Eighteen years have passed since the need for a comprehensive three-dimensional anatomical description of the cellular structure of an Arabidopsis thaliana (L.) Heynh. leaf was first formulated . The detailed characterisation of the cellular structure of the wild-type leaf would provide the factual basis for the identification of even the most subtle phenotypes of mutants and aid in the unravelling of growth mechanisms. A large number of Arabidopsis genotypes dramatically affected in leaf form or dimensions and in cell numbers or dimensions have been identified ; however, potentially many other genotypes have informative leaf phenotypes which are currently not detected. Besides genetic factors, environmental conditions affect leaf expansion rates and duration  and bring about morphological changes in leaf tissues which relate directly to leaf functioning in photosynthesis and transpiration [4–9]. Models of leaf size control integrate data on quantitative growth variables at the plant, organ and cell level, but for the latter, only epidermal cells or the sub-epidermal layer of palisade mesophyll cells are taken into account, not the whole leaf thickness [10–12]. A three-dimensional structural model of the developing leaf does not exist as yet, because the essential quantitative parameters of internal leaf tissues are missing.
The routine methods for leaf phenotype assessment at a cellular level use brightfield or differential interference contrast microscopy. Cell length measurements are performed on transverse sections obtained by classical histological sectioning with the inherent spatial inaccuracy [1, 13], while cell area or width is determined using epidermal peels or paradermal views of cleared leaves with measurements limited to the epidermis and one sub-epidermal layer of mesophyll cells [10–17]. Cell separation is another option for area measurement of mesophyll cells, but it implies the loss of any information on cellular organisation [1, 18, 19]. Important progress in the understanding of the processes of leaf cell proliferation and expansion was made using a cyc1At::GUS reporter, but imaging was limited to histological sectioning and paradermal views . Clearly, there exists a need for the visualisation and quantitative analysis of the cellular organisation of the intact leaf and all of its tissues in their three-dimensional context.
Currently, the most adequate and straightforward three-dimensional imaging methods for obtaining single-cell level resolution in plant biology are confocal and multiphoton laser scanning microscopy. Three-dimensional imaging of leaves has already been achieved by magnetic resonance imaging (MRI) [20–22], optical coherence microscopy (OCM) [23, 24], high-resolution X-ray computed tomography (HRCT)  and optical projection tomography (OPT) . However, none of these techniques provide the single-cell resolution required for the reconstruction and analysis of the cellular organisation of leaf tissues. They rather visualise the external morphology of the leaf at the organ level and thus allow the analysis of overall leaf volume and surface area. In vivo imaging using OCM does provide a tool for monitoring developmental changes at the organ level during leaf growth, while in OPT gene expression can be imaged, revealing for example the complete leaf venation pattern. MRI has been shown powerful in recording plant physiological processes such as water movement and the transport of assimilates and ions through vascular tissues. Particularly promising here is the recent development of MRI-positron emission tomography (PET) . In contrast to these techniques, which require highly specialised equipment and handling, laser scanning microscopy is a very accessible imaging method, both in terms of equipment and usage. Confocal microscopes have been adopted at most research institutes and multiphoton systems become increasingly available via institute or regional wide imaging platforms.
It is generally accepted that multiphoton microscopy provides deeper depth penetration than confocal microscopy [28, 29]. Plant leaves and various other plant organs are, however, notoriously difficult to image in depth because of weak penetration of light, which is due to an opaque cuticle and the presence of numerous light scattering molecules, mostly secondary metabolites, in the cuticle, cell wall and vacuoles . Even with a multiphoton microscope, one cannot penetrate deeper than the epidermis and one layer of mesophyll cells  (authors' experience). In contrast, strong clearing of plant tissues allows optical sectioning down to 200 μm in depth using both techniques and is, in principle, only limited by the working distance of the objective [29, 30]. This was recently shown by confocal imaging of Arabidopsis organs in gene expression analyses . Image quality, certainly in depth, is nonetheless technique-dependent and multiphoton microscopy outperforms confocal imaging when it comes to signal-to-noise ratio, contrast and thus effective resolution . Provided that optical sections are of high quality, i.e. high resolution and contrast, access is granted to volume-based quantitative data on leaf tissues and cell dimensions. Moreover, cells are shown within their three-dimensional context, which means that the structural interaction between cell types can be revealed.
In this paper, a procedure for the three-dimensional imaging and quantitative analysis of the structural properties of plant leaves is presented. It was developed and optimised for leaves of a wild-type accession of Arabidopsis (Col-4) grown in soil, and for a range of developmental stages, from initiation to senescence, representing differences in cell densities, cell dimensions and cuticle and cell wall properties. The resulting images covered the complete leaf thickness and were of high resolution and contrast throughout (when acquired by multiphoton laser scanning microscopy), allowing a detailed visualisation of the large diversity of leaf tissues. Here, the performance of the procedure was demonstrated for Arabidopsis leaf growth, focusing on four major structural and functional leaf tissues, the adaxial and abaxial epidermis and the palisade and spongy mesophyll. The procedure enabled the study of leaf expansion in surface area and thickness through specific tissue and cell variables such as the volumetric proportions of the epidermal and mesophyll tissues, cell density in these tissues, and cell dimensions, including volume, length and width.