Full appreciation of the roles of the plant cuticle in numerous aspects of physiology and development requires a comprehensive understanding of its biosynthesis and deposition; however, much is still not known about cuticle structure, trafficking and assembly. To date, assessment of cuticle organization has been dominated by 2D imaging, using histochemical stains in conjunction with light and fluorescence microscopy. This strategy, while providing valuable information, has limitations because it attempts to describe a complex 3D structure in 2D. An imaging technique that could accurately resolve 3D architecture would provide valuable additions to the growing body of information on cuticle molecular biology and biochemistry. We present a novel application of 3D confocal scanning laser microscopy for visualizing the architecture, deposition patterns and micro-structure of plant cuticles, using the fluorescent stain auramine O. We demonstrate the utility of this technique by contrasting the fruit cuticle of wild-type tomato (Solanum lycopersicum cv. M82) with those of cutin-deficient mutants. We also introduce 3D cuticle modeling based on reconstruction of serial optical sections, and describe its use in identification of several previously unreported features of the tomato fruit cuticle.
The cuticle, a complex hydrophobic membrane covering the aerial surfaces of plants, has many critical functions, including water regulation, pathogen resistance and the prevention of organ fusion during organogenesis (Nawrath, 2006; Riederer and Muller, 2006). However, despite extensive study of these important developmental and protective roles using a range of experimental approaches, many aspects of cuticle structure, particularly structure–function relationships, remain poorly understood.
Early research focused on cataloging the diversity of cuticle chemical constituents (Jetter et al., 2006), and more recent genetic and biochemical strategies have revealed some of the genes involved, particularly in the early stages of cuticle precursor biosynthesis (Pollard et al., 2008; Samuels et al., 2008). However, almost nothing is known about the mechanisms that mediate and regulate the trafficking and subsequent assembly of the cuticle components into a three-dimensional (3D) matrix, the micro-heterogeneity of cuticle ultrastructure, or the restructuring that must accompany organ growth and differentiation. As more cuticle-associated genes and molecular pathways are uncovered and additional forward or reverse genetic screens are developed, it will become increasingly important to characterize cuticle architecture with a greater degree of resolution. With this in mind, advanced imaging techniques with the ability to detect subtle changes in cuticle deposition and structural organization are needed.
To date, plant cuticle imaging has mainly been accomplished using a combination of bright-field, polarized-light and fluorescence microscopy, often incorporating lipophilic dyes, such as Nile blue A (Nile blue sulfate), the Sudan stains or the fluorescent dye auramine O (Considine and Knox, 1979; Pesacreta and Hasenstein, 1999; Dominguez et al., 2008) to enhance visualization. In addition, scanning electron microscopy and transmission electron microscopy can provide useful information about cuticle architecture. For example, scanning electron microscopy has been used to quantify differences in cuticle thickness and deposition (Isaacson et al., 2009), to visualize cuticle defects (Hovav et al., 2007) and to visualize epicuticular waxes (Pighin et al., 2004; Jeffree, 2006; Bird et al., 2007). On the other hand, transmission electron microscopy has revealed some of the structural heterogeneity in the cuticular layer, based on observed layering of electron-dense material (Emmons and Scott, 1998; Jeffree, 2006). It is presumed that this reflects differences in chemical composition arising from differential impregnation of primary and secondary cell walls with cutin and other materials (Jeffree, 2006).
Such studies have provided some insights into the complexity and diversity of cuticle architecture, although there are a number of limitations inherent to two-dimensional (2D) visualization, especially with respect to obtaining quantitative information. For example, perceived cuticle thickness can often be an optical artifact of cell topography; a misperception that is easily missed when looking at a single 2D image. Plant cuticles can have complex 3D architectures, sometimes penetrating several cell layers and even apparently encasing epidermal cells. However, it is not possible to unambiguously determine the entire cuticle structure around a single cell or within a tissue section using traditional techniques. As new cuticle-associated mutations are identified and reverse genetic analyses are used to explore the functions of cuticle-associated genes, there will be an increasing need for better tools to identify and study cuticle phenotypes that are not apparent when using current methodologies. Improved cuticle imaging will therefore be increasingly important in elucidating the underlying molecular pathways of cuticle biosynthesis and assembly.
In this paper, we present a novel application for confocal-based 3D fluorescent imaging and modeling as a means to study plant cuticle architecture, and demonstrate its utility in the characterization of tomato cuticle mutants.
Results and Discussion
2D cuticle imaging
Both Nile blue A and Sudan IV staining methods were tested on tomato fruit pericarp sections to assess their general utility in studying plant cuticles. Nile blue A is dichromatic when binding lipid material, and will differentially stain neutral fats and waxes red and fatty acids and acidic precursors blue (Gahan, 1984; Horobin, 2002b). This differentiation was observed in cross-sections of breaker-stage tomato fruit pericarp stained with Nile blue A (Figure 1a) within the cuticular layers, i.e. within the portion of the cuticle below the cuticle proper (the thin, outermost differentiated layer of the cuticle, which is assumed to contain no polysaccharides) (see Jeffree, 2006; for detailed review of terminology). Specifically, the external cuticular layer (ECL) stains pink, while the internal cuticular layer (ICL) and cuticularized anticlinal pegs (APs) stain light blue. The non-specific nature of Nile blue A was found to be suitable for distinguishing plant cuticles from the rest of the tissue, but because it also has an affinity for cellulosic cell walls and cytoplasmic contents, details of the cuticle were often obscured by intense coloration of the adjacent tissue: for example, the deep blue staining seen in the outer cellulosic periclinal cell walls. Sudan IV, which has a highly lipophilic structure and minimal solubility in water (Horobin, 2002a), was highly specific to cuticular material and did not generate background staining by occupying aqueous environments such as the cellulosic cell wall, cytosol and vacuole. Sudan IV staining (Figure 1b,c) allowed us to assess cuticle thickness, the extent of cuticularization of the anticlinal cell walls, and the penetration of cuticular material into underlying cell layers (sub-epidermal deposition). These characteristics are demonstrated in Figure 1(b,c), which shows Sudan IV staining of the cuticle of M82 tomato fruit (Figure 1b) and that of the cutin-deficient tomato mutant cd2 (Figure 1c) (Isaacson et al., 2009). The cuticle of M82 is extremely thick, has highly cuticularized APs and shows extensive deposition of cuticle material below the epidermal cell layer. In contrast, the cuticle of cd2 fruit was dramatically thinner (approximately 20 times) and showed neither cuticularization of APs nor any sub-epidermal deposition.
The contrast generated by Sudan IV between the cuticle and other cell material was improved upon by the use of the lipophilic fluorescent dye auramine O, which has a strong affinity for regions containing acidic and unsaturated waxes, as well as cutin precursors (Considine and Knox, 1979; Gahan, 1984; Pesacreta and Hasenstein, 1999) and a moderate affinity for other lipidic cytoplasmic contents. However, the occurrence of non-cuticular fluorescence did not appreciably hinder the identification of cuticular material (Figure 2).
It was previously observed that auramine O binds different regions of the cuticle with different affinity (Considine and Knox, 1979; Jeffree, 2006). We observed a dimly fluorescent central region sandwiched between two more strongly fluorescing layers (Figure 2a, inset); the bright outer layer is consistent with the presence of a cuticle proper, as has been described in grape (Considine and Knox, 1979). The dimmer fluorescence of the middle layer, the ECL, and the bright fluorescence of the internal layer, the ICL, as seen in Figure 2(a), mirror the dichromatic staining observed with Nile blue A (Figure 1a). The M82 cuticle in this particular sample was extremely thick, and cuticular material was observed to penetrate down as far as the third cell layer in some regions. In stark contrast, a fluorescence micrograph of another cutin-deficient mutant, cd1 (Figure 2b; described by Isaacson et al., 2009), showed highly reduced deposition of cuticle on the surface of the fruit, but a moderate amount of sub-epidermal deposition.
3D cuticle imaging
The above techniques have the advantages of being rapid routine approaches for identifying gross differences in cuticle thickness and deposition patterns. Moreover, they can be used in conjunction with fresh, cryo- or paraffin sections, and require only a light microscope for imaging. However, as mentioned, these methods are of limited value in accurately describing the complex 3D cuticle architectures exhibited by some plants and tissue types. We therefore investigated the potential of confocal scanning laser microscopy (CSLM) to obtain 3D images of plant cuticles.
CSLM has previously been described as a technique to image certain plant cuticles using intrinsic autofluorescence arising from the presence of phenolic compounds – specifically the flavonoid chalconaringenin, which is abundant in the cuticular membranes of some fruits (Fernandez et al., 1999). However, this compound only accumulates in ripening fruits (Fernandez et al., 1999), and therefore this approach is not suitable for studying early fruit cuticle development or the cuticles of vegetative tissues. One alternative approach is to use fluorescent stains, and the fact that auramine O shows high affinity for cuticular material (Figure 2) suggested that its use in conjunction with CSLM might allow the generation of high-resolution images in any optical plane within the sample. This is particularly important for cuticle imaging, due to the intrinsic density and optical opacity of the cuticle. Moreover, the dye can be applied to any cuticle, allowing comparative studies of developmental stages, tissues, genetic lines and plant species.
Tomato fruit sections were stained with both auramine O and Calcofluor White M2R, a fluorescence brightener commonly used in the visualization of crystalline cellulose in plant cell walls (Figure 3b,c). Fluorescence was detected in sequential scan mode, with only one excitation laser was active at a time to prevent cross-excitation of the dyes, and the fluorescence of each channel was assigned a different color, with the color intensity being responsive to the intensity of emission. A bright-field image of the same area was collected as a reference for each sample (Figure 3a), and the images obtained using these channels could then be assembled in any combination to yield either a composite fluorescence image (Figure 3d) or a fluorescence overlay on the bright-field image (results not shown). This allowed the flexibility to highlight various features of the cuticle within a single sample: if auramine O fluorescence is viewed alone, the structure of the cuticle can be studied with minimal interference from the surrounding tissue, while the incorporation of Calcofluor white M2R fluorescence places the cuticle in the context of the surrounding cells and helps delineate the interface between the cellulosic cell wall and the start of the cuticular layer.
z-stack collection and assembly
z-stacks of images up to 30 μm deep were collected in sequential scan mode through sections of tomato outer pericarp. When a suitable region of tissue has been selected (usually five epidermal cells), the resolution was increased to maximum, the line and frame averaging were increased, and the zoom factor was adjusted to obtain the highest-quality z-stack of images. The step size was optimized from 0.1–0.15 μm to achieve a high image density, resulting in a smooth 3D assembly. Fluorescence intensity decreased noticeably at deeper optical planes of the sample due to the relative optical opacity of the cuticle, and therefore, to compensate, gain and offset values were adjusted manually as the z-stack images were gathered to maintain constant fluorescence intensity.
When the z-stacks were complete, they were assembled into a 3D composite image using the 3D rendition feature of the LAS-AF software (Figure 4). This generated projections of the overall 3D structure through any angle of rotation. As with the 2D images, the auramine O fluorescence can be viewed alone to generate a clear view of the cuticle, but if low-intensity fluorescence is retained, the cellular membranes and nuclei remain visible (Figure 4 and Movie S1). The intactness of these cellular components and the overall tissue morphology demonstrated the lack of tissue disruption that occurred using our fixation and sectioning protocol. The low-intensity fluorescence can also be removed using the LAS-AF software, leaving only the cuticle and the brightest points of fluorescence from the nuclei. Alternatively, the results from the separate collection channels can be combined to yield composite projections revealing cuticle, membranes, nuclei and cell wall together (Figure 5 and Movie S2), providing more specific spatial information related to cuticle deposition patterns.
The 3D assemblies shown in Figures 4 and 5 provide examples of information that can be obtained using this approach; however, when the 3D images of M82 cuticle were compared with those of the cutin-deficient mutant cd1 (Figure 6 and Movies S3 and S4), the differences suggested by the 2D images became far more apparent. A dramatic reduction was seen in the amount of cuticular material at the fruit surface of cd1 and in the APs, which penetrate to only a fraction of the depth of the epidermal cell layer. In addition, elongated detached sub-epidermal globules were observed below the anticlinal walls of neighboring epidermal cells. Thus, the 3D technique provides far greater spatial and structural information than achieved using the 2D approach, allowing a more thorough and comprehensive comparison of cuticle architecture between lines and species than was previously possible.
3D reconstruction of cuticle architecture
z-stacks of auramine O fluorescence were assembled into opaque 3D models using the Volocity modeling program. The low intensity fluorescence from the nuclei and membranes was first removed from all images using the LAS-AF confocal software as each emission source would be rendered as a solid object by the program, regardless of fluorescence intensity. The models generated from this process could be manipulated in three dimensions, and as they were built from high-resolution images, an unusually high level of topographic detail was possible. The transparency of the 3D fluorescent assemblies (Figures 4–6 and Movies S1–S4) allows a view of structural details within each sample, and avoids the problem of objects toward the front of the focal plane obscuring those behind. Additionally, by rendering fluorescent boundaries within the sample as opaque surfaces, the Volocity modeling program made visible details of the cuticle architecture that have not been previously described, as far as we are aware (Figure 7).
Three-dimensional analysis of the M82 fruit cuticle revealed a complex internal surface topography of the cuticle at the interface with the anticlinal and outer periclinal cell walls (Figure 7c–e). This may reflect patterns of polysaccharide deposition or non-uniform trafficking and deposition of cuticular material. In addition, a repeating pattern of fluorescence intensity was apparent at the inner cuticle surface, where threads of low-intensity fluorescence penetrated in a radial direction through the ICL, an area of high fluorescence intensity (Figure 7c,d, arrowheads). As mentioned previously, the differential fluorescence intensity (and dichromatic staining pattern observed with Nile blue A) between the ECL and the ICL is thought to result from a high concentration of acidic cutin precursors in the ICL. The low intensity of fluorescence seen in the ECL is believed to be associated with regions of more highly polymerized cutin, resulting in a more neutral microenvironment (Considine and Knox, 1979), and consequently a reduced affinity for the cuticle stains. The threads of low-intensity fluorescence may indicate either localized deposition of cuticle precursors or localized regions of cutin polymerization within the cell wall. Both these features illustrate the complex heterogeneity that exists with the cuticle, and may shed light on mechanisms of deposition or trafficking of cuticle precursors in future work.
Another feature that only became apparent through the use of 3D rendering is a form of pitting in the anticlinal walls of the epidermal cells, which we term anticlinal cuticular channels (ACCs) to emphasize the fact that they penetrate the anticlinal extensions of the cuticle. In M82, these are depressions approximately 1–3 μm in diameter in the APs that separate adjacent epidermal cells (Figure 7e,f). In thin sections, these tapering indentations are seen to contain extensions of the protoplasts (Matas et al., 2004). We do not know whether a cellulosic wall layer lines these cavities, nor have we resolved a pit-closing membrane. The diameters of the ACCs distinguish them from plasmodesmata, which are 3–5 nm in diameter (Roberts and Oparka, 2003). The significance of the ACCs is not known, but we suggest that they may be important in water transport, preventing turgor pressure gradients between epidermal cells that are encased in the cuticle and are thus more apoplastically isolated.
Finally, the resolution of both attached and detached sub-epidermal globules in the M82 cuticle (Figure 7b,f) raises interesting questions about the mechanisms of deposition of cuticular material into underlying cell layers. It may be that cuticle biosynthesis occurs solely in the epidermal cell layer, and deposition below this cell layer results from lipophilic material being ‘squeezed’ into sub-epidermal regions. However, 3D image analysis indicated that, in some cases, the sub-epidermal deposits are completely isolated and show no connection with the primary, continuous portion of the cuticle. It remains to be seen whether this material is derived from the epidermal cells and selectively trafficked into the sub-epidermal wall, or whether sub-epidermal cells actually synthesize small amounts of cuticular components as well.
Some of the features that we describe above are so subtle that they would probably be dismissed as optical artifacts when viewing 2D images, if noticed at all. For example, accurate 3D views of the ACCs and detached sub-epidermal globules were easily obtained by our methods, whereas standard imaging techniques barely suggest the 3D nature of these structures. Furthermore, our methods have promise for quantitative studies. While no quantitative data are presented in this paper, such information can be readily obtained using computer modeling. Accurate, quantitative assessment of fluorescence intensity in discrete regions of the cuticle, variations in cuticle thickness over an area, and cuticle volume can be calculated, yielding highly detailed spatial analysis of the cuticle architecture to supplement developmental and comparative studies.
Plant cultivation and fruit harvesting
All tomato (Solanum lycopersicum) cultivars or mutants (M82, cd1 and cd2) were grown as described by Isaacson et al. (2009). Tomato fruit were harvested at the breaker stage (corresponding to the onset of ripening) and immediately dissected and fixed (see below).
Tissue fixation and embedding
A ring of pericarp, approximately 3 mm wide, was separated from the equatorial region of each tomato fruit and divided into 3 × 3 mm cubes using a razor blade. The tissue cubes were immediately transferred to FAA fixative [5% formalin (37% formaldehyde, aqueous), 5% glacial acetic acid, 45% ethanol, 45% distilled water] at a 1:10 volume ratio of tissue to fixative, and vacuum infiltrated for 15 min. The FAA was removed, replaced with fresh fixative, and the samples were left overnight at 4°C. They were then cryoprotected using a gradient of 10% sucrose followed by 20% sucrose in 100 mm PBS (11.5 g L−1 Na2HPO4, 2 g L−1 NaH2PO4·H2O, 90 g L−1 NaCl, pH 7.2). Samples were vacuum-infiltrated in the 10% sucrose solution for 15 min, which was replaced by fresh 10% sucrose solution, and the incubation continued at 4°C until the tissue cubes sank. This was repeated with the 20% sucrose solution. The tissue cubes were then washed gently, but thoroughly, in OCT medium (Sakura, http://www.sakura.com), and transferred to cryo-molds filled with OCT medium. Samples were added 3–5 per mold, and oriented such their cuticles were perpendicular to the bottom of the mold and parallel to one another. Molds were frozen in liquid nitrogen and stored at −80°C prior to sectioning.
Cryosectioning and slide preparation
Cryosections (8 and 30 μm) of each tomato line were produced using a Microm HM550 cryostat (ThermoFisher Scientific, http://www.thermofisher.com). The CryoJane tape-transfer system (Instrumedics, http://www.instrumedics.com) was used to transfer the sections to 0.5 × adhesive-coated slides, where they were adhered via UV-crosslinking. Slides could then be stored at −20°C until use. Each slide was post-fixed in room-temperature CryoJane aqueous slide fixative [40% glutaraldehyde solution (25% aqueous), 60% CryoJane salt buffer] for 45 sec, and rinsed gently with distilled water immediately prior to staining.
Sudan IV staining. Sudan IV (MP Biomedicals, http://mpbio.com) stock solution (0.1% w/v in isopropyl alcohol) was diluted 3:2 with distilled water, mixed well, allowed to sit at room temperature for 30 min, and filtered through a syringe filter to remove precipitates. The stain was added to 8 μm sections for 10 min, which were then rinsed first with 50% isopropyl alcohol, then with distilled water (Clark et al., 1973; Schneider, 1973). Slides were mounted in distilled water with a cover slip, sealed with nail polish, and viewed immediately.
Nile blue A staining. Nile blue A (Sigma, http://www.sigmaaldrich.com; 1% w/v in distilled water) was added to 8 μm sections for 30 sec. The stain was poured off, 1% acetic acid was added to the slides for another 30 sec, and the slides were rinsed with distilled water (Gahan, 1984). The slides were mounted in DABCO mounting medium (Alfa Aesar, http://www.alfa.com; 2.5% w/v 1,4-diazabicyclo[2.2.2]octane, 5% v/v 1 m Tris, pH 8.0, 90% v/v glycerol) with a cover slip, and sealed with nail polish.
Auramine O staining. Sections 8 μm thick were stained with Auramine O (Sigma; 0.01% w/v in 0.05 m Tris/HCl, pH 7.2) for 15 min (Gahan, 1984), and rinsed with distilled water. The slides were mounted as described for Nile blue A staining.
Calcofluor white M2R/auramine O staining. Sections 30 μm thick used for confocal imaging were stained with Calcofluor white M2R (Wyeth, http://www.wyeth.com, 0.1% w/v in distilled water) for 2 min (Gahan, 1984), rinsed with distilled water, and stained with auramine O for 15 min. Slides were rinsed with distilled water, mounted with a cover slip in DABCO mounting medium, and sealed with nail polish.
Microscopy and 3D modeling
Bright-field microscopy. Tissue sections 8 μm thick stained with Sudan IV, Nile blue A or auramine O were imaged using an Olympus BX60 epifluorescence microscope (Olympus, http://www.olympusmicro.com) equipped with Olympus UPlanApo 40 ×/1.0 and 60 ×/1.4 oil immersion objectives and a Sony 3CCD color video camera (model DXC9000, Sony, http://www.sony.com), and operated using Image-Pro Plus 6.3 software (Media Cybernetics, http://www.mediacy.com). Sections stained with auramine O (Figure 2) were excited using an Olympus BX-FLA mercury lamp, and emission was collected using a GFP broadpass filter cube (model U-M41017, Chroma Technology, http://www.chroma.com). The cuticle image shown in Figure 1(a) was obtained using a Zeiss AxioImager A1 microscope (Zeiss, http://www.zeiss.com) equipped with a Zeiss EC-Plan NeoFluar 100 ×/1.3 oil immersion objective, a Zeiss AxioCam MRc color video camera, and Zeiss AxioVs40 126.96.36.199 software.
Confocal microscopy. Confocal laser imaging was performed using a Leica TCS-SP5 confocal scanning laser microscope (Leica Microsystems, http://www.leica-microsystems.com) with a 63 ×/1.2 water immersion objective. Image processing was accomplished using LAS-AF 1.6.3 software (Leica Microsystems, http://www.leica-microsystems.com). Auramine O was excited using a 458 nm argon laser, and emission was collected between 491 and 563 nm. Calcofluor White M2R was excited with a 405 nm near-UV diode, and emission was collected between 415 and 448 nm. 2D images and z-stacks were obtained using a scan rate of 400 Hz in sequential scan mode to avoid cross-talk between fluorophores. The zoom factor was adjusted in the range 1.0–1.7, depending on the image subject, and step size was in the range 1.0–1.5 μm, and was determined via program auto-optimization based on the input parameters of the individual z-scan. Both line averaging and frame averaging were used on a case-by-case basis to improve image quality. 3D composite fluorescent images were generated using the 3D rendition feature of the LAS-AF software in maximum intensity mode.
3D modeling. Low-intensity auramine O fluorescence was subtracted from selected z-stacks of M82 tomato images using the LAS-AF software to remove fluorescence associated with cellular membranes and organelles. The resulting z-stacks, showing auramine O fluorescence from the cuticle only, were assembled into opaque, 3D models using Volocity 4.1 software (Improvision, http://www.improvision.com). Brightness and contrast were adjusted for each model to improve the resolution of cuticle architectural features.
The authors would like to thank Dr Karl Niklas for discussion and careful reading of this manuscript. The project was supported by the National Research Initiative of the United States Department of Agriculture Cooperative State Research, Education and Extension Service (grant number 2006-35304-17323) by Cornell University Agricultural Experiment Station Hatch Project NYC-184485, by National Science Foundation Plant Genome Research Program grant number DBI-0606595, and by United States–Israel Bi-national Science Foundation award number 2005168.