The adaxial surface of fully expanded, mature, undamaged leaves of E. camaldulensis Dehn and E. globulus Labill was analyzed either intact or after enzymatic isolation of the cuticle. Leaves were collected at late summer from the apex of external, west-facing shoots of 30–35 years old trees growing at the Forest Engineering School Arboretum (Technical University of Madrid, Spain). For each species, 6 and 30 leaves were selected for intact leaf and isolated cuticle analyses, respectively.

Prior to cutting the central part of the leaf blades in small pieces for cuticle isolation, leaf margins and midribs were removed with a scalpel. Leaf cuticles were subsequently isolated in a solution containing 5% cellulase and 5% pectinase (Novozymes, Bagsvared, Denmark), plus 1% polyvinylpyrrolidone (Sigma–Aldrich, Munich, Germany) and 2 mM sodium azide, which was set to pH 5.0 by adding sodium citrate (Guzmán et al., 2014a). Tissues were maintained in solution (changed after 2 weeks) at room temperature (23–25^∘C) for 1 month, manually shaking the flasks at frequent time intervals. After the extraction period, clean intact adaxial cuticles were selected, thoroughly washed in deionized water, air-dried, and stored for further use.

Isolated cuticles were cut into small pieces, wrapped in filter paper and successively extracted in 150 mL of dichloromethane (6 h; Sigma–Aldrich), ethanol (12 h; Sigma–Aldrich) and distilled water (24 h) using a Soxhlet apparatus. Solvents were subsequently evaporated to dryness with a rotary evaporator and the material extracted was weighed and stored. Soxhlet extracted (SE) cuticles were also stored for further analysis.

The cuticular material extracted in dichloromethane was analyzed by gas chromatography-mass spectrometry (GC-MS). Samples were filtered using polytetrafluoroethylene (PTFE) filters prior to injection in a GC-MS apparatus (GC Varian CP-3800 coupled to a MS ion trap Varian Saturn 2200, GC-ITMS, automatic injector COMBI-PAL; CTC Analytics, Zwingen, Switzerland). The chromatographic conditions were as follows (86 min per run): injection volume 1 μL, with Helium as carrier gas (1.2 mL min^-1), and injector temperature 290^∘C. The column (SLB-5 ms fused silica capillary column, 30 m × 0.25 mm × 0.25 μm film thickness; Supelco, Sigma–Aldrich) was set to 40^∘C (3 min), heating rate of 10^∘C min^-1 up to 290^∘C (18 min), and 20^∘C min^-1 up to 310^∘C (13 min). The MS conditions were: 70 eV ionization voltage, 150^∘C trap temperature, 20–550 units of mass scan range with 8 min solvent delay. The compounds were identified and quantified by comparing their mass spectra with NIST library spectra and analyzing the corresponding peaks of GC-MS ion chromatograms.

For the two species, 80 mg of dehydrated (oven-dried at 40^∘C before use), SE cuticles were used for cutin depolymerization. Samples were refluxed in 5 mL of 0.1 M sodium methoxide solution (Sigma–Aldrich) in methanol by heating in an oil bath at 80^∘C for 3 h, and continuous stirring. For separation of de-esterified (filtered) and residual (non-filtered) material from the methanolysis process, mixtures were vacuum filtered using 1 μm pore PTFE filters. The residual material was oven-dried under vacuum at 40^∘C overnight, weighted for the determination by mass-loss, of the total content of cuticle de-esterified material (one replicate) and saved for further examination.

The total volume of de-esterified material was divided for glycerol and cutin monomer and precursor identification (1:4, v:v), and handled separately. For glycerol analysis, the de-esterified aliquot was sampled as such. For cutin monomer identification, the respective aliquot was acidified to pH 6 with 0.05 M sulfuric acid in methanol, and the solvent was subsequently removed in a rotary evaporator at 50^∘C. The obtained residue was partitioned in dichlorometane:water (1:1, v:v; 35 mL each) using a separating funnel. The upper aqueous, clear phase was discarded and the organic, opaque phase located below was partitioned again with the same volume of water.

For GC-MS determination, aliquots (1 mL) of the de-esterified material and of the accumulated organic phases were collected. Aliquots were dried under a N₂ flow and derivatized with N,O-bis(trimethylsilyl)trifluoroacetamide with 1% trimethylchlorosilane (BSTFA with 1% TMCS, Sigma–Aldrich) and pyridine at a 1:1 (v:v) ratio. After 15 min staging in an oven at 60^∘C, solutions were analyzed with an Agilent 5973 MSD GC-MS (Agilent, Santa Clara, CA, USA) in a DB-5 ms column (60 m, 0.25 mm i.e., 0.25 μm film thickness; Agilent), with the following GC conditions: injector 320^∘C, initial oven temperature 100^∘C (5 min), heating rate of 8^∘C min^-1 up to 250^∘C, and 2.5^∘C min^-1 up to 320^∘C (20 min). The MS conditions were: source 220^∘C, quadrupole 150^∘C, electron ionization energy 70 eV. Cutin monomers were identified by library search (NIST 98, Wiley 6), comparison to standards and/or mass spectra interpretation (Graça et al., 2002). Monomer quantification was calculated from the areas of the corresponding peaks in the CG-MS ion chromatograms. As internal standards for glycerol and cutin monomers, we used 1,12-dodecanediol (0.05–0.3 mg; ≥98.5% GC, Sigma–Aldrich) and methyl 12-hydroxystearate (0.05–0.1 mg; ≥99.0% GC, Sigma–Aldrich), respectively. A response factor of 0.33 in relation to the internal standard (1,12-dodecanediol) was used for the quantitation of glycerol. The internal standards were diluted in methanol, homogenized in a warm bath, and added to the de-esterified material before derivatization (Graça and Lamosa, 2010).

Intact leaves, isolated cuticles, SE cuticles, and the residues obtained after de-esterification of SE cuticles (subsequently called SED residues) were examined by TEM and scanning electron microscopy (SEM).

Gold-sputtered adaxial leaf surfaces and both the inner and outer cuticle and residue surfaces (three replicates per sample) were observed with a Philips XL30 SEM (Eindhoven, The Netherlands) microscope.

For TEM observation, samples were cut into 4 mm² pieces and fixed in 2.5% glutaraldehyde-4% formaldehyde freshly prepared from paraformaldehyde (both from electron microscopy sciences (EMS), Hatfield, PA, USA) for 6 h at 4^∘C, rinsed in ice-cold phosphate buffer, pH 7.2, four times within a period of 6 h and left overnight. Tissues were post-fixed in a 1:1 2% aqueous osmium tetroxide (TAAB Laboratories, Berkshire, UK) and 3% potassium ferrocyanide (Sigma–Aldrich) solution for 1.5 h. They were then washed with distilled water (x3), dehydrated in a graded series of 30, 50, 70, 80, 90, 95, and 100% acetone (x2, 15 min each concentration) and embedded in acetone-Spurr’s resin (TAAB Laboratories) solutions [3:1, 2 h; 1:1, 2 h; 1:3; 3 h (v:v)] and in pure resin overnight at room temperature (25^∘C). Tissues were finally embedded in blocks which were incubated for 3 days at 70^∘C. Ultra-thin sections of three leaf and cuticle pieces per treatment were cut, mounted on nickel grids and post-stained with Reynolds lead citrate (EMS) for 5 min. They were subsequently observed with a Jeol 1010 TEM (Tokyo, Japan) operated at 80 kV and equipped with a CCD megaview camera.

The thickness of entire cuticle transversal sections and of the different cuticular regions was measured on TEM micrographs of leaves, intact and SE cuticles, and SED residues using siViewer (Olympus Soft Imaging Solutions, Münster, Germany) and ImageJ 1.45s (National Institutes of Health, Bethesda, MD, USA) softwares. Measurements were performed at periclinal areas, excluding zones located in the vicinity of cuticular pegs, and also the EW layer. Micrographs corresponding to extremely thick or thin cuticle sections or with no clear limits between adjacent cuticular regions were discarded. Finally, 15 micrographs of each sample (corresponding to three 4 mm² cuticle pieces) were selected for thickness measurement, with 10–20 repetitions.

Infrared spectra of cuticle samples (i.e., intact, SE cuticles, and SED residues), were obtained with an attenuated total reflectance (ATR) accessory (MIRacle ATR; PIKE Technologies, Madison, WI, USA) coupled to a FTIR spectrometer (Nicolet Nexus 670–870; Thermo Fisher Scientific, Waltham, MA, USA). Spectra of intact samples were recorded in transmission mode in the range 4000–400 cm^-1 with 4 cm^-1 resolution and accumulating 64 scans, and analyzed with Omnic v4.1b (Thermo Fisher Scientific) software. Two replicates per sample were analyzed.

The lamellate CP regions observed in intact leaf cuticles were replaced by amorphous, electron-lucent bands after Soxhlet extraction. These bands had a thickness in the same range or slightly below that of the corresponding intact CP. By contrast, no remarkable changes were noticed between the CL of intact and SE cuticles. X-ray diffraction and TEM studies performed on Hedera helix leaf cuticles both intact and after solvent extraction revealed similar results, suggesting an intermolecular disorganization of the cuticles, particularly in the CP (Viougeas et al., 1995). The appearance of the CL was also preserved (Viougeas et al., 1995), alike in Agave americana extracted cuticles (Wattendorff and Holloway, 1982). The CP thickness variation may be due to several phenomena which may also affect the CL region, such as the loss of cuticular material and structure, or the potential shrinkage of the tissues as affected by Soxhlet extraction or TEM sample preparation procedures. The influence of cuticle position within one or between different leaves should also be considered (Riederer and Schönherr, 1988).

No clear relationship between soluble lipids composition and cuticle internal morphology could be established. As an exception, E. camaldulensis epicuticular plates may correspond to hentetracontan-1-ol (Jeffree, 2006). The chemical nature of the CP lamellae has been attempted to be identified in a few investigations (see Jeffree, 2006), but remains unclear so far. Our results indicate that the electron-lucent lamellae of E. camaldulensis and E. globulus CP are partially composed of soluble lipids as suggested by Domínguez and Heredia (1999) for the Clivia miniata leaf cuticle, but the presence of other free or chemically bound compounds cannot be discarded. The granular structures covering the inner, intact cuticle surface which were removed during Soxhlet extraction may be associated with soluble intracuticular lipids. The remaining granules may correspond to insoluble lipids and/or cutin (i.e., cutinsomes; Heredia-Guerrero et al., 2008).

The de-esterification step led to substantial structural changes as compared to the Soxhlet extraction process. While FTIR spectra confirmed the major polysaccharide nature of the SED residues (Heredia-Guerrero et al., 2014), TEM micrographs enabled us to observe the cellulose architecture of the external cell wall, with a non-ordered, outer region and a larger multi-layered, helicoidal inner region (also visible in cuticular pegs). The relative position within the cuticles and measurement of TEM micrographs of intact, SE cuticles and such SED residues indicate that these regions may be ascribed to the CP and the CL, respectively.

One or the two cellulose fibril orientation patterns found in the SED residues have been previously reported for the epidermal cell wall of a few species and organs (e.g., Roland et al., 1982; Kutschera, 2008), and also for the H. helix leaf cuticle (Canet et al., 1996). Thereby, the structure of the cellulose network seems to leave an imprint on the morphology of cuticular cross-sections and may contribute to the different ultra-structure of the CP (possibly with randomly oriented cellulose) and CL (with ordered cellulose). However, it is also possible that the removal of cuticular material may have disrupted the orientation and arrangement of the cellulose network. Other factors should be also taken into account, such as those affecting SE cuticles.

Together with cellulose, it is likely that other cuticular polysaccharides, such as pectins or hemicelluloses, extend throughout the cuticle, as derived from cellulose and pectin distribution in intact leaf cuticles of three model species (Guzmán et al., 2014b). Under certain conditions, these polysaccharides might form a gel-like phase with embedded cellulose fibrils, as described for plant cell walls (Cosgrove, 2005; Burton et al., 2010). By analogy with the cell wall, the cellulose fibrils may constitute the reinforcing rods of the cuticle, while the non-cellulosic phase may provide flexibility, mechanical support (Burton et al., 2010), and facilitate the transport of hydrophilic, polar compounds and water through the cuticle (Guzmán et al., 2014a,b).

A significant amount of material was removed largely from the inner cuticle region, which experienced a thickness reduction of approximately 60–80%. Working with A. americana leaf cuticles, Wattendorff and Holloway (1982) reported a thickness reduction of 40% or more from the inner side after cutin depolymerization. Apart from the lower content and/or degree of polymerization of cutin located in the more exposed internal CL, which may yield this region more susceptible to degradation (Wattendorff and Holloway, 1982), the role of further chemical compounds should also be considered as discussed above.

The chemical and structural analyses performed with E. globulus and E. camaldulensis leaves, indicate that the cuticle of such species may be interpreted as a lipidized cell wall region. This result is in agreement with the viewpoint of Yeats and Rose (2013), who suggested that the cuticle can be considered a specialized lipidic modification of the cell wall. Our results show that a polysaccharide network may provide the basis for cuticle structure, a role which has been frequently assigned almost exclusively to cutin (Kolattukudy, 1970). Cuticular lipid constituents may impregnate and be intruded into such cell wall matrix, with waxes being also deposited on to the outer cuticle surface, early at epidermal cell ontogeny (Jeffree, 2006; Guzmán et al., 2014b). The occurrence of a polysaccharide network throughout the entire cuticle, with the exception of the EW layer, may have significant bio-mechanical implications and also contribute to the multi-directional transport of water and solutes as suggested by Guzmán et al. (2014a,b).

It is likely that the actual proportion of polysaccharides is seriously underestimated by current methods of cuticular handling and analysis. The importance of polysaccharides as integral, essential cuticle materials should be recognized for the development of methodologies which allow a proper interpretation of the plant cuticle in biological, chemical, and physical terms. In light of the present study, the prevailing model which conceives the cuticle as a lipidic and hydrophobic extra-cellular layer should be reconsidered, highlighting its cell wall nature and heterogeneous character in terms of polarity/apolarity and hydrophobicity/hydrophilicity. However, further studies with different plant organs, organ parts, and species will be required for proposing a broader plant cuticle model, and also for establishing a link between cuticle chemistry, structure, and function under an ecophysiological perspective.

While no specific cuticle structural patterns could be identified in association with chemical composition, the background cell wall architecture seems to have an effect on cuticle morphology. We hence conclude that the existing methodologies do not fully enable to establish a direct link between cuticle composition and structure. Therefore, the implementation of new experimental approaches also analyzing the performance of pure chemical compounds (e.g., cutin monomers) may help us elucidate the link between cuticle structure and chemistry also in relation with different plant species, organs, and habitats.