A tomato LATERAL ORGAN BOUNDARIES transcription factor, SlLOB1, predominantly regulates cell wall and softening components of ripening

Significance

A tomato fruit ripening–specific transcription factor, SlLOB1 predominantly influences fruit cell wall–related gene regulation and textural changes during fruit maturation and thus is distinct from broadly acting ripening transcription factors described to date that influence many ripening processes. As such, SlLOB1 is an intermediate regulator primarily influencing a physiological subdomain of the overall ripening transition.

Abstract

Fruit softening is a key component of the irreversible ripening program, contributing to the palatability necessary for frugivore-mediated seed dispersal. The underlying textural changes are complex and result from cell wall remodeling and changes in both cell adhesion and turgor. While a number of transcription factors (TFs) that regulate ripening have been identified, these affect most canonical ripening-related physiological processes. Here, we show that a tomato fruit ripening–specific LATERAL ORGAN BOUNDRIES (LOB) TF, SlLOB1, up-regulates a suite of cell wall–associated genes during late maturation and ripening of locule and pericarp tissues. SlLOB1 repression in transgenic fruit impedes softening, while overexpression throughout the plant under the direction of the 35s promoter confers precocious induction of cell wall gene expression and premature softening. Transcript and protein levels of the wall-loosening protein EXPANSIN1 (EXP1) are strongly suppressed in SlLOB1 RNA interference lines, while EXP1 is induced in SlLOB1-overexpressing transgenic leaves and fruit. In contrast to the role of ethylene and previously characterized ripening TFs, which are comprehensive facilitators of ripening phenomena including softening, SlLOB1 participates in a regulatory subcircuit predominant to cell wall dynamics and softening.

Tomato (Solanum lycopersicum) is a widely studied model of fleshy fruit development and ripening (1, 2). Softening is an important aspect of ripening physiology, as it determines palatability for frugivores and is a key factor in determining damage and loss in fruit food supply chains. Genotypes inhibited in ripening, early harvest, and controlled atmospheres limiting respiration and ethylene synthesis are deployed to maintain firmness and shelf life, often at the expense of quality. A clearer understanding of the genetic basis of fruit softening and ripening regulation are essential to optimize shelf life and quality for food and nutritional security.

Ripening-related textural changes are closely associated with cell wall metabolism, and extensive efforts have focused on understanding tomato fruit cell wall remodeling and the underlying genes (3, 4). Particularly notable wall modifications during ripening include depolymerization of pectins and hemicelluloses and pectin solubilization, which contribute to dissolution of the middle lamella, reduced cell adhesion, and cell wall swelling (3, 4). These are orchestrated by an array of cell wall–modifying proteins, the most studied of which is endo-polygalacturonase 2a (PG2a) from tomato. PG2a is encoded by a fruit-specific and ripening-induced gene that is responsible for up to 1% of messenger RNA (mRNA) in ripening pericarp (5), and its repression was the basis of the first commercialized transgenic plant, Flavr Savr (6). Numerous additional cell wall–degrading enzymes have been characterized, including pectate lyase (PL) pectin methylesterase (PE2, Pmeu1), β-galactosidase (TBG4) (3, 4), endoglucanase (CEL2) (7), and xyloglucan endotransglucosylase/hydrolases (XTH5) (8), also homologs from other fruit species (9). Additionally, expansin proteins [e.g., EXP1 10], which have no known enzymatic activity, contribute to cell wall loosening and textural changes (9, 10). However, altering the expression of these genes individually does not have substantive effects on softening with the notable exception of PL (11). Thus, a deeper understanding of fruit textural changes requires examination of higher order regulators that influence multiple cell wall–related genes.

Tomato-ripening mutants such as rin (ripening inhibitor, encoding a MADS-box transcription factor [TF]), Cnr (Colorless nonripening, encoding an SBP-box TF), and nor (nonripening, encoding a NAC TF) inhibit softening in addition to many other ripening characteristics, including color, flavor, ethylene hormone synthesis, and aroma (12–14). While the rin mutation has been shown to have dominant gain-of-function repressor action, the RIN gene is nevertheless critical in virtually all ripening activities (15, 16). Additional ripening genes have been identified through mutations or gene expression profiles and some functionally characterized, including the TF genes TAGL1, FUL1, FUL2, MADS1, NAC1, AP2a, and SlGRAS38 (1, 2, 17, 18). As with RIN, these influence a broad range of ripening processes, such as ethylene synthesis, pigmentation, time to initiation, and completion of ripening, and many cell wall–associated genes (e.g., PG2a, EXP1, PL, PE2, and TBG4) have altered expression in mutant or repression lines of these TFs (1, 2, 17–19). Two additional TFs, EREBP6 and LeHB1, are more specific to ethylene synthesis (1, 2). In the case of RIN, many are direct targets (20). Importantly, however, TFs that primarily target only the cell wall–modifying component of the ripening cascade have yet to be reported.

Earlier molecular studies revealed the genetic basis of ethylene synthesis (21, 22) and responses (23) and highlighted the essential role of this gaseous hormone in ripening phenotypes, including softening. Ethylene is a coordinator of ripening pathways acting in concert with many of the ripening TFs. Ethylene response factor (ERF) genes are TFs that reside at the end of the ethylene signaling pathway, and several tomato fruit ripening– and softening–related ERFs have been described. For example, AP2a RNA interference (RNAi) fruit are softer due to enhanced ethylene production, suggesting a negative regulatory effect (17). Additionally, overexpression of tomato LeERF1 was reported to accelerate ripening, including softening, while its repression extended shelf life (24). ERF2.2 underlies a firmness quantitative trait loci (Firs.p.QTL2.2) but has not yet been functionally characterized (25).

We searched public tomato fruit transcriptome data [https://tea.solgenomics.net (18, 26)] for TFs induced both in ripening pericarp and just prior to ripening in the locular gel surrounding the seeds. We hypothesized that genes expressed in the solubilizing locular gel immediately before ripening and in the pericarp at ripening initiation might participate more exclusively in cell wall–related activities. A tomato LATERAL ORGAN BOUNDRIES (LOB) domain gene, SlLOB1, matched this profile. LOB genes belong to a plant-specific TF family of 42 members in Arabidopsis thaliana and 35 in rice (Oryza sativa) (27). Based on the structure of the N terminus LOB domain, two subfamilies have been defined. Class I LOB proteins contain a complete LOB domain comprised of three conserved subdomains: the C domain involved in DNA binding, the GAS (Gly-Ala-Ser) domain, and the L domain for protein interaction. Class II LOB proteins lack the L domain (28–30). Most LOBs belong to the class I subfamily, including SlLOB1. LOB proteins have an essential role in lateral organ development, including lateral organ initiation and patterning, pollen and root development, plant regeneration capacity, pathogen responses, and secondary xylem and phloem growth in addition to metabolic process (anthocyanin and nitrogen metabolism) (28), while one has been associated with ripening banana fruit and expansin gene expression (31). Using the reference tomato genome sequence, 46 tomato LOBs were identified (32), and only one member, SlLBD40, has been functionally defined to date with evidence suggesting a role in drought tolerance (33).

Using RNAi repression and ectopic expression in transgenic tomato plants, we demonstrate that SlLOB1 acts as a transcriptional activator of a broad suite of cell wall–related genes and fruit softening. Additional ripening phenotypes were minimally affected including ripening initiation, onset of the ethylene burst, and full ripe fruit appearance, although elevated carotenoid levels were observed in ripe fruit of the repressed lines. In contrast to many previously described LOB genes, SlLOB1 repression revealed no significant phenotypes associated with organ development or differentiation, though such phenotypes were observed with ectopic overexpression. SlLOB1 is a ripening-related TF that is distinct from those described to date in that its primary targets are a suite of genes mediating cell wall and textural changes, a distinct subset of the late fruit development and ripening program.

Results

Tomato SlLOB1 Is Predominantly Expressed in Maturing Fruit.

SlLOB1 expression in the pericarp coincides with ripening. SlLOB1 is initially induced at the mature green (MG) stage, highly expressed at early ripening (breaker stage [BR]), and drops slightly after BR. Interestingly, in the MG locule, which becomes liquefied prior to pericarp ripening, SlLOB1 mRNA accumulates at three times the rate in pericarp (Fig. 1A). Expression of SlLOB1 occurred minimally in vegetative and floral tissues (Fig. 1A). Stems have the highest nonfruit expression yet less than 10% of levels in the immature green (IMG) locule. Locular gel expression is higher than in pericarp at all stages except BR in which levels are similar (Fig. 1A). Initial SlLOB1 expression in IMG locule is sixfold that of IMG pericarp. Notably, the RIN, NOR, and DML2 ripening regulator genes have been reported to show similar expression profiles (2).

Fig. 1.

Fruit gene expression and phenotypes following SlLOB1 repression. (A) Expression of SlLOB1 in WT tissues leaves (Le), root (R), and stem (St) of 2-wk-old seedlings, anthers (An), floral buds (B), prepollination carpel (Ca), sepals (Se), petals (Pe), seed (S), breaker stage (BR), immature green (IMG), mature green (MG), breaker + 7 d (B7), and breaker + 15 d (B15). (B) Relative transcript abundance of SlLOB1 in WT and two RNAi lines (LOB #3 and #6) at stages MG, BR, B7, and B15 of pericarp and locule (gel). (C) Fruit firmness measured by fruit compression and pericarp penetration at the indicated developmental stages. *P < 0.05, **P < 0.01. (D) WT and SlLOB1 RNAi (LOB #3) fruit at indicated developmental stages (see A) with pericarp partially removed to see locule development. Error bars indicate SE.

SlLOB1 Repression Results in Reduced Softening and Extended Shelf Life.

To investigate the function of SlLOB1 in fruit ripening, we generated seven independent SlLOB1 RNAi tomato lines, three of which (#1, #3, and #6) were assessed in the T1 generation. T2 generations were developed for the two mostly strongly repressed lines, #3 and #6 (SI Appendix, Fig. S1). SlLOB1 RNAi lines showed normal growth phenotypes as compared to wild type (WT), with no notable differences observed (SI Appendix, Fig. S2) and consistent with its expression being predominant in maturing fruit tissues. SlLOB1 mRNA levels in T2 RNAi BR pericarp and locule tissues were observed to be reduced to ≤10% of levels in untransformed WT (Fig. 1B).

To evaluate fruit softening, we measured both whole fruit compression and pericarp penetration force using a texture analyzer. Increased firmness was evident from BR through the red-ripe (RR) stage (Fig. 1C). Pericarp penetration force was the same in transgenic fruit as in WT at MG but was 25% higher at BR, and more than twice the force was needed at BR + 7 d (B7) and BR + 15 d (B15). A similar trend was observed for whole fruit compression (Fig. 1C). SlLOB1-repressed fruit displayed reduced collapse after 30 d of storage, although they were similar to WT after 60 d (Fig. 2A). Water loss from the fruit during storage, determined gravimetrically, was also substantially lower in the transgenic fruit (Fig. 2B). We also measured the fruit cuticle thickness (SI Appendix, Fig. S3A) and the force needed to penetrate the cuticle (SI Appendix, Fig. S3B), but neither showed a difference between genotypes, suggesting no major biomechanical role for the cuticle in the enhanced firmness phenotype of the SlLOB1 RNAi fruit.

Fig. 2.

Postharvest shelf life and water loss of T2 SlLOB1 RNAi fruit. (A) WT and LOB #3 and #6 fruit were harvested at B15 and stored at room temperature (25 °C) and photographed at the indicated days postharvest. (B) The same fruit shown in A were weighed at the indicated days postharvest to measure water loss (**P < 0.01). Error bars indicate SE.

SlLOB1-repressed fruit displayed less locule liquefaction (conversion of the locule tissue to a liquid or jelly-like state of normally ripe tomato fruit) than WT fruit at the same stage (Fig. 1D), and, similar to the prior description of fruit of the Cnr mutant (13), SlLOB1-repressed locule tissue also released water more readily than that from WT (SI Appendix, Fig. S4 A and B), similar to observations for the Cnr mutant (13). Cnr cell wall extracts do not swell when hydrated, reflecting their more intact structure (34), and repression of SlLOB1 had a similar effect, with swelling reduced by 20% compared to WT (SI Appendix, Fig. S4C). In addition, SlLOB1-repressed cell wall extracts from the locular gel showed a similar pattern of precipitation to those from Cnr, which was distinct from WT extracts (SI Appendix, Fig. S4B). We noted that, while Cnr fruit float in water, this was generally not the case in the repression lines. However, it was observed in fruit from a single overexpression line (SI Appendix, Fig. S4D) that showed the greatest repression of most cell wall genes (SI Appendix, Table S1) and greater fruit firmness than the RNAi fruit, presumably due to cosuppression (SI Appendix, Fig. S5). Moreover, SlLOB1 RNAi fruit floated in 3.6% sucrose solution while those from WT sank, indicating reduced density (SI Appendix, Fig. S4E). While the function of the CNR gene has recently been questioned (46), the Cnr mutant displays many attributes of fruit with altered cell walls (13). Our comparison to Cnr reinforces that SlLOB1 repression is consistent with extensive cell wall alterations but does not say anything regarding CNR gene function.

Identification of Cell Wall Genes Influenced by SlLOB1 Suppression.

We performed RNA sequencing (RNA-seq) transcriptome analysis of B7 pericarp and locule tissues from WT and two T1 repression lines (Datasets S1 and S2 and Fig. 3 A–C). Consistent with the phenotypic strength (SI Appendix, Fig. S1), line #6 had more DEGs (differentially expressed genes, cutoff: P < 0.05; ration > 2 or < 0.5) than line #3. In addition, the pericarp had more DEGs than locule tissue (Fig. 3 A–C). A gene ontology (GO) analysis of down-regulated DEGs revealed enrichment in cell wall–related transcripts (SI Appendix, Table S2). To identify genes with the strongest support for influence by SlLOB1, we focused on those with differential expression in both tissues. A total of 34 genes were commonly down-regulated in all four comparisons of pericarp and locular gel from the two RNAi lines compared to WT (Fig. 3B). Of these, 10 are related to cell wall modification (Fig. 3D), including expansin, endo-1,4-β-glucanase, xylosidase, pectate lyase, and mannanase. Additional putative cell wall genes that were differentially expressed in either pericarp or locule tissues are listed (SI Appendix, Fig. S6A). Notably, PG2a was up-regulated in SlLOB1 RNAi lines (SI Appendix, Fig. S6A), though we note prior investigations indicate a minimal role of this enzyme in tomato softening (3, 4, 36, 37). PL was not altered in pericarp but was repressed in the locule of SlLOB1 RNAi fruit (line #6) (SI Appendix, Fig. S6B).

Fig. 3.

Transcriptome analysis, cell wall gene expression, and EXP1 protein of SlLOB1 RNAi lines. (A) Overview of DEGs in SlLOB1 repression lines. (B) Venn diagram of overlapping down-regulated genes between SlLOB1 repression lines LOB #3 and LOB #6 pericarp and locule (gel) compared to WT. (C) Venn diagram of overlapping up-regulated genes of tissues as in B. (D) Subset of the 10 cell wall–associated DEGs in B, annotation, functional categories, and expression (log 2 of reads per kilobase million). (E) Time course qRT-PCR of EXP1 in pericarp and locule (gel) of WT and SlLOB1 RNAi fruit. MG, BR, B7, and B15 tissues were compared to MG WT as reference (defined as 1). (F) Detection of EXP1 protein in SlLOB1 RNAi and control Breaker fruit. (Upper) Total protein ponceau staining. (Bottom) Immunoblotting. 1, WT MG pericarp (peri); 2, WT BR peri; 3, WT BR locule (gel); 4, LOB #3 BR peri; 5, LOB #3 BR gel; 6, LOB #6 BR peri; 7, LOB #6 BR gel. (G) qRT-PCR validation of selected cell wall genes at MG, BR, B7, and B15 with WT MG used as reference (defined as 1). Error bars indicate SE.

Based on GO term enrichment, 47 TFs were down-regulated and five were up-regulated in fruit from both SlLOB1-repressed lines (SI Appendix, Table S2). SlLOB1 was the only differentially expressed LOB, supporting the specificity of RNAi-mediated gene suppression (SI Appendix, Fig. S7A). Among the down-regulated TFs, members of the zf-RING, F-box, and bHLH families accounted for 39% of the total (SI Appendix, Fig. S7A). Several TFs displayed substantial down-regulation in both locular gel and pericarp, including the HD-Zip genes SlANL2b (Solyc06g035940.2) and Solyc03g120910.2 as well as BSD (BTF2-like transcription factors, Synapse-Associated, and DOS2-like proteins) (Solyc07g022920.2), MYB-like (Solyc06g066340.2), WOX (Solyc02g082670.2), and zf-C2H2 (Solyc06g062670.2) (SI Appendix, Fig. S7A). SlANL2b (Solyc06g035940.2) is paralogous to CD2 (CUTIN DEFICIENT 2), a regulator of epidermal cell cuticle deposition (38) whose orthologous A. thaliana loss-of-function mutant increased cell wall polysaccharide content (39). Solyc03g120910.2 is orthologous to an A. thaliana gene involved in vascular development and parenchyma pith cell primary wall retention (40). SlGRAS38 (Solyc07g052960.1), a target of RIN and also a regulator of broad ripening phenomena including time to ripening initiation, was repressed in SlLOB1 repression fruit (SI Appendix, Fig. S7A) (18), while other functionally identified ripening TFs were not substantially influenced by SlLOB1 (SI Appendix, Fig. S7B).

Validation of Cell Wall Gene Repression.

Nine cell wall–related DEGs were selected for qRT-PCR validation in T2 generation RNAi fruit, and differential expression was confirmed in each case. EXP1 displayed the most substantial down-regulation (<1.5% of WT in pericarp and <3.3% in locule) (Fig. 3E). Notably, this degree of repression is greater than that observed in antisense EXP1 fruit (10). Immunoblot analysis with an EXP1 antibody detected the predicted 25 kDa protein in protein extracts from WT but not transgenic fruit (Fig. 3F). CEL2 (7), which encodes an endo-β-1,4-glucanase, was down-regulated in all four stages of pericarp and locule to as little as 5 and 3% of WT in lines #3 and #6 locular gel, respectively. Other repressed genes were XY (alpha-xylosidase), MAN (beta-1,4-endomannase), and PL1-27 (pectate lyase), three predicted cell wall–metabolizing enzymes with potential roles in xyloglucan side chain modification, galactomannan backbone hydrolysis, and homogalacturonan breakdown, respectively. Additionally, AGP2, a cell wall glycoprotein (35); E6, which is a candidate for firmness QTL2.2 (25); GASA (Gibberellic Acid-stimulated Arabidopsis), which has been shown to influence cell expansion in A. thaliana (41); and TBL (Trichome Birefringence-Like), involved in O-acetylation of hemicelluloses and pectic polysaccharides (42), were all down-regulated in SlLOB1-repressed fruit (Fig. 3G).

Ectopic Expression of SlLOB1 Promotes Softening.

We generated transgenic SlLOB1 ectopic expression plants; and six independent tomato lines were recovered, although three (OE1, OE5, and OE12) proved to be cosuppression lines, and only OE1 was further characterized in the context of gene suppression (SI Appendix, Figs. S4D, S5, and S8). The remaining three overexpression lines (OE2, OE6, and OE13) displayed similar pleiotropic phenotypes, including increased branching, dwarfism, enlarged pedicels, and reduced internode length (measured in the first three trusses), all consistent with altered organ boundary formation, though no discernable changes in leaf architecture were noted, and smaller fruit resulted with enlarged pedicel and seeds with abnormal seeds coats (Figs. S2 and S9 A–D). SlLOB1 RNAi fruit yielded seed of normal appearance (SI Appendix, Fig. S9E) and number that underwent normal germination, consistent with the low expression of SlLOB1 in seed [https://tea.solgenomics.net (18)]. Moreover, fruit became soft and the locule liquefied prior to ripening consistent with the reduced softening of SlLOB1 RNAi repression (Fig. 4 A–C and SI Appendix, Fig. S10).

Fig. 4.

Effects of SlLOB1 overexpression fruit. (A) Premature locule liquefaction occurs in locule of SlLOB1 overexpression fruit. WT (Left) and SlLOB1 OE2 (Right). (B) Softening as defined by pericarp penetration. (C) Softening as defined by whole fruit compression (*P < 0.05, **P < 0.01). (D) SlLOB1 expression in 35S: LOB1 OE2 and WT fruit tissues. (E) Heat map of cell wall gene expression in 35S: LOB1 OE2 leaves and fruit (log10 of reads per kilobase million [RPKM]). Error bars indicate SE. (F) Dual luciferase assay. EV indicates empty vector with output defined as 1. The dotted line indicates LUC/REN = 3. Error bars indicate SE.

A sufficient number of fruit for softening analyses were only available from lines OE2 and OE6. IMG of SlLOB1 OE locule liquefied prematurely (Fig. 4A). Additionally, both whole fruit compression and pericarp penetration assays showed that OE fruit were softer that WT fruit, with measurable differences detected as early as the IMG stage (Fig. 4 B and C and SI Appendix, Fig. S10). The firmness of the OE2 pericarp was 50% of WT at BR, though WT fruit rapidly softened as they matured, such that there was no significant difference at the B7 stage between transgenic and WT fuit (Fig. 4 B and C and SI Appendix, Fig. S10). Following harvest, OE fruit showed overt signs of water loss (wrinkling) in advance of WT (SI Appendix, Fig. S11A) and substantially greater water loss: 26% weight loss from OE6 fruit after 20 d storage, compared with 11% from WT (SI Appendix, Fig. S11B).

Transcriptome analysis (Dataset S3) of OE2 tissue showed that the SlLOB1 transcript in OE2 leaves was 357-fold greater than WT, and expression in OE2 MG pericarp was 19-fold that measured in WT (Fig. 4D). SlLOB1 mRNA accumulation increased 2.4-fold in the OE2 MG locule, consistent with its high endogenous expression. Modest increases in SlLOB1 expression were also observed in ripening fruit tissues (Fig. 4D), again consistent with high endogenous gene expression. We also checked the expression of nine cell wall genes in OE2 young leaves and fruit. In leaves, EXP1, TBL, and GASA transcripts were higher than in WT, with the others showing minimal differences. In fruit tissues, transcript accumulation of all nine genes was elevated in parallel with the relative increase in SlLOB1 expression (Fig. 4E). This was confirmed by qRT-PCR in OE6 (SI Appendix, Fig. S12). PL was induced to 5.5-fold in OE2 MG pericarp (SI Appendix, Fig. S6B). These results demonstrate that SlLOB1 is sufficient to induce expression in leaves for some genes, but for others, additional factors may be limiting. Finally, cell wall antibody probes were used to assess changes in cell wall polysaccharides and revealed reduced xyloglucan and homogalacturonan levels in BR stage SlLOB1 overexpression pericarp as compared to WT (SI Appendix, Fig. S13).

SlLOB1 Activates EXP1, CEL2, XY, AGP2, TBL, E6, and PL Promoters In Vitro.

To better place SlLOB1 function in the context of other LOB genes, a phylogenetic analysis of LOB1 orthologs was performed (SI Appendix, Fig. S14). LOB1 and LOB11 family members are close homologs in many species. A. thaliana AtLBD1 and AtLBD11 have 77% amino acid identity, while tomato SlLOB1 and SlLOB11 share 82% identity. SlLOB11 is mainly expressed in young roots and only minimally in fruit (SI Appendix, Fig. S14). AtLBD1 is functionally undefined, though citrus (Citrus sinensis) and poplar (Populus tremula × Populus alba) orthologs, CsLOB1 and PtaLBD1, participate in citrus bacterial canker susceptibility and secondary wood formation, respectively (43, 44).

SlLOB1 contains conserved domains, consistent with DNA binding and protein–protein interactions (SI Appendix, Fig. S14). LOB family members can bind the promoter and activate EXP gene expression in A. thaliana and banana (28, 29, 31). To better understand SlLOB1 function, we carried out promoter transactivation using transient expression in Nicotiana benthamiana. This revealed that SlLOB1 has activator activity on the EXP1, CEL2, XY, AGP2, TBL, E6, and PL promoters as defined by greater than threefold activity compared to controls (Fig. 4F). EXP1 transactivation was especially strong (67-fold induction). No SlLOB1 activation was observed for the MAN, PL1-27, and GASA promoters (Fig. 4F).

The abundance of EXP1 mRNA in response to ectopic SlLOB1 transgene expression and stronger promoter induction in the transactivation assay and substantial reduction of transcript and protein levels in SlLOB1 repression lines was particularly notable. To identify cis elements in the EXP1 promoter that bind SlLOB1, we generated an EXP1 promoter deletion series. These deletions, designated P1 to P5, were used to develop luciferase reporter constructs (SI Appendix, Fig. S15A). Promoter activity decreased to 72% of the full-length promoter activity in the first deletion (P1_−1,393 base pair [bp] to P2_−1,094 bp), while subsequent deletions (comparing P2 to P4 and P4 to P5) had little effect. With deletion P5, reporter activity decreased to only eightfold induction, suggesting at least two regulatory loci at −1,393∼ −1,094 and −389∼0. P3 (−1,094∼ −568) were not activated by SlLOB1 as predicted. Using the cis element prediction websites, PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) and Softberry (http://www.softberry.com/), we identified two putative cis elements: I (ACCTCAAAT) and II (ATTTTCTTCA), based on their presence in both regions (SI Appendix, Fig. S15A). We next generated deletion P6, missing predicted cis element I but carrying II, as well as deletion P7, with I but without element II. P7 showed only threefold activation, while P6 had no effect, indicating that SlLOB1 is functional without element I (SI Appendix, Fig. S15A).

Silencing SlLOB1 Altered Carotenoids but Not Additional Fruit-Ripening Phenotypes.

Suppression of SlLOB1 expression had profound effects on the cell wall–associated transcriptome, softening, fruit water loss, and shelf life but did not affect ripening initiation as measured in time to the BR stage following anthesis (SI Appendix, Fig. S16A). Ethylene production by the transgenic fruit was not different from the WT fruit in terms of time of initiation or stage of maximal production but remained higher during later fruit development when SlLOB1 was repressed (SI Appendix, Fig. S16A). The characterization of ACS and ACO ethylene synthesis genes in ripening fruit of WT and SlLOB1-repressed lines (SI Appendix, Fig. S16B) indicated fluctuation in several family members but nothing that could be readily interpreted as the molecular basis for elevated ethylene in later stage RNAi fruit. The repression of SlLOB1 resulted in noticeably darker red fruit, especially in the locular gel (SI Appendix, Fig. S16C). To better understand the basis of this color change, we characterized carotenoid profiles in both locule and pericarp tissue from B7 and B15 stage fruit by high-pressure liquid chromatography (HPLC) analysis. In the gel, lycopene and beta-carotene levels were approximately twice those in WT at both stages, while in pericarp, lycopene abundance was ∼60 and 40% higher in B7 and B15 fruit, respectively. No differences were noted in beta-carotene levels (SI Appendix, Fig. S16D).

Discussion

SlLOB1 Regulates Multiple Fruit Cell Wall–Associated Genes and Softening.

In addition to altering expression of genes encoding cell wall proteins, orthologs of A. thaliana cell wall–related TFs (e.g., Solyc06g0359400.2 and Solyc03g1209100.2) responded to SlLOB1 repression in fruit. Ectopic expression of SlLOB1 activated the same genes prematurely in green fruit, leading to precocious textural changes and softening. Together, these genes are associated with the modification of all three major cell wall polysaccharide classes (cellulose, pectins, and hemicellulose) in addition to cell wall glycoproteins. The corresponding synergistic activities, mediated through a single regulator, manifest in more substantive textural changes than previously observed when targeting a single cell wall–associated gene. It is noteworthy that the gene shown to exert the largest effect on softening to date, PL (11), is positively regulated by SlLOB1, and its promoter can be activated by SlLOB1 in vitro.

EXP1 Is a Direct Target of SlLOB1, and Additional EXP Genes Are Influenced by SlLOB1 Repression.

EXP1 is the most altered DEG by SlLOB1 manipulation. Expansins are widely studied in fleshy fruits (9, 10) with induction paralleling softening in all cases. It has been proposed that EXP disrupts noncovalent interactions between cellulose microfibrils and matrix polysaccharides, facilitating access to cell wall–modifying enzymes, and thus may be especially important to softening in the context of coregulated cell wall catalytic activities (45).

Similar to tomato SlLOB1 and EXP1, A. thaliana AtLBD18 binds to the AtEXP14 promoter, though in this case, it promotes lateral root emergence (29). CsLOB1 was observed to induce expansin expression following transient overexpression in sweet orange leaves (43), and banana MaLBD1/2/3 was reported to induce expansin promoter activity in tobacco BY-2 protoplasts (31). It is noteworthy that in addition to LeEXP1 (Fig. 3E), two additional EXP homologs (Solyc01g090810 and Solyc08g077910) were down-regulated in SlLOB1 repression lines (SI Appendix, Fig. S6A).

Gene Expression and Phenotypes Suggests SlLOB1 Acts Downstream of More Global Ripening Regulators.

The Cnr epiallele results in impaired ripening, reduced ethylene synthesis, extensive cell wall modification, and softening inhibition (13), although the degree of ripening inhibition has recently been questioned (46). SlLOB1 RNAi-repressed fruit are characterized by Cnr-like texture in that they are firmer and have less locule solubilization, reduced cell wall swelling, and decreased fruit density (SI Appendix, Fig. S4). SlLOB1 expression is reduced over fourfold in the Cnr mutant (Cnr 42DPA/WT 42DPA = 0.223, P value = 0.01), and pericarp DEGs of suppression line 6 (B7 pericarp) compared to published Cnr data [BR pericarp (20)] indicated 99 common down-regulated and 49 up-regulated genes. A total of 11 of the common down-regulated genes are cell wall associated (Dataset S4). The SlLOB1 promoter contains the GTAC binding motif of SPB proteins (47). However, the SlLOB1 promoter was not activated by CNR in a transient expression system (SI Appendix, Fig. S17). The similar firmness, locule liquefaction, fruit density, and cell wall swelling phenotypes of Cnr and SlLOB1 repression fruit suggest consequences of some, or all, of the common cell wall–associated DEGs operating together.

The results here demonstrate that SlLOB1 effects are targeted primarily to the fruit locule and pericarp cell wall and textural changes and, as such, affect ripening downstream of more global ripening regulators such as RIN. Indeed, SlLOB1 is repressed in both rin and nor mutants (20, 46). TFs encoded by Solyc04g081190_bZIP and Solyc06g035940.2_HD-Zip genes are also induced by SlLOB1 (SI Appendix, Figs. S7A and S17), indicating SlLOB1 operates in part via additional downstream TFs. Such genes might be responsible for altered expression of genes differentially expressed in response to SlLOB1 repression but whose promoters do not interact with SlLOB1. It is noteworthy that LOB genes have an intermediate placement in regulatory networks defined in other species. For example, in A. thaliana, LOBs can regulate AP2, WOX, or E2Fa positively and KNOX negatively, while BZR1, ARF, and NAC consensus binding sites are located upstream of different LOBs (28). Together, these interactions suggest that SlLOB1 operates downstream of more comprehensive ripening regulators (e.g., RIN and NOR) but upstream of other regulators, such as bZIP and the HD-Zip genes noted. Furthermore, members of several TF families, including MYB, bHLH, and LOB itself, are candidates for direct SlLOB1 interaction partners, as members of these families are known from studies of A. thaliana to form dimer or trimers with LOB proteins (28, 48). In these cases, they integrate regulatory connections between primary transcriptional regulators and downstream outputs (SI Appendix, Fig. S15B). Finally, a single tomato TF, SlMBP3, a member of the AGAMOUS subfamily of tomato MADS-box genes, was recently shown to be necessary for tomato seed development and locule liquefaction (49). SlMBP3 expression is predominant in the seed at anthesis and early development, strongly expressed in locular tissue postanthesis and through fruit development with much lower expression in carpel tissues [https://tea.solgenomics.net (18)]. The repression of SlMBP3 resulted in similar inhibition of locule liquefaction as with SlLOB1 repression reported here (Fig. 1D) but with additional phenotypes of altered seed coat development, reduced seed viability, and substantially reduced fruit size (49). Together, these results suggest SlLOB1 is more specific to locule liquefaction, while SlMBP3 has broader pleiotropic effects on fruit development. SlLOB1 presents a genetic target to more precisely modify locule liquefaction and texture absent effects on seed viability and fruit size.

SlLOB1 Influences Ethylene Production in Later Ripening in Addition to Carotenoid Profiles.

The repression of SlLOB1 had no effect on ethylene production during early ripening when fruit ethylene is at its highest, and the cascade of changes summing to render the ripe phenotype are initiated. As the fruit continues ripening, ethylene decreases, and this decrease was attenuated in SlLOB1 repression fruit (SI Appendix, Fig. S16A). Fruit overexpressing SlLOB1 began softening well before their WT counterparts prior to the induction of ripening ethylene (Fig. 4B and SI Appendix, Figs. S10 and S18). While prior data indicate a clear necessity for ethylene in tomato fruit softening (21, 22), the data presented here indicate that SlLOB1 is a more immediate effector of mature fruit textural changes through the activation of multiple cell wall–associated genes.

Unlike ethylene synthesis–inhibited tomato fruit or the ripening-repressed Cnr and rin mutant fruit, SlLOB1 RNAi fruit accumulate carotenoid pigments, although ultimately, they accumulate higher levels of lycopene and β-carotene than nontransgenic controls. Expression of carotenoid enzymatic genes was somewhat higher in both the pericarp and locule of the SlLOB1-repressed fruit, particularly for the rate-limiting step of phytoene synthase conferred by the PSY1 gene, consistent with enhanced carotenoid accumulation in both the pericarp and locule tissues (SI Appendix, Fig. S19). The fact that PSY1 is ethylene inducible (1) and most elevated in later stage fruit (SI Appendix, Fig. S19) suggests that the elevated carotenoid phenotype may be related to the elevated ethylene (SI Appendix, Fig. S16A). SlLOB1 did not interact directly with the PSY1 or PDS promoters in an in vitro activation assay (SI Appendix, Fig. S17). SlLOB1 ectopic expression resulted in reduced pericarp carotenoid accumulation but enhanced lycopene accumulation in locular gel, which is consistent in all lines (SI Appendix, Fig. S20 A and B). Additionally, five heat shock proteins (HSPs) were strongly induced in both SlLOB1 RNAi lines (SI Appendix, Fig. S21), including HSP21, which has previously been associated with lycopene accumulation (50).

In conclusion, our results demonstrate that SlLOB1 functions as a positive transcriptional regulator of fruit softening through its activities in both the locule and pericarp. The repression of SlLOB1 inhibits fruit softening via reduced expression of multiple cell wall–associated genes (in both locule and pericarp tissues), a number of which have promoter sequences capable of SlLOB1 interaction and which are activated when SlLOB1 is expressed ectopically in tomato leaves. SlLOB1 overexpression plants had contrasting phenotypes and gene expression, further supporting the role of this TF in directing fruit textural changes. While SlLOB1 ectopic expression plants displayed numerous nonfruit phenotypes, the limitation of fruit phenotypes in RNAi repression lines is consistent with SlLOB1 expression primarily in this tissue, suggesting a primary role in fruit ripening. In addition to cell wall phenotypes, SlLOB1 repression enhances carotenoid accumulation, possibly via ethylene and/or HSP stabilization of carotenoid pathway enzymes. These results raise the possibility of enhancing both texture and nutritional quality via targeted repression or selection of low-expression SlLOB1 alleles during breeding.

Materials and Methods

Plant Material.

Tomato plants were greenhouse grown under a 16-h light (26 to 29 °C) 8-h dark (17 to 20 °C) cycle. Fruit were tagged at 1-cm diameter (8 to 9 d post-anthesis [DPA]), and IMG (20 DPA), MG (35 DPA), BR (39 DPA), B7 (red ripe), and B15 (overripe) were harvested. Pericarp and locular gel were separated and frozen in liquid nitrogen and stored at −80 °C. Seeds were extracted from RR fruit.

Metabolite and Molecular Analysis.

Details of metabolite (ethylene and texture measurements, carotenoid extraction and quantification, cuticle staining, cell wall immunohistochemistry, cell wall material extraction, and wall swelling analysis) and molecular (DNA constructs and tomato transformation, RNA-seq library construction and qRT-PCR, Illumina read processing and GO enrichment analysis, protein gel-blot analysis, and dual luciference assasy) analyses are provided in SI Appendix. All primers used in this work are listed in SI Appendix, Table S3.

Data Availability

All study data are included in the article and/or supporting information.