Plant materials and growth conditions
Tomato (Solanum lycopersicum cv MicroTom) and other commercial accessions were grown in culture chambers (14 h day/10 h night cycle, 25/20 °C day/night temperature, 80% relative humidity). After 5% sodium hypochlorite sterilization, tomato seeds were sown on 0,5x Murashige and Skoog medium pH 5.9 with 0.8% (w/v) agar and transferred to the soil after about 16 days. Development and ripening stages refer to days post-anthesis (DPA) and breaker (BR), respectively, as determined by tagging flowers and fruits at anthesis and breaker stages.
Vector construction and plant transformation
CRISPR-P (http://cbi.hzau.edu.cn/crispr/) was used to design all the sgRNAs that target either SlMBP3, SlAGL11, our both genes named here dual SlMBP3/SlAGL11. Sequences of the sgRNAs and the induced genomic mutations are given in Supplementary Fig. 4 and Supplementary Data 11. Plasmids were assembled by the Golden Gate strategy29. To generate the transgenic plants for RNAi silencing lines, two DNA fragments specific to SlMBP3 and SlAGL11 were amplified by PCR from a tomato fruit cDNA library and cloned into pHellsgate12 system vector using the Gateway site-specific recombinational cloning protocols (Invitrogen, USA). For dual SlMBP3/SlAGL11 silencing, a DNA fragment corresponding to a conserved sequence in SlMBP3 and SlAGL11 cDNA was amplified and integrated into the pHellsgate12 system. SlMBP3 overexpression constructs was obtained after cloning the full length CDS of SlMBP3 and integration into a PMDC32 vector containing the cauliflower mosaic virus (CaMV) 35S promoter and the Nos terminator using the Gateway® technology (Invitrogen, USA). The 35S::SlMBP3-GFP construct was obtained by cloning SlMBP3 full length sequence into a modified pGreen vector containing the CaMV 35S promoter and the GFP coding sequence downstream of a SmaI cloning site19. ProSlMBP3 and ProSlAGL11 GUS constructs were obtained after amplification of 2,7 kb SlMBP3 promoter fragment and 1,6 kb SlAGL11 promoter fragment, respectively, from tomato genomic DNA and insertion into the pMDC162 vector using the Gateway site-specific recombinational cloning protocols (Invitrogen, USA). The pSlMBP3::SlMBP3 construct was built using Golden braid ligation technology (https://gbcloning.upv.es/)30. First, domestication steps have been performed as follows, DNA fragments containing the 2,7 kb promoter region of SlMBP3 and the full length CDS of SlMBP3 were amplified and cloned into pUPD2 vectors using the BsmBI restriction enzyme. In step two, the domesticated pSlMBP3, SlMBP3, and Nos terminator (https://www.addgene.org, GB0037) part was cloned into pDGBalpha2 vector (SlMBP3-alpha2 vector) using BsaI restriction enzyme. Then, the final construct, containing the Kanamycin resistance gene, was obtained by mixing the SlMBP3-alpha2 vector and Tnos:nptII:Pnos-pDBGalpha1R (https://www.addgene.org, GB0226), into final vector Omega-1 using BsmBI enzyme. All detail primers are given in Supplementary Data 11. Transgenic plants were generated by Agrobacterium-mediated transformation19.
DNA extraction and genotyping of tomato mutants
For genotyping the first generation (T0) of transgenic line, three different leaf samples from different branches of one plant were collected and total genomic DNA was extracted using the ReliaPrepTM gDNA Tissue Miniprep System (Promega, France). PCR was performed to select the transgenic lines containing the Cas9 construct in T0 generation, and then to select the lines that outcrossed the Cas9 construct in the T1 generation. T2 plants without Cas9 construct were genotyped by PCR using primers designed to amplify a region of 600 bp encompassing the two sgRNA sequences31. To further confirm the editing type and check whether the line is homozygous, more than ten plants were grown for each line. PCR-amplified DNA fragments from leaf each plant were cloned into pDonor207 vector using Gateway system, and twelve clones of each PCR product were sequenced to select only the lines with consistent mutation. At least three homozygous lines for each construct were selected for further study.
Processing data for the WGS of tomato
Total genomic DNA from young fruits of 12 tomato cultivars (Fig. 1) was extracted using the ReliaPrepTM gDNA Tissue Miniprep System (Promega, France). Library preparation and DNA sequencing using a HiSeq 3000 sequencer (Illumina) was done at Shanghai Majorbio Biopharm Technology Co., Ltd (Shanghai, China), operating in a 150 bp paired-end mode. The raw sequencing reads were first cleaned from their eventual remaining adapters using TrimGalore (a wrapper tool around cutadapt and FastQC, http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/ and http://www.bioinformatics.babraham.ac.uk/projects/fastqc), then mapped to MicroTom reference genome (http://tomatogenome.gbfwebtools.fr/) using BWA (Burrows–Wheeler Aligner)32 with default parameters. Only uniquely mapped reads were retained. Following the Broad institute’s Best Practices for SNP calling, the sample alignments were then organized by individual subsets called readgroups and cleaned from artifactual duplications caused by the PCR process. A last pre-process step is applied to the data before the SNP calling which consists in the correction of the Base quality scores, performed per-sample, using a machine learning process included in GATK (Genome Analysis Toolkit)33. As the base quality score plays an important role in weighing the evidence for or against possible variant, this step tends to correct those scores which correspond to the confidence emitted by the sequencer for each base. The actual variant calling step was managed with HaplotypeCaller from the same toolkit using recommended parameters (https://software.broadinstitute.org/gatk/best-practices/)33. The uniquely aligned reads at SlAGL11 locus were visualized in Integrative Genomic Viewer (IGV) at an interval of 14 kb region of Chromosome 11 (19,608,621-19,622,629). Subsequent genotyping of the SlMBP3 SNPs and SlAGL11 locus deletion was done through Sanger sequencing of PCR products. All primer sequences are listed in Supplementary Data 11.
Phylogenetic tree
Phylogenetic tree of Arabidopsis thaliana and Solanum lycopersicum MIKC type MADS-box transcription factors was built after alignment of full-length protein sequences using MUSCLE algorithm and Maximum Likelihood clustering. Analyses were conducted using MEGA634.
Histological observations
Fruit anatomy was observed using an Axio Zoom V16 microscope (Zeiss, Germany) (FR-AIB TRI imaging platform). Toluidine staining was obtained after dipping hand-cut fruit sections 15 s into 0.05% [w/v] aqueous toluidine blue (Sigma-Aldrich, USA) and water rinsing before mounting the samples under a cover slip.
Histochemical and statistical analysis
A higher resolution of the morphological differences of locular tissue cells between SlMBP3-KO and WT fruits was analyzed on paraffin sections. Whole fruits from SlMBP3-KO and WT at three different early development stages (3, 6, 9-DPA, each 9–15 fruits coming from six individual plants) were harvested by cutting above the crown and rapidly dissected in two halves with a razor blade and immersed in 50 mL Falcon tubes containing FAA fixative solution (10% formalin (37% formaldehyde solution, Sigma-Aldrich, Saint-Quentin Fallavier, France); 50% ethyl alcohol; 5% acetic acid; 35% distilled water). The fixative was infiltrated in the samples by performing ten cycles of 1 min vacuum and vacuum release, and the infiltrated samples were fixed for 16 h at 4 °C. Dehydration series and paraffin infiltration were then performed16,35. Large paraffin embedding molds were used to dispose 5–10 fruit halves from both genotypes of a given developmental stage, constituting so-called tissue microarrays. Twelve micrometer-thick serial sections of each tissue array enabled to simultaneously cut under the same conditions large number of fruit cross-sections to be compared. The sections were disposed on silane-coated microscopy slides. Sections were stained for 1 min in 0.05 % Toluidine blue in 0.1 M acetate buffer pH 4.6 for morphology. Following extensive washing, the slides were dried, mounted in Eukitt® (quick-hardening mounting medium, Sigma-Aldrich) and scanned in bright field mode with a 20× objective using a NanoZoomer HT scanner (Hamamatsu, Hamamatsu City, Japan). The scans were observed with NDPview (Hamamatsu, Hamamatsu City, Japan), enabling to extract 16–21 regions of interest (ROIs) containing locular tissue from 6–8 fruits from each developmental stage and genotype observed at 10×. The cells from the locular tissue in these ROIs were analyzed using ImageJ36.
GUS staining and assays
Homozygous T2 transgenic lines expressing the SlMBP3 promoter (ProSlMBP3) or the SlAGL11 promoter (ProSlAGL11) fused to the GUS reporter gene were selected based on kanamycin resistance and histochemical staining to confirm GUS activity. Flowers and young fruits samples were incubated at 37 °C in GUS staining solution (0.1% Triton X-Gluc, pH 7.2, 10 mM EDTA)37; the incubation time ranged from three hours (flowers, very young fruits) to twelve hours (20-DPA fruits). Samples were then depigmented by several washes of graded ethanol series with ordinal 30%, 50%,70%, 95% concentration.
Auxin-induced fruit set
Unfertilized flowers were emasculated, and the fruit set was induced by auxin treatment (50 mg/L auxin solution once every 2 days for a period of 7 days).
Seed germination
Seed germination was assayed using the following protocol: seeds were gently shaken (50 rpm) in distilled water overnight at 25 °C, and then transferred on moist filter paper in Petri dishes where they were incubated for 7 days at 25 °C. Assays were performed in triplicate (independent biological repeats) with at least 60 seeds.
Fruit weight
Fruit weight was estimated by weighing a minimum of 20 fruits harvested at 7 day post-breaker stage (Br + 7) from five different plants bearing the same number of fruit in order not to bias the data.
Fruit dry matter content
Dry matter content was estimated using 13 fruits harvested from each line at Br + 7 stage. The weight of each fresh fruit was determined before incubation (Thermosi, SR1000) at 65 °C for 48 h. Then the weight of each dry fruit was measured. Dry matter content was determined by estimating the dry/fresh fruit weight ratio.
Fruit firmness measurement
Fruit firmness was assessed using Harpenden calipers (British Indicators Ltd, Burgess Hill, UK)19. For each stage, the measures were performed at different development stages on a minimal set of ten fruits for each line.
Fruit shelf life
For assessing the shelf life, ten fruits were harvested from different lines at Br+7 stage, pooled and stored at room temperature. Photographs were taken after a storage period of 8, 20, 30, and 50 days.
Soluble sugar determination
For each line, ten different plants were used, all bearing the same number of fruit picked at the same ripening stage. Two to three fruits were harvested from each plant and frozen in liquid nitrogen. Glucose, fructose, and sucrose content were determined19.
RNA extraction and quantitative RT-PCR
Fruits from different tomato lines were harvested, frozen in liquid nitrogen, and stored at −80 °C. Total RNA samples were extracted using the RNeasy plant mini kit (Qiagen, Germany). Genomic DNA was removed by DNase treatment (Invitrogen, Cat. No. AM1906). cDNA synthesis was performed using Omniscript Reverse Transcription (Qiagen, Germany). Quantitative real-time PCR (qPCR) was performed in a 10 μL reaction volume using the Takyon PCR Master Mix (Eurogentec, Belgium) on a QS6 sequence detection system (Applied Biosystems, USA). Detailed qPCR primer sequences are listed in Supplementary Data 11.
RNA-seq analyses and data processing
Total RNA samples were extracted from tomato locular tissue and whole fruits of 10-DPA stage WT, SlMBP3-KO, and SlMBP3-OX fruits using Qiagen RNeasy Plant Mini Kit (Qiagen, Germany). For each sample, three independent biological replicates were performed and RNA quality was checked by Agilent 2100 Bioanalyzer to select only samples with rin >8.5. Paired-end RNA sequencing was performed at Shanghai Majorbio Biopharm Technology Co., Ltd (Shanghai, China), using a Truseq RNA Sample Preparation Kit (Illumina) and a Hiseq 2500 platform. The cleaned reads were mapped to the MicroTom reference genome sequence (http://tomatogenome.gbfwebtools.fr/) using HISAT2 2.1.038. HTSeq was then used to calculate raw counts39. Differential expression analysis was performed with the DESeq2 R package40. Genes were declared as differentially expressed genes (DEGs) if basemean > 5 and adjusted p-value (padj) < 0.05. DEGs were subjected to the MAPMAN software (version 3.5.1)41 for functional enrichment analysis. The cell wall-related gene annotation was performed according to www.polebio.lrsv.ups-tlse.fr/ProAnnDB/index.php/.
ChIP-seq assay and ChIP-qPCR
Chromatin immunoprecipitation protocol was adapted from ref.42 with minor modifications. Briefly, tomato fruit at 10-DPA stage expressing a GFP-tagged SlMBP3 protein (SlMBP3-ORF fused with a C-terminal GFP under 35S promoter) were collected from at least six different plants, sliced into small pieces and cross-linked by immersion under vacuum (650–700 mmHg) three times, for 5 min each time, in 1% formaldehyde 1x PBS solution. ChIP assay was performed in three independent biological replicates starting with 2 g of each sample tissue. During nuclei isolation step four washes with extraction buffer II (10 mM Tris-HCl pH 7.4, 0.25 M sucrose, 1% Triton X-100, 10 mM MgCl2, 1 mM PMSF, 50 µM MG132, 1x Roche complete protease inhibitor EDTA free) were needed to reduce chloroplast contamination. Chromatin was isolated by resuspending nuclei pellets in 400 µl of nuclei lysis buffer (20 mM Tris-HCl pH 7.4, 0.5% sarkosyl, 100 mM NaCl, 2 mM EDTA, 1 mM PMSF, 50 µM MG132, 1x Roche complete protease inhibitor EDTA free) and incubation 30 min at 4 °C with gentle agitation (10 rpm). Chromatin was sheared to 300~500 bp using Bioruptor® Pico (BO1060001) with four cycles of 30 s ON and 30 s OFF. Sonicated nuclear lysate was diluted four times with dilution buffer (50 mM Tris-HCl pH 7.4, 1.25% Triton X-100, 100 mM NaCl, 2 mM EDTA, 1 mM PMSF, 50 µM MG132, 1x Roche complete protease inhibitor EDTA free) and 30 µl were collected as Input sample. The remaining lysate was incubated overnight with 2 μL of GFP antibodies (Abcam, AB290, 1:500 dilution) or 5 μL IgG antibodies (Emd Millipore, 12370, 1:200 dilution) at 4 °C, using tube rotator (10 rpm), then coupled 1 h at 4 °C with PBS-BSA washed protein A magnetic beads (Dynabeads SKU10001D). Washing and elution were performed with low salt and high salt buffers42, and the immunoprecipitated DNA was eluted twice with 100 µl of elution buffer (50 mM Tris-HCl pH 8.0, 10 mM EDTA, 1% SDS). After reverse cross-linked overnight at 65 °C, ChIPed DNA was incubated 30 min at 37 °C with 1 μL RNase cocktail (Ambion, AM2286) then 2h30 at 50 °C with 2 μL Proteinase K (Ambion, AM2546) to remove RNA and protein contaminations. The Input and ChIP DNA samples were purified using ChIP-DNA clean kit (Zymo Research D5205) and quantified by Qubit® (Qubit dsDNA HS High Sensitivity Assay Kit- Q32851). Ten nanograms of immunoprecipitated DNA was used for library construction and sequencing.
ChIP-sequencing was performed at the GeT-PlaGe core facility (INRAe Toulouse). Sequence libraries were prepared using NEBNext® Ultra™ II DNA Library Prep Kit for Illumina®. Sequencing was performed on an Illumina HiSeq 3000 to produce paired-end reads of 150 bp. Cleaned ChIP-seq reads were aligned to MicroTom reference genome (http://tomatogenome.gbfwebtools.fr/) using BWA (Burrows–Wheeler Aligner)32 with default parameters, and only uniquely aligned reads were retained. Enriched peak regions in the non-redundant mapped reads were identified by MACS2 v1.4.2 (effective genome size = 770 Mb, p-value cutoff = 1.00e-05)43. Only the peaks overlapping with a gene in 3-kb upstream region of ATG were considered for further analysis, and the enriched motifs were analyzed by MEME-ChIP software (http://meme-suite.org)44. For ChIP-qPCR, the tomato fruit samples were harvested from six different plants and three biological replicates were performed. The primers used for ChIP-qPCR were designed around CArG-box in the enriched regions and are listed in Supplementary Data 11. The fold enrichment was calculated by comparing the Ct values of triplicate measurements between transgenic and wild-type plants.
Heatmap generation
Transcriptomic profiling of selected genes was extracted from the Tomato Expression Atlas22. The expression pattern in locular gel and pericarp tissue at different development stages was determined for 24 cell wall-related genes shown to be direct targets of SlMBP3. The normalized expression values associated with each tissue and stage were extracted from the TomExpress v2020 transcriptomic platform45. Heatmap representations were performed with the ComplexHeatmap R package46. The distance used for the clustering is based on the Pearson correlation, which clusters genes according to their expression pattern (log2 expression value+1).
Reporting Summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
