INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

The eukaryotic cell cycle is controlled by the ordered action of a protein complex composed of a catalytic subunit named cyclin-dependent kinase (CDK) and positive regulatory elements named cyclins. The association of the CDK with its cyclin partner determines the activity of the complex, its stability, its localization, and substrate specificity (Pines, 1994). The complex activity is also regulated by the phosphorylation status of the kinase (Dunphy, 1994), the controlled degradation of the cyclin subunit (Peters, 1998), and the binding of CDK inhibitors or regulatory factors (Lees, 1995).

The complexity of the plant cell cycle is reflected by the existence of several types of cyclins and CDKs. So far, plant cyclins have been classified into five major groups: A, B, C, D, and H (Renaudin et al., 1996; Yamaguchi et al., 2000). The A- and B-type cyclins known as mitotic cyclins accumulate during the S, G2, and early M phase and during the G2 and early M phase, respectively (Mironov et al., 1999). D-type cyclins control the progression through the G1 phase in response to growth factors and nutrients (Riou-Khamlichi et al., 2000). C- and H-type cyclins have been characterized recently in poplar (Populus tremula × tremuloides) and rice (Oryza sativa; Yamaguchi et al., 2000). Both of them were found to interact specifically with the rice CDK-activating kinase, but only Orysa;CycH;1 could activate the kinase, suggesting that it is the effective regulatory subunit.

In plants, five distinct classes of CDKs, CDKA through CDKE, have been defined according to phylogenetic, structural, and functional similarities with animal and yeast CDKs (Joubès et al., 2000a). This classification is mainly based on the conservation of the PSTAIRE motif in the cyclin-binding domain. The CDKA class groups functional homologs of the yeast p34^cdc2/CDC28 protein displaying the PSTAIRE canonical motif. Their expression and translation patterns are constitutive during the cell cycle. CDKBs possess a divergent motif: either PPTALRE or PPTTLRE, reflecting the existence of two subgroups, CDKB1 and CDKB2, respectively. Because both fail to functionally complement temperature-sensitive mutants of yeast CDC2/CDC28, they represent plant-specific CDKs. CDKAs are supposed to regulate both the G1-S and G2-M transitions, whereas CDKBs regulate the G2-M transition (Mironov et al., 1999). The three other CDK families (CDKC, D, and E) representing non-PSTAIRE kinases are poorly characterized and their function in the cell cycle regulation remains unclear. However, CDKDs were identified as CDK-activating kinases that activate the CDK/cyclin complexes by phosphorylation of the Thr residue within the T-loop region of CDKA (Yamaguchi et al., 1998).

Plant CDKCs display a PITAIRE motif in the cyclin-binding domain. So far, they have been found in alfalfa (Medicago sativa), pea (Pisum sativum), and Arabidopsis (Feiler and Jacobs, 1991; Newman et al., 1994; Magyar et al., 1997); no precise function could be assigned. In mammals, the PITAIRE motif identifies CDC2-related kinases such as the human cholinesterase-related cell division (CHED) kinase, which is required during hematopoieisis (Lapidot-Lifson et al., 1992). The CDK9 kinase displaying a PITALRE motif also belongs to this group. Its function has been well documented in human because the CDK9 kinase activity is involved in the regulation of gene transcription elongation (De Falco and Giordano, 1998; Bregman et al., 2000).

Early development of tomato (Lycopsersicon esculentum Mill.) fruit offers an interesting model for studying plant organogenesis, particularly the regulation of cell division and cell expansion phenomena that appear to account for two distinct developmental phases (Gillaspy et al., 1993). In an attempt to investigate the molecular mechanisms governing fruit organogenesis, we described the involvement of cell cycle genes encoding CDKA as well as mitotic and G1 cyclins, in the spatial and temporal regulation of mitotic activity in developing tomato fruits (Joubès et al., 1999, 2000b).

In this work, we present the characterization of new tomato CDK genes and their putative implication in fruit development. Three full-length cDNAs encoding a C-type CDK and B1- and B2-type CDKs were isolated from tomato fruit cDNA libraries. We analyzed the expression patterns of these genes during fruit development and in response to different hormone and nutrient regimes using tomato cell suspension and in vitro-cultured excised roots. We demonstrated that the CDKC protein does not interact with any of the mitotic and G1 cyclins isolated in tomato so far.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Isolation and Characterization of cDNAs Encoding Additional Tomato CDKs: CDKC, CDKB1, and CDKB2

To isolate fruit-specific cDNAs preferentially expressed during the cell expansion phase of development (Gillaspy et al., 1993; Joubès et al., 1999), we performed a reverse northern procedure based on a differential screen discriminating expanding fruits cDNAs from full-expanded leaves cDNAs. Among the selected clones, we identified a 1,827-bp cDNA harboring a complete ORF encoding a 512-amino acid peptide. The encoded protein displayed a high percentage of identical residues with CDKCs from alfalfa and Arabidopsis: 63% with Medsa;CDKC;1 (Magyar et al., 1997) and 59% and 64% with Arath;CDKC;1 and Arath;CDKC;2 respectively (Newman et al., 1994). It shared only 30% of identical residues with the tomato CDKA;1 (Joubès et al., 1999). The putative tomato CDKC comprised functional regions characteristic of CDKs (Fig. 1A): the cyclin-binding domain displaying a PITAIRE motif (residues 63-78), the catalytic domain (residues 163-173), and the T-loop region (residues 184-205) containing a Thr residue at position 199 (corresponding to T161 of CDKA), which phosphorylation stabilizes the cyclin binding. Furthermore, the primary sequence exhibits the three Arg residues (at positions 71, 164, and 188) involved in the stabilization of the phosphorylated T loop, and the two phosphorylation sites: T35 and Y36 corresponding to the conserved amino acids T14 and Y15 of CDKA. From the analysis of the predicted primary sequence and protein alignments, this cDNA is likely to code for a new C-type CDK from tomato. Thus, it was named Lyces;CDKC;1 according to the proposed plant CDK gene nomenclature (Joubès et al., 2000a).

View larger version (52K): [in this window] [in a new window]   Figure 1.   Comparison of amino acid sequences of CDKs. A, Multiple alignment of the deduced amino acid sequences of tomato CDKs (Lyces;CDKC;1, AJ294903; Lyces;CDKB1;1, AJ297916; Lyces;CDKB2;1, AJ297917; and Lyces;CDKA;1, Y17225), alfalfa CDKC (Medsa;CDKC;1, T50815), and the two Arabidopsis CDKCs (Arath;CDKC;1, T09572 and Arath;CDKC;2, BAA97308). Deduced amino acid sequences of the different CDKs were compared using the multiple alignment program CLUSTAL W version 1.7. The identical residues between Lyces;CDKC;1 and the different CDKs are represented by dots. Residues essential for activity are indicated by stars. Positions of the cyclin-binding domain (I), the catalytic domain (II), and the T-loop region (III) are boxed in black. Gaps () were introduced for maximizing the alignments. B, Phylogenetic analysis of CDKs. The most conserved part of the CDK core (from residue 11-276 of CDKAs) was compared using the multiple alignment program CLUSTAL W version 1.7. The Phylip package was used for the construction of the phylogenetic tree and comparison of parsimony and distance (Neighbor Joining algorithm) methods. The robustness of the tree was assessed by the bootstrap method on 1,000 replicates. The length of lines is proportional to the genetic distance between each node. Within the tree, the species are indicated by the first letters of the generic and specific names as described by Joubès et al. (2000a). Tomato CDKB1, CDKB2, and CDKC are highlighted in gray.

Using sequencing data from the The Institute for Genomic Research (Rockville, MD) Tomato Gene Index, we could identify two expressed sequence tags (ESTs) corresponding to cDNAs encoding respectively a member of the CDKB1 (EST 326331) and the CDKB2 subfamily (EST 336587). The corresponding full-length cDNAs were isolated after screening the "young tomato fruit" cDNA library (Joubès et al., 1999). The analysis of the predicted amino acid sequences (Fig. 1A) revealed the presence of the PPTALRE and PPTTLRE motives characterizing CDKB1 and CDKB2 proteins, respectively (Joubès et al., 2000a). The two cDNAs thus were named Lyces;CDKB1 and Lyces;CDKB2. The sequence of their respective encoded proteins shared 60.7% of identical residues, and only 46.2% and 44.4% with that of CDKA;1. The amino acid sequence of Lyces;CDKC;1 isolated herein showed a lower percentage of identity with these two B-type CDKs (25% with CDKB1 and 32% with CDKB2).

The identity of the tomato CDKC, CDKB1, and CDKB2 was confirmed by a phylogenetic analysis using the multiple alignment program CLUSTAL W (Fig. 1B). Lyces;CDKC;1 was classified clearly into the cluster containing the plant CDKCs (ms;c1, Medsa;CDKC;1; at;c1, Arath;CDKC;1; and at;c2, Arath;CDKC;2) distantly from the other clusters of plant CDKs. Within the same cluster, two other branches of the tree can be defined, displaying respectively the animal PITAIRE CDKs (human hs;CHED, accession no. Q14004, hs;CrkRs, accession no. AF227198, and Caenorhabditis elegans ce;9, accession no. P46551) and PITALRE CDK9 (human hs;9, accession no. P50750, and Drosophila melanogaster dm;9, accession no. AAB84112).

Expression of CDKC and CDKB Genes during Fruit Development and in Plant Organs

The relative transcript levels corresponding to the different tomato CDK genes were analyzed by semiquantitative reverse transcriptase (RT)-PCR using a combination of primers located in the coding sequence and in the 5'- or 3'-untranslated region sequence of the cDNAs, respectively (Table I). Specific primers for an actin cDNA were used as an internal control of RT-PCR.

In the course of fruit development (Fig. 2A), the Lyces;CDKC;1 transcripts were highly expressed at anthesis (A) and up to 5 DPA, i.e. during the cell division phase. After 5 DPA, their abundance decreased to reach a basal level during the cell expansion phase (8-20 DPA) and until the onset of ripening (MG stage). The expression profiles of Lyces;CDKC;1 and Lyces;CDKA;1 were similar, although the relative amount of Lyces;CDKA;1 transcripts was higher. Lyces;CDKB1;1 and Lyces;CDKB2;1 were also highly expressed from anthesis to 5 DPA. Their expression decreased abruptly at 8 DPA and almost disappeared at the MG stage. During fruit development, the expression of the actin cDNA was found to be constitutive. The transcript levels for these four tomato CDKs as well as the actin cDNA were enhanced in all plant organs displaying meristematic activity such as young leaves and roots. In stems, the level of CDKC and the two CDKB transcripts was very low. In full-differentiated leaves, i.e. in nondividing tissues, a very weak expression signal could be observed for Lyces;CDKC;1, whereas Lyces;CDKB1;1 and Lyces;CDKB2;1 transcripts were undetectable.

View larger version (78K): [in this window] [in a new window]   Figure 2.   Expression analysis of CDK genes in tomato. A, Semiquantitative RT-PCR analysis of CDK gene expression during fruit development and in vegetative organs. Total RNA to be used for RT was isolated from fruits harvested at the following developmental stages: anthesis (A), 3, 5, 8, 10, 15, and 20 DPA, and mature green (MG) stage, and from young leaves (YL), differentiated leaves (DL), roots (Ro), and stems (St). The specific amplification of cDNA fragments was detected after gel electrophoresis, Southern blotting, and hybridization to the corresponding 32P-labeled probes. B, Semiquantitative RT-PCR analysis of CDK gene expression in the different fruit tissues during development. Total RNA was isolated from dissected fruit tissues (epidermis, pericarp, and gel) at the following developmental stages: 10, 15, and 20 DPA; MG and red ripe (RR) stages.

We previously demonstrated that the distribution of mitotic activity is not only temporally but also spatially regulated in the tomato fruit (Joubès et al., 1999). Cell division occurs in the epidermis until the end of development, whereas the inner pericarp and gel (locular tissues) cells are characterized by an arrest of mitosis after 15 DPA and the concomitant endoreduplication of the nDNA content. Therefore, we investigated the expression of tomato CDK genes in the different tissues of the developing fruit (Fig. 2B). In the epidermis, the CDKC and actin transcripts appeared to be constitutively expressed between 10 DPA and the RR stage, whereas CDKA transcripts accumulated until the MG stage. Lyces;CDKB1;1 and Lyces;CDKB2;1 were weakly expressed when compared to the other CDK genes. However a longer exposition of the autoradiography showed similar patterns of expression than that of CDKA (data not shown). In the pericarp, mRNAs corresponding to CDKC and CDKA accumulated up to 20 DPA, then decreased to almost undetectable levels at the RR stage. Lyces;CDKB1;1 was weakly but constantly expressed between 10 DPA and the MG stage. Lyces;CDKB2;1 was expressed between 10 and 15 DPA, and then became barely detectable. The actin gene was almost constituvely expressed between 10 DPA and the MG stage, and then its expression decreased at the maturation stage. In the gel tissue, Lyces;CDKC;1 was always expressed during development, with very high levels between 10 and 15 DPA. The expression of Lyces;CDKA;1 was found to be constitutive during the gel development up to the MG stage. The genes encoding the tomato CDKB1 and CDKB2 were mainly expressed up to 15 DPA and almost disappeared thereafter. The transcript level for the actin cDNA decreased gradually in the course of the gel development.

Lyces;CDKC;1 Is Expressed in Floral Meristem and Young Fruits

To deepen the analysis of the temporal and spatial gene expression patterns, the accumulation of Lyces;CDKC;1 transcripts was further investigated by mRNA in situ hybridization (Fig. 3). In flower buds 1.5 to 2 mm long, a weak signal was uniformly distributed in all tissues (Fig. 3A). However, a stronger signal was observed in the primordia of petals and stamens, and in the L1 cell layer (Fig. 3A, inset). In fruits harvested at anthesis, a transient accumulation of the transcripts was detected in the pericarp, the placenta, and the ovules (Fig. 3C). At a higher magnification, the hybridization signal was observed clearly in the epidermis and subepidermis cell layers of the pericarp, i.e. the exocarp and the endocarp (Fig. 3E). In ovules, Lyces;CDKC;1 transcripts were detected in the cell layers of the integument and the nucellus. Then the hybridization signal decreased gradually in 5-DPA fruits (Fig. 3G).

View larger version (109K): [in this window] [in a new window]   Figure 3.   Analysis of Lyces;CDKC;1 transcript accumulation by in situ hybridization. Longitudinal sections of early stage flower buds (1.5-2 mm in size) and fruits harvested at anthesis and 5 DPA were hybridized with digoxigenin-labeled RNA probes. A and B, Section of floral bud; C and D, section of a fruit at anthesis; E and F, higher magnification of the section of a fruit at anthesis; G and H, section of a fruit at 5 DPA. A, C, E, and G were hybridized with the antisense probe; B, D, F, and H were hybridized with the sense probe (negative controls). en, Endosperm; ep, epidermis; in, integument; nu, nucellus; ov, ovule; pe, pericarp; pl, placenta; pp, petal primordium; se, seed; sp, stamen primordium. Bar scale = 100 µm.

The Expression of Lyces;CDKC;1 Is Not Controlled by Sugar and Phytohormones

To investigate the influence of sugar and phytohormones on the transcription level of the different CDK genes, we used tomato cell suspension cultures (Fig. 4A). When cells were incubated without Glc, BAP, and 2,4-D (Control 2), the levels of transcripts for CDKC, CDKBs, and CDKA were greatly reduced compared with un-starved cells (Control 1). After growth under various regimes of sugar and hormone combinations, the level of Lyces;CDKC;1 mRNAs was enhanced when compared with Control 2, thus suggesting that sugar starvation has little effect on the Lyces;CDKC;1 transcription. The refeeding of the cells with Glc induced the transcription of the CDKA and CDKB genes, independently of the hormonal status. However, after 72 h of sugar starvation, the addition of one or both hormones to Glc-free medium did not induce the transcription of CDKA and CDKB genes, which remained similar to that in Control 2. Sugar starvation exerted a similar and drastic effect on the accumulation of actin transcripts as previously observed by Sheu et al. (1994).

View larger version (46K): [in this window] [in a new window]   Figure 4.   Effect of hormones and sugars on CDK gene expression. A, Semiquantitative RT-PCR analysis of CDK gene expression in cell suspensions under the control of nutrient and hormonal factors. Freshly diluted tomato cell suspension cultures were grown for 72 h in a complete medium and subsequently depleted of auxin, cytokinin, and Glc. After depletion for 72 h, the cells were subdivided and grown in the presence (+) or absence () of Glc (G), 6-benzylaminopurine (BAP; B), and 2,4-dichlorophenoxyacetic acid (2,4-D; D). Twenty-four hours later, the cells were harvested. Controls 1 and 2 represent cells harvested before depletion and 72 h after depletion respectively. B, Semiquantitative RT-PCR analysis of CDK gene expression in excised tomato roots submitted to Suc starvation. Tomato roots were excised (control C) and cultivated in liquid medium in the presence (+) or absence () of Suc during 4 d. Starved roots were refed with Suc during 24 h after 1 or 3 d of starvation.

The ability of sugars to control the expression of the CDK genes was further analyzed using excised tomato roots submitted to sugar starvation (Fig. 4B). The expression of Lyces;CycA2;1 and Lyces;CycD3;1 cDNAs encoding a CycA2 mitotic cyclin and a CycD3 G1 cyclin, respectively (Joubès et al., 2000b) was also investigated. In agreement with the results obtained with cell suspensions (Fig. 4A), the level of CDKC transcripts in excised tomato roots was poorly affected by Suc starvation because it remains constant during the whole starvation period (Fig. 4B). The transcripts for CDKA and the two CDKBs, as well as actin, were rapidly and strongly affected by sugar starvation. The accumulation of the CycA2 and CycD3 cyclin transcripts were impaired similarly because they became almost undetectable after4 d of starvation. The addition of Suc back to the medium after 1 or 3 d of starvation induced the transcription of all the genes tested. It is interesting that when Suc was added back to the medium, the CDKC gene transcription was induced compared with the level of transcripts in the control corresponding to excised untreated roots, thus suggesting a positive effect of sugars on the CDKC gene transcription.

Lyces;CDKC;1 Does Not Interact with Tomato Mitotic and G1 Cyclins

To examine whether Lyces;CDKC;1 can interact with the already known tomato cyclins, a yeast two-hybrid approach was used. The putative protein-protein interactions between the different tomato cDNAs encoding CDKs (CDKA1, CDKA2, CDKB1, CDKB2, and CDKC; Joubès et al., 1999; this work) and cyclins (Lyces;CycA1;1, Lyces;CycA2;1; Lyces;CycA3;1, Lyces;CycB2;1, and Lyces;CycD3;1; Joubès et al., 2000b) were tested (Table II). To increase the stability of the putative complexes formed in the yeast cell, we used constructs in which the destruction box and PEST domain of the mitotic cyclins and the D-type cyclin, respectively, were deleted. The expression of fusion proteins between the GAL4 DBD and the different CDKs on one hand, and the GAL4 AD and the different cyclins on the other hand, was confirmed by western blot using anti-LexA and anti-Gal4 antibodies, respectively (data not shown). The interaction between the two GAL4 domains was revealed by the growth on a medium lacking His (His) and by measuring the -galactosidase activity resulting from the activated transcription of the lacZ gene.

As shown in Table II, the two CDKAs interacted very efficiently with CycA2, and to a lower extent to CycD3. They were unable to interact with CycA1, CycA3, and CycB2. CDKB2 interacted also with CycA2 but not with other kinds of cyclins. No interaction could be observed between CDKC or CDKB1 and any of the tested cyclins, suggesting the existence of other types of cyclin protein able to interact with these CDKs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Three cDNAs from tomato, named Lyces;CDKC;1, Lyces;CDKB1;1, and Lyces;CDKB2;1, representing new CDKs were isolated and their expression patterns in the course of early fruit development and under different nutritional regimes investigated.

Based on the characterization of their amino acid sequence and according to a phylogenetic analysis (Fig. 1), we identified Lyces;CDKC;1 as a member of the C-type CDK family characterized by the presence of a PITAIRE motif in the cyclin-binding domain, and Lyces;CDKB1;1 and Lyces;CDKB2;1 as members of different subfamilies of mitotic CDKBs (Joubès et al., 2000a). Analysis of the The Institute for Genomic Research Tomato Gene Index allowed the identification of two ESTs encoding CDKC: The first one (EST 401472) corresponds to Lyces;CDKC;1 isolated in this work and the second one (EST 307422) to a new CDKC. A genomic Southern-blot analysis using Lyces;CDKC;1 confirmed the existence of at least two genes in tomato (data not shown). The predicted translation products of these two tomato CDKC genes, as well as those deduced from the only two genes identified in the Arabidopsis genome, are characterized by the presence of solely a PITAIRE motif. This suggests that, unlike animal CDKCs such as CDK9, the PITALRE motif may not be found in the plant CDKC family. It is interesting that the two branches of the phylogenetic tree clustering animal PITAIRE and PITALRE CDKs seem to diverge after the formation of the plant CDKC cluster (Fig. 1B). Hence, this could imply that the plant CDKC group may contain indifferently functional homologs of both the animal PITAIRE and PITALRE CDKs.

The isolation of these new CDKC, CDKB1, and CDKB2 from tomato allowed us to investigate how their gene expression may be developmentally regulated during fruit organogenesis, and especially to determine their respective involvement in the cell division process. Cell division plays indeed a crucial role during fruit organogenesis because the number of cells resulting from mitotic activity is an essential determinant of the final size and sink strength of the fruit (Gillaspy et al., 1993; Frary et al., 2000). Early development of tomato fruit can be divided into two distinct phases (Gillaspy et al., 1993). The first one lasting for about 7 to 10 d after fertilization and fruit set is characterized by a very active period of cell divisions inside the ovary. During the second phase, fruit growth is mostly due to cell expansion, thus leading to a fruit that exhibits its almost final size and is able to ripen. Inside the developing tomato fruit, the distribution of mitotic activity is spatially and temporally regulated in relationship with the differential expression of CDKA and cyclins (Joubès et al., 1999, 2000b). At the transcriptional level, we already showed that the CDKA expression was associated with dividing tissues but also with expanding tissues (Joubès et al., 2000b). In the course of fruit development and in vegetative organs, the overall expression pattern of Lyces;CDKC;1, Lyces;CDKB1;1, and Lyces;CDKB2;1 were similar to that observed for Lyces;CDKA;1, i.e. a higher level of transcripts in dividing organs displaying meristematic activity (Fig. 2A). At the spatial level, the in situ hybridization experiment confirmed the preferential expression of Lyces;CDKC;1 in dividing tissues (e.g. floral organ primordia and ovules; Fig. 3). A uniformly distributed signal was observed in these dividing tissues, but clearly Lyces;CDKC;1 does not display a patchy pattern of expression as expected for a cell cycle-regulated gene.

It is interesting that a marked difference in the expression pattern of Lyces;CDKB1;1 and Lyces; CDKB2;1 was observed in differentiated leaves as the transcripts completely disappeared (Fig. 2A). This result is in accordance with the typical function of CDKBs in dividing cells (Mironov et al., 1999; Joubès et al., 2000a). This was further confirmed by analyzing the CDK gene expression in the different tissues of developing fruits, and especially in the gel tissue (Fig. 2B). After 15 DPA, the cells composing the gel stop dividing and expand concomitantly with nDNA endoreduplication (Joubès et al., 1999). Because the endoreduplication cycle is made of the succession of S and G phases without mitosis (for review, see Joubés and Chevalier, 2000), the expression of Lyces;CDKB1;1 and Lyces;CDKB2;1 was strongly reduced. A very faint signal could still be observed up to the MG stage for Lyces;CDKB1;1, whereas Lyces;CDKB2;1 transcripts became undetectable after 15 DPA. This slight difference in the gene expression may originate from the S/G2 and G2/M phase dependence at the transcriptional level of CDKB1 and CDKB2, respectively (Mironov et al., 1999; Mészáros et al., 2000). In Arabidopsis, the only CDKB gene so far studied, CDC2b, corresponds to a member of the CDKB1 family (Arath;CDKB1;1, Joubés et al., 2000a). This gene was found recently to play a role in regulating seedling growth in darkness independent of cell division or endoreduplication (Yoshizumi et al., 1999), thus suggesting that Arath;CDKB1;1 may be involved in regulating directly hypocotyl cell elongation or a specific phase of the cell cycle and/or overall chromosome spatial organization, critical for hypocotyl cell elongation and cotyledon development. These findings agree with the lack of participation in the promotion of endoreduplication we could also observe for Lyces;CDKB1;1 in the gel (Fig. 2B). Because Lyces;CDKB1;1 is still expressed at a low level in expanding cells of the gel, it may play a role in the regulation of cell expansion as suggested by Yoshizumi et al. (1999), implying a striking difference in the respective function of Lyces;CDKB1;1 and Lyces;CDKB2;1.

In the gel tissue, the Lyces;CDKC;1 transcripts accumulated at high levels in dividing cells (until 15 DPA). When only endoreduplication occurs (after 15 DPA), the accumulation of transcripts dropped to a basal level. Lyces;CDKA;1 behaved quite differently because the transcripts were highly expressed during the development of the gel tissue at very high levels to the onset of ripening. Hence, it is tempting to suggest a role for CDKA in the regulation of the endocycle (Joubés and Chevalier, 2000).

Very little is known about the control of plant CDKC gene expression. Because sugar and hormones have a role of prime importance during the process of fruit development (Gillaspy et al., 1993), we investigated the influence of these parameters on the regulation of Lyces;CDKC;1 transcription. The gene expression of Lyces;CDKC;1 was only slightly modulated by sugars or hormones (Fig. 4A). The transcription of Lyces;CDKA;1 was induced by the presence of Glc, whereas that of Lyces;CDKB1;1 and Lyces;CDKB2;1 were strictly dependent upon sugar availability. Such results dealing with the nutritional control of CDKA and CDKB gene transcription are in agreement with those described for Arabidopsis cell suspension by De Veylder et al. (1999). In excised tomato roots (Fig. 4B), even after 4 d in the absence of Suc, the Lyces;CDKC;1 gene expression was not affected by Suc starvation, whereas the CDKA and CDKB gene expressions were deeply affected, as well as those for mitotic cyclin CycA2;1 and G1 cyclin CycD3;1 used as controls of cell cycle- and sugar-regulated genes (Riou-Khamlichi et al., 1999, 2000; Burssens et al., 2000). The observation that Lyces;CDKC;1 expression is not modulated by nutritional or hormonal parameters and the constitutive presence of transcripts in cells that have stopped to divide suggests that the CDKC gene expression is not cell cycle-regulated. This is supported by the absence of a patchy pattern in the in situ hybridization experiments (Fig. 3), in tissues displaying an intense cell division activity, such as floral primordia, developing ovules, and very young fruits. Moreover, it was demonstrated that the expression of CDKC transcripts was constitutive throughout the cell cycle in alfalfa synchronized cells (Magyar et al., 1997), and that the gene expression of the human CDK9 kinase was not cell cycle-regulated and its kinase activity did not vary appreciably during the cell cycle (Bregman et al., 2000).

The next question we assessed was to determine whether Lyces;CDKC;1 could interact with the different cyclin partners available in tomato (Joubès et al., 2000b). Using the yeast two-hybrid system, different authors demonstrated the protein-protein interactions between CDKA or CDKB and various D-type cyclins (De Veylder et al., 1997, 1999; Nakagami et al., 1999; Meijer and Murray, 2000; Mészáros et al., 2000), CDKA and CycA2 (Roudier et al., 2000), and CDKD and its specific partners CycH and CycC (Yamaguchi et al., 2000). However, no exhaustive analysis of plant CDK/cyclin binding has been performed so far. Here, we were able to test a combination of 25 different CDK/cyclin interactions (Table II). Like in other plant species, we demonstrated that tomato CDKA interact with CycA2 and CycD3;1, but not with other mitotic cyclins (CycA1, CycA3, and CycB2). Moreover, the tomato CDKB2 could bind CycA2. In our hands, CDKB1 failed to interact with any of the tested cyclins, unlike the alfalfa CDKB1, which binds a cyclin D3 (Mészáros et al., 2000). We were unable to reveal any interaction between Lyces;CDKC;1 and mitotic and G1 cyclins. Therefore, the cyclin partner of plant CDKC remains to be identified among the numerous different cyclin genes present in a plant genome (e.g. more than 30 in Arabidopsis).

The function of CDKC in plant cells remains to be elucidated. In mammals, the function of PITAIRE kinases (such as the CHED kinase) is unknown. However the second type of CDKC-related kinases (CDK9) received much more attention as the kinase activity is associated with protein complexes implicated in HIV gene transactivation (Romano et al., 1999). Human CDK9 binds cyclins T and K, leading to the formation of different complexes with specific activity (Bregman et al., 2000). The CDK9/CycT complex, called the positive-transcription elongation factor b, possesses a carboxyl-terminal domain (CTD) kinase activity which was shown to activate the gene transcription in vivo (Napolitano et al., 2000). The CTD phosphorylation of the RNA polymerase II largest subunit by positive-transcription elongation factor b enables the polymerase to promote the transcription elongation by counteracting the effect of negative factors. Hence, CDK9 is a multifunctional kinase involved in the control of cell growth and/or cellular viability, even if its activity is not cell cycle-regulated (De Falco and Giordano, 1998). Assuming that Lyces;CDKC;1 may represent a plant orthologue of CDK9, it is tempting to hypothesize that its preferential gene expression in actively dividing cells rather than in expanding/endoreduplicating cells, could be associated with a putative CTD kinase activity responsible for the control of gene transcription. This putative function of Lyces;CDKC;1 may explain the intense signal of hybridization observed in floral primordia and developing ovules (Fig. 3). It may reflect the high transcription activity needed for the development of primordia and ovules because these tissues undergo an intense cell division activity and specific cell differentiation processes.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material and Growth Conditions

Cherry tomato (Lycopersicon esculentum Mill. cv West Virginia 106) plants and tomato cells derived from the pericarp of tomato (cv Sweet 100) fruit (Jean-Luc Montillet, Commissariat à l'Energie Atomique, Cadarache, France) were cultured as described by Joubés et al. (2000b). Cells were grown in modified Murashige and Skoog medium, containing 20% (w/v) Glc, 100 mg mL¹ 2,4-D, and 10 mg mL¹ BAP; pH was adjusted to 5.8. Cells were grown in the dark, at 25°C, and shaken at 150 rpm. They were maintained in exponential growth phase by 5-fold dilution every 6 d in new medium. Cells were harvested by filtration on glass wool under vacuum, frozen quickly in liquid nitrogen, and stored at 80°C.

Tomato roots were prepared as follows. Seeds were soaked in 90% (v/v) ethanol, washed twice in sterile water, and immersed in 2% (v/v) sodium hypochlorite solution containing 0.2% (w/v) SDS for 15 min. Seeds were then washed four times in sterile water. The sterilized seeds were let to germinate between two layers of paper (Whatman, Maidstone, UK) soaked in the mineral solution A (Saglio and Pradet, 1980) in sterile culture boxes (10 × 10 × 10 cm) for 3 d at 25°C and in darkness. After 3 d, the upper layer of Whatman paper was removed and the culture medium was replaced by the same solution supplemented with 0.1 M Suc. The boxes were then transferred to a growth chamber under the culture conditions as described by Joubès et al. (2000b). After 4 d, plantlets were harvested and roots excised under sterile conditions. The roots were subsequently incubated at 25°C in medium A (Saglio and Pradet, 1980) supplemented with an antibiotic-antimycotic mixture (ref. A-7292, 10 µL mL¹; Sigma, Saint Quentin Fallavier, France), under a continuous bubbling of a gas mixture containing 50% O₂ + 50% N₂ gas mixture, and in the presence or absence of 0.1 M Suc. The incubation medium was renewed daily. Excised roots were removed at different times, washed with sterile water, dried on filter paper, and promptly frozen in liquid nitrogen prior to total RNA isolation.

Extraction of Total RNA

Total RNA from whole fruits, fruit tissues, or various organs of tomato plants was extracted as previously described (Joubès et al., 1999). Total RNA from tomato cells was extracted using the RNeasy Plant Mini kit (Qiagen, Courtaboeuf, France). After extraction total RNA from tomato tissues and cells was dissolved in diethyl pyrocarbonate-treated water.

cDNA Library Screening

A cDNA library was constructed with poly(A⁺) mRNA prepared from total RNA extracted from tomato fruits harvested during the cell expansion phase (between 10-15 DPA), as previously described by Joubès et al. (1999). The cDNA library comprised 6.10⁶ recombinant plaques. After plating of the amplified cDNA library, 300 individualized bacteriophage plaques were picked randomly. The cDNA inserts were PCR amplified from the phage suspensions using the pBluescript universal M13-20 and reverse primers, separated after agarose gel electrophoresis, and blotted onto a Hybond N⁺ membrane (Amersham Pharmacia Biotech, Les Ulis, France). To screen for fruit-specific cDNAs, a reverse northern procedure was performed using complexed probes to hybridize duplicate membranes. As a positive probe, ³²P-labeled cDNAs were synthesized as described by Zegzouti et al. (1997) using total RNA from expanding fruits (15 DPA), and as a negative probe ³²P-labeled cDNAs using total RNA from full-expanded leaves. Clones showing a signal only or a stronger signal with the fruit probe were selected and rescued from the Uni-ZAP XR vector using the R408 helper phage following the manufacturer's instructions (Stratagene, La Jolla, CA). The complete nucleotide sequence of the inserts was determined allowing the isolation and identification of the clone Lyces;CDKC;1.

To isolate cDNAs encoding B-type CDKs, specific probes were generated by RT-PCR using total RNA from fruits at 3 DPA. The specific oligonucleotides used to amplify the cDNAs for B1-type and B2-type CDKs were as indicated in Table I. After cloning into pGEM-T vector (Promega, Lyon, France), the identity of the amplified cDNA fragments was confirmed by nucleotide sequencing. The amplified fragments were then used as ³²P-labeled probes to screen 300,000 plaques from the "young fruit" cDNA library (Joubés et al., 1999). After three rounds of screening, the positive clones were isolated, rescued from the Uni-ZAP XR vector and the complete nucleotide sequence of the inserts was determined.

Estimation of Relative Transcript Levels with RT-PCR

To determine specifically the relative transcript levels of each cDNA, RT-PCR assays were performed as previously described (Joubès et al., 1999, 2000b). For excised tomato roots, 5 µg of total RNA was used in the RT reaction, and a 10-fold dilution of the generated cDNAs was used subsequently in the PCR reaction. The specific sets of primers used for the amplification of each cDNA are summarized in Table I.

In Situ Hybridization

Flower buds and tomato fruits of different developmental stages were fixed in 2% (v/v) formalin, 5% (v/v) acetic acid, and 50% (v/v) alcohol for 4 h at room temperature. After fixation the tissues were dehydrated in ethanol series and ethanol was subsequently replaced by Histosol Plus (Life Sciences International, Cergy-Pontoise, France). The tissues were embedded in paraffin according to standard procedures. Sections (8 µm) were fixed in 3-aminopropyl-triathoxysilan-coated slides, deparaffinised in Histosol Plus. The slides were dehydrated in ethanol series for 5 min each time and air dried.

The Lyces;CDKC;1 DNA used for riboprobe synthesis was amplified by PCR using a specific set of primers (Table I) and cloned into the pGEM-T vector (Promega). The sense and antisense digoxygenin-labeled riboprobes were generated by runoff transcription using T7 and SP6 RNA polymerases according to the manufacturer's protocol (Roche Diagnostics, Meylan, France). For hybridization the sections on the pre-treated slides were incubated with probes labeled with digoxigenin-11-rUTP using the digoxigenin nucleic acid labeling kit (Roche Diagnostics) in 100 µL of the following mix: 50% (v/v) deionized formamide, 300 mM NaCl, 1 mM EDTA, 1× Denhardt's, 10% (w/v) dextran sulfate, and 10 mM Tris-HCl, pH 7.5. The slides were covered with coverslips, placed in a humid chamber, and incubated for 16 h at 45°C. After hybridization coverslips were removed and slides washed twice for 5 min in 2× sodium chloride/sodium phosphate/EDTA at room temperature. The slides were then washed twice for 5 min in 0.1× sodium chloride/sodium phosphate/EDTA at 55°C. Immunological detection of the hybridized probes was carried out using digoxigenin nucleic acid detection kit (Roche Diagnostics). The color reaction was stopped by washing the slides in water. Slides were air dried and sections were mounted in Neo-Entellan (Merck, Darmstadt, Germany).

Two-Hybrid Detection of Protein-Protein Interactions

The coding sequences of the different cDNAs were specifically amplified using Pfu DNA polymerase (Promega) and the resulting cDNAs were cloned into pCR4Blunt-TOPO (Invitrogen, Groningen, The Netherlands). They were subcloned into the yeast shuttle plasmids pLexA containing the DBD (Vojtek et al., 1993) and pGAD3S2X containing the AD, a modified version of pGAD1318 (Benichou et al., 1994). The recombinant plasmids were amplified in Escherichia coli, purified, and then introduced in the Saccharomyces cerevisiae strain L40 (Le Douarin et al., 1995) by sequential transformation using lithium acetate (Gietz and Woods, 1994). L40 recombinant cells were selected on plates containing 5 mM 3-amino-1,2,4-triazol lacking Leu and Trp for double transformants or lacking Leu, Trp, and His to reveal the protein-protein interactions. The human RAS and RAF coding sequences (Vojtek et al., 1993) were cloned into pLexA and pGAD3S2X, respectively, and used as a positive control of interaction. As a negative control of interaction, we used the murine laminin 1 sequence (Chang et al., 1996) cloned into pLexA in combination with the different recombinant pGAD3S2X. The quantitative -galactosidase assays were performed according to Urcuqui-Inchima et al. (1999).