UBC Faculty Research and Publications

Combined physiological, transcriptome, and cis-regulatory element analyses indicate that key aspects… Wong, D. C J; Lopez Gutierrez, R.; Dimopoulos, N.; Gambetta, G. A; Castellarin, S. D May 31, 2016

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

Item Metadata


52383-12864_2016_Article_2660.pdf [ 2.19MB ]
JSON: 52383-1.0361793.json
JSON-LD: 52383-1.0361793-ld.json
RDF/XML (Pretty): 52383-1.0361793-rdf.xml
RDF/JSON: 52383-1.0361793-rdf.json
Turtle: 52383-1.0361793-turtle.txt
N-Triples: 52383-1.0361793-rdf-ntriples.txt
Original Record: 52383-1.0361793-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessCombined physiological, transcriptome,and cis-regulatory element analysesindicate that key aspects of ripening,metabolism, and transcriptional program ingrapes (Vitis vinifera L.) are differentiallymodulated accordingly to fruit sizeD. C. J. Wong1, R. Lopez Gutierrez1, N. Dimopoulos1, G. A. Gambetta2 and S. D. Castellarin1*AbstractBackground: In wine grape production, management practices have been adopted to optimize grape and winequality attributes by producing, or screening for, berries of smaller size. Fruit size and composition are influenced bynumerous factors that include both internal (e.g. berry hormone metabolism) and external (e.g. environment andcultural practices) factors. Combined physiological, biochemical, and transcriptome analyses were performed toimprove our current understanding of metabolic and transcriptional pathways related to berry ripening andcomposition in berries of different sizes.Results: The comparison of berry physiology between small and large berries throughout development (from31 to 121 days after anthesis, DAA) revealed significant differences in firmness, the rate of softening, and sugaraccumulation at specific developmental stages. Small berries had significantly higher skin to berry weight ratio,lower number of seeds per berry, and higher anthocyanin concentration compared to large berries. RNA-sequencinganalyses of berry skins at 47, 74, 103, and 121 DAA revealed a total of 3482 differentially expressed genes betweensmall and large berries. Abscisic acid, auxin, and ethylene hormone pathway genes were differentially modulatedbetween berry sizes. Fatty acid degradation and stilbenoid pathway genes were upregulated at 47 DAA while cell walldegrading and modification genes were downregulated at 74 DAA in small compared to large berries. In the lateripening stage, concerted upregulation of the general phenylpropanoid and stilbenoid pathway genes anddownregulation of flavonoid pathway genes were observed in skins of small compared to large berries. Cis-regulatoryelement analysis of differentially expressed hormone, fruit texture, flavor, and aroma genes revealed an enrichment ofspecific regulatory motifs related to bZIP, bHLH, AP2/ERF, NAC, MYB, and MADS-box transcription factors.Conclusions: The study demonstrates that physiological and compositional differences between berries of differentsizes parallel transcriptome changes that involve fruit texture, flavor, and aroma pathways. These results suggest that, inaddition to direct effects brought about by differences in size, key aspects involved in the regulation of ripening likelycontribute to different quality profiles between small and large berries.Keywords: Aroma, Cell wall, Flavonoid, Grapevine, Hormone, Promoter, Quality, RNA-seq, Secondary metabolism,Transcriptomics* Correspondence: simone.castellarin@ubc.ca1Wine Research Centre, University of British Columbia, Vancouver, BC, CanadaFull list of author information is available at the end of the article© 2016 Wong et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Wong et al. BMC Genomics  (2016) 17:416 DOI 10.1186/s12864-016-2660-zBackgroundGrapes (Vitis vinifera L.) are a highly valued horticul-tural crop with production covering approximately 7million ha in 90 countries. The grape berry is extremelyrich in secondary metabolites ranging from anthocya-nins, carotenoids, norisoprenoids, tannins, terpenes, andother volatile organic compounds. These metabolites arevery important to wine production as they affect winequality by determining its color, aroma, and flavor [1].Wine-grape growers adopt various vineyard manage-ment practices in order to optimize grape and winequality attributes. Some of these practices involve produ-cing (e.g. via applied water deficits) and/or selecting (e.g.via postharvest sorting) berries according to size withthe belief that better wines are produced from small ber-ries due to a higher skin to berry weight ratio. The rea-soning being that this higher skin to berry weight ratioresults in higher concentrations of key secondary metab-olites accumulated in the skin [2, 3]. These practices arebecoming increasingly popular and machines are evensold to automatically sort berries based on size.The grape berry is a non-climacteric fruit with adouble-sigmoidal growth curve that can be separatedinto three major stages [4]. The first stage of develop-ment sees a rapid increase in berry size due to high ratesof cell division and expansion in the berry pericarp. Co-inciding with the rapid growth is the biosynthesis ofphenolic compounds, such as tannins and hydroxycinna-mates, and organic acids, such as tartaric and malic acid.In the second stage, the berry experiences a lag phase,where pericarp growth is arrested and the embryo com-pletes its development. At the end of the second stage,the berry undergoes veraison (the onset of ripening) andenters the third and final stage. During this phase, theberry experiences a second period of rapid cell expan-sion as the pericarp grows to its final size. Many changesin berry metabolism occur: accumulation of sugars,decrease in organic acid concentration, and productionof various secondary metabolites. Thus, berry size andcomposition will differ depending on the stage ofdevelopment.Hormones play central roles in berry ripening and asenvironmental mediators (reviewed in [5]). During thefirst phase of development, auxins are present at highlevels, and then decrease to very low levels as theberry undergoes veraison [6, 7]. Just prior to veraison,a small transient peak in ethylene is observed [8, 9] aswell as sharp increases in abscisic acid (ABA) [10–13].These observations indicate that ABA and ethyleneare strong candidates as promoters of ripening. Treat-ing berries with these hormones can affect the timingof veraison as well as other parameters, including fruitsize and composition (reviewed in [5]). For example,treating berries with auxin prior to veraison temporarilydelays berry growth, and sugar and anthocyanin accu-mulation [6].Many factors influencing berry size are intrinsic, beingrelated to the individual berry itself, such as seed num-ber [14] and seed weight [15]. Recent evidence hasshown that Pinot Noir and Cabernet Sauvignon seededberries showed greater cell division and expansion thanunseeded ones, resulting in larger berries [16]. Gouthuand Deluc [17] found a relationship between seed rela-tive abundance and hormone content, where there weredistinct differences in auxin and ABA levels betweenberries with different seed to berry weight ratios prior toveraison.External factors, such as the environment and culturalpractices, can also play a remarkable role in determiningberry size. Water deficit [18, 19] and some pruning strat-egies [20] can result in smaller berries with higher phen-olic content in the skin. However, other treatments maylower the concentration of some phenolics along with areduction in berry size. This is seen with differences inthe microclimate of clusters, where shading results insmaller berries with a lower phenolic content in the skincompared to light exposed clusters [21].There are contrasting conclusions regarding differ-ences in berry composition when comparing berriesstrictly on size. Some research indicates no differencesbetween sizes [19, 22], while other research showssmaller Sauvignon Blanc berries having lower methoxy-pirazine concentrations (herbaceous aromatics accumu-lated in several cultivated varieties) compared to largerberries [23]. Similarly, there is no consensus on whethersmaller berries make superior wines. Gil et al. [24] dem-onstrated that smaller grapes produced wines of deepercolor and that size is inversely correlated with the con-centration of phenolics, such as anthocyanins and stil-benes. In contrast, others have found that smallerberries do not translate into discernable differences inthe resulting wines [22].Changes in gene expression often parallel changes inberry skin metabolism throughout berry development[12, 25–28] and in response to stresses, such as waterdeficit [12, 29] and UV [30]. Many transcripts related toberry ripening and quality traits, such as those involvedin hormone, phenylpropanoid, terpenoid, fatty acid, andcell wall metabolism, are modulated during developmentand in response to the environment. However, despitethe strong impact on wine composition and quality, theregulation of transcripts/pathways related to ripeningand berry composition in skins of differently sized ber-ries is still unknown. In this study, our goal was to betterunderstand the relationship between berry size, fruit rip-ening, and berry metabolism. We characterized howfield grown Merlot berries differ in their ripening, me-tabolism, and transcriptional program in relation toWong et al. BMC Genomics  (2016) 17:416 Page 2 of 22berry size. We considered two extreme size classes andcompared berry physiology, composition, and wholegenome gene expression (RNA-sequencing) in the skinthroughout development in order to determine the bio-logical processes that discriminate between the two clas-ses of berries.MethodsSample collection and physiological measurementsV. vinifera L.’Merlot’ clone 181 vines were grown in avineyard located in Oliver (British Columbia, Canada;49°13′18.8″N 119°33′28.8″W). Three biological repli-cates of fruit samples were harvested from three separaterows. Each replicate had 120 to 170 berries randomlycollected from 30 vines per row at 31, 39, 47, 51, 54, 57,60, 64, 67, 74, 103, and 121 days after anthesis (DAA) in2014. Care was taken to avoid physical damage as berrieswere trimmed off the cluster at the pedicel and placedinto aluminized mylar zip-lock bags to prevent waterand turgor loss [31]. The bags were then immediatelyplaced into a cooling box at ambient temperature andtransported to the laboratory. Berry diameter and elasti-city (E), a precise quantification of berry firmness [13],were recorded for each berry [32] and berries were indi-vidually bagged, labeled, and stored at −80 °C until thefollowing analyses were conducted. From the populationof berries at each sampling point, for each biologicalreplicate seven berries were selected and pooled accord-ing to their size. Two pools were created: large berries,consisting of berries included in the top 90–95th per-centile and small berries, consisting of berries includedin the bottom 10–15th. These berries were used to cal-culate the evolution of E and total soluble solids (TSS),using a manual refractometer, across the season. Add-itional pools of seven berries selected from the samepercentile at 47 (before ripening, 4.9 °Brix), 74 (early rip-ening, 17.5 °Brix), 103 (ripening, 22.4 °Brix) and 121 (lateripening, 25.3 °Brix) were used for both metabolite (seefollowing paragraphs) and RNA-sequencing (RNA-seq)analyses. Three biological replicates for each treatmentwere considered for both metabolite and RNA-seqanalysis.Liquid chromatography (LC) and LC-massspectrophotometric (MS) analysisBerries were peeled while still frozen using a scalpel.Flesh tissue was then used for determining sugar and or-ganic acid concentration, and skin tissue for anthocyanincontent.For the determination of sugar (glucose and fructose)and organic acid (malic and tartaric acid) content, fleshtissues were grounded to a fine powder and centrifugedat 16,000 g for 10 min. Afterwards, the supernatantwas passed through a 0.22 μm PVDF 13 mm filter(Whatman Inc., Sanford, USA) and measured via HPLCon an Agilent 1100 HPLC system with a refractive indexdetector (Agilent Technologies, USA). An aliquot of20 μl of extract was injected into a NUCLEOGEL® ION300 OA column (300 mm× 7.8 mm ID, 10 μm) (Mach-ery-Nagel Inc., USA), maintained at 71 °C. The mobilephase was 2.5 mM H2SO4 with a flow rate of 0.55 ml/min.Quantification of metabolite concentration (expressed asmg/ml of juice) was based on calibration curves of au-thentic standards.For the extraction of anthocyanins, 0.2 g of skins weregrounded and extracted in 2 ml 50 % (v/v) methanol inwater for 3 h with rigorous shaking. The supernatantwas centrifuged, filtered and measured on an Agilent1100 LC/MSD Trap XCT Plus mass spectrometer. Sep-aration of anthocyanin compounds was achieved on anAgilent Zorbax SB-C18 column (150 mm × 4.6 mm ID,1.8 μm) held at 57 °C. The mobile phases used were:water-formic acid (2 %), solvent A, and acetonitrile-formic acid (2 %), solvent B. Flow rate was 0.8 ml/min.The solvent gradient program was 0.5 min, 6 % B;4 min, 10 % B; 13 min, 25 % B; 20 min, 35 % B; 25 min,60 % B; 30 min, 95 % B; and 32 min, 6 % B. The antho-cyanins mass spectra were analyzed after electrosprayionization (ESI) in alternating positive and negativeionization mode with a scan range between 50 and850 m/z. Quantification of single and total anthocyaninswas based on malvidin 3-glucoside equivalents (expressedas mg/g fresh weight of berry skin). The percentages of 3′4′-OH, 3′4′5′-OH, and methoxylated anthocyanins werecalculated among monoglucoside anthocyanins.RNA extraction, sequencing, and data analysisExtraction of total RNA was achieved with the SpectrumPlant Total RNA kit (Sigma-Aldrich) using ~80 mg ofgrounded skins according to the manufacturer’s proto-col. The integrity of the extracted RNA was determinedon an Agilent 2100 Bioanalyzer (Agilent) ensuring a RINscore >7.5 prior to library construction. Ribosomal RNAdepleted library construction was performed using anin-house workflow using customized kits from NEB atthe Canada’s Michael Smith Genome Sciences Center(Vancouver, Canada) followed by sequencing (V4 chem-istry) on an Illumina HiSeq 2500 platform (Illumina)and in-house quality control and filtering of 75-bppaired-end generated reads. Filtered pair-end reads werealigned against the 12× reference genome [33] usingBurrows-Wheeler Aligner with default parameters [34].Read summarization was performed with htseq-count(version 0.6.0) with intersection non-empty settings [35]using the grapevine 12× genome assembly available fromEnsemblPlants (http://plants.ensembl.org/Vitis_vinifera/Info/Index). Differential gene expression analysis wascarried out using edgeR (version 3.10.2) [36]. Genes wereWong et al. BMC Genomics  (2016) 17:416 Page 3 of 22deemed differentially expressed (DE) between pairwisecomparisons at a threshold of false discovery rate (FDR)< 0.05. Transcript abundance was calculated as Frag-ments Per Kilobase of exon per Million (FPKM) mappedreads using edgeR. Transcripts having FPKM values <0.5and assigned counts <5 were discarded.Clustering, functional enrichment, and promoter analysisClustering of DE genes using edgeR’s estimated gene ex-pression (normalized counts) log2 fold change (log2FC)between small/large berries was performed using the k-means method with 1000 iterations and the Speaman’srank correlation as the similarity metric. The latestgrapevine gene annotations based the 12× V1 modelswere obtained from [37]. A separate functional annota-tion of transcripts was performed using the Mercatorpipeline [38] to ascribe potential gene function andMapMan BINs [39] prior to gene enrichment analysis.MapMan BIN categories were considered significantlyenriched (adjusted P-value <0.05) as determined by Fish-er’s exact test adjusted with Bonferroni correction formultiple testing correction. Based on the 12× grapevinegenome assembly, all grapevine promoter sequences(1 kb upstream of the 5′ UTR) were retrieved from theCRIBI grape genome database (http://genomes.cribi.unipd.it/grape/) [40]. Target sequences (cis-regulatoryelements, CREs) of 63 plant transcription factors (TFs)representing 25 families recently characterized in Arabi-dopsis thaliana [41] and the ones from PLACE [42] wereretrieved. These sequences were scanned in promoterregions of DE genes within clusters and selected genefamilies. Motif overrepresentation was calculated accord-ing to [43] using hypergeometric test and adjusted withfalse discovery rate (FDR) for multiple testing correctionwhen appropriate. Putative CREs was considered signifi-cantly enriched if the associated FDR < 0.01.Statistical analysisA one-way ANOVA was performed using JMP 7 (SASInstitute Inc.) to detect significant differences (P < 0.05,P < 0.01, P < 0.001) in berry components, E, TSS, andsugar, organic acid, and anthocyanin levels betweensmall and large berries treatments at each samplingpoint.Results and discussionAnalysis of physiological and compositional parametersbetween small and large berriesThe diameters of all the berries considered in the experi-ment are reported in Fig. 1a. Berry diameter, total sol-uble solids (TSS, °Brix), and elasticity (E) evolution insmall and large berries during fruit development areFig. 1 Evolution of berry diameter, total soluble solids, and elasticity. (a) Berry diameter of the entire sampled population. (b) Diameter, (c) totalsoluble solids, and (d) elasticity of small and large berries during development. Green and purple indicates the recorded color of each individualberry. Means and standard errors are reported for each berry group at each developmental stage. * indicates a significant difference (P < 0.05)between small and large berriesWong et al. BMC Genomics  (2016) 17:416 Page 4 of 22reported in Fig. 1b, c, and d. Despite the differences inabsolute values, the evolution of berry diameter wassimilar between small and large berries throughout theseason (Fig. 1b). The TSS levels of small and large ber-ries increased from 51 DAA until harvest (Fig. 1c). Afaster increase in TSS was observed for large berriesfrom 60 to 67 DAA. TSS levels of small berriesremained relatively low until 67 DAA; afterward, a sharpincrease was observed. E measures berry firmness, so afirm berry has higher E values (i.e. more pressure is re-quired for a particular displacement), and as a berrysoftens E decreases [13, 31]. E of the small and largeberries strongly decreased from 47 to 74 DAA, afterwardthe decrease was gradual (Fig. 1d). Differences in E levelsand in the rate of E decrease between small and largeberries were apparent at certain developmental stages.For example, at 39 and 54 DAA, large berries had higherE than small berries, while between 60 and 74 DAA,small berries had higher E than large ones. Although nosignificant differences in E levels were observed at 54and 57 DAA, a steeper drop of E was observed in largethan in small berries from 54 to 74 DAA (Fig. 1d).Berries from 47, 74, 103, and 121 DAA were selectedfor further physiological and biochemical analysis (Figs. 2and 3). The skin to berry weight ratio was significantlyhigher in small than large berries at all four develop-mental stages (Fig. 2b); while large berries had signifi-cantly higher seed to berry weight ratios at 47 and 121DAA (Fig. 2c). In addition, significant differences in thenumber of seeds/berry were observed between sizes,where large berries have more seeds (Fig. 2d), consistentwith previous studies [15, 22].Glucose and fructose levels increased sharply from 47to 74 DAA and continue to increase until 121 DAA(Fig. 3a). Malic and tartaric acids (Fig. 3b and c) were athigh levels at 47 DAA, and progressively decreasedfrom 74 DAA onwards. The levels of sugars and or-ganic acids reported here are comparable with thelevels found in different cultivars at the same develop-mental stage [12, 21, 44]. No significant differences inthe concentrations of sugars, glucose and fructose(Fig. 3a), and malic acid (Fig. 3b) were observed be-tween small and large berries for any of the samplingdates. Tartaric acid concentration was significantlyhigher in small than in large berries at 47 and 103 DAA(Fig. 3c).The total anthocyanin content (mg/g of skin) was sig-nificantly lower in small berries at 74 DAA (Fig. 3d). Al-though the recorded veraison (50 % of the berriesdisplay red pigmentation) date in the vineyard was60 DAA, the large increase in sugar levels in large ber-ries prior to 70 DAA (Fig. 1c) could have stimulated anearlier accumulation of anthocyanin in these berries [45,46]. The accelerated drop in E observed in large berries,as well as the faster increase in sugar and anthocyaninlevels, suggest that large berries approach ripening fasterthan small berries. Although the anthocyanin concentra-tion in skins (mg/g skin) was similar between small andlarge berries at 103 and 121 DAA (Fig. 3d), small berrieshad a significantly higher anthocyanin concentrationexpressed as mg/g of berry (Fig. 3e). Vice versa, totalanthocyanin content, expressed as mg/berry, was higherin large berries (Fig. 3f). This suggests that the increase inanthocyanin concentration observed in small berries at103 and 121 DAA was not due to a higher synthesis ofthese pigments but to the higher skin to berry weight ratio(Fig. 2b). This higher proportion of skin tissue determinesthe higher concentration of anthocyanin in small berriesas observed in [19].Analysis of transcript differences between small and largeberriesTo determine the influence of berry size on the berryskin transcriptome, Illumina mRNA sequencing was per-formed on berry skins at the four selected developmen-tal stages (47, 74, 103, and 121 DAA). Using the 12× V1genome as reference [33], an average of 34 million high-quality paired-end reads per sample were successfullymapped, which typically corresponds to 92.5 % ofpaired-end reads for each library (Additional file 1:Table S1). An average of 29 million paired-end readsper sample was assigned to transcripts. We detectedthe presence of 23,012 unique transcripts expressedin at least one of four developmental stages. Approxi-mately 18,600, 17,800, 17,200, and 17,150 transcriptswere expressed in berry skins at 47, 74, 103, and 121DAA, respectively.A principal component analysis (PCA) was undertakento analyze the level of similarity of the transcriptomesanalyzed. The first three principal components explaineda cumulative variance of 92.7 %; with the first, second,and third principal component explaining 71.2, 17.8,3.7 % of the variance, respectively (Fig. 4a). Inspection ofthe PCA plots revealed a clear separation of berry skintranscriptome based on the developmental stage and noton the biological variation within a developmental stage.Furthermore, a separation of the transcriptomes drivenby the berry size was more evident at 47, 74, and121 DAA, while it was undetected at 103 DAA. A totalof 3482 unique genes (11.6 % of predicted transcrip-tome) were identified to be differentially (FDR < 0.05)expressed between small and large berries in at least onedevelopmental stage (Additional file 1: Table S2 and S3).A total of 2083 (557 downregulated, 1526 upregulated),1244 (983 downregulated, 261 upregulated), 298 (57downregulated, 241 upregulated) and 928 (375 downreg-ulated, 573 upregulated) were differentially expressed(DE) between small and large berries at 47, 74, 103, andWong et al. BMC Genomics  (2016) 17:416 Page 5 of 22121 DAA, respectively (Fig. 4b). Gouthu et al. [47] dem-onstrated that differences in the ripening program (i.e.asynchrony) was reflected at the transcript level withlarge differential expression of berry transcripts whencomparing between berry classes differing in softnessand coloration at mid-veraison (~69 DAA), regardless oftissue. These differences were drastically reduced to nearzero at maturity. Our study revealed that differences inthe skin transcriptomes between small and large berriesare larger at earlier stages of development (47 and 74DAA) than at later stages of development (103 and 121DAA). This difference, together with the differences infruit physiology and composition observed at the corre-sponding stages further suggests an asynchrony betweensmall and large berries at early stages of development(Figs. 1a, d and 3c, d). The reduction of DE (from 47 to103 DAA) suggests a process of resynchronization of thetranscriptome between small and large berries duringdevelopment. Interestingly, there was a second increaseof the number of DE transcripts at 121 DAA. This mightreflect the activation of an over-ripening/senescenceprogram in the skins of small berries [28]. Cramer et al.[28] investigated the evolution of the skin transcrip-tome at late developmental stages, including over-ripening stages. Approximately 15 % of the top 1000DE genes were the same when results were comparedbetween the DE genes between small and large ber-ries at 121 DAA (25 °Brix) in this study and between25 and 38 °Brix in Cramer et al. [28], suggesting thatsmall berries are ahead in ripening at these laterstages.Differentially expressed genes were clustered into 8clusters based on their log2 fold change differences insmall with respect to large berries at each developmentstage using the k-means clustering (Fig. 4c). Functionalannotation of DE genes according to high-level MapManontology categories [39] showed that the largest propor-tion was involved in RNA regulation and protein metab-olism (8–10 %), while transport, cell wall, signaling, andprimary and secondary metabolism categories contrib-uted between 5 and 8 % of the total DE genes (Table1, Additional file 1: Table S3). A total of 382 (of ~2213)transcripts annotated as potential grapevine transcrip-tional regulators [37] were DE between small and largeberries within each sampling date (Additional file 1:Table S3), indicating a difference in transcript regulatoryprograms between small and large berries. ValidatingFig. 2 Berry features. (a) Skin, flesh, and seed weight, (b) skin toberry weight ratio, (c) seed/berry weight ratio, and (d) seeds/berrynumber in small and large berries at 47, 74, 103, and 121 DAA. *, **,and *** indicate level of significance of P < 0.05, P < 0.01, andP < 0.001, respectivelyWong et al. BMC Genomics  (2016) 17:416 Page 6 of 22the observation that a large number of TF were DEbetween small and large berries, many cis-regulatoryelements (CREs) related to binding sites of various TFfamilies were also enriched among the DE genes(Table2, Additional file 1: Table S4), which may play animportant role in the regulation of DE genes. The fre-quent occurrences of multiple enriched CREs withinpromoter regions of DE transcripts will be discussedfurther in the context of metabolic pathway regulation(see below).Because of the observed differences between smalland large berries at the physiological level and of thepotential role of some of the DE genes on determiningfruit and wine quality, we focused our discussion onpathways related to hormone metabolism and signal-ing, cell wall modifications, and flavonoid, stilbenoid,and fatty acid biosynthesis.Modulation of hormone metabolism and signaling insmall and large berriesRegulation of fruit development and ripening involvesmajor metabolic changes regulated by complex interac-tions among hormones and not by a single hormone inisolation [48]. In grapes, ABA promotes berry ripeningand auxins delay this process, while results on ethyleneare mixed [5]. One hundred fifty-seven genes annotatedto function in hormone metabolism (i.e. biosynthesis,degradation, signaling and transcriptional regulation;BIN 17) were DE between small and large berries(Additional file 1: Table S5). A total of 38, 42, and 29transcripts related to ABA, auxin, and ethylene metabol-ism and signaling, respectively, were the most representedamong hormone related genes, followed by 18, 14, 11, and6 transcripts related to jasmonic/salicylic acid, cytokinin,brassinosteroid, and gibberellin metabolism and signaling,Fig. 3 Berry composition. (a) Glucose + fructose, (b) malic acid, and (c) tartaric acid concentration, expressed as mg/g berry, in small and largeberries at 47, 74, 103, and 121 DAA. Anthocyanin levels expressed as (d) mg/g skin, (e) mg/g berry, and (f) mg/berry in small and large berriesat 47, 74, 103, and 121 DAA. Means and standard errors are reported for each berry group at each sampling. *, **, and *** indicate level ofsignificance of P < 0.05, P < 0.01, and P < 0.001, respectivelyWong et al. BMC Genomics  (2016) 17:416 Page 7 of 22respectively. Due to their major role in the ripeningprocess [5], emphasis will be given to the transcripts in-volved in the biosynthetic pathways of ABA, auxin, andethylene (Fig. 5).Transcripts encoding structural enzymes of the ABAbiosynthetic pathway were modulated during fruit devel-opment and were DE between small and large berries(Fig. 5a and b). Transcripts encoding 9-cis epoxycarote-noid dioxygenase (NCED) regulate ABA biosynthesisand the alteration of NCED expression have direct im-plications in ABA concentration and ripening-relatedtraits in fruits [49]. The expression patterns of three DENCED transcripts in the skin were distinct (Fig. 5a). Forexample, NCED2 (VIT_10s0003g03750) transcriptspeak at 47 DAA and gradually decrease towards 121DAA, NCED3 (VIT_19s0093g00550) transcripts peakat 74 DAA and again at 121 DAA, while NCED4(VIT_02s0087g00910) transcripts progressively increaseFig. 4 Analysis of the berry skin transcriptome of small and large berries. (a) Principal component analysis (PCA) of the berry skin transcriptome ofsmall (filled circles) and large (empty circles) berries at 47 (green), 74 (pink), 103 (purple), and 121 (blue) DAA. (b) The Venn diagram represents thecommon and unique genes differentially expressed between small and large berries at 47, 74, 103, and 121 DAA. Differentially expressed genes ineach intersection of the Venn diagram are described in Additional file 1: Table S3. (c) The box plot and smoothed line plot represents the responseof small versus large berries and dynamic change of gene expression during berry development, respectively. Differentially expressed genes wereclustered using the k-means algorithm. The log2 fold changes between small and large and log2 (FPKM +1) values at 47, 74, 103, and 121 DAAwere used. Outlier log2 fold change values are represented as grey circlesWong et al. BMC Genomics  (2016) 17:416 Page 8 of 22towards 121 DAA. These observations were consistentwith previous studies showing that NCED transcripts areunder complex regulation in the berry [10–12, 50] andthat expression of NCEDs is required to maintain in situberry ABA biosynthesis. Several studies have highlightedthat NCED3 is the enzymatic isoform which is correlatedwith berry ABA accumulation [9, 11]. Therefore, the up-regulation of this gene in small berries at 47 DAA maycontribute to higher ABA levels at this stage; however,later during ripening when ABA levels are still high [11,13, 47] no significant difference in NCED3 transcriptswas observed in the skins of small and large berries. In-stead, NCED2 transcripts were highly downregulated at74 DAA in skins of small with respect to large berries.This might have contributed to differences in overallABA accumulation and lower ABA accumulation in smallberries at early stages of ripening would be consistentwith the delay in sugar and anthocyanin accumulationobserved at 67 and 74 DAA, respectively (Figs. 1 and 2).In addition, the downregulation of ripening-associatedTable 1 Enrichment of MapMan functional categories (BINs) of k-means assigned clusters containing differentially expressed transcriptscomparing skins of large and small berries. The table contains detailed information of the cluster size, BIN code and associateddescription, and the Bonferroni-adjusted P-value (Adj. P-value <0.05). Only high-level BINs are presented in this table (up to one decimalor depth = 1)Cluster Size BIN BIN name Adj. P-value1 472 16 secondary metabolism 0.00047616.8 secondary metabolism.flavonoids 2.78E-0718.2 Co-factor and vitamine metabolism.thiamine 0.04316320.2 stress.abiotic 0.00594733.1 development.storage proteins 0.0024192 342 5.3 fermentation.ADH 0.000283 265 NA NA NA4 884 10.7 cell wall.modification 0.0065527 RNA 0.00835527.3 RNA.regulation of transcription 2.15E-0730 signalling 0.04360830.3 signalling.calcium 7.64E-055 320 16 secondary metabolism 1.54E-0716.8 secondary metabolism.flavonoids 2.81E-086 553 10 cell wall 8.42E-1010.8 cell wall.pectin*esterases 0.00201526 misc 0.00169126.28 misc.GDSL-motif lipase 0.01483930 signalling 0.00175230.2 signalling.receptor kinases 0.0001687 480 10 cell wall 0.00282810.6 cell wall.degradation 0.03317217 hormone metabolism 7.42E-0517.1 hormone metabolism.abscisic acid 1.1E-0820.2 stress.abiotic 0.00228934 transport 0.02522634.4 transport.nitrate 0.0244228 166 15 metal handling 0.0084333 development 0.04159433.99 development.unspecified 0.01258734 transport 0.00518534.12 transport.metal 0.024644Wong et al. BMC Genomics  (2016) 17:416 Page 9 of 22NCED4 gene in small compared to large berries from 103DAA onwards might also reflect the lower ABA biosyn-thetic capacity of small compared to large berries at 103and 121 DAA. Nevertheless, NCED2 was observed to beupregulated in small compared to large berries at 121DAA. Collectively, different sized berries exhibit a differ-ent regulation of NCED transcripts, and significantlylower NCED transcripts levels were found in small berriesthan in large ones during berry ripening, especially at 74DAA, with potential effects on the ripening process.The genes encoding TRYPTOPHAN AMINOTRANS-FERASE OF ARABIDOPSIS1/TRYPTOPHAN AMINO-TRANSFERASE RELATED (TAA1/TAR) and YUCCA(YUC) are critical for regulating auxin levels and ripen-ing in young berries [7]. The expression of transcriptsencoding TAA1/TAR1 – 4 and YUC1 throughout berrydevelopment (2 to 16 weeks post flowering) have alsobeen reported in Shiraz berries [7]. These genes weremodulated during fruit development and were also DEbetween small and large berries in this study (Fig. 5aand b). Three TAA/TAR transcripts (TAR2, 3, and 4) andanother (TAR1) peaked in the skins of berries harvested at47 and 103 DAA, respectively. One YUC transcript(VIT_04s0008g03920) was specifically present at 47 DAA,another YUC transcript (VIT_07s0104g01250) peakedaround veraison (74 DAA), and other two YUCs(VIT_07s0104g01260 and VIT_04s0023g01480) weremostly present during late ripening. These observationswere consistent with previous studies showing TAA/TARand YUC transcript evolution and auxin accumulationTable 2 Enrichment of PLACE- and PBM-curated cis-regulatory elements (CRE) of selected k-means assigned clusters containingdifferentially expressed transcripts comparing skins of small and large berries. The table contains detailed information on the numberof promoters with the specified CRE, number of promoters in the genome containing the specified CRE, CRE sequence, false discoveryrate (FDR), and the designated CRE name and regulatory description. A full description on the corresponding genes containingthe associated CRE is available in Additional file 1: Table S4Cluster Matches in sample Matches in genome FDR Motif Motif name Regulation1 92 3331 9.67E-06 ACGTGKC ACGTABREMOTIFA2OSEM ABRE146 6199 9.67E-06 CACGTG CACGTGMOTIF bHLH/bZIP98 4120 0.000552 AGATATTT CCA1-1 MYB-related33 1008 0.0024 YACGTGGC ABREATCONSENSUS ABRE54 2115 0.004942 CACGTGG IRO2OS bHLH/bZIP4 109 1808 1.27E-10 CACGCG MYC2-5 bHLH397 10,397 7.44E-09 MACGYGB ABRERATCAL ABRE469 13,792 0.000232 GCCAC SORLIP1AT Light137 3331 0.000652 ACGTGKC ACGTABREMOTIFA2OSEM ABRE286 7976 0.000884 WTTSSCSS E2FCONSENSUS E2F227 6199 0.001528 CACGTG CACGTGMOTIF bHLH/bZIP120 3057 0.005201 GAGTGAG SORLIP5AT Light303 8835 0.005201 CTGACY WBOXNTCHN48 WRKY5 154 10,397 9.78E-05 MACGYGB ABRERATCAL ABRE6 131 4411 1.51E-06 TGTCGG ETT-1 ARF149 5482 1.95E-05 ACCGAC DRE2COREZMRAB17 AP2/ERF185 7231 1.95E-05 RCCGAC DRECRTCOREAT AP2/ERF285 12,862 0.000589 RYCGAC CBFHV AP2/ERF325 15,123 0.000887 CCWACC MYBPZM MYB248 11,172 0.002184 ACCWWCC BOXLCOREDCPAL MYB128 5288 0.006668 TGTCGA ETT-2 ARF46 1504 0.0082 GGNCCCAC TCP15 TCP7 67 2312 0.000349 GCCGAC RAP2.3-3 AP2/ERF153 7231 0.003952 RCCGAC DRECRTCOREAT AP2/ERF116 5288 0.005831 TGTCGA ETT-2 ARF202 10,397 0.008081 MACGYGB ABRERATCAL bZIP/bHLH241 12,862 0.008373 RYCGAC CBFHV AP2/ERFWong et al. BMC Genomics  (2016) 17:416 Page 10 of 22during berry development and ripening [7]. At 47 DAA,when auxins levels are high, transcripts encoding TAR4(VIT_18s0157g00170) and a YUC (VIT_04s0008g03920)were highly upregulated in small with respect to largeberries. Similarly, TAR1 (VIT_00s0225g00230) transcriptswere also upregulated at 74 DAA. However, TAR3(VIT_18s0157g00090) transcripts, which have been shownto relate with the accumulation of auxin at early stages [6],were downregulated in small with respect to large berriesat 47 DAA. Several studies have indicated that auxin conju-gation through the action of IAA-amido synthetases (GH3),which conjugates aspartic acid to auxin, is critical for ripen-ing initiation and the regulation of auxins in the berry dur-ing development [51, 52]. Interestingly, GH3-1 transcripts(VIT_03s0091g00310) which display a ripening-associateddevelopmental accumulation were significantly upregulatedin the skins of small compared to large berries. Therefore,significant differences in ripening physiology of the twoberry classes (Figs. 1, 2, and 3) and significant modulationof auxin metabolic pathways points to the possible involve-ment of auxin and its conjugates in determining asyn-chrony between berries of different sizes.The biosynthesis of ethylene is determined, in part,by the action of 1-aminocyclopropane 1-carboxylate(ACC) oxidase (ACO), a key enzyme involved in theproduction of ethylene using ACC as precursor [48].Several ACO-encoding members were also DE be-tween small and large berries (Fig. 5b), many of whichexhibit highest expression at 47 DAA followed by agradual decrease until 103 DAA (Fig. 5a). Althoughgrapevine is a non-climacteric fruit, previous findingshave shown an increase of ACO activity andethylene production at stages before veraison [8, 9].Two ACO transcripts (VIT_11s0016g02380 andVIT_00s2086g00010) displayed strong upregulation at47 DAA in small compared to large berries. Thismight provide evidence that in small berries at pre-veraison stages a higher production of ethylene may infact be delaying the ripening process and not enhan-cing ripening, as exemplified by the delayed sugar ac-cumulation at 68 DAA and the lower anthocyaninlevels at 74 DAA in small berries (Fig. 3). Indeed, re-cent studies have demonstrated application of ethyleneat pre-veraison stages delays ripening through the ac-tivation of auxin biosynthesis pathways, significantlypostponing sugar and anthocyanin accumulation, aswell as the berry growth [7, 53, 54]. Ethylene mayalso play a role later in ripening, especially in the skin,as the abundance of many transcripts involved inethylene biosynthesis and signaling peaked from23 °Brix onwards, stages when desirable flavor andaroma compounds accumulate [28]. In this study,the significant upregulation of two ACO transcripts(VIT_12s0059g01380 and VIT_00s2086g00010) from103 DAA onwards, indicates that ethylene might con-tribute to the induction of pathways important forA BFig. 5 Evolution during development and fold change between small and large berries of abscisic acid (ABA), auxin, and ethylene genesdifferentially expressed at 47, 74, 103, and 121 DAA. (a) Evolution, based on the mean log2 (FPKM + 1) of small and large berries, and (b) log2 fold(small/large) changes. The relative log2 (FPKM + 1) values registered in small and large berries on average during berry development in a aredepicted by green (high expression) and blue (low expression). Grey color indicates the absence(or low levels) of detectable transcripts at the corresponding stage. Blue and red colors in b indicate downregulated and upregulated transcripts,respectively, in small berries in relation to large berries. Boxes with bold margins indicate significant differences (adjusted P-value <0.05) betweenberry size treatments at a given developmental stage. The cluster column in b indicates the cluster number the associated transcript belongs toNCED, 9-cis-epoxycarotenoid dioxygenase; ABA2, xanthoxin dehydrogenase; TAA/TAR, TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS1/TRYPTOPHAN AMINOTRANSFERASE RELATED (TAA1/TAR); YUC, YUCCA; GH, IAA-amido synthetase; ACO, 1-aminocyclopropane-1-carboxylicacid oxidaseWong et al. BMC Genomics  (2016) 17:416 Page 11 of 22fruit flavor in the skin of small berries compared tolarge ones [28].Among hormone related transcription factors, thegrapevine bZIP TF, VvABF2, have been shown to modu-late ABA-dependent berry ripening processes throughthe induction of cell wall hydrolytic enzymes and com-plex modulation of multiple hormonal pathways includ-ing ABA, auxin, and ethylene [55]. Expression ofVvABF2 transcript (VIT_18s0001g10450) was highest at74 DAA corroborating its ripening-associated expressionand its role in berry ripening (Additional file 1: TableS3). Interestingly, VvABF2 was first upregulated in smallberries at 47 DAA and subsequently downregulated at74 DAA compared to large berries. This modulationmight mediate the altered hormonal metabolism profilesinvolving NCED, TAA/YUC, and ACO transcripts be-tween small and large berries at 47 and 74 DAA. Also,the regulation of cell wall degradation and expansion viaABF2 may explain the slower softening rate experiencedin small compared to large berries observed at the onsetof fruit ripening.In summary, the differential regulation of severalNCED, TAA/YUC, and ACO transcripts between theberry size classes suggest that hormonal pathways aredifferential regulated and might therefore contribute tovariations in ABA, auxin, and ethylene levels in the skinof small and large berries with potential effects on skinmetabolism and composition. However, additional stud-ies with intensive sampling strategies and measurementof these hormones in the two berry classes will beneeded to gain a deeper understanding on the relativecontribution of each hormone and its influence on berrysize and fruit composition traits.Modulation of cell wall metabolism in small and largeberriesIn fleshy fruits, the remodeling of cell wall architectureduring fruit development affects fruit softening [56] andinvolves concerted changes in cell wall-related gene ex-pression and multiple enzyme activities [48]. Influenceof berry skins in determining post-veraison berry growth[57, 58] and the role of cell wall degradation and modifi-cation enzymes in mediating this process [59] have beenpreviously reported. The altered transcriptional profilesof cell wall genes in skins between small and large ber-ries highlights the potential role of cell wall modificationgenes in modulating berry growth and softening.In this study, more than 120 annotated cell wall (BIN10) transcripts exhibited DE between small and largeberries during fruit development (Additional file 1: TableS5). Transcripts related to cellulose synthesis (BIN 10.2),cell wall modification (BIN 10.7), degradation (BIN10.6), and pectin esterification (BIN 10.8) were signifi-cantly modulated among the berry size classes in theseclusters. A decrease in berry E has been shown to be theearliest ripening-associated event observed at the onsetof ripening [13]. Emphasis will be given on the cell wallgenes differentially expressed at 74 DAA (Fig. 6a–d),given that this is the developmental stage that immedi-ately follows the large decrease in berry E and thestage when small and large berries displayed diffe-rences in E.We observed that 17 cell wall degradation genes (BIN10.6.2 and BIN 10.6.3), consisting of six xyloglucanendo-transglycosylase/trans-hydrolase, two 1,4-beta-mannan endohydrolase, five pectate lyases, and threepolygalacturonases, were downregulated in skins ofsmall compared to large berries at 74 DAA (Fig. 6a andb). This observation may reflect lower cell wall degrad-ation activity in small than in large berries, which sup-ports the slower rate of softening from 60 to 74 DAA.Pectin degrading enzymes, such as polygalacturonasesand pectate lyases, are two principal enzymes involvedin tomato fruit softening through active cell wall pectinde-polymerization [56]. Studies in tomato [60], apple[61], and strawberries [62] have demonstrated that silen-cing of polygalacturonase gene expression does notalways affect fruit softening. Pose et al. [62] found largedifferences in strawberry fruit firmness at harvest instrawberry polygalacturonase 1-suppressed lines, but thechanges in firmness in tomato during softening werevery small [60] and in apple most transgenic polygalac-turonase suppression lines had no effect on fruitfirmness [61]. Nevertheless, silencing of pectate lyasestranscripts result in firmer fruits both during softeningand at harvest [63, 64].In plants, the transcription factor APETALA2/Ethylene Responsive Factor (AP2/ERF) belongs to a mul-tigenic family involved in the control of metabolism,stress response, and plant development through thebinding of DRE and GCC-related motifs in promotersregions of regulatory targets (reviewed in [65]). Severalstudies in fruits have demonstrated that AP2/ERF TFscontribute in regulating ripening-related processes, espe-cially fruit softening, by targeting cell wall degradationgenes (reviewed in [48]). In this study, significantenrichment of ethylene related CREs, such as DRECRT-COREAT, CBFHV, and LTRECOREATCOR15 (Fig. 6e,Additional file 1: Table S4), in cluster 7 genes also corre-sponds to the presence of AP2/ERF CREs in promoterregions of seven (three 1,4-beta-mannan endohydrolase,two pectate lyases, and two polygaluacturonase) pre-dicted cell wall degrading transcripts. In agreement, weobserved eight DE AP2/ERF TFs which were signifi-cantly downregulated in small compared to large berriesat 74 DAA; four of which were ripening-associated(Additional file 1: Table S3). This suggest that AP2/ERFTFs may be critical in regulating cell wall degradingWong et al. BMC Genomics  (2016) 17:416 Page 12 of 22Fig. 6 Evolution during development, fold change values between small and large berries, and selected cis-regulatory element profile of cell walltranscripts differentially expressed at 47, 74, 103, and 121 DAA. The heat map represents the evolution (a and c), based on the mean log2 (FPKM+ 1) of small and large berries, and log2 fold (small/large) changes (b and d) of cell wall degradation (a and b) and modification (c and d) genes.The relative log2 (FPKM + 1) values registered in small and large berries on average during berry development in a and c are depicted by green (highexpression) and blue (low expression). Grey color indicates the absence (or low levels) of detectable transcripts at the corresponding stage. Blue and redboxes in b and d indicate downregulated and upregulated transcripts, respectively, in small berries in relation to large berries. Boxes with bold marginsindicate significant differences (adjusted P-value <0.05) between small and large berries at a given developmental stage. The cluster column inb and d indicates the cluster number the associated transcript belongs to. EM, 1,4-beta-mannan endohydrolase; PL pectate lyase; PG Polygalacturonase;XET/XTH Xyloglucan endotransglucosylase/hydrolase. (e and f) The heat map illustrates the presence of AP2/ERF and bHLH/NAC cis regulatory elementsin cell wall genes. Purple and white colors depict the presence and absence of the respective CRE in the promoter regions of the geneWong et al. BMC Genomics  (2016) 17:416 Page 13 of 22transcripts in grapevines, and differences in AP2/ERF TFregulation between small compared to large berries priorto ripening, may have affected the progression of soften-ing reaching this stage (Fig. 1d).Expansins act as primary cell wall loosening agentsand incorporators of new cell wall material during thefirst growth phase and as mediators of cell wall disas-sembly by facilitating the contact between cell wall ma-terial and degradation enzymes during stage III ofgrowth in grapevine berries [59, 66]. The grapevine gen-ome contains 29 expansin genes classified into four dis-tinct subfamilies (expansin A, expansin B, expansin-likeA, and expansin-like B) [67], of which two grapevineexpansins were demonstrated to promote cell expansion[68]. Many cell wall expansin transcripts (BIN 10.7) werehighly expressed at 74 DAA and six of these transcriptswere significantly downregulated in skins of small withrespect to large berries. Moreover, other six transcriptswere downregulated at 121 DAA when their level of ex-pression was on average low (Fig. 6c and d). The down-regulation of expansin transcripts observed in the skinsof small compared to large berries at 74 and 121 DAAindicates that cell wall loosening in the skin may be gen-erally reduced, potentially limiting skin and mesocarpexpansion and resulting in smaller berries. Indeed, fruitsof tomatoes overexpressing expansin displayed enhancedsoftening and were significantly larger compared to con-trols at various stages of fruit development [69].The regulation of some expansin transcripts such as EXPA19(VIT_18s0001g01130) and EXPB4 (VIT_15s0021g02700)by VvCEB1 have also been proposed [55, 70]. Analysisof VvCEB1, a fruit ripening-specific bHLH transcriptionfactor, gene expression from 10 different varieties exhi-biting differences in berry size revealed a strong correl-ation between fruit size and VvCEB1 transcriptaccumulation [71]. In this study, VvCEB1 transcript(VIT_01s0244g00010) reached a maximum at 74 DAAand slowly decreased until 121 DAA (Additional file 1:Table S3). VvCEB1 was downregulated in skins of smallcompared to large berries at 74 and 121 DAA, but wereonly significant at 121 DAA. The down regulation ofVvCEB1 in small berries at 74 and 121 DAA paralleledthe down regulation of many expansin transcripts. Pro-moter analysis performed on DE expansins for bHLH-related CRE revealed that the CATGTG element, a typ-ical DNA binding domain for bHLH TFs [72], wasenriched in the promoter region of 10 out of 13 DEexpansin genes (Fig. 6f, Additional file 1: Table S4).This suggests that upstream control of expansins differ-entially expressed in small and large berries may in partbe mediated by VvCEB1 during ripening. Nevertheless,previous studies have also shown that NAC TFs canbind specifically to CATGTG elements [73], whichmight provide evidence for the regulation of expansintranscripts by NAC TFs. We observed that the majorityof NACs, such as a ripening-associated NAC26(VvNAC26, VIT_01s0026g02710) which is highlyexpressed at 74 and 103 DAA, were downregulated insmall compared to large berries at 74 DAA in parallelwith downregulation of several expansin genes (Add-itional file 1: Table S3). A recent study has also shownthat polymorphisms in VvNAC26 were associated withberry size variation among 114 grapevine varieties [74].Together, this transcriptional network involving bHLHand NAC TFs may be critical for regulating berrygrowth and size determination in grapevine.Modulation of flavonoid and stilbenoid pathways in smalland large berriesFlavonoids and stilbenes are plant secondary metabolitescommonly found across the plant kingdom. These spe-cialized metabolites fulfill diverse physiological roles instress response, as antioxidants, and during reproduction[75]; in grapevine, they strongly affect grape and winequality [76]. We observed that differences in berry sizewere related to large differences in the expression of fla-vonoid genes (Fig. 7, Additional file 1: Table S5). The DEflavonoid genes were predominantly assigned to clusters1 (19 genes) and 6 (5 genes).Three chalcone synthases (CHSs; VIT_07s0151g01060,VIT_13s0067g02870, and VIT_13s0067g03820), andthree chalcone isomerases (CHIs; VIT_05s0136g00260,VIT_14s0068g00920, and VIT_14s0068g00930) -codinggenes were differentially modulated in small and largeberries. One CHS (VIT_13s0067g02870) and one CHI(VIT_05s0136g00260) were downregulated in small rela-tive to large berries at 47 DAA when the pathway pre-cursors are used for the production of flavan-3-ol andproanthocyanidins. Later during development, one CHS(VIT_13s0067g03820) was upregulated at 74 DAA, whiletwo (VIT_07s0151g01060 and VIT_13s0067g02870) outof three CHSs and all the CHIs mentioned above weredownregulated in small in comparison to large berries at121 DAA.Of the total 16 flavonoid 3′5′-hydroxylases (F3′5′Hs) found in the grapevine genome, we observed seven F3′5′H transcripts (VIT_06s0009g02810, VIT_06s0009g02840,VIT_06s0009g02860, VIT_06s0009g02880, VIT_06s0009g02920, VIT_06s0009g02970, and VIT_06s0009g03010)that were significantly downregulated in small berries at103 and 121 DAA (Cluster 1). Similarly, two flavonoid3′-hydroxylases (F3′Hs; VIT_17s0000g07200 and VIT_17s0000g07210) and one flavanone 3-hydroxylase (F3H;VIT_04s0023g03370) -coding genes, were also significantlydownregulated in small berries at 121 DAA (Cluster 1). F3′5′H and F3′H enzymes have been shown to be critical indetermining the accumulation of di- and tri-hydroxylatedflavonoids and the ratio between blue and red anthocyaninWong et al. BMC Genomics  (2016) 17:416 Page 14 of 22VIT_04s0023g03370Naringenin chalconeNaringeninEriodictyolStilbenes 4-Coumaryl-CoAVIT_10s0042g00920VIT_16s0100g00750VIT_16s0100g00770VIT_16s0100g00780VIT_16s0100g00800VIT_16s0100g00830VIT_16s0100g00840VIT_16s0100g00850VIT_16s0100g00900VIT_16s0100g00910VIT_16s0100g00930VIT_16s0100g00940VIT_16s0100g00950VIT_16s0100g00960VIT_16s0100g00990VIT_16s0100g01000VIT_16s0100g01010VIT_16s0100g01020VIT_16s0100g01030VIT_16s0100g01070VIT_16s0100g01110VIT_16s0100g01130VIT_16s0100g01140VIT_16s0100g01150VIT_16s0100g01160VIT_16s0100g01170VIT_16s0100g01190VIT_16s0100g01200VIT_00s2508g00010VIT_00s2849g00010VIT_06s0004g02620VIT_08s0040g01710VIT_13s0019g04460VIT_16s0039g01100VIT_16s0039g01110VIT_16s0039g01120VIT_07s0151g01060VIT_13s0067g02870VIT_13s0067g03820VIT_06s0009g02810VIT_06s0009g02840VIT_06s0009g02860VIT_06s0009g02880VIT_06s0009g02920VIT_06s0009g02970VIT_06s0009g03000VIT_06s0009g03010VIT_17s0000g07210VIT_17s0000g07200PentahydroxyflavononeDihydroquercetin Dihydrokaemferol DihydromyrecitinSTSF3’-H F3’5’-HF3HCHICHSF3’-H F3’5’-HF3HDFRDFRFLSFLS FLS VIT_18s0001g03430VIT_18s0001g03470Flavonols FlavonolsFlavonolsLeucocyanidin Leucopelargonidin LeucodelphinidinLDOXLARLAR LARFlavan-3-ols Flavan-3-olsFlavan-3-olsCyanidin Pelargonidin DelphinidinUFGTUFGT UFGTANRANR ANRVIT_02s0025g04720VIT_03s0038g04220DFRLDOX LDOXVIT_17s0000g04150VIT_00s0361g00040VIT_16s0039g02230Cyanidin-3-O-glucoside Pelargonidin-3-O-glucoside Delphinidin-3-O-glucosideAOMT AOMTPeonidin-3-O-glucoside Petunidin-3-O-glucosideVIT_05s0049g01010VIT_05s0049g01020VIT_07s0005g03340MYB13-15VIT_05s0136g00260VIT_14s0068g00930VIT_14s0068g00920Malvidin-3-O-glucosideF3HLog2 Fold change-1.5 0 1.5MYBA TFVIT_02s0033g00380VIT_02s0033g00390VIT_02s0033g00410VIT_02s0033g004504-CoumarateTrans-cinnamate4CLC4H VIT_11s0065g00350VIT_11s0078g00290VIT_02s0025g03660VIT_16s0039g02040VIT_16s0050g00390VIT_16s0039g01130VIT_16s0039g01170VIT_16s0039g01240VIT_16s0039g01280VIT_16s0039g01300VIT_16s0039g01320VIT_16s0039g01360PhenylalaninePALFig. 7 (See legend on next page.)Wong et al. BMC Genomics  (2016) 17:416 Page 15 of 22in grapes [77–79]. The simultaneous downregulation ofboth F3′H and F3′5′H transcripts in small compared tolarge berries, particularly at 121 DAA, did not affect therelative abundance of the different anthocyanin forms (datanot shown). Flavonol synthase (FLS) is the key enzymefor the production of flavonols in grapevines [80]. Ofthe five FLS transcripts encoded in the genome, thetwo transcriptionally active FLS (VIT_18s0001g03470 andVIT_18s0001g03430) exhibited a ripening-associated accu-mulation in the skin (Additional file 1: Table S5); similar tothat observed in previous studies in developing grapeberries [80, 81]. Transcripts encoding grapevine FLS5(VIT_18s0001g03430) were significantly upregulated insmall berries at 47, 103, and 121 DAA (Fig. 7). Similarly, theFLS4 (VIT_18s0001g03470) gene was significantly upregu-lated in small with respect to large berries at 74 DAA. Pre-vious studies have shown that FLS transcript expressionprovides an excellent indicator for cluster light exposureand marker for flavonol synthase activity given strong posi-tive relationship (correlation) of solar radiation intensitywith FLS transcripts and flavonol content in berry skins[82, 83]. In the current study, berries of different sizes wereharvested from the same clusters controlling for differencesin microclimate among berries. Thus differences in FLS ex-pression between berries of different sizes must have arosefrom endogenous mechanisms. Consistently, previous stud-ies reported that wine produced from small berries containmore flavonols compared to wines obtained from mediumand large berries [24].The expression level of key flavan-3-ol/proanthocyani-din genes codifying for leucoanthocyanidin reductase(LAR) and anthocyanidin reductase (ANR) were highlyexpressed at 47 DAA and decreased in the expression at74 DAA, consistently with the early flavan-3-ols andproanthocyanidin accumulation during berry develop-ment [84]. Downregulation of ANR transcript at 47DAA and upregulation of LAR transcripts at 74 DAAwere observed in skins of small compared to largeberries indicating a differential regulation of flavan-3-olbiosynthesis during stages that are critical for the accu-mulation of these compounds and their polymeric forms(proanthocyanidins) in the skin (Fig. 7).Glycosylation of cyanidin and delphinidin via UDPglucose:flavonoid 3-O-glucosyltransferase (UFGT) [85] iscritical in anthocyanin synthesis. The expression ofgrapevine UFGT (VIT_16s0039g02230) was significantlydownregulated at 121 DAA (Cluster 1). MYBA1 andMYBA2 TFs redundantly regulates UFGT expressionand modulates anthocyanin accumulation [86–88]. TheMYBA transcripts were specifically activated at 74 DAAand decreases approaching 121 DAA (Fig. 7, Additionalfile 1: Table S5). MYBA1 (VIT_02s0033g00410) was sig-nificantly downregulated in small compared to large ber-ries at 74 and 121 DAA, while the others MYBAs wereall downregulated at 121 DAA. The observed downregu-lation of grapevine MYBA1 in small compared to largeberries at 74 DAA is consistent with the lag of anthocya-nin accumulation in smaller berries suggesting a delayedonset of ripening. Although no concurrent downregula-tion of UFGT with MYBA1 was observed in small com-pared to large berries at 74 DAA, significant reductionof anthocyanin in small berries at 74 DAA might bestrongly related to altered regulation of anthocyaninmodification and transport pathways regulated byMYBA TFs [88]. Concurrent downregulation of MYBA1,MYBA2, and MYBA3 with many flavonoid pathwaystranscripts (three CHS, seven F3′5′H, and UFGT tran-scripts) targeted by MYBAs [88] did not significantly re-duce the concentration of anthocyanin in the skin at 121DAA. Anthocyanin accumulation ceased from 103 DAAin both small and large berries. This indicates that theprecursors for anthocyanin production (e.g. coumaroyl-CoA) are channeled to the production of other phenoliccompounds, such as stilbenes, at late stages of ripening[89].A marked upregulation of genes encoding enzymesinvolved in the general phenylpropanoid and stilbenoidpathway, namely phenylalanine ammonia lyase (PAL),cinnamic acid 4-hydroxylase (C4H), 4-coumarate:CoAligase (4CL), and stilbene synthase (STS), was observedat 47 and 121 DAA (Fig. 7, Additional file 1: Table S5).Of particular interests are transcripts encoding STSenzymes. In plants producing stilbenes, STS catalyzesthe production of cis- or trans-resveratrol using 4-coumaroyl-CoA and three molecules of malonyl-CoA.Interestingly, trends in gene expression of the majorityof transcripts encoding STS and PAL were very similar(Additional file 1: Table S5). Several studies have(See figure on previous page.)Fig. 7 Modulation in berry skin transcripts involved in the phenylpropanoid and flavonoid pathway in small and large berries at 47, 74, 103, and 121DAA. Blue and red boxes indicate downregulated and upregulated transcripts, respectively, in small berries in relation to large berries. Boxes with boldmargins indicate significant differences (adjusted P-value <0.05) between berry size treatments at a given developmental stage. Transcription factors(colored in yellow) involved in the regulation of the phenylpropanoid and/or the flavonoid pathway transcripts are depicted in dottedlines. PAL Phenylalanine lyase; C4H Cinnamate-4-hydroxylase; 4CL 4-Coumarate:coenzyme A ligase; CHI Chalcone isomerase; CHS Chalconesynthase; F3H Flavanone 3-hydroxylase; F3′H Flavonoid 3′-hydroxylase; F3′5′H Flavonoid 3′5′-hydroxylase; FLS flavonol synthase; DFR Dihydroflavonol4-reductase; LAR Leucoanthocyanin reductase; LDOX Leucoanthocyanin dioxygenase; UFGT UDPglucose:flavonoid 3-O-glucosyltransferase; and AOMTAnthocyanin O-methyltransferase. All other information is available at Additional file 1: Table S5Wong et al. BMC Genomics  (2016) 17:416 Page 16 of 22reported that multiple STSs are responsible for the pro-duction of stilbene/resveratrol accumulation in the skinduring berry development and in response to abiotic andbiotic stress [89–91]. A large proportion of PAL and STStranscripts were significantly upregulated in small ber-ries especially at 47 and 121 DAA (Fig. 7). Similar to thatof anthocyanin, stilbenes largely accumulate at the onsetof ripening [89–91]. However, no DE expression was ob-served at 74 DAA where their expression tends to peak.Concerted downregulation of CHS and upregulation ofSTS at 121 DAA in small compared to large berriesmight relieve the competition for coumaroyl-CoA andmalonyl-CoA substrates by CHS; thereby favoring theflux to stilbene production, since anthocyanin levels haveplateaued in both berries from 103 DAA onwards(Fig. 3d). Although the higher skin to berry weight ratiooften determines a higher concentration of the antho-cyanin and stilbenes accumulated in the skin [2, 3], con-certed upregulation of the stilbene branch in smallberries may enhance stilbene levels in wines made fromsmall berries compared to normal and large sized berriesharvested at the same time [24]. This induction of PALand STS genes in small compared to large berries at 121DAA might also be a consequence of altered hormonelevels. Hormones such as ethylene and jasmonic acidsare not only involved in the induction of key stilbenegenes (STS) and levels in grapevines [92–94] but also, inthe case of ethylene, might take part to the over ripeningprocess [28].Two grapevine MYB TFs, MYB14 (VIT_07s0005g03340)and MYB15 (VIT_05s0049g01020) directly regulate thebiosynthesis of stilbene in grapevines [90]. In this study,MYB15 had consistent expression trends, in both smalland large berries, peaking at 74 DAA, while MYB14 tran-script evolution was not consistent between small andlarge berries. In large berries, MYB14 transcripts progres-sively increase until 103 DAA before decreasing at 121DAA; while, in small berries, MYB14 expression peaked at47 DAA before decreasing slightly at 74 DAA and increas-ing again later during ripening (Additional file 1: TableS5). Nonetheless, grapevine MYB14 and MYB15 were sig-nificantly upregulated in small berries at 121 and 47 DAA,respectively (Fig. 7). This result indicates their role in theregulation of STS expression and stilbene production inthe skin.Modulation of fatty acid degradation pathways in smalland large berriesAromatic alcohols and aldehydes derived from the fattyacid metabolism pathways (Fig. 8a), such as hexanol/hexe-nol and heptanal/hexanal, also contribute to late-ripeningassociated flavor/aromas in fruits and finished wines [95].Several successive steps involving lipoxygenase (LOX), hy-droperoxide lyase (HPL), alcohol dehydrogenases (ADHs),and alcohol acyl transferases (AATs) enzymes regulate theproduction of these compounds in plant tissues [96]. Manytranscripts encoding enzymes of the aforementioned stepswere modulated during fruit development and differed inthe abundance between small and large berries (Fig. 8b andc, Additional file 1: Table S5). Lipoxygenase enzymes are in-volved in the formation of hydroperoxides from linolenicacid, which can lead to the accumulation of key winearomas in the grape. Two LOX genes differentiallyexpressed between small and large berries were highlyexpressed at 47 and 74 DAA before decreasing towards 121DAA. One LOX transcript (VIT_05s0020g03170) was sig-nificantly downregulated at 74 DAA and another LOX tran-script (VIT_09s0002g01080) was significantly upregulatedat 47 DAA in small compared to large berries. Hydroperox-ides are then converted to C6 related aldehydes, whichoften confer grassy-related aromas via HPL. One HPL tran-script (VIT_12s0059g01060), mostly accumulated at 47 and121 DAA, was downregulated in small compared tolarge berries. Subsequently, ADH converts C6 relatedaldehydes to C6 related alcohols, which can be fur-ther converted to produce volatile aromas via AATs.Interestingly, all seven ADH transcripts encoded inthe grapevine genome were DE. Four ADH transcriptswere highly expressed at 47 DAA and sharply de-crease from 74 DAA and onwards; one ADH tran-script peaked at 74 DAA while another two werehighly expressed at 121 DAA (Fig. 8b). The four earlydevelopmental related ADH transcripts (VIT_18s0001g15410, VIT_04s0044g01110, VIT_04s0044g01120,and VIT_04s0044g01130), as well as two (VIT_18s0001g15450 and VIT_14s0068g01760) of the threelate-ripening ADH genes, displayed significant upregu-lation in small compared to large berries at 47 DAA(Fig. 8c). On the contrary, at 121 DAA, one(VIT_18s0001g15450) of the two ADHs that areexpressed at high levels at this stage was downregu-lated in small berries. The accumulation profiles ofthe different aldehydes, alcohols, and esters in Caber-net Sauvignon grapes during berry development andripening has been reported and their formation isthought to involve a synergy between the biosynthesisand catabolism of various steps [97]. Although themolecular mechanisms are not fully understood, highlevels of volatile esters in young berries (6 weeks postflowering) and high levels of alcohol compounds ac-cumulated in later stages of berry development (12–14 weeks post flowering) are reported [97]. Therefore,the strong upregulation of four early ripening ADHand one LOX transcripts in small compared to largeberries might favor a higher production of C6 alde-hydes and subsequent turnover/flux to alcohols/esters(for fruitier aromas) during early development (47DAA) in small compared to large berries.Wong et al. BMC Genomics  (2016) 17:416 Page 17 of 22In tomato, a MADS-box transcription factor(SlRIN) is a negative regulator of fruit ripening andacts as a modulator of aroma production in fruits viadirect regulation of ADH, LOX, and HPL promoters[98]. We identified nine MADS-box TFs differentiallyexpressed between small and large berries: four geneswere allocated to clusters 2, two to clusters 6, andone to cluster 3, 5, and 7 individually. Of noteare the predicted homologs of tomato RIN(VIT_14s0083g01050 and VIT_01s0011g00110), allo-cated into cluster 2, that share high expression simi-larity with four ADH transcripts from 47 to 121 DAAin small and large berries (Additional file 1: TableS5). Analysis of the promoter regions of the total DEAB CDAATMotif name Motif LOXOLOXLADHADH1ADH2-1ADH2-2ADH2-3ADH3ADH7HPLCARGATCONSENSUSCARGCW8GATCARGNCATFig. 8 Berry skin transcript and selected cis-regulatory element profile of small and large berries of the fatty acid degradation/C6 volatile biosynthesispathway at 47, 74, 103, and 121 DAA. (a) Simplified pathway schematic. The heat map represents the transcript evolution (b), based on the mean log2(FPKM + 1) in small and large berries, and log2 fold (small/large) changes (c). The relative log2 (FPKM + 1) values from the four time points in bare depicted by green (high expression) and blue (low expression). Grey color indicates the absence (or low levels) of detectable transcripts atthe corresponding stage. Blue and red boxes in c indicate downregulated and upregulated transcripts, respectively, in small berries in relationto large berries. Boxes with bold margins indicate significant differences (adjusted P-value <0.05) between small and large berries at a givendevelopmental stage. The cluster column in c indicates the cluster the associated transcript belongs to. (d) The heat map illustrates the distributionof MADS box CREs in promoter regions of aroma-related transcripts differentially expressed between small and large berries. Purple and white colorsdepict the presence and absence of each CREs, respectively, in the promoter regions of the relevant transcripts. LOX Lipoxygenase; HPL Hydroperoxidelyase; ADH Alcohol dehydrogenase; AAT Alcohol acyl transferasesWong et al. BMC Genomics  (2016) 17:416 Page 18 of 22fatty acid degradation pathway genes (two LOX,seven ADHs, and one HPL) for various MADS-boxbinding site (CArG box) revealed that one motif(CWWWWWWWWG) was located within all testedpromoter regions except promoter region of an ADHtranscript (Fig. 8d, Additional file 1: Table S4). Theseobservations indicate that berry size may be associ-ated with a different accumulation of aromatics in thefruit, likely through the coordinate regulation ofMADS-box TFs and structural genes of the fatty aciddegradation pathway.ConclusionsSignificant differences in the physiology, biochemis-try, and transcripts are found in different sized ber-ries during fruit development. As small and largeberries approach the onset of ripening, clear differ-ences in the rate of development were apparent.Around the onset of ripening, the steeper drop inelasticity, more rapid accumulation of sugars, lowertartaric acid, and greater anthocyanin levels in largecompared to small berries suggest that fruit size isassociated to changes in the ripening physiology ofthe berry, where large berries approach ripeningfaster. These differences correspond to congruentchanges in the hormonal pathways related to ABA,auxin, and ethylene via genes encoding NCED forABA, TAA1/TAR and YUC for auxin, and ACO forethylene. Genes encoding pathways contributing tofruit texture, flavor, and aroma were also differentiallymodulated accordingly to the berry size. The modula-tion of cell wall degradation and modification genes(e.g. PG and EXP) may contribute to the differencesin elasticity decreases and berry growth/size. Upregu-lation of fatty acid degradation genes, especially ADH,during early development might favor production ofdesirable aromatics in small berries. In the late ripen-ing stages, concurrent upregulation of phenylpropa-noid and stilbenoid pathway genes with a paralleldownregulation of the flavonoid pathway in skins ofsmall compared to large berries indicates that smallerberries may have higher concentrations of aromaticsand stilbenes, major antioxidants produced in theberry. Further investigation on endogenous and ex-ogenous factors that regulate the fruit metabolism inberries of different sizes is necessary to identify thekey factors – e.g. hormone signals, advantage of pos-ition within a cluster that favors mineral or wateruptake, or a better microclimate – that determine theberry size itself as well as the transcriptome response.Finally, a deeper investigation of the fruit composition inrelation to berry size may lead to the adoption ofscreening strategies based on size for tailoring fruitand wine quality.Additional fileAdditional file 1: Table S1. Summary of RNA sequencing analysismetrics. RNA sequencing were carried in skins of small and large berriesat four berry developmental stages namely 47 (before ripening, 4.9 °Brix),74 (early ripening, 17.5 °Brix), 103 (ripening, 22.4 °Brix), and 121 (lateripening, 25.3 °Brix) days after anthesis (DAA). Table S2. Transcriptabundance of the DE genes, reported as log2 (FPKM + 1) values, for eachindividual biological replicate (R1, R2, R3) in each treatment (small andlarge berry) at 47, 74, 103, and 121 days after anthesis. Table S3.Summary of differentially expressed genes between small and largeberries at four berry developmental stages. All differentially expressedgenes (Adj. P-value <0.05) and detailed description of the 12×V1 gene ID,log2 fold change values (small vs large), differential expression calls acrossfour developmental stages, average log2 (FPKM + 1) values (3 replicates)of small and large berries together and individually, k-means assignedcluster (based on response and developmental stage), functional annotationsbased on Grimplet et al. [37] (including transcription factors), andMapMan pipeline. Table S4. Summary of PLACE- and PBM-curatedcis-regulatory elements (CRE) analysis of k-means assigned clustersand selected group of genes. All information on the number of promoterswith the specified CRE (match_in_sample), the number of genes withineach group, number of promoters in the genome containing the specifiedCRE (match_in_genome), P-value and FDR of enriched CRE (FDR < 0.01),motif name and description, and a short description on selected CREinstances in groups. Table S5. Summary of selected hormone, cell wall,flavonoid/phenylpropanoid, and aroma/flavor metabolic pathways transcriptsdifferentially expressed genes between small and large berries at fourberry developmental stages which have been divided into separateclasses and sub-categories. (XLSX 2418 kb)AcknowledgmentsWe thank Mission Hill Family Estate for providing the grapes used in thisstudy. We thank Dr. Patricia Bowen (Agriculture and Agri-Food Canada) forproviding laboratory space during the sample collection. We thank Canada’sMichael Smith Genome Sciences Centre for the sequencing services.FundingThis study was funded by the University of British Columbia, GenomeBC, and Natural Sciences and Engineering Research Council of Canada.Availability of data and materialsAll raw sequence reads have been deposited in NCBI Sequence ReadArchive (http://www.ncbi.nlm.nih.gov/sra). The BioProject and SRA accessionsare PRJNA316157 and SRP072255, respectively.Authors’ contributionsDCJW performed the physiological measurements, liquid chromatography(LC) and LC-mass spectrophotometric analysis, RNA extractions, transcriptomedata analysis, cis-regulatory sequence analysis, interpreted the results, anddrafted the manuscript. RLG performed the transcriptome and cis-regulatorysequence data analysis. ND performed part of the physiological measurements.GAG contributed to the experimental design, expert advice throughoutthe study, interpreted the results, and drafted part of the manuscript. SDCconceived the study, coordinated the field experiments, performed thephysiological measurements, supervised the entire study, interpreted theresults, and drafted part of the manuscript. All authors read and approvedthe final manuscript.Competing interestsThe authors declare that they have no competing interests.Author details1Wine Research Centre, University of British Columbia, Vancouver, BC,Canada. 2Bordeaux Sciences Agro, Institut des Sciences de la Vigne et du Vin,Ecophysiologie et Génomique Fonctionnelle de la Vigne, UMR 1287, F-33140Villenave d’ Ornon, France.Received: 4 February 2016 Accepted: 25 April 2016Wong et al. BMC Genomics  (2016) 17:416 Page 19 of 22References1. Lund ST, Bohlmann J. The molecular basis for wine grape quality–a volatilesubject. Science. 2006;311:804–5.2. Singleton VL. Effects on red wine quality of removing juice beforefermentation to simulate variation in berry size. Am J Enol Vitic.1972;23:106–13.3. Matthews MA, Anderson MM. Fruit ripening in Vitis vinifera L.: responses toseasonal water deficits. Am J Enol Vitic. 1988;39:313–20.4. Coombe BG. Research on development and ripening of the grape berry.Am J Enol Vitic. 1992;43:101–10.5. Kuhn N, Guan L, Dai ZW, Wu BH, Lauvergeat V, Gomès E, Li SH, Godoy F,Arce-Johnson P, Delrot S. Berry ripening: recently heard through thegrapevine. J Exp Bot. 2014;65:4543–59.6. Böttcher C, Harvey K, Forde CG, Boss PK, Davies C. Auxin treatment ofpre-veraison grape (Vitis vinifera L.) berries both delays ripening and increasesthe synchronicity of sugar accumulation. Aust J Grape Wine Res. 2011;17:1–8.7. Böttcher C, Burbidge CA, Boss PK, Davies C. Interactions between ethyleneand auxin are crucial to the control of grape (Vitis vinifera L.) berry ripening.BMC Plant Biol. 2013;13:222.8. Chervin C, El-Kereamy A, Roustan JP, Latché A, Lamon J, Bouzayen M.Ethylene seems required for the berry development and ripening in grape,a non-climacteric fruit. Plant Sci. 2004;167:1301–5.9. Sun L, Zhang M, Ren J, Qi J, Zhang G, Leng P. Reciprocity between abscisicacid and ethylene at the onset of berry ripening and after harvest. BMCPlant Biol. 2010;10:257.10. Owen SJ, Lafond MD, Bowen P, Bogdanoff C, Usher K, Abrams SR. Profiles ofabscisic acid and its catabolites in developing Merlot grape (Vitis vinifera)berries. Am J Enol Vitic. 2009;60:277–84.11. Wheeler S, Loveys B, Ford C, Davies C. The relationship between theexpression of abscisic acid biosynthesis genes, accumulation of abscisic acidand the promotion of Vitis vinifera L. berry ripening by abscisic acid. Aust JGrape Wine Res. 2009;15:195–204.12. Deluc LG, Quilici DR, Decendit A, Grimplet J, Wheatley MD, Schlauch KA,Mérillon J-M, Cushman JC, Cramer GR. Water deficit alters differentiallymetabolic pathways affecting important flavor and quality traits in grapeberries of Cabernet Sauvignon and Chardonnay. BMC Genomics.2009;10:212.13. Castellarin SD, Gambetta GA, Wada H, Krasnow MN, Cramer GR, PeterlungerE, Shackel KA, Matthews MA. Characterization of major ripening eventsduring softening in grape: turgor, sugar accumulation, abscisic acidmetabolism, colour development, and their relationship with growth. J ExpBot. 2015;67:709–22.14. Scienza A, Miravalle R, Visai C, Fregoni M. Relationships between seednumber, gibberellin and abscisic acid levels and ripening in CabernetSauvignon grape berries. Vitis 1978;17:361–815. Roby G, Matthews MA. Relative proportions of seed, skin and flesh, in ripeberries from Cabernet Sauvignon grapevines grown in a vineyard eitherwell irrigated or under water deficit. Aust J Grape Wine Res. 2004;10:74–82.16. Friend AP, Trought MCT, Creasy GL. The influence of seed weight on thedevelopment and growth of berries and live green ovaries in vitis vinifera L.cvs. pinot noir and cabernet sauvignon. Aust J Grape Wine Res.2009;15:166–74.17. Gouthu S, Deluc LG. Timing of ripening initiation in grape berries and itsrelationship to seed content and pericarp auxin levels. BMC Plant Biol.2015;15:1–16.18. Kennedy JA, Matthews MA, Waterhouse AL. Effect of maturity and vinewater status on grape skin and wine flavonoids. Am J Enol Vitic.2002;53:268–74.19. Roby G, Harbertson JF, Adams DA, Matthews MA. Berry size and vine waterdeficits as factors in winegrape composition: anthocyanins and tannins.Aust J Grape Wine Res. 2004;10:100–7.20. Holt HE, Francis IL, Field J, Herderich MJ, Iland PG. Relationships betweenberry size, berry phenolic composition and wine quality scores for CabernetSauvignon (Vitis vinifera L.) from different pruning treatments and differentvintages. Aust J Grape Wine Res. 2008;14:191–202.21. Dokoozlian NK, Kliewer WM. Influence of light on grape berry growth andcomposition varies during fruit development. J Am Soc Hortic Sci.1996;121:869–74.22. Walker RR, Blackmore DH, Clingeleffer PR, Kerridge GH, Rühl EH, Nicholas PR.Shiraz berry size in relation to seed number and implications for juice andwine composition. Aust J Grape Wine Res. 2005;11:2–8.23. Šuklje K, Lisjak K, Baša Česnik H, Janeš L, Du Toit W, Coetzee Z, Vanzo A,Deloire A. Classification of grape berries according to diameter and totalsoluble solids to study the effect of light and temperature onmethoxypyrazine, glutathione, and hydroxycinnamate evolution duringripening of sauvignon blanc (Vitis vinifera L.). J Agric Food Chem.2012;60:9454–61.24. Gil M, Pascual O, Gómez-Alonso S, García-Romero E, Hermosín-Gutiérrez I,Zamora F, Canals JM. Influence of berry size on red wine colour andcomposition. Aust J Grape Wine Res. 2015:n/a–n/a25. Ali MB, Howard S, Chen S, Wang Y, Yu O, Kovacs LG, Qiu W. Berry skindevelopment in Norton grape: distinct patterns of transcriptional regulationand flavonoid biosynthesis. BMC Plant Biol. 2011;11:7.26. Fasoli M, Dal Santo S, Zenoni S, Tornielli GB, Farina L, Zamboni A, PorcedduA, Venturini L, Bicego M, Murino V, Ferrarini A, Delledonne M, Pezzotti M.The grapevine expression atlas reveals a deep transcriptome shift drivingthe entire plant into a maturation program. Plant Cell. 2012;24:3489–505.27. Lijavetzky D, Carbonell-Bejerano P, Grimplet J, Bravo G, Flores P, Fenoll J, HellínP, Oliveros JC, Martínez-Zapater JM. Berry flesh and skin ripening features inVitis vinifera as assessed by transcriptional profiling. PLoS One. 2012;7.28. Cramer GR, Ghan R, Schlauch KA, Tillett RL, Heymann H, Ferrarini A,Delledonne M, Zenoni S, Fasoli M, Pezzotti M. Transcriptomic analysis of thelate stages of grapevine (Vitis vinifera cv. Cabernet Sauvignon) berryripening reveals significant induction of ethylene signaling and flavorpathways in the skin. BMC Plant Biol. 2014;14:1–21.29. Grimplet J, Deluc LG, Tillett RL, Wheatley MD, Schlauch KA, Cramer GR,Cushman JC. Tissue-specific mRNA expression profiling in grape berrytissues. BMC Genomics. 2007;8:187.30. Carbonell-Bejerano P, Diago M-P, Martínez-Abaigar J, Martínez-Zapater JM,Tardáguila J, Núñez-Olivera E. Solar ultraviolet radiation is necessary toenhance grapevine fruit ripening transcriptional and phenolic responses.BMC Plant Biol. 2014;14:183.31. Thomas T, Shackel K, Matthews M. Mesocarp cell turgor in Vitis vinifera L.berries throughout development and its relation to firmness, growth, andthe onset of ripening. Planta. 2008;228:1067–76.32. Wada H, Matthews MA, Shackel KA. Seasonal pattern of apoplastic soluteaccumulation and loss of cell turgor during ripening of Vitis vinifera fruitunder field conditions. J Exp Bot. 2009;60:1773–81.33. Jaillon O, Aury J-M, Noel B, Policriti A, Clepet C, Casagrande A, Choisne N,Aubourg S, Vitulo N, Jubin C, Vezzi A, Legeai F, Hugueney P, Dasilva C, HornerD, Mica E, Jublot D, Poulain J, Bruyère C, Billault A, Segurens B, Gouyvenoux M,Ugarte E, Cattonaro F, Anthouard V, Vico V, Del Fabbro C, Alaux M, Di GasperoG, Dumas V, et al. The grapevine genome sequence suggests ancestralhexaploidization in major angiosperm phyla. Nature. 2007;449:463–7.34. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–60.35. Anders S, Pyl PT, Huber W. HTSeq A Python framework to work with high-throughput sequencing data. Bioinformatics. 2014;31:166–9.36. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a bioconductor package fordifferential expression analysis of digital gene expression data.Bioinformatics. 2009;26:139–40.37. Grimplet J, Van Hemert J, Carbonell-Bejerano P, Díaz-Riquelme J, DickersonJ, Fennell A, Pezzotti M, Martínez-Zapater JM. Comparative analysis ofgrapevine whole-genome gene predictions, functional annotation,categorization and integration of the predicted gene sequences. BMC ResNotes. 2012;5:213.38. Lohse M, Nagel A, Herter T, May P, Schroda M, Zrenner R, Tohge T, FernieAR, Stitt M, Usadel B. Mercator: a fast and simple web server for genomescale functional annotation of plant sequence data. Plant Cell Environ.2014;37:1250–8.39. Thimm O, Bläsing O, Gibon Y, Nagel A, Meyer S, Krüger P, Selbig J, MüllerLA, Rhee SY, Stitt M. MAPMAN: a user-driven tool to display genomics datasets onto diagrams of metabolic pathways and other biological processes.Plant J. 2004;37:914–39.40. Vitulo N, Forcato C, Carpinelli EC, Telatin A, Campagna D, D’Angelo M,Zimbello R, Corso M, Vannozzi A, Bonghi C, Lucchin M, Valle G. A deepsurvey of alternative splicing in grape reveals changes in the splicingmachinery related to tissue, stress condition and genotype. BMC Plant Biol.2014;14:99.41. Franco-Zorrilla JM, López-Vidriero I, Carrasco JL, Godoy M, Vera P, Solano R.DNA-binding specificities of plant transcription factors and their potential todefine target genes. Proc Natl Acad Sci U S A. 2014;111:2367–72.Wong et al. BMC Genomics  (2016) 17:416 Page 20 of 2242. Higo K, Ugawa Y, Iwamoto M, Korenaga T. Plant cis-acting regulatory DNAelements (PLACE) database: 1999. Nucleic Acids Res. 1999;27:297–300.43. Ma S, Shah S, Bohnert HJ, Snyder M, Dinesh-Kumar SP. Incorporating motifanalysis into gene co-expression networks reveals novel modular expressionpattern and new signaling pathways. PLoS Genet. 2013;9.44. Sweetman C, Wong DC, Ford CM, Drew DP. Transcriptome analysis at fourdevelopmental stages of grape berry (Vitis vinifera cv. Shiraz) providesinsights into regulated and coordinated gene expression. BMC Genomics.2012;13:691.45. Gambetta GA, Matthews MA, Shaghasi TH, McElrone AJ, Castellarin SD.Sugar and abscisic acid signaling orthologs are activated at the onset ofripening in grape. Planta. 2010;232:219–34.46. Dai Z, Meddar M, Renaud C, Merlin I, Hilbert G, Delro S, Gomès E. Long-termin vitro culture of grape berries and its application to assess the effects ofsugar supply on anthocyanin accumulation. J Exp Bot. 2014;65:4665–77.47. Gouthu S, O’Neil ST, Di Y, Ansarolia M, Megraw M, Deluc LG. A comparativestudy of ripening among berries of the grape cluster reveals an alteredtranscriptional programme and enhanced ripening rate in delayed berries.J Exp Bot. 2014;65:5889–5902.48. Karlova R, Chapman N, David K, Angenent GC, Seymour GB, De Maagd RA.Transcriptional control of fleshy fruit development and ripening. J Exp Bot.2014;65:4527–41.49. Leng P, Yuan B, Guo Y, Chen P. The role of abscisic acid in fruit ripeningand responses to abiotic stress. J Exp Bot. 2014;65:4577–88.50. Sun L, Sun Y, Zhang M, Wang L, Ren J, Cui M, Wang Y, Ji K, Li P, Li Q, ChenP, Dai S, Duan C, Wu Y, Leng P. Suppression of 9-cis-epoxycarotenoiddioxygenase, which encodes a key enzyme in abscisic acid biosynthesis,alters fruit texture in transgenic tomato. Plant Physiol. 2012;158:283–98.51. Böttcher C, Keyzers RA, Boss PK, Davies C. Sequestration of auxin by theindole-3-acetic acid-amido synthetase GH3-1 in grape berry (Vitis vinifera L.)and the proposed role of auxin conjugation during ripening. J Exp Bot.2010;61:3615–25.52. Böttcher C, Boss PK, Davies C. Acyl substrate preferences of an IAA-amidosynthetase account for variations in grape (Vitis vinifera L.) berry ripeningcaused by different auxinic compounds indicating the importance of auxinconjugation in plant development. J Exp Bot. 2011;62:4267–80.53. Böttcher C, Harvey KE, Boss PK, Davies C. Ripening of grape berries can beadvanced or delayed by reagents that either reduce or increase ethylenelevels. Funct Plant Biol. 2013;40:566–81.54. Davies C, Nicholson EL, Böttcher C, Burbidge CA, Bastian SEP, Harvey KE,Huang A-C, Taylor DK, Boss PK. Shiraz wines made from grape berries (Vitisvinifera) delayed in ripening by plant growth regulator treatment haveelevated rotundone concentrations and “pepper” flavor and aroma. J AgricFood Chem. 2015;63:2137–44.55. Nicolas P, Lecourieux D, Kappel C, Cluzet S, Cramer G, Delrot S, LecourieuxF. The basic leucine zipper transcription factor ABSCISIC ACID RESPONSEELEMENT-BINDING FACTOR2 is an important transcriptional regulator ofabscisic Acid-dependent grape berry ripening processes. Plant Physiol.2013;164:365–83.56. Vicente AR, Saladié M, Rose JK, Labavitch JM. The linkage between cell wallmetabolism and fruit softening: looking to the future. J Sci Food Agric.2007;87:1435–48.57. Matthews MA, Cheng G, Weinbaum SA. Changes in water potential anddermal extensibility during grape berry development. J Am Soc Hortic Sci.1987:314–958. Huang XM, Huang HB, Wang HC. Cell walls of loosening skin in post-veraison grape berries lose structural polysaccharides and calcium whileaccumulate structural proteins. Sci Hortic (Amsterdam). 2005;104:249–63.59. Schlosser J, Olsson N, Weis M, Reid K, Peng F, Lund S, Bowen P. Cellularexpansion and gene expression in the developing grape (Vitis vinifera L.).Protoplasma. 2008;232:255–65.60. Powell ALT, Kalamaki MS, Kurien PA, Gurrieri S, Bennett AB. Simultaneoustransgenic suppression of LePG and LeExp1 influences fruit texture andjuice viscosity in a fresh market tomato variety. J Agric Food Chem.2003;51:7450–5.61. Atkinson RG, Sutherland PW, Johnston SL, Gunaseelan K, Hallett IC, Mitra D,Brummell DA, Schröder R, Johnston JW, Schaffer RJ. Down-regulation ofPOLYGALACTURONASE1 alters firmness, tensile strength and water loss inapple (Malus x domestica) fruit. BMC Plant Biol. 2012;12:129.62. Pose S, Paniagua C, Cifuentes M, Blanco-Portales R, Quesada MA, MercadoJA. Insights into the effects of polygalacturonase FaPG1 gene silencing onpectin matrix disassembly, enhanced tissue integrity, and firmness in ripestrawberry fruits. J Exp Bot. 2013;64:3803–15.63. Jiménez-Bermúdez S, Redondo-Nevado J, Muñoz-Blanco J, Caballero JL,López-Aranda JM, Valpuesta V, Pliego-Alfaro F, Quesada MA, Mercado JA.Manipulation of strawberry fruit softening by antisense expression of apectate lyase gene. Plant Physiol. 2002;128:751–9.64. Santiago-Domenech N, Jimenez-Bemudez S, Matas AJ, Rose JKC, Munoz-Blanco J, Mercado JA, Quesada MA. Antisense inhibition of a pectate lyasegene supports a role for pectin depolymerization in strawberry fruitsoftening. J Exp Bot. 2008;59:2769–79.65. Licausi F, Ohme-Takagi M, Perata P. APETALA2/Ethylene Responsive Factor(AP2/ERF) transcription factors: mediators of stress responses anddevelopmental programs. New Phytol. 2013:639–4966. Ishimaru M, Smith DL, Gross KC, Kobayashi S. Expression of three expansingenes during development and maturation of Kyoho grape berries. J PlantPhysiol. 2007;164:1675–82.67. Dal Santo S, Vannozzi A, Tornielli GB, Fasoli M, Venturini L, Pezzotti M,Zenoni S. Genome-wide analysis of the expansin gene superfamily revealsgrapevine-specific structural and functional characteristics. PLoS One.2013;8, e62206.68. Suzuki H, Oshita E, Fujimori N, Nakajima Y, Kawagoe Y, Suzuki S. Grapeexpansins, VvEXPA14 and VvEXPA18 promote cell expansion in transgenicArabidopsis plant. Plant Cell Tissue Organ Cult. 2014;120:1077–85.69. Brummell D, Harpster M, Civello P, Palys J, Bennett A, Dunsmuir P. Modificationof expansin protein abundance in tomato fruit alters softening and cell wallpolymer metabolism during ripening. Plant Cell. 1999;11:2203–16.70. Wong DCJ, Sweetman C, Drew DP, Ford CM. VTCdb: a gene co-expressiondatabase for the crop species Vitis vinifera (grapevine). BMC Genomics.2013;14:882.71. Nicolas P, Lecourieux D, Gomès E, Delrot S, Lecourieux F. The grape berry-specific basic helix-loop-helix transcription factor VvCEB1 affects cell size.J Exp Bot. 2013;64:991–1003.72. Simpson SD, Nakashima K, Narusaka Y, Seki M, Shinozaki K, Yamaguchi-Shinozaki K. Two different novel cis-acting elements of erd1, a clpAhomologous Arabidopsis gene function in induction by dehydration stressand dark-induced senescence. Plant J. 2003;33:259–70.73. Tran LP, Nakashima K, Sakuma Y, Simpson SD, Fujita Y, Maruyama K, FujitaM, Seki M, Shinozaki K, Yamaguchi-Shinozaki K. Isolation and functionalanalysis of Arabidopsis stress-inducible NAC transcription factors that bindto a drought-responsive cis-element in the early responsive to dehydrationstress 1 promoter. Plant Cell. 2004;16:2481–98.74. Tello J, Torres-Pérez R, Grimplet J, Carbonell-Bejerano P, Martínez-ZapaterJM, Ibáñez J. Polymorphisms and minihaplotypes in the VvNAC26 geneassociate with berry size variation in grapevine. BMC Plant Biol. 2015;15:253.75. Vogt T. Phenylpropanoid biosynthesis. Mol Plant. 2010;3:2–20.76. Castellarin SD, Bavaresco L, Falginella L, Gonçalves MIVZ, Di Gaspero G.Phenolics in grape berry and key antioxidants. In: Gerós H, Chaves MM,Delrot S, editors. The biochemistry of the grape berry. Bentham SciencePublishers, Sharjah, U.A.E.2012;89–110.77. Falginella L, Di Gaspero G, Castellarin SD. Expression of flavonoid genes inthe red grape berry of “Alicante Bouschet” varies with the histologicaldistribution of anthocyanins and their chemical composition. Planta.2012;236:1037–51.78. Falginella L, Castellarin SD, Testolin R, Gambetta GA, Morgante M, DiGaspero G. Expansion and subfunctionalisation of flavonoid 3′,5′-hydroxylases in the grapevine lineage. BMC Genomics. 2010;11:562.79. Castellarin SD, Di Gaspero G, Marconi R, Nonis A, Peterlunger E, Paillard S,Adam-Blondon A-F, Testolin R. Colour variation in red grapevines (Vitisvinifera L.): genomic organisation, expression of flavonoid 3′-hydroxylase,flavonoid 3′,5′-hydroxylase genes and related metabolite profiling of redcyanidin-/blue delphinidin-based anthocyanins in berry skin. BMCGenomics. 2006;7:12.80. Downey MO, Harvey JS, Robinson SP. Synthesis of flavonols and expressionof flavonol synthase genes in the developing grape berries of Shiraz andChardonnay (Vitis vinifera L.). Aust J Grape Wine Res. 2003;9:110–21.81. Castellarin SD, Pfeiffer A, Sivilotti P, Degan M, Peterlunger E, Di Gaspero G.Transcriptional regulation of anthocyanin biosynthesis in ripening fruits ofgrapevine under seasonal water deficit. Plant Cell Environ. 2007;30:1381–99.82. Downey MO, Harvey JS, Robinson SP. The effect of bunch shading on berrydevelopment and flavonoid accumulation in Shiraz grapes. Aust J GrapeWine Res. 2004;10:55–73.Wong et al. BMC Genomics  (2016) 17:416 Page 21 of 2283. Matus JT, Loyola R, Vega A, Pena-Neira A, Bordeu E, Arce-Johnson P, AlcaldeJA. Post-veraison sunlight exposure induces MYB-mediated transcriptionalregulation of anthocyanin and flavonol synthesis in berry skins of Vitisvinifera. J Exp Bot. 2009;60:853–67.84. Bogs J, Downey MO, Harvey JS, Ashton AR, Tanner GJ, Robinson SP.Proanthocyanidin synthesis and expression of genes encodingleucoanthocyanidin reductase and anthocyanidin reductase in developinggrape berries and grapevine leaves. Plant Physiol. 2005;139:652–63.85. Ford CM, Boss PK, Hæj PB. Cloning and characterization of Vitis viniferaUDP-glucose. Flavonoid 3- O-Glucosyltransferase, a homologue of theenzyme encoded by the maize Bronze- 1 locus that may primarily serve toglucosylate anthocyanidins in vivo. J Biol Chem. 1998;273:9224–33.86. Kobayashi S, Ishimaru M, Hiraoka K, Honda C. Myb-related genes of theKyoho grape (Vitis labruscana) regulate anthocyanin biosynthesis. Planta.2002;215:924–33.87. Cutanda-Perez M-C, Ageorges A, Gomez C, Vialet S, Terrier N, Romieu C,Torregrosa L. Ectopic expression of VlmybA1 in grapevine activates anarrow set of genes involved in anthocyanin synthesis and transport. PlantMol Biol. 2009;69:633–48.88. Rinaldo A, Cavallini E, Jia Y, Moss SMA, McDavid DAJ, Hooper LC, RobinsonSP, Tornielli GB, Zenoni S, Ford CM, Boss PK, Walker AR. A grapevineanthocyanin acyltransferase, transcriptionally regulated by VvMYBA, canproduce most acylated anthocyanins present in grape skins. Plant Physiol.2015;169(November):01255.2015.89. Gatto P, Vrhovsek U, Muth J, Segala C, Romualdi C, Fontana P, Pruefer D,Stefanini M, Moser C, Mattivi F, Velasco R. Ripening and genotype controlstilbene accumulation in healthy grapes. J Agric Food Chem.2008;56:11773–85.90. Höll J, Vannozzi A, Czemmel S, D’Onofrio C, Walker AR, Rausch T, Lucchin M,Boss PK, Dry IB, Bogs J. The R2R3-MYB transcription factors MYB14 andMYB15 regulate stilbene biosynthesis in Vitis vinifera. Plant Cell.2013;25:4135–49.91. Deluc LG, Decendit A, Papastamoulis Y, Mérillon JM, Cushman JC, CramerGR. Water deficit increases stilbene metabolism in cabernet sauvignonberries. J Agric Food Chem. 2011;59:289–97.92. Vezzulli S, Civardi S, Ferrari F, Bavaresco L. Methyl jasmonate treatment as atrigger of resveratrol synthesis in cultivated grapevine. Am J Enol Vitic.2007;58:530–3.93. Belhadj A, Telef N, Cluzet S, Bouscaut J, Corio-Costet MF, Mérillon JM.Ethephon elicits protection against Erysiphe necator in grapevine. J AgricFood Chem. 2008;56:5781–7.94. D’Onofrio C, Cox A, Davies C, Boss PK. Induction of secondary metabolismin grape cell cultures by jasmonates. Funct Plant Biol. 2009;36:323–38.95. Ebeler SE, Thorngate JH. Wine chemistry and flavor: Looking into the crystalglass. J Agric Food Chem. 2009;57:8098–108.96. Feussner I, Wasternack C. The lipoxygenase pathway. Annu Rev Plant Biol.2002;53:275–97.97. Kalua CM, Boss PK. Evolution of volatile compounds during thedevelopment of cabernet sauvignon grapes (vitis vinifera l.). J Agric FoodChem. 2009;57:3818–30.98. Qin G, Wang Y, Cao B, Wang W, Tian S. Unraveling the regulatory networkof the MADS box transcription factor RIN in fruit ripening. Plant J.2012;70:243–55.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:Wong et al. BMC Genomics  (2016) 17:416 Page 22 of 22


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