• Seed dormancy and dormancy cycling are poorly understood at the molecular level, but are central to plant community development. This study focuses on the embryonic axes of deeply dormant seeds of Prunus avium (wild cherry).
• Rapid amplification of cDNA ends (RACE), differential display and quantitative PCR were used to recover and monitor expression of cDNAs.
• cDNAs similar to two dormancy-imposing genes were isolated; a serine/threonine protein-phosphatase 2C homologous to ABSCISIC ACID INSENSITIVE 1 and 2 (Pa-PP2C1), and the transcription factor ABSCISIC ACID INSENSITIVE 3 (Pa-ABI3). Two germination-associated cDNAs were recovered; aconitase (Pa-ACO1) and eukaryotic translation initiation factor 3 subunit 8 (Pa-eIF3 SUBUNIT 8). Cold-treatment reduced expression of Pa-PP2C1 and Pa-ABI3, consistent with roles in establishing primary dormancy; neither was induced by imposition of secondary dormancy. Expression of these genes was distinct from expression of Pa-ACO1 and Pa-eIF3-SUBUNIT 8.
• Results were consistent with a role for ABI1/ABI2 and ABI3 homologues in primary dormancy of P. avium embryonic axes, but not secondary dormancy as control of germination appeared overridden by the tissues surrounding the embryo.
A wide range of dormancy mechanisms have evolved in higher plants to regulate germination of their seeds. Physiological seed dormancy is present throughout higher plants, and has a profound impact on the timing and periodicity of seedling emergence and, more widely, on the structure and development of plant communities across all major climatic regions (Baskin & Baskin, 1998). Consequently, the induction and loss of dormancy are triggered by diverse environmental cues activated through many different physiological mechanisms. The role of dormancy is, however, the same across all species – to spread germination across time but in synchrony with seasonal cycles to avoid unfavourable weather, maximize competitive advantage and ensure the establishment of plants (Baskin & Baskin, 1998). This conserved role suggests there may be a common underlying molecular mechanism. Studies with model species that have comparatively shallow dormancy have identified a number of genes involved in the regulation of dormancy and germination (Bewley, 1997; Holdsworth et al., 1999; Bentsink & Koornneef, 2002; Koornneef et al., 2002). However, there have been very few molecular studies in more deeply dormant species in which moist seeds require prolonged exposure to low temperature to release dormancy and induce germination. To address this, the presence and expression patterns of putative dormancy regulating gene homologues were investigated in the deeply dormant seeds of Prunus avium (wild cherry) exposed to dormancy-regulating temperatures.
The dormancy mechanisms in P. avium, like those of other species with deeply dormant seeds, appear to be complex. The embryo is dormant and the surrounding structures (endosperm, seed coat and endocarp) also impose some control (coat-enhanced dormancy). Embryo dormancy is deemed to be present when the embryo fails to grow after the removal of the tissues that enclose it. In P. avium, embryo dormancy is thought to occur early in development before seeds are fully mature, but shortly after seeds acquire the ability to germinate (Jensen & Eriksen, 2001). At this stage, growth can occur in excised embryos, but they produce physiologically dwarf plants and other abnormalities (Tukey, 1934). As development continues, the potential for embryo growth decreases (Abou-Zeid, 1972; Abou-Zeid & Gruppe, 1972; Jensen & Eriksen, 2001). Embryos extracted from fully mature seeds have very little capacity for growth, but this increases progressively when moist seeds are exposed to low temperature (Pollock & Olney, 1959). There are reports in the literature of variable potential for radicle growth between seed lots in the absence of low-temperature treatment (Michalska, 1982), but this may have resulted from using seeds that are not fully mature. Due to its fleshy fruit, the extent of seed maturity at harvest in P. avium is difficult to judge, and this can greatly influence the extent of seed dormancy and its response to low-temperature treatment (Finch-Savage et al., 2002). The presence of surrounding tissues further delays germination during low temperature treatment. A period of higher temperature (20–25°C) for 2 wk or more can re-impose full dormancy at any time before complete loss of dormancy in individual seeds (Suszka, 1962, 1967; Suszka et al., 1996). When dormancy is lost in any one seed, these same temperatures will increase the germination rate, as there is no separate treatment required for the induction of germination in this species. This well characterized system of progressive loss of primary dormancy and germination in the population, and then the induction of secondary dormancy as ambient temperatures rise in late spring, is thought to result in the spread of germination within and across years under natural conditions (Suszka, 1990).
The aim of the present study was to gain insight into the molecular control of dormancy in the deeply dormant seeds of P. avium by investigating expression of P. avium homologues of the dormancy-inducing genes ABI1/2 and ABI3 during low-temperature treatment. Their temporal expression patterns were compared to those of ACO1 and eIF3 SUBUNIT 8 as likely indicators of germination. The results were consistent with a role for ABI1/ABI2 and ABI3 homologues in primary dormancy of P. avium axes, but not secondary dormancy, as control of germination appeared overridden by the tissues surrounding the embryo.
Methods and Materials
Seed handling and dormancy-breaking treatment
The depth of dormancy and response to temperature treatment change in the late stages of seed maturity in wild cherry, and this may not be clearly indicated by changes in the surrounding fruit (Finch-Savage et al., 2002). In addition, the degree of seed maturity differs greatly between fruits on the same tree. Trees were therefore netted and care was taken to harvest only uniformly fully ripe fruit. Following harvest, the soft pericarp of the fruit was removed to leave the seed covered by the endocarp (henceforth collectively called the seed). Seeds were then dried to 10% water content before dry storage at 1°C. Before treatment, seeds were imbibed for 48 h on capillary matting saturated with distilled water at 20°C. Imbibed seeds were then mixed as 30% w/w with a 1 : 1 mixture of prewetted sharp sand and peat (Levingtons F1, Levington Horticulture Ltd, Ipswich, UK) in a layer no more than 5 cm deep. The moisture content of the mixture was checked periodically, and water at ambient temperature added as necessary. Control seeds were stored in the peat mix at 20°C throughout. The dormancy-breaking procedure used was 2 wk storage at 20°C followed by up to 17 wk storage at 3°C. Interruption of dormancy breaking was studied by removing seed samples from storage at 3°C after 6 and 9 wk and returning them to 20°C for 3 wk. Seeds were sampled at random from the bulk mixture in batches of 120 and the embryonic axes recovered, snap-frozen in liquid nitrogen and stored at −80°C until RNA extraction. The extent of dormancy loss was assessed at intervals during treatment by recording germination (defined in this study as 3 mm radicle protrusion) at 20°C on batches of 100 seeds removed from each of two bulk mixtures. Germination was also recorded in the same treatments on four replicates of 25 seeds each kept in 125 cc of medium. Previous experience had shown that the treatment procedures used maintained adequate aeration and moisture to the seeds (Finch-Savage et al., 2002).
Recovery and sequencing of P. avium ABI3, PP2C1, ACO1 and eIF3 SUBUNIT 8 cDNAs
DNA sequencing was performed by Sequiserve (Vaterstetten, Germany). Rapid amplification of cDNA ends (RACE; Frohman et al., 1988) was used to recover Pa-ABI3, Pa-PP2C1, Pa-ACO1 and Pa-eIF3 SUBUNIT 8 transcripts using a commercial kit (GeneRacer, Invitrogen, Paisley, UK) as detailed for each transcript. A two-step reverse transcription PCR using 2 µg P. avium total RNA (dephosphorylated, decapped and ligated to the GeneRacer 5′ RNA oligonucleotide), the GeneRacer-anchored poly T primer and Omniscript reverse transcriptase (Qiagen, Crawley, UK) was followed by PCR with Expand High-Fidelity polymerase (Roche, Mannheim, Germany). Gene-specific 3′ and 5′ RACE primers were used in conjunction with primers corresponding to the 3′ and 5′ sequences added to cDNA by the GeneRacer 3′-anchored poly T oligonucleotide and 5′ RNA Oligo, respectively. All custom RACE primers are shown in Table 1. RACE templates used for Pa-ABI3 and Pa-PP2C1 recovery were cDNA generated from primary dormant embryos which had never been chilled. Recovery of Pa-ACO1 and Pa-eIF3 SUBUNIT 8 RACE fragments was from embryonic axes that had been held at 3°C for 16 wk to break dormancy. RACE fragments were cloned in PCR Topo4 or 2.1 vectors and sequenced bidirectionally. At least two clones were sequenced for each fragment. Ambiguities were resolved by amplifying full-length or near-full-length coding sequence cDNAs using primers designed from the compiled 3′ and 5′ RACE fragments and direct sequencing.
Table 1. Primers used in 5′ and 3′ rapid amplification of cDNA ends (RACE) to recover cDNAs representing Prunus avium homologues of AB13, PP2C1, AC01 and eIF3 SUBUNIT 8
5′ RACE primer
3′ RACE primer
All sequences are shown 5′–3′.
Pa-eIF3 SUBUNIT 8
An internal fragment of P. avium ABI3 cDNA was recovered first by employing the primers and conditions of Fukuhara & Bohnert (2000). The corresponding genomic DNA fragment was also amplified from P. avium total DNA to allow some insight into genomic organization. Following cloning, the derived sequences were used to design a gene-specific primer (GSP) for use in 3′ RACE of Pa-ABI3 transcripts. The sequence recovered by 3′ RACE was used to design a GSP for use in 5′ RACE.
A Pa-PP2C1 3′ RACE GSP was designed by alignment of the cDNA sequences of the Fagus sylvatica and A. thaliana homologues. The recovered RACE fragment represented the 3′ 506 nucleotide pairs of a PP2C homologue plus 3′ untranslated region. This sequence was used to design a 5′ RACE primer. This generated a single c. 2 kb product corresponding to the remaining Pa-PP2C1 sequence.
Fragments of P. avium ACO1 and eIF3 SUBUNIT 8 were initially recovered from differential display gels following the protocol of Liang & Pardee (1992, 1997). Differential display was carried out to compare poly A+ RNA extracted from P. avium embryos that had undergone three treatments. These were 20 wk at 20°C (primary dormant); 2 wk at 20°C plus 15 wk at 3°C (nondormant, selected as split endocarp, endosperm not split but embryonic root expanded; germination tests showed 60% nondormant population); and germinated. Differential display analysis suggested high relative expression in embryos from seeds chilled for 16 wk compared to primary dormant seeds or germinated seeds (data not shown). The sequences derived from these bands (600 and 900 bp, respectively) were used to design 5′ RACE primers. The GSPs for 3′ RACE were designed from the sequences of the cloned 5′ RACE products.
Total RNA (2 µg) recovered using an RNeasy Plant Mini Kit with on-column DNase treatment (Qiagen) was converted to cDNA using an Omniscript system (Qiagen) and random hexanucleotide primers (Sigma-Aldrich, Dorset, UK) according to the manufacturer's protocol. Real-time PCR detection was carried out using 10 ng of cDNA template (assuming 100% conversion), SYBR Green Master Mix (Applied Biosystems, Warrington, UK) and 10 pmol each of the primers detailed in Table 2. Novel primers were designed using primer express software (Applied Biosystems) for use in two-step PCR, in a total volume of 15 µL. Sample normalization was carried out by simultaneous quantification of 18S ribosomal transcripts. Ribosomal transcripts were quantified using 2.5, 5.0, 10 and 20 ng of cDNA as template with 18S rRNA primers (van Hannen et al., 1998) to estimate PCR efficiency which in all cases approximated to 100%. An ABI Prism 7900HT Sequence Detection System (Applied Biosystems) was used for amplification and quantification. Each sample was analysed in triplicate reactions: time points 1 and 2 were carried out in biological triplicates; points 3–5 in biological duplicates for chilled seeds and as single points for control seed batches. Germinated embryos were assayed as a single point. PCR was carried out over 40 cycles with data collection during the extension phase of each cycle. All samples were run simultaneously, except for the germinated embryos. These were tested at a later date along with re-runs of two other samples.
Table 2. Primers used in real-time quantitative PCR experiments to monitor expression of Pa-AB13, Pa-PP2C, Pa-ACO1 and Pa-elF3 SUBUNIT 8
The means and variances of the threshold values for the three replicate reactions for the gene of interest and the 18S rRNA control were calculated on each sample. The data were normalized by subtracting the mean control value from the mean value for the gene. The variance of this normalized value for the sample was estimated by the sum of the variances of the threshold values for the gene and the control.
The normalized data were analysed using REML (Patterson & Thompson, 1971) in the statistical package GenStat (Payne, 2000). REML is a generalization of ANOVA which is also suitable for unbalanced designs. Dormancy-breaking treatment, time of storage, and their interaction were treated as fixed terms. Experimental run (germinated material and replicated samples vs the others) was taken as a random term, as was a residual term for the differences between biological replicates. An additional residual term, equal to the sample variance calculated for each sample in the normalization step, was also included. The statistical significance of differences between treatments were assessed using t-tests.
To produce the figures, the analysis was repeated with experimental run treated as a fixed term. This prevents the displayed standard errors of means being heavily influenced by the difference between the values of the repeated samples in the two experimental runs. The mean values from the analysis were then calibrated by subtracting the smallest value from each, and the relative amounts shown in the figures calculated by raising 2 to the power of the calibrated mean value. Therefore, normalization of data was first performed by relation to the 18S rRNA level to normalize for sample–sample variation, then by reference to the lowest expression point detected for each gene, such that the minimum expression for each gene became 1.
Genomic DNA was extracted from unopened buds of the same tree from which seeds were collected using a GenElute Plant Genomic DNA Extraction Kit (Sigma). Eight extractions were performed on 100 mg tissue each, and the purified DNA combined and concentrated by ultrafiltration (Centricon 100; Millipore, Watford, UK). DNA (10 µg) was digested overnight with the restriction endonucleases BamHI, EcoRI, HinDIII or PvuII with the manufacturer's buffers (Roche). Complete digestion was verified by adding 1 µg intact λ-DNA to 5% of each digestion mixture, overnight digestion and agarose gel electrophoresis to view the expected λ-digestion size ladder (Sambrook et al., 1989, data not shown). Digests were run on 1% agarose gels at 5 v cm−1 before depurination, denaturation, neutralization, and transfer to Hybond-N membranes (Sambrook et al., 1989). Hybridizations were carried out using Hybond-N membranes in MiracleHyb solution (Stratagene, Amsterdam, the Netherlands) at 68°C and washed twice in 2 × SSC, 1% SDS at 60°C. Probes consisted of PCR-amplified full-length sequences of each cDNA labelled with 50 µCi of 32P-α-dCTP (3000 Ci mmol−1) using a Ready Prime random labelling kit (both Amersham plc, Little Chalfont, UK). Images were captured using a phosphorescent imaging system (Molecular Dynamics PhosphorImager SI and imagequant 5.0 software, Amersham).
P. avium ABSCISIC ACID INSENSITIVE 3 (VIVIPAROUS 1)
The complete sequence and predicted primary structure of the Pa-ABI3 cDNA clone is shown in Fig. 1a. This cDNA is 2640 nucleotides long and encodes an open reading frame (ORF) from nucleotides 139–2553. The deduced polypeptide sequence consists of 804 amino acids with a predicted molecular mass of 89.985 kDa and a slightly acidic pI of 6.21. Pa-ABI3 has a structure highly similar to other members of the ABI3/VP1 family, an N-terminal domain rich in hydroxyl-amino acids (serine and threonine), and three putative DNA-binding domains rich in basic amino acids (B1, B2 and B3; McCarty et al., 1991; Giraudat et al., 1992) highlighted in Fig. 1a. The B3 domain is highly conserved and diagnostic of VP1/ABI3 transcription factors. The closest match to Pa-ABI3 in the current sequence database is the homologue recovered from Populus balsamifera (Rohde et al., 1998), a 736 amino acid polypeptide with 54% identity and 64% similarity of amino acids. Pa-ABI3 is the largest member of the VP1/ABI3 family recovered to date, being 14 amino acids longer than the previously largest member recovered from Mesembryanthemum crystallinum (Fukuhara & Bohnert, 2000). The partial genomic clone spanned two small introns (132 and 113 bp, respectively, 5′−3′) in highly conserved sites (Giraudat et al., 1992), the positions of which are indicated in Fig. 1a.
P. avium PROTEIN PHOSPHATASE 2C-1 (ABSCISIC ACID INSENSITIVE 1/ABSCISIC ACID INSENSITIVE 2: EC 126.96.36.199)
The complete sequence and predicted primary structure of the Pa-PP2C1 cDNA clone is shown in Fig. 1b(i). This cDNA is 1725 nucleotides long and encodes an ORF from nucleotides 121–1399. The deduced polypeptide sequence consists of 426 amino acids with a predicted molecular mass of 46.214 kDa and a slightly acidic pI of 5.62. The only recognizable domain within Pa-PP2C1 is the catalytic domain of the family 2C, serine/threonine phosphatases, which is underlined in Fig. 1b(i). The closest available match is the F. sylvatica homologue (Lorenzo et al., 2001), a 413 amino acid polypeptide with 66% identity and 76% similarity of primary sequence. Both A. thaliana homologues of Pa-PP2C1 (ABI1 and ABI2) were identified through a point mutation converting a glycine at position 180 or 168, respectively, to an aspartic acid residue, which reduces phosphatase activity and affinity for Mg2+ (Leung et al., 1997). The partial multiple alignment shown in Fig. 1b(ii) shows that the wild-type glycine is conserved in the P. avium sequence at position 175, and in the F. sylvatica homologue.
P. avium ACONITASE 1 (ACONITATE HYDRATASE, CITRATE HYDRO-LYASE: EC 188.8.131.52)
The complete sequence and predicted primary structure of the Pa-ACO1 cDNA clone is shown in Fig. 1c. This cDNA is 3113 nucleotides long and encodes an ORF from nucleotides 119–2827. The deduced polypeptide sequence consists of 902 amino acids with a predicted molecular mass of 98.88 kDa and slightly acidic pI of 6.01. The only recognizable domain within Pa-ACO1 are the ‘aconitase’ and ‘aconitase c’ domains, which have also been grouped together to form the single ‘AcnA’ (energy production and conservation) domain and span almost the entire protein, as indicated in Fig. 1c. The closest available sequence match to Pa-ACO1 is the 898 amino acid polypeptide encoded by the cytoplasmic A. thaliana homologue (Peyret et al., 1995), with 89% identity and 94% similarity.
The complete sequence and predicted primary structure of Pa-eIF3 SUBUNIT 8 cDNA is shown in Fig. 1d. This cDNA is 3114 nucleotides long and encodes an ORF from nucleotides 180–2984. The deduced polypeptide sequence consists of 934 amino acids with a predicted molecular mass of 106.301 kDa and an acidic pI of 5.65. The only recognizable domain within Pa-eIF3 SUBUNIT 8 is the PCI region (proteasome, COP9, initiation factor 3; also known as the PINT motif for proteasome, Int-6, Nip-1 and TRIP-15) region indicated in Fig. 1d. The closest available match to Pa-eIF3 SUBUNIT 8 is that encoded by the A. thaliana homologue p105 (Karniol et al., 1998), a 900 amino acid polypeptide showing 68% identity and 78% similarity of primary sequence.
Estimation of genomic copy number
Estimates of genomic copy numbers of homologues of these genes are available from studies on a variety of other plants. Table 3 shows the number of P. avium bands detected by southern blot hybridization (data not shown), predicted restriction enzyme-cutting sites from the cDNA sequences, the anticipated size of gene family, and references providing these estimates. The results of directly comparable experiments in other plants were available for ABI3 and ACONITASE genes. PP2Cs constitute a large protein family: sequence analysis of the A. thaliana genome suggests that it encodes 69 orthologous members (Kerk et al., 2002). However, Leung et al. (1997) recovered only two hybridizing sequences in an A. thaliana genomic library with an ABI1 probe: ABI1 itself and ABI2. It therefore appears that sequence degeneracy between PP2C orthologues is too high to reveal the extent of the family through this type of analysis. eIF3 SUBUNIT 8 cDNA sequences have been recovered from only two plant species to date: A. thaliana (Karniol et al., 1998) and Medicago truncatula (van Buuren et al., 1999), neither of which reports copy number estimates, but other components of eIF3 have been found to be represented by only one or two genomic sequences (Sabelli et al., 1999). Our estimates of P. avium gene copy numbers encoding the polypeptides described here are therefore consistent with those of other plant systems as far as they have been reported.
Table 3. Results of southern blot analysis of Prunus avium with the enzymes BamHI, EcoRI, HinDIII and PvuII
Estimated copy number
Expected copy number
The number of cutting sites within cDNA sequences is given to aid in estimation of genomic copy number. The partial genomic sequence of Pa-AB13 contained no cutting sites for any of the four enzymes used.
Expression of genes during dormancy breaking and germination
Figure 2a shows the pattern of dormancy loss in an initially dried and re-imbibed P. avium seed lot stored moist at 20°C for 2 wk, then at 3°C for 17 wk, a commonly used treatment in commercial practice to promote germination in this species. Seeds that had visibly germinated (> 3 mm radicle extension through the endosperm) during treatment at 3°C were discarded. At intervals, samples of seed were returned to 20°C to induce germination in fully nondormant seeds and induce secondary dormancy in those in which dormancy had not been fully lost. Nondormant seeds were defined as those that germinated within 3 wk of the return to 20°C. Seeds which have not completely lost dormancy are rendered secondary dormant by this procedure, as shown in Fig. 2b. This figure shows cumulative germination of cherry seeds over time during the standard dormancy-breaking treatment, and the effect of interrupting the 3°C period of this treatment with a 3 wk period at 20°C. In the latter case, germination was delayed by 8–9 wk, showing that the effect of previous cold treatment had been fully reversed by the warm interruption. No germination occurred when seeds were maintained at a constant 20°C.
Figure 3 shows the relative expression profiles of four selected genes in embryonic axes excised from seeds exposed to the following treatments: moist storage at 20°C (control); the standard dormancy-breaking treatment and germination at 3°C; and during secondary dormancy induced by 3 wk storage at 20°C, after 6 and 9 wk at 3°C.
Expression pattern of Pa-ABI3 The ABI3 protein is known to function as a transcription factor, with transcriptional activation and DNA-binding properties (McCarty et al., 1991; Suzuki et al., 1997), and is known to be essential in generating the state of primary embryo dormancy in other plant species (McKibbin et al., 2002 and references cited therein). Figure 3a shows that during storage at 3°C, after 12 wk expression of Pa-ABI3 fell to a basal level which was maintained in germinated seedlings. Expression in embryonic axes of control seeds stored at 20°C also fell initially, and by 11 wk stabilized significantly (P < 0.05) higher than in seeds stored at 3°C (c. fourfold after 17 wk of 3°C treatment). Inducing secondary dormancy by warming to 20°C for 3 wk following 6 or 9 wk at 3°C did not re-induce expression of Pa-ABI3, expression was indiscernible from that in axes of seeds stored at 3°C for the same total time.
Expression pattern of Pa-PP2C1Figure 3b shows that expression of Pa-PP2C1 fell rapidly in the embryonic axes of seeds stored at 20°C, stabilizing after 8 wk. Expression also fell in seeds stored at the dormancy-breaking temperature of 3°C. After 17 wk, expression in axes in seeds stored at 3°C was significantly (P < 0.05) lower than in the control, and remained repressed in germinated embryos. However, as for Pa-ABI3, inducing secondary dormancy by returning seeds to 20°C following 6 or 9 wk at 3°C did not induce Pa-PP2C1 to control values. Axes in secondary dormant seeds demonstrated marginally, though nonsignificantly, lower (c. 60%) expression of Pa-PP2C1 than embryos maintained at 3°C for the same total time.
Expression pattern of Pa-ACO1ACONITASE gene expression is linked to mobilization of storage compounds, and has previously been shown to undergo increases in expression during seed maturation and germination (Hayashi et al., 1995; Peyret et al., 1995; Eastmond & Graham, 2001). Figure 3c shows that results were consistent with this model, maximal expression being detected in untreated imbibed and fully treated germinated embryos. Expression was also elevated in primary dormant embryos following 6 wk at 3°C (c. twofold, P < 0.05), but then fell throughout the remaining period of dormancy breaking until germination. There was no significant change in Pa-ACO1 expression following induction of secondary dormancy.
Expression pattern of Pa-eIF3 SUBUNIT 8 Expression of eIF3 SUBUNIT 8 has been little studied in plants, but has been shown to be induced during the developmental changes involved in forming root associations with arbuscular mycorrhizal fungi (van Buuren et al., 1999). We hypothesized that an increase in Pa-eIF3 SUBUNIT 8 expression could be expected concomitant with loss of dormancy in the seed population, on the basis that implementing developmental change would involve the product of this gene. Figure 3d demonstrates that this was not the case. Pa-eIF3 SUBUNIT 8 expression was consistently high in primary dormant embryos stored at 20°C. Expression fell significantly during storage at 3°C (P < 0.05 at 6 wk chilling, the difference generally increasing with time), and rose significantly once germination had taken place (P < 0.05). Induction of secondary dormancy after 6 or 9 wk of chilling did not return Pa-eIF3 SUBUNIT 8 expression to control levels.
The germination behaviour of P. avium seeds used in the present study was consistent with that reported elsewhere, showing an absolute requirement for low temperature treatment and the induction of secondary dormancy by higher temperatures (Suszka, 1962, 1967; Suszka et al., 1996; Jensen & Eriksen, 2001; Finch-Savage et al., 2002). Fully cold-treated (nondormant) seeds will germinate at a wide range of temperatures, including 3 and 20°C as used here, without the need for further induction treatments. However, if the seeds are not fully cold-treated, significant exposure to the higher of these temperatures will induce secondary dormancy that is manifested by reversing any previous cold treatment. Thus secondary dormant seeds require the full period of cold treatment characteristic of primary dormant seeds (Fig. 2b). Previous work has shown this cycle can be repeated several times (Suszka, 1962, 1967; Suszka et al., 1996). Four genes likely to be involved in inducing dormancy and permitting germination were studied in embryonic axes excised from seeds exposed to these dormancy-breaking and dormancy-inducing temperature treatments.
The genes studied were shown to be highly similar in structure and genomic copy number to their counterparts in other, more studied, angiosperms. The similarity levels of Pa-PP2C1 to the A. thaliana ABI1- and ABI2-coding regions are equally high, with no obvious homology in the N-terminal 60 amino acids, followed by 45% identity and 58% similarity over the remainder of the protein. We cannot therefore define Pa-PP2C1 as a specific homologue of either ABI1 or ABI2.
The expression of the selected genes during loss of primary dormancy and germination closely mimics that expected from studies of their counterparts in other species. However, none of them revealed expression patterns suggestive of specific induction or repression reflecting initiation of the nondormant state. This is of particular note in the case of Pa-PP2C1 and Pa-ABI3, as the products of these genes are known to be important in imposing embryo dormancy in other systems (McCarty et al., 1991; Giraudat et al., 1992; Leung et al., 1994; Meyer et al., 1994; Leung et al., 1997; Rodriguez et al., 1998; Merlot et al., 2001), but transcript levels were low when 99% of seeds were still dormant (Fig. 2). This observation generates two hypotheses. Transcriptional analysis does not directly reflect the levels of active protein product, and it may be the case that Pa-ABI3 and Pa-PP2C1 activity declined to some dormancy-releasing level concomitant with the observed germinability of seed batches shown in Fig. 2. More likely, perhaps, is that embryo dormancy had indeed been lost some weeks before dormancy of the intact seed was broken. In this case dormancy would have been maintained through control exerted by the surrounding tissues (endosperm and seed coat). This second hypothesis is consistent with earlier morphological observations (Pollock & Olney, 1959; Michalska, 1982) that report a variable but enhanced capacity for root and leaf growth in isolated P. avium embryos after only 2–4 wk storage at 3°C.
Control levels of both Pa-ABI3 and Pa-PP2C1 transcripts also fell during storage at 20°C, but not to the levels induced by chilling, suggesting that some transcriptional activity continued at these loci at 20°C, or that the transcripts were more stable. However, inducing secondary dormancy by warming the seeds to 20°C after 6 or 9 wk chilling did not return transcript levels to control values (Fig. 3a,b). Again, this observation is consistent with a model of embryo dormancy having been lost after limited exposure to dormancy-breaking temperatures and subsequent control of germination during both continued primary dormancy and induction of secondary dormancy residing in the tissues surrounding the embryo. Involvement of an abi3 homologue in imposing secondary dormancy has been suggested only in whole seeds of A. fatua (Holdsworth et al., 1999).
Differential display analysis had suggested that P. avium ACONITASE and eukaryotic INITITAION FACTOR 3 SUBUNIT 8 gene expression were elevated in nondormant embryos relative to dormant and germinated embryos. On this basis, it was hypothesized that an elevation of expression concomitant with increasing nondormancy might be observed on the basis that entering and maintaining the nondormant state might require greater demands on energy reserves than primary dormancy (Fig. 2). However, analysis of ungerminated embryonic axes stored at 3°C for 6, 9, 12, 15 and 17 wk did not indicate any increased demand for the aconitase enzyme before a significant increase in expression in visibly germinated embryos, as expected from published studies (Hayashi et al., 1995; Peyret et al., 1995; Eastmond & Graham, 2001; Fig. 3c).
Expression of Pa-AC01 initially declined during storage at 20°C and later regained the higher expression shown in newly imbibed seeds. Some evidence of a peak in Pa-ACO1 expression shortly after the shift to 3°C was revealed (Fig. 3c). Expression of Pa-eIF3 SUBUNIT 8 rose between imbibition and the next sampling point after 2 wk at 20°C. Expression fell during chilling at 3°C, and was only significantly induced in visibly germinated embryos with an increased demand for protein synthesis (Fig. 3d). Expression remained significantly higher in control seeds stored at 20°C, suggesting that Pa-eIF3 SUBUNIT 8 is more actively transcribed or more stable in moist, primary dormant seeds than in those in which dormancy-breaking temperatures have been experienced. These findings are reminiscent of the rates of protein synthesis in whole P. avium seeds described by Michalski (1982). Michalski used 14C-leucine incorporation to monitor protein synthesis, and suggested a higher rate in seeds stored at 20°C than in those stored at 3°C.
eIF3 SUBUNIT 8 transcript in secondary dormant embryos induced by a period at 20°C did not increase; levels continued to fall rather than rising to control values (Fig. 3d). Wrzeœniewski (1985) demonstrated an increase in respiratory rate in cherry embryos on interruption of dormancy breaking with thermal increases to 20°C which returned to baseline within 1 wk. This burst in respiration has been associated with the onset of secondary dormancy, and may be expected to involve ACONITASE and/or eIF3 SUBUNIT 8 expression. As transcript levels were not measured in this study until 3 wk after the thermal shift to 20°C, no information on this possibility was recovered.
The function of eIF3 SUBUNIT 8 is complex and may involve both protein synthesis and degradation (Karniol et al., 1998; Kim et al., 2001; Seeger et al., 2001; Yahalom et al., 2001), and the results presented might suggest that it has an active role in maintaining the state of primary dormancy. These data suggest it has a function in mature P. avium embryos which is lost if they are exposed to dormancy-breaking conditions for even the relatively short period of 6 wk. The discrepancy between quantitative reverse-transcription PCR and expression patterns suggested by differential display analysis for these genes was probably caused by the nonquantitative nature of the differential display assay. It should be noted, however, that aconitase is a dual function protein which also functions as a translational regulator of iron-responsive element-containing transcripts (Haile et al., 1992; Peyret et al., 1995; Beinert et al., 1996), so varying iron availability may also have influenced levels of this transcript.
This study has provided an insight into the molecular mechanisms involved in the control of dormancy and early germination of P. avium embryos. The low genomic copy number of all genes studied here gives us confidence that the quantitative PCR approach detected transcripts from all relevant P. avium paralogues. The finding that homologues of ABI1/ABI2 and ABI3 are expressed in mature P. avium embryonic axes supports physiological evidence for dormancy in the embryo, and is the first molecular evidence for the existence of embryo dormancy in P. avium. Embryo dormancy therefore appears to be lost some time before the whole seed becomes nondormant, and is clearly separated from expression of the germination-indicator genes ACONITASE and eukaryotic INITIATION FACTOR 3 SUBUNIT 8. These findings emphasize the likely importance of the endosperm and seed coat in controlling germinability; indeed, they imply that secondary dormancy in P. avium seeds may not involve dormancy in the embryonic axis, but rather may prove to be entirely controlled by the surrounding endosperm and seed coat. This finding may demonstrate a major difference in dormancy cycling strategies between the woody P. avium and annual A. fatua in which a link between whole-seed ABI3 expression and secondary dormancy has been made (Holdsworth et al., 1999). It will be of interest to use the quantitative PCR methods developed here to monitor the expression of these indicator genes in P. avium endosperm and seed coat during dormancy-breaking treatments.
Support was provided by the UK Department for Environment, Food and Rural Affairs (DEFRA). We would like to thank Graham Seymour and Stephen Jackson for internal review of this manuscript, James Lynn for statistical analyses, Martin Sergeant for guidance on real-time PCR, Heather Clay for assistance with seed work, and Julia Brüggemann for translations (all HRI Wellesbourne). We also wish to thank Avital Yahalom (Department of Plant Sciences, Tel Aviv University) for advice on the potential functions of eIF3 subunit 8.