- © 2019 American Society of Plant Biologists. All rights reserved.
Abstract
The function of the plant hormone jasmonic acid (JA) in the development of tomato (Solanum lycopersicum) flowers was analyzed with a mutant defective in JA perception (jasmonate-insensitive1-1, jai1-1). In contrast with Arabidopsis (Arabidopsis thaliana) JA-insensitive plants, which are male sterile, the tomato jai1-1 mutant is female sterile, with major defects in female development. To identify putative JA-dependent regulatory components, we performed transcriptomics on ovules from flowers at three developmental stages from wild type and jai1-1 mutants. One of the strongly downregulated genes in jai1-1 encodes the MYB transcription factor SlMYB21. Its Arabidopsis ortholog plays a crucial role in JA-regulated stamen development. SlMYB21 was shown here to exhibit transcription factor activity in yeast, to interact with SlJAZ9 in yeast and in planta, and to complement Arabidopsis myb21-5. To analyze SlMYB21 function, we generated clustered regularly interspaced short palindromic repeats(CRISPR)/CRISPR associated protein 9 (Cas9) mutants and identified a mutant by Targeting Induced Local Lesions in Genomes (TILLING). These mutants showed female sterility, corroborating a function of MYB21 in tomato ovule development. Transcriptomics analysis of wild type, jai1-1, and myb21-2 carpels revealed processes that might be controlled by SlMYB21. The data suggest positive regulation of JA biosynthesis by SlMYB21, but negative regulation of auxin and gibberellins. The results demonstrate that SlMYB21 mediates at least partially the action of JA and might control the flower-to-fruit transition.
INTRODUCTION
In seed plants, successful reproduction involves coordinated flower organ development followed by fertilization and fruit development. During flower development, organ specification, growth, and patterning are tightly controlled. The female reproductive organ, the gynoecium, goes through a complex growth and patterning process, after which it ceases growth until fertilization occurs (Johri et al., 1992). Following successful fertilization, seeds start to develop within the ovary, the lower part of the gynoecium, and the ovary resumes growth and develops into a fruit (Gillaspy et al., 1993; Larsson et al., 2014). Seeds themselves, however, originate from ovules, which in turn emerge from the placental tissue as finger-like primordia (Smyth et al., 1990; Modrusan et al., 1994). Ovule structure is largely conserved, and three different regions can be distinguished along the proximal-distal axis: (1) the nucellus, which encloses the megaspore/megagametophyte lineage; (2) the chalaza, which initiates the inner and outer integuments that grow around the nucellus; and (3) the funiculus, which attaches the ovule to the placenta and provides a physical and vascular connection to it (Schneitz et al., 1995). In tomato (Solanum lycopersicum), ovules are anatropous, unitegmic, and tenuinucellar, the latter meaning that no hypodermal cell layer is present between the meiocyte and nucellus apex (Brukhin et al., 2003; Endress, 2011). In these so-called thin ovules, often an endothelium is formed on the inside of the integument, and the endothelium appears to supply the embryo sac with nutrients (Kapil and Tiwari, 1978; Endress, 2010, 2011). After fertilization, seed development is initiated leading to the formation of embryo and endosperm (Drews and Yadegari, 2002). The maternal tissue of the ovule also undergoes drastic changes, such as a rapid phase of cell division and expansion in the integuments (Haughn and Chaudhury, 2005). In several plant species, the proximal region of the nucellus undergoes programmed cell death (PCD) and partially or totally disappears (Lu and Magnani, 2018).
Gynoecium and ovule patterning, growth, and maturation are tightly controlled to achieve successful reproduction. Several plant hormones, as well as their crosstalk, are involved in the regulation of these processes (Marsch-Martínez and de Folter, 2016). The plant hormones auxin and gibberellins (GAs) are important for gynoecium and fruit development (Larsson et al., 2014; Pattison et al., 2014; Goldental-Cohen et al., 2017; Moubayidin and Østergaard, 2017). Another group of plant hormones involved in flower development are jasmonates (JAs) including jasmonic acid (JA) and its derivatives. These lipid-derived compounds are ubiquitously occurring plant signaling compounds that act in the response to biotic and abiotic stress, and in development (Wasternack and Hause, 2013). JA synthesized from α-linolenic acid released from plastid membranes is enzymatically converted into (+)-7-iso-jasmonoyl isoleucine (JA-Ile; Wasternack and Song, 2017), which represents the most biologically active form of JAs by mediating binding of the coreceptor proteins CORONATINE INSENSITIVE1 (COI1) and JASMONATE ZIM DOMAIN (JAZ; Chini et al., 2007; Thines et al., 2007; Fonseca et al., 2009). The JA response involves the activity of transcription factors (TFs), such as basic helix-loop-helix (bHLH) TFs like MYC2, which are repressed by JAZ proteins in the resting state of a plant. Rise in JA-Ile leads to the interaction of COI1 and JAZ, thereby mediating the proteasomal degradation of JAZ and freeing TFs such as MYC2 from repression (Wasternack and Hause, 2013).
Characterization of JA-insensitive or -deficient mutants provided strong evidence for the involvement of JA in flower development. Arabidopsis (Arabidopsis thaliana) JA-insensitive plants mutated in the JA-coreceptor COI1 (coi1) or JA-deficient plants mutated in JA-biosynthetic genes (e.g., allene oxide synthase, OPDA reductase3) are male sterile (Browse, 2009a). All of these mutants do not produce seeds due to an identical male-sterile phenotype with nonelongating stamen filaments, nondehiscent anthers, and nongerminating pollen (Browse, 2009b, 2009a). Several TFs have been shown to regulate JA-mediated stamen development, such as the R2R3-MYB-TFs MYB21, MYB24, and MYB108 (Mandaokar et al., 2006). Mutants affected in these TFs mirror to some extent the phenotype of JA-insensitive plants, exhibiting reduced male fertility that is associated with nonelongated filaments, delayed anther dehiscence, reduced pollen viability, and decreased fecundity relative to wild type (Mandaokar and Browse, 2009; Song et al., 2011). Interestingly, overexpression of AtMYB21 in coi1 or of AtMYB24 in OPDA reductase3 could partially rescue male fertility, suggesting a central role for both TFs in Arabidopsis stamen and pollen development (Song et al., 2011; Huang et al., 2017). AtMYB21 and AtMYB24 are targets of JAZ repressors (Song et al., 2011; Huang et al., 2017), but interact additionally with bHLH TFs of the IIIe clade, such as MYC2, MYC3, MYC4, and MYC5, to form a bHLH-MYB transcription complex that cooperatively regulates stamen development (Qi et al., 2015).
A JA-insensitive mutant in tomato (cv Micro-Tom), called jasmonic acid-insensitive1-1 (jai1-1), exhibits a 6.2-kb deletion in the tomato ortholog of AtCOI1 (Li et al., 2004). This mutant is, however, female sterile and does not produce seeds upon pollination with wild type or jai1-1 pollen, although fruit set and fruit development appear similar to wild-type plants (Li et al., 2004).The jai1-1 plants also exhibit some defects in the male reproductive function, such as a reduction in pollen viability and germination. The fertilization capability of pollen is, however, not affected (Li et al., 2004). Comparative transcript profiling of wild-type and jai1-1 stamens showed that genes encoding enzymes involved in the biosynthesis of ethylene (ET) and ET-related TFs as well as ET-response genes are expressed earlier during stamen development of jai1-1 in comparison with that of wild type (Dobritzsch et al., 2015). This premature ET function might cause enhanced dehiscence of the jai1-1 stamen in the open flower and misregulated pollen maturation and release. An additional phenotypic feature of jai1-1 mutant flowers is a swollen gynoecium, which in combination with the senescent stamen leads to a protrusion of the stigma from the anther cone of mature flowers (Li et al., 2004).
The female sterility of jai1-1 is in agreement with data showing that JA biosynthesis might occur predominantly in ovules, where one of the JA-biosynthetic enzymes, ALLENE OXID CYCLASE, is preferentially located (Hause et al., 2000). Moreover, JA and JA-Ile accumulate mainly in the carpel of flower buds highly exceeding the levels detected in nonstressed leaves (Hause et al., 2000). This organ-specific accumulation of JA/JA-Ile may result in organ-specific regulation of gene expression. Indeed, a number of JA-induced genes are specifically expressed within ovules (Hause et al., 2000), but their regulation by JA in gametophytic organs has not yet been proven. To address this question, we used wild type and jai1-1 flower buds at different developmental stages and performed an integrated approach by immunological detection of JA/JA-Ile and comparative transcript profiling of ovules. The obtained data and further histological analyses showed that the nucellus of jai1-1 flowers undergoes a premature PCD and that SlMYB21, an ortholog of AtMYB21, might be involved in its regulation. To test these results, three Slmyb21 mutants were identified by TILLING or generated by clustered regularly interspaced short palindromic repeats(CRISPR)/CRISPR associated protein 9 (Cas9) genome editing. Comparing transcript and hormone profiles of carpels from Slmyb21-2 and jai1-1 suggested that SlMYB21 regulates JA biosynthesis positively in female organs of tomato. In addition, SlMYB21 may mediate JA function in carpel and ovule development via regulation of auxin and GAs biosynthesis and signaling.
RESULTS
Wild-type Carpels of Large Flower Buds Contain Highest JA/JA-Ile Levels
Previous work revealed that the timing of flower development in wild-type and jai1-1 plants is very similar, showing the same developmental stages ranging from small flower buds up to open flowers (Dobritzsch et al., 2015). Six stages were classified using parameters such as bud size, sepal opening, and petal color. The youngest stage (1) represented a small bud completely enclosed by sepals; the mid–bud-stage (3) was characterized by slightly opened sepals and greenish-white petals; and the oldest stage (6) represented the open flower showing bright yellow petals. Open flowers of both genotypes differ by prominent phenotypic features of jai1-1, such as the protrusion of the stigma out of the stamen cone and the senescing tip of the stamen cone (Li et al., 2004; Dobritzsch et al., 2015). Additionally, detailed inspection of the carpels revealed that in jai1-1 they appear larger than in wild type (Figure 1A) reaching up to 100% higher fresh weight and 50% higher dry weight than in wild type (Supplemental Figures 1A and 1B). The enhanced growth of jai1-1 carpels might be one reason for the protrusion of the stigma from the stamen cone.
Classification of Stages and Levels of Jasmonates in Developing Carpels of Wild Type (WT) and jai1-1.
(A) Developmental stages of opened stamen cones and dissected carpels of wild type and jai1-1 showing the six developmental stages as defined by Dobritzsch et al. (2015). In the picture down left, bar = 5 mm for all photographs.
(B) JA and JA-Ile levels in developing carpels. Carpels of the respective stage were extracted, and contents of JA and JA-Ile were determined. Mean values ±sd are shown (n ≥ 3 independent pools of carpels). Data of the same developmental stage were compared between wild type and jai1-1 by Student’s t test (* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001).
(C) and (D) Immunocytochemical detection of JA/JA-Ile in ovaries (C) and ovules (D) of wild type and jai1-1. Carpels of wild type flowers at stages 1, 3, and 6 and of jai1-1 at stage 3 were harvested, fixed with 1-ethyl-3-(3-dimethyl aminopropyl)-carbodiimide hydrochloride (EDC), and processed for immunolabeling as described by Mielke et al. (2011). The occurrence of JA is indicated by the green fluorescence. Note the strongest label in wild-type carpels at stage 3, whereas fluorescence signal is faint in carpels and ovules of jai1-1. Bars = 50 µm for all micrographs. Abbreviations: ov, ovule; pl, placenta; ow, ovary wall.
Previous data suggested the synthesis of JAs specifically in ovules due to the preferential occurrence of the biosynthetic enzyme ALLENE OXID CYCLASE in these organs (Hause et al., 2000). Therefore, levels of JA and JA-Ile were determined in wild-type and jai1-1 carpels collected at each developmental stage (Figure 1B). The levels of both JA and JA-Ile exhibited a significant maximum in wild-type carpels at stage 3, but dropped down below the detection limit at stages 4 and 5, respectively (Figure 1B). Most importantly, the levels of both compounds were almost below the detection limit in jai1-1 carpels at all developmental stages.
To get insights into the possible occurrence of JAs in ovules, we used immunohistochemistry with a specific antibody against JA/JA-Ile and our established protocol (Mielke et al., 2011), JAs were visualized in cross sections of wild-type carpels of developmental stages 1, 3, and 6 covering low, high, and nondetectable levels of JA/JA-Ile, respectively. Carpels of jai1-1 flower buds at stage 3 lacking JAs served as negative control. The strongest label indicative for the occurrence of JA/JA-Ile occurred in wild-type carpels of stage 3, whereas carpels of wild-type stage 6 and jai1-1 did not show green fluorescence above the background (Figure 1C). Furthermore, JA and JA-Ile were detectable in ovules of wild-type flower buds at stage 1 and even stronger at stage 3, but not in ovules of jai1-1 flower buds (Figure 1D), suggesting that JAs may function in wild-type ovules of stage 3.
Jasmonate-Insensitivity Alters the Transcriptome in Ovules
To analyze the activity of JAs in transcriptional responses during ovule development, transcript profiling was conducted for wild-type and jai1-1 ovules of developmental stages with the maximal difference in JA/JA-Ile contents in wild type, i.e., stages 1, 3, and 6. RNA isolated from dissected ovules at these stages was used for hybridization of Agilent Tomato Arrays. The comparison between ovules of wild type and jai1-1 in each of the three analyzed stages revealed 383 differentially expressed genes (DEG) in ovules of both genotypes (Figure 2A; Supplemental Data Set 1). Almost two third of the identified genes showed a stage-specific, differential expression, whereas one third of genes appeared to be differentially expressed in two or three developmental stages. Among the stage-specific DEG, the number of DEG correlated with JA/JA-Ile levels: The lowest number of DEG was detected in stage 1 showing low JA/JA-Ile levels in wild type and no obvious phenotypical differences between wild type and jai1-1 (Figures 1A and 1B), whereas the highest number of DEGs occurred in stage 3. The identified DEG could be assigned to 22 functional classes (Figure 2B; Supplemental Data Set 1). The highest number of DEG occurred in the class “transcription factors and regulators” (36) followed by “secondary metabolism” (28), “defense” (26), and “proteinase inhibitors” (19).
Comparative Analysis of Transcript Accumulation in Dissected Ovules of Wild Type (WT) and jai1-1.
Total RNA isolated from three developmental stages of ovules of wild type and jai1-1 was subjected to transcript profiling using the Agilent-Tomato 44K-full genome chip.
(A) Venn diagram showing the number of significantly regulated genes (P ≤ 0.01, n = 3 independent pools of dissected ovules). Numbers in parentheses are related to genes with unknown functions. Note that the highest number of differentially regulated genes was found in stage 3.
(B) Classification of differentially expressed genes according to functional classes. The bold numbers indicate how many genes in total were differentially regulated in the respective developmental stage, whereas the regular numbers show how many of them exhibited decreased (dark blue) or increased transcript levels in jai1-1 (dark brown). Light blue and light brown colors mark the overall tendency of differential transcript accumulation in each developmental stage/functional class (blue higher in wild type, brown higher in jai1-1).
(C) and (D) Relative transcript levels of genes positively (C) and negatively (D) regulated by JA. Note that transcripts of the typical JA-regulated genes shown in (C) are nearly not detectable in ovules of jai1-1. All transcript levels were determined by RT-qPCR and set in relation to SlTIP41 (rel. expression). The inset in each diagram visualizes the absolute signal intensity (abs. signal int.) obtained from microarray analysis. Mean values ±sd are shown (n = 3). Data of the same developmental stage were compared between wild type and jai1-1 by Student’s t test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001). ▬○▬ wild type,—▵—jai1-1
Most of the DEG occurring in flower stages 1, 1+3, and 1+3+6 of the Venn diagram showed lower or almost no detectable expression in jai1-1 (blue-labeled numbers in Figure 2B), whereas DEG occurring in developmental stages 3+6 and 6 exhibited higher expression levels in jai1-1 than in wild-type ovules (brown-labeled numbers in Figure 2B). Most of the genes showing a strong downregulation in jai1-1 include well-characterized JA-induced genes, such as proteinase inhibitor II and cathepsin D inhibitor, which are involved in JA-mediated defense (Wasternack and Hause, 2013). Other strictly JA-regulated genes are the well-described genes encoding JAZ proteins (Thines et al., 2007), such as JAZ2 and JAZ6, and TFs, such as SlMYB21 and SlMYB14. Transcript levels of most of these genes exhibited a maximum at stage 3 of wild-type ovules correlating with the maximum JA/JA-Ile levels and are nearly undetectable in jai1-1 ovules as validated by quantitative RT-PCR with RNA from ovules of an independent experiment (Figure 2C). This indicates a positive regulatory role of JAs during ovule development including an improved defense status as previously suggested (Hause et al., 2000).
Several DEG occurring specifically in stages 3 and 6 exhibited increased accumulation in ovules of jai1-1 in comparison with wild type (Figures 2B and 2D). This suggests a negative regulatory role of JAs in late stages of ovule development. Mainly in stage 6 (open flower), several genes belonging to proteases (e.g., METACASPASE9 and SUBTILASES), nucleases (e.g., ENDONUCLEASE), and cell wall modifying enzymes (e.g., ENDOGLUCANASE and β-GLUCOSIDASE) are strongly upregulated in jai1-1 ovules (Figure 2D). It is possible that the enzymes encoded by these genes might mediate morphological changes in jai1-1 ovules, since proteases and nucleases have been implicated in nucellus degeneration by PCD (Lu and Magnani, 2018). Therefore, ovule morphology was analyzed in carpel cross sections (Figure 3). Ovules of early developmental stages (stage 1–3) were almost undistinguishable between wild type and jai1-1 (Supplemental Figure 2). At stages 4 and 5, however, the innermost cell layer of the nucellus of jai1-1 ovules showed a thickening of the cell walls, which was not visible in wild-type ovules (Figure 3A; Supplemental Figure 2). Indeed, jai1-1 ovules at stage 5 exhibited increased amounts of callose around the inner cells of the nucellus as visualized by immunostaining with an antibody against callose (Figure 3B), whereas in wild-type ovules callose was visible at plasmodesmata only (see inset in Figure 3B, wild type). Another obvious morphological difference was the occurrence of vacuoles in cells of the innermost cell layer of jai1-1 nucellus at stage 6 (Figure 3A). Both, increased callose deposition and cell vacuolation, might be features of PCD. Testing this by a terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay showed much higher TUNEL-positive cells in jai1-1 ovules, particularly in the innermost cells of the nucellus (Figure 3C). The abnormal morphology and the PCD occurring in these cells support the hypothesis drawn from transcript data that a premature elimination of the nucellus might occur in jai1-1 ovules. Nucellus elimination is typically dependent on fertilization (Xu et al., 2016). The lack of JA function in jai1-1 seems to uncouple fertilization and nucellus degeneration.
Ovules of jai1-1 Show Altered Morphology and Exhibit Enhanced Callose Accumulation and PCD in the Nucellus.
(A) Semi-thin cross sections of ovules stained with toluidine blue. The developmental stages are indicated by numbers. Note the thickened cell walls in stage 5 (arrow head) and vacuolated cells of the inner layer of nucellus in stage 6 (arrow) of jai1-1.
(B) Immunolabeling of callose in ovules of stage 5. Callose visualized by green fluorescence is detectable in small spots between all cells of wild type ovules pointing to plasmodesmata (see inset), but surrounds additionally the innermost cell layer of the nucellus of jai1-1 ovules.
(C) Cross sections of ovules at stage 6 were analyzed by the TUNEL assay. The wild type tissues only showed few TUNEL-positive signals. In contrast, a higher number of TUNEL-positive signals appeared in the innermost cell layer of the nucellus of the jai1-1 mutant. Positive and negative controls showed labeling of all and no nuclei, respectively (Supplemental Figure 2B). Bars = 50 µm in all micrographs, except the inset, where it = 10 µm.
Jasmonate-Regulated SlMYB21 Encodes a Flower-Specific Transcription Factor
Several genes encoding putative MYB-TFs were highly expressed in wild-type ovules, but were nearly undetectable in ovules of jai1-1 (Figure 2) including SlMYB21 (Solyc02g067760). The putative SlMYB21 protein shows 64% sequence identity with Arabidopsis MYB21 (AtMYB21), which is known as an important regulator of JA-mediated stamen development (Mandaokar and Browse, 2009; Song et al., 2011). SlMYB21 seems to be predominantly expressed in flower buds and open flowers as revealed by in silico analyses using tomato eFP browser (http://bar.utoronto.ca/) and the TomExpress database (Zouine et al., 2017; Supplemental Figures 3A and 3B). In dissected wild-type carpels, SlMYB21 transcripts accumulated with a transient maximum in flowers at stages 4 and 5 (Supplemental Figure 3C). To test, whether the protein encoded by Solyc02g067760 acts indeed as TF, we investigated its localization and transcriptional activity. Transient expression of a green fluorescent protein (GFP):SlMYB21 fusion in Nicotiana benthamiana leaves showed that the protein is located predominantly in the nucleus (Figure 4A). Moreover, SlMYB21 displayed transcription-inducing activity in yeast as its fusion to the GAL4 binding domain (BD) alone resulted in transcriptional activity (Figure 4B). The previously detected interaction of AtMYB21 with several JAZ proteins (Song et al., 2011; Huang et al., 2017) prompted us to investigate putative interaction of SlMYB21 with the tomato JAZ proteins. All 12 known JAZ proteins from tomato were tested for their interaction with SlMYB21 in a yeast-two-hybrid (Y2H) assay (Figure 4C). Surprisingly, SlMYB21 interacted only with SlJAZ9. Analyzing all combinations by bimolecular fluorescence complementation (BiFC), positive signals were found for JAZ9 and JAZ8 (Figure 4D). According to eFP browser, SlJAZ9 is mainly expressed in unopened flower buds, whereas SlJAZ8 is exclusively expressed in roots (not shown). Consequently, an interaction between SlMYB21 and JAZ8 may not be relevant in flowers. All other JAZ proteins did not show an interaction with SlMYB21 in the BiFC assays (not shown). The interaction of SlMYB21 with JAZ9 was further validated by the split transcription activator-like effector(TALE) system (Schreiber et al., 2019). The fusion of SlMYB21 to the activation domain (AD) and of JAZ9 to the BD of TALE did not result in induction of the reporter construct (Supplemental Figure 4). However, fusion of TALEBD and TALEAD to JAZ9 and SlMYB21, respectively, resulted in a TALE reconstitution leading to significantly increased β‑glucuronidase (GUS) activity (Figure 4E). Similar to yeast, SlMYB21 fused to TALEBD displayed transcriptional activity on its own, which was increased by coexpression of JAZ9-TALEAD. Removing the AD of SlMYB21 almost abolished the activity of the fusion construct, but coexpression of JAZ9-TALEAD increased the activity ∼4-fold. Thus, SlMYB21 exhibits transcriptional activity, and its interaction with one of the JAZ proteins of tomato suggests that this TF, like AtMYB21 (Song et al., 2011), might act at a hierarchical position in the JA signaling pathway (i.e., downstream of the COI1-JAZ coreceptor complex).
SlMYB21 Encodes an Active Transcription Factor That Interacts With SlJAZ9 and Complements the Arabidopsis Mutant myb21-5.
(A) SlMYB21 is located in the nucleus as shown by transient expression of a 35S:gMYB21:GFP construct in N. benthamiana leaves. The green fluorescence of the fusion protein is detectable in nuclei of the epidermal cells. Plastids are visible by their red autofluorescence. Protein gel blot performed with total protein extract from transformed leaves (empty vector and 35S:SlMYB21:GFP) and an anti-GFP antibody shows the correct size of the fusion protein as indicated by the molecular weight marker (M). Bar in the micrograph = 20 µm.
(B) Yeast assay used to detect the transcriptional activity of SlMYB21. Yeast cells transformed with SlMYB21 fused to the GAL4 BD and the empty vector encoding the GAL4 AD were grown on synthetic defined (SD) medium without tryptophan and leucine (-TL) or SD medium without tryptophane, leucine and histidine (-TLH). Transformation with the empty vector (e.v.) only served as negative control.
(C) Y2H assays testing the interaction of SlMYB21 with SlJAZ proteins. Full-length CDS of all 12 SlJAZ proteins (JAZ1–JAZ11, JAZ13) were fused with the GAL4 BD and CDS of SlMYB21 was fused with the GAL4 AD. Transformed yeast cells were grown on SD/-TLH to determine protein–protein interactions. Positive yeast transformations are visualized by growth on SD/-TL. Omitting SlMYB21 by using the empty vector (e.v.) served as control and did not show yeast growth on SD/-TLH.
(D) BiFC assays performed to test the interaction of SlMYB21 with SlJAZ proteins. CDS of all 12 SlJAZ proteins (JAZ1–JAZ11, JAZ13) and of SlMYB21 were introduced in the 2in1 vectors (Grefen and Blatt, 2012), transformed into N. benthamiana protoplasts, and analyzed by confocal laser scanning microscopy. Interaction of SlMYB21 with SlJAZ9 and SlJAZ8 was detectable, whereas all other JAZ proteins did not show interaction as exemplified shown for SlJAZ1. Interaction of AtMYC2 with AtJAZ1 served as positive control. Bars = 5 µm.
(E) splitTALE assay showing the interaction of SlMYB21 with SlJAZ9 in planta. CDS of SlMYB21 with or without AD (C-terminal deletion of 25 aa) and SlJAZ9 were fused either to the TALE BD (N-terminal) or to AD (C-terminal), respectively, and expressed together with the reporter construct 4xSTAP1:GUS in leaves of N. benthamiana. Transcriptional activity of full-length SlMYB21 fused to TALE BD is visible by a high GUS activity, which was enhanced by coexpression with SlJAZ9 fused to TALE-AD. Removal of AD of SlMYB21 resulted in lower basal activity, but was significantly increased by coexpression of SlJAZ9 fused to TALE-AD. Expression of complete TALE together with the reporter as well as 35S:GUS served as positive controls. Mean values ±se are shown (n = 3 different plants). Data were compared between expression of SlMYB21 alone and together with SlJAZ9 by Student’s t test (**P ≤ 0.01).
(F) Comparison of Arabidopsis flowers from different genotypes as indicated. As shown by one example out of eight and five independent lines, expression of pAtMYB21:AtMYB21 results in partial rescue of the phenotype leading to seed set, which is also visible by expression of pAtMYB21:SlMYB21. Bars = 1 mm for flowers and 1 cm for shoots. WT, wild type.
The high similarity of SlMYB21 and AtMYB21 prompted us to test whether SlMYB21 would be able to rescue the Arabidopsis mutant myb21-5. Flowers of Atmyb21-5 are characterized by short petals, short stamen filaments, and delayed flower opening and anther dehiscence (Reeves et al., 2012). Complementation of Atmyb21-5 by expressing AtMYB21 under the control of its endogenous promoter (pAtMYB21) resulted, however, only in a partial rescue that was characterized by an intermediate phenotype with increased filament length, dehiscent anthers, and opened flowers (Figure 4F). Out of eight independent lines, three developed seed-containing siliques. A similar partial and variable rescue of myb21-5 using a 35S:GFP-AtMYB21 construct has been previously observed (Reeves et al., 2012). Expression of tomato MYB21 under control of pAtMYB21 rescued Atmyb21-5 to a similar degree than Arabidopsis MYB21 (Figure 4F). Out of six independent lines, two were able to develop seed-containing siliques (Figure 4F). These data support the conclusion that SlMYB21 is a functional equivalent of AtMYB21.
SlMYB21 Promotes Fertility, Ovule Development, and Jasmonate Biosynthesis
For functional analysis of SlMYB21, a Slmyb21 mutant was selected by a TILLING approach (Okabe et al., 2011). Screening of the TILLING population resulted in four different mutants exhibiting a single base exchange within the two first exons of the SlMYB21 encoding gene (TOMJPE7564, TOMJPE7979, TOMJPE8245, and TOMJPW0469). Three of them were not altered in any property of MYB21, but TOMJPW0469 (Slmyb21-1) showed the exchange of a highly conserved Leu to a Lys residue within the first MYB domain, possibly affecting its transcription-inducing activity and/or DNA binding capability (Supplemental Figure 5A). Homozygous Slmyb21-1 mutant plants showed abnormal flower development with petals that did not fully open and hardly developed fruits (Supplemental Figure 5B). In addition, two independent mutants were obtained using the CRISPR/Cas9 gene editing system (Brooks et al., 2014) targeting the second exon of SlMYB21. Twenty-nine independent transformants were obtained expressing Cas9 and showing mutations in SlMYB21, 25 of which did not develop seeds. Among the eleven Cas9-positive and seed-producing lines, two different heterozygous offspring led to homozygous plants in the T2 generation (Figure 5). One of them showed a deletion of one T (Slmyb21-2) and the other an insertion of one T (Slmyb21-3), both of which are predicted to cause a premature stop codon and formation of a truncated, nonfunctional SlMYB21 (Figure 5A). The flower phenotype of both mutants was very similar to that of Slmyb21-1 and showed the same defects in flower morphology (Figure 5B). In contrast with Slmyb21-1, Slmyb21-2 and Slmyb21-3 plants developed fruits, which were all seedless. Slmyb21-2 plants were, however, capable to develop few seeds after hand pollination performed before flower opening. Ovules of all three Slmyb21 mutants showed a phenotype like jai1-1 ovules and exhibited accumulation of callose and vacuolated cells in the inner cell layer of the nucellus, although not so severe as jai1-1 ovules (Figure 5C; Supplemental Figures 5C and 5D). In addition, flower buds of myb21-2 at stage 3 were significantly longer than wild-type buds and myb21-2 carpels exhibited higher fresh weight than wild-type carpels similarly to carpels of jai1-1 (Supplemental Figures 1 and 6).
Slmyb21-2 and Slmyb21-3 Mutant Flowers and Fruits Show a Phenotype Similar to jai1-1, but Differ in the JA Content of Carpels.
(A) Sequence of genomic DNA encoding MYB21 showing the wild type (WT) sequence and two mutant sequences, which have either a deletion of one T (Slmyb21-2) or an insertion of one T (Slmyb21-3). The resulting peptide sequences are given below the DNA sequencing graph and show that both mutations result in a premature stop.
(B) Flowers of homozygous mutant plants exhibited defective opening and a protrusion of the stigma from the stamen cone of mature flowers. Fruits did not develop seeds.
(C) Cross-sections of mutant ovules show enhanced callose accumulation around and vacuolated cells in the inner layer of the nucellus as visualized by immuno stain at stage 5 and toluidine blue staining at stage 6, respectively. Bars = 50 µm in all micrographs.
(D) JA and JA-Ile levels in carpels of Slmyb21 were diminished in comparison with wild type, but were higher than in jai1-1 carpels. Carpels of stage 3 of wild type, jai1-1, and Slmyb21-1 (left) as well as of wild type, Slmyb21-2, and Slmyb21-3 (right) were extracted and contents of JA and JA-Ile were determined. Mean values ±se are shown (n = 5 independent pools of carpels). Different letters within data for compound indicate significant differences according to one-way-ANOVA with Tukey's HSD test (P < 0.05).
AtMYB21 was described to decrease JA levels through a negative feedback loop on the expression of JA biosynthesis genes (Reeves et al., 2012; Huang et al., 2017). We therefore determined JA and JA-Ile levels in carpels of the tomato myb21 mutants in comparison with wild type and jai1-1 (Figure 5D). For these determinations, carpels of flower buds at stage 3 were used, since they showed the highest levels of JA and JA-Ile in wild type (Figure 1B). Surprisingly, carpels of all three slmyb21 mutants exhibited diminished JA and JA-Ile levels in comparison with those of wild type (Figure 5D). This points to a putative positive regulatory function of SlMYB21 on JA biosynthesis, which is opposite to its function in Arabidopsis.
SlMYB21 Contributes to Jasmonate-Regulated Gene Expression in Carpels
The similar phenotype of Slmyb21 mutants and jai1-1 suggests that the JA-induced TF SlMYB21 might mediate at least some of the JA-regulated processes, which are important for female fertility. To explore the putative regulatory hierarchy, a transcript profiling by RNA sequencing (RNA-seq) was performed using dissected carpels at stage 3 from wild type, jai1-1, and Slmyb21-2 (Figure 6). When comparing transcript levels of mutants with those from with wild type, in total 2016 DEG were found in both mutants (Supplemental Data Set 2). Among them, 1220 and 316 were differentially expressed exclusively in jai1-1 and Slmyb21-1, respectively. Here, strictly JA-induced genes encoding defense proteins, such as PI-I, and encoding JA-signaling components, such as JAZ proteins and COI1, did not appear to be regulated by SlMYB21, since they were not found among the genes differentially regulated in Slmyb21-2 (Figure 6B). In contrast, genes encoding stress-related proteins, such as thaumatin and LEA proteins, as well as several TFs seemed to be regulated independently of JA by SlMYB21, since they appear exclusively among the DEGs of Slmyb21-2. Most importantly, however, there were 237 and 226 DEG commonly up- and downregulated, respectively, in both mutants. This supports the hypothesis that at least a part of JA-regulated genes might be regulated by SlMYB21 (Figure 6A).
Comparative Analysis of Transcript Accumulation in Carpels at Flower Stage 3 of Wild Type (WT), jai1-1, and slmyb21-2.
Total RNA isolated from carpels at stage 3 of wild type, jai1-1, and Slmyb21-2 was subjected to transcript profiling using RNA-seq.
(A) Venn diagram showing the number of significantly regulated genes (P ≤ 0.01, n = 3 from different plants with two carpels pooled each) in both mutants in comparison with wild type. Note that the highest number of differentially regulated genes was found in jai1-1, and in both mutants 226 and 237 genes were commonly down- and upregulated, respectively.
(B) to (D) Classification of differentially expressed genes according to Gene Ontology: B: regulation of jasmonic acid–mediated signaling pathway, C: auxin-activated signaling pathway, D: gibberellic acid–mediated signaling pathway. Selected nodes (biological processes, BP) from the Enrichment Map where the left half corresponds to the comparison of jai1-1 to wild type and the right half of Slmyb21-2 to wild type are shown. Note that light orange symbolizes a high q-value and therefore almost no impact on the enrichment (see Supplemental Figure 7). The corresponding heat maps are based on the row-scaled expression values (log2 FPKM+1) and list the set of genes, which contributed most to the enrichment of the selected biological process.
(E) Relative transcript levels of genes commonly upregulated in jai1-1 and Slmyb21-2.
(F) Relative transcript levels of genes commonly downregulated in jai1-1 and Slmyb21-2. All transcript levels in (C) and (D) were determined by RT-qPCR and set in relation to SlTIP41. The color code visualizes the ΔCt values from yellow representing high values (= almost no transcripts detectable) up to red representing low values (= high expression). The calculated, relative transcript levels are given in Supplemental Tables 3 and 4.
Gene Ontology analyses revealed that the DEGs occurring commonly in both mutants belong to various cellular processes (Supplemental Figure 7). Among them, the groups “auxin-activated signaling pathway” and “gibberellic acid mediated signaling pathway” showed several genes that were nearly not expressed in wild-type carpels, but significantly upregulated in carpels of both mutants (Figures 6C and 6D). The group of GA-related genes includes with ent-copalyl diphosphate synthase (Solyc06g084240), GA 20-oxidase-1, GA 20-oxidase-3, GA 3-β-hydroxylase, and GA 2-oxidase encoding genes, mainly GA-biosynthesis–related genes. Validation of RNA-seq-data by quantitative RT-PCR using RNA from carpels of an independent experiment showed that a selection of these genes indeed exhibits higher transcript levels in carpels of both mutants than in carpels of wild-type plants (Figure 6E; Supplemental Table 1). In addition, TFs, such as MYB12, a bHLH TF and an AP2-like TF, as well as pectin-related and fertilization-related genes, such as expansin45, two pectin esterases (Solyc03g083770 and Solyc11g019910) and one pectin acetyl esterase (Solyc08g014380), were among the commonly upregulated genes. In contrast, several TF-encoding genes, like the homolog of Arabidopsis BIGPETALp bHLH-TF (Solyc03g113560; Varaud et al., 2011) and S. lycopersicum fruit SANT/MYB-like1 (SlFSM1; Machemer et al., 2011), some auxin-related genes, and genes encoding proteins involved in cell-wall modification/cell organization, like two cellulose synthases (Solyc08g082640 and Solyc03g005450), were positively regulated by JAs and SlMYB21, since they exhibit high transcript levels in wild type and low transcript levels in both mutants (Figure 6F; Supplemental Table 2). Among them, SlMYB21 was drastically downregulated in carpels of both mutants. Moreover, the SlMYB21 homolog SlMYB24 is similarly downregulated suggesting the occurrence of a SlMYB21/SlMYB24 pair as described for Arabidopsis stamen development (Mandaokar et al., 2006).
Jasmonates and SlMYB21 Regulate GA Metabolism in Carpels
GAs play an important role in flower development, mainly in the developmental switch from flower to fruit after fertilization, since GA levels increase in the ovary upon pollination and exogenous application of GA to flowers induces parthenocarpy (Gorguet et al., 2005). This and the higher expression levels of GA biosynthesis genes in the mutants prompted us to determine levels of various gibberellin compounds in carpels of flower bud stage 3 from all three genotypes. We detected 15 different GAs including biosynthetic precursor, bioactive compounds, and inactive metabolites (Supplemental Table 3). A simplified biosynthetic pathway leading to the bioactive compounds GA1, GA3, GA4, and GA7 (Hedden and Phillips, 2000) from the precursors GA12 and GA53 is shown in Figure 7 along with their levels in carpels of wild type, jai1-1, and Slmyb21-2. The precursors GA12 and GA53 are oxidized in three steps in parallel pathways into GA9 and GA20 by the GA 20-oxidases. The formation of the bioactive compounds is then catalyzed by GA 3-β-hydroxylases, whereas the activity of GA 2-oxidases leads to formation of inactive compounds (GA51, GA34, GA29, GA8; Hedden and Thomas, 2012). In tomato, the major GA-biosynthesis pathway is represented by the early 13-hydroxylation pathway synthesizing GA1 as the bioactive form (García-Hurtado et al., 2012). The measurements revealed, however, enhanced levels of GA4 in jai1-1 in comparison with wild type. In addition, levels of inactive precursors GA53, GA44, and GA19 as well as of inactive catabolites GA34, GA29, and GA8 was higher in jai1-1 than in wild type, whereas the level of inactive intermediate GA20 was lower. The higher levels of GA53, GA44, and GA29 as well as the lower amount of GA20 might indicate decreased conversion of GA19 to GA20 and/or increased conversion of GA20 to GA29. The latter correlates with the enhanced expression of a GA 2-oxidase (Figure 6E), which might also contribute to the enhanced levels of GA34 in the parallel non–13-hydroxylation pathway (left side of the Figure 7).
GA Levels in Carpels From Flower Buds at Stage 3 From Wild Type (WT), jai1-1, and myb21-2.
Simplified GA biosynthesis pathway and levels of 14 GA isoforms determined in carpels from flower buds at stage 3. Data are means ±se (n = 5 independent pools of carpels). Different letters within data for each compound indicate significant differences according to one-way ANOVA with Tukey's Honestly Significant Difference test (P < 0.05).
Although showing similar gene expression data, carpels of Slmyb21-2 exhibited similar GA levels as carpels of the wild type, except increased levels of GA53 and GA19. Taken together, mainly GA intermediates representing precursors of bioactive GAs are increased in carpels of Slmyb21-2 at stage 3 in comparison with carpels of wild type flower buds. This makes it rather unlikely that enhanced levels of bioactive GAs might be the reason for the enhanced biomass of Slmyb21-2 carpels, whereas increased levels of GA4 in jai1-1 carpels might be causal for enhanced growth.
DISCUSSION
The survival of most plant species depends on flower development, successful fertilization, fruit, and seed set. All these processes require timely coordination between different organs by precise regulatory mechanisms that are, at least in part, mediated by phytohormones. For example, JAs are important for maintaining fertility, since the JA-insensitive mutants coi1 (Arabidopsis) and jai1 (tomato) are both sterile. However, like the great diversity in flower morphology and fruit variants between different species, the regulatory mechanisms seem to be different regarding tissue-specific functions: the tomato jai1 mutant is female sterile, whereas the Arabidopsis coi1 mutant is male sterile. Here, we addressed the question of species-specific action of JA in flower development by dissecting its function in tomato ovule development.
JA/JA-Ile levels of carpels and ovules determined via hormone measurements and immunological detection, respectively, peaked in the biggest but still closed flower bud in wild type (stage 3). This contrasts with jai1-1, where JA and JA-Ile were not detectable in carpels and ovules (Figure 1). Stage 3 is the developmental stage shortly before flowers fulfill their reproductive function. Consequently, coordination of anther dehiscence, ovule maturation, and petal opening is required. The lack of JA perception in jai1-1 ovules resulted in an abnormal development at later stages as visible by callose accumulation around cells of the inner layer of the nucellus accompanied by vacuolization of these cells (Figure 3). Both features are indicative of PCD, which was evidenced by a positive TUNEL assay in jai1-1 ovules. This premature PCD leading to an early nucellus breakdown and putative disruptions in endosperm development (Xu et al., 2016) might be the reason for the impaired embryo development of jai1-1 (Goetz et al., 2012). Indeed, comparative transcript analysis between wild-type and jai1-1 ovules revealed that in comparison with wild type, several genes involved in peptide cleavage and polysaccharide-related cell wall modifications were significantly upregulated in jai1-1 ovules dissected from open flower (Figures 2B and 2D). This indicates a negative regulatory role of JA in processes normally involved in cell death, which is induced after fertilization and has to be prevented before anthesis.
Regarding identification of putative mediators of the JA function, the transcriptomics data pointed toward SlMYB21, which was highly expressed in ovules from wild-type carpels at stage 3, but not at all in jai1-1 ovules. Its ortholog AtMYB21 is involved in JA-regulated filament and petal elongation in Arabidopsis (Mandaokar et al., 2006; Song et al., 2011; Reeves et al., 2012). Therefore, SlMYB21 was a promising candidate to be a putative mediator of JA function in tomato ovule development. Similarly to AtMYB21, expression of SlMYB21 is strictly JA dependent and specific to flowers, and the protein interacts with JAZ9 (Figure 4). The control of MYB21 protein activity by JA via a JAZ protein might lead to an even stronger loss-of-function in the JA-insensitive mutant due to the missing JA-induced degradation of JAZ proteins. To confirm that SlMYB21 is functionally equivalent to AtMYB21, SlMYB21 was expressed under the control of the AtMYB21 promoter in the Arabidopsis mutant myb21-5 (Reeves et al., 2012). This expression led to a complementation with respect to petal length and flower opening, but filament elongation showed varying results ranging from filaments that were slightly longer than those of myb21-5 (above sepal length) until complete rescue in two lines. Interestingly, such a partial, varying rescue is also achieved by complementation with the genomic sequence of AtMYB21 itself (Figure 4F) suggesting a complex regulation of MYB21 expression. Moreover, these results point to a different regulation of MYB21 expression in petals and stamen of Arabidopsis. Nevertheless, the same partial rescue visible with MYB21 from Arabidopsis and tomato supports the homology of the two MYB21 proteins.
The functional characterization of SlMYB21 and elucidation of its role in mediating JA function in tomato flower development were performed by identification of a TILLING mutant and generation of two independent mutants by CRISPR/Cas9 genome editing. All three mutants revealed a SlMYB21 function in flower buds and open flowers starting at the developmental stage 3. The mutant buds were bigger than those from wild type (Supplemental Figure 6), and the petals failed to open properly, even more severely than in jai1-1, indicating a JA-independent function of MYB21 in petal opening. Also, contrasting with the phenotype of jai1-1, there was no early senescence of stamen and hardly any impairment of pollen viability in Slmyb21-2 (data not shown; Li et al., 2004; Dobritzsch et al., 2015), supporting a MYB21-independent role of JA in the process of stamen and pollen development in tomato. This is different to tomato plants expressing an artificial repressor, AtMYB24-SUPERMAN REPRESSION DOMAIN X, which was expected to impede the function of SlMYB21 and did result in flowers with aborted flower opening, but additionally exhibited abnormal pollen grains (Niwa et al., 2018). Moreover, we found that SlMYB21 acts as a positive regulator of JA biosynthesis as shown by the diminished JA/JA-Ile levels in all three Slmyb21 mutants. This contrasts with AtMYB21, which acts as a negative regulator of JA biosynthesis, because of increased JA levels in flowers of Atmyb21 myb24 double mutants, which in turn result in enhanced transcript levels of JAZ genes, LIPOXYGENASE2, LIPOXYGENASE4, ALLENE OXIDE SYNTHASE, and OPDA REDUCTASE3 (Reeves et al., 2012; Huang et al., 2017).
The development of carpels and ovules in the Slmyb21 mutants was very similar to jai1-1 as indicated by the size of carpels and the morphology of ovules, which accumulated callose and showed vacuolated cells in the innermost cell layer of the nucellus (Figure 5; Supplemental Figure 5). It is tempting to speculate that JA acts through SlMYB21 to prevent nucellus senescence processes before anthesis. Surprisingly, however, Slmyb21-2 plants were able to develop a few seeds after hand pollination performed before flower opening, suggesting that the obvious defects in ovule development of Slmyb21 mutants are not as severe as in jai1-1. Thus, we hypothesized that defects in MYB21 protein function led to an uncoupling of fertilization and fruit development, since under normal conditions Slmyb21-2 and Slmyb21-3 plants produced fruits similarly to wild type but did not develop seeds.
To gain insight into how the development of tomato carpels is regulated by JA and MYB21, a comparative transcript profiling was performed using wild type, jai1-1, and myb21-2 carpels from flower buds at stage 3. The commonly deregulated genes in both mutants contained gene groups related to GA-biosynthesis and auxin signaling (Figures 6C to 6F). Both hormones are known to be essential signaling molecules mediating the switch from flower to fruit development, since exogenous application of both hormones initiates fruit set without fertilization (Serrani et al., 2007). On the one hand, a strong interference with the auxin homeostasis regulation became obvious–several AUX/IAA proteins and auxin response factors (ARFs) encoding genes were upregulated in carpels of both mutants, among them, ARF5 and ARF7 known to act as regulators for fruit set (de Jong et al., 2009; Liu et al., 2018) and ARF9 known to be induced after fertilization or after auxin treatment (de Jong et al., 2015). Their upregulation might contribute to the formation of parthenocarpic fruits, which were also found upon downregulation of the ARF7 repressor IAA9 (Wang et al., 2009). Therefore, the formation of parthenocarpic fruits might at least in part be due to the loss of fine-tuned auxin signaling.
GAs represent another important signal in flower and fruit development. GA levels increase in the ovary upon pollination (Gorguet et al., 2005; Serrani et al., 2007) and a constitutive GA signaling results in parthenocarpic fruits (Carrera et al., 2012). Our data point to a negative feedback of JA and MYB21 on expression of GA-biosynthetic genes in carpels of closed flower buds (Figures 6D and 6E) and indicate a GA-mediated signaling in mutant buds that is similar to wild-type flowers after fertilization. The levels of bioactive GAs were, however, not elevated in carpels of jai1-1 and myb21-2 except that a higher content of GA4 was observed in jai1-1 (Figure 7). In tomato, the early 13-hydroxylation pathway leading to the bioactive form GA1 is the major GA-biosynthetic pathway. It was reported, however, that overexpression of GA 20-oxidases shift the metabolism to the parallel non–13-hydroxylation pathway, resulting in biosynthesis of GA4 (García-Hurtado et al., 2012). The increased transcript levels of GA 20-oxidases in carpels of jai1-1 might therefore cause the accumulation of GA4.
GA and auxin commonly regulate fruit initiation by the interplay of the GA-signaling repressor SlDELLA and auxin-signaling components, such as SlIAA9 and SlARF7, which together repress parthenocarpy by coregulation of genes involved in fruit growth and, additionally, GA and auxin metabolism (Hu et al., 2018). Therefore, fine-tuning of auxin and GA signaling seems to be essential for controlling the signal for switching from flower into fruit development. The deregulation of GA and auxin signaling in carpels of jai1-1 and myb21-2 flower buds at stage 3 points toward a negative regulatory role of JA and MYB21 on GA and auxin action, thereby preventing fruit initiation before fertilization (Figure 8). This was clearly visible in both mutants by the early ovary growth (Supplemental Figures 1 and 6) as well as by the up- and downregulation of genes that are involved in cell separation and in cell shape maintenance, respectively (Figures 6E and 6F).
Jasmonates and SlMYB21 Regulate Ovule Development in Tomato.
During flower development in tomato, levels of JA and JA-Ile increase transiently resulting in upregulation of JA-regulated genes including this encoding SlMYB21, which in turn regulates specific genes. In addition, action of SlMYB21 positively feed back into JA/JA-Ile biosynthesis. Both JA- and MYB21-regulated gene expression are necessary to ensure proper ovule development and contribute to coordination of floral organ development. In addition, JA/JA-Ile and SlMYB21 repress biosynthesis and/or function of auxin and GA, thereby preventing a premature switch into fruit development.
In summary, MYB21 mediates JA function mainly during carpel and ovule development in tomato, but is additionally involved in the coordination of flower opening (Figure 8). In doing so, there seems to be conserved functions of MYB21 during petal development and flower opening in Arabidopsis and tomato. Regarding the development of reproductive organs, however, tomato MYB21 might function as a hub in the interplay of JA, auxin and GA in the carpel by coordination of the timely death of the nucellus after fertilization and the initiation of fruit formation, both being essential for successful seed set and development.
METHODS
Plant Material, Growth Conditions, Harvest of Carpels, and Dissection of Ovules
Tomato (Solanum lycopersicum) plants cv MicroTom wild type, jai1-1 (Li et al., 2004), and mutants generated during this work were grown in a controlled growth chamber with 16 h light (200 µmol photons*m−2*s−1; Metal Halide Lamp MT250DL, IWASAKI Electric Co.) at 26°C and 8 h dark at 20°C, both at 70% humidity. Homozygous jai1-1 mutants were selected by PCR according to (Li et al., 2004). Phenotypic markers such as missing anthocyanin production in first leaves and protrusion of stigma were additionally used.
Harvest of carpels was performed using 5- to 6-week-old plants showing the first open flowers. Carpels of defined developmental stages were harvested in a very strict time window of 30 min starting at 7 h after the onset of light period, collected on dry ice, transferred to liquid nitrogen, and stored at −80°C. Carpels of primary inflorescences were used for dissecting ovules. Hand-cut sections of carpels were transferred immediately into ice-cold ethanol and vacuum infiltrated. Ovules were dissected using a stereomicroscope and stored in ethanol at −20°C. After freeze drying at −20°C for 4 h, material was directly used for RNA isolation. Harvesting time for carpels subjected to phytohormone quantification was expanded to 2 h, and carpels of primary and secondary inflorescences were used.
Determination of Fresh and Dry Weight, and Calculation of Water Content
Carpels were harvested and fresh weight (FW) was determined immediately. After drying at 50°C for one week, the dry weight was determined and the water content (WC) calculated according to the formula: WC = (FW − dry weight)/FW.
Determination of JA, JA-Ile, and GAs
Quantitative analysis of JA and JA-Ile was done using 50 mg of homogenized plant material per sample as described (Balcke et al., 2012). In brief, plant material was extracted with 500 μL pure methanol supplied with [2H6]JA, and [2H2] JA-Ile (50 ng each) as internal standards. After centrifugation, the supernatant was diluted with 9 volumes of water and subjected to solid phase extraction on HR-XC (Chromabond, Macherey-Nagel) column. Elution was done with 900 μL acetonitrile. Ten μL of the eluate were subjected to ultraperformance liquid chromatography–tandem mass spectrometry according to Balcke et al. (2012). The contents of JA, and JA-Ile were calculated using the ratio of analyte and internal standard peak heights.
Samples were analyzed for GA content according to Urbanová et al., 2013 with some modifications. Carpels of stage 3 from wild type, jai1-1, and myb21-2 (30 mg of fresh weight) were homogenized with 1 mL 80% (v/v) acetonitrile containing 5% (v/v) formic acid and 19 internal GA standards ([2H2]GA1, [2H2]GA3, [2H2]GA4, [2H2]GA5, [2H2]GA6, [2H2]GA7, [2H2]GA8, [2H2]GA9, [2H2]GA12, [2H2]GA12ald, [2H2]GA15, [2H2]GA19, [2H2]GA20, [2H2]GA24, [2H2]GA29, [2H2]GA34, [2H2]GA44, [2H2]GA51 and [2H2]GA53; OlChemIm) and extracted at 4°C overnight with constant stirring. The homogenates were centrifuged for 10 min at 4°C, and supernatants were purified using mixed mode anion exchange cartridges (Waters, www.waters.com). Analysis was done by ultra-high-performance liquid chromatography (Acquity UPLC System; Waters) coupled to triple-stage quadrupole mass spectrometer (Xevo TQ MS, Waters) equipped with electrospray ionization interface. GAs were detected using multiple-reaction monitoring mode based on transition of the precursor ion [M-H]− to the appropriate product ion. Data were acquired and processed by Masslynx 4.1 software (Waters), and GA levels were calculated using the standard isotope-dilution method (Rittenberg and Foster, 1940).
Histochemical Analyses
For light microscopic analysis carpels were fixed in 3% (v/v) sodium cacodylate-buffered glutardialdehyde (pH 7.2), dehydrated in an ethanol series, and embedded in epoxy resin (Spurr, 1969). Semi-thin sections (1 µm) were stained with toluidine blue. To stain callose, sections were treated with 0.5% (w/v) NaOH in ethanol for 5 min followed by two washing steps with water. Immunostaining was done using a mouse anti-(1→3)-β-glucan antibody (Biosupplies, http://www.biosupplies.com.au/, Cat# 400-2) diluted 1:500 in phosphate buffer saline (PBS) containing 5% (w/v) BSA at 4°C overnight followed by a goat-anti-mouse-IgG antibody conjugated with AlexaFluor488 (Invitrogen, www.thermofisher.com/) at 37°C for 90 min. Sections were mounted in anti-fading reagent Citifluor (Science Services GmbH).
For immunocytochemical analysis of JAs, carpels were fixed with 4% (w/v) 1-ethyl-3-(3-dimethyl aminopropyl)-carbodiimide hydrochloride (Merck KgaA) in PBS, embedded in polyethylene glycol (PEG) 1500, and immunostained with an antibody against JA as described (Mielke et al., 2011).
Fixation of carpels with 4% (w/v) paraformaldehyde/0.1% (v/v) Triton X-100 in PBS and embedding in paraplast (Sigma-Aldrich) was used for TUNEL. The 12-µm-thick sections were pretreated with 2 µg mL−1 proteinase K in 10 mM Tris-HCl (pH 8.0) at 37°C for 30 min. After washing with PBS, TUNEL reaction was carried using the in situ Cell Death Detection Kit (Sigma-Aldrich) according to the supplier’s instructions. Positive controls were generated by pretreatment of sections with 1 µg mL−1 DNase I in PBS at room temperature for 10 min. Negative controls were obtained by omitting the terminal-deoxynucleotidyl-transferase. After washing, sections were mounted in the anti-fading reagent Citifluor.
Micrographs were taken using a Zeiss ‘AxioImager’ microscope (Zeiss) equipped with an AxioCam (Zeiss) and were combined using Photoshop 12.0.4 (Adobe Systems).
RNA Isolation and RT-quantitative PCR
RNA isolation from homogenized material was performed using the RNeasy Plant Mini Kit (Qiagen, www.qiagen.com) according to the supplier’s instructions including on-column digestion of DNA for 30 min or routine DNase treatment using DNA-free Kit (Thermo Scientific, /www.thermofisher.com/) afterward. RNA quality was tested by capillary electrophoresis using QIAxcel Advanced System (Qiagen). First strand cDNA synthesis was performed in a final volume of 20 μL with M-MLV Reverse Transcriptase, RNase H Minus, Point Mutant (Promega, /www.promega.de/) or ProtoScript II First Strand cDNA Synthesis Kit (NEB, /www.neb-online.de/) according to the supplier’s protocol using oligo(dT)19 primer.
The 3 μL of diluted cDNA was mixed with 2 μL 5x EvaGreen QPCR Mix II (Bioandsell), 2 pmol forward primer, and 2 pmol reverse primer and dH2O (ad 10 µl). QPCR primers for candidate genes were designed with the software CloneManager (Sci-Ed Software) and Primer3 (http://bioinfo.ut.ee/primer3/) using the corresponding sequences of the tomato genome (sol genomic network, http://solgenomics.net and GeneBank, https://www.ncbi.nlm.nih.gov/genbank/; for primer sequences see Supplemental Table 4). PCR was done using RT-qPCR-System CFX Connect (Bio-Rad, www.bio-rad.com) with the following protocol: denaturation (95°C for 15 min), amplification (40 cycles of 95°C for 15 s and 56°C for 30 s), and melting curve (95°C for 10 s, 60°C heating up to 95°C with a heating rate of 0.05°C s−1). Data were analyzed with CFX Manager Software (Bio-Rad). Relative gene expressions were calculated by the comparative quantitation cycle method (Schmittgen and Livak, 2008) using S. lycopersicum TAP42 INTERACTING PROTEIN (SlTIP41; Expósito-Rodríguez et al., 2008) as constitutively expressed gene. Each reaction was measured in duplicate or triplicate.
Transcript Profiling using GeneChip Analysis
RNA isolated independently from three ovule samples per developmental stage (1, 3, and 6), all from wild type and jai1-1, was analyzed using Agilent-Tomato 44K-full genome chips. Synthesis and purification of cDNA; synthesis, labeling, purification, quality control, and fragmentation of coding RNA; as well as hybridization, washing, and scanning of the chips were done by the service partner Atlas Biolabs (http://www.atlas-biolabs.de/) according to the supplier’s protocols.
Data analysis was performed using ArrayStar (www.dnastar.com). All samples were quantile normalized. To identify genes that were differentially expressed, pairwise comparison between wild type and jai1-1 at the three different developmental stages was done. P-values were corrected according to the Benjamini-Hochberg method using the statistical package of ArrayStar. Genes were considered as differentially expressed when adjusted p-values were ≤ 0.01 and fold change ≥ 8. Sets of gene showing differential expression were obtained for each developmental stage. Gene annotation was done using data from ‘sol genomic network’ and mapping results obtained by MapMan (www.mapman.gabipd.org) followed by manual check and correction using nucleotide BLAST (www.ncbi.nlm.nih.gov).
Transcript Profiling using Illumina Sequencing
RNA isolated from carpels of stage 3 from wild type, jai1-1, and myb21-2 was used for Illumina sequencing. A 1 μg subsample of total RNA from each of 9 RNA extracts (1 stage×3 genotypes×3 biological replicates) was sent to the GATC Biotech AG (https://www.eurofinsgenomics.eu/) and used to produce Illumina HiSeq 2500 strand-specific 2×150 bp paired-end reads. Total RNA quality was determined using an Agilent 2100 Bioanalyzer, and quality of the sequenced raw reads were assessed by FastQC (version 0.11.5, http://www.bioinformatics.babraham.ac.uk/projects/fastqc/).
Read adapters were removed with cutadapt (version 1.16), and quality trimming was performed with sickle (version 1.33). Cleaned reads were mapped by STAR (–alignIntronMin 40–alignIntronMax 5000, version 2.5.2) against the reference genome assembly of tomato SL2.50 downloaded from the Plant ENSEMBL database. Gene counts were obtained with subread’s featureCounts (version 1.5.1) by counting unique exonic overlaps. The reproducibility of the biological replicates was confirmed by hierarchical clustering using the 1-correlation (spearman) as distance measures between samples. For detecting differential gene expression edgeR was used to perform likelihood ratio test on the linear models. Library sizes were scaled using the TMM method in egdeR, and dispersion was estimated tagwise (genes). Low expressed genes with counts per million (cpm) < 1 within each replicate group were excluded from statistical analysis. P-values were adjusted using Benjamini and Hochberg false discovery rate procedure. Differently expressed genes were identified by a significance threshold of 0.05 and a minimal log2-fold-change of ±1.
Further analysis by enrichment of Gene Ontology terms was performed based on gene set enrichment analysis (Subramanian et al., 2005) using the enrichment map app from Cytoscape (Merico et al., 2010) with a false discovery rate–q-value and similarity cutoff of 0.15 and 0.3, respectively.
Cloning of MYB21 and JAZ-Encoding Genes for Interaction Studies
Gateway recombination was used for Y2H and BiFC approaches. Therefore, PCR was performed with AccuPrime Pfx SuperMix (Invitrogen) and primers containing attB1/B2 sites using cDNA from tomato shoot material treated with JA-methyl ester to amplify JAZ1, JAZ2, JAZ6, JAZ7, JAZ8, JAZ9, JAZ10, and JAZ13. For amplifying the coding sequence (CDS) of JAZ3, JAZ4, JAZ11, MYB21, and JAZ5, cDNA from stamen and JA-methyl ester-treated root material, respectively, was used. The PCR products were cloned into vector pDONR221-P1P2 and pDONR221-P1P4/P3P2 for Y2H and BiFC, respectively, using the BP-Clonase Kit II (Invitrogen). Primer pairs for the construction of the entry vectors are listed in Supplemental Table 5.
For Y2H, all JAZ coding sequences were fused to DNA BD and MYB21 CDS to AD of GAL4 by recombination into pDEST32 and pDEST22 (Invitrogen), respectively. BiFC binary vectors were achieved using 2in1 vectors according to Grefen and Blatt, 2012. For splitTALE assay (Schreiber et al., 2019), the CDS of MYB21 with or without AD (C-terminal deletion of 25 aa) and JAZ9 were cloned into level 1 vectors using Golden Gate method (Werner et al., 2012), resulting in a fusion either to the TALE BD (N-terminal; Act2p:TALE(BD)-MYB21) or to AD (C-terminal; 35S:JAZ9-TALE[AD]; Schreiber and Tissier, 2016). For splitTALE assays, constructs were transformed into Agrobacterium tumefaciens strain GV3101 and used for transient expression in Nicotiana benthamiana leaves.
Primer pairs are listed in Supplemental Table 6. All constructs were transformed into Escherichia coli strain DH10B, from which plasmids were isolated using a plasmid isolation kit (Macherey-Nagel).
Transcriptional Activation Analysis in Yeast Cells
For testing the transcriptional activity of MYB21, its CDS was fused to GAL4 BD into pDEST32. Yeast transformation was performed as described below, and pDEST22 was added as an empty vector. Transcriptional activity was deduced from yeast growth on synthetic defined (SD) medium –Trp/-Leu/-His (plus 2 mM 3-amino-1,2,4-triazol).
Yeast-Two-Hybrid Assays
Transformation of competent pJ69 yeast cells was done as described previously (Gietz and Schiestl, 2007). Yeast cells were incubated with bait and prey plasmids, PEG 3350, 0.1 M lithium acetate, and single-stranded carrier DNA at 42°C for 1 h. Yeast cells were grown for 6 d on SD medium -Trp/-Leu. Single colonies were selected and cultured overnight at 30°C to an OD600 of 1, harvested, and resuspended in sterile water. The 5 μL of undiluted and 1:10 and 1:100 diluted suspensions were dotted either on SD medium -Trp/-Leu or on SD medium -Trp/-Leu/-His (plus 2 mM 3-amino-1,2,4-triazol) and incubated at 30°C for 3 d. Cells grown on SD medium -Trp/-Leu/-His were indicative for an interaction of the proteins under study.
BiFC in N. benthamiana Protoplasts
Plasmid DNA was isolated from 50 mL E. coli overnight culture using the PureYield Plasmid Midiprep Kit (Promega). DNA purification was done by filtration, precipitation using PEG, washing with 70% (v/v) ethanol, and resolving in 20 μL water. Protoplast transformation was done according to Yoo et al., 2007 with some modifications. Protoplasts were isolated from two young leaves of 4-week-old N. benthamiana plants that were cut in small pieces and vacuum infiltrated with 5% (w/v) cellulase Onozuka R-10 and 0.4% (w/v) macerozyme R.10 (both from Yakult, www.yakult.co.jp/ypi/en/) in 0.4 M mannitol, 20 mM KCl, 20 mM MES (pH 5.7), 110 mM CaCl2, and 0.1% (w/v) BSA, and incubated for 4 h followed by shaking for 30 min. The cells were directly passed through a nylon mesh filter, the flow-through centrifuged for 1 min at 200 g, and resuspended first in W5 (154 mM NaCl, 125 mM CaCl2, 5 mM KCl, 2 mM MES, pH 5.7) and finally in mannitol-magnesium buffer (0.4 M mannitol, 15 mM MgCl2, 4 mM MES, pH 5.7) to get a density of 105 protoplasts per mL. For transformation, 200 μL of protoplast suspension were incubated with 10 µg plasmid and 40% (w/v) PEG in 0.2 M mannitol and 100 mM CaCl2 for 5 min. After sedimentation at 200 g for 1 min, cells were resuspended in 200 μL W1 buffer (0.5 mM mannitol, 20 mM KCl, 4 mM MES, pH 5.7) and incubated at room temperature overnight. Images were captured by a Confocal Laser Scanning Microscope (LSM700, Zeiss). Upon excitation of 488 nm, yellow fluorescent protein fluorescence and chlorophyll autofluorescence were recorded at 493-531 nm and 644-800 nm, respectively.
SplitTALE Assays in N. benthamiana
Overnight cultured Agrobacteria transformed with the respective plasmids were resuspended in infiltration media consisting of 10 mM MES, 10 mM MgCl2, and 100 µM acetosyringone to OD600 = 0.4. Agrobacteria containing the splitTALE constructs and the reporter construct 4xSTAP1:GUS (Schreiber and Tissier, 2016; Schreiber et al., 2019) were mixed in ratio 1:1:1 and used to transiently transform N. benthamiana leaves through syringe-mediated infiltration (Sparkes et al., 2006). At 3 d after inoculation, leaf material was harvested, frozen in liquid nitrogen, homogenized, and mixed with extraction buffer containing 50 mM Na2HPO4 (pH 7), 10 mM EDTA, 0.1% (w/v) SDS, 0.1% (v/v) Triton X-100, and 10 mM β-mercaptoethanol. GUS assay was performed as described (Kay et al., 2007). GUS activity was measured after incubation with 4-methylumbelliferyl-β-d-glucuronide trihydrate (Duchefa, www.duchefa-biochemie.com) at 37°C for 85 min and calculated in relation to the total content of proteins determined by Bradford method (Bradford and Trewavas, 1994).
Subcellular Localization in N. benthamiana
For localization studies, the Golden Gate method (Weber et al., 2011; Werner et al., 2012; Engler et al., 2014) was used to clone the genomic sequence of MYB21 with the second intron shortened (2204 bp deleted) with a C-terminal GFP fusion under the control of 35S promoter in level 1 vector (see primer pairs in Supplemental Table 6). The construct was infiltrated in N. benthamiana leaves mediated by A. tumefaciens GV3101 pMP90 (OD600 = 0.5). Pictures were taken 3 d after infiltration using AxioImager (Zeiss) equipped with the proper filters for GFP and an AxioCam camera (Zeiss). The infiltrated area was harvested, and total protein extracted (Meyer et al., 1988) and subjected to SDS-PAGE followed by immunoblot analysis according to standard protocols. Immunodetection was performed with an anti-GFP antibody (Santa Cruz Biotechnology, https://www.scbt.com) as primary antibody and anti-mouse IgG conjugated with alkaline phosphatase (EMD Millipore Corporation) as secondary antibody. Immunodecorated GFP protein was stained with p-nitroblue tetrazoline chloride and 5-bromo-4-chloro-3-indolylphosphate.
Cloning of CRISPR/ Cas9 and Complementation Construct
Constructs were designed harboring two single guide RNAs targeting two specific 20-bp sequences (chosen using CRISPRdirect crispr.dbcls.jp) in the second exon of MYB21 under the control of the Arabidopsis (Arabidopsis thaliana) U6 consensus promoter (Nekrasov et al., 2013). All cloning steps were done using Golden Gate method (Werner et al., 2012; Engler et al., 2014). All cloning cassettes including the Cas9 endonuclease CDS (Nekrasov et al., 2013) under the control of 35S promoter were assembled into the level 2 vector pAGM4723.
For the complementation of myb21-2, a sequence of 1.5 kb upstream of the ATG start codon was cloned to drive the expression of MYB21 (genomic sequence with shortened second intron) with a C-terminal fusion to a 3xFlag tag. All cloning cassettes were assembled into the level 2 vector pAGM4673. The constructs were finally used for transformation of A. tumefaciens GV3101 followed by stable tomato transformation. All primer pairs are listed in Supplemental Table 6.
Stable Tomato Transformation
Tomato was stable transformed via somatic embryogenesis using cotyledon pieces of 8-d-old seedlings germinated on phytoagar (Duchefa) for 6 d in darkness and 2 d under 16 h photoperiod. The cotyledon pieces with 2 cutting sites were placed upside down on KCMS media (pH 5.7; 4.4 g/L MS including vitamins, 20 g/L Suc 200 mg/l KH2PO4, 8 g/L agar [Sigma-Aldrich], 0.9 mg/l thiamine, 100 µg/L kinetin, 0.5 mg/l IAA) and preconditioned overnight in the dark (Čermák et al., 2015). Overnight cultures of Agrobacterium were pelleted and resuspended in KCMS liquid-media (4.4 g/L MS including vitamins, 20 g/L Suc, 200 mg/l KH2PO4) to OD600 = 0.08 and used for transformation according to Fernandez et al., 2009. After 2 d in the dark, cotyledon pieces were transferred to 2Z-media with the abaxial side in contact to the media (4.4 g/L MS including vitamins, 30 g/L Suc, 100 mg/l myoinositol, 8 g/L agar, 1 mL/l Nitsch-vitamins [1000×; Duchefa], 375 mg/l timentin [ticarcillin and clavulanate 15:1; Duchefa], 100 mg/l kanamycin, 0.1 mg/l IAA, 2 mg/l trans-zeatin riboside; pH 5.8) and cultivated in 16-h photoperiod at 24°C. Developing calli and plantlets were transferred every 2 weeks to Z-media containing decreased trans-zeatin riboside concentrations from 1 to 0.5 mg/l and finally 0.1 mg/l GA instead of trans-zeatin riboside (Eck et al., 2006). After another 2 weeks on rooting media (4.3 g/L MS without vitamins, 30 g/L Suc, 100 mg/l myoinositol, 7 g/L agar, 1 mL/l Nitsch-vitamins, 375 mg/l timentin, 100 mg/l kanamycin), plants with developed roots were transferred into soil and slowly adjusted to 70% of relative humidity.
Generation and Characterization of Transgenic Arabidopsis Plants
The genomic sequence of MYB21 of tomato (second intron shortened at 2204 bp) and Arabidopsis was amplified from genomic DNA using primers containing attB1/B2 sites and cloned into pDONR221 to generate entry vectors attR1-MYB21genomic-attR2. A 7281-bp region upstream of the start codon of the AtMYB21 gene was cloned in pUC57-L1-KpnI-XmaI-R1, a pUC57 vector carrying KpnI and XmaI restriction sites between attL4 and attR1 sites for Gateway cloning (Invitrogen); this vector was kindly provided by the group of Niko Geldner in the University of Lausanne. The sequence immediately upstream of the AtMYB21 start codon is AT rich, impairing primer design. Therefore, the first six upstream nucleotides (a string of six adenines) were excluded from cloning. PCR amplification of the entire 7281 bp was unsuccessful; therefore, the region was cloned in two steps. First, primers P230+P194u amplified a 2277-bp region containing an endogenous KpnI restriction site at the 5′-end. The fragment was digested with KpnI and XmaI and cloned into pUC57-L1-KpnI-XmaI-R1 to generate pMP60. Second, primers P231+P197u amplified a 5558-bp fragment that overlaps by 546 bp with the insert in pMP60. This fragment was digested with KpnI and ligated to KpnI-linearized pMP60 to generate pMP71, the final pENTRY attL4-MYB21pro-attR1. The sequences cloned in pMP60 and pMP71 were confirmed by Sanger sequencing. The corresponding entry vectors were used for further recombination into pDG27 with OLE1-tagRFP for transformant selection (Shimada et al., 2010). The resulting binary vectors were used to transform A. tumefaciens strain GV3101 to mediate the transformation of Arabidopsis myb21-5 (Reeves et al., 2012) heterozygous plants by the floral dip method (Clough and Bent, 1998). All primer pairs are listed in Supplemental Table 7.
Transformed seeds were identified by checking for red fluorescence due to OLE1:tagRFP expression. Segregating homozygous Atmyb21-5 lines were selected by using a cleaved-amplified polymorphic sequence marker for the mutant single nucleotide polymorphism. For this, the PCR products (see primer pair Supplemental Table 8) were digested with EcoRI (high fidelity, NEB) for 2 h.
Mutant Screen by TILLING
Lines mutated in the gene encoding MYB21 (Solyc02g067760) were screened from 9216 EMS-mutagenized lines by TILLING using LI-COR DNA analyzer according to the procedure described by Okabe et al., 2011. A 606-bp region encompassing the first two exons of the gene were amplified by PCR using the fluorescent DY-681- and DY-781-labeled primers (Biomers; Supplemental Table 9).
Genotyping of Slmyb21 Tomato Mutants
PCRs were performed using Phire Plant Direct PCR Master Mix (Thermo Scientific) with 0.3-mm leaf discs as template. To select myb21 TILLING mutants cleaved-amplified polymorphic sequence markers (http://helix.wustl.edu/dcaps/dcaps.html) were used to identify the wild-type single nucleotide polymorphisms. The SlMYB21 PCR products were digested using HphI for line TOMJPE7979 and Eco31I for Slmyb21-1 (fast digest, Thermo Fisher Scientific). TILLING mutant line TOMJPE8245 was genotyped by high-resolution melt curve qPCR (from 60° to 90°C each 0.1°C 10-s detection). The knock-out mutants myb21-2 and myb21-3 generated via CRISPR/Cas9 were characterized by sequencing (Eurofins) the purified 500 bp PCR product flanking the single guide RNA target sites in the second exon of SlMYB21. In comparison with the wild-type sequence, either a T- deletion or T-insertion was detected. To identify homozygous myb21-2 plants in the complementation approach, the silent mutation at the beginning of the second exon within the transgene was checked additionally in the PCR product sequence (primer myb21-2_for and 8245_rev). All primer pairs are listed in Supplemental Table 8.
Statistical Analysis
One-way analysis of variance (one-way ANOVA) followed by Tukey Honestly Significant Difference Test for statistical significance was performed using the VassarStats website for Statistical Computation (VassarStats) or R (R Core Team; https://www.r-project.org/). Two group comparisons between wild type and mutant samples were done using the Student's t test. Means were considered significantly different based on threshold value corresponding to P < 0.05 (ANOVA) or P < 0.05, P < 0.01, and P < 0.001 as indicated by (*), (**) and (***), respectively, (Student's t test). Detailed results of tests are shown in Supplemental Data Set 3.
Accession Numbers
The original Agilent GeneChip data as well as normalized data from this study are publicly available at ArrayExpress database (http://www.ebi.ac.uk/arrayexpress/) under accession number E-MTAB-7544. Sequence data (FASTQ files from RNA-seq experiments) from this article can be found at in the ArrayExpress database under accession number E-MTAB-7545.
Sequence data from key genes in this article can be found in the GenBank/EMBL/Solgenomics databases under the following accession numbers: SlMYB21, Solyc02g067760; SlJAZ1, Solyc07g042170; SlJAZ2, Solyc12g009220; SlJAZ3, Solyc03g122190; SlJAZ4, Solyc12g049400; SlJAZ5, Solyc03g118540, SlJAZ6, Solyc01g005440; SlJAZ7, Solyc11g011030; SlJAZ8, Solyc06g068930; SlJAZ9, Solyc08g036640; SlJAZ10, Solyc08g036620; SlJAZ11, Solyc08g036660; SlJAZ13, Solyc01g103600; SlCOI1, Solyc05g052620; AtMYB21, At3g27810.
Supplemental Data
Supplemental Figure 1. Carpels of jai1-1 show enhanced growth in comparison to wild type.
Supplemental Figure 2. Morphology of wild type and jai1-1 ovules (stages 1 – 4).
Supplemental Figure 3. SlMYB21 encodes a flower specific MYB domain protein.
Supplemental Figure 4. splitTALE assay testing the interaction of SlMYB21 with JAZ9 fused to activation and binding domain of TALE, respectively.
Supplemental Figure 5. TILLING line Slmyb21-1 flowers and fruits show a phenotype similar to jai1-1.
Supplemental Figure 6. Wild type and Slmyb21-2 flower buds at stage 3 differ in length and carpel weight.
Supplemental Figure 7. Gene ontology analysis of differentially regulated genes in carpels of jai1-1 and Slmyb21-2 in comparison to wild type.
Supplemental Table 1. Relative transcript levels in relation to SlTIP41 of genes commonly down-regulated in jai1-1 and Slmyb21-2 as determined by RT-qPCR.
Supplemental Table 2. Relative transcript levels in relation to SlTIP41 of genes commonly up-regulated in jai1-1 and Slmyb21-2 as determined by RT-qPCR.
Supplemental Table 3. Levels of gibberellins in carpels at stage 3 from wild type, jai1-1 and Slmyb21-2.
Supplemental Table 4. Primer sequences for RT-qPCR.
Supplemental Table 5. Primer sequences used for gateway cloning of JAZs and MYB21-CDS.
Supplemental Table 6. Primer sequences used for golden gate cloning.
Supplemental Table 7. Primer sequences for generation of Arabidopsis transformation construct using gateway cloning.
Supplemental Table 8. Primer sequences used for genotyping of jai1-1, myb21 TILLING lines, Slmyb21-2 and Atmyb21-5.
Supplemental Table 9. Primer sequences used for TILLING.
Supplemental Data Set 1. Tomato genes found in the microarray data to be differentially regulated in ovules of wild type and jai1-1 plants.
Supplemental Data Set 2. Tomato genes found in the RNA-seq data to be differentially regulated in carpels of wild type, jai1-1 and Slmyb21-2 plants.
Supplemental Data Set 3. Results of student’s t tests and ANOVA of quantitative data.
Dive Curated Terms
The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:
Acknowledgments
We thank Hagen Stellmach and Merve Kuru (IPB Halle) for help in JA and JA-Ile quantification, BiFC approaches, and qPCR. Simone Fraas (Martin-Luther-University Halle-Wittenberg) is acknowledged for performing sectioning of tomato carpels, and Gregg Howe (Michigan State University) and Jason Reed (University of North Carolina) for providing seeds of Sljai1-1 and Atmyb21-5, respectively. We thank Edelgard Wendeler (MPIPZ Cologne) for technical support. Tomato seeds (TOMJPE7564, TOMJPE7979, TOMJPE8245, TOMJPW0469) were provided by University of Tsukuba, Gene Research Center, through the National Bio-Resource Project (NBRP) of the AMED. Claus Wasternack (IPB Halle) is highly acknowledged for critical reading of the manuscript. This work was supported by Deutsche Forschungsgemeinschaft (DFG) (grant HA2655/12-1), the Czech Science Foundation (Grantová agentura České republiky) (grant 18-10349S), and the European Regional Developmental Fund Project “Centre for Experimental Plant Biology” (grant CZ.02.1.01/0.0/0.0/16_019/0000738).
AUTHOR CONTRIBUTIONS
R.S., S.D., S.M., and B.H. designed the research; R.S., S.D., C.G., G.H., T.S., Y.O., H.E., I.F.A., and D.T. performed research; R.S., S.D., B.A., D.T., and B.H. analyzed data; and R.S. and B.H. wrote the article with input from all authors.
Footnotes
↵[OPEN] Articles can be viewed without a subscription.
- Received January 2, 2019.
- Revised February 19, 2019.
- Accepted March 19, 2019.
- Published March 20, 2019.
References
