A mutant allele of the flowering promoting factor 1 gene at the tomato BRACHYTIC locus reduces plant height with high quality fruit.

Reduced plant height due to shortened stems is beneficial for improving crop yield potential, better resilience to biotic/abiotic stresses, and rapid crop producer adoption of the agronomic and management practices. Breeding tomato plants with a reduced height, however, poses a particular challenge because this trait is often associated with a significant fruit size (weight) reduction. The tomato BRACHYTIC (BR) locus controls plant height. Genetic mapping and genome assembly revealed three flowering promoting factor 1 (FPF1) genes located within the BR mapping interval, and a complete coding sequence deletion of the telomere proximal FPF1 (Solyc01g066980) was found in the br allele but not in BR. The knock-out of Solyc01g066980 in BR large-fruited fresh-market tomato reduced the height and fruit yield, but the ability to produce large size fruits was retained. However, concurrent yield evaluation of a pair of sister lines with or without the br allele revealed that artificial selection contributes to commercially acceptable yield potential in br tomatoes. A network analysis of gene-expression patterns across genotypes, tissues, and the gibberellic acid (GA) treatment revealed that member(s) of the FPF1 family may play a role in the suppression of the GA biosynthesis in roots and provided a framework for identifying the responsible molecular signaling pathways in br-mediated phenotypic changes. Lastly, mutations of br homologs also resulted in reduced height. These results shed light on the genetic and physiological mechanisms by which the br allele alters tomato architecture.

INTRODUCTION

Tomato (Solanum lycopersicum) is the most valuable horticultural crop worldwide (Food and Agriculture Organization of the United Nations, 2016) and provides essential micronutrients (U.S. Department of Agriculture, 2016). Fresh‐market and processing tomatoes are the two most widely consumed types of tomatoes and account for more than $2.6 billion in annual farm cash receipts in the United States alone, with fresh‐market tomatoes accounting for more than $1.2 billion (U.S. Department of Agriculture, 2016).

In particular, large‐fruited fresh‐market tomatoes cultivated in the United States (also called round tomatoes or beefsteak tomatoes) represent a unique type (fruit class) of tomato that is bred for direct consumption (Bhandari & Lee, 2021). These tomatoes account for more than $600 million per year in farm cash receipts (Florida Tomato Exchange, 2022). In this type of tomato fruit class, fruits that are medium‐sized (medium fruit size > 5.715 cm in diameter; U.S. Department of Agriculture, 2005) or larger and those with high exterior standards (i.e., flawless skin and perfectly symmetrical fruit shape) are generally classified as marketable and are desirable in the fresh‐market tomato industry and sold in the consumer market.

Most fresh‐market tomato varieties have determinate vines (driven by the homozygous sp allele at the SELF‐PRUNING locus; Barton et al., 1955; Pnueli et al., 1998). However, their stems grow off of conventional raised plastic beds (typically 71.12 cm in width), which expose the fruit/vegetative tissue to the soil unless the stems are tied on vertical stakes (e.g., Figure 1d in Lee et al., 2018). The displacement of plants, especially fruit laying on the soil, significantly reduces fruit yield and quality via damages from human activity, machinery, and soil‐borne pathogens (Adelana, 1980; University of Florida, Institute of Food and Agricultural Sciences [UF/IFAS], 2022). Thus, the majority of US large‐fruited fresh‐market tomatoes are grown with the support of stakes and ties on raised plastic beds in the open field cultivation system. However, the shift toward farm machinery to increase productivity and market value (Pardey et al., 2016; Zahara & Johnson, 1981) necessitates the development of a ground cultivation system for shortened plants, which does not require stakes and ties on raised plastic beds to enable mechanical harvesting (discussion in Gardner & Davis, 1991; Florida Tomato Committee, 2022; Frasca et al., 2014; Kemble et al., 1994a; Kemble et al., 1994b). Essentially, a reduced plant height enables a higher plant density, which can compensate for the loss in vertical space utilization, and has the potential to increase commercially acceptable yield potential (Ozminkowski et al., 1990; Scott et al., 2010). Therefore, it would be desirable to develop shortened fresh‐market tomato varieties that maintain the marketable fruit size for commercially acceptable yield potential.

FIGURE 1

DNA sequence variation in a flowering promoting factor 1 gene (FPF1, Solyc01g066980) at the BRACHYTIC (BR) locus causes a reduced tomato height. (a) Diagram of the BR locus. A scaffold (a sequence assembly carrying genomic DNA from the br genetic interval, which is 481‐kb in size; scaffold ID 143086) spanning the BR locus (175‐kb) was compared to the genome of Heinz 1706. Numbers and black marks indicate tomato genes (e.g., Solyc01g066915 according to the annotation ITAG4.0; Fernandez‐Pozo et al., 2015). Three FPF1 genes are shown by black‐filled marks. Red arrow indicates a complete coding sequence deletion in a FPF1 gene in the br allele. Black arrows indicate sequence polymorphisms in introns. The prefix brM indicates molecular markers used for fine mapping. brM10 and brM12 indicate the beginning and end of the mapping interval, respectively, supported by two different recombinant inbred line (RIL) populations and brM11 indicates a recombination breakpoint supported by a single RIL population. C indicates the direction toward the centromere. (b) Generation of Solyc01g066980 mutations by the CRISPR‐Cas9 using two independent single‐guide RNAs (sgRNA1 in red and sgRNA2 in green). Black horizontal lines and a block icon indicate untranslated regions and the coding sequence of Solyc01g066980 at the BR allele, respectively. A complete coding sequence deletion found in the br allele is indicated by a dashed rectangle. Nucleotide sequences of Solyc01g066980 mutant alleles br.8.1 and br.8.2 are shown with the corresponding amino acid sequences, and deletions are indicated by red or green dashes. The sequence gap length is shown in parentheses. (c) A mature transgene‐free, homozygous mutant plant (br.8.2 ; right) was compared to its wild‐type large‐fruited fresh‐market tomato Fla. 8059 (wild‐type allele BR/BR; left). Fla. 8059 is one of the two parents of a commercial hybrid (F1), Tasti‐Lee™ (Bejo Seeds, Oceano, CA), currently in the US market. The soil level in a 5‐gallon soil bag is indicated by horizontal dashes. Each plant was supported by seven strings. White vertical bar represents 50 cm. (d) Representative picture of fresh‐market tomato plants with the support of stakes and ties on raised plastic beds. Three plants of each genotype are shown. An approximation of the plant height is indicated by horizontal dashes. (e) The stem lengths of 6‐week‐old plants were measured. WT, wild‐type. Mutants are transgene‐free, homozygous M2 generation. The n value represents the total number of plants evaluated. Statistical significance is indicated by ***P < 0.001, as determined by one‐way ANOVA in conjunction with a two‐tailed Tukey's HSD multiple comparison test. Error bars indicate 95% confidence intervals. (f) Scanning electron microscope images of the first internode of WT (left image) and mutant (right) plants. White bar indicates 100 μm.

A shorter stem driven by shortened internodes is a detectable phenotypic property (phenotype) of plants typically deficient in endogenous gibberellin (GA) biosynthesis or defective in perception of GA (Binenbaum et al., 2018; Yamaguchi, 2008). In tomato, multiple genes essential for GA synthesis are functionally characterized (Carvalho et al., 2011; Illouz‐Eliaz et al., 2019; Koornneef et al., 1990; Li et al., 2012). Similarly, this phenotype can be observed in tomatoes with different hormonal mutations, such as mutations in genes involved in brassinosteroid (Bishop et al., 1999; Li et al., 2016; Martí et al., 2006; Nie et al., 2017) or indole‐3‐acetic acid syntheses (Higashide et al., 2014; Leyser, 2018; Zhao, 2018). However, such mutation‐derived phenotypes with reduced height have not been utilized commercially, particularly in large‐fruited fresh‐market tomato varieties, often due to extreme reductions in the fruit size (medium‐sized or larger fruits are classified as typical marketable fruits in large‐fruited fresh‐market tomato in the United States; Florida Tomato Committee, 2022) and marketable yield, which were observed for tomato germplasm releases and varieties (Scott & Harbaugh, 1989; J.W. Scott and R.G. Gardner, personal communications; a comparison of fruit size shown in Figure S1), and experimental tomato lines (Higashide et al., 2014; Liu et al., 2018; Srivastava & Handa, 2005) that harbor mutant allele(s).

Unlike typical dwarfism in which all organs are reduced in size, the plant BRACHYTIC (BR) gene typically influences stem growth, which was observed in both monocots, such as barley (Rossini et al., 2006) and maize (Balzan et al., 2018; Knöller et al., 2010; Multani et al., 2003; Xing et al., 2015), and dicots, such as chickpea (Gaur et al., 2008) and soybean (Cui et al., 2007). Mutations in an ATP binding cassette type B auxin transporter decreases internode length in br2 mutant maize (Multani et al., 2003; Xing et al., 2015). In peach, a GA receptor GID1c mutant causes the brachytic dwarfism trait, but does not impair fruit development (Hollender et al., 2016). Collectively, these studies suggest that different mechanisms/genetic loci affect organ development.

Tomato BR is a known genetic locus and defined by phenotype (Balint‐Kurti et al., 1995; Barton et al., 1955; MacArthur, 1931), although its DNA sequence is not published yet. Plants with long (hereafter referred to as the wild‐type phenotype), intermediate, and short (br phenotype) stems have homozygous BR allele (BR/BR), heterozygous alleles, and homozygous br allele (br/br), respectively (Lee et al., 2018). The br is the primary determinant of the shortened internode phenotype, which inevitably reduces stem length, and is a focus of fresh‐market tomato breeding programs (Frasca et al., 2014; Scott et al., 2010; Tigchelaar, 1986; UF/IFAS, 2022). Furthermore, the br has been used to generate a shorter architecture in indeterminate fresh‐market tomato lines facilitating horticultural practices, as the production can be on shorter vertical stakes rather than the tall growing indeterminate tomatoes but still maintain indeterminate habit for its benefits (Panthee & Gardner, 2013; North Carolina State University [NCSU], 2022).

Notably, there is no evidence for a negative correlation between marketable fruit yields and the br (Frasca et al., 2014; Gardner & Davis, 1991). However, in the absence of a completely known pedigree for tomato materials, the effect of the genetic background on quantitative yield traits of br plants is unclear. Moreover, although mapping of the tomato BR locus was reported (Lee et al., 2018), the genetic and molecular characteristics have not been reported in a peer‐reviewed forum. Therefore, knowing the genetic basis of br would contribute to knowledge‐based breeding, by which an ideotypic plant architecture can be designed and the appropriate gene targets can be manipulated to create such improved varieties.

Therefore, br is a key target for the optimization of the fresh‐market tomato architecture for various breeding goals. In this study, we mapped and identified sequence variants at the BR locus in large‐fruited fresh‐market tomatoes (Fla. 8044 [BR/BR], Fla. 8624 [BR/BR], Fla. 8834 [br/br], and Fla. 8916 [br/br]) and found that one of three flowering promoting factor (FPF) genes within the fine mapping interval is directly responsible for reduced plant height, as validated using CRISPR‐Cas9‐driven mutants. Using CRISPR‐Cas9‐driven br mutants and a pair of sister lines (one with and the other without the br allele, respectively) derived from conventional crossing/selection methods, we evaluated the effects of the br trait on fruit yields and horticultural traits of a large‐fruited fresh‐market tomato in two growing seasons in the greenhouse or open field in the Southeastern United States, where this tomato was originally bred. We compared gene expression patterns across tissues, genotypes, and GA treatments by a network analysis and evaluated the role of FPF genes in br‐mediated traits. We created new sources of fresh‐marketable tomato plants harboring mutated versions of FPF homologs adjacent to br. We discussed the implications of this gene for future breeding approaches.

RESULTS AND DISCUSSION

A DNA sequence polymorphism in a flowering promoting factor 1 gene is present in the fine‐mapping interval

The tomato BR was mapped to a 763‐kb interval on chromosome 1 (Lee et al., 2018). To narrow this interval, three recombinant inbred line (RIL) populations segregating for plant height were used for further mapping of the BR locus. Four newly identified RILs had crossing overs mapping to two sites within the previous BR interval (Table S1). The BR locus determined by two different populations corresponded to a 174.8‐kb interval (spanning molecular markers brM10 and brM12) in the genome (SL4.0 version, Tomato Genome Consortium 2012) of the fully sequenced tomato Heinz 1706 variety (BR/BR) (Figure 1a). An additional recombination breakpoint was identified at brM11 in three RILs derived from a single population, which further narrowed down the interval (spanning molecular markers brM11 and brM12). However, no recombinant(s) were identified in an independent population that could confirm the results.

In the genome annotation ITAG4.0 (Fernandez‐Pozo et al., 2015), there were 10 predicted genes within the 174.8‐kb interval in which BR was fine mapped. To generate compelling gene candidates, we characterized the interval containing the br allele derived from a br plant. Using the Illumina HiSeq system and 10× Genomics Chromium library, we analyzed the genome sequence of a single br plant (tomato line 3040717). Tomato line 3040717 is a large‐fruited, determinate (sp/sp) inbred breeding line that carries the homozygous br allele derived from conventional crossing/selection methods (UF/IFAS, 2022). Additionally, five inbred lines (three lines with BR/BR and two with br/br) were sequenced using the Illumina HiSeq system to obtain high density DNA sequence polymorphisms, which include single nucleotide polymorphisms (SNPs), insertions and deletions. Both the assembly (scaffold ID 143086 from the 10× Genomics Chromium library, 481‐kb in size) and alignment data of br plants showed very high sequence similarity in the interval when compared to that of the wild‐type plants, except for a complete coding sequence deletion of an FPF1 gene (Solyc01g066980) (Figure 1a, Figure S2a, S2b). No sequence polymorphisms, except the deletion in that gene, were found in the protein‐coding exons in the predicted genes within the interval. Sequence polymorphisms found in the intergenic regions of three other genes (Solyc01g066915, Solyc01g066940, and Solyc01g067020) were not consistently detected across the independently sequenced inbred lines (Figure S2c). Interestingly, two additional FPF1s (Solyc01g066950 and Solyc01g066970) were present immediately adjacent to the FPF1 (Solyc01g066980), and the DNA sequence of each gene of the two phenotypes was identical.

Gene‐edited deletion mutations in an causes a reduced tomato height

The absence of DNA polymorphisms in annotated genes, except in a single FPF1, and the observed deletion of the coding sequence of the telomere‐proximal FPF1 (Solyc01g066980) in the mutant allele of the BR gene (br) with three intact FPF1 genes in the wild‐type allele of BR suggest that the telomere‐proximal FPF1 carrying a single exon is the only candidate for the br allele‐mediated reduced plant height. Furthermore, expression polymorphism of one or more of FPF1s may possibly contribute to a similar or identical br‐mediated phenotypic change.

To validate whether Solyc01g066980 is the causative gene for stem length differences between br and wild‐type plants, we used the CRISPR‐Cas9 system to create mutation(s) in Solyc01g066980 (Figure 1b). Expectedly, plants with mutations in Solyc01g066980, which no longer produced the wild‐type Solyc01g066980 protein, were significantly shorter than wild‐type plants (Figures 1c–e and S3a), indicating that Solyc01g066980 is directly responsible for reduced plant height. Scanning electron microscopy (SEM) images of epidermal cells of the first internode showed approximately 30% increase in the number of cells within a 1‐mm2 area in the mutant compared to that of the wild‐type (Figure 1f). Furthermore, knocking‐out the same gene, Solyc01g066980, in the processing tomato M82 also shortened its architecture (Figure S4).

The br allele in tomato lines used in this study was derived from the inbred tomato line 823125‐1‐3 with the br phenotype (Gardner, 2000). We further evaluated the genomic sequences of nine tomato accessions known to show br phenotype and maintained by the Tomato Genetics Resource Center (https://tgrc.ucdavis.edu) to verify if other accessions harbor similar genetic lesions based on the alignment of whole‐genome sequencing (WGS) reads. Of the nine accessions, five showed the sequence deletion of Solyc01g066980 unique to the br allele, whereas the remaining four accessions did not show such deletion (Figure S5, Table S2). All nine accessions were S. lycopersicum, which is a modern (domesticated) tomato species. In four accessions without the sequence deletion of Solyc01g066980, other loci and/or expression polymorphism of one or more of FPF1s may confer a similar or identical br phenotype.

brachytic reduces yield but does not negatively impact the average fruit and heaviest‐fruit weights

Large fruit size is a particularly important trait for fresh‐market tomato production, especially in the United States, based on market demand. Fruit weight and size are analogous to yield, because fresh‐market tomato fruits can be sold in packages that meet net standard weight and fruit size requirements as per USDA market standards (Guan et al., 2015; Scott et al., 2013; U.S. Department of Agriculture, 2005). Thus, there is a demand for fruits classified as larger than medium in size including extra‐large fruits (fruit size >6.985 cm in diameter), where size is defined in accordance with the shipping point and market inspection instructions for tomatoes (U.S. Department of Agriculture, 2005). The br‐mediated reduced height phenotype has been used in tomato breeding programs because of the lack of significant negative impacts of the br on the marketable fruit size and yield, which has been observed in such programs. Therefore, combining high yields with this source of a reduced plant height may be a feasible objective for breeding. However, other factors may contribute to the observed phenotypes, especially linkage disequilibrium around br and/or genetic recombination (i.e., background genes) during artificial crossing. Accordingly, the horticultural traits, fruit weights, and yield of CRISPR‐Cas9‐driven br mutants were tested.

Plant FPF1 typically promotes the flowering time—for example, Arabidopsis (Kania et al., 1997; Melzer et al., 1990), cotton (Gossypium L.) (Wang et al., 2014), rice (Guo et al., 2020), tobacco (Nicotiana tabacum) (Smykal et al., 2004). In two separate greenhouse trials, we did not observe statistically significant differences in the days to first flower and leaves to first flower among the three genotypes (the wild‐type and two mutants) (Figures 2a,b and S3b,c). Furthermore, there was no difference in the days to first fruit (Figures 2c and S3d,e). The differences in the fruit Brix and acidity among genotypes were not significant (Figure 2d).

FIGURE 2

The days to first flower or first fruit, and Brix/acidity levels of the brachytic mutant tomato. Days to first flower is defined as the number of days from sowing to the first full bloom (i.e., when petals create a 180° angle) (a), leaves to first flower the number of leaves produced before initiation of the primary inflorescence (b), days to first fruit is defined as days from sowing to the first fruit with 1 cm in diameter (c), and fruit Brix (%) and acid (%) (d). WT, wild‐type. br.8.1 and br.8.2 , transgene‐free, homozygous M2 generation mutant lines. The n value represents the total number of plants for each genotype evaluated during each trial; ns indicates no significant difference (ANOVA at P > 0.05) was found between any genotypes, except for trial 2 in c; the Welch's test (P > 0.05) was used for the comparison in trial 2 in c. Error bars indicate 95% confidence intervals. GH, greenhouse (a through d).

There was wide distribution of fruit weights in the representative large‐fruited fresh‐market tomato variety at the time of harvest (Figure 3a). We did not find differences in the average weights of all fruits harvested or in the heaviest‐fruit weight among the three genotypes (Figure 3b,c). Additionally, there were no differences in the average fruit weight stratified by fruit color among all the genotypes (Figure 3d,e).

FIGURE 3

brachytic reduced fruit yields of the large‐fruited fresh‐market tomato, with no adverse effects on fruit weights. (a) Distribution of fruit weights in wild‐type (WT) and mutant plants, br.8.1 and br.8.2 . The entire fruit harvested from each of the three different genotypes tested in a greenhouse trial are displayed. Medium‐sized fruit was defined as described by U.S. Department of Agriculture, 2005. Number of fruits was calculated for a 1 g fruit weight window. (b) Average fruit weight (g) per plant. (c) Average heaviest fruit weight (g). An example of an extra‐large fruit [as per the USDA standard size designations for tomatoes (U.S. Department of Agriculture, 2005)] harvested from br.8.2 (right). White bar represents 2 cm. (d) Average green fruit weight (g) and (e) average red [breakers or later fruit color classification as per the equipment catalog for fresh and processed product inspections (U.S. Department of Agriculture, 2017)] fruit weight (g). (f) Total number of fruits per plant; (g) total fruit yield (kg) per plant; (h,i) yield of medium‐, large‐, or extra‐large‐sized fruits per plant evaluated in the greenhouse and field, respectively. (j) yield of extra‐large fruit evaluated in the field. (k) Dry weight of above ground tissues per plant; (l) harvest index. Mutants are transgene‐free, homozygous M2 generation (a through l). The n value represents the total number of plants for each genotype evaluated during each trial (a through l). Statistical significance (*P < 0.05, **P < 0.01) was determined by one‐way ANOVA in conjunction with a two‐tailed Tukey's HSD multiple comparison test (a through l). ns indicates no statistical difference (ANOVA at P > 0.05) among any of the genotypes (b through l). Error bars indicate 95% confidence intervals (b through l). Negative numbers (g, h, i, j, and l) indicate % decreases in the mutant compared to WT. GH and field trials indicate the evaluation of traits in the greenhouse and field, respectively (a through l).

The number of fruits tended to be lower in the two mutants than that in the wild‐type plant in the two growing seasons in the greenhouse trials (Figure 3f). Both mutants showed lower total fruit yields per plant than that of the wild‐type (Figure 3g). With respect to the yield of medium‐, large‐, or extra‐large‐sized fruits per plant, genotype effects were not consistently significant in the two growing seasons (Figure 3h). However, in greenhouse trial 1, both mutants showed a significant reduction in the medium‐sized or larger fruit yields. A reduction in yield was observed in greenhouse trial 2 for both mutants (32% or 45% reduction compared to that in the wild‐type). Similar results were observed in concurrent field evaluations (Figure 3i). Furthermore, a clear negative impact of the br mutation on the extra‐large fruit yield was found in the field trials (Figure 3j). Interestingly, there were no differences in the dry weight of aboveground tissues, including stems and leaves (Figure 3k), which resulted in a reduction in the harvest index (HI; total fruit yield/dry weight) in mutants (Figure 3l). The lack of a difference in dry weight between the wild‐type and mutants may be attributed to the denser vegetative tissues observed in br plants than that in the wild‐type plant.

Association between high extra‐large‐sized fruit yield of ‐mediated shortened tomato and other genes

Marketable fruit yield losses due to the br allele have not yet been reported. In contrast, we observed reductions in the extra‐large and medium‐sized or larger fruit yields of CRISPR‐Cas9‐driven br mutants. This raises the intriguing possibility that background genes may have compensated for the negative impact of br on marketable fruit yield by linkage disequilibrium and/or genetic recombination. We further evaluated variation in tomatoes with BR/BR or br/br alleles. We used two sister lines developed from Fla. 8653 (BR/BR) × Fla. 8916 (br/br), with a completely known pedigree to transfer the br allele to a US‐adapted large‐fruited fresh‐market cultivar (UF/IFAS, 2022) for yield evaluations. Unlike the CRISPR‐Cas9‐driven single gene mutant, a sister line can carry several to many different genes (traits) regardless of whether those can be easily observed by researchers when it compared to its mate (i.e., BR sister line vs. br sister line in Figure 4) other than the trait of interest. We focused on the extra‐large and medium‐sized or larger fruit yields, which are routinely evaluated by tomato breeders during germplasm/cultivar development (Hutton et al., 2015, 2017; Scott et al., 2008, 2009). Interestingly, during the yield evaluation, no negative impact of the br was detected compared to the line with the BR and control Fresh‐market tomato ‘Sanibel’ (BR/BR) (Figure 4). The yield of extra‐large fruit in the br sister line was significantly (P < 0.05) higher than that of the BR sister line in the two growing seasons (Figure 4b). Earlier observations by breeders are generally consistent with these results. This variation could explain the current widespread use of the br‐derived phenotypes. While the br reduces the yield of marketable fruit (i.e., large size fruit) in large‐fruited fresh‐market tomatoes, our data clearly revealed that artificial selection contributes to commercially acceptable yield potential in br tomatoes.

FIGURE 4

Field evaluation of extra‐large and medium‐, large‐, or extra‐large‐sized fruit yields of a pair of sister lines with or without the brachytic (br) allele during two growing seasons. (a) Yield of medium‐, large‐, or extra‐large‐sized fruits per plant; (b) yield of extra‐large fruit per plant. A pair of sister lines (BR/BR or br/br) derived from conventional crossing/selection methods using parental lines homozygous for either BR or br alleles at the BR locus. Sanibel (BR/BR) is a commercial large‐fruited fresh‐market tomato cultivar currently in the U.S. market (Seminis, MO). The n value represents the total number of plants for each genotype evaluated during each trial; ns indicates no statistical difference (ANOVA at P > 0.05) among genotypes; *P < 0.05 determined by one‐way ANOVA in conjunction with a two‐tailed Tukey's HSD multiple comparison test. Error bars indicate 95% confidence intervals.

Five homologs in tomato show tissue‐specific expression

To identify the FPF gene family in Solanaceae, we performed a hidden Markov model (HMM) search using the PFAM FPF model against the 11 Solanaceae annotated protein datasets, which include three tomato species, one modern (domesticated) (S. lycopersicum) and two wild tomato species (Solanum pimpinellifolium and Solanum pennellii). We identified 57 protein sequences (including five modern tomato sequences) matching the model (Table S3). For each species, multiple sequences were identified in the datasets used in this study (ranging from three FPFs in Capsicum annuum cv. CM334 to eight in Nicotiana N. benthamiana). A maximum likelihood phylogenetic analysis revealed that five modern tomato sequences can be clustered into two categories (Figures 5a and S6). One contained all three FPFs on chromosome 1, and the other category clustered all three tomato species, including a single modern tomato gene, Solyc06g005530, close to a single terminal branch. Interestingly, both the wild tomatoes had five FPF1s, which is the same number as that found in the modern tomato. However, the modern tomato and its closest relative S. pimpinellifolium carried three FPF1s on chromosome 1, while S. pennellii carried four FPF1s on chromosome 1, implying molecular divergence in the FPF1 family in Solanum.

FIGURE 5

Network analysis of gene expression patterns across tissues, genotypes, and gibberellic acid (GA) treatments. (a) Maximum likelihood phylogenetic tree of Solanaceae flowering promoting factor 1 (FPF1) families. Colored dots represent five modern tomato (S. lycopersicum) FPF1s identified by sequence similarity to the families in Solanaceae species. Wild tomatoes (S. pimpinellifolium and S. pennellii) are indicated by asterisks. Scale bar represents 1.0 substitutions per site. A full version of this tree with individual annotations and bootstrap support values can be found in Figure S6 and Data S1. (b) Expression of tomato FPF1s in different tissues. Each gene is color‐coded according to figure a. WT, wild‐type. M, transgene‐free, homozygous M2 generation mutant line. (c) Comparison of differential gene expression between genotypes, tissues, or treatments (untreated vs. GA3 treated). Numbers in parentheses indicate up or down‐regulated genes (e.g., four genes were upregulated in the mutant fruit compared to the wild‐type) at a false discovery rate (FDR) of <0.05. (d) Module‐level differential expression (left, tissue‐specific co‐expression modules; right, gene expression data adjusted for tissue with respect to GA3 treatment) was found (e.g., genes in a module GTM46 were up‐regulated in the mutant tissues compared to the wild‐type). Two tomato FPF1s found in modules are color‐coded according to figure a. Gibberellin metabolism gene(s) are listed in the corresponding module. Each module passes the FDR < 0.05 threshold in at least one of the multiple comparisons. (e) Gene Ontology enrichment analysis (top, tissue‐specific ontologies; bottom, ontology adjusted for tissue with respect to the GA3 treatment). The panel shows the highest‐enriched set of differentially expressed genes. Red line represents the threshold P Bonferroni = 0.05.

To obtain an overview of the expression profiles of the five tomato FPF1s, RNA‐seq libraries were constructed from different tissue types, the first internode (stem), leaf, and root at the 6‐week‐old growth stage (the growth stage used in conventional brachytic phenotyping; Lee et al., 2018) and small green fruit tissue in both phenotypes (wild‐type and mutant [br.8.2 ]) for sequencing. Additionally, first internodes collected 3 h after the GA3 treatment at the 6‐week‐old stage were used for library construction. Comparing the expression profiles among homologs, both BR gene (Solyc01g066980) and its immediately adjacent gene Solyc01g066970 were expressed (Figure 5b). Notably, both genes were highly expressed in roots and their expression levels were not significantly affected by the GA3 treatment. The other three homologs had low expression levels in most or all tissue types.

Identification of modules of coexpressed genes across tissues, genotypes, and the GA3 treatment by network analysis

A comprehensive catalog of expressed genes was identified from the same RNA‐seq libraries outlined above. In fruits, 40 differentially expressed (DE) genes (up‐ or down‐regulated combined; <0.10% of genes in the genome annotation ITAG4.0) at a false discovery rate (FDR) of <0.05 were found in a comparison between genotypes (Figure 5c, Table S4). The number of significant DE genes between GA3‐treated and untreated plants was high in the stem tissues for both genotypes.

To characterize the biological pathways related to expressed genes, we performed a weighted gene coexpression network analysis (WGCNA) and identified several genotype‐specific (Figure 5d, Table S5) or the GA3 treatment‐specific coexpression modules (Figure 5d, Table S6). In an analysis of genotype‐specific modules (left panel in Figure 5d), the module GTM46 included highly up‐regulated genes in the mutant that was shared across the tissue types, while a comparison of fruit tissues showed less positive correlation. An analysis of tissue‐adjusted data (i.e., identification of modules driven by the GA3 treatment effect in stem) revealed 30 modules (right panel in Figure 5d). In particular, modules GAM12 and GAM35 included highly up‐ and down‐regulated genes, respectively, in the GA3‐treated stem for both genotypes, reflecting the effect of the GA3 treatment, respectively. Furthermore, those two modules were enriched for distinct Gene Ontology (GO) terms (bottom panel in Figure 5e, Table S6); module GAM12 was enriched for RNA modification and module GAM35 was enriched for integral component of the membrane. Interestingly, the module GAM14 included up‐regulated genes in the GA3‐treated mutant compared to the GA3‐treated wild‐type, while comparisons between with and without the GA3‐treatment for any genotype showed a converse correlation. We found that 319 genes were assigned to the module GAM14, but enrichment of significantly DE genes in the module GAM14 did not pass a specified threshold (P Bonferroni = 0.05) in this study. Highly down‐regulated genes in a particular GA3‐treated genotype were also found in modules GAM17 and GAM28. None of the FPFs were included in modules with Z‐values ≥4.0 or ≤4.0.

GA is a key regulator of plant growth and phase transitions during plant development (Binenbaum et al., 2018; Yamaguchi, 2008), as it stimulates plant stem elongation. In Arabidopsis, the overexpression of FPF1 results in an internode elongation in response to GA3 (Kania et al., 1997). In peach, the degree of a GA receptor GID1c silencing corresponds to the degree of the brachytic dwarfism (Hollender et al., 2016). In rice (Oryza sativa), antagonistic regulatory genes (one with an unknown function and a zinc‐finger transcription factor annotation) in response to GA3 determine internode elongation (Nagai et al., 2020). As both rice genes showed relatively low sequence identity to tomato sequences in the Phytozome (https://phytozome-next.jgi.doe.gov) and Solanaceae Genomics Network (https://solgenomics.net) databases based on nucleotide and protein blast searches (https://blast.ncbi.nlm.nih.gov/Blast.cgi), we were unable to identify highly homologous tomato genes. Furthermore, the rice FPF1 genes partially match tomato FPF1s (https://phytozome-next.jgi.doe.gov). However, peach GID1c and the tomato gene Solyc01g098390 showed 80% identity. The Solyc01g098390 was assigned to a module GAM35, which included down‐regulated genes in the GA3‐treated mutant. Taken together, these results may reflect the divergence of stem elongation‐related genes along with the monocot‐dicot divergence and the shared biochemical mechanisms of these genes in closely related species.

In an analysis of tissue‐adjusted data, 2‐oxoglutarate (2OG) and Fe (II)‐dependent oxygenase superfamily protein (Solyc06g082030) was assigned to the module GAM12 that included highly up‐regulated genes in the GA3‐treated stem tissues for both genotypes, whereas oxygenase family protein (Solyc03g025490) was assigned to the module GAM3 that was rather weakly associated with the GA3 treatment. Another gene, Solyc07g061720 (gibberellin 2‐beta‐dioxygenase 1), was assigned to the module GAM35 with highly down‐regulated genes in the GA3‐treated stem for both genotypes. Taken together, genes including Solyc06g082030 and Solyc07g061720 in both modules GAM12 and GAM35 may establish GA metabolism pathway and/or hormonal regulation (e.g., homeostasis).

In our study, stem elongation in young br plants was observed in response to both the GA3 and GA4+7 treatments (Figure S7). In addition, more than 10,900 DE genes were found in the stem tissues of the mutant in a comparison between with and without the GA3 treatment. At least two FPF1s were expressed predominantly in the roots. The rice FPF1‐like protein 4 preferentially accumulated in the leaf blade and roots in the seedling, and altered the root system architecture (Guo et al., 2020). An analysis of genotype‐specific modules revealed that two genes encoding oxygenase (2‐oxoglutarate [2OG] and Fe(II)‐dependent oxygenase superfamily protein, Solyc06g082030; oxygenase family protein, Solyc03g025490) were assigned to a heme binding‐enriched module, which included both Br (Solyc01g066980) and its homolog (Solyc01g066970) (module GTM3 in the left panel in Figure 5d and top panel in Figure 5e). GA biosynthetic steps were observed in Arabidopsis roots (Mitchum et al., 2006). Heme binding GO annotation suggests that the genes may be related to the catalytic oxidation reactions of organic compounds (Howe et al., 2020). As we only investigated gene expression in the stem tissue (specifically the first internode) collected 3 h after the GA3 treatment at the 6‐week‐old stage, we cannot rule out the possibility that changes in FPF expression are associated with developmental states. However, members of the tomato FPF1 family may play a role in the suppression of the GA biosynthesis in roots, which controls diverse aspects of growth and development. Furthermore, a gibberellin oxidase (Solyc01g093980) was up‐regulated (FDR < 0.05) in the stem of the mutant compared to that of the wild‐type stem (module GTM14), indicating that this gene could be responsible for the shortened stem length.

Mutated br homologs present new sources of a reduced plant height

Considering the observed sequence variation and expression patterns of the FPFs adjacent to the br on chromosome 1, we investigated phenotypes associated with mutated versions of the two br homologs, Solyc01g066950 and Solyc01g066970. Using a single‐guide RNA targeting a sequence region only differentiated by a single nucleotide, three different mutants were obtained simultaneously (Figure 6a): br.7 , with a 1‐bp insertion in Solyc01g066970; br.57.1 , with a 5‐bp deletion in both Solyc01g066950 and Solyc01g066970; br.57.2 , with a 1‐bp insertion in Solyc01g066950 and a 5‐bp deletion in Solyc01g066970. None of these mutants had DNA sequence variation in BR (Solyc01g066980). Expectedly, all three mutants showed significantly reduced height (Figure 6b). As the number of knocked out genes increased, the stem length reduced accordingly. Further experimental validation is needed to determine whether a reduced height can be obtained solely by mutation(s) in the centromere‐proximal FPF (Solyc01g066950), which showed a relatively low expression level detected mainly in roots. However, our collective findings indicate that multiple br homologs confer a br plant‐like shortened stem length. Targeting homologous sequence for simultaneous gene editing via one sgRNA using CRISPR‐Cas9 system required considerably less time to generate double mutants in this study. However, this method does not necessarily allow the identification of sole gene contribution unlike single gene mutant.

FIGURE 6

Reduced plant height in plants harboring mutated brachytic homologs. (a) Diagram of two flowering promoting factor 1 (FPF1) genes (Solyc01g066950 and Solyc01g066970), the centromere‐proximal homologs of brachytic. sgRNA3 in blue indicates a single‐guide RNA used in the CRISPR‐Cas9 system, which targeted a sequence region with only a single nucleotide difference between the two homologous FPF1s (i.e., “T” at 68,005,223 bp on Solyc01g066950 and “C” at 68,057,560 bp on Solyc01g066970). Black horizontal lines and block icons indicate untranslated regions and the coding sequences, respectively. The first nucleotide position of each start codon is given. Nucleotide sequences of three different mutants (br.7 , br.57.1 , and br.57.2 ) are shown with the corresponding amino acid sequences, and deletions and insertions are indicated by blue dashes and underlines, respectively. The sequence gap length between two genes is shown in parentheses. WT, wild‐type. (b) The stem lengths of 6‐week‐old plants. Mutants are transgene‐free, homozygous M2 generation. The n value represents the total number of plants for each genotype evaluated. **P < 0.01 based on one‐way ANOVA in conjunction with a two‐tailed Tukey's HSD multiple comparison test. Error bars indicate 95% confidence intervals.

High levels of genetic variation (e.g., copy number variation of DNA segments [CNV]) have been observed in plant genomes (Alonge et al., 2020; Cao et al., 2011; Hanikenne et al., 2013; Iovene et al., 2013; McHale et al., 2012; Swanson‐Wagner et al., 2010; Zheng et al., 2011), and emerging evidence indicates that CNVs mediate a number of valuable crop traits—for example, CNV (1 to 11 copies)‐mediated soybean cyst nematode (SCN) resistance (Cook et al., 2012; Cook et al., 2014; Lee et al., 2015; Lee et al., 2016), increased SCN resistance in high‐copy‐number (11 copies) individuals from a population of a single cultivar (Lee et al., 2016). Together with our results, this suggests further approaches to diversify the genetic resources of this large‐fruited fresh‐market tomato, such as the creation of tomato lines that carry diverse mutations in the FPF1 genes (e.g., knock‐outs of all the three FPFs on chromosome 1 and/or homologs on other chromosomes). Expression polymorphism of these genes by either gene editing‐driven knock‐out(s) or recombination event(s) may result in considerably reduced plant architectures than those obtained by single or double mutants and should be further optimized for the ground cultivation system (e.g., compact growth habit tomatoes) and/or for other breeding goals (e.g., developing tomatoes for indoor agriculture). The impact(s) of such complex mutations on the fruit marketability should be evaluated.

Our work indicates that br‐mediated reduced tomato height is attributable to a DNA polymorphism in the FPF1. Knock‐out(s) of br homologs can confer a br plant‐like short architecture. The br reduced the HI, while retaining the production of heavy fruits. Shared and divergent gene expression patterns associated with the genetic differences between genotypes (wild‐type and br plants), and between GA3‐treated and untreated tissues were observed in both genotypes. Given that the two FPF1s were highly expressed in roots and not in the stem, developmental and cell‐to‐cell signaling events from the roots to the stem are likely to drive phenotypic change(s) in br plants. The FPF1 homologs may have a wide range of functions. Genetic engineering of plant architecture is an effective approach for increasing marketable crop yield, and br plays a key role in reducing the plant height in tomato breeding programs (Frasca et al., 2014; NCSU, 2022; Scott et al., 2010; Tigchelaar, 1986; UF/IFAS, 2022). Clearly, future investigations of the genetic association between the br‐mediated reduced plant height and the marketable fruit yield, especially the extra‐large‐sized fruit yield, are necessary to successfully develop shortened fresh‐market tomato varieties.

EXPERIMENTAL PROCEDURES

Fine mapping of the locus

The BR locus was fine mapped following the same basic procedures used for the initial mapping of this locus described in our previous study (Lee et al., 2018). Further details are available in Supporting Information S1.

Genome assembly from haplotype

WGS of the tomato line 3040717, harboring the homozygous br allele, was conducted using Illumina technology. Detailed steps for genome assembly are available in Supporting Information S2.

WGS of tomato accessions obtained from the Tomato Genetics Resource Center and Fla. lines was conducted using an Illumina HiSeq instrument as described in our previous study (Bhandari & Lee, 2021). The sequencing libraries were sequenced on average of 23 Gb for each tomato.

CRISPR‐Cas9 gene editing

Guide RNAs (gRNAs) targeting the FPF1 genes were designed using CRISPR‐P (Lei et al., 2014) and CRISPR‐PLANT (Xie et al., 2014), and each of the gRNA was cloned into a binary vector following the same basic procedures described by Xie and Yang (2013) (Table S7). Duplex oligos carrying BsaI sites in binary vectors were synthesized (IDT, www.idtdna.com). The binary vector pHSN401 (www.addgene.org)‐gRNA plasmid was introduced into Agrobacterium tumefaciens (A. tumefaciens) strain LBA4404 (Takara, www.takarabio.com) according to the manufacturer's instructions. A. tumefaciens‐mediated transformations of Fla. 8059 (a parental line of “Tasti‐Lee F1” [Bejo, Seeds, Oceano, CA], which does not carry br; Scott et al., 2008; Tasti‐Lee F1 is a fresh‐market tomato cultivar currently in the US market [e.g., Publix Super Markets, Inc., www.publix.com]) were performed as described by Van Eck et al. (2019), with the following modifications in the preculture medium and selective regeneration medium steps: cotyledon explants from 7‐ to 9‐day‐old seedlings were precultured and 3 mg L−1 or 6 mg L−1 hygromycin was used.

Potential Cas9‐gRNA‐introduced mutations were examined by Sanger sequencing of PCR products and the T7 Endonuclease I assay (NEB, www.neb.com) (Figure S8a; Table S7). Details of the experimental procedures are available in Supporting Information S3. To identify homozygous transgene‐free mutants, four primer pairs targeting the Cas9 gene in the binary vector or the Hyg gene were used (Figure S8b and S8c). Potential transgene‐free mutants were further validated by WGS (Figures S8d–g). Potential off‐target sites (i.e., up to four mismatches compared to each target region) were predicted using the Cas‐OFFinder (Bae et al., 2014). A lack of off‐target activity was verified (Figure S9, Table S8).

CRISPR brachytic mutants in the M82 background were developed by the Zach Lippman laboratory. Detailed steps are available in Supporting Information S4.

Greenhouse trial

Greenhouse trials were conducted during two growing seasons at the University of Florida's Gulf Coast Research and Education Center (UF GCREC; Wimauma, FL). Young plants were grown as previously described (“Phenotype analysis” in Lee et al., 2018). Each plant was transplanted into a 5‐gallon soil bags 6 weeks after sowing. Soil bags were randomly placed at a distance 70 cm between the bags. Plants were managed as described by Hochmuth (2018). A total of 10 g of Nutri‐Leaf® 20‐20‐20 (Miller Chemical, www.millerchemical.com) was applied weekly to each 5‐gallon soil bag. Day and night temperatures were set to 26.6°C and 18.3°C, respectively. No artificial pollination or pruning was performed during the trials. No rotten fruits were observed at the time of harvest during the greenhouse trials. During the spring of 2020, seed sowing in the greenhouse (S) and fruit harvest (H) were performed on January 31 and July 9, respectively. For the second growing cycle, S and H were performed on September 3, 2020, and January 10, 2021.

The stem lengths of 6‐week‐old plants were measured, as described in our previous study (Lee et al., 2018). Days to first flower was defined as the number of days from sowing to the first full bloom (i.e., when petals create a 180° angle; Lee & Hutton, 2021). Leaves to first flower was defined as the number of leaves produced before initiation of the primary inflorescence. Days to first fruit was defined as the days from sowing to the first fruit (of 1 cm in diameter). Days to first flower and fruit color data were collected by the same individual throughout the seasons in this study. Ripe fruits were harvested and the tomato fruit Brix and acidity were measured immediately after harvest using digital refractometers PAL‐1 (ATAGO; www.atago.net). The Brix or acidity values were measured for each individual fruit sample according to the manufacturer's instructions, and the average value from replicates for each genotype was calculated.

All fruits (>0.5 cm in diameter) that developed in each seasonal trial were harvested on a single harvesting date, regardless of fruit weight, quality (e.g., irrespective of whether the fruits had defects such as cracks), or color. Fruit color (green vs. breakers; U.S. Department of Agriculture, 2017) was visually examined on the day of harvest (Figure S10). Fruits were sorted by size into four classes (i.e., any fruit smaller than medium size, medium, large, and extra‐large) using the USDA Tomato Sizer (U.S. Department of Agriculture, 2017), and descriptions of medium‐, large‐, and extra‐large‐sizes were in accordance with the shipping point and market inspection instructions for tomatoes (U.S. Department of Agriculture, 2005). For each genotype and plot, the fruit weight and number were calculated from the average values for the weight and number of fruits from the plants in each trial, respectively.

Field trial

Field trials of the CRISPR‐Cas9‐driven mutants were conducted during two growing seasons at the UF GCREC, where conditions are representative of typical field‐grown fresh‐market tomato production environments in the Southeastern United States. Plants were grown as described previously (“Phenotype analysis” section of Lee et al., 2018). A recommended fertilizer, pesticide, and irrigation program (Freeman et al., 2015) was followed throughout the growing season until harvest. In each season, the experiment was conducted using a randomized complete block design, as described in our previous study (Lee & Hutton, 2021). Six blocks were included in each season. Damaged or diseased plants were marked throughout the season, and data were not collected from such plants. During the spring of 2020, S and H were performed on January 22 and May 23, respectively, and For the second growing cycle during the fall of 2020, S and H were performed on August 3 and December 8, respectively.

Field trials of sister lines were conducted during two growing seasons at the UF GCREC. A pair of sister lines from a cross between Fla. 8653 (BR/BR) × Fla. 8916 (br/br) and directionally selected (one with br and the other without) for the br allele were used. Both parental lines are large‐fruited, determinate inbred lines. Three blocks were included in each season. Parental lines and the tomato cultivar “Sanibel” (BR/BR) (Seminis, www.seminis.com) were included as controls. F6 and F6:7 sister lines, with or without br, were used for yield evaluation in 2017 and 2018, respectively. The date of harvest was determined as the day when at least 70% of the fruit in a plot had attained a mature color. During the fall of 2017, S and H were performed on July 31 and December 30, respectively. For the second growing cycle during the spring of 2018, S and H were performed on February 2 and May 26, respectively. Fruit collection and yield evaluations were performed, as described in our previous study (Lee & Hutton, 2021).

Statistical analysis

Two independent experiments (i.e., trial 1 and trial 2) were performed to collect phenotypic data. For each experiment, the statistical significance level (P < 0.05) for comparisons between any two genotype mean values was determined by one‐way analysis of variance (ANOVA) in conjunction with a two‐tailed Tukey's HSD multiple comparison test, or the Welch's test. Error bars indicate the 95% confidence intervals.

Gibberellic acid treatment

Plants used for hormone treatments were grown simultaneously with those used in the greenhouse trial in the fall of 2020. At 14 days after sowing, plants were sprayed generously for 7 days with 10−4 M Gibberellin A3 (GA3; PhytoTech Labs, https://phytotechlab.com) or 10−4 M GA4+7 (PhytoTech Labs) containing 0.02% Tween 20 (Fisher Scientific). Control plants were treated with a solution containing 0.02% Tween 20. The stem lengths of 6‐week‐old plants were measured, as described by Lee et al. (2018).

Identification of BR orthologs

Putative orthologs of Solyc01g066980 were identified as described in our previous study (‘Identification of C. sativa kelch‐motif orthologs using hidden Markov models’ section in Zhang et al. (2020)). Detailed steps are available in Supporting Information S5.

RNAseq and expression analysis

In both genotypes, wild‐type and mutant (M2 generation of br.8.2 ), tissue samples were collected from individual plants grown simultaneously with those used in the greenhouse trial in the fall of 2020. Five different tissue types were collected: stem without the GA3 treatment (specifically the 1st internode) at the 6‐week‐old stage, stem (specifically the 1st internode) collected 3 h after the GA3 treatment at the 6‐week‐old stage, leaf at the 6‐week‐old stage, root at the 6‐week‐old stage, and fruit at the time of harvest. The leaf, stem with or without the GA3 treatment, and root samples were collected from 6‐week‐old plants. Green fruits (5.1 ± 1.0 g and 5.2 ± 1.1 g [mean ± standard deviation] for the wild‐type and mutant, respectively) were collected at the time of harvest in the greenhouse trial. For each biological replication, the stem, leaf, and root were collected from the same individual plant, and four biological replications (four different plants) were collected for each genotype and tissue type. The samples were flash‐frozen in liquid nitrogen immediately after excision.

Total RNA was extracted from a single plant per tissue using a Qiagen RNeasy Plant Mini Kit (Qiagen, www.qiagen.com). The yield of mRNA was quantified using a NanoDrop Spectrophotometer (Thermo Fisher Scientific, www.thermofisher.com), and the quality was verified using an Agilent 2100 Bioanalyzer (Agilent, www.agilent.com). The libraries (250–300 bp insert cDNA library) were constructed and sequenced using Illumina HiSeq technology (PE 150) by the Novogene (https://novogene.com).

A total of 981.0 million Illumina reads were generated. After removing reads with adaptor contamination, uncertain nucleotides constitute (>10% of either read), or low quality nucleotides (base quality <5 and >50% of reads), alignments of the raw reads to the tomato reference genome assembly SL4.0 (Fernandez‐Pozo et al., 2015) were performed using STAR (Dobin et al., 2013). HTSeq v0.6.1 (Anders & Huber, 2010) was used to count reads mapped for each annotation (ITAG4.0; https://solgenomics.net). Then the expected number of Fragments Per Kilobase of transcript sequence per million base pairs sequenced (FPKM; Trapnell et al., 2010) for each gene was calculated. Furthermore, genes with counts per million aligned reads of at least 0.5 in at least four biological replications were retained, resulting in 23,869 genes for further analysis. A DE gene analysis was performed using DESeq2 (1.28.1; Love et al., 2014). DE was evaluated for the following groups of contrasts: (1) genotype (wild‐type and mutant genotypes for the same tissue and treatment) and (2) treatment (GA3‐treated and untreated for stem tissue). DE was evaluated based on the log2 fold change, Z‐statistic (log2 fold change divided by its standard error), p value, and FDR estimate. FDR was calculated independently for each test.

Gene coexpression network analysis

A gene coexpression network analysis was performed using the WGCNA R software package, as described by Langfelder and Horvath (2008). Two separate datasets were created: (1) one without any data adjustment and (2) one adjusted for tissue types (i.e., identification of modules driven by the GA3 treatment in stem). Enrichment of significantly DE genes in each module was identified at a threshold of Bonferroni correction <0.05 for 3778 GO terms for tomato (EnsemblPlants Biomart release 49; Kinsella et al., 2011; Howe et al., 2020).

CONFLICT OF INTEREST

The Authors did not report any conflict of interest.

AUTHOR CONTRIBUTIONS

M.B.L. created CRISPR brachytic mutants in the Fla. 8059 background and performed some greenhouse evaluation. R.S. conducted the mapping. S.F.H. supplied seed, interpreted mapping data, and provided knowledge of fresh‐market tomato breeding. T.G.L. designed the study, obtained funding, performed experiments, supervised and conducted data analysis, prepared figures, and wrote the manuscript.

Data S1. ML tree in the Newick format.

Click here for additional data file.

Figure S1. A comparison of fruit size in different tomatoes.br.8.2 is a CRISPR‐Cas9‐driven BRACHYTIC gene‐edited mutant tomato. br.8.2 has a genetic background of large‐fruited fresh‐market tomato, Fla. 8,059. Fla. 8,059 is one of the two parents of a commercial hybrid (F1), Tasti‐Lee™ (Bejo Seeds, Oceano, CA), currently in the U.S. market.

Figure S2. DNA sequence polymorphisms in genes annotated in the BRACHYTIC (BR) fine mapping region. a, The sequence of BR allele (Solyc01g066980) is obtained from the genome of Heinz 1706 (the tomato reference genome assembly SL4.0; Fernandez‐Pozo et al., 2015) in the Phytozome database (https://phytozome-next.jgi.doe.gov). Two base pair positions, 68,064,387 bp and 68,064,993 bp, indicate the centromeric and telomeric ends of the deletion of the FPF1 gene at the br/br allele (Figure S5). b, Results of PCR using primers tagging the deletion in Solyc01g066980. For primers br_deletion_F and br_deletion_R see Table S7. c, A scaffold (a sequence assembly carrying genomic DNA from the br genetic interval, which is 481‐kb in size; scaffold ID 143086) spanning the BR fine mapping interval (175‐kb) was compared to the genome of Heinz 1706. Fla. 8,834 and Fla. 8,916 are large‐fruited, determinate inbred lines, and each carries homozygous br (Lee et al., 2018). Fla. 7907B, Fla. 8,059 and Fla. 8111B are large‐fruited, determinate inbred lines that do not carry the br allele. Numbers and black marks indicate tomato genes (e.g., Solyc01g066915 according to the annotation ITAG4.0; Fernandez‐ Pozo et al., 2015). Three Flowering Promoting Factor 1 (FPF1) genes are shown by black‐filled marks. Red arrow indicates a complete coding sequence deletion in a FPF1 gene in the br allele. Black arrows indicate sequence polymorphisms in introns. The prefix brM indicates molecular markers used for fine mapping. brM10 and brM12 indicate the beginning and end of the mapping interval, respectively, supported by two different recombinant inbred line (RIL) populations and brM11 indicates a recombination breakpoint supported by a single RIL population. C indicates the direction toward the centromere.

Figure S3. Greenhouse evaluation of the CRISPR‐Cas9‐driven brachytic (br) mutants.a, The stem lengths of 6‐week‐old plants. WT, wild‐type (BR/BR allele). Heterozygous, F1 plant derived from a cross between WT and br.8.1 ; br.8.1 , transgene‐free homozygous M1 generation mutant. b, Days to first flower. br.8.1 and br.8.2 , transgene‐free, homozygous M3 generation mutant lines. c, leaves to first flower. br.8.1 and br.8.2 , transgene‐free, homozygous M3 generation mutant lines. d and e, days to first fruit. br.8.1 and br.8.2 , transgene‐free, homozygous M1 and M3 generation mutant lines, respectively. The n value represents the total number of plants for each genotype evaluated during each trial; ns indicates no significant difference (ANOVA at p > 0.05) was found between any genotypes, except for d and e; the Welchs test (p > 0.05) was used for the comparison in d and e. Error bars indicate 95% confidence intervals.

Figure S4. Knocking‐out of Solyc01g066980 in the processing market tomato M82 reduced the plant height. a, A representative mutant plant (right) was compared to its wild‐type (left). b, CRISPR‐Cas9‐driven mutations in Solyc01g066980. Red and blue sequences indicate protospacer‐adjacent motifs and an insertion, respectively. The sequence gap length is shown in parentheses.

Figure S5. Illumina whole‐genome sequence coverage of the BRACHYTIC (BR) locus in nine tomato accessions. Illumina data showing the deletion in the Solyc01g066980 (a), the centromeric (b, 68,064,387 bp on chromosome 1) and telomeric (c, 68,064,993 bp on chromosome 1) ends of the deletion. 3,040,717 is a single br plant used for the 10 × Genomics Chromium library in this study.

Figure S6. Maximum likelihood phylogenetic tree of Solanaceae flowering promoting factor 1 families.

Figure S7. Gibberellic acid (GA) regulates the stem elongation of young brachytic (br) plants. br plants sprayed with GA3 and GA4 + 7 were shown by filled bars. C‐1, the distance between the cotyledonary node and the first node; 1–2, the first and the second; 2–3, the second and the third; 3–4, the third and the fourth; and sum, the cotyledonary node to the fourth node (on the top). Untreated, GA untreated plants; GA3 and GA4 + 7, plants with the GA3 and GA4 + 7 treatments, respectively. *p < 0.05, **p < 0.01, and ***p < 0.001 based on one‐way ANOVA in conjunction with a two‐tailed Tukeys HSD multiple comparison test. Error bars indicate 95% confidence intervals.

Figure S8. Development of transgene‐free mutants. a, M0 tomato transformants were tested for the presence of deletion(s) using the T7 Endonuclease I assay. b and c, PCR genotyping of the M0 generation for the absence of T‐DNA using two different markers (b, Cas9 gene selectable marker; c, hygromycin selectable marker). d, A single Illumina sequence read was detected in the mutant plant #1. We mapped the Illumina whole‐genome sequencing reads to the complete vector (pHSN401) sequence. The mutant number (#1 or #2) indicates the same individual transformant throughout figures b, c, and d. The mutant plant #1 was discarded as unreliable. Confirmation of the desired CRISPR‐Cas9‐driven mutation using the Illumina technology: e, Solyc01g066980; f, Solyc01g066950; and g, Solyc01g066970. WT, wild‐type. br.8.2 , br.7 , br.57.1 , and br.57.2 , transgene‐free, homozygous mutant plants.

Figure S9. Illumina whole‐genome sequence coverage of the potential off‐target site on chromosome 8.

Figure S10. Tomato color classification. Fruits with a light green color [empty arrow; column #1; USDA color classification ‘Green’ (U.S. Department of Agriculture, 2017)], pink color (filled arrow; column #2; ‘Breakers’), half red (column #3; ‘Pink’), and full red (column #4; ‘Red’). White bar represents 2.5 cm. Tasti‐Lee™ (Bejo Seeds, Oceano, CA) fruits are sampled.

Click here for additional data file.

Supporting Information S1. Three mapping populations from crosses Fla. 8,624 × Fla. 8,834, Fla. 8,044 × Fla. 8,834, and Fla. 8,044 × Fla. 8,916 were used. Fla. 8,834 and Fla. 8,916 are large‐fruited, determinate (sp/sp; homozygous at the self‐pruning locus) inbred lines, and each carries the homozygous br allele. Fla. 8,044 and Fla. 8,624 are large‐fruited, determinate inbred lines that do not carry the br allele. Recombinant screening was initiated by testing all available F2 progeny from the mapping populations, which combines four new RILs selected in this study (Table S1). DNA sequence polymorphisms identified from the tomato Illumina Infinium array initiated by the Solanaceae Coordinated Agricultural Project (SolCAP) and WGS were selected to develop molecular markers as described by Lee et al., 2018 to screen for recombinants and to saturate the targeted intervals with molecular markers. The primer sequences for the molecular markers are presented in Table S7.

Supporting Information S2. Genomic DNA was extracted as described in our previous study (‘Fosmid library construction’ section in ‘Supplementary Materials’ in Cook et al. (2012)). A single plant was used for DNA extraction. DNA was treated with RNase (Roche, www.roche.com) by incubation 25 μg/mL RNase at 4 °C overnight. The presence of br was investigated using molecular markers spanning the locus. 1 ug of high‐molecularweight genomic DNA was sequenced at the DNA Services Lab at the University of Illinois Roy J. Carver Biotechnology Center. A single DNA sequencing library was prepared using the Chromium Genome V2 Library Kit (10X Genomics, www.10xgenomics.com). The library was loaded into a single lane and sequenced using version 4 of the Illumina HiSeq SBS Kit (Illumina, www.illumina.com). In total, 689,977,878 reads (about 76 × coverage of the 0.9 Gb tomato genome) were generated. The reads were assembled using Supernova (10X Genomics, www.10xgenomics.com). Candidate scaffolds were identified by local BLAST searched based on the BR interval of the Heinz 1706 reference sequence. Molecular marker sequences that matched the anticipated region of the scaffold were confirmed.

Supporting Information S3. Total genomic DNA of each transformed plant in the M0 generation was extracted from young leaves using the DNeasy Plant Mini Kit (Qiagen, www.qiagen.com). PCRs were performed to examine mutations in the targeted region. PCR cycling and running parameters were as follows: initial denaturation step at 95 °C for 7 min, 30 cycles at 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 1 min, followed by a final extension at 72 °C for 7 min. For the T7 Endonuclease I assay, genomic DNA extracted from individual plants was used as the template. A pair of targeted region‐specific primers and Q5 Hot Start High‐Fidelity 2X Master Mix (NEB, www.neb.com) were used for PCR. The cycling and running parameters were as follows: initial denaturation step at 98 °C for 30 s, 35 cycles at 98 °C for 5 s, 60 °C for 10 s, and 72 °C for 20 s, followed by a final extension at 72 °C for 2 min. PCR products were purified using a QIAquick PCR Purification Kit (Qiagen, www.qiagen.com), and 200 ng of the PCR products was digested with T7E1 according to the manufacturers instructions.

Supporting Information S4. Targeted mutagenesis of BRACHYTIC (Solyc01g066980) using the CRISPR‐Cas9 system for processing tomato cultivar M82 was performed as previously described (Brooks et al., 2014; Swartwood & Van Eck, 2019; Van Eck et al., 2019). Briefly, the binary vectors were constructed through Golden Gate cloning as described by Werner et al. (2012), and introduced into tomato by ‐mediated transformation as described by Swartwood and Van Eck (2019) and Van Eck et al. (2019). 1st generation transgenic plants were transplanted in soil and genotyped to validate potential Cas9‐gRNA‐introduced mutations by Sanger sequencing of PCR products (Xu et al., 2015; Table S7).

Supporting Information S5. The Solanaceae annotated protein dataset was obtained from the Solanaceae Genomics Network (https://solgenomics.net). Protein sequences were as follows: ITAG4.0_proteins.fasta ( ), Spimpinellifolium_genome.protein.fa (Solanum pimpinellifolium), Spenn‐v2‐aa‐annot.fa (Solanum pennellii), P 53 GSC_DM_v3.4_pep.fasta (Solanum tuberosum), Eggplant_V3.protein.putative_function.fa (Solanum melongena), Pepper.v.1.55.PEP.fa ( L. CM334), Capsicum.annuum. L_Zunla‐1_v2.0_PEP.fa ( L. Zunla‐1), Capsicum.annuum.var.glabriusculum_Chiltepin_v2.0_PEP.fasta ( ), Petunia_axillaris_v1.6.2_proteins.fasta ( ), Petunia_inflata_v1.0.1_proteins.fasta (Petunia inflata), and Niben101_annotation.proteins.fasta (Nicotiana benthamiana). Detailed steps for orthologs identification are as follows: The seed alignment model was generated using the MAFFT (mafft 7.450; Katoh & Standley, 2013) for the Linux platform. Plant protein sequences were as follows: three , seven Zea mays, six , nine , five , two Vitis vinifera, seven , five , and three sequences from PANTHER (Thomas et al., 2003) family ID PTHR33433 (data access on October 30, 2019). The canonical proteins of Solanaceae were identified using a hidden Markov model search (HMMER 3.1b2; http://hmmer.org) applying 10–1 e‐value cutoff. No manual curation of the Solanaceae sequences was performed. To increase the confidence that true orthologs were detected, a reciprocal BLAST search of the predicted proteomes was performed using full‐length sequences. BLAST sequences were performed using tools provided at the NCBI (www.ncbi.nlm.nih.gov) and Solanaceae Genomics Network. All the Solanaceae proteins that matched Solyc01g066980 were detected with an e‐value of 0.0 using BLAST. Alignments were performed using the MUSCLE tool (muscle 3.8.31; Edgar, 2004) for the Linux platform. A maximum likelihood (ML) phylogenetic analysis of the protein sequence alignment was performed using RAxMLHPC‐PTHREADS‐SSE3 (version 8.2.3; Stamatakis, 2014) for the Linux platform using the VT model of amino acid substitutions and gamma distribution parameters estimated by the software. A total of 100 bootstrap replicates were performed. Phylogenetic trees were visualized and edited using Geneious R10.2.2 (www.biomatters.com). The ML tree annotated by RAxML is provided in the Newick format (Data S1).

Click here for additional data file.

Table S1. Fine mapping of the BRACHYTIC locus.

Click here for additional data file.

Table S2. Presence/absence of the sequence deletion of a flowering promoting factor 1 gene in the BRACHYTIC locus in nine tomato accessions.

Click here for additional data file.

Table S3. Flowering promoting factor 1 annotations (57) identified in Solanaceae.

Click here for additional data file.

Table S4. Gene expression profiles.

Click here for additional data file.

Table S5. Tissue‐specific co‐expression modules and Gene Ontology overview.

Click here for additional data file.

Table S6. Co‐expression modules adjusted for tissue with respect to gibberellic acid (GA) treatment and Gene Ontology overview.

Click here for additional data file.

Table S7. Sequence information for molecular markers, guide RNAs, and PCR primers used in this study.

Click here for additional data file.

Table S8. List of potential off‐targets.

Click here for additional data file.