Integration of genomics, transcriptomics and metabolomics identifies candidate loci underlying fruit weight in loquat

Abstract

Fruit weight is an integral part of fruit-quality traits and directly influences commodity values and economic returns of fruit crops. Despite its importance, the molecular mechanisms underlying fruit weight remain understudied, especially for perennial fruit tree crops such as cultivated loquat (Eriobotrya japonica Lindl.). Auxin is known to regulate fruit development, whereas its role and metabolism in fruit development remain obscure in loquat. In this study, we applied a multi-omics approach, integrating whole-genome resequencing-based quantitative trait locus (QTL) mapping with an F1 population, population genomics analysis using germplasm accessions, transcriptome analysis, and metabolic profiling to identify the genomic regions potentially associated with fruit weight in loquat. We identified three major loci associated with fruit weight, supported by both QTL mapping and comparative genomic analysis between small- and big-fruited loquat cultivars. Comparison between two genotypes with contrasting fruit weight performance through transcriptomic and metabolic profiling revealed an important role of auxin in regulating fruit development, especially at the fruit enlarging stage. The multi-omics approach identified two homologs of ETHYLENE INSENSITIVE 4 (EjEIN4) and TORNADO 1 (EjTRN1) as promising candidates controlling fruit weight. Moreover, three single nucleotide polymorphism (SNP) markers were closely associated with fruit weight. Results from this study provided insights from multiple perspectives into the genetic and metabolic controls of fruit weight in loquat. The candidate genomic regions, genes, and sequence variants will facilitate understanding the molecular basis of fruit weight and lay a foundation for future breeding and manipulation of fruit weight in loquat.

Figure 1

Distribution of markers in the high-density genetic map and quantitative trait loci associated with fruit weight. (A) The X-axis corresponds to each linkage group; the Y-axis represents the genetic distance (cM). (B) The X-axis represents each linkage group; the blue line corresponds to the LOD score on the left Y-axis; the red line corresponds to the phenotypic variation explained (%) on the right Y-axis. The horizontal grey line indicates the threshold of LOD with 99.5% confidence interval (LOD = 6.2).

Figure 2

Genetic relationship and population differentiation of small- and large-fruited cultivars, and comparison of QTL regions and top F_st regions potentially associated with fruit weight. (A) and (B) The black color indicates small-fruited cultivars; the red color indicates large-fruited cultivars. The phylogenetic tree was constructed using PHYLIP with 100 bootstraps and visualized using MEGAX software. Principal component analysis (PCA) was carried out using PLINK. (C) Distribution of F_st comparing small- and large-fruited cultivars. Each dot represents a sliding window of 10 kb, within which the average F_st value was calculated. The blue line is the top 1% F_st line (F_st = 0.31); the brown line is the top 0.1% F_st line (F_st = 0.42). (D) Comparisons for Chr8, Chr12, and Chr15, which contain three major QTLs for fruit weight.

Figure 3

Observation of fruit development and DEGs identified from the RNA-seq experiment. (A) Observations of fruit development of ZP44 and ZP65 at 10 time points, including 0 days past anthesis (0D, S1), 7D (S2), 14D, 28D (S4), 42D, 56D (S6), 63D, 77D (S8), 84D, and 91D. (B) Comparison of fruit diameter (transverse diameter) between ZP44 and ZP65. The values are the mean fruit diameter of 15 fruits. Bars are standard errors. (C) Principal component analysis using transcriptome profiles of all 30 replicates. (D) Summary of DEGs between ZP44 and ZP65 at each stage. (E) Comparison of DEGs at each stage.

Figure 4

K-means clustering of DEGs. The number in black font is the number of genes in each sub-class; the number in red font is the number of DEGs between ZP44 and ZP65 for the “star” labeled time point.

Figure 5

The clustering of auxin metabolites and significantly different metabolites between ZP44 and ZP65. (A) A heat map showing auxin metabolite levels. Data were processed using the z-score normalization method. (B)-(E) Comparisons of metabolite levels for L-tryptophan (TRP), 3-indoleacetonitrile (IAN), N-(3-indolylacetyl)-L-alanine (IAA-Ala), and 2-oxindole-3-acetic acid (OxIAA).

Figure 6

Distribution of QTL regions, top 1% F_st regions, and candidate genes potentially associated with fruit weight in two reference genomes of loquat. The explanations for color codes of regions and genes are at the bottom of the figure. The connecting lines between the two genomes indicate the same chromosomes.

Figure 7

Single nucleotide polymorphism markers closely associated with fruit weight. (A) The distributions of fruit weight for F₁ individuals with predicted genotypes from GATK for Marker 1 and 2. The predicted genotypes of Marker 2 were used for the comparison because there were more available data for the minor class of genotype combinations than for Marker 2′. The genotypes of Dafang were “C/G” at Marker 1 and “A/C” at Marker 2. The genotypes of Ninghaibai were “G/G” at Marker 1 and “A/A” at Marker 2. The number of plants in each column corresponds to the genotype combination in that column. (B) Three genotypes at the four SNP loci based on Sanger sequencing from the validation panel. (C) The genotypes obtained from Sanger sequencing for the validation panel at Markers 1, 2, 2′, and 3.

Figure 8

The differential expression patterns and polymorphisms of EjTRN1 between ZP44 and ZP65. (A) Relative expression levels of EjTRN1 at five developmental stages of ZP44 and ZP65. (B) An insertion/deletion identified at the promoter of EjTRN1 with PCR validation. (C) Comparison of protein sequences of EjTRN1 between ZP44 and ZP65. The functional domains were predicted using the conserved domain search tool at NCBI.