Cutin deficiency in the tomato fruit cuticle consistently affects resistance to microbial infection and biomechanical properties, but not transpirational water loss


*(fax +1 607-2555407; e-mail


Plant cuticles are broadly composed of two major components: polymeric cutin and a mixture of waxes, which infiltrate the cutin matrix and also accumulate on the surface, forming an epicuticular layer. Although cuticles are thought to play a number of important physiological roles, with the most important being to restrict water loss from aerial plant organs, the relative contributions of cutin and waxes to cuticle function are still not well understood. Tomato (Solanum lycopersicum) fruits provide an attractive experimental system to address this question as, unlike other model plants such as Arabidopsis, they have a relatively thick astomatous cuticle, providing a poreless uniform material that is easy to isolate and handle. We identified three tomato mutants, cutin deficient 1 (cd1), cd2 and cd3, the fruit cuticles of which have a dramatic (95–98%) reduction in cutin content and substantially altered, but distinctly different, architectures. This cutin deficiency resulted in an increase in cuticle surface stiffness, and in the proportions of both hydrophilic and multiply bonded polymeric constituents. Furthermore, our data suggested that there is no correlation between the amount of cutin and the permeability of the cuticle to water, but that cutin plays an important role in protecting tissues from microbial infection. The three cd mutations were mapped to different loci, and the cloning of CD2 revealed it to encode a homeodomain protein, which we propose acts as a key regulator of cutin biosynthesis in tomato fruit.


The aerial organs of all plants are covered by a thin lipophilic layer, the cuticle, which defines their boundaries and provides protection against biotic and abiotic stresses (Jeffree, 2006; Nawrath, 2006). Although the chemical composition of plant cuticles varies between species, and even between organs of a single plant, they are all composed of two main components: cutin, a polymer of mainly C16 and C18 hydroxy fatty acids or diacids, and waxes (Jeffree, 2006; Nawrath, 2006). The higher order structural organization of these polymers also shows a similar generic pattern: a thin epicuticular wax layer, beneath which is a mixture of intracuticular waxes and cutin, and within which are embedded polysaccharides (Jeffree, 2006; Nawrath, 2006). Decades of research have provided insights into the diversity of cuticle composition and architecture; however, most studies have been descriptive, comparative surveys, and relatively little is known about the biosynthesis, transport and assembly of cuticular compounds. Similarly, there is a limited understanding of the relative importance and differing roles of waxes and cutin.

The cuticle provides a critical and typically very efficient barrier against water loss, although it is not completely impermeable (Burghardt and Riederer, 2006). Determining the relative contribution of waxes and cutin to water retention has been challenging, largely because of the difficulties in separating the polymeric constituents to yield a material that retains its native architecture. Studies using solvent-extracted, enzymatically isolated cuticles resulted in the conclusion that water permeability is limited primarily by waxes (Schönherr, 1976, 1982). However, such chemical treatments may alter the structure of the remaining matrix, and should be interpreted cautiously. Alternatively, a number of Arabidopsis mutants have now been identified with cuticle phenotypes and a high permeability to water. For example, the double mutant gpat4/gpat8 has an abnormal cutin composition, but a normal wax profile (Li et al., 2007), whereas wax2 shows the opposite trend (Chen et al., 2003; Goodwin and Jenks, 2005; Rowland et al., 2007), and yet others, such as bodyguard and fatb-ko are altered in both cutin and waxes (Bonaventure et al., 2003, 2004; Kurdyukov et al., 2006a). The phenotypes of these mutants suggest that both cutin and wax are important in resisting desiccation. However, it is important to note that in many cases defects in stomatal morphology have been noted, which could explain the altered water status.

Similarly, the relative importance of cutin and waxes to resistance against microbial infection is not well understood. Many fungal pathogens secrete cutinases during infection (Kolattukudy, 1985); however, none of the knock-out lines reported to date, in various fungal species, has provided clear evidence that cutinase plays a direct role in cuticle penetration (e.g. Yao and Koller, 1995; van Kan et al., 1997; Reis et al., 2005). Evaluations of Arabidopsis mutants with cuticle phenotypes have also provided inconclusive, or even contradictory, information (Xiao et al., 2004; Bessire et al., 2007; Chassot et al., 2007; Li et al., 2007; Tang et al., 2007). Here too, interpretation of the results is complicated by the fact that the stomata of these mutants are suggested to be defective, and thus might provide a gateway for pathogen entry. In summary, although some researchers have concluded that the cuticle does not present a major barrier to pathogens, this remains an open question.

One of the major obstacles towards functional dissection of the cuticle has been the identification of a robust research model. Arabidopsis stems and leaves have played this role for more than a decade, and many cuticle-associated mutants have been identified (Nawrath, 2006; Pollard et al., 2008; Samuels et al., 2008). However, Arabidopsis cuticles are extremely delicate and contain stomata (Franke et al., 2005): two major experimental limitations. Tomato (Solanum lycopersicum) fruits, on the other hand, offer a system that is potentially more suitable, as the cuticle is relatively thick and astomatous, thereby providing a uniform intact surface (Vogg et al., 2004). Additionally, several studies in tomato have shown the same molecular pathways are present in vegetative and reproductive tissues, but are regulated by different suites of genes (Fraser et al., 1999; Ronen et al., 2000; Galpaz et al., 2006). This can allow a phenotype to be exhibited even if the mutation abolishes an essential function that would result in lethality if the gene was specific to vegetative tissues.

In this paper we describe three tomato mutant lines [cutin deficient 1 (cd1), cd2 and cd3], the fruit cuticles of which are severely cutin deficient, providing an excellent experimental resource to investigate the contribution of cutin to various aspects of cuticle function and fruit physiology. We describe the consequences of the massive reduction in cutin on cuticle architecture and biomechanical properties, and susceptibility of the fruit to water loss and microbial infection. We also report on the mapping of the three mutants and the cloning of CD2.


Identification of three independent tomato mutant lines with fruit cuticle-associated phenotypes

Seeds of three mutant lines (cd1, cd2, cd3) were obtained from the ‘Genes that Make Tomatoes’ collection (Menda et al., 2004; that were annotated as having cuticle-associated phenotypes. Their fruits had a highly glossy phenotype (Figure 1) and a ‘rubbery’ surface texture compared with those from the M82 background cultivar, but no visible phenotype in other organs. The three cd lines were back-crossed with M82 and then crossed with each other, and the segregation ratios of the glossy trait in the progeny indicated that the mutations occur at three different single loci. Moreover, the trait was inherited in a recessive mode in all three lines, although it is possible that heterozygous plants have a weaker intermediate phenotype.

Figure 1.

 Two red ripe-stage fruit from the cutin deficient (cd) lines cd1, cd2 and cd3, and their background line M82. Scale bar: 1 cm.

The cd mutant fruit cuticles have dramatically reduced cutin content

The composition of isolated fruit cuticles from three key stages of fruit development (small green, SG; mature green, MG; and red ripe, RR) was analyzed to determine whether the glossy phenotypes could be associated with quantitative or qualitative differences in wax and/or cutin content. The most striking difference between M82 and the mutant lines was in the total level of cutin per unit surface area. At the RR stage, when the cuticle reaches full maturity, the cutin levels of M82 fruits were, on average, 985 μg cm−2, whereas drastically lower quantities were detected in fruits of cd1 (50 μg cm−2), cd3 (38 μg cm−2) and cd2 (15 μg cm−2), corresponding to a reduction of 95, 96 and 98%, respectively (Table 1). This severe cutin deficiency was detected throughout development, from the early SG stage (Table S1). In addition, whereas the levels of cutin per surface area were seen to increase as the fruit developed and ripened in M82, the levels in the cd fruits remained relatively constant. Interestingly, the ratios between the main cutin monomers in the cuticles of the cd lines were similar to those of M82, with 9(10), 16-dihydroxyhexadecanoic acid being the most abundant monomer. One exception was hexadecanoic acid, which was detected at much higher levels in the cd lines at all developmental stages (Tables 1 and S1).

Table 1.   Cutin acid monomer levels (μg cm−2), and percentages, of isolated cuticles from M82, cd1, cd2 and cd3 ripe tomato fruit
 M82 (%)cd1 (%)cd2 (%)cd3 (%)
  1. Data represent the means of three replicates ± SE.

  2. aIsomer composition not determined.

  3. bDouble-bond position not determined. Previously reported at Δ12 carbon (Baker et al., 1982).

Hexadecanoic0.9 ± 0.3 (0.1)0.9 ± 0.5 (1.8)1.8 ± 0.3 (12.1)7.9 ± 0.8 (20.6)
16-OH Hexadecanoic35 ± 1.6 (3.7)0.8 ± 0.4 (1.6)0.4 ± 0 (2.7)0.6 ± 0.2 (1.6)
10,16-diOH Hexeadecanoica760.9 ± 55 (80.2)28.7 ± 8.5 (57.6)7 ± 0.3 (47.0)16.4 ± 1.6 (42.7)
Hexadecane-1,16-dioic12 ± 1.6 (1.3)0.5 ± 0.1 (1.0)0.1 ± 0 (0.7)0.8 ± 0.3 (2.1)
18-OH Octadecanoic25.6 ± 9.3 (2.7)1.5 ± 1 (3.0)0.3 ± 0.1 (2.0)1.2 ± 0.2 (3.1)
9,18-diOH Octadecanoic11.2 ± 0.4 (1.2)1.4 ± 0.2 (2.8)0.3 ± 0.1 (2.0)0.4 ± 0 (1.0)
9,10,18-triOH Octadecanoic5.4 ± 0.7 (0.6)2 ± 1.4 (4.0)1.2 ± 0.5 (8.1)0.3 ± 0.2 (0.8)
9,10,18-triOH Octadecenoicb5.8 ± 1.1 (0.6)1.7 ± 0.8 (3.4)0 ± 0 (0.0)0.5 ± 0.3 (1.3)
p-coumaric7.2 ± 0.6 (0.8)1.3 ± 0.3 (2.6)0.5 ± 0.1 (3.4)1.6 ± 0.2 (4.2)
m-coumaric15.6 ± 1.9 (1.6)2 ± 0.5 (4.0)0.5 ± 0.2 (3.4)2.3 ± 0.3 (6.0)
Not identified105.2 ± 16 (11.1)9 ± 1.5 (18.1)2.8 ± 1.5 (18.8)6.5 ± 0.4 (16.9)
Total984.8 ± 4850 ± 8.414.9 ± 238.4 ± 3.1

Explanations for the dramatic reduction in cutin content of the enzymatically isolated cuticle might include a blockage in the transport of cutin monomers within the cell, their export across the plasma membrane, trafficking through the apoplast to the cell surface or subsequent polymerization, rather than a disruption in the biosynthetic pathway. To test this possibility, and investigate whether the mutants accumulate normal levels of cutin monomers, which are then lost during preparation of isolated cuticles, the lipid profiles of entire dewaxed MG fruit peels of all four genotypes were compared. The same trend was observed as for the enzymatically isolated cuticles, in that fruit peels of all cd lines contained very low levels of cutin monomers compared with those of M82, and among the three mutants, the cutin content was highest in cd1 and lowest in cd2 (Table 2). We note that although the total cutin monomer values obtained for the peel extracts using acid-catalyzed depolymerization (Table 2) were slightly higher than those obtained with the base-catalyzed depolymerization of the enzymatically isolated cuticles (Table 1), the same pattern was apparent.

Table 2.   Cutin acid monomer levels (μg cm−2) from epidermal peel of M82, cd1, cd2 and cd3 mature green (MG) fruit
  1. As detected after acid-catalyzed (MeOH-HCl) depolymerization. Data represent means of three replicates ± SE.

  2. aIsomer composition not determined.

  3. bDouble-bond position not determined. Previously reported at Δ12 carbon (Baker et al., 1982).

Hexadecanoic16.3 ± 0.515.3 ± 0.913.5 ± 0.323.9 ± 2.5
16-OH Hexadecanoic26.8 ± 1.60.6 ± 0.10.8 ± 0.10.4 ± 0.0
10,16-diOH Hexeadecanoica678.6 ± 28.759.6 ± 4.510.3 ± 3.119.6 ± 1.3
Hexadecane-1,16-dioic6.9 ± 0.30.6 ± 0.00.3 ± 0.00.4 ± 0.0
Octadecanoic2.9 ± 0.13.7 ± 0.61.2 ± 0.64.0 ± 0.5
Octadecenoic (C18:1/C18:2/C18:3)35.9 ± 5.232.8 ± 5.930.9 ± 5.027.5 ± 3.0
18-OH Octadecanoic223.8 ± 9.914.2 ± 1.55.6 ± 3.916.2 ± 1.7
9,18-diOH Octadecanoic5.3 ± 0.21.3 ± 0.20.2 ± 0.10.2 ± 0.0
9,10,18-triOH Octadecanoic10.1 ± 0.71.0 ± 0.14.7 ± 1.21.2 ± 0.1
9,10,18-triOH Octadecenoicb6.9 ± 0.21.1 ± 0.01.5 ± 0.31.0 ± 0.0
p-coumaric3.3 ± 0.31.3 ± 0.30.7 ± 0.31.1 ± 0.3
m-coumaric23.3 ± 0.36.2 ± 0.51.0 ± 0.16.4 ± 0.4
Not Identified56.0 ± 3.514.6 ± 1.66.1 ± 0.48.9 ± 0.5
Total1096.2 ± 49.2152.48 ± 9.977.0 ± 2.5102.0 ± 2.3

cd fruits have normal levels of wax, but each line has a different composition

The wax composition of M82, cd1, cd2 and cd3 was measured at the RR and MG developmental stages (Tables 3, 4, S2 and S3). M82 fruit wax at the RR stage was dominated by alkanes, which comprised about one-third of the total waxes, and the average total fruit wax coverage was similar in the mutants (Table 3).

Table 3.   Wax class composition of red ripe M82, cd1, cd2 and cd3 fruit
 Total loadAcidsAldehydesAlkanolsAlkenolsAlkenes
  1. Values are given as μg cm−2. Data represent the means of three or four replicates ± SE; N.D., below the limit of detection.

M828.41 ± 0.590.51 ± 0.070.29 ± 0.090.43 ± 0.050.32 ± 0.141.22 ± 0.12
cd18.77 ± 0.620.84 ± 0.110.19 ± 0.030.82 ± 0.150.17 ± 0.070.80 ± 0.09
cd27.72 ± 0.990.60 ± 0.110.28 ± 0.090.18 ± 0.050.03 ± 0.010.69 ± 0.29
cd38.74 ± 0.460.51 ± 0.030.18 ± 0.030.52 ± 0.020.17 ± 0.050.64 ± 0.08
 AlkanesIso-Alkanesanteiso-AlkanesEstersTriterpenoids & SterolsUnidentified
M822.79 ± 0.100.38 ± 0.070.08 ± 0.01N.D.1.62 ± 0.180.77 ± 0.20
cd12.50 ± 0.290.52 ± 0.140.08 ± 0.020.06 ± 0.012.17 ± 0.340.62 ± 0.08
cd23.94 ± 0.470.70 ± 0.070.11 ± 0.01N.D.0.96 ± 0.300.23 ± 0.03
cd32.82 ± 0.180.28 ± 0.050.09 ± 0.020.19 ± 0.052.67 ± 0.270.67 ± 0.33
Table 4.   Wax composition of red ripe M82, cd1, cd2 and cd3 fruit
 C chain length/compoundM82cd1cd2cd3
  1. Values are given as μg cm−2 × 102. Data represent the means of three or four replicates ± SE; N.D., below the limit of detection.

Alkanoic acids162.5 ± 0.56.7 ± 1.04.7 ± 1.311.1 ± 1.9
181.6 ± 0.74.9 ± 1.44.5 ± 1.45.6 ± 0.8
202.3 ± 0.91.3 ± 0.70.8 ± 0.50.5 ± 0.1
222.0 ± 0.28.3 ± 2.06.7 ± 2.52.7 ± 0.2
2418.7 ± 5.319.5 ± 3.120.6 ± 5.510.2 ± 1.9
267.7 ± 2.212.8 ± 3.49.5 ± 2.55.9 ± 0.3
281.2 ± 0.44.7 ± 0.61.9 ± 0.81.7 ± 0.2
307.5 ± 1.65.9 ± 0.55.1 ± 1.56.0 ± 0.5
328.0 ± 0.919.3 ± 4.66.3 ± 0.77.3 ± 1.0
Aldehydes2420.4 ± 6.55.7 ± 2.820.2 ± 6.910.1 ± 2.1
264.6 ± 1.91.5 ± 0.25.5 ± 1.92.8 ± 0.6
323.5 ± 0.812.0 ± 2.92.2 ± 0.24.9 ± 0.1
Alkanols200.6 ± 0.30.3 ± 0.10.3 ± 0.20.4 ± 0.1
220.5 ± 0.32.1 ± 1.20.4 ± 0.10.5 ± 0.2
241.6 ± 0.54.0 ± 2.20.6 ± 0.21.5 ± 0.2
251.2 ± 0.50.5 ± 0.10.2 ± 0.10.3 ± 0.1
263.6 ± 1.23.4 ± 0.81.1 ± 0.71.1 ± 0.4
271.5 ± 0.04.1 ± 1.01.2 ± 0.62.8 ± 0.4
284.2 ± 1.511.3 ± 2.04.2 ± 0.77.5 ± 0.9
3010.7 ± 2.319.3 ± 2.75.4 ± 1.914.9 ± 0.8
3215.0 ± 2.231.5 ± 7.74.9 ± 1.416.5 ± 2.1
343.6 ± 0.75.4 ± 1.90.0 ± 0.06.7 ± 0.8
Alkenols221.9 ± 1.01.6 ± 0.90.5 ± 0.11.5 ± 0.4
247.1 ± 3.46.6 ± 4.20.4 ± 0.14.0 ± 1.5
2623.2 ± 9.99.2 ± 2.92.1 ± 1.111.6 ± 3.4
Alkanes230.3 ± 0.10.4 ± 0.10.3 ± 0.10.5 ± 0.1
252.6 ± 0.42.5 ± 0.43.7 ± 1.51.8 ± 0.4
261.7 ± 0.32.9 ± 0.32.5 ± 0.24.3 ± 2.2
274.4 ± 0.810.8 ± 0.87.7 ± 3.04.1 ± 0.6
288.3 ± 2.213.1 ± 0.610.9 ± 2.111.0 ± 1.4
2942.8 ± 14.473.2 ± 13.368.2 ± 19.769.9 ± 6.4
3020.2 ± 1.721.0 ± 2.615.2 ± 7.327.3 ± 2.0
31140.4 ± 14.390.1 ± 9.1187.5 ± 25.9118.8 ± 10.6
3223.3 ± 4.010.2 ± 1.536.5 ± 3.017.7 ± 1.6
3332.5 ± 3.321.0 ± 6.857.5 ± 5.823.8 ± 5.0
342.9 ± 0.64.4 ± 1.53.7 ± 0.83.0 ± 0.3
Alkenes33105.4 ± 9.166.6 ± 8.253.1 ± 21.451.3 ± 6.6
343.7 ± 0.54.3 ± 0.33.2 ± 2.24.0 ± 0.6
3512.5 ± 3.69.5 ± 0.912.3 ± 5.18.2 ± 0.7
iso-Alkanes296.4 ± 1.111.1 ± 3.67.3 ± 1.95.6 ± 3.3
304.5 ± 1.54.7 ± 0.98.0 ± 0.53.0 ± 0.5
3117.4 ± 2.221.2 ± 4.544.2 ± 5.415.4 ± 1.9
323.8 ± 0.23.1 ± 0.810.8 ± 1.13.4 ± 0.2
335.7 ± 4.512.3 ± 5.3N.D.0.8 ± 0.5
anteiso-Alkanes304.1 ± 1.13.1 ± 0.82.7 ± 0.43.6 ± 0.5
311.0 ± 0.31.0 ± 0.21.4 ± 0.20.8 ± 0.2
322.9 ± 0.73.7 ± 0.96.4 ± 0.94.6 ± 1.2
Triterpenoids & Sterolsα amyrin39.1 ± 5.544.6 ± 5.814.8 ± 4.632.7 ± 4.5
β amyrin27.2 ± 3.445.5 ± 11.815.9 ± 3.446.5 ± 9.4
δ amyrin52.4 ± 5.873.8 ± 11.537.1 ± 21.351.9 ± 6.7
multiflorenol8.3 ± 0.511.4 ± 1.33.8 ± 1.57.5 ± 0.9
taraxerol8.3 ± 0.511.4 ± 1.33.8 ± 1.57.5 ± 0.9
taraxasterol4.8 ± 0.45.5 ± 1.42.0 ± 0.74.5 ± 0.6
ψ taraxasterol8.5 ± 0.18.0 ± 2.02.9 ± 1.46.3 ± 0.9
Unidentified13.6 ± 2.217.0 ± 2.115.7 ± 3.6109.6 ± 15.2
Alkyl EstersN.D.6.1 ± 0.6N.D.19.4 ± 4.9
Unidentified76.9 ± 20.061.7 ± 7.723.4 ± 3.266.6 ± 33.3

The cd1 fruits showed altered levels of the two main classes of waxes: alkanes and triterpenoids. A decreased alkane content was attributed to reductions in the C31, C32 and C33 alkanes, specifically, despite elevated levels of C26–C29 alkanes (Tables 3 and 4), whereas a higher triterpenoid content was mainly the result of an increase in all types of amyrins (α, β and δ). The cd2 fruits also showed abnormal levels of alkanes and triterpenoids, but the trend was opposite to that observed for cd1. In cd2 alkanes were elevated, again primarily because of elevated levels of C31, C32 and C33 alkanes, and amyrin levels were reduced. Only minor differences were detected in cd3 waxes, other than the higher levels of unidentified triterpenoids.

The cd fruits show dramatic differences in cuticle architecture

A range of microscopic techniques were used to determine whether the remarkably low levels of cutin result in differences in cuticle morphology and ultrastructure. Scanning electron microscopy (SEM) and light microscopy of fruit pericarp cross-sections from RR and MG fruit, respectively, revealed striking differences between the M82 fruit cuticles and those of the cd lines (Figure 2a,b, respectively). Tomato fruits typically have a relatively thick cuticle that surrounds the epidermal cell layer and extends down the anticlinal walls, often permeating through the walls of additional underlying cell layers (Bargel and Neinhuis, 2004; Matas et al., 2004). Dramatic differences in gross cuticle architecture between the mutants and M82 were evident using either SEM (Figure 2a) or light microscopy of sections treated with the lipid-specific stain Sudan IV (Figure 2b). The cd1 cuticle, although still far less substantial than that of M82, nonetheless penetrated between the cells, forming clear anticlinal pegs, and in some cases extended into the periclinal walls underlying the epidermal cell layer, although in no cases were the epidermal cells surrounded with cuticular material, as was observed in M82 (Figure 2b). In contrast, the cuticle of cd3 typically had small protrusions, rather than true anticlinal pegs, and little staining was detected other than in the outer epidermal wall. The cd2 mutant showed the most dramatic phenotype, and the cuticle comprised an extremely thin, superficial surface layer. The differences in cuticle architectures were also evaluated using high-resolution three-dimensional imaging and tomography (see Buda et al., 2009).

Figure 2.

 Images of breaker-stage fruit pericarp sections from M82 and cd lines obtained using: (a) scanning electron microscopy; and (b) light microscopy showing cuticles stained with Sudan IV. Scale bars: (a) 10 μm; (b) 20 μm.
Abbreviations: AP, anticlinal peg; CM, cuticular membrane; Col, collenchyma cell; EC, epidermal cell.

Transmission electron microscopy (TEM) imaging again emphasized the major reduction in cuticle thickness in the cd mutants, but also revealed substantial differences in the ultrastructure of the cuticle and epidermal cell wall (Figure 3). The M82 cuticle showed a typical gradual transition from a darker, electron-dense region (the internal cuticular layer; ICL), to the more opaque external cuticular layer (ECL). In contrast, the cuticles of all three cd mutants were characterized by a thin dark layer. Differences were also seen between the mutants, as although the electron dense region in the cd2 cuticle had a sharply defined interface with the underlying wall, those of cd1 and cd3 had intermediate layers containing dark fibrillar structures and more diffuse interfaces.

Figure 3.

 Transmission electron microscopy images of red ripe-stage fruit pericarp sections from M82 and cd lines. Scale bars: 2 μm.
Abbreviations: CL, cuticular membrane; Cyt, cytoplasm; ECL, external cuticular layer; ICL, internal cuticular layer; PCW, polysaccharide cell wall.

The cd mutant cuticles exhibit differences in biomechanical properties

Enzymatically isolated tomato cuticles were imaged using atomic force microscopy (AFM) in tapping mode (Figure 4), to investigate the consequences of the cutin deficiency on the nanobiomechanical properties of the cuticle surface, and to compare those properties between the three mutants. The topographic images showed clear differences between the wild-type and cd samples, which are likely to result from the compositional differences described above. On a larger scale (Figure 4a,d,g,j), whereas cellular outlines were visible for M82, they were less discernable in the mutants, with the exception of cd1, consistent with their reduced formation of anticlinal pegs (Figure 2b). The higher resolution images (Figure 4b,e,h,k) showed amorphous lipid clusters, typical of fruit cuticular membranes, which were similar in all samples. Locally, the surface roughness was dominated by these lipid clusters with a root-mean-square roughness of approximately 40–50 nm in all four lines (data not shown). Because of the fragile nature of the cuticles, their biomechanical properties were explored by AFM nanoindentation measurement, by applying a simple Hertzian contact mechanics model (Round et al., 2000). From this, the elastic modulus of each of the samples was determined, and substantial differences were seen between the M82 and cd samples, where the M82 cuticle had a far lower Young’s modulus value than those of the mutants (Table 5). To compare the relative contributions of the cuticular waxes and cutin to cuticle biomechanical properties, Soxhlet extraction was used to dewax the tomato cuticular samples, and the mechanics were measured by AFM nanoindentation, as described above. The dewaxed surfaces exhibited a smooth texture (Figure 4c,f,i,l), and the Young’s modulus values of the cd cuticles were similar to those of the M82 cuticle (Table 5) and earlier studies of isolated dewaxed tomato fruit cuticles (Round et al., 2000). This suggests that the differences in nanobiomechanical properties of the cuticle are not caused by significant differences in the make-up of the polymerized cutin matrix, but rather by its interplay with the varying lipid content.

Figure 4.

 Atomic force microscopy tapping mode topographical images of cuticles from red ripe fruit of M82 (a–c), cd1 (d–f), cd2 (g–i) and cd3 (j–l), with Δz values of 6089, 115, 65, 2326, 199, 95, 6836, 157, 128, 2762, 219 and 149 nm, respectively. Samples were analyzed at magnifications of 70 x 70 μm2 (a, d, g and j), 2 x 2 μm2 (b, e, h and k) and 2 x 2 μm2, after removal of epicuticular waxes (c, f, i and l).

Table 5.   Surface elastic modulus (Young’s modulus) values for isolated cuticles from M82, cd1, cd2 and cd3 fruit, taken at different locations on the surface, before and after the removal of waxes
 Cuticles with waxDewaxed cuticles
Low (MPa)High (MPa)Low (MPa)High (MPa)
  1. The low value corresponds to the center of the surface depressions, whereas the high value corresponds to the ridges of the cellular outlines on the surface.

M8240 ± 937 ± 553 ± 857 ± 15
cd1240 ± 33187 ± 5522 ± 319 ± 2
cd2303 ± 41234 ± 4752 ± 2347 ± 14
cd3162 ± 26195 ± 2646 ± 349 ± 2

The cd mutant fruit show increased susceptibility to microbial infection, but not necessarily increased water loss

Despite the fact that the cd mutant cuticles were shown to be severely cutin deficient, and to have dramatically perturbed and distinctly different architectures, the fruits appeared remarkably normal, other than having an increased glossiness (Figure 1). Under controlled glasshouse growth conditions, no differences were seen in fruit ripening or other aspects of plant physiology, such as susceptibility to wilting. We therefore investigated whether some of the cuticle functions associated with protection against biotic and abiotic stresses were compromised.

An important role of the cuticle is to limit water loss from plant tissues, and so to determine whether the cd fruit show altered transpiration, the weights and external appearance of detached ripe fruits held at room temperature (21°C) were evaluated over a 10-week period (Figure 5). cd1 fruits showed rapid symptoms of desiccation within a few days of harvest, and by 3 weeks were notably shriveled, whereas those of cd2 and cd3 appeared similar to M82 fruits (Figure 5a). The phenotypes correlated with water loss, as cd1 fruits showed approximately twofold greater sustained water loss rates than those of M82. The values for cd2 and cd3 were only marginally greater than the wild type (Figure 5b), and it was notable that they were also different from each other. To further test the permeability of the fruit cuticles to aqueous solutions, drops of Toluidine Blue solution were placed on the surface for up to 8 hours, as was reported in a study of the other cuticle mutants (Tanaka et al., 2004). However, no staining of the underlying tissue was seen in fruits from M82 or any of the cd lines (data not shown).

Figure 5.

 Analysis of transpirational water loss from M82 and cd fruit.
(a) Sets of two ripe fruits of M82, cd1, cd2 and cd3 3 weeks following fruit detachment; (b) percentage weight loss of M82 and cd fruits over a 10-week period.

Another important function of the cuticle is to protect the plant from pathogens: to test the susceptibility of the cd fruits to opportunistic microbes, ripe fruit were stored at room temperature in high humidity conditions, and infection rates were recorded. Fruits of the cd lines started to show symptoms of infection after 12 days of storage, and most were infected by 25 days (Table 6). The first signs of infection on any M82 fruit were not apparent until 22 days, and by 30 days only 13% of M82 fruits were infected (Table 6), with most fruit showing no symptoms even after 10 additional days. To test pathogen–fruit interactions in a more controlled way, and using a tomato-associated pathogen of commercial importance, ripe M82 and mutant fruits were inoculated with spores of the fungus Botrytis cinerea, the causal agent of gray mold. Droplets of inoculum were applied ectopically to the fruits, which were then stored at room temperature in high humidity conditions. No infection or any other symptoms were observed with M82 fruits, whereas some fruits of each cd line became infected, although cd1 fruits (15% infected) were far less susceptible than those of cd2 or cd3 (38 and 56%, respectively) (Table 6).

Table 6.   Percentage of fruits infected by opportunistic pathogens after harvest and storage at 100% humidity at room temperature, and percentage of fruits that were infected by Botrytis cinerea 7 days after inoculation
LineOpportunistic microbial infectionB. cinerea inoculation
No. of fruit% infected by 20 days from harvest% infected by 25 days from harvest% infected by 30 days from harvestNo. of fruit% of infected fruits
  1. The numbers of fruits are given in parentheses.

M82300 (0)3 (1)13 (4)34 0 (0)
cd11741 (7)53 (9)76 (13)3415 (5)
cd22737 (10)70 (19)93 (25)3438 (13)
cd31338 (5)62 (8)69 (9)3456 (19)

NMR analysis of cd fruit cuticles suggests differences in the organization of hydrophobic and hydrophilic domains

One factor that might contribute to the differences in dehydration rate among the tomato fruit lines is the relative distribution and interactions between hydrophobic and hydrophilic domains. To test this hypothesis, the abundance of potentially important functional groups and polymeric interactions in isolated dewaxed cuticles from RR stage M82 and cd fruits were analyzed by NMR spectroscopy.

Typical solid-state CPMAS 13C NMR spectra for M82 and mutant (cd2) tomato cutin are shown in Figure 6, illustrating the ability to resolve chemically distinct carbon moieties [C=O, aromatics/alkenes, CHO, CH2O and (CH2)n], and to discern clear differences in relative signal intensities for long-chain aliphatic structures (A, 30 ppm) and various oxygenated carbon moieties (O, 50–100 ppm). High-fidelity DPMAS experiments provided integrated signal intensity ratios between these carbon-containing functional groups (Figure 6). Two NMR spectral comparisons were made for enzymatically isolated cuticles. First, the ratio of oxygenated aliphatic carbons (CHO + CH2O) to long-chain aliphatics [(CH2)n] (O/A; Figure 6b) was measured, and was found to be elevated by more than twofold for each mutant: an observation that is attributable to a large percentage of both CHO and CH2O groups, compared with the total carbon signal (not shown). For cd2, the O/A ratio was also increased by an anomalously small percentage of aliphatics (not shown). Second, this trend was similar for carbonyl groups (C=O) to aliphatic ratios, which were higher in the mutants than in M82, but were similar between the mutants (Figure 6c). These analyses also showed an increase in multiply bonded or cross-linked structural elements with respect to aliphatic chains (Figure 6d).

Figure 6.

 NMR analysis of isolated red ripe fruit cuticles from M82 and the cd lines.
(a) 75-MHz CPMAS 13C NMR spectra and peak assignments for tomato cutins from M82 and cd2 fruits; 13C NMR peak ratios of (b) oxygenated aliphatic carbons (CHO + CH2O); (c) carbonyl groups (C=O); (d) or multiply bonded to long-chain aliphatics [(CH2)n]; (e), variation in the fraction of liquid-like aliphatic chains (13C NMR resonances at 20–40 ppm), for cutins (solid) and cuticles (empty) of M82 and cd fruits.

Finally, our observations of anomalies in chemical composition for the cd lines were extended to examine possible consequences for molecular organization. For instance, we questioned whether aliphatic chains with protruding hydroxyl groups or midchain cross links might pack less efficiently, or if well-mixed cutin and wax constituents could form a flexible, resilient composite. The fractions of liquid-like alkyl chains for both enzymatically isolated M82 tomato fruit cuticles and dewaxed cutins (Figure 6e) were at least threefold greater in the mutant cuticles compared with M82. For the cuticles, the greater average flexibility of the cutin and wax alkyl chains in the cd samples (from NMR) contrasted with enhanced surface stiffness (from AFM), a reasonable result because the latter nanoindentation measurements sense the mechanical resistance of the waxy cuticular overlayer, which is most abundant in the mutants.

The three cd mutations map to different loci, and positional cloning of CD2 suggests it encodes a homeodomain protein

The three mutants (cd1, cd2 and cd3) were each crossed with the wild tomato species Solanum pimpinellifolium (LA1589), and F2 plants were genotyped using genetic markers across the tomato genome, whereas fruits from the F2 populations were phenotyped for water loss rates and cuticle thickness. CD1 was mapped to the top end of chromosome 11, CD2 was mapped to chromosome 1 and CD3 was mapped to chromosome 8 (Figure 7a). Further fine mapping of cd2 localized it within a single BAC (SL_MboI0026F20), and finally localized it to a region of approximately 8 kb (Figure 7a). This area contained only one predicted gene, which was annotated as a homolog of the Arabidopsis gene ANTHOCYANINLESS2 (ANL2), a homeodomain gene that has been associated with anthocyanin distribution in epidermal cells (Kubo et al., 1999) (Figure 7b). A comparison of the sequence of the 8-kb regions from M82 and cd2 revealed only one nucleotide difference: a G → A substitution that is predicted to cause an amino acid substitution of a conserved glycine to an arginine at position 736 of the protein (Figure 7b,c). The CD2 protein contains two predicted functional domains: a ‘homeodomain’ DNA binding domain and a ‘StAR (steroidogenic acute regulatory protein) related lipid-transfer (START)’ domain, the function of which in plants is not clear, although it has been suggested to be involved in the binding of regulatory lipids/sterols (Schrick et al., 2004). In the Arabidopsis genome, 16 genes encode proteins with a similar structure, forming the class-IV homeodomain-leucine zipper family, or HD-GL2 (Nakamura et al., 2006). Interestingly, the expression of many of these genes, and the functions of the corresponding proteins, has been associated with the development of cells in the outer layers of the plant (Nakamura et al., 2006). Examples include ANTHOCYANINLESS2 (ANL2), which was connected to anthocyanin distribution in epidermal and subepidermal cells (Kubo et al., 1999), and GLABRA2 (GL2), which is involved in determining the developmental fate of trichomes and non-root hair cells (Di Cristina et al., 1996; Rerie et al., 1994). Moreover, promoter-GUS expression analyses of genes from this family found that although each is expressed in different organs, most are expressed in the outermost cell layers (Nakamura et al., 2006). It has been suggested that members in the HD-GL2 family regulate transcription, and, accordingly, nuclear localization has been shown for GL2 (Szymanski et al., 1998), and DNA binding sites were found for three members of the family: GL2 (Ohashi et al., 2003), MERISTEM LAYER1 (ATML1; Abe et al., 2001) and PROTODERMAL FACTOR2 (PDF2; Abe et al., 2003). The functions of START domains are less known, and although they have been shown to bind cholesterol, lutein or ceramide in mammals, the amino acid sequences of these domains are not sufficiently conserved with those of plants to allow for the prediction of ligand(s) (Schrick et al., 2004). The cd2 mutation is in the C terminus of the protein, not in the homeodomain or START domain (Figure 7c), which provides no insight into the consequences of the mutation, but we note that the mutated glycine is completely conserved in all HD-GL2 family members from Arabidopsis and homologs from numerous plant species.

Figure 7.

 Mapping of the cd mutants. (a) cd2 was mapped to chromosome 1 between CAPS markers T1162 and C2_At4g01210, cd3 to chromosome 8 between markers C2_At4g11560 and TG294, and cd1 to chromosome 11 between TG497 and C2_At5g04420. Fine mapping of cd2 on chromosome 1 was performed with the CAPS markers indicated on the map. BAC MboI026F20 was selected and cd2 was mapped to a DNA fragment of approximately 8 kb; (b) Schematic representation of CD2 showing domain organization and the location of the cd2 mutation. aa positions are indicated below. The total predicted length of CD2 is 821 aa.


Three tomato mutants with a glossy fruit phenotype were identified, and allelic tests showed that each line carries a mutation in a different genetic locus. An analysis of the chemical composition of fruit cuticle from the three lines revealed a dramatic reduction in cutin content (Table 1), resulting in extremely thin cuticles with abnormal ultrastructures (Figures 2 and 3), and a greater surface stiffness. NMR analysis further suggested altered molecular organization and packing, as increases in oxygen-linked, potentially cross-linked, and multiply bonded carbon moieties of the mutant cuticle constituents were observed. To our knowledge, such a large decrease in cutin content has not been previously described in any other mutant. We suspect that such a reduction of cutin content in vegetative tissues would result in lethality, or a very severe phenotype, illustrating the potential of the tomato fruit as a model system to study cuticle biosynthesis, assembly and function.

The influence of the cutin matrix on cuticular transpiration

Despite the striking reduction in cutin content, the water loss rates of cd2 and cd3 fruits were not very different from M82. It is important to note that although the cutin content of the fruit cuticle of all mutants was dramatically reduced, the residual cutin was still sufficient to form a visible cuticle. For comparison, the cutin content of the cd2 fruit cuticle, the mutant with the lowest levels, was approximately 15 μg cm−2, which is <2% of M82 fruit, and yet is still much more than the values reported for Arabidopsis stems or leaves (0.2–3.3 μg cm−2) (Xiao et al., 2004; Franke et al., 2005; Kurdyukov et al., 2006b; Bird et al., 2007; Panikashvili et al., 2007). This suggests that either the cutin matrix does not play a central role in limiting water loss, inferring that waxes are the primary barrier, or that the residual level was sufficient, and that tomato fruit have far more cutin than is necessary for resisting desiccation. Indeed, of the three mutants, cd1 fruits showed both the highest water loss rate and level of cutin, further demonstrating the lack of correlation between cutin content and water permeability. Images of cd1 fruit cuticles obtained by light microscopy, electron microscopy and AFM did not reveal any microfissures, as has been described for a more rapidly desiccating Cwp1 tomato genotype (Hovav et al., 2007), or other structural disruptions that might explain this phenotype.

Another explanation for the high water loss rates of cd1 fruits might be the abnormal wax composition. Previous studies have not identified a clear relationship between wax composition and water permeability (Riederer and Schneider, 1990; Hauke and Schreiber, 1998; Kissinger et al., 2005), but characterization of the lecer6 tomato mutant, which is deficient in very-long-chain fatty acid β-ketoacyl-CoA synthase (Millar et al., 1999) suggested such a correlation (Leide et al., 2007). lecer6 fruit cuticles have elevated levels of amyrins but lower levels of alkanes of chain length >30, compared with the wild type, and higher water loss rates were reported in the mutant fruit (Vogg et al., 2004; Leide et al., 2007). Furthermore, a negative correlation was found between water permeance and the levels of cuticular alkanes. Interestingly, RR-stage cd1 fruit cuticles have similarly altered levels of amyrins and alkanes of chain length >30 (Tables 3 and 4). Leide et al. (2007) referred to a model in which the waxes are organized as islands of crystalline platelets in an amorphous material (Geyer and Schonherr, 1990; Reynhardt and Riederer, 1994; Riederer and Schreiber, 1995; Kosma and Jenks, 2007), and water molecules move through the amorphous matrix but not through the crystalline regions. Furthermore, triterpenoids are predicted to be localized in the amorphous matrix, and alkanes are predicted to be localized in the crystalline regions. Hence, a reduction in alkanes, accompanied by an increase in triterpenoids, would result in an increase in the amorphous portion of the wax, and a decrease in the crystals, potentially leading to increased water flux. This model is supported by our results, as is the idea that cutin does not function as a significant hydrophobic barrier, but rather provides a framework into which the intracuticular wax molecules are deposited, providing a structure that can effectively restrict water movement. We note that, unlike cd1, the lecer6 has a thicker cuticle than its background genotype (Vogg et al., 2004; Leide et al., 2007), and the two mutations map to different chromosomes (data not shown), so we exclude the possibility that cd1 is altered in the LeCER6 gene.

Cutin and plant pathogen resistance

Fruits of the cd lines showed an increased susceptibility to opportunistic saprophytes, and to B. cinerea, under controlled inoculation conditions (Table 6). Susceptibility to pathogens has been shown to be influenced by the roughness of the surface of a plant organ (Zabka et al., 2008); however, the roughness of the cd cuticles was shown by AFM to be similar to that of M82 (Figure 4b,e,h,k and data not shown), so it is less likely that the increased susceptibility of their fruit was caused by changes in surface roughness as a result of the different wax composition. Rather, the data suggest that the increased susceptibility resulted from a massive reduction in cutin content. This phenomenon may, at least in part, explain the relatively high thickness of the tomato cuticle, and the relatively low occurrence of microbial infection of nutritionally rich ripe fruits under normal conditions.

The contributions of cutin and wax content to cuticle biomechanical properties

Biomechanical studies of isolated whole cuticles from tomato fruits (Lopez-Casado et al., 2007) have suggested that cutin imparts the cuticle with its viscoelastic behavior (low elastic modulus and high strain values). In its absence, the cuticle is predicted to be stiffer and less elastic. The cuticles of the cd lines were too fragile to perform macro-scale biomechanical tests, but nano-scale AFM analyses showed that the membrane is harder in the presence of waxes, although dewaxed samples from the mutants and M82 showed similar biomechanical behavior. Based on the compositional differences between the mutants and M82, the observed changes in biomechanical behavior can most likely be attributed to the relative cutin content, as well as the wax composition of the samples. The high values of the surface modulus for the cd mutants correlated with their extremely low cutin content, and with the greater proportions of CHO + CH2O and carbonyl species, compared with aliphatics (Figure 6). In this scenario, the cuticles with the greater proportions of these species would be more crystalline, and would thus exhibit larger moduli.


The data presented here help resolve some of the questions regarding the contribution of cutin to cuticle properties and function; however, the identification of these cutin-deficient mutants also provides an opportunity to better understand cutin biosynthesis. We found cd2 to be mutated in a gene that is presumed to encode a transcription factor of the HD-GL2 family. This, together with the observation that most HD-GL2 genes are predominantly expressed in surface cell layers, supports a model wherein CD2 is a key regulator of cutin biosynthesis in tomato fruit, and this will provide a platform for future studies. We also anticipate that the identification of the mutated loci for cd1 and cd3 will reveal two more genes that are essential in this poorly understood pathway, and genetic mapping is already underway.

Experimental Procedures

Plant materials

Seeds of the tomato (S. lycopersicum) lines described here [wild-type cultivar M82, and the derived mutated lines e4247m1 (cd1), e4393m2 (cd2) and n3056m1 (cd3)] were obtained from the Genes that Make Tomatoes germplasm collection (Menda et al., 2004; Plants were grown in a glasshouse in Ithaca, New York, under 16-h of light and 8-h of dark, using standard practices. Fruits were harvested at the SG developmental stage when the fruit length reached 2–3 cm, at the MG stage when fruits reached full size, but were still green, and at the RR stage 4–5 days after the color break. F2 mapping populations were grown in the field (Freeville NY, summer 2007).

Genetic mapping

Three F2 mapping populations were created by crossing each cd mutant to the wild species S. pimpinellifolium (accession LA1589). Genomic DNA was extracted from a young leaf of each seedling according to the method described by Fulton et al. (1995). For cd1, 94 plants of the F2 population were phenotyped for fruit water loss: 21 plants that exhibited high water loss rates and nine plants that exhibited low water loss rates were genotyped for 48 markers across the tomato genome. For cd2, fruit cuticles from 94 plants were phenotyped and the plants were genotyped for 60 markers across the genome. For cd3, cuticles from 47 F2 plants were phenotyped and 13 plants were found to have the cd3 phenotype. Of these, 11 plants as well as an additional seven plants that had normal (S. pimpinellifolium) cuticles were genotyped for 48 markers across the genome. Genotype and phenotype data were analyzed using qgene software (Nelson, 1997). For the fine mapping of cd2, 940 plants of the cd2 F2 population were screened for recombination events between CAPS markers T1162 (position 1.059) and C2_At4g01210 (position 1.081). Fine mapping located the cd2 mutation on a single BAC (SL_MboI0026F20). Using the BAC sequence (GenBank Accession AC217915), new CAPS markers were designed and cd2 was mapped to a region of approximately 8 kb.

Water loss measurements

A total of 41–48 fruits of each line were picked at the RR stage, and were stored at room temperature for 10 weeks. Fruit weight was recorded every week, and water loss was calculated as a percentage of weight loss.

Cutin monomer and wax analysis

For compositional analysis, cuticles were enzymatically isolated (Schönherr and Riederer, 1986) from tomato fruit exocarp discs by incubation at 35°C in 2% (v/v) pectinase (EC; Sigma-Aldrich,, 0.1% (w/v) cellulase (EC; TCI America, in 20 mm citric acid, pH 3.7, containing 0.001% (w/v) phenylmercuric nitrate to prevent microbial growth. Isolated cuticles were washed in an acetone series and refluxed for 24 h in a Soxhlet apparatus with chloroform:methanol (1:1) and 50 mg L−1 butylated hydroxytoluene. After washing with methanol to remove chloroform, cuticle composition was analyzed as described in Saladiéet al. (2007).

The cutin monomer content of epidermal peels, including intracellular lipids, was analyzed based on a depolymerization method (Franke et al., 2005), with modification. Briefly, MG fruits were dewaxed by a 1-min submersion in chloroform. Epidermal tissues were peeled from the fruit equator and 0.6–1-cm2 disks were punched out, blotted dry and subjected directly to transmethylation reactions, consisting of 6.5 ml of 2.7 N methanolic hydrochloride (MeOH-HCl) containing 7% (v/v) methyl acetate at 60°C. Methyl heptadecanoate was used as an internal standard. After 16 h, reactions were allowed to cool to room temperature, and were terminated by adding 6 ml of saturated, aqueous NaCl, followed by two extractions (10 ml) with distilled dichloromethane to remove methyl ester monomers (Bonaventure et al., 2004). The organic phase was washed three times with 0.9% aqueous NaCl (w/v), then dried with 2,2-dimethoxypropane under nitrogen gas. Derivatization and subsequent GC analysis was performed as described above for cutin monomer analysis. All values represent the mean of four replicates. The presence of C18:1, C18:2 and C18:3 fatty acids indicated that intracellular lipids and potential intracellular cutin were not substantially removed by the 1-min chloroform submersion used for wax extraction.

Cuticular wax analysis was performed as described in Saladiéet al. (2007) with slight modification. GC was carried out with temperature-programmed automatic injection at 80°C, then held for 2 min at 80°C, raised by 40°C min−1 to 200°C, held for 2 min at 200°C, raised by 3°C min−1 to 320°C and held at 320°C for 30 min. Triterpenoids were putatively identified based on spectra described by Bauer (2002).

Pathogen treatments

Botrytis cinerea (isolate BCT-862) isolated from tomato fruit was grown on potato dextrose agar (PDA) plates until they reached a state of heavy sporulation. Inoculum was prepared by gently rinsing the plates with water, and inoculum droplets containing approximately 2000 spores were applied ectopically to a group of 8–10 RR fruits of the M82 and cd lines. Fruits were stored at room temperature in a sealed container at saturating humidity. To assess opportunistic microbial infection, 8–10 RR fruit from the M82 and cd lines were gently rinsed with water, dried and stored as described above.

Scanning electron microscopy (SEM)

Cubes of RR-stage fruit pericarp were frozen in liquid propane, freeze dried, mounted on aluminum stubs and sputter coated with gold : palladium 60:40 using a Denton Desk II coater (to a metal coat thickness of approximately 10 nm). Microscopic observations at the edges of the tissue break-line were made with a Leica 440 scanning electron microscope using an accelerating voltage of 5 kV (Leica,

Transmission electron microscopy (TEM)

Pieces of pericarp (cubes of approximately 5-mm per side) were excised from the equator of three replicate fruits of each genotype at the SG, MG and RR stages. Samples were prepared for TEM using microwave technology (oven model 3450; Ted Pella, under vacuum, to improve cuticle fixation. Samples were fixed in primary fixative containing 2.5% (v/v) glutaraldehyde and 2% (v/v) formaldehyde in 0.05 m phosphate buffer, pH 6.8 (PB) (Karnovsky, 1965), and were then subjected to two 40-sec microwave treatments using the low power setting, allowing 3 min between treatments. Samples were washed with PB and post-fixed in 1% (v/v) osmium tetroxide and 1.5% potassium ferricyanide in water, followed by one 40-sec microwave treatment with vacuum, as described above, for osmium fixation. Samples were washed with water (2 × 40-sec microwave treatments) and then dehydrated through a gradient ethanol series (40 sec for each change). Samples were infiltrated with propylene oxide-Spurr resin mixtures and polymerized in 100% resin for 48 h at 60°C. Ultrathin sections (80–100 nm) were stained with 2% uranyl acetate and Reynold’s lead citrate, and were viewed with a Philips CM-10 Biotwin transmission electron microscope (FEI Co., Images were analyzed using ImageJ software (Abramoff et al., 2004).

Light microscopy

Light microscopy and staining were conducted as described in Buda et al. (2009), accompanying paper).

Enzymatic cuticle isolation

To obtain isolated cuticles for AFM and NMR analyses, fruit pericarp sections were incubated in a mixture of 40 U ml−1 cellulase (EC; TCI America), 10 U ml−1 pectinase (EC; TCI America) in sodium citrate buffer (50 mm, pH 4.0) and 1 mm NaN3 for 7–10 days at 32°C. Cuticles were then washed in water incubated again in fresh enzymatic buffer until clean, and were then washed in water again and dried at room temperature.

Atomic force microscopy (AFM)

Surface topography and nanoindentation properties of isolated cuticular samples were determined using a commercial AFM (Alpha300 S; WITec, that includes a confocal (Raman and fluorescence) microscope. The enzymatically isolated cuticle samples were attached to glass slides with double-sided tape and imaged under ambient conditions (20 ± 1°C) at approximately 45% relative humidity. Images were collected in both tapping mode and contact mode. Tapping mode images were collected using ultrasharp silicon AFM tips (VISTA probes; NanoScience Instruments,, with nominal tip radii of <10 nm and a lever resonance frequency of approximately 300 kHz (force constant of approximately 40 N m−1). To determine the representative roughness of the surface, 0.5 × 0.5 μm2 AFM topographic images were taken at random locations on the cuticle surface, and the root-mean-square roughness of the surfaces were then determined.

For contact mode imaging and nanoindentation experiments, Si tips were used (Mikromasch, As the spring constants and tip dimension can vary between probes, the tip size and lever spring constant were determined experimentally for each cantilever tip used. The normal force constants of the levers were determined based on the measurements of the cantilever spring constant and a standardized lever of a known spring constant (Tortonese and Kirk, 1997), and were found to exhibit a typical force constant of 0.35 ± 0.06 N m−1. The tip shape and radius of curvature were approximated by the imaging of an Sr3TiO3 (305) single crystal (Sheiko et al., 1993), and were found to exhibit a typical radius of curvature of 68 ± 9 nm. AFM images were collected with scan sizes ranging between 2 × 2 μm2 and 70 × 70 μm2, at a typical scan rate of 0.2 Hz.

Measurements of the surface mechanical properties were obtained by nanoindentation measurements. Cuticle samples were first imaged in contact mode, and then their nanomechanical properties were evaluated by indentation, with reference to an ideally hard surface: in this case a cleaned Si(100) wafer. To determine the extent of tip indentation into the cuticular sample, the slope of the force–distance spectrum during retraction was taken while the tip was in contact, and the indentation was determined at a fixed load of 0.5 nN for each sample. Based on these indentation experiments, the Young’s modulus of the cuticular surface could be determined assuming a Hertzian contact (Round et al., 2000). Measurements were made in at least 10 different locations both in the surface depression on the cuticle and on the ridges that corresponded to the boundaries of the underlying cells. Soxhlet extraction (Pacchiano et al., 1993) was used to dewax the tomato cuticular samples, which involved successive refluxing (approximately 6 h for each solvent) with hot methanol, dichloromethane and tetrahydrofuran (THF).

NMR analysis

Intact or dewaxed enzymatically isolated cuticles (7–10 mg of powdered samples) were analyzed using NMR spectroscopy. To obtain quantitative estimates of the various carbon types, integrated areas were measured from direct polarization-magic-angle spinning (DPMAS) 13C NMR (Duer, 2004) using a Varian 600-MHz DirectDrive NMR System (, with a recycle time of 100 sec between successive acquisitions. Similar compositional trends for replicate tomato samples and error limits of 3–18% in repeated experiments with M82 and cd lines were established using cross polarization (CPMAS) 13C NMR spectra obtained at both 300 MHz (on a Varian Unityplus spectrometer) and 600 MHz (NMR System). These latter experiments used a 2-ms mixing time and a 2-sec recycle time in order to ensure consistent cross polarization of the carbons and spin relaxation of the 1H nuclei, between acquisitions, respectively. DPMAS and CPMAS NMR spectra were acquired with a comparable signal-to-noise ratio of >50:1 for the (CH2)n groups. The two-pulse phase-modulation (TPPM) decoupling method and a 1H field of 50 kHz were used, unless otherwise stated. The typical spinning speeds were 8 kHz.

The proportions of liquid-like and solid-like aliphatic chain carbons were estimated by comparing integrated areas of the spectral region between 20 and 40 ppm for DPMAS spectra acquired with low-power decoupling (1H field strength corresponding to 5 kHz, mobile carbons only) and high-power decoupling (1H field strength corresponding to 50 kHz, all carbons) during NMR signal acquisition. To determine 1H rotating-frame relaxation times, T(H), the decay of the carbon signal intensity, was monitored with increasing 13C–1H contact time (τ) in a CPMAS experiment (Schaefer and Stejskal, 1979).


The authors would like to thank Dr J. Lorbeer for providing the B. cinerea culture, Dr D. Zamir for tomato germplasm, and Dr S. Tanksley and Emma Flemmig for assistance with the CD2 mapping. The project was supported by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant number #2006-35304-17323; by the CUAES Hatch Project, NYC-184485; by the United States–Israel Binational Science Foundation (award #2005168); and by NSF grants #MCB-0134705, MCB-0843627 and DBI-0606595. TI was supported by a postdoctoral award no. FI-375-2005 from the United States–Israel Binational Agricultural Research and Development Fund.

Accession numbers: sequence data from this article for the cDNA of CD2 can be found in the GenBank/EMBL data libraries under accession number GQ222185.