Abstract Background Early ripening is an important desirable attribute for fruit crops. ‘Kyoho’ is a popular table grape cultivar in many Asian countries. ‘Fengzao’ is a bud mutant of ‘Kyoho’ and ripens nearly 30 days earlier than ‘Kyoho’. To identify genes controlling early fruit development and ripening in ‘Fengzao’, RNA-Seq profiles of the two cultivars were compared at 8 different berry developmental stages in both berry peel and flesh tissues. Methods RNA-Seq profiling of berry development between ‘Kyoho’ and ‘Fenzhao’ were obtained using the Illumina HiSeq system and analyzed using various statistical methods. Expression patterns of several selected genes were validated using qRT-PCR. Results About 447 millions of RNA-Seq sequences were generated from 40 RNA libraries covering various different berry developmental stages of ‘Fengzao’ and ‘Kyoho’. These sequences were mapped to 23,178 and 22,982 genes in the flesh and peel tissues, respectively. While most genes in ‘Fengzao’ and ‘Kyoho’ shared similar expression patterns over different berry developmental stages, there were many genes whose expression were detected only in ‘Fengzao’ or ‘Kyoho’. We observed 10 genes in flesh tissue and 22 genes in peel tissue were differentially expressed at FDR ≤ 0.05 when the mean expression of ‘Fengzao’ and ‘Kyoho’ were compared. The most noticeable one was VIT_214s0030g00950 (a superoxide dismutase gene). This ROS related gene showed lower expression levels in ‘Fengzao’ than ‘Kyoho’ in both peel and flesh tissues across various berry developmental stages with the only exception at véraison. VIT_200s0238g00060 (TMV resistance protein n-like) and VIT_213s0067g01100 (disease resistance protein at3g14460-like) were the two other noticeable genes which were found differentially expressed between the two cultivars in both peel and flesh tissues. GO functional category and KEGG enrichment analysis of DEGs indicated that gene activities related to stress and ROS were altered between the two cultivars in both flesh and peel tissues. Several differentially expressed genes of interest were successfully validated using qRT-PCR. Conclusions Comparative profiling analysis revealed a few dozens of genes which were differentially expressed in the developing berries of ‘Kyoho’ and its early ripening mutant ‘Fengzao’. Further analysis of these differentially expressed genes suggested that gene activities related to ROS and pathogenesis were likely involved in contributing to the early ripening in ‘Fengzao’. Electronic supplementary material The online version of this article (doi:10.1186/s12864-016-3051-1) contains supplementary material, which is available to authorized users. Keywords: Grape, Early ripening, RNA-Seq, Bud mutant, Kyoho Background Fruit ripening is a complex and highly coordinated developmental process. A series of changes in physiological and biochemical catabolism are involved during fruit ripening, which in turn affect fruit quality, such as flavor, texture, color and aroma [[35]1]. Fleshy fruits have been divided into climacteric and non-climacteric categories. Grapes are non-climacteric fruit and their berry development follows a double sigmoid pattern, involving two phases of rapid growth separated by a lag phase [[36]2]. The growth and developmental stages of grape berry have been well characterized and described in the modified E-L system [[37]2]. Recently, a number of transcriptional and metabolomic analyses have been carried out to study the grape berry ripening process at various berry developmental stages, among different cultivars, and in distinct environmental conditions [[38]3–[39]6]. These studies have uncovered a wealth of developmentally regulated genes in grape berry [[40]7, [41]8]. However, much remains to be understood about the molecular and biochemical events leading to grape ripening [[42]7]. Most of our knowledge about fruit development and ripening mechanisms come from climacteric fruits [[43]1, [44]7], especially from the characterization of monogenic tomato mutants, including ripening inhibitor (rin), non-ripening (nor), colorless non-ripening(Cnr), green-ripe (Gr), green flesh (gf), high pigment 1 (hp1), high pigment 2 (hp2), and never-ripe (Nr) [[45]1]. In perennial fruit species, it is not very easy to generate and screen such ripening mutants as what have been done in tomato. Nevertheless, mutants affecting fruit development have been reported in several fruit species, including pear [[46]9] and sweet orange [[47]10]. Some studies have been attempted to understand the molecular and biochemical processes of wild types and their mutants in these perennial fruit species. For example, based on a combination of two-dimensional electrophoresis (2-DE) and matrix assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) analysis, Liu et al. [[48]9] detected an increase of proteins related to cell-wall modification, oxidative stress and pentose phosphate metabolism and a decrease of proteins related to photosynthesis and glycolysis during the fruit development process in both ‘Zaosu’ pear and its early-ripening bud sport. However, all these proteins increased or decreased much faster in the early-ripening bud sport than its wild type ‘Zaosu’. Another example is the comparative profiling study of the late-ripening orange mutant of ‘Jincheng’ (C. sinensis L. Osbeck) and its wild type, discovering that differentially expressed unigenes (DEGs) were mainly clustered into five pathways: metabolic pathways, plant-pathogen interaction, spliceosome, biosynthesis of plant hormones and biosynthesis of phenylpropanoids [[49]10]. From these and other work, many genes and pathways have clearly been demonstrated to involve in the fruit ripening processes of perennial species. ‘Kyoho’is a tetraploid interspecific hybrid grape derived from a cross of Vitis vinifera x Vitis labrusca. It is a well-known table grape cultivar widely grown in China, Japan and several other Asian countries. It has large berries and high yield, and can grow well in high temperature, rainy, wet and other adverse environments. However, ‘Kyoho’ is a mid-late ripening grape, which limits it availability for consumers in the early seasons. Recently, we identified an early ripening bud mutation of ‘Kyoho’ and released it as an early table grape, named as ‘Fengzao’ [[50]11]. ‘Fengzao’ ripens in early July in Henan province, China, nearly one month earlier than ‘Kyoho’, while there are no noticeable changes of other fruit and horticultural traits between the two cultivars [[51]12]. As a step to understand the genetic basis of early ripening in ‘Fengzao’, we compared the RNA-Seq profiles of “Kyoho’ and ‘Fengzao’ at 8 different berry developmental stages in both berry peel and flesh tissues. We identified several genes which likely play critical roles in accelerating the berry ripening process in ‘Fengzao’. Materials and methods Plant material and RNA preparation Both ‘Kyoho’ (hereafter WT) and ‘Fengzao’ (hereafter MT, clone number: 200203 F9) vines were cultivated in the same vineyard with the same viticulture management practices on the farm of Henan University of Science & Technology (the county of Yanshi, Luoyang, China (34.41° N, 112.46° E)). The mean annual temperature is 14.2 °C. During the period of early April and late September, the average day length is 13.8 h. Fruit samples from three vines in 2013 were harvested at the developmental stages corresponding to EL 27, 29, 31, 33, 34, 35, 37, and 38 (EL refers to the modified Eichhorn and Lorenz developmental scale as described by Coombe [[52]2]). The characteristics of these developmental stages are as follows: EL 27 at the beginning of berry setting; EL 31 pea-size berries; EL 32 beginning of bunch closure, berries touching; EL33 characterized by hard green berries; EL 34 just before véraison characterized by green berries, which are starting to soften; EL35 corresponding to véraison; EL 37 involving sugar and anthocyanins accumulation, and active growth; and EL 38 corresponding to harvest time [[53]2]. We collected berry samples from both MT and WT on the same dates when MT or WT reached its eight berry developmental stages. Because MT rapidly accelerated its ripening process after E-L 31, the time (days) required to reach to the E-L stages 33, 34, 35, 37 and 38 in MT were different from that in WT (Fig. [54]1). As a result, we have 20 berry bulk samples taken. These 20 bulk samples included 9 (F1-F9) from MT and 11 (K1-K11) from WT. Among the 9 MT berry samples, F1, F2, F3, F4, F6, F7, F8 and F9 each represented one of the 8 E-L berry developmental stages of MT (Fig. [55]1). F5 did not correspond to any of the 8 E-L stages in MT. It was taken at the same date (6/21) when WT reached its E-L stage 33 and the K5 sample of WT was harvested. Similarly, among the 11 WT berry samples, K1, K2, K3, K5, K8, K9, K10 and K11 each represented one of the 8 E-L berry developmental stages of WT (Fig. [56]1). The other 3 bulk samples, K4 (6/17), K6 (6/27) and K7 (7/4), were taken when MT reached its E-L stages 33, 34 and 35, and the F4, F6, and F7 samples of MT were harvested, respectively. Fig. 1. Fig. 1 [57]Open in a new tab Berry developmental stages and sampling time points. The E-L system follows that of Coombe [[58]2]. F1-F9 and K1-K11 represent the sampling time points for MT and WT, respectively. Because WT and MT had different ripening dates, the corresponding dates for reaching their E-L system stages, as indicated in the brackets after F or K, were different. For example, F4 and K5 both represented the samples taken at the E-L system stage 33, but K5 was sampled later (6/21) than F4 (6/17). This is because ‘Fengzao’ reached the E-L system stage 33 about 5 days earlier than ‘Kyoho’ The RNA-Seq libraries from the peel and flesh tissues of WT were labeled as KP and KF, respectively. Similarly, the corresponding libraries of MT were labeled as FP and FF. Ten representative berries were sampled at each developmental stage from 3 individual vines in 2013. Flesh and peel tissues were separated and immediately frozen in liquid nitrogen and stored at -80 °C until used for RNA extraction. Representative berries were similarly sampled and processed in 2014 for qRT-PCR validation of the expression levels of certain genes of interest. Library preparation and sequencing RNA-Seq libraries were constructed according to the protocols of Zhong et al. [[59]13] and Wang et al. [[60]14]. Briefly, 20 μg total RNA was used to enrich mRNA by using the oligo (dT) magnetic beads. After adding first strand synthesis buffer, the mRNA was fragmented by incubation at 94 °C for 5 min. The first strand cDNA was synthesized with random hexamer-primer using the fragmented mRNAs as templates by Superscript®III Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA). cDNAs were purified with Agencourt RNAClean XP beads (Beckman Coulter Genomics, Danvers, MA, USA) followed by end repair and dA-tailing (NEB, Ipswich, MA, USA). The short fragments were then ligated with Y-shaped adapters using high concentrated T4 ligase (Enzymatics, Beverly, MA, USA). The adaptor ligated cDNAs were size selected with Ampure XP beads (Beckman Coulter Genomics, Danvers, MA, USA) before PCR amplification with indexed primers. RNA-Seq libraries were sequenced using the Illumina HiSeq system at the Biotechnology Resource Center of Cornell University (Ithaca, NY, USA). In total, forty libraries were sequenced from fruit peel and flesh of WT and MT sampled at different developmental stages. All sequences were deposited in the Short Read Archive at NCBI under accession number SRR1557134 and SRR1558172. Sequence and expression analysis The raw reads were cleaned by removing adaptor sequences, empty reads and reads with unknown or low-quality bases. The clean reads were aligned to the grape reference genome [[61]15] using TopHat v2.0.9 with default parameters. The coordinates of the mapped putative transcripts were then compared with the current grape genome V2.1 annotation [[62]16] ([63]http://genomes.cribi.unipd.it/). The software of featureCounts was used to obtain raw read counts from the alignments which can be unambiguously assigned to genomic features (exon) for each sample [[64]17]. The R package, DESeq2 [[65]18], was employed to identify differentially expressed genes (parameters: p value ≤ 5 %) based on the read count for each gene at different developmental stages. Cluster analysis and gene annotation Following alignments, raw counts of individual genes were normalized to Reads Per Kilobase of exonmodel per Million mapped reads (RPKM) based on the 12X v2.1 gene annotation [[66]16] using featureCounts [[67]17]. To correct potential scale effect of gene expression and avoid working with negative expression values, we added 1 to the RPKM values of the expressed genes and then transformed the modified PRKM values using 2 as the log base. A mean expression value was calculated across different berry developmental stages for each expressed gene. The mean calculation did not include those data points with log2 value being 0. Those data points which had no detectable expression were replaced with the mean for the purpose of pattern comparison. The deviation from the mean expression was calculated for each expressed gene at individual berry developmental stages. The expression change patterns of individual genes were characterized by K-means clustering in both MT and WT. All statistical analyses were performed in R version 2.15.3. Clustering of transcript expression patterns based on RPKM levels was carried out using the k-means method and with the Euclidean similarity metric. Gene Ontology (GO) categorization was carried out using Blast2GO (version 2.3.5) ([68]http://www.blast2go.de/) with the 12X V2.1 predicted transcripts as references [[69]16]. Then, WEGO software