Cuticle Biosynthesis in Tomato Leaves Is Developmentally Regulated by Abscisic Acid^,

Abscisic acid regulates cutin and wax abundance and composition in the cuticles of tomato leaves.

Abstract

The expansion of aerial organs in plants is coupled with the synthesis and deposition of a hydrophobic cuticle, composed of cutin and waxes, which is critically important in limiting water loss. While the abiotic stress-related hormone abscisic acid (ABA) is known to up-regulate wax accumulation in response to drought, the hormonal regulation of cuticle biosynthesis during organ ontogeny is poorly understood. To address the hypothesis that ABA also mediates cuticle formation during organ development, we assessed the effect of ABA deficiency on cuticle formation in three ABA biosynthesis-impaired tomato mutants. The mutant leaf cuticles were thinner, had structural abnormalities, and had a substantial reduction in levels of cutin. ABA deficiency also consistently resulted in differences in the composition of leaf cutin and cuticular waxes. Exogenous application of ABA partially rescued these phenotypes, confirming that they were a consequence of reduced ABA levels. The ABA mutants also showed reduced expression of genes involved in cutin or wax formation. This difference was again countered by exogenous ABA, further indicating regulation of cuticle biosynthesis by ABA. The fruit cuticles were affected differently by the ABA-associated mutations, but in general were thicker. However, no structural abnormalities were observed, and the cutin and wax compositions were less affected than in leaf cuticles, suggesting that ABA action influences cuticle formation in an organ-dependent manner. These results suggest dual roles for ABA in regulating leaf cuticle formation: one that is fundamentally associated with leaf expansion, independent of abiotic stress, and another that is drought induced.

The plant cuticle, a hydrophobic extracellular layer that coats the epidermis of aerial organs, is composed of a cutin matrix of ester-linked ω-hydroxylated fatty acids that is covered by epicuticular waxes and infiltrated with intracuticular waxes. Polymeric cutin has also been reported to contain relatively low levels of phenolic compounds, such as caffeic, ferulic, and coumaric acid esters (Baker et al., 1982; Rautengarten et al., 2012). The cuticle is a key barrier to desiccation and pathogen entry and so is required from the onset of organ formation throughout ontogeny (Yeats and Rose, 2013). In addition to being an intrinsic part of organ developmental programs, cuticle formation is dynamic as its composition and coverage change in response to environmental conditions, such as light/dark, ultraviolet radiation, and drought (Hooker et al., 2002; Cameron et al., 2006; Shepherd and Wynne Griffiths, 2006; Kosma et al., 2009; Go et al., 2014; Martin and Rose, 2014). For example, water deficit in Arabidopsis (Arabidopsis thaliana) triggers an increase in the accumulation of both cutin monomers and waxes, resulting in a thicker, less permeable cuticle (Kosma et al., 2009).

A key factor in the mediation of responses to water deficit is the phytohormone abscisic acid (ABA; Liu et al., 2005), and the consequences of ABA treatment on cuticle composition and associated gene expression have been investigated (Kosma et al., 2009). While the levels of waxes, and in particular alkanes, were reported to increase following ABA application, no such change in the abundance of cutin monomers was observed, suggesting that other factors trigger changes in the cuticle following water stress (Kosma et al., 2009). Furthermore, the expression levels of numerous wax biosynthesis-related genes are known to respond to ABA treatment (Kosma et al., 2009; Seo et al., 2009, 2011; Wang et al., 2012). For example, the expression of ECERIFERUM1 (CER1), which encodes an enzyme that generates alkanes from very-long-chain acyl-CoA (Bernard et al., 2012), is induced by ABA treatments (Kosma et al., 2009). In addition, the presence of cis-ABA-responsive elements in the promoters of some wax-related genes, such as CER6 (Hooker et al., 2002), suggests that ABA directly regulates wax synthesis. Moreover, some Arabidopsis transcriptions factors that are associated with cuticle biosynthesis, such as MYB96, are known to be regulated by ABA and to be part of the ABA-mediated signaling pathway (Cominelli et al., 2008; Seo et al., 2009; Lee et al., 2015). Lastly, Cui et al. (2016) showed that the expression of some cuticle related genes was affected in Arabidopsis ABA mutants, further suggesting a role for ABA in cuticle development.

Several recent reports indicate that the relationship between ABA and cutin is complex, and ABA biosynthesis and signaling are impaired in mutants with disrupted cutin biosynthesis, suggesting that cutin components may be involved in mediating osmotic stress signaling (Wang et al., 2011). In addition, the transcription factor NF-X LIKE2, which binds to the promoters of a cutin-related gene and a transcription factor regulating cutin formation (Lisso et al., 2012), suppresses ABA accumulation and ABA responses, possibly to avoid osmotic stress response under normal conditions (Lisso et al., 2011). Thus, NF-X LIKE2 suppresses both ABA signaling and cutin biosynthesis. However, ABA has not been shown to have a direct effect on cutin formation.

Collectively, these studies demonstrate that ABA plays a role in the regulation of cuticular wax formation, but it is not known whether this is only in the context of responses to environmental stress or whether ABA influences the developmental regulation of cuticle biosynthesis. Indeed, in addition to its role in plant responses to water stress, ABA is also known to regulate many aspects of development, including the maintenance of shoot growth in well-watered plants (Sharp et al., 2000), the promotion of seed dormancy and the synthesis of seed storage proteins and lipids (Finkelstein et al., 2002). Consequently, a goal of this current study was to test the hypothesis that ABA is required for cuticle formation independent of water stress.

To this end, we studied the cuticle structure, composition, and properties of three tomato (Solanum lycopersicum) ABA-deficient nonallelic mutants: notabilis, flacca, and sitiens. The notabilis (not) mutant has a mutation in a 9-cis-epoxycarotenoid dioxygenase gene, which is believed to encode a key regulatory enzyme of the ABA biosynthesis pathway (Burbidge et al., 1999; Thompson et al., 2000). The mutation in sitiens (sit) affects an aldehyde oxidase that catalyzes the last step of the ABA synthesis pathway (Harrison et al., 2011), while flacca (flc) has a mutation in a molybdenum cofactor sulfurase that is required for the activation by sulfuration of the molybdenum cofactor of aldehyde oxidase (Sagi et al., 2002). The severity of the phenotypes of these mutants correlates with the ABA concentrations in shoots of the 5-week-old plants: not has the mildest phenotype and the highest level of ABA, while sit has the most severe phenotype and the lowest ABA levels (Tal and Nevo, 1973; Taylor and Tarr, 1984).

In this study, the role of ABA in cuticle formation was investigated by assessing the structure and composition of leaf and fruit cuticles from the three ABA-deficient mutants. Additionally, the expression of genes involved in cuticle biosynthesis, transport, and regulation was evaluated to determine the consequence of ABA deficiency on the cuticle biosynthesis pathway. Finally, to determine whether the functions of the ABA-deficient fruit cuticles were impaired, we measured water loss, from which we inferred changes in cuticle permeability. Based on the results, we propose a dual, but organ-specific, role for ABA in the control of cuticle formation.

RESULTS

ABA Deficiency Leads to a Reduction in Plant and Organ Size

For the purposes of this study, we compared not and flc in the Ailsa Craig (AC) genetic background [not(AC) and flc(AC), respectively] as well as sit and flc in the Rheinlands Ruhm (RR) background [sit(RR) and flc(RR), respectively]. We also included the RR and AC wild-type genotypes in the comparison. Importantly, this allowed an assessment of conserved phenotypes in different genetic backgrounds. Most of the published analyses of ABA levels in the ABA-deficient not, sit, and flc tomato mutants have focused on their leaves (Tal, 1966; Tal et al., 1979; Nagel et al., 1994). We observed that this reduction in ABA levels was consistent throughout leaf development and in ripening fruits (Fig. 1), indicating a constitutive perturbation of ABA biosynthesis. Proportionally, the reduction in ABA levels was greater in Pink (Pk) stage fruit, ranging from a 89% decrease in flc(AC) to a 99% decrease in sit(RR), than in leaves. However, the wild-type levels of ABA were substantially lower in Pk fruit than they were in leaves, regardless of the developmental stage. Furthermore, leaf ABA levels were lower in sit(RR) than in flc(RR), and lower in flc(AC) than in not(AC), as was previously reported (Tal and Nevo, 1973; Taylor and Tarr, 1984). Plant growth was generally reduced in the ABA-deficient plants (Supplemental Fig. S1), which had smaller curled leaves (Fig. 2A) and smaller fruits (Fig. 2B; Supplemental Fig. S2) than those of the respective wild types. A reduction in organ size, including roots, of these mutants has previously been reported (Rančić et al., 2010; Sharp et al., 2000; Nitsch et al., 2012).

Figure 1.

ABA levels in fruit and four leaf developmental stages of ABA-deficient tomato lines. The label “emerging” corresponds to developing leaves of fully mature plants (leaf size <2 cm). The labels “small,” “medium,” and “large” represent three successive developmental stages of leaf development of 5-week-old plants, as described in the “Materials and Methods” section. The label “Fruit” corresponds to Pk stage fruit pericarp. Differences between the mutants and their corresponding wild-type genotypes were assessed using Dunnett test if the equal variances hypothesis of the Levene test was not rejected at α = 0.05, and by a Steel test otherwise, with *P < 0.05, **P < 0.01, and ***P < 0.001, respectively. The error bars represent ses of the mean for n = 4. ng/g FW, ng of ABA per g of fresh weight; AC, Ailsa Craig; not(AC), notabilis in AC background (AC); flc(AC), flacca in (AC); RR, Rheinlands Ruhm; flc(RR), flacca in (RR); sit(RR), sitiens in (RR).

Figure 2.

ABA-deficient mutant leaf and fruit phenotypes. A, Photographs of the youngest mature (large) leaf of 5-week-old seedlings are represented. Scale bar = 5 cm. B, Pictures of representative Red Ripe stage fruits of the ABA-deficient genotypes and their respective wild types. Scale bar, 1 cm. AC, Ailsa Craig; not(AC), notabilis in AC background (AC); flc(AC), flacca in (AC); RR, Rheinlands Ruhm; flc(RR), flacca in (RR); sit(RR), sitiens in (RR).

Effect of ABA Deficiency on Leaf and Fruit Cuticle Ultrastructure and Distribution

The ultrastructure of the leaf and fruit cuticles of flc(AC) and sit(RR) and their respective wild-type genotypes were assessed using transmission electron microscopy (TEM). The cuticles of small expanding mutant fruits of similar diameters (∼13.5 mm) were thicker than those of the wild-type fruit (Supplemental Fig. S3); specifically, an increase of 364% and 220% for flc(AC) and sit(RR), respectively (Supplemental Table S1). In contrast, the cuticles of the adaxial and abaxial fully expanded leaf surfaces were thinner in the mutants than in the wild-type genotypes (Fig. 3; Supplemental Fig. S4). We observed a reduction in the leaf adaxial cuticular thickness of 31% and 48% (significant for P < 0.05) for flc(AC) and sit(RR), respectively, and of 21% and 42% for the abaxial surface of flc(AC) and sit(RR), respectively (Supplemental Table S1).

Figure 3.

TEM micrographs of the cuticle of the adaxial epidermis of fully expanded leaflets. A, C, E, and G, Outer epidermal cell junction. B, D, F, and H Outer periclinal wall of an epidermal cell. Scale bars = 500 nm. Cl, Chloroplast; Cp, cuticle proper; cut, cuticular membrane; cyt, cytoplasm; dis, disrupted cuticle; gl, cuticular globules; pcw, polysaccharide cell wall; V, vacuole. AC, Ailsa Craig; flc(AC), flacca in (AC); RR, Rheinlands Ruhm; sit(RR), sitiens in (RR).

The ultrastructure of the wild-type and the mutant fruit cuticles were indistinguishable (Supplemental Fig. S3), but various abnormalities were observed in the mutant leaf cuticles. Most strikingly, while the cuticles of the wild-type leaves had a uniformly reticulate appearance between the polysaccharide cell wall and the cuticle proper layer, the cuticles of the mutants showed an accumulation of electron-translucent globules (Fig. 3, C, G, and H; Supplemental Fig. S4G). In addition, the structure of the leaf cuticles was occasionally disrupted (Fig. 3H). Irregularities in the appearance of the cuticle and the polysaccharide cell wall have previously been reported for sit in the Moneymaker background (Curvers et al., 2010), but in that analysis, the mutant leaf cuticle was reported to be thicker than the wild-type cuticle, which is contrary to the results of this study. This apparent contradiction may be explained by insufficient contrast in the TEM image shown in Figure 2 of Curvers et al. (2010) between the cuticle and the polysaccharide cell wall of the wild type, making it difficult to distinguish the cuticle. Indeed, it appears that the thin region that the authors defined as the cuticle may have been inadvertently mislabeled, and a re-examination of the images published in the article suggests to us that the wild-type cuticle was not thinner than that of the mutants.

To further assess possible cuticular abnormalities in the ABA mutants, we examined the cuticles of fully expanded fruit at the Mature Green (MG) stage using light microscopy (Supplemental Fig. S5). The mutant cuticles did not show a major difference in thickness, based on the region overlying the center point of the outer epidermal cells, compared with their respective wild types: flc(RR) and sit(RR) cuticles were 41% (P < 0.0001) and 40% (P < 0.0001) thicker, respectively, while those of not(AC) and flc(AC) were not significantly thicker than the wild type (Supplemental Table S2). However, the cuticles of the mutants showed more substantial penetration through the anticlinal and periclinal walls of additional subepidermal layers (Supplemental Fig. S5). Image analysis indicated an increased penetration of 35% (P < 0.0001), 40% (P < 0.0001), and 22% (P < 0.001) for not(AC), flc(RR), and sit(RR), respectively, but the average increase (13%) measured in flc(AC) was not statistically significant from wild type (Supplemental Table S2).

Effect of ABA Deficiency on Cutin and Wax Chemical Composition of Fruit and Leaf Cuticles

To confirm the observation, made by light microscopy, that the distribution of cuticles of the mutant MG fruit was more extensive than in the respective wild-type fruit, we performed a compositional analysis of isolated cuticles. The amount of total cutin per square cm of the mutant MG fruit was significantly higher for not(AC) and flc(RR) with an increase of 38% and 21%, respectively, and slightly higher for flc(AC), with an increase of 17%, but not for sit(RR) (Supplemental Fig. S6). Similarly, the increases of levels of coumaric acids (125%) and of 10(9),16-dihydroxyhexadecanoic acid (33%) were statistically significant for not(AC), and no differences were observed in the levels of hexadecanoic acid (C₁₆) and 1,16-hexadecanedioic acid (C₁₆ di-ac). The only cutin-associated compound that was consistently different in the four mutant genotypes compared with their respective wild types was ω-hydroxyhexadecanoic acid (ω-OH C₁₆): not(AC), flc(AC), and flc(RR) had increases of 16%, 160%, and 99%, respectively, but sit(RR) showed a decrease of 18% (Supplemental Fig. S6). These biochemical data were consistent with the light microscopy analysis, indicating that the mutant fruit cuticles were somewhat thicker than those of the respective wild types.

The total amounts of fruit cuticular waxes per square cm did not significantly differ between the mutants and their respective wild-type genotypes (Supplemental Fig. S7). However, some compositional differences were observed, as the mutant cuticles tended to have lower amounts of alkanes, with a significant decrease in the levels of the C₃₀ and C₂₉ alkanes in either three [flc(AC), flc(RR), and sit(RR)]or one [flc(AC)] of the mutant genotypes, respectively. In contrast, levels of amyrins were generally higher in the mutants, with a significant increase, compared to the wild-type fruits, in the amounts of α-amyrin and δ-amyrin in two [not(AC) and flc(RR)] and one [flc(RR)] of the mutants, respectively. These results suggest that wax components of MG fruit cuticles are only moderately affected by ABA deficiency.

We then measured cutin levels in leaves at three distinct developmental stages (the youngest leaf of a 5-week-old plant, the youngest fully expanded leaf, and the leaf located in between them, referred as small, large, and medium in Fig. 4, respectively), which revealed four distinct patterns of compound accumulation (Fig. 4). The levels of coumaric acids, caffeic acid, ω-OH C₁₆, 10(9),16-diOH C₁₆, 2-OH C₂₄ and C₃₂ were consistently lower in the mutant genotypes at all developmental stages than those in the wild types. These decreases resulted in a decrease in total cutin levels for the three stages of development in the mutants. In contrast, the levels of C₁₆, dimethylester C₁₆, C₁₈ diene, and C₁₈ triene were consistently higher. The levels of 2-OH C₂₅ and 2-OH C₂₆ were unaffected in the mutants and those of ferulic acid, C₁₈, C₁₆, and C₃₀ showed no consistent difference through leaf development.

Figure 4.

Cutin analysis of ABA-deficient lines for three leaf developmental stages. Amount of cutin monomers per cm² of small (A), medium (B), and large (C) leaves of 5-week-old plants (n = 6). Significant differences between the mutants and their wild types were given by a Dunnett test if the equal variances hypothesis of the Levene test was not rejected at α = 0.05 and by a Steel test otherwise (*P < 0.05, **P < 0.01, and ***P < 0.001). The error bars represent se of the mean. “Total identified” refers to the sum of the compounds shown in the graphs, and “total” refers to the sum of the peaks detected by GC. AC, Ailsa Craig; not(AC), notabilis in AC background (AC); flc(AC), flacca in (AC); RR, Rheinlands Ruhm; flc(RR), flacca in (RR); sit(RR), sitiens in (RR) and ac, acid.

Unlike the effect on cutin, the effect of ABA deficiency on wax levels became more marked during leaf development. Amyrins and alkanes of chain length from C₃₁ to C₃₃ were the only constituents showing a consistent change between the small leaf mutants and their respective wild types (Supplemental Fig. S8). Medium and large mutant leaves also generally had lower levels of these compounds (Supplemental Fig. S9; Fig. 5). Additionally, in medium and large leaves, alkanes from C₂₅ to C₃₀, isoalkanes from C₂₉ to C₃₃, OH-C₂₂ and the fatty acid C₂₂ tended to be more abundant in the mutants than in their wild types (Supplemental Fig. S9; Fig. 5). In contrast, the increase of the levels of aiC₂₈, aiC₃₀, and aiC₃₁ anteisoalkanes was the most pronounced for the mutant medium leaf and the decrease of the longer anteisoalkanes levels (aiC₃₂, aiC₃₃, and aiC₃₄) was the most consistent in large leaves. These results indicate that the composition of the leaf cuticular waxes was altered by ABA deficiency. In summary, ABA deficiency had a moderate effect on the cuticles of MG fruits and resulted in a reduction in the levels of cutin and changes in wax and cutin composition in leaf cuticles. Notably, the observed cuticle compositional changes showed opposite patterns in fruits and leaves: The C₂₉ and C₃₀ alkanes were less abundant in the mutant fruit cuticles but more abundant in mutant leaf cuticles, while α- and δ-amyrins, coumaric acids, 10(9),16-diOH C₁₆, and ω-OH C₁₆ were more abundant in the mutant fruit cuticles but less abundant in leaf cuticles.

Figure 5.

Analysis of large leaf cuticular waxes from ABA deficient mutants. Amount of waxes per cm² of the youngest fully expanded leaf of 5-week-old plants (n = 5). Significant differences between the mutants and their wild types were given by a Dunnett test if the equal variances hypothesis of the Levene test was not rejected at α = 0.05, and by a Steel test otherwise (*P < 0.05, **P < 0.01, and ***P < 0.001). The error bars represent se of the mean. Outlier values were removed. “Total identified” refers to the sum of the compounds shown in the graph, and “total” refers to the sum of the peaks detected by GC. AC, Ailsa Craig; not(AC), notabilis in AC background (AC); flc(AC), flacca in (AC); RR, Rheinlands Ruhm; flc(RR), flacca in (RR); sit(RR), sitiens in (RR). ‘i’ refers to the iso form of the alkane and ‘ai’ to the anteiso form.

Effect of ABA Application on ABA-Deficient Leaves

Exogenous application of ABA to the ABA-deficient mutants resulted in a slight, but consistent increase in leaf total cutin levels, with an increase of 20% and 11% for flc(AC) and sit(RR), respectively (Fig. 6). More specifically, ABA application resulted in an increase in the abundance of coumaric acids (93% and 50%), caffeic acids (83% and 18%), and 10(9),16-diOH C₁₆ (58% and 30%), for flc(AC) and sit(RR), respectively, suggesting a partial rescue of the cuticle deficiency (Fig. 6). Application of ABA treatment to wild-type leaves resulted in a minor increase in total cutin for AC (16%), but not for RR (−5%), but had little effect on the wild-type cuticular waxes (Fig. 7). This result was unexpected, as it was reported that applying ABA to Arabidopsis leaves resulted in an increase in the levels of alkanes (Kosma et al., 2009). We considered the possibility that the amount of ABA penetrating the tomato leaves in our study was insufficient to induce the expected increase in alkane levels. However, the same ABA treatment was able to partially restore the normal amounts of waxes in the leaves of the mutant lines (Fig. 7). Indeed, ABA application resulted in an increase in the abundance of most of the compounds that were reduced by ABA deficiency (C₃₁–C₃₃ alkanes, C₃₂–C₃₄ anteisoalkanes, amyrins and taraxasterol) and in a decrease in the abundance of most of the compounds that were increased by ABA deficiency (C₂₆ and C₂₇ alkanes and C₂₉–C₃₁ isoalkanes). These results confirmed that ABA deficiency alters the cuticular waxes composition of tomato leaves (Fig. 7).

Figure 6.

Leaf cutin phenotype is partially rescued by ABA application. Cutin analysis was performed on the youngest mature (large) leaf of 5-week-old plants. Black and red stars represent statistical differences with wild type-ctr and with the corresponding control genotype, respectively. Control plants (-ctr) were sprayed with water + 0.1% ethanol; -ABA plants were sprayed with 100 µm ABA + 0.1% ethanol. Significant differences between the treated plants and their controls were given by a Dunnett test if the equal variances hypothesis of the Levene test was not rejected at α = 0.05 and by a Steel test otherwise (*P < 0.05, **P < 0.01, and ***P < 0.001; n = 6). The error bars represent se of the mean. ctr, Control. “Total identified” refers to the sum of the compounds shown in the graph, and “total” refers to the sum of the peaks detected by GC. AC, Ailsa Craig; flc(AC), flacca in (AC); RR, Rheinlands Ruhm; sit(RR), sitiens in (RR).

Figure 7.

Leaf wax phenotype is partially rescued by ABA application. Wax analysis was performed on the youngest mature (large) leaf of 5-week-old plants. Black and red stars represent statistical differences with wild type-ctr and with the corresponding control genotype, respectively. Control plants (-ctr) were sprayed with water + 0.1% ethanol; -ABA plants were sprayed with 100 µm ABA + 0.1% ethanol. Significant differences between the treated plants and their controls were given by a Dunnett test if the equal variances hypothesis of the Levene test was not rejected at α = 0.05 and by a Steel test otherwise (*P < 0.05, **P < 0.01, and ***P < 0.001; n = 6). The error bars represent ses of the mean. ctr, Control. “Total identified” refers to the sum of the compounds shown in the graph, and “total” refers to the sum of the peaks detected by GC. AC, Ailsa Craig; flc(AC), flacca in (AC); RR, Rheinlands Ruhm; sit(RR), sitiens in (RR).

The Cuticle Biosynthesis Pathway Is Down-Regulated in ABA-Deficient Lines

To better understand the complex changes in leaf cutin and cuticular wax composition resulting from ABA deficiency, we investigated whether there were corresponding variations in the expression of cuticle biosynthesis genes. Some of these genes have previously been characterized in tomato (CUTIN SYNTHASE1 [CUS1]; Yeats et al., 2012; CUTIN DEFICIENT3 [CD3]; Isaacson et al., 2009; Shi et al., 2013; and CER6; Vogg et al., 2004), but other genes that we targeted were identified using the amino acid sequences of cuticle-related proteins from Arabidopsis to search the Sol Genomics Network database (http://solgenomics.net/) with the BLAST tool, thereby obtaining the closest tomato homolog. Most of the surveyed genes involved in cutin and/or wax monomer biosynthesis, transport, deposition, and regulation were expressed at lower levels in the emerging leaves of the mutants than in those of the wild types (Supplemental Fig. S10A). Of particular note, the expression levels of CD3 and GLYCEROL-3-PHOSPHATE ACYLTRANSFEREASE6 (GPAT6), which catalyze cutin monomer formation (Shi et al., 2013; Yang et al., 2012), were significantly lower (2- to 4.5-fold) in the ABA-deficient mutants than in their respective wild-type genotypes (Supplemental Fig. S10A). Additionally, the expression levels of wax biosynthesis genes, such as CER3 or CER6, and of the CUTICLE MONOMER ATP-BINDING CASSETTE TRANSPORTER11 (ABCG11), were also substantially lower in the mutants than in the wild-type leaves (Supplemental Fig. S10A; Samuels et al., 2008). In addition, the mean expression levels of CUS1, CYP77A6, ABCG12, GPAT4, LONG-CHAIN ACYL-COA1 (LACS1), and CER1 were consistently lower in the emerging leaves of the mutants, although in each case this was statistically significant for only one, or none, of the mutant genotypes. The expression levels of HOTHEAD, FIDDLEHEAD, WSD1, LACS2, GPAT4, and of the gene expression regulators SHINE1, CD2, and CER7 did not differ between the mutants and their wild types. However, the overall decrease in the expression of wax- and cutin-related genes in the ABA-deficient mutants correlated with a decrease in the expression of three transcription factors that are known to regulate cuticle formation: MYB41, MYB30/96, and MYB16/106 (the same tomato gene is the closest homolog of both AtMYB30 and AtMYB96, and the same is the case for AtMYB16 and AtMYB106; Supplemental Fig. S10A).

As a consequence of the ABA deficiency, the mutant genotypes are more likely to suffer from water stress than the wild-type plants (Tal, 1966), which might be expected to induce the expression of genes involved in cuticle biosynthesis (Kosma et al., 2009). However, the fact that most of the tested cuticle-related genes showed either no change in expression or decreased expression in the mutants suggested that the ABA deficiency, and not the secondary effects of water stress, was responsible for the observed differences in expression. To further test this possibility, we evaluated the expression of a set of genes that are known to be responsive to drought (Supplemental Fig. S11A). We did not detect the expression of the drought-responsive gene Solyc03g116390 (Gong et al., 2010) in emerging leaves of the mutant or wild-type genotypes, suggesting that none of the plants were water stressed. Moreover, the expression levels of two other drought-responsive genes Solyc02g063520 and Solyc02g076690 (Gong et al., 2010) were not affected or were lower in the ABA mutants than in their respective wild-type genotypes. This further indicates that the plants were not drought stressed and/or that these two genes are likely regulated by ABA. In addition, the expression of Solyc06g050520, which is known to be drought induced, and the expression of Solyc06g065970, which is suppressed by drought (Liu et al., 1998; Gong et al., 2010), were not consistently affected in the mutants. The only expression pattern that was consistent with water stress was observed in not in the case of Solyc06g050520. Taken together, these data indicate that the leaves of the wild-type and mutant genotypes that were subjected to the various cuticle analyses described above were not experiencing water stress.

To examine the effects of ABA deficiency on cuticle-related genes in leaf development in more detail, we selected a subset of the genes that were tested in emerging leaves of mature plants and evaluated their expression in the three stages of leaf development used in this study (Supplemental Fig. S12). The expression of the selected genes was lower in the mutants than in the wild-type genotypes at the small stage of development, as was previously observed (Supplemental Fig. S10). In addition, the decrease of expression of CUS1 and LACS1, which was suggested but was not statistically significant in Supplemental Figure S10, was confirmed in this subsequent experiment (Supplemental Fig. S12). Two distinct expression patterns were observed: CUS1 and CER3 had consistently lower expressions in the mutants in the three stages of leaf development, while the relative expression of CD3, GPAT6, ABCG11, LACS1, CER6, and MYB30/96 steadily increased as the leaf developed (Supplemental Fig. S12).

The expression of cuticle associated genes in 15 d post-anthesis fruits generally showed a similar pattern to that in leaves (Supplemental Fig. S10), although the reduction in their expression levels was less significant, and no difference was seen for the transcription factor genes (Supplemental Fig. S10B). Indeed, several cuticle-associated genes (CUS1, CD3, CYP77A6, GPAT6, ABCG12, and CER3) were expressed at lower levels, in the mutants compared with their respective wild types, as was the case in leaves. Quantitative PCR (qPCR) analysis of the drought-responsive genes described above showed no differences in expression between the mutants and their wild types, suggesting that the fruits did not experience water stress (Supplemental Fig. S11B).

Taken together, these gene expression analyses suggest that ABA deficiency resulted in down-regulation of genes in the cuticle biosynthesis pathway, particularly in leaves, which is consistent with the original hypothesis that ABA regulates cuticle formation during organ development independently of water stress.

Effect of ABA Application on the Expression of Cuticle-Associated Genes

The cutin and cuticular wax phenotypes were partially rescued following an application of ABA (Figs. 6 and 7). To determine the effect of such treatment on cuticle-associated gene expression, we examined the expression patterns of seven genes in the four mutant genotypes (Fig. 8) and of 16 genes in the mutants of RR genetic background (Supplemental Fig. S13) throughout leaf development. To increase the likelihood that ABA penetrated the plant surface, we dipped the leaves into an ABA solution, rather than spraying them. The expression of CER3 was strongly increased by the ABA treatment, in all genotypes at all stages of leaf development. The expression of CD3, CYP77A6, GPAT6, ABCG11, CER1, and CER6 was increased by the application of ABA to a lesser extent, while the expression of LACS1 and CER4 was not affected by the ABA treatment. The transcription factors MYB41 and MYB30/90 were strongly induced to the ABA treatment in the mutants and wild-type genotypes, whereas the ABA treatment only slightly affected MYB16/106 expression in the mutant genotypes and not at all in the wild-type genotypes. We concluded that ABA treatment induced the expression of genes contributing to several steps of the synthesis of cutin precursors (specifically, the hydroxylation of fatty acids and the production of acylglycerols), the elongation of C₁₆ and C₁₈ fatty acids into very long-chain fatty acids for wax biosynthesis, the formation of alkanes, and to the transport of cutin precursors and waxes into the apoplast.

Figure 8.

Effect of ABA treatment on cuticle-related gene expression of ABA deficient lines through leaf ontogeny. In small (A), medium (B), and large (C) leaves of 5-week-old plants. The REST-mcs software was used to normalize gene expression on RPL2 and ACTIN and to determine their relative expression using the corresponding wild types. The log₂ values of the ratios are plotted in this figure. Significant differences between the mutants and their wild types were given by the pairwise fixed reallocation randomization test incorporated in REST-mcs (*P < 0.05, **P < 0.01, and ***P < 0.001; n = 3). AC, Ailsa Craig; not(AC), notabilis in AC background (AC); flc(AC), flacca in (AC); RR, Rheinlands Ruhm; flc(RR), flacca in (RR); sit(RR), sitiens in (RR).

ABA Deficiency Affects the Water and Barrier Functions of the Cuticle

The wilting phenotype of the ABA-deficient mutants has been attributed to their inability to prevent water loss by stomatal closure (Tal, 1966). Nevertheless, water loss through the leaf cuticle was reported to be slightly higher in the three tomato mutants (Tal, 1966), and the sit leaf cuticle has been reported to be more permeable to water (Curvers et al., 2010). We compared the cuticle permeability of detached Red stage fruits, which lack stomata, from the wild-type and ABA-deficient lines, by measuring water loss under controlled temperature and humidity conditions. The stem scar was sealed to ensure that any transpirational water loss occurred through the cuticle. The mutant fruits showed greater resistance to water loss than the respective wild-type fruits, with not(AC) having an intermediate phenotype, and flc(AC), flc(RR), and sit(RR) fruits losing the least water (Supplemental Fig. S14).

DISCUSSION

ABA Regulates Cuticle Biosynthesis during Leaf Development

The key question underlying this study was whether ABA regulates cuticle formation during the course of organ development, in addition to its established role in modifying cuticle formation as a consequence of abiotic stress. We addressed this question using a range of ABA-deficient tomato mutants. The biochemical analyses indicated that ABA regulates cutin biosynthesis during leaf development and exogenous ABA application resulted in a slight but consistent increase in the levels of the predominant tomato cutin monomer, 10(9)-16,dihydroxyhexadecanoic acid, and of coumaric acids, in the two tested mutants, flc(AC) and sit(RR). We conclude that ABA is required for the synthesis of wild-type levels of cutin during leaf development. The reduced expression in the ABA mutants, compared to their respective wild-type genotypes, of a number of genes involved in cutin formation revealed that the pathway as a whole is influenced by ABA, from the formation of cutin precursors (LACS1, CD3, CYP77A6, and GPAT6) to monomer transportation (ABCG11) and polymerization (CUS1). The regulatory role of ABA on these genes was further demonstrated for CD3, CYP77A6, GPAT6, and ABCG11, whose expression was induced by the ABA treatment (Fig. 8; Supplemental Fig. S13). We note that not all cutin-associated genes showed altered expression in the ABA mutants or responsiveness to ABA. This was not unexpected, given the likely complexity in the regulatory pathways involved in cuticle formation, and future studies may uncover additional regulatory mechanisms.

ABA application did not result in consistent differences in wax composition in the two wild-type genotypes following ABA application. This result was somewhat unexpected since an ABA treatment of Arabidopsis was reported to cause an increase in cuticular alkane levels (Kosma et al., 2009). We do not believe the lack of such a response in our study was due to insufficient diffusion of ABA into the leaf since we observed a partial rescue of the wax composition phenotypes of the ABA-deficient lines, indicating that ABA is needed for cuticular wax deposition during leaf ontogeny. The ABA mutants showed an overall increase in the levels of C₂₅ to C₃₀ alkanes and of the C₂₉ to C₃₃ isoalkanes in medium and large leaves, compared to their respective wild types. In contrast, the longer-chain alkanes (C₃₁–C₃₃), which are the main wax constituents, were less abundant in the three stages of leaves of the ABA mutants, as were the amyrins. Our data suggest that during leaf development, ABA promotes very-long-chain fatty acid elongation and the accumulation of triterpenoids, while repressing isoalkane production. In contrast, Kosma et al. (2009) suggested that ABA treatment mainly triggers an increase of alkane accumulation. This discrepancy may indicate distinct effects of ABA action in leaves as a response to water stress versus during normal leaf development or differences in ABA effects between tomato and Arabidopsis.

The consequences of ABA deficiency on the profiles of the leaf cuticular waxes were complex but can be partially explained by the gene expression analysis. Notably, the expression levels of CER1, CER3, and CER6 were reduced in the ABA mutants and increased in response to the ABA application. AtCER1 and AtCER3 are known to interact to synthesize alkanes (Bourdenx et al., 2011; Bernard et al., 2012), and overexpression of AtCER1 was reported to substantially increase the levels of isoalkanes (Bourdenx et al., 2011). The decreased expression of CER1 and CER3 in the mutant leaves may therefore partly explain their abnormal levels of alkanes and isoalkanes. The reduced expression levels of CER6 in the ABA mutant leaves may underlie the decrease in the abundance of longer-chain alkanes and the increased levels of shorter-chain alkanes, as this shift in chain length suggests a reduction in very-long-chain fatty acid elongation. Indeed, CER6 is a component of the fatty acid elongase complex, which has previously been shown to be involved in the production of alkanes longer than C₃₀ in tomato leaves (Vogg et al., 2004). In summary, the data partially explain some of the differences in cutin and wax composition observed in the mutant leaves, although a broader survey of gene expression may more fully explain the compositional variations.

Our analysis also identified three transcription factors (MYB41, MYB30/96, and MYB16/106) whose expression was lower in the ABA-deficient mutants, suggesting an indirect regulation of the cuticle-associated genes by ABA. Moreover, the expression of two of these transcription factors, MYB30/96 and MYB41, was responsive to ABA treatment in all tested genotypes. However, the expression of other regulatory factors, such as CER7, CD2, and SHINE1, was unaffected, suggesting that ABA regulates cuticle formation by influencing the expression of a subset of the transcription factors associated with cuticle formation, but not the entire regulatory network. In Arabidopsis, AtMYB41, AtMYB30, AtMYB16, AtMYB106, and AtMYB96 have been shown to regulate cuticular wax formation and, with the exception of AtMYB96, to also regulate cutin synthesis (Cominelli et al., 2008; Raffaele et al., 2008; Seo et al., 2011; Oshima et al., 2013). Since the tomato gene Solyc03g116100 was identified as the closest homolog of both AtMYB30 and AtMYB96, it may regulate both leaf cutin and wax formation; however, this remains to be tested.

ABA Deficiency Has Distinctly Different Effects on Tomato Fruit and Leaf Cuticles

A major conclusion from our study was that the fruit cuticle was not as affected by ABA deficiency as the leaf cuticle: No ultrastructure abnormalities were detected, and cuticle-related gene expression did not differ substantially from wild-type levels. TEM analysis indicated that the cuticles of expanding ABA-deficient mutant fruits were thicker than those of the wild-type genotypes of the same size. However, since the mutant fruits did not expand to the same extent as those of wild-type fruit, our comparison of similarly sized fruits was likely of younger wild-type and older mutant fruits. The latter therefore had more time to deposit a cuticle, hence the thicker appearance. However, light microscopy imaging and cutin biochemical analysis showed that the mutant mature fruits tended to accumulate more cuticle than the wild-type fruits. In particular, the mutant fruit cuticles tended to penetrate deeper into the subepidermal cell layers (Supplemental Fig. S5). This phenotype may be related to the observation that the mutant fruits generally lost less water than those of their respective wild types (Supplemental Fig. S14). These results are contrary to what was seen for the leaf cuticle of the mutants, which had thinner cuticles (Fig. 4) and increased permeability (Tal, 1966). Chemical analysis of fruit cuticles also revealed opposite trends for a range of cutin and cuticular wax components, including examples with higher or lower levels in the mutant fruits than in the respective wild types. Notably, these trends were reversed for the leaf cuticles, highlighting the organ-specific regulation of cuticle formation by ABA.

The lack of a substantially altered fruit cuticle composition in the mutants may be considered counterintuitive, given that the reduction in ABA levels was greater in fruit than in leaves (Fig. 1) and that the severity of the shoot phenotypes of the ABA mutants has been correlated with the magnitude of ABA reduction (Tal and Nevo, 1973; Taylor and Tarr, 1984). This apparent contradiction suggests that cuticle formation in fruits is less correlated with ABA levels than it is in leaves, and indeed, the levels of ABA in Pk stage fruits were substantially lower than those of leaves. It may be that fruit cuticle does not need to be as dynamically regulated by water stress as the leaf cuticle, since the surface/volume ratio of leaves is far higher than that of fruits and the leaves have stomata, both of which are factors that make leaves more prone to desiccation. Moreover, fruit cuticles of various species are, on average, thicker than leaf cuticles, but are also more permeable (Becker et al., 1986; Araus et al., 1991; Schreiber and Riederer, 1996), suggesting that limiting water loss is more important in leaves than in fruits. Indeed, it has been suggested that transpirational water loss may be necessary for fruit growth (Lee, 1989; Schreiber and Riederer, 1996). Given these factors, it might be expected that water stress would have less of an effect on fruit cuticles than on leaf cuticles. This hypothesis has yet to be tested in fleshy fruit, although water stress was reported to result in an increase in total cuticular wax levels of cotton (Gossypium hirsutum) leaves and bracts, but not bolls (Bondada et al., 1996), illustrating the potential for differences in cuticle responses to water stress. We propose that the leaf cuticle biosynthesis pathway is more sensitive to changes in ABA levels than the analogous pathway in fruits.

Previous studies have shown that environmental stresses, such as drought, increase ABA levels, resulting in the increased production of cuticular waxes (Kosma et al., 2009; Seo et al., 2009, 2011; Wang et al., 2012). In this study, we conclude that ABA regulation of cuticle formation is an intrinsic aspect of leaf development. Specifically, our data suggest that ABA regulates the structure of the cuticle, the amount of cutin, and the composition of cutin and waxes during tomato leaf development, while abiotic stress exerts a secondary level of ABA-mediated control.

MATERIALS AND METHODS

Plant Materials

The tomato wild-type genotypes AC and RR and their mutants, not, flc, and sit, were obtained from the University of California Davis/C.M. Rick Tomato Genetics Resource Center (http://tgrc.ucdavis.edu/). Plants were grown in a greenhouse under 16 h of light at 22°C during the day and 19°C during the night with 150 ppm of 15-5-15 (N-P-K), 4Ca 2Mg fertilizer (Jack’s Professional LX; J. R. Peters) being applied twice a day. The leaves of 5-week-old plants provided the small, medium, and large developmental stages, with “large” being the youngest fully expanded leaf, “medium” being leaves halfway through expansion, and “small” being leaves starting their expansion. Emerging leaves were very young leaves (length <2 cm) harvested from mature plants. Ripening stages were determined as follows: MG stage, the fruit has reached its full size and the surface and interior tissues are entirely green; Pk stage, 30% to 60% of the fruit surface is pink or red in color; Red stage, more than 90% of the surface is red; and Red Ripe, the fruit surface is entirely red.

ABA Measurements

Plant material was flash frozen and stored at −80°C before being ground into a fine powder in liquid nitrogen. Around 200 to 300 mg of this powder was mixed with 500 µL of 1-propanol:water:concentrated HCl (2:1:0.002). Aliquots of 80 ng of d₆-ABA (Toronto Research Chemicals) were added to each sample as an internal standard. The samples were shaken at 4°C for 30 min, centrifuged 5 min at 13,000 rpm, and 800 µL of dichloromethane was added to the supernatant. The shaking and centrifugation steps were repeated as above and the lower organic phase was then collected and dried at 35°C under a gentle stream of nitrogen gas. The extracts were resuspended in 100 µL of methanol and analyzed using a triple-quadrupole liquid chromatography-tandem mass spectrometry system (Quantum Access, Thermo Scientific) as described by Thaler et al. (2010) at the Chemical Ecology Group Core Facility, with the selected reaction monitoring of compound-specific parent/product ion transitions ABA 263 → 153; d₆-ABA 269 → 159.

Exogenous ABA Treatments

For the biochemical analyses, 3-week-old tomato seedlings were sprayed with 100 µm (±)-ABA (Sigma-Aldrich) in 0.1% ethanol every 5 d for 2 weeks, to the extent that entire leaf adaxial surfaces were covered, as described by Achuo et al. (2006) and Curvers et al. (2010). Control plants were similarly sprayed with 0.1% ethanol. Cutin and wax samples were extracted as described in the “cutin monomer and wax analysis” section from the youngest fully expanded leaf of 5-week-old plants, using the two leaflets closest to the distal leaflet. The six biological replicates per genotype of cutin analysis were constituted of leaflets from independent plants. Each biological replicate of the wax analysis were constituted of the leaflets of at least three independent plants. For the statistical analyses of these experiments, the Levene test was first used to test for equal variances. If the null hypothesis was not rejected at α = 0.05, comparison of means for all pairs using Tukey-Kramer’s HSD test was performed. If the variances were unequal for α = 0.05, nonparametric comparisons for each pair using Wilcoxon method were done.

For the gene expression experiments, 5-week-old tomato plants were submerged for few seconds into 100 µm (±)-ABA (Sigma-Aldrich) in 0.1% ethanol or into a 0.1% ethanol solution for controls. The treatment was repeated 3 h later. An hour after the second treatment, the leaves were harvested and flash-frozen in liquid nitrogen. The samples were processed according to the protocol described in “Gene Expression Analysis.”

Gene Expression Analysis

Total RNA samples were extracted from emerging, small, medium, and large leaves (as described in “Plant Materials”) and from the fruit pericarp using TRIzol reagent (Life Technologies) and following the manufacturer’s directions. cDNAs were generated using SuperScript II reverse transcriptase (Life Technologies) according to the manufacturer’s instructions. A Life Technologies/ABI Viia7 instrument at Cornell’s Institute of Biotechnology Genomics Facility was used to generate the data. A list of the primers used in this study is given in Supplemental Table S3, together with gene identification numbers (SGN; http://solgenomics.net/). The REST-MC software was used to determine gene expression levels and to apply its pairwise fixed reallocation randomization statistic test on the results (Pfaffl et al., 2002).

Water Loss Measurements

Red fruits were harvested, their pedicel scars were covered with high vacuum grease (silicone lubricant, Dow Corning) and they were placed in an incubation chamber at 35% humidity and 22°C in the dark. Fruit surface area was determined by measuring fruit diameter. The fruits were weighed every day, and decreases in weight were considered to be equivalent to the amount of water loss.

Light Microscopy

Tissue fixation and embedding were performed as by Buda et al. (2009). A saturated solution of Oil Red O (Alfa Aesar) in isopropyl alcohol was diluted 3:2 with distilled water, mixed well, left for 30 min at room temperature, and then filtered with a syringe filter of 0.8/0.2 µm pore size (Acrodisc syringe filters; Pall Corporation) to remove precipitates. The solution was then applied on 6 µm sections, generated as described in Yeats et al. (2012), for 30 min at 100% humidity. The slides were rinsed with a series of 50, 30, 22, 15, and 8% isopropyl alcohol and mounted in the last dilution. The stained slides were viewed on an AxioImager A1 microscope (Zeiss) using Zeiss EC-Plan NeoFluar 40×/0.75 dry objective, a Zeiss AxioCam MRc color video camera, and Zeiss Axio Vs40 4.6.3.0 software. Photoshop CS5 software (Adobe) was used to adjust the image levels and color balance.

TEM

Leaf material was harvested from the youngest fully expanded leaf of 6-week-old plants, using the two leaflets closest to the distal leaflet. Leaflets were cut in pieces of ∼5 × 5 mm, and for the fruit pericarp samples, ∼3 × 3-mm cubes of tomato pericarp were excised from pericarp sections using a razor blade. The harvested tissue was placed in a fixative consisting of 1% glutaraldehyde in 0.1 m sodium cacodylate buffer (pH 7.2) for 60 min on ice. The samples were washed three times in wash buffer (0.1 m sodium cacodylate buffer, pH 7.2) for ∼5 min each and subsequently postfixed in 0.5% OsO₄ in 0.1 m sodium cacodylate buffer (pH 7.2) for 60 min on ice. The washing step was then repeated. Samples were dehydrated with a series of acetone dilutions (10, 30, 50, 70, 90, and 100%) for 20 min each, with the 100% solution being replaced by fresh acetone after 10 min. Samples were then infiltrated by immersion in 75% acetone-25% Spurr’s Low Viscosity Resin (Electron Microscopy Sciences) overnight, followed by 2 h immersions in 50% resin/50% acetone and then 75% resin/25% acetone. Finally, the samples were infiltrated with 100% resin overnight. Polymerization of the resin was performed at 60°C for 10 h. Sixty-nanometer sections were cut with a diamond knife on a Reichert Jung Ultracut E ultramicrotome and collected on Formvar-coated nickel grids. The grids were stained in uranyl acetate/lead citrate, washed with deionized water, and dried. The sections were viewed with a Zeiss Libra 120 transmission electron microscope (Zeiss) at 120 kV.

The cuticle and the underlying cell wall were visually distinguished based on differences in their optical contrast and by looking at multiple images of the same sample. Measurements of cuticular thickness were based on the TEM micrographs of leaflets from three independent plants per genotype and of three to six fruits per genotype. At least three measurements per leaflet were taken. Wild-type and mutant fruits of ∼14.4-, ∼13.5-, and ∼12.7-mm diameters were compared using from 10 to 22 measurements per size and genotype. Student’s t test was used to determine statistical significance of the results.

Cutin Monomer and Wax Analysis

Cutin and waxes were extracted from the youngest fully expanded leaf of 6-week-old plants, using the two leaflets closest to the distal leaflet. For the leaf samples, cuticular waxes were extracted by dipping the leaflets in chloroform containing 50 or 100 µg of tetracosane, for 30 s. Cutin samples were isolated as described in Li-Beisson et al. (2013). For the fruit samples, the fruit cuticle was isolated as described in Isaacson et al. (2009), with the peels being incubated at 42°C instead of 35°C. Isolated cuticles were washed in distilled water, air dried, and rinsed three times with chloroform to remove cuticular waxes and contaminating lipid residues. Fruit cuticular waxes were extracted by dipping intact fruits in two successive baths of chloroform for 1 min each, the first bath containing 50 or 100 µg of tetracosane as internal standard. The baths were pooled, anhydrous sodium sulfate was added to remove any trace of water, and the solutions were passed through filter paper. Leaf and fruit cutin samples were depolymerized as described in Li-Beisson et al. (2013). Cutin monomers and cuticular waxes were derivatized as previously described (Li-Beisson et al., 2013)

The cutin and wax extracts were resuspended in 100 µL chloroform and injected into a gas chromatography (GC) 6850 GC system (Agilent) with a HP-1 (30 m × 320 µm × 0.1 µm) column (Agilent). The wax samples were injected with an oven temperature of 50°C, and the temperature was held constant for 2 min. The temperature was then increased by 40°C per min up to 200°C, and then by 4°C/min up to 235°C and held for 15 min, prior to an increase of 10°C per min up to 315°C, which was then maintained for 15 min. The cutin monomer samples were injected with an oven temperature of 50°C, which was then held for 2 min. The temperature was then increased by 40°C per min up to 120°C and held for 2 min prior to an increase of 10°C per min up to 320°C, which was then maintained for 15 min. Compounds were identified as described in Yeats et al. (2012).

Statistical Analyses

Unless stated otherwise in the text, the statistical tests were performed using JMP software, by first performing an ANOVA on the mutants and their wild type. If the variances were equivalent for α = 0.05, a pairwise comparison test using Dunnett’s method (using a control group) was performed. If the variances were unequal for α = 0.05, the nonparametric multiple comparison test “Steel with control” was used.

Accession Numbers

Sequence data from this article can be found in the SGN data libraries under the accession numbers Solyc11g006250 (CUS1), Solyc08g008610 (BDG), Solyc08g081220 (CD3), Solyc05g055400 (CYP77A6), Solyc01g094700 (GPAT4), Solyc09g014350 (GPAT6), Solyc01g079240 (LACS2), Solyc06g062600 (HTH), Solyc03g005320 (FDH), Solyc03g019760 (ABCG11), Solyc11g065350 (ABCG12), Solyc03g065250 (CER1), Solyc03g117800 (CER3), Solyc06g074390 (CER4), Solyc02g085870 (CER6), Solyc01g079240 (LACS1), Solyc01g011430 (WSD1), Solyc05g047420 (CER7), Solyc01g091630 (CD2), Solyc03g116610 (SHN1), Solyc10g005460 (MYB41), Solyc03g116100 (MYB30/96), Solyc02g088190 (MYB16/106), Solyc06g050520 (DREB2A), Solyc03g116390, Solyc02g076690, Solyc02g063520, Solyc06g065970, Solyc05g054480 (ACTIN), Solyc10g006580 (RPL2).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Whole-plant phenotype of the ABA-deficient mutants.

• Supplemental Figure S2. ABA-deficient mutant fruit phenotype.

• Supplemental Figure S3. TEM micrographs of the cuticle of small expanding fruit.

• Supplemental Figure S4. TEM micrographs of the cuticle of the abaxial epidermis of fully expanded leaflets.

• Supplemental Figure S5. Light microscopy micrographs of fruit cuticles.

• Supplemental Figure S6. Analysis of ABA-deficient mutant fruit cutin.

• Supplemental Figure S7. Analysis of fruit cuticular waxes in ABA-deficient mutants.

• Supplemental Figure S8. Amounts of small leaf cuticular waxes from ABA-deficient mutants.

• Supplemental Figure S9. Amounts of medium leaf cuticular waxes from ABA-deficient mutants.

• Supplemental Figure S10. Cuticle related gene expression in developing leaves and fruits of ABA-deficient lines.

• Supplemental Figure S11. Expression of drought-responsive genes detected by qPCR.

• Supplemental Figure S12. Expression of cuticle-related genes in ABA-deficient lines in three stages of leaf development.

• Supplemental Figure S13. Response of cuticle-related genes to ABA treatment of ABA-deficient lines through leaf ontogeny.

• Supplemental Figure S14. Analysis of transpirational water loss from the ABA-deficient lines and their respective wild types.

• Supplemental Table S1. Fruit and leaf cuticle thickness measured on TEM images.

• Supplemental Table S2. MG fruit cuticle penetration in between subepidermal cells and thickness, measured on light microscopy images.

• Supplemental Table S3. IDs and specific qPCR primer pairs of the tomato genes analyzed in this study.

Glossary

Pk

Pink stage fruit

MG

Mature Green stage fruit