In the present study, we used the peanut genotypes Jihuatian 1 and PI478819. The high-sucrose variety Jihuatian 1 is a Spanish-type cultivar developed by the Hebei Academy of Agriculture and Forestry Sciences, China, whereas the low-sucrose line PI478819 is a Virginia-type variety introduced from the United States. These germplasms were used as the female and male parents, respectively, which were crossed to obtain an F₁ population. KASP molecular marker technique was used to identify true and false hybrids of F₁ seeds. The KASP markers with obvious differences between parents were designed, and the genomic DNA of parents and F₁ seeds were extracted and detected by KASP molecular markers. The homozygous type with the same genotype as the parent was false hybrid, and the heterozygous genotype was expressed as true hybrid (Qin et al., 2020). A subsequent F₂ segregating population consisting of 831 lines was obtained by selfing. In addition, a population of recombinant inbred lines (RILs) was obtained based on single-seed descent. F₂ population and parental individuals were planted on the experimental farm of Henan Academy of Agricultural Sciences in Xinxiang (Henan province) in May 2017, and a total of 251 lines of the RIL population were planted in Xinxiang (Henan), Kaifeng (Henan), and Zhumadian (Henan) in May 2021. Seeds of both the F₂ and RIL populations were planted individually in holes. For the RIL population, RILs were planted in a randomized complete block design with two replications. Each RIL comprising 10 plants in one replicate was planted in a single row with inter-plant spacing of 0.2 m and inter-row spacing of 0.5 m in each of the testing environments. Crop management was conducted following regular agricultural practices (Zhang et al., 2022).

Mature pods harvested from the experimental plants were naturally sun-dried, and the sucrose contents of peanut kernels were measured using a near infra-red (NIR) spectrometer (DA7200; Perten). For measurement, we selected three replicates of approximately 20 uniform kernels, which were evenly packed into a sample cup, and NIR spectral information was collected in the wavelength range 950–1,650 nm (Qin et al., 2016).

Young leaves were collected from all F₂ population lines and the parent plants, from which genomic DNAs were extracted using a plant genomic DNA extraction kit (DP305-03; TianGen), followed by determination of DNA quality. On the basis of the determined sucrose contents of F₂ individuals, we selected 20 plants with extremely high sucrose content and 20 with low sucrose content, the respective DNAs of which were mixed in equal quantities to give two extreme phenotypic mixed pools. DNAs from these two mixed pools and both parents were then subjected to whole-genome resequencing using the Illumina HiSeq/DNBSEQ platform in conjunction with a double-terminal 150-bp sequencing strategy. The sequencing depth was 20×, and the reference genome used was the Tifrunner_V20190521 version of cultivated peanut (https://www.peanutbase.org/).

For the detection of SNP and insertion/deletion (InDel) variants, we used GATK software (McKenna et al., 2010), and SnpEff software (Cingolani et al., 2012) was used to perform variant annotation and predict variant impact. In order to obtain high-quality SNPs for association analysis, the SNPs were initially filtered by removing SNP loci with multiple genotypes, and then those loci with read support values of less than 4. Parent SNP information was then used filter out those sites with different phenotypes derived from the same parent. The SNPs that remained were deemed credible.

SNP-index is a marker association analysis method based on differences in genotype frequencies of mixed pools (Fekih et al., 2013). The main purpose of this method is to detect significant differences in the genotype frequencies of mixed pools using Δ (SNP-index) statistics (Fekih et al., 2013). The stronger the correlation between marker SNPs and traits, the closer Δ (SNP-index) is to 1. The Euclidean distance (ED) algorithm is a method where by sequencing data is used to detect significant differences between the markers of mixed pools and to evaluate the intervals associated with traits (Hill et al., 2013). In the context of the present study, with the exception of differences in sucrose content-related sites, other sites of the two mixed pools constructed using the BSA should be relatively consistent, and consequently, the ED values of non-target sites should be approximately 0. In contrast, the higher the ED value, the greater is the difference of the marker between the two mixed pools. In this study, we used G statistics for the purpose of gene detection. The G value of each SNP is calculated according to the allele sequencing depth, and is weighted according to the physical distance of the adjacent SNP (Magwene et al., 2011). In addition, given that G values are close to the lognormal distribution, the non-parametric estimation of the zero distribution of G values can be used to estimate the p-value of each SNP (Mansfeld and Grumet, 2018). When the values of P and the false discovery rate are both less than 0.01, it is considered that this interval may be the main effect area affecting the trait of interest (i.e., sucrose content in the present study).

Associations among the Δ (SNP-index), ED, G statistics, and p values of the SNP loci were analyzed using the website https://github.com/xiekunwhy/bsa. The four methods used are all based on a 2-Mb sliding window with a step size of 10 kb, which is applied to calculate the average and smooth the map. A 99% confidence level was selected as the threshold for screening, and the window above the confidence level was defined as the area associated with sucrose content. The intervals obtained using the four correlation analysis methods were compared, and the overlapping interval was regarded as the QTL interval associated sucrose content. The genes and polymorphic sites in the candidate interval were annotated using the website https://www.peanutbase.org/.

Young leaves were collected from RIL population plants, and genomic DNAs were extracted using a plant genomic DNA extraction kit (DP305-03, TianGen). On the basis of the differential SNP information obtained for the two parents Jihuatian 1 and PI478819 in the initial mapping interval of the QTL, we designed 23 pairs of KASP primers using Primer Premier 5.0. FAM or HEX fluorescent splice sequences were attached to the 5′ ends of the primers and synthesized by the LGC Genomics company. The PCR reaction mixtures used contained the following: 1 μl of template DNA at a concentration of 50–100 ng/μl and 1 μl of a mixture of 1× Master Mix and Primer Mix. We performed LGC water bath PCR amplification, using the following amplification program: pre-denaturation at 94°C for 15 min; 10 cycles of denaturation at 94 °C and extension at 55°C–61°C for 1 min; 26 cycles of denaturation at 94°C and extension at 55°C for 1 min; and preservation at 10 °C. After the reactions were completed, the genotypes of each site were determined using the SNPline genotyping platform (Majeed et al., 2018).

QTLs for the sucrose contents in plants cultivated in each environment and at different stages of growth were detected based on the replication mean using QTL IciMapping (Li et al., 2007; Meng et al., 2015), setting the mapping step size as 1 cM and the logarithm of odds (LOD) threshold as 3.0. The QTL region of LG06 was drawn using MapChart 2.3 (Voorrips, 2002). QTLs were designated as follows: q+ the abbreviated trait name + linkage group number, or named as q+ the abbreviated trait name + linkage group number + a number designating one of multiple QTLs in a single linkage group, following the International Rules of Genetic Nomenclature (Liu et al., 2008).

On the basis of BSA-seq analysis and fine mapping combined with gene annotation information, we performed a preliminary determination of candidate genes. Following a previously described procedure (Pattee et al., 1974), kernel tissue were collected from both parents at 20, 35, 50, and 60 days after flowering (stages S1–S4), with three biological replicates for each period. S1 is the early development stage, S2 and S3 are the developing stages, and S4 is the seed maturity stage. Total RNA was extracted from the collected tissues using a RNAprep Pure Plant Plus Kit (DP441, TIANGEN) and the concentration and purity of the extracted RNA were examined. High-quality RNA samples were selected based on the obtained purity values and concentration values were used to determine the amount of RNA template. The isolated RNA was subsequently reversed transcribed to cDNA using a FastKing RT Kit (With gDNase) (KR116, TIANGEN), and the cDNA thus obtained was diluted with sterile double-distilled water. qPCR reaction systems were prepared according to the requirements of a PowerUp SYBR Green Master Mix kit. A Quant Studio 5 real-time quantitative PCR instrument was used to run the reactions, and the 2^−ΔΔCT method was used to determine gene expression levels (Livak and Schmittgen, 2001). For each sample, we assessed three biological replicates, for each of which, we also analyzed three technical replicates. The relative expression of candidate genes at the different developmental stages of Jihuatian 1 and PI4788 was determined based on normalization analysis of the gene expression data, using the ADH3 gene as an internal reference gene (Brand and Hovav, 2010). The cDNA sequences of candidate genes and ADH3 were downloaded from the Peanutbase website (https://www.peanutbase.org/), and corresponding primers were designed using Primer Premier 5.0.

NIR spectrometric analysis indicated that the sucrose contents of the female parent Jihuatian 1 and male parent PI478819 were 8.96% and 3.70%, respectively. For the sucrose content of the F₂ population, we obtained maximum and minimum values of 11.03% and 2.52%, respectively, with a coefficient of variation of 0.28%, (Supplementary Table S1). The sucrose content per plant in the F₂ population showed continuous variation and an approximate normal distribution, which is typical of a quantitative character (Figure 1A). The sucrose content of the RIL population was measured in three environments, with mean values of 5.89%, 5.64%, and 5.91% and coefficient of variation ranging from 0.29% to 032% being obtained (Table 1). In each of the three growth environments, we detected a continuous frequency distribution of sucrose content in the RIL population, indicating that the population may contain multiple major genes or QTLs associated with the control of sucrose content (Figures 1B–D). ANOVA revealed that sucrose content is influenced by genotype, the environment, and genotype–environment interactions (Table 2).

Using the measured phenotype data, individuals with extreme phenotypes were used to form two mixed pools (Supplementary Table S1). The original data obtained from whole-genome re-sequencing of the two mixed pools and two parents were filtered to obtain a total of 336.63 Gbp clean reads, with a Q30 value ≥84.09%, GC content ranging from 36.66% to 38.26%, and the distribution of insert sizes showing a unimodal normal distribution. The average comparison efficiency between samples and the reference genome was 97.14%, the average sequencing depth was 32.92×, and we obtained 98.72% genome coverage (Table 3). The values of these parameters indicated the sufficiently good quality of sequencing and a high percentage matches with the peanut reference genome, thereby indicating that the obtained sequences could be used for subsequent variant detection and analysis.

Prior to bulked segregant analysis, we filtered out low-quality SNPs, and thereby finally obtained 318,057 high-quality credible SNPs. These high-quality SNPs were subjected to Δ (SNP-index) (Figure 2A), ED (Figure 2B), G-value (Figure 2C), and Fisher’s exact test (Figure 2D) association analyses, and plotted according to the chromosomal distribution of each parameter. Using a 99% confidence level as the screening threshold for associated chromosomal intervals, all four methods identified multiple candidate intervals on multiple chromosomes (Supplementary Table S2). The three overlapping intervals obtained using these four methods were identified as candidate intervals associated with peanut sucrose content (Table 4).

According to the different SNP information of Jihuatian 1 and PI478819 in three overlapping QTL intervals, the KASP primers were designed (Supplementary Table S3). The markers were genotyped in 251 RIL lines grown in three environments and subjected to genetic linkage analysis. The results revealed that the QTL detected in the initial mapping interval on chromosome A03 was not identified in the RIL population, indicating that this locus might be a false positive locus (Table 5). For all three assessed environments, we detected a candidate interval on chromosome A06 with phenotypic variance explained (PVE) and LOD values of 31.95%–41.05% and 28.70–44.84, respectively, which was considered to be a major QTL, which we designated qSUCA06 (Figure 3). The genetic distance of the qSUCA06 interval was 2.01 cM and the physical distance was 0.29 Mb (112367085–112662675 bp) (Table 5).

The total of 23 genes were identified in the qSUCA06 interval (Figure 3, Supplementary Table S4). And then we detected eight genes changed in exon regions by further amplification and identification in this interval. Among which, six and three genes characterized by SNP and InDel differences, respectively, between Jihuatian 1 and PI478819 (Supplementary Table S5). We have made in-depth functional annotation on several databases and identified two genes related to protein synthesis and metabolism, three genes related to signal transduction, two genes related to cell cycle, and one gene encoding transcription factor through the analysis of the biological process of gene expression products.

The candidate gene designated Arahy.Y2LWD9, which encodes acyl-CoA-binding domain 3 (ACBD), is a domain of acyl-CoA-binding proteins, a class of lipid transporter family proteins, which may be associated with sucrose. Given the detected correlation between Arahy.Y2LWD9 and sucrose accumulation, we analyzed the levels of Arahy.Y2LWD9 expression in the two parents. On the basis of the cDNA sequences of Arahy.Y2LWD9, we designed primers (Table 6) and performed qRT-PCR analyses of candidate genes, using ADH3 as the internal reference control (Brand and Hovav, 2010). The results showed that whereas there were no significant difference between two parents at the S1 stage of seed development with respect to the relative expression of Arahy.Y2LWD9, we detected significant differences in expression at stages S2, S3, and S4 (Figure 4). Overall, the expression of Arahy.Y2LWD9 in the two parents showed an upward trend, which was opposite to the observed accumulation of sucrose, thereby tending to indicate this gene may play a negative regulatory role in the accumulation of sucrose in peanut (Li et al., 2021).