• Open Access

Down-regulation of Glucan, Water-Dikinase activity in wheat endosperm increases vegetative biomass and yield

Authors

Errata

This article is corrected by:

  1. Errata: Down-regulation of glucan, water-dikinase activity in wheat endosperm increases vegetative biomass and yield Volume 11, Issue 3, 390–391, Article first published online: 18 January 2013

(Tel +61 2 6246 5245; fax +61 2 6246 5345; email jean.ral@csiro.au)

Summary

A novel mechanism for increasing vegetative biomass and grain yield has been identified in wheat (Triticum aestivum). RNAi-mediated down-regulation of Glucan, Water-Dikinase (GWD), the primary enzyme required for starch phosphorylation, under the control of an endosperm-specific promoter, resulted in a decrease in starch phosphate content and an increase in grain size. Unexpectedly, consistent increases in vegetative biomass and grain yield were observed in subsequent generations. In lines where GWD expression was decreased, germination rate was slightly reduced. However, significant increases in vegetative growth from the two leaf stage were observed. In glasshouse pot trials, down-regulation of GWD led to a 29% increase in grain yield while in glasshouse tub trials simulating field row spacing and canopy development, GWD down-regulation resulted in a grain yield increase of 26%. The enhanced yield resulted from a combination of increases in seed weight, tiller number, spikelets per head and seed number per spike. In field trials, all vegetative phenotypes were reproduced with the exception of increased tiller number. The expression of the transgene and suppression of endogenous GWD RNA levels were demonstrated to be grain specific. In addition to the direct effects of GWD down-regulation, an increased level of α-amylase activity was present in the aleurone layer during grain maturation. These findings provide a potentially important novel mechanism to increase biomass and grain yield in crop improvement programmes.

Introduction

With a total production of over 600 million tons of per annum, wheat, along with maize and rice, provides a significant proportion of the calories and nutrients required to support the global population. The continuing rapid increase in the world’s population, predicted to reach nine billion by 2050, and the growing loss of arable land to urbanisation and alternative end uses such as biofuels are factors contributing to food security concerns. Considerable focus has been devoted to identifying mechanisms for increasing the genetic potential for both biomass and yield in cereals. While a range of tools including marker assisted selection, introgression of genes from diverse germplasms and hybrid technology have been applied in the quest for ever increasing yields, genetic modification (GM) provides a further powerful tool for increasing productivity. The role of GM in improving yield in maize is highlighted by the market penetration of GM in the USA where in 2008 over 80% of the maize planted contained at least one GM trait (Fraley, 2009), whereas none of the wheat planted contained a GM trait. As a consequence of the widening productivity gap between crops, acreages devoted to wheat are under pressure.

Approaches to defining genetic mechanisms for increasing wheat yield through Quantitative Trait Loci analysis and the identification and pyramiding of physiological traits contributing to yield have been employed (Quarrie et al., 2006). Recently, a number of mechanisms for increasing the yield of cereals have been reported including (i) increasing starch synthesis in grain through increasing ADP-glucose pyrophosphorylase activity (Dauvillee et al., 2006; Meyer et al., 2004; Smidansky et al., 2002, 2003), (ii) modifying the levels of brassinosteroids (Wu et al., 2008), (iii) modifying inflorescence architecture (Huang et al., 2009) and (iv) the use of GM to produce disease, pest and herbicide-resistant crops (Cao et al., 1992). This report provides evidence for a novel yield enhancement mechanism resulting from the reduction of an enzyme involved in starch degradation, the Glucan, Water-Dikinase (GWD; EC 2.7.9.4).

Starch represents up to 70% of the wheat grain and is therefore a major source of carbohydrates for the human diet (for review Ball and Morell (2003). Studies of transitory starch degradation in dicots demonstrates that phosphorylation of starch plays a crucial role in initiating degradation of the granule (Smith et al., 2005). All starches are phosphorylated; however, the degree of phosphorylation varies considerably between plant species (Blennow et al., 2002).

Three enzyme activities are involved in the control of starch phosphorylation level. GWD and Phosphoglucan, Water-Dikinase (PWD; EC 2.7.9.5) have been demonstrated to phosphorylate starch in plants (Kotting et al., 2005; Mikkelsen et al., 2005), catalysing the transfer of the β-phosphate from ATP to a glucosyl residue, predominantly in the C-6 and C-3 positions, respectively. A third enzyme, a chloroplastic glucan binding phosphatase, has been shown to be required for normal starch degradation in Arabidopsis leaves (Kotting et al., 2009). Phosphate groups are thought to decrease the local order within the starch granule, facilitating the access of degrading enzymes including Endo- (α) and Exo- (β) amylases and debranching enzymes, which depolymerise starch molecules such that maltose and glucose are transported from the chloroplast to the cytosol for metabolism.

Although the involvement of starch phosphorylation in transitory starch catabolism is established (Zeeman et al., 2007), the role of phosphoglucans in starch metabolism in cereal endosperm has not been experimentally defined. The lack of a marked diurnal synthesis/degradation cycle and the low level of phosphate in the reserve starch of cereals have not provided a compelling rationale for investigating the role of GWD in the cereal endosperm to date. Our interest in manipulating starch phosphate levels to manipulate the physicochemical properties of wheat starch provided the impetus to conduct the current study.

In this study, we describe the impact of grain-specific down-regulation of GWD in wheat endosperm and the unexpected consequences of this endosperm-specific down-regulation of GWD in increasing both plant biomass and grain yield in subsequent generations.

Results

Reducing GWD expression in wheat endosperm decreases starch phosphate content

Two approaches were taken to establish that GWD is expressed in the developing wheat endosperm. Analysis of transcripts in the EST database using the Wheat Estimated Transcript Server (http://www4.rothamsted.bbsrc.ac.uk/whets/cgi-bin/whets1.3/whets_home.pl) established that GWD transcript was present in all tissues analysed including wheat endosperm (data not shown). GWD transcript was demonstrated to be expressed in the wheat endosperm by real-time quantitative PCR. GWD transcript was present during early developmental stages, with transcript levels peaking at 4 days postanthesis (DPA) and then declining during the period when starch synthesis is most active, before increasing again after 20 DPA (Figure 1a).

Figure 1.

 Analysis of Glucan, Water-Dikinase (GWD) expression in transgenic lines. (a) Expression levels of GWD during seed development in Bobwhite 26. Relative abundance of GWD transcripts compared to tubulin and agarose gel showing the relative level of GWD expression during seed development. Each reaction contained the same amount of total RNA extracted from seeds at the developmental stages indicated. (b) Schematic representation of the pGWDBx17 construct for wheat transformation. Bx17HMWG is the high-molecular-weight glutenin promoter. GWD, cDNAF and cDNAR are the sense and anti-sense regions respectively from a cDNA for wheat GWD. Rice SBEI intron 9 is a sequence corresponding to intron 9 of a rice starch branching enzyme I. NOS terminator is from the nopaline synthase gene. (c) Immunodetection of GWD isoforms in wheat endosperm. Proteins were extracted from endosperm of T2 plants and probed with an anti-GWD antibody. Bw26 parental line, taGWD (neg) is a negative segregant from taGWD4. All other lines contain the silencing construct. (d) Phosphate levels in starch isolated from mature seeds of the transgenic lines. Glc6-phosphate content is expressed in ng per mg of starch.*denote the significant difference at P < 0.05.

RNAi was used to down-regulate GWD activity in an endosperm-specific manner using the Bx17 high-molecular-weight glutenin promoter (Figure 1b, Figure S1). To confirm the specificity of the Bx17 promoter, we have performed in silico analysis of the expression pattern using information deposited in the wheat EST database of The Gene Index Project (http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/gimain.pl?gudb=wheat) and the Bx17 unigene set at NCBI (http://www.ncbi.nlm.nih.gov/UniGene/clust.cgi?UGID=2790903&TAXID=4565&SEARCH=Ta.50490). Version 12 of the wheat gene index (update 18-04-10) contains 1 053 965 different EST from all tissues of wheat. All ESTs derived from the Bx17 gene have been found in cDNA libraries derived from developing heads or endosperm. No Bx17 transcripts have been identified in any other tissue. The NCBI unigene set for Bx17 is derived from 251 individual transcripts; of these, 245 are from libraries derived from developing heads or endosperm, four are from libraries derived from both stem and seeds, one is from a specified tissue and one is from stem tissue. This clearly demonstrates that the Bx17 promoter is at least highly grain if not endosperm specific in wheat.

The construct expressed a double-stranded RNA molecule for inhibition of the GWD homologous genes in wheat endosperm, targeting the conserved starch binding domain. Thirteen geneticin-resistant independent transgenic wheat lines containing pBx17-GWD_IR were generated from biolistic-mediated transformation of wheat immature embryos (cv. Bobwhite 26). The presence of the pBx17-GWD_IR construct in the geneticin-resistant plant genomes was confirmed by PCR (data not shown). Western blotting analysis with antisera raised against wheat GWD peptides revealed that the GWD level was significantly reduced in the endosperm of seven T2 transgenic lines. In three of these lines, GWD protein expression was at the limit of detection (Figure 1c). The reduction in GWD protein levels in the endosperm paralleled both reduction in the target endogenous GWD transcript (Table S1) and reduction in the starch phosphorylation levels in the endosperm (Figure 1d). No reduction in GWD transcript levels was observed in the leaf or stem tissues assayed. In addition, analysis of the transgene expression confirmed its seed specificity as construct expression was only detected in transgenic grain (Figure S2). The data confirm that GWD is active in the wheat endosperm and contributes to the phosphorylation of wheat starch at the C-6 position.

Enhanced biomass production and yield in GWD RNAi plants

During the growth of the T1 plants, it was noted that GWD transgenic lines consistently appeared larger than the control lines. T0 plants did not display a comparable phenotype. Comprehensive growth analysis in pots was conducted on five independent events at the T3 generation. The transgenic lines analysed (taGWD1-1, taGWD3-1, taGWD4-1, taGWD4-6 and taGWD4-7) displayed increased early vigour compared to the controls (untransformed Bobwhite 26 and negative segregants). Analysis by PCR confirmed the presence of the construct in all taGWD plants tested. At stages 13 and 15 of the Zadoks scale (third and fourth leaf unfolded, respectively) (Zadoks et al., 1974), all five lines showed an increase in leaf area from 58% to 81% compared to the parental line (Tables 1 and S2). Interestingly, the transgenic lines showed a slight delay in germination compared to the transgenic control lines in tissue culture conditions (Figure 2a–b). This delay was accentuated when the germination occurred in the light. Tiller emergence and flowering time were delayed by approximately 1 week in the transgenic lines. In pot studies, each transgenic plant had 2–3 more tillers at maturity than the controls and produced a greater number of seeds per head and larger seeds, resulting in biomass and yield increases of 20%–54% and 16%–69%, respectively (Table 1 and Table S2). The increase in seed size was accompanied by a concomitant increase in embryo size (Figure 2c–d). Despite the increase in seed weight, the relative proportion of starch per seed was not significantly altered (Tables 1 and S2).

Table 1.   Growth characteristics of transgenic Glucan, Water-Dikinase RNAi plants
 Leaf area (cm2)Dried weight (g/plant)Heads per plantSeed weight (g/100 seeds)Seed productionStarch content (% of total seed weight)
Stage 13 (Zadoks)Stage 15 (Zadoks)Stage 40 (Zadoks)g/headg/plant
  1. For the single pot trials, n = 5 for all lines and for the simulated plot trial, n = 60 for all lines. Zadoks is the Zadoks scale of plant stages *, ** and *** denote the significant difference at P < 0.05, 0.01 and 0.001, respectively.

Single pot studies
 BW26 (parent)10.64 (±0.31)48.44 (±6.86)NDND5.75 (±1.89)3.4 (±0.36)1.33 (±0.21)9.03 (±2.1)64.44 (±2.6)
 taGWD 1-1 T217.5 (±2.10)**76.75 (±10.16)***NDND9.8 (±1.48)**4.86 (±0.34)**1.75 (±0.21)**15.25 (±2.7)*59.63 (±2.4)
 taGWD 3-1 T216.86 (±1.22)**81.96 (±5.59)***NDND8.2 (±1.64)*4.60 (±0.28)**1.65 (±0.24)*11.52 (±3.7)65.57 (±3.43)
 taGWD 4-1 T219.95 (±2.82)**77.91 (±13.77)***NDND8 (±1.22)*4.27 (±0.21)*1.58 (±0.06)*10.56 (±2)66.23 (±3.49)
 taGWD 4-6 T216.43 (±1.07)***87.75 (±6.74)***NDND8.4 (±1.67)*4.82 (±0.33)*1.63 (±0.23)*11.89 (±1.5)*67.43 (±4.78)
 taGWD 5-9 T218.65 (±2.30)***82.88 (±12.57)***NDND8.2 (±1.48)*4.49 (±0.22)*1.70 (±0.13)**12.66 (±2.8)*65.66 (±3.95)
 taGWD 4 (neg)11.87 (±1.75)59.93 (±5.59)NDND7.2 (±0.83)3.89 (±0.12)1.26 (±0.13)9.88 (±1.4)63.85 (±1.64)
Simulated plot trials
 BW26 (parent)9.20 (±1.21)ND451.62 (±124)3.74 (±1.59)1 (±0.18)*3.13 (±0.39)2.22 (±0.36)2.22 (±0.6)63.24 (±1.6)
 taGWD 1-1 T415.74 (±1.36)***ND732.93 (±154)**5.71 (±1.99)**1.45 (±0.25)**3.29 (±0.41)**1.98 (±0.35)2.88 (±0.64)*62.84 (±2.6)
 taGWD 4-6 T412.99 (±1.30)***ND632.72 (±117)**5.76 (±1.83)**1.23 (±0.23)3.48 (±0.35)*2.29 (±0.33)2.82 (±0.58)*68.57 (±3.03)
 taGWD 4-7 T416.13 (±2.04)***ND667.20 (±119)**4.51 (±1.6)*1.40 (±0.22)**3.46 (±0.44)**2.54 (±0.26)2.54 (±0.53)*65.55 (±2.78)
 taGWD 4 (neg)9.80 (±1.73)ND521.69 (±107)3.85 (±1.53)1.17 (±0.19)2.93 (±0.37)2.02 (±0.24)2.38 (±0.55)66.61 (±1.35)
Figure 2.

 Percentage Coleorhiza emergence in both constant light (a) and dark (b) conditions. Numbers indicated are coleorhiza emergence as a percentage of the number of grains that had germinated at the final time point assayed. Error bars are standard error values. **P < 0.01, *P < 0.05; P values were determined with Student’s t test (two-tailed). Embryo size estimation. Areas (c) and lengths (d) are expressed in mm2 and mm respectively.

To investigate the impact of GWD down-regulation under plant growth conditions more representative of field production, large deep tubs were used in greenhouses to allow canopy closure and more extensive root development. A planting density similar to that under field conditions was adopted in bays containing four 2.4-m long rows of wheat plants. To minimise boundary effects, data were collected on the middle two rows only. Over 60 plants for each of the three transgenic lines amongst those with the highest degree of suppression of GWD (taGWD1-1, taGWD4-6 and taGWD4-7) were grown alongside two controls, untransformed Bobwhite 26 and a negative segregant. One-third of the plants for each line were tested, by PCR, for the presence of the construct, 98% of the plants tested were PCR positive, which strongly suggests the insertions were stable. In these lines, plant leaf area at Zadoks stage 13 and also at Zadoks stage 40 (completion of booting) was enhanced by 40%–62%, while aerial dry weight per plant after harvest increased by 20%–54% (Figure 3, Tables 1 and S2). In each of these three transgenic lines, the heads were longer and seed production per spike was increased by 10%–20% (Figure 4). The 100-seed weight of the transgenic lines was also higher than the controls, with the seeds being 5%–10% larger. The combined effect of these traits was to increase individual plant yield by 14% (taGWD4-7), 27% (taGWD4-6) and 29% (taGWD1-1), respectively (Figure 4, Table S2). The statistical analysis package ‘asreml’ (Butler et al., 2009) was used to model the effect of GWD on yield. Based on the predicted values from the model, the yield per row for the GWD lines was 85.1 g and for the non-GWD lines was 67.2 g, indicating from this experiment a yield advantage of 26.6%.

Figure 3.

 Growth comparison under simulated plot conditions Plant growth: (a) Comparison of leaf area between wild-type and T4 GWD RNAi lines at stage 13 on the Zadoks scale (3rd leaf is unfolded). (b) Comparison of total leaf area between Wild-type and transgenic RNAi GWD lines at stage 40 of the Zadoks scale (flag leaf extension). (c) Comparison of total aerial tissue dry weight of control and transgenic RNAi GWD lines. (d) Growth comparison of transgenic lines (taGWD1-1, taGWD4-6 and taGWD4-7), parental line and a negative segregant (taGWD4 ctrl), under simulated field densities in the glasshouse. TaGWD1-1, taGWD4-6 and taGWD4-7 are transgenic plants; taGWD4 (neg) is a negative segregant and bw26 is the parental line. T4 plants were analysed. Leaf area: cm2. Dry weight: g per plant. Data represent the mean ± SE of 70 individual plants for each line assayed. **P < 0.01, *P < 0.05; P values were determined with Student’s t test (two-tailed).

Figure 4.

 Seed production under simulated plot conditions. Seed production: (a) Comparison of head length (mm) between wild-type and transgenic lines. (b) Average seed weight of wild-type and T4 GWD RNAi lines. Seed weight: g per 100 seeds. (c) Total seed yield per plant. (d) Comparison of head and seed morphology between T4 GWD RNAi lines (taGWD1-1, taGWD4-6 and taGWD4-7) and parental line and negative segregant control [taGWD4 (neg)]. Data are mean ± SE from 70 individual plants for each line assayed. **P < 0.01, *P < 0.05; P values were determined with Student’s t test (two-tailed).

An out of season field trial was conducted in 2009/2010 to bulk-up seeds for future replicated field experiments and also to collect preliminary information on the plant development under field conditions. Because of unavoidable delays associated with necessary regulatory requirements, the trial was not planted until mid-spring rather than the standard winter planting window. This resulted in the trial being conducted under high-temperature stress and abnormal photoperiods. A design involving six replicates of the lines involved in the large scale glasshouse trial allowed rigorous analysis of key phenotypes such as leaf area at early stage of plant development, total biomass, grain production and grain size. Despite the unrepresentative regime, significant increases in biomass, seed number per head and seed weight were observed. Transgenic GWD lines consistently showed better establishment and an increase in leaf area per plant at Zadoks stage 13 (3rd leaf unfolded) relative to the controls (from 41% to 75%) (Figure 5a). Mature transgenic plants exhibited similar increases in estimated biomass, 100 grain weight and overall seed production per head (Figure 5b–d) to those seen in glasshouse conditions. Transgenic lines also showed similar delay in tiller emergence and flowering time as seen in glasshouse condition. However, the transgenic GWD plants appeared to produce less tillers than their controls. In particular, no secondary tillers developed in the GWD lines under high-temperature stress. However, the increases in seed size and seed number per head compensated for the decrease in tiller number, and no overall yield penalty was observed (Table S3). A second replicated field trial was conducted in 2010 near Narrabri, New South Wales. In this field trial, different nitrogen (N) treatments were tested. As in the first trial, significant increases in leaf area and individual plant vegetative biomass were observed in the GWD RNAi plants at all developmental stages examined (Figure 5e–g). Nitrogen treatment had no impact on the observed phenotypes. Unfortunately, extreme rainfall events occurred between anthesis and maturity (322 mm of rain in a 6 week period), and no reliable data relating to plot yield or seed size and number was able to be obtained.

Figure 5.

 Influence of the GWD RNAi constructs on wheat plant development and morphology in preliminary field trial [Australian Capital Territory and in North New South Wales (Australia)]. (a) Comparison of leaf area between wild-type and T5 GWD RNAi lines at Zadoks stage 13 (3rd leaf unfolded). Leaf area: cm2. (b) Comparison of total aerial tissue dry weight of control and transgenic RNAi GWD lines. Dry weight: g per plant. (c) Average seed weight of wild-type and T4 GWD RNAi lines. Seed weight: g per 100 seeds. (d) Total grain yield expressed in g per head. Comparison of plant morphology has been established between T5 GWD RNAi lines (taGWD1-1, taGWD4-6 and taGWD4-7) and parental line and negative segregant control [taGWD4 (neg)]. Values are the average of six replicated plots for each of three biological replicates. *, ** and *** denote the significant difference at P < 0.05, 0.01 and 0.001. (ns) is nonsignificant. Influence of the GWD RNAi constructs on wheat plant development and morphology. (e) Total leaf area has been monitored at the Zadoks stage 15 (4th leaf unfolded). (f) Flag leaf area has been measured at anthesis. Leaf area is expressed in cm2. (g) Aerial dried biomass at harvest is expressed in gram. Comparison of plant morphology between T6 taGWD RNAi lines (taGWD4-6), parental line (bw26) and negative segregant control [taCtl (neg)]. Values are the average of five replicated plots for each of three biological replicates. *, ** and *** denote the significant difference at P < 0.05, 0.01 and 0.001 respectively. (ns) is nonsignificant.

To further investigate the cause of the enhanced growth and yield phenotype, the levels of starch and water-soluble carbohydrates (WSC) were determined in different plant tissues at different developmental stages and at the end of the photoperiod when the carbohydrate accumulation is the highest. At 25 days DPA, both starch and WSC levels, including free fructose, glucose and sucrose, were significantly greater in the stems and heads of the transgenic lines (Table 2). Starch and sucrose levels were increased two to threefold, while free glucose and fructose concentrations were increased five to 10-fold. No significant differences were observed in the starch and WSC content of flag leaves or early vegetative tissue.

Table 2.   Carbohydrate composition of the flag leaf and stem at 25 DPA
 Free glucoseFree fructoseSucroseStarch
  1. Values are the average of three technical replicates for each of three biological replicates. Results are expressed in ng of carbohydrates per mg of chlorophyll. * and ** denote the significant difference at P < 0.05 and 0.01, respectively.

BW26 leaf5.75 ± 1.84.77 ± 2.619.21 ± 6.78.02 ± 1.7
taGWD1-1 leaf8.26 ± 2.6*4.41 ± 2.823.03 ± 5.610.41 ± 3.4
taGWD4-6 leaf6.29 ± 1.43.69 ± 1.416.92 ± 35.81 ± 2.3*
taGWD4-7 leaf5.87 ± 2.54.20 ± 1.716.02 ± 4.16.05 ± 1.8*
Bw26 stem3.25 ± 0.613.59 ± 2.932.49 ± 11.517.88 ± 6
taGWD1-1 stem20.02 ± 2.2**29.29 ± 8.9**88.37 ± 19.3**61.55 ± 12.9**
taGWD4-6 stem19.21 ± 5**24.02 ± 4.9**57.5 ± 17.24**33.2 ± 10.4**
taGWD4-7 stem35.25 ±  5.1**91.32 ± 20**81.38 ± 14.6**51.29 ± 14.4**

Decreased activity of GWD is correlated with an increase in α-amylase activity

To establish whether the reduction in the level of GWD in the endosperm resulted in changes in the level of other enzymes involved in starch degradation, we assayed the activities of enzymes implicated in carbohydrate degradation and/or germination. In all the transgenic lines assayed at late development stage (25 DPA), there was a significant increase in total α-amylase activity relative to the controls. A twofold increase was seen in taGWD1-1 and 3-1, while in taGWD4-1, 4-6 and 4-7 the activity was 4.8, 6.3 and 7.8 fold higher, respectively (Figure 6a, Table S4). For all other enzymatic activities assayed (D-enzyme, β-amylase, α-glucosidase, β-glucanase, cellulase and lichenase), either no consistent changes or no statistically significant changes were detected between the different transgenic lines (Table S4).

Figure 6.

 Expression of α-amylase in late mature seeds in the GWD RNAi plants. (a) Alpha-amylase activity in wholemeal grain samples. Amylases activity is expressed in ceralpha unit (CU per g wholemeal). (b) Relative levels of α-amylase activity in different seed fractions. Embryos, endosperm and the seed coat were separated from seeds harvested at 25 DPA and assayed for α-amylase activity. Activity is expressed in CU and compared to the stable xylanase activity. (c–h) Confocal images of aleurone cells isolated from grain of a non-late mature alpha-amylase wheat cultivar, BW26 (c–e) and a transgenic line, taGWD4-6 (f–h) at 25, 28 and 31 DPA, respectively. Cells fluorescing green represent live cells, whereas those fluorescing red are dead. Scale bar = 50 μm, is shown in all frames. Transgenic lines and control lines are those described in Figure 2.

Measurement of α-amylase activity in the different tissues of the developing seed showed that accumulation of α-amylases was localised to the pericarp and aleurone layer of the developing grain (Figure 6b), as almost no amylase activity was detected in the starchy endosperm or in the embryo. To confirm the location of the α-amylase activity on the outer layer of the grain, we used a technique that associates high level of amylase with programme cell death (PCD) in the late maturity α-amylase (LMA) phenotype extensively described in wheat (Mrva et al., 2006). In grains affected by LMA, high levels of the high-PI α-amylase is associated with an increase in PCD in the aleurone layer (Mrva et al., 2006). Using confocal microscopy and staining with propidium iodide, which fluoresces when inside non-living cells but is unable to penetrate an intact cell membrane, the presence of PCD in the aleurone layer was examined. In the control lines, no PCD was observed at 25 or 28 DPA and only low levels of PCD were present at 31 DPA (Figure 6c–e). In the transgenic lines, PCD was observed extensively in the aleurone layer at 25, 28 and 31 DPA (Figure 6f–h).

Discussion

Previous studies have demonstrated a key role for GWD in the phosphorylation of starch as an essential element of transitory starch metabolism (Baunsgaard et al., 2005; Edner et al., 2007; Kotting et al., 2005, 2009; Mikkelsen et al., 2006; Yu et al., 2001). The results described in this study characterise for the first time the effect of a specific alteration of GWD expression in cereal grain. Unexpectedly, the results confirmed not only the direct role of GWD in starch phosphorylation, but also demonstrated that down-regulation of GWD in the developing grain results in profound changes in the growth and development of the subsequent generation, from germination through vegetative growth, inflorescence and grain development.

Glucan, Water-Dikinase phosphorylates starch in wheat endosperm

Phosphorylation of the starch plays a crucial role in initiating degradation of the granule by digestion (Smith et al., 2005). The charged and hydrophilic phosphate groups are thought to increase the starch solubilisation at the granule surface and then facilitate the access of degrading enzymes releasing maltose and glucose from the chloroplast to the cytosol for metabolism (for review see (Blennow et al., 2002).

Although the involvement of starch phosphorylation in transitory starch catabolism in dicots is well understood (Zeeman et al., 2007), the role of phosphoglucans in starch degradation in cereal endosperm remains unknown. To investigate the role of GWD in endosperm starch metabolism, this study used RNAi to down-regulate GWD activity in a grain-specific manner using the Bx17 high-molecular-weight glutenin promoter, a promoter considered to be grain specific in wheat and rice (Furtado et al., 2008). In silico analysis of expression data for wheat supports this specificity profile (data not shown). In the 6 transgenic GWD RNAi wheat lines characterised, endosperm GWD transcripts were reduced by 41%–72% compared to the controls, but no reduction in GWD transcript level was seen in the leaves or stem. Consistent with this finding, no evidence of expression of the GWD RNAi construct was found in any tissue other than the grain. On the basis of this evidence, it is unlikely that the non-grain phenotypes observed in this study are a direct result of expression of the GWD RNAi construct in non-grain tissues. Nevertheless, we cannot completely rule out a potential effect of the GWD RNAi construct on other grain tissues. The use of alternate grain-specific promoters will provide additional information on the role of GWD during grain development. As these lines also contain the nptII gene, there is a possibility that expression of this gene causes the phenotype. However, in glasshouse studies, we have not observed this phenotype in any other wheat transgenics produced that target different traits. Lines for two of these traits were grown in the field in 2009 alongside the GWD RNAi lines, and the GWD RNAi phenotype was not observed (data not shown). It is highly improbable that the phenotype is a result of expression of the nptII gene.

Although GWD is clearly involved in dicots in transient starch degradation in leaf and in starch degradation in tubers, the results presented here show that GWD is expressed and contributes to starch phosphorylation during grain development in wheat. GWD has also been shown to be expressed in developing barley grains, particularly in the early stage of grain development with high levels of expression in the pericarp, leading to the proposition that the pericarp acts as a major short-term storage starch tissue ensuring sink strength of the grain (Radchuk et al., 2009). A starch degradation pathway involving α-amylases and one β-amylase is thought to occur in the pericarp and nucellus as they undergo programmed cell death, facilitating relocation of accumulated sugars into the starchy endosperm.

The presence of increased levels of α-amylase in the seed of GWD RNAi lines parallels the increase in α-amylase seen in LMA susceptible wheat genotypes (Mrva et al., 2006). In both GWD RNAi and LMA genotypes, a programmed cell death response in the aleurone layer accompanies the increase in α-amylase expression. In Arabidopsis, changes in starch phosphorylation have been linked to increased α-amylase expression (Yu et al., 2001). This suggests that there may be a conserved relationship between modified starch phosphorylation and α-amylase expression in wheat as is the case in Arabidopsis (Yu et al., 2005). It is noteworthy that in Arabidopsis the primary pathway of starch degradation involves GWD, PWD and β-amylases, and not α-amylases. Yet in response to decreases in GWD level, α-amylase expression is increased, possibly as a mechanism for compensating for the decrease in starch degradative capacity (Delatte et al., 2006). In rice, it has been established that sugar starvation enhances α-amylase transcription via a protein kinase activation and phosphorylation process involving a sugar response complex including Snf1-related protein kinases SnrK1 and a transcription factor MYBS1 (Lu et al., 2007). Therefore, inactivating GWD in the endosperm during seed development may lead to increase the expression of α-amylase through a feedback response to the reduction in available sugars via reduced starch degradation capacity.

Grain-specific down-regulation of GWD increases biomass and yield in subsequent generations

While the level of starch phosphorylation in cereal endosperm is low relative to transient or tuber starches in dicots, it is nonetheless not surprising that this study confirms that GWD plays a role in starch phosphorylation in the cereal endosperm. However, it was surprising to observe that inactivation of GWD in the developing wheat grain led to increases in grain size, plant biomass and overall yield in glasshouse conditions in subsequent generations of the six GWD RNAi lines tested. The increase in grain size includes an increase in embryo size, but the percentage of starch per total dried weight was unaltered. Increased biomass was observed from an early two leaf stage and maintained throughout plant development in both glasshouse and the two field trials. Three factors contribute to the final yield of a plant: number of seeds per head, seed size and number of tillers per plant. In glasshouse trials, all three factors were enhanced in the transgenic lines relative to the controls. In the first field trial, the increase in seed number per head and the average seed weight was confirmed. Reductions in tiller number, delay in tiller emergence and flowering time for GWD RNAi lines relative to controls were observed in both field trials and under two N regimes. An approximately 1-week delay in flowering time was observed in GWD RNAi lines relative to controls. Delayed flowering and restricted tillering may contribute to the growth and development phenotypes observed, and analysis of the trait in diverse genetic backgrounds will be required to define the interaction between the GWD RNAi-induced phenotype and other aspects of wheat phenology.

Constitutive and grain-specific down-regulation of GWD result in different phenotypes

The phenotypes resulting from down-regulation of GWD observed in this study differ markedly from those seen in dicotyledonous plants, where mutations in GWD or whole plant silencing of GWD led to a starch excess phenotype (Edner et al., 2007; Mikkelsen et al., 2006; Yu et al., 2001). In Arabidopsis, alterations in starch phosphorylation are associated with a marked reduction in starch granule breakdown reducing carbohydrate availability at night. As a result, the starch excess phenotype in leaves and stems is associated with a reduction in growth (Edner et al., 2007). Recently, starch turn over and GWD activity have been functionally linked to cell division and differentiation in embryos, and the oil content in mature seeds (Andriotis et al., 2010). In addition, seed-specific inactivation of plastidial glucose-6-phosphate/phosphate antiporter GPT1 causes a drastic reduction in oilseed associated with an increase in starch content and embryo abortion in Arabidopsis (Andriotis et al.). In Solanum tuberosum, whole plant silencing lines showed a reduction in tuber size compared to their control lines (Vikso-Nielsen et al., 2001). Moreover, some pollen sterility has been reported in Lotus japonicus and tomatoes with reduced levels of GWD (Nashilevitz et al., 2009; Vriet et al., 2010). Very recently, reduction in GWD transcript level in entire Maize plant using RNAi technology leads to an increase in transitory starch in leaf without affecting plant biomass or morphology (Weise et al., 2012). Therefore, while studies in which GWD expression has been reduced through whole down-regulation of plant or mutation have shown adverse effects on plant growth, it is probable that the use of a tissue-specific promoter in the current study rather than any fundamental difference in the overall biochemical or physiological function of GWD between dicots and monocots underpins the differences in the respective phenotypic responses. Given that the monocot and dicot genomes sequenced to date have a single GWD gene, there is limited prospect of identifying mutations that have differential impact in particular tissue types as might be the case for other genes (e.g. the monocot starch synthase I, II and III genes), which have distinct isoforms expressed in different tissue types. Use of tissue-specific promoters to engineer differential tissue-specific expression of GWD will in future provide additional information on the physiological impact of modifying GWD function in particular plant tissues.

Potential mechanisms through which grain-specific GWD down-regulation increases growth and development

No mechanism definitively linking GWD down-regulation in the seed to the range of phenotypic effects described in this work has been proven; however, a number of observations provide avenues for future investigation. Two effects of GWD down-regulation may contribute to enhanced early vigour. Firstly, GWD RNAi lines have increased embryo size. The importance of embryo size as a key contributor for early vigour in cereal has been clearly demonstrated (LopezCastaneda et al., 1996). Amongst a range of factors studied, including emergence, utilisation of seed reserve during germination, shoot biomass and leaf expansion rate, embryo size was the single most important factor to account for differences in seedling vigour amongst the studied cereal species. Secondly, increased amylase levels in the aleurone may lead to enhanced availability of carbohydrate to the emerging plant (while not enhancing the rate of germination), promoting early growth.

The phenotype observed in the GWD RNAi lines generated in this study has similarities to the phenotype obtained through the over-expression of the ADP-glucose pyrophosphorylase (AGPase) in wheat endosperm (Smidansky et al., 2002) and rice (Smidansky et al., 2003). AGPase is the key enzyme controlling flux into the starch biosynthetic pathway. Wheat plants overexpressing AGPase in seeds had increased plant biomass and seed yield (by a combination of increased grain size and grain number). It was hypothesised that carbohydrate availability during seed establishment and development may be a key factor in determining the number of seeds set and the extent of seed filling. Similar to the findings of this study, these plants contained elevated levels of WSC in the stem and also in the first head seeds at different stages of development (Smidansky et al., 2007).

In an analogous manner to the AGPase overexpression plants, photosynthate accumulation in the GWD RNAi lines could be expected to be stimulated by the increase in plant biomass and leaf area, leading to an increase in photosynthates available to the floral meristem, the developing inflorescence and ultimately, the developing grain. This is consistent with the hypothesis that availability of carbohydrates just before and at anthesis is a key determinant of kernel number per unit area (Bindraban et al., 1998). While the mechanism through which leaf area and biomass are stimulated remains to be defined, a number of studies have highlighted the interaction between sugar and phytohormone signalling pathways, emphasising the importance of sugar signalling in plant growth and development (Roitsch, 1999). Starch and sucrose have been defined as the major factors influencing carbon partitioning and plant growth in Arabidopsis (Roitsch, 1999; Sulpice et al., 2009).

Inactivation of GWD in a seed-specific manner would be expected to lead to a reduction in starch degradation during grain development. While the degree to which starch degradation accompanies grain filling is not defined, any significant change in balance between these processes may lead to an increase in sink strength and the transport of additional photosynthates into the grain to promote larger grain size.

While the details of the manner in which the down-regulation of GWD in developing endosperm translates into enhanced biomass and grain yield requires further investigation, this study opens new perspectives on the importance of the interplay between degradation and synthesis in seed storage tissue, affecting not only the development of the seed, but markedly influencing the growth and development of the subsequent generation. This work provides a promising mechanism for enhancing biomass and grain yield in wheat, with potential further application to other crops.

Experimental procedures

Vector construction and wheat transformation

The region corresponding to nucleotides 581–1020 of the full length wheat GWD cloned cDNA (GenBank accession #GU250878) was cloned into pBx17IRcasNOT (Figure S1) in the sense and antisense orientation on either side of intron 9 from starch branching enzyme I. The cloned region is flanked by the Bx17 high-molecular-weight glutenin promoter and the NOS terminator (Figure 1b).

Co-transformation was carried out using the method of Pellegrineschi et al. (Pellegrineschi et al., 2002) and pCMneoSTLS2 (Maas et al., 1997) which contained the selectable marker neomycin transferase (nptII).

Nomenclature of the lines is as follow: ‘Ta’ for Triticum aestivum, ‘GWD’, the first number indicates the transformation day and the last number is the independent event selected from the transformation day. For example, taGWD4-6 and taGWD4-7 are the 6th and 7th independent events selected for the transformation 4.

Plant growth and data collection

All plants were grown in glasshouses at CSIRO Plant Industry, Canberra, Australia, late spring, under natural light with temperatures regulated to 20 °C during the day and 14 °C at night. Plants in single pots were watered daily to maintain pot capacity. Plants in simulated plot trials were watered automatically at a rate equivalent of 10 mm of water every 3 days.

A field bulk-up, under licence DIR092 from the Office of the Gene Technology Regulator (OGTR) (http://www.ogtr.gov.au/internet/ogtr/publishing.nsf/Content/dir092) was conducted at the Ginninderra Experiment Station, Canberra, Australia from September 2009 to January 2010. Three transgenic lines were sown along with the parental line and a negative segregant of one of the transgenic lines. Each plot contained the same number of seeds. Two nitrogen treatments were used with three randomised replicates per N treatment group (Common agricultural practice: 0 and 80 kg N per Ha). Plots were irrigated as needed.

A second field trial was conducted near Narrabri, NSW, Australia, under licence DIR099 from the OGTR (http://www.ogtr.gov.au/internet/ogtr/publishing.nsf/Content/dir099) from July to December 2011. In this trial, a transgenic line was planted along with a negative segregant and the parental line. Each plot contained the same number of seeds. Two nitrogen treatments (added N and no-addition) were used with five randomised replicates per N treatment group. Plots were irrigated as needed.

The presence of the transgene was monitored by PCR, using the primers described below.

Seed weight, seed number, leaves and stems were counted or weighed after drying. Leaf area was estimated using a ‘leaf area metre’ (LI-COR Biosciences Model LI-3000C Portable Area Metre with conveyor belt).

Percentage coleorhizae emergence over time in both constant light and dark conditions in quadruplicate sets

Twenty grains were placed in a Petrie dish with two layers of 8.5-cm Whatman #1 filter paper circles. Then, 5 mL of distilled water were added to each, sealed with parafilm and placed at 20 °C. Plates in constant light conditions received 126 μmol/m²/s light intensity, whereas those in dark conditions were wrapped in two layers of heavy duty aluminium foil to exclude any light. Plates were examined at indicated time points for emergence of coleorhizae.

Thirty imbibed Embryos from mature seeds were dissected after 12 h and photographed using a Moticam 2500 camera on an Olympus SZ51 microscope and analysed using Motic Image Plus 2.0 from Motic China Group Co., LTD.

Statistical analysis

Means for plants from positive and negative groups were compared using the two-sample t test with unequal variances and α = 0.05.

Randomised designs for growth study in pots were developed using the method developed by Coombes et al. (2002).

The R package ‘asreml’ (Butler et al., 2009) was used for statistical analysis to model the effect of GWD on yield. The model consisted of a fixed effect component for GWD, ‘plants per row’ (two rows per line) was used as a covariate, and the individual lines were considered random effects.

PCR primers

PCR primers were designed based on the wheat GWD cDNA reported here. PCR primers used for Real-time PCR were IB_GWD5 for (5′-CCGAATACCCTGTCTGAATT-3′) and IB_GWD5rev (5′-GCTTCACCACCTTGTGC-3′) for GWD; tub1for (5′-AGTGTCCTGTCCACCCACTC-3′) and tub1rev (5′- CAAACCTCAGGGAAGCAGTCA-3′) for tubulin control. PCR primers used for GWD RNAi construct were GWD_IR_for (5′-AAAAGGATCCGGTACCGCCTTCTGGCTCAACAGTTC-3′) and GWD_IR_rev (5′-AAAAGAATTCACTAGTATCACCTTCACCTCCACGAC-3′).

PCR screening was performed using ZLBx17pro3′ (5′-CAACCATGTCCTGAACCTTCACC-3′) and GWD_IR_rev.

Western blot analysis

Western blot analyses were performed to estimate the level of GWD in starch from developing and mature grains. Same amount of extracted proteins for each sample was loaded onto a 7.5% SDS–PAGE and separated according to established procedures (Laemmli, 1970). After electrophoresis, electroblotting was performed using the method described by Dauvillee et al. (2006) using a polyclonal antibody against a selected wheat GWD peptide diluted 1 : 5000 (Genscript, Piscataway, NJ). The complex antigen-antibody was visualised by chemiluminescence (Amersham-Biosciences).

Real-time quantitative PCR

Total RNA was extracted from plant tissues using an Rneasy Plant Mini Kit (Qiagen, Hilden, Germany) following the supplier’s instructions. cDNA synthesis was performed using the Invitrogen first strand synthesis kit with oligo-dT Primer.

Specific amplifications were detected using the Brilliant SYBR Green QPCR MasterMix (Stratagene, LaJolla, CA). Amplifications were performed using the Rotor-gene 6000 from Corbett research Lifescience (Qiagen, Doncaster, Vic., Australia). The specific fluorescence was detected at 520 nm and analysed with the Rotorgene 6 analysis software by comparison with specific standard curves. GWD transcripts were quantified relatively to the transcript level of tubulin present in the total RNA extracted.

Enzymatic assays

Alpha-amylase, β-amylase, α-glucosidase, β-glucanase, cellulase and lichenase activities in wholemeal samples were determined using their respective assay kits from Megazyme International Ireland Ltd, according to the manufacturer instructions. The results were expressed in AU (activity unit) per g flour or extract.

Starch phosphate content was measured using the method described by Fettke et al. (2008).

D-enzyme activity was measured using the adapted protocol used for Arabidopsis thaliana (Zeeman et al., 2004).

Starch measurements were performed following the protocol described in Ral et al. (2008).

Confocal microscopy

Aleurone cells were stained with propidium iodide and carboxyfluorescein diacetate (CFDA). Propidium iodide is non-fluorescent in aqueous solution and produces a red fluorescence upon entering the cells. It is unable to penetrate intact cells and can only enter ruptured cells. CFDA is non-fluorescent in aqueous solution, but is able to penetrate intact cells where it fluoresces green.

Aleurone-programmed cell death was observed by confocal microscopy using Confocal laser scanning microscope (Leica SP2 on DMRB upright microscope: Leica Microsystems Pty Ltd, North Ryde, NSW, Australia) as described (Mrva et al., 2006).

Carbohydrate measurement

Leaves and stems were harvested for each transgenic line and a control at 25 days postanthesis. The starch, total fructose, free hexoses and sucrose were extracted and measured as described in Campbell et al. (1990), respectively. Chlorophyll content was determined using the method described in Winterma (1969). All spectrophotometric measurements were performed using a Thermo Multiscan Spectrum plate reader.

Acknowledgements

The authors acknowledge Frank Gubler, Tony Condon, Colin Jenkins, Richard Richards, Elizabeth Dennis and Jim Peacock for input into experimental designs and for helpful discussions; Jenny Thistleton, Emma Anschaw, Geoff Ellacott, Freddie Loyman and Jos Mieog for their technical assistance; Russell Heywood, Phil Dunbar, Byron Cochran, Greg Constable, Shiming Liu, Chris Tyson, Warwick Stiller and Mick Poole for assistance with the field trials and Rosemary White and Mark Talbot for the microscopy. Financial support for the research was provided by the Grains Research and Development Corporation of Australia.

Ancillary