not in the endosperm (Manz et al., 2005; Penfield et al., 2006). It has been proposed that ABA inhibits Arabidopsis seed germination by limiting the availability of energy and nutrients by preventing seed storage protein degradation (Garciarrubio et al., 1997), but not by inhibiting storage lipid mobilization (Penfield et al., 2005).
Stored proteins in seeds are not only an important source of amino acids during early germination, but are also important for energy production (Angelovici et al., 2011). Their activation is already prepared in the dry seeds: stored proteinases mobilize storage proteins in legume radicles (reviewed by Müntz et al., 2001; Müntz, 2007). Early degradation of protein bodies also occurs in the micropylar endosperm of Arabidopsis (Bethke et al., 2007). Aspartate and glutamate are among the most abundant amino acids in seed storage proteins. They are substrates for aspartate and alanine aminotransferases (AspAT and AlaAT) that are activated during imbibition and thought to participate in respiratory pathways (Fig. 8) (Obroucheva and Antipova, 1997; Miyashita et al., 2007; Rocha et al., 2010). AspAT could also contribute with oxaloacetate production to a unique system to explain ATP synthesis in seeds, termed ‘Perl's pathway’ (Perl, 1986; Botha et al., 1992). It depends on the fact that cytosolic malate dehydrogenase (MDHc) and PEPCK activities are already high in some seeds during the early phase of germination. In this ATP-synthesizing system MDHc provides NADH which is split by NADH-pyrophosphorylase yielding ADP. The latter is converted to ATP by pyruvate kinase (PK; Fig. 8). That PEPCK activity increases in germinating seeds is known from Arabidopsis (Penfield et al. 2004) and several other species (Botha et al., 1992; Ratajczak et al., 1998). MDHc and PEPCK protein also accumulate in Arabidopsis seeds (Supplementary Table S1, and references therein). In support of ‘Perl's pathway’, transcripts of MDHc, PK, and AspAT accumulate during early germination in Arabidopsis seeds (Fig. 8). Further research is needed to elucidate the possible role of this pathway in ATP production during seed germination. Cold stratification induces aspartate accumulation, but accumulation of TCA cycle metabolites derived from it is only evident upon subsequent incubation at 21 °C in the light (Angelovici et al., 2011). ABA induces the expression of glutamate decarboxylase (GAD; Supplementary Table S1 and Fig. S2) which produces γ-aminobutyric acid (GABA) associated with stress responses, and the GABA shunt for energy production is also evident in seeds (Shelp et al., 1995; Bouche et al., 2003).
Seeds store not only protein, oil, and starch, but also essential metals such as iron (Fe). In an innovative approach, Lanquar et al. (2005) identified the importance of vacuolar metal storage and activation during early germination in Arabidopsis seeds. Metal, in particular Fe, is mobilized during early germination by the redundant broad-range metal transporters NRAMP3 and 4. Germination of the double mutant is inhibited under conditions of Fe deficiency, as the seeds fail to retrieve Fe from the vacuole even though they contain as much Fe as the wild type. Fe, zinc, provitamin A (‘Golden Rice’), and folate are the most important micronutrients for which malnutrition can be improved by biofortification (Mayer et al., 2008). Research and breeding programmes are underway to enrich these compounds in crop seeds and depend on understanding seed metabolic engineering.
Novel directions and techniques for studying early seed germination
In the coming years, novel methods will lead to significant advances in our understanding of seed biology and plant evolution. New technologies situated at the interface of biology and disciplines such as material sciences, physical chemistry, and engineering offer the possibility to tackle new questions with interdisciplinary approaches.
Huge advances have been made in the area of imaging. Environmental scanning electron microscopy (eSEM) offers the opportunity to take high magnification images of living seeds (Fig. 4) (Muscariello et al., 2005; Windsor et al., 2000). These tools (eSEM and similar high resolution imaging techniques) can be used to tackle questions surrounding the structure of seeds and changes in these during germination in different seed types. eSEM is an excellent method to show the diversity in seed structures in different species and through the observation link back to questions concerning the evolution of morphologically different seed types.
New tools based on 1H-NMR imaging technology can be used not only to visualize and quantify water uptake (Manz et al., 2005; Wojtyla et al., 2006; Koizumi et al., 2008), but also for non-invasive imaging of seed oils (Neuberger et al., 2009). This enables scientists to assess spatial water distribution and oil content. Microsensors can be used to provide spatial and temporal oxygen maps of seeds (Rolletschek et al., 2009) which can be an important hint to the answer to the question of where and when fermentation processes occur in seeds. Kranner et al. (2010a) applied non-invasive infrared thermography to seeds and demonstrated that viability and even biochemical processes such as the dissolution of low molecular weight compounds could be assessed. This promising method is able to link biophysical with biochemical parameters and seed viability and could, for example, be used to distinguish different seed types by their thermographic behaviour during germination. MALDI-MS (matrix-assisted laser desorption/ionization mass spectrometry) imaging involves the visualization of the spatial distribution of proteins, peptides, metabolites, biomarkers, or other chemicals within thin tissue sections and might be a powerful tool for exploring the spatial distribution of nutrients in seeds. It has been used to visualize GABA in eggplant fruit sections where it localizes to the seeds (Goto-Inoue et al., 2010). 1H-NMR imaging, microsensors, MALDI-MS imaging, and thermography all can be used to quantify global changes in seeds in a spatial and temporal manner.
Bringing together confocal microscopy and computer-based image analyses, Sliwinska et al. (2009) created informative 3D images of an elongating embryo, which led to the localization of an elongation zone of the embryo. This could be further used for a cross-species approach to investigate conservation and biodiversity of embryo elongation zones in combination with established methods that can identify genes and proteins involved in cell expansion growth during germination. Quantitative phase tomography was used to elucidate structural details of Arabidopsis seeds (Cloetens et al., 2006). This technique uses a synchrotron-based approach to generate 3D, high-resolution images of a specimen to the cell level.
There have been great advances in the area of sequencing and epigenetics, which will greatly enhance our knowledge of germination in a wide array of species. Combined epigenetic (ChIP-seq) and transcriptome (RNA-seq) analyses with next-generation sequencing technologies will make it possible to analyse plants without a sequenced genome on a genomic scale (Bräutigam and Gowik, 2010). A combination of high-throughput sequencing with more classical methods can greatly advance our knowledge about developmental processes. Recently such a combined approach was used to study the developmental dynamics in maize leafs and identified 180 transcription factors for which now functional genomics studies would be interesting (Li et al., 2010). To describe dynamic networks from such results ‘Systems Biology Graphical Notation’ can be used, and corresponds to an engineer's view on regulation as it was published for seed development (Junker et al., 2010).
In addition, new databases and platforms designed specifically to collect and analyse information from high-throughput approaches to seed germination are now in place or being developed. The seed-specific gene ontology system TAGGIT facilitates the identification and visualization of the germination signature (Holdsworth et al., 2008), and the seed-specific eFP-browser and the eNorthern tool at the Toronto bar website (www.bar.utoronto.ca) visualizes Arabidopsis transcript expression patterns in seeds (Winter et al., 2007; Bassel et al., 2008). The co-expression tool (CORNET), similar to the Genemania browser from the Toronto bar website, provides an easy to use and helpful tool in finding interactions either in already published experiments or in user-supplied data, and thus helps to handle the generated amount of data better (De Bodt et al., 2010). These tools can help to generate hypotheses by supplying in silico data from already published experiments and help curate the massive amount of data generated by high-throughput analyses.
Seeds are starting to be used in cross-species systems biology approaches and interdisciplinary collaborations such as the European ‘virtual SEED’ network project (www.vseed.eu) where a molecular approach is combined with biophysical and morphological data, enabling the assembly of a more comprehensive model of seed germination. Understanding this process as a whole from the very beginning—from early seed germination to the establishment of the seedling—can help engineer and select for better and more robust crop species, thus increasing crop yield and quality.
Plant species developed a huge morphological and physiological diversity in seed types and states to match local environmental demands for germination timing. Darwin was aware of what he called the ‘vitality of seeds’ and made a connection to plant evolution: ‘The power in seeds of retaining their vitality when buried in damp soil may well be an element in preserving the species, and therefore seeds may be specially endowed with this capacity’ (Darwin, 1855a). Cross-species approaches in seed science and other areas of plant science will further increase our understanding of the evolution of plants. ‘Appropriate germination responses to environmental factors are the first requirement for successful growth and adaptation in any life-history trait; no subsequent life-history trait can even be expressed if the plant does not first survive past the germination stage. As such, germination timing can be a stringent selective sieve, determining which genotypes can establish in particular conditions.’ (Donohue, 2005). Seed germination and dormancy are indeed the most important early life-history traits.
Supplementary data are available at JXB online.
Figure S1. Regulation of GA- and ABA-related transcripts by cold stratification of Arabidopsis thaliana seeds.
Figure S2. Regulation of transcripts for the energy metabolism of Arabidopsis thaliana seeds by imbibition, cold stratification, GA, and ABA.
Table S1. Transcriptome analysis for energy metabolism genes during Arabidopsis thaliana germination sensu strictu and its regulation by hormones and cold stratification.
We thank Dr Ralf Thomann (FMF, Freiburger Materialforschungszentrum, University of Freiburg, Germany, www.polymermicroscopy.com) for expert advice and the seed eSEM images, and our colleague Kai Gräber for his critical reading and helpful comments. Our work is funded by the ERA-NET Plant Genomics grant vSEED (grant no. DFG LE720/8) to GLM and by a postdoctoral fellowship of the Deutsche Forschungsgemeinschaft to KM (grant no. MU3114/1-1); these are gratefully acknowledged.