• Open Access

Integrated analysis of seed proteome and mRNA oxidation reveals distinct post-transcriptional features regulating dormancy in wheat (Triticum aestivum L.)

Authors


Summary

Wheat seeds can be released from a dormant state by after-ripening; however, the underlying molecular mechanisms are still mostly unknown. We previously identified transcriptional programmes involved in the regulation of after-ripening-mediated seed dormancy decay in wheat (Triticum aestivum L.). Here, we show that seed dormancy maintenance and its release by dry after-ripening in wheat is associated with oxidative modification of distinct seed-stored mRNAs that mainly correspond to oxidative phosphorylation, ribosome biogenesis, nutrient reservoir and α-amylase inhibitor activities, suggesting the significance of post-transcriptional repression of these biological processes in regulating seed dormancy. We further show that after-ripening induced seed dormancy release in wheat is mediated by differential expression of specific proteins in both dry and hydrated states, including those involved in proteolysis, cellular signalling, translation and energy metabolism. Among the genes corresponding to these proteins, the expression of those encoding α-amylase/trypsin inhibitor and starch synthase appears to be regulated by mRNA oxidation. Co-expression analysis of the probesets differentially expressed and oxidized during dry after-ripening along with those corresponding to proteins differentially regulated between dormant and after-ripened seeds produced three co-expressed gene clusters containing more candidate genes potentially involved in the regulation of seed dormancy in wheat. Two of the three clusters are enriched with elements that are either abscisic acid (ABA) responsive or recognized by ABA-regulated transcription factors, indicating the association between wheat seed dormancy and ABA sensitivity.

Introduction

Dormancy is an adaptive trait that confers seeds a mechanism to remain quiescent until conditions are optimal for germination (Finkelstein et al., 2008). Modern cereal cultivars, however, are characterized by reduced seed dormancy due to selection by breeders for rapid and uniform germination and seedling establishment (Simpson, 1990). As a result, they are susceptible to preharvest sprouting (PHS), which causes substantial loss in grain yield and quality (Gubler et al., 2005), especially in years when wet conditions occur along with harvest maturity. Given that wheat (Triticum aestivum L.) is one of the most important crops worldwide, incorporation of a certain degree of dormancy into commercial cultivars is essential to mitigate the economic losses associated with sprouting damage. However, this requires detailed dissection of the molecular mechanisms underlying dormancy in wheat seeds.

During the late maturation phase, seeds of many plant species acquire primary dormancy, and its decay is regulated by a number of endogenous and exogenous cues that exert both synergistic and competing effects (Finkelstein et al., 2008). Seed dormancy decay by after-ripening, a period of dry seed storage, has been demonstrated in many plant species (Iglesias-Fernandez et al., 2011); however, the underlying molecular features are still not well known. Thermodynamic analyses of water sorption isotherms have shown that seed dormancy release by dry after-ripening is associated with changes in molecular mobility and seed water status (Bazin et al., 2011a). It has been proposed by the same authors that the change in seed water status is triggered by nonenzymatic oxidative processes and can induce active metabolic reactions including gene transcription. Consistently, changes in transcript abundance appear to occur during dry after-ripening of seeds in various species (Bazin et al., 2011b; Bove, 2005; Gao et al., 2012; Liu et al., 2013), although de novo transcription is not indispensable for seed germination (Rajjou et al., 2004). After-ripening also induces imbibition-mediated transcriptional changes related to various biological processes. For example, transcripts of genes involved in cell organization and biogenesis, proteolysis and those associated with the protein synthesis machinery are over-represented in imbibing after-ripened (AR) seeds of Nicotiana plumbaginifolia (Bove, 2005), Arabidopsis (Arabidopsis thaliana, Cadman et al., 2006) and wheat (Gao et al., 2012), whereas the expressions of genes involved in abscisic acid (ABA) biosynthesis, gibberellin (GA) catabolism and stress response are enriched in the corresponding dormant (D) seeds. Furthermore, the transcription of genes associated with the protein synthesis machinery was shown to be repressed in imbibing D seeds. These results highlight the significance of transcriptional changes in controlling de novo protein synthesis, which is essential for seed germination (Rajjou et al., 2004).

Gene expression is also regulated by a variety of post-transcriptional mechanisms, and pictures about the role of such mechanisms in regulating seed dormancy are emerging (Bazin et al., 2011b). Oxygen permeates through a glassy matrix such as vitreous cytoplasm (Andersen et al., 2000), indicating its ability to diffuse into dry seeds during storage and thereby acts as a major source of reactive oxygen species (ROS). This is because respiratory and enzymatic activities are repressed when the seeds are in the dry state. In agreement with this, after-ripening-induced seed dormancy release in sunflower (Helianthus annuus) is associated with the accumulation of ROS (Oracz et al., 2007). Apart from lipids and proteins, ROS are able to oxidize nucleic acids; and RNA molecules are more susceptible to oxidative damage than DNA (Kong and Lin, 2010). It has been found by Bazin et al. (2011b) that mRNAs are more sensitive to after-ripening-induced oxidation than other RNA species in seeds. Oxidized mRNAs mainly derive from the oxidation of base guanine that produces 8-oxo-7, 8-dihydroguanine (8-OHG). The 8-OHG serves as a marker for changes in the degree of mRNA oxidations upon detection with anti-8-OHG antibodies. Seed dormancy release by after-ripening has recently been shown to be tightly associated with the oxidation of specific seed-stored mRNAs, which always leads to decreased protein synthesis (Bazin et al., 2011b). As the anatomical, physiological and metabolic bases of dormancy and germination are distinct between dicot and monocot species, comparative analysis of after-ripening-induced oxidation of seed-stored transcripts between the two species provides valuable insights into unique and conserved post-transcriptional mechanisms regulating dormancy release. To date, however, after-ripening-induced oxidation of seed-stored mRNAs has not been studied in any monocot species.

Imbibed D and AR seeds exhibit similar capacity but distinct pattern of protein synthesis (Chibani et al., 2006), highlighting the importance for differential regulation of translational activity. Accordingly, proteomic studies have identified specific proteins involved in the regulation of seed dormancy in different species. For example, comparative proteomic analysis between dry D and AR seeds of Arabidopsis indicated that after-ripening is associated with differential regulation of seed storage proteins, such as cruciferin precursors, and those related to energy and protein metabolism (Chibani et al., 2006). Other studies with seeds of tree species have also indicated the importance of specific proteins related to signal initiation and transduction, transcription, translation, energy metabolism and cell cycle in regulating seed dormancy (Pawłowski, 2007; Pawłowski, 2009; Pawłowski, 2010). These results demonstrated the importance of selective but not global de novo synthesis of seed proteins in controlling dormancy.

Furthermore, using a candidate protein approach, Schoonheim et al. (2007) have shown that 14-3-3 proteins, which act as regulators of signalling networks (van Hemert et al., 2001), mediate ABA action through their interaction with ABI5 and ABF, important mediators of ABA signalling in seeds (Nambara et al., 2010). This result implicates a role for 14-3-3s in regulating seed dormancy. The germination of AR or nondormant (ND) seeds is characterized by imbibition of dry seeds and subsequent protrusion of the expanding embryo axis through seed covering layers (Finch-Savage and Leubner-Metzger, 2006). Proteomic studies in Arabidopsis (Gallardo et al., 2002), barley (Bønsager et al., 2007), wheat (Mak et al., 2009) and rice (Oryza sativa L., He et al., 2011) have provided valuable insights into the mechanisms underlying the transition of seeds from the state of metabolic quiescence induced by desiccation to its activation during imbibition. However, the specific proteins that underlie after-ripening-induced translational changes during imbibition of wheat seeds are not well unknown.

To gain insights into the gap of knowledge on the molecular features of seed dormancy in wheat, we previously examined transcriptional changes triggered by after-ripening. To extend our understanding of the mechanisms acting at the post-transcriptional levels, this study investigated after-ripening-induced oxidative modification of seed-stored mRNAs in the dry state and changes in seed proteome between AR and D wheat seeds in both dry and imbibed states. Co-expression analysis of the genes differentially expressed (Gao et al., 2012) and oxidized during dry after-ripening, and those representing proteins differentially regulated between D and AR seeds in both dry and imbibed states was performed to identify more candidate genes and transcriptional regulatory mechanisms involved in controlling dormancy in wheat seeds.

Results and discussion

Germination of dormant vs. after-ripened seeds

Comparison between the D and AR seeds of wheat cv. AC Domain for their germination performance revealed that 95% of the AR seeds germinated following 1-day imbibition, when the corresponding D seeds did not show any sign of germination (Figure 1). Only approximately 15% of the D seeds germinated after 5-day imbibition.

Figure 1.

Effect of after-ripening on germination of dormant and after-ripened seeds of wheat cv. AC Domain imbibed in water. Seeds were considered germinated when the coleorhiza was visible beyond the seed coat. Germination was scored daily for 5 day. Data are means ± SD, n = 3, where n refers to a batch of 25 seeds.

Oxidative modification of seed-stored mRNAs during after-ripening

To gain insights into post-transcriptional mechanisms regulating seed dormancy in wheat, we investigated oxidation of seed-stored mRNAs in response to dry after-ripening. Our analysis identified 120 differentially oxidized probesets, which refers to the 8-OHG fraction, between the dry D and AR seeds (at cut-off values of twofold change and ≤ 0.05) in which 80 probesets were highly oxidized in dry AR and the remaining 40 probesets in the corresponding D seeds (Table S1). The presence of oxidized mRNAs in D seeds suggests the initiation of mRNA oxidation during seed maturation. Our results confirm that mRNA oxidation during dry after-ripening is highly selective and is associated with seed dormancy release (Bazin et al., 2011b). These authors have proposed that the selective nature of this oxidative process could be related to the mRNAs' location relative to the site of ROS production in the cell or their functional state within the messenger ribonucleoproteins, as they found no association between oxidized mRNAs and their guanine content or abundance.

The results of our study along with that of Bazin et al. (2011b), who used sunflower seeds as their experimental material, suggest cross-species evolutionary conservation of mRNA oxidation as a post-transcriptional seed dormancy–regulating mechanism. Following annotation by Arabidopsis genes, at least 20% of probesets oxidized in our dry AR whole seed samples are found to be expressed in imbibing wheat embryos (Bassel et al., 2011). However, comparison of mRNAs highly oxidized in our system with those oxidized in dry AR sunflower embryos (Bazin et al., 2011b) identified no wheat–sunflower orthologues, implying that the mRNAs targeted for after-ripening-induced oxidation are distinct between species.

Gene ontology of probesets corresponding to oxidized seed-stored transcripts

As mRNA oxidation always leads to decreased protein synthesis (Bazin et al., 2011b), our results suggest the significance of biological processes represented by the oxidized probesets in regulating seed dormancy. Probesets corresponding to mRNAs highly oxidized in AR seeds are over-represented in nutrient reservoir activity (GO: 0045735, = 1.8e-56; Table S1) and α-amylase inhibitor activity (GO: 0015066, = 8.3e-20; Table S1). Thirty-seven of the nutrient reservoir activity–related probesets represent genes encoding gliadin and glutenin, which account for approximately 80% of seed storage proteins in wheat (Payne et al., 1982). These proteins accumulate during the late stages of seed development (Bewley and Black, 1994) and are proteolytically degraded during germination to serve as a source of amino acids and nitrogen (Spencer, 1984). Thus, it is likely that oxidative modification of the related mRNAs forms a mechanism by which their synthesis is repressed during imbibition. Consistently, probesets over-represented in nutrient reservoir activity exhibited down-regulation during imbibition (Gao et al., 2012). Nine of the probesets highly oxidized in AR seeds are enriched in α-amylase inhibitor activity, suggesting the significance of enhanced hydrolytic starch degradation in regulating dormancy release. In agreement with this hypothesis, after-ripening led to imbibition induced up- and down-regulation of probesets corresponding to α-amylase and its inhibitors, respectively (Gao et al., 2012; Potokina et al., 2002). Furthermore, transcripts of peroxidases and ribosomal protein are also oxidized during after-ripening (Table S1), implying that post-transcriptional control of ROS production and synthesis of metabolically active proteins plays an important role in regulating seed dormancy in wheat.

Probesets highly oxidized in the D seeds are over-represented in oxidative phosphorylation (GO: 0006119, = 6.2e-09; Table S1) and ribosome biogenesis (GO: 0042254, = 1.4e-06; Table S1). Oxidative phosphorylation plays a major role in the synthesis of ATP, which is used as a source of energy for metabolic processes, during seed germination (Moroashi and Sugimoto, 1988). Our data therefore suggest the significance of post-transcriptional repression of this biological process in regulating the maintenance of seed dormancy. The oxidation of ribosome biogenesis–related mRNAs in dry D seeds is consistent with a recent report by Zhang et al. (2012), who showed the association of inhibition of seed germination with transcriptional repression of rRNA that is involved in the biogenesis of protein-synthesizing ribosomes.

Proteins differentially expressed during dry after-ripening

Our proteomic analysis revealed after-ripening mediated differential abundance of specific storage proteins, proteases/enzyme inhibitors and those involved in cellular signalling (Table 1; Figure S1). After-ripening induced a decrease in the abundance of storage protein triticin (7.0-fold, ≤ 0.05; Table 1), leading to dormancy release. In contrast, the level of another storage protein designated as 27K exhibited an increase (over twofold, ≤ 0.05), although such proteins mainly accumulate in the later stages of seed development (Bewley and Black, 1994).

Table 1. Proteins differentially expressed between dry after-ripened and dry dormant seeds
SpotaIdentificationID numberbMS/MS MASCOT peptides (score)cAR-0/D-0d (P-valuee)
  1. a

    Protein spot names correspond to the 2D gels in Figure S1.

  2. b

    GenBank IDs of the matching proteins. The ID names starting with ‘CL’ are from the local wheat EST database.

  3. c

    Amino acid sequences of the top two peptides matching the MS/MS spectra. Ions score is −10*Log (P), where P is the probability that the observed match is a random event. Individual ions scores >50 indicate identity or extensive homology (P < 0.05).

  4. d

    Normalized protein spot volumes in dry after-ripened (AR-0) seeds divided by the corresponding normalized spot volume in dry dormant (D-0) seeds. Data are means of 2–3 independent biological replicates.

  5. e

    Significant differences between samples were determined using Student's t-test at P-value of ≤0.05.

  6. f

    Protein spot identified as aspartate aminotransferase with less confidence as the matching peptides exhibited relatively less score (<50).

a1Triticin ACB41345.1

R.LLAEALGTSGK.I (90)

R.CTGVFAIR.R (59)

0.14 (0.020)
a2Unnamed protein product cystatin CAI84619.1 R.NEQAGIVGHKDVTDVTGDR.G (64)0.33 (0.036)
a3α-amylase/trypsin inhibitor CM16 CAA35596.1 K.SRPDQSGLMELPGCPR.E + Oxidation (M) (64)0.20 (0.014)
a4Manganese superoxide dismutase AAB68036.1

K.ALEQLDAAVSK.G (81)

K.NLKPISEGGGEPPHGK.L (70)

0.36 (0.009)
a5Hv14-3-3b CAA63658.1

K.AAQEIALAELPPTHPIR.L (77)

K.TVDSEELTVEER.N (73)

0.05 (0.000)
a6Hv14-3-3b CAA63658.1

K.TVDSEELTVEER.N (98)

R.IISSIEQKEESR.G (77)

0.09 (0.003)
a714-3-3 protein homologue CAA44259.1

K.QAFDEAIAELDSLGEESYK.D (94)

K.SAQDIALADLPTTHPIR.L (93)

0.21 (0.006)
a8fAspartate aminotransferase CV777781

K.EYLPITGLADFNKLSAK.L (39)

R.LEGGVGGGGGGR.V (26)

0.06 (0.017)
a9No good hitCL1Contig1401-.VATVSIPR.T (65)0.11 (0.040)
b127K protein BAC76688.1 R.GHNLSLEYGR.Q (55)2.31 (0.046)

Protease inhibitor cystatin decreased in abundance during after-ripening (3.0-fold, ≤ 0.05; Table 1). Cystatin is involved in controlling the activity of gliadain, a cysteine protease associated with wheat seed germination (Kiyosaki et al., 2007). Thus, it is likely that after-ripening renders a mechanism to degrade cystatin and in turn activate gliadain-mediated proteolysis. Consistently, repression of cystatin in barley aleurone treated with the germination promoting GA is associated with enhanced degradation of storage proteins (Martinez et al., 2009). The level of trypsin/α-amylase inhibitor designated as CM16, because of its solubility in chloroform/methanol, was also reduced by after-ripening (5.0-fold, ≤ 0.05). These results, overall, reflect the association of enhanced proteolysis and hydrolysis of storage reserves with seed dormancy release.

After-ripening also decreased the abundance of proteins related to cellular signalling, including 14-3-3 proteins and superoxide dismutase (SOD) (over twofold, ≤ 0.05; Table 1). The 14-3-3s are involved in the activation of seed ABA response (Schoonheim et al., 2007); the decrease in their level likely leads to loss of seed ABA sensitivity and dormancy. Whereas the reduction in the level of SOD, one of the antioxidative enzymes, might form a mechanism to maintain cellular homoeostasis of ROS that acts as a signal in governing embryo transition from a D to ND status (Oracz et al., 2007, 2009).

Regulation of proteins differentially expressed during after-ripening

After-ripening is associated with accumulation of ROS derived from nonenzymatic reactions (Oracz et al., 2007), and oxidation of storage proteins occurs in dry seeds (Job et al., 2005). This along with the unlikely occurrence of transcriptional and translational activities in dry D seeds (Bazin et al., 2011a) might suggest that the decreased abundance of triticin, cystatin, SOD and 14-3-3 during after-ripening (Table 1) results from their oxidative degradation. No differential expression and oxidation of probesets annotated as cystatin, SOD and 14-3-3 was evident in response to after-ripening; however, the transcript abundance of specific triticin probesets was decreased (over threefold, ≤ 0.05; Figure 2), implying the decay of triticin mRNAs. The reduction in the level of CM16 during after-ripening (Table 1) is accompanied by oxidation of its corresponding probesets (Table S1), and this might suggest its translational repression initiated possibly by the change in seed water status due to nonenzymatic oxidative reactions (Bazin et al., 2011a,b). The accumulation of 27K during after-ripening (Table 1) with no apparent differential expression of its probesets between dry AR and D seeds (Figure 2) can be explained by its enhanced solubility and/or extractability due to oxidative processes that increase protein–water interactions (Cherian and Chinachoti, 1997).

Figure 2.

Fold change in the expression of probesets corresponding to proteins differentially expressed between dry after-ripened (AR-0) and dry dormant (D-0) seeds (AR-0/D-0). Asterisks indicate significant difference in expression (≥ twofold change and ≤ 0.05). Abbreviations: Tri, triticin; Cys, cystatin; CM16, α-amylase/trypsin inhibitor; SOD, superoxide dismutase; 14-3-3, 14-3-3 proteins; AST, aspartate aminotransferase; Uk, unknown (no good hit); 27K, 27K protein. Fold changes in the expression of these probesets between D and AR seeds in both dry and imbibed states are shown in Table S2.

Proteins differentially regulated by imbibition in dormant and after-ripened seeds

To identify proteins that possibly regulate dormancy during imbibition, changes in protein abundance were compared between dry and imbibed seeds in each of the D and AR samples, and also between imbibed D and imbibed AR seeds. Our comparative study was focused on 24-h imbibed seeds, as changes in the proteome of cereal seeds appear to occur mainly at/after this stage (Bønsager et al., 2007; Yang et al., 2007). Furthermore, 95% of the AR but none of the D seeds completed germination at 24 h after imbibition, reflecting that differences in seed proteome at this stage arose from the phase of germination sensu stricto.

Proteins differentially expressed by imbibition specifically in dormant seeds

Imbibition down-regulated 14-3-3, calreticulin (CRT)-like and serpin proteins specifically in D seeds (Table 2, Figure S2a,b). The abundance of a 14-3-3 homologue protein decreased by after-ripening (over fourfold, ≤ 0.05; Table 1) and this level was maintained during imbibition (Figure S2c,d). In D seeds, however, this protein was down-regulated only after imbibition (over fivefold, ≤ 0.05; Table 2), reinforcing our hypothesis that 14-3-3 protein plays important role in regulating wheat seed dormancy. The level of CRT-like proteins decreased to undetectable level during imbibition in D but not in AR seeds (Table 2). Calreticulin is a major Ca2+ sequestering protein implicated in regulating seed ABA response and dormancy (Lorenzo et al., 2002; Pawłowski, 2007). Our data, thus, imply that maintenance of seed ABA sensitivity and dormancy in wheat requires CRT degradation. Consistent with this hypothesis, treatment of seeds with ABA decreases the level of CRT (Pawłowski, 2007).

Table 2. Proteins differentially expressed during imbibition specifically in dormant seeds, and their comparison with those in imbibed after-ripened seeds
SpotaIdentificationID numberbMS/MS MASCOT Amino acids sequence (score)cD-24/D-0d (P-valuee)D-24/AR-24f (P-value)
  1. a

    Protein spot names correspond to the 2D gels in Figure S2a,b.

  2. b

    GenBank IDs of the matching proteins. The ID names starting with ‘CL’ are from the local wheat EST database.

  3. c

    Amino acid sequences of the top two peptides matching the MS/MS spectra. Ions score is −10*Log (P), where P is the probability that the observed match is a random event. Individual ions scores >50 indicate identity or extensive homology (P < 0.05).

  4. d

    Normalized protein spot volumes in imbibed dormant (D-24) seeds divided by normalized spot volumes in the corresponding dry (D-0) seeds. Data are means of 2–3 independent biological replicates; ≤0.01 indicates that the abundance of the corresponding protein in D-24 seeds was close to background.

  5. e

    Significant differences between samples were determined using Student's t-test at P ≤ 0.05; NA = not available.

  6. f

    Normalized protein spot volumes in D-24 seeds divided by normalized spot volumes in the corresponding after-ripened (AR-24) seeds. Data are means of 2–3 independent biological replicates; ≤0.01 indicates that the abundance of the corresponding protein in D-24 seeds was close to background.

d1Serpin CAB52709.1

K.GAWTDQFDSSGTK.N (98)

R.VSSVFHQAFVEVNEQGTEAAASTAIK.M (96)

≤0.01 (NA)≤0.01 (NA)
d214-3-3 protein homologue CAA44259.1

K.QAFDEAIAELDSLGEESYK.D (94)

K.SAQDIALADLPTTHPIR.L (93)

0.19 (0.03)0.77 (0.019)
d3Calreticulin-like protein AAW02798.1

K.SGTLFDNILITDDAALAK.T (112)

R.FYAISAEYPEFSNK.D (81)

≤0.01 (NA)≤0.01 (NA)
d4Calreticulin-like protein AAW02798.1

R.FYAISAEYPEFSNK.D (105)

K.SGTLFDNILITDDAALAK.T (93)

≤0.01 (NA)≤0.01 (NA)
e1Serpin 3 ACN59485.1

R.VSSVFHQAFVEVNEQGTEAAASTAIK.M (113)

K.YKAETQSVDFQTK.A (96)

2.39 (0.001)2.86 (0.035)

While the abundance of a specific serpin in D seeds was repressed to undetectable level during imbibition, the level of another one (serpin 3) was up-regulated (over twofold, ≤ 0.05), leading to a nearly threefold more accumulation than that found in AR seeds (≤ 0.05; Table 2). Serpins, a super family of versatile proteins, take part in the regulation of complex proteolytic systems, and those in cereal seeds are likely to inhibit endogenous and exogenous proteases irreversibly (Østergaard et al., 2000). Our data therefore may suggest decreased proteolysis, which might attribute at least partially to the failure in germination of D seeds.

Regulations of proteins differentially expressed by imbibition specifically in dormant seeds

The 14-3-3 protein appeared to be regulated post-transcriptionally as its abundance decreased during imbibition (Table 2) while expression of its probesets either increased or remained unaffected (Figure 3a). Expression of all probesets annotated as CRT remained unaffected during imbibition of D seeds; however, three specific probesets exhibited over twofold down-regulation relative to AR seeds (≤ 0.05; Figure 3a,b). Given that imbibition substantially reduced CRT level in D but not in AR seeds (Table 2), our data imply its transcriptional regulation. Both transcriptional and post-transcriptional mechanisms likely regulate serpins as the two serpins showed opposite pattern in abundance (Table 2), while expression of their probesets either decreased or did not change during imbibition of D seeds or in imbibed D relative to AR samples (Figure 3a,b). Probesets corresponding to all proteins down-regulated by imbibition specifically in D seeds are not among those oxidized in dry D seeds, suggesting that their regulation during imbibition is independent of seed maturation–associated mRNA oxidation.

Figure 3.

Fold change in the expression of probesets corresponding to proteins differentially expressed during imbibition specifically in dormant (D) seeds (D-24/D-0; a), and their respective comparison between imbibed D and the corresponding after-ripened (AR) seeds (D-24/AR-24; b). Asterisks indicate significant difference in expression (≥ twofold change and ≤ 0.05). Fold changes in the expression of these probesets between D and AR seeds in both dry and imbibed states are shown in Table S2.

Proteins differentially expressed by imbibition specifically in after-ripened seeds

Imbibition differentially regulated specific seed storage proteins and protease inhibitors and those involved in energy metabolism, translation, protein folding and amino acid metabolism specifically in AR seeds (Table 3, Figure S2c,d).

Table 3. Proteins differentially expressed during imbibition specifically in after-ripened seeds, and their comparison with those in imbibed dormant seeds
SpotaIdentificationID numberbMS/MS MASCOT Amino acids sequence (score)cAR-24/AR-0d (P-valuee)AR-24/D-24f (P-value)
  1. a

    Protein spot names correspond to the 2D gels in Figure S2c, d.

  2. b

    GenBank IDs of the matching proteins. The ID names starting with ‘CL’ are from the local wheat EST database.

  3. c

    Amino acid sequences of the top two peptides matching the MS/MS spectra. Ions score is −10*Log (P), where P is the probability that the observed match is a random event. Individual ions scores >50 indicate identity or extensive homology (P < 0.05).

  4. d

    Normalized protein spot volumes in imbibed after-ripened (AR-24) seeds divided by normalized spot volumes in the corresponding dry (AR-0) seeds. Data are means of 2–3 independent biological replicates; ≤0.01 indicates that the abundance of the corresponding protein in AR-24 seeds was close to background.

  5. e

    Significant differences between samples was determined using Student's t-test at P ≤ 0.05; NA = not available.

  6. f

    Normalized protein spot volumes in AR-24 seeds divided by normalized spot volumes in the corresponding dormant (D-24) seeds. Data are means of 2–3 independent biological replicates; ≤0.01 indicates that the abundance of the corresponding protein in AR-24 seeds was close to background.

  7. g

    Inconsistent spots between biological replicates of imbibed D seeds. No clear comparison of glucose and ribitol dehydrogenase accumulation between imbibed AR and D samples.

f1Triticin ACB41345.1

R.LLAEALGTSGK.I (93)

R.SSQLHSSQNIFSGFDVR.L (59)

0.49 (0.016)0.66 (0.010)
f2Globulin 3 (cupin domain contained)CL1Contig7398 

R.QASEGGQGHHWPLPPFR.G (83)

R.DTFNLLEQRPK.I (73)

≤0.01 (NA)≤0.01 (NA)
f3Serpin CAA72274.1

K.AFVEVNETGTEAAATTIAK.V (126)

K.AAEVTAQVNSWVEK.V (97)

0.42 (0.007)0.63 (0.025)
f4Serpin CAA72274.1

K.DILPAGSIDNTTR.L (66)

K.GAWTDQFDPR.A (56)

≤0.01 (NA)≤0.01 (NA)
f5Serpin 2 ACN59484.1

R.VAFANGVFVDASLSLKPSFQELAVCNYK.S (112)

K.GLWTEKFDESK.T (87)

0.38 (0.018)0.42 (0.044)
f6Serpin 2 ACN59484.1

K.ISFGFEATNLLK.S (96)

K.VVVDQFMLPK.F + Oxidation (M) (82)

0.18 (0.037)0.12 (0.033)
f7Glucose and ribitol dehydrogenase homologT06212

K.GNATLLDYTATK.G (85)

K.VALVTGGDSGIGR.A (82)

0.19 (0.040)0.74 (0.638)g
f8Granular bound starch synthase I BAA88511.1

K.GPDVMIAAIPEIVK.E + Oxidation (M) (88)

K.EEDVQIVLLGTGK.K (81)

0.42 (0.000)0.56 (0.032)
f9Putative eukaryotic translation initiation factor 6 BAC45212.1

K.ATEELIADVLGVEVFR.Q (80)

R.IQFENNCEVGVFSK.L (63)

≤0.01 (NA)≤0.01 (NA)
f10Eukaryotic translation initiation factor 5A1 AAZ95171.1

K.LPTDDVLLGQIK.T (94)

K.TYPQQAGAIR.K (81)

0.05 (0.000)0.16 (0.019)
f11Protein disulfide isomerase 2 precursor AAK49424.1

K.AYYGAVEEFSGK.D (85)

R.KSEPIPEANNEPVK.V (64)

0.41 (0.025)0.37 (0.014)
f12Protein disulfide isomerase 3 precursor AAK49425.1

K.APEDATYLEDGK.I (97)

K.LAPILDEAAATLQSEEDVVIAK.M (70)

0.44 (0.039)0.44 (0.033)
f13Protein disulfide isomerase 3 precursor AAK49425.1

R.TADEIVDYIK.K (82)

K.VVVADNVHDVVFK.S (82)

0.32 (0.045)0.40 (0.044)
f14Unnamed protein product CAA34060.1

K.TAIAIDTILNQK.Q (101)

R.AAELTTLLESR.M (96)

0.34 (0.007)0.59 (0.017)
g1N-acetyl-gamma-glutamyl-phosphate reductase

CL187Contig2

CL187Contig3

K.IVDLSADFR.L (69)1.51 (0.046)1.82 (0.031)
g2Globulin 2CL1Contig7433

R.AQPESVFVAGPQQQR.R (94)

K.ALAFPQQAR.E (60)

2.10 (0.026)1.37 (0.048)

Seed storage proteins and protease inhibitors

Storage protein globulin 3 decreased to undetectable level during imbibition of AR seeds, but remained unaffected in D seeds (Table 3), implying the role of after-ripening in inducing imbibition-mediated proteolysis of this protein and in turn dormancy decay. Imbibition of AR seeds also decreased the abundance of storage protein triticin (twofold, ≤ 0.05) but increased that of globulin 2 (twofold, ≤ 0.05). The up-regulation of globulin 2 in imbibing AR seeds may imply the significance of re-induction of specific seed maturation programmes in enhancing seed vigour (Rajjou et al., 2008), leading to dormancy decay.

Two serpin proteins decreased in abundance during imbibition specifically in AR seeds (over twofold, ≤ 0.05), leading to their significantly lower abundance in imbibed AR than in the corresponding D seeds (Table 3). These results may imply the role of after-ripening in repressing the action of serpins in imbibing seeds and thereby activate proteolytic degradation of storage proteins and dormancy release.

Proteins involved in energy metabolism

Granule-bound starch synthase I (GBSSI) was down-regulated during imbibition specifically in AR seeds (2.4-fold, ≤ 0.05), leading to a significant reduction in its accumulation relative to that found in the corresponding D seeds (nearly twofold, ≤ 0.05; Table 3). These results suggest that after-ripening induces imbibition-mediated suppression of starch synthesis, and thereby dormancy release. Imbibition of AR seeds also led to substantial repression of glucose/ribitol dehydrogenase (GRD; over fivefold, ≤ 0.05). Consistently, GRD was down-regulated in imbibing D seeds of Trollius ledebouri treated with GA, which stimulates dormancy breaking and germination in this species, but not in the untreated seeds (Bailey et al., 1996), implying the association of after-ripening induced suppression of GRD with dormancy decay.

Proteins involved in translation, protein folding and amino acid metabolism

While maintained at a similar level in D seeds, two eukaryotic translation initiation factors (eIFs), eIF6 and eIF5A1, decreased in their abundance during imbibition of AR seeds (Table 3). The eIF6 protein was virtually undetectable in imbibed AR seeds, while the level of eIF5A1 showed substantial reduction (over 20-fold, ≤ 0.05), leading to a significantly lower abundance when compared with that found in imbibed D seeds (over sixfold, ≤ 0.05). The eIF6 acts as a ribosome dissociation factor because it can bind to the 60S ribosomal subunit and, in turn, prevent the association between the 60S and 40S subunits (Russell and Spremulli, 1980). Our data, thus, imply the role of after-ripening in inducing imbibition-mediated activation of specific translational programmes, and thereby dormancy release.

Imbibition of AR seeds also caused suppression of two precursor forms of protein disulfide isomerase (PDI), PDI2 and PDI3, that acts to catalyse protein folding (Ellgaard and Ruddock, 2005), leading to a lower accumulation than that found in D seeds (over twofold, ≤ 0.05; Table 3). As increased abundance of PDIs is associated with the accumulation of seed storage proteins (Xia and Kermode, 1999), our data might suggest the significance of after-ripening in repressing PDIs, which promotes proteolysis, and in turn seed dormancy release and germination.

N-Acetyl-gamma-glutamyl-phosphate reductase (AGPR), which is involved in the synthesis of proline, arginine and glutamine (Kishor et al., 2005), was slightly up-regulated during the imbibition of AR seeds, leading to a nearly twofold more accumulation than that detected in D seeds (≤ 0.05; Table 3). This result suggests the significance of seed produced amino acids in dormancy release. Indeed, antisense repression of a proline biosynthetic gene, Δ1-pyrroline-5-carboxylate synthetase, in Arabidopsis caused a delay in seed germination (Hare et al., 2003).

Regulation of proteins differentially expressed by imbibition specifically in AR seeds

Seed dormancy release by dry after-ripening is associated with targeted mRNA oxidation, which could lead to translational repression of the corresponding proteins during imbibition and thereby regulate the germination process (Bazin et al., 2011b). However, probesets of proteins down-regulated by imbibition specifically in AR seeds were not oxidized during dry after-ripening, except a specific probeset annotated as GBSSI (Table S1). It is likely that most of the proteins annotated by oxidized probesets are expressed at a level below the detection limit of our 2D gel system.

Although their proteins showed increased or decreased abundance during imbibition of AR seeds or in imbibed AR relative to D seeds (Table 3), probesets corresponding to globulin 2, globulin 3, eIF5A1, eIF6, PDIs and AGPR exhibited either no differential expression or a pattern opposite to that shown by their respective protein (Figures 4a,c,d; S3a,c,d). These data imply post-transcriptional regulation of these proteins by mechanisms other than dry after-ripening-induced mRNA oxidation. Consistent with their protein expression pattern, however, either all or specific probesets of triticin, serpin, GBSSI and GRD were repressed during imbibition of AR seeds (over twofold, ≤ 0.05; Figure 4a,b). Furthermore, specific serpin and GBSSI probesets exhibited suppression in imbibed AR relative to D seeds (over twofold, ≤ 0.05; Figure S3a,b), suggesting their transcriptional regulation. The decreased abundance of GBSSI in imbibing AR seeds (Table 3) can also be attributed to the oxidation of one of its probesets during dry after-ripening.

Figure 4.

Fold change in the expression of probesets corresponding to the proteins differentially expressed during imbibition specifically in after-ripened (AR) seeds (AR-24/AR-0; a-d). Asterisks indicate significant difference in expression (≥ twofold change and ≤ 0.05). Fold changes in the expression of these probesets between dormant and AR seeds in both dry and imbibed states are shown in Table S2. Abbreviations: GRD, glucose/ribitol dehydrogenase; GBSSI, granule-bound starch synthase I; eIF6, eukaryotic translation initiation factor 6; eIF5A1, translation initiation factors 5A1; PDI, protein disulfide isomerase; AGPR, N-acetyl-gamma-glutamyl-phosphate reductase.

Previous proteomic analyses of rice and wheat embryos and different tissues of barley seeds have shown that most storage proteins and proteases/enzyme inhibitors are mainly localized in the endosperm/aleurone tissues, whereas those involved in translation and protein folding such as eIF5A1 and PDIs, and cellular signalling such as 14-3-3s and CRT, in the embryos (Bønsager et al., 2007; Kim et al., 2009; Mak et al., 2009). These results suggest the tissue specificity of the related proteins identified in this study, which involved whole seed samples.

Gene co-expression analysis

To identify more candidate genes involved in the regulation of seed dormancy in wheat, probesets that were found to be differentially expressed (58 probesets; Gao et al., 2012) and oxidized (120 probesets; Table S1) during dry after-ripening and those representing proteins differentially expressed between dry D and AR seed samples (20 unique probesets; Table S2) were subjected to gene co-expression analysis using Web tools in WheatNet (http://aranet.mpimp-golm.mpg.de/wheatnet). This analysis revealed that probesets down-regulated or oxidized during dry after-ripening are enriched in two co-expression clusters designated as cluster 116 and cluster 132 (Figure 5). Whereas those corresponding to the proteins down-regulated by dry after-ripening are enriched exclusively in cluster 132, suggesting that the two co-expression clusters represent further candidate genes regulating after-ripening-mediated seed dormancy release in wheat. Given that change in seed water status capable of inducing active metabolic reactions appear not to occur in dry seeds that maintain dormancy, the genes in the two clusters are more likely to be regulated by nonenzymatic post-transcriptional and post-translational mechanisms (Figure 5). Co-expression clusters 116 and 132 contain a total of 82 and 72 probesets (Table S3), respectively. Annotation of these probesets by HarvEST WheatChip and GO analysis revealed that the probesets in both clusters are enriched in nutrient reservoir (GO: 0045735, ≤ 8.6e-34) and enzyme regulatory (GO: 0030234, ≤ 4.5e-15) activities (Table S3). Furthermore, the two clusters consist of probesets corresponding to protease inhibitors, storage proteins and NAC domain containing proteins, suggesting the linkage of these genes with after-ripening-induced seed dormancy release.

Figure 5.

Co-expression analysis of probesets differentially expressed and oxidized by dry after-ripening and those corresponding to proteins that exhibited differential abundance between dormant (D) and after-ripened seeds (AR) in both dry and imbibed states by WheatNet. These probesets appear to be regulated at transcriptional, post-transcriptional and post-translational levels and are enriched in co-expression clusters 95, 116 and 132. It is likely that these co-expression clusters represent further candidate genes regulating seed dormancy in wheat. Probesets in clusters 116 and 132 are enriched in nutrient reservoir and enzyme regulatory activities. No GO category is enriched with probesets contained in cluster 95.

To extend our search for more candidate genes regulating seed dormancy in wheat, we performed co-expression analysis of probesets of proteins differentially expressed between imbibed D and AR seeds (57 unique probesets; Table S2). These probesets are enriched in cluster 95 and cluster 116 (Figure 5). As active metabolic reactions occur during seed imbibition and both similar and differential expression patterns were evident between the proteins and their respective probesets, it is likely that the genes in the two clusters are regulated by enzymatic transcriptional, post-transcriptional and post-translational mechanisms (Figure 5). However, we cannot rule out the possibility that nonenzymatic regulatory mechanisms operate during seed imbibition. Cluster 95 contained a total of 82 probesets (Table S3) with no enriched GO category, but consists of probesets corresponding to more candidate genes that potentially involve in the regulation of seed dormancy, including those representing dehydrin family proteins such as responsive to ABA 18 (RAB18), late embryogenesis (LEAs) and XERO1, and caleosins such as responsive to desiccation 20 (RD20). The presence of probesets representing ABI5, an important regulator of ABA signalling and dormancy in seeds (Nambara et al., 2010), in this cluster indicates the validity of this approach for the discovery of dormancy-related genes.

Transcriptional regulation of co-expressed genes

To identify candidate transcription regulators controlling the co-expressed genes, the probesets in each cluster were annotated by rice genes using HarvEST WheatChip. Analysis of the 1-kb upstream region of the identified rice genes by Osiris (http://www.bioinformatics2.wsu.edu/cgi-bin/Osiris/cgi/visualize_select.pl; Morris et al., 2008) revealed that two motifs, ACGTOSGLUB1 and PROLAMINBOXOSGLUB1 (found in the upstream region of rice GluB-1 gene), are enriched in clusters 95 and 116, respectively (Table 4). The ACGT motif, which is found in the promoter regions of genes encoding storage proteins, acts as a putative binding site for AtbZIP10 and AtbZIP25 transcription factors that interact with ABI3 (Lara et al., 2003). Consistently, the expression of seed storage protein genes exhibits a significant reduction in abi3 mutants (Parcy et al., 1994). Furthermore, two more ABA-regulated cis-elements, ABREOSRAB21 and ACGTABREMOTIFA2OSEM, are enriched in the co-expression cluster 95 (Table 4). All these data suggest that the genes assigned in co-expression cluster 95 are regulated by ABA and play important roles in controlling seed dormancy. The cis-element PROLAMINBOX (5′-TGTAAAG-3′) is recognized by DNA binding with One Finger (DOF) transcription factor (Yanagisawa, 2004) that mediate ABA-induced transcriptional changes in its target genes in germinating seeds (Moreno-Risueno et al., 2007), suggesting that the genes in cluster 116 might also be controlled by ABA and play roles in regulating dormancy release by after-ripening.

Table 4. Sequences of cis-elements over-represented in co-expression clusters 95 and 132a
Co-expression Cluster #Motif sequenceMotif IDP-valueb
  1. a

    Cis-elements were identified using the OSIRIS Web-based resource.

  2. b

    Significance was determined at P < 0.001.

95ACGTSSSC (S = G/C)ABREOSRAB21<10−4
GTACGTGACGTOSGLUB1<10−4
ACGTGKC (K = G/T)ACGTABREMOTIFA2OSEM<10−10
116TGCAAAGPROLAMINBOXOSGLUB1<10−4

In summary, our data show that selective oxidation of seed-stored mRNAs form a mechanism to control seed dormancy in wheat and the changes in patterns of seed transcripts and proteins between D and AR samples before and during imbibition highlight the significance of specific post-transcriptional and post-translational features regulating seed dormancy maintenance and release. These findings contribute to a better understanding of the molecular mechanisms controlling wheat seed dormancy, which is necessary to design strategies for the improvement in PHS tolerance in wheat.

Experimental procedures

Plant materials and growth conditions

Seeds of common wheat cv. AC Domain were used in this study. AC Domain is hard red spring wheat characterized by a high degree of PHS tolerance at harvest and is adapted to the Canadian prairies (Townley-Smith and Czarnecki, 2008). Plant growth conditions, seed harvesting and generation of AR seeds are as described before (Gao et al., 2012). Briefly, mature air-dried seeds were harvested from the middle region of spikes of wheat plants grown in a greenhouse at 18–22 °C/14–18 °C (day/night) under a 16/8-h photoperiod. A portion of the seeds was immediately stored at −80 °C, whereas the remaining portion was first stored at room temperature and ambient relative humidity for 10 months (which gave a seed moisture content of 0.07 g water g DW−1) and then stored at −80 °C until further use.

Seed germination assays

Germination assays were performed as described previously (Gao et al., 2012). Briefly, the D and AR seed samples were surface-sterilized and then imbibed between two layers of filter papers moistened with sterile water in a Petri-plate (25 seeds per plate) at 22 °C in darkness. Seeds were considered germinated when the coleorhiza was visible beyond the seed coat. Unimbibed D and AR seeds were used for analyses of mRNA oxidation and seed proteomes in the dry state.

Immunoprecipitation of oxidized mRNAs

Total and mRNA samples were extracted from three independent biological replicates of the same dry D and AR seeds used for our germination and transcriptomic studies following the protocol described before (Gao et al., 2012). Oxidized mRNAs were immunoprecipitated according to Bazin et al. (2011b) and Shan et al. (2003). Briefly, the mRNA samples (approximately 3 μg) were incubated at 65 °C for 2 min and then on ice to remove secondary structures. Subsequently, the mRNA samples were mixed with monoclonal anti-8-OHG antibody 15A3 (2 μg; QED Bioscience, San Diego, CA) in 200 μL 1× PBS buffer and incubated at room temperature for 2 h. No antibody was added to the negative control experiments. Washed Dynabeads (Invitrogen, Carlsbad, CA) were then added followed by overnight incubation of the mix at 4 °C to capture the antibody. The beads were washed three times with 1× PBS with 0.02% (v/v) Tween-20, and the immunoprecipitated RNA extracted from the pellet by adding 150 μL of 1× PBS, 0.02% (v/v) Tween-20 and 1% (w/v) SDS and incubating the mixture at 70 °C for 10 min. The Dynabeads were then captured by a magnet and the supernatant precipitated with ethanol/NaCl in the presence of 1 μL of 15 mg/mL glycoblue (Invitrogen) as coprecipitant. The pellet containing the mRNA was dissolved in RNase-free H2O and used for microarray hybridization.

DNA microarray analysis

Labelling and hybridizing the oxidized mRNA samples (100 ng mRNA in 3 μL) to the Affymetrix GeneChip Wheat Genome Array (Affymetrix, Santa Clara, CA), verification of the reproducibility of the data from the three independent biological replicates and subsequent analysis were performed as described previously (Gao et al., 2012). Differentially oxidized probesets were identified using FlexArray software (http://genomequebec.mcgill.ca/Flex-Array; Blazejczyk et al., 2007) by analysis of variance (≥ twofold change and ≤ 0.05). Validation of the microarray data was performed with qPCR as described before (Gao et al., 2012) using eight randomly chosen probesets with or with no differential oxidation (Table S1). The microarray data derived from the mRNA oxidation experiment have been deposited in the Gene Expression Omnibus (GEO) database with the accession number GSE41949.

Protein extraction and two-dimensional gel electrophoresis

Total soluble protein was extracted from three independent biological replicates of dry and 24 h imbibed whole seeds (0.5 g fresh weight) of the same D and AR samples used for germination, mRNA oxidation and transcriptomic studies as described previously (El-Bebany et al., 2010). To perform the two-dimensional electrophoresis, protein samples [500 μg/450 μL isoelectric focusing (IEF) solution] were used to rehydrate the 24-cm IEF strips (pH 4–7, GE Healthcare, Little Chalfont, UK) overnight at 22 °C under a layer of dry strip fluid. The IEF strips were then focused for a total of 58.3 kVh (Multiphor II: GE Healthcare) and then equilibrated before run on second dimension 12% sodium dodecyl sulfate (SDS)-polyacrylamide gels. Subsequently, gels were stained with Coomassie Brilliant Blue R-250 and the protein patterns compared between samples. Estimation of the abundance of spots representing the targeted proteins was performed with Quantity One quantification software (Bio-Rad, Hercules, CA) after taking the background signal into account. Significant differences in protein abundance between samples were determined using Student's t-test at P ≤ 0.05.

Mass spectrometry

Gel pieces containing the differential protein spots were excised and processed as described by El-Bebany et al. (2010). Following digestion in situ with trypsin (modified sequencing grade: Promega, Madison, WI), the peptide samples were analysed with a linear ion trap mass spectrometer (LTQ XL; ThermoFisher Scientific, Waltham, MA) coupled with a nano-HPLC system (Ultimate 3000; Dionex, Germering, Germany) using the parameters described in El-Bebany et al. (2010). Following conversion to Mascot generic format files, the output files were queried with Mascot (v2.2; Matrix Science, London, UK) against the protein NCBI nonredundant (nr) protein and a local nonredundant wheat EST databases.

Conversion of proteins into probesets and co-expression analysis

To investigate whether the proteins differentially expressed between D and AR seeds are controlled transcriptionally, and to perform gene co-expression analysis, the differential proteins were converted into their respective probesets using WheatNet Blast2probesets tool (http://aranet.mpimp-golm.mpg.de/wheatnet; cut-off value e ≤ 1e-50). The expression profiles of the resulting probesets were extracted from transcriptomic data derived from the same D and AR wheat seed samples (GEO accession number GSE32409). Analysis of the transcriptomic data was performed as described previously (Gao et al., 2012), and probesets were considered to be differentially expressed if they exhibit ≥ twofold change at ≤ 0.05.

Probesets corresponding to proteins differentially expressed by after-ripening in both dry and imbibed states (Table S2) and those differentially expressed (Gao et al., 2012) and oxidized (Table S1) during after-ripening were queried to generate co-expression clusters using Web tools in WheatNet (http://aranet.mpimp-golm.mpg.de/wheatnet). Following the co-expression analysis, probesets in each co-expression clusters were annotated by rice genes using HarvEST WheatChip. Analysis of the 1-kb upstream region of the identified rice genes was performed with Osiris (http://www.bioinformatics2.wsu.edu/cgi-bin/Osiris/cgi/visualize_select.pl; Morris et al., 2008).

Acknowledgements

This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada to BTA. The authors thank Zhen Yao, Brenda Oosterveen and Jacqueling Ching for their technical assistance. The authors have no conflict of interest to declare.

Ancillary