Plant materials
Due to tree, fruit and pollen availability, we used different French blood-flesh cultivars for the experimentation: R6041, Sanguine Super Tardive and Montarsa clone S7634, belonging to the same genetic group and all carrying the bf gene. The same way, different non blood-flesh cultivars were used as control depending on the analyses: Lovell (the peach sequence reference), S6146 (Pamirskji), S2678 (Weeping Flower Peach), O’Henry and SF11332 (Fig. S1).
The description over time of the bf phenotype and the biochemical anthocyanin measurements by High-Performance Liquid Chromatography (HPLC) were carried out using the cultivar R6041, a blood-flesh nectarine of French type. It inherited the blood-flesh character of Sanguine Pilat cultivar thanks to the self-fertilization of hybrid N°3 resulting from the cross between Sanguine Pilat and Snowqueen.
An F2 peach progeny of 155 individuals, designated ‘173²’, was first developed and used to map the bf locus. It was then extended to 1600 individuals for fine mapping. 173² comes from the self-pollination of a single F1 tree (n° 173) resulting from a cross between the non blood-flesh cultivar O'Henry and the blood-flesh cultivar Sanguine Super Tardive [21]. Sanguine Super Tardive is an old French blood-flesh peach cultivar (Elias, 1987) homozygous for bf trait (bf/bf), O'Henry is a high-quality, yellow, fondant, freestone peach released in 1968 by Grant Merrill (California, USA) from an open pollination of Merrill Bonanza. Trees of 173² were grown in greenhouses and then in orchards at INRAE in the Unit for Genetics and Improvement of Fruit and Vegetables in Avignon (France).
Another blood-flesh nectarine of French type, NECTAVIGNE® Montarsa clone S7634 (http://catalogue.starfruits-diffusion.com/variete/nectavignemontarsa/) (bf/bf), was used for the genomic and expression studies. It was compared to the selection SF11332 from a French breeding program, a white-flesh nectarine present on the same orchard and with same tree age and same maturation period.
Plant phenotyping
In order to describe the appearance of the coloration in the fruit, we collected fruits from R6041 in orchards at 7 stages from fruit thinning to ripening. Three batches of samples from 5 to 10 fruits each were peeled, and the flesh were cut in pieces, frozen in liquid nitrogen and stored at -80°C for further anthocyanin analyses. Midrib samples were also stored the same way.
The 173² progeny was characterized by phenotyping the color of stem, leaf midrib and fruit according to the scoring system defined by Chaparro et al [18].
In addition, to measure the proportion of red pigmentation in the mesocarp, we collected fruits from 76 individuals of the 173² progeny, at maturity stage. Six to 12 fruits were cut into two parts perpendicularly to the suture (the stone cut in half or removed according to the stage) (Fig. S2). Both sections were scanned for further analysis using Image Jsoftware (available at http://rsb.info.nih.gov/ij/, developed by Wayne Rasband, National Institutes of Health, Bethesda, MD). Along the radius of the fruit, the total thickness of flesh and blood-flesh thickness were measured to estimate the proportion of blood-flesh (blood-flesh thickness / total flesh thickness). An average ratio per individual was calculated from 6 to 12 measurements. In parallel, three samples of mixed flesh from three fruit at maturity from 70 individuals over the 76 were frozen for further determination of anthocyanin content.
Anthocyanin determination
Frozen samples of fruit flesh and midrib of R6041 were ground to obtain a homogeneous powder in liquid nitrogen which was then freeze-dried. For anthocyanin extraction, 1200 µL of methanol acidified with 1% acetic acid (v/v) were added in a tube containing approximately 100 mg, but precisely weighed, of freeze-dried flesh powder. After homogenization, the tube was sonicated for 15 minutes at low temperature (in the presence of crushed ice) to limit temperature increase that would degrade the polyphenols. The resulting methanolic extract was then filtered (0.45 µm PTFE filter) and 20 µl were then injected in high-performance liquid chromatography (HPLC). The HPLC-diode array detection (DAD) analyses were performed using an Agilent 1050 (Palo Alto, CA, USA). Chromatographic separation was carried out using a (250 × 4 mm i.d.) Licrospher PR-18 5 µm column (Merck, Darmstadt, Germany) with a guard column (Licrospher PR-18 5 µm column, Merck, Darmstadt, Germany), both thermostated at 30°C. The mobile phase consisted of water/formic acid (98/2, v/v) (eluent A) and acetonitrile (eluent B) at a flow rate of 1 ml min− 1. The elution program was as follows: 3–9% B (0–5 min), 9–16% B (5–15 min), 16–50% B (15–45 min), 50–90% B (45–48 min), 90–90% B (48–52 min), 90–3% B (52–55 min), 3–3% B (55–60 min). Anthocyanin compounds were quantified at 520 nm by comparisons with external standards.
The evaluation of anthocyanin content of fruit at maturity from the 173² progeny was performed by colorimetric assay, using a spectrophotometer (700/535 nm). An average content was calculated from the 3 samples of mixed flesh frozen for each of the 70 individuals.
DNA isolation and development of markers (SSR, SNP-KASP™) for mapping in 173² progeny
DNA from young leaves of the parents and each individual of the mapping population was extracted using the DNeasy 96 Plant kit (Qiagen, MD, USA), according to the manufacturer’s instructions. DNA quality and quantity were evaluated in 1.5% agarose gel electrophoresis and by spectrophotometer (Thermo Scientific NanoDrop™) before DNA normalization to a concentration of 10 ng/ml. DNA for genomic sequencing (SANGER and NGS) from blood-flesh and non-blood-flesh accessions was extracted as previously described by Bernatzky and Tanksley [22] to obtain a sufficient amount of DNA. DNA concentrations were quantified by Qubit microfluorometry using the Molecular Probes QuantiT™ PicoGreen® dsDNA Assay (Life Technologies, Frederick, MD, USA, Cat no. P1149).
SSR primer pairs used in this study were the same as those used by Lambert and Pascal [23] except new SSR developed from Prunus persicav1.0 genome sequence (http://www.rosaceae.org/species/prunus_persica/
genome_v1.0) using MAPMAKER/EXP V3.0 software ([24]). A total of 84 SSRs were screened with the protocol from Zhijun Shen [17] on 155 individuals from the 1732 progeny.
Development of markers (SSR, SNP-KASP™) for fine-mapping in 173² progeny
For fine mapping of the bf locus, KASP™ (allele-specific competitive PCR) markers were designed based on the SNPs of the peach SNP 9K array, 40K SNP array and the genomic sequence of nectarine clone S7634 of blood-flesh NECTAVIGNE® Montarsa (http://catalogue.starfruits-diffusion.com/
variete/nectavignemontarsa/). The primer sequences were sent to MWG-Biotech (www.mwg-biotech.com/) which synthesized the corresponding KASP™ markers. The reaction mixture for each marker was prepared according to the protocol described by LGC Genomics Ltd (www.lgcgenomics.com/genotyping/kasp-genotypingreagents/). After validation using the DNA of S7634 (blood-flesh cultivar), S2678 and S6146 (non-blood-flesh cultivars) clones, fluorescence genotyping of 237 individuals from the 1732 progeny was carried out using the multilabel reader 2103 EnVision® (Perkin Elmer Company, USA). For the KASP™ test, the reaction mixture was prepared according to the protocol described by LGC Genomics (http://www.lgcgenomics.com/genotyping/kasp-genotypingreagents/). The following cycling conditions were used: 15 min at 94°C, 10 touchdown cycles of 20 s at 94°C and 1 min at 65 − 57°C, 23 cycles of 20 s at 94°C and 1 min at 57°C Fluorescent endpoint genotyping was performed using the 2103 EnVision® Multilabel Reader (Perkin Elmer Company, USA). A total of 24 SNP markers (Table_S1) were screened on 153 individuals, 7 SNP markers on 666 additional individuals and 2 SNP markers (SNP_IGA_386862 and SNP_IGA_386961) on 960 extra individuals to search for recombinants and reduce the interval.
Map construction and genetic mapping in 173² progeny
The first subset of 155 individuals from the 1732 progeny was used to develop a genetic map. Linkage groups (LG) were initially established using a critical log likelihood cut-off (LOD) of 8.0 and a recombination fraction of 0.30. The distances between the markers were calculated using the Kosambi mapping function [25]. After mapping, Mapmaker's “error detection” option was used to clean the dataset. Map figures were obtained using MapChart software [26].
For 76 individuals from the 1732 progeny for which data were available for four traits, a QTL detection was performed using Rqtl package (R Development Core Team 2006, ‘qtl’ library). For the internode length (displaying a normal distribution) the genome scan was performed via Haley-Knott regression and for the presence/absence of coloration, a model for binary trait was used. Since the two other traits, proportion of blood-flesh and anthocyanin content, did not follow a normal distribution, a nonparametric test was used based on the method described by Kruglyak and Lander[27].
RNA Isolation, cDNA library preparation and Illumina sequencing
Fruit from the two cultivars S7634 and SF11332 were collected at four stages of fruit development: i) small fruit stage at 60 days after bloom (DAB), ii) unripe fruit stage at 80 DAB, just before apparition of blood coloration, iii) unripe fruit stage at 100 DAB, after the peak of cyanidin accumulation, iv) ripe fruit stage at 120 DAB. Four biological replicates (pooled flesh from different fruits) for each cultivar and stage (32 samples) were frozen and used for total RNA extraction using Fruit-Mape™ for RNA Purification (Takara, Japan) and NucleoSpin®RNA Plant (Macherey-Nagel) protocols. The yield and purity of RNA were checked by Agilent RNA 6000 Nano kit and by agarose gel electrophoresis. The RNA samples with RNA integrity number (RIN) greater than 8.5 were sent to sequencing at Get-PlaGe platform in Toulouse, France ([email protected]).
For cDNA library preparation, one µg of total RNA from each pool were used to isolate poly(A) mRNA. The poly(A) mRNA was used to prepare a non-directional Illumina RNA-Seq library with a Tru-Seq RNA Kit according to the manufacturer’s instructions (Illumina, CA, USA). One library using MiSeq platform (Illumina) in paired-end (2x150pb) was sequenced.
RNA was prepared in the same way for the Real-time quantitative PCR analysis. The two same cultivars were studied but samples were collected here at seven stages (60 DAB, 75 DAB, 95 DAB, 115 DAB, 135 DAB, 155 DAB and 170 DAB).
Transcriptome analysis, mapping and differentially expressed genes analysis
RNA sequencing resulted in a total of 170Gb sequencing data. The quality control of raw sequencing libraries obtained was analyzed with FastQC version V0.11.8 [28] (https://www.bioinformatics.babraham.ac.uk/projects/
fastqc), program integrated in a pipeline described below. The reads were trimmed for base quality considering that Q ≤ 20 and reads with lengths ≤ 15 pb were removed. The pipeline RNA Seq snakemake was used to analyze the raw data (https://forgemia.inra.fr/gafl/pipelines_snakemake/rnaseq/). Reads were mapped against the Prunus persica reference genome sequence (Prunus persica Whole Genome Assembly v2.0.a01 [29]) from the Genome Database for Rosaceae (GDR, http://www.rosaceae.org/peach/genome) using STAR (star_v2.7.3) and samtools (samtools_v1.9). STAR was used once to perform indexation on all genes (including those non annotated) and a second time to map the reads on the Reference genome with the new indexation.
Transcript abundance and differential gene expression were calculated with the DiCoExpress R script ([30], https://forgemia.inra.fr/Gnet/dicoexpress). This script used different R packages to evaluate the quality of the RNA seq data, filter and normalize the raw data, define the statistical model generating possible contrasts and perform the differential expression analysis using linear models and their generated extensions for negative distributions (GLM) for all contrasts of interest. A list of Differential Genes Expression (DGEs) was generated and an analysis of co-expression was performed using Gaussian mixture models. A GO enrichment analysis was performed, with successfully annotated genes classified according to the next level of the three main GO categories, namely biological process, cellular component and molecular function with the agriGOv2 program (http://systemsbiology.cau.edu.cn/agriGOv2/). Mercator4v.2 coupled with Mapman software was used to combine annotation and gene expression data on metabolic pathway patterns or other processes (https://plabipd.de/portal/mer, https://mapman.gabipd.org/tutorial).
Genomic sequencing, polymorphism study and Deletion KASP™ marker development
The sequencing of the genomes of French blood-flesh S7634 and the two non blood-flesh clones S6146 and S2678 (Pamirskji and Weeping Flower Peach) was performed by the MGX-Montpellier Genomix (c/o Institut de Génomique Fonctionnelle, Montpellier, France). Preparation of gDNA libraries was made with the NextEra DNA sample preparation kit of Illumina following the supplier’s instruction. Libraries were validated by a quantification of the DNA chip Agilent DNA 1000, as well as by qPCR (Applied Biosystems 7500 with SYBR Green chemistry). Sequencing paired-end 150 was performed using an Illumina HiSeq 2500 thanks to the technical SBS (Sequence By Synthesis). A depth of 42X was obtained. The raw reads were cleaned by removing adaptor sequences and low-quality reads and the quality of the sample was analyzed with FastQC software. AMAPS filter ≥ 3 was used. Clean reads were aligned to the peach reference genome database (http://www.rosaceae.org/species/prunus_persica/
genome_v2.0), using the Integrative Genomics Viewer, IGV ([31], https://igv.org).
To confirm the polymorphism observed in S7634 cultivar, peach genomic DNA from blood-flesh and non blood-flesh cultivars were amplified by GoTaq® DNA Polymerase protocol PCR using PCR primers (Table_S1) located in the PpPSAK gene. All PCR amplicons were checked on 1.5% agarose gel in TBE buffer with a band visualization with a standard ethidium bromide.
A deletion KASP™ marker was developed to verify that the blood-flesh trait was ligated to the deletion on PpPSAK gene. The deletion KASP™ marker was designed by targeting a 7 amino acid deletion in exon 3 of Prupe.4G088200 (4329424.4330281) found in the blood-flesh cultivars. The marker was validated in different genetic backgrounds, including the collections of French blood-flesh peaches and blood-flesh peaches from Italy and China, representing 1945 individuals (1455 descendants of the extended F2 173² plus 405 accessions from the INRAE peach collection, old French cultivars from the Rhône Alpes region, American-Canadian, Italian, Hungarian, German (Lavalady) and Chinese blood-fleshed cultivars presented in Table S2.
Gene expression analysis at bf locus via cDNA synthesis and Real-time quantitative PCR
Each of the three biological replicates (pooled flesh from different fruit) for both cultivars remind which ones and the 7 stages (60 DAB, 75 DAB, 95 DAB, 115 DAB, 135 DAB, 155 DAB and 170 DAB) was measured three times (three technical replicates).
cDNA was synthesized from 1µg of total RNA using the Affinity Script Multiple Temperature cDNA Synthesis Kit (Agilent) with 2.5µM oligonucleotide dT primer and 5µM nucleotide reverse specific, following the manufacturer’s instructions. All cDNA were stored at -20°C before being used as template in qRT-PCR analysis. The expression analysis was performed by quantitative real-time PCR (qRT-PCR) using a ‘Stratagen MxPro 3005’ and the Brilliant III SYBR Green QPCR Master Mix with High ROX (Agilent Technologies).
The genes of the bf region, the regulatory and structural genes of the anthocyanin biosynthesis pathway in plants, and the genes linked to photosynthesis and anthocyanins were selected for RT-qPCR expression analysis. The primers for 27 genes were designed using GenScript Real-time PCR Primer Design (https://www.genscript.com/tools/real-time-pcr-taqman-primer-design-tool) and their sequence is given in Table_S1. The reaction was set up in a total volume of 15µl, following the instructions of the Brilliant III SYBR Green QPCR Master Mix kit with High ROX.
The relative expression levels of each gene were calculated using the 2-ΔΔCT method. All RT-qPCR reactions were normalized using the Ct value corresponding to the Peach Translation elongation factor 2 (peach TEF2), Initiation Factor EiF4-Gamma (peach EiF4γ ) and Peach 60S ribosomal protein L13 (peach RPL13) genes. Expression analyses were performed in accordance with MIQE standards [32]. An R script (adapted from [33]) was used to calculate the normalized expression of the 27 genes studied. Analysis of variance (ANOVA) was applied to transcript variations.
Yeast Two-Hybrid assays
Yeast two-hybrid analysis was performed using Matchmaker Gold Yeast One‐Hybrid Library Screening System (Clontech, http://www.clontech.com/). The full-length coding sequence of 5 genes were amplified, then cloned by GATEWAY Cloning Technology and inserted into vectors pGBKT7 and pGADT7 for all constructions. The interactions were tested on the two-way interaction. The fusion construct with pGBKT7 were introduced into yeast strain Y187 and the fusion construct with pGADT7 were introduced into yeast strain Y2HGOLD according the manufacturer’s instructions (Yeastmaker®, Clontech). The transformed yeast cells were grown on the SD–Leu-Trp plates to check transactivation activity. After mix of the strains and mating, yeast was plated onto SD-Leu-Trp plates. Positives colonies were validated by PCR amplification and then streaked on new SD-Leu-Trp and SD-Leu-Trp-His plates. Finally, positive colonies were transferred onto Quadruple Drop-out (QDO) medium (YPD media lacking leucine, tryptophan, histidine and adenine) plates to validate the positive interaction between the two tested genes. Photographs were taken 2–3 days after inoculation on media.