Proteomic analysis of the development and germination of date palm (Phoenix dactylifera L.) zygotic embryos

RESEARCH ARTICLE

Proteomic analysis of the development and germination

of date palm (Phoenix dactylifera L.) zygotic embryos

Besma Sghaier-Hammami1, Luis Valledor2, Noureddine Drira1 and Jesús V. Jorrin-Novo3

1 Laboratoire des Biotechnologies Végétales Appliquées à l’Amélioration des Cultures, Faculté desSciences de Sfax, Sfax, Tunisie

2 ERIPHYSAGE Research Group, Plant Physiology Area, Department of B.O.S., University of Oviedo,Oviedo, Spain

3 Agricultural and Plant Biochemistry and Proteomics Research Group, Department of Biochemistryand Molecular Biology, University of Córdoba, Córdoba, Spain

By using a comparative proteomic approach (2-DE coupled to MS/MS), the development,maturation, and germination of date palm zygotic embryos, have been studied. Proteins weretrichloroacetic acid (TCA)–acetone–phenol extracted and resolved by 2-DE in the 5–8 pH range.The total protein content and the number of spots resolved increased from early (12 weeks afterpollination (WAP); 68.96 mg/g DW: 207 spots) to late (17 WAP; 240.85 mg/g DW: 261 spots)stages, decreasing upon germination (from 120.8 mg/g DW: 273 spots in mature embryos to26.35 mg/g DW: 87 spots in 15 days after germination). Up to 194 spots showed qualitative orquantitative differences between stages. Statistical analysis of spot variation was performed byPCA, obtaining a more accurate grouping of the samples and determining the most discriminantspots. Samples were also clustered based on Pearson distance and Ward’s minimum distance.Sixty-five variable spots were subjected to MS analysis, resulting in 21 identifications. The iden-tified proteins belong to the following functional categories: enzymes of glycolysis, tricarboxylicacid cycle, and carbohydrate biosynthesis, protein translation, storage (glutelin), and stress-related proteins. The evolution pattern of the functional groups was examined and discussed interms of metabolism adaptation to the different embryogenic and germination stages.

Received: June 23, 2008Revised: January 7, 2009

Accepted: January 7, 2009

Keywords:

Date palm proteomics / Phoenix dactylifera L. / Seed germination / Zygotic embryos

Proteomics 2009, 9, 2543–2554 2543

1 Introduction

Proteomics has nowadays become a very important andpowerful tool in plant biology research. Thus, by using a

comparative proteomics approach, a number of proteins –differentiating genotypes, organs, or involved in growthand development and responses to biotic and abiotic stres-ses – have been identified. Arabidopsis thaliana, Oryzasativa, Medicago truncatula, and other plant species of agri-cultural or forestry interest have been used to performthese studies, mainly by using 2-DE combined with MS[1–5].

We are conducting a research project aimed at studyingzygotic and somatic embryogenesis, as well as seed germi-nation, in date palm (Phoenix dactylifera L.), one of the mosteconomically important fruit crops across North Africa andthe Middle-East. A deeper knowledge of proteins and genesinvolved in embryogenesis would be very useful for theimprovement of vegetative clonal propagation techniques of

Correspondence: Professor Jesús V. Jorrin-Novo, Departamentode Bioquímica y Biología Molecular, Universidad de Córdoba;Campus de Rabanales, Ed. Severo Ochoa, 14071 Cordoba, SpainE-mail: [email protected]: 134-957-218439

Abbreviations: ADH, alcohol dehydrogenase; DAG, days aftergermination; m ZE, mature zygotic embryos; PC, principal com-ponent; SE, somatic embryos; TCA, trichloroacetic acid; UDPG,

UDP-glucose; UTP-GPUT, UTP-glucose-1-phosphate uridylyl-transferase; WAP, weeks after pollination; ZE, zygotic embryos

DOI 10.1002/pmic.200800523

© 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

2544 B. Sghaier-Hammami et al. Proteomics 2009, 9, 2543–2554

elite genotypes, a challenge for this and other woody plantspecies [6]. Although somatic embryo (SE) production is awell-established process for date palm [7], there is still roomfor improvement. The SE differs from the zygotic embryo(ZE) in lacking a dormancy phase, a seed integument andan endosperm, all of which are required for seed survivaland germination. This may explain why SE seedlings areless vigorous than those raised from true seeds. In thisrespect, more knowledge is required of SE physiology, andespecially of the differences in protein content between SEand ZE [8].

Zygotic embryogenesis is a complex process that beginswith a single cell and results in the formation of matureembryos, which will give rise to a plant with full multi-cellularity, sexuality and structure [9]. One factor vital toembryogenic development is the maturation phase, since amajor change occurs during this period: a switch from aregional – and cell-specific program to a storage accumula-tion program in order to prepare the embryo for post-embryonic development [10]. At the final stage, desiccationtolerance is acquired and metabolic activity is reduced to aquiescent state [11]. Accumulation of reserve products playsan essential role in seedling survival, providing compoundsthat are utilized by the germinating embryo until the devel-opment of autotrophy occurs [12].

Seed tissue development and germination have beenstudied by using transcriptome and proteomic analysis,either independently or in combination, in legumes [13–16],A. thaliana [17], cereals [18–21], and other species [22]. Byusing these approaches, key proteins or genes have beenidentified, the major ones corresponding to storage proteins(representing up to 83% of total proteins in the mature avo-cado, Persea americana Mill., embryos [23]), carbohydratecatabolism and biosynthesis, and stress related proteins. Theidentified proteins showed a complex evolution pattern,making them biochemically potential markers for embry-ogenesis both in situ and in vitro (embryo cultures, somaticand gametic embryogenesis) [24].

In a previous paper [8], we reported changes in the pro-tein pattern during ZE development and germination byusing 1-DE. The work presented here takes a step forwardusing the more powerful 2-DE, which, combined with MS,allowed the identification of some proteins that were differ-entially accumulated in their different development andgermination stages. The use of proteomics with date palmhas been reported in at least a couple of papers. In the firstone, by Chandra Sekhar and DeMason [25], 1- and 2-DEwere used to analyze the protein content of endosperm andembryo in mature seeds, revealing a number of tissue-spe-cific protein bands or spots. The second one, by Gómez-Vidal et al. [26], is a methodological paper in which theoptimization of an extraction protocol for leaf protein anal-ysis is reported. In both works, no MS analysis was per-formed. As in other woody plant orphan species, almostabsent in DNA/protein databases, date palm proteomeanalysis is a challenging task [27].

2 Materials and methods

2.1 Plant material

Seeds from the Deglet Nour date palm cultivar were used.Protein extracts were obtained from developing, mature, andgerminating ZE [8]. Seeds were collected after the transitionstage, occurring at 10 weeks after pollination (WAP), oncethe endosperm had hardened, at 12, 14, and 17 WAP, andonce they were mature (maturation took place at 23 WAP).Embryos (30–50 per replicate) were excised from seeds,weighed, frozen in liquid nitrogen, and stored at 2707C untilprotein extraction. Three biological replicates were extractedat each stage. Mature seeds were imbibed in water and keptin the dark at 25 6 17C in order to induce germination.Embryos (20 per replicate; three biological replicates persample) were excised from seeds at 9, 12, and 15 days aftergermination (DAG). A photograph of the fruits, seeds, andembryos at the sampled times is presented in SupportingInformation Fig. 1.

2.2 Protein extraction

Three different extraction protocols had been previouslyevaluated in terms of protein yield, spot resolution, andnumber of resolved spots: buffer extraction (Tris–Maleate–Triton X-100), and protein precipitation by using trichloro-acetic acid (TCA)–acetone, and TCA–acetone-phenol, withthe latter, as previously reported by Maldonado et al. [28],providing the best results in terms of protein yield andnumber of resolved spots (Supporting Information Fig. 2).Embryos were ground to a fine powder with liquid nitrogenusing a mortar and pestle. The powder (0.1 g) was subjectedto a protein extraction as indicated by Wang et al. [29]. Thefinal pellet was vacuum-dried and dissolved in 40–200 mL,depending on the amount of starting material, of 8 M urea,2% CHAPS, 20 mM DTT, and 0.5% ampholytes (Bio-Lyte 3/10, BioRad). Once the pellet was solubilized and the insolu-ble material eliminated by centrifugation, the protein con-tent was quantified by the method of Bradford [30], usingBSA as standard.

2.3 2-DE

In preliminary 2-DE experiments with IEF in the 3–10 pHrange, it was observed that most of the spots were con-centrated in the pH 5–8 region (Supporting InformationFig. 3). Considering this, and in order to increase the resolu-tion by preventing protein comigration as far as possible, IEFwas performed using 5–8 pH linear range strips. For thesecond dimension, SDS-PAGE, 12% polyacrylamide gelswere utilized.

Protein extracts (total protein 500 mg), were subjected to2-DE [28]. Gels were stained with Coomassie (CBB G-250;Merck, Darmstadt, Germany) as reported by Mathesius et al.[31]. Images were digitized with a GS-800 Calibrated Densi-


Proteomics 2009, 9, 2543–2554 2545

tometer (BioRad) and analyzed with the PDQuest™ software(BioRad), using a guided protein spot detection method [32].Normalized spot volumes based on total quantity in validspots were calculated for each 2-DE gel and used for statis-tical calculations of protein abundance. Experimental Mr

values were calculated by mobility comparisons with proteinstandard markers (Mr-SDS-70L; Sigma, St. Louis, MO, USA)run in a separate marker lane on the SDS-gel, while pI wasdetermined by using a 5–8 linear scale over the total dimen-sion of the IPG strips. For all the spots, the pI, Mr, normal-ized volumes (mean values, SD, and CV) were determined.

2.4 Statistics

The structure of raw proteomic data is known to be very dis-turbing for multivariate statistics and clustering algorithms[33]. The nature of proteomic experimental conditions causesa certain number of missing spots, which alters and causesdisparities between data sets. To deal with this problem, weonly considered “consistent spots,” which are those presentin the three biological replicates, thus preventing the assign-ment of normalized volume values to missing spots formultivariate analysis.

Another issue related to proteomics data is the correla-tion between spot normalized volume and spot variance,described in some proteomic studies [34–36], which meansthat the higher the mean intensity of a spot, the higher thevariance, this being explained by a scale phenomenon relatedto data acquisition [37]. To reduce this variance-mean de-pendence between different spot intensities and sample sets,a cubic root transformation was applied. After that, spotvolumes were standardized (spot volume/Sgel spotvolumes6100).

Differentially expressed spots were defined after apply-ing a one-way ANOVA test. Spot values passed the Kolmo-gorov–Smirnov normality test.

A multivariate analysis was performed over the whole setof spots and on those showing differences. PCA was appliedto the correlation matrix to reduce its dimensionality (SPSSv. 15 package). Using un-rotated principal component (PC)scores, the relation between the different sampling timeswas studied by determining the spots with the highest loadon the variance.

Samples were clustered employing Ward’s minimumvariance method over a Pearson distance-based dissimilaritymatrix. Heat maps were also plotted employing PermutMa-trix software v.1.9.3 [38].

2.5 MS analysis and protein identification

Spots were automatically excised (Investigator ProPic,Genomic Solutions), transferred to Multiwell 96 plates, anddigested with modified porcine trypsin (sequencing grade;Promega) by using a ProGest (Genomics Solution) digestionstation. The digestion protocol used was that of Schevchenkoet al. [39], with minor variations. Gel plugs were destained by

incubation (twice for 30 min) with 200 mM ammoniumbicarbonate in 40% ACN at 377C, then subjected to threeconsecutive dehydration/rehydration cycles with pure ACNand 25 mM ammonium bicarbonate in 40% ACN, respec-tively, and finally dried at room temperature for 10 min.20 mL trypsin, at a final concentration of 12.5 ng/mL in25 mM ammonium bicarbonate was added to the dry gelpieces and the digestion proceeded at 377C overnight.

Peptides were extracted from gel plugs by adding 10 mLof 1% TFA (15 min incubation), and were then purified byZipTip. Peptides were deposited onto MALDI plate using thedry droplet method (ProMS, Genomic Solutions) and theCHCA as matrix at 5 ng/mL concentration in 70% ACN,0.1% TFA. Samples were analyzed in a 4700 ProteomicsAnalyzer MALDI-TOF/TOF mass spectrometer (AppliedBiosystems), in the m/z range 800–4000, with an accelerat-ing voltage of 20 kV, in reflectron mode and with a delayedextraction set to 120 ns. Spectra were internally calibratedwith peptides from trypsin autolysis (M 1 H1 = 842.509,M 1 H1 = 2211.104). The three most abundant ions werethen subjected to MS/MS analysis, providing informationthat could be used to determine the peptide sequence.

A combined PMF search (MS plus MS/MS) was per-formed using GPS Explorer™ software v 3.5 (Applied Biosys-tems) over nonredundant MSDB and NCBI databases usingthe MASCOTsearch engine (Matrix Science, London; http://www.matrixscience.com). The following parameters wereallowed: taxonomy restrictions to Viridiplantae, one missedcleavage, 100 ppm mass tolerance in MS, and 0.5 Da for MS/MS data, cysteine carbamidomethylation as a fixed mod-ification and methionine oxidation as a variable modifica-tion. The confidence in the PMF matches (p,0.05) was basedon the MOWSE score and confirmed by the accurate over-lapping of the matched peptides with the major peaks of themass spectrum.

3 Results and discussion

3.1 Protein extraction and 2-DE

The 2-DE protein profiles of date palm ZE during maturationand after germination were compared. Embryos were takenfrom seeds at 12, 14, and 17 WAP (development), oncematured (23 WAP) and 9, 12, and 15 DAG (germination),(Supporting Information Fig. 1). It is important to empha-size that ZE are difficult to access because they are minuteand deeply embedded in maternal tissues, it being almostimpossible to separate both parts, which explains why earliertimes than 12 WAP were not analyzed. For protein extrac-tion, the Wang et al. [29] TCA–acetone–phenol method wasutilized. With this protocol, the protein yield varied from26.35 (15 DAG) to 240.85 (17 WAP) mg/g DW, with adecreasing tendency at earlier times of development and ongermination (Table 1). This variation has been previouslyreported in other embryogenic and germination studies



Table 1. Protein yield and number of spots detected in Coo-massie-stained 2-DE gels

Stage Protein content(mg/g DW)

Number of spots

Development stage

12 WAP 068.96 6 03.6 207.66 6 1.1514 WAP 169.46 6 03.0 243.66 6 0.5717 WAP 240.85 6 10.7 261.00 6 1.73m ZE 120.80 6 09.4 273.33 6 1.15

Germination stage

9 DAG 101.63 6 08.5 170.33 6 0.5712 DAG 040.48 6 03.6 099.33 6 0.5715 DAG 026.35 6 0.80 087.33 6 0.57

Values correspond to the different stages analyzed: embryodevelopment (12, 14, and 17 WAP), mature embryos (m ZE), andgermination (9, 12 and 15 DAG). Values are mean 6 SD of threebiological replicates. All the differences were statistically signifi-cant (ANOVA test, p,0.001).

[24, 40–42], and is mainly related to the accumulation (dur-ing embryo maturation) and mobilization (during germina-tion) of reserve proteins and biochemical reprogramming.

Proteins (500 mg) were separated by 2-DE with IEF carriedout in the 5–8 pH range (Fig. 1, Supporting InformationFig. 4). The number of resolved spots (Table 1) increased from207 (12 WAP) to 273 (mature embryos), then decreasing dur-ing germination up to 87 (15 DAG). The 20 most abundantspots at the different stages are listed in the in SupportingInformation Table 1. This evolution pattern closely resemblesthe one observed for the total protein content. When inter-preting the results presented, it is important to realize that byusing 2-DE only a minimal fraction of the total cell proteomecan be visualized, and that the protein evolution pattern dis-cussed here only refers to those proteins solubilized under theexperimental conditions used, with 5–8 pI, 6–120 Mr values,and above the detection limit for Coomassie staining. Thus, asan example, dehydrins, important in embryo developmentand generally very basic, will not be visualized, requiringalternative protocols. However, Aberlenc-Bertossi et al. [43]have reported the existence of a dehydrin gene encoding for aprotein with 90 amino acid and pI 6.14.

As the same amount of protein was loaded onto the gel,following the Bradford protocol estimation, a greater proteindiversity in mature embryos and a higher content of specificspots can be deduced. Similar results and conclusions werereported on the basis of our previous 1-DE analysis in datepalm [8], and in barley [44], Cyclamen persicum [22] and avo-cado [23], with noticeable changes in the storage proteincontent observed during maturation in the avocado system.

3.2 Statistical analysis

Following a traditional univariate analysis made with 2-DEdata, 194 spots differentially expressed in the developmental

Figure 1. 2-DE master gel of date palm embryos containing allthe spots visualized at the different stages analyzed (develop-ment, once mature, and during germination). The Mr is given onthe left, while the pI is given at the top. Proteins (500 mg) wereseparated in the first dimension on an immobilized, linear, 5–8 pHgradient and in the second dimension on a 12% acrylamide-SDSgel. Gels were Coomassie-stained. Only identified protein spotsare numbered (Table 4).

stages studied (ANOVA test, 95% confidence level) werefound. Variable spots were only considered for analysis ifthey were consistent and their CV was �25%, correspondingto the mean analytical variance of the system (unpublishedresults). These variable spots were either absent in at leastone stage (qualitative variable spots) or differentially accu-mulated between stages (quantitative variable spots), (Table 2;Fig. 2 and Supporting Information Table 2). As expected, thelargest number of differences was found at the most distantsampling time points, earlier developmental, and later ger-mination stages (Table 2). Similar results were found in M.truncatula seeds, in which 615 varied significantly at specificstages out of 790 polypeptides profiled during development[14].

The importance of studying the whole protein patternand performing a multivariate analysis has also been con-sidered [32, 45]. First of all, a data reduction to the wholedataset and to the 194 differential spots by means of PCAanalysis was applied. Out of the potential 301 or 194 PCsextracted, the first eight or seven PCs accounted for 97% ofthe biological variability of each whole or differential spotdataset, respectively (Table 3). When PCs were tested for dif-ferences between groups, PCs 1–3 were significant in the twostudied datasets (one-way ANOVA, .95% confidence). Theemployment of these components in a 2-D (plotting PC1 andPC2) or three dimensional (plotting PC1, PC2, and PC3)


Proteomics 2009, 9, 2543–2554 2547

Figure 2. Differences in the 2-DEmap between embryo stages(development, once mature,and during germination). A realgel, corresponding to matureembryos, is shown on the left.Areas in the gel are magnified inorder to visualize differences inspot intensity. Only identifiedprotein spots are shown andnumbered (Fig. 1, Table 4).

Table 2. Number of spots showing qualitative (absent in at leastone of the stages) or quantitative differences

Stage Qualitativedifferences

Quantitativedifferences

Nondetected

Absentin m ZE

Moreintensein m ZE

Lessintensein m ZE

Nonvariablespots

12 WAP 51 10 76 45 0814 WAP 18 06 79 26 4517 WAP 13 08 35 71 599 DAG 51 04 38 68 21

12 DAG 112 03 30 25 1315 DAG 126 03 28 20 08

The 2-DE map of embryos at 12, 14, and 17 WAP, and 9, 12, and 15DAG, was compared with that of the mature embryos. WAP,weeks after pollination; DAG, days after germination; m ZE,mature zygotic embryos.

representations allowed the effective separation of samplesinto their original groups (Supporting Information Fig. 5),being the plot structure not greatly different between wholeand differential spots datasets. The two dimensional PCAplots (Supporting Information Figs. 5a and b) show thatsamples of 12 and 15 DAG were closely grouped in bothplots, indicating similarity in the spot map. 17 WAP andmature zygotic embryo (m ZE) samples were also grouped inthe plot corresponding to the differential spots but not in thewhole spot dataset. These differences between groups indi-

cate that some key spots with high loading over PCs andcommon to both samples are lost after univariate analysis.The remaining spots are different enough to differentiatesamples. This is not surprising, because after the filtering ofnondifferential spots we can expect a change in the structureof the dataset. The inclusion of a third dimensional PCA inthe plot (Supporting Information Figs. 5c and d) did not alterremarkably their structure, but 12 WAP samples are out ofthe group constituted by 12 and 15 DAG samples after add-ing PC3 to the plot. Due to the biological dissimilarities be-tween these developmental stages this is a more accuraterepresentation, also indicating that PC3 can be used as amarker to differentiate them. This application of PCA hadbeen successfully used before by other authors [46, 47].

The selection of spots that could be used as biomarkers ofeach developmental stage would be of great interest. Theloading of each particular spot to PC 1, 2, and 3 was deter-mined from the factor matrix generated during the PCA. Thefive spots showing the highest correlation (above )0.9)) witheach PC were determined (spots number 82, 91, 147, 144,and 162 for PC1; 5, 29, 54, 104, and 134 for PC2; 47, 70, 99,111, and 118 for PC3). These spots can be used to differ-entiate the stages studied and, once identified, to improvethe explanation of the variability contained in those groups.This method is commonly used in biomarker selection pro-cesses [48]. Out of these spots, only spot 111 was identifiedafter MS analysis, corresponding to a phosphoglyceratekinase (discussed later).

Differential spots were clustered employing Ward’sminimum variance method over a Pearson distance-based



Table 3. PCs calculated from the 301 spots matched between samples and from the 194 differential spot set

Component Eigenvalues (301 spots) Eigenvalues (194 spots)

Total % ofvariance

Cumulative%

Total % ofvariance

Cumulative%

PC 1 151.133 50.210 50.210 78.673 40.553 40.553PC 2 52.025 17.284 67.494 41.898 21.597 62.150PC 3 38.372 12.748 80.242 30.414 15.677 77.827PC 4 28.989 09.631 89.873 23.331 12.026 89.853PC 5 12.951 04.303 94.176 08.796 04.534 94.387PC 6 08.585 02.852 97.028 03.762 01.939 96.327PC 7 01.917 00.637 97.665 02.001 01.032 97.358PC 8 01.090 00.362 98.027

dissimilarity matrix (Fig. 3), which has proved to be the mostaccurate procedure for proteomics data [33]. There was ashort distance, similar to that of replicates of the same group,between samples corresponding to 12 and 15 DAG, indicat-ing that these two time samples have a fairly similar 2-DEmap, as the PCA analysis showed before. Spots in the sametree were compared, employing a clustering method and arepresentation of quantitative variations between stages, de-termining 10 main groups based on the Pearson dissim-ilarity coefficient. The different expression dynamics of eachspot could be distinguished in the heat map, allowing a defi-nition of some expression groups, the presence or absence ofsome of them being a characteristic of a particular stage(group II, only present in m ZE; group VI, present in allstages except 12 and 15 DAG).

3.3 Protein identification and evolution pattern

Sixty-five variable spots, including those showing qualitativechanges and quantitative ones with the highest maximum/minimum value ratios, were subjected to MALDI-TOF-TOFMS analysis. Protein identification was accomplished by acombined PMF MS/MS strategy, consulting the non-redundant NCBI and MSDB databases using the MASCOT(Matrix Science; http://www.matrixscience.com) algorithm.This search resulted in 21 confidence hits (Table 3). For therest of the spots, a low PMF score, poor MS/MS fragmenta-tion, or no hits, were obtained. This low identification per-centage is common in species that, like P. dactylifera, are verypoorly represented in databases (32 protein entries and 9ESTs by December 2008 at NCBI), such as Prunus avium [49],Picea abies [50], Craterostigma plantagineum [51], and Vignaunguiculata [52].

The identified proteins were grouped into the followingcategories: (i) enzymes of glycolysis (6 proteins), tricarboxylicacid cycle (2), and carbohydrate biosynthesis (3); (ii) proteintranslation (1); (iii) storage proteins (3); (iv) stress-relatedproteins (3). For the enzymes aldose reductase and glyox-ylase I, a specific pathway cannot be assigned, these being

categorized as stress-related protein [44]. Some of the spotsmatched the same protein and differed in pI and/or Mr,hence corresponding to isoforms or multiple forms/PTMvariants of the same gene product: phosphopyruvate hydra-tase (spots 98 and 117), alcohol dehydrogenase (ADH) (spots36 and 138), UTP-glucose-1-phosphate uridylyltransferase(UTP-GPUT) (spots 94 and 96), 70 kDa HSP (spots 76 and103) and glutelin (spots 6, 56, and 89).

The presence of protein isoforms with different expres-sion patterns during seed maturation has been reported inbarley [44]. For some identified proteins, close theoreticaland experimental Mr values were obtained, but for others(starch synthase II, spot 154 and 17.6 kDa HSP, spot 148),lower experimental than theoretical values were obtained,these probably being degradation products of the native pro-tein.

The different identified proteins and their evolution pat-tern during seed development and germination have beendiscussed, in order to establish tendencies and correlatethem with functional categories. Spot intensity at each stagewas expressed as a percentage of the corresponding value inmature embryos (Fig. 4). For 45.3% of the most abundantspots (Supporting Information Table 1) and 9 out of the 21identified proteins (Table 4, Fig. 4), the time pattern observedfollowed the same tendency for total proteins and number ofspots (Table 1), which is increasing from early to late embryodevelopmental stages, with a maximum at 17 WAP or at themature stage and then decreasing on germination. Excep-tions to this general tendency were observed for: (i) phos-phoglycerate kinase (spot 111), starch synthase II (spot 154),70 HSP (spot 76), and 12% of the most abundant spots; thisgroup showed the highest values at the earliest embryogenicstage analyzed (12 WAP); (ii) 17.6 kDa HSP (spot 148), pyr-uvate decarboxylase (spot 120), UTP-GPUT (spots 94 and96), and 5.3% of the most abundant spots; they showed nochanges throughout the assayed period; (iii) malate dehy-drogenase (spot 137), and 10.6% of the most abundant spots,which exhibited high accumulation at the germination stage(12 DAG).


Proteomics 2009, 9, 2543–2554 2549

Figure 3. Two-way hierarchicalcluster of differentially ex-pressed spots. A heat maprepresentation of the clusteredspots shows the protein valueaccording to the standardizedspot percent in each replicate,with light grey intensity indicat-ing up-accumulation and blackabsence.

The evolution of glycolytic and tricarboxylic acid cycleenzymes during seed development has been reported in anumber of studies [14, 21, 53, 54]. Our data on these proteinsindicate active energy metabolism during embryogenesis,with fermentation of pyruvate to ethanol, as revealed by thedifferential presence of ADH. Due to their limited diffusionalgas exchange, embryos grow in an environment of low oxy-gen, so that their respiration is inhibited. In broad bean andpea seeds it has been reported that during early develop-mental stages ethanol emission and ADH activity were high,indicating fermentative metabolism. Upon date palm germi-nation, malate dehydrogenase protein (spot 137) increasedand ADH (spot 36) disappeared, indicating a change fromanaerobic/fermentative to aerobic/respiration pathways.Also, the reduction in glycolytic enzymes at late germinationstages could correspond to the beginning of some photo-synthesis activity. A similar evolution pattern of glycolyticenzymes has been reported in maize endosperm [21].

Glyoxalase I (spot 12) and aldose reductase (spot 108),showed similar pattern to that of glycolytic enzymes andADH, their increase being related to the production of toxicby-products such as aldehydes [55] and methylglyoxal [56] asa consequence of the high ATP demand linked to fermenta-tion [55, 56]. In a proteomic study of seed maturation carriedout in barley [44], it has been shown that these two proteinsgradually increased during development. It has been pro-

posed that glyoxalase I may play a role in cell proliferationand differentiation [57]. Yang et al. [58] reported in the Indicarice cultivar that aldose reductase began to accumulate 15days after pollination, reaching a peak at seed maturity, andthen disappeared upon imbibition [58]. In tobacco, the ec-topic expression of an alfalfa aldehyde reductase clone resul-ted in tolerance to oxidative stress and dehydration [59].

Starch is the main storage carbohydrate in plants, beingsynthesized in seeds during early developmental stages, andthen degraded for embryo development [60]. Plant starchsynthase contributes to the elongation of glucan chains dur-ing starch biosynthesis, with multiple isoforms being foundin the endosperm of higher plants [61]. Starch synthase II(spot 154) showed maximum values at 12 WAP, but waspresent at very similar levels at the other sampled times. Thiscould indicate active starch synthesis at the early stages ofembryo development, also reported in cypress ZE [40], inwhich early stages of embryo development were character-ized by the predominant elaboration of starch. As theirdevelopment proceeded, starch content gradually decreased.Starch content increased continuously during pedunculateoak ZE development from early to late stages [62] and Gal-lardo et al. [14] reported a transient expression of starch syn-thesis in M. truncatula seeds during seed filling.

Two proteins (spots 94 and 96) matched with UTP-GPUT, and they were detected in all the samples with no



Figure 4. Evolution pattern of the identified proteins. Changes in spot intensity of the different proteins identified (classified according totheir function), throughout the different developmental stages, are expressed as a percentage of the intensity in m ZE.

significant differences between stages. This enzyme cata-lyzes the reversible production of UDP-glucose (UDPG) andpyrophosphate (PPi) from Glucose-1-P and UTP, the latterbeing an important enzyme in carbohydrate metabolism. Inleaves, it is primarily involved in sucrose biosynthesis, pro-viding UDPG for sucrose-P synthethase, whereas in otherorgans that like seeds depend to some extent on importedcarbon, it may take part in sucrose breakdown, using UDPGproduced by sucrose synthase. In cereal seed endospermUDPG is coupled to the activity of cytosolic ADP-glucose

pyrophosphorylase, resulting in an equimolar production ofADP-glucose, the immediate precursor of starch in all plants.This ensures a close interaction of the reactions of sucrosemetabolism [63]. In addition to being used in sucrose/starchpathways, UTP-GPUT participates, directly or indirectly, inthe biosynthesis of cell wall polysaccharides and in the syn-thesis of the carbohydrate moiety of glycolipids and glyco-proteins [64].

Stress-related proteins of the HSP family have beenidentified in this work, concretely the 17.6 kDa (spots 76 and


Proteomics 2009, 9, 2543–2554 2551

Table 4. List of identified proteins

Protein namea) Species/accessionnumberb)

Spotnum-berc)

MASCOT scoresd) MPe) Experimental Theoretical

Protein score/protein score CI%

Mrf) pIg) Mr

f) pIg)

Glycolysis

Phosphopyruvate hydratase(EC 4.2.1.11)

Zea mays, S16257 098 162/100 11 58.89 6.0 48.26 5.2

Phosphopyruvate hydratase(EC 4.2.1.11)

Ricinus communis, S39203 117 101/100 10 57.97 6.5 48.11 5.6

Triose phosphate isomerase Solanum tuberosum, Q38JI4 107 135/100 04 26.18 6.4 27.22 5.7Phosphoglycerate kinase

(EC 2.7.2.3)Nicotiana tabacum, T03660 111 173/100 13 42.03 6.3 50.31 8.5

ADH (EC 1.1.1.1) Populus tremula, Q5ZFY1 036 108/100 07 43.96 5.8 33.09 6.1ADH 1. Putative expressed O. sativa, Q2R8Z6 136 115/100 05 44.24 6.8 38.14 6.2

Tricarboxylic acid pathway

Pyruvate decarboxylase(EC 4.1.1.1)

O. sativa, T03295 120 082/99.99 05 67.02 6.4 65.34 5.9

Putative mitochondrialNAD-dependant malatedehydrogenase

S. tuberosum, Q8L5C8 137 208/100 10 40.04 6.6 36.42 8.5

No specific pathway assigned/stress relatedh)

Glyoxalase I (EC 4.4.1.5) Z. mays, Q6XC06 012 116/100 06 36.24 5.6 32.45 5.6Aldose reductase ALDRXV4 Xerophyta viscosa, Q9XF81 108 067/99.76 06 38.52 6.3 35.87 7.7

Carbohydrate biosynthesis

UTP-glucose-1-phosphate,uridylyltransferase(EC 2.7.7.9)

Hordeum vulgare, JC4785 094 116/100 13 54.34 6.0 51.78 5.2

UTP-glucose-1-phosphate,uridylyltransferase 1 a

M. truncatula, Q1SZZ3 096 111/100 11 54.40 6.1 51.48 5.2

Starch synthase II H. vulgare, Q8H1Y8 154 060/98.84 12 35.15 6.8 87.80 7.7

Protein translation

Elongation factor 1a GTP-binding

Saccharum hybrid cultivarCP65-357, Q8SAT2

194 154/100 12 52.10 7.7 49.70 9.1

Storage protein

Glutelin Elaeis guineensis, Q9M4R4 006056089

061/98.92118/100157/100

030302

33.3933.1233.33

5.66.06.2

52.98 6.6

Stress-related

HSP70 O. sativa, Q6L509 076 298/100 20 76.78 5.8 71.90 8.7HSP70 Triticum aestivum, Q9SAU8 103 170/100 12 78.08 6.1 71.38 5.117.6 kDa class I small HSP Lycopersicon esculentum,

Q9SYV0148 128/100 06 06.28 7.0 17.63 5.8

With no assigned function

Sequence 1 fromPatent EP1217068

Elaeis oleifera, CAD42140 087 130/100 06 24.62 6.1 21.53 6.0

a) Name of the protein obtained via the MASCOT software (www.matrixscience.com) from the MSDB database.b) Species and accession number from the MSDB database.c) Spot number represents the number on the master gel (Fig. 1).d) MOWSE score probability (protein score) for the entire protein and for ions complemented by the percentage of the confidence index

(CI).e) MP, matched peptides.f) Molecular weight.g) Isoelectric point.h) Finnie et al. [44], categorized them as stress related protein.



103) and 70 kDa (spot 148) HSPs. These proteins, productsof a multigene family, act as ATP-dependent molecularchaperones, with some of them being constitutively presentin some tissues and others induced by heat or other abioticstresses [65]. They functionally cooperate in minimizingaggregation of newly synthesized proteins and facilitatingthe actual folding process [66]. A high level at early develop-mental stages for 17.6 kDa HSP spot 76, and no very bigchanges for the other two HSP (spots 103 and 148) wasobserved. These and other members of the HSP family havebeen reported to be highly expressed during zygoticembryogenesis, initiation of SE from somatic cells, micro-spores and developing pollen in alfalfa and tobacco, andfrom hypocotyls in carrots [67, 68].

Glutelin, together with prolamins, are the major storageproteins in monocots. Thus, in rice seeds, glutelin can reachup to 80% of total endosperm protein [69]. Glutelins are amultigene family, and they are initially synthesized as a57 kDa precursor and then sorted into the vacuole, wherethey are proteolytically processed into acidic (pI 5–6, 35 kDa)and basic (pI around 9, 22 kDa) subunits [70]. In our pre-vious report [8], a 22-kDa band was identified by Edman se-quencing as glutelin, corresponding to the basic subunit. Itis not observed in the 2-DE gels in the 5–8 pH range utilized,while three acidic forms were observed in this work, with pIof 5.6, 6, and 6.2. The evolution of the glutelin storage pro-teins observed here is in good agreement with that pre-viously published [8]. The presence of the glutelin storageprotein at an early developmental stage in ZE of otherangiosperms has been reported [71]. Raghavan [72] showedthat avocado embryos (P. americana Mill.) accumulated 40-,45-, and 48-kDa albumins at early developmental stages. Thesame pattern was described in carrot, in which the synthesisof storage proteins occurs as soon as embryogenesis is initi-ated and independently of any maturation background [73].In germinating tomato embryos, and after 2-DE, two mainreserve proteins were identified, vicilin and legumin;whereas the spot intensity of the former decreased duringgermination that of the latter did not change or slightlyincreased up to 120 h [42].

4 Concluding remarks

Dramatic changes in the 2-DE protein profile occurred dur-ing embryogenesis and germination in date palm ZE. Anincrease in the amount of proteins and resolved spots wasobserved from early to late embryogenesis stages, thesevalues being maximal in mature or close-to-maturityembryos. Upon germination, a decrease in both parameterswas observed.

The 2-DE map analysis revealed the existence of qualita-tive and quantitative differences in spot intensity betweenstages. The differential spots were subjected to a PCA-basedmultivariate analysis, studying the relationships betweenspots and different sampling times, and, also, the spots were

heat-map clustered to study quantity dynamics throughoutdevelopment.

Some of the variable spots were identified after MALDI-TOF-TOF analysis, and grouped into functional categories.The spot evolution pattern in the different stages was corre-lated with the following main biochemical changes: (i) stor-age protein (glutelin) accumulation during embryogenesisand consumption during germination; (ii) starch accumula-tion at the earliest embryogenic stages; (iii) change in energymetabolism from fermentation during embryogenesis torespiration after germination.

Thanks to Dr. Samuel Ogueta, Dr. Consuelo Gómez, Dr. M.A. Castillejo (Proteomics Service, University of Córdoba) for MSanalysis and technical assistance and to Dr. Ana Maldonado, Dr.Sira Echevarría-Zomeño and Prof. David Hopwood for criticalreview of the manuscript.

The authors have declared no conflict of interest.

5 References

[1] Canovas, F. M., Dumas-Gaudot, E., Recorbert, G., Jorrin, J. etal., Plant proteomic analysis. Proteomics 2004, 4, 285–298.

[2] Rossignol, M., Peltier, J. B., Mock, H. P., Matros, M. et al.,Plant proteome analysis: A 2004–2006 update. Proteomics2006, 6, 5529–5548.

[3] Jorrín, J. V., Rubiales, D., Dumas-Gaudot, E., Recorbet, G. etal., Proteomics: A promising approach to study biotic stres-ses in legumes. A review. Euphytica 2006, 147, 37–47.

[4] Jorrín, J. V., Maldonado, A., Castillejo, M. A., Plant proteomeanalysis: A 2006 update. Proteomics 2007, 7, 2947–2962.

[5] Jorrin-Novo, J. V., Plant proteomics in Europe. COST Action.Abstracts of the II Meeting, Proteomica 2008, 1, 116–155.

[6] Cairney, J., Pullman, G. S., The cellular and molecular biol-ogy of conifer embryogenesis. New Phytol. 2007, 176, 511–536.

[7] Al-Khayri, J. M., in: Jain, S. M., Gupta, P. K. (Eds.), Protocolsof Somatic Embryogenesis in Woody Plants, Springer, TheNetherlands 2005, pp. 309–320.

[8] Sghaier, B., Bahloul, M., Bouzid, G. R., Drira, N., Develop-ment of zygotic and somatic embryos of Phoenix dactyliferaL. cv. Deglet Nour: Comparative study. Sci. Hortic. 2008, 116,169–175.

[9] Raghavan, V., in: Sharma, K. K., Thorpe, T. A. (Eds.), In VitroEmbryogenesis in Plants, Kluwer Academic Publishers,Dordrecht 1995, pp. 73–115.

[10] Yadegari, R., Goldberg, R., in: Larkins, B. A., Vasil, I. K. (Eds.),Cellular and Molecular Biology of Plant Seed Development,Kluwer Academic Publishers, Dordrecht 1997, pp. 3–52.

[11] Buitink, J., Leger, J. J., Guisle, I., Vu, B. L. et al., Tran-scriptome profiling uncovers metabolic and regulatory pro-cesses occurring during the transition from desiccation-sensitive to desiccation-tolerant stages in Medicago trunca-tula seeds. Plant J. 2006, 47, 735–750.


Proteomics 2009, 9, 2543–2554 2553

[12] Merkle, S. A., Parrott, W. A., Flinn, B. S., in: Thorpe, T. A.(Ed.), In Vitro Embryogenesis in Plants, Kluwer AcademicPublishers, Dordrecht 1995, pp. 155–203.

[13] Gallardo, K., Le Signor, C., Vandekerckhove, J., Thompson,R. D., Proteomics of Medicago truncatula seed developmentestablishes the time frame of diverse metabolic processesrelated to reserve accumulation. Plant Physiol. 2003, 133,664–682.

[14] Gallardo, K., Firnhaber, Ch., Zuber, H., Héricher, D. et al., Acombined proteome and transcriptome analysis of devel-oping Medicago truncatula seeds: Evidence for metabolicspecialization of maternal and filial tissues. Mol. Cell. Prote-omics 2007, 6, 2165–2179.

[15] Firnhaber, C., Pühler, A., Küster, H., ESTsequencing and timecourse microarray hybridizations identify more than 700Medicago truncatula genes with developmental expressionregulation in flowers and pods. Planta 2005, 222, 269–283.

[16] Hajduch, M., Ganapathy, A., Stein, J. W., Thelen, J. J., Asystematic proteomic study of seed filling in soybean.Establishment of high-resolution two-dimensional refer-ence maps, expression profiles, and an interactive proteomedatabase. Plant Physiol. 2005, 137, 1397–1419.

[17] Ruuska, S. A, Girke, T., Benning, C., Ohlrogge, J. B., Contra-puntal networks of gene expression during Arabidopsisseed filling. Plant Cell 2002, 14, 1191–1206.

[18] Finnie, C., Maeda, K., Ostergaard, O., Bak-Jensen, K. S. et al.,Aspects of the barley seed proteome during developmentand germination. Biochem. Soc. Trans. 2004, 32, 517–519.

[19] Zhang, H., Sreenivasulu, N., Weschke, W., Stein, N. et al.,Large-scale analysis of the barley transcriptome based onexpressed sequence tags. Plant J. 2004, 40, 276–290.

[20] Grimanelli, D., Perotti, E., Ramirez, J., Leblanc, O., Timing ofthe maternal-to-zygotic transition during early seed devel-opment in maize. Plant Cell 2005, 17, 1061–1072.

[21] Méchin, V., Thévenot, C., Le Guilloux, M., Prioul, J. L. et al.,Developmental analysis of maize endosperm proteomesuggests a pivotal role for pyruvate orthophosphate diki-nase. Plant Physiol. 2007, 143, 1203–1219.

[22] Winkelmann, T., Heintz, D., Dorsselaer, A. V., Serek, M. et al.,Proteomic analyses of somatic and zygotic embryos ofCyclamen persicum Mill. reveal new insights into seed andgermination physiology. Planta 2006, 224, 508–519.

[23] Sánchez-Romero, C., Perán-Quesada, R., Barceló-Muñoz, A.,Pliego-Alfaro, F., Variations in storage protein and carbohy-drate levels during development of avocado zygoticembryos. Plant Physiol. Biochem. 2002, 40, 1043–1049.

[24] Pretová, A., Obert, B., Hajduch, M., Gregová, E., Total proteinand isozyme characterization in the flax zygotic embryoduring development. Sex Plant Reprod. 2001, 13, 329–334.

[25] Chandra Sekhar, K. N., DeMason, D. A., Quantitative ultra-structure and protein composition of date palm (Phoenixdactylifera) seeds: A comparative study of endosperm vs.embryo. Am. J. Bot. 1988, 75, 323–329.

[26] Gómez-Vidal, S., Tena, M., Lopez-Llorca, L. V., Salinas, J.,Protein extraction from Phoenix dactylifera L. leaves, arecalcitrant material, for two-dimensional electrophoresis.Electrophoresis 2008, 29, 448–456.

[27] Carpentier, S. C., Panis, B., Vertommen, A., Swennen, R.,Sergeant, K. et al., Proteome analysis of nonmodel plants: Achallenging but powerful approach. Mass Spectrom. Rev.2008, 27, 354–377.

[28] Maldonado, A. M., Echevarría-Zomeño, S., Jean-Baptiste,S., Hernández, M. et al., Evaluation of three different precip-itation protocols of protein extraction for Arabidopsis thali-ana leaf proteome analysis by two-dimensional electropho-resis. J. Proteomics 2008, 71, 461–472.

[29] Wang, W., Vignani, R., Scali, M., Cresti, M., A universal andrapid protocol for protein extraction from recalcitrant planttissues for proteomic analysis. Electrophoresis 2006, 27,2782–2786.

[30] Ramagli, L. S., Rodríguez, L. V., Quantitation of microgramamounts of protein in two-dimensional polyacrylamide-gelelectrophoresis sample buffer electrophoresis. Electropho-resis 1985, 6, 559–563.

[31] Mathesius, U., Keijzers, G., Natera, S. H. A., Djordjevic, M. A.et al., Establishment of a root proteome reference map forthe model legume Medicago truncatula using the expressedsequence tag database for peptide mass fingerprinting.Proteomics 2001, 1, 1424–1440.

[32] Chich, J. F., David, O., Villers, F., Schaeffer, B. et al., Statisticsfor proteomics: Experimental design and 2-DE differentialanalysis. J. Chromatogr. B 2007, 849, 261–272.

[33] Meunier, B., Dumas, E., Piec, I., Béchet, D. et al., Assessmentof hierarchical clustering methodologies for proteomic datamining. J. Proteome Res. 2007, 6, 358–366.

[34] De Vienne, D., Bost, B., Fiévet, J., Zivy, M. et al., Geneticvariability of proteome expression and metabolic control.Plant Physiol. Biochem. 2001, 39, 271–283.

[35] Jorge, I., Navarro, R. M., Lenz, Ch., Ariza, D. et al., The HolmOak leaf proteome: Analytical and biological variability inthe protein expression level assessed by 2-DE and proteinidentification tandem mass spectrometry de novo sequenc-ing and sequence similarity searching. Proteomics 2005, 5,222–234.

[36] Valledor, L., Castillejo, M. A., Lenz5et, C., Rodríguez, R. et al.,Proteomic analysis of Pinus radiata needles: 2-DE map andprotein identification by LC/MS/MS and substitution-toler-ant database searching. J. Proteome Res. 2008, 7, 2616–2631.

[37] Gustafsson, J. S., Robert, C., Glasbey, C. A., Blomberg, A.,Rudemo, M., Statistical exploration of variation in quantita-tive two-dimensional gel electrophoresis data. Proteomics2004, 4, 3791–3799.

[38] Caraux, G., Pinloche, S., Permutmatrix: A graphical envi-ronment to arrange gene expression profiles in optimal lin-ear order. Bioinformatics 2005, 21, 1280–1281.

[39] Schevchenko, A., Wilm, M., Vorm, O., Mann, M., Massspectrometric sequencing 23 of proteins silver-stainedpolyacrylamide gels. Anal. Chem. 1996, 68, 850–858.

[40] Sallandrouze, A., Faurobert, M., El Maâtaoui, M., Character-ization of the developmental stages of cypress zygoticembryos by two-dimensional electrophoresis and by cyto-chemistry. Physiol. Plant 2002, 114, 608–618.

[41] Horikoshi, M., Morita, Y., Effect of cycloheximide on proteinbreakdown in germinating rice seeds. Agric. Biol. Chem.1981, 45, 527–529.

[42] Sheoran, I. S., Olson, D. J. H., Ross, A. R. S., Sawhney, V. K.,Proteome analysis of embryo and endosperm from germi-nating tomato seeds. Proteomics 2005, 5, 3752–3764.

[43] Aberlenc-Bertossi, F., Sané, D., Daher, A., Borgel, A., Duval,Y., Aptitude à la déshydratation des embryons zygotiques de



palmier à huile et de palmier dattier: étude de l’expressionde gènes LEA. Les Actes du BRG 2006, 6, 401–413.

[44] Finnie, C., Bak-Jensen, K. S., Laugesen, S., Roepstorff, P. etal., Differential appearance of isoforms and cultivar varia-tion in protein temporal profiles revealed in the maturingbarley grain proteome. Plant Sci. 2006, 170, 808–821.

[45] Jacobsen, S., Grove, H., Jensen, K. N., Sørensen, H. A. et al.,Multivariate analysis of 2-DE protein patterns – Practicalapproaches. Electrophoresis 2007, 28, 1289–1299.

[46] Verhoeckx, K. C. M., Bijlsma, S., Van der Greef, J., Witkamp,R. F. et al., In search of secreted protein biomarkers for theanti-inflammatory effect of beta2-adrenergic receptor ago-nists: Application of DIGE technology in combination withmultivariate and univariate data analysis tools. J. ProteomeRes. 2005, 4, 2015–2023.

[47] Rodriguez-Piñeiro, A. M., Rodríguez-Berrocal, F. J., Páez deLa Cadena, M., Improvements in the search for potentialbiomarkers by proteomics: Application of principal compo-nent and discriminant analyses for two-dimensional mapsevaluation. J. Chromatogr. B 2007, 849, 251–260.

[48] Smit, S., van Breemen, M. J., Hoefsloot, H. C. J., Smilde, A.K. et al., Assessing the statistical validity of proteomicsbased biomarkers. Anal. Chim. Acta 2007, 592, 210–217.

[49] Chan, Z., Wang, Q., Xu, X., Meng, X., Qin, G. et al., Functionsof defense-related proteins and dehydrogenases in resist-ance response induced by salicylic acid in sweet cherryfruits at different maturity stages. Proteomics 2008, 8, 4791–4807.

[50] Valcu, C. M., Lalanne, C., Plomion, C., Schlink, K., Heatinduced changes in protein expression profiles of Norwayspruce (Picea abies) ecotypes from different elevations.Proteomics 2008, 8, 4287–4302.

[51] Röhrig, H., Colby, T., Schmidt, J., Harzen, A., Facchinelli, F. etal., Analysis of desiccation-induced candidate phosphopro-teins from Craterostigma plantagineum isolated with amodified metal oxide affinity chromatography procedure.Proteomics 2008, 8, 3548–3560.

[52] Führs, H., Hartwig, M., Elisa, L., Molina, B., Heintz, D. et al.,Early manganese-toxicity response in Vigna unguiculata L. –A proteomic and transcriptomic study. Proteomics 2008, 8,149–159.

[53] Vensel, W. H., Tanaka, C. K., Cal, N., Wong, J. H. et al.,Developmental changes in the metabolic protein profiles ofwheat endosperm. Proteomics 2005, 5, 1594–1611.

[54] Mack, A., Clark, T. K., Forrey, R. C., Balakrishnan, N. et al.,Cold He1H2 collisions near dissociation. Phys. Rev. 2006,74, 052718.

[55] Barthels, D., Targeting detoxification pathways: An efficientapproach to obtain plants with multiple stress tolerance?Plant Sci. 2001, 6, 284–286.

[56] Veena, V. S., Reddy, S. K., Sopory, glyoxalase I from Brassicajuncea: Molecular cloning, regulation, and its over-expres-sion confer tolerance in transgenic tobacco under stress.Plant J. 1999, 17, 385–395.

[57] Thornalley, P. J., The glyoxalase system: New developmentstowards functional characterization of a metabolic pathwayfundamental to biological life. Biochem. J. 1990, 269, 1–11.

[58] Yang, W., Tan, B., Huang, D., Rautiainen, M. et al., MODISleaf area index products: From validation to algorithmimprovement. IEEE Trans. Geosci. Rem. Sens. 2006, 44,1885–1889.

[59] Karuna, S. B., Rajendrakumar, C. S., Reddy, A. R., Aldosereductase in rice (Oryza sativa L.): Stress response anddevelopmental. Plant Sci. 2000, 160, 149–157.

[60] Nakamura, Y., Towards a better understanding of metabolicsystem for amylopectin biosynthesis in plants: Rice endo-sperm as a model tissue. Plant Cell Physiol. 2002, 43, 718–725.

[61] Wang, T. L., Bogracheva, T. Y., Hedley, C. L., Starch: As sim-ple as A, B, C. J. Exp. Bot. 1998, 49, 481–502.

[62] Prewein, C., Endemann, M., Reinohl, V., Salaj, J. et al.,Physiological and morphological characteristics duringdevelopment of pedunculate oak (Quercus robur L.) zygoticembryos. Trees 2006, 20, 53–60.

[63] Kleczkowski, L. A., Glucose activation and metabolismthrough UDPglucose pyrophosphorylase in plants. Phy-tochemistry 1994, 37, 1507–1515.

[64] Kleczkowski, L. A., Geisler, M., Ciereszko, I., Johansson, H.,UDP-glucose pyrophosphorylase. An old protein with newtricks. Plant Physiol. 2004, 134, 912–918.

[65] Lee, J. H., Schöffl, F., An Hsp70 antisense gene affects theexpression of HSP70/HSC70, the regulation of HSF, and theacquisition of thermotolerance in transgenic Arabidopsisthaliana. Mol. Gen. Genet. 1996, 252, 11–19.

[66] Fink, A. L., Chaperone-mediated protein folding. Physiol.Rev. 1999, 79, 425–449.

[67] Kitamya, E., Suzuki, S., Sano, T., Nagata, N., Isolation of twogenes that were induced upon the initiation of somaticembryogenesis on carrot hypocotyls by high concentrationof 2,4-D. Plant Cell. Rep. 2000, 19, 551–557.

[68] De Rocher, A. E., Vierling, E., Developmental control of smallheat shock protein expression during pea seed maturation.Plant J. 1994, 5, 93–102.

[69] Takaiwa, F., Ogawa, M., Okita, T. W., in: Shewey, P. R., Casey,R. (Eds.), Seed Proteins. Kluwer Acadamic Publishers,Dordrecht 1999, pp. 401–425.

[70] Yamagata, H., Sugimoto, T., Tanaka, K., Kasai, Z., Biosyn-thesis of storage proteins in developing rice seeds. PlantPhysiol. 1982, 70, 1094–1100.

[71] Rahman, S., Shewry, P. R., Miflin, B. J., Differential proteinaccumulation during barley grain development. J. Exp. Bot.1982, 33, 717–728.

[72] Raghavan, V., in: Raghvan, V. (Eds.), Molecular Embryologyof Flowering Plant. Cambridge University Press, Cambridge1997, pp. 440–464.

[73] Dodemann, V. L., Ducreux, G., Kreis, M., Zygotic embryo-genesis versus. J. Exp. Bot. 1997, 48, 1493–1509.


Documents

Proteomic analysis of the development and germination of date palm (Phoenix dactylifera L.) zygotic embryos