Proteomes of hard and soft near-isogenic wheat lines reveal that kernel hardness is related to the amplification of a stress response during endosperm development

Abstract

Wheat kernel texture, a major trait determining the end-use quality of wheat flour, is mainly influenced by puroindolines. These small basic proteins display in vitro lipid binding and antimicrobial properties, but their cellular functions during grain development remain unknown. To gain an insight into their biological function, a comparative proteome analysis of two near-isogenic lines (NILs) of bread wheat Triticum aestivum L. cv. Falcon differing in the presence or absence of the puroindoline-a gene (Pina) and kernel hardness, was performed. Proteomes of the two NILs were compared at four developmental stages of the grain for the metabolic albumin/globulin fraction and the Triton-extracted amphiphilic fraction. Proteome variations showed that, during grain development, folding proteins and stress-related proteins were more abundant in the hard line compared with the soft one. These results, taken together with ultrastructural observations showing that the formation of the protein matrix occurred earlier in the hard line, suggested that a stress response, possibly the unfolded protein response, is induced earlier in the hard NIL than in the soft one leading to earlier endosperm cell death. Quantification of the albumin/globulin fraction and amphiphilic proteins at each developmental stage strengthened this hypothesis as a plateau was revealed from the 500 °Cd stage in the hard NIL whereas synthesis continued in the soft one. These results open new avenues concerning the function of puroindolines which could be involved in the storage protein folding machinery, consequently affecting the development of wheat endosperm and the formation of the protein matrix.

Endosperm texture, grain hardness, grain softness, near-isogenic lines, oxidative stress, puroindolines, seed development, unfolded protein response

Introduction

Kernel hardness in bread wheat (Triticum aestivum L.), a staple food source for humans, is a major grain quality trait as it determines the physical properties and hence the end-uses of flour. Flour obtained from hard-textured grains is more suitable for bread-making, whereas soft grain flour is better for making pastry and biscuits. Grain hardness is defined by the resistance of kernels to crushing or by the particle size distribution of ground grain. The trait is associated with a major QTL located on the short arm of chromosome 5D, where the genes encoding puroindoline-a (PINA) and puroindoline-b (PINB) co-localize with the hardness (Ha) locus. A deletion or mutation in either Pina or Pinb has been linked to hard kernel texture in wheat (Bhave and Morris, 2008). PINA and PINB are low-molecular-weight (about 13 kDa) cationic proteins (Blochet et al., 1993) whose folded helicoidal structure is stabilized by five disulphide bonds. Some other plant proteins have similar structural characteristics like amylase/protease inhibitors, 2S storage proteins, and lipid transfer proteins (Douliez et al., 2000). The most distinctive feature of puroindolines is a tryptophan-rich domain (TRD), comprising 5 and 3 tryptophan residues in PINA and PINB, respectively (Gautier et al., 1994). The TRD is involved in lipid binding and the destabilization of biological membranes, which gives these proteins their antimicrobial properties (Charnet et al., 2003; Jing et al., 2003).

Puroindolines are specifically expressed in the caryopsis of plants from the Triticeae and Avenae tribes (Douliez et al., 2000). They start to accumulate from 8–12 d after anthesis (DAA), increasing rapidly until 15–18 DAA to reach a maximum around 26–33 DAA (Gautier et al., 1994; Turnbull et al., 2003). It was recently shown that puroindolines and storage proteins followed the same route within the grain, moving from protein bodies to the storage vacuole and finally to become embedded in the protein matrix at the end of endosperm development (Lesage et al., 2011). It was also shown that puroindolines influence storage protein polymer size and polydispersity.

Although numerous reports describe the influence of puroindolines on the technological end-uses of flour, their antimicrobial activities and the genetic bases of hardness, their biological functions during cereal grain development are still not understood. To gain an insight into the cellular function of puroindolines, the metabolic and amphiphilic proteomes of developing wheat grains of two near-isogenic lines (NILs), one hard and one soft, of Triticum aestivum cv. Falcon were analysed. Differences in the proteomes during grain development were related to ultrastructural observations of developing endosperm cells in the genetic context of the absence or presence of Pina. The results suggest puroindolines affect protein folding in the secretory pathway (endoplasmic reticulum) and, consequently, the aggregation of storage proteins and the formation of the protein matrix in drying endosperm.

Materials and methods

Genetic material

Experiments were carried out on hard and soft Australian near-isogenic lines (NILs) of the bread wheat Triticum aestivum cv. Falcon, obtained after six generations of back-crossing with selection for differences in grain hardness only, which was evaluated by particle size index (Symes, 1969). The soft allele was contributed by the bread wheat Triticum aestivum cv. Heron, genetically related to Falcon.

Hard and soft Falcon NILs differ mainly by the presence or absence of the Pina gene; hard Falcon has a Pina-D1b (Pina null) and Pinb-D1a genotype (Giroux and Morris, 1998) and soft Falcon has a Pina-D1a and Pinb-D1a genotype. These NILs were grown during Spring 2008 at INRA Plant Breeding Station under normal greenhouse conditions. Grain hardness values were 98 for the Falcon hard NIL and 42 for the Falcon soft NIL, respectively, evaluated by NIRS (AACC, 1999).

Culture conditions and sampling

Seeds were sown in pots in a greenhouse. After 3 weeks, plantlets were placed in a growth chamber at 6 °C with an 8/16 h day/night photoperiod for 2 weeks. Plants were then grown in a plastic greenhouse with the natural photoperiod and irrigated without fertilizer. At anthesis, wheat ears were tagged and air temperatures were recorded every 30 min in four locations in the greenhouse near the spikes. Daily average temperatures were calculated and summed, allowing the expression of grain development in thermal time (°Cd). For proteomics analysis developing caryopses were collected from the middle of the ear at 180, 300, 500, and 750 °Cd, corresponding in our conditions to 11, 19, 32, and 44 DAA. These four stages correspond to the end of endosperm cellularization, the fast starch accumulation period, the slow starch accumulation period and maturity before desiccation, respectively. For each of the four developmental stages, three samples of 12 kernels (two kernels from six different plants) were collected, immediately frozen in liquid nitrogen, and stored at –20 °C.

Extraction of albumins and globulins from developing kernels

The three samples of each developmental stage and genotype (i.e. 24 samples) were weighed then ground in liquid nitrogen. Before proteins were extracted according to the method of Debiton et al (2011), wholemeal flour was first washed with acetone to eliminate chlorophyll, particularly abundant in young kernels. For this, 500 μl of cold extraction solution (–20 °C) containing cold acetone, 0.07% (v/v) β-mercaptoethanol and 0.34% (w/v) protease inhibitor cocktail (Product number P 9599, Sigma, Steinheim, Germany), which displays a broad specificity for the inhibition of serine, cysteine, aspartic, metalloproteases, and aminopeptidases, were added to 250 mg of freshly powdered material. After vortexing, proteins were precipitated at –20 °C for 24 h. After centrifugation (15 000 rpm, 30 min, 0 °C), the supernatant was discarded. The pellet was washed three times with cold extraction solution until no longer green and dried at room temperature for 2 h, before albumins and globulins were extracted together as described by Debiton et al. (2011).

Extraction of amphiphilic proteins from developing kernels

Amphiphilic proteins were extracted from the pellet after extraction of the albumin/globulin fraction according to Debiton et al. (2011). Amphiphilic proteins present in the 300, 500, and 750 °Cd samples were solubilized in 200 μl of 2-DE solution buffer (4% CHAPS, 7 M urea, 2 M thiourea, 1% IPG buffer pH 3–11, and 70 mM DTT) by vortexing and shaking for 2 h. The suspensions were sonicated at 20 W for 20 s and kept for 30 min at 20 °C. Samples were centrifuged (10 000 g for 5 min at 20 °C). Amphiphilic proteins, mainly hydrophobic membrane proteins, are not easily solubilized and are proportionally more abundant in samples from the early developmental stages. To improve the solubilization of amphiphilic proteins in 180 °Cd samples, two detergents ASB14 at 1.5% and Triton X 100 at 0.5% were included in the 2-DE solution buffer whereas CHAPS concentration was reduced at 2.5%, keeping a total detergent concentration of 4%, without any other change in the protocol.

Proteins in the two fractions were quantified using the Bradford protein assay (Bradford, 1976) on 5 μl from replicates of each developmental stage using bovine serum albumin (Sigma-Aldrich) as the standard.

Two-dimensional electrophoresis (2-DE)

Two proteomic analyses were carried out, one for each fraction of proteins extracted from developing kernel with 48 samples in each analysis (2 NILs×4 stages×3 biological replicates×2 technical replicates). The first dimension of protein separation (isofocusing) was carried out on 24-cm long Immobiline immobilized pH gradient (IPG) DryStrips pH 3–11 NL (GE Healthcare, Uppsala, Sweden). Strips were first rehydrated overnight in a reswelling tray at room temperature with 460 μl of a solution containing 4% (w/v) CHAPS, 7 M urea, 2 M thiourea, 1% (v/v) IPG buffer pH 3–11, 1.2% (v/v) Destreak reagent (GE Healthcare, Uppsala, Sweden), and a few grains of bromophenol blue. Protein extracts (150 μg) were cup-loaded on the acidic side of the strip and isofocused in an Ethan IPGphor II apparatus (GE Healthcare, Uppsala, Sweden) for a total of 90 kVh for the albumin/globulin fraction and 70 kVh for the amphiphilic fraction. After pH equilibration for 15 min in a solution of 6 M urea, 50 mM TRIS-HCl pH 8.8, 30% (v/v) glycerol, and 2% (w/v) SDS containing 1% (w/v) DTT, the proteins were alkylated for 15 min in a solution that was the same except that DTT was replaced by 2.5% (w/v) iodoacetamide. The strips were deposited on SDS-polyacrylamide gels (14% T, 2.1% C) and proteins separated in the second dimension (5 W gel⁻¹ for 30 min and then 10 W gel⁻¹ for 6 h). Gels were stained overnight using G250 colloidal Coomassie Brilliant Blue.

Image analysis

Gels were scanned using a GS-800 scanner and Quantity One software (Bio-Rad, Richmond, VA, USA). Images were analysed using Samespots v 4.1 (Nonlinear Dynamics, UK). Statistical tests were done on mean values computed for normalized volumes of four to six replicates on the basis that the total spot volume from gel to gel should be equal since the same amount of protein was loaded, thus correcting for possible variations between gels. Quantitative differences between protein spots of the two NILs were evaluated for each developmental stage and considered significant when the P-value of the ANOVA test was less than 0.05. Relative abundance of the same spot in hard and soft NILs was calculated using means of normalized volumes.

Protein identification by mass spectrometry

Coomassie-stained spots of interest were excised using pipette tips. Spots were then destained with 100 μl of 25 mM NH₄HCO₃:acetonitrile (95:5, v/v) for 30 min, washed twice in 100 μl of 25 mM NH₄HCO₃:acetonitrile (50:50, v/v), and then dehydrated in 100% acetonitrile. Gel spots were completely dried using a SpeedVac before digestion at 37 °C for 5 h with 15 μl of trypsin (10 ng μl⁻¹; V5111, Promega) in 25 mM NH₄HCO₃. Peptide extraction was optimized by adding 8 μl of acetonitrile, followed by 10 min of sonication.

For LC-MS/MS analysis of peptide mixtures, on-line nanoflow liquid chromatography was performed using the Ultimate 3000 RSLC (Dionex, Voisins le Bretonneux, France) with 15 cm nanocapillary columns of an internal diameter of 75 μm (Acclaim Pep Map RSLC, Dionex). The solvent gradient from 4% to 50% acetonitrile in 0.5% formic acid was run at a flow rate of 300 nl min⁻¹ for 30 min. The eluate was electrosprayed into an LTQ Velos mass spectrometer (Thermo Fisher Scientific, Courtaboeuf, France) through a nanoelectrospray ion source. The LTQ Velos was operated in a CID top 10 mode (i.e. one full scan MS from which 10 major peaks are selected for MS/MS). Raw data files were processed using version 1.2 of the Thermo Proteome Discoverer. For protein identification, the NCBInr Viridiplantae protein database was combined with sequences of possible human keratin contaminants. Variable modifications were oxidation (M) and carbamidomethylation (C). Peptide mass tolerance was set to 1.5 Da and fragment mass tolerance was set to 0.8 Da. Two missed cleavages were allowed. Protein identification was validated when at least two peptides originating from one protein showed significant Mascot™ scores (P < 0.01) (http://www.matrixscience.com/search_form_select.html). When several proteins were identified in the same spot, the one with the best score and the most peptides was chosen. When proteins were identified from only two peptides, spectra were verified to assess their validity.

Specimen preparation for transmission electron microscopy

For ultrastructural observations, samples of endosperm were cut into 1 mm³ pieces with a razor blade and fixed in 3% glutaraldehyde in 0.1 M phosphate buffer pH 7.4 for 4 h at 4 °C. After several washes in phosphate buffer then deionized water, samples were post-fixed in 1% osmium tetroxide in water for 1 h at room temperature and rinsed several times in deionized water.

Tissue samples were transferred to 30% (v/v) ethanol for 30 min at room temperature then dehydrated through a series of ethanol solutions: 50% ethanol for 30 min, then 70%, 85%, 95%, and 100% ethanol for 1 h each. Samples were stored overnight in 100% ethanol. Specimens were then infiltrated with ethanol:London Resin White (v/v) mixtures as follows: 4:1 for 45 min, 3:2 for 60 min, 2:3 for 90 min, 1:4 for 2 h, and then pure resin 3 times for 1 h each. Samples were incubated overnight in pure resin then transferred to capsules filled with resin and allowed to polymerize for 4 days at 55 °C. For electron microscopy, 80 nm sections were prepared using an ultramicrotome (Microm MT-7000, Oklahoma City, USA). The sections were collected on carbon-coated 150-mesh copper grids and stained with uranyl acetate (2% (w/v) in deionized water filtered through a 0.45 μm sieve) by flotation on 50 μl droplets in a dark wet chamber. Finally, the grids were washed in deionized water, air-dried and stored. Samples were observed using a JEOL JEM 1230 transmission electron microscope (TEM) running at 80 keV.

Results

Proteomic analyses of developing kernels

The proteomes of near-isogenic hard and soft wheat endosperms of the Falcon wheat cultivar were compared on 12 gels for each of four key developmental stages: the end of endosperm cellularization (at 180 °Cd), during the fast starch accumulation period (at 300 °Cd), during the slow starch accumulation period (at 500 °Cd) and at maturity before desiccation (at 750 °Cd). Table 1 displays, for each developmental stage, the numbers of albumin/globulin and amphiphilic proteins and the number of significantly different spots that were then picked and identified.

Table 1.

Spots numbers in each proteomic experiment

Developmental stages	Albumins/globulins					Amphiphilic proteins					Total
(°Cd)	180	300	500	750	Total	180	300	500	750	Total
Total number of spots	2113	2156	2175	2058	8502	1404	1370	1312	1306	5392	13894
Varying spotsa	8	8	31	7	54	0	19	7	7	33	87
Picked spots	7	6	27	4	44	0	18	7	7	32	76
Identified spots	4	6	22	0	32	0	11	5	6	22	54

a

Numbers of spots significantly different between hard and soft kernels (P <0.05).

The percentages of albumin/globulin spots that differed significantly between hard and soft NILs were 0.57, 0.58, 2.36, and 0.54% of the total number of spots, for the 180, 300, 500, and 750 °Cd stages, respectively. For the amphiphilic proteins, significantly different spots represented 0, 0.88, 0.32, and 0.34% of the total number of spots for the 180, 300, 500, and 750 °Cd stages, respectively. A total of 54 proteins were identified among the 87 proteins expressed differentially between hard and soft endosperms at the four developmental stages. Images of 2-DE gels showing the positions of identified proteins are supplied in Supplementary Figs S1–Supplementary Data at JXB online.

Of the proteins that were differentially expressed in hard and soft kernels, most up-regulation occurred in the early developmental stages in the hard line and in the late stages of kernel development in the soft line (Fig. 1). Differentially expressed proteins were involved in several cellular processes: protein synthesis and assembly, stress-defence, cell structure, carbohydrate metabolism, photosynthesis, energy, and storage (Table 2).

Table 2.

LC-MS/MS identification of differentially expressed proteins in the developing wheat kernels of the hard and soft NILs

Cell process	Function	Protein name	180 °Cd	300 °Cd	500 °Cd	750 °Cd	Taxonomy	Reference	Protein fraction	Spot number
PS&A	Folding	Protein disulphide isomerase	+48%/H				Aegilops tauschii	gi|47118046	AG	899
PS&A	Folding	HSP70		+98%/H			Hordeum vulgare	gi|476003	Amph	261
PS&A	Folding	Lumenal binding protein cBiPe2		+95%/H			Zea mays	gi|1575128	Amph	291
PS&A	Folding	Protein disulphide isomerase			+46%/H		Triticum aestivum	gi|508975	AG	918
PS&A	Folding	Calnexin			+45%/H		Triticum aestivum	gi|296524380	AG	590
PS&A	Folding	HSP80-2 cytoplasmic			+48%/S		Triticum aestivum	gi|5123910	AG	582
PS&A	Folding	HSP80-2 cytoplasmic			+57%/S		Triticum aestivum	gi|5123910	AG	687
PS&A	Folding	HSP80-2 cytoplasmic			+64%/S		Triticum aestivum	gi|5123910	AG	694
PS&A	Targeting to membrane	Signal recognition particle 54 kDa subunit	+46%/S				Hordeum vulgare	gi|1020000	AG	1514
PS&A	Translation	40S ribosomal protein S7	+79%/S				Hordeum vulgare	gi|49036474	AG	2499
PS&A	Translation	Similar to 60S ribosomal protein L32A	+56%/H				Oryza sativa	gi|115479981	AG	2972
PS&A	Translation	60S ribosomal protein L5-1		+58%/H			Triticum aestivum	BJ290988	AG	1767
PS&A	AA synthesis	Aspartate aminotransferase		+82%/H			Triticum aestivum	gi|164471780	Amph	1192
PS&A	AA synthesis	Aspartate aminotransferase		+87%/S			Triticum aestivum	gi|164471780	Amph	816
PS&A	AA synthesis	Aspartate aminotransferase		+99%/S			Triticum aestivum	gi|164471780	Amph	841
PS&A	Translation	Cold shock protein-1			+66%/S		Triticum aestivum	gi|259439834	AG	2655
PS&A	Translation	Acidic ribosomal protein			+49%/S		Triticum aestivum	gi|32400796	AG	2955
PS&A	Translation	Putative ribosomal protein S18			+66%/S		Triticum aestivum	gi|22204120	Amph	1947
PS&A	Translation	Elongation Factor 1α			+53%/S		Gymnadenia conopsea	gi|89276321	AG	1538
PS&A	AA synthesis	Alanine aminotransferase2			+48%/S		Unknown species	gi|1703227	AG	1160
S-D	Antioxidant enzyme	Peroxidase 1 chain A cristal structure		+88%/H			Hordeum vulgare	gi|57830301	AG	1939
S-D	Antioxidant enzyme	Glutathione S-transferase 19E50		+62%/H			Triticum aestivum	gi|22022400	AG	2254
S-D		Pathogenesis related protein		+60%/H			Unknown species	gi1588926	AG	3205
S-D	Transducer	Guanine nucleotide-binding subunit		+89%/H			Triticum aestivum	BJ281061	Amph	898
S-D	Protease inhibitor	Serpin Z1C			+74%/H		Unknown species	gi|75313848	AG	1408
S-D	Endopeptidase inhibitor	CM 17 protein precursor			+54%/H		Triticum aestivum	gi|21711	AG	2793
S-D	Protease inhibitor	Serpin Z1B			+49%/S		Unknown species	gi|75279910	AG	1364
C S	Cytoskeleton	β-tubulin 3			+55%/H		Hordeum vulgare	gi|146760211	AG	1010
C S	Cytoskeleton	β-tubulin 3			+72%/H		Hordeum vulgare	gi|146760211	AG	1019
C S	Cytoskeleton	Actin			+54%/S		Triticum aestivum	gi|48927618	AG	2009
C S	Lipid storage	Oleosin			+71%/S		Hordeum vulgare	gi|479082	Amph	1921
C S	Cell wall synthesis	UDP-D-glucuronate decarboxylase		+76%/S			Hordeum vulgare	gi|50659030	Amph	918
C M	Starch synthesis	Glucose-1P adenylyltransferase large subunit 1			+92%/S		Unknown species	gi|1707930	AG	1089
C M	Starch synthesis	Glucose-1P adenylyltransferase large subunit 1			+51%/S		Unknown species	gi|1707930	AG	1121
C M	Starch catabolism	Limit dextrinase type starch debranching enzyme			+54%/H		Triticum aestivum	gi|119633045	AG	299
Phot	Carboxylase/oxygenase	Rubisco large subunit		+86%/H			Podocarpus aff. pilgeri	gi|13549392	Amph	590
Phot	Carboxylase/oxygenase	Rubisco large subunit		+91%/H			Deschampsia cespitosa	gi|144583548	Amph	607
Phot		Oxygen-evolving enhancer protein 2, chloroplastic			+46%/S		Triticum aestivum	gi|131394	AG	2391
Ener	Oxidoreductase	Glyceraldehyde-3-phosphate dehydrogenase		+45%/S			Triticum aestivum	gi|148508784	AG	1720
Ener	Oxidoreductase	5,10 methylene-tetrahydrofolate dehydrogenase		+48%/S			Triticum monococcum	gi|115589734	AG	1890
Ener	Ion transport	Vacuolar proton-ATPase subunit A			+77%/S		Triticum aestivum	gi|90025017	AG	682
Ener	Decarboxylase	Putative 2-oxoglutarate dehydrogenase, E1 subunit			+45%/S		Oryza sativa	gi|115474297	AG	325
Ener	Ion transport	Mitochondrial outer membran porin		+89%/H			Unknown species	gi|1172553	Amph	1197
Stor		γ-Gliadin		+78%/H			Triticum aestivum	gi|15148385	Amph	1922
Stor		Puroindoline-a			+332%/S		Triticum aestivum	gi|408873	Amph	2072
Stor		Puroindoline-a			+137%/S		Triticum aestivum	gi|408873	Amph	2078
Stor		HMW glutenin subunit Dy			+67%/S		Triticum aestivum	gi|140169817	Amph	258
Stor		Globulin 3			+50%/S		Triticum aestivum	gi|215398470	AG	839
Stor		HMW glutenin subunit Bx				+69%/S	Triticum aestivum	gi|109452233	Amph	150
Stor		γ-Gliadin				+134%/S	T. turgidum ssp. dicoccoides	gi|209971793	Amph	793
Stor		Grain softness protein-1D				+90%/S	Triticum aestivum	gi|60652224	Amph	1526
Stor		Puroindoline-a				+271%/S	Triticum aestivum	gi|408873	Amph	1697
Stor		Puroindoline-b				+91%/S	Triticum aestivum	gi|201071365	Amph	1708
Stor		Puroindoline-a				+99%/S	Triticum aestivum	gi|408873	Amph	1759

Relative amount of each protein is indicated for the relevant stage of kernel development (180, 300, 500, and 750 °Cd), followed by a letter specifying the genotype displaying the highest mean of the spot volume (H, hard; S, soft). PS&A, protein synthesis and assembly; S-D, stress-defence; C S, cell structure; C M, carbohydrates metabolism; Phot, photosynthesis; Ener, energy; Stor, storage. AG, albumin/globulin fraction; Amph, amphiphilic protein fraction.

Fig. 1.

Numbers of over-expressed proteins at each developmental stage (°Cd) in the hard line (dark bar) and in the soft line (white bar).

Effect of hard/soft genotype on endosperm abundance of translation, folding, stress, and cell redox proteins

It was noted that proteins involved in the processes of folding, translation, and stress account for nearly 50% of the differentially regulated spots in developing hard kernel. At the first stage (180 °Cd), corresponding to the end of endosperm cellularization, protein disulphide isomerase (PDI) and 60S ribosomal protein were more abundant in the hard NIL (+48% and +56%, respectively) whereas signal recognition particle and 40S ribosomal protein were more abundant in the soft line (+46% and +79%, respectively). Each of these proteins is involved in protein translation and folding (Table 2). During the fast starch accumulation stage (300 °Cd), HSP70 and cBiPe2, both protein-folding-related proteins extracted in the amphiphilic fraction, were much more abundant in hard kernel (+98% and +95%, respectively). Two other proteins with functions related to protein folding, PDI and calnexin, were more abundant in hard kernel (+6% and +45%, respectively) during the later slow starch accumulation stage (500 °Cd).

At the 300 °Cd and 500 °Cd stages, nine of the proteins differentially expressed between hard and soft NILs have functions relating to translation. Among them, seven were more abundant in the soft NIL, up-regulated by 49% to 99%: elongation factor 1α, acidic ribosomal protein, putative 18S ribosomal protein, two aspartate aminotransferases, cold shock protein-1, and alanine aminotransferase 2. Two proteins were more abundant in the hard line: 60S ribosomal proteins L5-1and aspartate aminotransferase.

At the 300 °Cd and 500 °Cd stages, significant differences were also found in the abundance of proteins related to stress processes, including cell redox homeostasis, in hard and soft NILs. There was more peroxidase 1 chain A, glutathione S-transferase, pathogenesis-related protein, guanine nucleotide-binding subunit, serpin Z1C, CM 17 protein precursor (a member of the plant amylase/protease inhibitor family) in hard kernels, but less serpin Z1B (Table 2).

The hard/soft genotype affects levels of proteins involved in photosynthesis, energetics, and storage

At the 300 °Cd stage, kernels of both NILs are still green so are probably still photosynthesizing. Two spots corresponding to Rubisco large subunits were more abundant in the hard NIL (Table 2), suggesting peripheral layers of hard kernels have a greater capacity for carbon fixation at this stage. Another chloroplast protein, oxygen-evolving enhancer protein 2, was more abundant in the soft line at 500 °Cd. Four spots involved in energy processes were also more abundant in the soft line, glyceraldehyde 3P dehydrogenase and 5, 10 methylene-tetrahydrofolate dehydrogenase at 300 °Cd, and vacuolar proton-ATPase subunit A and a putative 2-oxoglutarate dehydrogenase E1 subunit at 500 °Cd. The mitochondrial outer membrane porin was over-expressed in the hard line by 89%.

Two spots of glucose-1P adenylyltransferase (AGPase) were more abundant in the soft line during the period of slow starch accumulation whereas a limit-dextrinase-type starch debranching enzyme was more abundant in the hard endosperm. Starch metabolism seems, therefore, to be influenced by the hard/soft genotype.

Some storage proteins were extracted in the amphiphilic protein fraction. It is noteworthy that there was 78% more of a γ-gliadin at the 300 °Cd stage in the hard kernel. In the soft line, HMW glutenin subunit Dy and globulin 3 were more abundant at the 500 °Cd stage whereas HMW glutenin subunit Bx and another γ-gliadin were more abundant at the 750 °Cd stage. The greatest difference between hard and soft proteomes was the PINA content, as there was 332% and 271% more PINA at the 500 °Cd and the 750 °Cd stages, respectively, in the soft NIL than in the hard one. Differential amount of PINB in the NILs at 750 °Cd (+91% in the soft line compared to the hard one) is consistent with ELISA quantification (24.3±1.1 mg PINB and 12.7±0.5 mg PINB in 100 g freeze-dried soft and hard kernels, respectively) (Lesage et al., 2011).

Kinetics of protein accumulation and ultrastructure of developing kernels

The total amount of protein in the albumin/globulin (Fig. 2) and amphiphilic fractions (Fig. 3) increased from 180 °Cd to 500 °Cd in both NILs. From 500 °Cd, however, the kinetics began to differ as protein accumulation reached a plateau in the hard NIL whereas both protein types continued to accumulate in the soft line until 750 °Cd.

Fig. 2.

Albumin/globulin content in developing kernels (μg grain⁻¹). Solid line, Falcon hard; dashed line, Falcon soft. Confidence intervals are from replicated data.

Fig. 3.

Amphiphilic proteins content in developing kernels (μg grain⁻¹). Solid line, Falcon hard; dashed line, Falcon soft. Confidence intervals are from replicated data.

The ultrastructure of developing grain from each NIL was observed by TEM. ER and mitochondria showed an unusual appearance in hard endosperm (Fig. 4). At 300 °Cd, ER appeared as distended beads (Fig. 4B, C) and mitochondria seemed to be swollen showing whitish spots, low density matrix. Multivesicular body (MVB)-like organelles were only observed in hard endosperm (Fig. 4D). It was also found that protein bodies were still merging in the soft endosperm at 500 °Cd (Fig. 5A) whereas the protein matrix was already formed in hard endosperm at the same stage (Fig. 5B). The protein matrix in 750 °Cd soft endosperm (Fig. 5D) appeared similar to that of 500 °Cd hard endosperm (Fig. 5C), showing that hard endosperm completed its development earlier.

Fig. 4.

Unusual organelles appearance in wheat hard endosperm observed by TEM. (A) 300 °Cd soft endosperm (×8000). (B) 300 °Cd hard endosperm (×6000). (C) Swollen mitochondria and distended ER in 300 °Cd hard endosperm (×25 000). (D) Multivesicular body-like in 300 °Cd hard endosperm (×25 000). PB, protein body; ER, endoplasmic reticulum; St, starch; cw, cell wall; Mt, mitochondria; MVB, multivesicular body.

Fig. 5.

Kinetics of the protein matrix formation. (A) Soft endosperm at 500 °Cd, protein bodies are still merging, (×5000). (B, C) Hard endosperm at 500 °Cd, protein bodies are no longer visible (B, ×5000; C, ×10 000). (D) Soft endosperm at 750 °Cd, (×15 000) Appearance of protein matrix is similar to hard endosperm at 500 °Cd (C). PB, protein body; St, starch; cw, cell wall; mr, membrane remnants.

Discussion

The two Falcon NILs phenotypically selected for hardness, theoretically after six generations of back-crosses, are 99.22% identical (Symes, 1969). DArTs polymorphism markers were used to compare the genomes of these two lines revealing that they are 99.38% identical based on 2236 markers tested (data not shown). Two-dimensional electrophoresis of albumin/globulin and amphiphilic proteins revealed that, on average, only 0.69% of protein spots differed significantly between hard and soft NILs. A lower level of evaluated genetic difference between the two NILs, shown by two independent analyses (DArTs and proteomics), compared with the theoretical value, is likely to be related to the low degree of polymorphism between Falcon and Heron. However, the small genetic differences between the two NILs, which were located on different chromosomes, might influence several developmental plant traits not directly related to puroindolines.

The two NILs tested here are known to differ at the Pina locus—the gene is absent in the hard NIL—so it was expected that the major proteomic difference found between the two lines would be in the amount of PINA. The extent of the difference, however, seemed less than expected. This fact can be explained by the presence of another protein at the same location on 2-DE gels of the hard NIL proteome, which is masked by the large spot area of PINA on 2-DE gels of the soft NIL proteome. Indeed, the cytoplasmic ribosomal protein S15a was identified on 2-DE gels of the hard NIL proteome where the PINA spot is located (see Supplementary Fig. S6, spot 1697, at JXB online) on 2-DE gels of the soft NIL proteome (data not shown). This superposing of proteins means that the quantitative difference in Pina expression between hard and soft NILs may be underestimated. For comparison, the levels of PINB in the two NILs concur with results obtained independently by ELISA (Lesage et al., 2011).

A stress response is amplified in developing kernels of the hard NIL

Proteomics analysis indicated that several chaperones and enzymes resident in endoplasmic reticulum (ER) were more abundant in hard kernel compared with the soft kernel, leading to the hypothesis that there is an increase in protein misfolding in the developing hard NIL grain. As spontaneous polypeptide folding is faster than protein synthesis itself, shown in vitro (Landry and Gierasch, 1994), nascent polypeptide chains are prevented from premature folding and/or aggregation by the binding of molecular chaperones from several conserved protein families, initially described as heat shock proteins (HSP), particularly HSP70/BiP. Compared with the cytoplasm, the ER then provides an oxidizing environment that facilitates disulphide bond formation and protein folding in the lumen. This feature is important in wheat endosperm where numerous storage proteins, i.e. α-, β- and γ-gliadins and glutenins contain disulphide bonds in their non-repetitive domains (Shewry and Tatham, 1997).

It was found that there was almost twice as much of both BiP and HSP70 in hard endosperm as in soft endosperm at 300 °Cd, possibly indicating that more proteins are misfolded in hard endosperm. Zhang and Boston (1992) demonstrated that prolamins are misfolded in the maize endosperm of the floury-2 mutant, correlating this with an increase in BiP content. Increases of up to almost 50% in the amounts of other proteins involved in protein folding, like PDI and calnexin, were found in hard kernels at different stages of development. Up-regulation of BiP, HSP70, PDI, and calnexin could indicate that an unfolded protein response (UPR) is activated early in the hard NIL. It is well established that the UPR can be triggered when an imbalance in the ratio between client proteins and chaperones induces ER stress. As demonstrated in yeast and animal cells, the UPR enhances the expression of many genes involved in various processes such as chaperone action, ER associated degradation, N-linked glycosylation, mannosylation, synthesis of cytoskeleton proteins, lipid biosynthesis, cell metabolism, vacuolar function, and trafficking (Jonikas et al., 2009). The UPR in plants has been less documented although recent data confirm similar molecular mechanisms may operate in this response (Vitale and Boston, 2008; Deng et al., 2011; Gupta and Tuteja, 2011).

In our results, differences in the intensities of 11 spots from 180 °Cd to 500 °Cd could relate to differences in translation. For example, down-regulation of elongation factor and ribosomal proteins in the hard line compared with the soft one might be expected to have a direct effect on the translation capacity and cold shock proteins might inhibit translation by their cold shock domains binding to mRNA (Sommerville, 1999). The two 60S ribosomal proteins that were up-regulated in the hard line would normally be recruited after fixation of 40S ribosomal protein by mRNA, but were probably redundant as 40S subunits were down-regulated. Three spots more abundant in soft kernels at 500 °Cd were identified as HSP 80-2, cytosolic chaperones known to form complexes with HSP 70 (Frydman, 2001). Their decrease in the hard line may be coupled to the down-regulation of proteins involved in translation. Activation of the UPR is known to be associated with attenuated translation (Trusina et al., 2008), thus reducing the polypeptide load entering the ER, so it is possible that the differentially expressed translation-related proteins in hard NIL may be an effect of the UPR.

Several features of the hard NIL proteome suggest that the redox status of the endosperm is perturbed. Glutathione S-transferase 19E50, an antioxidant enzyme, was up-regulated in hard kernels, as were peroxidase 1 chain A, involved in the response to oxidative stress, and a guanine nucleotide-binding subunit, a signal transducer acting as regulator of ROS production (Nakashima et al., 2008). Differential expression (described below as up-regulated or down-regulated in the hard line) of several other proteins indicates that the hard endosperm is subject to redox stress. Mitochondrial porin (up-regulated), which modifies mitochondrial permeability is involved in oxidative stress signalling, leading to apoptosis (Jabs, 1999). This possibly enhanced permeability might explain mitochondria’s aspect. Glyceraldehyde 3-phosphate dehydrogenase (down-regulated), an oxidoreductase essential in carbohydrate metabolism, is a major target of oxidative stress (Hwang et al., 2009) regulating transcription (Sheng and Wang, 2009). Cytoskeleton proteins actin (down-regulated) and β-tubulins (up-regulated) are known to be targets of glutathione S-transferase in plants under oxidative stress (Colville and Kranner, 2010). This apparently amplified stress response in hard kernels is consistent with the fact that pathogenesis-related proteins, like serpins and amylase-protease inhibitor CM17, are up-regulated. Incidentally, the UPR mechanism is linked to the cell’s redox status through enhancement of the glutathione pathway and regulation of reactive oxygen species (ROS) production (Sevier and Kaiser, 2008) so the phenotypes of redox stress and UPR could be causally linked in the hard NIL.

The hard/soft genotype affects the kinetics of kernel development

Several lines of evidence from both endosperm protein expression and ultrastructure indicate that the developmental kinetics is altered within a shorter time frame in hard kernels than in soft kernels. While differential expression consisted mostly of up-regulation of proteins in the early developmental stages in the hard kernel, from 500 °Cd onwards it was mostly due to down-regulation of proteins in the hard kernel compared with the soft one (Fig. 1). Up-regulation of Rubisco in the hard NIL at the 300 °Cd stage is a specific example illustrating that the shift in developmental kinetics was not limited to the endosperm but also involved the peripheral layers of the developing grain.

It is known that wheat endosperm undergoes programmed cell death (PCD) in a stochastic manner (Young and Gallie, 2000) from 16 DAA to 30 DAA and that PCD affects the entire endosperm except for the aleurone layer. ROS can activate PCD in animals and plants (Jabs, 1999). In our work, oxidative stress response proteins and ultrastructural features were indicators that PCD began earlier in hard endosperm than in soft endosperm. Synthesis of proteins, both metabolic and amphiphilic ones, also ceased earlier in hard kernels. The down-regulation of proteins involved in energy processes, like vacuolar proton-ATPase subunit A, from 300 °Cd onwards in the hard NIL is consistent with the early onset of apoptosis.

The overall picture is that there is likely to be higher oxidative stress within cells lacking Pina (i.e. hard kernel) resulting from probable failure to restore protein folding homeostasis. This would induce aggregation of misfolded proteins in rough ER causing it to appear distended and undergo accelerated cell death (Frydman, 2001). MVBs were observed only in hard endosperm. As MVBs are involved in degradation (Piper and Katzmann, 2007), this observation strengthens the hypothesis that the hard NIL has a heavier load of misfolded proteins and/or greater degradation of no longer functional mitochondria than the soft NIL.

Putative physiological function of puroindolines

Storage protein polymer size was found to be significantly higher in the hard Falcon NIL compared with the soft one (Lesage et al., 2011), so the absence of PINA might affect the aggregation of storage proteins. It is shown here that folding and cell redox processes probably differ in hard kernels compared with soft ones, such that the folding capacity of hard endosperm cells is limited, leading to the aggregation of prolamins. Some storage proteins were recovered from amphiphilic protein extracts, suggesting that they could interact with puroindolines in the same cellular compartment. Puroindolines have in vitro lipid and membrane binding properties (Wilde et al., 1993; Kooijman et al., 1998; Le Guernevé et al., 1998) owing to their amphiphilic TRD (Evrard et al., 2008; Jing et al., 2003). It was previously found that PINA localizes to vesicular membranes in developing endosperm cells before they merge with protein bodies (Lesage et al., 2011). It is conceivable then that puroindolines could interact both with membranes and prolamins. If so, puroindolines could be seed-specific proteins that contribute to the proper folding or prevent early aggregation of storage proteins entering the ER. Like BiP, they are very abundant and follow the same pathway of intracellular trafficking from the ER to protein bodies (Lesage et al., 2011). According to this hypothesis, absence of PINA might impair the folding capacity of hard endosperm cells inducing ER stress. If an increase in ER stress leads to the UPR cascade of signals, this could explain pleiotropic effects related to the hardness genotype. The amplification of oxidative stress would be expected to result in storage protein aggregation, an increase in protein polymer size, and hardness.

Conclusion

This study challenges the current view that the main function of puroindolines is in defence against pathogens. By evaluating developmental differences between two wheat NILs differing in their grain hardness and the presence/absence of Pina, it was found that stress-related and folding proteins are more abundant in hard kernel and that endosperm development of the hard NIL reached completion earlier than the soft NIL. This suggests that puroindolines affect endosperm development and storage protein polymerization via the protein folding machinery in particular, with a putative amplified UPR accelerating PCD. The interaction between storage proteins and puroindolines would be very important to explore for understanding the protein matrix formation.

Abbreviations

Abbreviations
 
• 2-DE

  two-dimensional electrophoresis

 
• Pina

  puroindoline-a gene

 
• PINA

  puroindoline-a protein

 
• Pinb

  puroindoline-b gene

 
• PINB

  puroindoline-b protein

 
• NILs

  near isogenic lines

 
• DAA

  days after anthesis

 
• LMW

  low molecular weight

 
• HMW

  high molecular weight

 
• UPR

  unfolded protein response

 
• ROS

  reactive oxygen species

 
• PCD

  programmed cell death

 
• TRD

  tryptophan rich domain

 
• SDS

  sodium dodecyl sulphate

 
• CHAPS

  3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulphonate

 
• IPG

  immobilized pH gradient

 
• DTT

  dithiothreitol

 
• ASB14

  amidosulphobetaine-14

 
• LC-MS/MS

  liquid chromatography-tandem mass spectrometry

 
• CID

  collision induced dissociation

We thank Marie-Claire Debote for growing plants. We are grateful for the use of plant growth facilities at the GDEC Unit, INRA Clermont-Ferrand, France and for the use of microscopy facilities at the Biopolymers-Structural Biology platform, INRA Nantes, France. This work was supported by grants from Institut National de la Recherche Agronomique.