Deciphering gene regulatory networks that control seed development and maturation in Arabidopsis


*(fax +33 1 3083 3111; e-mail


Seeds represent the main source of nutrients for animals and humans, and knowledge of their biology provides tools for improving agricultural practices and managing genetic resources. There is also tremendous interest in using seeds as a sustainable alternative to fossil reserves for green chemistry. Seeds accumulate large amounts of storage compounds such as carbohydrates, proteins and oils. It would be useful for agro-industrial purposes to produce seeds that accumulate these storage compounds more specifically and at higher levels. The main metabolic pathways necessary for oil, starch or protein accumulation are well characterized. However, the overall regulation of partitioning between the various pathways remains unclear. Such knowledge could provide new molecular tools for improving the qualities of crop seeds (Focks and Benning, 1998, Plant Physiol. 118, 91). Studies to improve understanding of the genetic controls of seed development and metabolism therefore remain a key area of research.

In the model plant Arabidopsis, genetic analyses have demonstrated that LEAFY COTYLEDON genes, namely LEC1, LEC2 and FUSCA3 (FUS3), are key transcriptional regulators of seed maturation, together with ABSCISIC ACID INSENSITIVE 3 (ABI3). Interestingly, LEC2, FUS3 and ABI3 are related proteins that all contain a ‘B3’ DNA-binding domain. In recent years, genetic and molecular studies have shed new light on the intricate regulatory network involving these regulators and their interactions with other factors such as LEC1, PICKLE, ABI5 or WRI1, as well as with sugar and hormonal signaling. Here, we summarize the most recent advances in our understanding of this complex regulatory network and its role in the control of seed maturation.

Introduction to seed development and maturation

Evolutionary aspects

Seed development and maturation represent an evolutionary advantage that allows most plants to cope with unfavorable environmental conditions by interrupting their life cycle and resuming growth when placed under favorable conditions (Bentsink and Koornneef, 2002; Bewley, 1997). There is considerable variability among plants in terms of seed morphology, physiology and maturation, leading to the accumulation of various compounds such as oil, seed storage proteins (SSPs) and starch in the various tissues of the seed. In fact, seed maturation is not even a compulsory step in the plant life cycle. Lower plants (bryophytes, pteridophytes or algae) do not produce seed, and embryo maturation is consequently alleviated in these species (Harada, 1997; Vicente-Carbajosa and Carbonero, 2005). Even among angiosperms, several species produce embryos that are able to grow when excised in an immature state from the maternal tissues. Similarly, viviparous mutants that are defective in seed maturation can produce viable seedlings. Together with phylogenetic analyses, these observations strongly suggest that seed maturation constitutes an intrusive developmental and metabolic phase introduced into the ancestral plant life cycle, i.e. uninterrupted embryo growth and seedling development (Figure 1). It may be hypothesized that pre-existing physiological responses and developmental processes allowing plants to cope with unfavorable environmental conditions such as dehydration have been retained to develop this pathway and permit seed maturation. Indeed, several seed proteins such as late embryogenesis-abundant (LEA), oleosins and SSPs are also found in angiosperm pollen (Huang, 1996; Wise and Tunnacliffe, 2004; Zakharov et al., 2004) or in spores of bryophytes (Schallau et al., 2008). In addition, molecular analyses support the conservation of regulatory pathways controlling gene expression in fern spores and in seeds of both gymnosperms and angiosperms (Schallau et al., 2008).

Figure 1.

 Proposed model of genetic and molecular interactions in the regulatory network involved in the control of seed development and maturation in Arabidopsis thaliana. Arrows and T bars indicate positive and negative effects, respectively. The factors that induce and/or maintain seed maturation are shown in red. The factors that promote cell growth and differentiation are shown in blue. The numbers indicate the various targets of the regulators.

Genetic control of the maturation phase

The maturation processes occurring in various seed tissues (i.e. seed coat, endosperm and embryo) contribute to seed quality, allowing efficient dispersal and establishment of the seedlings (Welbaum et al., 1998). Seed quality relies, therefore, on the tight control of embryo morphogenesis, maturation and germination. This raises the question of the genetic and molecular relationships that exist between the various developmental and metabolic phases. Several observations suggest that the various processes can be disconnected to some extent. Various mutants affected in embryogenesis can express seed maturation-specific genes and accumulate some storage compounds (Devic et al., 1996; Lepiniec et al., 2005; Yadegari et al., 1994). It has also been suggested that maturation and post-germinative growth could occur simultaneously in the embryo (Finkelstein and Crouch, 1986), although expression of maturation- and germination-specific genes may not overlap at the cellular level. Furthermore, we do not know whether the maturation processes occurring in angiosperms have arisen progressively during evolution through the elaboration of several intricate regulatory mechanisms or whether they have arisen from a small number of pleiotropic modifications such as the mutation of homeotic genes. Simple genetic switches may exist to shift from embryogenesis to a maturation program, and subsequently to germination. These genetic programs may be exclusive at the cellular level but could occur simultaneously in the same embryo. Finally, the nature and origin of the molecular mechanisms that control both the entrance into the maturation phase and those that prevent cell growth and division remain to be elucidated.

Arabidopsis as a model to study seed development and maturation

In Arabidopsis, seed development and maturation are well documented (Baud et al., 2002; Goldberg et al., 1994; Harada, 1997; Laux and Jurgens, 1997; Laux et al., 2004; Lehti-Shiu et al., 2005; Lepiniec et al., 2005, 2006; Mansfield and Briarty, 1991; Mansfield et al., 1991; Mayer et al., 1991; Weijers and Jurgens, 2005; West and Harada, 1993). Briefly, embryo morphogenesis is initiated by double fertilization of the embryo sac, giving rise to the endosperm and the zygote. The zygote divides asymmetrically to form apical and basal cells leading to the embryo proper and the suspensor, respectively. A precise order of events ensures the correct relative positioning of the various tissues and organs (meristems, cotyledons and hypocotyl) and the arrangement of cell types within each tissue of the embryo. At the heart stage of embryo development (about 7 days after fertilization), most of the cell division and differentiation events have already occurred, i.e. the protoderm has differentiated into the epidermis, the pro-vascular bundles are ready to form the vascular system, and the overall shape of the embryo is determined with organization of both the apico-basal (shoot and root meristems) and lateral (cotyledons) symmetries. During this first phase, the triploid endosperm develops through a syncytial phase that is followed by cellularization and differentiation events.

Starch and hexoses accumulate only transiently during seed development and very low amounts remain in dry seed. In contrast, sucrose and some oligosaccharides gradually accumulate at the end of the maturation phase (Baud et al., 2002; Focks and Benning, 1998). Interestingly, the decrease in the hexose to sucrose ratio correlates with transition to the maturation phase. This observation is consistent with the recently demonstrated role of the AtSUC5 sucrose transporter in the Arabidopsis endosperm (Baud et al., 2005) and of APETALA2 in the control of the hexose to sucrose ratio and seed mass (Ohto et al., 2005). Taken together, these data suggest a signaling function for sugars during the transition from embryo morphogenesis to maturation in Arabidopsis, as already suggested for other plant species (Gutierrez et al., 2007; Weber et al., 2005). However, the role of the hexose to sucrose ratio in oilseeds remains under debate (Tomlinson et al., 2004). Trehalose-6-phosphate is also thought to play a critical role in triggering seed maturation, but the molecular mechanism involved remains unclear (Eastmond et al., 2002; Gomez et al., 2006).

During the maturation phase, embryo growth and cell cycle activities are stopped (Raz et al., 2001). The embryo goes through a period of cellular expansion and differentiation, concomitant with reduction of the endosperm to one cell layer and the onset of maturation (until around 17–20 days after fertilization). The embryo accumulates nitrogen compounds (proteins) and carbohydrate storage compounds (lipids) that each account for 30–40% of the seed dry matter (Baud et al., 2002; Mansfield and Briarty, 1992). Lipids accumulate in the form of triacylglycerols (TAG) in cytosolic oil bodies, which occupy about 60% of the cell volume in the cotyledons of mature embryos. During late maturation, the seed becomes metabolically quiescent and tolerant to desiccation. The metabolic pathways leading to accumulation of the main storage compounds, including oil, are well characterized (Baud et al., 2002; Fait et al., 2006; Ruuska et al., 2002; Schwender et al., 2004a,b). The tremendous interest in using oils for industrial applications is emphasized and reviewed elsewhere in this special issue (Durrett et al., 2008; Dyer et al., 2008). However, genetic engineering has routinely resulted in only a low accumulation of the desired fatty acids (Dyer and Mullen, 2008), and the genetic, cellular and molecular mechanisms that regulate the accumulation and partitioning between various pathways remain unclear. Nevertheless, a few transcription factors that appear to collectively control the various facets of seed maturation have been characterized during recent years and are described below.

Characterization of ABI3, FUS3, LEC2 and LEC1

Identification of the loci

Mutations of ABI3 (Giraudat et al., 1992; Koornneef et al., 1984), LEC1 (Lotan et al., 1998; Meinke, 1992; West et al., 1994), LEC2 (Meinke et al., 1994; Stone et al., 2001) and FUS3 (Baumlein et al., 1994; Gazzarrini et al., 2004; Keith et al., 1994; Luerssen et al., 1998) genes lead to similar pleiotropic effects on seed phenotype (Harada, 2001; Holdsworth et al., 1999; Parcy et al., 1997; Vicient et al., 2000). Early during embryogenesis, the LEC1, LEC2 and FUS3 genes are required to maintain embryonic cell fate and to specify cotyledon identity. Mutant cotyledons display some of the characteristics of young leaves, exhibiting both trichomes on their surface and a complex vascular pattern. The mutant embryos may also present abnormal suspensor phenotypes, precocious cell-cycle activation, and growth of apical and root meristems. Later during embryogenesis, these genes, together with ABI3, are also involved in initiation and maintenance of the maturation phase. Mutant seeds are less tolerant to desiccation, accumulate lower levels of storage compounds, and instead accumulate anthocyanin pigments and/or are affected in chlorophyll breakdown. The dormancy of mutant seeds is also modified. In a humid environment, they can display precocious germination, and mutant combinations show extreme viviparity.

Nevertheless, the mutant phenotypes are not identical. For instance, abi3 is the only mutant that is highly resistant to abscisic acid (ABA) and the mutation does not affect cotyledon identity (Giraudat et al., 1992). However, abi3 mutations strongly affect the accumulation of SSPs (more so than fus3), and both the quantity and quality of seed lipids are also modified. The lec2 phenotype is the least severe. The mutant seeds are less sensitive to desiccation and only slightly affected in the accumulation of storage compounds (Meinke et al., 1994; Stone et al., 2001). There are also differences in the tissues affected (To et al., 2006). ABI3 is important for the expression of some SSP genes (e.g. At2S3) in the embryo axis, the center of the cotyledons and the endosperm. FUS3 is necessary in the axis and the endosperm. Although the lec2 mutant lacks SSPs in some sectors of the cotyledons, LEC2 is required throughout the embryo. Finally, analyses of mutant combinations have demonstrated that these regulatory proteins can act at least partially redundantly and sometimes synergistically, depending on the traits and the tissues studied (Meinke et al., 1994; Parcy et al., 1997; Raz et al., 2001; To et al., 2006).

Structure of the LEC and ABI3 genes

Consistent with their partial functional redundancy, the three genes LEC2, FUS3 and ABI3 encode related transcription factors of the B3 domain family (Giraudat et al., 1992; Luerssen et al., 1998; Stone et al., 2001). The B3 DNA-binding domain of 120 amino acid residues was originally identified as the third basic region of the ABI3 protein and of its maize homolog VP1 (McCarty et al., 1991). This domain is also found in other classes of transcription factors (Qu and Zhu, 2006), namely the ARFs (auxin response factors; Ulmasov et al., 1997), the RAVs (‘related to ABI3/VP1’; Kagaya et al., 1999) and the HSI2/VAL family (Suzuki et al., 2007; Tsukagoshi et al., 2005, 2007). This domain shares some similarities with the DNA-binding domain of a prokaryotic endonuclease. The structure of the domain has been determined, allowing a model of its interaction with DNA to be proposed (Yamasaki et al., 2004). LEC1 belongs to a different class of proteins and is homologous to HAP3 subunits of the CAAT box-binding factors (CBFs), a family of heteromeric transcription factors (Lee et al., 2003; Lotan et al., 1998).

Expression of the LEC and ABI3 genes

Expression patterns have been characterized for these four genes (Figure 2). LEC1 is specifically expressed in seed and is detected in both the embryo and the endosperm, early during embryogenesis (Lee et al., 2003; Lotan et al., 1998). LEC2 is mainly expressed during early embryo development, although it is also detected occasionally in vegetative tissues (Kroj et al., 2003; Stone et al., 2001). Expression of FUS3 has mainly been detected in the protodermal tissue of the embryo (Gazzarrini et al., 2004; Tsuchiya et al., 2004). Consistent with this localization, specific expression of FUS3 in the protodermal cell layer can rescue most of the fus3 phenotypes, including accumulation of SSPs in various cell layers of the embryo (Gazzarrini et al., 2004). This L1-specific expression pattern is difficult to reconcile with a direct effect of FUS3 on the promoter of SSP genes throughout the embryo. It has been proposed that FUS3 acts indirectly through the regulation of TTG1 expression and hormonal levels. Recent data have shown that ABI3 is expressed in the whole embryo, which is consistent with the expression pattern of its target genes (To et al., 2006). ABI3 is also detected in vegetative organs, and has a role in lateral meristem development (Rohde et al., 2000). However, the level of gene expression is very low for all the regulatory genes studied, making it difficult to characterize mRNA accumulation by using in situ hybridization. Most of the data available to date have been obtained using reporter genes controlled by native promoters in transgenic plants. Although these approaches have given consistent results (To et al., 2006), complementary experiments are required to describe more comprehensively the tissue and cellular expression of these regulatory genes and take into account possible post-transcriptional regulation. To date, only a pFUS3:FUS3:GFP construct has been used, providing results consistent with epidermal localization of the protein (Gazzarrini et al., 2004). The site of protein accumulation at the tissue and intracellular levels remains to be investigated or confirmed for the other factors.

Figure 2.

 Tissue-specificity of the B3 regulatory network in the developing embryo of Arabidopsis thaliana. The areas where gene expression is detected are indicated in red. The expression patterns were determined using a promoter:GUS fusion for LEC2, a promoter:cDNA:GFP fusion for FUS3, and in situ hybridization (ISH) for both ABI3 and LEC1.

A complex and intricate regulatory network

Cross-talk and feedback regulation

Although lec2, fus3 and abi3 mutants display similar alterations of the maturation process, they also exhibit some specific phenotypes that appear to be additive in double mutants, suggesting that the three proteins belong to parallel regulatory pathways that partially overlap (Keith et al., 1994; Meinke et al., 1994; West et al., 1994). Various genetic analyses have established the existence of interactions between these genes, but their true nature remains to be elucidated at the molecular level (Brocard-Gifford et al., 2003; Nambara et al., 2000; Parcy et al., 1997; Raz et al., 2001). FUS3 and LEC2 were shown to act in partially redundant pathways to locally regulate FUS3 expression (Kroj et al., 2003). In addition, analyses of double mutants with lec1, lec2 and fus3 suggested that LEC1 could act upstream of LEC2 and FUS3 (Meinke et al., 1994). Finally, a recent exhaustive genetic analysis of the regulation of gene expression in single and multiple mutant backgrounds (To et al., 2006) demonstrated that expression of both FUS3 and ABI3 is controlled by a complex network of local and redundant regulations involving LEC1 and LEC2, and also FUS3 and ABI3 themselves (Figure 3). For instance, FUS3 expression is regulated by LEC2 and FUS3 itself in the root tip, by LEC2 and ABI3 in the embryo axis, and by the four regulators in cotyledons. The expression of ABI3 is also controlled by the four regulatory proteins in cotyledons. However, ABI3 expression is independent of any of these regulators in the embryo axis. These results are in agreement with molecular studies demonstrating that ectopic expression of LEC1 or LEC2 can induce FUS3 and ABI3 expression (Kagaya et al., 2005b; Santos Mendoza et al., 2005). It would be interesting to know to what extent ABI3 and FUS3 can complement lec1 mutation, as previously tested for the lec2 mutant (To et al., 2006). Nevertheless, such experiments cannot prevent non-specific ‘B3’ effects due to constitutive (non-physiological) expression of FUS3 or ABI3 that would alter their binding specificity. Additional experiments conducted in mutant backgrounds are necessary to confirm these results (e.g. ectopic expression of FUS3 in abi3 lec2 double mutants or ABI3 in a lec2 fus3 background). Similarly, direct induction of various target genes by LEC2 (Braybrook et al., 2006; Kroj et al., 2003; Santos Mendoza et al., 2005) should be confirmed in the fus3 abi3 mutant background. Interestingly, at least two members of the recently characterized HIS/VAL family of transcriptional repressors (i.e. HSI2 and HSL1) redundantly inhibit the B3 maturation network in seedlings (Suzuki et al., 2007; Tsukagoshi et al., 2005, 2007). The fact that these regulators may function by repressing sugar-inducible genes is consistent with a role of sugar signaling in the maturation process (see above and Figure 1).

Figure 3.

 A complex network of local and redundant regulations. Schematic representation of the local genetic controls in various embryo tissues (Lotan et al., 1998; To et al., 2006; Tsuchiya et al., 2004, and references therein).

Hormonal signaling

The key role of auxin in plant embryogenesis has been well described (Jenik and Barton, 2005; Weijers and Jurgens, 2005). Nonetheless, only minor auxin-related phenotypes have been observed during early embryogenesis of the lec mutants, such as the abnormal development of the suspensor detected occasionally in lec1 and lec2 (Lotan et al., 1998; Stone et al., 2001) and reduced length of the axis. Nevertheless, recent data suggest that auxin signaling may interfere with the B3 regulatory network. The ability of lec1 and lec2 mutants to form somatic embryos is strongly reduced (Gaj et al., 2005). Conversely, ectopic expression of LEC1 and LEC2 in vegetative cells can trigger the formation of embryo-like structures (Lotan et al., 1998; Stone et al., 2001). Interestingly, some key proteins in auxin signaling (i.e. ARFs) also belong to the B3 family, and an ABI3 homolog of bean (Phaseolus vulgaris) is able to bind the auxin-responsive element (Nag et al., 2005). Consistent with the latter result, ABI3 appears to be involved in auxin signaling as well as lateral root development (Brady et al., 2003). Additionally, ectopic expression of LEC2 causes rapid activation of an auxin-responsive gene (Braybrook et al., 2006). Ectopic expression of FUS3 in the epidermal cells leads to a reduced venation pattern in cotyledons, and auxin can induce FUS3 expression (Gazzarrini et al., 2004). Finally, it has been proposed that the role of LEC1 in the promotion of embryonic cell identity and division requires auxin and sucrose (Casson and Lindsey, 2006). Despite this wealth of potential links between auxin signaling and seed maturation regulators, their roles remain to be established.

Mutant analyses have not been sufficient to demonstrate the role of both ABA and gibberellin (GA) during embryo morphogenesis. Indeed, either some traces of ABA are found in the mutant seeds or the supply of GA is required for obtaining flowers and seeds. Nevertheless, ABA immunomodulation in transgenic tobacco seeds (Phillips et al., 1997) and ectopic expression of a pea GA2 oxidase in Arabidopsis (Singh et al., 2002) support an essential role for ABA during maturation and for GA in seed development, respectively. In addition, it has been firmly established that the ABA/GA ratio is a key regulator of both maturation and germination processes (Bentsink and Koornneef, 2002; Debeaujon and Koornneef, 2000; Dubreucq et al., 1996; Finkelstein et al., 2002; Giraudat et al., 1994; Karssen et al., 1983; Koornneef et al., 1982, 2002; Ogawa et al., 2003). As the embryo enters the maturation phase, ABA content increases in seed, and the resulting high ABA/GA ratio promotes maturation, induces dormancy and inhibits cell-cycle progression, embryo growth and germination. FUS3 and LEC2 have been shown to inhibit GA biosynthesis by repressing the expression of GA biosynthetic genes (Curaba et al., 2003; Gazzarrini et al., 2004). Consequently, in lec2 and fus3 mutants, the ABA/GA ratio is lower than in wild-type seeds. This is fully consistent with the defects observed during the maturation phase and with the precocious cell differentiation and growth of mutant embryos. Lack of GA biosynthesis was shown to be epistatic to the lec mutations for the formation of trichomes, suggesting that GAs act downstream of the LEC genes. Similarly, ABA is required for the induction of some SSPs in seedlings ectopically expressing FUS3 (Kagaya et al., 2005a,b) or ABI3 (Parcy et al., 1994). Conversely, it has been shown that the activity of ABI3 and FUS3 can be regulated at the post-translational level by ABA and/or GA. AIP2 (ABI3-interacting protein 2) is an E3 ligase, whose expression is under the control of ABA, that can trigger the degradation of ABI3 (Zhang et al., 2005). This regulation could ensure rapid degradation of ABI3 during imbibition, thus promoting germination. Similarly, it has been suggested that ABA and GA could regulate the stability of FUS3 (Gazzarrini et al., 2004).

Other regulators of seed maturation

Other important regulators for seed development and/or maturation have been identified by genetic analyses. Interestingly, PICKLE (PKL), a chromatin-remodeling factor (CHD3), acts in concert with GA to repress embryonic traits during and after germination, including the expression of LEC1, LEC2 and FUS3 in roots (Dean Rider et al., 2003; Henderson et al., 2004; Li et al., 2005; Ogas et al., 1997, 1999; Rider et al., 2004). Consistent with these results, it has been shown recently that PKL is also necessary to repress ABI3 and ABI5 expression during germination, in response to ABA (Perruc et al., 2007). Some reports have suggested that epigenetic mechanisms (i.e. histone modifications) could repress the expression of seed-specific genes during Arabidopsis seed germination (Tai et al., 2005) or control the activation of maturation-specific genes by an ABI3 homolog in bean (Ng et al., 2006). This type of regulation has been recently confirmed by the demonstration that two histone deacetylases, namely HDA6 and HDA19, redundantly inhibit embryonic properties during germination via the repression of LEC1, FUS3 and ABI3 (Tanaka et al., 2008). It has also been demonstrated that the expression of FUS3 could be regulated by some members of the polycomb protein family involved in the epigenetic control of seed and plant development (Makarevich et al., 2006). LEC1-like (L1L) is closely related to LEC1, but it has a different function during embryogenesis due to its specific pattern of gene expression (Kwong et al., 2003). In effect, L1L is expressed earlier than LEC1 during embryo development and is necessary for early embryogenesis. Mutations of ABI4 and ABI5 also affect seed maturation, although the effects of such mutations are limited compared to those of abi3 (Finkelstein and Lynch, 2000; Finkelstein et al., 1998, 2002). The ABI4 protein belongs to a family of transcription regulators that contain a plant-specific APETALA2 (AP2) domain (Finkelstein et al., 1998) and acts downstream of ABI3 (Brocard-Gifford et al., 2003; Soderman et al., 2000). ABI5 is a transcription factor of the bZIP family (Finkelstein and Lynch, 2000; Lopez-Molina and Chua, 2000) that acts in the same signaling pathway as ABI4 (see below). The TAN gene encodes a protein with a WDR motif that could interact with other proteins to control various aspects of both early and late phases of embryo development (Yamagishi et al., 2005). The tan mutant shares many characteristics with the lec mutants, suggesting that these genes may have overlapping roles during embryogenesis. Finally, the homeobox GLABRA2, which is known to regulate the formation of trichomes and root hairs, has recently been shown to also be involved in the control of oil accumulation in seeds (Shen et al., 2006), but its role in the regulatory network is still unknown.

Identifying the targets of the B3 regulatory network

Target promoters and the role of RY motifs

It has been shown that B3-type transcription factors can act directly on the expression of genes encoding storage proteins (Vicente-Carbajosa and Carbonero, 2005). LEC2 has been shown to exert direct control over At2S1–S4 and 2S-like gene expression (Braybrook et al., 2006; Kroj et al., 2003). FUS3 and ABI3 can also directly contribute to the induction of storage protein gene expression (Ezcurra et al., 2000; Reidt et al., 2000). The three B3-type regulatory proteins directly activate the target genes by binding to RY motifs present in the promoters (Ezcurra et al., 2000; Monke et al., 2004; Reidt et al., 2000; Reinders et al., 2002). Indeed, the RY box (CATGCA) is necessary for the correct expression of several seed-specific genes in Arabidopsis (Baumlein et al., 1986, 1992; Conceicao Ada and Krebbers, 1994; Ellerstrom et al., 1996; Stalberg et al., 1993), legumes (Bobb et al., 1997; Dickinson et al., 1988) and maize (Suzuki et al., 1997), for example. Furthermore, it has been shown that FUS3, LEC2 and ABI3 can bind an RY motif in vitro (Braybrook et al., 2006; Monke et al., 2004; Reidt et al., 2000), and that LEC2 and FUS3 bind an RY motif in a yeast one-hybrid assay (Kroj et al., 2003).

However, RY elements probably do not act alone in seed promoters (i.e. they may not be sufficient to confer the correct expression of target genes). It has been shown that other DNA motifs (e.g. G-box) and factors binding these motifs are required for the correct expression of target genes during seed maturation (Bensmihen et al., 2002; Brocard-Gifford et al., 2003; Ezcurra et al., 1999; Hobo et al., 1999; Kurup et al., 2000; Nakashima et al., 2006; Sakata et al., 1997; Vicente-Carbajosa and Carbonero, 2005; Wobus and Weber, 1999). Several bZIP transcription factors have been shown to play a role during seed development. ABI5, for example, affects ABA sensitivity and controls the expression of some LEA genes in seeds (Carles et al., 2002; Finkelstein and Lynch, 2000; Lopez-Molina and Chua, 2000). The ABI5 protein binds to abscisic acid-responsive element (ABRE) cis-elements that are present in the promoters of several LEA genes such as AtEM1 or AtEM6. Interestingly, ABI5 expression is itself regulated by ABI3, and both proteins interact in yeast two-hybrid assays, suggesting that ABI3 induces the expression of one of its partners in a transcriptional complex (Lopez-Molina et al., 2002; Nakamura et al., 2001). ABI5 belongs to a bZIP subfamily (Jakoby et al., 2002) that includes other members involved in seed maturation or ABA signaling in vegetative tissues (Bensmihen et al., 2002; Brocard-Gifford et al., 2003; Choi et al., 2000; Finkelstein et al., 2005; Johnson et al., 2002). Some bZIP proteins from another subgroup (i.e. bZIP10 and bZIP25) that are homologous to maize OPAQUE2 also play a role during seed maturation. These transcription factors act together with ABI3 to regulate the expression of SSP genes (Lara et al., 2003).

Limitations of genetic approaches and interest in inducible systems

It is important to note that, due to partial functional redundancy, the intricate regulatory network and conservation of the B3 domain in the three proteins LEC2, FUS3 and ABI3, it is difficult to draw conclusions from genetic analyses regarding the precise mechanisms and specific function for each regulatory protein in planta. Ectopic expression of a regulatory protein can also lead to misinterpretation of the real network (providing an non-specific B3 effect). In order to decipher the specific effects of the various regulators, several groups have taken advantage of inducible systems to control the expression of LEC1, LEC2 and FUS3 to identify direct target genes (Braybrook et al., 2006; Kagaya et al., 2005a,b; Santos Mendoza et al., 2005; Wang et al., 2007). These methods, coupled to quantitative PCR experiments and/or transcriptomic analyses, have confirmed some previously identified putative direct targets and revealed a set of new potential target genes.

Among the target genes identified using an inducible system (and confirmed in vitro) is AGAMOUS-LIKE15 (AGL15; Braybrook et al., 2006), which encodes a MADS box protein that is expressed preferentially in seed (Lehti-Shiu et al., 2005). Although the agl15 mutant does not display abnormal seed phenotypes, ectopic expression of AGL15 affects the embryonic program and enhances the competency of shoot apical meristems to undergo somatic embryogenesis (Adamczyk et al., 2007). Interestingly, it has been shown that AGL15 directly regulates expression of a gene encoding an enzyme that is involved in the oxidation of active gibberellin (Wang et al., 2004), providing some clues about the putative role of AGL15 in seed development.

Other direct target genes include oleosins, which are involved in the formation of oil bodies. Nevertheless, none of the genes required for the synthesis of TAGs stored in these oil bodies appeared to be regulated by LEC2. This result was all the more intriguing as the induction of LEC2 in maturing embryos correlates well with the onset of oil deposition. Moreover, the ectopic expression of LEC2 in developing leaves was shown to be sufficient to trigger TAG accumulation in these tissues (Santos Mendoza et al., 2005). These results suggested that intermediate regulators might be involved in the control of TAG biosynthesis. Indeed, it has been shown that ectopic expression of FUS3 can trigger the expression of fatty acid biosynthetic genes (Wang et al., 2007). In addition, we have recently demonstrated that the transcriptional activator WRINKLED1 (WRI1), a direct target of LEC2, is necessary for the regulation of oil biosynthesis by LEC2 (Baud et al., 2007a).

Control of WRI1 and the TAG biosynthetic pathway

In contrast to the B3-type master regulators that exhibit a broad control on seed maturation, WRI1, a transcription factor of the AP2/EREB family, has an impact on more specific aspects of the maturation process (Cernac and Benning, 2004). wri1 mutants produce wrinkled seeds with severe depletion of TAGs (Baud et al., 2007a; Cernac and Benning, 2004). In this seed mutant, which has a low oil content, carbohydrate metabolism is compromised (Baud and Graham, 2006; Focks and Benning, 1998), rendering maturing embryos unable to efficiently convert sucrose into TAGs. Microarray experiments and quantitative RT-PCR analyses on mutant wri1 seeds and Pro35Sdual:WRI1 lines led to isolation of some putative targets of WRI1 (Andre et al., 2007; Baud et al., 2007a,b; Ruuska et al., 2002). These include several genes encoding enzymes from glycolysis and the fatty acid biosynthetic network (Figure 4). Taken together, these data indicate that WRI1 specifies the regulatory action of LEC2 towards the metabolic network involved in the production of storage fatty acids. These results exemplify how metabolic and developmental processes affecting the maturing embryo can be coordinated at the molecular level. Further analyses are now required to identify the cis-regulatory element recognized by WRI1 and determine both the transcriptional and post-transcriptional regulation of this factor. At the transcriptional level, LEC1 may participate in regulation of WRI1 expression (Casson and Lindsey, 2006). Likewise, sucrose may play a role in triggering the induction of WRI1, as well as of LEC2, FUS3 and ABI3 (Masaki et al., 2005; Tsukagoshi et al., 2007).

Figure 4.

 Control of storage compound synthesis and accumulation in maturing seeds of Arabidopsis thaliana. The precursors for de novo fatty acid synthesis in maturing embryos are derived from sucrose through the glycolytic pathway and/or the OPPP. Plastidial pyruvate kinases (PKps) play a key role in provision of these precursors. Fatty acids produced in the plastids are then exported towards the cytosol in the form of acyl CoAs and used to form triacylglycerols, ultimately stored in oil bodies. The amino acids required for the synthesis of storage proteins during the maturation process are either directly imported from the maternal tissues or synthesized/modified in the embryonic tissues. Storage proteins are ultimately stored in specific vacuoles. Solid arrows represent positive transcriptional regulations. Target genes encoding members of the metabolic network leading to storage compound synthesis/accumulation are indicated in italics. FA, fatty acids; OPPP, oxidative pentose phosphate pathway; SSP, seed storage proteins; TAG, triacylglycerides; TCA cycle, tricarboxylic acid cycle.


As exemplified in this special issue, there is a strong interest in using plant storage compounds for industrial applications (biomass for biofuels and biomaterials). For instance, plant oil is one of the most energy-rich and abundant forms of reduced carbon available from nature and represents a possible substitute for conventional diesel (Durrett et al., 2008). However, for industrial and economic reasons, the accumulation of specific storage compounds needs to be increased. Interestingly, master seed regulators can directly control the expression of SSPs and/or activate secondary transcription factors that are able to trigger other transcription programs. Among the latter, WRI1 is specifically involved in the regulation of oil biosynthesis and is therefore an interesting candidate for biotechnology applications. However, our understanding of the gene regulatory networks that control seed development and maturation is still limited. Several transcription factors, their interactions and their target genes remain to be characterized. In addition, other levels of regulation (e.g. post-translational or metabolic) will have to be taken into account for efficient metabolic engineering. Finally, the model network will have to be extended to other plant species.

The current data indicate a complex regulatory scheme in which LEC1 and LEC2 initiate and control seed maturation and prevent germination, together with FUS3 and ABI3. Recent genetic and molecular analyses have shed new light on this intricate regulatory network and led to the elaboration of a model that seems to be fully coherent with phenotypic analyses of single and double mutants (Figure 1). The four regulators act in concert with hormones (auxin, ABA and GA), epigenetic mechanisms (e.g. PKL protein) and target regulatory proteins such as WRI1, ABI5 or AGL15. In addition, other master regulatory genes have been identified (e.g. TAN and L1L), although their exact function in the network is not known.

It may be hypothesized that this complex network provides robust and tight control of seed maturation. These results also emphasize that phenotypic analyses must be carried out at the cellular level to unravel complex regulatory traits. Furthermore, they highlight some of the limitations and drawbacks of the various analyses carried out so far, making additional experiments necessary to firmly establish the specificity and precise function (i.e. identification of their target genes) of the various factors in planta. The use of inducible systems coupled to transcriptomic analyses, as well as analyses of the molecular interactions (protein–protein and protein–DNA) in planta using, for instance, chromatin immunoprecipitation (ChIP) or bimolecular fluorescence complementation (BiFC) could provide interesting data to reinforce the model.

An interesting and still open question is the origin of this complex regulatory network. It is possible that the three B3-type regulators (LEC2, FUS3 and ABI3) derive from a common ancestor with auto-regulatory properties (To et al., 2006). ABI3 is present in various vegetative tissues and fulfils several functions related to plastid development, control of flowering time, outgrowth of axillary meristems and lateral root formation (De Meutter et al., 2005; Horvath et al., 2003). ABI3 also appears to be conserved among plants, as homologous genes have been found in monocots (Hattori et al., 1994; McCarty et al., 1991), gymnosperms (Zeng and Kermode, 2004) and in the moss Physcomitrella (Marella et al., 2006). Interestingly, the homologous maize protein (VP1) has been shown to complement the abi3 mutation (Suzuki et al., 2001), suggesting that ABI3 is closely related to the ancestral B3 protein. Interestingly, recent findings strongly support the conservation of regulatory pathways controlling gene expression in fern spores and in seeds of both gymnosperms and angiosperms (Schallau et al., 2008). Nevertheless, despite the recent identification of a putative FUS3 homolog in monocots (Moreno-Risueno et al., 2008), data from more distantly related plants (e.g. gymnosperms and lower plants) are still lacking, impeding a robust phylogenetic analysis of the evolutionary history of the B3 factors involved in the control of seed maturation.


We wish to acknowledge several colleagues: F. Berger, M. Delseny, M. Devic, J. Giraudat, A. Marion-Poll, M. Miquel, C. Rochat and T. Roscoe for discussions about seed development and maturation; N. Berger, E. Harscoët, J. Kronenberger and A. To, for their invaluable contribution to parts of the work presented here; and finally H. North and the reviewers for helpful comments and correcting the manuscript. Part of our work on Arabidopsis seed development and transcriptional regulation is supported by grants from the Agence Nationale pour la Recherche (ANR) Genoplante ‘Arabidoseed’ (TRIL-033) and the ANR ‘TF code’ (ANR-07-BLAN-0211-02). S.B. is a Chargé de Recherche at CNRS.