- © 2005 American Society of Plant Biologists
Abstract
Postanthesis growth of tomato (Solanum lycopersicon) as of many types of fruit relies on cell division and cell expansion, so that some of the largest cells to be found in plants occur in fleshy fruit. Endoreduplication is known to occur in such materials, which suggests its involvement in cell expansion, although no data have demonstrated this hypothesis as yet. We have analyzed pattern formation, cell size, and ploidy in tomato fruit pericarp. A first set of data was collected in one cherry tomato line throughout fruit development. A second set of data was obtained from 20 tomato lines displaying a large weight range in fruit, which were compared as ovaries at anthesis and as fully grown fruit at breaker stage. A remarkable conservation of pericarp pattern, including cell layer number and cell size, is observed in all of the 20 tomato lines at anthesis, whereas large variations of growth occur afterward. A strong, positive correlation, combining development and genetic diversity, is demonstrated between mean cell size and ploidy, which holds for mean cell diameters from 10 to 350 μm (i.e. a 32,000-times volume variation) and for mean ploidy levels from 3 to 80 C. Fruit weight appears also significantly correlated with cell size and ploidy. These data provide a framework of pericarp patterning and growth. They strongly suggest the quantitative importance of polyploidy-associated cell expansion as a determinant of fruit weight in tomato.
In Angiosperms, fruit typically develops from ovary after flower pollination and fertilization. In fleshy fruits, cells in the ovary wall undergo a long series of divisions and expansion, which give the fruit its final size before the onset of ripening (Esau, 1962; Coombe, 1976; Gillaspy et al., 1993; Giovannoni, 2004). Knowledge of these mechanisms associated with fruit growth and of their relationship with environmental factors increases our basic understanding of fruit development as well as paving the way for potential applications in agriculture (Giovannoni, 2004; Tanksley, 2004).
The tomato (Solanum lycopersicon) berry is one of the most studied fleshy fruits. The extensive genetic resources available for tomato and related species are illustrated by a wide variability of many characters of tomato fruit (Causse et al., 2002). The characterization of many mutant lines with specific alterations in fruit development is currently under way in various laboratories to unravel the mechanisms underlying fruit development (Mazzucato et al., 1998; Giovannoni, 2004; Tanksley, 2004). As far as growth is concerned, these studies require a detailed knowledge of fruit patterning. Few studies to date have given quantitative details of the kinetics and localization of cell division and of cell expansion in pericarp, a prominent tissue of tomato fruit formed from ovary wall (Smith, 1935; Gillaspy et al., 1993). In addition, although many studies have compared fruit growth in mutants and various wild-type tomato lines (Bohner and Bangerth, 1988; Mazzucato et al., 1998; Cong et al., 2002; Bertin et al., 2003; Liu et al., 2003), data from such varying genetic backgrounds remain difficult to reconcile into an integrated model of pericarp formation.
Pericarp becomes polyploid in several fleshy fruit species (Coombe, 1976), including tomato, where C values span from 2 to 256 C at the end of growth (Bergervoet et al., 1996; Joubès et al., 1999). In addition, Joubès et al. (1999) showed that another fruit tissue, the locular gel, which develops from the placenta after fertilization, is also polyploid, but not the epidermis. Endoreduplication is a frequent, somatic event in many plant organs and tissues (Kondorosi et al., 2000; Larkins et al., 2001; Barow and Meister, 2003; Sugimoto-Shirasu and Roberts, 2003). Its wide occurrence in plant and animal cells is largely assumed to be associated with metabolic activity and with cell expansion (Edgar and Orr-Weaver, 2001; Storchova and Pellman, 2004). This latter phenomenon is especially important in plants as compared with animals, as the volume of plant cells may increase by several orders of magnitude. In this way, cell expansion plays a critical role in the control of plant organ size as well as cell division (Dolan and Davies, 2004). One role of endoreduplication in the fruit could be to trigger the huge cell expansion that takes place in these organs (Coombe, 1976), but to our knowledge no data have demonstrated this hypothesis as yet. Although some studies have shown the importance of cell number for fruit size in tomato (Bohner and Bangerth, 1988; Frary et al., 2000), so far there is very little data dealing with the relationship between cell expansion and fruit size in this material (Tanksley, 2004), as well as in other species.
This study considers the variability of tomato fruit size to address the question of its dependence on cell size and polyploidy. It includes a kinetic analysis of pericarp development in a cherry tomato line, and a comparative analysis of pericarp in 20 tomato lines displaying a wide range of fruit size and genetic origins. Both approaches emphasize a significant correlation between polyploidy and cell size for this material. The contribution of cell size to final fruit weight is demonstrated, and the putative role of endoreduplication in this phenomenon is strongly suggested. These data provide a framework for the analysis of specific mutant lines. Moreover, they set the basis for a genetic approach to cell expansion and endoreduplication in tomato fruit.
RESULTS
Fruit Characteristics of 20 Tomato Lines
Twenty tomato lines were selected in the Institut National de la Recherche Agronomique collection of tomato genetic resources in Avignon (Table I). All lines are from S. lycopersicon, except cherry tomato Wva700 from S. pimpinellifolium. All lines have indeterminate growth, except Caline. These lines were chosen because of the very different fruit sizes and various genetic backgrounds. Mean fruit weights ranged from 3.8 to 431 g (Table I). Part of this variability was related to the number of carpellar locules, which varied from two to 22 (Table I). A set of 12 lines displayed fruit weights from 3.8 to 130 g while keeping two or three carpellar locules (exceptionally four; lines 1–12 in Table I).
Properties of the 20 tomato lines
The number before the name of each line identifies it in Figures 2 and 10. Fruit weight of ripe fruit (mean ± sd, n = 5–7) and locule number (n = 9–17; when more than one number is indicated, a bold character shows the most frequent locule number) were measured in fruit grown in Avignon, spring 2003, as detailed in “Materials and Methods.”
Organization of Pericarp at Anthesis, and Growth Initiation
In tomato, fertilization occurs within 20 h after anthesis (Picken, 1984). At that time, the carpellar wall of Wva106, a cherry tomato line, displays eight to nine cell layers, in which all cells have a roughly cubic shape and have nearly the same size, except for vascular bundles (Fig. 1A). The characteristics of ovaries at anthesis were compared in the 20 tomato lines (Fig. 2). Ovary diameter varied from 1.3 to 4.7 mm among the 20 lines, and from 1.3 to 2.4 mm between the 12 lines with only two to three locules per fruit (data not shown). Yet, pericarp pattern was found to be dramatically conserved between the 20 lines, as inter-line sds of pericarp thickness (Fig. 2A), mean cross-sectional pericarp cell area (Fig. 2B), and number of cell layers across pericarp (Fig. 2C) were less than 10% of the mean values of these parameters. The conserved structure of pericarp at anthesis is illustrated in Figure 1 for Wva 106 (Fig. 1A); for Ferum 26, which has the largest fruits with only two to three locules (Fig. 1B; Table I); and for Grosse de Gros, which shows the largest fruits among the 20 lines (Fig. 1C; Table I).
Structure and development of pericarp after anthesis. A to C, Cross sections of ovary wall at anthesis for Wva106 (A), Ferum 26 (B), and Grosse de Gros (C) lines. D, Pericarp cross section of 4-DPA Wva106 fruit, showing anticlinal (vertical arrows), periclinal (horizontal arrows), and randomly oriented (arrowheads) cell divisions in inner epidermal and subepidermal layers and in central cells. E, Enlarged portion of the outer pericarp of the same fruit as in D showing anticlinal divisions (vertical arrows) in outer epidermal layer and periclinal divisions (horizontal arrows) in outer subepidermal layer. oe, Outer epidermis; ie, inner epidermis; vb, vascular bundle. Scale bars: 50 μm (A–D) or 20 μm (E).
Cellular parameters of pericarp at anthesis and breaker stages. Tomato plants from 20 lines were grown in a greenhouse (Avignon). For each line, three to seven fruits were sampled for structural and cytological analyses at the time of anthesis (white columns) and at the beginning of breaker stage (gray columns). A, Pericarp thickness at anthesis (mm, left) and breaker (mm, right) stages. B, Mean cross-sectional area of one pericarp cell at anthesis (μm2, left) and breaker (mm2, right) stages. C, Number of cell layers across pericarp. D, MCV of pericarp cells at breaker stage.
The increase in pericarp thickness became detectable at 4 DPA in Wva106 (Figs. 1, A and D, and 3B). This was due both to cell division and to cell expansion. After anthesis, mitoses occurred at the highest rate in the two outer epidermal and subepidermal layers (Fig. 1, D and E), as well as in the two inner epidermal and subepidermal ones, but to a lower extent (Fig. 1D). In epidermal layers, only anticlinal cell divisions were detected, whereas they were mostly periclinal in the two subepidermal layers. Thus, the outer subepidermal layer, and to a lesser degree the inner subepidermal one, are the major sources of new cell layers during pericarp growth. Mitotic activity also occurred in more central cells, but to a much lesser extent and with various division planes. In Wva106, new cell layers arise very early, between 3 and 5 DPA (Fig. 3D). Cell expansion also resumed very early, as it was detectable at 4 DPA (Figs. 1 and 3C).
Kinetics of pericarp growth in the Wva106 line. Six Wva106 plants were grown in a greenhouse (Bordeaux). At each developmental stage, three to five fruits (total no. 102) were selected for measurement (mean ± sd). A, Fruit diameter. B, Pericarp thickness. C, Mean cross-sectional area of one pericarp cell. D, Number of cell layers across pericarp. E, MCV of pericarp cells, as described in Figure 5. Insets in C and E show enlarged kinetics for the 0 to 8 DPA period. Arrows show the breaker stage.
Pericarp Growth and Pericarp Structure at Breaker Stage
Fruit diameter and pericarp thickness of Wva106 fruits increased steadily from anthesis, and both parameters leveled off at green mature stage (Fig. 3, A and B). Then, a secondary increase of pericarp thickness by 1.5 times was evident at the transition between green mature and breaker stage, at the time when chlorophyll is degraded and carotenoids accumulate (Fig. 3B). In contrast, no increase of fruit diameter was detected at that time (Fig. 3A). When the pericarp thickness of the 20 tomato lines was compared at breaker stage, dramatically contrasted values were found (Fig. 2A). The increase of pericarp thickness from anthesis to breaker stage ranged from 10 times for cherry tomato lines as Wva106 to more than 50 times for Ferum 26. These variations were accounted for by generation of new cell layers and by cell expansion, as detailed below.
The production of new cell layers after anthesis varied from five in two cherry tomato lines, Wva700 and Wva106, to 17 in three larger fruit lines, Ferum 26, Montfavet 315, and Grosse de Gros (Fig. 2C). In 13 lines, the average number of new cell layers was between nine and 12. New cell layers originated predominantly from the outer subepidermis layer and less from the inner one, as suggested by mitotic activity in both layers and by counting cell layers above and below vascular bundles (data not shown).
Examples of fruit structure and pericarp pattern at breaker stage are shown in Figure 4. Most pericarp cells expanded a lot during fruit growth, to reach diameters of 200 μm and beyond (Fig. 4, A–C, right). In contrast, outer epidermal and subepidermal cells kept a size close to the one they had at anthesis (Figs. 1E and 4D). The mean size of pericarp cells was determined in cross sections of parenchymatous (not vascular) parts of the mesocarp (see the location of these measurements in Fig. 4C, right). The most outer and inner layers of pericarp and the vascular bundles were excluded from these measurements.
Pericarp structure at breaker stage. A to C, Cross sections of fruit (left) and pericarp (right) at breaker stage are shown for Grosse de Gros (A), Ferum 26 (B), and Wva106 (C) lines. D, Enlargement of Wva106 outer pericarp at the same magnification as Figure 1E; these show that epidermal and subepidermal cells have mainly enlarged tangentially since anthesis. The rectangle in A, right, is an example of an area within which mean cell size has been estimated, as detailed in “Materials and Methods.” Bars: 1 cm (A–C, left sections), 1 mm (A–C, right sections), and 20 μm (D).
In Wva106, pericarp cell expansion followed a two-step increase very similar to the kinetics of pericarp thickness (Fig. 3, B and C). In a first step, from anthesis to green mature stage, the mean cellular cross section area increased from 150 to 18,000 μm2. After a transient stationary level, cell expansion resumed rapidly at the transition with breaker stage, and mean cross section cell area attained 34,000 μm2 in ripening fruits. Quantitative differences in the extent of cell expansion were obvious between tomato lines at the breaker stage (Fig. 4, A–C, right). At that time, the mean cross-sectional area of mesocarp cells varied from 20,000 μm2 in Wva700 cherry tomato line to 100,000 μm2 in Ferum 26, Montfavet 133-5, and Montfavet 135-11 (Fig. 2B). It should be noted that the smallest mean cell sizes, below 40,000 μm2, were encountered both in cherry tomato lines and in some lines with the largest fruit, such as Grosse de Gros and Jaune Grosse Lisse (Figs. 2B and 4, A–C).
Ploidy Analysis in Tomato Fruit
Ploidy of cells from pericarp, locular gel, and central columella was analyzed in Bubjekosoko fruits at breaker stage (Fig. 5, A, C, and E). The three tissues display different ploidy profiles, with the largest C values (256 C) in pericarp. Sepals become also polyploid during fruit growth, but to a lower extent, with C values from 2 to 32 C (Fig. 5G).
Endoreduplication in tomato fruit. Ploidy has been analyzed by flow cytometry in pericarp (A and B), locular gel (C and D), columella (E and F), and sepals (G and H) of Bubjekosoko fruit grown in Bordeaux and harvested at breaker stage. Left sections show histograms of one representative fruit. Right sections (gray bars) show the frequency of each C value class as calculated from measurements performed on 29, 20, 19, and 12 fruits for B, D, F, and H, respectively. In B, the results from another set of three fruits from plants of the same line grown in Avignon are also shown (white bars, MCV').
Ploidy histograms and mean C value (MCV) are fairly reproducible, as relative sd of MCV is lower than 15% between similar fruits, except for columella (Fig. 5, B, D, F, and H). Pericarp and locular gel have a similar MCV of 35 to 40, larger than columella (MCV = 18) and sepals (MCV = 8). MCVs as well as ploidy profiles can usually be reproduced in pericarp of similar fruit of the same line, grown in different conditions (Fig. 5B). Very similar data were obtained with the Wva106 line (data not shown).
The kinetics of pericarp MCV during fruit development was investigated in the Wva106 line. At anthesis, the whole ovary contains mostly 2- and 4-C, and few 8-C nuclei (MCV = 3.2; Figs. 3E and 6A). In pericarp, MCV increased slightly to 7 within 10 DPA (Fig. 3E) because of the disappearance of 2-C (from anthesis), the increase of 8-C (from 3 DPA), and the appearance of 16-C nuclei (from 6 DPA). Then, it increased more steadily up to 32 at the breaker stage (Fig. 3E), owing to the successive appearance of 32-C (from 10 DPA), 64-C (from 13 DPA), 128-C (from 20 DPA), and 256-C (from 33 DPA) nuclei (Fig. 6B). Very similar data were obtained with the Bubjekosoko line (data not shown).
Developmental kinetics of C-value classes in Wva106 pericarp. Data are from the same experiment as in Figure 3. A, Evolution of 2-C (white triangles, dashed line), 4-C (white squares, dashed line), 8-C (black triangles, solid line), and 16-C (black squares, solid line) classes in fruit pericarp. B, Evolution of 32-C (white circles, dashed line), 64-C (white diamonds, dashed line), 128-C (black circles, solid line), and 256 C (black diamonds, solid line) classes in fruit pericarp.
The pericarp ploidy of the 20 tomato lines was analyzed at breaker stage (Fig. 2D). MCVs varied from 24 to 68 according to the line. Ploidy profiles of two lines with a low MCV and two lines with a high MCV are shown in Figure 7. C values from 4 to 128 C were systematically encountered in all lines. Interestingly, the lowest MCVs were encountered both in cherry tomato lines and in some lines with very large fruits, such as Jaune Grosse Lisse, Marmandaise, and Grosse de Gros. High MCVs were due to relatively high frequencies of 128-, 256-, and 512-C nuclei, but not to a significant decrease of nuclei with lower C values (Fig. 7). In nine lines, 256- + 512-C nuclei represented more than 5% of all nuclei. The highest MCVs were found in the three lines Montfavet 136-11, Ferum 26, and Saint-Pierre Clause in which 512-C nuclei have a frequency above 1% (Fig. 7).
Ploidy distribution in pericarp of four tomato lines. Data are mean ± sd of the frequency of each C-value class of three to seven fruits at breaker stage for four selected tomato lines, two lines with a low MCV value, Wva106 (white bars) and Grosse de Gros (light-gray bars), and two lines with a high MCV, Montfavet 136-11 (dark-gray bars) and Ferum 26 (black bars). Growth conditions and line characteristics are detailed in Figure 2.
Evidence for Correlation between Ploidy, Cell Size, and Fruit Size
The two sets of measurements of cell size and ploidy during fruit development in Wva106 (Fig. 3, C and E) and at breaker stage in 20 tomato lines (Fig. 2, B and D) were used to investigate the relationship between these two parameters. For this purpose, cell size was expressed as cell diameter, calculated from cross-sectional areas by assuming round-shaped cells.
Figure 8A shows that there is a significant correlation between ploidy and cell size in pericarp of 102 Wva106 single fruits collected at various developmental stages. Figure 8B shows that ploidy and cell diameter are also significantly correlated in fruit pericarp of 20 tomato lines at breaker stage. Although the correlation is somewhat weaker than in Figure 8A, because of more dispersed individual values, the most suitable regression is polynomial as in Figure 8A.
Relationship between cell size and ploidy. Each point shows the MCV and cell diameter in pericarp of a single fruit. The mean cell diameter was calculated from mean cross-sectional cell areas estimated as described in “Materials and Methods,” by approximating cells as spheres. White symbols in A and C report data from the same experiment as in Figure 3, where fruit development was analyzed in Wva106 line from anthesis to ripening (n = 102 fruits). Black symbols in B and C report data from the same experiment as in Figure 2, where 20 tomato lines were compared at breaker stage (all lines are represented by the same symbol; n = 81 fruits). Dashed lines show the polynomial regression curves for each set of data. R2 = 0.93 and 0.69 for A and B, respectively. The equation in C is y = 0.05x2 + 8.9x − 7.0, R2 = 0.96, n = 183.
The combination of both sets of data in Figure 8, A and B, reveals a unique relationship between ploidy and cell size in tomato pericarp (Fig. 8C). Cell diameter is positively correlated with ploidy by a polynomial, almost linear, over a wide range of variation: 35-fold for cell diameter and 27-fold for MCV.
As cell size is obviously one of the regulators of fruit size, we investigated the quantitative relationship between these two parameters. During Wva106 fruit growth, fruit weight increases from 3 to 4 mg at anthesis to 8 to 9 g at the late mature stage, having a close, positive correlation with the increase of mean pericarp cell diameter (Fig. 9A). No correlation was found between fruit weight and pericarp cell size when fruit from all 20 lines were analyzed at the breaker stage (Fig. 9B, white and black symbols). Because fruit weight is also dependent on the number of carpellar locules, the relationship between cell size and fruit weight was investigated for fruits with only two to three carpellar locules, and a significant correlation was found (Fig. 9B, black symbols). The combination of both sets of data in Figure 9, A and B, again reveals a unique correlation between fruit weight and mean pericarp cell diameter in tomato fruit with only two to three carpellar locules, whatever their developmental stage. As expected, fruit weight is approximately a cubic function of cell diameter (Fig. 9C).
Relationship between cell size and fruit size. Each point shows the weight of one fruit as a function of its mean pericarp cell diameter. Mean cell diameters were calculated from mean cross-sectional cell areas estimated as described in “Materials and Methods,” by approximating cells as spheres. Note the log scale of fruit weight in all panels. A, Same experiment as in Figure 3, where fruit development was analyzed in Wva106 line from anthesis to ripening. The dashed line is the regression curve (R2 = 0.96, n = 107). B, Same experiment as in Figure 2, where fruit from 20 tomato lines was compared at the breaker stage. Black symbols in B represent fruits with only two to three carpellar locules and the dashed line is the regression curve (R2 = 0.85, n = 41). White symbols in B show 40 fruits with four or more carpellar locules. C, Combination of all data from A (white symbols) and of data from fruits with only two to three carpellar locules in B (black symbols). The dashed line is the regression curve (y = 3.10−6x2.79, R2 = 0.96, n = 148).
The relationship of endoreduplication with fruit size is illustrated in Figure 10, which shows mean pericarp ploidy and fruit weight of each of the 20 tomato lines. As mentioned for cell size, ploidy is not correlated with fruit size when the 20 lines are compared (black and white symbols in Fig. 10), but a positive correlation becomes prominent when the 12 lines with two to three carpellar locules are considered (black symbols in Fig. 10).
Relationship between fruit size and ploidy. Fruit weight is shown as a function of MCV in pericarp. Each point represents one tomato line (mean of three to seven fruits per line, 20 tomato lines). Data are the same as in Figure 2. The number beside each symbol is the line number shown in Table I. Black symbols represent lines with only two to three carpellar locules, and the associated dashed line shows the polynomial regression curve (y = 0.0264x2 + 0.2622x − 21.758, R2 = 0.87, n = 12). White symbols show eight lines with four or more carpellar locules.
DISCUSSION
The Ovary Wall Structure Is Highly Conserved in Several Tomato Lines, and Different Mechanisms Control Pericarp Growth
This study shows that the pattern of ovary wall, including the number of cell layers and cell size, is dramatically conserved at the time of anthesis in 20 tomato lines, including 19 S. lycopersicon lines and one wild relative, S. pimpinellifolium. This phenomenon is remarkable with respect to the large variability among the 20 lines in ovary size and locule number on one hand, and in overall fruit growth and final pericarp pattern and thickness on the other hand (Table I; Fig. 2).
Our data indicate the cooccurrence of two distinct mechanisms of cell division in tomato pericarp after anthesis, as previously reported in grape (Vitis vinifera; Considine and Knox, 1981). Periclinal cell divisions, located mostly in the outer subepidermal cell layer and, to a lesser extent, in the inner one, generate five to 17 new cell layers according to the line. These will be referred to as histogenic cell divisions. In addition, randomly oriented cell divisions occur in many cell layers to accommodate pericarp growth with fruit growth; we refer to these as growth-related cell divisions. Several lines of evidence point to different controls for histogenic and growth-related cell divisions in tomato pericarp. Our data show that histogenic cell divisions proceed rapidly in the Wva106 line, as they are completed within 5 DPA, whereas mitotic activity remains significant in the pericarp up to 20 DPA. Similar data were reported in other lines (Joubès et al., 1999; Cong et al., 2002). The major quantitative trait locus fw2.2 has been shown to control fruit size directly through cell division, but not cell expansion (Cong et al., 2002; Liu et al., 2003). The authors showed that cell layer production is not regulated by fw2.2, which argues for different genetic controls of the two modes of division in pericarp. The variation between lines of the number of cell layers generated after anthesis suggests a genetic determinant for histogenic cell division. To which extent this pericarp-located genetic mechanism also operates in more central parts of the tomato fruit remains unknown.
The accumulation of some metabolites is heterogeneous throughout pericarp. For instance, starch predominantly accumulates in inner pericarp cells, and carotenoid synthesis during ripening is often more intense in outer pericarp (Smith, 1935). Variation in pericarp patterning, such as number and location of new cell layers, may thus modify the balance between metabolic pathways related to fruit quality. Cell layers are only defined here according to their position, and no assumption is made as to their homogeneity with regard to cell size and cell content.
Confocal analysis has revealed the volume and shape of inner mesocarp cells in grape berry (Gray et al., 1999). As in grape lines with almost round-shaped fruits, tomato pericarp cells are roughly isodiametric or egg shaped (C. Cheniclet, unpublished data). Thus, convenient approximations of cell diameters and volumes can be extrapolated from cross-sectioned cell areas. However, single-cell size measurements cannot be simply compared, e.g. as histograms, because of variability in the location of cross section planes. For this reason, we have only estimated the mean size of mesocarp parenchymatous cells. The large variation in this parameter during fruit growth and between different lines has added significant new data on the process of cell expansion in tomato pericarp.
Our data point to the rapid increase of cell size as early as at 4 DPA, i.e. before the end of histogenic and growth-related cell divisions (Figs. 1, A and D, and 3C). Cell expansion then occurs during 3 to 4 weeks up to the green mature stage and may be accompanied by cell division during 2 to 3 weeks. At the mature green stage, as compared with anthesis, the mean extrapolated pericarp cell volume has increased between 2,000 times in three cherry tomato lines, Wva700, Wva106, and Cervil, and 22,000 times in four lines, Montfavet 133-5, Kondine Red, Ferum 26, and Montfavet 136-11 (Fig. 2B). These data are in good agreement with the small amount of data available elsewhere concerning tomato pericarp (Bohner and Bangerth, 1988; Cong et al., 2002) and other fleshy fruits (Coombe, 1976; Gray et al., 1999; Higashi et al., 1999; Harada et al., 2005). They also indicate that the 20 tomato genotypes display a broad range of values, from 1 to 14, for their mean pericarp cell volume at the end of fruit growth. These results provide a solid basis for further genetic analysis of cell size regulation in tomato fruit.
A detailed time-course analysis of pericarp thickness and cell size reveals a rapid and transient step of cell enlargement at the transition between green mature and breaker stages (Fig. 3C). To our knowledge, this phenomenon has not yet been reported. It is not related to environmental changes of growth conditions, and it is reproducible (data not shown). It appears to be confined only to pericarp, as fruit size does not increase significantly at the same time. A decrease in tomato pericarp cell turgor has been reported at the beginning of ripening (Shackel et al., 1991). It is tempting to make this event the consequence of cell expansion at the same time. Although the underlying mechanism has not yet been explored, it could be related to cell wall alterations at the onset of ripening (Giovannoni, 2004).
Besides some results showing the importance of cell number for fruit size in tomato (Bohner and Bangerth, 1988; Frary et al., 2000), very few studies have addressed the quantitative contribution of cell expansion to the growth of fleshy fruit (Coombe, 1976; Tanksley, 2004). This study shows that the dramatic increase in pericarp cell size during tomato fruit development correlates nicely with the increase in fruit weight (Fig. 9A). This same correlation holds true when those lines that share a similar number of carpellar locules, so that there is no effect of locule number on fruit weight, are compared at the breaker stage (Fig. 9, B and C). This suggests that cell expansion in other tissues, such as locular gel and central columella, parallels that of pericarp cells. Indeed, cell expansion also occurs in these two tissues to a large extent, with columella cells remaining smaller than pericarp or locular gel cells (C. Cheniclet, unpublished data). These data demonstrate that the huge potential of plant cells for expansion is actually a strong determinant of fruit size in tomato.
Endoreduplication Occurs to Various Extents in Different Fruit Tissues and in Different Tomato Lines
Pericarp, locular gel, central columella, and sepals display significant differences in ploidy profiles with respect to distribution, highest value, mean, and mode of C values (Fig. 5). These data extend previous results on the same material (Bergervoet et al., 1996; Joubès et al., 1999; Bertin et al., 2003). The broad distribution of C values in a given tissue may relate partly to the presence of various cell types, as is the case for pericarp. Five C value classes were found in locular gel, despite the apparent lower heterogeneity of cell types in this tissue. A similar result was reported for maize (Zea mays) endosperm (Larkins et al., 2001; Dilkes et al., 2002) and in older reports (Buvat, 1965). The complexity of ploidy profiles in given tissues suggests discrete endogenous regulations within tissues, with cell age probably playing an influential role (List, 1963; Melaragno et al., 1993).
The tomato ovary comprises an equal number of 2 and 4 C cells at anthesis. Because 4-C ploidy can be attributed both to G2 or to the first endocycle, the unambiguous detection of polyploidy is only by 8-C nuclei, due to the second endocycle. This event has already started to a limited extent at anthesis, and it resumes at 3 DPA in tomato pericarp. These data agree with those of Bertin et al. (2003) to suggest that the first and second endocycles begin prior to anthesis in ovary cells. Then, the third, fourth, and fifth endocycles occur successively every 3 d up to 13 DPA (Fig. 6), which suggests that the mean endocycle duration in this material is 3 d. To our knowledge, such a value has not yet been reported for plant materials. The sixth (to 128 C) and seventh (to 256 C) endocycles appear much later, at 20 and 33 DPA, respectively, and only concern a small number of nuclei. This suggests a decrease in the efficiency of these endocycles, rather than the disappearance of a polyploidy-inductive signal after 13 DPA, as a significant increase of 32- and 64-C nuclei still occurs up to 40 DPA. As a whole, pericarp MCV increases steadily from anthesis to ripening in a way that confirms and extends previous reports (Bergervoet et al., 1996; Bertin et al., 2003). Notably, no significant change of ploidy occurs during the second phase of cell expansion at the end of the green mature stage (Fig. 3, C and E).
Significant variations in pericarp MCV were found between the 20 tomato lines at the breaker stage (Fig. 2D). They originated from variations in the largest C values, namely 128, 256, and 512 C. The large variation of MCV between lines suggests that a genetic component regulates the ability of each line to proceed through high levels of polyploidy in pericarp, i.e. through the sixth to eighth endocycles. A similar situation has recently been demonstrated in maize endosperm, with a 2-fold variation of MCVs between most Midwestern dent types and maize popcorns (Dilkes et al., 2002).
Endoreduplication Could Be a Regulator of Cell and Fruit Size
A positive correlation between cell size and ploidy has been demonstrated in numerous instances in a wide range of organisms (Day and Lawrence, 2000; Kondorosi et al., 2000; Edgar and Orr-Weaver, 2001; Sugimoto-Shirasu and Roberts, 2003; Storchova and Pellman, 2004). We have combined kinetic and genetic variations to demonstrate that such a correlation also occurs in developing tomato pericarp. The data of Figure 8C show a striking quantitative agreement with many data from different plant species, such as developing xylem cells of monocotyledonous roots (List, 1963), pea (Pisum sativum) cotyledon cells (Lemontey et al., 2000), Arabidopsis (Arabidopsis thaliana) epidermal pavement cells (Melaragno et al., 1993), and maize endosperm cells (Leiva-Neto et al., 2004). In these studies, the reported cell size and MCV values fit with the first part (MCV below 20) of the regression curve of Figure 8C (data not shown), which strongly supports the relevance of the correlation between ploidy and cell size. Tomato fruit offer the opportunity to extend this correlation to much larger ploidy values. In situ analysis of ploidy and cell size in tomato pericarp are under way to assess the location of polyploid cells in this tissue.
Although widely assumed, the correlation between cell size and ploidy is not systematic. Cell size is tissue specific, in a way unrelated to ploidy. In Arabidopsis, the ploidy of cortical root cells is not related to their size in ecotypes differing by organ size (Beemster et al., 2002), and the ploidy of hypocotyl cells can be uncoupled from cell expansion during seed germination (Gendreau et al., 1998). Ploidy and cell size of maize endosperm cells have also been shown to vary separately (Vilhar et al., 2002; Leiva-Neto et al., 2004). We have found only diploid nuclei in grape fleshy tissues (J.P. Renaudin, unpublished data), as already suggested (Ojeda et al., 1999), but grape mesocarp cells become almost as large as in ripe tomato pericarp (Gray et al., 1999; J.P. Carde, unpublished data). Large and exclusively diploid cells have also been reported in apple fruit (Malus sp.; Harada et al., 2005). In tomato fruits of varying size, pericarp ploidy was not modified because of their position in a truss, but a 30% variation of pericarp cell size was found (Bertin et al., 2003). An increase of cell volume by 50% to 100% occurs in tomato pericarp at the breaker stage in the absence of any ploidy increase (Fig. 3, C and E).
Endoreduplication appears to start a few days prior to cell expansion in tomato ovary, in a manner similar to that which occurs in Arabidopsis hypocotyls cells during germination (Gendreau et al., 1998) and in Arabidopsis trichomes (Hulskamp, 2004). This and the previous data are consistent with the hypothesis that endoreduplication is likely to be a driving regulator of cell expansion in these materials, although direct evidence awaits further demonstration. It appears also clearly that, alternatively, other phenomena are also able to promote cell expansion. Moreover, one additional function of endoreduplication, notably in fleshy fruits, could be to enhance cell growth rate, so as to shorten the duration of fruit growth, or to increase fruit size.
The contribution of polyploidy to the control of organ size has long been assumed from the observation of many constitutively polyploid plants (Day and Lawrence, 2000; Sugimoto-Shirasu and Roberts, 2003), and much less from developmentally controlled endoreduplication (Lemontey et al., 2000). This study shows that, in addition to well known parameters regulating fruit size in tomato, such as carpellar locule number, the mean ploidy level achieved in pericarp also correlates with fruit size. This suggests that the variation of the mean ploidy of the whole fruit parallels that of the pericarp. Quantitative trait loci analysis of fruit size in tomato (Causse et al., 2002; Tanksley, 2004) would be a valuable approach for identifying some genetic determinants of endoreduplication in this material.
This study provides a framework of pericarp patterning and growth for forthcoming genetic and functional genomic analyses of processes involved in tomato fruit development and quality. In particular, we reveal the dramatic extent of cell expansion, and we propose endoreduplication to play a driving role in this process in tomato.
MATERIALS AND METHODS
Plant Material
In a first set of experiments, seeds from 20 tomato lines (Solanum lycopersicon; Table I) were sown in January 2003. Five to eight plants per line were picked out and grown in the soil of a greenhouse in Avignon. In a second set of experiments, seeds from Wva106 and Bubjekosoko lines were sown in pots in January 2004. Five plants from each line were picked out in 25-cm pots with vermiculite and grown in a greenhouse in Bordeaux. In both experiments, the plants were grown under greenhouse conditions, with average daily minimal, medium, and maximal temperatures, respectively, of 15°C, 20°C, and 24°C from anthesis (March) to ripening (May). Air relative humidity was stable at 80% in both places. The plants were watered daily with a nutrient solution (Algospeed 1 g L−1, containing 13N-13P-24K-3Mg + oligoelements). Lateral shoots were removed regularly. Flowers were pollinated with an electrical bee. In the Avignon experiment, three to seven ovaries at anthesis or fruits at the transition between green mature and breaker stages were taken from position two to five of second to fourth truss for each line to perform cytology and ploidy analyses. In the Bordeaux experiment, three to six ovaries or fruits were taken at various stages from 0 to 72 DPA, from position two to five of first to seventh truss for cytology and ploidy analyses.
Cytological Analyses
Ovaries from the 20 lines sampled at anthesis and young fruits of the Wva106 line were prepared for cytological analysis by a resin-embedding method. After removal of floral organs, ovaries were cut at equatorial level and the two halves immersed in 2.5% glutaraldehyde in a phosphate buffer (0.1 m pH 7.2) for 2 h at room temperature. For the young fruit, an equatorial slice was excised and cut into fragments less than 4-mm wide before immersion in the fixative. During fixation, a partial vacuum was applied to extract intercellular gas. Samples were rinsed, dehydrated through an ethanol series, and embedded in Technovit 7100 (Kulzer) in 0.5-mL microtubes. Sections (1–3 μm thick) were made with glass knives on a Reichert 2040 microtome, stained with toluidine blue, and photographed on a Zeiss Axiophot microscope with a Spot digital color camera (Diagnostic Instruments).
In most of the 20 lines, the pericarp thickness of developed fruits exceeded the width of glass knives. Additionally, since embedding methods are time consuming and have low throughput, we developed a quick method for pericarp cytological analysis. Thin pericarp slices (0.3–0.6 mm thick, 1–2 cm long) were handmade with a razor blade in the fruit equatorial plane, by avoiding septa, and placed on the surface of a drop of 0.04% toluidine blue. After 10 to 15 min staining, they were rinsed briefly in water and immersed, with the colored face turned upside, into a small layer of water in a petri dish. Pericarp fragments were observed with a Leica FLIII stereomicroscope with illumination from above. Images were acquired with a Leica DC300F color digital camera.
Images acquired with both methods were analyzed with ImagePro-Plus software (Media Cybernetics). For each fruit, three to 10 portions of pericarp were analyzed. The number of cell layers from the outer epidermis to the inner epidermis was estimated in pericarp areas devoid of vascular bundles. The mean pericarp cell size was estimated using a method similar to that of Cong et al. (2002). A rectangular area (width = 1.5 mm, height = pericarp thickness × 80%) was drawn between outer and inner pericarp epidermis and centered in a zone containing no vascular bundles, as shown in Figure 4C. The mean pericarp cell size, excluding peripheral zones and vascular bundles, was calculated by dividing the rectangle area by the number of cells included in it.
Ploidy Analysis
Nuclei were prepared from whole ovaries at anthesis, and from various tissues of developing fruits by gentle chopping with a razor blade of 0.1–0.2 g fresh weight in 0.5 mL of Cystain UV ploidy solution (Partec). The suspension was filtered through a 100 μm nylon mesh and the remaining sample was reextracted with 0.5 mL of the same solution. The combined filtrates were analyzed on a Partec PAS-II flow cytometer. Data were plotted on a semilogarithmic scale. Calibration of C values was made with nuclei from young leaves and ovaries at anthesis, and from the observation of endosperm triploid nuclei when young seeds were analyzed (data not shown).
Ploidy histograms were quantitatively analyzed with DPAC software (Partec), after manual treatment to exclude noise. The MCV of each histogram was calculated as the sum of each C value class weighed by its frequency. Although this parameter overemphasizes high ploidy levels because of the exponential increase of DNA content during endoreduplication (Barow and Meister, 2003), it was retained for comparison with cell size, because calculation of mean cell size has shown also the same bias toward the largest cells. Moreover, calculation of cycle value, defined as the mean number of endoreduplication cycles per nucleus (Barow and Meister, 2003), gave very similar results (data not shown).
Acknowledgments
The technical assistance of A.M. Cassalter, J. Leonetti, and V. Rouyère in growing tomato plants is acknowledged.
Footnotes
-
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Jean-Pierre Renaudin (jean-pierre.renaudin{at}bordeaux.inra.fr).
-
Article, publication date, and citation information can be found at www.plantphysiol.org/cgi/doi/10.1104/pp.105.068767.
-
↵1 This work was supported by Région Aquitaine (contract no. 2004 0307002A).
- Received July 24, 2005.
- Revised September 25, 2005.
- Accepted September 28, 2005.
- Published November 25, 2005.
LITERATURE CITED
