Pleiotropic phenotypes of the sticky peel mutant provide new insight into the role of CUTIN DEFICIENT2 in epidermal cell function in tomato.

Plant epidermal cells have evolved specialist functions associated with adaptation to stress. These include the synthesis and deposition of specialized metabolites such as waxes and cutin together with flavonoids and anthocyanins, which have important roles in providing a barrier to water loss and protection against UV radiation, respectively. Characterization of the sticky peel (pe) mutant of tomato (Solanum lycopersicum) revealed several phenotypes indicative of a defect in epidermal cell function, including reduced anthocyanin accumulation, a lower density of glandular trichomes, and an associated reduction in trichome-derived terpenes. In addition, pe mutant fruit are glossy and peels have increased elasticity due to a severe reduction in cutin biosynthesis and altered wax deposition. Leaves of the pe mutant are also cutin deficient and the epicuticular waxes contain a lower proportion of long-chain alkanes. Direct measurements of transpiration, together with chlorophyll-leaching assays, indicate increased cuticular permeability of pe leaves. Genetic mapping revealed that the pe locus represents a new allele of CUTIN DEFICIENT2 (CD2), a member of the class IV homeodomain-leucine zipper gene family, previously only associated with cutin deficiency in tomato fruit. CD2 is preferentially expressed in epidermal cells of tomato stems and is a homolog of Arabidopsis (Arabidopsis thaliana) ANTHOCYANINLESS2 (ANL2). Analysis of cuticle composition in leaves of anl2 revealed that cutin accumulates to approximately 60% of the levels observed in wild-type Arabidopsis. Together, these data provide new insight into the role of CD2 and ANL2 in regulating diverse metabolic pathways and in particular, those associated with epidermal cells.

Plants are continually exposed to environmental changes that impact their fitness and survival. Changes in light intensity, light quality, temperature, and water availability occur on a daily basis and insect pests and pathogens pose a constant threat. Consequently, plants have evolved a suite of physical and chemical adaptations and defenses against abiotic and biotic stresses that have facilitated their colonization of diverse environments. The surface properties of plants are crucial to their successful adaptation to stress. The epidermal cell layer and associated structural appendages such as trichomes, together with the cuticle form the primary physical and chemical barriers that protect plants against multiple stresses (Glover, 2000;Sieber et al., 2000;Schilmiller et al., 2008;Javelle et al., 2011b). Furthermore, the surface properties of plants are dynamic and can change in response to stress, therefore influencing the physiology of the plant (Lu et al., 1996;Ingram et al., 2000;Abe et al., 2003;Ingram, 2008;Kosma et al., 2009;Wang et al., 2011).
The metabolism of epidermal cells is programmed for the synthesis of lipids that form the cuticle, a heterogeneous lipid-based barrier comprised of cutin, intra, and epicuticular waxes and polysaccharides that covers the aerial surfaces of all terrestrial plants. In addition to serving as the principal barrier to water loss (Kolattukudy, 1980;Riederer and Schreiber, 2001;Pollard et al., 2008) the cuticle provides structural support, possesses antiadhesive properties that limit pathogen infection, resists insect feeding and oviposition, and has reflective properties that reduce heat load and limit the effect of UV radiation (Bargel et al., 2006). Furthermore, the chemical properties of the cuticle are dynamic with both cuticle deposition and wax composition altering in response to water stress and abscisic acid Wang et al., 2011). The epidermal and subepidermal cells are also sites of phenylpropanoid biosynthesis, and in particular the synthesis and accumulation of flavonoids and anthocyanins (Martin and Gerats, 1993;Hichri et al., 2011;Matas et al., 2011). Anthocyanins protect plants through their UV radiation absorbing properties and their synthesis is influenced by stress and can vary with light intensity and wavelength (Beggs et al., 1987;Li et al., 1993;Christie et al., 1994;Landry et al., 1995;Fuglevand et al., 1996;Steyn et al., 2002). Together, these studies highlight the importance of the epidermal cell layer in the synthesis of compounds important for stress adaptation in plants that likely contributed to colonization of the terrestrial environment by early land plants (Edwards et al., 1996;Cooper-Driver, 2001).
Many enzymes, transporters, and regulatory factors required for the biosynthesis and deposition of cuticular lipids have been identified and characterized in Arabidopsis (Arabidopsis thaliana; Li et al., 2007Li et al., , 2010aPanikashvili et al., 2007Panikashvili et al., , 2009Samuels et al., 2008;Li-Beisson et al., 2010;McFarlane et al., 2010;Seo et al., 2011;Wu et al., 2011). Tomato (Solanum lycopersicum) fruit also serve as a powerful system for investigating the physical properties and biosynthesis of the cuticle fostered in large part by the abundance of cuticular lipid deposited during fleshy fruit development, its astomatous nature, and relevance in fruit cracking and postharvest storage Scott, 1997, 1998;Matas et al., 2004;Bargel and Neinhuis, 2005). Genomics-and proteomics-based approaches have identified genes and proteins preferentially expressed in the fruit peel, and perturbation of several genes through mutagenesis has revealed altered fruit cuticle phenotypes (Vogg et al., 2004;Hovav et al., 2007;Mintz-Oron et al., 2008;Isaacson et al., 2009;Yeats et al., 2010;Matas et al., 2011).
Fruit of the sticky peel (pe) mutant of tomato have a rubbery surface texture rather than the typical smooth surface associated with wild-type fruit, rendering fruits sticky to the touch (Butler, 1952). In addition, pe fruits have a highly glossy fruit surface, a phenotype recently associated with cutin deficiency (Isaacson et al., 2009). In this study, a combination of microscopy and chemical analysis revealed that pe fruit are cutin deficient and have an altered wax profile. In addition, several phenotypes attributed to altered epidermal cell function are apparent in pe, including cutin deficiency and altered wax deposition in leaves, together with increased cuticular permeability, lower trichome density, and reduced anthocyanin accumulation. Genetic mapping indicated that pe encodes a new allele of CUTIN DEFICIENT2 (CD2), which encodes a member of the class IV homeodomain-Leu zipper (HD-ZIP IV) family, and was previously only associated with a fruitspecific reduction in cuticle biosynthesis (Isaacson et al., 2009). In addition, we show that mutation in a CD2 homolog at the anthocyaninless2 (anl2) locus of Arabidopsis also causes a cutin-deficient phenotype in rosette leaves. Together, these data identify additional roles for CD2 and ANL2, defining a regulatory link between cuticle and flavonoid biosynthesis, two pathways that operate within epidermal cells that are critical for plant responses to stress.

Introgression of the pe Allele into the Ailsa Craig Genetic Background
At the initiation of this research, the genetic background of the pe mutant was unknown. Therefore, we initiated a backcross strategy to introgress the pe mutation into the Ailsa Craig (AC) genetic background, a cultivar previously selected for introgression of varied morphological mutants of tomato (Darby et al., 1978). The mutant and wild-type plants used in the following experiments were derived from a BC 3 F 2 population that we estimate contains approximately 93.75% of the AC genetic background. Therefore, the influence of the genetic background on the data presented is likely to be minimal.

Altered Morphology and Trichome Chemistry of the pe Mutant
In addition to the previously documented sticky peel phenotype of pe fruit (Butler, 1952), the pe mutant exhibits a short stature with pale-green leaves and stems (Fig.  1A). Five-week-old wild-type AC plants had an average dry weight of 6.45 g whereas the dry weight of pe plants was 3.86 g. Glandular trichomes are epidermal cell appendages that are prevalent on the surface of tomato plants and synthesize a suite of specialized metabolites known, or hypothesized to play a role in plant defense responses against pests and pathogens (Schilmiller et al., 2008). Type VI glandular trichomes are the most abundant glandular trichomes on the tomato leaf surface (Kang et al., 2010) and these are reduced greater than 3-fold in pe (Fig. 1, B-D). The type VI glandular trichomes of cultivated tomato synthesize a mixture of mono-and sesquiterpenes of which b-phellandrene predominates (Schilmiller et al., 2009). The volatile terpene levels in leaf dips of AC and pe were compared (Supplemental Fig.  S1). Consistent with the reduced abundance of type VI glandular trichomes on the surface of pe leaves, volatile terpene levels are also significantly reduced to approximately 16% of that observed in wild type.

Reduced Anthocyanin Accumulation in pe
To investigate the pale phenotype of pe in more detail, chlorophyll and anthocyanin levels were determined in fully expanded and meristematic leaves. Chlorophyll levels in AC and pe are not significantly altered but anthocyanin content is reduced by approximately 85% in pe (Table I). The anthocyanins represent an end point of the phenylpropanoid pathway and are known to accumulate in epidermal and subepidermal cell layers (Martin and Gerats, 1993;Hichri et al., 2011). The distribution of anthocyanins within the stems of AC and pe was compared by confocal laser-scanning microscopy and stems of the high pigment-1 (hp-1) and entirely anthocyaninless (ae) mutants were included as positive and negative controls, respectively ( Fig. 1, E-I). Images reveal two main patterns of anthocyanin distribution in AC, with accumulation observed in the epidermal and subepidermal cells and also surrounding the vasculature (Fig. 1F). In contrast anthocyanin accumulation is greatly reduced in stems of pe (Fig. 1G).

Lignin Content Is Not Altered in pe
Like anthocyanins, lignin also represents an end point of the phenylpropanoid pathway (Vogt, 2010). Reduced lignin synthesis in plants is associated with stunted growth, mediated by increased salicylic acid levels (Brown et al., 2001;Li et al., 2010b;Gallego-Giraldo et al., 2011). To investigate whether the semidwarf phenotype of pe is related to reduced accumulation of lignin, the total lignin content in leaves and stems of AC and pe was analyzed. The total percent acetyl bromide soluble lignin (% ABSL) as well as individual lignin monomers of AC and pe leaves showed no significant differences between the two genotypes (Supplemental Table S1).

The pe Fruit Cuticle Has Altered Physical and Chemical Properties
Fruits of the pe mutant are highly glossy and are sticky to the touch when compared to wild type (Fig. 2, A and B;Butler, 1952). In addition, we observed that ripe pe fruits are generally crack resistant. The average Young's elastic modulus (Y) is approximately 2-fold greater than wild type, indicating increased stiffness of the pe fruit peel (Fig. 2C). Glossiness and increased Figure 1. Phenotypic variation between AC and pe. A, Whole-plant phenotype of AC and pe. B and C, Light micrographs of trichomes on leaves of AC and pe. Type VI glandular trichomes are indicated by arrowheads, scale bar = 2 mm. D, Density of glandular type VI trichomes on the adaxial leaf surface of AC and pe. Data are presented as the mean of n = 8 6 SE on adaxial side of AC and pe leaves. Asterisks denote significant differences (***, P , 0.001) as determined by Student's t tests. E, Comparison of anthocyanin accumulation in the stems of AC and pe. Stems of the hp1 and ae mutants are shown for comparison. F to I, Autofluorescence of anthocyanins in AC, pe, hp1, and ae visualized by confocal laser-scanning microscopy (scale bar = 200 mm). stiffness in tomato fruit peels is associated with cutin deficiency (Isaacson et al., 2009). Scanning electron microscopy (SEM) images of pe fruit indicated reduced cuticle deposition compared to AC fruit (Fig. 2, D-G). Staining of cryosections of the fruit peel of AC and pe with the lipid reactive stain, Sudan IV confirmed reduced cuticle lipid deposition in the pe mutant (Fig. 2, H and I).
The chemical composition of pe fruit cuticles is significantly altered compared to that of AC fruits. The cutin monomer load (mass/area) in ripe pe fruits is approximately 2% of that in AC and 6.5% of AC levels in green fruits (Table II). Furthermore, the large reduction in cutin monomer content is observed for all hexadecanoic-acid-derived cutin monomers, together with cis-and trans-coumaric acid, products of the phenylpropanoid pathway. The notable exception to the general reduction of cutin monomers in pe is the abundance of the hexadecanoic acid itself, which has slightly increased abundance in mutant fruit, particularly at the green stage of development.
Altered glossiness of plant surfaces is often associated with changes in wax composition (Chen et al., 2003;Aharoni et al., 2004;Bourdenx et al., 2011). The total wax load is not significantly different between AC and pe fruits at either the green or ripe stages of development (Supplemental Table S2). However, the cutin monomer to wax ratio, signifying relative proportion of wax within the cutin matrix, falls dramatically from approximately 75:1 (1,086.5:14.42 mg cm 22 ) in ripe fruits of AC to 1.6:1 (26.1:15.7 mg cm 22 ) in pe (Table  II; Supplemental Table S2). Also, the composition of individual wax components is significantly altered. For example, alkanes constitute approximately 40% of the total wax load in AC fruits but between 55% and 62% in pe fruits due largely to increases in C31 to C33 alkanes at both the green and ripe stages of fruit development ( Fig.  3; Supplemental Tables S2 and S3). Fatty acids constitute a relatively minor component of the total wax load ranging from 2% to 5% in both AC and pe and there are no significant differences at the green stage, although ripe fruits of the mutant have a significantly higher fatty acid content due mainly to a 2-fold increase in hexadecanoic acid. In contrast to the general increase in the levels of long-chain alkanes and fatty acids in the epicuticular wax of pe, the fatty alcohols (alkanols and alkenols) together with coumarates, sterols, and triterpenoids show a general reduction in pe compared to AC. In particular, the amyrins, which accumulate to similar levels as the alkanes in wild-type fruit, are reduced by approximately 86% in green fruit of pe ( Fig.  3; Supplemental Tables S2 and S3). Alkanes are well known to constitute much of the epicuticular wax, whereas triterpenoids are almost exclusively intracuticular waxes (Vogg et al., 2004). Thus the reduction in triterpenoids but not alkanes with the loss of cutin is consistent with the differential localization of these waxes.

Altered Surface Chemistry and Increased Water Conductance in pe Leaves
To investigate whether the altered surface chemistry observed in pe is restricted to fruit or is a more general phenotype of the mutant, the cutin and wax composition of the leaves of AC and pe were compared. As was observed in fruit, the major monomer identified in tomato leaf cutin is also 9(10),16-dihydroxyhexadecanoic acid, constituting between 20% and 25% of the total cutin monomer load (Fig. 4). Additional monomers identified include several fatty acids and the phenylpropanoids cisand trans-coumaric acid together with caffeic acid. The overall cutin monomer load in leaves of pe is 54% of that in AC, with an approximately 5-fold reduction in 9(10), 16-dihydroxyhexadecanoic acid and significant reductions in additional monomers (Fig. 4A).
Tomato leaf cuticular wax components identified in this study comprise n-alkanes, isoalkanes, anteiso alkanes, petacyclic triterpenoids, sterol derivatives, and fatty acids. Among these, alkanes and branched alkanes constituted the major fraction (up to 90%) of all the identified wax composition in wild-type leaves. The total leaf wax load is reduced by 30% in the mutant. This is mainly attributed to significant reduction in alkanes. Interestingly all the straight-chain alkanes in the pe mutant were reduced by approximately 50% compared to wild type, whereas branched-chain alkanes are essentially unaltered (Fig. 4B). As both wax and cutin loads decrease in leaf, unlike in fruit (Table  II; Supplemental Table S2), the cutin monomer to wax ratio changes only fractionally in the pe mutant, from 2.3 (5.85:2.53 mg cm 22 ) to 1.8 (3.20:1.80 mg cm 22 ; Fig.  4). This clearly shows a differential regulation of straight-chain and branched hydrocarbons in fruit and leaf cuticular wax, suggesting they may be independently regulated.
Cuticular waxes play a pivotal role in limiting water loss in plants. The levels of alkanes within cuticular waxes are often positively correlated with increased Table I. Anthocyanin and chlorophyll content of AC and pe leaves Data represent the mean of n = 5 6 SE. Asterisks denote significant differences between genotypes of the same developmental stage (***, P , 0.001) as determined by Student's t tests. Anthocyanin content (AU 535 nm g 21 FW) 12.4 6 2.1*** 1.9 6 0.6 13.1 6 1.0*** 5.4 6 0.63 Chlorophyll content (mg mL 21 ) 54.8 6 2.0 53.3 6 3.1 67.3 6 0.6 66.4 6 0.8 resistance to cuticular water loss and an increase in cuticular wax abundance, particularly alkanes can occur during water stress (Grncarevic and Radler, 1967;Kosma et al., 2009;Bourdenx et al., 2011;Seo et al., 2011). Furthermore, altered cutin composition may disrupt the intermolecular packing of waxes within the cutin matrix, although this relationship remains unclear (Pollard et al., 2008;Schreiber, 2010;Buschhaus and Jetter, 2011). Given the reduction in alkane and cutin monomer content in the leaves of the pe mutant, the rate of leaf water loss from mutant leaves was determined. The pe mutant does not have an obviously wilty phenotype however leaf water conductance in intact pe leaves was approximately 3-fold higher than in AC (Fig. 5A). Similarly, an increase in the rate of chlorophyll leaching was observed in pe leaves (Fig.  5B), a phenomena previously shown to be associated with cuticle permeability . Previous research indicated that defects in cutin biosynthesis can lead to alterations in stomatal structure including impaired development of cuticular ledges that lie between adjacent guard cells (Li et al., 2007). Scanning electron micrographs of the leaf surface did not reveal any structural differences in the stomata of the pe mutant ( Fig. 5C). Furthermore, the average stomatal density on the abaxial leaf surface in wild-type leaves (30 in 0.55 mm 2 ) is higher than in pe leaves (21 in 0.55 mm 2 ; Fig. 5D). Together, these data suggest that enhanced rate of water loss from the leaves of pe mutant is due to cuticular water loss and not to altered stomatal structure or density.

Root Suberin and Wax Composition Are Not Dramatically Altered in pe
The altered cutin and wax composition observed in the leaves and fruits of pe suggested the possibility of a general perturbation in synthesis of fatty-acid-derived polymers. This was investigated by determining the suberin and wax content in roots of pe. The aliphatic components that account for nearly 90% of the total suberin, together with the aromatic components are not significantly altered in pe (Supplemental Fig. S2A). In addition, the overall root wax load is largely unaltered in pe although small increases in abundance of b-C22:0 monoacyl-glycerol, C29:2 sterol, and C18:0 and C22:0 primary alcohols are observed (Supplemental Fig.  S2B).

Mapping of the pe Locus and Candidate Gene Identification
The pe locus was provisionally mapped to chromosome 1 of the classical genetic map of tomato (Mutschler et al., 1987). An F2 mapping population segregating for the pe mutant allele was generated through crosses between tomato (pe/pe; LA2467) and Solanum pimpinellifolium (PE/PE; LA1589). The mutant plants were pale green in color and this phenotype cosegregated with the cutin-deficient, sticky and glossy fruit phenotype in the F2 population. The F2 population was genotyped with chromosome 1 molecular markers, revealing that the pe locus is located within a 424-kb interval Table II. Cutin monomer composition in fruit cuticles of AC and pe in mg cm 22 (%) Data represent the mean of n = 5 6 SE. Asterisks denote significant differences between genotypes of the same developmental stage (*, P , 0.05; ***, P , 0.001) as determined by Student's t tests. ND, Not detected.
To confirm that a mutation in the CD2 gene is responsible for the pe mutant phenotype, virus-induced gene silencing (VIGS) experiments were performed using two constructs targeting separate regions of the CD2 gene. CD2-silenced lines showed a 20% to 30% reduction in height compared with TRV2 empty vector control lines and leaves showed sectors of pale coloration characteristic of the pe mutant (Fig. 6, B-D). Anthocyanin levels were reduced in the CD2-silenced lines by approximately 30% to 35% (Fig. 6E). The major leaf cutin monomer 9 (10),16-dihydroxyhexadecanoic acid was reduced by 50% in the silenced lines and there was a concomitant increase in the leaf water conductance (Fig. 6, F and G). Together, these data indicate that the pe mutant phenotype is caused by a mutation in CD2.

Mutation at the anl2 Locus Leads to Cutin Deficiency
Phylogenetic analysis indicated that CD2 is closely related to a number of HD-ZIP IV proteins including several that impact epidermal cell development and cuticle biosynthesis (Nakamura et al., 2006;Javelle et al., 2010Javelle et al., , 2011a Fig. 7A). HD-ZIP IV genes are typically expressed in epidermal cells and expression analysis indicated that CD2 transcripts are enriched in stem peels, which are comprised of a mixture of epidermal and subepidermal cells, compared to the levels observed in whole stem and stem core with the peel removed (Fig. 7B).
ANL2 is the closest Arabidopsis homolog to CD2 (Fig. 7A) and is also preferentially expressed in epidermal cells (http://efp.ucr.edu/cgi-bin/absolute.cgi). Furthermore, both pe and anl2 mutants have reduced anthocyanin accumulation (Kubo et al., 1999; Fig. 1). These similarities led us to hypothesize that anl2 mutants would also be cutin deficient. This was confirmed in rosette leaves of anl2, which exhibit approximately a 40% reduction in cutin monomer load together with a 25% reduction in the alkane load of the cuticular waxes compared to Columbia-0 (Fig. 8).

Characterization of pe Identifies Additional Roles for CD2
Phenotypic characterization indicates that the pe mutant has altered fruit surface chemistry, leading to reduced cutin levels and a modified surface wax composition ( Fig. 2; Table II; Supplemental Table S2). The altered chemical profile leads to a change in the physical properties of the fruit peel including increased (1) alkanes, (2) branched chain alkanes, (3) fatty acids, and (4) triterpenoids. Inset graphs indicate total cutin and wax loads in AC and pe. Values represent mean values (n = 5) 6 SE. Asterisks denote significant differences (*, P , 0.05; **, P , 0.01; ***, P , 0.001) as determined by Student's t tests.
glossiness and a higher Young's elastic modulus (Fig. 2, A-C). These phenotypes are identical to those previously described in cd mutants of tomato where cutin deficiency was correlated with altered biomechanical and structural properties together with enhanced susceptibility to microbial infection (Isaacson et al., 2009). A combination of genetic mapping and gene cloning revealed that pe represents a new mutant allele of CD2 that truncates the predicted protein by 160 amino acids and therefore likely represents a null allele (Fig. 6A).
Previously, the cd2 phenotype was associated with fruit-specific alteration in cuticle biosynthesis (Isaacson et al., 2009). However, as shown in this study, many additional phenotypes are evident in pe including a semidwarf stature, reduced anthocyanin accumulation, altered cutin and wax deposition in leaves, increased permeability of the leaf cuticle, and reduced trichome density together with lower volatile terpene production (Figs. 1, 4, and 5; Table I; Supplemental Fig.  S1). These phenotypes were confirmed in the cd2 allele and VIGS lines targeting CD2 ( Fig. 6; Supplemental Figs. S3 and S4). Together, these data reveal a broad role for CD2 in plant development, and particularly in relation to metabolism within epidermal and subepidermal cells, which is supported by the epidermal enriched expression pattern of CD2 in stems (Fig. 7B) and in tomato fruit (Matas et al., 2011).

HD-ZIP IV Proteins and Cuticle Biosynthesis
CD2 belongs to a subgroup of HD-ZIP IV proteins, several of which have defined roles in epidermal cell development and are preferentially expressed within the epidermis (Fig. 7B; Javelle et al., 2011a). Proteins of this family are characterized by HD-ZIP domain followed by a steroidogenic acute regulatory lipid transfer domain. The fact that HD-ZIP IV proteins are widely distributed in lower plants, gymnosperms, and angiosperms but are absent in algae is consistent with a role in the evolution of land plants (Ponting and Aravind, 1999;Mukherjee et al., 2009). Several closely related CD2 homologs influence cuticle composition and/or regulate genes involved cuticle biosynthesis. For example, OUTER CELL LAYER1 (OCL1) of maize (Zea mays) is expressed early in embryo development, prior to protoderm formation and is subsequently expressed in the L1 cell layer in the shoot apical meristem (Ingram et al., 1999). Overexpression of OCL1 in maize altered the wax composition in leaf blade as well as sheath and increased the expression of several genes involved in wax biosynthesis and transport (Javelle et al., 2010). Furthermore, transient assays suggested that activation of the expression of a wax transporter and a nonspecific lipid-binding protein likely occurs through direct binding of OCL1 to regulatory regions within these genes (Javelle et al., 2010). Similarly, Arabidopsis HDG1 directly regulates the expression of BODYGUARD and FIDDLEHEAD, two genes involved in cuticle biosynthesis and disruption of HDG1 function leads to cuticles with increased permeability . Furthermore, Arabidopsis ANL2 and HDG1 are preferentially expressed in epidermal or subepidermal cells (Suh et al., 2005;Kubo et al., 2008; http://efp.ucr.edu/cgi-bin/ absolute.cgi) and our analysis indicates that anl2 has a cutin-deficient phenotype (Fig. 8).
Overall, the characterized members of the CD2 subclade all possess a common role in the regulation of cuticle biosynthesis, even though the composition of cutin can vary between species and tissues. For A, Leaf water conductance in AC and pe (n = 5) 6 SE. B, Chlorophyll leaching expressed as percent of total leaf chlorophyll extracted after 24 h (n = 5) 6 SE. C, Scanning electron micrographs of representative stomata on the abaxial leaf surface of AC and pe (scale bar = 5 mm). D, Stomatal density on abaxial leaf surface area of 0.55 mm 2 (n = 10 6 SE). Asterisks denote significant differences (*, P , 0.05; ***, P , 0.001) as determined by Student's t tests.
example, while tomato cutin is predominantly comprised of hexadecanoic-acid-derived monomers, Arabidopsis leaves have a C18:2 dicarboxylic-acid-rich cutin (Table II; Figs. 4 and 8). These cutins require different sets of biosynthetic genes (Li et al., 2007;Pollard et al., 2008;Li-Beisson et al., 2009), suggesting that while the regulation of cuticle biosynthesis may be generally conserved, these regulatory proteins may target distinct sets of biosynthetic genes in diverse species or tissues contributing to their different cuticle Figure 6. Characterization of the pe locus and silencing of CD2. A, Genetic map of the pe locus. Genetic markers and the number of recombinant individuals between adjacent markers from a total of 114 F2 plants are shown. The approximate physical distance between the flanking markers and the pe locus is indicated. Sequence analysis of the CD2 gene from AC and pe revealed a single G nucleotide insertion following nucleotide 2045, leading to a frame-shift mutation. B to G, Silencing of CD2 recreates the pe mutant phenotype. B, Pale-leaf phenotype in a CD2-2 VIGS line. C, Short stature of a CD2-2 VIGS line. D, Plant height of CD2 VIGS lines. E, Anthocyanin content of CD2 VIGS lines. F, The major leaf cutin monomer 9(10),16-dihydroxyhexadecanoic acid is reduced in CD2-silenced lines. G, CD2 silencing impairs leaf cuticle barrier properties and increases water conductivity. All values in D to G, represent mean values (n = 5) 6 SE. TRV2 empty vector control lines were used for all comparisons. Asterisks denote significant differences (*, P , 0.05; ***, P , 0.001) as determined by Student's t tests.
composition. Furthermore, given that several members of the CD2 subclade influence cuticle biosynthesis (Fig.  7A), it is possible that the phylogenetically closely related and as-yet-uncharacterized genes, may also have a similar role.

A Role for CD2 in Reducing Leaf Water Loss
The cuticle regulates nonstomatal water loss and the amount and composition of cuticular wax is associated with cuticle permeability to water. For example, mutants or transgenic lines with reduced levels of verylong-chain alkanes often display elevated rates of water loss (Vogg et al., 2004;Leide et al., 2007;Qin et al., 2011). Furthermore, enhancing the levels of these wax components can restrict water loss, leading to improved drought tolerance (Bourdenx et al., 2011;Seo et al.,   A, A neighborjoining phylogenetic tree derived from a multiple sequence alignment of the deduced full-length amino acid sequences of CD2 homologs constructed using MEGA V5.0 software (Tamura et al., 2011). HDG7 is included as an outgroup. Bootstrap values are from 2,000 replicates, indicated above the nodes. Proteins highlighted in bold have defined functions in cuticle biosynthesis. B, Quantitative reverse transcription PCR analysis of CD2 expression in stem peel of AC relative to stem core and whole stem. Three biological and three technical replications for each sample were analyzed. Data are presented as the mean expression values 6 SE relative to that observed in whole stem. Values labeled with different letters are significantly different (least square means, P , 0.05). 2011). Similarly, drought-stressed plants can increase cuticular wax biosynthesis as a defense mechanism to reduce further water loss .
The wax composition of leaves and fruits vary in pe when compared to AC. For example, hydrocarbons are elevated in pe fruits whereas in leaves the straightchain alkanes are significantly reduced compared to wild type ( Fig. 4B; Supplemental Table S3). The same trend for cutin and wax is observed in the fruit (Isaacson et al., 2009) and leaves (Supplemental Fig.  S3) of cd2 when compared to the M82 parental line. The basis for the differential CD2-mediated control of wax accumulation in leaves and fruits is currently unclear but may occur through altered gene regulation between these tissues or possibly altered fluxes of precursors in mutant tissues. For example, a proportion of the hexadecanoic acid precursor, typically utilized in cutin biosynthesis may be channeled toward alkane biosynthesis when cuticle biosynthesis is limited in pe and cd2 fruits. However, this is unlikely to be a general phenomenon as an increase in alkane accumulation is not observed in cd1 and cd3 fruits (Isaacson et al., 2009).
Our data suggest a correlation between cuticular wax composition in pe and leaf cuticle permeability. For example, leaf water conductance and chlorophyll leaching are increased in pe leaves compared to AC (Fig. 5). A similar trend was observed in CD2-silenced lines (Fig. 6G). In contrast, fruits of the pe and cd2 mutants, which have higher levels of long-chain alkanes, do not exhibit enhanced rates of postharvest water loss (Supplemental Table S3; Isaacson et al., 2009). Although disruption of individual cutin or wax components can alter permeability, the physical consequences of these changes on the structure of the cutin matrix, is not well understood. Alkanes are proposed to influence cuticle permeability by forming waterimpermeable crystalline regions within the cuticle and the packaging of the wax crystals is likely dependent upon the cutin structure (Pollard et al., 2008;Buschhaus and Jetter, 2011). Therefore, while it is not possible to more exactly define the underlying reason for the increased permeability of pe leaves, it could be due to one, or a combination of the following factors, including decreased alkane levels, a decreased cutin load, an alteration in the cutin to wax ratio, or an asyet-undefined biochemical change. A comparison of wax biosynthesis between wild-type and pe leaves and fruit tissues, together with a more in-depth understanding of wax deposition in these genotypes and tissues, may provide insight into how CD2 influences cuticular permeability.
Leaf water loss is also intricately linked to stomatal structure and physiology and several lines of research have highlighted possible links between cuticle composition and stomatal density and structure. Cutin deficiency in Arabidopsis leads to altered stomatal morphology including a reduction in the cuticular ledges (Li et al., 2007). No structural alterations were observed in stomata of the pe mutant, suggesting that CD2 may not influence cuticle biosynthesis in leaf guard cells (Fig. 5C). Similarly, previous reports have shown that altered cuticular wax composition can positively or negatively influence stomatal density in Arabidopsis (Gray et al., 2000;Chen et al., 2003;Yang et al., 2011). The exact mechanisms through which this process occurs are unknown although it is possible that the altered composition of leaf cuticular waxes in pe may exert a compensatory influence, leading to the observed reduction in stomatal density (Fig. 5D).

CD2 and Phenylpropanoid Biosynthesis
Flux into the phenylpropanoid pathway through the enzyme Phe ammonia lyase represents a branch point between primary and specialized metabolism in plants.
The phenylpropanoid pathway synthesizes a wide range of compounds including flavonols, anthocyanins, lignins, suberin and cutin aromatics, and tannins, which have diverse roles in plants as structural polymers, pigments, UV protectants, and signaling molecules (Hahlbrock and Scheel, 1989;Vogt, 2010). In agreement with the epidermal enriched expression pattern of CD2 (Fig. 7B), the pe mutant exhibits reduced accumulation of a subset of phenylpropanoids that typically are synthesized and accumulate in the epidermal and subepidermal cells. In addition to the anthocyanins that are severely reduced in pe tissues, coumarates, which comprise between 3% and 5% of the total cutin monomer load in wild-type tomato fruits, are also reduced in pe ( Fig. 1; Tables I and II). The reduction in coumarate is readily ascribed to the reduction of the cutin aliphatic hydroxy fatty acid monomers, to which coumarate is likely esterified. In contrast, the abundance of lignin and suberin ferulates, two biopolymers derived from phenylpropanoid precursors that are not typically associated with epidermal cells, are unaltered in pe (Supplemental Fig. S2; Supplemental Table S1). It will be of interest to determine if additional phenylpropanoid-derived metabolites are altered in the epidermal cells of pe and to understand the flux through the pathway in these cells in the pe mutant. In addition, the role of CD2/ANL2 in directly regulating anthocyanin biosynthesis, as well as understanding their relationship to known transcriptional regulators of anthocyanin biosynthesis warrants additional investigation.

CD2 as a Regulator of Epidermal Cell Function
HD-ZIP IV proteins are master regulators of epidermal cell fate and function in higher plants and have roles in cuticle biosynthesis together with patterning of trichome and stomata (Javelle et al., 2011a). Our biochemical characterization of the pe mutant of tomato has revealed alterations in multiple pathways associated with epidermal cell function that have evolved to facilitate plant adaptation to stress, including altered cuticle biosynthesis in leaves and fruits, reduced anthocyanin accumulation, and lower trichome and stomatal densities. The subsequent identification of pe as an allele of CD2 extends current knowledge of the role of CD2 beyond cuticle biosynthesis in tomato fruit to include a spectrum of altered phenotypes associated with epidermal cell function. While mutant alleles of CD2 influence a broad range of epidermal cell phenotypes, it is difficult to determine whether these alterations constitute primary or secondary effects. For example, altered stomatal and trichome densities can result from perturbation of cuticular wax biosynthesis, including mutations that affect catalytic enzymes, suggesting that disruption of cuticle structure and/or physiology induces secondary phenotypes (Gray et al., 2000;Chen et al., 2003;Kurata et al., 2003;Aharoni et al., 2004). Establishing cause and effect relationships with respect to the altered stomatal and trichome density phenotypes observed in pe will be complex.
Our data suggest that CD2 is a master regulator of epidermal cell function in tomato. Although the exact role of CD2 in specifying epidermis-related phenotypes remains to be defined, it is possible that CD2 directly regulates the expression of the biosynthetic genes involved in synthesis of diverse epidermal cell associated molecules. Alternatively, it is equally probable that CD2 acts upstream of these pathways, within a transcriptional network that maintains epidermal cell identity and function. Defining the immediate targets of CD2 will provide insight into these questions and will help identify the regulatory networks that control specialized metabolism within plant epidermal cells.

Plant Material and Growth Conditions
Seeds of the pe (pe/pe; LA2467), hp-1 (hp-1/hp-1; LA3538), and ae (ae/ae; LA3612) mutants, together with Solanum pimpinellifolium (LA1589) were obtained from the Tomato Genetics Resource Center, University of California, Davis. The cultivar AC was originally obtained from the Glasshouse Crops Research Institute (Littlehampton, Sussex, UK). As the genetic background of LA2467 was not known, a segregating BC 3 F 2 population was generated from an initial cross of LA2467 (pe/ pe) with AC (PE/PE) as the recurrent parent. The mutant plants (pe/pe) were selected based on phenotype; short stature, pale green color of the stem and leaves, and stickiness of the fruit surface, all of which cosegregated. This population was used for all experiments. Unless otherwise stated, plants were grown in peat-based compost supplemented with fertilizer in greenhouses at Michigan State University under 16 h day (25°C) and 8 h night (20°C). Arabidopsis (Arabidopsis thaliana) seeds of anl2 (SALK_000196) were described previously (Nakamura et al., 2006) and sown in soil (1:1:1 Sure mix:Medium vermiculite:Perlite, MI Grower Products Inc., www. suremix.com) and after 3 d at 4°C, maintained in a growth chamber under a 16 h photoperiod (145 mmol m 22 s 21 ) at 22°C at a relative humidity of 65%. Tomato (Solanum lycopersicum) fruits were harvested when they reached 35 to 40 mm diameter for green stage or at 7 d after the onset of ripening (breaker + 7 d) for ripe stage of development. All experiments involving leaves, stems, and roots were performed on 5-week-old plants unless otherwise stated. Rosette leaves from 7-week-old Arabidopsis plants were used for cutin analysis.

Extraction and Analysis of Cuticular Lipids
Extraction of surface waxes from leaf, root, and fruit tissues of tomato or from leaves of Arabidopsis was accomplished through dipping tissue samples in chloroform containing 1% methanol for 2 min in the case of fruits and 1 min for other tissues. The surface areas of leaves and fruits were determined prior to wax extraction using a LI-3100C area meter (Li-Cor, http://www.licor.com) and Tomato Analyzer version 2.1.0.0., respectively (Brewer et al., 2006). Internal standards of tetracosane, hexadecanol and heptadecanoic acid (5 mg/mL each), and v-pentadecalactone and methyl heptadecanoate (10 mg/mL each; Sigma-Aldrich, http://www.sigmaaldrich.com) were added to each wax and cutin extract, respectively. Preparation of trimethyl-silyl derivatives of wax components, followed by delipidation, base-catalyzed transmethylation, and acetyl derivatization of cutin monomers from all the tissues was performed as previously described (Molina et al., 2006). Gas chromatography (GC)-flame ionization detector (6890N, Agilent Technologies) analysis of the derivatized wax or fatty acid methyl ester products used a DB-5 capillary column. GC temperature was programmed from 140°C to 310°C for cutin and to 330°C for wax, at 5°C min 21 . Samples were injected in split mode (330°C injector temperature). Quantification was based on flame ionization detector ion current using peak areas relative to internal standard peak areas. When necessary, compound identification was performed by GC-mass spectrometry (MS; EI mode) using an Agilent Technologies 6850 GC, equipped with an HP-5 MS column connected to an Agilent 5975 mass spectrometer. Helium was used as carrier gas at 2 mL min 21 and oven temperature was programmed from 120°C to 340°C at 10°C min 21 . Splitless injection was used and ions were collected in scan mode (40-800 atomic mass units) with peaks quantified on the basis of their total ion current.

Imaging of Cuticular Lipids and SEM
Cryosections of fruit pericarp, 8-mm thick, were generated using a Microm HM550 cryostat (Thermo scientific, http://www.thermoscientific.com) and stored at 220°C. Sections were stained using Sudan IV, 0.1% w/v in isopropyl alcohol. Bright-field images were collected using an Olympus IX81 inverted microscope (Olympus, www.olympus.com) configured with UPlan SApo 203 objective, NA0.75. The images were captured using an Olympus DP72 color camera and analyzed with DP2-BSW software. For imaging of fruit cuticles by SEM, fruits were hand sectioned and fixed for 30 min in 4% glutaraldehyde followed by 30 min incubation in 0.1 M phosphate buffer (pH 7.4). Further processing of samples and imaging are according to Li et al. (2007). The same procedures were followed for collecting leaf stomatal images.

Imaging Anthocyanin Distribution
Images of hand-cut cross sections of stem tissue were collected using an Olympus FluoView FV1000 confocal laser-scanning microscope (Olympus) configured on IX81 inverted microscope with a 103 UPLSAPO objective, NA0.4. An excitation wavelength of 559 nm was generated using solid state laser, while the fluorescence emission was collected using a 570 to 640 nm band pass filter, to eliminate the detection of chlorophyll fluorescence. The differential interference contrast transmitted laser light image was collected using the 559-nm laser line and the images presented as overlays of confocal and differential interference contrast images.

Determination of Stomatal and Trichome Densities
For stomatal density measurements, leaf surface imprints on abaxial sides of the leaf surface were taken using transparent nail polish and placed onto microscope slides. Imaging was performed using a Zeiss Axiophot microscope (Carl Zeiss, Inc., http://www.zeiss.com) at 203, image processing was accomplished using Image-pro plus image analysis software (Media Cybernetics, http://www. mediacy.com). Stomatal density was expressed as number of stomata in five different fields of 0.55 mm 2 per leaflet of five individual leaflets. Trichome density was analyzed on 3-week-old plants of AC, M82, pe, and cd2 using plant growth conditions and methods previously described (Kang et al., 2010).

Analysis of Volatile Terpenes from Tomato Trichomes
Volatile terpene levels were determined in 3-week-old plants. Briefly, a leaflet from the second newly emerging leaf was dipped in 1 mL of methyl tertbutyl ether containing 5 ng/mL of tetradecane as an internal standard, and allowed to rock for 1 min. GC-MS analysis was performed as previously described (Schilmiller et al., 2009). Terpene identification was based on comparison of mass spectra and retention times with those available in an essential oil library (Adams, 2009).

Determination of Fruit Glossiness
The glossiness of red-ripe fruits was measured using a custom-made gonioreflectometer consisting of a Fire-iTM digital camera (http://www.unibrain. com) and two 60-W incandescent light bulbs located in a light-impervious container. Thirty images were taken for each fruit, as it rotated 360°. The average number of saturated blue pixels for all the images was recorded as an index of fruit gloss measured in gloss units (a gloss unit is defined as number of pixels above the saturation threshold that is fixed based on specular reflectance of the sample image) programmed using MATLAB (MathWorks, www.mathworks. com). The instrument was calibrated using standard glossy and nonglossy green and red spheres prior to data collection.

Mechanical Analysis of the Fruit Cuticle
Young's elastic modulus (Y) of isolated tomato cuticles of 3 3 0.5 cm dimension was measured using a dynamic mechanical analyzer Q800 (TA Instruments http://www.tainstruments.com) in constant stress ramp mode at room temperature until breakage was observed. Y values were calculated using Universal Analysis 2000 software (TA Instruments).

Determination of Anthocyanin and Chlorophyll Content
Anthocyanins were extracted from leaves according to Peters et al. (1998). The absorbance (A 535 ) of the aqueous phase was determined spectrophotometrically using a Hitachi U3000 spectrophotometer (Hitachi High Technologies America, Inc., http://www.hitachi-hta.com). Chlorophyll content was sampled from three leaf discs. Discs were incubated in darkness for 48 h at 4°C in 100% dimethyl sulfoxide. A 647 and A 664 were determined spectrophotometrically as described above. Chlorophyll a and b were calculated according to Moran (1982).

Determination of Leaf Epidermal Permeability
Leaf water conductance was measured using Li-Cor 6400 (Li-Cor). Plants were dark acclimated for 3 h prior to measurements being made to eliminate the effect of stomatal conductance. Chlorophyll-leaching assays were performed on fully expanded leaves. The amount of chlorophyll extracted into the solution was quantified every 30 min as described above and data were expressed as a percentage of the total chlorophyll extracted after 24 h according to Kosma et al. (2009).

Lignin Analysis
Lignin was extracted as previously described (Foster et al., 2010). Results are reported as % ABSL based on the dry weight of the alcohol insoluble residue according to York et al. (1986), Fukushima et al. (1991), and Robinson and Mansfield (2009). For the calculation of % ABSL content of tomato, the molar extinction coefficient for Arabidopsis (23.35 g 21 L cm 21 ) was used as an estimate given that no tomato-specific coefficients were available at the time of analysis (Chang et al., 2008).

Genetic Mapping
An interspecific F2 population of 114 individuals derived from a cross between tomato (pe/pe; LA2467) and S. pimpinellifolium (PE/PE; LA1589), was phenotyped for paleness of the vegetative tissues and stickiness of the fruit peel and simultaneously genotyped with chromosome 1 molecular markers, previously selected to be polymorphic between LA2467 and LA1589 (Supplemental Table S4). Genomic DNA was extracted from expanding leaves as previously described (Barry et al., 2005). Details of genetic maps and molecular markers can be accessed through the Sol Genomics Network (http://solgenomics.net).

Cloning of CD2 and Phylogenetic Analysis
Total RNA was extracted from expanding leaves of AC and pe using the RNeasy mini kit (Qiagen, http://www.qiagen.com). One microgram of RNA was used as a template for oligo dT primed first-strand cDNA synthesis using the transcriptor first-strand cDNA synthesis kit (Roche Applied Science, https://www.roche-applied-science.com). The coding region of CD2 was amplified from cDNA using Pfu Ultra DNA polymerase (Agilent technologies, http://www.home.agilent.com) using the primers CD2-F and CD2-R (Supplemental Table S5). PCR fragments were purified using the PureLink PCR purification kit (Invitrogen, http://www.invitrogen.com) and cloned into the pCR4Blunt-TOPO vector (Invitrogen). Sequencing of inserts was accomplished using vector primers and the gene-specific primers CD1 through CD6 (Supplemental Table S5). Sequences were assembled using Sequencher 4.7 software (Gene Codes Corporation, http://genecodes.com). A neighborjoining phylogenetic tree was constructed from a multiple sequence alignment of the deduced full-length amino acid sequences of selected CD2 homologs using MEGA V5.0 software (Supplemental Table S6; Tamura et al., 2011).

Construct Assembly and Silencing of CD2
Two constructs, CD2V1 and CD2V2, were assembled in the TRV2-LIC vector as previously described (Dong et al., 2007) using the primers CD2V1-F, CD2V1-R, CD2V2-F, and CD2V2-R (Supplemental Table S5). VIGS experiments were performed as described by Velasquez et al. (2009). Tomato cotyledons were infiltrated with Agrobacterium tumefaciens strain GV3101 cultures before the appearance of the first true leaves and tissues were harvested 4 weeks after infiltration. TRV2-LIC empty vector plants were used as controls and TRV2:: PHYTOENE DESATURASE plants were used to monitor the progression of gene silencing. The VIGS plants were maintained in a growth chamber under a 16 h photoperiod (145 mmol m 22 s 21 ) at a constant temperature of 27°C and a relative humidity of 60%.

Quantitative Reverse Transcription PCR Analysis
For CD2 expression analysis stems were divided into stem peel, stem core (stem devoid of peel), and whole stem. Total RNA was isolated as described above. RNA was treated on column with DNase (Qiagen) and 1 mg of RNA was used for reverse transcription using SuperScript III first-strand synthesis system (Invitrogen). Gene-specific primers for the analyzed genes were designed by Primer Express 3.0 (ABI; listed in Supplemental Table S5). The PCR reactions were performed with FAST SYBR master mix, 23 (ABI, http://www.appliedbiosystems.com) in a 25 mL volume using an Applied Biosystems StepOnePlus real-time PCR system (ABI) with the following cycling program: 10 min at 95°C, 15 s at 95°C followed by 40 cycles of 1 min at 60°C, 15 s at 95°C, and 1 min at 60°C. The comparative DDCT method was performed according to Balaji et al. (2008).

Statistical Analyses
Statistical analyses were performed using SAS (SAS Institute, www.sas. com). The genotypic constituents were evaluated by Student's t test and least square means.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Volatile terpene levels in AC and pe.
Supplemental Figure S2. Suberin and root wax composition in AC and pe.
Supplemental Figure S3. Comparison of leaf cutin and wax composition in M82 and cd2.
Supplemental Figure S4. Trichome density on leaves and anthocyanin accumulation in leaves and stem peels of M82 and cd2.
Supplemental Table S1. Lignin composition of leaves and stems in AC and pe.
Supplemental Table S2. Total wax load in fruits of AC and pe.
Supplemental Table S3. Fruit wax composition of AC and pe.
Supplemental Table S4. PCR-based genetic markers flanking the pe locus.
Supplemental Table S5. Oligonucleotide primers used in the study.
Supplemental Table S6. CD2-related genes utilized in phylogenetic analysis.