Three varieties of Solanum lycopersicum L., i.e., cv. Moneymaker, Rio Grande, and Ailsa Craig, were used in this study. Moneymaker and Ailsa Craig are indeterminate type and Rio Grande is a determinate type tomato. Tomato seeds were sterilized by 70% EtOH and 10% sodium hypochlorite solution (Antiformin, Wako, Japan) including 0.005% Triton X-100 (octylphenol ethoxylate, Wako, Japan). Sterilized seeds imbibed water overnight and were sown on germination medium [1.5% sucrose, 1/2 × MS (pH 5.7)]. Expanded cotyledons 7 days after sowing were used for transformation. Tomato calli and shoots were grown in tissue culture in a growth chamber under conditions of 21–25°C with 15 h light at 30–50 μmol m⁻² s⁻¹/9 h dark. T1 or T2 tomato plants imbibed water overnight, germinated in Jiffy-7 (Jiffy), and grown in a growth chamber (CLE-305, TOMY, Japan) or plant growing rack (CR-400L, TOMY, Japan) under conditions of 21–25°C with 15 h light at 40–70 μmol m⁻² s⁻¹ (low light conditions) or 150–250 μmol m⁻² s⁻¹ (normal light condition)/9 h dark.

In our previous study (Ueta et al., 2017), the SlIAA9 (Solyc04g076850) target sequence, 5'-GAGCTCAGGCTCGGTCTACC-3' (gRNA2) was used for genome editing of all four tomato cultivars. The CRISPR-Cas9 expression vector, pEgP237-2A-GFP (Ueta et al., 2017; Supplementary Figure S1A), was used, which comprises a gRNA expression cassette under the control of the Arabidopsis U6 snRNA-26 (AtU6-26) promoter, and a Cas9 expression cassette for an Arabidopsis codon-optimized spCas9 (AtCas9) fused to GFP via a 2A peptide, which is driven by the 2 × CaMV35SΩ promoter. The AtCas9 contains the 3 × NLS at its C-terminus. In this vector, a kanamycin-resistance marker was used for selection of transgenic plants.

The CRISPR-Cas9 expression vector was introduced into Agrobacterium tumefaciens (GV2260) and used to transform tomato as described previously (Ueta et al., 2017; Hashimoto et al., 2018). Briefly, Agrobacterium harboring CRISPR-Cas9 vector was cultured on LB medium or AB-MES medium. Leaf discs prepared from cotyledons were subjected to Agrobacterium infection. After co-culture for several days in the dark, the leaf discs were transferred to callus induction medium (CIM) 1 (3% sucrose, 1 × MS, 100 mg/L kanamycin, 1.5 mg/L trans-zeatin, and 25 mg/L meropenem) and cultured for 4 weeks for Moneymaker and Ailsa Craig. After the appearance of calli, shoot induction medium (SIM) 1 (3% sucrose, 1 x MS, 100 mg/L kanamycin, 1.0 mg/L trans-zeatin, and 25 mg/L meropenem) was used for further incubation. Leaf discs from Rio Grande were incubated in CIM2 (3% sucrose, 1 × MS, 50 mg/L kanamycin, 1 mg/L BA, 0.1 mg/L IAA, and 25 mg/L meropenem) for 4–5 weeks after co-culture, then calli were transferred to SIM2 (3% sucrose, 1 × MS, 50 mg/L kanamycin, 1.0 mg/L trans-zeatin, 0.5 mg/L IAA, and 25 mg/L meropenem) and incubate for at least a further 5 weeks. Transgenic shoots were transferred to root induction medium (1.5% sucrose, 0.5 × MS, 50 mg/L kanamycin, and 25 mg/L meropenem) for 2–4 weeks and then to soil pots.

Genomic DNA was isolated from tomato leaves using Nucleospin Plant II (TaKaRa, Japan), and used for amplification of target sequences by PCR using PrimeSTAR GXL DNA Polymerase (TaKaRa, Japan). An AccI site is located on the SlIAA9 target sequence, and PCR-RFLP was used for detection of the mutation. In PCR-RFLP, PCR fragments were digested with AccI (NEB, Japan) and analyzed by agarose gel electrophoresis. For Sanger sequencing analysis, PCR fragments purified from agarose gels were cloned by the Seamless ligation cloning extract (SLiCE) method (Motohashi, 2015) into cloning vector pNEB193 (NEB, Japan). All primers used for PCR are listed in Supplementary Table S1.

Around 20–50 plants of the T1 generation and eight plants of the T2 generation were selected randomly for PCR-based detection of T-DNA regions. Preparation of total DNA was as described above. Nine regions in the T-DNA in the CRISPR-Cas9 vector were selected and amplified using PrimeSTAR GXL polymerase (TaKaRa, Japan). All primers used for PCR and the corresponding amplicon sizes are indicated in Supplementary Table S1. PCR experiments were replicated three times for each line.

For Southern blot, total DNA was extracted from leaves of WT and sliaa9 mutants by the modified cetyltrimethylammonium bromide (CTAB)-based DNA extraction method (Murray and Thompson, 1980). Briefly, tomato leaves were ground in liquid nitrogen and incubated in 2% CTAB solution [100 mM Tris-HCl (pH8.0), 20 mM EDTA (pH8.0), 1.4 M NaCl, and 2% (w/v) CTAB]. The lysates were extracted with chloroform:isoamyl alcohol (24:1), and then 1% CTAB solution [50 mM Tris-HCl (pH8.0), 10 mM EDTA (pH8.0), 1% (w/v) CTAB] was added to the water layer. Total DNAs were obtained by CsCl-EtOH precipitation. Total DNA was digested with HpaI, which has a single restriction recognition site in the vector (Supplementary Figure S1A), fractionated on 1% agarose gels, and transferred to nylon membrane (Zeta-Probe, Bio-Rad, United States) by capillary transfer. After UV cross-linking (UVP CL-1000 crosslinker, Analytik Jena, Upland, CA, United States), the membranes were hybridized with DIG-labeled probes for gRNA, Cas9, and LHCB (Solyc02g070970) genes. Probes were amplified using pEgP237-2A-GFP (for gRNA and Cas9) or tomato genomic DNA (for LHCB) using the primers listed in Supplementary Table S1 and labeled with DIG using a DIG High Prime DNA Labeling and Detection Kit (Roche, Basel, Switzerland). Probe labeling, membrane hybridization, and detection were carried out according to the manufacturer’s instructions. Hybridization signals were detected and imaged with PXi (SYNGENE, Bangalore, India).

Amplicon sequencing was performed using MiSeq (Illumina, Japan) and MiSeq Reagent Kit v2 Nano (Illumina, Japan) was used for analysis. Genomic DNAs were used to amplify the region of the CRISPR-Cas9 target sites or off-target sequences by PCR. One off-target candidate sequence (SL2.50 ch6:26946923-26946946) was selected by Cas-OT to examine the mutation. PCR products were separated by electrophoresis, purified from the agarose gel using the Wizard SV Gel and PCR Clean-Up System (Promega, Japan), and used as templates for a second round of PCR. Second PCR was performed using TruSeq HT primer (Illumina, Japan). All primers used for PCR are listed in Supplementary Table S1. MiSeq data were analyzed using CLC Genomics Workbench software version 7.5.1 (CLC bio, Japan) and mapped on the off-target candidate using Integrative Genomics Viewer (IGV; Broad Institute).

Shoot lengths of wild-type and sliaa9 mutants under low or normal light conditions were measured until 48 days after germination. Hypocotyl length in young seedlings of the sliaa9 mutants and wild-type were measured at 21 days after germination. The third to tenth internode length of the sliaa9 mutants and the wild-type were measured at 48 days after germination. Several parameters of the fifth to tenth matured leaves from each plant of the sliaa9 mutants and the wild-type were measured 48 days after germination. The parts of leaves measured are shown in Supplementary Figure S4A. Leaf area was measured using ImageJ. Average values were determined using 2–6 lines of the same mutant allele. All experiments repeated 2–3 times.

SPAD-502Plus (Konica Minolta, Japan) was used to measure leaf chlorophyll concentration in leaves. SPAD values were measured at 3–8 sites near the principal vein on the wide part of the tomato leaf on the third or higher fully elongated leaves. In total, 20–30 SPAD values in a plant were analyzed; average values were determined using 3–6 lines of the same mutant allele. All experiments repeated three times.

Hypocotyls of 21-days-old seedlings and the third internodes of 35-days-old seedlings grown under low or normal light intensity were harvested, respectively. Total RNAs were extracted using RNeasy Plant Mini Kit on the QIAcube automated system (Qiagen), followed by reverse-transcribed using a ReverTra Ace qPCR RT Master Mix with gDNA remover (TOYOBO) following the manufacturer’s instructions. Quantitative PCR was performed with GoTaq qPCR Mater Mix (Promega) on LightCycler 96 (Roche). The expression levels of SlIAA3 (5'-GCAATTCAAATCAATCATTCTTTCT-3' and 5'-GTTTATTATCCCAGGCAAACCTAAT-3') were normalized relative to those of a reference gene, Slactin7-like (5'-TGTCCCTATCTACGAGGGTTATGC-3' and 5'-AGTTAAATCACGACCAGCAAGAT-3'; Hashimoto et al., 2018). Relative expression was calculated by two to the power of relative expression (log₂), calculated by subtracting Ct values of IAA3 from those of Slactin7-like.

A CRISPR-Cas9 vector targeting exon 2 of the SlIAA9 gene (+217 to +236 from ATG, 5'-GAGCTCAGGCTCGGTCTACC-3'), which was used to generate mutants in tomato cultivars, Micro-Tom and Ailsa Craig, in a previous study (Ueta et al., 2017), was introduced into Moneymaker and Rio Grande via Agrobacterium transformation. PCR-RFLP analysis using an AccI restriction enzyme site located at the CRISPR-Cas9 target was performed to detect CRISPR-Cas9-induced mutations in the T0 regenerated tomato shoots. Non-digested fragments in PCR-RFLP analysis indicated 100% somatic mutations in all cultivars (Figures 1A,B). The progenies of sliaa9 T0 100% mutant lines (T1 lines) were isolated by self-pollination from T0 mutant lines. Sanger sequencing analysis of cloned PCR fragments at the target sites in the T1 mutant lines revealed only one or two types of mutation, which were found at similar levels in all the sequenced clones in these lines (Figure 1C; Supplementary Figure S1). These results suggest the generation of bi-allelic mutants, in which the SlIAA9 mutations introduced a frame shift just after domain I, and a stop codon was generated to produce truncated SlIAA9. The progenies of sliaa9 knockout mutant lines (T1 mutants) were isolated by self-pollination from T0 mutant lines, and T1 mutants showed homozygotes or heterozygotes that were segregated from the parental T0 generation (Supplementary Figure S1). Since T1 mutants were generated from T0 lines by self-pollination, the T-DNA, including the CRISPR-Cas9 cassette, can be segregated out. To detect T-DNA insertion in the genomic DNA of sliaa9 T1 progenies, PCR analysis designed to amplify nine T-DNA regions and Southern blot analysis were performed using specific primers and DIG-labeled probes, respectively (Supplementary Table S1; Supplementary Figure S1A). Several null-segregant mutants were isolated, in which no T-DNA regions were detected by PCR or Southern blot analysis (Supplementary Figures S1B,C). T-DNA regions were also undetected in the T2 progenies of MM#2, i.e., MM#2-13 (Supplementary Figure S1B). An off-target candidate sequence was analyzed using next generation sequencing, Miseq, and the results suggest that off-targets were not detected in mutant plants (Supplementary Figure S2). The T2 progenies of these null-segregant sliaa9 T1 mutant lines MM#2-13, RG#4-1, and AC#1-1 were named as MM-sliaa9,” “RG-sliaa9,” and “AC-sliaa9,” respectively, and used for subsequent experiments.

We next analyzed the effects of SlIAA9 disruption on plant growth among cultivars under different light conditions: low (40–70 μmol m⁻² s⁻¹) or normal (150–250 μmol m⁻² s⁻¹) light intensities. MM-sliaa9 and RG-sliaa9 showed slower rates of shoot elongation growth in the early growth stages compared with wild-type under low light intensity, whereas AC-sliaa9 did not show clear difference in growth rates (Figures 1D, 2A). Shoot lengths of wild-type and sliaa9 mutants in Moneymaker and Rio Grande grown under normal light intensity were mostly longer than those grown under low light intensity, on the contrary, Ailsa Craig plants did not show large differences between light conditions (Figures 2A,B). Moneymaker sliaa9 showed a strong phenotype in the internode elongation, and this phenotype of Rio Grande mutant was apparent in low light intensity (Figure 2C). On the contrary, there were no significant differences in the internode length of Ailsa Craig sliaa9 and wild-type under the different light intensity (Figure 2C). These results suggest that sliaa9 mutants showed different shoot elongation growth phenotypes among cultivars.

Next, we analyzed the effect of light intensity on hypocotyl growth of young seedlings of tomato sliaa9 mutants grown in different light intensity to identify the AUX/IAA9 function in light response in the tomato cultivars. The results indicated a different response to light conditions among mutants of the three cultivars (Figure 2D). Thus, the hypocotyl length of MM-sliaa9 was smaller than that of wild-type under both light intensity, whereas, AC-sliaa9 showed a longer hypocotyl than that of wild-type in low light intensity; however, the effect on hypocotyl growth was rescued in normal light, and hypocotyl length of mutant and wild-type were similarly decreased (Figure 2D). There was no clear differences between RG-sliaa9 and wild-type; hypocotyl length of mutant and wild-type were similarly elongated under low light intensity (Figure 2D). Furthermore, hypocotyl elongation in Moneymaker is strongly affected by SlIAA9 knockout and did not show any further effects of light intensity. These results might suggest that photomorphogenesis in the hypocotyl elongation of young seedlings differs in different tomato cultivars, and the relationship of the auxin signaling pathway mediated by SlIAA9 and photomorphogenesis might be clearer in Ailsa Craig than in Moneymaker and Rio Grande.

To understand molecular mechanisms of elongation growth controlled by SlIAA9 and the downstream pathway, we focused on the expression of SlIAA3 gene. The SlIAA3 expression levels were analyzed in the third internodes and hypocotyls of young seedlings grown under low or normal light intensity (Supplementary Figure S3). The results indicated that there were not close relationship of the SlIAA3 expression levels, elongation growth mediated by SlIAA9, and the light intensity among cultivars, although weak downregulation of the SlIAA3 expression was observed in the SlIAA9 Rio Grande mutants. The results suggest that complicated pathways might be existed in the downstream of SlIAA9 in elongation growth of the tomato cultivars.

We next analyzed the leaf morphology of tomato sliaa9 mutants. Previous studies had shown that disruption of SlIAA9 in tomato led to altered leaf morphology, with the compound leaf changing to a single leaf (Wang et al., 2005; Berger et al., 2009; Saito et al., 2011; Ueta et al., 2017). In the present study, the sliaa9 mutants in the three cultivars exhibited morphological changes to the single leaf (Figure 3A). To analyze the phenotypes in elongation growth in leaves, we measured several parameters of fully expanded leaves in young plants (Supplementary Figures S4, S5; Figures 3B–E). The length of sliaa9 mutant leaves in Moneymaker and Rio Grande showed no significant differences when compared with those of wild-type in normal lights, whereas those of the Moneymaker mutant in low light conditions and the Ailsa Craig mutant in both light conditions were shorter than those of wild-type (Figures 3A,B), indicating different mechanisms related to elongation in leaves and shoots between cultivars. When the ratio of petiole length to full length of leaves was compared among the mutants, all three cultivars exhibited a higher ratio in sliaa9 mutants compared with wild-type in both light conditions (Figure 3C). Leaf width in sliaa9 mutants of all three cultivars was narrower than that of wild-type, and the mutants of Moneymaker and Ailsa Craig also showed a smaller leaf area than wild-type especially in low light conditions (Figures 3D,E). These results suggest that SlIAA9 knockout affected the several parameters in leaf shapes that might be differentially controlled in various tomato cultivars.

To analyze any further effects of SlIAA9 disruption on plant growth, we measured leaf numbers and chlorophyll content in mature leaves of mutant plants using a Soil Plant Analysis Development (SPAD) chlorophyll meter. Leaf numbers increased similarly in sliaa9 mutants and the wild-type of all three cultivars growing under both light conditions (Supplementary Figure S5). There were no significant differences in SPAD values between sliaa9 mutants and the wild-type plants, except that Moneymaker plants showed larger values under normal light than in low light conditions (Supplementary Figure S4E).

We next analyzed flower and fruit phenotypes of the sliaa9 mutants of Moneymaker and Rio Grande. Flower development in all these mutants showed no differences when compared with wild-type plants (Figure 4A). Fruit shape in sliaa9 mutants was also similar to that of the wild-type; on the contrary, several fruits had no, or only few, seeds, suggesting a parthenocarpic phenotype as shown in previous studies using cultivars Micro-Tom and Ailsa Craig (Figure 4A; Ueta et al., 2017). The total number of fruit, and seed numbers, were measured 2 months after the first fruit was fully ripened, indicating that parthenocarpic fruits were generated only in sliaa9 mutants (Figures 4B,C). These results indicate that the role of SlIAA9 in fruit development is common to the three cultivars, and that parthenocarpic phenotypes can be efficiently introduced into various cultivars by CRISPR-Cas9 technology.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.