Enhanced soluble sugar content in tomato fruit using CRISPR/Cas9-mediated SlINVINH1 and SlVPE5 gene editing

Plant material and growth conditions

We use cultivated tomato M82 as the experimental material. The tomatoes were grown in a greenhouse at the seedling stage (environmental conditions: 14 h light/10 h dark, daytime temperature of 25 °C, night temperature of 18 °C, and daily automatic irrigation by sprinkler equipment). The seedlings were transplanted to the field before entering the flowering stage. The management of the test field was the same as that of the general field with drip irrigation once a week, nitrogen as the main fertilizer at the seedling stage, and P and K as the main fertilizer at the reproductive stage. The test field was located at the comprehensive testing field (87°47′E, 43°95′N) of Xinjiang Academy of Agricultural Sciences in Urumqi, China.

Sample preparation

The genotypes of the transgenic seedlings and wild-type seedlings were identified when the plants grew stronger, with at least four leaves, and genomic DNA from the leaves was extracted for the detection of mutations. We tested for soluble sugar and TSS content in the fruit when the fruit was red and ripe (about 50 days post anthesis). At least six uniform fruits (two from each replicate with three replications) were collected from the second and third spikes of each plant, crushed, and homogenized into a sauce. The sauce was then analyzed for soluble sugar and TSS content.

Phylogenetic analysis

The protein sequences of INVINH and VPE from tomato (Solanum lycopersicum), Arabidopsis thaliana, pepper (Capsicum annuum), citrus (Citrus sinensis), soybean (Glycine max), tobacco (Nicotiana tabacum), rice (Oryza sativa), potato (Solanum tuberosum), grape (Vitis vinifera), and maize (Zea mays) were collected using the BLAST database search tool at NCBI (Fig. 1). The following protein sequences were downloaded from GenBank or SGN: Sl INVINH1, accession number AJ010943; Os INVINH1, AK288558; Os INVINH6, AK110798; Zm CWINVINH1, XP_008668976; Zm INVINH2, AX214357; ZM INVINH3, AX214336; At INVINH, AT1G48010; NT VINVINH, AY594179; NT CWINVINH, Y12805; St INVINH1, JQ043180; St INVINH2, GU980594; St VPE, XP_006355320; St VPE2, ACC68681; Nt VPE1b, BAC54828; NtVPE1a,BAC54827; NtVPE3, BAC54830; Nt VPE2, BAC54829; Gm VPE2, NP_001236564; Cs VPE, NP_001275777; Zm VPE, ACG34144; Zm VPE1, AAL58571; Zm VPE2, ACG47915; At VPE, NP_195020; At VPE2, NP_180165; At VPE3, NP_176458; At VPE4, NP_001189938; Vv VPE,AGC54786; Os VPE1, XP_015636414; Ca VPE1b, CA12g18500; Sl VPE1b, Solyc08g065610; Sl VPE5, Solyc12g095910; Sl VPE4, Solyc08g079160; Sl VPE1, Solyc08g065790; Sl VPE2, Solyc08g065780; Sl VPE10, Solyc08g065690; Sl VPE6, Solyc08g065530; Sl VPE11, Solyc08g065710; Sl VPE12, Solyc08g065720; Sl VPE8, Solyc08g065590; Sl VPE13, Solyc08g065740; Sl VPE7, Solyc08g065570; Sl VPE9, Solyc08g065600; Sl VPE14, Solyc08g065750. Multiple alignments of the INVINH and VPE protein sequences were processed using MEGA7 (Kumar, Stecher & Tamura, 2016). A phylogenetic tree was constructed with the neighbor-joining statistical method using 1,000 bootstrap replicates (Kumar, Stecher & Tamura, 2016).

Design of sgRNA and construction of CRISPR/Cas9 vector

CRISPR-P (http://cbi.hzau.edu.cn/cgi-bin/CRISPR) was utilized to design the target sites of the target genes SlINVINH1 (Solyc12g099200.1) and SlVPE5 (Solyc12g095910.1). Each gene was provided with two target sites for a more efficient gene knockout. SlINVINH1 target sites were selected in the first exon, and SlVPE5 target sites were selected in the first and second exon (Fig. 2B). The CRISPR/Cas9 original vector was obtained from Prof. Cai-xia Gao (Chinese Academy of Science, Beijing, China). The optimization of CRISPR/Cas9 gene codon and the synthesis of sgRNA sequence was provided by the Biological Company of Hong Xun, Suzhou, China. We used the construction method described by Yu et al. (2017), adding the 35S promoter driving CRISPR/Cas9 gene, the U6 promoter of Aribidopsis thaliana, and tomato control the sgRNA1 and sgRNA2, separately. The cloned fragments were assembled using the Circular Polymerase Extension Cloning (CPEC) method (Quan & Tian, 2009). The pCAMBIA1301 binary vector (AtU6-sgRNA1-SlU6-sgRNA2-35S-Cas9) was constructed and used for gene knockout (Fig. 2A).

Plant transformation

We used the Agrobacterium-mediated transformation method as described previously (Van Eck, Kirk & Walmsley, 2006) to transform pCAMBIA1301 vectors containing the CRISPR/Cas9 and sgRNA cassette into M82. Tomato seeds were germinated after sterilization with 10% NaClO on ½ MS medium. The apical segments of hypocotyls were punctured with Agrobacterium suspension after 9 to 12 days of culturing (OD600 = 0.5–0.6). The explants were then inoculated on a selective medium containing three mg/L hygromycin until the transgenic plants were regenerated from the callus. After the plants were grown in vitro, they were transplanted in a pot with soil and placed in light growth chamber.

DNA extraction and mutation detection

Genomic DNA from fresh frozen leaves was extracted using a high-efficiency plant genome DNA extraction kit (Tiangen, Beijing, China). Specific primers were used to amplify the flanks of genomes containing the target site (Table S1) and the PCR products were isolated by 1% agarose. The PCR products were linked to the pZERO-T vector after cutting and purifying (Transgen, Beijing, China), and 26 positive clones were randomly selected for mutagenesis from each plant. The Sanger method was used for sequencing with M13 primers. We obtains T1 and T2 lines for genotyping analysis by strict self-pollination of the T₀ plants. The target fragments were directly sequenced. The genotyping of F₂ plants was obtained from the F₁ lines using strict self-pollination. The F₁ lines were obtained from the two cross-pollinated T₀ lines. Their target fragments were also directly sequenced.

Determinations of soluble sugars, TSS and lycopene

Soluble sugars were determined as described previously by Freschi et al. (2010). Glucose, fructose, and sucrose levels were determined using high-performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD; Dionex, Sunnyvale, CA, USA). We used a Carbopac PA1 column (250 × 4 mm, five μm particle size, Dionex, Sunnyvale, CA, USA) as an isocratic run with 18 mM NaOH as the mobile phase. A standard curve made with pure glucose, fructose, and sucrose was used to calculate sugar content. TSS were measured with a refractometer DR201-95 (Kruess, Germany).

For determination of lycopene, pigments were extracted from two g of fresh ripe fruits using acetone/petroleum ether (1:1, v/v), then dried under a stream of N₂ and dissolved in 100% dichloromethane. The HPLC analysis was performed with 10 μL dichloromethane-dissolved pigments on ACQUITY UPLC (Waters, Milford, MA, USA). Lycopene was identified by the characteristic absorption spectra (472 nm) and distinctive retention times, compared with the standard substance (Sigma, St. Louis, MO, USA). Each lycopene content was calculated through the linear regression equation generated from the corresponding calibration curve, which was made using the corresponding standard substance.

Each sample contained three replicates with two fruits per replicate.

Characteristics and phylogenetic analysis of genes

The invertase inhibitor (INVINH1) protein consists of 171 amino acid residues, with a signal peptide of 19 amino acid residues at the N-terminus. The differences between INVINH1 and another type of invertase inhibitor in tomato, SolyCIF, include distinctions in protein sequences and protein structures (Reca et al., 2008). Compared with the sequence of the INVINH1 protein in other crops, four cysteine residues were conserved, which was a significant feature of all invertase inhibitors (Rausch & Greiner, 2004). Cluster analysis revealed that the INVINH1 protein of tomato had the closest relationship with that of potato and tobacco, which are also solanaceous crops, but was distantly related to Arabidopsis, corn, and rice (Fig. 1A).

The vacuolar processing enzyme (VPE) has been classified in the cysteine protease family and is mainly involved in the regulation of post-translational levels of proteins in vacuoles. The VPE proteins are highly conserved in most plants and animals (Hara-Nishimura, Takeuchi & Nishimura, 1993). Among the numerous VPE proteins in tomatoes, SlVPE3 and SlVPE5 have been reported to regulate the accumulation of soluble sugars. The soluble sugar content was higher in RNA interference lines (RNAi-SlVPE3 and RNAi-SlVPE5) compared with the control plants. However, the soluble sugar content of RNAi-SlVPE5 line was significantly higher than that of RNAi-SlVPE3 line. Therefore, only SlVPE5 was chosen as the target gene for genome editing in this study (Ariizumi et al., 2011). Cluster analysis revealed that SlVPE3 and SlVPE5, which are involved in sugar metabolism, did not cluster with other tomato VPE proteases but were rather the most closely related to tobacco (Nicotiana tabacum), the NtVPE3 protein, and the sweet pepper CaVPE1b protein (Fig. 1B). This suggests that there may be a class of proteases among the VPE family proteins that specifically regulate the accumulation of sugars.

Characterization of targeted editing in transgenic plants

The transformation system of the SlINVINH1 gene knock-out finally yielded 14 T₀ generation lines positive for the Cas9 gene. Genomic DNA was extracted from the leaves of each line. PCR using primers designed for the target region was then conducted to amplify the mutant region fragments, and each clone of the amplified products was sequenced and analyzed to obtain information on mutations in the target region of the INVINH1 gene. The sequencing results demonstrated that among the 14 lines that were positive for Cas9, only six had a mutation in the target sequence of the SlINVINH1 gene (probability of mutation = 42.86%). Among the mutants, one line was a homozygous mutant (both alleles had the same type of mutation), two were biallelic mutants (both alleles were mutated, but with different mutations), one was a heterozygous mutant (only one allele was mutated), and two were chimeric mutants (only some of the cells are mutated) (Figs. 2C and 2D; Table S2). Detailed mutations in the target sequence were mainly the deletions of single or multiple nucleotides (Fig. 2C).

The transformation system of the SlVPE5 gene knock-out yielded 13 T₀ generation lines that were positive for the Cas9 gene. The results of mutation detection revealed that among the 13 Cas9-positive lines, eight lines had mutations in the target sequence of the SlVPE5 gene (probability of mutation = 61.54%). Among the mutants, one line was a homozygous mutant, another line was a biallelic mutant, five lines were heterozygous mutants, and one line was a chimeric mutant (Figs. 2C and 2D; Table S2). Detailed mutations in the target region were mainly the deletion and replacement of single or multiple nucleotides (Fig. 2C). Further analysis revealed that the mutation types of the SlINVINH1 gene were mainly biallelic mutations and chimeras, while the mutation type of the SlVPE5 gene was mainly heterozygotic (Figs. 2C and 2D; Table S2).

CRISPR/Cas9-mediated mutants exhibited soluble sugar and TSS content increase

The T₀ mutant line was obtained as described above. Soluble sugar and TSS contents were determined after maturation with three replicates for each line. Among the T₀ mutant lines of the SlINVINH1 gene four lines (INH-1, INH-2, INH-3, and INH-4) (Fig. 2C) were selected. The INH-5 and INH-6 lines were not selected as they were chimeras and displayed an extremely low probability of passing the mutations on to the next generation. We observed that the contents of both glucose and fructose in the fruits from lines INH-1, INH-2, and INH-4 were significantly increased compared to the wild-type. INH-1 showed a 40.19% increase in glucose and a 42.42% increase in fructose; INH-2 showed a 36.39% increase in glucose and a 35.69% increase in fructose; INH-4 showed a 40.82% increase in glucose and a 42.76% increase in fructose (Fig. 3A; Table S3). In addition, the TSS content also showed a significant increase (31.90% in INH-1, 30.17% in INH-2, 32.76% in INH-4) (Fig. 3A; Table S3), indicating that the SlINVINH1 gene mainly regulated the accumulation of glucose and fructose in the fruits. However, the content of soluble sugars in the fruit of heterozygous INH-3 did not increase significantly compared to the wild-type (Fig. 3A; Table S3), which could be due to the recessive function of the mutated slinvinh1 gene.

Among the T₀ lines that carried the mutations of the SlVPE5 gene, three lines, VPE-1, VPE-2, and VPE-7, were selected. The VPE-8 line was not selected as it was chimera and showed an extremely low probability of passing the mutations to the next generation. Among the five heterozygous, only one line, namely VPE-2, was selected as a representative for testing. The test revealed that the glucose and fructose content in the lines fruits of VPE-1 and VPE-7 increased significantly (glucose increased 35.20% in VPE-1, and increased 35.83% in VPE-7; fructose increased 37.33% in VPE-1, and increased 43.00% in VPE-7). The TSS content also increased significantly (30.63% in VPE-1, 32.43% in VPE-7) (Fig. 3B; Table S3), indicating that the SlVPE5 gene also regulated the accumulation of glucose and fructose in fruits. However, the soluble sugar content in the fruits of heterozygous INH-2 did not increase significantly compared to the wild-type (Fig. 3B; Table S3), which might also be caused by the recessive gene characteristics of the mutant slvpe5 gene.

The double homozygous mutants further increased the soluble sugar and TSS content

The T₀ generation homozygous mutant lines INH-4 and VPE-7 were cross-pollinated to generate the F₁ generation. The F₂ generation was obtained by self-pollination of the F₁ generation, and 515 well-grown lines were obtained after sowing the F₂ generation seeds. We screened for the gene segregation and obtained 32 lines with homozygous mutated SlINVINH1 and SlVPE5 sites (genotype: inh/inh-vpe/vpe) from 515 lines. We identified two lines without the exogenous Cas9 genes by further screening for the exogenous Cas9 genes among the 32 lines. The two lines were numbered F₂–1 and F₂–2.

The soluble sugar and TSS in fruit from the F₂–1 and F₂–2 lines were determined after maturation, each line with three replicates. Glucose and fructose content in F₂–1 and F₂–2 fruits significantly increased compared to the wild-type. In F₂–1, glucose increased by 64.86%, and fructose increased by 68.40%, while in F₂–2, glucose increased by 67.41%, and fructose increased by 69.44% (Fig. 4; Table S4). The TSS content also showed a significant increase (55.17% in F₂–1 and 62.07% in F₂–2) (Fig. 4; Table S4).

We observed that both the SlINVINH1 and SlVPE5 gene mainly regulated the accumulation of glucose and fructose in the fruit. In addition, both glucose and fructose content was significantly synergistically increased in the F₂–1 and F₂–2 fruit (compared to the average of INH-4 and VPE-7 lines, glucose increased by an average of 18.12% in two F₂ lines; and fructose increased by 14.06%) (Table S5). The TSS content was also significantly higher than the single locus-mutated homozygote (which increased by 22.31%) (Table S5). This indicated that the SlINVINH1 and SlVPE5 genes displayed a synergistic effect in regulating the content of soluble sugar.

Dynamic variation patterns of fruit development and coloration

SlINVINH1 is an invertase inhibitor gene and the loss of invertase function may delay fruit ripening. However, the loss of SlVPE5 gene function may also cause changes in other commercial traits like the size and color of fruits. We followed the development of two lines (F₂–1 and F₂–2), starting from flowering to fruit ripening, in real time. The results indicated that tomatoes of the F₂–1 and F₂–2 lines were similar to the wild-type tomato fruits, starting from flowering to the physiological fruit expansion stage, and including both shape and size of the fruits. However, during the breaker stage and the beginning of the ripping stage, fruits of the F₂–1 and F₂–2 lines were significantly delayed the break stage compared to the wild-type fruits, although they were fully ripened at the same time as wild-type fruits (Fig. 5A). This indicated that although the knock-out of the SlINVINH1 and SlVPE5 genes did delay the break stage and the color change of tomato fruits, it neither affected the ripening and harvesting time of the fruits nor the color of fruits after ripening nor did it affect fruit weight (Fig. 5B; Table 1).

The effects of SlINVINH1 and SlVPE5 in increasing the soluble sugar and the TSS content of tomato

SlINVINH1 is an invertase inhibitor gene, which specifically inhibits the activity of cell wall invertase. Invertase hydrolyzes sucrose into glucose and fructose and plays a major role in fruit development. The suppression of SlINVINH1 gene expression increases the glucose and fructose content in fruits (Sturm, 1999; Jin, Ni & Ruan, 2009). SlVPE5 belongs to the family of VPE genes, which targets various types of vacuolar hydrolases (such as β-glycosidase, α-mannnosidases, and α-galactosidases) for degradation and the negative regulatory mechanism for soluble sugar accumulation (Rojo et al., 2003; Ariizumi et al., 2011). We observed that the both genes SlINVINH1 and SlVPE5 mainly regulated the accumulation of glucose and fructose in fruits (Figs. 3A, 3B, and 4), confirming the findings of previous research (Jin, Ni & Ruan, 2009; Ariizumi et al., 2011; Xu et al., 2017). Previous research also found that the SlVPE5 gene also regulates the accumulation of sucrose in fruits (Wang et al., 2016). However, in this study no sucrose content was detected in mature fruits (detection threshold was ≥0.2 mg/g fw) in the T₀ or F₂ lines. This may be due to the low sucrose content in mature fruits and because the sucrose contribution to the TSS of mature fruits is very small (Tieman et al., 2017). The knock-out of both loci in homozygotes resulted in a significant increase in both tested sugars in the fruit, and the TSS content was also significantly higher than that in single locus-mutated homozygotes, indicating that there was a synergistic effect of SlINVINH1 and SlVPE5 genes in regulating the content of soluble sugar. This proves once again that the increase of soluble sugar content in fruits is a complex process and not regulated by a single gene (Beckles et al., 2012; Patrick, Botha & Birch, 2013). This finding is crucial for the improvement of quantitative genetic traits of soluble sugar.

The knockout efficiency of two genes

The knock-out efficiency of both genes were different. In the knock-out experiment on SlINVINH1, we obtained six lines with the successful knock-out of the target gene from 14 positive lines. Its knock-out efficiency was 42.86%. In the knock-out experiment on SlVPE5, eight lines with the successful knock-out of the target gene were obtained from 13 positive lines, with a knock-out efficiency of 61.54%. Furthermore, the mutation types of the SlINVINH1 gene were mainly biallelic mutations and chimeras, while the mutation types of the SlVPE5 gene were mainly heterozygous (Figs. 2C and 2D; Table S2). This may be linked to the different types and structures of the genes (Pan et al., 2016; Li et al., 2018). According to previous research vector compontents, such as codon optimization of CRISPR/Cas9 gene, promoter, and structure of gRNA, may affect gene knockout efficiency (Dang et al., 2015; Ma et al., 2015; Čermák et al., 2017; Feike et al., 2019; Shao et al., 2020; Grützner et al., 2021).

Effective improvements to the commercial traits of tomato

We obtained two homozygous lines (F₂–1 and F₂–2) with mutations in both the functional gene loci, excluding exogenous Cas9, and found that both the soluble sugar and TSS content in tomato fruits were significantly increased (Fig. 4), facilitating the rapid improvement of the commercial traits of tomatoes. Our study confirmed the feasibility of target gene editing by the transfer of Cas9 and sgRNA, and the subsequent application of conventional methods such as hybridization and inbreeding. Two mutation sites may be brought together, however, the mutation site could be made homozygous and the exogenous gene Cas9 could be eradicated. This method greatly shortened the breeding cycle to improve tomato varieties. It also completely eliminated various risk factors associated with the exogenous genes. This system will improve technological innovation and progress in tomato breeding, and has tremendous potential in a variety of applications.