•Seeds can enter a state of dormancy, in which they do not germinate under optimal environmental conditions. Dormancy can be broken during seed after-ripening in the low-hydrated state.
•By screening enhancer trap lines of Arabidopsis, we identified a role for the NADPH-oxidase AtrbohB in after-ripening. Semiquantitative PCR was used to investigate AtrbohB transcripts in seeds. These methods were complemented with a pharmacological approach using the inhibitor diphenylene iodonium chloride (DPI) and biomechanical measurements in the Brassicaceae seed model system cress (Lepidium sativum) as well as protein carbonylation assays.
•atrbohB mutants fail to after-ripen and show reduced protein oxidation. AtrbohB pre-mRNA is alternatively spliced in seeds in a hormonally and developmentally regulated manner. AtrbohB is a major producer of superoxide in germinating Arabidopsis seeds, and inhibition of superoxide production by diphenylene iodonium (DPI) leads to a delay in Arabidopsis and cress seed germination and cress endosperm weakening.
•Reactive oxygen species produced by AtrbohB during after-ripening could act via abscisic acid (ABA) signalling or post-translational protein modifications. Alternative splicing could be a general mechanism in after-ripening: by altered processing of stored pre-mRNAs seeds could react quickly to environmental changes.
In the field, dormancy is broken by a species-specific set of conditions, including changes in ambient temperature, light penetration into the soil and soil hydration. In the laboratory, dormancy of Arabidopsis seeds is often broken by cold stratification of dark-imbibed seeds, which is known to raise levels of gibberellins (GA; Penfield et al., 2005). Other means are light regimes whose action is mediated by phytochrome-induced GA biosynthesis, provision of the seeds with nutrients such as nitrate or the direct application of hormones counteracting dormancy, usually GA. Dormancy release also occurs during seed after-ripening (i.e. air-dry storage of seeds at room temperature for several months).
In addition to breaking dormancy and widening the window of environmental conditions in which seeds will complete germination, characteristics of after-ripening are a loss of sensitivity to the germination-inhibiting hormone abscisic acid (ABA) and an increase in sensitivity to germination-promoting hormones such as GA or ethylene (Finch-Savage & Leubner-Metzger, 2006). Fresh and after-ripened seeds differ not only in hormone signal transduction, but also in hormone contents. Changes in hormone contents are regulated both on the biosynthesis and degradation levels (Kucera et al., 2005; Cadman et al., 2006).
After-ripening leads to more synchronized and faster seed germination. In several species and conditions, dormant seeds do not germinate at all, while after-ripened seed populations quickly complete germination to 100%. Examples are the grains of wild oat (Poaceae) described by Adkins & Ross (1981) and barley (Poaceae), which do not germinate in the light in the dormant state in contrast to the after-ripened state (Gubler et al., 2008). In all these cases, after-ripening takes place at room temperature in seeds stored in the air-dry state. It has been suggested that after-ripening and loss of dormancy are two distinct processes, as ABA-deficient, nondormant Arabidopsis seeds still show transcriptome changes characteristic for after-ripening after several months of storage at room-temperature (Carrera et al., 2008; Holdsworth et al., 2008).
Arabidopsis ecotypes vary widely in the depth of their dormancy, as measured in their capacity for after-ripening. Deeply dormant ecotypes such as Cape Verde Islands (Cvi) require many months of after-ripening to lose dormancy, while the ecotype Columbia (Col), commonly used in laboratories around the world and in this study, has a very shallow dormancy. The ecotype Cvi was used by Bentsink et al. (2006) for a quantitative trait locus (QTL) analysis of seed dormancy. In this QTL analysis, delay of germination 1 (DOG 1) was identified as a major player in seed dormancy control. It later emerged that in addition to expression differences between accessions differing in dormancy, a complex pattern of alternative splicing is needed for DOG1 function (Schwab, 2008).
Once dormancy is broken, germination can proceed if conditions are favourable. Arabidopsis and the closely related Brassicaceae species garden cress (cress, Lepidium sativum) germinate in two sequential steps: first the testa ruptures, revealing the underlying endosperm still covering the radicle. Only after a lag of several hours or even days is the endosperm ruptured and the radicle emerges (Müller et al., 2006). By definition, endosperm rupture is the end of the germination process which begins with seed imbibition (Bewley, 1997). Endosperm rupture of Arabidopsis and cress can be specifically delayed by the addition of physiological concentrations of ABA to the germination medium, an effect that can be countered by GA and ethylene (Kucera et al., 2005; Müller et al., 2006). This delay is correlated with a delay in weakening of the endosperm layer which covers the radicle tip. Endosperm weakening is a prerequisite for radicle emergence (Müller et al., 2006). Both radicle growth potential and endosperm properties determine germination behaviour.
Seed after-ripening and germination can alter gene expression in a tissue-specific manner. In the work described here, we used Arabidopsis enhancer trap lines to identify genes regulated specifically during germination in an after-ripening and/or ABA-dependent manner in the embryo or endosperm. Enhancer trap techniques use a reporter gene construct driven by a minimal promoter which is only activated if the insertion site is close to an enhancer element in an endogenous promoter of the transgenic plant. A large-scale approach using a CaMV35S–minimal promoter–Glucuronidase (GUS) construct in Arabidopsis was initiated almost 10 yr ago with the aim of characterizing gene expression patterns in inflorescences (Campisi et al., 1999). Seeds from this library were used by Liu et al. (2005a) to screen for seeds with a GUS signal during germination. This helpful resource has already been used to identify a transcription factor which positively regulates seed germination (Liu et al., 2005b).
By using seeds from this collection we found, in the present work, that the NADPH-oxidase AtrbohB plays a role in seed germination and after-ripening. NADPH-oxidases, also known as respiratory burst oxidase homologues (Rboh) are named after a homology with the gp91phox domain of the animal respiratory burst oxidase which is involved in the immune response (Keller et al., 1998). Rbohs localize to the plasma membrane, where they can transfer electrons from cytosolic NADPH or NADH to apoplastic oxygen, leading to the production of apoplastic superoxide (Sagi & Fluhr, 2006). The gene family has 10 known members in Arabidopsis; AtrbohA-J (Torres & Dangl, 2005) and homologues have also been found in a number of other angiosperm species (Simon-Plas et al., 2002; Yoshioka et al., 2003; Sagi et al., 2004; Kobayashi et al., 2006; Sagi & Fluhr, 2006). The individual Rbohs differ in their expression pattern across plant tissues and organs and have been shown to be involved in a range of processes including pollen tube elongation (Potockýet al., 2007), response to pathogen attack (Torres et al., 2002) and wounding (Sagi et al., 2004) and root elongation and root hair development (Foreman et al., 2003). Superoxide and other reactive oxygen species (ROS) have been proposed to play roles in seed germination and dormancy (Bailly, 2004; Oracz et al., 2007; Müller et al., 2009). However, it is not known which roles Rbohs play in seeds.
Materials and Methods
Three lots of 50 seeds of Arabidopsis (Col-0; atrbohA-I (from Miguel Angel Torres and Jonathan Jones); enhancer trap lines (Liu et al., 2005a) ordered from NASC) or Nicotiana tabacum L. (cv Xanthi, WT and ntrbohD; Simon-Plas et al., 2002), were placed on 1/10 Murashige–Skoog (MS) salts solidified with 1% (w : v) agar. Germination took place at 24°C in continuous light. Lepidium sativum L. seeds (cress, ‘Gartenkresse einfache’; Juliwa, Heidelberg, Germany) were incubated in Petri dishes on two layers of filter paper soaked with 1/10 MS salts at 18°C in continuous light as described by Müller et al. (2006). Where indicated, 1 μm (Arabidopsis) or 10 μm (cress) cis-S(+)-abscisic acid (ABA; Duchefa, Haarlem, the Netherlands) was added to the medium or cold stratification was conducted for 24 h at 4°C in darkness. Germination was scored under a binocular microscope by counting the incidence of endosperm rupture in the seed populations. Seeds were considered fresh for 3 wk after harvest and after-ripened after at least 6 months of air-dry storage at room temperature. Diphenylene iodonium chloride (DPI; Sigma-Aldrich) was dissolved in dimethyl sulphoxide (DMSO) as a 10 mm stock and added to the germination media in the concentrations indicated. Control media for comparison had an equal volume of DMSO added.
GUS staining and photographic documentation
Seeds were placed in Petri dishes on two layers of filter paper and covered in 5 ml staining solution (100 mm phosphate buffer (pH 7.0), 0.1% (v : v) Triton X-100 and 2 mm X-Gluc). After 24 h of incubation (dark, 24°C) the testa was removed. The staining was examined and photographed another 24 h later. All photographs of seeds were taken with the software IM1000 (Leica Microsystems, Wetzlar, Germany) and a Leica DCF480 digital camera attached to a Leica12.5 binocular microscope.
Identification and verification of T-DNA insertion sites in enhancer trap lines
Genomic DNA was extracted using the DNeasy Plant Mini kit (Qiagen). Genome-walking PCR was performed using the GenomeWalker and Advantage 2 Polymerase Mix Kits (Clontech-Takara Bio Europe, Saint-Germain-en-Laye, France). Genomic DNA of CS24365 was digested with StuI. Two specific primers for the T-DNA insert were designed, the first (5′-AACTTAATCGCCTTGCAGCACA-3′) being used in the first round of PCR, the second (5′-ATGAGACCTCAATTGCGAGCT-3′) serving as a nested primer in the second round of PCR. The resulting fragments were cloned into pGEM-TEasy vectors and sequenced. Thermal asymmetric interlaced PCR and arbitrary degenerate primer design was carried out according to Liu & Huang (1995). The two primers mentioned above plus another nested primer (5′-ATCGCCTTGCAGCACATCCC-3′) were used as insert-specific primers. Fragments were purified from an agarose-gel using the DNA Gel Extraction Kit (Fermentas, St Leon-Rot, Germany) and directly sequenced.
Reverse transcription and semiquantitative PCR
RNA was extracted from seedlings of Arabidopsis wild type (WT) and mutants with the Plant RNeasy Kit (Qiagen) including the optional DNAse step and reverse transcribed (SuperscriptIII; Invitrogen) with a mix of oligodT and random hexamer primers (1 : 1). The resulting cDNA was used as a template with rRNA-primers RR-FOR (5′-CGAGCTGATGACTCGCGCTTA-3′) and RR-REV (5′-GAGTGGAGCCTGCGGCTTA-3′) and AtrbohB-primers B-FOR (5′-GGAATTCC CGTCAGAAGGTGAGACAAACA-3′) and B-REV (5′-CGGGATCCCG CCAGAATTCAAACAATTCAGTTTT-3′). To confirm splice variants of AtrbohB-ß, we used forward primers B-intr-1-for (5′-CACTGCAATACTCAACTTGTTT-3′) and B-intr-1-for-2 (5′-CTCTCTGTAATTTCACTGCAA-3′) with reverse primers B-ex2/ex3-rev (5′-GTCCAAATTTTTGTCCACCATA-3′) and B-ex3/ex4-rev (5′-GCAATTATCT CCTTAACCTCAT-3′). Primers used for testing alternative splicing in other Atrbohs were A-FOR (5′-ACGGATTATTACACCGATCTAGA-3′), A-REV (5′-ATAA TCTCTCTTACTTCGGCTT-3′), D-FOR (5′-CCTC TTACTCTCTGCCAAGT-3′), D-REV (5′-AATAATCTCAGCCACCTCTTC-3′), E-FOR (5′-GGATTGCGATT GCGATCAGA-3′), E-REV (5′-AAGTTCTTTGATTTCTTCTCTAG-3′), F-FOR (5′-TCGCTCCGATTTCGCTCAAT-3′), F-REV (5′-ATTATCTCTTTTACTTCCTCTTC-3′), I-FOR (5′-GGAATCGTTGATTGGAACGAT-3′) and I-REV (5′-TCTCTTTTACTTCATTCTC TGTG-3′). For all primer pairs, we used an annealing temperature of 60°C. A total of 35 PCR cycles were used for rboh-primers and 25 cycles for the rRNA. Both were confirmed to be in the linear amplification phase.
In-situ RNA hybridization
In-situ RNA hybridization was performed as described previously (Mayer et al., 1998). The AtrbohB–probe cDNA was amplified from reverse-transcribed RNA of seedlings using primers B-FOR and B-REV (see earlier). After digestion with EcoRI and BamHI, the cDNA was ligated into pBluescript II KS+. For the antisense probe, the plasmid was linearized with EcoRI and transcribed with T7 RNA polymerase (Promega) using a digoxigenin-labelling kit (Roche Diagnostics); for the sense probe, the plasmid was linearized with BamHI and transcribed with T3 RNA polymerase (Promega).
Puncture force measurements
Puncture force was measured as described previously (Müller et al., 2006). In short, cress seeds were cut in half, the radicle removed without damaging the endosperm cap and a metal probe slowly lowered into the empty endosperm cap. The force it took to rupture the endosperm was recorded with a custom-made machine.
Genotyping of atrbohB
The atrbohB insertion mutant was identified by performing a PCR reaction on genomic DNA extracted from pools of Col-0 plants containing dSpm transposon insertions (Tissier et al., 1999). Primers used include dSpm11 and dSpm1 (Tissier et al., 1999) and specific primers for AtrbohB 154B (5′-GAATAATGTAATTGTAGTGAATGCG-3′) and 75 B (5′-ACAAATTCGCTAGATTCAACCAT-3′). The PCR products, 154B/dSpm1 and 75B/dSpm1, were 150 bp and 500 bp respectively. Insertions in the gene were confirmed by sequencing PCR products spanning the insertion. The line identified was homozygous and derived from a BASTA-resistant heterozygous parent.
Superoxide localization by nitroblue tetrazolium (NBT) staining
Arabidopsis embryos were dissected from imbibed seeds under a binocular microscope, equilibrated for 10 min in 20 mm phosphate buffer, pH 6.5, and transferred to a staining solution containing 5 mm NBT in 20 mm phosphate buffer, pH 6.5. Seeds were left in the staining solution until staining was visible.
Extraction of total soluble proteins and Oxyblots
Two biological replicates of 100 μl Arabidopsis seeds were ground with mortar and pestle in liquid nitrogen and thawed in thiourea–urea buffer (Harder et al., 1999) with the protease inhibitor cocktail ‘Complete Mini’ (Roche), 60 U ml−1 DNAse and 6 U ml−1 RNAseA. After shaking for 30 min at 4°C, samples were centrifuged at 14800 g for 15 min at 4°C until the supernatant was clear. The final supernatant corresponds to the soluble protein extract. Protein concentration was determined with a Bio-Rad assay kit.
Bovine gamma globulin was used as a standard protein samples (10 μg) proteins were separated in one-dimensional electrophoresis on 11% polyacrylamide gels and stained with Coomassie brilliant blue. 2,4-Dinitrophenylhydrazone (DNP) labelling and detection of carbonylated proteins was performed as described in Job et al. (2005). Sodium dodecyl sulphate (SDS) was added to the extracts with a final concentration of 0.8% (w : v) and dialysis against water performed over night at 4°C. Four volumes of 10 mm dinitrophenylhydrazine (DNPH) in 2 mm HCl were added after dialysis and samples mixed for 30 min at room temperature. The labelled proteins were then precipitated with 20/80 TCA/acetone containing 1 mM dithiothreitol (DTT) and resolubilized in thiourea–urea buffer (Harder et al., 1999) with 2% (v : v) Triton X and 20 mm DTT. After separation on one-dimensional gels, proteins (80 μg) were transferred by semi-dry blot onto nitrocellulose membranes. The DNP-labelled carbonylated proteins were detected with a rabbit anti-DNP antibody (Serologicals, Norcross, GA, USA) followed by a secondary antibody coupled to horseradish peroxidase (Sigma-Aldrich). Detection was performed with the ECL Plus kit (GE Healthcare, Little Chalfont, UK).
Sequences were aligned and analysed with the software geneious 4.0.2 (Biomatters Ltd, Auckland, New Zealand). The implemented muscle algorithm (Edgar, 2004) was used for alignments.
After-ripening promotes seed germination in Brassicaceae and other families
We compared the germination process of fresh and after-ripened cress seeds, fresh and after-ripened cress seeds by time-course analyses of their endosperm rupture. Mature cress seeds displayed only a shallow dormancy. Even immediately after harvest, c. 80% of the seeds germinated readily within 4 d (Fig. 1). After the seeds were after-ripened for several months, the final germination percentage rose to almost 100%, the kinetics of endosperm rupture was more uniform and the overall germination speed increased: all seeds completed germination within 1 d. The effect of after-ripening on germination can be demonstrated in a range of different species. We compared the t50% values of fresh and after-ripened seeds (i.e. the mean times required for the populations to reach 50% endosperm rupture) of four endospermic species. After-ripening reduced the t50% values about twofold for the Brassicaceae seeds of cress (Fig. 1) and Arabidopsis (Fig. 2). After-ripening also promoted the germination of Solanaceae seeds such as tobacco (Nicotiana tabacum) and tomato (Lycopersicon esculentum), where the after-ripened : fresh t50% ratios were 0.81 and 0.23, respectively (Groot & Karssen, 1992; Leubner-Metzger, 2003).
The extent of the after-ripening mediated changes is not only species specific, but also ecotype specific and to a certain extent even batch specific. Seed after-ripening is thus an important developmental process that occurs in low-hydrated seeds and determines the seeds’ responses to environmental cues. We aimed to identify genes that play a role in this process by screening a selection of enhancer trap lines with reporter-gene expression in seeds for their germination phenotypes in the fresh and after-ripened state.
Seeds of an enhancer trap line with an insertion in AtrbohB have an after-ripening phenotype
We obtained several lines from the seed–GUS-expression library donated by the group of Hiro Nonogaki to the Arabidopsis Biological Resource Center. These seeds had been prescreened for GUS staining during germination (Liu et al., 2005a,b). We propagated and rescreened these lines for their GUS expression patterns in order to determine the spatial activity of the trapped enhancer in after-ripened germinating seeds. Seeds were dissected to clearly localize staining and grouped as showing staining in the embryo and/or the endosperm (Table S1). All lines were also screened for a germination phenotype, and only those that differed in the kinetics of endosperm rupture when compared with the wild type (WT, Col-0) were analysed further. As after-ripening includes changes in ABA-sensitivity, we conducted germination kinetics of fresh as well as after-ripened seeds on medium with and without ABA, respectively. Table S1 lists lines, localization of staining and the germination phenotypes.
Line CS24365 showed a particularly interesting phenotype with regard to after-ripening (Fig. 2): this is evident from detailed time-course analyses of the endosperm rupture of fresh and after-ripened seeds of CS24365 in medium without (CON; Table S1) and with ABA (Fig. 2). We did not observe differences in CON, but fresh seeds completed germination earlier than WT seeds on medium with ABA (Fig. 2a). Moreover, while WT seeds became less sensitive to ABA during after-ripening, the germination kinetics of CS24365 did not change: for after-ripened seeds, the insertion line completed germination later than the WT (Fig. 2b). A comparison of the associated t50% values showed that while the t50% decreased over twofold for the WT, hardly any change could be observed for the mutant (Fig. 2c). The after-ripening mediated decrease in the ABA sensitivity of endosperm rupture thus seems to be blocked in the CS24365 line.
Genome walking led to the identification of the T-DNA insertion sites in enhancer trap line CS24365 (Fig. 3a), which was confirmed by thermal asymmetric interlaced (TAIL)-PCR. The GUS-construct localized to chromosome 1, basepair 2932468, which is in the promoter region 300 bp upstream of the NADPH-oxidase AtrbohB (At1g09090). CS24365 showed a clear GUS staining in the whole embryo (Fig. 3b) but not the endosperm layer. Semiquantitative PCR confirmed that the expression of AtrbohB is lower in imbibed after-ripened CS24365 seeds than in the corresponding WT (Fig. 3c). The transcript of BIP3 (At1g09080), the other gene close to the insertion site, was not affected (data not shown). In order to confirm that the phenotype was actually caused by the insertion in the AtrbohB gene promoter, we compared the germination phenotype of CS24365 with that of the insertion mutant atrbohB, which carries a dSpm insertion in the coding region itself (Torres et al., 2002). Both lines behaved in the same way (see atrbohB-mutant in Table 1). Thus, the reduced expression of AtrbohB led to an after-ripening phenotype, where the ABA sensitivity does not change during after-ripening.
Table 1. Some atrboh mutants germinate differently from wild type (WT) seeds
Germination fresh CON
Germination fresh ABA
Germination after-ripened CON
Germination after-ripened ABA
RNA localization in WT after-ripened CON
Fresh and after-ripened Arabidopsis mutant seeds were germinated at 24°C in continuous light without (CON) or with 1 μm ABA added and after 1 d of stratification at 4°C in darkness. Symbols: +, germinated faster than WT; 0, germination equals WT; −, germinated slower than WT. RNA localization of the corresponding gene by in situ hybridization: EM, embryo; ES, endosperm; ND, not determined.
AtrbohB is differentially expressed and the pre-mRNA alternatively spliced in germinating Arabidopsis seeds depending on after-ripening status and ABA
We investigated the expression pattern of the AtrbohB transcript by semiquantitative reverse-transcription polymerase chain reaction (RT-PCR) on WT seed RNA extracted from fresh and after-ripened seeds imbibed for 24 h in the absence and presence of 1 μm ABA. At this point, neither fresh nor after-ripened seeds exhibited endosperm rupture. Interestingly, we observed two bands of AtrbohB-transcripts (Fig. 4a) which seem to be the products of alternative mRNA splicing (Fig. 4b): sequencing of the two fragments showed that in the transcript leading to the larger band (635 bp –AtrbohB-β), intron 1 is retained, while intron 1 is spliced out in the smaller fragment (485 bp –AtrbohB-α). In order to confirm the identity of the larger band as a product of alternative splicing and exclude the possibility of genomic DNA contamination, we did a PCR with a forward primer in intron 1 and reverse primers spanning the exon border of exons 2/3 and exons 3/4, respectively. Both sets of primers yielded PCR bands of the expected sizes, confirming that AtrbohB-α and AtrbohB-β are alternative splicing variants of the AtrbohB pre-mRNA. A sequence alignment of the AtrbohB genomic sequence with the AtrbohB-α and AtrbohB-β cDNAs showed that the retention of intron 1 leads to a premature termination codon (PTC; Figs 4b and S1). It is highly unlikely that AtrbohB-β is translated into a protein, as a PTC at such an early point will almost certainly lead to nonsense-mediated mRNA decay (NMD; Kertész et al., 2006).
Figure 4a shows that the expression of the two mRNA splicing variants AtrbohB-α (without intron 1) and AtrbohB-β (including intron 1) is regulated differentially in fresh and after-ripened WT-seeds in the presence and absence of ABA (Fig. 4a): both transcript variants are expressed in fresh seeds imbibed in CON medium. The addition of 1 μm ABA to the medium caused the accumulation of AtrbohB-ß mRNA in fresh seeds and prevented AtrbohB-α mRNA expression. In after-ripened seeds, in which AtrbohB-α was found to be the dominant splicing form in both CON and ABA medium, ABA prevented AtrbohB-β mRNA expression. Thus, it appears that developmental factors (after-ripening) and hormonal regulation (ABA) both affect the relative expression of the two AtrbohB splicing variants. Alternative splicing of intron 1 was not evident for other Atrbohs investigated which demonstrates a specificity of the effect for AtrbohB.
We also checked which splicing variants are present in dry seeds. We observed that only splice variant AtrbohB-ß or pre-mRNA, which would yield the same band, is present in dry fresh seeds (Fig. 4a). Thus, 24 h imbibition of fresh seeds probably activated splicing of intron 1 in medium without, but not with ABA. In after-ripened dry seeds, we observed both bands in approximately equal strength. The transcript profiles of after-ripened imbibed seeds can thus to result from altered splicing efficiency. Nonsense-mediated mRNA decay may play an additional role in removing AtrbohB-ß transcripts.
A database search on NCBI for AtrbohB cDNAs provided two sequences, in which a second instance of alternative splicing was found: NM_100780.3 retains intron 9, which is not present in NM_202070.1. Retention of intron 9 also leads to a premature translation stop (Fig. S2), leaving out the NADPH binding sites. It is still unclear which of the splicing variants are translated into proteins.
AtrbohB is a major producer of superoxide in embryos of germinating Arabidopsis seeds
RNA in-situ hybridization in WT seeds showed that AtrbohB transcript expression localizes to the embryo (Fig. 5a). This finding is in agreement with the GUS-staining in the enhancer-trap line (Table S1).
While we cannot quantify the amount of ATRBOHB-protein as no specific antibodies are available and we cannot extract sufficient membranes from seeds, we can detect the product of their enzymatic activity: RBOHs are NADPH-oxidases that produce apoplastic superoxide. Superoxide reacts with NBT to a blue-ish precipitate. Histostains with NBT showed that superoxide is produced in the embryo of 24 h-imbibed Arabidopsis WT seeds (Fig. 5b). Superoxide production was strongly reduced in the insertion mutant atrbohB (Fig. 5b) and the enhancer trap line CS24365 (data not shown), which points to a major role of RBOHB in the superoxide production in the embryo during seed germination. In accordance with this, DPI, an inhibitor of NADPH-oxidases and other flavin-containing enzymes, inhibited superoxide production in WT seeds (Fig. 5c).
After-ripened atrbohB seeds show reduced protein carbonylation
Our findings indicate that ATRBOHB is largely responsible for superoxide production in Arabidopsis embryos. Oracz et al. (2007) showed that targeted protein oxidation occurs during after-ripening as well as artificial breakage of dormancy. A possible explanation for the after-ripening phenotype of the atrbohB mutant is thus that its reduced superoxide production during the after-ripening period leads to a reduction of the required protein oxidation.
We therefore decided to study protein carbonylation, an identifier of protein oxidation (Möller et al., 2007), in fresh and after-ripened WT and atrbohB seeds by immunoblots with antibodies against derivatives of carbonyls in soluble proteins (Fig. 6). We compared dry and 6-h imbibed seeds. Levels of protein carbonylation were slightly higher in fresh dry WT seeds than in atrbohB seeds. This difference became more pronounced during after-ripening: we observed a lower overall protein oxidation signal in dry atrbohB seeds compared with WT seeds that had after-ripened for the same amount of time. After 6 h imbibition, protein oxidation levels were reduced in WT seeds, presumably by proteolysis of the carbonylated proteins, while those of the mutants remained constant or even increased slightly. Thus, superoxide production by ATRBOHB could contribute to protein oxidation involved in seed after-ripening.
Insertion mutants in other Atrboh genes show a germination phenotype
In order to test if other Atrboh genes are involved in seed germination, we also tested germination of dSpm insertion mutants for nine other known Arabidopsis rbohs (Table 1). Indeed, while atrbohB was the only mutant in which germination in the presence of ABA was accelerated in fresh seeds, mutants carrying an insertion in AtrbohE, AtrbohH or AtrbohI germinated faster than WT in the fresh and after-ripened state on medium without ABA, and on medium with ABA in the after-ripened state. Interestingly, atrbohH showed a very strong sensitivity to ABA in the fresh state, when it germinated slower than any other mutant or the WT. AtrbohF was observed to germinate slower than the WT in all conditions tested, while atrbohD germinated more slowly only in the after-ripened state. No clear germination phenotype was detected for the remaining atrbohs in the conditions we used. We performed in-situ RNA hybridization to check for expression of the rbohs whose mutation led to a germination phenotype in after-ripened WT seeds (Table 1). There did not appear to be a connection between RNA localization and phenotype. The fact that the germination phenotypes differ between these mutants points to a variety of roles the Atrbohs might play during germination.
We examined whether alternative splicing of intron 1 in fresh and after-ripened seeds is evident for the Atrbohs for which we found germination phenotypes of the mutants (AtrbohA, D, E, F, I; Table 1). In this independent RT-PCR experiment with forward primers in exon 1 and reverse primers over the exon3/exon4 border, AtrbohB was also included as a positive control and, as expected, there was an amplification of AtrbohB-β (intron 1 retained) and AtrbohB-α (intron 1 spliced) as shown in Fig. 4. By contrast, under the same experimental conditions no PCR bands for AtrbohA, D, E, F and I were amplified with the predicted size of a transcript retaining intron 1 (β-type). The PCR bands with the expected size for the α-type transcripts (intron 1 spliced) of AtrbohA, D, E, F and I were amplified in after-ripened seeds in the dry and 24-h imbibed states without and with ABA added. The α-type transcripts, but no β-type transcripts, were detected for AtrbohD, E, F and I in 24 h-imbibed fresh seeds. α-Type AtrbohE transcripts (but no β-type) were also detected in dry fresh seeds. AtrbohE appeared to be most abundant, whereas AtrbohF and AtrbohI were of low abundance, and AtrbohA was not detected in fresh seeds. This demonstrates that the after-ripening status and ABA did not cause a general effect on intron 1 splicing, but that this effect is specific for AtrbohB.
After-ripened seeds of the tobacco mutant ntrbohD (Simon-Plas et al., 2002) germinated slower than the corresponding WT on media with and without ABA, just as the corresponding Arabidopsis mutant atrbohD does (data not shown). Thus, Rbohs might play a role in seed germination in other species besides Arabidopsis.
DPI delays Arabidopsis and cress endosperm rupture and cress endosperm weakening
In a pharmacological approach, we tested the effect of the NADPH-oxidase inhibitor DPI on seed germination. 100 μm DPI delayed the germination of after-ripened Arabidopsis WT seeds in the presence and absence of ABA, while atrbohB-mutant seeds showed a strongly reduced sensitivity to DPI (Table 2). We conclude that rbohB is a major target of DPI in Arabidopsis seed germination, although other Atrbohs may also contribute to the regulation of seed germination.
Table 2. Diphenylene iodonium chloride (DPI) delays endosperm rupture of Arabidopsis and cress and endosperm weakening of cress
Time and treatment
Endosperm rupture (%)
Endosperm resistance (mN)
ND, not determinable (seeds are too small). Endosperm rupture (%): means of three lots of 50 seeds ± SE are given, germination at 24°C (Arabidopsis) : 18°C (cress) in continuous light. Endosperm resistance: means of at least three lots of 25 seeds ± SE are given.
Arabidopsis thaliana, wild type after-ripened
30 h control
95 ± 1
30 h 100 μm DPI
51 ± 2
374 h 1 μm ABA
76 ± 2
374 h 1 μm ABA 100 μm DPI
22 ± 1
Arabidopsis thaliana, rbohB after-ripened
374 h 1 μm ABA
25 ± 1
374 h 1 μm ABA 100 μm DPI
16 ± 0
Lepidium sativum (cress), after-ripened
18 h control
78 ± 3
19.5 ± 2.6
18 h 150 μm DPI
19 ± 5
29.8 ± 2.6
96 h 10 μm ABA
75 ± 4
21.0 ± 1.3
96 h 10 μm ABA 150 μm DPI
12 ± 4
Some Atrbohs, such as AtrbohD (Penfield et al., 2006), are known to be expressed in the Arabidopsis endosperm layer. We therefore examined whether endosperm weakening is affected in DPI-treated seeds. As Arabidopsis seeds are too small to allow for biomechanical approaches, we used the closely related species cress whose seeds are larger than those of Arabidopsis but have a very similar seed anatomy with one to two cell layers of endosperm and similar germination behaviour (Müller et al., 2006). Treatment with DPI delayed the endosperm rupture of cress seeds just as that of Arabidopsis seeds, although a higher concentration (150 μm) had to be used. We measured a strong effect on endosperm weakening: at 18 h after imbibition, it took a force of 19.5 mn to puncture the endosperm cap covering the radicle in untreated seeds, but 29.8 mn in DPI-treated seeds (Table 2). We hypothesize that Atrbohs play roles both in the embryo and the endosperm of germinating Brassicaceae seeds.
Seed after-ripening is a process of great relevance both in wild species and crops (Baskin & Baskin, 1998; Cadman et al., 2006; Finch-Savage & Leubner-Metzger, 2006). While the first might rely on after-ripening to time their germination with changing environmental conditions, dealing with dormancy and after-ripening of the latter is an important aspect of seed management in agriculture and ex-situ seed banks. It is thus relevant to further our understanding of this important process and the genes and proteins involved.
By screening a set of Arabidopsis enhancer trap lines for after-ripening and germination phenotypes, we found that a decrease in AtrbohB-transcript in embryos of mature seeds leads to changes in after-ripening: mutant seeds do not after-ripen with respect to their sensitivity to ABA. There are two possible interpretations for this phenotype, and these are not mutually exclusive. (1) AtrbohB plays a role in ABA perception or signalling. The signal would in this case be either superoxide, produced by AtrbohB or the product of its dismutation, hydrogen peroxide. Hydrogen peroxide would be more likely to serve as messenger in a signalling cascade because of its comparatively long lifespan and ability to cross membranes (Hancock, 2001; Laloi et al., 2004). The associated sensors could either react directly to the hydrogen peroxide molecule or sense changes in the redox state of the cell. Other Atrbohs have already been shown to be involved in hormone signalling: AtrbohD and AtrbohF have been shown to be involved in ABA-signal transduction pathways in guard cells (Kwak et al., 2003), while AtrbohF is also involved in ethylene action on stomatal closure (Desikan et al., 2006). (2) AtrbohB promotes after-ripening by producing superoxide, which leads to post-translational modifications by carbonylation of proteins. After-ripening and loss of dormancy have been associated with an increase in reactive oxygen species (ROS; superoxide and hydrogen peroxide) production in sunflower axes (Oracz et al., 2007). This accumulation of ROS leads to targeted protein carbonylation during after-ripening as well as during artificial breaking of dormancy by the application of cyanide or methylviologen. Our Oxyblots showed that atrbohB mutant seeds indeed have a lower overall protein carbonylation after an equal period of after-ripening than the corresponding WT.
We found that the AtrbohB pre-mRNA is alternatively spliced in fresh and after-ripened seeds, with variant AtrbohB-ß retaining intron 1. This process is developmentally and hormonally regulated: it depends on the state of the seed (fresh/after-ripened) and is ABA-dependent. The only other published report of alternative splicing of a plant Rboh is, to our knowledge, ZmrbohB (highest sequence similarity in Arabidopsis with AtrbohF), which occurs in two splicing variants in different maize tissues (Lin et al., 2009). Interestingly, the expression pattern of the splicing variants could be changed by subjecting the plants to stress, which is known to lead to ABA production. One of the variants retains intron 11, which leads to a PTC, and the authors suggest that this leads to NMD. Just as the splicing variant of AtrbohB with a retained intron 9 we found on NCBI, the variant of ZmrbohB retaining intron 11 would miss its NADPH-binding domain. As the domains relevant to electron transport are still present, this protein might still display activity, if the mRNA is translated and not degraded by NMD mechanisms. Interestingly, the animal NADH-Oxidase NOX1 also displays alternative splicing with one variant leading to a protein without NADH binding sites (Geiszt et al., 2004). While intron retention is rare in animal systems (Kan et al., 2002; Nagasaki et al., 2005), it has been found to be the most common form of alternative splicing in Arabidopsis (Ner-Gaon et al., 2004; Nagasaki et al., 2005; Kertész et al., 2006).
Our discovery that AtrbohB alternative mRNA splicing is differentially regulated in fresh and after-ripened seeds is the first instance of alternative splice forms being differentially expressed in seeds of different after-ripening status. DOG1, a gene cloned for a major QTL of Arabidopsis seed dormancy (Bentsink et al., 2006), also occurs in various splice forms, but no relation to after-ripening is known. It seems that the ratios of DOG1 transcript variants in combination with the actual amounts of the different forms contribute to the regulation of dormancy, and these ratios change during seed maturation when dormancy is induced (Schwab, 2008). The changes in transcript levels of AtrbohB-α and AtrbohB-ß we observed in our semiquantitative PCR involve changes both to the ratio and possibly also to the total amount of AtrbohB transcripts.
We hypothesize that alternative splicing is an attractive mechanism for dormancy and after-ripening regulation in seeds: seeds are known to store mRNAs. A global analysis of stored mRNAs in seeds by Nakabayashi et al. (2005) identified over 12 000 mRNAs in dry seeds, among which those with ABA responsive elements (ABREs) in the 1 kb promoter region were over-represented. A scan for promoter motifs of AtrbohB with the place signal scan software (Higo et al., 1999) showed two ABRE motives. If seeds stored pre-mRNAs, those could be alternatively spliced depending on the developmental status and environmental conditions, when the seed encounters conditions allowing rehydration and germination. This would be a flexible and fast way of reacting to changes in the environment.
Several cases in which ABA plays a role in the regulation of alternative splicing have been published. For example, PIMT2, an l-isoaspartyl methyltransferase involved in repairing proteins damaged in stress conditions has two splicing variants whose expression is regulated by ABA (Xu et al., 2004). l-Isoaspartyl methyltransferases also play important roles in maintaining seed vigour and longevity during seed aging (Oge et al., 2008). Thus, in accordance with our hypothesis, alternative splicing is used by the plants to react quickly to changes in their environment, especially to stress in which ABA signalling is known to play a major role. This fits with the fact that ABREs are over-represented in promoters of genes whose RNA is stored in seeds.
We show that DPI delays germination of after-ripened Arabidopsis and cress seeds and endosperm weakening of cress. A possible interpretation of these effects is that superoxide, together with its dismutation product hydrogen peroxide, can react at apoplastic peroxidases (Chen & Schopfer, 1999) and/or transition metal ions in the cell wall (Fry et al., 2002) to form hydroxyl radicals in the Fenton reaction. These highly reactive radicals can in turn cleave cell wall polymers, leading to cell wall loosening, which is a prerequisite of endosperm weakening as well as radicle growth (Müller et al., 2009). As AtrbohB is expressed in the embryo and not in the endosperm, its superoxide production could act on radicle elongation, implying that a different Rboh is responsible for effects in the endosperm. The possibility has to be kept in mind that the delaying effect of DPI on germination has to be interpreted separately from the role of Rbohs in after-ripening, as DPI leads to a decrease in ROS production during the actual germination process as opposed to a decrease during after-ripening in the low-hydrated state caused by a mutated rboh.
The embryo of an imbibed seed reacts to ABA with a decrease in growth potential (Schopfer & Plachy, 1985). The endosperm, on the other hand, reacts to ABA with a delay in tissue weakening as was shown in cress (Müller et al., 2006) and tomato (Toorop et al., 2000). We could detect AtrbohB RNA only in the embryo, but not the endosperm both with the GUS-reporter in the enhancer trap lines and by in-situ RNA hybridization in the WT. One other Atrboh, namely AtrbohD, has been investigated, if briefly, in connection with seed germination. After-ripened seeds of the double mutant atrbohD/F have a lowered sensitivity to ABA in the germination medium (Kwak et al., 2003). Penfield et al. (2006) pinpointed this phenotype to being caused by AtrbohD only. Their transcriptome data showed that the AtrbohD transcript is much more abundant in the endosperm than in the embryo of seeds treated with ABA and in seeds just after radicle protrusion. Based on these data, they suggested a role for AtrbohD in ABA signalling in the Arabidopsis endosperm. Under our conditions, atrbohD seeds germinated at the same speed as the WT in the fresh state and more slowly when they were after-ripened, thus also displaying an after-ripening phenotype. It would be an interesting option if different Rbohs mediated ABA signalling in different seed tissues and under different environmental conditions.
We are indebted to Jonathan Jones for his cooperation on behalf of the atrboh-mutants. We would like to thank Lydia Buller for expert technical assistance and the German Research Foundation (DFG) and the Wissenschaftliche Gesellschaft Freiburg for funding (DFG LE720/6). M.A.T. received the BIO 2007-66806 Grant from the Plant Nacional de I + D (Spain) and the IRG RTD REG/T.2 (2007)D/562971 from the European Community. We are grateful to Kai Gräber and Peter Schopfer for comments and discussions. Seeds of the tobacco mutant ntrbohD were kindly provided by Francoise Simon-Plas (INRA, Dijon, France).