Proteomic Analysis of Near-Isogenic Sunflower Varieties Differing in

Near-isogenic sunflower lines containing 25% (inbred RHA280) and 48% (RHA801) oil by seed dry mass were comparatively analyzed in biological triplicat...
0 downloads 0 Views 881KB Size
Download PDF
Proteomic Analysis of Near-Isogenic Sunflower Varieties Differing in Seed Oil Traits Martin Hajduch,†,# Jill E. Casteel,† Shinxue Tang,§ Leonard B. Hearne,‡ Steve Knapp,§ and Jay J. Thelen*,† Department of Biochemistry and Statistics Department , 109 Christopher S. Bond Life Sciences Center, University of MissourisColumbia, Columbia, Missouri 65211 and Department of Agronomy, University of GeorgiasAthens, Athens, Georgia 30602 Received March 16, 2007

Near-isogenic sunflower lines containing 25% (inbred RHA280) and 48% (RHA801) oil by seed dry mass were comparatively analyzed in biological triplicate at 18 days after flowering using two-dimensional (both pI 3-10 and 4-7) Difference Gel Electrophoresis. Additionally, two inbred lines varying in oleic acid content, HA89 (18% oleic) and HA341 (89% oleic), were also analyzed in the same manner. Statistical analyses of these sunflower lines was performed beginning with fitting a mixed effects linear model to the log-transformed optical volume of each spot to account for gel variation, followed by testing the significance between varieties for mean transformed optical spot volumes. The p-values from the spot analysis procedures were then used to find the cutoff point for differential expression using a 10% false-discovery rate (FDR). Comparison of the oil content and oleic acid composition lines revealed 77 and 42 protein spots below the 10% FDR cutoff, respectively, and were therefore declared differentially expressed. Liquid chromatography-tandem mass spectrometry analysis of each of these protein spots resulted in assignments for 44 and 17 spots, respectively. Fructokinase, plastid phosphoglycerate kinase, and enolase proteins were determined to be up-regulated in the high oil line, while phosphofructokinase, cytosolic phosphoglucomutase, and cytsolic phosphoglycerate kinase were up-regulated in the low oil variety. Additionally, four activities involved in amino acid synthesis were up-regulated in the low oil variety in addition to 12S storage proteins and a protein similar to legumin storage protein. Interestingly, two 2-DE spots identified as 14-3-3 proteins were found to be up-regulated in high oleic acid variety. Alteration of glycolytic and amino acid biosynthetic enzymes, as well as storage protein levels, suggests seed oil content is tightly linked to carbohydrate metabolism and protein synthesis in a complex manner. Keywords: oilseeds • seed oil content • DIGE • quantitative proteomics • statistical proteomics

Introduction Oilseed crops vary considerably in oil content ranging from 16% to 45%.1 This wide variation is in part due to anatomical and morphological differences, particularly with respect to seed coat, embryo, and endosperm contribution to overall seed structure.2,3 Differences in metabolic regulation leading up to and including de novo fatty acid synthesis,4,5 as well as structural packaging of storage triacylglycerol,6 may also contribute to this wide variation. However, these differences can be difficult to unveil amidst a background of evolutionarilydivergent traits including embryo size. To better understand this complex metabolic trait, targeted analysis of isogenic or * To whom correspondence should be addressed: Jay J. Thelen, University of MissourisColumbia, Department of Biochemistry, 109 Bond Life Sciences Center, Columbia, Missouri 65211, Phone, 573 884-1325; fax, 573 884-9676; e-mail, [email protected]. † Department of Biochemistry, University of MissourisColumbia. ‡ Statistics Department, University of MissourisColumbia. § Department of Agronomy, University of GeorgiasAthens. # Present address: Institute of Plant Genetics and Biotechnology, Slovak Academy of Sciences, Nitra, Slovak Republic.

3232

Journal of Proteome Research 2007, 6, 3232-3241

Published on Web 06/20/2007

near-isogenic cultivars from the same plant species is a more direct approach. Sunflower has been bred for over 60 years primarily for increased seed kernel (referred to as achene in sunflower) size and oil content. Edible sunflower (Helianthus annus) oil is characterized by high levels of oleic acid, although low oleichigh linoleic varieties are also cultivated. The availability of near-isogenic cultivars varying in achene oil content, from 25% to 48%, as well as oleic acid composition (18-89%), presented an opportunity to query the biochemical basis for these agronomically important traits which in the case of oil content has proven to be highly elusive using traditional biochemical and molecular genetic approaches. Additionally, sunflower germplasm isogenic for high and low oleic acid was previously shown to involve a mutation in the stearoyl-acyl carrier protein desaturase;7 however, segregation analyses suggest additional genes influence this trait. This finding is surprising as the complete pathway for de novo fatty acid synthesis is known in plants, although there is a general lack of understanding about regulatory factors that may influence metabolic traits. 10.1021/pr070149a CCC: $37.00

 2007 American Chemical Society

Proteomic Comparison of Sunflower Seed Oil Varieties

As a contemporary systems biological approach, proteomics allows one to directly investigate changes in protein expression rather than predict them from transcript expression data. Moreover, some proteomics approaches including two-dimensional electrophoresis (2-DE) allow the direct detection of differences in post-translational status arising from the host of protein modifications in eukaryotic cells. A variation on the traditional 2-DE approach employs the use of sensitive, fluorescent Cyanine dyes that covalently react with primary amines through an activated N-hydroxysuccinimide ester moiety, also referred to as Difference Gel Electrophoresis (DIGE).8,9 This technology simplifies the challenges of spot matching among biological samples and also dramatically reduces the amount of protein required for gel loading which results in higher quality spot resolution and appearance. Although DIGE has been employed in bacterial, animal, and yeast quantitative proteomics investigations, there are very few reports in plants10,11 and no example of this approach to analyze near-isogenic germplasm for biomarkers responsible for quantitative trait loci such as oil content and composition.

Materials and Methods Sunflower Germplasm and Genetic Background. Four sunflower (H. annuus L.) inbred lines RHA280 (PI 552943),12 RHA801 (PI 599768),13 HA89 (PI 599773), and HA341 (PI 509051)14 were grown in the field at Corvallis, OR, in 2002. Seeds of each were acquired from the USDA National Plant Germplasm System North Central Regional Plant Introduction Station (http://www.ars-grin.gov/). RHA280 has low seed oil concentrations, brown-striped confectionery-type seeds, an opaque white hypodermis (Hyp Hyp), and unpigmented phytomelanin (p p), whereas RHA801 has high seed oil concentrations, black oil-type seeds, a transparent hypodermis (hyp hyp), and black phytomelanin (P P).12,13,15 RHA280 and RHA801 are the parents of a recombinant inbred line (RIL) mapping population segregating for apical branching (B),16 pericarp hypodermis pigment (Hyp),17 and pericarp phytomelanin pigment (P) loci, in addition to quantitative trait loci (QTL) for several genetically correlated seed traits, for example, seed oil concentration, 100-seed weight, and kernel-to-pericarp weight ratio.15,18,19 Tang et al.15 reported seed oil concentrations of 254 g/kg for RHA280 and 481 g/kg for RHA801 from seed samples harvested from plants grown at Corvallis, OR, in 2000 and 2001. HA341, a mutant (high-oleic) inbred line, was developed by selecting among BC1F2 and BC1F3 progeny from a cross between the wild-type (low-oleic acid) inbred line HA89 and Pervenets, an open-pollinated high-oleic acid cultivar.12,20 Pervenets (PI 483077) is the source (donor) of a dominant oleoylphosphatidylcholine desaturase (FAD2-1) mutation necessary for producing high-oleic acid concentrations in sunflower seed oils.7,21,22 The fatty acid concentrations of HA89 and HA341 were previously quantified by gas chromatography.7 HA89 produced 181.1 ( 12.8 g/kg oleic acid and 697.8 ( 15.8 g/kg linoleic acid, and HA341 produced 886.8 ( 12.8 g/kg oleic acid and 27.8 ( 10.8 g/kg linoleic acid. The primary inflorescences of six plants per inbred line were bagged pre-anthesis and throughout flowering to exclude foreign pollen and prevent outcrossing. Physiologically immature (developing) kernels were harvested into liquid nitrogen 18 days after flowering (DAF) and stored at -80 °C. Protein Isolation, CyDye Labeling, and Two-Dimensional Gel Electrophoresis. Total protein was isolated from developing

research articles seed and subjected to two-dimensional protein electrophoresis (2-DE) as described previously.23 Protein quantification was performed in triplicate using the Coomassie dye binding assay (Bio-Rad, Hercules, CA) against standard curve of chicken γ-globulin. For difference gel electrophoresis (DIGE), the protein pellet was reconstituted in lysis buffer (30 mM Tris HCl, pH 8.5, 7 M urea, 2 M thiourea, and 4% (w/v) CHAPS) with gentle mixing for 30 min at room temperature followed by centrifugation for 15 min at 14 000g to remove insoluble material. Then, 50 µg of protein was taken and adjusted to final volume of 10 µL with lysis buffer. One microliter of CyDye (100 pmol) was added and incubated on ice for 30 min in the dark. High oleic acid (HA341) and oil (RHA801) inbreds were labeled with Cy5 and low oleic acid (HA89) and oil (RHA280) inbreds with Cy3. The labeling reaction was terminated with 1 µL of 10 mM lysine on ice for an additional 10 min in the dark. For isoelectric focusing (IEF), 50 µg of protein was mixed with equal volume of 2× sample buffer (8 M urea, 130 mM DTT, and 4% (w/v) CHAPS), incubated 10 min on ice, mixed with 2.25 µL of IPG buffer (GE Healthcare, Piscataway, NJ), and adjusted to total volume of 450 µL with sample buffer. For preparative, colloidal Coomassie G-250 (CBB) stained gels, the protein pellet was resuspended in IEF sample extraction buffer (8 M urea, 2 M thiourea, 2% (w/v) CHAPS, 2% (v/v) Triton X-100, and 50 mM DTT) with gentle mixing. For IEF, 1 mg of total protein was mixed with 2.25 µL of appropriate IPG buffer in a total volume of 450 µL of IEF extraction buffer. DIGE Imaging, 2-DE Analysis, and Protein Identification. DIGE gels were scanned using an FLA-5000 dual photomultiplier laser scanner equipped with a Cy3/Cy5 dual emission filter (FUJI Medical, Stamford, CT). The photomultiplier tubes of this scanner collect 65 536 shades of gray (16 bit grayscale). MultiGauge software (FUJI Medical, Stamford, CT) automatically assigned a false color to each channel (Cy3, channel 1, red; Cy5, channel 2, green). DIGE images (16 bit TIFF) were analyzed using ImageMaster 2-D Platinum software (version 5.0, GE Healthcare, Piscataway, NJ). For pI 3-10 DIGE gels, only pI 7-10 region was analyzed; no spot groups were observed from pI 3-4. Protein abundance was expressed as relative volume according to the normalization method provided by ImageMaster software. Statistically significant, differentially expressed proteins were selected based on statistical modeling and false-discovery rate (FDR) as described below. The differentially expressed protein spots (FDR e10%) were excised from corresponding preparative CBB-stained 2-DE gel and trypsin-digested as described previously.23 Liquid chromatography-tandem mass spectrometry using a ProteomeX LTQ ion trap (Thermo-Fisher, San Jose, CA) was performed according to manufacturer’s instructions for high-throughput protein identification as described previously.24 Statistical Analysis. Spot intensities are surrogates for the concentration of the proteins and peptides in the experimental samples. By the construction of a statistical model of the experiment, it is possible to account for both the random and fixed influences of experimental conditions and differences between varieties on the experimental results. The spot intensities from different gels but the same experimental varieties can be combined to give a refined estimate of the optical volume for the matched spots by treatment. The statistically corrected Journal of Proteome Research • Vol. 6, No. 8, 2007 3233

research articles spot intensities can then be compared to determine a statistical measure of equality of protein concentration by treatment for each spot. Given the optical volume measurements for matched spots, the first task is to test if a variance stabilizing transform of the spot volume data is suggested. The Box-Cox procedure25 was applied to the experimental data Y. This procedure finds the λ such that the variance of the transformed data will be minimized. That is, Y′ ) Yλ , where s2{Y′} e s2{Y}. This procedure was used to find the optimal λ, which was λ ) 0, which implies a logarithmic transformation. The observed optical volume was then transformed Y′ ) log2(Y). Next, a mixed linear model was fitted to the data. There is a random effect term in the model to account for the differences in optical volume for spots across gels. There is also a fixed effect term in the model to account for the difference between sunflower varieties, the term that is of experimental interest. This gives an expected optical volume for each spot for each variety. Also, t tests for all pair wise variety comparisons were computed. This associates with each comparison a p-value for each spot that is testing the hypothesis that the amount of protein in any given spot from both cultivars (or treatments) is equal. Sorting the list of spots into ascending order based on the p-values gives an ordered list of proteins that are present in different amounts from the statistically most likely to the least likely. Given an ordered list of experimental spots, we can define the false discovery rate (FDR)26,27 to be the probability that entries in the set of the first j spots are in error. If φ is the desired FDR, and j is the largest j such that p-valuej e φj/n, then φ is the FDR for the first j list entries. The FDR is the probability that an entry in the ordered set of j proteins is not present in different concentrations between experimental varieties. FDR specifies the error rate that we expect in the first j experimental spots. Database Querying with Mass Spectral Data. Tandem mass spectral data were searched in parallel against three databases: the National Center for Biotechnology Information (NCBI, ftp://ftp.ncbi.nih.gov/blast/) non-redundant database; NCBI UniGene database for Helianthus annuus (ftp://ftp.ncbi.nih.gov/repository/UniGene/Helianthus_annuus/); and The Institute for Genomic Research (TIGR) tentative consensus (TC) database for Helianthus annuus (http://www.tigr.org/tigrscripts/tgi/T_index.cgi?species)sunflower) (March 2005 curation for all three databases). For each excised 2-DE spot, the protein assignment with the highest number of unique matching peptides was selected. If two protein assignments gave the same number of matching peptides, the assignment with higher protein coverage was selected. Analysis of LC-MS/MS data was performed as described earlier.24 Briefly, database searches were performed on a local licensed copy of Sequest28,29 as part of the BioWorks 3.1SR1 software suite (ThermoFisher, San Jose, CA). Search parameters were set as follows: enzyme, trypsin; number of internal cleavage sites, 2; mass range, 400-1600; threshold, 500; minimum ion count, 35; peptide mass tolerance, 1.50 Da; variable modifications, oxidation (M); static modification, carboxyamidomethylation (C). Matching peptides were filtered according to correlation scores (XCorr at least 1.5, 2.0 and 2.5 for +1, +2, and +3 charged peptides, respectively). For all protein assignments, a minimum of two unique, nonoverlapping peptides was required. 3234

Journal of Proteome Research • Vol. 6, No. 8, 2007

Hajduch et al.

Results 2-DE Gel Analyses Reveal that the Majority of Sunflower Seed Proteins Migrate between pI 4 and 7. To detect differentially expressed proteins, total proteins isolated from 18 DAF developing sunflower seeds of RHA280 (low oil), RHA801 (high oil), HA89 (low oleic acid), and HA341 (high oleic acid) inbreds were analyzed by 2-DE DIGE in biological triplicate (Figure 1). Initial experiments were performed using broad range, pI 3-10 immobilized pH gradient (IPG) strips, which revealed that the majority of sunflower seed proteins migrated between pI 4 and 7. No reproducible 2-DE spots were observed from pI 3 to 4 and the spots between pI 7 and 10 represented less than 20% of the total spot volume. Proteins within pI range of 4-7 exhibited significant overlap, which is known to interfere with quantification and identification.30 To alleviate this problem, medium range pI 4-7 IPG strips were employed to quantify and identify the bulk of differential protein spots. Using this strategy, it was possible to resolve proteins within pI 4-7 clearly, and the spot overlap problem was minimized, a critical first step for differential proteomics screens. Application of Statistical Modeling and False Discovery Rate to Identify Differentially Expressed 2-DE Protein Spots. Spots were detected and quantified independently from each CyDye TIFF image using ImageMaster Platinum. Protein expression data were subjected to comprehensive statistical analysis as outlined in Figure 2. To ensure quality of data, protein spots needed to be present in each of the biological triplicate gels. Raw protein spot volumes were converted to relative volumes to account for sample loading variation and efficiency differences between the two fluorophores. Relative volumes were used to test for the optimal variance-stabilizing transform of the spot volume data. The Box-Cox procedure was applied to the experimental data Y to find the optimal variance stabilizing transform, the logarithm. Next, a mixed linear model was fit to the transformed data. Normally, a fixed effects term for the dye effect would be included in the model, but for this experiment, there was no dye-swap in the experimental design. However, an earlier study showed similar intensities and variation between NHS-activated Cy3 and Cy5 dyes, suggesting dye labeling and variation is inconsequential compared to the inherent technical and biological variation associated with 2-D DIGE.9 However, this is not true for all fluorophores, as recent studies indicate that inclusion of a Cy2 internal standard introduces more variation than simple two-dye analyses, presumably due to the poor molar absorptivity and high background observed with Cy2.31-33 In this statistical model, there are terms to account for the differences in optical volume for spots across gels and also a fixed effect term in the model to account for the treatment effect, the term that is of experimental interest. This associates, with each treatment comparison, a p-value for each spot that is testing the hypothesis that the amount of protein from each treatment for a spot is equal. By sorting the list of spots into ascending order based on the p-values gives an ordered list of proteins that are present in different amounts from the statistically most likely to the least likely. This list of proteins is available for gels of pI 3-10 and pI 4-7 of RHA280/801 and HA89/341 sunflower inbreds (Supplemental Tables 1, 2, Supporting Information). The list of protein spots sorted in ascending order by p-values was used to select 2-DE spots within 10% FDR upper bound. In the DIGE comparison of low versus high oil content lines (RHA280 vs RHA801), 69 2-DE spots were found within 10%

Proteomic Comparison of Sunflower Seed Oil Varieties

research articles

Figure 1. Two-dimensional difference gel electrophoretic (2-DE DIGE) analysis of developing seed from sunflower inbred lines. (A) 2-DE DIGE analysis of total protein isolated from developing seed (18 DAF) from low oil (RHA280) and high oil (RHA 801) sunflower inbreds. (Left) DIGE overlay image of Cy3-red/Cy5-green channels. Cy3, 50 µg of total protein isolated from low oil RHA280 inbred. Cy5, 50 µg of proteins isolated from high oil RHA801 inbred. (B) 2-DE DIGE analysis of total protein isolated from low oleic acid (HA89) and high oleic acid (HA341) sunflower inbreds. (Left) DIGE overlay image of Cy3-red/Cy5-green channels. Cy3, 50 µg of total protein isolated from low oleic acid HA89 inbred. Cy5, 50 µg of proteins isolated from high oleic acid HA341 inbred.

FDR in the gel of pI 4-7 and an additional 8 in the pI range 7-10 of the gel pI 3-10 (Supplemental Table 1). In the case of low versus high oleic acid lines (HA89 vs HA341), 41 spots satisfied 10% FDR from the gel of pI 4-7 and only one from the pI region of 7-10 of gel pI 3-10 (Supplemental Table 2). Analysis of these differential proteins by plotting as log expression ratios indicated a higher proportion of proteins upregu-

lated in both the reduced oil and oleic acid lines when compared to their higher oil and oleic acid counterparts (Figure 3). All differential spots were excised from the corresponding preparative 2-DE gels, digested with trypsin, and identified by LC-MS/MS. Out of 119 2-DE statististically significant spots from the RHA280/801 and HA89/341 pairwise analyses, 61 spots Journal of Proteome Research • Vol. 6, No. 8, 2007 3235

research articles

Hajduch et al.

of the differential proteins between these sunflower inbreds. Although sunflower has relatively limited gene resources for data mining, a 51% identification frequency was achieved by searching three different databases that contains sunflower cDNA sequences: NCBInr, UniGene, and TIGR. Despite some level of redundancy, querying all three databases together enabled a higher number of protein assignments. The 44 differential proteins between the oil content lines belonged to 10 different functional classes and 16 subclasses. Metabolism, energy, and protein storage related proteins accounted for 18, 27, and 20% of the identified proteins, respectively. Two differential spots (247, 268) mapped to specific gene products that did not show strong homology to any entries in NCBI. Interestingly, the transcript for Han.2388 (spot 247) was found exclusively in developing embryo libraries. In contrast, the 17 differential proteins for varieties altered in oleic acid composition mapped to 8 functional classes and 12 subclasses.

Discussion

Figure 2. Statistical analysis of 2-DE DIGE gels. Proteins isolated from sunflower inbreds were labeled with CyDyes and resolved by 2-DE in biological triplicate. Each CyDye 2-DE 16-bit TIFF image was analyzed separately using ImageMaster Platinum software (v.5.0). The gels were matched and relative volumes for each spot were calculated. Statistical modeling was used to determine predicted value with error of prediction for each analyzed 2-DE spot. This value was used to calculate expected difference of expression and its standard error for each spot. Two-sample t test was used to calculate probability of 2-DE spots to be differentially expressed (p-value). Finally, to generate list of statistically confidently differentially expressed 2-DE spots, a cutoff false discovery rate of 10% was applied to the p-values for the spot list. These proteins were excised from preparative CBB-stained gels, digested with trypsin, and analyzed by LCMS/MS.

Figure 3. Differentially expressed proteins from statistical analysis of DIGE gels. Protein spots were statistically analyzed as described in Figure 2 and all differential spots were expressed as a log(base 10) ratio of high/low oil (upper panel) and high/ low oleic acid (lower panel). The number of differential proteins was 77 and 42 for the oil content and oleic acid composition lines, respectively.

were successfully identified (Supplemental Tables 3, 4, Supporting Information). Therefore, this is only a fractional view 3236

Journal of Proteome Research • Vol. 6, No. 8, 2007

Seed oil content is a complex trait that has been selectively bred into numerous crop plants but so far uncharacterized at the molecular level. The reasons for this are various but largely due to the infancy of systems biological approaches and the lack of requisite resources (e.g., microarray chips) to perform such research with crop plants. However, it has been demonstrated that proteomics research requires only the availability of large, annotated cDNA datasets for data mining. With the recent development of these resources for sunflower, we have initiated the first proteomic analysis of oil content and composition in this oilseed crop. A Three Step Statistical Approach to Characterize Differentially Expressed Proteins. Two-dimensional DIGE has only recently been used as a quantitative proteomics approach in plants.10,11 We demonstrate here the use of this technique for identifying seed oil content and composition markers from near isogenic sunflower lines altered for these traits by breeding. Although there have been numerous statistical analyses performed with DIGE investigations, there is presently no consensus strategy for identifying differential proteins.31,34,35 We present here an alternative that is simple yet robust for minimizing false-positives. We started with a list of spots common to multiple gels. Then we modeled the global differences between gels as a statistical random effect and the dye that is being observed as a statistical fixed effect. Once these known sources of measurement error are accounted for, we compared the common spot by treatment (type of sample) effect across multiple gels using a t test. This gave us a spot by treatment p-value measure of dissimilarity that is adjusted for the reproducibility of the spot intensities across multiple gels. If this list is sorted into ascending order on p-value, then we have an ordered list of spots with those that are most likely to be present in different concentrations appearing at the top of the list. A false discovery rate (FDR) of 10% was then used to characterize the entire data set. FDR is used extensively in microarray transcript profiling experimentation and has also been used by a few groups to characterize proteome data sets.9,31,34,35 In this paper, we applied FDR for selection of deferentially expressed spots. FDR is a statistical procedure that is effective in correcting for multiple simultaneous hypothesis tests. The 2-DE DIGE-based differential proteomics approach aims to

Proteomic Comparison of Sunflower Seed Oil Varieties

identify protein spots with different concentrations on the same gel. For spots that are common to a set of gels, it is possible to compute the mean intensity and variance in intensity, correcting for the individual gel variability. With this information, a t-type test together with p-value for each protein 2-DE spot can be performed. However, the p-value from this type of test is correct only for the hypothesis tested. Because, in this case, there are many spots per gel, testing of multiple hypotheses simultaneously occur. The FDR procedure allowed us to specify the probability of elements on an ordered list being incorrect.26,27 It should be noted that the statistical workflow described here was designed to minimize false-positive differentially expressed proteins. A Student t test analysis (as standard for ImageMaster software) to discover differential proteins, by itself, may suggest additional differential protein spots. However, the confidence associated with Student t test analyses is lower which could yield a higher frequency of false-positives. Because the purpose of this investigation was to discover markers for seed oil content, with the intent of pursuing these at the biochemical and molecular levels, it was important to have a high burden of proof for differential expression. Differential Expression of Glycolysis, Amino Acid Biosynthesis, and Storage Proteins in Low versus High Oil Sunflower Lines Suggests Cross-Talk between Protein and Oil Storage. Oil content is approximately 2-fold higher in mature seeds from RHA801 versus RHA280 cultivars. The potential metabolic shift that produced this difference in oil content was predicted to be most prominent during mid-seed development. This is based upon previous proteomic studies of seed filling in soybean and rapeseed that revealed enzymes involved in sucrose assimilation, glycolysis, and de novo fatty acid synthesis are all prominent at this period of seed development.23,24 Because oil content is a complex trait possibly involving multiple intermediary metabolic pathways in addition to de novo fatty acid synthesis, we expected to observe more protein expression differences between the RHA280 and 801 lines than a parallel analysis of inbred lines varying in oleic acid composition, which is largely determined by a single desaturase activity, FAD2.7 Indeed, 77 protein spots were differentially expressed in the oil content inbred lines versus 42 differential proteins for the two inbred lines varying in oleic acid composition. These proteins were landmarked on a preparative 2-DE gel and analyzed by LC-MS/MS resulting in assignments for 44 and 17 for the lines varying in oil and oleate, respectively. The 44 and 17 differentially expressed proteins identified between RHA280/801 and HA89/341 comparisons, respectively, were classified based on their function (Table 1; Supplemental Table 3 and 4). The two main metabolic processes that were affected in the oil content lines were glycolysis and amino acid metabolism. Glycolytic enzymes fructokinase (spots 357, 362) and plastid phosphoglycerate kinase (PGA kinase, spot 406) were slightly higher in RHA801, whereas enolase 2 (spot 139) was strongly upregulated in the high oil line (Figure 4). Previous proteomic analysis of seed filling in soybean23 and rapeseed24 indicated high expression levels of sucrose synthase (SuSy) but not invertase, suggesting that sucrose assimilation proceeds principally through SuSy. Because fructose is one of the products of SuSy (UDP-glucose is the other), it is possible that fructokinase, in collaboration with SuSy, could play an important role in phloem unloading during seed filling in sunflower. Interestingly, tomato fructokinase has previously been shown

research articles to have unusual regulatory properties and may play a major role in sink strength of developing fruit.36,37 Because phosphoenolpyruvate (PEP) is the likely precursor for plastid de novo fatty acid synthesis during seed development,23,24,38 enolase could be important in “pulling” glycolyticand RuBisCO-generated 3-phosphoglycerate (3-PGA) into PEP and, therefore, de novo fatty acid synthesis. This model would presume that 3-PGA and 2-PGA are in equilibrium through phosphoglycerate mutase. Although a second enolase form (spot 487) was reduced in RHA801, this was only a 65% reduction compared to a 296-fold induction of enolase 2, one of the most strongly induced protein spots in this entire study. The appearance and differential expression of multiple enolase isoforms may point to complex regulation for PEP production in sunflower. It is less clear how PGA kinase is involved, because the plastid form (spot 406) is 48% higher in RHA801, whereas the cytosolic form (spot 378) is approximately 200-fold lower. Because plastid PGA mutase and enolase activities have never been reported in plants and there are no apparent gene products in Arabidopsis, RuBisCO-generated 3-PGA must exit the plastid for conversion to PEP. If the bulk flow of 3-PGA is toward PEP, reduction of cytosolic PGA kinase (a reversible reaction) could limit retrograde glycolytic flow, particularly when coupled to increased enolase expression. Cytosolic phosphoglucomutase (PGM, spot 564) and phosphofructokinase (PFK, spot 542) were 51- and 124-fold induced, respectively, in the low oil RHA280 line. Because PGM is involved in starch synthesis/breakdown, which is transiently stored in developing embryos, it is highly probable that a temporal and perhaps diurnal relationship exists with PGM and/or PFK activity. Such scenarios will likely require further developmental analysis of these activities to fully understand their role in developing sunflower seed. Alternatively, induction of these activities in the low oil line could be a response to the years of breeding this confection variety for increased seed size, which was produced at the expense of oil content dry mass. For example, induction of some metabolic activities in RHA280 could be a compensatory response for reducing seed oil below the genetically programmed setpoint (e.g., beta-ketoacyl-ACP synthetase 1, spot 430) or a concomitant response to redirect carbon from oil into starch or protein. Four enzymes associated with amino acid metabolism (spots 474, 483, 524, and 601) were highly upregulated in RHA280. Additionally, the 12S CRA1 storage protein (spots 284, 297), and a protein with similarity to a legumin protein (spot 238) were each induced in low oil RHA280 line, whereas an 11S globulin storage protein form was more prominent in RHA801. These data suggest amino acid and storage protein synthesis are perturbed either in response or in conjunction with oil content alterations. Furthermore, aspartic (spot 279) and cysteine proteases (spot 350) and proteasome subunit (spot 271) were also differential in these oil lines, perhaps suggesting protein turnover and processing are also affected. Collectively, these data suggest that glycolysis is at least fractionally induced in high oil RHA801. In contrast, amino acid biosynthesis and seed storage proteins appear to be enhanced in the low oil RHA280 variety. This apparent metabolic crosstalk between oil and protein content in seeds is not particularly surprising as it has been observed in breeding lines of soybean,39-41 Brassica,42,43 and sesame44 that oil and protein levels are negatively correlated. It was also shown that increased expression of the major seed storage proteins contribute to the Journal of Proteome Research • Vol. 6, No. 8, 2007 3237

research articles

Hajduch et al.

Table 1. Summary of Some Differentially Expressed Proteins in Low/High Total Oil (RHA280/801) and Low/High Oleic Acid (HA89/ 341) Sunflower Inbredsa Low Oil RHA280/High Oil RHA801 protein name

group ID

Metabolism of amino acids Cytosolic glutamine synthetase Similar to probable alanine aminotransferase Similar to probable alanine aminotransferase Homologue to acetohydroxy acid isomeroreductase 3-methylcrotonyl-CoA carboxylase precursor protein Homologue to cobalamine-independent methionine synthase Metabolism of fatty acids Homologue to beta-ketoacyl-ACP synthetase 1-2 Glycolysis Similar to enolase 2 Similar to fructokinase Similar to fructokinase Similar to phosphofructokinase Homologue to phosphoglycerate kinase cytosolic Similar to phosphoglycerate kinase Phosphoglucomutase, cytoplasmic 1 Membrane proteins Similar to probable major protein body membrane protein MP27 Similar to probable major protein body membrane protein MP27 Similar to probable major protein body membrane protein MP27 Translation factors Similar to probable elongation factor 1-gamma 2 Similar to probable elongation factor 1-gamma 2 Cofactors Similar to phosphoribosylaminoimidazolecarboxamide formyltransferase

accession number

p-value

upregulated in

384 474 483 524 601 610

Han.564 TC12311 TC12311 TC9724 4587569 TC12298

1.20E-02 1.20E-02 1.00E-03 2.10E-04 7.10E-03 1.00E-04

high oil low oil low oil low oil low oil high oil

430

TC12151

4.00E-03

low oil

139 357 362 542 378 406 564

TC14784 TC9203 TC9203 30725422 TC11511 Han. 1686 12585309

1.90E-04 1.90E-02 2.40E-02 1.20E-02 1.10E-02 2.90E-03 7.90E-03

high oil high oil high oil low oil low oil high oil low oil

332

TC9755

1.20E-02

low oil

341

TC9755

1.30E-02

low oil

342

TC9755

1.10E-04

low oil

105 108

Han.592 Han.592

1.30E-02 6.60E-06

high oil high oil

539

TC14634

3.70E-03

low oil

Low Oleic Acid HA89/High Oleic Acid HA341 protein name

Metabolism of amino acids Similar to probable tyrosine aminotransferase Glycolysis Similar to cytosolic triosephosphatisomerase Similar to cytosolic triosephosphatisomerase Pentose phosphate 6-phosphogluconate dehydrogenase Membrane proteins Similar to probable major protein body membrane protein MP27 Similar to probable major protein body membrane protein MP27 Similar to probable major protein body membrane protein MP27 Signal transduction 14-3-3-like protein 14-3-3-like protein Transcription factors Similar to putative transcription factor Translation factors Similar to Eukaryotic initiation factor 5A a

group ID

accession number

p-value

upregulated in

329

TC11664

4.90E-06

low oleic acid

187 191

Han.1424 Han.1424

5.40E-03 4.70E-04

high oleic acid high oleic acid

385

21928135

4.80E-03

low oleic acid

59

TC9755

5.00E-03

high oleic acid

281

TC9755

1.10E-06

low oleic acid

566

TC9755

2.00E-04

low oleic acid

45 50

Han.15 Han.15

3.30E-03 2.40E-05

high oleic acid high oleic acid

145

TC13454

4.20E-03

high oleic acid

140

Han.1764

3.80E-03

high oleic acid

See supplementary tables for complete list. The protein name, 2-DE spot number, and p-value calculated based on two-sample t test are shown.

high-protein trait observed in high protein soybean varieties.40 Surprisingly, enzymes involved in de novo fatty acid synthesis were not apparently induced in the high oil RHA801 line. This could be due to their under-representation by 2-DE or developmentally late regulated control of these activities in sunflower. Interestingly, three 2-DE spots identified as probable major protein body membrane protein MP27 (spots 332, 341, and 342 on Figure 4; spots 281 and 566 on Figure 5) were detected to 3238

Journal of Proteome Research • Vol. 6, No. 8, 2007

be more abundant in the low oil RHA280 and low oleic acid HA89, when to compared to high oil RHA801 and high oleic acid HA341, respectively. In both RHA801 and HA341 inbreds, different translational factors were found to be more prominent than in RHA280 and HA89 (spots 105, 108, Figure 4; spot 140, Supplementary Table 2). Interestingly, one transcription factor was also found to be associated with high oleic acid content of HA341 (spot 145, Supplementary Table 2). Additionally, it appears that 14-3-3 proteins might be associated with changes

Proteomic Comparison of Sunflower Seed Oil Varieties

research articles

Figure 4. Detailed view of some differentially expressed 2-DE spots for low oil (RHA280) and high oil (RHA 801) sunflower inbreds. Only statistically significant differentially expressed and identified protein spots are shown. Differential spots are circled and spot numbers are noted.

in high oleic acid content of HA341, as this was one of the most strongly induced proteins in this study (spots 45 and 50, Figure 5). It is known that 14-3-3 proteins bind to phosphoproteins and are involved in the regulation of diverse processes including starch biosynthesis in plastids.45 Interestingly, two isoelec-

tric species of stearoyl-ACP desaturase was recently identified from a phosphoproteomic screen of developing rapeseed proteins.46 Perhaps 14-3-3 proteins are also involved in the regulation of fatty acid and more specifically oleic acid biosynthesis. Journal of Proteome Research • Vol. 6, No. 8, 2007 3239

research articles

Hajduch et al.

Figure 5. Detailed view of some differentially expressed 2-DE spots for low oleic acid (HA89) and high oleic acid (HA341) sunflower inbreds. Only statistically significant differentially expressed and identified protein spots are shown. Differential spots are circled and spot numbers are noted.

Conclusions Seed oil content is a complex trait, likely influenced by multiple genes and tightly controlled to maximize individual seed fecundity and total seed production. This is evident based upon the study of a seed-specific, dominant negative mutant for plastid heteromeric acetyl-CoA carboxylase (ACCase), the committed step for de novo fatty acid synthesis.4 Early first and second generations of this transgenic mutant had an increased frequency of seed abortion while viable seed was 25% lower in oil content on average. Therefore, due to the central importance of triacylgycerol as a storage reserve for embryo germination, it is possible that some of the protein expression changes observed in the sunflower oil content lines could be pleiotropic, a consequence of the low (or high) oil trait instead of the cause. This remains the major challenge in analyzing proteomics data from comparative studies such as these, because the data at present does not distinguish between these two possibilities. During seed filling, there is a dramatic level of protein turnover and de novo synthesis;23,24 therefore, it will be important to perform a developmental analysis of the seed proteins expressed in these sunflower varieties to get a more complete depiction of the protein regulatory changes occurring either directly or indirectly as a result of targeted oil content breeding programs. Global shifts or alterations in the developmental timing of reserve deposition could also partly explain the near doubling of oil content in RHA801. For this reason and also to potentially identify each of the differential protein spots, an in-depth developmental study of these two oil content varieties is warranted. This work demonstrates that proteomics is a useful approach to characterize complex traits using near isogenic germplasm. This proof-of-principle example demonstrates that unlike genomics research, proteomics provides the plant biochemist the ability to systematically analyze crop plants in the absence of genome information and prior to the development of microarray chips. 3240

Journal of Proteome Research • Vol. 6, No. 8, 2007

Abbreviations: 2-DE, two-dimensional gel electrophoresis; DAF, days after flowering; DIGE, difference gel electrophoresis; FDR, false discovery rate; IPG, immobilized pH gradient; PEP, phosphoenolpyruvate; PGA, phosphoglycerate; PGM, phosphoglucomutase; PFK, phosphofructokinase; RuBisCO, ribulose 1,5-bisphosphate carboxylase/oxygenase.

Acknowledgment. This investigation was supported by National Science Foundation-Plant Genome Research Program Young Investigator Award DBI-0332418 and GEPR Award DBI0604439. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. Supporting Information Available: Supplementary Tables 1-4. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Thelen, J. J.; Ohlrogge, J. B. Metabolic engineering of fatty acid biosynthesis in plants. Metab. Eng. J. 2002, 4, 12-21. (2) Li, Y.; Beisson, F.; Pollard, M.; Ohlrogge, J. Oil content of Arabidopsis seeds: The influence of seed anatomy, light and plant-to-plant variation. Phytochemistry 2006, 67, 904-915. (3) Borisjuk, L.; Nguyen, T. H.; Neuberger, T.; Rutten, T.; Tschiersch, H.; Claus, B.; Feussner, I.; Webb, A. G.; Jakob, P.; Weber, H.; Wobus, U.; Rolletschek, H. Gradients of lipid storage, photosynthesis and plastid differentiation in developing soybean seeds. New Phyt. 2005, 167, 761-776. (4) Thelen, J. J.; Ohlrogge, J. B. Both antisense and sense expression of biotin carboxyl carrier protein isoform two inactivates the plastid acetyl-coenzyme A carboxylase in Arabidopsis. Plant J. 2002, 32, 419-431. (5) Cernac, A.; Andre, C.; Hoffmann-Benning, S.; Benning, C. WRI1 is required for seed germination and seedling establishment. Plant Physiol. 2006, 141, 745-757. (6) Siloto, R. M. P.; Findlay, K.; Lopez-Villalobos, A.; Yeung, E. C.; Nykiforuk, C. L.; Moloney, M. M. The accumulation of oleosins determines the size of seed oilbodies in Arabidopsis. Plant Cell 2006, 18, 1961-1974.

research articles

Proteomic Comparison of Sunflower Seed Oil Varieties (7) Schuppert, G. F.; Tang, S. X.; Slabaugh, M. B.; Knapp, S. J. The sunflower high-oleic mutant Ol carries variable tandem repeats of FAD2-1, a seed-specific oleoyl-phosphatidyl choline desaturase. Mol. Breed. 2006, 17, 241-256. (8) Unlu, M.; Morgan, M. E.; Minden, J. S. Difference gel electrophoresis: a single gel method for detecting changes in protein extracts. Electrophoresis 1997, 18, 2071-2077. (9) Tonge, R.; Shaw, J.; Middleton, B.; Rowlinson, R.; Rayner, S.; Young, J.; Pognan, F.; Hawkins, E.; Currie, I.; Davison, M. Validation and development of fluorescence two-dimensional differential gel electrophoresis proteomics technology. Proteomics 2001, 1, 377-396. (10) Mooney, B. P.; Miernyk, J. A.; Greenlief, C. M.; Thelen, J. J. Using quantitative proteomics of Arabidopsis roots and leaves to predict metabolic activity. Physiol. Plant. 2006, 128, 237-250. (11) Casati, P.; Zhang, X.; Burlingame, A. L.; Walbot, V. Analysis of leaf proteome after UV-B irradiation in maize lines differing in sensitivity. Mol. Cell. Proteomics 2005, 4, 1673-1685. (12) Fick, G. N.; Zimmer, D. E.; Kinman, M. L. Registration of six sunflower parental lines. Crop Sci. 1974, 14, 912. (13) Roath, W. W.; Miller, J. F.; Gulya, T. J. Registration of RHA801 sunflower germplasm. Crop. Sci. 1981, 21, 479. (14) Miller, J. F.; Zimmerman, D. C.; Vick, B. A.; Roath, W. W. Registration of sixteen high oleic sunflower germplasm lines and bulk population. Crop. Sci. 1987, 27, 1323. (15) Tang, S. X.; Leon, A.; Bridges, W. C.; Knapp, S. J. Quantitative trait loci for genetically correlated seed traits are tightly linked to branching and pericarp pigment loci in sunflower. Crop. Sci. 2006, 46, 721-734. (16) Putt, E. D. Recessive branching in sunflower. Crop. Sci. 1964, 4, 444-445. (17) Leon, A. J.; Lee, M.; Rufener, G. K.; Berry, S. T.; Mowers, R. P. Genetic mapping of a locus (hyp) affecting seed hypodermis color in sunflower. Crop. Sci. 1996, 36, 1666-1668. (18) Tang, S.; Yu, J. K.; Slabaugh, M. B.; Shintani, D. K.; Knapp, S. J. Simple sequence repeat map of the sunflower genome. Theor. Appl. Gen. 2002, 105 1124-1136. (19) Yu, J. K.; Tang, S.; Slabaugh, M. B.; Heesacker, A.: Cole, G.; Herring, M.; Soper, J.; Han, F.; Chu, W. C.; Webb, D. M.; Thompson, L.; Edwards, K. J.; Berry, S.; Leon, A. J.; Olungu, C.; Maes, N.; Knapp, S. J. Towards a saturated molecular genetic linkage map for cultivated sunflower. Crop. Sci. 2003, 43, 367387. (20) Soldatov, K. I. Chemical mutagenesis in sunflower breeding. In Proc. 7th Intl. Sunflower Conf., Krasnodar, USSR, June 27-July 3, 1976; Intl. Sunflower Assoc.: Vlaardingen, the Netherlands, 1976; pp 352-357). (21) Hongtrakul, V.; Slabaugh, M. B.; Knapp, S. J. A seed specifc ∆12 oleate desaturase gene is duplicated, rearranged, and weakly expressed in high oleic acid sunflower lines. Crop. Sci. 1998, 38, 1245-1249. (22) Lacombe, S.; Berville´, A. A dominant mutation for high oleic acid content in sunflower (Helianthus annuus L.) seed oil is genetically linked to a single oleate-desaturase RFLP locus. Mol. Breed. 2001, 8, 129-137. (23) Hajduch, M.; Ganapathy, A.; Stein, J. W.; Thelen, J. J. A Systematic Proteomic Study of Seed-Filling in Soybean: Establishment of High-Resolution Two-Dimensional Reference Maps, Expression Profiles, and an Interactive Proteome Database. Plant Physiol. 2005, 137, 1397-1419. (24) Hajduch, M.; Casteel, J. E.; Hurrelmeyer, K. E.; Song, Z.; Agrawal, G. K.; Thelen, J. J. Proteomic analysis of seed filling in Brassica napus: Developmental characterization of metabolic isozymes using high-resolution two-dimensional gel electrophoresis. Plant Physiol. 2006, 141, 32-46. (25) Box, G. E. P.; Cox, D. R. An Analysis of Transformations. J. R. Stat. Soc. 1964, B26, 211-252. (26) Benjamini, Y.; Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. 1995, B57, 289-300. (27) Benjamini, Y.; Yekutieli, D. False discovery rate - adjusted multiple confidence intervals for selected parameters. JASA 2005, 100, 71-81.

(28) Eng, J. K.; McCormack, A. L.; Yates, J. R. III. An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. J. Am. Soc. Mass Spectrom. 1994, 5, 976-989. (29) Yates, J. R. III; Eng, J. K.; McCormack, A. L.; Schieltz, D. Method to correlate tandem mass spectra of modified peptides to amino acid sequences in the protein database. Anal. Chem. 1995, 67, 1426-36. (30) Campostrini, N.; Areces, L. B.; Rappsilber, J.; Pietrogrande, M. C.; Dondi, F.; Pastorino, F.; Ponzoni, M.; Righetti, P. G. Spot overlapping in two-dimensional maps: a serious problem ignored for much too long. Proteomics 2005, 5, 2385-2395. (31) Fodor, I. K.; Nelson, D. O.; Alegria-Hartman, M.; Robbins, K.; Langlois, R. G.; Turteltaub, K. W.; Corzett, T. H.; McCutchenMaloney, S. L. Statistical challenges in the analysis of twodimensional difference gel electrophoresis experiments using DeCyder. Bioinformatics 2005, 21, 3733-3740. (32) Karp, N. A.; Lilley, K. S. Maximizing sensitivity for detecting changes in protein expression: Experimental design using minimal CyDyes. Proteomics 2005, 5, 3105-3115. (33) Kultima, K.; Scholz, B.; Alm, H.; Skold, K.; Svensson, M.; Crossman, A. R.; Bezard, E.; Andren, P. E.; Lonnstedt, I. Normalization and expression changes in predefined sets of proteins using 2D gel electrophoresis: A proteomic study of L-DOPA induced dyskinesia in an animal model of Parkison’s disease using DIGE. BMC Bioinformatics 2006, 7, 475. (34) Alm, H.; Scholz, B.; Fischer, C.; Kultima, K.; Viberg, H.; Ericksson, P.; Dencker, L.; Stigson, M. Proteomic evaluation of neonatal exposure to 2,2′,4,4′,5-pentabromodiphenyl ether. Env. Health Perspect. 2006, 114, 254-259. (35) Meunier, B.; Bouley, J.; Piec, I.; Bernard, C.; Picard, B.; Hocquette, J.-F. Data analysis methods for detection of differential protein expression in two-dimensional gel electrophoresis. Anal. Biochem. 2005, 340, 226-230. (36) MartinezBarajas, E.; Randall, D. D. Purification and characterization of fructokinase from developing tomato (Lycopersicon esculentum Mill) fruits. Planta 1996, 199, 451-458. (37) MartinezBarajas, E.; Luethy, M. H.; Randall, D. D. Molecular cloning and analysis of fructokinase expression in tomato (Lycopersicon esculentum Mill). Plant Sci. 1997, 125, 13-20. (38) Schwender, J.; Ohlrogge, J. B. Probing in vivo metabolism by stable isotope labeling of storage lipids and proteins in developing Brassica napus embryos. Plant Physiol. 2002, 130, 347-361. (39) Burton, J. W. Quantitative genetics: results relevant to soybean breeding. In Soybeans: Improvement, Production and Uses; Wilcox J. R., Ed.; American Society of Agronomy: Madison, WI, 1987; pp 211-247. (40) Yaklich, R. W. β-Conglycinin and glycinin in high-protein soybean seeds. J. Agric. Food Chem. 2001, 49, 729-735. (41) Hyten, D. L.; Pantalone, V. R.; Sams, C. E.; Saxton, A. M.; LandauEllis, D.; Stefaniak, T. R.; Schmidt, M. E. Seed quality QTL in a prominent soybean population. Theor. Appl. Genet. 2004, 109, 552-561. (42) Grami, B.; Stefansson, B. R. Genetics of protein and oil content in summer rape: Heritability, numbers of effective factors, and correlations. Can. J. Plant Sci. 1977, 57, 937-943. (43) Mahmood, T.; Rahman, M. H.; Stringam, G. R.; Yeh, F.; Good, A. G. Identification of quantitative trait loci (QTL) for oil and protein contents and their relationships with other seed quality traits in Brassica juncea. Theor. Appl. Genet. 2006, 113, 1211-1220. (44) Culp, T. W. Inheritance and association of oil and protein content and seed coat type in sesame, Sesamum indicum L. Genetics 1959, 44, 897-909. (45) Sehnke, P. C.; Chung, H. J.; Wu, J.; Ferl, R. J. Regulation of starch accumulation by granule-associated plant 14-3-3 proteins. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 765-770. (46) Agrawal, G. K.; Thelen, J. J. Large scale identification and quantitative profiling of phosphoproteins expressed during seed filling in oilseed rape. Mol. Cell. Proteomics 5, 2044-2059.

PR070149A

Journal of Proteome Research • Vol. 6, No. 8, 2007 3241