Abstract

   Ripening is one of the key processes associated with the development of
   major organoleptic characteristics of the fruit. This process has been
   extensively characterized in climacteric fruit, in contrast with
   non-climacteric fruit such as grape, where the process is less
   understood. With the aim of studying changes in gene expression during
   ripening of non-climacteric fruit, an Illumina based RNA-Seq
   transcriptome analysis was performed on four developmental stages,
   between veraison and harvest, on table grapes berries cv Thompson
   Seedless. Functional analysis showed a transcriptional increase in
   genes related with degradation processes of chlorophyll, lipids,
   macromolecules recycling and nucleosomes organization; accompanied by a
   decrease in genes related with chloroplasts integrity and amino acid
   synthesis pathways. It was possible to identify several processes
   described during leaf senescence, particularly close to harvest. Before
   this point, the results suggest a high transcriptional activity
   associated with the regulation of gene expression, cytoskeletal
   organization and cell wall metabolism, which can be related to growth
   of berries and firmness loss characteristic to this stage of
   development. This high metabolic activity could be associated with an
   increase in the transcription of genes related with glycolysis and
   respiration, unexpected for a non-climacteric fruit ripening.

Introduction

   Fruit ripening is a complex, finely coordinated process, that involves
   a series of important physiological changes that affect the fruit
   palatability [[42]1,[43]2]. Changes occurring during fruit ripening
   include epidermis discoloration, increase in soluble sugars
   concentration and acidity reduction, pulp softening and aroma
   intensification [[44]3]. In climacteric fruits these changes are
   associated with a characteristic increasing in respiration, accompanied
   by an increase in ethylene production [[45]4]. Given the economic
   relevance of many climacteric fruits, where these changes are
   associated with a characteristic increasing in respiration and ethylene
   production, several studies have been performed in order to
   characterize the metabolic and transcriptional changes associated to
   changes in the perception of fruit sensory attributes [[46]5–[47]7].

   Grapes belong to the group of non-climacteric fruit, commonly
   associated with a lack of a marked increase in respiration or ethylene
   generation during ripening, despite that recent findings pointed to a
   differential ethylene response during grape development, and increased
   expression at veraison of ethylene biosynthesis related genes in
   Thompson Seedless [[48]8–[49]10]. Grape berry development shows a
   double-sigmoidal growth curve defining three growth phases, two
   fast-growing (phase I and phase III), separated by a slower lag-growth
   stage (phase-II) known as veraison [[50]11,[51]12]. During this phase,
   the accumulation of organic acids reaches its maximum level and
   chlorophyll is degraded giving rise to the beginning of color change
   characteristic of each cultivar [[52]13,[53]14]. This process marks the
   beginning of ripening characterized among other by berry softening and
   cell expansion due to the active accumulation of water in the vacuole
   [[54]12].

   Recently there has been an increase in the amount of transcriptome
   studies which have helped to unravel some of the main molecular
   mechanisms involved in the ripening of grape berries [[55]15–[56]18].
   Most of these studies have been conducted in wine grape cultivars and
   focused to answer questions related to biological processes of interest
   to the winery industry. However, quality attributes of table grape
   berries differ from wine grapes in several aspects, some of them in an
   opposite way. One example is the higher firmness in table grape berries
   which contribute to desirable crunchy texture, while in wine grapes is
   correlated with lower phenolic compounds extraction [[57]19,[58]20].
   Therefore, the application of the knowledge derived from wine berries
   into table berries is not straightforward. The present study aims to
   increase the knowledge gap among table and wine berries ripening by
   means of a transcriptome analysis using the white-green table grapes
   cultivar Thompson Seedless.

Material and methods

Plant material and maturity parameters

   Grape bunches of table grapes (Vitis vinifera) cultivar Thompson
   Seedless samples were obtained from a commercial vineyard located in
   Llay-Llay (Valparaiso Region, Chile) during the 2012 season. The grape
   bunches were obtained and evaluated weekly from veraison to harvest.
   The latter was determined based on total soluble solids (18–19% w/w
   sucrose), according to commercial standard for Thompson Seedless
   cultivar. Each sampling was performed collecting five grape bunches
   from different individuals, considering a total of six evaluated points
   ([59]Table 1). Firmness was determined in 25 grape berries, with five
   berries from five bunches picked from five different plants, using a
   Texture Analyser TA-XT plus equipped with the Volodkevich Bite Jaw
   probe (Stable Micro Systems Ltd., Surrey, UK) by penetration test of
   whole grape berry (including skin) Complete force-distance curve was
   obtained at 1 mm s^-1 of penetration to 15 mm and firmness was
   calculated by area under the curve (N ● mm) [[60]21,[61]22].

Table 1. Maturity parameters of Thompson Seedless table grape cultivar.

   TA[62]^a DPA[63]^b Firmness
   (N mm) Diameter
   (mm) Soluble Solids
   (w/w sucrose) Acidity
   (g L^-1)
   A1 43 240 ± 5 a[64]^* 14.3 ± 0.2 c 5.9 ± 0.1 e 1.7 ± 0 a
   A2 52 199 ± 10 b 14.3 ± 0.3 c 9 ± 0.3 d 1.4 ± 0 b
   - 58 117 ± 4 c 16.4 ± 0.3 b 12.1 ± 0.2 c 1.3 ± 0 b
   A3 65 99 ± 3 cd 18.5 ± 0.3 a 13.6 ± 0.2 b 0.8 ± 0 c
   - 72 81 ± 2 d 19.0 ± 0.2 a 16.4 ± 0.2 a 0.6 ± 0 cd
   A4 78 80 ± 2 d 19.0 ± 0.2 a 17.2 ± 0.2 a 0.5 ± 0 d
   [65]Open in a new tab

   ^a Selected point for transcriptomic analysis;

   ^b Days Post Anthesis;

   * Letter represent the results of Tukey test p Value < 0.05.

   The size of the 25 grape berries was determined according to equatorial
   diameter measurement using a caliper expressing the value in
   millimeters. The soluble solids content of the selected berries was
   determined using a temperature-compensated digital refractometer (HI
   96811, Hanna Instruments Inc., Woonsocket USA) expressing the results
   in % w/w of sucrose per 100 g solution. The acidity was determined by
   titration of three-pooled juice of 10 berries per bunch previously
   selected for each sampling point. The titration was performed with 0.1
   N NaOH (pH 8.2) and reported as g L^-1 of tartaric acid.

Total RNA isolation

   Four representative ripening points were selected for transcriptomic
   analysis (TA) from the total of six sampling points in a temporarily
   equidistant manner identified as A1 (veraison), A2, A3 and A4 (harvest)
   ([66]Table 1). Grape berries from five bunches belonging from different
   plants, were frozen in liquid nitrogen and stored at -80°C immediately
   after harvest. Frozen berries were grinded until fine powder using
   liquid nitrogen in a mortar. Total RNA was extracted from entire grape
   berry according to Gudenschwager et al. [[67]23], and concentration was
   determined using spectrophotometric method (Epoch, Biotek, VT, USA).
   For each sampling point it were performed five total RNA extractions.
   Each RNA extraction was composed of 10 selected berries of five bunches
   from five plants. The quality of each RNA extraction was confirmed by
   electrophoresis on 1.2% agarose gel and using capillary electrophoresis
   Fragment Analyzer Automated CE System (Advanced Analytical
   Technologies, Inc., IA, USA). The amount of total RNA was determined by
   fluorometer using Qubit RNA BR assay kit (Invitrogen, CA, USA)
   according to manufacturer indication. Two RNA pools were constructed
   for each sampling point by taking 5 μg of RNA per each extraction. The
   quality and concentration of each RNA pool was confirmed again by
   capillary electrophoresis and Qubit kit before the library
   construction.

RNA-Seq library construction and sequencing

   One microgram of total RNA of each sampled pool was used to isolate
   poly(A) mRNA and to prepare a non-directional Illumina RNA-Seq library
   with a TruSeq RNA sample preparation according to the manufacturer
   instructions (Illumina, Inc., San Diego, CA, USA). Library quality
   control was performed with a Fragment Analyzer Automated CE System
   (Advanced Analytical Technologies, Inc., IA, USA) using the protocol
   indicated by the manufacturer. Paired-end libraries were sequenced
   using HiSeq 2000 platform (Illumina, Inc., San Diego, CA, USA). A total
   of 100 nucleotides paired-ends reads were generated. Two technical
   replicates of each A1, A2, A3 and A4 samples were sequenced.

Sequences processing

   Reads sequences were processed according to Q20 quality criteria,
   trimming reads sequence with Q ≤ 20 and length ≤ 15 nucleotides.
   Paired-end reads that overlapped in at least 10 nucleotides were
   selected. Processed reads were mapped against Vitis vinifera cv
   Thompson Seedless reference genome [[68]24] using Tophat 2.0.10
   [[69]25]. Counting of mapped reads was performed using the HTSeq
   package
   ([70]http://www.huber.embl.de/users/anders/HTSeq/doc/-overview.html)
   [[71]26].

Expression and functional analysis

   The expression of each transcript was determined by calculating
   Fragments Per Million per Kilobase (FPKM) of exon mapped reads
   [[72]27]. FPKM cut-off value above 2, which corresponds to at least 20
   reads matched, was applied in order to consider a gene as expressed and
   be further analyzed. Differential expression analysis was performed at
   all evaluated instances considering A1 as reference condition using the
   Edge R package [[73]26,[74]28]. The genes that showed a significant
   differential expression value in at least one comparison (A2 vs A1; A3
   vs A1 or A4 vs A1) were selected according to Baggerly test results
   (FDR < 0.05) [[75]29]. A false discovery rate threshold (FDR < 0.05)
   was used in order to determine significant differences in gene
   expression. Clustering analysis was performed according to FPKM values
   of each of the genes using K-means method considering the Euclidean
   distance between them and the figure of merits [[76]30,[77]31]. Gene
   Ontology (GO) terms enrichment analysis of each cluster was performed
   using AgriGO [[78]32] and complemented with REVIGO tool in order to
   remove redundant terms based on semantic similarities [[79]33]. For the
   identification of the most significant GO terms in each cluster, the
   “Tree Map” visualization of REVIGO was used, considering the term with
   the most significant p-Value within each sub-set of non-redundant
   related terms [[80]33]. PlantCyc of Plant Metabolic Network database
   ([81]http://pmn.plantcyc.org) was used for metabolic pathway analysis
   [[82]34]. Metabolic Domains enrichment analysis seeks to find if genes
   associated to a given metabolic domain are overrepresented in a cluster
   using a hypergeometric test. Pathway enrichment analysis is also based
   on a hypergeometric test, being constrained to genes associated to
   pathways.

Validation by qRT-PCR

   qRT-PCR was carried out using the Fast EvaGreen qRT-PCR MasterMix
   (Applied Biotium, Hayward, CA, USA) and the Mx3000P Real-Time PCR
   System (Stratagene, La Jolla, USA). Reactions were performed in
   triplicate containing 50 ng of cDNA, 500 nM of primers, 2X
   concentration of Fast EvaGreen qRT-PCR MasterMix and nuclease free
   water with a final volume of 20 μL per reaction. The Ct values for all
   genes were normalized to the Ct value of gene GSVIVG01000037001
   (mevalonate kinase) in each experimental condition, obtaining the best
   results when compared other evaluated references genes (transcriptional