The tomato SlSHINE3 transcription factor regulates fruit cuticle formation and epidermal patterning


Author for correspondence:

Asaph Aharoni

Tel: +972 8 934 3643



  • Fleshy tomato fruit typically lacks stomata; therefore, a proper cuticle is particularly vital for fruit development and interaction with the surroundings. Here, we characterized the tomato SlSHINE3 (SlSHN3) transcription factor to extend our limited knowledge regarding the regulation of cuticle formation in fleshy fruits.
  • We created SlSHN3 overexpressing and silenced plants, and used them for detailed analysis of cuticular lipid compositions, phenotypic characterization, and the study on the mode of SlSHN3 action.
  • Heterologous expression of SlSHN3 in Arabidopsis phenocopied overexpression of the Arabidopsis SHNs. Silencing of SlSHN3 results in profound morphological alterations of the fruit epidermis and significant reduction in cuticular lipids. We demonstrated that SlSHN3 activity is mediated by control of genes associated with cutin metabolism and epidermal cell patterning. As with SlSHN3 RNAi lines, mutation in the SlSHN3 target gene, SlCYP86A69, resulted in severe cutin deficiency and altered fruit surface architecture. In vitro activity assays demonstrated that SlCYP86A69 possesses NADPH-dependent ω-hydroxylation activity, particularly of C18:1 fatty acid to the 18-hydroxyoleic acid cutin monomer.
  • This study provided insights into transcriptional mechanisms mediating fleshy fruit cuticle formation and highlighted the link between cutin metabolism and the process of fruit epidermal cell patterning.


The epidermis of all aerial plant organs is covered with an extracellular layer termed the cuticle, which is largely composed of cutin embedded in and overlaid with waxes (Kolattukudy, 2001). The organic soluble waxes mainly include aliphatic compounds derived from very long chain fatty acids (Jetter et al., 2006), whereas the insoluble cutin matrix comprises interesterified hydroxy fatty acids and derivatives (Kolattukudy, 2001). Biosynthesis of plant cuticle components in the epidermis and their secretion to the extracellular matrix involve the coordinated induction of several metabolic pathways, in which transcription factors (TFs) may play key roles (Broun, 2004; Suh et al., 2005).

The first identified cuticle regulator was SHN1/WIN1, one of the three members of an Arabidopsis APETELA2 (AP2)-domain TF superfamily (Aharoni et al., 2004; Broun et al., 2004). Overexpression of AtSHN1/WIN1 promotes cutin biosynthesis, which is preceded by coordinated rapid induction of several cutin biosynthetic genes, while its down-regulation has the opposite effect (Kannangara et al., 2007). Recently, we reported that the activity of AtSHNs extends beyond the regulation of cuticle metabolic pathway and that they take part in the genetic program that mediates floral organ development and function (Shi et al., 2011). By cosilencing all three SHN clade members, we demonstrated that AtSHNs act redundantly in patterning reproductive organ surface, modulating processes associated with cell elongation, adhesion, and separation. In barley, mutation in the SHN1/WIN1 ortholog (Nud) turned the typically hulled caryopses to naked ones (Taketa et al., 2008). When AtSHN2 is overexpressed in rice, it coordinately regulates the biosynthesis of cellulose and lignin, and enhances cellulose deposition while decreasing lignin deposition (Ambavaram et al., 2011). Furthermore, two Arabidopsis MYB TFs, AtMYB41 and AtMYB96, were recently reported to control cuticle deposition under abiotic stress (Cominelli et al., 2008; Park et al., 2011). In addition, homeodomain-leucine zipper IV (HD-ZIP IV) family TFs were found to be additional regulators of cuticle metabolism, as revealed by the investigation of OUTER CELL LAYER1 (OCL1) in maize (Javelle et al., 2010; Depege-Fargeix et al., 2011) and CUTIN DEFICIENT2 (CD2) in tomato (Isaacson et al., 2009; Nadakuduti et al., 2012).

In fruit crops, the cuticle provides structural support for the integrity of the whole fruit and appears to influence both growth and ripening (Saladié et al., 2007). The mechanical and rheological properties of the cuticle are of considerable economic significance, and the variation in fruit cuticle composition may underlie differences in quality traits such as resistance to desiccation, microbial infection and cracking (Matas et al., 2004; Isaacson et al., 2009). Despite the critical roles of the cuticle in fruit development, ripening and postharvest behavior, molecular tools have only recently been applied to examine the genetic bases of its assembly and functional roles in fruit development (Mintz-Oron et al., 2008; Yeats et al., 2010; Matas et al., 2011). Information regarding genes encoding cuticle-related enzymes is accumulating in tomato as well as other fruit species (Vogg et al., 2004; Hovav et al., 2007; Leide et al., 2007, 2011; Saladié et al., 2007; Wang et al., 2011; Girard et al., 2012; Yeats et al., 2012). A set of proteins associated with lipid metabolism have also been identified at the tomato surface (Yeats et al., 2010; Catalá et al., 2011). Recent work in tomato pointed out that TFs play crucial roles in regulating cuticle biosynthetic pathways in fleshy fruit (Adato et al., 2009; Isaacson et al., 2009; Ballester et al., 2010; Nadakuduti et al., 2012). SlMYB12 appears to control the pathway leading to the biosynthesis of the flavonoid naringenin chalcone that is embedded, to a large extent, in the tomato fruit cuticle, providing the typical yellow pigmentation. The CD2 protein contains an HD DNA binding motif and a START domain reported in nonplant systems to bind regulatory lipids.

In this study we identified three tomato orthologs of the AtSHN clade and focused on characterizing the SlSHN3 member, because its transcript was highly enriched in the exocarp of MG fruit, which is similar to other cuticle-related genes' expression (Mintz-Oron et al., 2008). Transgenic Arabidopsis plants overexpressing SlSHN3 displayed a similar phenotype to those obtained by overexpressing the Arabidopsis gene orthologs. Silencing SlSHN3 in tomato resulted in striking alterations in cuticle-associated phenotypes in the epidermis. Transcriptome analysis revealed that besides the regulation of genes associated with cutin metabolism, SlSHN3 possibly controls the expression of regulatory genes associated with epidermal cell patterning. One of the SlSHN3 downstream direct targets is SlCYP86A69, a putative homolog of a protein previously associated with Arabidopsis cutin biosynthesis. Through the identification and characterization of the slcyp86a69 mutant, we demonstrated that mutation in SlCYP86A69 resulted in dramatic changes to the cuticular layer. Moreover, cutin deficiency in slcyp86a69 significantly affected the fruit response to abiotic and biotic stress in a way similar to the SlSHN3 silenced plants. Thus, our study revealed a SlSHN3 regulated transcriptional network that acts in epidermal patterning and fruit cuticle formation and links cutin metabolism with the more global program of epidermal cell patterning and organ formation in tomato fruit.

Materials and Methods

Plant materials and transformation

Seeds from the slcyp86a69 mutant (Solanum lycopersicum L. cv MicroTom, MT) were obtained from the EMS mutant population of C. Rothan (Okabe et al., 2011). Seeds of the cd3 mutant (S. lycopersicum L. cv M82) were obtained from the Genes that Make Tomatoes germplasm collection (Isaacson et al., 2009). The heteroexpression of SlSHN3 was on the Arabidopsis (Arabidopsis thaliana, L.) Col-0 background, while the RNAi silencing of SlSHN3 was in the MT background. All tomato fruits were harvested at indicated stages. F2 mapping populations were grown in a local field at Rehovot, Israel.

For generation of the 35S:SlSHN3 construct, a 603 bp SlSHN3 cDNA fragment was amplified, subcloned into the pFLAP100 vector and then cloned into the pBIN PLUS binary vector. The transformation to Arabidopsis was performed as described previously (Clough & Bent, 1998). For generation of the 35S:SlSHN3 RNAi construct, a 255 bp cDNA fragment specifically targeting SlSHN3 was amplified and integrated into pENTR D TOPO (Invitrogen). The LR Clonase (Invitrogen) was used to recombine this fragment into the pK7GWIWG2(II) binary vector (Karimi et al., 2002). Cotyledon transformation in cv MT tomato was performed as previously described (Dan et al., 2006). The primers used in construct creation in this study are listed in Supporting Information, Table S2.

Gene expression analysis

Gene expression analysis was carried out as previously described (Mintz-Oron et al., 2008). Total RNA was extracted with the Trizol Reagent (Invitrogen) from manually dissected mesocarp and exocarp tissues pooled from five to six fruits. The cDNA was synthesized by AMV Reverse Transcriptase (EurX Ltd., Gdansk, Poland). Quantitative real-time PCR (qRT-PCR) analysis using gene-specific oligonucleotides was performed on an ABI 7300 instrument (Applied Biosystems, Norwalk, CT, USA) with the PlatinumR SYBR SuperMix (Invitrogen) in three biological replicates. Sequences of gene-specific oligonucleotides are provided in Table S2.

Dual luciferase transient assay

The dual luciferase transient assay was performed as previously described (Shi et al., 2011), which was initially modified from Hellens et al. (2005).

Histological staining and electron microscopy observation

Microtom sections of fixed and embedded fruit peels were used for lipid staining with Sudan IV (Isaacson et al., 2009) and Auramine O (Buda et al., 2009). For scanning electron microscopy (SEM) and transmission electron microscopy (TEM), tissues were collected and processed using standard protocols (Chuartzman et al., 2008), and performed using an XL30 ESEM FEG microscope (FEI) at 5–10 k, and a Technai T12 transmission electron microscope (FEI), respectively.

Cuticular lipid analyses

Fruit exocarp discs were prepared as previously described (Hovav et al., 2007). Cuticular waxes were extracted by immersing the isolated exocarp discs twice, each in 5 ml of CHCl3 at room temperature for 1 min. The collected solution was then spiked with 5 μg tetracosane (Fluka, Ronkonkoma, NY, USA) as an internal standard and analyzed as previously described (Kurdyukov et al., 2006a,b). As for cutin measurement, exocarp discs after the wax extraction were exhaustedly delipidated with methanol : chloroform (1 : 1, v/v) mixture, air-dried and analyzed as previously described (Franke et al., 2005).

Enzyme activity assay

A yeast expression system specifically developed for the expression of P450 enzymes was utilized to induce the expression of SlCYP86A69 and extract microsomes (Pompon et al., 1996). Enzymatic activities of SlCYP86A69 from transformed yeast microsomes were determined by following the formation rate of metabolites. The standard assay (0.1 ml) contained 20 mM sodium phosphate (pH 7.4), 1 mM NADPH, and substrate (100 μM). The reaction was initiated by the addition of NADPH and was stopped after 30 min. Metabolites generated were methylated with diazomethane and trimethylsilylated with N,O-bistrimethylsilyltrifluoroacetamide containing 1% (v/v) trimethylchlorosilane (1 : 1, v/v), and then subjected to GC-MS analysis.

Responses of harvested tomato fruit to fungal infection and dehydration

Inoculation with the fungus Colletotrichum coccodes isolate 138 on freshly harvested breaker tomato fruits from wildtype, slcyp86a69, and 35S:SlSHN RNAi plants was carried out as previously reported (Alkan et al., 2008). For water loss measurement, a total of 30–40 fruits of each line were picked at the red stage, and were stored at room temperature for c. 1 month. Fruit weight was recorded periodically, and water loss was calculated as a percentage of weight loss.


The tomato SHINE orthologs exhibit fruit epidermis-associated expression

A putative tomato SHN ortholog that displayed epidermal-enriched transcript abundances at the early stages of fruit development was identified in a previous study (Mintz-Oron et al., 2008). This tomato SHN ortholog appeared to be closest to AtSHN3 (At5g25390; Aharoni et al., 2004), with 59% identity and 75% similarity, respectively, at the amino acid level. Two additional tomato SHN-like genes, SlSHN1 and SlSHN2, were subsequently identified by in silico analysis of the tomato genome sequence (Fig. 1a). Similar to their Arabidopsis (Aharoni et al., 2004) and barley orthologs (Taketa et al., 2008), the three tomato SHNs share two conserved domains (‘mm’ and ‘cm’ domains) outside the AP2 domain (Fig. 1b). As in Arabidopsis, SlSHN3 and SlSHN1 contain a single intron 82 bp from the start codons (Aharoni et al., 2004); however, SlSHN2 has an additional intron 897 bp from its start codon.

Figure 1.

SlSHN3 is the tomato (Solanum lycopersicum, Sl) ortholog of Arabidopsis (Arabidopsis thaliana, At) AtSHN3. (a) Phylogenetic analysis of the SHINE (SHN) proteins from Arabidopsis (At), barley (Hordeum vulgare, Hv), and tomato (Sl) species. (b) Sequence alignment of the characteristic conserved middle motif (termed ‘mm’) and C-terminal motif (termed ‘cm’) of SHN proteins from Arabidopsis (At), barley (Hv), and tomato (Sl) plants. (c) Predominant expression of SlSHN3 in the exocarp tissues of immature green fruit examined by real-time PCR (qRT-PCR). IG, immature green; MG, mature green; Br, breaker; O, orange; R, red. Values represent mean ± standard error of the mean (n = 3). (d) Heterologous expression of SlSHN3 in Arabidopsis displayed typical shiny and curled rosette leaves (right) as reported in plants overexpressing AtSHNs.

The expression of SlSHN3 was consequently evaluated in 21 different tomato plant tissues and its highest expression level was detected in immature green (IG) fruit exocarp (Fig. 1c). In fruit tissues, SlSHN2 expression pattern was similar to that of SlSHN3, while SlSHN1 exhibited a more constitutive exocarp-associated expression pattern during development (Fig. S1a). Normalized digital expression based on deep transcriptome sequencing of samples from laser-captured fruit pericarp cell/tissue types (Matas et al., 2011) also showed that SlSHN3 exhibits the highest expression level of the three tomato SHNs at the IG stage of fruit development (Fig. S1b).

To further examine the functional similarity between the tomato SlSHN3 and the Arabidopsis SHNs, it was overexpressed in Arabidopsis. The transgenic plants (35S:SlSHN3) displayed characteristic phenotypes similar to those observed previously in overexpressers of either of the three Arabidopsis SHNs (Aharoni et al., 2004), including brilliant and shiny green rosette leaves, leaf curling (Fig. 1d), and the accumulation of leaf waxes and cutin (Fig. S1c). These results provided the first evidence that SlSHN3 is an ortholog of AtSHN3 and that it likely plays an important role in patterning the tomato fruit epidermis.

Silencing of SlSHIN3 results in profound effects on the tomato fruit epidermis

To provide direct evidence regarding the role of SlSHN3 in tomato fruit surface, we generated 35S:SlSHN3-RNAi (RNAi) tomato lines that specifically target the SlSHN3 gene. Fruit of the RNAi lines exhibited several altered phenotypes, including glossier surface (Fig. S2a), and much less enzymatically isolated cuticle (Fig. S2b). TEM examination of the mature green (MG) fruit epidermis confirmed that the cuticle layer from RNAi fruit was much thinner than that of WT (Fig. 2a,b). Typically, the WT fruit cuticle contains both external and internal cuticular layers (ecl and icl, respectively; Buda et al., 2009); by contrast, the RNAi fruit cuticle contained only a thinner icl layer (Fig. 2a,b). Notably, epidermal cells in the RNAi fruit exhibited profound morphological alterations, displaying a thicker and more sharply defined primary cell wall as compared with the WT (Fig. 2a,b).

Figure 2.

Cuticular abnormalities observed in 35S: SlSHN3 RNAi tomato (Solanum lycopersicum) fruit. (a, b) Transmission electron microscopy (TEM) images showing the difference in the structure of the cuticle layer and the primary cell wall in mature green (MG) fruit epidermis between wildtype (WT) (a) and 35S:SlSHN3 RNAi (b). ecl, external cuticular layer; icl, internal cuticular layer; pcw, primary cell wall. (c, d) Auramine O (c) and Sudan IV (d) staining demonstrating the difference in cuticular lipid deposition in MG fruit epidermis between WT (upper panel) and 35S:SlSHN3 RNAi (bottom panel). ap, anticlinal peg; c, cuticle layer; ec, epidermal cell; sd, subepidermal deposition.

Staining fruit surfaces with histological lipid stains, either Sudan IV or Auramine O, also revealed a substantial reduction in cuticle deposition, both at the surface and between the epidermal cells, in the RNAi MG fruit epidermis (Fig. 2c,d). While the cuticle of WT fruit displayed a typical pattern of highly cuticularized anticlinal pegs (ap) and extensive deposition of cuticle material below the epidermal cell layer (subepidermal deposition, sd), that of the RNAi fruit displayed discontinuous ap and much less sd staining (Fig. 2c,d). Altogether, these findings suggested that SlSHN3 strongly affects the patterning of the fruit cuticle and the morphology of the epidermal cells.

Silencing of SlSHIN3 reduces the amounts of cutin and waxes in the tomato fruit cuticle

Gas chromatography-mass spectrometry analysis was carried out to characterize the chemical composition of the enzymatically isolated cuticle from MG fruit of the RNAi lines. Overall, total cutin monomer abundance in the MG fruit cuticle in the RNAi line was reduced to c. 40% of wildtype values (Table 1). The significant reductions in the amounts of aromatics, dicarboxylic fatty acids, midchain and terminal-hydroxylated fatty acids, and 2-hydroxylated fatty acids contributed to the remarkable reduction of cutin monomers in the SlSHN3 RNAi line.

Table 1. Cutin monomers identified in enzymatically isolated cuticles derived from 35S:SlSHN3 RNAi tomato (Solanum lycopersicum) and its corresponding wildtype (WT, S. lycopersicum. cv MicroTom) fruit epidermis (mature green stage)
Cutin monomersμg cm−2
  1. Significantly altered monomers are colored gray (Student's t-test, < 0.05, n = 4).

 cis-coumaric acid0.1930.0470.2270.064
 Benzoic acid5.7051.72314.4623.065
 trans-coumaric acid66.54819.664154.82322.772
 Coumaric acid derivative0.2220.0460.3010.085
Saturated fatty acids (FAs)
 C16:0 FA1.0940.2481.9870.368
 C18:0 FA0.2210.0460.2900.021
 C22:0 FA0.0590.0100.1160.037
Dicarboxylic fatty acids (DFAs)
 C16:0 DFA7.5922.53923.0132.724
Midchain hydroxylated fatty acids (HFAs)
 C16-9/10,16-Di HFA364.922136.645829.41430.218
 C16-9/10-H DFA85.39732.358220.0134.414
 C18-9,10,18-tri HFA0.0190.0080.0220.002
Terminal-hydroxylated fatty acids (ω-HFAs)
2-Hydroxylated fatty acids (2HFAs)
Unidentified monomers (UIs)

The amount of total cuticular waxes (Table S1) was also significantly reduced in the enzymatically isolated MG fruit cuticles of the RNAi line, including amounts of most free fatty acids, β-amyrin and two sterols. But, although not significant, concentrations of some alkanes increased in the RNAi line. Thus, the metabolic analyses clearly indicated that reduced SlSHN3 expression affects tomato fruit cuticle composition.

Silencing of SlSHIN3 results in down-regulated expression of an array of cutin metabolism and epidermal patterning regulatory genes

To better understand the molecular mechanisms by which SlSHN3 regulates fruit cuticle assembly, we examined the expression of 25 putative tomato orthologs of known Arabidopsis cuticle-related and epidermis-associated genes in fruit epidermis of the SlSHN3 silenced line. This set of genes was selected from a reported epidermis-enriched transcript dataset (Mintz-Oron et al., 2008) and by in silico sequence alignment and gene prediction (Table S2). The selection of those genes was also based on a recent study in Arabidopsis where AtSHNs' putative target genes were identified (Shi et al., 2011). The expression of 14 out of the tested 25 genes was significantly down-regulated in MG fruit epidermis of the RNAi plants (Fig. 3a). These included genes coding for three cytochrome P450s (SlCYP77A1, SlCYP86A68 and SlCYP86A69), two GSDL-motif lipases (SlGDSLa and SlGDSLb/SlGDSL1, Girard et al., 2012), two putative acyltransferases (SlDCR and SlGPAT6), one long chain acyl-CoA synthetase (SlLACS2), and an oxidoreductase (SlHTH).

Figure 3.

Identification and characterization of putative SlSHN3 target genes. (a) Quantitative real-time PCR analyses of cutin- and epidermis-related transcripts in wildtype (WT) and 35S:SHN3 RNAi tomato (Solanum lycopersicum) exocarp (mature green stage). Values represent mean ± standard error of the mean (n = 3; *, < 0.05; **, < 0.01, Student's t-test). (b) Examining SlSHN3 activation of putative target gene promoter regions by means of a dual luciferase (LUC) transient assay. Values represent means ± standard error of the mean (n = 4; *, < 0.05; **, < 0.01, Student's t-test). Gene identifiers and oligonucleotides are listed in Table S2. LUC, activity of firefly luciferase; REN, activity of renila luciferase.

The expression of genes encoding five putative TFs was also significantly reduced in MG fruit epidermis of the RNAi plants (Fig. 3a). These included SlSHN2, SlMIXTA, and three HD-ZIP IV genes (SlGL2, SlHDG11a and SlANL2c). The recently reported cuticle-associated TF CD2 belongs to the same protein family (Isaacson et al., 2009; Nadakuduti et al., 2012); nevertheless, the expression of its coding gene SlANL2a was not altered in the SlSHN3 silenced lines. We note here that these down-regulated TFs from other species were all associated with epidermal cell patterning (Ariel et al., 2007; Javelle et al., 2011).

SlSHN3 activates the promoters of several putative downstream target genes

We used a dual luciferase (LUC)-based transient expression assay (Shi et al., 2011) to evaluate whether SlSHN3 can directly activate the promoters of five putative downstream target genes that were significantly down-regulated in the SlSHN3 silenced lines. Co-infiltration of 35S:S1SHN3 and promoter–LUC constructs resulted in significant transactivation of the promoters of four of the five tested putative S1SHN3 targets (Fig. 3b). These activated promoters included those of two TFs (SlMIXTA and SlGL2a) and two cytochrome P450s (SlCYP86A68 and SlCYP86A69). The results suggested that SlSHN3 affects fruit cuticle formation and that epidermis patterning may be, in part, a result of direct action on gene expression of cutin biosynthetic genes and on regulatory factors related to epidermal cell patterning.

Expression of SlSHN3 and its putative downstream target genes in the fruit epidermis during development

To further understand the association of SlSHN3 with its putative downstream gene targets, their expression patterns were examined in fruit exocarp and mesocarp tissues in five stages of fruit development. We found that all three CYP86A members (SlCYP86A33, SlCYP86A68, and SlCYP86A69) and a CYP77A member (SlCYP77A1) exhibited higher expression in the exocarp tissue than in the mesocarp during fruit development (Fig. S3a). A similar expression pattern was observed for four of the five TFs down-regulated in the RNAi fruit exocarps (SlSHN2, S1MIXTA, SlHDG11a and SlANL2c; Figs S1a, 3b). SlGL2 displayed significant epidermis-associated expression only at late stages of fruit development (Fig. S3b).

Normalized digital expression based on deep transcriptome sequencing of samples from laser-captured fruit pericarp cell/tissue types was used to determine the spatial expression of these genes. At the IG stage, SlSHN3 and most of its putative target genes exhibited the highest expression levels in the epidermis. Of these, SlSHN2, SlMIXTA, SlGDSLb/SlGDSL1, SlCYP86A69, SlHTH, SlDCR, and SlLGPAT6 were exclusively expressed in libraries obtained from the inner and outer epidermal cell layers (Figs S1b, S4). It is noteworthy that SlGPAT6 expression was higher in the inner epidermis than in the outer epidermal cells, while SlGL2 and SlGDSLa, which showed identical expression patterns, were not expressed in the fruit outer epidermis. Taken together, the qRT-PCR and digital transcriptome analysis indicated that both SlSHN3 and its putative targets are associated with the fruit epidermis (outer and inner epidermal layers) at the early stages of tomato fruit development.

Identification of two mutant alleles corresponding to the putative SlSHN3 target gene, SlCYP86A69

In the framework of studying cuticle formation in tomato fruit, we screened an ethyl methyl sulfonate (EMS) mutant population (cv MT) and identified a new mutant that displayed altered surface phenotype. Fruit of this mutant had a pink shiny appearance (Fig. S2c). We tested whether this mutant might correspond to one of the three cuticle-deficient loci (cd1-cd3) reported by Isaacson et al. (2009). Linkage analysis of F2 plants of a cross between the EMS mutant and WT (cv M82) revealed that the surface phenotype cosegregated with single nucleotide polymorphism (SNP) markers mapped to the cd3 locus on chromosome 8 (Isaacson et al., 2009). Segregation of two SNPs tested in our analysis narrowed the linkage interval from over 1.5 million bp to < 300 kb, a region that harbors 33 predicted genes. One of these 33 genes was the putative SlSHN3 target, SlCYP86A69, which is predicted to encode a protein with 86% amino acid sequence similarity to the Arabidopsis CYP86A8 and 96% similarity to the petunia CYP86A22 (Han et al., 2010). Sequencing of the complete SlCYP86A69 gene in the mutant revealed a homozygous G to A substitution, which was predicted to cause a Cys460Tyr missense mutation in the active site of the heme binding cysteine of the enzyme (Fig. S2d). This SNP cosegregated with the surface phenotype among > 110 members of the F2 population.

In a parallel study of an unrelated segregating population of the original cd3 mutant (Isaacson et al., 2009), through map positional cloning, we narrowed down the cd3 linkage interval to a 5 kb region containing the SlCYP86A69 gene (Fig. S5a) that was found to harbor a 2 bp (‘CC’) deletion (at positions 1327–1328) and a 1131 bp insertion (Fig. S5b). This deletion and insertion, derived from a genomic DNA sequence on chromosome 8, are predicted to cause a shift in the SlCYP86A69 open reading frame, resulting in the translation of a truncated protein that includes the original first 402 amino acids, the aberrant addition of nine amino acids and a premature stop codon.

The slcyp86a69 mutant displays abnormal fruit surface phenotypes

More detailed examination of enzymatically isolated slcyp86a69 cuticles revealed that the isolated cuticle from either the MG or Red (R) stage mutant fruit was very thin and readily disintegrated (Fig. 4a,b). Light and electron microscopy observations of epidermal sections revealed striking differences between the slcyp86a69 and WT fruit cuticle and epidermal cells (Fig. 4c–f). In WT fruit epidermis section, TEM observation revealed that both cuticular layers (ecl and icl) were significantly reduced in thickness of the MG stage epidermis section (Fig. 4c,d). Furthermore, when compared with WT, the primary cell wall of the mutant epidermal cell was much thicker and more defined. The significant reduction of the cuticular layers in the mutant was further confirmed by light microscopy with Sudan IV staining (Fig. 4e,f) and cryo-SEM (Fig. S2e,f) analyses.

Figure 4.

Identification of cuticle-related tomato (Solanum lycopersicum) cytochrome P450 gene SlCYP86A69. (a) Enzymatically isolated cuticle discs of mature green (MG) stage fruit. While the wildtype (WT) discs stay intact during the isolation process, those of the slcyp86a69 mutant fall apart easily. (b) The enzymatic isolated and exhaustedly extracted cutin matrix of WT and slcyp86a69 fruit epidermis at the breaker (Br) stage. Values represent means ± standard error of the mean (n = 5, **, < 0.01, Student's t-test). (c–d) Transmission electron microscopy (TEM) images of WT (c) and the slcyp86a69 mutant (d) epidermis. ecl, external cuticular layer; icl, internal cuticular layer; pcw, primary cell wall. (e, f) Sudan IV staining images of WT (e) and slcyp86a69 mutant (f) fruit epidermis.

An additional abnormal surface phenotype was closely associated with epidermal cell morphology. Compared with WT, the mutant epidermal cells were significantly elongated (Fig. S2e,f), displaying a more flattened epidermal surface (Fig. S2g,h). Furthermore, the slcyp86a69 mutant phenotypes (in the cv MT background) appeared to be more severe than those observed in the cd3 mutant (cv M82 background; Isaacson et al., 2009). Thus, the phenotypes detected upon mutation in SlCYP86A69, one of SlSHN3 putative downstream targets, largely resembled the epidermal phenotypes detected in the SlSHN3 silenced fruit (Figs 2, S2a,b). At this stage, it is important to mention that expression of the SlCYP86A69 gene was significantly down-regulated in the SlSHN3 silenced lines and that its promoter was activated by SlSHN3 (Fig. 3a,b).

The amount of cutin monomers is substantially reduced in the slcyp86a69 fruit cuticle

The distorted cuticle characteristics described in the previous section indicated that chemical changes might have occurred in the slcyp86a69 cuticle. GC-MS analysis revealed that the amount of cutin monomers was significantly reduced in the slcyp86a69 fruit (red) cuticle. Specifically, the substantial decline in the concentrations of midchain hydroxylated fatty acids (such as C16-9,10/16 DiHFA, C16-ω-HFA, C16:0 DFA), 2-hydroxylated fatty acids and flavonoids had a major contribution to the total reduction of cutin in the mutant fruit epidermis; consequently, there was a substantial accumulation of C16:0 FA (Table 2).

Table 2. List of enzymatically isolated cutin monomers identified after BF3 depolymerization with their respective concentrations in slcyp86a69 mutant tomato (Solanum lycopersicum) and its corresponding wildtype (WT, S. lycopersicum. cv MicroTom) fruit epidermis (red stage)
Cutin monomersμg mg−1
slcyp86a69 WT
  1. Monomers that show significant changes are colored gray (Student's t-test, < 0.05, n = 4).

Saturated fatty acids (FAs)
 C16:0 FA0.7530.1410.1140.012
 C18:0 FA0.0060.0010.0080.004
Dicarboxylic fatty acids (DFAs)
 C16:0 DFA0.0180.0070.1490.001
Terminal hydroxylated fatty acids (ω-HFAs)
Midchain hydroxylated fatty acids (HFAs)
 C16-9,10/16 Di HFA14.3043.82130.1881.484
 C18:1-9,10,18-Tri HFA0.0000.0000.0010.000
 C18-9,10,18-Tri HFA0.0130.0040.0950.003
 C18 Penta HFA0.0100.0010.0440.001
2-Hydroxylated fatty acids (2HFAs)
 C18:1 2HFA0.0980.0270.4150.027
 C23 2HFA0.0040.0010.0020.000
 Naringenin chalcone0.0290.0020.1190.033
Unidentified (UI) monomers

Recombinant SlCYP86A69 enzyme activity

We examined the activity and substrate specificity of the recombinant enzymes encoded by the WT and mutant CYP86A69 genes. The WT enzyme hydroxylated C18:1 as well as C14 and C16, albeit with low activity, while the mutated form showed no such activities (data not shown). In a more detailed examination, oleic acid was incubated with microsomes from yeast expressing CYP86A69 in the absence or presence of NADPH. The reaction products were resolved by GC-MS after derivatization. A metabolite was detectable after 20 min incubation in the presence of NADPH, which was not formed in the absence of NADPH (Fig. 5), with a fragmentation pattern that was characteristic of the derivative of 18-hydroxyoleic acid (Pinot et al., 1992). This enzyme activity was in accordance with the chemical analysis data, confirming the hydroxylase activity of SlCYP86A69.

Figure 5.

Recombinant enzyme activity of SlCYP86A69. (a, b) GC resolution of metabolites generated in incubations of oleic acid with microsomes from yeast expressing SlCYP86A69 in the absence (a) or presence (b) of NADPH. (c) Mass spectrum of derivatized metabolite (peak 1 in Fig. 5b) generated in incubation with oleic acid showed ions at m/z (relative intensity %): 73 (100%), (CH3)3Si+; 146 (17%), (CH2=C+ (OSi(CH3)3-OCH3); 159 (20%), (CH3-O+=C+(OSi(CH3)3)CH=CH2); 337 (51%), (M-47) (loss of methanol from the (M-15) fragment); 369 (19%), (M-15) (loss of CH3 from TMSi group); 384 (12%), (M). This fragmentation pattern is characteristic of derivatized 18-hydroxyoleic acid (M = 384 g mol−1; Pinot et al., 1992).

Cutin deficiency in tomato fruit cuticle enhances postharvest water loss and susceptibility to the fungus C. coccodes

Two important physiological roles of fruit cuticles in pre- and postharvest tomato fruit are thought to be restricting water loss and defending against pathogen (Isaacson et al., 2009). We therefore evaluated susceptibility of WT, slcyp86a69, and SlSHN3-RNAi fruits to water loss and fungal infection. Postharvest WT, slcyp86a69, and SlSHN3-RNAi fruits were kept at room temperature under constant humidity conditions and their weight losses were monitored daily. Fruit of both slcyp86a69 and SlSHN3-RNAi lost significantly more water than those of their corresponding WT at all examined time points (Fig. 6a).

Figure 6.

Responses of tomato (Solanum lycopersicum) fruit to dehydration and fungal infection. (a) Water loss of tomato fruits kept at room temperature up to 1 month. Values represent means ± standard error of the mean (n = 5 replicates; each replicate contains six to eight fruits from at least three different plants). (b) Decay development after fungal inoculation with Colletotrichum coccodes conidia. After surface sterilization in 0.3% (v/v) hypochlorite, rinsing and drying, the fruit was inoculated with 7 μl of 106 conidia ml−1 on the peel. The incubation was carried out at 22°C and 95% relative humidity in covered plastic containers. Each tomato line contained at least 20 fruits, each with two inoculations. Decay diameter was measured daily postinoculation. (c) Representative pictures of tomato fruits 5 d after conidial inoculation.

The black dot disease in tomato (Dickson, 1926) is caused by the plant pathogen C. coccodes. Susceptibility to C. coccodes of whole red fruit and isolated cuticle discs from both the slcyp86a69 and SlSHN3 RNAi plants was tested by ectopic application of C. coccodes conidia. From the third day onwards, the decay diameters on both slcyp86a69 and SlSHN3 RNAi fruits were significantly higher than those of WT fruits. The slcyp86a69 fruits were significantly more susceptible to the fungus than SlSHN3 RNAi fruit; 7 d postinoculation, the decay diameter of the former was three times that of the latter (Fig. 6b,c). Interestingly, the different susceptibility of slcyp86a69 and SlSHN3 RNAi fruits to C. coccodes was in accordance with the different reductions in the amounts of cutin in their fruit cuticles (Figs 2, 4; Tables 1, 2).


The plant cuticular layer varies significantly among organs of even the same plant species, tightly associating with the role of each organ in development and growth and the interaction with the surroundings (Koch & Barthlott, 2009). Variation in organ surfaces is largely a result of altered composition of the cuticular matrix and its structure. Other factors that play a major role in patterning the cuticular layer are the underlying epidermal cells that generate building blocks for the construction of the cuticle and their characteristic cell wall adjoining the cuticular layer. The surfaces of many fleshy fruits are unusual compared with most other aerial plant organs, as they lack stomata. This study and recently published work (Curvers et al., 2010; Leide et al., 2011; Girard et al., 2012; Yeats et al., 2012) aimed to unravel the genetic factors that mediate the assembly of the fleshy fruit surface. Here, we centered on the transcriptional regulation of the cuticle biosynthetic pathway in fruit largely through the characterization of SlSHN3. This protein appears to be an important element in the control of cuticle formation in epidermal cells and in the regulation of cell patterning during tomato fruit development.

SlSHN3 is an AtSHN3 ortholog that is required for tomato fruit cutin biosynthesis and possibly for cutin polymer assembly

Subsequent to the discovery of the Arabidopsis SHN1/WIN1 clade (Aharoni et al., 2004; Broun et al., 2004), Taketa et al. (2008) reported on the HvNUDUM gene (HvNud), a barley ortholog of AtSHN1, and suggested that HvNud regulates lipid formation associated with caryopsis and hull adhesion. As with three AtSHNs and three SlSHNs that are also linked to the regulation of surface lipids, HvNud contains two conserved motifs outside the AP2 domain, termed the ‘mm’ and ‘cm’ domains (Aharoni et al., 2004). The study in barley provided the first evidence for the functional importance of a highly conserved valine residue in the ‘mm’ motif (Taketa et al., 2008). It is therefore possible that the ‘mm’ and ‘cm’ motifs have an important role in the activity of this subclass of AP2 domain factors. Hence, the presence of both sequences in a given protein can serve as a good indicator for the identification of SHN orthologs from multiple species (Fig. S6). At this stage, the role of these motifs in SHNs is not clear, which merits further investigation.

The presence of the SHN signature domains in SlSHN3 provides a clue for its role in regulating cutin biosynthesis in tomato fruit. Although SlSHN3 transcripts were detected in various tissues, it was predominantly expressed in the epidermal tissues of the IG stage fruit. More detailed transcriptomic analysis in five isolated fruit cell layers showed that the SlSHN3 transcript is predominantly expressed in the outer and inner epidermal layers (Fig. S1b). It appears that many of the cuticle-associated genes display a spatial expression pattern in fruit that is similar to SlSHN3 (Matas et al., 2011). This characteristic transcript profile is congruent with the presence of a cuticle coating the locular surface of the inner epidermis of tomato fruit (Mintz-Oron et al., 2008; Matas et al., 2011). Apart from the SlSHN3 transcript profile, its heterologous expression in Arabidopsis and silencing in tomato clearly pointed to its involvement in the control of the cuticle biosynthetic pathway, which is required for cuticle formation during the early stages of fruit development. As genes encoding GDSL-motif lipases (SlGDSLa and SlGDSLb/SlGDSL1) and acyl transferases (SlGPAT6 and SlDCR), previously suggested to be involved in either cutin polymer oligomerization, assembly, modification or recycling (Pollard et al., 2008; Panikashvili et al., 2009; Yeats et al., 2010; Girard et al., 2012), were significantly down-regulated in the SlSHN3 silenced fruit, it is likely that SlSHN3 does not merely regulate the biosynthetic genes involved in cutin monomer production.

SlSHN3 action is associated with regulatory proteins controlling epidermal cell patterning

During land colonization, plants acquired metabolic pathways as part of the epidermal cell patterning process to generate cell wall and cuticular constituents (including those needed for synthesis of the polymers lignin, cellulose, cutin, and suberin), providing mechanical strength for maintaining their architecture and transporting water as well as for forming a barrier from the surroundings. It is therefore not surprising that the regulatory network controlling the patterning of epidermal cells might act on several epidermal pathways in concert. Indeed, work with cuticular mutants suggested that the process of epidermal cell patterning is tightly associated with the metabolism of cuticle constituents, as mutations in cuticle genes often result in altered trichome, epidermal pavement cells and stomata development (Yephremov et al., 1999; Bird & Gray, 2003; Kurdyukov et al., 2006b). In addition, overexpression of AtSHNs in Arabidopsis severely affected epidermal cell development (Aharoni et al., 2004).

Recent studies in tomato fruit (Isaacson et al., 2009; Nadakuduti et al., 2012) and maize (Javelle et al., 2010) and this report provide a possible insight into the interaction between the cuticle and epidermal development. In these three studies, genes encoding TFs of the HD-ZIP IV family were proposed to act on the transcriptional regulation of genes involved in cuticular lipid biosynthesis. Several members of the HD-ZIP IV family were shown to exhibit enriched or specific expression in the outer cell layer of various organs and were predicted to function in epidermal processes. For example, AtGL2 regulates both the synthesis of seed mucilage (Rerie et al., 1994) and the root cell wall (Tominaga-Wada et al., 2009), AtPDF2 controls cell fate in the epidermal layer (Abe et al., 2003), and AtHDG11 restrains the outgrowth of trichomes (Nakamura et al., 2006). The tomato homologs of these Arabidopsis HD-ZIP genes (i.e. SlGL2, SlPDF2d and SlHDG11a) displayed a significant down-regulation in the SlSHN3 silenced lines. Interestingly, SlANL2a/CD2, which was associated with the regulation of tomato fruit cutin biosynthesis (Isaacson et al., 2009) and epidermal cell function (Nadakuduti et al., 2012), was not altered in the SlSHN3 silenced fruit, suggesting that it is likely not a direct downstream target of SlSHN3. A very recent study by Wu et al. (2011) reported that the rice and Arabidopsis HDG1 regulates two cuticle-associated genes (BDG and FDH) by binding to the cis-element L1 box in their regulatory regions. CFL1 modulates the function of HDG1 through protein–protein interactions; however, it regulates the expression of SHN1 through a route other than HDG1. These findings, together with the data reported here, provide insights into the regulatory hierarchy among CFL1, SHNs and HDGs in the tomato fruit cuticle.

Another regulatory protein SlMIXTA, whose Arabidopsis ortholog AtMIXTA is associated with epidermal cell development (Baumann et al., 2007), was also significantly down-regulated in SlSHN3 silenced fruit. We previously showed that SlMIXTA transcripts are highly enriched in the epidermis in five stages of fruit development (Mintz-Oron et al., 2008), and expression analysis in specific tomato fruit cells and tissues showed that it is particularly enriched in the outer and inner epidermal layers, as observed for many cuticle-associated genes (Fig. S4). At least two of the epidermal cell development-associated factors (SlGL2 and SlMIXTA) could be direct downstream targets of SlSHN3, as their promoters were strongly activated by SlSHN3 in transient promoter activation assays (Fig. 3b).

An additional process in the epidermal layer that might be partially controlled by SHNs and the HD-ZIP proteins is polysaccharide cell wall metabolism. Shi et al. (2011) showed that genes associated with pectin metabolism and cell wall structural proteins are likely direct targets of SHNs in Arabidopsis. Meanwhile, Ambavaram et al. (2011) reported that AtSHN2 controls secondary cell wall biosynthesis (lignin and cellulose). Key evidence for this was provided by overexpression of AtSHN2 in rice, which resulted in increased cellulose and decreased lignin content in culms. Here we observed that reduced activity of SlSHN3 resulted in changes in the shape and thickness of the outer epidermal cells (Fig. 2a,b), suggesting that the epidermal cell wall has been perturbed. In Arabidopsis, Tominga-Wada et al. (2009) provided evidence that GL2 regulates cell wall metabolism through direct action on promoters of CELLULOSE SYNTHASE 5 and XYLOGLUCAN ENDOTRANSGLUCOSYLASE 17 genes. Microarray analysis showed that the expression of genes putatively encoding epidermal patterning regulators, cell wall-related and structural cell wall proteins was significantly altered in the SlSHN3 RNAi MG fruit peel (data not shown).

Taken together, these data suggest that SlSHN3 might take part in a regulatory network controlling cuticular lipid biosynthesis that involves the action of epidermal cell patterning TFs. SlSHN3 may control epidermal cuticle and cell wall metabolism either indirectly through the HD-ZIP or MIXTA proteins or directly by acting on their promoters, or alternatively through both pathways (Fig. 3).

The SlSHN3 downstream target gene encoding a CYP86A subgroup member is crucial for tomato fruit surface formation

In the course of studying SlSHN3, we identified an EMS-induced mutant that corresponded to one of its putative downstream target genes encoding a cytochrome P450 of the CYP86A subgroup (i.e. SlCYP86A69). Members of this subgroup were previously reported to act as fatty acid hydroxylases in the cutin biosynthetic pathway (Wellesen et al., 2001; Xiao et al., 2004). They were also proposed to be putative downstream targets of Arabidopsis SHNs (Kannangara et al., 2007; Shi et al., 2011). Both the EMS mutation and cd3, a second slcyp86a69 mutant allele, displayed a similar cuticle-deficient phenotype to the SlSHN3 silenced lines, albeit more severe (Fig. S2; Isaacson et al., 2009).

In terms of enzyme activity, despite the relatively higher level of the recombinant CYP86A69 activity towards C18:1, the most dramatic reduction in amounts of the C16:0 derivatives in both slcyp86a69 mutant alleles suggests that in vivo CYP86A69 acts on both C18:1 and C16:0 acids as a substrate for ω-hydroxylation. The reduced but not complete loss of C16-9/10, 16 DiHFA in slcyp86a69 mutant fruit (Table 2) points to an additional activity that might be catalyzed by the SlCYP86A68 enzyme. We further propose that SlCYP77A1, which is also significantly down-regulated in the SlSHN3 silenced fruit (Fig. 3a), acts as a midchain hydroxylase in tomato. The activities of both SlCYP86A68 and SlCYP77A1 are currently under study.

SlSHN3 and CYP86A69 are required for fruit pathogen and dehydration response

Studies of Arabidopsis and tomato fruit mutants and transgenic lines with abnormal cuticles demonstrated that modifications of cuticle composition, structure and permeability may lead to changes in response to both biotic or abiotic stresses (Bessire et al., 2007; Isaacson et al., 2009). Isaacson et al. (2009) proposed that cutin provides protection against microbial infection but that cuticular waxes and triterpenoids, rather than cutin, have a key influence on water loss. Fruits from both the SlSHN3 silenced lines and the cyp86a69 mutants displayed increased water loss and sensitivity to the fungus C. coccodes and this was coupled to various degrees of modification in both cuticular waxes and cutin monomers (Fig. 6 and Tables 1, 2). Apart from compositional changes, the thickness of the cuticle was most dramatically reduced, particularly in the cyp86a69 fruit. This physical change might have facilitated the penetration of C. coccodes into the fruit surface and induced appresoria formation as observed in the cyp86a96 fruit.

The altered response to stress in both genotypes might have been initiated by changes to any of the chemical or physical factors described earlier that alter the diffusive properties of the cuticle. However, besides its passive protective role, the cuticle also has sensing capabilities that induce signaling cascades associated with the plant activation of defense (L'Haridon et al., 2011; Liu et al., 2011). Fruit cuticle-associated mutants and transgenic lines generated here and in previous studies could serve as an excellent genetic material to discover the active role of the cuticle in the induction of stress responses. The same kinds of studies should also incorporate the investigation of ABA biosynthesis and signaling in the cuticular context, as this hormone seems to be crucial for cuticular-mediated signaling (Curvers et al., 2010; Wang et al., 2011).


We thank Eugenia Klein and Eyal Shimoni for help with the electron microscopy assays, Dana Elyahu and Dorit Levy for technical assistance, and Ester Feldmesser for bioinformatics analyses. We thank David Nelson for naming of the CYP450s analyzed in this study and Laetitia Martin for careful reading of this manuscript. A. Aharoni is the incumbent of the Adolpho and Evelyn Blum Career Development Chair of Cancer Research. The work was supported by the Israel Science Foundation and the European Research Council (ERC) project SAMIT (FP7 program) to A.A., Agence Nationale de la Recherche (SUIT project: ANR-09-KBBE-006-001) to B.G., NSF Plant Genome Research Program (DBI-0606595), the United States-Israel Binational Agricultural Research and Development Fund (IS-4234-09), and CUAES-Hatch grant (NYC-18442) to J.K.C.R. All authors declare that there is no conflict of interest.