A natural variation contributes to sugar accumulation in fruit during tomato domestication

Tomato is the most widely consumed fruit and vegetable crop in the world, serving as an important source of micronutrients in human diet (Zhu et al., 2018). The impact of sugar on the taste of tomato fruits is generally estimated by determining their total soluble solids (TSS) accumulation (Kader, 2008), therefore, the determination of TSS is responsible for the fruit quality of tomato designed for fresh market. However, domestication has resulted in a decline in fruit taste from wild ancestors to modern tomato cultivars (Tieman et al., 2017). To understand the genetic basis causing this decline, we measured the TSS contents in fruits from a population consisting of 46 wild Solanum pimpinellifolium (SP), 94 semi-domesticated S. lycopersicum var. cerasiforme (SLC) and 148 fully domesticated S. lycopersicum var. lycopersicum (SLL) with large natural variations (Figure S1a; Data set S1). We conducted a genome-wide association study (GWAS) for the TSS using this tomato population with a total of 7 632 172 common SNPs (Zhu et al., 2018). The P-value of 1.31 × 10⁻⁷ was set as the significance threshold after Bonferroni-adjusted correction. Two significant associations with TSS levels were identified on chromosomes 8 and 9 (Figure 1a). An extracellular invertase encoding gene Lin5 (Solyc09g010080) was found from 11 018 to 6892 bp upstream of the leading SNP (SNP_t, P = 1.10 × 10⁻⁷) on chromosome 9 (Figure 1b; Data set S2). The Lin5 facilitates the cleavage of sucrose in apoplast, impacting sugar supply from source organs to fruits in tomato. The variation for Brix9-2-5 of Lin5 resulted in the conversion of asparagine to glutamic acid at position 348 in S. lycopersicum, which was considered to be a major reason for the decrease in enzyme activity and fruit sink strength compared to the green-fruited S. pennellii (Fridman et al., 2004). Sequence analysis revealed the variation of Brix9-2-5 was not involved in the red-fruited tomato population, while another significant variation SNP2458 residing in Lin5 coding region (2458 bp relative to the start codon) was in strong linkage disequilibrium (r² = 0.89) with SNP_t, thus closely associating with TSS levels (Figure 1b; Data set S3). Based on the reference genome, we found that the SNP2458 variation causes a conversion of adenine (A) to guanine (G), resulting in the substitution of asparagine (N) with aspartic acid (D) at position 366 of Lin5 (Figure 1b). All accessions were subsequently classified into two haplotype groups according to the SNP2458 variation. Accessions with alternative Lin5^2458G belong to haplotype 1 (Hap1) group, whereas genotypes with reference Lin5^2458A are representatives of Hap2 group. Statistically, the accessions in Hap1 group exhibited dramatically higher TSS levels in fruits than those in Hap2 group (Figure 1c; Data set S1). Allelic distribution and TSS levels in three groups revealed that sugar accumulation was gradually reduced in fruit during tomato domestication of selection for larger fruits from SP to SLC and then from SLC to SLL (Figures 1d,e and S1b).

Figure 1

Open in figure viewerPowerPoint

A sugar-associated variation during tomato domestication. (a) Manhattan plot of the GWAS. Dashed line indicates the Bonferroni-adjusted significance threshold. SNPs residing within significant associations are marked in red. (b) Physical distance of SNP2458 and SNP_t. (c) TSS levels in fruits of accessions in two haplotype groups. n indicates the number of accessions. (d, e) Distribution of Lin5 alleles (d) and TSS levels (e) in SP, SLC, and SLL. n represents the number of accessions. (f) Schematic representation of prime editors developed in this study. (g) Sequence chromatograms of edited lines at SNP2458 site in TS-21 and AC. Arrows mark the edited bases. (h) Statistics of prime editing efficiency and hereditability at SNP2458 site. (i–k) Phenotype (i), biomass (j), and TSS levels (k) of fruits in homozygous edited plants. n shows fruit number in (j) and biological replicate in (k). Scale bars: 1 cm. Values are means ± SD; statistical significance determined by two-sided t-test. GWAS, genome-wide association study; SNP, single nucleotide polymorphism; SP, Solanum pimpinellifolium; SLC, Solanum lycopersicum var. cerasiforme (SLC); SLL, Solanum lycopersicum var. lycopersicum; TSS, total soluble solids.

Prime editing has been successfully applied to introduce various types of precise base changes in monocots (Hua et al., 2020). However, prime editing technology has been barely employed to achieve precise base substitutions in dicots. To verify the variation role in fruit sugar accumulation, we generated two prime editors by expressing the plant codon-optimized prime editing system in tomato, designed as tomato PE2max (tPE2max) and tPE4max. A composite promoter consisting of polymerase II (CaMV 35S enhancer-CmYLCV) and III (Arabidopsis U6-26) promoter was used to initiate pegRNA transcription while the expression of a Moloney murine leukaemia virus reverse transcriptase (M-MLV RT) fused with the SpCas9 (R221K/N394K/H840A) nickase was driven by two tandem repeat CaMV 35S (2× 35S) promoters (Yourik et al., 2019). An expression cassette SlPMS1dn-amiR-SlMSH2 upstream of the rbcsE9 terminator driven by 2× 35S promoters was specially added to the tPE4max editor to inhibit DNA mismatch repair pathway. Two pegRNAs were designed for the SNP2458 variation to generate the substitution of D366N in wild tomato TS-21 and N366D in cultivated tomato Ailsa Craig (AC), respectively (Figure 1f). The tPE4max exhibited a significantly higher editing efficiency of 12.7% compared to 7.4% of tPE2max in regenerated TS-21 plants, while the efficiency of tPE4max up to 20.3% was similar to the 20.7% of tPE2max in regenerated stable AC plants (Figure 1g,h). Sanger sequencing showed that all six potential off-target loci did not contain mutations in the gene editing lines (Table S1). Inheritability analysis displayed that all offspring of homozygous edited lines maintain the editing base at SNP2458, while descendants of heterozygous edited plants undergo genetic segregation at this locus (Figure 1h), indicating that two prime editors could achieve inheritable precise base substitution in tomato genome but the editing efficiency can be further improved.

The negative correlation between fruit weight and sugar content in tomatoes was considered to be likely associated with the polymorphism of Lin5 gene (Tieman et al., 2017). We thus measured the fresh weight and TSS levels of the fruits in homozygous edited tomato lines. The results revealed TSS levels in fruits decreased in edited TS-21 plants and increased in edited AC plants, while comparable fruit biomass compared with wild-type plants (Figure 1i–k), supporting that the SNP2458 in Lin5 is likely responsible for fruit sugar accumulation rather than size during tomato domestication. Overall, our findings indicate that a natural variation in the coding region of Lin5 contributing to the TSS content decline in the cultivated tomato during domestication. The wild Lin5 variant provides valuable natural resource and genetic marker for improvement in tomato fruit quality. We have successfully overcome the obstacles of employing prime editing in tomato, which will undoubtedly enable a rapid de novo domestication of wild tomato.