UBC Faculty Research and Publications

Generation of a predicted protein database from EST data and application to iTRAQ analyses in grape (Vitis… Lücker, Joost; Laszczak, Mario; Smith, Derek; Lund, Steven T Jan 26, 2009

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12864_2008_Article_1934.pdf [ 1.34MB ]
JSON: 52383-1.0220539.json
JSON-LD: 52383-1.0220539-ld.json
RDF/XML (Pretty): 52383-1.0220539-rdf.xml
RDF/JSON: 52383-1.0220539-rdf.json
Turtle: 52383-1.0220539-turtle.txt
N-Triples: 52383-1.0220539-rdf-ntriples.txt
Original Record: 52383-1.0220539-source.json
Full Text

Full Text

ralssBioMed CentBMC GenomicsOpen AcceResearch articleGeneration of a predicted protein database from EST data and application to iTRAQ analyses in grape (Vitis vinifera cv. Cabernet Sauvignon) berries at ripening initiationJoost Lücker1, Mario Laszczak2, Derek Smith3 and Steven T Lund*1Address: 1Wine Research Centre, Faculty of Land and Food Systems, University of British Columbia, Vancouver, BC, Canada, 2Bioinformatics Group, Canada's Michael Smith Genome Sciences Centre, BC Cancer Agency, Vancouver, BC, Canada and 3University of Victoria-Genome British Columbia Proteomics Centre, Victoria, BC, CanadaEmail: Joost Lücker - jlucker@gmail.com; Mario Laszczak - mlaszczak@hotmail.com; Derek Smith - derek@proteincentre.com; Steven T Lund* - stlund@interchange.ubc.ca* Corresponding author    AbstractBackground: iTRAQ is a proteomics technique that uses isobaric tags for relative and absolutequantitation of tryptic peptides. In proteomics experiments, the detection and high confidenceannotation of proteins and the significance of corresponding expression differences can depend onthe quality and the species specificity of the tryptic peptide map database used for analysis of thedata. For species for which finished genome sequence data are not available, identification ofproteins relies on similarity to proteins from other species using comprehensive peptide mapdatabases such as the MSDB.Results: We were interested in characterizing ripening initiation ('veraison') in grape berries at theprotein level in order to better define the molecular control of this important process for grapegrowers and wine makers. We developed a bioinformatic pipeline for processing EST data in orderto produce a predicted tryptic peptide database specifically targeted to the wine grape cultivar, Vitisvinifera cv. Cabernet Sauvignon, and lacking truncated N- and C-terminal fragments. By searchingiTRAQ MS/MS data generated from berry exocarp and mesocarp samples at ripening initiation, wedetermined that implementation of the custom database afforded a large improvement in highconfidence peptide annotation in comparison to the MSDB. We used iTRAQ MS/MS in conjunctionwith custom peptide db searches to quantitatively characterize several important pathwaycomponents for berry ripening previously described at the transcriptional level and confirmedexpression patterns for these at the protein level.Conclusion: We determined that a predicted peptide database for MS/MS applications can bederived from EST data using advanced clustering and trimming approaches and successfullyimplemented for quantitative proteome profiling. Quantitative shotgun proteome profiling holdsgreat promise for characterizing biological processes such as fruit ripening initiation and may befurther improved by employing preparative techniques and/or analytical equipment that increasepeptide detection sensitivity via a shotgun approach.Published: 26 January 2009BMC Genomics 2009, 10:50 doi:10.1186/1471-2164-10-50Received: 2 August 2008Accepted: 26 January 2009This article is available from: http://www.biomedcentral.com/1471-2164/10/50© 2009 Lücker et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 17(page number not for citation purposes)BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50BackgroundThe field of protein discovery through mass spectrometry(MS) continues to grow rapidly but the number of speciesfor which finished (i.e. > 98% complete) whole genomesequence data are available is currently not keeping pace.For a large number of laboratories worldwide studyingproteomes in 'non-mainstream' organisms, annotationsof tandem mass spectra data must rely on open readingframe (ORF) predictions from expressed sequence tag(EST) data from their species of interest or a phylogeneti-cally close relative. ESTs generated from single passsequencing reactions are frequently not full length and thereading frames are unknown. Proteolytic peptidesequence databases derived from multiple, truncated pre-dicted ORFs per each of thousands of ESTs can hamper theability of search engines such as MASCOT [1] and algo-rithms such as Paragon in ProteinPilot software [2] tomake statistically robust protein identifications from MS/MS spectrum data [3]. Protein identifications from MS/MS spectra may be further complicated when the EST datathat are used to build a peptide sequence database are cre-ated based on one genotype for a given species. We reporthere on the development of scripts for the generation of apredicted tryptic peptide sequence database based on ESTdata in grapevine. Our computational approach accountsfor multiple open reading frames, truncated predictedORFs, and the presence of N-terminal signal peptides, andmay be useful for MS/MS-based protein discovery in anyspecies for which EST data are available.Quantitative protein expression profiling analyses inplants have increasingly implemented stable isotopiclabeling as an advance or complement to two dimen-sional gel electrophoresis (2DGE) methods. Isotopecoded affinity tagging (ICAT) reagents are used to cova-lently label cysteine residues with heavy or light hydrogenor carbon in two complex peptide samples, for example,wild type versus mutant genotypes. The ICAT chemistry isused to purify labeled peptides via affinity chromatogra-phy and then samples are mixed and subjected to LC-MS/MS [4]. One of the first reports on an ICAT application inplants was in wheat (Triticum aestivum L.) where relativeexpression in monosomic deletion mutants was used tobegin to clarify the influence of ancestral genomes on dif-ferential seed protein expression for breeding applications[5]. The ICAT technique is limited, however, by the tag-ging of cysteine residues only, as well as the need for affin-ity purification of labeled peptides; invariably,information is lost through these steps. An improvementto the ICAT technique involves the labeling of aminegroups using a set of four or more isobaric tags. Theadvantages of this technique, isobaric tagging for relativeand absolute quantitation (iTRAQ), are that most pep-identical peptides that are differentially tagged, therebyenriching detection sensitivity and accuracy in compari-son to ICAT [6]. Few reports of iTRAQ implementation inplant proteome studies have been reported but pioneeringwork in this field has been successful, for example, in fur-ther defining the organellar proteome in Arabidopsis [7],characterizing pathogen defense mechanisms in Arabi-dopsis [8], and clarifying micronutrient stress responses inbarley (Hordeum vulgare) [9].We were interested in characterizing ripening initiation ingrape berries at the level of differential protein expressionin order to better define the molecular control of thisimportant process for grape growers and wine makers.Grape berry ripening is non-climacteric and ethylene doesnot act as a major signal initiating this process, as it doesin climacteric species such as tomato (Lycopersicon esculen-tum). Abscisic acid (ABA), hexoses, and brassinosteroids(BRs) have previously been implicated in non-climactericripening regulation but how these and potentially othersignaling pathways interact to effect major changes inberry biochemistry at ripening initiation is poorly under-stood. The tissues in the grape berry consist of the seeds,the mesocarp (flesh), and the exocarp (skin); the pericarprefers to the mesocarp and exocarp, collectively. Primaryand secondary compounds important for grape and wineproducts begin to accumulate in the exocarp and/or themesocarp at ripening initiation [10], so we consideredthat it was important to evaluate changes in the berry pro-teome separately in these tissues. To date, a limitednumber of reports on proteome profiling in grapevineand grape berries have been published in which 2DGEwas employed [11-14]. We considered that the iTRAQtechnique could be useful in surmounting some technicallimitations encountered with 2DGE and allow us to detecta greater number of proteins per sample. In this report, wedemonstrate the application of our computationalapproach to tryptic peptide sequence database develop-ment from a large collection of grapevine EST data andvalidate its usefulness by showing improved detectionand annotations of MS/MS data derived from grape exo-carp and mesocarp total protein extracts. We further pro-vide new quantitative information on differential proteinexpression during ripening initiation in grape berries. Thisis the first report in which iTRAQ has been used to studydifferential protein expression in any fruit.MethodsPlant materialGrape clusters were sampled from V. vinifera cv. CabernetSauvignon clone 15 grafted on rootstock 101-14 in a com-mercial vineyard near Osoyoos, British Columbia, in the2004 and 2005 seasons. Sampling dates during each sea-Page 2 of 17(page number not for citation purposes)tides are labeled, no affinity purification step is required,and the isobaric nature of the tags allows co-elution ofson were focused on the developmental stages undergoingripening initiation. Clusters were sampled on a single dateBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50in 2004, August 12th, which was the timing of approxi-mately 50% ripening initiation based on a turning pinkcolor phenotype. For the 2005 season, the ripening initia-tion stage was sampled over a longer period (August 10ththrough August 16th), since in this growing season, ripen-ing advanced slowly due to lower atmospheric tempera-tures. Five clusters from five different vines were sampledin each season and snap frozen directly in liquid nitrogenin the vineyard and then transported on dry ice to UBCVancouver where they were stored at -80°C.Individual grapes from each of the 2004 and 2005 clusterswere developmentally staged based on a visual pigmenta-tion assessment and were segregated for each season intogreen, pink/turning, fully turned red, and fully turnedpurple phenotypic classes. For the 2005 samples, greengrapes were only taken from clusters collected on August10th, since for this date and August 12th there was no visi-ble change in color present in any of the grape clusters.Thirty grapes of similar sizes per pigmentation class peryear were segregated for experimentation. Prior to totalprotein extraction, individual grapes were partiallythawed in gloved hands and then, using a forceps, the exo-carp tissue was carefully peeled away from the mesocarpand placed immediately into liquid nitrogen. Seeds werethen carefully removed while keeping the remaining mes-ocarp tissue frozen in liquid nitrogen. Exocarp and meso-carp samples were ground to a powder under liquidnitrogen and then used for total protein extractions.Tissue preparation for protein extractionPreparation of exocarp tissue samples for protein extrac-tion was performed according to a previously describedprotocol for olive leaf [15] with some modificationsdescribed here. The procedure was carried out on ice andcentrifugations were performed at 4°C. Throughout theprocedure, each wash was done by complete resuspendingof the tissue pellet. Four hundred mg of powdered exocarptissue was placed in a 2 mL G-tube (Fisher Scientific Can-ada, Ottawa, ON). The tissue was suspended in 1.5 mL ofa cold (-20°C) ethyl acetate:ethanol (1:2 (v:v)) solutionby vortexing for 30 s; the ethyl acetate:ethanol extractionwas previously found to be useful for removing pectins aswell as pigments such as chlorophylls [16]. Following cen-trifugation for 3 min at 21000 × g, the supernatant wasremoved and the ethyl acetate:ethanol extraction and cen-trifugation steps were repeated on the remaining tissue.The sample was next extracted twice with cold (-20°C)100% acetone by vortexing and centrifuging, as before.Subsequently, the tissue with added acetone was trans-ferred from the G-tube to a mortar using a 1 mL pipettewith the tip end excised to increase diameter and then theacetone was evaporated from the tissue at room tempera-was ground to an even finer powder. The powder wastransferred back to a clean 2 mL G-tube by suspending thetissue in 1.5 mL of cold (-20°C) TCA:acetone (1:9 (v/v))and vigorously mixed and centrifuged, as before. Extrac-tion with 10% TCA:acetone was repeated five to seventimes, or until no more anthocyanins (red-pigmented fla-vonoids) could be extracted from the tissue. This was fol-lowed by three washes with chilled (4°C) 10% TCA inwater by vigorous mixing and centrifugation, as before, toextract the pectins and remaining anthocyanins from thetissue. After this, the tissue was washed twice with cold (-20°C) 80% acetone and centrifuged, as before. Proteinextraction was performed after drying the tissue pellet tocompletion in a speed vacuum extractor (SPD131DDA,Thermo Scientific, Milford, MA, USA).For the preparation of the mesocarp tissue, the same pro-cedure for the exocarp was used with the following modi-fications. Three g of starting material was used per sampleand the first extractions up to the grinding step with whitequartz were done in 50 mL Oakridge tubes. Since someprotein can be extracted from the mesocarp via TCA:ace-tone extraction alone [14], a 20 min incubation time at -20°C was introduced after the first 100% acetone step andincluded in the subsequent TCA:acetone containing stepsto ensure that all of the protein remained precipitated. Inthe TCA:H2O step, the 20 min incubation was done onice. Since no anthocyanins are present in mesocarp, onlytwo TCA:acetone extractions were carried out for the mes-ocarp tissue.Total protein extractionTwo hundred to 300 mg of pre-extracted and dried exo-carp or mesocarp tissue contained in a 2 mL G-tube wasextracted by resuspending the pellet in 0.75 mL cold Tris-buffered phenol, pH 7.9. Then, 0.75 mL of dense SDSbuffer (30% sucrose, 2% SDS, 0.1 M Tris-HCL, pH 8.0)was added. The mixture was vortexed for 30 s and incu-bated on ice for 40 min with intermittent vortexing. Thephenol phase containing the protein as the top phase wasseparated by centrifugation at 21000 × g for 5 min andtransferred into a clean 2 mL G-tube. The remaining SDSphase was re-extracted with another 0.75 mL Tris-bufferedphenol and incubated for 20 min before centrifuging andsubsequent transfer and combination of the two phenolphases. Protein was precipitated by adding a minimum of5 vol cold methanol plus 0.1 M ammonium acetate to thecombined phenol phase. Precipitation was carried out at -20°C for 30 min or overnight. After centrifugation at21000 g for 10 min, the pellet was washed twice with coldmethanol containing 0.1 M ammonium acetate and sub-sequently with 80% acetone twice. Pellets were next dis-solved in 200–300 μL fresh buffer containing 6 M urea,Page 3 of 17(page number not for citation purposes)ture. After the addition of 1/3 vol of white quartz sand(Sigma-Aldrich, Oakville, ON, Canada) to the tissue, it2% CHAPS, 5 mM EDTA, and 30 mM HEPES, pH 8.1, toBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50obtain a concentration of approximately 1.0 μg/μL. Care-ful sonication on ice was used to dissolve the samples.Protein quantitation was done using a bicinchoninic acid(BCA) absorption assay (Sigma-Aldrich Canada Ltd.,Oakville, ON) and read in a Victor V plate reader (Perk-inElmer Life and Analytical Sciences, Woodbridge, ON,Canada) equipped with a photometric filter of 560 nmand 10 nm bandwidth. The quality of each protein samplewas checked via SDS-PAGE; all samples were devoid ofindications of degradation and showed good resolutionwith low background. Total protein samples were thenshipped on dry ice to the University of Victoria-GenomeBC Proteomics Centre in Victoria, BC, for iTRAQ analyses.Using a second BCA assay, each protein sample was re-quantified just before aliquoting 100 μg of each samplefor iTRAQ labeling steps.Experimental design and labeling of peptides with iTRAQ reagentsThe experimental design consisted of the four develop-mental stages described earlier for each of exocarp 2004,mesocarp 2004, exocarp 2005, and mesocarp 2005. Twobiological replicates were employed for each stage and tis-sue for the 2005 samples, whereas one 2004 sample wasused for each stage of mesocarp or exocarp. An additionaltechnical replicate was carried out for exocarp 2004, rep-resenting separate iTRAQ labeling reactions and analysesstarting from the same protein sample.Labeling of peptides with iTRAQ reagents (Applied Bio-systems Canada, Streetsville, ON) was performed accord-ing to the manufacturer's recommendations as follows.One hundred μg of each protein sample in a maximumvolume of 200 μL was precipitated overnight using 100%acetone and dissolved in 20 μL of denaturing buffer con-taining 1 μL denaturant and 2 μL reducing reagent (TCEP)as provided in the iTRAQ kit, followed by vortexing andincubation at 60°C for 1 h. One μL of cysteine blockingsolution (MMTS) was then added to each sample, fol-lowed by incubation at room temperature for 10 min.These protein samples were digested overnight withtrypsin (Promega, Madison, WI, USA) at 37°C. iTRAQlabeling was carried out by adding iTRAQ reagents 114,115, 116, and 117 to either the exocarp or the mesocarpsamples representing the four developmental stages,green, pink/turning stage, red/fully turned, and purple,respectively. Subsequently, these four samples were mixedby vortexing and further incubated at room temperaturefor 1 h.The four iTRAQ-labeled peptide samples were pooledtogether, diluted 1:10 with cation exchange sample bufferthis acidification step, it is important to remove pectinsprior to total protein extractions; we found in previous tri-als that the pectins likely polymerized and precipitatedout of solution, converting samples mostly to a gelatinousstate unsuitable for further analyses (data not shown). Thecombined peptide mixture was fractionated by strong cat-ion exchange (SCX) chromatography on a BioCAD work-station (Applied Biosystems), using a 4.6 mm × 20 cmpolysulfoethyl aspartamide column (PolyLC Inc, Colum-bia, MD, USA). First, the mixed samples were loaded inbuffer A at a flow rate of 0.2 mL/min. Once completelyloaded, the column was washed for 20 min with buffer A.Peptides were eluted by a linear gradient of 0 to 350 mMKCl in buffer B (20 mM KH2PO4, 25% acetonitrile, pH3.0). Sixty-nine fractions were collected over the course of70 min at a flow rate of 1 mL/min. Of these fractions, only12 fractions containing the eluted labeled peptides asmeasured by optical density monitoring at 214 nm werechosen for analysis on a 2 h LC-MS/MS program. The frac-tionated samples were reduced to 150 μL in a speed-vac(Thermo-Savant, Holbrook, NY, USA) and transferred toautosampler tubes (LC Packings, Amsterdam, The Nether-lands).Liquid chromatography and mass spectrometryThe samples were analyzed for identification and quanti-tation on a QSTAR Pulsar i hybrid tandem mass spectrom-etry (LC-MS/MS) system (Applied Biosystems, MDSSciex), fitted with a nano-electrospray ionization source(Proxeon, Odense, Denmark) using a 10 μm fused silicaemitter tip (New Objectives, Woburn, MA, USA) andinterfaced with an integrated LC system consisting of aFamos autosampler, SwitchOS II switching pump, andUltimate micropump (LC Packings). Individual fractionscontaining peptides were injected onto a 300 μm × 5 cmC18 PepMap guard column (5 μm, 100A; LC Packings),resolved using a 75 μm × 150 mm analytical column (3μm, 100A; LC Packings), and eluted using an automatedbinary gradient (200 nL/min) from 100% buffer A (2%acetonitrile (ACN), 0.05% formic acid in H2O) to 40%buffer B (0.05% formic acid in 98% ACN) in 40 min, thenfrom 40% to 80% buffer B for 5 min. MS time of flight(TOF) scans were acquired from m/z 400 to 1200 for onesecond with up to two precursors selected for MS/MS fromm/z 100 to 1500 using information-dependent acquisi-tion at 2.5 seconds per scan; rolling collision energy wasused to promote fragmentation.Custom predicted tryptic peptide databaseA schema showing the pipeline for production of the pre-dicted peptide database in support of this subsection isshown in Figure 1. All publicly available EST data for eachVitis species (AS, all sequences), including those from allPage 4 of 17(page number not for citation purposes)(A) containing 25% acetonitrile in 10 mM KH2PO4, andthen adjusted to pH 3.0 using phosphoric acid. Because ofV. vinifera (wine grape) cultivars, were downloaded inAugust 2007 as FASTA files from the National Center forBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50Biotechnology Information (NCBI, Bethesda, MD, USA).These data were parsed on the basis of reported Vitis spe-cies of origin with the vast majority being from V. viniferacultivars. Since we were specifically interested in studyingthe proteome in V. vinifera cv. Cabernet Sauvignon (CS)pericarp tissue, an additional, more rigorous approach tothe parsing of the CS ESTs was carried out in order toreduce or eliminate the potential for subsequent assemblyof paralogous CS sequences into invalid contigs, therebythe NCBI Genbank database or from an in-house ESTproject [17] and subdivided into the following categoriesbased upon the reported source tissues for the cDNAsused for single pass sequencing: Whole berry includingseed (CSB), berry without seed (pericarp, CSP), skin with-out seed or flesh (exocarp, CSE), seed only (CSS), andother tissues (CSO) including leaf, flower, tendril, androot. Because the in-house ESTs were also present in theNCBI Genbank database, the corresponding entries inSchema of the workflow in construction of the predicted peptide database based upon Vitis spp. ESTsFigure 1Schema of the workflow in construction of the predicted peptide database based upon Vitis spp. ESTs. Steps in EST selection (yellow), EST curation (orange), contig assembly (green), and translation, start methionine prediction, tryptic cleavage site prediction, and removal of predicted truncated N- and/or C-terminal peptides (blue) are shown. The in-house "GrapeGen" EST database was derived from the V. vinifera cv. Cabernet Sauvignon and Muscat Hamburg cDNAs [17]. Where a CS EST was duplicated between the Genbank and in-house databases, the Genbank EST was removed and the EST from the in-house database, containing the phred scores, was used for clustering. CS = Cabernet Sauvignon ESTs; MH = Muscat Hamburg ESTs; Vv = Vitis vinifera; Wild = Vitis spp. ESTs other than V. vinifera. A key to the codes used in 'Cluster ORF ID' and 'Protein Annotation' in Additional files 1 through 8 is presented in Additional file 10 and may be printed out as a quick reference when examining the iTRAQ results; tissue types (i.e. CS_) are described in Methods and Additional file 10.Page 5 of 17(page number not for citation purposes)striving to strengthen the validity of protein identificationin our iTRAQ experiments. CS ESTs were obtained fromGenbank were removed since the Genbank entries do nothave sequence quality (phred [18]) scores. The followingBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50files containing EST data comprised each of the abovementioned groups: VV (VV.fasta representing all V. vinif-era ESTs including in-house ESTs); WS (WS.fasta includ-ing ESTs from all available wild species, V. aestivalis; V.cinerea × V. riparia; V. cinerea × V. rupestris; Vitis hybrid(species not indicated in Genbank); V. labrusca; V. pseudor-eticulata; V. riparia; V. rotundifolia; V. shuttleworthii), CSO(Bud.fasta; Flower_leaf _root.fasta; Leaf_blade.fasta; Peti-ole.fasta; Root.fasta; Flower-Pre-bloom.fasta;Inflorescence_including_flowers.fasta; Stem.fasta;Nectary_of_flowers.fasta; Flower_Bloom.fasta; Leaf.fasta;Inflorescence.fasta), CSS (Seed.fasta), CSP (Pericarp.fasta;Fruit_with_seeds_removed.fasta;Fruit_without_seeds.fasta), CSE (Fruit_skin.fasta), andCSB (Berry.fasta; Fruit.fasta).Sequences were processed using cross_match (minmatch12, penalty -2, minscore 20; http://www.phrap.org) andtrim2 (G. Williams, http://emboss.sourceforge.net/apps/release/5.0/emboss/apps/trimseq.html) in order toremove vector sequences as well as ambiguous nucle-otides at the sequence ends. To successfully perform theabove cleanup analyses, phred quality scores were usedwhere available; otherwise, 'place-holder' quality scoreswere generated for any sequences for which no phredscores were available, as was the case for most of the ESTsin Genbank. Place-holder quality scores were also usedlater in the cluster assembly process as discussed in moredetail, below. Following the cross_match and trim2processing, the sequences were further trimmed using Perlscripts designed in-house to eliminate known invalidsequences (e.g. microbial sequences, simple sequencerepeats) and trim polyA/T tails, if present in a givensequence. PolyA/T stretches were limited to 12 bp in orderto prevent subsequent chimeric contig assembly based onthose repeats. If polyA was followed by a > 30 bp stretchof AC, AT, GC, or GT repeats, the polyA stretch wastrimmed to 12 bp and all sequence 3' to this was dis-carded; if polyT was preceded by a > 30 bp stretch of AC,AT, GC, or GT repeating sequence, the polyT stretch wastrimmed to 12 bp and all sequence 5' to this was dis-carded. If polyA started at least two thirds of the ESTsequence length, it was trimmed to 12 bp; if polyT startedat less than one third of the EST sequence, it was trimmedto 12 bp. Any part of a sequence that started or ended with> 30 bp of repeats of AC, AT, GC, or GT was deleted. If asequence started or ended with 'N's (indicating ambigu-ous base calls), the 'N's were deleted and the correspond-ing quality scores were also removed.To better ensure that contig assemblies were based onhigh quality nucleotide sequence data, percent 'N'(ambiguous base call) content was determined for eachscanned for 'N's and, if present, were trimmed to excludethe 'N's, thereby lowering the total 'N' percentage.Sequences shorter than 200 bp were trimmed to the firstand last occurrences of an 'N'. For resulting sequenceslonger than 50 bp, the 'N' percentage was recalculatedand, if still > 0.3%, a record of the sequence was made.Each of these sequences was then compared with othersequences in a combined dataset using BLASTN to deter-mine its uniqueness. If a given sequence was already rep-resented in the dataset by another sequence with a lower'N' content, the sequence in question was eliminated.The curated sequence datasets were next clustered usingPCAP software [19] with parameters of 95% overlap iden-tity and 60 bp overlap length [17]. PCAP was used insteadof CAP3 in order to take advantage of parallelized process-ing. Parallelization provided the ability to distribute eachdataset assembly workload across 100 CPUs for signifi-cantly faster processing time. The PCAP assembly programwas modified and recompiled with EST_flag set at 1 (thedefault is 0, which indicates genomic reads). The PCAPassembly step was followed by a series of post-assemblysteps (bdocs -y 100 -z 0, bclean -y 100 -w 1, bcontig -y 100-p 95, bconsen -y 100 -z 0 -p 95, bform -y 100). We per-formed two clustering permutations in order to test theeffects of database design on peptide identification usingour iTRAQ data. First, we clustered all sequences togetherto create the "AS" database, including WS, VV, and all CS_files; all sequences were weighted evenly. Second, CSB,CSE, CSP, CSO, CSS, WS, and VV (including CSsequences) were clustered separately with higher weight-ing (place-holder scores) placed on CS sequences in theVV build and the original phred scores retained for the in-house CS sequences. Weighting was accomplished byassigning higher quality scores such that when polymor-phisms were encountered by PCAP in an assembly, prefer-ence was given for selection of the CS nucleotide for theresulting contig. Following assemblies, the generated con-tigs and singletons were merged into one file for eachdataset (AS, CSB, CSE, CSP, CSO, CSS, VV, WS). Anysequences longer than 2500 bp were suspected to be chi-meric, so they were parsed to a separate file, translated inall 6 frames, and peptides with a minimal size of 80amino acids before a predicted stop codon were submit-ted to a BLASTX search against the nr database. The result-ing multiple peptides predicted within long contigs werecoded with "LC", as well as with "F" for the translationalframe, with the frame number (either positive for forwardor negative for reverse) and the peptide number desig-nated from among the multiple peptides (separated by aperiod from the frame number).A BLASTX analysis was next performed on each contig andPage 6 of 17(page number not for citation purposes)sequence. If the percentage was > 0.3 (i.e. four or more'N's per 1000 bp), the flanking 100 bp regions wheresingleton sequence against the nr database in order toidentify the best frame for subsequent in silico translation.BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50The frame identified via BLASTX analysis was then used togenerate the predicted ORF (i.e. amino acid translation)for a given contig or singleton. In order to further curatepredicted ORFs, each was subjected to in silico cleavage atany 'unknown' amino acid ('X') or stop codon and thencompared to a similarly generated list of peptides from thecorresponding best scoring protein sequence identified inthe BLASTX search. The 'best peptide' was then identifiedin the translation frame as the peptide with an exactmatch to the BLASTX peptide. If no such peptide could beidentified, the longest peptide generated by in silico cleav-age of the sequence at each occurrence of an unknownamino acid and/or stop codon was used. All sequenceswhich resulted in "no hit found" in the BLASTX resultswere subsequently translated in all six frames andappended to the end of the 'best peptide' file. In all caseswhere a six-frame translation was applied, the resultingpeptides (designated as 'NH' in the database) werecleaved in silico at every unknown amino acid and/or stopcodon and only those sequences 80 amino acids or longerwere kept.The resulting list of 'best peptides' for each of thesequences was then subjected to BLASTP analysis usingthe UniProtKB database in order to determine thesequence identity. The five highest BLASTP hits for eachquery sequence were aligned using an in-house Perl scriptto identify putative N- and/or C-termini. If no consensussite could determined for an N- or C-terminus via align-ment with similar sequences, then these sequences weretrimmed at the tryptic digestion site nearest to the ends ofthe predicted ORF to eliminate potentially truncated pre-dicted tryptic peptides from the database. The parametersprogrammed into the scripts included: 1) BLASTP e-val-ues, where E = H indicates a less significant hit (> 1e-05)and E = L indicates a stronger hit (E ≤ 1e-05), 2) the differ-ence in length of the Vitis query sequence versus each ofthe top five subject sequences, and 3) the length of theexact match of amino acids to the top hit. These parame-ters improved automation of accurate predictions ofmethionine (M) sites and identification of likely full-length amino acid sequences without requiring manualinspection of the BLASTP results. Sequences identifiedwith a predicted methionine at the N-terminus werecoded with '(M)'. Determinations of C termini were donebased on a small range cutoff (± 2) of amino acidsbetween the stop codon in each top hit and the predictedstop codon in each corresponding Vitis ORF; if the differ-ence was greater than two amino acids and deemedunclear, the C-terminal end of the predicted ORF wastrimmed at the nearest upstream tryptic cleavage site.The detailed process, above, was applied to each data setsue-specific sequences were analyzed for uniquenessbased on comparing each sequence to every othersequence and discarding all shorter sequences for eachexact match. Once all CS duplicate sequences wereremoved, this dataset was merged with the remaining twosets, VV and WS. This final set consisting of all of thesequences was then subjected to yet another uniquenesstest where each sequence was compared to every othersequence but this time CS sequences where intentionallynot removed, even if an exact duplicate existed in eitherthe VV or the WS set and was of greater length. Thisallowed for preferential retention of the CS sequences inorder to keep information about the tissue of origin of adetected protein. Out of a total of 113243 sequences sub-mitted to the uniqueness test, 52394 were identified withCS duplicates present due to the preferential retention ofthose sequences. From the resulting sequences, only thosethat started with a predicted methionine were then sub-mitted for SignalP analysis http://www.cbs.dtu.dk/services/SignalP/ and processing which allowed for the iden-tification and ultimate trimming of signal peptides givingrise to predicted mature protein sequences. Thosetrimmed sequences that had a predicted cleavable target-ing signal were coded '(SP)'. After removal of the pre-dicted signal peptides, a final uniqueness test wasperformed.Analysis of MS/MS dataiTRAQ MS/MS data were analyzed using ProteinPilot soft-ware v. 2.0.1 (Applied Biosystems) for both tryptic pep-tide identification and quantification. The peptides andcorresponding relative abundances were obtained in Pro-teinPilot using a confidence cutoff (called a 'Prot Score')of > 1.3 (> 95%). Database searching for each sample wasdone on predicted tryptic peptide sequence data usingeither the MSDB database (Release 20063108, Hammer-smith Campus of Imperial College London) or in-housedatabases (Vitis_spp_ORF_db_v1.0, untrimmed, or ASdatabases). Annotations and annotated protein namesindicated in ProteinPilot output files were coded to indi-cate several parameters specific to the ORF identified aswell as the EST or contig from which the ORF sequencewas predicted.iTRAQ data representing the four ripening initiationstages in each of the three exocarp samples (2004, 2005-1, 2005-2) were combined into a single tab delimited file.Likewise, iTRAQ data representing each of the three mes-ocarp samples (2004, 2005-1, 2005-2) were combinedinto a second tab delimited file. Duplicate entries amongexocarp or mesocarp files were identified using an in-house script in the R environment with 'Custom ORF ID'as the search string. Then, ratiometric data at each of thePage 7 of 17(page number not for citation purposes)individually (AS, WS, VV, CSO, CSS, CSP, CSE and CSB).In preparation for the merger of the datasets, the CS tis-three comparisons using 'green' as the reference stage (i.e.pink/green; red/green; purple/green) were averaged priorBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50to export for cluster analyses. Entries with the same namebut different template cDNA sources were not averagedsince these may represent isoforms from different sourcetissues and/or cultivars. We chose to cluster all proteinsdetected in the exocarp or mesocarp in order to capture allinformation on expression patterns detected, withoutrestricting our analyses to only those proteins that werereplicated amongst the individual exocarp or mesocarpfiles. K-means clustering into four partitions was carriedout on ratiometric data for the exocarp and mesocarp filesseparately using MultiExperiment Viewer (MeV) software(The Institute for Genome Research; http://www.tm4.org/mev.html). We used a 1.5-fold threshold for biologicalsignificance which was validated by consistencies betweentrends in protein expression presented here as increasingor decreasing with corresponding patterns of gene expres-sion identified in previous publications (see Figure 4 leg-end for citations).ResultsProtein detection and annotationIn order to strengthen the ability of MS/MS spectra anno-tation software such as ProteinPilot to accurately identifypeptide sequences in complex total protein samples, weperformed weighted clustering on a large EST collectiongenerated from Vitis spp. to create the most cultivar-spe-cific peptide map database possible for Cabernet Sauvi-gnon. In addition, we created Perl scripts in order to findthe N and C termini in translated ORFs or otherwise tocarry out trimming at known tryptic digestion sites. Thiswas done to help decrease the number of incorrectlyannotated, non-tryptic (i.e. truncated) N- and/or C-termi-nal peptides incorporated into protein abundance quanti-tations. To further increase the number of identifiablepeptides, our in-house scripts searched all of the predictedORFs designated as containing a likely start methionineusing SignalP to identify and remove putative N-terminalsignal peptides, resulting in a database containing matureprotein sequence data for the predicted full length pro-teins.To assess whether the custom tryptic peptide databaseimproved protein discovery, we examined the number ofhigh confidence proteins and peptides identified by Pro-teinPilot using this database versus a common approachto annotation, a search of the MSDB. We performed thisanalysis on two iTRAQ data sets, mesocarp 2005-1 andexocarp 2005-1, derived from four stages of Cabernet Sau-vignon berries at ripening initiation. At a confidence levelof 95% (p < 0.05), 1424 proteins were identified in themesocarp using the custom database, whereas only 1184proteins were identified in a search of the same data setusing the MSDB (Table 1). At a confidence level of 95%,1493 proteins were identified in the exocarp using the cus-tom database, whereas 1390 proteins were identified in asearch of the same data set using the MSDB (Table 1).These results indicate that in these two iTRAQ data sets,the use of the custom tryptic peptide database improvedhigh confidence protein discovery by 20.2% and 7.4%.The number of high confidence peptides detected usingthe custom database was 1.9-fold and 1.8-fold higher inmesocarp and exocarp, respectively, in comparison tosearches of the same iTRAQ data sets using the MSDB(Table 1). The greater difference in the number of highconfidence peptides versus proteins detected using thecustom database in comparison to the MSDB indicatesthat the most important impact of implementing the cus-tom database was that it afforded the identification ofmore high confidence peptides per protein than could beachieved using the MSDB.The effects of weighting and trimming on high confidencepeptide and protein detection were analyzed using the2005 exocarp iTRAQ data set. We performed a secondbuild of all Vitis spp. ESTs using PCAP. For the purposesof this comparative analysis, all ESTs in this second buildwere equally weighted for consensus sequence determina-tion using an arbitrary phred score; no higher weighting ofCabernet Sauvignon-derived ESTs was used. Thisunweighted Vitis EST database was then used to generatea second predicted tryptic peptide database, includingTable 1: Comparison of numbers of high confidence peptides and proteins detected by ProteinPilot based on the amino acid sequence database searched.Sample Database searched Proteins detected Distinct peptides detected2005 mesocarp MSDBa 1184 56792005 mesocarp Custom Vitis DB with weightingb and trimmingc 1424 107932005 exocarp MSDB 1390 69152005 exocarp Custom Vitis DB with weighting and trimming 1493 129452005 exocarp Custom Vitis DB with trimming but no weighting 1447 129562005 exocarp Custom Vitis DB with weighting but no trimming 1382 8845aRelease 20063108, Hammersmith Campus of Imperial College LondonbPage 8 of 17(page number not for citation purposes)EST clustering weighted higher scores for ESTs derived from Cabernet Sauvignon in order to avoid the potential loss of SNPs specific to this cultivar during consensus sequence determination by the PCAP softwarecPredicted tryptic peptides that were determined to likely be truncated were trimmed and removed from the custom peptide databaseBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50trimming of end-truncated peptides, as before. Thenumber of high confidence peptides and proteinsdetected were similar using the ORF databases createdfrom the weighted versus the unweighted EST databases(Table 1), suggesting that the effect of simple nucleotidepolymorphisms (SNPs) among the Vitis spp. ESTs is ulti-mately negligible at the level of peptide identification viaMS/MS in grapevine. The effectiveness of trimming pre-dicted end-truncated peptides, on the other hand, was evi-dent in the increased number of peptides identified (Table1) by ProteinPilot in searches of the trimmed (default)ORF database versus a third, untrimmed custom Vitis ORFdatabase that we created for this comparison by omittingthe trimming step shown in Figure 1; both of these data-bases were weighted for Cabernet Sauvignon ESTs in theoriginal PCAP builds. These findings indicate that auto-mated searching for the best predicted methionine startsite and removing truncated N- and/or C-terminal trypticpeptide sequences in predicted ORFs based upon EST dataincreases the ability of MS/MS spectra annotation soft-ware to identify peptide sequences with high statisticalconfidence.Technical and biological replication of expression ratiosProteins identified in iTRAQ data sets from two technicalreplicates were analyzed. Two 100 μg aliquots from thesame total protein sample, exocarp 2004, were separatelylabeled with iTRAQ reagents, subjected to nanoLC-MS/MS, and the custom predicted tryptic peptide databasewas searched with the resulting spectra using ProteinPilot.Only proteins that were detected in each of the four ripen-ing initiation stages were retained for the comparativeanalysis. In-house custom scripts written in R language[20] were employed to detect protein entries that wereduplicated in the two ProteinPilot output files represent-ing the technical replicates, using 'Custom ORF ID' as thesearch string. Out of a total of 1741 proteins detected inboth iTRAQ data sets, 507 or 29% of these proteins weredetected in both files (Additional file 1), whereas theremainder were unique to either technical replicate with547 proteins in one file and 687 proteins in the other. Wewere interested in estimating consistency in trends inexpression data along ripening initiation for each repli-cated protein entry that we detected (Additional file 1).We determined that strict estimation of replicability viaPearson's correlation coefficient or concordance correla-tion coefficient analyses of expression ratios did not pro-vide relevant quantitative information (data not shown),since even small deviations in expression ratios can resultin poor correlative data. We considered that identifyingconsistent trends in protein accumulation above an arbi-trary cutoff such as 1.5-fold was most relevant for the pur-poses of discovering protein candidates for hypothesissystem was devised, whereby expression ratios identifiedby manual inspection as differing by less than log2(0.3)for each duplicate protein entry for at least two of thethree ratios received a score of 1, expression ratios differ-ing by between log2(0.3) and log2(0.6) for at least two ofthe three ratios received a score of 2, and expression ratiosdiffering by log2(0.6) or greater for at least two of the threeratios received a score of 3. The mean replication score forthe 507 duplicated protein entries in the technical repli-cates was determined to be 1.7 ± 0.7 (Additional file 1);17% of the protein entries received a score of 3. These dataindicate that if a 1.5-fold change or greater is consideredas biologically relevant, over 80% of duplicate proteinentries but only approximately one quarter of all proteinsdetected in two technical replicates would be expected tobe identified as significantly changing along ripening ini-tiation with similar quantitative trends, based on the tech-nical replicates analyzed here.Proteins identified in iTRAQ analyses of two biologicalreplicates, exocarp 2004 and exocarp 2005-1 were nextanalyzed in an identical manner to the two technical rep-licates. Out of a total of 2187 proteins detected in the twoiTRAQ data sets, 718 or 33% of these proteins weredetected in both files (Additional file 2), whereas theremainder were unique to either biological replicate with733 proteins in one file and 736 proteins in the other. Thesame replication scoring system was implemented forproteins detected in both iTRAQ data sets (Additional file2) as was done for the technically replicated proteins(Additional file 1), above. The mean replication score forthe 718 proteins common to both biological replicateswas determined to be 1.6 ± 0.7 (Additional file 2); 12% ofprotein entries received a score of 3.To further explore the biological significance of replicabil-ity analyses, proteins common to either the technical rep-licates (Additional file 1; 507 proteins) or the biologicalreplicates (Additional file 2; 547 proteins) were comparedand it was determined that 343 protein entries were com-mon to the technical replicates and the biological repli-cates, i.e. were detected in all three data sets analyzed forreplication (Additional file 3); we considered thisapproach as valid since all data analyzed here werederived from the exocarp. The technical and biologicalreplication results indicate that the 'shotgun' LC-MS/MSapproach employed here with iTRAQ-labeled exocarptotal proteins is capable of repeatedly identifying approx-imately one third of the total number of proteins, whetherthey are technical or biological replicates.Expression profiling of proteins along grape berry ripening initiationPage 9 of 17(page number not for citation purposes)formulation related to ripening control. As an alternateapproach to correlation estimation, a replication scoringFour clusters were generated for the exocarp in whichtrends in protein accumulation were increasing stronglyBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50(91 proteins), increasing gradually (621 proteins), notchanging significantly (1156 proteins), or decreasing (570proteins), respectively, from the green through the fullyturned ripening initiation stages (Figure 2). Additional file4 lists proteins by cluster number along with correspond-ing log2-transformed ratiometric data for each proteinentry. Ratiometric data were calculated relative to thegreen stage for each of the pink/turning, red, and purple/fully turned stages. Several protein isoforms significantlyincreasing along ripening initiation were identified withannotated functions in anthocyanin flavonoid biosynthe-sis and storage (e.g. flavanone-3-hydroxylase, flavonoid-3'-hydroxlase, flavonoid-3'5'-hydroxylase and its associ-ated cytochrome b5 protein, anthocyanin synthase,anthocyanidin 3-O-glucosyl transferase, putative anthocy-anin O-methyltransferase), defense (e.g. pathogenesis-related protein 4, thaumatin (VvTL1)), and cell expansion(e.g. expansin, aquaporin). Conversely, several compo-nents of the photosynthetic machinery were identified asdecreasing greater than 1.5-fold along ripening initiation,which is consistent with a reduction in photosynthesis atthis period of berry development.We mined the exocarp data (Additional file 4) for proteinsthat were increasing in abundance relative to the greenstage and annotated as enzyme or transporter compo-nents of pathways leading to hypothesized regulators ofripening initiation in grapes, ABA, glucose, and BR. Aputative LytB (IspH) protein (Q9FEP0; Adonis aestivalis cv.palaestina) increased 1.6-fold in abundance and isresponsible for the last step of the plastidic pathway toisopentenyl diphosphate (IDP), leading in part to the pro-K-means cluster analysis of expression data for exocarp proteinsFigure 2K-means cluster analysis of expression data for exocarp proteins. Four partitions were used to classify proteins that were increasing strongly (top left panel), increasing gradually (top right panel), not changing significantly (bottom left panel), or decreasing (bottom right panel) along ripening initiation. Ripening initiation stages corresponding to the expression data are depicted by the photographs on the two lower x-axes. X-axes on the lower two panels also correspond identically to the top two panels. Y-axes in the left two panels correspond identically to the right two panels. Numbers of proteins are shown in the Page 10 of 17(page number not for citation purposes)top left corner of each panel.BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50duction of the plant hormones, ABA and gibberellic acid.Other proteins of this MEP pathway were also detectedbut only as slightly increasing in abundance along ripen-ing initiation. An isopentenyl-diphosphate δ-isomerase Iprotein was detected as increasing 2-fold along ripeninginitiation (Q39472; Clarkia breweri; Q6EJD1; Puerariamontana); this enzyme controls a major early step in iso-prenoid biosynthesis and is likely localized to the chloro-plast. We identified one component specific to the ABAbiosynthetic pathway, a protein similar to violaxanthinde-epoxidase from tea (Q8S4C2; Camellia sinensis), whichwas stably expressed along ripening initiation. CytosolicIDP, potentially also formed in the plastid and exportedto the cytoplasm, is incorporated in the biosynthesis ofBRs. A putative grapevine ortholog to a BR biosyntheticprotein from pea, PsLKB, (Q9ATR0; Pisum sativum), wasidentified as increasing 1.6-fold, peaking at the third (red)stage tested. A grapevine hexose transporter (VvHT6;Q4VKB3) was identified as increasing 1.5-fold, which isconsistent with hexose accumulation during ripening ini-tiation. Few proteins annotated with signal transductionfunctions were detected in any of the four clusters thatcould hypothetically be involved in ABA, glucose, and/orBR signaling. Abscisic stress response protein (VvASR;Q94G23) was previously implicated in cross-talk betweenABA and glucose signaling [21]; expression data were var-iable, with some isoforms increasing up to 4-fold alongripening initiation and others showing stable expression.A putative pirin protein of unknown function increased4.5-fold along ripening initiation. A putative ortholog tothe Malus spp. TTG1 WD40 repeat protein (Q9M610),which has previously been implicated in Arabidopsis inthe regulation of anthocyanin biosynthesis [22], wasdetected as increasing, indicating that VvTTG1 could playa similar role in the grape exocarp; a caveat is that the con-fidence level for VvTTG1 unique peptide detection wasless than 95%. Curiously, we did not identify peptidesrepresenting VvMYBA1, which is highly expressed at thetranscriptional level during ripening initiation andencodes a transcription factor previously demonstrated topositively regulate anthocyanin biosynthesis genes ingrape [23].Four clusters were generated for the mesocarp in whichtrends in protein accumulation were increasing strongly(58 proteins), increasing gradually (502 proteins), notchanging significantly (1148 proteins), or decreasing (491proteins), respectively, from the green through the fullyturned ripening initiation stages (Figure 3). Additional file5 lists proteins by cluster number along with correspond-ing log2-transformed ratiometric data for each proteinentry, relative to the green stage. Similar to the exocarpdata, we identified > 2-fold accumulation of several pro-polygalacturonase, pectate lyase) and defense (e.g. patho-genesis-related proteins) concomitant with a significantreduction in proteins inhibitory to fruit ripening (e.g.polygalacturonase-inhibiting protein).We mined the mesocarp data (Additional file 5) for pro-teins annotated as enzyme or transporter components ofpathways leading to ABA, glucose, and brassinosteroidaccumulation. We detected VvNCED2 (9-cis-epoxycarote-noid dioxygenase 2; Q5SGD0) increasing over 2-foldalong ripening initiation, which represents expression ofa key committed step specifically in ABA biosynthesis inthe plastid [24]. A protein annotated as ABA glucosyl-transferase increased 1.25-fold along ripening initiationin the mesocarp. As in the exocarp, an isopentenyl-diphosphate δ-isomerase I was detected as increasing 2-fold along ripening initiation (Q6EJD1; Pueraria lobata),plus, a putative cytosolic isoform was detected as stablyexpressed (Nicotiana tabacum; Q9AVG7) [25]. Farnesyldiphosphate synthase, a cytoplasmic enzyme leading tothe BR biosynthetic pathway via squalene biosynthesis,was detected in the mesocarp as increasing 1.4-fold alongripening initiation. A putative grapevine ortholog to theBR biosynthetic protein in pea, PsLKB, (Q9ATR0; Pisumsativum), and in cotton, GhDWARF1 (Q2QCX8; Gossyp-ium hirsutum), was detected as increasing 1.4-fold. Thehexose transporter, VvHT6, was detected as increasing 2-fold, similar to its expression pattern that was detected inthe exocarp. We mined the mesocarp data further to iden-tify proteins annotated with signal transduction func-tions. No plasma membrane receptor candidates wereidentified as increasing but several were identified as notchanging or decreasing along ripening initiation, mostnotably a receptor-like kinase similar to PERK1-like pro-tein from rice (Q6ZIG4) that decreased 1.5-fold at theonset of pigment accumulation to 2.5-fold at the purplestage; this PERK1-like protein was also identified as signif-icantly decreasing in the exocarp (Additional file 4). Thesame pirin protein identified in the exocarp was similarlydetected as increasing in the mesocarp along ripening ini-tiation. Inconsistent iTRAQ data were obtained for VvASRin the mesocarp with isoforms showing an increase, stableexpression, or strong down-regulation.A custom script written in the R environment was used tosearch for proteins common to the combined exocarp andmesocarp files (Additional files 4 and 5) using 'CustomORF ID' as the search string. Protein isoforms detected incommon between exocarp and mesocarp (1147 in total)are shown in Additional file 6. Protein isoforms detectedonly in exocarp (1149 in total) are shown in Additionalfile 7. Protein isoforms detected only in mesocarp (905 intotal) are shown in Additional file 8. Isoforms of anthocy-Page 11 of 17(page number not for citation purposes)tein isoforms annotated with functions in cell enlarge-ment (e.g. expansins, aquaporins), fruit softening (e.g.anin flavonoid biosynthetic proteins were only detectedin the exocarp, which is consistent with induction timingBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50and tissue localization of these pigments during ripeninginitiation and demonstrates the efficacy of our separationof exocarp and mesocarp for proteomic analyses. Addi-tional file 9 shows the source tissues and number ofunique high confidence peptides per protein for thoseproteins indicated here in the Results section and in Figure4.DiscussionThe iTRAQ data obtained with exocarp and mesocarptotal proteins confirmed previous ratiometric transcriptabundance data for key components of ABA and BR bio-synthesis, as well as the influx and accumulation of sugarsduring ripening initiation (Figure 4). We confirmed forincluding a putative anthocyanin O-methyltransferase.The detection of VvNCED2 accumulation confirmed pre-vious real-time RT-PCR data [26] and further supports arole specifically for this NCED family member in ABA bio-synthesis in the mesocarp during berry ripening initiation.Based on the sensitivity limitations to the shotgun pro-teomic technique employed here, we cannot concludethat VvNCED2 accumulation, activity, and, consequently,ABA biosynthesis are localized to the mesocarp;VvNCED2 may be expressed in the exocarp but we did notdetect it. The previous detection of ζ-carotene desaturasetranscripts in the exocarp [27] argues against a model inwhich ABA is synthesized in the mesocarp and trans-ported to the exocarp whereupon it activates, in part,K-means cluster analysis of expression data for mesocarp proteinsFigure 3K-means cluster analysis of expression data for mesocarp proteins. Four partitions were used to classify proteins that were increasing strongly (top left panel), increasing gradually (top right panel), not changing significantly (bottom left panel), or decreasing (bottom right panel) along ripening initiation. Ripening initiation stages corresponding to the expression data are depicted by the photographs on the two lower x-axes. X-axes on the lower two panels also correspond identically to the top two panels. Y-axes in the left two panels correspond identically to the right two panels. Numbers of proteins in each cluster are shown in the top left corner of each panel.Page 12 of 17(page number not for citation purposes)the first time the strong accumulation of several compo-nents of anthocyanin biosynthesis at the protein level,anthocyanin biosynthesis. If the primary site of ABA pro-duction in the developing fruit is the seed, however, andBMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50ABA transported from the seed to the surrounding peri-carp is an early signaling event initiating ripening, as wehave previously hypothesized [26], it is reasonable thatthe first tissue in the pericarp in which ABA is synthesizedis the mesocarp, given its closer proximity to the seed thanceeds from the seed through the mesocarp and, finally, inthe exocarp. The presence and possible moderate accumu-lation of a putative ABA glucosyltransferase indicates thatan attenuating mechanism for the ABA signal may operateduring ripening initiation to control ABA homeostasisModel showing transcript (squares) and protein (circles) expression trends annotated with functions in ABA, BR, and anthocy-anin biosynthesis in the grape berry pericarp at ripening initiationFigure 4Model showing transcript (squares) and protein (circles) expression trends annotated with functions in ABA, BR, and anthocyanin biosynthesis in the grape berry pericarp at ripening initiation. Transcript accumulation data are based on previous studies [26,27,35-37], whereas all protein accumulation data presented here are new findings for grape berries. Green indicates an increase in transcript or protein abundance during ripening initiation. Yellow indicates no significant change in protein accumulation. ABA-O = abscisic aldehyde oxidase; ANS = anthocyanin synthase; BR6OX1 = brassinosteroid-6-oxidase; BRI1 = receptor-like kinase, brassinosteroid insensitive 1; β-hyd = β-carotene hydroxylase; CYD = cyanidin; CYTb5 = cytochrome b5; DFR = dihydroflavonol reductase; DHK = dihydrokaempferol; DHM = dihydromyricetin; DHQ = dihydro-quercitin; DMADP = dimethylallyl diphosphate; DPD = delphinidin; DWF1 = dwarf 1; F-OMT = putative anthocyanin flavonoid O-methyltransferase; FDPS = farnesyl diphosphate synthase; F3H = flavonoid-3-hydroxylase; F3'H = flavonoid-3'-hydroxylase; F3'5'H = flavonoid-3'5'-hydroxylase; GCR2 = G protein-coupled receptor 2; GGDP = geranylgeranyl diphosphate; GGDPS = geranylgeranyl diphosphate synthase; IDP = isopentenyl diphosphate; IPI = isopentenyl diphosphate isomerase; LCY-β = lyco-pene-β-cyclase; MVD = malvidin; NCED = 9-cis-epoxycarotenoid dioxygenase; PDS = phytoene desaturase; PND = peonidin; PSY = phytoene synthase; PTD = petunidin; SDR = short-chain alcohol dehydrogenase/reductase; VDE = violaxanthin de-epox-idase; ZDS = ζ-carotene desaturase; ZEP = zeaxanthin epoxidase.Page 13 of 17(page number not for citation purposes)the exocarp. It remains to be determined whether a gradi-ent in ABA biosynthesis during early ripening stages pro-[28]. The accumulation of the pirin protein is also intrigu-ing, given this protein's previously demonstrated interac-BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50tion with G-protein alpha in Arabidopsis seeds and apotential role in modifying ABA action via negative feed-back control [29].We used a relatively new quantitative MS/MS-basedapproach, iTRAQ, to advance our understanding of differ-ential protein expression underlying non-climacteric rip-ening initiation in grape berries. The iTRAQ approachoffered several advantages over 2DGE methods for pro-tein discovery, including increased detection sensitivitybased on our findings reported here in comparison to pre-vious reports on grape berry proteomics. By using strongcation exchange and reverse phase column microcapillarychromatography coupled with nanospray MS/MS detec-tion with total protein extracts from grape berries, we wereable to resolve three-fold or more proteins per samplethan would be expected using 2DGE [11,12,14]. One cur-rent limitation to an MS-based proteomic approach withgrapevine is that there are no finished genome sequencedata for grapevine, although two projects [30,31] areundertaking assembly and annotation from which a highquality ORFeome database can eventually be derived.Although there are over 300,000 Vitis spp. ESTs depositedin Genbank, V. vinifera is a highly heterozygous species[31]; therefore, we considered that it might be importantto weight builds of these ESTs via manipulation of phredscores so as to favor sequence data corresponding to ourcultivar of interest, Cabernet Sauvignon, when SNPs wereencountered by PCAP, where applicable for a given contigassembly. Although we determined that weighting ESTs tothe genotype of interest for EST assembly provided noclear advantages in this study, we conclude that producinga tryptic peptide database targeted to Vitis sequences andincluding removal of predicted truncated peptidesimproved protein detection and annotation. Further-more, our findings indicate that a tryptic peptide databasebased on finished Pinot Noir whole genome sequencedata will be valid to implement for proteome studies withany V.vinifera cultivar or Vitis species, with the exceptionof cases where deletions have occurred in the Pinot Noirhomozygous line [30].We chose to not include genome sequence data availablefor V. vinifera cv. Pinot Noir [30,31] due to significantgaps in current assemblies and the potential for inaccurateautomated gene predictions. Until grapevine genomesequence assembly and annotation are finished, we pro-pose that the predicted ORF database presented here willbe of value to the grapevine community in two significantways. While gaps exist in the genome sequence assem-blies, the protein database presented here may provideinformation for 'missing' proteins either not yet predictedfrom the Pinot Noir genome sequence data and/or otherther, protein predictions based upon expressed sequences(ESTs) represent 'real' proteins, so our database couldpotentially be used to validate ORF predictions based onwhole genome sequence data alone.While iTRAQ labeling coupled with nanoLC-MS/MSproved overall to be an advance over 2DGE in sensitivityand quantitation, consistent detection of predicted pro-tein sequences between technical or biological replicatesfrom grape exocarp was limited. Consistency in trends inratiometric data along ripening initiation for those pro-teins that were detected in replicate exocarp samples wasfurther limited. Inconsistent ratiometric data for someproteins detected in both biological replicates may repre-sent differences in expression due, for example, to varia-bility in seasonal growing conditions and not technicalvariability. Nonetheless, limited replicable detection ofproteins among biological samples for technical reasonsis commonly encountered with liquid chromatographyand MS-based proteomics [32] and could arise from vari-ation in preparation of total proteins and/or iTRAQ-labeled peptides from sample to sample. Underlying theapparent variation arising from sample preparations areconstraints on detection imposed by the mass spectrome-ter, both with respect to the dynamic range of the instru-ment and the nature of selections of peptides in the firstMS by the MS software for export to the collision cell priorto amino acid detection via the second MS. Furthermore,our finding that two thirds of the replicated exocarp pro-teins were detected both in technical and biological repli-cates may have reflected the higher probability ofdetecting these mainly abundant, housekeeping-type pro-teins in any given total protein sample using a shotgunapproach.Over-sampling of abundant peptides in digested total pro-tein samples is a limitation of the shotgun approach toquantitative proteomics [33] and likely precluded ourability to discover more proteins annotated with signaltransduction functions that could regulate ABA, BR, andhexose responses during ripening initiation. In order toincrease detection sensitivity of low abundance regulatoryproteins using shotgun proteomics techniques, it willlikely be helpful to isolate membrane and nuclear pro-teins separately from cytoplasmic proteins prior to diges-tions. Affinity chromatography of berry protein extractsusing antibodies directed against abundant proteins suchas thaumatin detected in both exocarp and mesocarp mayalso improve detection sensitivity for low abundance pro-teins by selectively removing these proteins prior toiTRAQ labeling steps. Similarly, advances in detectionsensitivity in devices such as the Fourier transform ioncyclotron resonance MS (FTICR-MS) [34] should allow usPage 14 of 17(page number not for citation purposes)Vitis spp. not represented in the Pinot Noir genomesequence data, e.g. due to chromosomal deletions. Fur-to delve deeper into the grape berry proteome in order tobetter understand the molecular control of non-climac-BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50teric ripening in this species. Nonetheless, the searchablequantitative data presented here on over 3000 proteinsdetected in two separate berry tissues along four ripeninginitiation stages can serve as a framework public resourcecontributing towards our understanding of the dynamicgrapevine proteome.ConclusionWe determined that a predicted peptide database can bederived from grapevine EST data using advanced cluster-ing and trimming approaches and successfully imple-mented for quantitative proteome profiling. Wedemonstrated in grapevine that by implementing a pre-dicted peptide database targeted to an organism of inter-est, a significant increase in the number of highconfidence peptides identified and annotated from MS/MS data was gained in comparison to more commonlyused MSDB searches. Furthermore, by using a shotgunquantitative proteomics approach in combination with atargeted predicted peptide database, we showed thatgreater numbers of high confidence peptides weredetected than would be expected using 2D gel electro-phoresis techniques. We verified for the first time at theprotein level, quantitative expression patterns for compo-nents of isoprenoid and flavonoid metabolism importantfor grape compositional chemistry, as well as identifiednew signal transduction candidate proteins associatedwith non-climacteric ripening initiation in grape berries.AbbreviationsABA: abscisic acid; BLAST: basic local alignment andsearch tool; BR: brassinosteroids; EST: expressed sequencetag; iTRAQ: isobaric tagging for relative and absolutequantitation; ORF: open reading frame; PCAP: parallelcontig assembly programAuthors' contributionsJL designed the experiments, carried out protein extrac-tions and data analyses, assisted in designing the scriptsfor the predicted peptide database, and drafted the manu-script. ML designed the scripts and carried out program-ming to produce the predicted peptide databasespresented here. DS carried out mass spectrometry anddatabase searches using the iTRAQ MS/MS data. STL per-formed primary data analyses, assisted in drafting themanuscript, and was responsible for overseeing theproject.Additional materialAdditional file 1Duplicated proteins and associated replication scores for exocarp 2004 technically replicated iTRAQ experiments. Ratiometric data are given in log2 scale.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S1.xls]Additional file 2Duplicated proteins and associated replication scores for exocarp 2004 and exocarp 2005-1 biologically replicated iTRAQ experiments. Rati-ometric data are given in log2 scale.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S2.xls]Additional file 3Proteins identified in all of the three iTRAQ data sets analyzed (exo-carp 2004 (two technical replicates) and exocarp 2005-1) and tested for replication in expression trends along ripening initiation.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S3.xls]Additional file 4Exocarp proteins listed for each K-means cluster along with corre-sponding log2-transformed ratiometric data for each protein entry. Exocarp proteins from the 2004, 2005-1, and 2005-2 datasets were com-bined into a single file. Duplicate entries were identified using an in-house script in the R environment with 'Custom ORF ID' as the search string. Then, ratiometric data at each of the three comparisons using 'green' as the reference stage were averaged prior to export for K-means cluster anal-yses.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S4.xls]Additional file 5Mesocarp proteins listed for each K-means cluster along with corre-sponding log2-transformed ratiometric data for each protein entry. Mesocarp proteins from the 2004, 2005-1, and 2005-2 datasets were combined into a single file. Duplicate entries were identified using an in-house script in the R environment with 'Custom ORF ID' as the search string. Then, ratiometric data at each of the three comparisons using 'green' as the reference stage were averaged prior to export for K-means cluster analyses.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S5.xls]Additional file 6Protein isoforms detected in common between the exocarp and meso-carp (Additional files 4 and 5, respectively). A custom script written in the R environment was used to search for proteins common to the com-bined exocarp and mesocarp files using 'Custom ORF ID' as the search string.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-Page 15 of 17(page number not for citation purposes)2164-10-50-S6.xls]BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50AcknowledgementsThe authors wish to thank Monica Elliot for assistance with protein prepa-rations and iTRAQ labeling. We thank Dr. Fred Y. Peng for assistance with iTRAQ data processing and cluster analyses. The authors gratefully acknowledge funding for this research from Genome Canada and project management support from Genome British Columbia as part of the Genome Canada-Genoma España collaborative research and development initiative. The authors also wish to thank the Province of British Columbia and the BC wine industry for their continued support of research activities in the University of British Columbia Wine Research Centre.Resources and raw data developed through the research presented here can be downloaded freely at: http://www.landfood.ubc.ca/people/ste ven.lund/ These are: 1) scripts used for the customized grapevine protein database creation, 2) the protein database, and 3) the ProteinPilot output files for exocarp 2004, exocarp 2005-1, exocarp 2005-2, mesocarp 2004, mesocarp 2005-1, and mesocarp 2005-2. ProteinPilot software can be cies different to V. vinifera; the corresponding author may be contacted for additional information.References1. Perkins DN, Pappin DJC, Creasy DM, Cottrell JS: Probability-basedprotein identification by searching sequence databases usingmass spectrometry data.  Electrophoresis 1999, 20:3551-3567.2. Shilov IV, Seymour SL, Patel AA, Loboda A, Tang WH, Keating SP,Hunter CL, Nuwaysir LM, Schaeffer DA: The Paragon algorithm:A next generation search engine that uses sequence temper-ature values and feature probabilities to identify peptidesfrom tandem mass spectra.  Mol Cell Prot 2007, 6:1638-1655.3. Choudhary JS, Blackstock WP, Creasy DM, Cottrell JS: Matchingpeptide mass spectra to EST and genomic DNA databases.Trends Biotech 2001, 19:S17-S22.4. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH: Quantitative anal-ysis of complex protein mixtures using isotope-coded affinitytags.  Nature Biotech 1999, 17:994-998.5. Islam N, Tsujimoto H, Hirano H: Wheat proteomics: Relation-ship between fine chromosome deletion and protein expres-sion.  Proteomics 2003, 3:307-316.6. Pierce A, Unwin RD, Evans CA, Griffiths S, Carney L: Eight-channeliTRAQ enables comparison of the activity of 6 leu-kaemogenic tyrosine kinases.  Mol Cell Prot 2007.7. Dunkley TPJ, Svenja H, Shadforth IP, Runions J, Weimar T, Hanton SL,Griffin JL, Bessant C, Brandizzi F, Hawes C, Watson RB, Dupree P, Lil-ley KS: Mapping the Arabidopsis organelle proteome.  Proc NatAcad Sci USA 2006, 103(17):6518-6523.8. Jones AME, Bennett MH, Mansfield JW, Grant M: Analysis of thedefense phosphoproteome of Arabidopsis thaliana using dif-ferential mass tagging.  Proteomics 2006, 6:4155-4165.9. Patterson J, Ford K, Cassin A, Natera S, Bacic A: Increased abun-dance of proteins involved in phytosiderophore productionin boron-tolerant barley.  Plant Physiol 2007, 144:1612-1631.10. Lund ST, Bohlmann J: The molecular basis for wine grape qual-ity – A volatile subject.  Science 2006, 311:804-805.11. Deytieux C, Geny L, Lapaillerie D, Claverol S, Bonneu M, Donèche B:Proteome analysis of grape skins during ripening.  J Exp Bot2007, 58:1851-1862.12. Giribaldi M, Perugini I, Sauvage F-X, Schubert A: Analysis of proteinchanges during grape berry ripening by 2-DE and MALDI-TOF.  Proteomics 2007, 7:3154-3170.13. Castro AJ, Carapito C, Zorn N, Magné C, Leize E, Van Dorsselaer A,Clément C: Proteomic analysis of grapevine (Vitis vinifera L.)tissues subjected to herbicide stress.  J Exp Bot 2005,56:2783-2795.14. Sarry JE, Sommerer N, Sauvage FX, Bergoin A, Rossignol M, AlbagnacG, Romieu C: Grape berry biochemistry revisited upon pro-teomic analysis of the mesocarp.  Proteomics 2004, 4:201-215.15. Wang W, Scali M, Vignani R, Spadafora A, Sensi E, Mazzuca S, CrestiM: Protein extraction for two-dimensional electrophoresisfrom olive leaf, a plant tissue containing high levels of inter-fering compounds.  Electrophoresis 2003, 24:2369-2375.16. Jain AK, Basha SM: A capillary electrophoretic method for iso-lation and characterization of grape xylem proteins.  Afr J Bio-tech 2003, 2:66-70.17. Peng FY, Reid KE, Liao N, Schlosser J, Lijavetzky D, Holt R, MartínezZapater JM, Jones S, Marra M, Bohlmann J, Lund ST: Generation ofESTs in it Vitis vinifera wine grape (Cabernet Sauvignon) andtable grape (Muscat Hamburg) and discovery of new candi-date genes with potential roles in berry development.  Gene2007, 402:40-50.18. Ewing B, Hillier L, Wendl MC, Green P: Base-calling of automatedsequencer traces using phred. I. Accuracy assessment.Genome Res 1998, 8:175-185.19. Huang X, Wang J, Aluru S, Yang S-P, Hillier L: PCAP: A whole-genome assembly program.  Genome Res 2003, 13:2164-2170.20. Development Core Team: R: A Language and Environment forStatistical Computing. Vienna, Austria.  2006 [http://www.R-project.org].21. Çakir B, Agasse A, Gaillard C, Saumonneau A, Delrot S, AtanassovaR: A grape ASR protein involved in sugar and abscisic acidAdditional file 7Protein isoforms detected only in exocarp samples. Proteins excluded from Additional file 6 that were detected in the combined exocarp file (Additional file 4) are shown.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S7.xls]Additional file 8Protein isoforms detected only in mesocarp samples. Proteins excluded from Additional file 6 that were detected in the combined mesocarp file (Additional file 5) are shown.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S8.xls]Additional file 9Protein source tissues and number of unique high confidence peptides per protein. Proteins presented in the Results and Discussion sections as well as in Figure 4 are shown. The source tissue for each protein hit shown in Column A corresponds to a ProteinPilot output file, which can be down-loaded at http://www.landfood.ubc.ca/people/steven.lund The number of unique > 95% confidence peptides determined by the ProteinPilot soft-ware for each protein is shown in Column E. Ratiometric data shown in Columns G, H, and I are relative to the green stage (Column F) and are not log2-transformed.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S9.xls]Additional file 10Reference key for coding terms used in Column A (Cluster ORF ID) and Column B (Protein Annotation) in Additional files 1 through 8. The key is provided as a printable rapid reference for data mining in Addi-tional files 1 through 8, as well as in the protein database available online, as shown in the Acknowledgments.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2164-10-50-S10.doc]Page 16 of 17(page number not for citation purposes)obtained by contacting Applied Biosystems. The scripts for protein data-base creation will need to be modified prior to implementation with a spe-signaling.  Plant Cell 2003, 15:2165-2180.22. Walker AR, Davison PA, Bolognesi-Winfield AC, James CM, Srini-vasan N, Blundell TL, Esch JJ, Marks MD, Gray JC: The TRANSPAR-Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Genomics 2009, 10:50 http://www.biomedcentral.com/1471-2164/10/50ENT TESTA GLABRA1 locus, which regulates trichomedifferentiation and anthocyanin biosynthesis in Arabidopsis,encodes a WD40 repeat protein.  Plant Cell 1999, 11:1337-1350.23. Kobayashi S, Goto-Yamamoto N, Hirochika H: Retrotransposon-induced mutations in grape skin color.  Science 2004, 304:982.24. Nambara E, Marion-Poll A: Abscisic acid biosynthesis and catab-olism.  Annu Rev Plant Biol 2005, 56:165-185.25. Phillips MA, D'Auria JC, Gershenzon J, Pichersky E: The Arabidopsisthaliana Type I isopentenyl diphosphate isomerases are tar-geted to multiple subcellular compartments and have over-lapping functions in isoprenoid biosynthesis.  Plant Cell 2008,20:677-696.26. Lund ST, Peng FY, Nayar T, Reid KE, Schlosser J: Gene expressionanalyses in individual grape (Vitis vinifera L.) berries duringripening initiation reveal that pigmentation intensity is avalid indicator of developmental staging within the cluster.Plant Mol Biol 2008, 68(3):301-315.27. Grimplet J, Deluc LG, Tillett RL, Wheatley MD, Schlauch KA, CramerGR, Cushman JC: Tissue-specific mRNA expression profiling ingrape berry tissues.  BMC Genomics 2007, 8:187.28. Xu Z-J, Nakajima M, Suzuki Y, Yamaguchi I: Cloning and character-ization of the abscisic acid-specific glucosyltransferase genefrom adzuki bean seedlings.  Plant Physiol 2002, 129:1285-1295.29. Lapik YR, Kaufman LS: The Arabidopsis cupin domain proteinAtPirin1 interacts with the G Protein α-subunit GPA1 andregulates seed germination and early seedling development.Plant Cell 2003, 15:1578-1590.30. The French-Italian Public Consortium for Grapevine Genome Char-acterization: The grapevine genome sequence suggests ances-tral hexaploidization in major angiosperm phyla.  Nature 2007,449:463-468.31. Velasco R, Zharkikh A, Troggio M, Cartwright DA, Cestaro A, PrussD, Pindo M, FitzGerald LM, Vezzulli S, Reid J, Malacarne G, Iliev D,Coppola G, Wardell B, Micheletti D, Macalma T, Facci M, Mitchell JT,Perazzolli M, Eldredge G, Gatto P, Oyzerski R, Moretto M, Gutin N,Stefanini M, Chen Y, Segala C, Davenport C, Demattè L, Mraz A, Bat-tilana J, Stormo K, Costa F, Tao Q, Si-Ammour A, Harkins T, LackeyA, Perbost C, Taillon B, Stella A, Solovyev V, Fawcett JA, Sterck L,Vandepoele K, Grando SM, Toppo S, Moser C, Lanchbury J, BogdenR, Skolnick M, Sgaramella V, Bhatnagar SK, Paolo F, Gutin A, Peer YVan de, Salamini F, Viola R: A high quality draft consensussequence of the genome of a heterozygous grapevine vari-ety.  PloS One 2007, 2:e1326.32. Thelen JJ, Peck SC: Quantitative proteomics in plants: Choicesin abundance.  Plant Cell 2007, 19:3339-3346.33. Reinhardt TA, Lippolis JD: Developmental changes in the milkfat globule membrane proteome during the transition fromcolostrum to milk.  J Dairy Sci 2008, 91:2307-2318.34. Zimmer JSD, Monroe ME, Qian WJ, Smith RD: Advances in pro-teomics data analysis and display using an accurate mass andtime tag approach.  Mass Spec 2006, 25:450-482.35. Symons GM, Davies C, Shavrukov Y, Dry IB, Reid JB, Thomas MR:Grapes on steroids. Brassinosteroids are involved in grapeberry ripening.  Plant Physiol 2006, 140:150-158.36. Deluc LG, Grimplet J, Wheatley MD, Tillett RL, Quilici DR, Deluc LG,Grimplet J, Wheatley MD, Tillett RL, Quilici DR: Transcriptomicand metabolite analyses of Cabernet Sauvignon grape berrydevelopment.  BMC Genomics 2007, 8:429.37. Pilati S, Perazzolli M, Malossini A, Cestaro A, Demattè L, Fontana P,Dal Ri A, Viola R, Velasco R, Moser C: Genome-wide transcrip-tional analysis of grapevine berry ripening reveals a set ofgenes similarly modulated during three seasons and theoccurrence of an oxidative burst at veraison.  BMC Genomics2007, 8:428.yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 17 of 17(page number not for citation purposes)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items