Barley mutants with low rates of endosperm starch synthesis have low grain dormancy and high susceptibility to preharvest sprouting


Author for correspondence:
Alison M. Smith
Tel: +44 1603 450622


  • Studies of embryo dormancy in relation to preharvest sprouting (PHS) in cereals have focused on ABA and other hormones. The relationship between these phenomena and the rate of grain filling has not been investigated.
  • A collection of barley mutants impaired in starch synthesis was assessed for preharvest sprouting in the field. In subsequent glasshouse experiments, developing grains were assayed for germination index, sugars, abscisic acid (ABA) and the effects of temperature and exogenous ABA on germination.
  • Mutant lines displayed greater preharvest sprouting in the field than parental lines. In the glasshouse, nondeep physiological dormancy was reduced in developing grains of five lines with mutations affecting proteins involved in endosperm starch synthesis. Inhibition of germination by exogenous ABA and elevated temperature was decreased in developing mutant grains. Sugar concentrations were high but embryo and endosperm ABA contents were unaltered.
  • We reveal a direct connection between grain filling and the extent of grain dormancy. Impaired endosperm starch synthesis directly influences the acquisition of embryo dormancy, perhaps because endosperm sugar concentrations modulate the ABA responsiveness of the embryo. Thus environmental or genetic factors that reduce grain filling are likely to reduce dormancy and enhance susceptibility to PHS.


Seed dormancy is of central importance to seed fitness. Dormancy imposed during embryo development prevents germination before the completion of seed maturation (vivipary). Dormancy that persists beyond seed maturation contributes to mechanisms that permit germination when conditions are favourable for seedling establishment (Bewley, 1997). Seed dormancy is also an important economic trait. Breeders have selected against dormancy in order to promote quick and uniform germination in wheat and barley. However, this means that during adverse weather conditions, grains are susceptible to preharvest sprouting (PHS) (Simpson, 1990; Gubler et al., 2005). Under such conditions the degree of grain dormancy is not sufficient to prevent partial or complete on-ear germination, leading to a reduction in grain quality, yield and viability, with severe economic consequences (Mohan, 2008).

The major factor implicated in the imposition and maintenance of seed dormancy is the phytohormone abscisic acid (ABA). In cereals, exogenously applied ABA inhibits germination (Walker-Simmons, 1987; Wang et al., 1995); viviparous (precocious germination) mutants have defects in ABA biosynthesis or signalling (McCarty, 1995; Holdsworth et al., 1999); and altered rates of ABA catabolism correlate with changes in dormancy status (Kushiro et al., 2004; Okamoto et al., 2006; Gubler et al., 2008; Barrero et al., 2009). Sensitivity to ABA is also important in determining degrees of dormancy in cereals. Changes in dormancy status through development of cereal grains are frequently reported to be correlated with changes in ABA concentrations and/or sensitivity of the embryo to inhibition of germination by ABA (e.g. Walker-Simmons, 1987; Benech-Arnold et al., 1999). Studies in several species indicate that the effects of ABA on dormancy are mediated by complex signalling networks that may also involve GA, ethylene and sugars (Smeekens, 2000; Feurtado & Kermode, 2007; Linkies et al., 2009). It is generally assumed that adverse weather conditions induce PHS in the field through direct effects on ABA and/or GA signalling and hence on embryo dormancy. However, the relationship between susceptibility to PHS, ABA signalling and embryo dormancy in cereals remains poorly understood.

It is well established that adverse weather conditions that increase the incidence of PHS in the field – for example, high temperature and high rainfall – also reduce the rate and extent of accumulation of storage products in the endosperm (i.e. grain filling: Kettlewell et al., 2003; Zahedi & Jenner, 2003). It is therefore possible that grain filling, like PHS, is influenced by environmentally induced alterations in ABA signalling. Consistent with this idea, there are strong correlations between ABA concentrations, grain filling and the capacity for starch synthesis in developing grains of wheat (Ahmadi & Baker, 1999; Yang et al., 2004, 2006), barley (Seiler et al., 2011) and rice (Zhu et al., 2011). However, it is also possible that adverse weather conditions directly reduce the rate of grain filling, and that this reduction then leads to loss of embryo dormancy and PHS. Such a possibility has not been addressed in studies of PHS to date.

To investigate whether grain filling can influence embryo dormancy and susceptibility to PHS, we studied a collection of barley lines carrying mutations that reduce starch accumulation in the grain. The collection includes lines with mutations in three different genes encoding proteins required for normal rates of starch synthesis in the endosperm. It thus allows unequivocal determination of the effect of reduced grain filling, independent of environmental influences on the developing grain.

We observed that the low starch mutants were more susceptible to PHS in the field than their parental lines, consistent with a direct effect of grain filling on embryo dormancy. In glasshouse studies, developing mutant grains displayed greater precocious germination than their parental lines, reduced imposition of dormancy by the endosperm on the embryo, and reduced inhibition of germination by exogenous ABA and by elevated temperature. Our results show unambiguously that reductions in starch synthesis in the endosperm reduce the acquisition of dormancy in the embryo. We thus reveal a direct and previously unreported connection between grain filling, the imposition of grain dormancy and the incidence of PHS.

Materials and Methods

Chemicals and reagents

Phenyl β-D-glucopyranoside, butyl-hydroxy-toluol, methoxyamine hydrochloride, N-methyl-N-trimethylsilyltrifluoroacetamide and ABA were from Sigma-Aldrich, Poole, UK. Deuterated ABA (d4-ABA) was from the NRC Plant Biotechnology Institute, Saskatchewan, Canada. α-Amylase and α-amyloglucosidase were from Megazyme, Bray, Ireland.

Plant material and growth conditions

Barley starch mutants  The lines used in this study and the nature of the mutations (where known) are shown in Table 1. Lines sex7 and sex8 were a kind gift from Professor Jerome Franckowiak, North Dakota State University, USA.

Table 1.   Shrunken-grain mutants and corresponding parental lines
  1. References are to publications describing gene identification. ‘sex’ stands for shrunken endosperm xenia. The original sex7 and sex8 mutants were not derived from Bowman: the mutant lines used here have been backcrossed to Bowman and are thus near-isogenic to this line (Franckowiak, 1995).

Risø13BomiADPglucose transporterPatron et al. (2004)
Risø16BomiADPglucose pyrophosphorylase small subunit 1Johnson et al. (2003)
Risø17BomiIsoamylaseBurton et al. (2002)
Risø29Carlsberg IIADPglucose transporterPatron et al. (2004)
sex7BowmanADPglucose transporterS. Comparot-Moss & K. Denyer, John Innes Centre, unpublished

Glasshouse-grown material  Grains were germinated on damp filter paper at 17°C. Seedlings were transferred to soil (John Innes compost no. 2, 30% grit, adjusted to pH 7.5 with lime, containing thiacloprid for insect control) and grown in a glasshouse with supplemental light in winter (December–April) and without supplemental light in summer (May–August).

Field-grown material  Lines were grown in the field at the National Institute of Agricultural Botany (NIAB), Cambridge, UK, during 2007. Grains were sown on 7 May and harvested by combine on 7 September. Grains were threshed, air-dried and stored at room temperature.

Hagberg falling number (HFN)

Samples of 25 g of grain were milled to flour using a Laboratory Mill 3100 (Perten, Stockholm, Sweden). Falling numbers were determined using the FN1700 apparatus (Perten, Stockholm, Sweden) on samples of 7.0 ± 0.05 g of flour (moisture corrected), according to the manufacturer’s instructions.

α-Amylase activity

Activity was determined using the ‘Ceralpha’ method (Megazyme, Bray, Ireland). Samples of 0.5 g of milled grain were extracted at room temperature in 5 ml Ceralpha Extraction Buffer for 5 min with occasional mixing, then centrifuged at 1000 g for 10 min. The supernatant was transferred to a fresh tube and assayed according to the manufacturer’s instructions. One Ceralpha unit of activity is the amount of enzyme required to release 1 μmol of product (p-nitrophenol) in 1 min.

Germination assays

Ears of glasshouse-grown plants were tagged when the first anther was visible or at the point of ear emergence, whichever occurred first. Grains were harvested 3–8 wk after flowering, and FW and DW were determined and the percentage moisture content was calculated. For each time point, grains were removed from between five and 10 ears to provide 50–60 grains per assay. Grains from the extreme ends of the ear were avoided. Between 10 and 25 grains were sown in 90-mm-diameter Petri dishes on 292 grade filter paper (Sartorius-Stedim Biotech, Aubagne, France) with 3 ml of distilled water or ABA solution as indicated, and incubated at the temperature described in the text. Water or ABA solution was added to prevent drying, typically every 2–3 d. Germinated grains were counted daily, then discarded. The germination index (GI) was calculated using the method of Walker-Simmons (1988) as follows:


where n1, n2... n10 are numbers of grains germinated on day 1 to …10 and 10, 9 and ….1 are the weightings given to grains which have germinated on the first, second and subsequent days. Thus if all grains germinate within the first 24 h, the GI is 1.

Metabolite analyses

Sample preparation for sugar and starch analysis  Samples were frozen in liquid nitrogen immediately after harvesting, freeze-dried and homogenized in a Geno/Grinder 2000 (SPEX Certiprep, Stanmore, London, UK) at 500 rpm for 90 s. Known amounts of material (20–50 mg DW) were extracted three times by incubation in 2 ml 70% (v/v) aqueous ethanol containing 50 mg l−1 phenyl β-D-glucopyranoside (as internal standard) for 60 min. Soluble and insoluble materials were separated by centrifugation. The supernatants were dried using a vortex evaporator and stored at −20°C before analysis for sugars. The pellet was retained for starch assays.

Sugar analysis  Dried samples were derivatized for GC-MS analysis by addition of 100 μl methoxyamine hydrochloride, and the three extracts for each sample (see previous section) were pooled. Samples were incubated at 30°C for 90 min with continuous shaking, followed by addition of 200 μl N-methyl-N-trimethylsilyltrifluoroacetamide and incubation at 37°C for 30 min, then at room temperature for 120 min. Analyses were performed on an Agilent (Agilent Technologies, Wilmington, DE, USA) GC 6890N coupled to a Mass Selective Detector 5973inert. Splitless injections (1 μl) were made using an Agilent 7683 automatic sampler. Conditions of chromatography were: inlet temperature, 250°C; helium carrier gas at 0.9 ml min−1; nominal inlet pressure, 60 kPa; temperature programme of 80°C for 2 min, 10°C min−1 to 380°C then hold for 3 min. The column was a ZB-5HT Inferno (Zebron: 7HG-G015-02, Phenomenex, Macclesfield, UK) 30 m × 0.25 mm × 0.25 μm with a 5 m guard column. Mass spectrometry employed electron ionization in positive mode (70 eV), with a source temperature of 230°C and a quad temperature of 150°C set to the manufacturer’s defaults. Total ion scans were made from 50 to 500 amu and data were processed via the Agilent GC Chemstation software (D.03.00) in conjunction with the NIST Mass Spectral Library, V8.0 (National Institute of Standards and Technology, Gaithersburg, MD, USA).

Starch analysis  Starch was assayed by digestion to glucose followed by enzymatic assay according to Smith & Zeeman (2006).

ABA extraction and analysis  Freeze-dried samples (10–20 embryos; 25–50 mg DW), or single endosperm (30–60 mg DW) were homogenized in a Geno/Grinder 2000 (500 rpm, 90 s), then extracted with 500 μl 80% (v/v) methanol containing 20 or 40 ng d4-ABA (as internal standard) and 50 mg ml−1 butyl-hydroxy-toluol (as an anti-oxidant) at 4°C overnight. The suspension was centrifuged at 3000 g for 10 min and the supernatant removed to a fresh tube. The pellet was re-extracted twice with 80% methanol and the supernatants combined and dried under vacuum. The residue was dissolved in 90 μl 100% methanol, mixed thoroughly and sonicated for 5 min, followed by addition of 600 μl diethyl ether and incubation overnight at 4°C. After centrifugation at 3000 g for 10 min, the supernatant was removed to a fresh tube. The pellet was re-extracted with 60 μl 100% methanol, mixed thoroughly and sonicated, followed by addition of 400 μl diethyl ether. After centrifugation at 3000 g for 10 min, the supernatant was combined with that from the first extraction. Samples were centrifuged at 20 800 g for 5 min then applied to Strata NH2 columns (8B-S009-EAK, Phenomenex) pre-equilibrated with 2 ml diethyl ether. Columns were washed with two 800 μl aliquots of chloroform: isopropanol (2 : 1 v/v), allowed to stand for 5 min, then eluted with 2 ml diethyl ether containing 4% acetic acid. Elutes were dried under nitrogen before analysis using LC-MS.

Liquid chromatography-mass spectrometry analyses were performed on an Agilent 1100 single quadrupole LC-MS system. After resuspension in 200 μl 20% MeOH, samples of 10 μl were subjected to chromatography on a 50 × 2 mm Luna C18(2) 3 μm column (Phenomenex) at 23°C and a flow rate of 400 μl min−1, with the following gradient of methanol in 0.1% (v/v, aq.) formic acid: from 20% methanol to 80% methanol in 6 min, 80% methanol for 1 min, 80% to 20% methanol in 0.2 min, 20% methanol for 2.3 min. Negative electrospray mass spectrometry was in single ion mode, measuring ABA at m/z 263, and d4-ABA at m/z 267; fragmenter voltage of 70 V; spray chamber conditions were 11 l min−1 drying gas at 350°C; 200 kPa nebulizer pressure; 3500 V capillary voltage. Standardization was with a six-point standard curve, spiked with d4-ABA at the same concentration as was present in the samples. Values were corrected for extraction and measurement efficiencies (averaging 55 and 58%, respectively).


Barley starch mutants have enhanced susceptibility to PHS

We examined the occurrence of PHS in field-grown plants of five barley mutants with lesions known to affect the pathway of starch synthesis in the developing grain, and their parental lines. Four of the mutant lines have mutations in genes necessary for the supply of ADPglucose for starch synthesis (mutations affecting cytosolic ADPglucose pyrophosphorylase (AGPase) and the ADPglucose transporter: lines Risø13, Risø16, Risø29 and sex7), and one (line Risø17) has a lesion in a gene encoding the starch debranching enzyme isoamylase1 (Table 1). These lesions all result in low starch contents and shrunken grains (Burton et al., 2002; Johnson et al., 2003; Patron et al., 2004), and both AGPase and ADPglucose transporter mutants have been shown to have reduced rates of starch synthesis during grain development (Doll, 1983; Johnson et al., 2003). Conditions during the field trial were generally cool and wet. There was rainfall on 12 consecutive days during grain filling (from 24 to 14 d before harvest), and further rain 6 d before harvest. The temperature was below 15°C for almost all of the growing period, including the entire grain-filling period (Supporting Information, Fig. S1).

Harvested grains that exhibited PHS were placed into three classes according to the degree of sprouting, from class 1 (mild) to class 3 (severe) (Fig. S2). Four of the mutant lines had a higher proportion of sprouted grains, and more severe sprouting, than their parental control lines. The Risø17 mutant had fewer sprouted grains but more severe sprouting of those grains than its Bomi parent (Fig. 1a). Harvested grains were also analysed for α-amylase activity, production of which is an early event in germination. All of the mutants tested had higher α-amylase activity in harvested grain than the parental lines (Fig. 1b). A further parameter that can reveal the onset of premature germination is the HFN. HFN is a measure of the viscosity of flour suspensions after partial gelatinization. The presence of α-amylase in the flour results in partial degradation of the gelatinized starch and hence low viscosity, measured as a low HFN. HFN was reduced in some of the mutant lines, consistent with the greater degree of PHS and higher α-amylase activity, but it was strongly increased relative to the parental line in Risø13 and Risø16 (Fig. 1c). The amylopectin component of starches of both Risø13 and Risø16 has a different pattern of chain-length distribution from that of the parental line (T.P. Howard et al., unpublished), so the behaviour of the starches in the HFN assay is expected to be intrinsically different from that of the parental line.

Figure 1.

The occurrence and severity of preharvest sprouting (PHS) in field-grown low-starch mutants of barley. On each graph, mutants are grouped with their parental line or equivalent non-mutant background line: thus Risø8, Risø13, Risø16, Risø17 and Risø1508 are derived from Bomi, Risø29 is derived from Carlsberg II, and sex7 and sex8 are in the Bowman background. (a) Grains from mutants with known lesions in genes encoding proteins involved in starch synthesis were examined for PHS. For each line, between 100 and 120 grains were chosen randomly for examination from a sample harvested from multiple plants. The total height of the bar shows the percentage of grains exhibiting PHS. These grains were classed according to the degree of PHS: class 1, coleoptile emerging/enlarging (white section of the bar); class 2, coleoptile lengthening/greening (light grey section of the bar); class 3, coleoptile > 3/4 length of grain/roots developed (black section of the bar). Photographs of examples of each class are in Fig. S2. (b) α-Amylase activity for some of the lines shown in (a). Risø13, Carlsberg II and Risø29 grains were not analysed. Values are in Ceralpha units (CU), and are means of measurements on three independent samples of a single batch of flour for each line. Error bars are SEM. (c) Hagberg falling number for the lines shown in (a). Values are means of measurements on two independent samples of a single batch of flour from 25 g mature grain for each line. The dashed line represents the threshold of the assay (62 s): the viscosity of water. (d) Grains from lines with unknown mutations affecting grain filling were put into three classes according to the degree of PHS, as for (a). (e) α-Amylase activity for the lines shown in (d).

To investigate further the relationship between grain filling and PHS, we examined field-grown material of three more barley mutants defective in grain filling. The mutations in these lines are not known, but all have shrunken grains, two are reported to have low grain starch contents, and one (Risø1508) has been shown to have a reduced rate of starch accumulation during grain development (Table 1; Kreis, 1978; Doll, 1983; Tester et al., 1993). As was the case for the known starch synthesis mutants, all three mutants displayed more severe PHS (Fig. 1d) and higher α-amylase activity (Fig. 1e) than parental lines. Taken together, these data provide clear evidence that impaired starch synthesis in developing barley grains enhances susceptibility to preharvest sprouting.

Developing grains of starch mutants have higher germination indices than those of parental lines

To discover whether increased susceptibility to PHS in low-starch mutants is related to changes in the dormancy of the developing grain, we used a subset of the mutants to study the ability of developing grains to germinate. The subset consisted of four of the lines carrying known mutations in the pathway of starch synthesis, and the three shrunken-grain lines with unknown causal mutations. When grown in a glasshouse, grains from all of the mutants had lower starch contents at harvest than grains of their parental lines (Fig. S3a,b). Intact grains were removed from glasshouse-grown plants and analysed for their ability to germinate at 17°C in the dark (expressed as a germination index; see the Materials and Methods section) at weekly intervals from 3 wk after anthesis. This period covered the end of the grain filling period, through desiccation to harvest maturity (Fig. S3c–f). As expected, germination indices varied between years and seasons. However, all of the mutant lines had higher germination indices than their parental lines at some or all of the time points studied (Fig. 2). For most mutant lines, grains were able to germinate at an earlier developmental stage than those of parental lines, and germination indices were substantially higher than those of parental grains until the maximum value was reached. These data demonstrate that grains of the mutant lines have a lower degree of dormancy than those of parental lines, particularly at early stages of development.

Figure 2.

Germination kinetics of developing grains of low-starch barley mutants. Grains were harvested at intervals from 3 wk after flowering, from glasshouse-grown plants. Assays for estimation of germination indices were conducted in the dark at 17°C. In each graph, mutant lines are compared with their parental line (Bomi or Bowman) grown in the same place at the same time. For experiments in (a), (c) and (e), plants were grown in summer in two different years; for experiments in (b) and (d), plants were grown in winter. Values are means from four independent samples; error bars are SEM.

Embryo dormancy is imposed by the endosperm/pericarp in parental lines, but not in starch mutants

Barley displays nondeep physiological dormancy, meaning that dormancy is imposed upon the embryo by surrounding tissues: embryos will readily germinate if removed from dormant seeds (Simpson, 1990; Benech-Arnold et al., 1999; Finch-Savage & Leubner-Metzger, 2006). We therefore investigated whether the higher germination capacity of developing grains of the low-starch mutants is the result of an altered interaction between the embryo and the adjacent starch-storing tissues. Accordingly we compared the germination of isolated embryos, dehulled grains (grains retaining the embryo, pericarp and endosperm but without the outer glumellea, palea and lemma layers) and intact grains of mutant lines Risø16 and sex7 and their respective parental lines. Isolated embryos of all lines germinated readily, with little difference in germination indices between mutant and parental lines at both 3 and 5 wk after flowering. Germination indices were much lower for intact grains. As observed earlier, intact grains of mutant lines had higher germination indices than their parental lines at both 3 and 5 wk after flowering (Fig. 3). Removal of the hull increased the germination index for all lines. At 3 wk, the germination indices of dehulled grains of parental lines were only half that of isolated embryos. However, dehulled grains of mutant lines had essentially the same germination indices as isolated embryos (Fig. 3a). Thus the capacity of the endosperm and/or pericarp to impose dormancy on the embryo at this developmental stage is absent in low-starch mutants.

Figure 3.

The effect of the embryo covering structures on germination indices. Assays were on grains sampled 3 wk after flowering (a) and 5 wk after flowering (b) from plants grown in a glasshouse in winter, and were conducted at 17°C in the dark. Values are means from three independent samples; error bars are SEM. White bars, isolated embryo; grey bars, grain minus hull (i.e. embryo plus endosperm and pericarp); black bars, intact grain.

The importance of the endosperm and/or pericarp in imposing dormancy on the embryo declined over the next 2 wk, coinciding with the completion of the desiccation process and the death of the starchy endosperm. At 5 wk after flowering, dormancy in both parental and mutant lines was imposed almost exclusively by the hull (Fig. 3b). These data indicate that the impact of the endosperm on embryo dormancy is greatest during the period of starch synthesis, and drops following endosperm death.

Thermodormancy is reduced in starch mutants

Although mature, wildtype barley grains germinate rapidly when temperatures are below c. 20°C (as in the germination assays reported earlier), they germinate progressively less well at higher temperatures (Walker-Simmons, 1988). This so-called primary or thermodormancy is acquired during grain development. To discover whether acquisition of thermodormancy is altered in the low-starch mutants, mature grains were germinated at 28°C. The germination index at this temperature was markedly lower than at 17°C for both mutant and parental lines, but the effect on parental lines was greater than on mutant lines. Thus in the experiment shown in Fig. 4, germination index fell to c. 0.2 at 28°C for the parental line Bomi, but only to 0.55 for the mutant Risø16. Similarly, the germination index was c. 0.02 for the parental line Bowman at 28°C, but c. 0.3 for the mutant sex7. Although absolute values for germination index varied from one experiment to another, further experiments on these and three more mutant lines also showed greater thermodormancy in parental than in mutant lines (Fig. S4).

Figure 4.

The effect of elevated temperatures on germination indices. Assays were on mature grains (8 wk after flowering) from plants grown in a glasshouse in winter, and were conducted in the dark. Values are means from three independent samples; error bars are SEM. White bars, germination at 17°C; black bars, germination at 28°C.

Developing grains from low-starch mutants have normal ABA contents but reduced sensitivity to ABA

We examined whether the altered dormancy characteristics of the mutants correlated with alterations in their endogenous ABA contents, and/or the sensitivity of germination to treatment with exogenous ABA. Embryo and endosperm material was harvested rapidly from grains c. 4 wk after flowering, and frozen immediately in liquid nitrogen. ABA was extracted with methanol and diethyl ether, and assayed by LC-MS. Although there were differences in ABA contents between individual lines, there was no correlation between ABA content and dormancy characteristics. The ABA content was higher in Risø16 grain than in grain of the parental line Bomi, whereas it was lower or unaffected in sex7 grain relative to grain of the parental line Bowman (Fig. 5).

Figure 5.

ABA content of embryo and endosperm from developing grain. Measurements were made on grains sampled 4 wk after flowering from plants grown in a glasshouse in summer. Values are means of measurements on six independent samples; error bars are SEM. White bars, embryo; black bars, endosperm.

By contrast, the germination of developing mutant grains was much less sensitive to ABA treatment than that of parental grains. This difference was particularly striking at 3 wk (Fig. 6a), when the germination indices of the parental lines in the presence of 1 mM ABA were only 0.05 and 0.15 (for Bomi and Bowman, respectively), whereas those for the mutants were 0.37 and 0.53 (for Risø16 and sex7, respectively). These data indicate that the reduced degree of dormancy in the low starch lines is not related to changes in ABA content, but may result from a loss of sensitivity to ABA.

Figure 6.

The impact of exogenous ABA on germination indices. Measurements were made on grains sampled 4 wk after flowering (a) and 8 wk after flowering (b) from plants grown in a glasshouse in summer. Assays were conducted in the dark at 17°C. In (a), white, grey and black bars represent treatments with water, 5 μM ABA and 1 mM ABA, respectively. In (b), white bars represent treatment with water and black bars represent treatment with 1 mM ABA. Values are means of measurements on four independent samples; error bars are SEM.

Developing mutant grains have low amounts of starch and high amounts of sugar

Previous reports show that developing grains of barley starch mutants have elevated amounts of sugar (e.g. Kreis & Doll, 1980; Doll, 1983; Tyyneläet al., 1995) and it is known that the ABA sensitivity of a number of plant processes is strongly influenced by sugar concentrations (Ramon et al., 2008). We therefore compared amounts of starch and sugar with amounts of major organic and amino acids in developing grains of two mutant lines with those of their parental lines. As expected, mutant grains had lower amounts of starch than grains from parental lines (Fig. 7a). These were accompanied by substantially higher amounts of the predominant soluble sugar, sucrose (Fig. 7b). Concentrations of glucose, fructose (Fig. 7c) and mannose (Fig. S5) were also higher in mutant grains. Overall, soluble sugar concentrations were three times higher in the sex7 mutant than in Bowman, and 1.8 times higher in the Risø16 mutant than in Bomi (Fig. 7d). Other major classes of soluble metabolites were not generally affected by the mutations at this stage of grain development. There were no consistent differences between mutant and parental lines in contents of the major amino acids aspartate and glutamine, or the major organic acids citrate and malate (Fig. S5). Thus at 3 wk after flowering low rates of starch synthesis result specifically in elevated contents of soluble sugars, without strongly affecting the content of other major soluble metabolites.

Figure 7.

Starch and sugar content of grains during grain filling. Values are means of measurements on six independent samples taken c. 3 wk after flowering from plants grown in a glasshouse in summer. Error bars are SEM. (a) Starch; (b) sucrose; (c) glucose (black bars) and fructose (white bars); (d) total sugars: sum of sucrose, glucose, fructose, mannose and maltose. Mannose and maltose values are shown in Fig. S5.


The rate of endosperm starch accumulation affects embryo dormancy

Our results reveal that the acquisition of dormancy in developing barley grains is strongly influenced by the rate of starch accumulation in the endosperm. First, low-starch mutant lines of barley lacking specific, individual proteins involved in endosperm starch synthesis showed reduced dormancy during grain development. This was true for four lines, carrying mutations in genes encoding three different proteins that each participate in a different aspect of starch synthesis. Other low-starch mutants, each lacking a different, unknown protein necessary for normal rates of endosperm starch synthesis, also showed reduced imposition of dormancy during grain development. Secondly, the endosperm/pericarp of low-starch lines with known lesions in starch synthesis was unable to impose dormancy on developing embryos, whereas the endosperm/pericarp of parental lines strongly repressed embryo germination. Thirdly, germination of mature grains was less inhibited by high temperature in low-starch mutants than in parental lines. Taken together, these results show that dormancy is imposed upon the developing embryo by endosperm/pericarp factor(s) that are determined by the rate of endosperm starch accumulation.

Mechanisms linking starch accumulation to embryo dormancy

How might the rate of starch accumulation in the endosperm influence the dormancy status of the embryo? Our results point to two promising lines of enquiry. First, altered rates of starch accumulation may influence dormancy via mechanisms involving sugar signalling. There are numerous reports of elevated sugar concentrations in developing grains of low-starch mutants of barley, and we confirmed that developing grains of sex7 and Risø16 have elevated sucrose and hexose concentrations. Sugar signalling is crucial in many aspects of the integration of growth and environmental responses with carbon availability in the plant, and its effects are often interlinked with those of hormone signalling networks (Rolland et al., 2006; Ramon et al., 2008; Cho et al., 2010). However, despite detailed attention to hormone signalling pathways in developing seeds (e.g. Gubler et al., 2005; Finkelstein et al., 2008), there is almost no information on the effects of sugar concentrations on the acquisition of dormancy in developing embryos. This is an important subject for further research.

Secondly, altered rates of starch accumulation may influence dormancy by affecting the sensitivity of the embryo to ABA inhibition of germination. In Arabidopsis seedlings, sugar concentrations influence the expression of transcription factors that mediate ABA-induced gene expression (reviewed in Ramon et al., 2008). Our data indicate that a similar phenomenon may occur in developing barley grain. Relative to parental lines, mutant lines with high amounts of sugar in the endosperm showed reduced acquisition of nonphysiological dormancy during grain development, low degrees of thermodormancy in the mature grain, and insensitivity of the developing embryo to inhibition of germination by ABA. These observations are consistent with the idea that different endosperm sugar concentrations bring about different levels of expression of ABA signalling components in the adjacent embryo, hence different degrees of sensitivity to imposition of dormancy by both endogenous and exogenous ABA. Sensitivity to ABA is regarded as a major factor that determines embryo dormancy in cereals. For example cereal grains with low sensitivity to ABA have lower degrees of thermodormancy than grains that are sensitive to ABA (Corbineau & Côme, 2000). Furthermore, PHS-tolerant and PHS-susceptible wheat cultivars contain similar concentrations of ABA but exhibit different sensitivities to inhibition of germination by exogenous ABA (Walker-Simmons, 1987).

A recent study of the composition of developing grains of the low-starch mutant Risø16 provides strong support for the idea that reduced rates of starch synthesis in developing grains can bring about changes in hormone signalling. Faix et al. (2012) show that relative to the parental line Bomi, Risø16 grains have elevated sugar concentrations and altered concentrations of ABA, cytokinins and transcripts for several ABA-responsive genes at some developmental time points. The largest alteration in expression of ABA-responsive genes occurred at time points at which ABA concentrations were similar in mutant and parental grains, indicating that ABA sensitivity as well as ABA synthesis may be altered in these grains.

Our data are not consistent with the idea that reduced synthesis of ABA is responsible for the reduced dormancy of the mutant grain. Neither mutant had reduced ABA concentrations in endosperm or embryo at 4 wk after flowering. Changes in endogenous ABA concentrations are in any case unlikely to account for the reduced sensitivity of the mutant embryo to suppression of germination by exogenous ABA.

The rate of starch accumulation influences susceptibility to PHS

Our results reveal a causal link between the rate of starch accumulation and susceptibility to PHS in barley. Grains from mutant lines known to have low rates of starch accumulation are more likely to exhibit PHS in the field than grains from parental lines with normal rates of starch accumulation. Increased susceptibility to PHS of low-starch grains is highly likely to result from the low degree of embryo dormancy in these grains. As far as we are aware, this link between starch accumulation, embryo dormancy and susceptibility to PHS has not been reported previously. It has important implications for attempts to understand and control the incidence of PHS: genetic or environmental factors that reduce starch accumulation during grain development are likely to result in low degrees of embryo dormancy and hence high susceptibility to PHS in the field. Indeed, we suggest that the known effects of high temperature on the incidence of PHS may be brought about at least in part by this mechanism. High temperatures during the period of grain filling can decrease dormancy duration and increase susceptibility to PHS in wheat and barley (Rodriguez et al., 2001; Trethowan, 2001; Lunn et al., 2002). There is also ample evidence that high temperatures (> 25–30°C) specifically reduce the rate of starch synthesis in developing wheat grains, most likely as a result of the unusual thermal responses of some of the enzymes of starch synthesis (summarized in Denyer et al., 1994; Jenner, 1994). Thus high temperatures during grain filling may increase the incidence of PHS because they directly reduce the rate of starch synthesis, resulting in reduced embryo dormancy.

Our results shed a radically new light on dormancy and PHS sensitivity in cereals. Most research on these topics has focused on a central and causal role for ABA signalling in the embryo, with no consideration of a role for accumulation of reserves in the endosperm (discussed in Kermode, 2005; Kucera et al., 2005; Finch-Savage & Leubner-Metzger, 2006; Gerjets et al., 2010). Indeed, in several instances ABA concentrations or ABA sensitivity have been proposed to control the rate of grain filling. For example, a set of shrunken-grain (therefore likely to be low-starch) barley mutants that displayed symptoms of premature loss of dormancy were shown to have altered ABA and GA concentrations, leading to the conclusion that these hormones were responsible for the reduced grain filling (Green et al., 1997). The shrunken-grain, low-starch phenotype of the seg8 mutant of barley was attributed to observed perturbations of ABA synthesis during grain development (Sreenivasulu et al., 2010). Modifications to the watering regime during grain development in barley and wheat affect both ABA concentrations and starch accumulation, leading to the suggestion that ABA concentrations control the rate of starch synthesis (Yang et al., 2006; Seiler et al., 2011). We suggest that correlations between ABA signalling, dormancy and the rate of grain filling require re-examination in the light of our discovery that that starch synthesis can itself generate signals that influence ABA sensitivity and the imposition of dormancy in the developing embryo.

There is no doubt that ABA signalling mechanisms are central to the imposition and maintenance of dormancy in wheat and barley grains, and that alterations in these mechanisms can result in altered sensitivity to PHS (e.g. Walker-Simmons, 1987; Morris et al., 1989; Kawakami et al., 1997; Benech-Arnold et al., 1999; Corbineau et al., 2000; Noda et al., 2002; Gubler et al., 2008; Leymarie et al., 2009). However, our results show that these mechanisms may lie downstream of primary signals generated by the rate of grain filling. This factor must be taken into account in future research on the complex interactions between genotype and environment that determine the incidence of PHS in the field.


We thank John Flintham and James Simmonds (John Innes Centre) for their advice on preharvest sprouting, and Lionel Hill and Alan Jones (John Innes Centre Metabolite Service) for LC- and GC-MS analysis, respectively. A.C. and R.M. thank the Nuffield Scholarship programme for supporting their work. This project was funded by Biotechnology and Biological Sciences Research Council Grant BB/E007015/1.