•Seed longevity is an important trait in many crops and is essential for the success of most land plant species. Current knowledge of its molecular regulation is limited. The Arabidopsis mutants abscisic acid insensitive3-5 (abi3-5) and leafy cotyledon1-3 (lec1-3) have impaired seed maturation and quickly lose seed viability. abi3-5 and lec1-3 were used as sensitized genetic backgrounds for the study of seed longevity.
•We exploited the natural variation of Arabidopsis to create introgression lines from the Seis am Schlern and Shahdara accessions in, respectively, the abi3-5 and lec1-3 backgrounds. These lines carry natural modifiers of the abi3 and lec1 phenotypes. Longevity tests and a proteomic analysis were conducted to describe the seed physiology of each line.
•The modifier lines showed improved seed longevity. The Shahdara modifiers can partially re-establish the seed developmental programs controlled by LEC1 and restore the accumulation of seed storage proteins that are reduced in abi3-5 and lec1-3.
•The isolation and characterization of natural modifiers of the seed maturation mutants abi3-5 and lec1-3, and the analysis of their seed proteomes, advance our current understanding of seed longevity.
Seed longevity defines the life span of seeds and is a crucial trait for the germplasm conservation of endangered and cultivated species. Seed longevity also influences the uniformity of germination and crop establishment. When stored seeds deteriorate, they lose vigour, become more sensitive to stresses during germination and ultimately become unable to germinate (Finch-Savage & Leubner-Metzger, 2006; Holdsworth et al., 2008). Seed longevity is influenced by a combination of genetic factors (Miura et al., 2002; Clerkx et al., 2004a), interactions with the environment throughout seed maturation and deteriorative events that occur before or during storage (Bewley & Black, 1994). Species that yield orthodox seeds (which remain viable when desiccated), such as Arabidopsis, have evolved strategies to resist deteriorative biotic and abiotic factors. The seed coat provides protection against mechanical damage and restricts water and oxygen uptake, while the embryo contains seed maturation-specific proteins that confer tolerance to desiccation. Arabidopsis is an excellent model plant for the genetic and molecular dissection of seed longevity, in particular because of the availability of mutants affected in this trait (Debeaujon et al., 2000; Clerkx et al., 2004a; Sattler et al., 2004).
Seed longevity in Arabidopsis is influenced by events that occur during seed development. This process is regulated by the integration of genetic programmes with hormonal and metabolic pathways (Gutierrez et al., 2007). Four master regulator genes with partially redundant functions, ABSCISIC ACID INSENSITIVE3 (ABI3), LEAFY COTYLEDON1 (LEC1), LEC2 and FUSCA3 (FUS3), play a central role during seed maturation. ABI3, LEC2 and FUS3 are transcription factors with a B3 DNA-binding domain that physically binds to the Sph/RYcis-elements in the promoter region of seed-specific genes (Suzuki et al., 1997; Kroj et al., 2003; Monke et al., 2004). ABI3 is one of the main transducers of the abscisic acid (ABA) hormone signal, which is necessary for the expression of maturation genes, such as seed storage protein (SSP) genes (Parcy et al., 1997; Kroj et al., 2003), late embryo abundant (LEA) genes (Giraudat et al., 1992), genes with antioxidant functions (Haslekås et al., 2003) and heat shock protein (HSP) genes (Kotak et al., 2007). LEC1 is part of an oligomeric transcriptional activator (Lotan et al., 1998) that co-ordinates embryo morphogenesis and maturation (Harada, 2001). Mutations in any of the ABI3, LEC1, LEC2 and FUS3 genes affect multiple processes of maturation and consequently seed longevity. Loss of function mutants of ABI3 and LEC1 produce seeds that lose their viability during the first few weeks after harvest (Meinke, 1992; Nambara et al., 1995). The abi3-5 mutation greatly reduces ABA sensitivity at germination and prevents proper seed maturation, the acquisition of desiccation tolerance and the degradation of chlorophyll (Ooms et al., 1993). The lec1-3 mutation causes embryo abnormalities, such as the accumulation of anthocyanins, an undersized hypocotyl and cotyledons with trichomes and stomata. Seeds of the lec1-3 mutant lack protein and oil reserves and tend to be viviparous (Meinke, 1992). The abi3-5 and lec1-3 mutants provide two sensitized genetic backgrounds that allow rapid identification of factors affecting seed longevity.
Arabidopsis grows in different habitats over a wide geographical range and is adapted to different environments, thereby showing natural variation for many traits (Koornneef et al., 2004). Differences in seed longevity between accessions can be exploited for genetic studies and quantitative trait locus (QTL) identification. Previous work revealed genetic variation for Arabidopsis seed traits such as dormancy and longevity (Bentsink et al., 2000; Alonso-Blanco et al., 2003; Clerkx et al., 2004b). We have crossed several Arabidopsis accessions with the abi3-5 and lec1-3 mutants (which are in a Landsberg erecta (Ler) background) and selected introgression lines with enhanced seed longevity compared with the Ler mutants. Genetic mapping and seed physiology assays, including proteome analysis, were performed to characterize the modifier lines.
Materials and Methods
The mutant lines abi3-5 and lec1-3 are both in the Ler genetic background and have been previously described (Ooms et al., 1993; Raz et al., 2001). The following Arabidopsis accessions were used: C24 (C24, CS906), Columbia (Col, CS907), Cape Verde Islands (CVI, N8580), Drahonin (Dra, CS1116), Dijon (Di-0, CS1106), Eilenburg (Eil-0, CS6693), Enkheim (En-2, CS1138), Eringsboda (Eri-1, CS22548), Gudow (Gd-1, N1184), Gluckingen (Gu-0, N1212), Koln (Kl-5, N1284), Landsberg erecta (Ler, CS20), Niederzenz (Nd-1, CS1636), Nossen (No-0, CS1394), Rschew (RLD-1, CS913), Slavice (Sav-0, N1514), Seis am Schlern (Sei-0, N1504), Shahdara (Sha, CS929), Warschau (Wa-1, CS6885) and Wassilewkija (Ws-0, CS1602). All accessions originated from the Arabidopsis Biological Resource Center in Ohio (ABRC) and the Nottingham Arabidopsis Stock Centre (NASC) or were collected and donated to the ABRC by M. Koornneef (CS22548).
Segregating populations were genotyped using simple sequence length polymorphisms (SSLP) and cleaved amplified polymorphic sequences (CAPS). The primers used to amplify these markers were designed for flanking short polymorphic sequences based on the Monsanto Landsberg-Columbia polymorphisms database (http://www.arabidopsis.org/browse/Cereon/index.jsp). After the sixth round of backcrossing and selfing, the density of the genetic maps was further increased using a custom-made Illumina Golden Gate SNP Assay® (Illumina Inc., San Diego, CA, USA) with 384 single nucleotide polymorphisms (SNPs) performed by Service XS (Leiden, the Netherlands).
The modifier lines were grown together with abi3-5, lec1-3 and wild-type Ler. After harvest, the seeds were stored in paper bags at room temperature in a closed, dark container, and their viability was monitored periodically. Seed germination tests were performed by sowing 70–150 seeds on water-soaked filter paper (Macherey-Nagel, Düren, Germany) in 6-cm-diameter Petri dishes (Greiner Bio-one, Kremsmünster, Austria). Seeds were then incubated in a germination chamber (van den Berg, Montfoort, the Netherlands) for 6 d (14 h light 25°C; 10 h darkness 20°C). For each genotype a minimum of five biological replicates were tested. A controlled deterioration test was performed to accelerate deterioration of wild-type seeds (Tesnier et al., 2002). Aliquots of c. 300 seeds were placed in open 200-μl PCR tubes (Biozym, Hessisch Oldendorf, Germany) and incubated in a sealed exicator at 37°C. The exicator contained a saturated solution of KCl to maintain the relative humidity at 83%. After incubation the seeds were dried at room temperature for 3 d and a germination test was performed as described above.
Protein extraction from dry seeds
Total protein extracts were prepared from dry mature seeds as described previously (Rajjou et al., 2008). Each seed sample was obtained from 10 plants per genotype. Fifty milligrams of seeds was ground in liquid nitrogen using a mortar and pestle. Total soluble proteins were extracted at 4°C in 640 μl of a lysis buffer containing 7 M urea, 2 M thiourea, 4% (w/v) CHAPS (GE Healthcare, Münich, Germany) and 1% (v/v) Pharmalyte carrier ampholytes, pH 3–10 (GE Healthcare). This extraction buffer also contained 18 mM Tris-HCl, 14 mM Trizma base (Sigma, Steinheim, Germany), the protease inhibitor cocktail Complete Mini (Roche Diagnostics, Mannheim, Germany), 53 units ml−1 DNase I (Roche Diagnostics), 4.9 units ml−1 RNase A (Sigma), and 0.2% (v/v) Triton X-100. After 10 min at 4°C, 14 mM dithiothreitol (DTT) was added and the protein extracts were stirred for 20 min at 4°C, and then centrifuged (35 000 g for 10 min) at 4°C. The supernatant was submitted to a second clarifying centrifugation as above. The final supernatant corresponded to the total soluble protein extract. Proteins were cleaned up following a trichloroacetic acid/acetone cold precipitation protocol (Plomion & Lalanne, 2007). Protein concentrations in the various extracts were measured according to Bradford (1976). Bovine serum albumin was used as a standard.
Proteins were first separated by electrophoresis according to charge. Isoelectric focusing was performed with 150 mg of protein for each sample. Proteins from the various extracts were separated using gel strips forming an immobilized nonlinear pH gradient from 3 to 10 (Immobiline DryStrip pH 3–10 NL, 18 cm; GE Healthcare) (Bjellqvist et al., 1982). Strips were rehydrated for 14 h at 20°C with the thiourea/urea lysis buffer containing 2% (v/v) Triton X-100, 20 mM DTT, and the protein extracts. Isoelectric focusing was performed at 20°C in the Multiphor II system (GE Healthcare) for 1 h at 300 V and for 7 h at 3500 V. Proteins were then separated according to size. The gel strips were equilibrated twice for 30 min each in 2 × 100 ml of equilibration solution containing 6 M urea, 30% (v/v) glycerol, 2.5% (w/v) sodium dodecyl sulphate (SDS), 0.15 M 1,3-bis(tris(hydroxymethyl)methylamino) propane (BisTris), and 0.1 M HCl. DTT (50 mM) was added to the first equilibration solution, and iodoacetamide 4% (w/v) was added to the second. Equilibrated gel strips were placed on top of vertical polyacrylamide gels (10% (v/v) acrylamide, 0.33% (w/v) piperazine diacrylamide, 0.18 M Trizma base, 0.166 M HCl, 0.07% (w/v) ammonium persulphate, and 0.035% (v/v) N,N,N′,N′-tetramethylethylenediamine (TEMED)). A denaturing solution (1% (w/v) low-melting agarose (Invitrogen, Karlsruhe, Germany), 0.4% (w/v) SDS, 0.15 M BisTris, and 0.1 M HCl) was loaded onto gel strips. After agarose solidification, electrophoresis was performed at 10°C in a buffer (pH 8.3) containing 25 mM Trizma base, 200 mM taurine, and 0.1% (w/v) SDS for 1 h at 35 V and for 14 h at 100 V. The gels for each protein sample (200 × 250 × 1.0 mm) were run in parallel (Isodalt system from Amersham Biosciences) and four technical replicates were produced. 2D gels were stained with silver nitrate for densitometric analyses (Rabilloud, 1999).
Protein identification and quantification
Digitized images (16-bit greyscale; 300 dpi) of the stained gels were aligned using Progenesis SameSpots (Nonlinear Dynamics, Newcastle, UK) and the intensity of each protein spot was quantified. The spots that corresponded to previously identified seed proteins localized on 2D electrophoresis Arabidopsis seed proteome reference maps (Gallardo et al., 2001, 2002; Rajjou et al., 2004, 2006, 2008; Job et al., 2005; http://www.seed-proteome.com) were identified by visual comparison. With the Progenesis SameSpots program (Nonlinear Dynamics) the seed proteomic profiles were compared with each other. The spot volume ratios between different genotypes were calculated using the average spot volume of the four technical replicates. The program performed a one-way analysis of variance (ANOVA) for each spot and returned a P-value that takes into account the mean difference, variance and sample size. The P-value is a measure of the probability of the spot data being obtained if no real difference exists.
Selection and genotyping of the abi3-5 and lec1-3 seed longevity modifiers
The existing natural variation in Arabidopsis was exploited to identify modifiers that improve the poor seed longevity of the abi3-5 and lec1-3 seed maturation mutants. Twenty Arabidopsis accessions (listed in the Materials and Methods) were crossed with both the abi3-5 and lec1-3 mutants. In the resulting F2 populations, homozygous abi3-5 and lec1-3 plants with the highest seed longevity were selected and backcrossed to Ler. In the F2 progenies of these backcrosses, one line derived from a cross between the Seis am Schlern (Sei) accession and abi3-5 and one line derived from a cross between the Shahdara (Sha) accession and lec1-3 showed consistently improved longevity compared with, respectively, the abi3-5 and lec1-3 mutants. Both lines were backcrossed five more times with Ler. From every F2 population, homozygous abi3-5 and lec1-3 plants with the highest seed longevity were used for the next backcross. The resulting modifier lines were genotyped with polymorphic SSLP and SNP markers to determine the positions of the Sei and Sha introgressions (Fig. 1). The resulting abi3-5 modifier line had a Sei introgression at the top of chromosome 5 at between 0 and 8 Mb and was named abi/Sei. The lec1-3 modifier line contained two Sha introgressions, one at the top of chromosome 4 at between 0 and 8 Mb and one at the bottom of chromosome 5 at between 19 and 26 Mb. An additional backcross was made with this lec1-3 modifier line to separate the Sha introgressions. Subsequent genotyping and phenotyping showed that the Sha introgression at the top of chromosome 4 actually contained two independent modifiers with different strengths (see next paragraph and Fig. 2b). Six different modifier lines of Sha in the lec1-3 mutant background were generated for further analyses and named lec/Sha1 (containing a Sha introgression at between 0 and 1.5 Mb on chromosome 4), lec/Sha2 (containing a Sha introgression at between 19 and 26 Mb on chromosome 5), lec/Sha3 (containing a Sha introgression at between 1.4 and 8 Mb on chromosome 4), lec/Sha1+2 (containing the Sha1 and Sha2 introgressions), lec/Sha1+3 (containing the Sha1 and Sha3 introgressions) and lec/Sha1+2+3 (the original line containing all three Sha introgressions).
The modifier lines show enhanced seed longevity
The seed longevity of the modifier lines and the abi3-5 and lec1-3 mutants was determined by periodic measurement of the germination percentage of their seeds, which were stored in dry conditions after harvest. The abi/Sei modifier line showed a higher germination percentage than abi3-5 directly after harvest, but did not show a reduced speed of deterioration (Fig. 2a). Therefore, the abi/Sei modifier increases the percentage of viable seeds directly after harvest, but not seed longevity.
Seeds from the modifier lines lec/Sha1, lec/Sha2 and lec/Sha3 exhibited higher germination percentages after storage than those from the lec1-3 parent (Fig. 2b). In addition, they showed a release of dormancy similar to Ler wild-type seeds, which was not observed in lec1-3 (Fig. 2b,c). The lec/Sha1 line had the highest longevity and strongest dormancy phenotypes. These two traits were only moderately increased in the lec/Sha2 and lec/Sha3 lines, but reached much higher values in the lec/Sha1+3 and lec/Sha1+2 lines (Fig. 2c). The lec/Sha1+2+3 line showed the greatest restoration of longevity and dormancy.
An artificial aging treatment (Tesnier et al., 2002) was used to determine whether the parental accessions, from which the modifiers were derived, displayed differences in seed longevity. Three-year-old seeds, which were fully after-ripened and 100% viable, were stored at high relative humidity (85%) and temperature (37°C) to create an environment in which all seeds quickly deteriorated. Seeds from the Sha accession showed the highest resistance to this artificial aging treatment, while Sei did not differ from Ler (Fig. 3).
The lec/Sha modifiers promote development of the embryonic hypocotyl
Seeds of the lec1-3 mutant accumulated anthocyanins, showed occasional vivipary, and carried embryos with an underdeveloped hypocotyl and rounded cotyledons (Fig. 4a,f), as was shown by Meinke et al. (1994) for the lec1-1 mutant. The embryo hypocotyls of the lec/Sha1 and lec/Sha2 modifier lines showed improved development and resembled wild-type hypocotyls. This improvement was not fully penetrant and was only observed in some of the seeds. In contrast, the cotyledons of the lec/Sha modifiers were similar to the lec1-3 cotyledons. An analysis of 12 lec/Sha1 plants showed that 74 ± 17% of the harvested seeds had a wild-type-like hypocotyl. Interestingly in the lec/Sha2 line this phenotype was only observed in a small fraction of the seeds (< 5%). The lec/Sha3 line did not show any improvement of the seeds, but Sha3 had a positive effect on embryo hypocotyl development in combination with the lec/Sha1 and lec/Sha2 modifiers (Fig. 4b,c,g,h). Nearly 100% of the seeds of the lec/Sha1+2+3 line had a wild-type-like hypocotyl (Fig4d,i). Therefore, all three modifiers contribute to hypocotyl development.
The abi3-5 and lec1-3 mutations influence the seed proteome
ABI3 and LEC1 have distinct and overlapping functions in the regulation of seed maturation. In order to study their general significance for the seed proteome and to elucidate their role in seed longevity, the proteomic profiles of freshly harvested seeds of the abi3-5 and lec1-3 mutants were compared with that of Ler wild-type by 2D gel electrophoresis.
The proteome analysis led to the identification of 165 protein spots with a significantly different abundance in the abi3-5 and lec1-3 mutants compared with wild-type Ler (P-value < 0.05). One hundred and thirteen of these spots corresponded to proteins, or protein fragments, that were previously identified and positioned on the Arabidopsis seed proteome reference maps (Gallardo et al., 2001, 2002; Rajjou et al., 2004, 2006, 2008; Job et al., 2005; http://www.seed-proteome.com; listed in Tables S1 and S2).
The abi3-5 seed proteome had a reduced amount of 12S seed storage proteins and other storage proteins, compared with Ler (Fig. 5). Moreover, it was characterized by increased amounts of chloroplast proteins, such as the two ribulose bisphosphate carboxylase subunits and ferritin 1, consistent with the green seed phenotype of the abi3-5 mutant (Table S1).
Compared with Ler, the lec1-3 seed proteome was characterized by a marked reduction in seed maturation proteins, such as 12S seed storage proteins and their precursors, LEAs and seed-specific small HSPs (Fig. 5, Table 1). The lec1-3 proteomic profile also showed a higher abundance of proteins that typically increase in germinating seeds, such as tubulin, aconitate hydratase, glycosyl hydrolase, the cell division cycle protein CDC48 and the enzyme phosphoenolpyruvate carboxykinase. Consistent with a germinative physiology, several chloroplast proteins, including ribulose bisphosphate carboxylase, were abundant in the lec1-3 mutant seeds. Proteins related to the stress response (mitochondrial HSP70, HSP90.5 and HSP90.7) and enzymes involved in oxygen radical detoxification of cells (a putative superoxide dismutase and thioredoxin-dependent peroxidase) were also increased in lec1-3 and were probably induced as a response to the seed deterioration processes that affect lec1 mutant seeds during desiccation (Meinke et al., 1994). In general, the lec1-3 mutation had a stronger influence on the abundance of seed proteins than the abi3-5 mutation.
Table 1. Proteins with different abundance in lec1-3 and the lec/Sha modifiers relative to Landsberg erecta (Ler)
Proteins were isolated from freshly harvested seeds, separated by 2D gel electrophoresis and quantified with the Progenesis SameSpots program.
lec1-3/Ler, normalized spot volume in the lec1-3 mature seeds divided by the normalized spot volume in the Ler mature seeds: lec/Sha1+3/Ler, lec/Sha1+2/Ler and lec/Sha1+2+3/Ler, normalized spot volume in the lec/Sha modifier mature seeds divided by the normalized spot volume in the Ler mature seeds. All values have a P-value < 0.05 (Table S2). The calculation of the ratios and P-values is described in the Materials and Methods.
Heat shock protein 17.4
Late embryogenesis abundant (LEA)
12S seed storage protein (precursor)
12S seed storage protein (precursor)
Alpha-globulin 12S fragment
Alpha-globulin 12S fragment
Beta-cruciferin 12S (precursor)
Late embryogenesis abundant (LEA)
Cupin family protein
Cupin family protein
Heat shock protein 17.6A
Mildew resistance locus O 6
Cell division cycle protein (CDC48)
At3g09840 or At5g03340
Jacalin-related lectin 30
Putative myrosinase-associated protein
Translation elongation factor Tu
Pyruvate orthophosphate dikinase
Ribulose bisphosphate carboxylase large subunit
Superoxide dismutase (Mn), putative
Shepherd protein (SHD). HSP90.7
The lec/Sha modifiers partially restore the seed proteomic profile of lec1-3
The seed proteomic profiles of the modifiers were compared with those of the abi3-5 and lec1-3 mutants. The profile of the abi/Sei modifier line only showed a small change compared with that of abi3-5 (Fig. 5), which was expected because of its weak influence on seed longevity and the relatively small difference in proteomic profiles between abi3-5 and wild type. By contrast, the lec/Sha1+2, lec/Sha1+3 and lec/Sha1+2+3 modifier lines showed considerable differences, compared with the lec1-3 profile (Fig. 5). Nearly all detected proteins showed restoration towards wild-type levels. For instance, most of the SSPs, such as 12S globulins (Fig. 6b, spots 70, 71, 82 and 83), and LEAs were consistently restored (Table 1). The storage proteins were more abundant in the lec/Sha1+2 modifier line than in the lec/Sha1+3 line, and most abundant in the triple modifier line, which showed close to wild-type levels (Fig. 6b, Table 1). Several proteins highly expressed in the lec1-3 proteome, such as the germination-related protein jacalin-related lectin 30 and the putative myrosinase-associated protein (Fig. 6b, spots 58 and 1126), showed reduced expression in the lec/Sha modifiers (Table 1).
Seed longevity is an important ecological and commercial trait, but its molecular mechanisms are poorly understood. Arabidopsis seeds need to be stored for years to observe a reduction in viability. To accelerate the evaluation of seed longevity, seeds can be stored under conditions that enhance aging (controlled deterioration), or mutants with reduced seed longevity can be used. In this work, we have exploited the latter strategy to identify natural modifiers that enhance the poor survival capacity of lec1-3 and abi3-5 mutants. A proteomic analysis of these modifiers and the lec1-3 and abi3-5 mutants revealed specific changes in storage- and germination-related proteins.
The proteomic profiles of the lec1-3 and abi3-5 mutants confirm their position in the gene network regulating seed maturation
The LEC1 transcription factor acts upstream of the main genetic network of seed maturation (To et al., 2006). Consistent with this, the proteome of freshly harvested lec1-3 seeds deviates much more from the wild type than the abi3-5 seed proteome does. LEC1 controls the expression of the SSP genes in a hierarchical manner, which involves ABI3 and FUS3 proteins (Kagaya et al., 2005). ABI3 and FUS3 are B3 domain transcription factors that act in a partially redundant way to activate each other (To et al., 2006). Consequently, the lack of ABI3 protein in the abi3-5 mutant might be compensated by FUS3.
Seeds from the lec1-3 mutant are characterized by a reduced abundance of seed storage proteins and their precursors, compared with Ler. Furthermore, they lack several LEAs and two small HSPs (Table 1 and Table S1). Small HSPs and LEAs protect membranes, proteins and nucleic acids from the deleterious removal of water (Wehmeyer & Vierling, 2000; Delseny et al., 2001; Boudet et al., 2006) and are required for seed longevity. The lack of protective proteins in lec1-3 probably causes stress symptoms in seeds, which is consistent with the accumulation of anthocyanins (Fig. 4), and with the increased abundance of three proteins of the HSP70 and HSP90 families and radical scavenging enzymes (Table 1). Earlier studies demonstrated that different kinds of stress cause increased HSP expression (Vierling, 1991).
The proteome analysis revealed that proteins related to seed germination, such as the enzymes involved in the tricarboxylic acid cycle (aconitate hydratase), gluconeogenesis (phosphoenolpyruvate carboxykinase), cell division (CDC48) and photosynthesis (rubisco precursors), were increased in lec1-3 seeds, consistent with their nondormant phenotype.
Modifiers that improve the seed longevity of lec1-3 and abi3-5 mutants could only be detected in the Sha accession and partially overlap with previously detected QTLs
To obtain natural modifiers of seed longevity, the lec1-3 and abi3-5 mutants were crossed with 20 accessions. One line with improved longevity of abi3-5 was recovered from the cross with the Sei accession. However, this line only showed improved germination directly after harvest and did not show an improvement in the deterioration rate of the seeds. Modifiers that truly improved seed longevity were solely obtained from the Sha accession. The Sha modifiers were identified in lec1-3, but not in the abi3-5 background, which suggests that they need the ABI3 protein to exert their function. The lec/Sha modifiers were originally selected together in a single line, as a result of their strong combined phenotype. After separation, the individual modifiers showed weaker phenotypes. Modifier alleles from other accessions with similar strength might have been lost during the selection process. In fact, additional abi3-5 and lec1-3 modifiers from four other accessions were obtained in the first selection round, but could not be reliably confirmed in subsequent backcross populations.
The lec/Sha modifiers can either be components of the genetic pathways for seed development controlled by the LEC1 transcription factor, or contribute to seed longevity independently of LEC1. It is possible that the modifiers have a general effect on longevity because Sha seeds showed more resistance to artificial aging than Ler seeds (Fig. 3). Previous work on a Ler/Sha recombinant inbred line population identified a QTL for resistance to artificial seed aging at the top of chromosome 4, for which the Sha allele increased longevity (Clerkx et al., 2004b). The lec/Sha1 and lec/Sha3 modifiers that we detected are located in the same chromosomal region and one of them might coincide with this QTL. Nevertheless, two other QTLs that were formerly found for controlled deterioration and a QTL for natural aging do not co-localize with any of the lec/Sha modifiers. Furthermore, the lec/Sha2 modifier maps in a region where no QTL for seed longevity has previously been detected. This indicates that our approach did not identify the same set of loci as QTL analysis.
Analysis of the natural modifiers
The abi/Sei modifier line showed a higher percentage of germination than abi3-5 directly after harvest, but they did not differ in the rate at which seed viability decreased during storage. Therefore, the beneficial effects on longevity of the Sei introgression are restricted to the pre-harvest period, during which the modifier might reduce some of the desiccation damage that is characteristic of maturing abi3-5 seeds (Ooms et al., 1993). The abi/Sei modifier was similar to abi3-5 in every physiological aspect we investigated, including the seed proteome. Consequently, our analysis does not explain the abi/Sei seed germination phenotype. Previous work showed that the severity of desiccation damage in abi3 mutants correlates with the accumulation of sugars (Ooms et al., 1993). An analysis of the seed sugar composition and content in the abi/Sei modifier line might reveal differences with the abi3-5 mutant.
Seeds of the lec/Sha modifier line survived storage substantially better than those of lec1-3, and showed an improvement of additional traits that are disturbed in lec1-3, such as dormancy, hypocotyl development and the abundance of seed storage proteins. The lines with two or three Sha introgressions (lec/Sha1+3, lec/Sha1+2 and lec/Sha1+2+3) produced better developed seeds with improved longevity compared with the single lec/Sha modifier, indicating that lec/Sha1, lec/Sha2 and lec/Sha3 contain independent modifiers with additive effects. Phenotypic analyses of lines heterozygous for lec/Sha1 and lec/Sha2 showed that these modifiers behave in a co-dominant fashion.
The lec/Sha modifiers function redundant to LEC1 or downstream to LEC1 in seed maturation and longevity
The lec/Sha modifiers can partially restore the seed maturation programmes of the LEC1 genetic pathway in the absence of functional LEC1 protein and enhance seed longevity. The effects of the lec/Sha modifiers on the embryo were limited to hypocotyl development, while lec1 features were maintained in the cotyledons. This indicates that only a subset of the LEC1 functions are restored by the lec/Sha modifiers and suggests that they act downstream of LEC1. Alternatively, the lec/Sha modifiers could be redundant with LEC1 if they are only expressed in the hypocotyl. The genetic network that controls seed development is regulated locally (Santos-Mendoza et al., 2008) and it was previously shown that the ABI3 and FUS3 promoters have residual activity in the lec1 embryo hypocotyl (To et al., 2006). The lec/Sha modifiers might strengthen the activation of seed maturation programmes in the hypocotyl co-operatively with FUS3 and ABI3. The lec/Sha modifiers partially restore the accumulation of seed maturation proteins such as 12S globulins, small HSPs and LEAs in a redundant manner. The germination- and stress-related proteins that accumulate in lec1-3 seeds are reduced by the modifiers, but their levels still remain significantly higher than in wild type. This could be attributable to the spatial confinement of the function of the modifiers and/or to their incomplete penetrance.
LEC1 is homologous to the HAP3 subunit of a trimeric CCAAT-binding factor that is widely distributed from fungi to mammals (Lotan et al., 1998). LEC1 may function in a similar way in Arabidopsis, as the subunit of a complex that regulates a specific set of genes related to embryo development, including ABI3 and FUS3. Although the downstream genes of LEC1 are known, neither direct targets nor the cis-elements of the putative LEC1 complex have been identified. The lec/Sha modifiers might be allelic variants of regulatory elements that act downstream of the LEC1 complex. The lec/Sha1 modifier has the strongest phenotype and is a first choice for fine mapping and cloning. The molecular identification of the lec/Sha modifiers may make a significant contribution to elucidation of both the poorly understood seed longevity trait and the downstream function of LEC1.
We thank Hetty Blankestijn-de Vries and Gerda Ruys for technical assistance during the selection of the modifier lines. This work was supported by a Marie Curie Host Fellowships for Early Stage Researchers Training grant to MS, by grant WBI.4737 from the Technology Foundation STW to EJMC and by the Max Planck Society.