The specific overexpression of a cyclin-dependent kinase inhibitor in tomato fruit mesocarp cells uncouples endoreduplication and cell growth


  • Mehdi Nafati,

    1. Institut National de la Recherche Agronomique (INRA), Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    2. Université de Bordeaux 2, Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author
  • Catherine Cheniclet,

    1. Institut National de la Recherche Agronomique (INRA), Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author
  • Michel Hernould,

    1. Institut National de la Recherche Agronomique (INRA), Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    2. Université de Bordeaux 2, Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author
  • Phuc T. Do,

    1. Max-Planck-Institut für Molekulare Pflanzenphysiologie, Am Mühlenberg 1, 14467 Potsdam-Golm, Germany
    Search for more papers by this author
  • Alisdair R. Fernie,

    1. Max-Planck-Institut für Molekulare Pflanzenphysiologie, Am Mühlenberg 1, 14467 Potsdam-Golm, Germany
    Search for more papers by this author
  • Christian Chevalier,

    Corresponding author
    1. Institut National de la Recherche Agronomique (INRA), Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
      (fax +33 557122541; e-mail
    Search for more papers by this author
  • Frédéric Gévaudant

    1. Institut National de la Recherche Agronomique (INRA), Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    2. Université de Bordeaux 2, Unité Mixte de Recherche 619 sur la Biologie du Fruit, BP 81, F-33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author

(fax +33 557122541; e-mail


The size of tomato fruit results from the combination of cell number and cell size, which are respectively determined by the cell division and cell expansion processes. As fruit growth is mainly sustained by cell expansion, the development of fleshy pericarp tissue is characterized by numerous rounds of endoreduplication inducing a spectacular increase in DNA ploidy and mean cell size. Although a clear relationship exists between endoreduplication and cell growth in plants, the exact role of endoreduplication has not been clearly elucidated. To decipher the molecular basis of endoreduplication-associated cell growth in fruit, we investigated the putative involvement of the tomato cyclin-dependent kinase inhibitor SlKRP1. We studied the kinetics of pericarp development in tomato fruit at the morphological and cytological levels, and demonstrated that endoreduplication is directly proportional to cell and fruit diameter. We established a mathematical model for tissue growth according to the number of divisions and endocycles. This model was tested in fruits where we managed to decrease the extent of endoreduplication by over-expressing SlKRP1 under the control of a fruit-specific promoter expressed during early development. Despite the fact that endoreduplication was affected, we could not observe any morphological, cytological or metabolic phenotypes, indicating that determination of cell and fruit size can be, at least conditionally, uncoupled from endoreduplication.


Endoreduplication is an alternative cell cycle characterized by the absence of mitosis and consisting of repeated DNA replication rounds, thus leading to an exponential increase in the amount of genomic DNA. This process, commonly associated with cell differentiation, is widespread in plants and particularly in angiosperms where it arises in most tissues (Nagl, 1976; D’Amato, 1984).

Several hypotheses have been proposed in an attempt to decipher the physiological role of endoreduplication. Since endoreduplication leads to nuclear DNA amplification, and therefore a multiplication of the gene copy number, this process could support an increase in transcriptional and metabolic activities (Galitski et al., 1999; Kondorosi and Kondorosi, 2004). However, this has not yet been clearly demonstrated in plants (Leiva-Neto et al., 2004). The increase in gene copy number could additionally provide a means to protect the genome against DNA damage caused by adverse environmental factors (Hase et al., 2006; Ramirez-Parra and Gutierrez, 2007).

The frequently observed positive correlations between endoreduplication and cell size in many different plant species, organs and cell types (Joubès and Chevalier, 2000; Sugimoto-Shirasu and Roberts, 2003; Kondorosi and Kondorosi, 2004) are commonly interpreted as evidence that endoreduplication is the driver of cell expansion. It is believed that successive rounds of DNA synthesis during endoreduplication consequently induce hypertrophy of the nucleus, thus influencing the final size of the cell which may therefore adjust its cytoplasmic volume with respect to the DNA content of the nucleus (according to the ‘karyoplasmic ratio’ theory; Sugimoto-Shirasu and Roberts, 2003). In tomato (Solanum lycopersicum), evidence for such a positive correlation between cell size and ploidy level has indeed been provided during fruit development (Cheniclet et al., 2005). In a recent survey (Bourdon et al., 2010), we showed that the ability to develop large cells in various fleshy fruit species is not restricted to endopolyploidizing fruits. Hence endoreduplication obviously does not appear to be a pre-requisite for cell expansion as previously reported (De Veylder et al., 2001; Beemster et al., 2002; Leiva-Neto et al., 2004). Nevertheless, it appears that endoreduplication clearly participates in modulating the rate of cell expansion and/or organ growth, such as in fruit development. Indeed when endopolyploidizing fruits are compared, it becomes apparent that the largest cells are present in fruits which undergo the greatest number of endocycles (e.g. tomato, pepper, melon), and that endoreduplication occurs in fruit from species exhibiting rapid fruit development (in <10 weeks) while it is absent in fruit from species where fruit development lasts for a very long period of time (over 17 weeks) (Bourdon et al., 2010).

In fruits of Cucurbitaceae and Solanaceae, mesocarp cells commonly undergo six rounds of DNA duplication (endocycle), and the highest ploidy levels for these cells are reached in tomato fruits where eight endocycles (up to 512C) can be observed (Bourdon et al., 2010). This high level of endopolyploidy in tomato fruits and the numerous data reported on this process in this species (Chevalier, 2007) makes it an outstanding model for studying endoreduplication and its physiological role during fruit development.

Part of the existing molecular control for classical cell cycle regulation is conserved in the endoreduplication cycle, especially that targeting the activity of cyclin-dependent kinase/cyclin (CDK/CYC) complexes which phosphorylate various protein targets, allowing the transition from one phase of the cell cycle to the next. The commitment to endocycle and the consequent lack of mitosis has been proposed to occur in the absence of a mitosis-inducing factor (MIF) which normally governs the passage through the G2–M transition (Inzé and De Veylder, 2006). Down-regulation of M-phase CDK activity is sufficient to drive cells into the endocycle (Vlieghe et al., 2007), and given that M-phase CDKB1;1 activity is required to prevent premature entry into the endocycle (Boudolf et al., 2004) it was proposed as a likely candidate kinase to be a component of a MIF. Various distinct mechanisms may account for the loss of CDK activity. The CDK phosphorylation status, the availability of the cyclin regulatory subunit and the involvement of specific CDK inhibitors represent just a few potential regulatory mechanisms. We could provide evidence of the direct implication of the two former mechanisms in the regulation of endoreduplication and its impact on tomato fruit development (Gonzalez et al., 2007; Mathieu-Rivet et al., 2010), and have previously hypothesized that CDK inhibitors could also play a part in endoreduplication-driven fruit growth in tomato (Bisbis et al., 2006).

In plants, most of the functional studies on CDK inhibitors have been focused on a particular class called interactors of cyclin-dependent kinase/Kip-related proteins (ICK/KRP) (Wang et al., 2007). ICK/KRP genes belong to a multigene family, and so far no direct phenotypic effect induced by the down-regulation or mutation of ICK/KRP genes has been reported, most probably because of functional gene redundancy. In addition, the simultaneous down-regulation of multiple ICK/KRPs shows a hyperplasic growth phenotype (Moulinier Anzola et al., 2010). However, numerous studies have reported the effects of ICK/KRP over-expression in Arabidopsis (Wang et al., 2000; Schnittger et al., 2003; Verkest et al., 2005; Weinl et al., 2005; Bemis and Torii, 2007), maize (Coelho et al., 2005), tobacco (Jasinski et al., 2002), rice (Barrôco et al., 2006) and Brassica (Zhou et al., 2002). A common phenotype associated with an ICK/KRP over-expression is an impairment of the cell cycle, resulting in plant dwarfism. At the cellular level, it appears that in all cases analysed, endoreduplication and cell size are affected. A low level of ICK/KRP over-expression appears to block cell division and induce nuclear endoreduplication, together with only a slight alteration in cell size and a slight decrease in final plant size (Verkest et al., 2005; Weinl et al., 2005). By contrast, a high level of ICK/KRP over-expression negatively affects both cell division and endoreduplication, generating cells of larger size, and results in an overall plant dwarfism (De Veylder et al., 2001; Jasinski et al., 2002; Zhou et al., 2003).

In the present work, we aimed to perform a functional analysis of the tomato ICK/KRP gene SlKRP1 during fruit development. As an initial step, we demonstrated that endoreduplication is directly proportional to cell diameter and fruit diameter, allowing us to establish a mathematical model for tissue growth according to the number of divisions and the number of endocycles. However, the over-expression of SlKRP1 under the control of a promoter specifically expressed during the phase of cell expansion of early fruit development revealed no general morphological, cytological or metabolic defects but did display a decreased level of endoreduplication. Thus for the conditions of study our data demonstrate that cell size and fruit size determination can be uncoupled from the endoreduplication process. The implications of these findings with respect to the relative role of both processes during fruit development will be discussed.


Fruit growth, cell enlargement and endoreduplication are linearly correlated during fruit development in tomato

Measurements of mean fruit diameter resulting from the geometric mean of the three diameters according to the X, Y and Z axes (Figure 1a), pericarp width and mesocarp cell area (Figure 1b), and endoreduplication levels were performed throughout fruit development (from anthesis to the onset of maturation) using fruits harvested from wild-type (WT) plants. As shown in Figure 2(a), the mean fruit diameter increased at a relatively constant rate from anthesis to 20 days post-anthesis (dpa). Interestingly, this indicates that no difference in growth rate occurred at the transition between the phase of cell division (0–6 dpa) and the phase of cell expansion (6–20 dpa) during the early development of tomato fruit. The mean fruit diameter is highly positively correlated and directly proportional to endoreduplication within pericarp cells during the entire period of fruit growth (Figure 2b). The correlation between endoreduplication in a particular tissue such as pericarp and fruit growth is due to the constant proportionality between pericarp thickness and fruit diameter (Figure 2c). At the cytological level, the same proportionality was observed between endoreduplication and the mean cell size in mesocarp (Figure 2d). From these observations, we conclude that fruit growth, pericarp thickness, mesocarp cell growth and pericarp cell endoreduplication all evolve according to a common proportionality law during fruit development. It is noteworthy that this relationship still occurred whatever the seasonal period of fruit growth (winter or summer) (Figure S1).

Figure 1.

 Definition of measurement axes used for morphometric analyses of fruit and cell size.
(a) Description of the axes used for measurements of mean fruit diameter.
(b) Median transverse section of a fruit and histological view of the pericarp. The larger arrow encompasses the pericarp as a whole; the smaller arrow encompasses the mesocarp where cell size was measured.

Figure 2.

 Development of wild-type tomato (cv. Wva106) fruits.
(a) Growth curve established by measuring the mean fruit diameter in the course of fruit development (from anthesis to 40 days post-anthesis, dpa) (= 71).
(b) Relationship between the mean fruit diameter and the endoreduplication index (EI) (= 65).
(c) Relationship between pericarp thickness and mean fruit diameter (= 23).
(d) Relationship between mean cell size in the mesocarp and the endoreduplication index (EI) (= 30).

Interestingly, the relative rates of fruit growth (Rf) and endoreduplication (Re) calculated across fruit development (see Experimental Procedures for calculations) (Figure 3a) were quite similar, from around 10 to 40 dpa. The observed discrepancy between anthesis and 10 dpa could be explained by active cell divisions which would be anticipated to contribute to the increased development of pericarp during this period (Figure 3b). This was confirmed when the relative rate of cell production (Rc) and Re were combined, as it resulted in a perfect match with Rf (Figure 3c). Hence, on the basis of the number of cell division cycles and the number of endocycles alone, we managed to represent the variations in fruit growth during the whole period of development. These data strongly support the existence of a linear quantitative relationship between the number of endocycles, cell growth and fruit growth, which could be expressed as follows: = (leiEI + l0) × 2DI (in axis), where L is the tissue size being considered (e.g. pericarp thickness), EI is the endoreduplication index corresponding to the mean number of endoreduplication cycles per nucleus, DI (in axis) is the division index corresponding to the mean number of successive cell divisions according to a considered axis, lei is a constant reflecting the mean gain in cell size resulting from each endoreduplication cycle and l0 is the mean size of 2C cells according to the axis considered.

Figure 3.

 Comparison of relative fruit growth rate (Rf) with relative endoreduplication rate (Re) (a), relative cell production rate (Rc) (b), and cumulative Re and Rc (c), in the course of wild type fruit development. dpa, days post-anthesis.

In order to test the significance of the model, we retrieved the data published by Cheniclet et al. (2005) describing the cellular parameters of developing pericarp. Indeed, the resin-embedding method for tissue sample preparation offers better accuracy for counting the number of sub-epidermal cell layers than toluidine blue-stained fresh tissues. We thus confirmed the existence of a linear correlation between endoreduplication and cell size (Figure 4a). In addition, we could also demonstrate that such a linear correlation occurs when using data from fruits harvested from 20 different tomato varieties displaying a large range in fruit size (Figure S2). As shown in Figure 4(b), on the basis of our model we could predict values for pericarp thickness (Table S2) which closely matched the experimentally measured pericarp thickness (Figure 3b) across the entire course of pericarp development.

Figure 4.

 Modeling pericarp growth.
(a) Relationship between endoreduplication index (EI) and mean cell size in the pericarp. Data used to establish the linear regression were obtained from developing Wva106 fruits and extracted from Cheniclet et al. (2005).
(b) Evolution of pericarp thickness during fruit development. Experimentally determined values for pericarp thickness from Cheniclet et al. (2005) were compared with predicted values using the following equation: predicted value = (EI × 0.0271−0.0048) × 2DI (DI is the division index). See Table S1 for predicted values used for EI and DI.

The fruit-specific overexpression of SlKRP1 during the cell expansion phase induced a dramatic decrease in endoreduplication

The expression patterns of the four different genes encoding ICK/KRP identified so far in tomato (Bisbis et al., 2006; Nafati et al., 2010), were analysed in various vegetative organs and across the course of tomato fruit development, using real-time quantitative PCR (qPCR) (Figure 5). All tomato KRP genes were ubiquitously expressed. However, their level of expression displayed significant differences. SlKRP1 and SlKRP3 were indistinctly expressed in immature and mature leaves, while SlKRP2 and SlKRP4 showed a preferential expression in immature dividing leaves. Interestingly, the transcripts for SlKRP2 and SlKRP4 were more abundant in the very early development of tomato fruit, when cell divisions are maximal inside the young developing fruit (Cheniclet et al., 2005), while the transcripts for SlKRP1 and SlKRP3 accumulated during the course of fruit development to reach a maximum at 20 and 30 dpa, respectively, when cell expansion accounts essentially for fruit growth.

Figure 5.

 Gene expression analysis of the ICK/KRP gene in tomato.
Quantitative real-time PCR analysis of SlKRP1 (a), SlKRP2 (b), SlKRP3 (c) and SlKRP4 (d) expression in tomato vegetative organs and developing fruits were performed using total RNA isolated from roots (Ro), immature (IL) and mature (ML) leaves, flowers at the 7 mm stage (Fl), and fruits harvested at 3, 5, 8, 10, 15, 20, 25 and 30 days post-anthesis (dpa) and the Breaker (Br) and Red Ripe (RR) stage. All values were normalized to the SlActin and SleiF4A housekeeping genes. The ΔΔCt method was used for relative quantification of mRNA abundance. Data are mean ± standard deviation (= 3).

We next initiated the functional analysis of SlKRP1 during tomato development. For this purpose a gain-of-function strategy was applied using constructs referred to hereafter as ProPEPC2:SlKRP1OE, in which SlKRP1 was expressed in the sense orientation under the control of the tomato PhosphoEnolPyruvate Carboxylase 2 (PEPC2) promoter (Guillet et al., 2002). The PEPC2 promoter is a fruit-specific promoter leading to high levels of expression during the phase of cell expansion and more specifically within the expanding cells of fruit mesocarp (Fernandez et al., 2009).

We ended up with three independent ProPEPC2:SlKRP1OE T2 lines, namely lines 2, 5 and 30, and measured the expression level of SlKRP1 in leaves and fruits from 6 to 40 dpa (Figure 6). A negligible expression of the transgene could be detected in leaves and young fruits at 6 dpa. After 10 dpa, the expression of SlKRP1 was increased to a 30-fold level in fruits from line 30 and to a 20-fold level in those from line 5, in comparison to WT fruits. Fruits from line 2 showed only a twofold increase for SlKRP1 expression at 10 dpa. The over-expression level remained stable until 20 dpa in each transgenic line during early fruit development but decreased thereafter.

Figure 6.

 Quantitative real-time PCR analysis of SlKRP1 expression in leaf and pericarp from developing fruits (at 6, 10, 15, 20, 30 and 40 days post-anthesis, dpa) from ProPEPC2:SlKRP1OE T1 lines 2, 5 and 30 compared with wild type (WT).
Quantification and normalization of expression are as in Figure 5.

Ploidy levels in fruit pericarp tissues from anthesis to 40 dpa were determined by flow cytometry (Figure 7). Between anthesis (Figure 7a) and 6 dpa (Figure 7b), no significant differences could be detected between WT and any of the ProPEPC2:SlKRP1OE plants. From 10 dpa to the end of fruit growth (40 dpa), endoreduplication was strikingly reduced in the strongest ProPEPC2:SlKRP1OE lines 5 and 30 (Figure 7c–g). As shown by the evolution of EI during fruit growth (Figure 7h), the rate of endoreduplication was halved between 10 and 20 dpa in the strong ProPEPC2:SlKRP1OE lines 5 and 30 in comparison with WT. After 20 dpa, the rate of endoreduplication in fruits from lines 5 and 30 matched that of the WT. However, the endoreduplication level in fruits from lines 5 and 30 did not overcome its initial delay (Figure 7g).

Figure 7.

 Effects on endoreduplication of SlKRP1 overexpression during fruit development.
(a–g) Ploidy level distribution in ProPEPC2:SlKRP1OE fruit pericarp as measured by flow cytometry during fruit development at anthesis (a), 6 (b), 10 (c), 15 (d), 20 (e), 30 (f) and 40 days post-anthesis (dpa) (g).
(h) Evolution of the endoreduplication index during fruit development. Values are mean ± standard error (> 4).

We next compared WT and the representative ProPEPC2:SlKRP1OE line 30 during the course of fruit development. In this analysis we documented the frequency of appearance of each DNA ploidy level, referred to as EF and calculated as described in Experimental Procedures (Figure 8). When compared with WT, the kinetics of the appearance of each class of DNA ploidy levels was obviously lower in ProPEPC2:SlKRP1OE line 30. Interestingly, a second wave of production of each ploidy level apparently occurs in the ProPEPC2:SlKRP1OE line 30. This second wave of endoreduplication was confirmed by estimating the endoreduplication ratio between lines 5 and 30 and WT (as described in Experimental Procedures). It was drastically decreased between 10 and 20 dpa in the ProPEPC2:SlKRP1OE lines but thereafter increased in comparison with WT (Figure S3a). Intriguingly, the levels of the SlCYCD3;1 transcript, used as a G1–S marker, and the SlCYCA3;1 transcript, used as an S-phase and endoreduplication marker (Mathieu-Rivet et al., 2010), decreased drastically at 10 dpa in ProPEPC2:SlKRP1OE fruits and similarly increased at 20 dpa when compared with WT (Figure S3b).

Figure 8.

 Kinetics of endoreduplication in the representative ProPEPC2:SlKRP1OE line 30 compared with wild type (WT), in the course of fruit development.
The frequency of appearance of each DNA ploidy level referred to as EF was calculated as described in Experimental procedures.

The decrease in endoreduplication does not alter fruit morphology, cell division activity or cell size determination in ProPEPC2:SlKRP1OE fruits

We next intended to evaluate the model postulated above for fruit growth in developing fruits of the ProPEPC2:SlKRP1OE lines (Figure 9). The EI in WT fruits at two developmental stages, namely 10 and 20 dpa, was experimentally determined by flow cytometry from fruits harvested at two independent seasonal periods. In parallel, EI was predicted using linear regression of the data of mean fruit diameter (as described in Figure S1; Figure 9a). Irrespective of the cultivation season, the predicted values for the WT were found to perfectly match with the experimental values. When applied to fruits from ProPEPC2:SlKRP1OE line 30, however, the model was no longer valid since the experimentally determined EI values diverted significantly from the predicted values, the experimentally determined EI being considerably smaller than would be anticipated (Figure 9a). Whatever the growth conditions were, values obtained from ProPEPC2:SlKRP1OE fruits deviated from the linear regression established with WT fruit values (Figure 9b), thus indicating that fruit growth, as represented by the mean fruit diameter, was no longer correlated to endoreduplication in SlKRP1 over-expressing fruits.

Figure 9.

 Phenotypical analysis of ProPEPC2:SlKRP1OE lines compared with wild type (WT).
(a) Comparison between experimentally determined endoreduplication index (EI) and predicted EI. The EI were determined experimentally by flow cytometry using nuclear preparations from 10 and 20 days post-anthesis (dpa) pericarp from WT and ProPEPC2:SlKRP1OE fruits harvested from plants cultivated in summer (July–August 2009) and winter (December 2009–January 2010). The predicted EI were obtained using the linear regression obtained from data for mean fruit diameter (as described in Figure S1). Star symbols above bars indicate that Student’s t-tests are significantly different ( 0.001) from WT.
(b) Absence of relationship between the mean fruit diameter of ProPEPC2:SlKRP1OE fruits and the EI. Data were obtained from 10 and 20 dpa fruits harvested from plants cultivated in summer (July–August 2009) and winter (December 2009–January 2010).

To explain these unexpected results, we performed a careful analysis of morphological and cytological parameters during fruit development in three selected lines (Figure 10). Fruit size as expressed by the mean fruit diameter (Figure 10a), pericarp width on the equatorial plane (Figure 10b), the number of cell layers across the pericarp (Figure 10c) and the mean mesocarp cell area (Figure 10d) were thus determined. For each of these parameters, no significant differences could be observed between the ProPEPC2:SlKRP1OE lines and WT plants. Interestingly, this was also the case when pericarp width and mean mesocarp cell area were determined on the antero-posterior axis (X, Z as shown in Figure 1; data not shown).

Figure 10.

 Morphological and cytological analysis of ProPEPC2:SlKRP1OE fruits compared with wild type (WT) in the course of fruit development.
(a) Evolution of mean fruit diameter.
(b) Evolution of pericarp thickness.
(c) Number of cell layers in the Z-axis of the pericarp.
(d) Mean cell size in the Z-axis in the pericarp.

We next investigated whether the observed decrease in endoreduplication had affected nuclear size. Nuclei were prepared from 10 and 20 dpa pericarp tissues from WT and ProPEPC2:SlKRP1OE line 30; the mean area of isolated nuclei was then measured by imaging and ploidy levels were determined by flow cytometry (Figure S4). As expected the size of the nuclei correlated perfectly with the ploidy levels in both WT and ProPEPC2:SlKRP1OE line 30 fruits. Therefore, the decrease in ploidy levels and consequent decrease in nuclear size in ProPEPC2:SlKRP1OE fruit occurred while cell size was unaffected (Figure 9d), indicating that a disruption in the nuclear-to-cytoplasmic ratio was modified. Taken together, these results revealed that the PEPC2-driven over-expression of SlKRP1 during the phase of cell expansion induced neither morphological nor cytological effects on fruit development. In addition, while the mathematical model for fruit growth is fully relevant for WT fruits, genetic modification can clearly induce a complete uncoupling of endoreduplication from cell size.

Fruits from ProPEPC2:SlKRP1OE lines displayed no consistent changes in their metabolite content

Metabolite profiling using GC-MS was performed to compare the metabolite content in fruits from WT and ProPEPC2:SlKRP1OE lines 5 and 30, at 15, 20, 30 and 40 dpa (Table S2). Globally no significant differences could be observed between the ProPEPC2:SlKRP1OE lines and WT plants. In keeping with this statement line 5 was even more closely related to the WT (Pearson coefficient = 0.9946) than to line 30 (Pearson coefficient = 0.9857), when evaluated on a global basis. That said, when the data were assessed on a metabolite-by-metabolite basis, the majority of metabolites did not display dramatic or consistent differences across the genotypes studied.


Establishing a model for mesocarp growth according to cell cycle and endocycle values

In early developmental stages, tomato fruit growth results from the combination of cell number and cell size which are respectively determined by cell division and cell expansion processes according to two successive developmental phases (Gillaspy et al., 1993). In the course of fruit development, Cheniclet et al. (2005) demonstrated the existence of a clear correlation between endoreduplication and cell and fruit size. We here confirm these data and reveal a linear relationship between endoreduplication, cell growth and fruit growth during the phase of cell expansion in tomato fruit development (Figures 2–4). Interestingly, this indicates that the fruit growth rate is constant throughout development, whatever the mode of growth according to cell proliferation or cell expansion associated with endoreduplication (Figure 2). This observation is consistent with that recently documented in the tomato cultivar MoneyMaker as well as the wild tomato species Solanum pennellii (Steinhauser et al., 2010), and is in full accordance with those obtained from Arabidopsis leaf and sepal epidermis (Horiguchi et al., 2006; Roeder et al., 2010).

This led us to propose the following formula to predict fruit growth from endoreduplication and cell division parameters: = (leiEI + l0) × 2DI (in axis). According to the established linear regression (Figure 2b), the value for lei (representing the mean gain in cell size resulting from each endoreduplication cycle) was estimated to be 0.0271 mm and the value for l0 (representing the mean cell size of a population solely composed of diploid cells with EI = 0), was estimated as a negative value (−0.0048 mm) using the data provided by Cheniclet et al. (2005) (Figure 4a). One would have expected a positive value for l0 and we therefore postulate that the proposed formula applies only to cells under endoreduplication. This suggests that the increase in cell size encountered at the end of the G2 phase which transiently reaches the 4C state in proliferating cells, is less than the increase in cell size induced by a round of endoreduplication (lei), found at a constant value.

In the present work our kinetic analysis allowed us to model tomato pericarp growth on the basis of the number of cell divisions and endocycles. To our knowledge, this linkage has never been demonstrated before. It would be interesting to test this function in other types of fruit tissue such as the jelly-like locular tissue undergoing extended endoreduplication (Joubès et al., 1999), as well as in other plant species, in order to assess its generality.

The regulation by the PEPC2 promoter restricts the effect of SlKRP1 over-expression to endoreduplicating cells in tomato mesocarp

Over the last decade, several reports have described the effects of ICK/KRPs on cell division and endoreduplication (for a review see Wang et al., 2007). With the exception of the specific over-expression of ICK/KRPs in Arabidopsis trichomes (Schnittger et al., 2003), all these reports made use of promoters that are active in both dividing and endoreduplicating cells, precluding clear conclusions about the independent effect of ICK/KRPs on cell division or endoreduplication in a particular developing tissue context. To investigate the effect of endoreduplication on fruit growth independently from cell divisions, we aimed at deregulating the expression of the tomato ICK/KRP gene SlKRP1, which was shown to be associated with DNA ploidy increase in fruit development (Bisbis et al., 2006). For this purpose we used the tomato PEPC2 promoter (Fernandez et al., 2009) which drives a high level of transgene expression in tomato fruit pericarp between 6 and 30 dpa, i.e. essentially during the cell expansion phase of fruit development (Figure 4).

The PEPC2-driven specific over-expression of SlKRP1 in fruit pericarp did not produce any apparent phenotype at the morphological and cytological levels (Figure 9). We could not detect any decrease in the number of pericarp cell layers between the transgenic and WT lines, indicating that cell division was not affected. While most cell divisions have already been achieved prior to 6 dpa, some divisions continue to occur in the external cell layers of mesocarp until 15 dpa (Cheniclet et al., 2005). Given that the over-expression of SlKRP1 remains very high from 10 to 20 dpa in ProPEPC2:SlKRP1OE fruits (Figure 4), we could expect an effect on these divisions in the external mesocarp. However, Fernandez et al. (2009) showed that PEPC2 promoter expression occurs preferentially in the largest cells of the central mesocarp of tomato fruits, thus explaining the absence of any effect on cell divisions within the mesocarp of ProPEPC2:SlKRP1OE fruits.

The use of the PEPC2 promoter to alter the endoreduplication process in tomato fruit mesocarp was fully relevant, as DNA ploidy levels were clearly lowered (Figures 7 and 8). Thus the effect of SlKRP1 over-expression on ploidy levels in the endoreduplicating cells of mesocarp was quite similar to that obtained in other plant models and organs such as Arabidopsis leaf mesophyll cells (De Veylder et al., 2001; Verkest et al., 2005; Weinl et al., 2005) or rice endosperm cells (Barrôco et al., 2006).

The fruit-specific over-expression of SlKRP1 results in an uncoupling of endoreduplication and cell growth

Numerous studies in plants have documented a clear numerical relationship between endoreduplication and cell growth, suggesting a causal relationship between these two cellular processes. Nowadays, the function of an accelerator for organ growth, such as in fruit (Bourdon et al., 2010), or for plant growth in response to environmental constraints (Barow and Meister, 2003) is a favored role for endoreduplication. Indeed we have been able to demonstrate that the over-expression of the anaphase complex activator CCS52A, required for proper destruction of cyclins, had an impact on fruit development by sustaining and even enhancing fruit growth via an induction of endoreduplication (Mathieu-Rivet et al., 2010). Nevertheless the role of endoreduplication in modulating the rate of organ growth and/or cell expansion has been disputed (John and Qi, 2008), mainly because functional analyses aimed at deregulating target genes related to endoreduplication and/or cell size have also affected progress in the cell cycle and therefore also affected cell division. Four different situations have emerged from the literature. (i) Endoreduplication is induced while cell divisions are inhibited. This scenario occurs when progress into M phase is hampered by down-regulation of B-type CDKs (Boudolf et al., 2009), A-type cyclins (Cebolla et al., 1999) or by a mild up-regulation of KRPs (Wang et al., 2000; De Veylder et al., 2001; Jasinski et al., 2002; Zhou et al., 2003). In such a situation the premature blockage of mitosis induces endoreduplication, generating plants with fewer larger size cells (Cebolla et al., 1999; Wang et al., 2000; Jasinski et al., 2002) or cells of approximately equal size (De Veylder et al., 2001; Zhou et al., 2003) to those of the WT. (ii) Endoreduplication is decreased while cell divisions are promoted. This is the case when D-type cyclins are up-regulated (Dewitte et al., 2003; Qi and John, 2007). At the organ level no significant difference in size can be observed when compared with WT; however, tissues are composed of more cells of a smaller size than those observed in the WT. (iii) The dual decrease of endoreduplication and cell division can be achieved by targeting genes involved in S-phase or in both S and M phases (Hemerly et al., 1995; De Veylder et al., 2001; Jasinski et al., 2002; Zhou et al., 2003). In this extreme situation, plants are characterized by dwarfism; the plant organs are composed of very few cells which are generally diploid and far larger than WT. (iv) Endoreduplication is decreased independently from cell divisions which are not affected. This scenario was reported in plants over-expressing a dominant negative form of CDKA in the endoreduplicating cells of maize endosperm using the 27 kDa γ zein promoter (Leiva-Neto et al., 2004). As such this represented the first report showing that cell size determination can be independent of the level of endoreduplication. As the case was unique and observed in a highly differentiated tissue, the authors proposed a specific role for endoreduplication in maize endosperm given the importance of this tissue as a source of nitrogen reserves that can be mobilized to fuel development.

Here we provide a second demonstration that cell size can be uncoupled from endoreduplication, this time in tomato fruit pericarp. Indeed, the use of the PEPC2 promoter to drive the specific over-expression of SlKRP1 in endoreduplicating cells within fruit pericarp provided data which are remarkably similar to those from Leiva-Neto et al. (2004). Although normal pericarp development is characterized by the close relationship between endoreduplication and cell growth (Figures 2–4), in full agreement with the ‘karyoplasmic ratio’ theory (Sugimoto-Shirasu and Roberts, 2003), the reduced levels of endoreduplication in ProPEPC2:SlKRP1OE fruits (Figures 7 and 9) do not result in any morphological or cytological defects (Figure 9a–c), any reduction in cell size (Figure 9d) or any consistent differences in biochemical composition (Table S2). Interestingly, Leiva-Neto et al. (2004) showed that lower levels of endoreduplication in maize endosperm have only minor effects on the content of starch and storage protein, as well as on the associated transcribed genes. As far as maize endosperm and tomato fruit are concerned, these data thus suggest that endoreduplication is unlikely to contribute to the regulation of transcriptional activity and subsequent metabolic activity by increasing the availability of DNA templates for gene expression. It is noteworthy that the metabolic modifications observed for tomato fruits over-expressing the endoreduplication-associated gene SlCCS52A were mostly secondary effects resulting from the observed alteration in fruit growth (Mathieu-Rivet et al., 2010).

Endoreduplication or cell growth: what comes first?

The mean level of endoreduplication in various plant organs has repeatedly been found to be correlated with organ size. From the literature two mechanisms accounting for organ growth are opposed: first, the final size of an organ can result from a given balance of cell-based (autonomous) growth relying on division, expansion and endoreduplication; second, growth itself is the dominant regulator of cell proliferation, and the determinant of final cell size and ultimate organ size, according to an organismal level of regulation (Mizukami, 2001; John and Qi, 2008). In view of the organismal control, the synthesis of cytoplasm is the primary process and cell division and endoreduplication-driven cell enlargement are secondary processes to maintain the karyoplasmic ratio (Cookson et al., 2006; John and Qi, 2008). Although the correlation between cell size and endoreduplication is obvious, the direction of causality remains very much a matter of debate. Since endoreduplication corresponds to successive rounds of DNA duplication in the absence of mitosis, it would appear likely that a minimal cell size must be required to commit to a subsequent round of DNA replication, thus implying cell growth. However once DNA synthesis is completed, doubling of the quantity of DNA can in turn promote cell growth, according to the ‘karyoplasmic ratio’ theory (Sugimoto-Shirasu and Roberts, 2003).

The strong proportionality between ploidy level and cell size during normal plant development reported in this work is obviously valid at the scale of a cell population within a tissue or an organ, and is thus experimentally determined according to means of parameters. At the cellular level, a high degree of variability occurs within each population studied, which impairs the application of our model. Indeed, individual cells cannot be assigned to clearly defined subpopulations of a given size, since cell size shows a continuous distribution. Hence it is highly probable that for a given cell size, several distinct ploidy levels may occur. When referring to the mean number of endoreduplication cycles as the definition for EI (Figure 7h), the population of cells in ProPEPC2:SlKRP1OE fruit mesocarp performed one endoreduplication round under WT cells. Nevertheless, the attained levels of endoreduplication were still sufficient to support the optimal cell size increase observed in WT (Figure 10d). These observations suggest that endoreduplication would be likely support a range of cell sizes rather than a defined one. This assertion would infer that endoreduplication precedes cell growth, as reported during the elongation of the hypocotyl in Arabidopsis (Traas et al., 1998).

In conclusion, we have demonstrated here that the fruit-specific overexpression of SlKRP1 under the control of the PEPC2 promoter has a negative impact on endoreduplication within fruit pericarp, similar to that previously observed for strong KRP overexpressors. However, final fruit size and mean pericarp cell size were not affected, revealing that it was possible to uncouple endoreduplication and cell growth in ProPEPC2:SlKRP1OE fruit. Although endoreduplication was not totally impaired in ProPEPC2:SlKRP1OE fruit, it is likely that enough ploidy still occurred to support cell growth. This would thus infer that endoreduplication does not exert direct control on cell growth but would rather be a limiting factor for cell growth, in accordance with the previous model from Schnittger et al. (2003). In the absence of such a direct control, we propose that endoreduplication and cell growth may be co-regulated by a common upstream factor, since both processes follow the very same dynamics during fruit development.

Experimental Procedures

RNA extraction, reverse transcription and real-time quantitative PCR

Total RNA was extracted from vegetative organs and pericarp from fruits harvested at the following developmental stages: anthesis, 3, 5, 6, 10, 15, 20, 30 and 40 dpa, using TRI® Reagent (Sigma-Aldrich, After extraction, RNA samples were DNAse-treated using the TURBO DNA-free kit (Applied Biosystems, Complementary DNA was synthesized from 1 μg of total RNA using IScript cDNA Synthesis kit (Bio-Rad, Real-time quantitative PCR was performed with a CFX96 Real-Time PCR Detection System (Bio-Rad) using a final volume of 25 μl of reaction mixture containing 12.5 μl of iQ SYBR Green Supermix (Bio-Rad), 0.2 mm of each primer and 1 μl of the template diluted to the tenth. For all real-time qPCR experiments, each reaction was performed in triplicate. To determine relative fold differences for each sample, the Ct value for each gene was normalized to the Ct value for internal references (cDNAs encoding actin and eiF4A).

Generation of ProPEPC2:SlKRP1OE lines and plant growth conditions

The sequence encoding KRP1 from the pET28-KRP1 vector (Bisbis et al., 2006) was amplified in a two-stage PCR reaction and inserted into the GATEWAY vector pDONR201 (Invitrogen, by attB recombination following the manufacturer’s protocol. An error-free entry clone was confirmed by sequence analysis before recombination into a pK2GW7 vector ( to produce the pK2GW7-his-KRP1 construct. A 2-kb DNA fragment containing the SlPEPC2 promoter (ProPECPC2) (Fernandez et al., 2009) was amplified using specific primers flanked by SacI/SpeI restriction sites. The pK2GW7-his-KRP1 construct and ProPECPC2 DNA fragment were both separately digested using SacI and SpeI, and recombined by ligation. The ligation product, the pK2GW7-pPEPC2-KRP1 vector, was introduced into Agrobacterium tumefaciens strain GV3101 by transformation and subsequently into Solanum lycopersicum cv. WVA106 plants using the cotyledon transformation method as described previously (Gonzalez et al., 2007). Transgenic plants were selected on kanamycin-containing medium and later transferred to soil. Homozygosis was verified by real-time qPCR on genomic DNA of T2 plants. For each transgenic line, three T2 plants were analysed and compared with three independent WT plants. Tomato plants were grown in a greenhouse under a thermoperiod of 25°C/20°C and a photoperiod of 14/10 h (day/night).

Cytological methods

Fruits were harvested at various developmental stages determined according to dpa and weight. Fruit diameters according to the X, Y and Z axes (as shown in Figure 1) were measured as to determine the mean fruit diameter (MFD) calculated as the geometric mean of the three diameters according to the formula: inline image. Morphometric analyses were performed by using a gauge of 2.5 × 103 μm2. To visualize the size of mesocarp cells, the surface of a freshly sectioned fruit was incubated in 0.1% toluidine blue, washed in water and dried on a filter paper. Images of pericarp sections were taken using a microscope (Zeiss Axioplan, and then analysed with imagepro-plus software (Media Cybernetics, To evaluate the mean cell size in mesocarp, cell surface values were square rooted. Mean cell length in pericarp was directly measured according to the Z-axis. Data from morphometric measurements were statistically analyzed and Student’s t-test was used to evaluate the significance of the results.

Flow cytometry analysis

Ploidy profiles of isolated nuclei from tomato pericarp and ovaries at anthesis were determined as described (Cheniclet et al., 2005).

Calculations for endoreduplication parameters

The endoreduplication index (EI) representing the mean number of endoreduplication cycles was calculated according to Barow and Meister (2002) using the formula


where Fx is the frequency of the peak x considered.

The frequency of appearance of each DNA ploidy level in a defined time interval (EF) was calculated according to the formula


where x is the considered DNA ploidy level, Fn is the frequency of the ploidy level n, tN is the developmental stage and tN−1 is the preceding developmental stage. The calculated EFx thus represents the number of nuclei reaching the following ploidy level.

The endoreduplication ratio (ER) between ProPEPC2:SlKRP1OE and WT fruit was calculated for each developmental stage using EI according to the formula


where tN is the development stage (in dpa) and tN−1 is the preceding stage.

Calculations of relative rates of fruit growth, endoreduplication and cell production

The relative rates of fruit growth (Rf), endoreduplication (Re) or cell production (Rc) per day were calculated according to the formula


where F is the considered factor (mean fruit diameter for Rf, EI for Re or number of cell layers for Rc), tN is the developmental stage (in dpa) and tN−1 is the preceding developmental stage.

Metabolite profiling

Metabolite extraction, derivatization and GC-MS analysis were performed as described previously (Lisec et al., 2006) with modifications specific to tomato fruit (Schauer et al., 2006). Data processing was performed using TagFinder software (Luedemann et al., 2008), identifying metabolites by comparison with database entries for authentic standards (Schauer et al., 2005).


This research was supported by the 6th Framework Program of the European Commission, within the European Solanaceae Integrated project, EU-SOL (grant no. FOOD–CT–2006–016214), and by funding from the Region Aquitaine; MN was supported by grant no. 24220–2006 from the Ministère de l’Enseignement Supérieur et de la Recherche (France). We are indebted to Dr Christophe Rothan for very stimulating scientific discussions within the EU-SOL project, and Mrs Patricia Ballias and Aurélie Honoré for excellent technical work in taking care of plants.