Skip to main content
NCBI home page
The Plant Cell logoLink to The Plant Cell
. 2021 Aug 2;33(10):3293–3308. doi: 10.1093/plcell/koab201

A mutation in a C2H2-type zinc finger transcription factor contributed to the transition toward self-pollination in cultivated tomato

Lele Shang 1, Jianwen Song 1, Huiyang Yu 1, Xin Wang 1, Chuying Yu 1, Ying Wang 1, Fangman Li 1, Yongen Lu 1, Taotao Wang 1, Bo Ouyang 1, Junhong Zhang 1, Robert M Larkin 1, Zhibiao Ye 1, Yuyang Zhang 1,2,✉,2
PMCID: PMC8505859  PMID: 34338777

Abstract

The degree of stigma exsertion has a major influence on self-pollination efficiency in tomato, and its improvement is essential for raising productivity and for fixing advantageous traits in cultivated tomato. To study the evolution of stigma exsertion degree in tomato, we searched for genes associated with this trait and other aspects of flower morphology, including the lengths of anthers, styles, and ovaries. We performed a genome-wide association on 277 tomato accessions and discovered a novel stigma exsertion gene (SE3.1). We reannotated the structure of the gene, which encodes a C2H2-type zinc finger transcription factor. A mutation of the lead single nucleotide polymorphism creates a premature termination codon in SE3.1 and an inserted stigma in cultivated tomatoes. SE3.1 is essential for the conversion of flush stigmas to inserted stigmas. This conversion has a major impact on the rate of self-fertilization. Intriguingly, we found that both SE3.1 and Style2.1 contribute to the transition from stigma exsertion to insertion during the domestication and improvement of tomato. Style2.1 controls the first step of exserted stigmas to flush stigmas, and SE3.1 controls the second step of flush stigmas to inserted stigmas. We provide molecular details for the two-step process that controls the transition from stigma exsertion to insertion, which is of great agronomic importance in tomato.

A mutation in SE3.1 leads to the development of inserted stigmas in cultivated tomatoes, which together with Style2.1 contributed to the transition from exserted to inserted stigmas.

 

graphic file with name koab201f9.jpg

Open in a new tab

Introduction

Variation in floral morphology influences the mating system by promoting either allogamy (cross-pollination) or autogamy (self-pollination; Stebbins, 1970; Goodwillie et al., 2010). Such variation is common in flowering plants and has a major effect on mating strategies (Kalisz et al., 1999; Sicard et al., 2011; Vallejo-Marin et al., 2014; Carleial et al., 2017; Wozniak et al., 2020). Herkogamy is described as the spatial separation of pollen and stigmas within flowers and prevents self-pollination (M�ller, 1883). The reduction of herkogamy can substantially improve the efficiency of self-pollination in self-compatible species (Luo and Widmer, 2013; Torang et al., 2017; Opedal, 2018).

Thirteen tomato (Solanum sp.) species have evolved different mating systems encompassing obligatory allogamous, facultative allogamous, and autogamous species (Rick, 1988; Chen and Tanksley, 2004; Peralta et al., 2008). One of the major differences in their flowers is the degree of stigma exsertion (Rick, 1995), with the highest degree of stigma exsertion being observed in obligatory allogamous species with self-incompatibility (SI), presumably to promote outcrossing (Rick, 1982). Mutations at the S locus led to a breakdown in SI in several species, giving rise to self-compatibility (SC) and contributing to the shift from obligate to facultative allogamy in tomato (Li and Chetelat, 2010; Tovar-Mendez et al., 2014; Li and Chetelat, 2015). Solanum pimpinellifolium is a self-compatible species that is categorized as a facultative allogamous species because of its diverse outcrossing rates associated with different degrees of stigma exsertion (Rick et al., 1977, 1978). Solanum lycopersicum var. cerasiforme is thought to be the ancestor of cultivated tomato (S. lycopersicum) and was probably domesticated from the wild relative S. pimpinellifolium (Ranc et al., 2008; Lin et al., 2014). It is worth noting that during the domestication process, the degree of stigma exsertion diminished so that stigmas became flush with anthers to promote autogamy (Bai and Lindhout, 2007). The stigmas of most S. lycopersicum accessions became further inserted to increase the efficiency of autogamy (Rick, 1988). Thus, the mating strategy of tomato species relies on the degree of stigma exsertion.

The degree of stigma exsertion is mainly determined by the relative lengths of anthers, ovaries, and styles. To determine the genetic basis underlying the variation in outcrossing rates seen in S. pimpinellifolium, previous studies characterized several floral traits such as the lengths of petals, anthers, styles, sepals, and ovaries in both self-pollinating and outcrossing accessions (Rick et al., 1977; Georgiady and Lord, 2002). Using two accessions with outcrossing rates of 0% and 37%, respectively, quantitative trait locus (QTL) analysis identified a QTL on chromosome 3 that explained ∼35% of the observed variation in anther length. A second QTL was identified on chromosome 8 with larger effects on style length (Michael et al., 2002). The progeny from crosses between wild species and S. lycopersicum—a cultivated species with an inserted stigma—allowed the mapping of several QTLs controlling the degree of stigma exsertion to chromosomes 2, 5, and 9 (Fulton et al., 1997; Gorguet et al., 2008). Further analysis of the genomic region on chromosome 2 that harbors the major QTL se2.1 influencing stigma exsertion led to the identification of at least five tightly linked genes—one controlling style length, three controlling anther length, and the last one affecting anther dehiscence (Chen and Tanksley, 2004). The causal gene for se2.1 associated with style length was later cloned and named Style2.1. A 450-bp deletion in the promoter region of Style2.1 converts long styles to short styles and thus, is responsible for the transition from stigma exsertion to stigma insertion that accompanied the evolution of self-pollination in cultivated tomato (Chen et al., 2007).

Previous studies have explored the distribution of the 450-bp deletion in the Style2.1 promoter across tomato species. The full Style2.1 promoter was mainly distributed in wild species, while a majority of accessions from S. lycopersicum var. cerasiforme and S. lycopersicum harbored the 450-bp deletion (Chen et al., 2007; Vosters et al., 2014), which reduced style length and stigma exsertion degree (Chen et al., 2007). However, style length and stigma exsertion degree were not strictly associated with the presence of the 450-bp deletion in S. lycopersicum var. cerasiforme and S. lycopersicum (Vosters et al., 2014), as stigma exsertion degree still showed variation among S. lycopersicum var. cerasiforme and S. lycopersicum accessions carrying the deletion, and manifested as flush or inserted stigmas. The conversion from flush to inserted stigmas raised the efficiency of fruit setting in tomato (Rick and Dempsey, 1969). However, to date, no genes have been identified controlling the transition from flush stigmas to inserted stigmas in tomato.

Genome-wide association studies (GWAS) are powerful and efficient in uncovering the molecular basis underlying complex traits in different crops (Lin et al., 2014; Si et al., 2016; Zhang et al., 2019; Wang et al., 2020). In this study, we analyzed the lengths of anthers, styles, and ovaries in a population of 277 tomato accessions, followed by the quantification of the degree of stigma exsertion by calculating the ratio between anther length and pistil length, which we then used for GWAS and identified the stigma exsertion gene SE3.1. Indeed, the lead single-nucleotide polymorphism (SNP) identified by GWAS mapped to the SE3.1 coding region after a careful reannotation of the locus. This polymorphism converted a C to T and introduced a premature termination codon in SE3.1 that caused the transition from flush stigmas to inserted stigmas in the population. In addition, we determined that the loss of SE3.1 in an accession carrying the presumptive functional allele (SE3.1C) results in the development of inserted stigmas and thus, promotes self-fertilization. Furthermore, we studied the functions of SE3.1 and Style2.1 in controlling the degree of stigma exsertion and propose a two-step model for the regulation of the degree of stigma exsertion during the domestication and improvement of tomato. Thus, our results provide greater understanding of how tomato acquired exerted stigma over the course of its evolution.

Results

Reduction in the degree of stigma exsertion during the domestication and improvement of tomato

The evolution of mating systems among flowering plants is often accompanied by changes in floral morphology that facilitate the transition from allogamy to autogamy. The stigma exsertion degree (SED) contributes to mating strategies in tomato and can be classified as exserted, flush, or inserted relative to the position of anthers (Chen and Tanksley, 2004). To explore stigma exsertion, we focused on a diversity panel of 277 tomato accessions consisting of 25 wild accessions comprising one S. cheesmaniae and 24 S. pimpinellifolium (PIM), 103 domesticated S. lycopersicum var. cerasiforme accessions (CER), and 149 improved S. lycopersicum accessions (BIG; Figure�1A;Supplemental Figure S1). We classified SED in each accession as either exserted, flush or inserted (Figure�1B;Supplemental Data Set S1 and “Materials and Methods”). To obtain a more quantitative measure of these phenotypes, we determined the lengths of anthers (AL), styles (SL), and ovaries (OL). In addition, we defined pistil length (PL) as the sum between style length and ovary length and calculated the ratio between anther length and pistil length (AL/PL) (Figure�1C;Supplemental Figure S2 and Supplemental Data Set S1).

Figure 1.

Figure 1

Open in a new tab

Reduction in the degree of stigma exsertion during tomato domestication and improvement. A, Composition of the 277 tomato accessions used here: 25 wild accessions (one S. cheesmaniae and 24 S. pimpinellifolium [PIM]), 103 S. lycopersicum var. cerasiforme (CER), and 149 S. lycopersicum (BIG). B, Top view (upper) and internal section (lower) of flowers from representative the tomato accessions TS-19 (PIM, exserted stigma), TS-278 (BIG, flush stigma), and TS-9 (BIG, inserted stigma). C, Schematic representation of a flower illustrating the parameters measured. PL is the SL + OL from the same flower. AL/PL is the ratio between anther length and pistil length from the same flower. D, Percentages of the three SED categories in wild, CER, and BIG accessions. E–H, Distributions of pistil (E), ovary (F), style (G), and anther (H) lengths from wild, CER, and BIG accessions. (I) Pearson’s correlation coefficients (r) between floral traits across the 277 tomato accessions. Statistically significant correlations are indicated with an asterisk (*P < 0.001). J, Distribution of the ratio between anther length and pistil length from wild, CER and BIG accessions. All distributions are shown as box plots of the range of percentiles from the total data, as determined using Tukey’s method. Central line, median; whiskers, interquartile range; outer dots, outliers. Different letters indicate significant differences (P < 0.01, one-way ANOVA with Tukey’s post hoc test).

We scored the three SED classes in the three groups of accessions. The percentage of accessions with exserted stigmas decreased from 96% (wild accessions) to 4.7% (BIG accessions; Figure�1D). We observed flush stigmas from accessions belonging to all three groups, with the highest proportion being in the CER group (57.3%) and in the BIG group to a lesser degree (32.2%). Few accessions from the CER group had inserted stigmas (16.5%), while a majority of accessions from the BIG group had inserted stigmas (63.1%; Figure�1D;Supplemental Data Set S1). None of the wild accessions showed an inserted stigma.

Although we did not notice a trend for pistil lengths, ovary lengths, and anther lengths gradually increased from the wild group to the CER group and from the CER group to the BIG group (Figure�1, E, F, and H). Similarly, although style length did not vary significantly in the CER group relative to the BIG group, styles were shorter in the CER and BIG group relative to the wild group (Figure�1G). During domestication (as seen by comparing the wild and CER groups) and improvement (as seen by comparing the CER and BIG groups), the length of both styles and ovaries positively correlated with pistil length. Additionally, pistil length was positively correlated with anther length. Together these effects contributed to variation in the ratio between anther length and pistil length (Figure�1I).

We next evaluated the degree of stigma exsertion using the ratio between AL/PL: mean values were 0.85 in group of wild accessions, 0.95 in the CER group, and 1.05 in the BIG group. When sorting accessions based on their SED phenotype, the mean AL/PL ratio for accessions with exserted stigmas was below 0.90, while it was between 0.90 and 1 in accessions with flush stigmas and over 1.0 for accessions with inserted stigmas (Figure�1J;Supplemental Data Set S1). Indeed, the AL/PL ratio was strongly and negatively correlated with SED (r = –0.84) and therefore represented an appropriate estimation of this phenotype (Figure�1I). Taken together, these data indicate that a flower trait associated with stigma exsertion degree was reduced during the domestication and improvement of tomato.

Identification of a major locus that influences the degree of stigma exsertion using GWAS

The AL/PL ratio ranged from 0.72 to 1.38 across the 277 tomato accessions used in this study. As mentioned above, the AL/PL ratio of accessions from the wild group was lower than 0.9. The AL/PL ratio of most accessions from the CER group varied from 0.9 to 1.0 and was greater than 1.0 in the accessions from the BIG group (Figure�2A). Using 4,856,876 SNPs with a minimum allele frequency (MAF) >5% (minor allele with homozygosity in at least 10 accessions), we performed GWAS using the efficient mixed-model association expedited (EMMAX) algorithm and the AL/PL ratio as phenotype (Kang et al., 2010). After a Bonferroni-adjusted correction, we set the P-values for the suggestive and significant thresholds to 2.06 � 10−7 and 1.03 � 10−8, respectively (Figure�2, B and C;Supplemental Figure S3). The linkage disequilibrium (LD) block containing the lead SNP, ch03_53874338 (P = 1.15 � 10−19), harbored a single gene that encoded a C2H2-type zinc finger protein (Solyc03g098070). In fact, the lead SNP was located in Solyc03g098070 (Figure�2D;Supplemental Data Set S2), with the cytosine (C) allele being associated with a low AL/PL ratio and the thymine (T) allele with a high AL/PL ratio (Figure�2E;Supplemental Data Sets S1 and S3). Therefore, we concluded that Solyc03g098070 is the most likely candidate gene and hereafter refer to Solyc03g098070 as Stigma Exsertion 3.1 (SE3.1). We monitored GUS activity in transgenic plants harboring a SE3.1pro:GUS reporter construct and detected robust GUS activity in anthers, styles and ovaries (Supplemental Figure S4). Taken together, we concluded that the T/C alleles of SE3.1 at the lead SNP, ch03_53874338, (i.e. SE3.1T/C) are likely associated with variation in the degree of stigma exsertion.

Figure 2.

Figure 2

Open in a new tab

Identification of a major locus that influences the degree of stigma exsertion by genome-wide association. A, Distribution of the ratio between AL/PL across 277 tomato accessions. B, Manhattan plot from GWAS using the AL/PL ratio as phenotype. The red and blue dashed horizontal lines indicate the Bonferroni-adjusted significant and suggestive thresholds, P�=1.03�10−8 and 2.06�10−7, respectively. The red arrow points to the lead SNP, ch03_53874338 (P�=1.15�10−19) in SE3.1. The y-axis indicates –log10-transformed P-values. C, Quantile–quantile plot of expected versus observed P-values for GWAS. D, The 2-Mb region surrounding the lead SNP (red arrow) in SE3.1. The red vertical lines and blue horizontal lines below the x-axis show gene models. Magenta vertical lines, SNPs located in LD blocks; light blue vertical lines, SNPs not located in LD blocks; gray vertical lines, no SNP present. The green lines indicate the position of the SNPs in the LD blocks. E, Distribution of AL/PL ratio as a function of genotype at the lead SNP (ch03_53874338), represented as either cytosine (C) or thymine (T) (n: number of accessions). Statistical significance was determined using a two-tailed t test. The physical position is based on version SL2.40 of the tomato genome.

A polymorphism in SE3.1 leads to premature termination of translation

Based on the ITAG2.3 annotation, the lead SNP (ch03_53874338) is located 3-bp upstream of the predicted start codon of Solyc03g098070 (Figure�3A; “Materials and Methods”). With the ITAG3.2 annotation update, another gene model (Solyc03g098075) was predicted immediately upstream of Solyc03g098070, on the same coding strand, and overlapping with the 5′-untranslated region (5′-UTR) of Solyc03g098070 (Figure�3B). Of the two SE3.1 alleles at the lead SNP, the T allele introduced a premature stop codon in the predicted gene model of Solyc03g098075 (Figure�3B), raising the possibility that Solyc03g098075 may in fact be part of the Solyc03g098070 gene model. The true coding sequence of SE3.1 would then begin with the start codon predicted for Solyc03g098075 at position –312 relative to the current start codon called for Solyc03g098070. If this hypothesis is correct, a cDNA starting at the ATG of Solyc03g098075 and ending at the stop codon of Solyc03g098070 should be amplified from tomato accessions. In addition, the T allele would then result in a premature termination codon (PTC) and a truncated SE3.1 protein (Figure�3C).

Figure 3.

Figure 3

Open in a new tab

The C/T polymorphism in SE3.1 leads to premature termination of translation. A, Gene model for SE3.1 (Solyc03g098070) from an older annotation of the reference genome (SL2.40). The lead SNP (ch03_53874338, red vertical line) is located 3-bp upstream of the predicted start codon (ATG) of Solyc03g098070. The reference genome (ITAG2.3) carries a T at this position. B, Gene models at and near SE3.1 in the current reference genome (SL3.20). Horizontal red arrows indicate the positions of primers used to amplify gDNA and cDNA that extend from the ATG of Solyc03g098075 to the stop codon (TAG) of Solyc03g098070. C, Proposed structure of Solyc03g098070 (SE3.1). The red vertical line indicates the position of the lead SNP. The C to T change converts a CAG codon (Gln) to a TAG causing premature termination of translation. The gray solid lines indicate the genomic regions PCR amplified to generate constructs 1–3. D, gDNA and cDNA amplicons from accessions carrying the T allele of SE3.1 (TS-9, TS-7, and TS-46) or the C allele of SE3.1 (TS-278, TS-136, and TS-225). gDNA length: 2,490 bp, cDNA length: 1,305 bp. E, Immunoblot analysis of N. benthamiana leaves transiently infiltrated with the SE3.1 constructs shown in (C). 1-C, SE3.1C; 3-C, SE3.1C; ORF, SE3.1C-ORF; 2-T, SE3.1T, respectively. The red arrow indicates the predicted SE3.1C protein. F, Final reannotation of the SE3.1 locus. Green boxes, exons; blue boxes, untranslated regions. The positions of the translation start codon (ATG) and stop codon (TAG) of SE3.1 are shown.

Accordingly, we amplified genomic DNA (gDNA) and cDNA sequences beginning from the start codon of Solyc03g098075 and ending at the stop codon of Solyc03g098070 from accessions carrying the T allele (TS-9, TS-7, TS-46) or the C allele (TS-278, TS-136, TS-225; Figure�3, B and D) and determined the sequence of the resulting polymerase chain reaction (PCR) amplicons. First, we amplified cDNAs of the same length in all accessions tested. Second, their sequences revealed the splicing of the intron from Solyc03g098075 (at positions –209 to –108 relative to the ATG from Solyc03g098070 of ITAG3.2) and the intron from Solyc03g098070 (at positions +294 to +942) into a single transcript (Figure�3D;Supplemental Figure S5), supporting the idea that the Solyc03g098075 and Solyc03g098070 gene models in fact belong to the same gene model. We next determined the translation potential of the SE3.1C and SE3.1T transcripts by transient infiltration assays in Nicotiana benthamiana leaves. To this end, we generated the following constructs: constructs 1 and 2, starting at the predicted ATG of Solyc03g098075 and ending before the stop codon from Solyc03g098070 for the C allele (SE3.1C, construct 1-C) or the T allele (SE3.1T, construct 2-T); construct 3-C starting 508-bp upstream of the predicted ATG from Solyc03g098070 in ITAG3.2 and ending before the stop codon of Solyc03g098070 for the C allele; and construct SE3.1C-open reading frame (ORF) (without the stop codon) (Figure�3C). All clones include a FLAG tag sequence at the C terminus for detection by immunoblot. Importantly, we detected a protein of the same size in protein extracts from leaves infiltrated with construct 1-C, construct 3-C, and construct SE3.1C-ORF (Figure�3E). These data indicate that the Solyc03g098075 and Solyc03g098070 form a single gene model, with a single ATG located at –312 bp relative to the current predicted ATG for Solyc03g098070. Notably, we failed to detect a band for construct 2-T, expressing the SE3.1T allele, presumably because of the PTC in SE3.1T (Figure�3E). These findings are consistent with our hypothesis on SE3.1: the SE3.1C allele encodes a protein with three C2H2 zinc fingers with a zinc binding site (Supplemental Figures S6A and B), whereas the current gene model for Solyc03g098070 only contains the two C2H2 zinc fingers and lacks the zinc binding site (Supplemental Figure S6B). Moreover, a tertiary structure analysis indicated that SE3.1C may bind more zinc ions (Supplemental Figure S6C; Waterhouse et al., 2018). In conclusion, these results provide evidence that the lead SNP associated with the AL-to-PL ratio (and thus the degree of stigma exsertion) is located in the second exon of the updated SE3.1 gene model, which includes the coding sequence of Solyc03g098070 and Solyc03g098075 (ITAG3.2). Additionally, the lead SNP converts a Gln in the C allele to a PTC in the T allele (Figure�3F).

SE3.1 controls flush stigma by regulating the development of anther and style

To independently confirm that the SE3.1C/T polymorphism regulates the degree of stigma exsertion, we transformed the accession TS-9, which has an inserted stigma and is homozygous for the T allele at SE3.1, with four transgenes. Construct 1 (35Spro:3070C) overexpressed the genomic region of Solyc03g098070 beginning at the 5′-UTR from the ITAG2.3 gene model, and thus does not include the true predicted ATG of Solyc03g098070; Construct 2 (35Spro:SE3.1C or 35Spro:SE3.1T) overexpressed the genomic region of the reannotated SE3.1 gene model harboring either the C or the T allele; Construct 3 (SE3.1Cpro:SE3.1C) used 2,000 bp of SE3.1 promoter to drive the expression of the SE3.1C genomic region (Figure�4A). All SE3.1C-overexpressing lines developed flush stigmas and had AL/PL ratios below 1.0, in contrast to the inserted stigma of the TS-9 parental accession (Figure�4, B and D). The increased degree of stigma exsertion in the SE3.1C-overexpressing lines was caused by shorter anthers and longer styles (Figure�4, E–H). In contrast, the 3070C-overexpressing lines developed inserted stigmas similar to those of the wild-type TS-9 (Figure�4C;Supplemental Figure S7, F–J). As a control, we also transformed TS-9 with the SE3.1T allele: the resulting overexpression lines developed the same inserted stigmas as in the wild-type (Figure�4C;Supplemental Figure S7, A–E). These data provided evidence that the SE3.1C/T polymorphism controls whether stigmas are flush or inserted, while also independently validating our reannotation of the SE3.1 gene model. The degree of stigma exsertion was the same in the SE3.1C-overexpressing lines and in the transgenic lines harboring the SE3.1Cpro:SE3.1C transgene, indicating that the function of SE3.1 depends on the presence of the C allele rather than on SE3.1 transcript levels (Figure�4B).

Figure 4.

Figure 4

Open in a new tab

SE3.1 controls flush stigma development by regulating anther and style lengths. A, Schematic representation of constructs generated. B and C, Internal sections of flowers from the wild-type (TS-9) harboring 35Spro:SE3.1C or SE3.1Cpro:SE3.1C (B) and 35Spro:SE3.1T (as control) or 35Spro:3070C (C). The images were collected with a Leica DFC7000 GT microscope. (D–H) Distribution of the AL-to-PL ratio (D) and the length of anthers (E), styles (F), ovaries (G), and pistils (H) from the wild-type (TS-9) and transgenic lines harboring 35Spro:SE3.1C, shown as box plots. I, The generation of se3.1C mutants by CRISPR/Cas9. The sequences of SE3.1C in the wild-type (TS-278, carrying the SE3.1C allele) and in the se3.1-1, se3.1-3, se3.1-4 mutants are shown. J, Internal sections of flowers from the wild-type (TS-278), se3.1-1, se3.1-3, and se3.1-4. The images were collected with a Leica DFC7000 GT microscope. K–O, Distributions of the AL-to-PL ratio (K) and the length of anthers (L), styles (M), ovaries (N), and pistils (O) from the wild-type (TS-278) and se3.1. Different letters indicate statistically significant differences (P < 0.05, one-way ANOVA with Tukey’s post hoc test).

Next, we wished to obtain complementary confirmation by mutagenizing SE3.1C in the flush stigma accession TS-278 using CRISPR/Cas9-mediated genome editing (Figure�4I). We sequenced the resulting se3.1 alleles in independent se3.1 mutants and selected three mutants (se3.1-1, se3.1-3, se3.1-4) for further characterization (Figure�4I). The se3.1 mutants all developed inserted stigmas and had AL/PL ratios over 1.0 (Figure�4, J and K). The reduced degree of stigma exsertion in these se3.1 mutants was related to longer anthers and shorter styles (Figure�4, L–O). Taken together, these results indicate that SE3.1 controls the degree of stigma exsertion mainly by regulating the length of anthers and styles. Moreover, the SE3.1C allele contributes to the development of flush stigmas, while the loss-of-function SE3.1 alleles lead to the development of inserted stigmas.

To gain insight into the morphogenesis of the degree of stigma exsertion, we dissected the flowers at different developmental stages from one SE3.1C-overexpressing line and its wild-type TS-9, as well as from the se3.1-1 mutant and its wild-type TS-278. We selected inflorescences containing one flower at anthesis. The developmental process included nine stages: anthesis (A) and eight stages prior to anthesis (Figure�5). In contrast to the wild-type TS-9, which developed inserted stigmas at all developmental stages, the SE3.1C-overexpressing line displayed flush stigmas from the A –4 stage to anthesis and inserted stigmas from the A –8 to the A –5 stages (Figure�5A). Conversely, the loss of SE3.1C led to the development of inserted stigmas at all stages. In contrast, the wild-type TS-278 developed flush stigmas from the A –4 stage to anthesis (Figure�5B). These results indicate that the degree of stigma exsertion is determined around the middle stages of flower development.

Figure 5.

Figure 5

Open in a new tab

Stigma exsertion degree at different flower developmental stages. A, Internal sections of flowers at different developmental stages from a line overexpressing SE3.1C and the wild-type (TS-9). B, Internal sections of flowers at different developmental stages from a se3.1-cr line and the wild-type (TS-278). All flowers from each genotype were from a single inflorescence. One representative flower at each of the eight stages (A-8 to A-1) prior to anthesis and at anthesis is shown. The images were collected with a Leica DFC7000 GT microscope.

SE3.1 promotes self-fertilization by reducing stigma exsertion degree in cultivated tomato

Different degrees of herkogamy are strongly correlated with self-fertilization rates in Arabidopsis (Luo and Widmer, 2013). A previous study showed that fruit set is greater in cultivated tomatoes with inserted stigmas relative to those with flush stigmas (Rick and Dempsey, 1969). To assess the influence of SE3.1 on both the degree of stigma exsertion and the self-fertilization rate, we quantified fruit sets by normalizing the numbers of fruits produced by an inflorescence to the number of flowers borne by that inflorescence. We discovered that the self-fertilization rates of se3.1 mutants are higher relative to the wild-type TS-278 (Figure�6). Combined with the reduced degree of stigma exsertion seen in se3.1 mutants (Figure�4, J and K), these data indicate that SE3.1 influences the self-fertilization rate by regulating the degree of stigma exsertion.

Figure 6.

Figure 6

Open in a new tab

Elevated self-fertilization rates in se3.1 mutants. A, Mean self-fertilization rates from the wild-type (TS-278) and se3.1 mutants. Fruit numbers were counted from the second or third inflorescence (n = 9–20 inflorescences per line). Error bars indicate standard error. Different letters indicate statistically significant differences (P < 0.0001, one-way ANOVA with Tukey’s post hoc test). B, Representative images of fruits on an inflorescence from the wild-type (TS-278) and the se3.1 mutant.

The two-step transition from stigma exsertion to insertion requires SE3.1 and Style2.1

It was previously reported that Style2.1 is associated with a major quantitative trait locus directly regulating stigma exsertion, whereby a 450-bp deletion in the Style2.1 promoter contributed to the evolution of self-pollinating varieties with inserted stigmas from cross-pollinating varieties with exserted stigmas (Chen et al., 2007). Here, our results demonstrated that SE3.1 may have also been selected during the breeding process. To specifically address this question, we measured nucleotide diversity (π) in PIM, CER, and BIG accessions and calculated the ratios of nucleotide diversity between groups to detect putative domestication (πPIM/πCER) and improvement (πCER/πBIG) sweeps (Supplemental Data Sets S4 and S5). We detected a single drastic reduction in the genomic region harboring SE3.1 when comparing nucleotide diversity between CER and BIG accessions (πCER/πBIG=13.87; Figure�7A;Supplemental Data Set S4–S7 and “Materials and Methods”). These data provide evidence that SE3.1 is included in an improvement sweep in this region. In agreement, the group of wild accessions consisted of 84% accessions with the SE3.1C allele and 16% with the SE3.1T allele, with a gradual increase in the relative representation of the SE3.1T allele to 20.4% in the CER and 51% in the BIG groups (Figure�7B).

Figure 7.

Figure 7

Open in a new tab

Two-step transition from exserted to inserted stigmas requires SE3.1 and Style2.1. A, Ratios of nucleotide diversity (π) between PIM and CER, or between CER and BIG accessions along chromosome 3. The red horizontal lines indicate top 5% threshold for entire genome (πPIM/πCER = 2.48 for domestication and πCER/πBIG = 8.52 for improvement). Red arrows, SE3.1. B, Distribution of SE3.1C and SE3.1T alleles among wild, CER, and BIG accessions. C, Distribution of the ratio between anther length and pistil length as a function of the genotype at Style2.1 and SE3.1, shown as box plots. Style2.1-Long allele, full Style2.1 promoter; Style2.1-Short allele, Style2.1 promoter with the 450-bp deletion. The number of accessions number is given on the graph. D, Distribution of the three groups of accessions, Style2.1 and SE3.1 alleles across the 277 tomato accessions as a function of the ratio between anther length and pistil length. E, Genotyping of various accessions for the 450-bp InDel in the Style2.1 promoter. F, Internal sections of flowers from the representative accessions from the three groups genotyped in (E). G, Internal sections of flowers from the wild-type (TS-19) and mutants generated by CRISPR/Cas9 (se3.1, style2.1, and the se3.1 style2.1 double mutant), collected with a Leica DFC7000 GT microscope. H, Distribution of the ratio between anther length and pistil length from the wild-type (TS-19) and mutants, shown as box plots. Different letters indicate statistically significantly differences (P < 0.0001, one-way ANOVA with Tukey’s post hoc test)

To determine the association between both SE3.1 and Style2.1 and the degree of stigma exsertion, we genotyped the 277 tomato accessions at SE3.1 and Style2.1 (Supplemental Data Sets S3 and S8). In general, the SE3.1C allele and the long allele of Style2.1 (without the 450-bp deletion in its promoter) were both present in the accessions with exserted stigmas, mainly in the wild and CER groups. The 450-bp Style2.1 deletion was, however, detected in accessions with flush stigmas from the CER and BIG groups. Furthermore, stigmas were inserted when the 450-bp Style2.1 deletion and the SE3.1T allele were both present in BIG accessions (Figure�7, C–F). To independently test this association between genotype and phenotype, we generated se3.1 and style2.1 single mutants using CRISPR/Cas9 genome editing in the representative exserted stigma accession TS-19 from the PIM group (Supplemental Figure S8). We then obtained se3.1 style2.1 double mutants by crossing se3.1 and style2.1 single mutants. From the exserted stigmas of the wild-type TS-19, style2.1 mutants exhibited flush stigmas, while the se3.1 style2.1 double mutants had inserted stigmas. However, se3.1 single mutants in the TS-19 background only showed the exserted stigmas of its parental background (Figure�7, G and H). Based on these data, we conclude that mutations in both genes were required to induce a transition from exserted to inserted stigmas in PIM accessions, first inactivating Style2.1, then SE3.1. Furthermore, we overexpressed Style2.1 in the inserted stigma accession TS-9 from the BIG group, but did not observe any variation in the degree of stigma exsertion (Supplemental Figure S9). Taken together, these results indicate that the transition from exserted to inserted stigmas required two steps. The first step—from exserted to flush stigmas—required a loss-of-function mutation in Style2.1. The second step—from flush to inserted stigmas—required a loss-of-function mutation in SE3.1 (Figure 8).

Figure 8.

Figure 8

Open in a new tab

Proposed two-step model for the variation in the degree of stigma exsertion in tomato. Schematic representations of tomato flowers with exserted (S. pimpinellifolium and the other wild species), flush (S. lycopersicum var. cerasiforme), or inserted (S. lycopersicum) stigmas based on the relative differences in anther, style, and ovary lengths. The drawing illustrates the changes in degree of stigma exsertion that occurred during the domestication and improvement of tomato. The loss of Style2.1 in S. lycopersicum var. cerasiforme resulted in the transition from exserted to flush stigmas. Subsequently, the SE3.1T allele in S. lycopersicum caused the transition from flush to inserted stigmas. Introducing the SE3.1C allele in accessions homozygous for the SE3.1T allele converted inserted stigmas to flush stigmas. Introducing the wild-type Style2.1 in accessions homozygous for the SE3.1T allele did not affect the degree of stigma exsertion. In contrast, introducing the wild-type Style2.1 in accessions homozygous for the SE3.1C allele resulted in exserted stigmas. The function of SE3.1C might be suppressed by another gene (SSE3) in particular accessions of processing tomato.

Discussion

Although S. pimpinellifolium can self-pollinate, its outcrossing rates are tremendously variable among accessions, ranging from 0% to 80%. The outcrossing rates were strongly correlated with both anther length and the degree of stigma exsertion (Rick et al., 1977). A comparison with fertility revealed a negative correlation between the degree of stigma exsertion and fruit set. In accessions with stigmas that exserted more than 1.5 mm above the surface of the anther tube, the mean fruit set was 51.7%. In accessions with stigmas that exserted less than 1.5 mm, the mean fruit set was 75.2%. These findings are consistent with particular degrees of stigma exsertion interfering with self-pollination and, therefore, increasing the rates of cross-pollination (Rick et al., 1978). Thus, the efficiency of self-pollination is highly variable in S. pimpinellifolium. During the domestication process, the position of the stigma shifted from the outside to the inside of the anther tube (Rick, 1950), which is consistent with our observations of the populations used in this study (Figure�1, D and I). Using two cultivated tomatoes, one with a flush stigma and one with an inserted stigma, a previous study showed that fruit sets were higher for plants with more inserted stigmas relative to those with flush stigmas. These data are consistent with the effective promotion of self-pollination from accessions with stigmas developing within anther tubes, resulting in greater fruit bearing capacity (Rick and Dempsey, 1969). In this study, we identified one gene that controls the transition from flush to inserted stigmas and named this gene SE3.1. We used CRISPR/Cas9 genome editing to genetically inactivate SE3.1 in an accession with flush stigmas and observed that se3.1 mutants develop inserted stigmas (Figure�4J) and that fruit sets increase in the mutants (Figure�6). Therefore, we conclude that the degree of stigma exsertion strongly influenced the mating system in tomato, which provides a molecular basis for selecting materials with inserted stigmas to improve production (Rick and Dempsey, 1969).

The gene controlling the conversion from flush stigmas to inserted stigmas was identified by GWAS. The lead SNP (ch03_53874338) was located in Solyc03g098070 (SE3.1) and was significantly associated with variation in the AL/PL ratio (Figure�2). We revised the gene model for SE3.1 by a combination of reverse transcription polymerase chain reaction (RT-PCR), sequencing, and transient infiltration assays and detection of the encoded protein. Using the correct gene model, we showed that one of the alleles represented by the lead SNP creates a premature termination codon in SE3.1 (Figure�3). Additional functional complementary experiments demonstrated that SE3.1 helps convert flush stigmas to inserted stigmas (Figure�4). We also noticed that the degree of stigma exsertion is determined at a middle stage of flower development, before anthesis (Figure�5). These findings are consistent with previous work indicating that the position of the stigma can be ascertained at an early stage of flower development, which helps breeders in selecting materials with inserted stigmas prior to anthesis (Rick and Dempsey, 1969).

Tomato accessions with inserted stigmas became predominant during the domestication and improvement of tomato (Figure�1, D and I). Our analysis of tomato genomes indicated that the loss-of-function allele of SE3.1 (SE3.1T) was selected during tomato improvement (Figure�7A;Supplemental Data Sets S4–S7). Furthermore, our study of the function of Style2.1 and SE3.1 led us to propose a two-step model for the transition from exserted stigmas to inserted stigmas during the domestication and improvement of tomato. A previous study demonstrated that Style2.1 regulates the development from exserted to inserted stigma (Chen et al., 2007). In our study, we mutagenized Style2.1 in TS-19, an accession from the PIM group with an exserted stigma. The style2.1 mutants developed flush stigmas. The stigmas became inserted in style2.1 se3.1 double mutants (Figure�7, G and H). Thus, Style2.1 contributed to the transition from exserted to flush stigmas, while SE3.1 contributed to the transition from flush to inserted stigmas. We conclude that the inactivation of both Style2.1 and SE3.1 was necessary and sufficient to convert exserted stigmas to inserted stigmas during the domestication and improvement of tomato (Figure 8).

During crop domestication, several typical traits have been selected to meet the requirements of human society and agricultural practices, such as larger fruits or grains, more robust growth with reduced lateral branching and more determinate growth (Doebley et al., 2006). In tomato, plant morphology and fruit size have undergone tremendous changes during domestication and improvement (Lin et al., 2014). TS-19 plants have a bushy growth phenotype and produce small fruits (less than 1cm in diameter). In our study, the mutants in the TS-19 background were less bushy and produced larger fruits. Indeed, se3.1 style2.1 double mutants displayed fewer branches and larger fruits relative to the single mutants (Supplemental Figure S10). Therefore, the function of SE3.1 in controlling stigma exsertion degree and the potential function in controlling plant morphology and fruit size suggest the selection of SE3.1 during tomato improvement.

Processing tomatoes have high soluble solid and lycopene contents and they were bred for the production of tomato paste and other processed products. Processing tomatoes are distinct at both the phenotypic and genomic levels compared to other tomatoes that are cultivated for fresh consumption (Lin et al., 2014). In the majority of processing tomatoes, the degree of stigma exsertion is mostly associated with SE3.1, which we also observed in the se3.1 mutant lines in the TS-278 (processing tomato) background (Supplemental Data Set S1; Figure�4J). However, we determined that 22 tomatoes and 16 accessions known as processing tomatoes (e.g. M82) harbor the SE3.1C allele and yet have inserted stigmas (Supplemental Data Sets S1 and S3). The processing tomato accession M82 was previously demonstrated to develop exserted stigmas when transformed with a Style2.1 transgene (Supplemental Figure S2; Chen et al., 2007). When we overexpressed Style2.1 in the TS-9 background (a fresh consumption accession harboring the SE3.1T allele), the degree of stigma exsertion was not affected (Supplemental Figure S9). These data indicate that Style2.1 function might depend on SE3.1C in cultivated tomato and is consistent with our two-step model (transition from inserted to exserted stigmas), which indicates that SE3.1 controls the transition from inserted stigmas to flush stigmas, and that the presence of both SE3.1 and Style2.1 promote the development of exserted stigmas. Additionally, we speculate that in some processing tomatoes, SE3.1C function might be suppressed by an alternative gene that we refer to as SUPPRESSOR OF SE3.1 (SSE3). We propose that in this sense, the genetic background of processing tomatoes is different from other accessions in the BIG group. More experiments are required to test this hypothesis.

In summary, we identified the stigma exsertion gene SE3.1 using GWAS and revised its gene model. We determined that a SNP created a premature termination codon in SE3.1 that leads to the development of inserted stigmas in S. lycopersicum. The transition to inserted stigmas substantially improved fruit production. In addition, the degree of stigma exsertion is established during an early stage of flower development in tomato. Intriguingly, we discovered that SE3.1 and Style2.1 both contributed to the transition from exserted to inserted stigmas during the domestication and improvement of tomato. Style2.1 controls the first step—the conversion of exserted to flush stigmas—while SE3.1 regulates the second step—the conversion of flush to inserted stigmas. Taken together, our findings support a model for a two-step transition from exserted to inserted stigmas (Figure 8) and the idea that reduced stigma exsertion degree in tomato has a strong influence on the efficiency of self-pollination.

Materials and methods

Plant materials and growth conditions

A population consisting of 277 tomato accessions was used for GWAS. The accessions included one S. cheesmaniae (wild), 24 S. pimpinellifolium (PIM), 103 S. lycopersicum var. cerasiforme (CER), and 149 S. lycopersicum (BIG) accessions (Supplemental Data Set S1 and Supplemental Figure S1). These accessions are part of a diversity panel of 360 accessions and the groups indicated in parentheses are the same abbreviations used previously (Lin et al., 2014). Tomato plants were grown in Wuhan, China, at two independent locations: an open field at Huazhong Agriculture University and a greenhouse at Zhongdu Seed Company from March to June in 2013. Tomato seeds were sown in 5 � 10 plugs in the greenhouse; the plants were transplanted to the field after one month. For each accession, about 12 seedlings were planted at a distance of 35 cm between plants and 50 cm between rows in six rows. Tomato plants were grown from March to June, mainly the long day period of a year. For growth and phenotyping, two replicates were performed at each location. In the greenhouse, plants were grown under a photoperiod consisting of 16 h of light (150 μmol m−2 s−1) and 8 h of darkness at 25 � 2�C and a relative humidity of 70%.

Plant phenotyping

We used two indices for evaluating the phenotype. To calculate the ratio between anther length and pistil length, we measured the lengths of the relevant flower parts (i.e. anthers, styles, and ovaries) in a minimum of three individuals for each accession. At least three fully open flowers from the second or third inflorescence were freshly dissected and the lengths of the anthers, styles and ovaries were measured using Vernier calipers. AL was measured from the top of the anther to the bottom of the ovary; SL was measured from the top of stigma surface to the top of the ovary, and OL was estimated from the longitudinal diameter of the ovary. PL was defined as style length plus ovary length (SL + OL; Figure�1C). We then divided anther length by pistil length to calculate the AL/PL ratio for each flower. As a second index, we visually inspected the flowers and determined an SED based on the positions of the stigmas relative to the surface of anthers. SED was divided into three categories: inserted, flush, and exserted (Figure�1B).

Phylogenetic analysis

A subset of 16,503 SNPs at four-fold-degenerate sites (MAF >5% and missing rates <10%) across the 277 tomato accessions was obtained from a previous study (Lin et al., 2014) and was used to build a maximum likelihood tree. We constructed the phylogenetic tree using the IQ-TREE software with a GTR+ASC model and 1,000 bootstrap (Nguyen et al., 2015).

Genome-wide association

Using 4,856,876 SNPs with MAFs >5% (minor allele with homozygosity in at least 10 accessions) from the 277 tomato accessions were obtained from a previous study (Lin et al., 2014), and were used for genome-wide association. The Beagle software was used to impute missing genotypes (Browning and Browning, 2007). The AL/PL ratio from the two independent locations and the averages of the two locations were used as phenotypes. The EMMAX algorithm was used to perform the association analyses (Kang et al., 2010). The matrix of pairwise genetic distance was used as the variance–covariance matrix for random effects. The first 10 principal components were included as fixed effects. Based on the number of SNPs analyzed (4,856,876), the suggestive and significant thresholds were estimated to be approximately 2.06 � 10−7 (1/4,856,876) and 1.03 � 10−8 (0.05/4,856,876), respectively. LD and haplotype blocks were constructed using the LDBlockShow software (Dong et al., 2020). The physical positions of the SNPs were identified from the tomato genome sequence, version SL 2.40 (http://solgenomics.net/).

Gene reannotation

In the current ITAG2.3 annotation (genome release SL2.40), the lead SNP (ch03_53874338) is located 3-bp upstream of the predicted start codon (ATG) of Solyc03g098070. The position of the first base of the predicted start codon in Solyc03g098070 is defined as 1. Therefore, the position of the lead SNP is –3. The sequence of the reference genome is from tomato accession Heinz 1706 (TS-253; Consortium, 2012). The reference genome harbors a thymine (T) at the position of the lead SNP, but is replaced by a cytosine (C) in other accessions (Figure�3A). In the later ITAG3.2 annotation, another gene model (Solyc03g098075) was added immediately upstream of Solyc03g098070 and overlaps with its 5′-UTR (Figure�3B). Positions in Figure�3C are all relative to the position of the first base of the predicted ATG of Solyc03g098070 from the ITAG2.3 and ITAG3.2 annotations.

Quantitative real-time PCR analysis

Total RNA was extracted using the TRIzol reagent (Invitrogen, USA). The HiScript II 1st Strand cDNA Synthesis Kit (+gDNA wiper) (Vazyme, China) was used to remove trace genomic DNA and synthesize first-strand cDNAs according to the manufacturer’s protocol. The relative transcript levels of specific genes were determined by RT-qPCR on a LightCycler480 System (Roche, Switzerland). The Q-actin (Solyc11g008430) gene was used as reference. Primer sequences are listed in Supplemental Table S1.

Agrobacterium-mediated transient infiltration

The genomic sequences, beginning from the ATG (i.e. from –312 bp according to current ITAG2.3 annotation for Solyc03g098070) and ending before the stop codon (i.e. +1,741 bp) of SE3.1C and SE3.1T, or beginning from the 5′-UTR (i.e. from –508 bp) and ending before the stop codon (i.e. +1,741 bp) of SE3.1C, and the ORF of SE3.1C (without the stop codon) were cloned individually into the pH7LIC4.1 vector in-frame with a sequence encoding a 3xFLAG tag to yield the constructs 1-SE3.1C, 2-SE3.1T, 3-SE3.1C, and SE3.1C-ORF (Figure�3C). All primer sequences are listed in Supplemental Table S1. The resulting constructs were introduced into Agrobacterium (Agrobacterium tumefaciens) strain GV3101 by electroporation. Agrobacteria were then grown in LB medium overnight and resuspended in infiltration buffer (10-mM MgCl2, 10-mM MES-KOH, pH 5.7, and 150–200-μM acetosyringone) to a final OD600 of 0.5 before infiltrating N. benthamiana leaves (Song et al., 2021). The leaf tissue was harvested for assays 2 days after infiltration.

Protein extraction and immunoblotting

Total proteins were extracted in extraction buffer (50-mM Tris–HCl, pH 7.5, 150-mM NaCl, 10% [v/v] glycerol, 1% [w/v] PVPP, 2-mM DTT, 5-mM EDTA, and 10 μL�mL−1 plant protease inhibitor cocktail). After centrifugation at 13,000g for 30 min at 4�C, supernatants were mixed with 5� sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) loading buffer and boiled for 5 min. The boiled samples were resolved by SDS–PAGE and transferred to PVDF membranes for 1–2 h. The membranes were blocked with blocking buffer (1� Tris buffered saline (TBS) with 0.05% [v/v] Tween-20 and 5% [w/v] nonfat dry milk) for 1–2 h at room temperature. Subsequently the blocked membranes were incubated with either mouse anti-actin (1:1,000 dilution; Catalog number A01050, Abbkine, China) or anti-FLAG (1:2,000 dilution; Catalog Number A8592, Sigma, Germany) antibodies for another 1–2 h. The membranes were washed six times with 1� TBST buffer for 5 min each time. The membrane that was incubated with anti-actin antibody was subsequently incubated for 1 h in blocking buffer containing an anti-mouse IgG (H + L) HRP Conjugate (MBL, Japan).

GUS staining

The promoter sequence (3,000-bp upstream of the reannotated ATG) from SE3.1 was PCR amplified from TS-278 genomic DNA and cloned into the pHELLSGATE8 vector lacking the CaMV 35S promoter to yield a GUS reporter construct. Primer sequences are listed in Supplemental Table S1. To detect GUS activity, transgenic lines harboring the SE3.1pro:GUS construct were incubated in GUS staining solution (100-mM sodium phosphate buffer, pH 7.0, 2-mM X-gluc, 10-mM Na2EDTA, 0.5-mM K3Fe(CN)6, 0.5-mM K4Fe(CN)6, 0.1% [v/v] Triton X-100) at 37�C overnight in the dark. The incubated samples were then cleared in 95% (v/v) ethanol and 70% ethanol.

Generation of transgenic tomato

The genomic regions, starting from the 5′-UTR of Solyc03g098070 based on the ITAG2.3 annotation (Figure�3A), or from the 5′-UTR of the reannotated SE3.1 gene structure and ending at the 3′-UTR, were PCR amplified from TS-278 genomic DNA and cloned downstream of the cauliflower mosaic virus (CaMV) 35S promoter in pHELLSGATE8 to yield the 35Spro:3070C and 35S:proSE3.1C constructs, respectively. The genomic region starting from the 5′-UTR of the reannotated SE3.1 gene structure and ending at the 3′-UTR was PCR amplified from TS-9 genomic DNA and cloned into pHELLSGATE8 downstream of the 35S promoter to yield the 35Spro:SE3.1T construct. Finally, the genomic region starting from 2,000-bp upstream of the 5′-UTR from SE3.1C and ending at the 3′-UTR was PCR amplified from TS-278 genomic DNA and cloned into pHELLSGATE8 lacking the 35S promoter to yield the SE3.1Cpro:SE3.1C construct (Figure�4A). The tomato accession TS-9 was transformed with these four constructs using Agrobacterium (strain C58)-mediated transformation (Ouyang et al., 2005). We used the CRISPR-Cas9 binary vector pTX to introduce mutations in the second and third exons of SE3.1C using the simple guide RNAs (sgRNA) sgRNA1 and sgRNA2 (Figure�4I). The sgRNAs were designed at CRISPR-P v2.0 (http://crispr.hzau.edu.cn/CRISPR2/).

The Style2.1 coding sequence was amplified from TS-19 cDNA and cloned into pHELLSGATE8 downstream of the CaMV 35S promoter to yield the overexpression construct 35Spro:Style2.1. TS-9 was transformed with 35S:Style2.1 using Agrobacterium strain C58. We used the CRISPR-Cas9 binary vector pTX to generate style2.1 mutants in the TS-19 background by targeting two sites in the first exon of Style2.1 using sgRNA3 and sgRNA4 (Supplemental Figure S8B), which were designed at CRISPR-P v2.0 (http://crispr.hzau.edu.cn/CRISPR2/).

To generate the se3.1 style2.1 double mutant in the TS-19 background, we first used the CRISPR-Cas9 binary vector (pTX) to generate the se3.1 mutant in the TS-19 background by targeting two sites in the third exon of SE3.1C using sgRNA5 and sgRNA6 (Supplemental Figure S8A), which were designed at CRISPR-P v2.0 (http://crispr.hzau.edu.cn/CRISPR2/). We then crossed the style2.1 mutant to the se3.1 mutant. We propagated the progeny from this cross and identified homozygous double mutants lacking the Cas9 transgene in the T2 generation by genotyping PCR. Primer and sgRNA sequences are listed in Supplemental Table S1.

Detection of domestication and improvement sweeps

To detect genomic regions underlying tomato domestication and improvement, we measured the level of nucleotide diversity (π) in PIM, CER and BIG lines using VCFtools in 50-kb windows, with a step size of 5 kb (Danecek et al., 2011). We calculated the ratios of nucleotide diversity between PIM and CER (πPIM/πCER) and between CER and BIG (πCER/πBIG). We selected windows with the top 5% of ratios (2.48 and 8.52 for domestication and improvement, respectively) for further analysis.

Statistical analysis

All statistical analyses were performed in GraphPad Prism version 8.3.0 (GraphPad Software, Inc., La Jolla, CA: http://www.graphpad.com/). One-way ANOVA with Tukey’s post hoc test and two-tailed t test were used in this study. The results of the statistical tests are provided in Supplemental Data Set S9.

Accession numbers

The sequence data used in this article can be found in the Sol Genomics Database (https://solgenomics.net/) with the following accession numbers: SE3.1 (Solyc03g098070), Style2.1 (Solyc02g087860), Q-actin (Solyc11g008430).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Phylogenetic analysis of the 277 tomato accessions.

Supplemental Figure S2. Frequency distribution of AL, SL, OL and PL in the population.

Supplemental Figure S3. Loci that regulate the AL/PL ratio identified using GWAS using phenotypic data collected from two independent locations.

Supplemental Figure S4. Expression pattern of SE3.1 in flowers.

Supplemental Figure S5. cDNA sequence from SE3.1 amplified from different accessions.

Supplemental Figure S6. Analysis of the protein sequence and structure of SE3.1.

Supplemental Figure S7. Floral traits from the wild type (TS-9) and transgenic lines, 35Spro:SE3.1T and 35Spro: 3070C.

Supplemental Figure S8. Generation of different mutants in TS-19 using the CRISPR-Cas9 system.

Supplemental Figure S9. Functional characterization of Style2.1.

Supplemental Figure S10. Potential function of SE3.1 during tomato domestication and improvement.

Supplemental Table S1. List of primers used in this study.

Supplemental Data Set S1. Accessions used in this study and phenotypic data.

Supplemental Data Set S2. Significant SNPs within a 2-Mb window of the lead SNP on chromosome 3.

Supplemental Data Set S3. Distribution of SE3.1 alleles in the population.

Supplemental Data Set S4. Putative domestication sweeps.

Supplemental Data Set S5. Putative improvement sweeps.

Supplemental Data Set S6. Genes within putative domestication sweeps.

Supplemental Data Set S7. Genes within putative improvement sweeps.

Supplemental Data Set S8. Detection of the 450-bp InDel in the population.

Supplemental Data Set S9. Statistical analysis.

Supplementary Material

koab201_Supplementary_Data
Click here for additional data file. (9.9MB, zip)

Acknowledgments

We thank Prof Hanhui Kuang for critical reading and editing of the manuscript. The pH7LIC4.1 vector was kindly provided by Dr Feng Li.

Funding

This work was supported by grants from the National Natural Science Foundation of China (31991182; 31672149; 31972426), National Key Research and Development Program (2018YFD1000800; 2017YFD0101902;), Fundamental Research Funds for the Central Universities (2662018PY073), China Postdoctoral Science Foundation (2021M691174), and Wuhan Frontier Projects for Applied Foundation (2019020701011492).

Conflict of interest statement. None declared.

 

These authors contributed equally (L.S., J.S.).

 

L.S., J.S., Y.Z., and Z.Y. designed the experiments; L.S., and J.S. performed experiments and wrote the manuscript; L.S., H.Y., and X.W. analyzed data; J.S., C.Y., Y.W., and F. L. helped perform some of the experiments; J.Z., B.O., T.W., Y.L., L.R., Z.Y., and Y.Z. supervised the project and revised the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell) is: Yuyang Zhang (yyzhang@mail.hzau.edu.cn).

References

  1. Bai Y, Lindhout P (2007) Domestication and breeding of tomatoes: what have we gained and what can we gain in the future? Ann Bot 100: 1085–1094 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Browning SR, Browning BL (2007) Rapid and accurate haplotype phasing and missing-data inference for whole-genome association studies by use of localized haplotype clustering. Am J Hum Genet 81: 1084–1097 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Carleial S, van Kleunen M, Stift M (2017) Small reductions in corolla size and pollen: ovule ratio, but no changes in flower shape in selfing populations of the North American Arabidopsis lyrata. Oecologia 183: 401–413 [DOI] [PubMed] [Google Scholar]
  4. Chen KY, Tanksley SD (2004) High-resolution mapping and functional analysis of se2.1: a major stigma exsertion quantitative trait locus associated with the evolution from allogamy to autogamy in the genus Lycopersicon. Genetics 168: 1563–1573 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen KY, Cong B, Wing R, Vrebalov J, Tanksley SD (2007) Changes in regulation of a transcription factor lead to autogamy in cultivated tomatoes. Science 318: 643–645 [DOI] [PubMed] [Google Scholar]
  6. Consortium TG (2012) The tomato genome sequence provides insights into fleshy fruit evolution. Nature 485: 635–641 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Danecek P, Auton A, Abecasis G, Albers CA, Banks E, DePristo MA, Handsaker RE, Lunter G, Marth GT, Sherry ST, et al. (2011). The variant call format and VCFtools. Bioinformatics 27: 2156–2158 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Doebley JF, Gaut BS, Smith BD (2006) The molecular genetics of crop domestication. Cell 127: 1309–1321 [DOI] [PubMed] [Google Scholar]
  9. Dong S-S, He W-M, Ji J-J, Zhang C, Guo Y, Yang T-L (2020) LDBlockShow: a fast and convenient tool for visualizing linkage disequilibrium and haplotype blocks based on variant call format files. Brief Bioinform 22:�bbaa227 [DOI] [PubMed] [Google Scholar]
  10. Fulton TM, Beck-Bunn T, Emmatty D, Eshed Y, Lopez J, Petiard V, Uhlig J, Zamir D, Tanksley SD (1997) QTL analysis of an advanced backcross of Lycopersicon peruvianum to the cultivated tomato and comparisons with QTLs found in other wild species. Theoret Appl Genet 95: 881–894 [Google Scholar]
  11. Georgiady MS, Lord EM (2002) Evolution of the inbred flower form in the currant tomato, Lycopersicon pimpinellifolium. Int J Plant Sci 163: 531–541 [Google Scholar]
  12. Goodwillie C, Sargent RD, Eckert CG, Elle E, Geber MA, Johnston MO, Kalisz S, Moeller DA, Ree RH, Vallejo-Marin M, et al. (2010). Correlated evolution of mating system and floral display traits in flowering plants and its implications for the distribution of mating system variation. New Phytol 185: 311–321 [DOI] [PubMed] [Google Scholar]
  13. Gorguet B, Eggink PM, Oca�a J, Tiwari A, Schipper D, Finkers R, Visser RGF, Van Heusden AW (2008) Mapping and characterization of novel parthenocarpy QTLs in tomato. Theoret Appl Genet 116: 755–767 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kalisz S, Vogler D, Fails B, Finer M, Shepard E, Herman T, Gonzales R (1999) The mechanism of delayed selfing in Collinsia verna (Scrophulariaceae). Am J Bot 86: 1239–1247 [PubMed] [Google Scholar]
  15. Kang HM, Sul JH, Service SK, Zaitlen NA, Kong SY, Freimer NB, Sabatti C, Eskin E (2010) Variance component model to account for sample structure in genome-wide association studies. Nat Genet 42: 348–354 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Li W, Chetelat RT (2010) A pollen factor linking inter- and intraspecific pollen rejection in tomato. Science 330: 1827–1830 [DOI] [PubMed] [Google Scholar]
  17. Li W, Chetelat RT (2015) Unilateral incompatibility gene ui1.1 encodes an S-locus F-box protein expressed in pollen of Solanum species. Proc Natl Acad Sci USA 112: 4417–4422 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lin T, Zhu G, Zhang J, Xu X, Yu Q, Zheng Z, Zhang Z, Lun Y, Li S, Wang X, et al. (2014) Genomic analyses provide insights into the history of tomato breeding. Nat Genet 46: 1220–1226 [DOI] [PubMed] [Google Scholar]
  19. Luo Y, Widmer A (2013) Herkogamy and its effects on mating patterns in Arabidopsis thaliana. PLoS One 8: e57902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Michael SG, Richard WW, Elizabeth ML (2002) Genetic analysis of traits distinguishing outcrossing and self-pollinating forms of currant tomato, Lycopersicon pimpinellifolium (Jusl.) Mill. Genetics 161: 333–344 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. M�ller H (1883) The Fertilisation of Flowers. Macmillan and Company, New York [Google Scholar]
  22. Nguyen LT, Schmidt HA, von Haeseler A, Minh BQ (2015) IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol Biol Evol 32: 268–274 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Opedal OH (2018) Herkogamy, a principal functional trait of plant reproductive biology. Int J Plant Sci 179: 677–687 [Google Scholar]
  24. Ouyang B, Chen Y, Li H, Qian C, Huang S, Ye Z (2005) Transformation of tomatoes with osmotin and chitinase genes and their resistance to Fusarium wilt. J Horticult Sci Biotechnol 80: 517–522 [Google Scholar]
  25. Peralta IE, Spooner DM, Knapp S (2008) Taxonomy of wild tomatoes and their relatives (Solanum sect. Lycopersicoides, sect. Juglandifolia, sect. Lycopersicon; Solanaceae). Syst Bot Monogr 84: 1–186 [Google Scholar]
  26. Ranc N, Munos S, Santoni S, Causse M (2008) A clarified position for Solanum lycopersicum var. cerasiforme in the evolutionary history of tomatoes (solanaceae). BMC Plant Biol 8: 130. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Rick CM (1950) Pollination relations of Lycopersicon esculentum in native and foreign regions. Evolution 4: 110–122 [Google Scholar]
  28. Rick CM (1982) Genetic relationships between self-incompatibility and floral traits in the tomato species. Biol Zent Bl 101: 185– 198. [Google Scholar]
  29. Rick CM (1988) Evolution of mating systems in cultivated plants. In Gottlieb LD, Jain SK, eds, Plant Evolutionary Biology. Springer Netherlands, Dordrecht, pp 133–147 [Google Scholar]
  30. Rick CM (1995) Tomato: Lycopersicon esculentum (Solanaceae). Longman Scientific & Technical, London [Google Scholar]
  31. Rick CM, Dempsey WH (1969) Position of the stigma in relation to fruit setting of the tomato. Bot Gazette 130: 180–186 [Google Scholar]
  32. Rick CM, Fobes JF, Holle M (1977) Genetic variation inLycopersicon pimpinellifolium: evidence of evolutionary change in mating systems. Plant Syst Evol 127: 139–170 [Google Scholar]
  33. Rick CM, Holle M, Thorp RW (1978) Rates of cross-pollination in Lycopersicon pimpinellifolium: Impact of genetic variation in floral characters. Plant Syst Evol 129: 31–44 [Google Scholar]
  34. Si L, Chen J, Huang X, Gong H, Luo J, Hou Q, Zhou T, Lu T, Zhu J, Shangguan Y, et al. (2016). OsSPL13 controls grain size in cultivated rice. Nat Genet 48: 447–456 [DOI] [PubMed] [Google Scholar]
  35. Sicard A, Stacey N, Hermann K, Dessoly J, Neuffer B, Baurle I, Lenhard M (2011) Genetics, evolution, and adaptive significance of the selfing syndrome in the genus Capsella. Plant Cell 23: 3156–3171 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Song J, Shang L, Wang X, Xing Y, Xu W, Zhang Y, Wang T, Li H, Zhang J, Ye Z (2021) MAPK11 regulates seed germination and ABA signaling in tomato by phosphorylating SnRKs. J Exp Bot 72: 1677–1690. [DOI] [PubMed] [Google Scholar]
  37. Stebbins GL (1970) Adaptive radiation of reproductive characteristics in angiosperms, I: pollination mechanisms Annu Rev Ecol Syst 1: 307–326 [Google Scholar]
  38. Torang P, Vikstrom L, Wunder J, Wotzel S, Coupland G, Agren J (2017) Evolution of the selfing syndrome: anther orientation and herkogamy together determine reproductive assurance in a self-compatible plant. Evolution 71: 2206–2218 [DOI] [PubMed] [Google Scholar]
  39. Tovar-Mendez A, Kumar A, Kondo K, Ashford A, Baek YS, Welch L, Bedinger PA, McClure BA (2014) Restoring pistil-side self-incompatibility factors recapitulates an interspecific reproductive barrier between tomato species. Plant J 77: 727–736 [DOI] [PubMed] [Google Scholar]
  40. Vallejo-Marin M, Walker C, Friston-Reilly P, Solis-Montero L, Igic B (2014) Recurrent modification of floral morphology in heterantherous Solanum reveals a parallel shift in reproductive strategy. Philos Trans R Soc Lond B Biol Sci 369: 20130256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Vosters SL, Jewell CP, Sherman NA, Einterz F, Blackman BK, Moyle LC (2014) The timing of molecular and morphological changes underlying reproductive transitions in wild tomatoes (Solanum sect. Lycopersicon). Mol Ecol 23: 1965–1978 [DOI] [PubMed] [Google Scholar]
  42. Wang Z, Hong Y, Zhu G, Li Y, Niu Q, Yao J, Hua K, Bai J, Zhu Y, Shi H, et al. (2020) Loss of salt tolerance during tomato domestication conferred by variation in a Na(+)/K(+) transporter. EMBO J 39:e103256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Waterhouse A, Bertoni M, Bienert S, Studer G, Tauriello G, Gumienny R, Heer FT, de Beer TAP, Rempfer C, Bordoli L, et al. (2018) SWISS-MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res 46: W296–W303 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Wozniak NJ, Kappel C, Marona C, Altschmied L, Neuffer B, Sicard A (2020) A similar genetic architecture underlies the convergent evolution of the selfing syndrome in Capsella. Plant Cell 32: 935–949 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Zhang M, Liang X, Wang L, Cao Y, Song W, Shi J, Lai J, Jiang C (2019) A HAK family Na(+) transporter confers natural variation of salt tolerance in maize. Nat Plants 5: 1297–1308 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

koab201_Supplementary_Data
Click here for additional data file. (9.9MB, zip)

Articles from The Plant Cell are provided here courtesy of Oxford University Press

RESOURCES