Cortex wedges from a 1 cm thick equatorial slice of apple (cv. 'Royal Gala') were cut with a paring knife. To avoid the small cells found near the skin, a block of approximately 1 cm³ of cells from the central cortex were selected (figure 1A). To reduce the number of damaged cells from the initial cut approximately 0.1 cm of tissue was trimmed from each surface of each wedge using a fine edged scalpel blade. The resulting cortex tissue was then cubed into approximately 2 mm³ tissue blocks with the fine edge scalpel. Single cell isolation was achieved in small 50 ml glass beakers using 40 ml of 0.05 M Na₂CO₃ in 0.3 M mannitol. Mannitol was used to osmotically buffer the apple cells in their physiologically normal osmotic range 11. To delocalise gases present in the air gaps and to aid separation, the cubes were gently boiled on a magnetic hot-plate stirrer (180 rpm) for 20–30 minutes stirring with a 2 cm magnetic stirrer, the heat was then reduced until boiling stopped and the tissue was stirred until free cells could be observed in the solution (a cloudy appearance). Residual cellular clumps of vascular bundles and unseparated cells were removed by passing through a 1 mm mesh sieve into a 50 ml falcon tube. These clumps typically represented 10 to 20% of the volume of chopped apple pieces. The separated cells were allowed to settle for an hour, after which excess supernatant was removed leaving an equal amount of liquid as settled cells. At this point the cell homogenate could be stored at 4°C before imaging. Utilising a chemical rather than an enzymic treatment allowed the use of boiling to speed up cell separation, greatly reducing the time used in other tissue maceration methods 172324.

Prior to microscopic imaging, harvested cell preparations were re-suspended by gentle flicking of the tube, 30 ul aliquots of suspended tissue were spotted onto clean glass microscopy slides and viewed at 4× under bright field with contrast maximised. Images were collected in a grid like manner to reduce biased selection of cells, using a CoolSnap digital camera and captured using RSImage software, version 1.9.2 (Roper Scientific Ltd, Tucson, Arizona). Images were saved as 24 bit Tiff files.

The size of cells was analysed using the public domain ImageJ software package http://rsb.info.nih.gov/ij/. For each image an unadjusted image was kept open to check that measurement of cells were consistent with the raw image. The threshold was set so the outline of each cell was clearly differentiated from the background, and then each image was converted to binary (black and white). Each cell was then filled using "Fill holes". Where cells had not been completely separated then the "Watershed" separation was used to separate out the single cells from the clumps. Occasionally intact cells were not filled completely due to small gaps in the outline. If this occurred then the gaps were manually filled and before proceeding with the "Fill holes" again. Areas were calculated using "analyse particles" with a particle size cut-off threshold of 1100 pixels. A skeletonised image was obtained and visually checked that the cells analysed were whole and single (Figure 1B–D).

It has previously been shown that in some cultivars the blush side (sun exposed) of apples are often firmer than the non blush side (shaded) 26. To assess the differences between sun and shade sides of fruit and between fruit cultivars, the cell size of 5 common Malus × Domestica Borkh. (apple) cultivars with different textures were measured. Three similarly sized apples from standard industry cold stored conditions of 'Braeburn', 'Cripps Pink/Pink Lady'™, 'Scifresh/Jazz'™, 'SciRos/Pacific Rose'™ and 'Royal Gala' cultivars were measured for firmness on the sun and shade side using a penetrometer. For all apple cultivars there was no statistically significant difference in firmness between the sun and shade side (Table 1). There was however, a statistically significant difference in firmness between cultivars, with 'SciFresh' and 'Pink Lady' being firmer than 'Royal Gala' and 'SciRos' (Table 1). Individual apple cells were isolated from each variety (Figure 2). For each apple cells were isolated from cortex tissue adjacent to the penetrometer wound site and approximately 170 individual cells were measured (Table 1, Figure 2). There was no correlation between cell size and firmness either within a cultivar or between cultivars. However it was noted that the firmer apples ('SciFresh' and 'Cripps Pink') had more angular cells compared to the softer cultivars 'SciRos' and 'Royal Gala' (Figure 2).

To further investigate differences in cell size between cultivars, size measurements from each cultivar were combined and histograms of the distributions were plotted Figure 3. Using "rnorm" in the statistical software package "R", a normal distribution of cell sizes was calculated with the mean and standard deviation of sizes from each cultivar using a random generator of 10,000 points. This modelled distribution was plotted on the actual cell distribution and it was found that for all species a normal distribution of cell size occurred (Figure 3A–E). This implies that apples have only one population of cell types in the middle of the cortex tissue. The modelled normal distributions for each cultivar were compared and it was found that each apple cultivar showed a distinct distribution of cell sizes (Figure 3F). The range of cell sizes observed for these cultivars were consistent with previous studies (e.g19) which have shown cells with a diameter in the range of 250 uM, assuming a circular area would produce approx 50,000 uM² (Table 1).

When separating cells there is a possibility that cell size and shape can be altered either mechanically or osmotically. To investigate the difference in shapes observed in cells extracted from 'SciFresh' (more angular cells) and 'SciRos' (more rounded cells), these cultivars were chosen for confocal microscopic analysis. Blocks of tissue were taken from the equatorial regions of firm apples of 'SciFresh' and 'SciRos' within 3 weeks of harvest. Each piece comprised a 5–6 mm thick cross-section extending from 5 mm to 15 mm from the fruit surface. Tissue was fixed in 20 g L^-1 fresh formaldehyde, 25 g L^-1 glutaraldehyde in 0.1 M sodium phosphate buffer pH7.2. A light vacuum was applied to remove as much air as possible from the tissue which was then stored at 4°C in the fixative. Samples for confocal microscopy were washed in 0.1 M sodium phosphate buffer for 2–3 h (3 changes of buffer) and were then sectioned at 800 μm using a Vibratome 1000 (Technical Products International, St Louis Mo) and stained with 0.001 g L^-1 acroflavin in 0.1 M buffer for 15 min, washed 3 times in phosphate buffer and mounted on slides with 800–900 μm deep chambers to prevent the coverslip compressing the tissue. Sections were viewed using an Olympus FV1000 confocal microscope (Olympus Corporation, Tokyo, Japan). For each area imaged a stack of approximately 50 individual images was taken at 6 μM intervals. These stacks were used to produce single z projection images. From these confocal images it is clear that the rounded and angularity morphologies seen in the 'SciRos' and 'SciFresh' isolated cells are consistent with the un-separated tissue (Figure 4B, D). To identify any size changes that might have occurred during cell separation, the average longest dimension of the 'SciRos' cells from the confocal tissue section, and 'SciRos' isolated cells were measured. Cells in the confocal sections averaged 268.08 μM (± 50.3), compared to isolated cells that averaged 276.47 μM (± 46.55), demonstrating there was no significant change in size during extraction.

To test whether this method could be used in other plant tissue, we assessed fleshy tissue from unripe fruit purchased from the local supermarket, including green mature/breaker stage tomato (Solanum lycopersicum) (Outer pericarp including skin) (Figure 5A), ripe rock melon (Cucumis melo CV cantalupensis) (rind extending into flesh tissue) (figure 5B), and mature unripe kiwifruit (Actinidea deliciosa) (outer pericarp, inner pericarp and core tissue extracted separately) (figure 5D–F). Additionally, cortex tissue from immature apple fruit was also tested (approximately 80 DAFB) (figure 5C). It was found that all these showed a good cell separation of flesh tissue: the tomato skin and the rind tissue of the water melon did not separate into individual cells. However, the core tissue from kiwifruit broke down rapidly into individual cells. The immature apples also showed rapid cell disassociation, suggesting cellular linkages are less well formed in rapidly growing fruit. Unlike the apple cells which are fairly homogeneous in morphology, kiwifruit have been previously documented as having a range of cell types in the fruit 27, with the outer pericarp having large cells and small cells, the inner pericarp having long thin cells and idioblasts (cells containing crystalline oxylate) and the core having regular smaller cells 27. All these classes of cells were observed (Figure 5) suggesting that this method did not exclude certain cell types.