Dormant seeds are capable of remaining alive in the hydrated state for extended periods of time without losing vigor, until environmental cues or after-ripening result in the release of dormancy. Here, we investigated the possible role of the regulatory subunit of the sucrose non-fermenting-related kinase complex, MtSNF4b, in dormancy of Medicago truncatula seeds. Expression of MtSNF4b and its involvement in a high-molecular-weight complex are found in dormant seeds, whereas imbibition of fully after-ripened, non-dormant seeds leads to dissociation of the complex. MtSNF4b is capable of complementing the yeast Δsnf4 mutant and of interacting with the MtSnRK1 α-subunit in a double hybrid system. Transcriptome analyses on freshly harvested and after-ripened RNAi Mtsnf4b and wild-type embryos implicate MtSNF4b in the defense response in hydrated dormant embryonic tissues, affecting the expression of genes encoding enzymes of flavonoid and phenylpropanoid metabolism, WRKY transcription factors and pathogenesis-related proteins. Silencing MtSNF4b also increased the speed of after-ripening during dry storage, an effect that appears to be related to a change in base water potential. No significant difference in ABA content or sensitivity was detected between mutant and wild-type seeds. Pharmacological studies using hexoses and sugar analogs revealed that mannose restored germination behavior and expression of the genes PAL, CHR and IFR in RNAi Mtsnf4b seeds towards that of the wild-type, suggesting that MtSNF4b might act upstream of sugar-sensing pathways. Overall, the results suggest that MtSNF4b participates in regulation of a constitutively activated defense response in hydrated, dormant seeds.
The end of seed development is characterized by termination of seed filling, a decrease in seed water content and progressive arrest of metabolism. During this final phase, seeds such as those of Medicago truncatula acquire the ability to survive the dry state for long periods of time, also referred to as longevity, and enter into a dormant state. Dormancy is a genetically and environmentally determined developmental state that is imposed during imbibition of freshly harvested seeds, during which metabolism is active, but growth processes are repressed (Finch-Savage and Leubner-Metzger, 2006; Bradford and Nonogaki, 2007). Dormancy is gradually lost during the time-dependent process of after-ripening (AR) that occurs in the dry seed. In the dormant state, seeds are capable of remaining hydrated for extended periods of time without losing vigor (Finch-Savage and Leubner-Metzger, 2006; Bradford and Nonogaki, 2007). In addition, fluctuations of environmental conditions will cause the seeds to undergo hydration and dehydration cycles in the soil, which they can do without losing their ability to germinate. Dormant seeds have developed an array of protective mechanisms in order to remain viable until environmental cues are provided to release dormancy, leading to germination and seedling establishment.
The sucrose non-fermenting related kinase (SnRK1) is known for its central role in energy metabolism, linking stress, sugar and developmental signals to regulate plant metabolism, energy balance, growth and stress responses (Schwachtje et al., 2006; Ananieva et al., 2008; Baena-Gonzalez and Sheen, 2008). The complex has also been shown to play a role during seed maturation, affecting accumulation of storage proteins and occasionally leading to precocious germination (Radchuk et al., 2006; Rosnoblet et al., 2007). The SnRK1 complex comprises an auto-inhibitory catalytic α subunit that is activated upon interaction with two other non-catalytic subunits, a β subunit (SnRKβ), which has been suggested to play an essential role in subcellular localization and the specificity of recognition between the kinase complex and its targets, and a γ subunit (SNF4/SnRKγ), which is thought to regulate the activity of the complex (Halford et al., 2000; Kleinow et al., 2000). Several regulatory β and γ subunits exist in plants, and are differentially regulated at the transcriptional level according to stress and developmental stage (Bouly et al., 1999; Buitink et al., 2004). In seeds, several SnRKβ and SnRKγ subunits are co-expressed, suggesting co-existence of several SnRK1 complexes formed by different regulatory subunits (Buitink et al., 2004). This might explain the strong pleiotropic effects observed when the catalytic subunit is down-regulated (Radchuk et al., 2006; Baena-Gonzalez and Sheen, 2008).
The expression of one subgroup of SnRKγ subunits, Pv42/LeSNF4/MtSNF4b, is correlated with the final stage of maturation, when longevity and dormancy are acquired (Bradford et al., 2003; Buitink et al., 2004). MtSNF4b is expressed in parallel with the induction of desiccation tolerance, and its silencing affects oligosaccharide metabolism and longevity during maturation (Bradford et al., 2003; Rosnoblet et al., 2007). LeSNF4 transcripts are ABA-inducible in tomato seeds, and their accumulation is linked to the inability of seeds to germinate, with levels remaining high when radicle protrusion is blocked by ABA, water stress, far-red light or dormancy (Bradford et al., 2003).
In this study, we investigated the possible role of MtSNF4b in seed germination and dormancy of M. truncatula seeds. We show that MtSNF4b expression and complex formation are related to the AR status of the imbibed seeds. A transcriptomic analysis on imbibed mutant and wild-type embryos demonstrated that MtSNF4b silencing reduces gene expression related to biotic stress and secondary metabolism in dormant embryonic tissues. Silencing MtSNF4b also increased the speed of AR independently of ABA. A role of sugar sensing pathways downstream of SnRK1/SNF4b is suggested by the observation that mannose both delays germination and increases PAL, CHR and IFR expression in RNAi Mtsnf4b seeds. The results suggest that SNF4b plays a role in constitutive defense responses during hydration of dormant M. truncatula seeds.
MtSNF4b expression and complex formation are related to dormancy
To investigate the role of MtSNF4b in dormancy of M. truncatula seeds, we first characterized its dormancy behavior. Seeds of M. truncatula do not germinate readily under optimal conditions when imbibed 4 days after harvest, and exhibit non-deep dormancy comparable to that of Nicotiana plumbaginifolia (Grappin et al., 2000; Faria et al., 2005) (Figure 1a). Subsequent storage of freshly harvested seeds under controlled, dry conditions led to the gradual release of dormancy, referred to as AR, and a concomitant increase in germination speed (Figure 1a). Seeds that are imbibed after two weeks of AR take 10–14 days to germinate at 20°C in the dark, whereas breaking of dormancy either by one year of AR or an incipient cold imbibition (48 h at 4°C) results in a seed population that germinates within 24 h under similar imbibition conditions. Incubation in fluridone to block ABA and carotenoid synthesis released dormancy and increased the germination speed such that it was comparable with that of fully after-ripened seeds (Figure 1a). To investigate whether MtSNF4b gene expression is regulated in relation to dormancy, as found for the LeSNF4 gene of tomato (Bradford et al., 2003), its expression was analyzed by quantitative RT-PCR. MtSNF4b transcript levels in embryos remain high in freshly harvested, dormant seeds (Figure 1b). In contrast, transcript levels decreased in imbibed non-dormant seeds prior to germination.
Previous studies using yeast and plant native protein extracts demonstrated, using chromatography techniques, that the catalytic subunits, together with their regulatory counterparts, were present in high-molecular-weight fractions, and that this assembly corresponded to the activity of the SNF1 complex (Estruch et al., 1992; Sugden et al., 1999). We previously demonstrated that MtSNF4b is incorporated into a high-molecular-weight complex during the final stages of maturation (Rosnoblet et al., 2007). To investigate the link between MtSNF4b complex formation and dormancy, soluble proteins were extracted from hydrated dormant seeds after imbibition for 48 h at 20°C, or from hydrated seeds in which dormancy was released by imbibition at 4°C. No seeds had germinated after 48 h under either imbibition condition, but cold-treated seeds germinated within 12 h compared to 12–14 days for seeds that were imbibed directly at 20°C. Freshly extracted native proteins were separated by gel filtration, after which the presence of the SNF4b protein in the various fractions was analyzed by Western blot (Figure 1c). SNF4b was not only visible in those fractions corresponding to its own molecular weight (approximately 42 kDa), but a high-molecular-weight complex of 400–600 kDa containing SNF4b (fractions 1–4) was also detected in dormant seeds. In contrast, imbibition of seeds in which dormancy was released resulted in partial dissociation of the complex, as evidenced by the almost complete absence of SNF4b in these same fractions. Further imbibition of non-dormant seeds was previously found to lead to complete disappearance of SNF4b in fractions >200 kDa (Rosnoblet et al., 2007). This suggests that the SNF4b-associated complex remains intact during hydration of dormant seeds.
MtSNF4b is a functional SNF4 subunit and interacts with the α subunit of Medicago truncatula
To verify that the complex in which MtSNF4b participates is a SnRK1 complex, we first tested whether MtSNF4b indeed encodes a functional homolog of the corresponding yeast gene by complementation of a mutant deficient in SNF4 (snf4Δ2). In yeast, the SNF4 protein is required for activation of SNF1 kinase, which is essential for de-repression of sugar-metabolizing enzymes in the absence of glucose. As a consequence, snf4Δ2 mutants are able to grow on glucose but not on galactose or other carbohydrate sources (Figure 2a,b). Expression of MtSNF4b in a yeast strain with a deletion in the SNF4 gene restored growth on galactose to a level essentially equivalent to that in the wild-type (Figure 2a).
In addition, a possible interaction with the catalytic α subunit (SnRK1) was tested using a bacterial double hybrid system. Screening of the EST databank Medicago truncatula Gene Index (MTGI) as well as the Medicago truncatula genome revealed the presence of one SnRK1 gene with 96 and 92.8% similarity to AKIN10 (Q38997) and AKIN11 (P92958) of Arabidopsis thaliana, respectively, and 96.6% similarity with the SnRK1 of tomato (Q9M726, LeSNF1). The sequence of this SnRK1, hereafter named MtSnRK1, was cloned into the pTRG vector, and the MtSNF4b sequence into the pBT vector. No auto-activation of either subunit could be detected, but MtSNF4b and MtSnRK1 were capable of interacting with each other (Figure 2c). These data suggest that, in M. truncatula seeds, MtSNF4b interacts with the SnRK1 subunit and is likely to play a role in activation of the SnRK1 catalytic function. These data confirm the results obtained for another member of the PV42/SNF4b family, LeSNF4, which is capable of complementing yeast snf1 and snf4 mutants and physically interacting with LeSNF1 and LeSIP1 in a glucose-dependent manner in yeast two-hybrid assays (Bradford et al., 2003).
Transcriptome profiling indicates a role for MtSNF4b in gene regulation as a function of after-ripening time
To detect genes that are potentially regulated by MtSNF4b, we made use of previously produced lines in which MtSNF4b was silenced by RNAi (Rosnoblet et al., 2007), together with a set of controls to verify the effects of transformation and RNAi-mediated effects. Silencing MtSNF4b reduced protein levels of SNF4b to almost undetectable levels compared to wild-type seeds (Figure S1) (Rosnoblet et al., 2007). A transcriptome analysis was performed on 6 h-imbibed embryos of wild-type and RNAi Mtsnf4b lines at 2–3 weeks of AR, referred to as the fresh (F) dataset. As the SNF4b remained complex only in freshly harvested imbibed seeds, we also compared wild-type versus RNAi Mtsnf4b seeds at 6 months of AR, in which the complex dissociated upon imbibition (referred to as the stored, S, dataset). A third analysis was added to identify the genes that are differentially expressed in dormant (freshly harvested) versus non-dormant seeds (after AR), as this dataset is not yet available for M. truncatula seeds. For this purpose, a comparison was made between wild-type seeds after 2–3 weeks of AR versus 6 months of AR, using RNA from the F and S datasets (referred to as the dormancy, D, dataset). Sets of genes that showed enhanced or decreased expression in each of the three datasets were identified, and their overlap is presented in a Venn diagram (Figure 2 and Appendix S1).
A total of 217 genes are present in the F dataset, compared to only 37 genes in the S dataset, confirming a dormancy-related link as found for MtSNF4b expression and complex formation (Figure 3). Comparison between the F and S datasets showed that, irrespective of the duration of AR, only seven genes were differentially up-regulated in both datasets, such as 12-oxophytodionoate (OPDA) reductase, a Myb transcription factor and two genes encoding plant lipid transfer proteins (MtN5). Interestingly, when a threshold P value of 0.05 was used instead of 0.01, an additional two genes, encoding 12-OPDA reductase aand glutathione-S-transferase, were detected.
Twenty-one genes overlapped between the D dataset and the F dataset. Twenty were up-regulated in wild-type seeds at 2–3 weeks of AR, encoding several pathogenesis-related proteins (ABR17/18) that are developmentally regulated during seed development (Iturriaga et al., 1994), as well as genes encoding chalcone synthase and chalcone reductase, peroxidases or caffeoyl CoA 3-O-methyltransferase (Appendix S1). Only one gene overlapped between the down-regulated F and D datasets, encoding a small heat-shock protein. An additional 175 genes were found in the D dataset that were not detected in the F dataset, including legumins and caleosins amongst the up-regulated genes and 126 down-regulated genes, such as ribosomal RNAs or several phosphatase 2C proteins that function as negative regulators in ABA signaling (Figure 3 and Appendix S1) (Gosti et al., 1999). Silencing MtSNF4b changes the expression of a subset of genes that are regulated prior to AR.
MtSNF4b silencing reduces gene expression related to secondary metabolism and biotic stress
From the comparison between the datasets, it is evident that the largest set of genes that are regulated differentially between mutant and wild-type are up-regulated genes of the F dataset (Figure 3). Silencing MtSNF4b results in a decrease in the expression of these genes when seeds are not yet fully after-ripened. The main classes characterizing the 162 up-regulated genes in the F dataset are related to secondary metabolism, biotic stress, the cell wall, transcription factors and transport (Table 1). A large number of genes are implicated in secondary metabolism, encoding enzymes related to flavonoid and phenylpropanoid biosynthesis. A closer look among them revealed that 27 genes encode various isoforms of 11 of the 13 enzymes involved in medicarpin biosynthesis (Appendix S2 and Figure S3). The class of transcription factors contains five WRKY transcription factors and two ethylene-responsive transcription factors. In addition, four genes related to protein fate encode RING-H2 finger proteins that belong to the ATL sub-family, which has been suggested to be involved in the early steps of the plant defense signaling pathway (Serrano and Guzman, 2004). These results were confirmed for several genes by quantitative RT-PCR (Figures 4 and S2). In wild-type seeds, transcript levels of two genes found in the F dataset, chalcone reductase (CHR) and a pathogenesis-related protein (PR10), increased sharply during imbibition after only 2 weeks of AR, whereas transcript levels remained low when seeds were imbibed after complete AR (Figure 4a,d). In addition, for both CHR (Figure 4a–c) and PR10 (Figure 4d–f), the quantitative RT-PCR analysis confirmed reduced expression in seeds of three independent RNAi lines compared to wild-type seeds and seeds transformed with a vector containing RNAi construction targeted against the proximal region of the MtENOD40-1 promoter, which serves as control for RNAi-induced silencing. Comparable observations were made on a number of other genes involved in secondary metabolism (IFR, IFS and PAL) at two harvests before and after AR (Figure S2). Using a PR10 family-specific antibody (Ramos and Bisseling, 2003), it was verified that differences in PR10 gene expression resulted in differences at the protein level (Figure 3g). For three independent biological replicates, Western blots confirmed that the PR10 protein level was higher in control seeds than in mutant seeds.
Table 1. Classification of up-regulated or down-regulated genes in wild-type/RNAi Mtsnf4b embryos at 2–3 weeks of AR (F dataset)
Number of genes
Fold change (WT/RNAi) >1.5 (P < 0.01)
Fold change (WT/RNAi) >−1.5 (P < 0.01)
Amino acid metabolism
Lipid, fatty acid and isoprenoid
Disease, virulence and defense
RNA regulation of transcription
Classification of the 55 down-regulated genes in the F dataset revealed that 26 genes are related to abiotic stress, encoding eight heat-shock proteins, five late embryogenesis abundant (LEA) proteins and four detoxification enzymes (Table 1 and Appendix S2). It is possible that the absence of SNF4b provokes a stress response, which is also evident from the up-regulation of a heat shock transcription factor in the mutant seeds. Overall, these data suggest a role for MtSNF4b in activation of constitutive defense responses to biotic stress when seeds are in a dormant state.
Silencing MtSNF4b affects the speed of AR of freshly harvested seeds
Given the potential link of MtSNF4b with dormancy, we tested the effect of MtSNF4b silencing on germination speed in relation to AR. Seeds of various lines grown under similar conditions were imbibed in the dark at 20°C for various times (Figure 5a–d). For wild-type seeds, freshly harvested seeds germinated only partially (>20%) over 3 weeks of imbibition, whereas seed lots that were after-ripened for 2–3 weeks germinated within 8–16 days. Compared to wild-type seeds, RNAi Mtsnf4b seeds germinated approximately twice as fast after 2–3 weeks of AR, but were equally dormant when imbibed straight after harvest. This difference in germination speed after 2–3 weeks of AR was observed for several independent harvests in a reproducible manner (data not shown). AR durations of 3 months and 1 year led to a comparable germination speed between mutant and control lines (Figure 5c,d).
Progress towards completion of germination is a function of the mean base water potential [ψb(g)] during imbibition (Finch-Savage and Leubner-Metzger, 2006; Bradford and Nonogaki, 2007). To assess how sub-optimal water conditions influence germination time during the course of AR, wild-type and mutant seeds after various durations of AR were imbibed at −0.27 MPa using a polyethylene glycol solution (Figure 5e–g). Imbibition of seeds after 2–3 weeks of AR under these conditions exacerbated the differences in germination behavior between wild-type and mutant seed batches (Figure 5e). This difference was still observed but to a lesser extent after 3 months of AR, and had completely disappeared after 1 year of AR (Figure 5f,g). These data suggest that silencing MtSNF4b might be related to a change in ψb(g). While weakening of enclosing tissues is a prerequisite for germination, embryo growth potential is a second factor that determines whether a seed germinates or not (Finch-Savage and Leubner-Metzger, 2006; Bradford and Nonogaki, 2007). No difference in the penetration of tetrazolium was detected between mutant and wild-type seeds (data not shown), discarding the hypothesis that cell-wall permeability is at the origin of the reduced dormancy phenotype, as found for Arabidopsis testa mutants (Debeaujon et al., 2000). Collectively, these data indicate that the phenotype observed as a result of silencing MtSNF4b is under the influence of a time-dependent component controlled by dry storage and thus AR.
Action of SNF4b occurs downstream of ABA
Given the difference in dormancy and speed of AR induced by silencing MtSNF4b, we investigated the possible link between SNF4b and ABA signaling pathways. ABA concentrations above 10 μm induced rapid accumulation of the SNF4b protein in immature seeds (Figure 6a). ABA contents (Figure 6c,d) were assessed in seeds obtained from harvests at 1 week and 8 months of AR, respectively, with the corresponding germination behavior shown in Figure 6(b). No significant difference in ABA content after 1 week of AR was detected between wild-type or RNAi seeds, regardless of whether they were dry or imbibed for 24 h (Figure 6c), although a twofold difference in germination speed was observed. In addition, in the batch of seeds obtained after 8 months of AR, which almost completely relieved dormancy, no difference between wild-type and mutant ABA content was observed either. Imbibition of these seeds resulted in a slight but significant decrease in ABA for the wild-type and one of the RNAi lines (Figure 6d), but no large degradation of ABA was detected as found for non-dormant seeds of N. plumbaginifolia (Grappin et al., 2000). ABA sensitivity was studied by evaluating the final percentage germination of seed lots after 8 months of AR as a function of ABA concentration. A comparable sensitivity to ABA is observed for two independent seed lots of RNAi Mtsnf4b lines (Figure 4e). Fluridone was found to efficiently break dormancy of both wild-type and mutant M. truncatula seeds, decreasing the time for 50% germination (T50) from 152 to 20 h for wild-type seeds, and from 80–102 to 20–23 h for two RNAi Mtsnf4b lines (Table S1). Incubation in paclobutrazol (100 μm), a gibberellic acid (GA) inhibitor, doubled the germination speed in both wild-type and mutant lines. In contrast, GA3 up to concentrations of 100 μm had little effect on germination speed and did not change the germination behavior between mutant or wild-type seeds (Figure S4). Thus, ABA is capable of inducing gene and protein expression of MtSNF4b, but signaling pathways downstream of MtSNF4b are not implicated in changing the content or sensitivity of ABA during imbibition of M. truncatula seeds.
Role of MtSNF4b in sugar signaling
We next investigated a possible role of MtSNF4b in sugar sensing pathways. Mannose and 2-deoxyglucose, glucose analogs that are phosphorylated by hexokinase but not further metabolized, cause repression of photosynthetic gene expression at low physiological levels, and inhibit germination in Arabidopsis (Pego et al., 1999). Glucose produced via a glycolysis-dependent pathway is capable of inducing the expression of PR1 and PR5, which belong to the family of pathogenesis-related proteins (Xiao et al., 2000; Jossier et al., 2009). 3-O-methylglucose (3-OMG), another glucose analog, is taken up by the plant with the same efficiency as glucose, but, unlike glucose, mannose or 2-deoxyglucose, cannot be phosphorylated by hexokinase. This analog can be used to activate a hexokinase-independent signaling pathway via sugar uptake that activates expression of genes such as PAL (Rolland et al., 2006). As several PAL genes were differentially expressed between mutant and wild-type embryos (Appendix S1), we first determined the effect of 3-OMG on germination and gene expression in wild-type seeds. Surprisingly, 3-OMG was capable of further delaying the germination of dormant, 2-week AR seeds, but had no further effect on germination of non-dormant seeds (Figure S5). Seeds with an intermediate dormancy level showed an intermediate delay of germination by 3-OMG. Similarly, 3-OMG repressed the expression of PAL, CHR and IFR in dormant seeds, but increased their expression in non-dormant seeds (Figures 7a and S6a,g). The delay of germination in 2-week AR seeds of RNAi Mtsnf4b lines by 3-OMG was comparable to that of wild-type seeds (Figure S7). 3-OMG has no significant effect on gene expression (Figures 7b and S6b,h), comparable to that of 3-OMG on gene expression in wild-type seeds at intermediate dormancy level.
In dormant and non-dormant wild-type seeds, mannose did not affect germination or gene expression (Figures 7e, S5 and S6e,k). However, when MtSNF4b was silenced, mannose delayed germination and increased gene expression in 2-week AR seeds. Likewise, glucose increased gene expression in RNAi Mtsnf4b seeds but had no effect in wild-type seeds (Figures 7c and S6c,i). The effect of glucose on germination was dependent on the dormancy level, resulting in an increase in the speed and homogeneity of germination in both dormant wild-type and RNAi seeds at 2 weeks of AR (Figures S5 and S7). Glucose had no effect on germination in non-dormant wild-type seeds, as is the case for Arabidopsis (Pego et al., 1999). The fact that these observations did not result from an osmotic effect brought about by the sugars was verified by using a polyethylene glycol solution with a comparable osmotic water potential (Figure S8).
This analysis shows that MtSNF4b, a functional subunit of the SnRK1 complex in M. truncatula, is involved in seed dormancy-related events. Silencing MtSNF4b affects the expression of biotic stress-related genes that are expressed in seeds during imbibition only when they are dormant. In addition, silencing of MtSNF4b also resulted in faster AR of freshly harvested seeds. We previously demonstrated a role for MtSNF4b during maturation (Rosnoblet et al., 2007), and the present study indicates an additional role for MtSNF4b during the germination phase of seeds. At the final stages of maturation, concomitant with the formation of a SnRK1/SNF4b complex, oligosaccharide accumulation was affected in RNAi Mtsnf4b seeds, in addition to the longevity of mature seeds (Rosnoblet et al., 2007). The effect of MtSNF4b on AR may have an indirect origin via modification of cell walls through activation of biotic defenses. In MtSNF4b-deficient embryos, transcriptomic analysis revealed an up-regulation of genes involved in cell-wall metabolism and rigidity, such as enzymes involved in the breakdown of galactomannans (α-galactosidase and β-mannosidase), as well as down-regulation of a polygalacturonase inhibitor (Table 1 and Appendix S1). However, several studies have suggested a role for SNF4b in dormancy or AR (Bradford et al., 2003; Argyris et al., 2008; Carrera et al., 2008). A homolog of MtSNF4b in tomato, LeSNF4, was previously identified as being highly expressed in non-germinating tissues, and was suggested to indicate a high effective ABA/GA ratio in seeds (Argyris et al., 2008). Indeed, genes associated with the synthesis and deactivation of GA and ABA are reciprocally regulated in association with the imposition and release of dormancy (Cadman et al., 2006). Although we confirmed that the regulatory subunit SNF4b is ABA-inducible, our study failed to demonstrate a direct downstream link with ABA, either by pharmacological approaches or transcriptomic analyses (Figure 6 and Appendix S1). Recently, a study was performed with the aim of separating the action of ABA in seed dormancy from AR-regulated gene expression using Arabidopsis thaliana mutants defective in ABA synthesis or perception (Carrera et al., 2008). Interestingly, the homolog of MtSNF4b in Arabidopsis, At1g15330, was identified as one of the 103 genes that were down-regulated by AR independently of ABA signaling (Carrera et al., 2008). These data support the suggestion that SNF4b plays a role in AR rather than dormancy, and confirm the ABA-independent pathway downstream of MtSNF4b.
Given the suggested link between SnRK1 and sugar signaling pathways (Tiessen et al., 2003; Jossier et al., 2009), it is possible that the altered soluble sugar composition in RNAi seeds observed by Rosnoblet et al. (2007) is related to the phenotypes observed in this study. We first verified whether this difference in sugar content had an effect on osmotic potential, which might explain the germination phenotype. However, when expressed on a molar basis, no difference exists in sugar content between mutant and wild-type seeds, indicating that the final difference in osmotic pressure due to differences in sugar composition is negligible. Our pharmacological studies and quantitative RT-PCR experiments indicated a complex relationship between sugar perception, signaling and dormancy. Whereas glucose and mannose delay seed germination of Arabidopsis (Pego et al., 1999), this is not the case for M. truncatula, raising the question of whether differences in mechanisms of sugar sensing exist between species. After-ripening modified the perception of 3-OMG in M. truncatula seeds. In non-dormant seeds, 3-OMG induced expression of PAL, CHR and IFR, whereas expression of these genes was repressed by 3-OMG in dormant seeds. It remains to be investigated whether this effect implies sugar sensing via a hexokinase-independent pathway or perturbed metabolism due to a putative accumulation of 3-O-methylglucose-6-phosphate (Cortès et al., 2003), which may have a different effect on dormant versus non-dormant seeds. The effect of 3-OMG on gene expression appears to be independent of MtSNF4b. The difference between wild-type and RNAi seeds is most likely due to a lower dormancy level in the 2-week AR RNAi seeds, which is comparable to that of wild-type seeds after 4 weeks of AR.
Our results rule out the possibility that a hexose signaling pathway acts upstream of MtSNF4b, because this implies that these sugars would activate sugar signaling pathways in wild-type seeds but not in RNAi Mtsnf4b seeds. However, glucose and mannose appear to act downstream of MtSNF4b in activating gene expression. Both sugars are capable of restoring PAL, CHR and IFR expression in Mtsnf4b mutants to levels comparable to those of wild-type seeds after 2 weeks of AR. Also, mannose delays germination of RNAi seeds but not of wild-type seeds. These results suggest that activation of a hexokinase-mediated signaling pathway is capable of restoring the phenotype of the mutants and coupling germination behavior with defense gene expression. However, glucose does not delay germination, leaving the possibility open that AR/germination and gene expression are controlled by two independent signaling pathways that are both influenced by MtSNF4b. An alternative explanation is that mannose de-regulates carbon and energy metabolism more profoundly when MtSNF4b is silenced.
The effect of MtSNF4b on biotic stress-related gene expression during hydration of dormant seeds appears to be uncoupled from the maturation phenotype. We showed that expression of these genes is activated during imbibition of freshly harvested dry wild-type seeds, and not the result of accumulation of transcripts during maturation. In SNF4b-deficient seeds, this activation is largely impeded. The genes that are differentially expressed as a result of silencing MtSNF4b are classified in the groups ‘disease, virulence and defense’, ‘cell wall’ and ‘secondary metabolism’. MtSNF4b silencing-induced reduction in PR10 transcripts was also found at the translational level. However, the role for PR10 proteins in biotic stress is circumstantial. PR10 genes are known to be induced by pathogens as well as abiotic stresses (McGee et al., 2001; Hashimoto et al., 2004). Silencing of PR10 resulted in an increased resistance to Aphanomyces euteiches, probably due to increased expression of PR-5b proteins (Colditz et al., 2007). On the other hand, over-expression of pea ABR17 in Arabidopsis rendered the seeds and seedlings more tolerant to abiotic stresses (Srivastava et al., 2006). The functionality of the largest group of genes, which are related to phenylpropanoid metabolism (PAL, C4H and CCoAOMT) and flavonoid and isoflavonoid biosynthesis genes (CHS, CHR, CHI, IFS, HID, IFR and VR), remains to be tested in M. truncatula seeds. Most of these genes were found to be induced by pathogens or elicitors in M. truncatula and correlate with the accumulation of medicarpin (Deavours and Dixon, 2005; Dhaubhadel et al., 2007; Naoumkina et al., 2007). Medicarpin is an anti-fungal pterocarpan phytoalexin that is produced by many legumes (Mundodi et al., 2001). Furthermore, medicarpin has been found to delay alfalfa seed germination (Miller et al., 1988;Dornbos et al., 1990). It may be that this phytoalexin is present in seeds primarily to avoid predation or microbial infections, but also has an effect on the regulation of germination in legume seeds.
What might be the downstream targets of the SnRK1/SNF4b complex that are responsible for the differential gene expression in relation to biotic stress? Direct targets are most likely (de)activated through phosphorylation by the catalytic α subunit and will not be identified by the transcriptome analysis. Nonetheless, expression of several WRKY genes in the F dataset points suggests possible downstream regulation by MtSNF4b. A recent study on a kinase–protein interaction in rice identified SnRK1 as a potential phosphorylation candidate of WRKY (Ding et al., 2009). WRKY proteins are involved in the regulation of plant defense responses (Eulgem and Somssich, 2007). A massive induction of WRKY transcription factor genes is correlated with the induction of genes involved in the central phenylpropanoid pathway and the downstream steps in the biosynthesis of medicarpin (Naoumkina et al., 2007). Over-expression of the same WRKY genes as those found in our transcriptomic analysis (W100630 and W108715) in transgenic tobacco enhances typical defense responses such as PR protein production and accumulation of soluble and wall-bound phenolic compounds (Naoumkina et al., 2008). In addition, they were shown to be induced by either yeast elicitor or methyl jasmonate (Naoumkina et al., 2008). Our transcriptomic analysis also revealed a link to jasmonic acid (JA). Three 12-OPDA reductase genes are down-regulated by silencing MtSNF4b, irrespective of the AR status of the seeds. These genes are responsible for the reduction of 12-OPDA to form OPC-8:0, which is subjected to β-oxidation to yield JA. JA and OPDA are both active as signaling molecules and induce the expression of overlapping but distinct sets of genes involved in biotic stresses (Taki et al., 2005).
Although a number of genes regulated by MtSNF4b are also found in the D dataset, additional genes related to defense are also differentially regulated when MtSNF4b is silenced, suggesting that this subset of genes is directly related to the action of SNF4b and not simply to the reduced dormancy status of the mutant embryos. Imbibed dormant M. truncatula seeds have seed coats that remain devoid of any contaminating fungi and bacteria for months, whereas isolated seed coats are readily infected (data not shown). Expression studies in dormant Arabidopsis seeds demonstrated that genes related to defense and protection are also highly expressed (Cadman et al., 2006). It remains to be tested whether the activation of defense gene expression by the SNF4b/SnRK1/β complex in seeds renders them more resistant to pathogen attack. An interesting parallel can be made with the developmentally induced dormancy in the drought-tolerant plant Retama raetam, a desert legume (Pnueli et al., 2002). Dormancy in this plant is accompanied by the accumulation of transcripts encoding PR10-like proteins and WRKY transcription factors, whereas non-dormant plants subjected to stress conditions contained transcripts encoding small heat shock proteins.
Detailed experimental procedures used in this study are supplied in Appendix S3.
Plant material and treatments
Seeds of Medicago truncatula Gaertn. (R108) were harvested at the same time for all genotypes from plants that were grown side by side under identical environmental conditions as described by Rosnoblet et al. (2007). After-ripening occurred during storage at 60% RH, 20°C in the dark. After various durations of AR, batches of 50–100 seeds were scarified using sand paper and imbibed in water or in a polyethylene glycol solution (−0.27 MPa) at 20°C in the dark, and were considered to have germinated when the radicle protruded from the surrounding envelope. Statistical analysis of the germination data was performed using the χ2 test, and differences were considered significant at P < 0.05. To release dormancy, seeds were either imbibed for 48 h at 4°C and subsequently transferred to 20°C, or imbibed in 100 μm fluridone [1-methyl-3-phenyl-5-(3-trifluoromethyl-(phenyl))-4-(1H)-pyridinone] (Duchefa, http://www.duchefa.com). ABA sensitivity was tested on seeds after 8 months of AR that were imbibed at various concentrations of ABA (Sigma, http://www.sigmaaldrich.com/) (control in 0.1% EtOH) in the dark at 20°C, and the final percentage germination was determined. For the microarray analysis, seeds at 2–3 weeks and 6 months of AR were imbibed in the dark for 6 h. For pharmacological studies, seeds were imbibed at 20°C in the dark in water or 50 mm 3-O-methylglucose (Sigma), 60 mm glucose (Sigma), 20 mm mannose (Sigma) or 40 mm glucosamine (Sigma) at various durations of AR. For quantitative RT-PCR studies, 20 h-imbibed embryos were removed from the seed coat and endosperm and frozen in liquid nitrogen.
Plasmid construction and plant transformation
SNF4b RNAi plasmid construction using the binary vector pFGC5941/Gateway (pFRB), transformation of the plasmid into M. truncatula and selection of positive lines were performed as described by Rosnoblet et al. (2007). Four homozygous lines (T2) were retained for further analysis. Two additional transformants were used as negative controls: an R108 line transformed with the empty vector (RNAi empty) and an R108 line transformed with the pFGC5941 vector containing an RNAi construct targeted against the proximal region (1.2 kb) of the MtENOD40-1 promoter (RNAi control). All studies were performed on seeds of the T2 or T3 generation.
Gel filtration chromatography and Western blot
Seeds at 1–2 weeks of AR were imbibed for 48 h at 4 or 20°C in the dark, after which native proteins were extracted and separated according to their molecular weight as described by Rosnoblet et al. (2007). Fractions were analyzed by Western blotting using an anti-SNF4b antibody as described by Rosnoblet et al. (2007). To study the effect of ABA, immature embryos at 16 days after pollination were incubated for 6 h in various concentrations of ABA. Soluble proteins were extracted and analyzed by Western blotting (10 μg/lane) using the anti-SNF4b antibody. For PR10 protein induction, soluble proteins were extracted from 36 h imbibed embryos after 2 weeks of AR and analyzed by Western blotting using an antibody against pea PsRH2 that recognizes MtPR10-1 (Ramos and Bisseling, 2003).
The cDNA of MtSNF4b was cloned into pENTR/D-TOPO and recombined with pYES-DEST52 (Invitrogen, http://www.invitrogen.com/). The plasmid pYES-DEST52-MtSNF4b and the empty plasmid were introduced into the Δsnf4 mutant of Saccharomyces cerevisiae MCY2634 (MATa his3-Δ200 leu2-3112 ura3-52 snf4Δ2). Strains were plated on yeast nitrogen base (YNB)/2% glucose or YNB/2% galactose.
Bacterial two-hybrid analysis
MtSNF4b was cloned into the pBT vector fused with a lamba-cl DNA binding site and MtSNF1 (TC116507) was cloned into the pTRG vector fused with RNA polymerase-α. Reporter competent cells were co-transformed with both vectors using bacterioMatch II (Stratagene, http://www.stratagene.com/) according to the manufacturer’s protocol, and spotted on selective (3-amino-1,2,4-triazole and streptomycin) and non-selective media by 20 μl of saturated culture (1/1, 1/10, 1/100, 1/1000) and incubated at 37°C for 1 and 2 days, respectively. Negative controls were bacteria co-transformed with pTRGGal11 and pBTMtSNF4b or pTRGMtSNF1 and pBT vectors.
ABA content was measured in dry and imbibed seeds at 1 week and 8 months of AR, using two harvests. Imbibition times were 24 h (1 week AR) or 18 h (8 months AR). Extraction was performed on three biological replicates of 25 seeds as described by Jiang and Zhang (2001). ABA concentration was determined using the Phytodetek ABA immunoassay kit according to the manufacturer’s instructions (Idetek).
Three microarray experiments were conducted to compare the transcriptome in embryos of (i) wild-type versus two RNAi lines at 2–3 weeks of AR, (ii) wild-type versus one RNAi line at 6 months of AR, and (iii) wild-type embryos at 2 weeks versus 6 months of AR (dormant versus nondormant). Microarray analysis replicates consisted of three independent biological replicates of 50 embryos, obtained from different plants. Probe labeling using the CyScribe cDNA post-labeling kit (GE Healthcare, http://www.gelifesciences.com), hybridization and image acquisition were performed as described by Buitink et al. (2006). Dye-specific variations were accounted for using a dye swap between the various wild-type RNA samples. Hybridization was performed using Mt16kOLI1Plus chips (Hohnjec et al., 2005). Image data processing was performed using GENESPRING software (Agilent). Genes with a significant difference (P value <0.01) and >1.5 fold change in expression were considered.
Real-time quantitative PCR analysis
Reverse transcription reactions were performed on 1 μg of total RNA using a QuantiTect reverse transcription kit (Qiagen, http://www.qiagen.com/). Quantification of cDNA by quantitative RT-PCR was performed using an ABI PRISM 7100 sequence detection system (Applied Biosystems, http://www.appliedbiosystems.com/) and SYBR Green. Values are based on three repetitions. The constitutively expressed mRNA Msc27 was used as a housekeeping gene for standardizing data and to assess the efficiency of quantitative RT-PCR (Gallardo et al., 2007). Amplification primers are listed in Table S2. Changes in transcript abundance were estimated as the fold change relative to control conditions (Livak and Schmittgen, 2001).
We thank Dr Marten Denekamp (Syngenta Seeds, Enkhuizen, the Netherlands) for providing the pFRB RNAi binary vector and Dr P. Ratet (Institut des Sciences Vegetales, Gif-sur-Yvette, France) for the gift of seeds of Medicago truncatula cv. R108. We are indebted to Dr H. Franssen (Wageningen University, the Netherlands) for providing the PsRH2 antibody to perform the PR10 analysis. Lysianne Brocard and Martin Crespi (Institut des Sciences Végétales, CNRS, Gif sur Yvette, France) are gratefully acknowledged for the gift of mutants containing an RNAi construct targeted against the proximal region (1.2 kb) of the MtENOD40-1 promoter. The yeast Δsnf4 strain MCY2634 was kindly provided by Dr Maria Carlson (Columbia University, NY, USA).