Plant material.

Fruit of the sweet cherry cultivars Adriana, Dönissens Gelbe, Flamengo Srim, Frühe Rote Mecklenburger, Hedelfinger, Rainier, Sam, Staccato, and Sweetheart, all grafted on Gisela 5 rootstocks (Prunus cerasus L. × Prunus canescens Bois), were sampled at commercial maturity as indexed by color, size, and taste. In addition, ‘Regina’ fruit was sampled during development from 27 d to maturity at 97 DAFB. Trees were grown in a greenhouse and an experimental orchard of the Horticultural Research Station of the Leibniz University in Ruthe (lat. 52°14′N, long. 9°49′E). Fruit were selected for freedom from defects and for uniformity, and transferred to the laboratory. Unless otherwise specified fruit was processed within 48 h of sampling.

General experimental procedure.

Sections of the fruit skin comprising epidermal and hypodermal cells were prepared using a razor blade, blotted with tissue paper or rinsed by dipping briefly (5 s) in deionized water. A section was immediately transferred to a microscope slide, covered with a cover slip, and incubated for 30 min in an aqueous solution containing one of a broad range of concentrations of an osmolyte. Unless otherwise specified, sucrose was used as the standard osmolyte. After 30 min, the slide was transferred to the stage of a microscope and viewed in incident transmitted light at ×40 magnification using a fluorescence microscope (BX-60; Olympus Europa, Hamburg, Germany). Occasionally, differential interference contrast was used to enhance contrast. Unless otherwise specified, cells were observed only in sections that appeared healthy as indexed by the absence of coagulated protoplasts. Calibrated digital images were taken (DP 71, Olympus Europa) and the percentage of plasmolyzed cells was determined. In some experiments, swelling of the cell walls was quantified by determining cell wall thickness using image analysis (Cell-P, Olympus Europa). The osmolarity of incubation media and of juice extracted from the same fruit as that used in the plasmolysis and cell wall swelling assays was quantified by vapor pressure osmometry (VAPRO^® 5520 and 5560; Wescor, Logan, UT).

Experiments.

The effects of osmolyte type and concentration on the percentage plasmolysis were investigated in ‘Adriana’ sweet cherry using sucrose, mannitol, or a mix of osmolytes that mimicked the composition of extracted sweet cherry juice. We refer to this solution as artificial cherry juice. Where solubility permitted, osmolarities ranged from 0 to 5 MPa (equivalent to 0 to 2000 mmol·kg⁻¹). For the artificial juice, a 1000 mmol·kg⁻¹ solution was composed of glucose (431.8 mm), fructose (393.5 mm), sorbitol (76.9 mm), malic acid (13.8 mm), and potassium malate (56.3 mm) (Herrmann, 2001). These five osmolytes accounted for 98% of the osmolarity recorded for sweet cherry juice. The pH of the artificial juice (pH 3.4) was close to that of natural sweet cherry juice (pH 3.6). The percentage of plasmolyzed cells was quantified as described above on sections excised from five fruit per osmolyte type.

Differences in the osmolarity of the skin and flesh were investigated in ‘Hedelfinger’, ‘Sam’, and ‘Sweetheart’ fruit. The osmolarity of the epidermis was assessed by quantifying the percentage of plasmolysis of fruit skin sections after 30 min of incubation in sucrose solutions and that of the flesh by water vapor pressure osmometry of expressed juice. Sections were prepared from 10 fruits per cultivar.

Plasmolysis was also determined following incubation of skin sections in juice extracted from flesh of the same fruit in the immediate vicinity of the position where the skin section was excised. Plasmolysis was assessed after 30 min. Segments incubated in silicone oil (AK 10; Wacker Chemie, Munich, Germany) served as controls and were inspected immediately after excision. To broaden the database, this experiment was also conducted with ‘Dönissens Gelbe’, ‘Flamengo Srim’, ‘Frühe Rote Mecklenburger’, ‘Hedelfinger’, ‘Rainier’, ‘Sam’, and ‘Staccato’. Two sections per fruit from a total of 10 fruit per cultivar were assessed.

The effect of development on the osmolarity of epidermal cells as indexed by plasmolysis was studied in developing ‘Regina’ (27 to 97 DAFB). Sections of the fruit skin were prepared, incubated in sucrose solutions (30 min), photographed, and the percentage of plasmolyzed cells and thickness of cell walls were quantified as described above. This procedure allowed direct comparisons between cell wall thickness and the extent of plasmolysis on an individual skin segment basis. The number of replicates was 10. Data for the osmolarity of the flesh of the fruit (obtained from the same batch of fruit) were taken from Schumann et al. (2014).

The time course of cell wall swelling was established using skin sections from ‘Sam’ sweet cherry (n = 5). Because cells burst when incubated in water for extended periods of time, incubation was ended after 3 h. By this time, the majority of protoplasts were still intact and alive so that swelling could be established on an individual segment basis.

Potential relationships between the swelling of cell walls and the initial cell wall thickness were identified in ‘Sam’ sweet cherry by quantifying cell wall thickness within 5 min of excision of skin segments and then repeating the measurement on the same segment at the same position 3 h later. Fruit used in this experiment was held at 2 °C for 5 to 6 d.

Whether cell wall swelling was affected by the health (or otherwise) of neighboring epidermal cells was studied in ‘Sweetheart’. In these experiments, a polyethylene glycol solution was used that was osmotically buffered at an osmolarity of 200 mmol·kg⁻¹. A 6-h incubation period was selected. Under these conditions, some protoplasts collapsed as indexed by their coagulated cytoplasm and epidermal sections contained both healthy and collapsed cells. Cell wall thickness was measured between two healthy epidermal cells, between a living cell and a dead cell and between two dead cells. Three measurements were made per pair of cells and section on a total of 20 fruit.

Data analyses.

Results in tables and figures are presented as means ± se. Where error bars are not visible, they are smaller than the symbols. Analyses of variance (PROC GLM) and regression analyses (PROC REG, PROC NLIN) were carried out using SAS (version 9.1.3; SAS Institute, Cary, NC). Percentage plasmolysis data were arcsine transformed before analysis of variance. The osmolarity at 50% plasmolysis was estimated as the point of inflection of a logistic regression model. At this osmolarity, the turgor pressure is zero (the point of insipient plasmolysis) and the osmolarity of the symplast is assumed to be equal to that of the incubation solution.

Evidence for a less negative osmotic potential in skin than flesh.

The less negative osmotic potential of skin than flesh is consistent with earlier data by Moing et al. (2004) who reported a difference in Ψ_Π of 0.5 MPa (range in ΔΨ_Π = 0.4 to 0.7 MPa). The ΔΨ_Π values in our study were 2- to 3-fold larger than those of Moing et al. (2004) probably because of our much higher spatial resolution achievable by a microscope assessment of epidermal plasmolysis, whereas Moing et al. (2004) quantified juice osmolarity from 1.0 ± 0.2-mm thick tissue blocks, where only the outer ≈100 µm (≈10%) was epidermal (Glenn and Poovaiah, 1989) and most of the rest (≈90%) was parenchymatous. Hence, in their samples, the juice from the true skin (having a less negative Ψ_Π^S) was mixed with that from some adhering parenchyma cells beneath (having a more negative Ψ_Π^F), probably resulting in an underestimate of the magnitude of ΔΨ_Π, the difference in osmotic potential between skin and flesh.

The conclusion that osmotic potential differs between skin and flesh is based on different techniques (i.e., plasmolysis for the skin vs. osmometry of the flesh). Unfortunately, it was technically impossible to observe plasmolysis in a parenchyma cell of the flesh (cells far too large) or, conversely, to use a vapor pressure osmometer to record the osmotic potential of an epidermal cell (cells far too small). The large, thin-walled parenchyma cells prevent critical assessment of the onset of plasmolysis in the narrow focal plane of the microscope. Moreover, the soft flesh tissues of a mature fruit are difficult to handle in thin sections that require the absence of mechanical stress and minimal juice leakage. Nevertheless, we believe a methodological artifact is an unlikely explanation for the ΔΨ_Π value we infer between skin and flesh. Furthermore, the observation that epidermal cells consistently plasmolyze when incubated in juice from the very same fruit would seem to offer conclusive and direct evidence for the existence of a major difference in Ψ_Π between skin and flesh. Lastly, the significant value for ΔΨ_Π we report here is not a singular event at one point in the course of development, but is present throughout stage III during which time it increases continuously (Fig. 4C).

Methodological artifacts.

It is important to understand potential methodological limitations of this study as they affect the estimates both of Ψ_Π^S and of Ψ_Π^F and, hence, of ΔΨ_Π.

• Ψ_Π^S: The excision of a skin strip for microscopy allows its relaxation (release of elastic strain) and this may have altered the water potential. Relaxation is rapid [halftime of ≈2.7 min (Grimm et al., 2012)] and this is accompanied by a decrease in planar area of the sample and consequently in the planar area of each epidermal cell. Cell wall strain release allows each epidermal cell to assume a more spherical shape and hence a decrease in turgor and thus a more negative water potential than before. If the skin sample is now bathed in an excess volume of an external solution, epidermal water potential will soon come into balance with this and the result will be a generally too-negative estimate of epidermal water potential. This suggests, the true difference in osmotic potential between flesh and skin is actually greater than that quantified above—in other words our estimate is conservative and the true difference will more likely be greater than (not less than) that we suggest.

• Ψ_Π^F: Juice extraction by crushing the flesh tissue will destroy any in vivo compartmentation and hence will create a potential for autolysis. What is the likelihood that enzyme reactions will significantly alter (decrease) juice osmotic potential? We have already seen that the predominant juice osmolytes are the hexose monomers glucose and fructose (Herrmann, 2001) and these are less likely to be affected with any rapidity. Sucrose is a possible candidate for rapid enzymatic cleavage but this moiety occurs at a too-low concentration for its possible autolytic cleavage to markedly decrease juice osmotic potential (Herrmann, 2001).

We suggest that the net effect of any methodological artifacts will more likely decrease the measured osmotic potential difference (ΔΨ_Π) we report, than increase it. Hence, our inferences are satisfactorily conservative.

Are skin and flesh at water potential equilibrium?

A central question is whether skin and flesh are at water potential equilibrium or could their water potentials differ significantly?

A water potential equilibrium implies the absence of a net movement of water and/or of osmolytes between flesh and skin. Under equilibrium conditions, Ψ^F = Ψ^S and, hence,

View Expanded

where the much more negative value taken by Ψ_Π^F than by Ψ_Π^S requires Ψ_P^F to be correspondingly much higher (i.e., much more positive) than Ψ_P^S. Rearranging the above equation

View Expanded

shows that under equilibrium conditions, ΔΨ_Π and ΔΨ_P must be numerically equal. This would seem to imply that Ψ_P^F must greatly exceed Ψ_P^S and that it would do so by an average of 1.1 MPa (Table 1). However, recorded values of Ψ_P^F of mature fruit are always very much less than 1.1 MPa, indeed they usually take values of less than 10% of this value [≈0.1 MPa (Knoche et al., 2014; Schumann et al., 2014)]. This would seem to imply a highly negative value for Ψ_P^S, which inference is difficult to accept as being real. Thus, a high value for Ψ_P^F can be excluded as a mechanism for balancing the very negative value of Ψ_Π^F.

In grape berries, high values for Ψ_P^F as a consequence of similarly very negative values for Ψ_Π^F have been shown to be prevented by the accumulation of osmolytes in the apoplast (Wada et al., 2008, 2009). Here, the high levels of apoplastic osmolytes roughly balance the high symplastic ones, thereby preventing the development of a high Ψ_P^F. If this situation was also to occur in sweet cherry, then Eq. [1] must be modified by relaxing its implicit assumption that the apoplasts of the flesh and skin are essentially composed of pure water. This means we must include the opportunity for an apoplastic osmolyte in the notation for the flesh tissue (Ψ_Π^Fa), not just a symplastic one (Ψ_Π^Fs) and likewise in the skin (Ψ_Π^Sa and Ψ_Π^Ss).

View Expanded

This shows that the essential absence of Ψ_P^Fs (obtained by measurement) requires Ψ_Π^Fs to be about equal to Ψ_Π^Fa (Schumann et al., 2014). The value of Ψ_Π^Ss needs not be similar to Ψ_Π^Sa or the epidermal cells would not be turgid.

Next, because Ψ_Π^Fs is much more negative than Ψ_Π^Ss (i.e., a major result presented in this study) Ψ_Π^Fa must be much more negative than Ψ_Π^Sa. Such a difference in the osmotic concentrations of the two adjoining apoplasts (i.e., that of the flesh and that of the skin) requires postulation of a sharp gradient in osmolyte concentration along the planar interface between the apoplasts of the flesh and skin tissues. This gradient will likely generate a powerful concentrative driving force for the diffusion of osmolytes from flesh apoplast to skin apoplast. Similarly, it will generate a powerful water potential driving force for the diffusion of water in the reverse direction, from skin apoplast to flesh apoplast. Although plausible, or even inevitable, direct evidences for these counter-directional diffusions of osmolyte and water are lacking.

If a large difference in osmolyte concentration is to be sustained between the apoplasts of the flesh and skin (ΔΨ_Π), two possibilities remain. The first is that there is a diffusion barrier in the apoplast (equivalent to an endodermis) at the flesh–skin tissue interface that significantly impedes the counter-directional diffusions of osmolytes and water. The other possibility is that a standing difference (Ψ^F < Ψ^S) is maintained actively by a continuous supply of carbohydrates to the flesh from the tree via the pedicel phloem, and/or on the other by a strong metabolic sink for carbohydrates in the skin. Although there does not seem to be a vascular system within the thin (100 µm) skin layer, there is a prolific but fine network of ventral and dorsal vascular bundles in the outer flesh (mesocarp). The distribution of the vascular system is consistent with this hypothesis, but it would also require that the skin be very active metabolically so as to “use up” the major proportion of the centrifugally diffusing osmolytes (predominantly sugars).

Returning to the idea of an apoplastic barrier, the swollen cell walls of the skin could represent a significant diffusion resistance that slows progress toward osmotic and thus water potential equilibrium. Unlike the internal cuticle of tomato [Solanum lycopersicum L. (Matas et al., 2011)], there is no evidence for an internal transport barrier in sweet cherry fruit that could hydraulically “isolate” skin from flesh.

Based on the above arguments, the existence of a gradient in Ψ between skin and flesh is a more likely explanation for the ΔΨ_Π, but definitive evidence is lacking. It is important to note that in both cases, the consequences would be similar, i.e., a driving force for diffusion of osmolytes from flesh to skin and for flow of water in the reverse direction from skin to flesh.

The broad range of osmotic potentials of bathing solutions (up to 1 MPa and more) causing plasmolysis of fractions of epidermal cells of the very same fruit deserves some further comments (Fig. 2). Based on the arguments presented above, this range reflects the range of Ψ_Π^S of the epidermal cells indicating considerable heterogeneity within the population of cells in a given fruit. Assuming water potential equilibrium within the epidermis, we would expect a correspondingly broad range in their Ψ_P^S. Unfortunately, due to the small size of epidermal cells, direct evidence for this inference using pressure probe techniques is difficult to obtain.

Swelling of cell walls on plasmolysis.

Swelling of cell walls is commonly observed in ripening fruit that has a “soft and melting” texture, but not in fruit having a “crisp and fracturable” texture (Redgwell et al., 1997). Our observations in sweet cherry are consistent with this conclusion. According to Redgwell et al. (1997) swelling is thought to result from pectin solubilization and a subsequent increase in volume when water moves into voids of the cellulose–hemicellulose network formed on pectin solubilization and extraction. We observed increased swelling beyond 55 DAFB as compared with before 55 DAFB, which is in agreement with increased activity of pectinases during stage III of fruit development [Fig. 3 (Brummell, 2006; Kondo and Danjo, 2001)]. Unfortunately, little is known about the physical properties of swollen cell walls and we are not aware of published information quantifying the “osmotic” potentials and pressures exerted by the swelling and hydrating cell walls. Our data suggest 1) that the swelling pressure is probably low, whereas 2) the volume increase due to swelling is quite significant [+ 0.64 µm in width (Fig. 5A)]. First, cell wall swelling occurred only when living cells plasmolyzed and hence, turgor was absent. Second, cell wall swelling was greatest in cell walls separating two dead cells, intermediate for cell walls separating a dead from a living cell, and least when cell walls separated two living and turgescent cells. Apparently, in turgescent cells, the turgor balanced the swelling of cell walls. Because the turgor of cells of the outer mesocarp is lower than 0.1 MPa at maturity (Knoche et al., 2014; Schumann et al., 2014), the pressure developed by swollen cell walls is expected to be of the same order of magnitude or slightly lower. This is of a similar magnitude to the swelling pressure reported for the starch biopolymer in potato [Solanum tuberosum L. (Jarvis et al., 1992)]. The volume increase of the apoplast due to cell wall swelling indicates significant water uptake, the forces driving this uptake are not yet known.