Bypassing Negative Epistasis on Yield in Tomato Imposed by a Domestication Gene

•Mutations in two MADS-box genes were selected in tomato domestication and breeding•Interaction of the alleles caused extreme inflorescence branching and sterility•Natural and gene-edited MADS-box alleles provide a range of inflorescence types•Dosage from selected alleles allows improved inflorescence architecture and yield Selection for inflorescence architecture with improved flower production and yield is common to many domesticated crops. However, tomato inflorescences resemble wild ancestors, and breeders avoided excessive branching because of low fertility. We found branched variants carry mutations in two related transcription factors that were selected independently. One founder mutation enlarged the leaf-like organs on fruits and was selected as fruit size increased during domestication. The other mutation eliminated the flower abscission zone, providing “jointless” fruit stems that reduced fruit dropping and facilitated mechanical harvesting. Stacking both beneficial traits caused undesirable branching and sterility due to epistasis, which breeders overcame with suppressors. However, this suppression restricted the opportunity for productivity gains from weak branching. Exploiting natural and engineered alleles for multiple family members, we achieved a continuum of inflorescence complexity that allowed breeding of higher-yielding hybrids. Characterizing and neutralizing similar cases of negative epistasis could improve productivity in many agricultural organisms.Video AbstracteyJraWQiOiI4ZjUxYWNhY2IzYjhiNjNlNzFlYmIzYWFmYTU5NmZmYyIsImFsZyI6IlJTMjU2In0.eyJzdWIiOiIwYjhjYWZhMjJiY2Q1NjRjMDU4NmM3Yjk1NzkxYzU5NCIsImtpZCI6IjhmNTFhY2FjYjNiOGI2M2U3MWViYjNhYWZhNTk2ZmZjIiwiZXhwIjoxNjc5MjMwODQwfQ.nSvcV41bviI9EzCZWSdkVokL1gol3kvN7FeIQye1W8TxTGGNq_sbPnSlW6mPXqxt5VZw7rZVBK0nsZRfAqffQLjj23qESbU30e-X_HeUzM5Qt1n2ApRL3Kw5Xn50QLwrDjm0b27LgzZ7YLL1DrpqbQbc1ANZ390oMY2rv7BeKgzr7mTy1_YxwWmYsf3moVFUFcxyiLWO1AdPd_M_yka_LCdJJxWczGREKwpQ2_QMjYnqwWz1kUV9ZwunJa_bogAvPTRJJXApgon83sjWDCoypznXQmLeFauzIihAu3ibrCymSHMvVyPljV0uS2npnBd-PwerpEr9PbnvO8PGf7Ov6g(mp4, (54.32 MB) Download video Selection for inflorescence architecture with improved flower production and yield is common to many domesticated crops. However, tomato inflorescences resemble wild ancestors, and breeders avoided excessive branching because of low fertility. We found branched variants carry mutations in two related transcription factors that were selected independently. One founder mutation enlarged the leaf-like organs on fruits and was selected as fruit size increased during domestication. The other mutation eliminated the flower abscission zone, providing “jointless” fruit stems that reduced fruit dropping and facilitated mechanical harvesting. Stacking both beneficial traits caused undesirable branching and sterility due to epistasis, which breeders overcame with suppressors. However, this suppression restricted the opportunity for productivity gains from weak branching. Exploiting natural and engineered alleles for multiple family members, we achieved a continuum of inflorescence complexity that allowed breeding of higher-yielding hybrids. Characterizing and neutralizing similar cases of negative epistasis could improve productivity in many agricultural organisms. The architectures of plant reproductive shoot systems—inflorescences—are major determinants of crop yield, and modified inflorescence complexity was a recurring target during crop domestication and improvement (Meyer and Purugganan, 2013Meyer R.S. Purugganan M.D. Evolution of crop species: genetics of domestication and diversification.Nat. Rev. Genet. 2013; 14: 840-852Crossref PubMed Scopus (618) Google Scholar). Prominent examples include the cereal crops barley, maize, rice, and wheat, for which humans selected variants with greater branching to increase flower and grain production (Boden et al., 2015Boden S.A. Cavanagh C. Cullis B.R. Ramm K. Greenwood J. Jean Finnegan E. Trevaskis B. Swain S.M. Ppd-1 is a key regulator of inflorescence architecture and paired spikelet development in wheat.Nat. Plants. 2015; 1: 14016Crossref PubMed Scopus (130) Google Scholar, Doebley et al., 1997Doebley J. Stec A. Hubbard L. The evolution of apical dominance in maize.Nature. 1997; 386: 485-488Crossref PubMed Scopus (1090) Google Scholar, Huang et al., 2009Huang X. Qian Q. Liu Z. Sun H. He S. Luo D. Xia G. Chu C. Li J. Fu X. Natural variation at the DEP1 locus enhances grain yield in rice.Nat. Genet. 2009; 41: 494-497Crossref PubMed Scopus (707) Google Scholar, Ramsay et al., 2011Ramsay L. Comadran J. Druka A. Marshall D.F. Thomas W.T.B. Macaulay M. MacKenzie K. Simpson C. Fuller J. Bonar N. et al.INTERMEDIUM-C, a modifier of lateral spikelet fertility in barley, is an ortholog of the maize domestication gene TEOSINTE BRANCHED 1.Nat. Genet. 2011; 43: 169-172Crossref PubMed Scopus (236) Google Scholar). Yet for many crops, particularly fruit-bearing species such as grape and tomato, inflorescence architecture has changed little from wild ancestors and, therefore, has been underexploited in breeding (Mullins et al., 1992Mullins M.G. Bouquet A. Williams L.E. Biology of the Grapevine. Cambridge Univ. Press, 1992Google Scholar, Peralta and Spooner, 2005Peralta I.E. Spooner D.M. Morphological Characterization and Relationships of Wild Tomatoes (Solanum L. Section Lycopersicon).Missouri Botanical Garden Press. 2005; 104: 227-257Google Scholar). Variation in inflorescence architecture is based on changes in the activity of meristems, small groups of stem cells located at the tips of shoots (Kyozuka et al., 2014Kyozuka J. Tokunaga H. Yoshida A. Control of grass inflorescence form by the fine-tuning of meristem phase change.Curr. Opin. Plant Biol. 2014; 17: 110-115Crossref PubMed Scopus (44) Google Scholar, Park et al., 2014aPark S.J. Eshed Y. Lippman Z.B. Meristem maturation and inflorescence architecture--lessons from the Solanaceae.Curr. Opin. Plant Biol. 2014; 17: 70-77Crossref PubMed Scopus (51) Google Scholar). During the transition to flowering, vegetative meristems gradually mature to a reproductive state and, depending on the species, terminate immediately in a flower or give rise to a variable number of new inflorescence meristems that become additional flowers or flower-bearing branches. In domesticated tomato (Solanum lycopersicum) and its wild progenitor S. pimpinellifolium, a new inflorescence meristem emerges at the flank of each previous meristem. Several reiterations of this process give rise to inflorescences with multiple flowers arranged in a zigzag pattern, resulting in the familiar “tomatoes-on-the-vine” architecture (Figure 1A) (Park et al., 2012Park S.J. Jiang K. Schatz M.C. Lippman Z.B. Rate of meristem maturation determines inflorescence architecture in tomato.Proc. Natl. Acad. Sci. USA. 2012; 109: 639-644Crossref PubMed Scopus (125) Google Scholar). Despite a rich resource of wild relatives that develop weakly branched inflorescences with high fertility, improving tomato inflorescence architecture to boost flower production and yield has remained challenging due to genetic incompatibilities and the challenge of transferring complex polygenic traits (Lemmon et al., 2016Lemmon Z.H. Park S.J. Jiang K. Van Eck J. Schatz M.C. Lippman Z.B. The evolution of inflorescence diversity in the nightshades and heterochrony during meristem maturation.Genome Res. 2016; 26: 1676-1686Crossref PubMed Scopus (35) Google Scholar, Lippman et al., 2008Lippman Z.B. Cohen O. Alvarez J.P. Abu-Abied M. Pekker I. Paran I. Eshed Y. Zamir D. The making of a compound inflorescence in tomato and related nightshades.PLoS Biol. 2008; 6: e288Crossref PubMed Scopus (155) Google Scholar, Macarthur and Chiasson, 1947Macarthur J.W. Chiasson L.P. Cytogenetic Notes on Tomato Species and Hybrids.Genetics. 1947; 32: 165-177PubMed Google Scholar). Another potentially valuable source of inflorescence variation is rare natural and induced highly branched mutants in domesticated germplasm. We previously showed that branching in one of these variants and in a related wild species is due to an extended meristem maturation schedule, which allows additional inflorescence meristems to form (Lemmon et al., 2016Lemmon Z.H. Park S.J. Jiang K. Van Eck J. Schatz M.C. Lippman Z.B. The evolution of inflorescence diversity in the nightshades and heterochrony during meristem maturation.Genome Res. 2016; 26: 1676-1686Crossref PubMed Scopus (35) Google Scholar, Park et al., 2012Park S.J. Jiang K. Schatz M.C. Lippman Z.B. Rate of meristem maturation determines inflorescence architecture in tomato.Proc. Natl. Acad. Sci. USA. 2012; 109: 639-644Crossref PubMed Scopus (125) Google Scholar). These findings suggested that subtle modification of meristem maturation could provide beneficial changes in inflorescence architecture (Park et al., 2014aPark S.J. Eshed Y. Lippman Z.B. Meristem maturation and inflorescence architecture--lessons from the Solanaceae.Curr. Opin. Plant Biol. 2014; 17: 70-77Crossref PubMed Scopus (51) Google Scholar). Yet breeders typically select against even moderate branching, primarily due to imbalances in source-sink relationships that cause flower abortion and low fruit production, especially in large-fruited varieties (Stephenson, 1981Stephenson A.G. Flower and fruit abortion: proximate causes and ultimate functions.Annu. Rev. Ecol. Syst. 1981; 12: 253-279Crossref Google Scholar). In this study, we explored a new class of branched variants from a large core collection and discovered mutations in two closely related MADS-box transcription factor genes, one of which arose during domestication and the other within the last century of crop improvement. Each mutation was selected separately for improved flower morphology and fruit retention traits. However, redundant roles in meristem maturation caused undesirable branching upon combining both mutations. Breeders overcame this negative epistasis by selecting suppressors of branching, but in so doing, they limited the potential to improve flower production through weak branching. By dissecting this interaction, we discovered a dosage relationship among natural and gene-edited mutations in multiple regulators of meristem maturation. Combining these mutations in homozygous and heterozygous states allowed us to create a quantitative range of inflorescence types and develop weakly branched hybrids with higher flower and fruit production. To explore the challenges with improving tomato inflorescences, we screened a core collection of 4,193 wild and domesticated accessions for deviation from the typical inflorescence architecture of multiple flowers arranged along a single branch (Figure 1A) (https://unity.phenome-networks.com, see STAR Methods). We previously reported 23 extremely branched accessions that were all defective in the gene COMPOUND INFLORESCENCE (S, homolog of Arabidopsis WUSCHEL-RELATED HOMEOBOX 9, WOX9) (Figure 1B) (Lippman et al., 2008Lippman Z.B. Cohen O. Alvarez J.P. Abu-Abied M. Pekker I. Paran I. Eshed Y. Zamir D. The making of a compound inflorescence in tomato and related nightshades.PLoS Biol. 2008; 6: e288Crossref PubMed Scopus (155) Google Scholar). However, we also found three rare variants not allelic to s that branched less frequently and also lacked the abscission zone on the stems (pedicels) of flowers known as the “joint” (Figures 1C, 1D, and S1A–S1F ). Searching other germplasm sources provided one additional branched jointless mutant derived from an X-ray mutagenesis (Figures S1C and S1F) (Stubbe, 1972Stubbe H. Mutanten der Kulturtomate Lycopersicon esculentum Miller VI.Kulturpflanze. 1972; 16: 185-230Crossref Scopus (10) Google Scholar). Crosses among all four accessions failed to complement (Figures S1G–S1I). Thus, we collectively named these accessions compound inflorescence 2 (s2) and designated one accession as a reference (LA4371, see STAR Methods). An analysis of higher-order mutants between s and s2 showed an additive genetic relationship, indicating that the gene(s) underlying s2 function separately from the S gene (Figures 1C and S1J). We noted during the generation of s s2 plants that s2 segregated at a ratio of ∼1/16 (Figure 1E), suggesting that two unlinked recessive mutations underlie s2 phenotypes. Consistent with this, jointless plants (unbranched and branched) segregated as a single recessive mutation. This jointless trait resembled two classical jointless-2 (j2) mutants reported 50 years ago. The original j2 was discovered in the unbranched wild tomato species S. cheesmaniae from the Galapagos Islands (Rick, 1956aRick C.M. Genetic and Systematic Studies on Accessions of Lycospersicon from the Galapagos Islands.Am. J. Bot. 1956; 43: 687-696Crossref Google Scholar). A second allele arose spontaneously in an agricultural field, but this mutation was also associated with inflorescence branching that caused excessive flower production and poor fruit set due to epistatic interactions with the domesticated germplasm (Reynard, 1961Reynard G.B. New Source of the j2 Gene Governing Jointless Pedicel in Tomato.Science. 1961; 134: 2102Crossref PubMed Scopus (17) Google Scholar, Rick, 1956bRick C.M. A new jointless gene from the Galapagos L. pimpinellifolium.TGC Report. 1956; 6: 23Google Scholar). Breeders selected and utilized unbranched j2 because it reduced fruit dropping and enabled large-scale machine harvesting of processing tomatoes while maintaining good fruit set (Zahara and Scheuerman, 1988Zahara M.B. Scheuerman R.W. Hand-harvesting jointless vs. jointed-stem tomatoes.Calif. Agric. 1988; 42: 14Google Scholar). Notably, the jointless phenotype of s2 was allelic to j2 (Figure S1K), and we failed to find s2 plants with normal pedicels, suggesting that branching required the j2 mutation. We therefore designated the second locus enhancer-of-jointless2 (ej2). To better understand the developmental basis of s2 branching, we examined stages of meristem maturation during early inflorescence development. Tomato inflorescences develop according to the sympodial growth program (Park et al., 2014aPark S.J. Eshed Y. Lippman Z.B. Meristem maturation and inflorescence architecture--lessons from the Solanaceae.Curr. Opin. Plant Biol. 2014; 17: 70-77Crossref PubMed Scopus (51) Google Scholar), in which each vegetative meristem matures into a transition meristem (TM) and terminates in a floral meristem (FM) that produces the first flower of the inflorescence. Additional flowers arise from iterative formation of specialized axillary (sympodial) inflorescence meristems (SIM), resulting in a multi-flowered inflorescence (Figure 1F). In s mutants, both TM and SIM maturation are severely delayed, allowing multiple SIMs to form at each cycle (Figure 1G) (Lippman et al., 2008Lippman Z.B. Cohen O. Alvarez J.P. Abu-Abied M. Pekker I. Paran I. Eshed Y. Zamir D. The making of a compound inflorescence in tomato and related nightshades.PLoS Biol. 2008; 6: e288Crossref PubMed Scopus (155) Google Scholar, Park et al., 2012Park S.J. Jiang K. Schatz M.C. Lippman Z.B. Rate of meristem maturation determines inflorescence architecture in tomato.Proc. Natl. Acad. Sci. USA. 2012; 109: 639-644Crossref PubMed Scopus (125) Google Scholar). Additional SIMs also formed in s2 plants, but fewer than in s (Figure 1H). To determine if s2 was delayed in maturation, we performed RNA-seq on sequential s2 meristem maturation stages and compared transcriptome dynamics with existing maturation profiles for s and wild-type (WT) (see STAR Methods). A principal component analysis (PCA) using 2,582 maturation marker genes (Lemmon et al., 2016Lemmon Z.H. Park S.J. Jiang K. Van Eck J. Schatz M.C. Lippman Z.B. The evolution of inflorescence diversity in the nightshades and heterochrony during meristem maturation.Genome Res. 2016; 26: 1676-1686Crossref PubMed Scopus (35) Google Scholar) showed that meristem maturation in s2 was delayed like in s, and subsets of TM and FM marker genes showed that this delay was weaker than s consistent with less branching in s2 inflorescences (Figures 1I–1K and S2). The j2 mutant was previously mapped to the centromere of chromosome 12, but poor recombination prevented identification of the responsible gene (Budiman et al., 2004Budiman M.A. Chang S.B. Lee S. Yang T.J. Zhang H.B. de Jong H. Wing R.A. Localization of jointless-2 gene in the centromeric region of tomato chromosome 12 based on high resolution genetic and physical mapping.Theor. Appl. Genet. 2004; 108: 190-196Crossref PubMed Scopus (56) Google Scholar, Yang et al., 2005Yang T.J. Lee S. Chang S.-B. Yu Y. de Jong H. Wing R.A. In-depth sequence analysis of the tomato chromosome 12 centromeric region: identification of a large CAA block and characterization of pericentromere retrotranposons.Chromosoma. 2005; 114: 103-117Crossref PubMed Scopus (46) Google Scholar). To clone the genes underlying j2 and ej2, we generated two F2 populations from crossing s2 with the jointed (J2/J2) cultivar M82 and the wild ancestor of tomato, S. pimpinellifolium. In the intra-species F2 population, s2 plants segregated at the expected ratio of ∼1/16, but this segregation was substantially lower in the S. pimpinellifolium population, suggesting unknown modifier loci can suppress s2 branching (Figures S3A–S3C). To map j2 and ej2 simultaneously, we performed genome sequencing on pools of DNA from s2, j2, and WT F2 segregating plants (see STAR Methods). Comparing SNP ratios between s2 and WT pools in both populations revealed a region near the bottom of chromosome 3 and the centromere of chromosome 12 with a strong bias for SNPs from the s2 parent (Figures 2A, S3D, and S3E). SNP ratios between s2 and j2 revealed a bias only on chromosome 3. These results confirmed that j2 is located near the chromosome 12 centromere and revealed that ej2 resides on chromosome 3.Figure 2Mutations in Two SEPALLATA MADS-Box Genes Cause s2 BranchingShow full caption(A) Mapping-by-sequencing of s2. Ratio of SNP-ratios (s2/M82) between different pools of segregating phenotypic classes (top: s2/WT; middle: s2/j2; bottom: j2/WT) is shown for chromosome 3 and 12.(B) The j2 mapping interval includes the SEP4 homolog Solyc12g038510.(C) Genomic sequencing reads (left) and PCR (right) showing a Copia-like Rider transposon insertion in the first intron of Solyc12g038510 in s2 mutants.(D) Sashimi plots of Solyc12g038510 RNA-seq reads in WT (top) and s2 (bottom) floral meristems. An intronic transcriptional start site leads to out-of-frame Solyc12g038510 transcripts in s2 mutants. Numbers indicate reads per million (RPM) supporting splice-junctions, and alternative s2 splicing is highlighted in red.(E) Generation of j2CR null mutations by CRISPR/Cas9 using two single-guide RNAs (sgRNA, target1, and target2; red arrows). Black arrows indicate forward (F) and reverse (R) primers used for genotyping and sequencing. Sequences of j2CR allele 1 (a1) and a2 are shown. sgRNA targets and protospacer-adjacent motif (PAM) are indicated in red and bold font, respectively, and deletions by blue dashes. Insertions are indicated in blue, and sequence gap length is shown in parentheses.(F) Unbranched inflorescences and fruits from WT and j2CR mutants showing WT jointed (green asterisks) and j2CR jointless (red asterisks) pedicels.(G) Complementation test between j2CR and j2TE (jointless pedicels; red asterisks).(H) The ej2 mapping interval includes the SEP4 homolog Solyc03g114840.(I) Genomic sequencing reads (left) and PCR (right) in s2 mutants, revealing a 564 bp insertion in the fifth intron of Solyc03g114840.(J) Sashimi plots, as in (D), for Solyc03g114840 RNA-seq reads in WT and s2 floral meristems, indicating partial exon skipping and intron retention in s2 mutants.(K) Generation of ej2CR null mutations by CRISPR/Cas9.(L) Unbranched ej2CR mutant inflorescences with extremely long sepals (green arrowheads) and pear-shaped fruits. Scale bars, 1 cm.(M) Unopened flowers showing that the weak natural ej2w allele causes longer sepals and fails to complement ej2CR.(N) Quantification of relative sepal length (sepal length/petal length ± SEM, N, number of flowers) for genotypes in (M). P, two-tailed, two-sample t test compared to WT.See also Figure S3 and STAR Methods.View Large Image Figure ViewerDownload Hi-res image Download (PPT) (A) Mapping-by-sequencing of s2. Ratio of SNP-ratios (s2/M82) between different pools of segregating phenotypic classes (top: s2/WT; middle: s2/j2; bottom: j2/WT) is shown for chromosome 3 and 12. (B) The j2 mapping interval includes the SEP4 homolog Solyc12g038510. (C) Genomic sequencing reads (left) and PCR (right) showing a Copia-like Rider transposon insertion in the first intron of Solyc12g038510 in s2 mutants. (D) Sashimi plots of Solyc12g038510 RNA-seq reads in WT (top) and s2 (bottom) floral meristems. An intronic transcriptional start site leads to out-of-frame Solyc12g038510 transcripts in s2 mutants. Numbers indicate reads per million (RPM) supporting splice-junctions, and alternative s2 splicing is highlighted in red. (E) Generation of j2CR null mutations by CRISPR/Cas9 using two single-guide RNAs (sgRNA, target1, and target2; red arrows). Black arrows indicate forward (F) and reverse (R) primers used for genotyping and sequencing. Sequences of j2CR allele 1 (a1) and a2 are shown. sgRNA targets and protospacer-adjacent motif (PAM) are indicated in red and bold font, respectively, and deletions by blue dashes. Insertions are indicated in blue, and sequence gap length is shown in parentheses. (F) Unbranched inflorescences and fruits from WT and j2CR mutants showing WT jointed (green asterisks) and j2CR jointless (red asterisks) pedicels. (G) Complementation test between j2CR and j2TE (jointless pedicels; red asterisks). (H) The ej2 mapping interval includes the SEP4 homolog Solyc03g114840. (I) Genomic sequencing reads (left) and PCR (right) in s2 mutants, revealing a 564 bp insertion in the fifth intron of Solyc03g114840. (J) Sashimi plots, as in (D), for Solyc03g114840 RNA-seq reads in WT and s2 floral meristems, indicating partial exon skipping and intron retention in s2 mutants. (K) Generation of ej2CR null mutations by CRISPR/Cas9. (L) Unbranched ej2CR mutant inflorescences with extremely long sepals (green arrowheads) and pear-shaped fruits. Scale bars, 1 cm. (M) Unopened flowers showing that the weak natural ej2w allele causes longer sepals and fails to complement ej2CR. (N) Quantification of relative sepal length (sepal length/petal length ± SEM, N, number of flowers) for genotypes in (M). P, two-tailed, two-sample t test compared to WT. See also Figure S3 and STAR Methods. MADS-box transcription factors are known to contribute to pedicel abscission zone development in tomato (Liu et al., 2014Liu D. Wang D. Qin Z. Zhang D. Yin L. Wu L. Colasanti J. Li A. Mao L. The SEPALLATA MADS-box protein SLMBP21 forms protein complexes with JOINTLESS and MACROCALYX as a transcription activator for development of the tomato flower abscission zone.Plant J. 2014; 77: 284-296Crossref PubMed Scopus (75) Google Scholar, Mao et al., 2000Mao L. Begum D. Chuang H.W. Budiman M.A. Szymkowiak E.J. Irish E.E. Wing R.A. JOINTLESS is a MADS-box gene controlling tomato flower abscission zone development.Nature. 2000; 406: 910-913Crossref PubMed Scopus (255) Google Scholar, Nakano et al., 2012Nakano T. Kimbara J. Fujisawa M. Kitagawa M. Ihashi N. Maeda H. Kasumi T. Ito Y. MACROCALYX and JOINTLESS interact in the transcriptional regulation of tomato fruit abscission zone development.Plant Physiol. 2012; 158: 439-450Crossref PubMed Scopus (108) Google Scholar, Shalit et al., 2009Shalit A. Rozman A. Goldshmidt A. Alvarez J.P. Bowman J.L. Eshed Y. Lifschitz E. The flowering hormone florigen functions as a general systemic regulator of growth and termination.Proc. Natl. Acad. Sci. USA. 2009; 106: 8392-8397Crossref PubMed Scopus (255) Google Scholar). The jointless1 mutant (j1) was mapped to chromosome 11 and was found to be mutated in a homolog of the Arabidopsis MADS-box flowering regulator SHORT VEGETATIVE PHASE (SVP) (Mao et al., 2000Mao L. Begum D. Chuang H.W. Budiman M.A. Szymkowiak E.J. Irish E.E. Wing R.A. JOINTLESS is a MADS-box gene controlling tomato flower abscission zone development.Nature. 2000; 406: 910-913Crossref PubMed Scopus (255) Google Scholar). We therefore searched the ∼6 Mbp j2 mapping interval for MADS-box genes, and among the 164 genes in this region, we found only one candidate, Solyc12g038510, a homolog of the Arabidopsis floral organ identity MADS-box gene SEPALLATA4 (SEP4) (Figure 2B) (Ditta et al., 2004Ditta G. Pinyopich A. Robles P. Pelaz S. Yanofsky M.F. The SEP4 gene of Arabidopsis thaliana functions in floral organ and meristem identity.Curr. Biol. 2004; 14: 1935-1940Abstract Full Text Full Text PDF PubMed Scopus (587) Google Scholar). Previous transcriptional silencing of Solyc12g038510 resulted in jointless pedicels, but it was suggested that Solyc12g038510 and J2 were different genes because the published j2 mapping interval did not coincide with Solyc12g038510, likely from unreliable centromeric marker resolution (Budiman et al., 2004Budiman M.A. Chang S.B. Lee S. Yang T.J. Zhang H.B. de Jong H. Wing R.A. Localization of jointless-2 gene in the centromeric region of tomato chromosome 12 based on high resolution genetic and physical mapping.Theor. Appl. Genet. 2004; 108: 190-196Crossref PubMed Scopus (56) Google Scholar, Liu et al., 2014Liu D. Wang D. Qin Z. Zhang D. Yin L. Wu L. Colasanti J. Li A. Mao L. The SEPALLATA MADS-box protein SLMBP21 forms protein complexes with JOINTLESS and MACROCALYX as a transcription activator for development of the tomato flower abscission zone.Plant J. 2014; 77: 284-296Crossref PubMed Scopus (75) Google Scholar). However, our genomic sequencing of s2 and j2 mutants exposed a Copia-like Rider-type transposable element (TE) in the first intron of Solyc12g038510 that was absent in WT (Figure 2C). Furthermore, our s2 RNA-seq revealed that most Solyc12g038510 transcripts initiated in the first intron, resulting in an early nonsense mutation (Figures 2D and S3H). To validate that Solyc12g038510 is J2, we used CRISPR/Cas9 to engineer loss-of-function mutations, and the resulting j2CR plants developed jointless unbranched inflorescences (Figures 2E and 2F). Moreover, progeny from crossing j2CR with s2-derived j2 had jointless and unbranched inflorescences (Figure 2G), and sequencing Solyc12g038510 in the original j2 S. cheesmaniae accession revealed an early stop codon (Figures S3F–S3H). Thus, the SEP4 gene Solyc12g038510 is J2, and two natural mutations arose independently (hereafter designated j2TE and j2stop) (Reynard, 1961Reynard G.B. New Source of the j2 Gene Governing Jointless Pedicel in Tomato.Science. 1961; 134: 2102Crossref PubMed Scopus (17) Google Scholar, Rick, 1956aRick C.M. Genetic and Systematic Studies on Accessions of Lycospersicon from the Galapagos Islands.Am. J. Bot. 1956; 43: 687-696Crossref Google Scholar). Both j2 and ej2 are required for s2 branching, suggesting that the underlying genes function redundantly, similar to SEP genes in Arabidopsis that control floral organ identity (Ditta et al., 2004Ditta G. Pinyopich A. Robles P. Pelaz S. Yanofsky M.F. The SEP4 gene of Arabidopsis thaliana functions in floral organ and meristem identity.Curr. Biol. 2004; 14: 1935-1940Abstract Full Text Full Text PDF PubMed Scopus (587) Google Scholar, Pelaz et al., 2000Pelaz S. Ditta G.S. Yanofsky M.F. B and C foral organ identity functions require SEPALLATA MADS-box genes.Nature. 2000; 405: 200-203Crossref PubMed Scopus (1085) Google Scholar). We searched the 66 genes in the 500 kbp ej2 mapping interval for MADS-box genes and found the tandemly arranged Solyc03g114830 and Solyc03g114840 (Figure 2H). Solyc03g114830 is a homolog of Arabidopsis FRUITFULL and knockdown of this gene causes subtle fruit-ripening defects (Bemer et al., 2012Bemer M. Karlova R. Ballester A.R. Tikunov Y.M. Bovy A.G. Wolters-Arts M. Rossetto P.d.B. Angenent G.C. de Maagd R.A. The tomato FRUITFULL homologs TDR4/FUL1 and MBP7/FUL2 regulate ethylene-independent aspects of fruit ripening.Plant Cell. 2012; 24: 4437-4451Crossref PubMed Scopus (227) Google Scholar). Our genomic sequencing of s2 mutants did not reveal any Solyc03g114830 coding or noncoding SNPs or large indels, and s2 fruits ripened normally. In contrast, Solyc03g114840 is another homolog of SEP4, and we found a 564 bp insertion in the 5th intron of s2 mutants, which was absent in WT (Figure 2I). Notably, RNA-seq reads from s2 revealed a third of Solyc03g114840 transcripts was misspliced, suggesting that the insertion caused a partial loss of function (Figure 2J). To test this and uncover the phenotypic consequences of strong loss of EJ2 function, we engineered new alleles with CRISPR/Cas9 and found ej2CR inflorescences were unbranched, but the sepals (outermost leaf-like organs of the flowers) were exceptionally large and fruits were pear-shaped (Figures 2K and 2L). To determine if the original ej2 mutation impacted flower or fruit morphology, we backcrossed ej2 into M82 and measured relative sepal length (defined by sepal/petal length ratio). Notably, whereas there was no obvious change in fruit shape or size, ej2 sepals were 50% longer than WT but shorter than ej2CR, consistent with a weak allele (Figures 2M, 2N, and S3I). Importantly, flowers of F1 progeny from crossing ej2 and ej2CR also developed long sepals. Thus, Solyc03g114840 is EJ2, and the natural ej2 mutation is a weak loss-of-function allele (hereafter designated ej2w). Finally, we verified that the other s2 accessions carried mutations in both j2 and ej2. PCR genotyping showed all but