Functional characterization of the tomato cyclin-dependent kinase inhibitor SlKRP1 domains involved in protein–protein interactions

Authors

  • Mehdi Nafati,

    1. Institut National de la Recherche Agronomique (INRA), Université de Bordeaux, 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
  • Nathalie Frangne,

    1. Institut National de la Recherche Agronomique (INRA), Université de Bordeaux, 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), Université de Bordeaux, 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
  • Christian Chevalier,

    1. Institut National de la Recherche Agronomique (INRA), Université de Bordeaux, 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
  • Frédéric Gévaudant

    1. Institut National de la Recherche Agronomique (INRA), Université de Bordeaux, 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

Author for correspondence:
Christian Chevalier
Tel: +33 557 122693
Email: chevalie@bordeaux.inra.fr

Summary

  • Cyclin-dependent kinase (CDK) inhibitors (kip-related proteins, KRPs) play a major role in the regulation of plant cell cycle in antagonizing its progression, and are thus regulators of development. The primary sequence of KRPs is characterized by the existence of conserved motifs, for which we have limited information on their functional significance.
  • We performed a functional analysis of various domains present in KRPs from tomato. A series of deletion mutants of SlKRP1 was generated and used in transient expression assays to define the relevance of conserved protein domains in subcellular and subnuclear localizations. Specific interactions of SlKRP1 and its deletion variants with cell cycle proteins were investigated using two-hybrid assays and bimolecular fluorescent complementation.
  • Plant KRPs are distributed into two phylogenetic subgroups according to the presence of conserved motifs. Members of subgroup 1 represented by SlKRP1 share 6 conserved motifs whose function in protein localization and protein–protein interactions could be identified. A new interaction motif was localized in the central part of SlKRP1 that targets SlCDKA1 and SlCYCD3;1 to the nucleus.
  • Our results bring new insights to the functional role of particular domains in KRPs relative to subcellular localization or proteolytic degradation.

Introduction

Progression of the eukaryotic cell cycle relies on remarkably conserved molecular mechanisms. It involves kinase activities from cyclin-dependent kinase (CDK)/cyclin (CYC) complexes that coordinate the transition from one phase of the cell cycle to the next. Several types of CDKs and cyclins exist in plants reflecting the complexity of the plant cell cycle. Twenty-nine CDK-related sequences have been identified in Arabidopsis and classified into seven distinct classes (CDKA to CDKF plus a group of CDK-like proteins, CKL), defined according to phylogenetic, structural and functional similarities with animal and yeast CDKs (Joubès et al., 2000a; Vandepoele et al., 2002; Menges et al., 2005). CDKAs are functional homologues of the yeast p34CDC2/CDC28 protein and thus correspond to the canonical CDK that regulate both the G1–S and G2–M transitions whereas CDKBs are plant-specific CDKs that regulate the G2–M transition (Mironov et al., 1999). The Arabidopsis genome encodes 49 cyclins that have been classified into eight classes and 23 subgroups (CYCA1-3, CYCB1-3, CYCC, CYCD1-7, CYCH, CYCL, CYCP1-4 and CYCL) (Wang et al., 2004). From this complex family of regulators, the A- and B-type cyclins are known as mitotic cyclins regulating the progression through the S-, G2- and early M-phase, while D-type cyclins control the progression through the G1 phase in response to growth factors and nutrients (Inzé & De Veylder, 2006).

CDK/CYC complex activity is under tight post-translational regulations through phosphorylations mediated both positively by CDK/CYC activating kinases (CAK) on residue Thr161 or negatively by the WEE1 kinase on residue Tyr15 (Shimotohno et al., 2006), through the proteolytic degradation of the cyclin moiety (Genschik & Criqui, 2007) or the stable binding of specific CDK inhibitors (CKI) (Wang et al., 2007).

In plants, two families of CKI proteins have been identified to date, and are referred to as the interactor of Cdc2 kinase (ICK)/kip-related protein (KRP) family (Wang et al., 1997; De Veylder et al., 2001), and the SIAMESE protein family (SIM) (Churchman et al., 2006). ICK/KRPs were identified on the basis of a slight sequence homology with one type of animal CKI, namely the p27KIP1 protein, localized at the C-terminal end and corresponding to the motif of interaction with CDK/CYC complexes. Outside of this functional domain, ICK/KRPs show no significant homology with their animal counterparts (Wang et al., 2007). All ICK/KRPs studied so far are able to bind CDKA and D-type cyclins. The interaction with D-type cyclins can occur within complexes or with the cyclin subunit on its own (Nakai et al., 2006). In addition, putative binding to and inhibition of CDKB complexes have also been reported in vitro (Nakai et al., 2006).

Single mutations in ICK/KRP genes or loss-of-function strategies do not produce any phenotype, mostly because of gene redundancy. In contrast, the overexpression of ICK/KRPs leads to plant dwarfism as the progression within the cell cycle is deeply altered (Wang et al., 2000; Jasinski et al., 2002a; Verkest et al., 2005). This phenotype can be partly complemented by the co-overexpression of a D-type cyclin (Jasinski et al., 2002b; Schnittger et al., 2003). Interestingly, the phenotypes reported at the cytological level in ICK/KRP overexpressor plants can be classified into two categories according to the level of the ICK/KRP expression (Verkest et al., 2005). A low level of overexpression leads to an increase in nuclear ploidy, according to the process of endoreduplication, at the expense of cell divisions, but with only slight cell size modifications. Conversely, a high overexpression levels leads to a decrease in both mitotic activity and endoreduplication level, concomitantly with an increase in cell size, probably owing to an organismal control tending to achieve a proper organ size. This relationship between the level of ICK/KRP overexpression and the cell cycle alteration led to the hypothesis that G2/M-specific CDK/CYC complexes are more sensitive to ICK/KRP inhibition than G1/S complexes (Verkest et al., 2005; Pettko-Szandtner et al., 2006).

In animal cells, the regulation of p27KIP1 at the post-translational level has been extensively studied because many types of cancer originate in a deregulation of its inhibitory function (Vervoorts & Luscher, 2008). In plants, the first evidence of a post-translational activation of ICK/KRP has been reported in Medicago, as a calmodulin-like kinase is able to stimulate the inhibitory activity of MtKRP (Pettko-Szandtner et al., 2006). A CDK/CYC complex harbouring CDKB1;1 has been shown to phosphorylate ICK2/KRP2 in Arabidopsis thaliana, thus leading to its proteolytic degradation, in order to prevent exit from the cell cycle towards endoreduplication (Verkest et al., 2005). Substantial evidence has recently emerged arguing for ICK/KRP degradation via the E3-ubiquitin ligase SCF pathway involving the 26S proteasome (Kim et al., 2008; Liu et al., 2008; Ren et al., 2008). However, the motifs responsible for the post-translational regulation of ICK/KRPs are yet to be described, although it has been reported that the first 108 amino acids of ICK1/KRP1 (Zhou et al., 2003b) as well as a C-terminus located putative motif (Jakoby et al., 2006) could be responsible for the protein instability.

In the proteolytic degradation and signal transduction pathways the signalosome subunit, JAB1/CSN5 (for COP9 SigNalosome subunit 5), is responsible for the nucleus-to-cytoplasm translocation of p27KIP1 at the G0/G1 transition in animal cells, to promote this CKI degradation (Tomoda et al., 1999). Again, a deregulation in p27KIP1 cellular compartmentalization is associated with cancers, which emphasizes the importance of this process in proper cell cycle regulation (Vervoorts & Luscher, 2008). Whether this type of regulation for ICK/KRP degradation and/or subcellular addressing occurs in plants is highly probable, as an interaction between the tobacco NtKIS1a protein and the NtCSN5 signalosome subunit has been reported (Le Foll et al., 2008). All ICK/KRPs studied to date show a clear nuclear localization (Wang et al., 2007). However, the pattern of localization inside the nucleus varies among the different ICK/KRPs: in Arabidopsis, ICK2/KRP2, ICK4/KRP6 and ICK5/KRP7 are homogeneously localized within the nucleus, while ICK1/KRP1, ICK6/KRP3, ICK7/KRP4 and ICK3/KRP5 show punctuate subnuclear distributions (Bird et al., 2007).

We have previously reported the isolation and biochemical characterization of two ICK/KRPs in tomato, namely SlKRP1 and SlKRP2, and have shown that SlKRP1 contributes to the control of endoreduplication through the inhibition of mitotic CDKA/CYC complex activity during tomato fruit development (Bisbis et al., 2006). To expand our knowledge on the regulatory role of CDK-specific inhibitors in tomato fruit development, we performed a functional analysis of the conserved protein domains present in tomato ICK/KRPs. We showed that ICK/KRPs can be classified into two phylogenetic groups correlated with the existence of short conserved motifs. In the present study, we describe the role of several of these motifs in the subcellular and subnuclear localization of tomato ICK/KRPs and investigate the extent of their putative interaction with candidate cell cycle proteins.

Materials and Methods

Phylogenetic analysis of plant ICK/KRPs

The sequence alignment of ICK/KRPs was performed using clustalw (http://align.genome.jp/) and manually adjusted to improve the best fit of short conserved motifs. The phylogenetic tree was constructed with the minimum evolution algorithm using mega3 (Kumar et al., 2004). Bootstrap analysis with 1000 replicates was performed to test the significance of the nodes.

Plasmid and construct preparation

cDNAs encoding the different target genes were ordered from the Sol Genomics Network Database (http://www.sgn.cornell.edu/search/clone-order.pl), amplified with attB-flanked specific primers and cloned into pDonr201 plasmid using the BP clonase reaction (Invitrogen).

For each mutant form of SlKRP1, either deleted or mutated inside the original sequence, two independent PCRs were first made to amplify the two different portions of the cDNA. The purified PCR products were then mixed at equal molarities and used as template for a third PCR with attB-flanked specific primers. The PCR conditions were standard using a temperature of 50°C for annealing and combinations of gene-specific primers for the different constructs.

cDNAs inserted in the pDonr201 plasmid were then cloned into different sets of destination vectors : pDest22 and pDest32 (Invitrogen) for two-hybrid assays; pBS-35S-YFP-attR and pBS-35S-attR-YFP (kindly provided by Dr Von Arnim, University of Tennessee, Knoxville, USA) for yellow fluorescent protein (YFP) fusions; p2CGW7 and p2GWC7 (purchased from the Plant Systems Biology Laboratory, VIB, Ghent University, Belgium) for cyan fluorescent protein (CFP) fusions; and nEYFP/pUGW2, cEYFP/pUGW2, nEYP/pUGW0 and cEYFP/pUGW0 (kindly provided by Dr Tsuyoshi Nakagawa, Shimane University, Japan) for bimolecular fluorescent complementation (BiFC) assays.

Transient expression in tomato leaf protoplasts and cultured cells, and in onion epidermal cells

Transient expression analyses were performed using homologous systems, namely leaf protoplasts and cultured cells from tomato, and onion epidermal cells as a heterologous system.

Protoplasts were prepared from 7-d-old leaves from tomato plants and transformed using polyethylene glycol (PEG) according to Di Sansebastiano et al. (1998). To perform biolistic transformations, 1 ml of 4-d-old tomato SWEET 100 cultured cells (Rontein et al., 2002) or onion epidermis were placed on Murashige and Skoog (MS) basal medium in Petri dishes. For subsequent bombardment, 1 μm and 1.6 μm gold particles (60 mg ml−1) were used for SWEET100 cells and onion epidermal cells, respectively. Gold particles were suspended in 50% ethanol and 25 μl of the suspension were mixed with 5 μl (5 μg) plasmid DNA, 25 μl of 2.5 M CaCl2, 10 μl of 0.1 M spermidine. Pelleted gold particles were washed with consecutively 70% and 100% ethanol and resuspended in 30 μl of 100% ethanol, loaded on macrocarriers (8 μl of gold particle suspension per macrocarrier) for transformation with the particle delivery system using a rupture disc of 1100 psi (7.58 MPa) (PDS-1000/He; Bio-Rad). The distance between macrocarrier and the tissue was 6 cm. After gene delivery, cultured cells or epidermal tissue slices were incubated overnight on MS basal medium at room temperature in the dark, before analysis. Each transformation assay was performed in triplicate and each experiment was replicated at least twice.

For 4,6-diamidino-2-phenylindole (DAPI) treatment experiments, the onion epidermis tissues were incubated in 1 μg ml−1 DAPI for 2 min and washed in distilled water before visualization.

Microscopy techniques and imaging

Images for subcellular localization, protein colocalization and BiFC assays were obtained by using the TCS SP2 AOBS confocal scanning microscope from Leica (Gennevilliers, France). CFP was excited using an argon laser at 458 nm and emissions were collected from 465 to 500 nm; YFP was excited by an argon laser at 514 nm and the emission collected from 525 to 600 nm. A 406 nmHg lamp was used for DAPI and emissions were collected from 465 to 500 nm. Images of DAPI colocalization were obtained by using a Nikon Eclipse E800 epifluorescence microscope (Nikon, Champigny sur Marne, France). YFP was visualized using a GFP filter (Ex: 460–500 nm, DM: 505 nm, BA: 510 nm), DAPI was visualized using a DAPI filter (Ex: 340–380 nm, DM: 400 nm, BA: 435–485 nm) and images were recorded using a camera Spot RTke (Diagnostic Instruments Inc., Sterling Heights, MI, USA). All images were processed using Photoshop Software (version CS3, Adobe Systems Incorporated, San José, USA). During subcellular and BiFC assay, the YFP signal was false coloured in glow. During colocalization experiments, the YFP signal was false coloured in yellow and the CFP signal was false coloured in cyan.

Two-hybrid assay

Yeast strains (MAV203) were transformed using the Proquest two-hybrid system for GATEWAY technology according to the manufacturer’s instruction (Invitrogen). After transformation, the Petri dishes were incubated for 3 d at 30°C, and isolated colonies were diluted in 40 μl water. Four microlitre of each diluted colony were then deposited on a Petri dish containing SD-LTH medium supplemented with a variable amount of 3-amino1,2,4triazol (3-AT) (20–80 mM) and reincubated for three further days before evaluation of growth.

Results

Identification of KRPs from tomato

Previously we described the characterization of two KRPs from tomato SlKRP1 and SlKRP2 (Bisbis et al., 2006). We report here the identification of two new tomato KRP sequences by blast searches using the Solanaceae Genomics Network (SGN) Unigene database (http://sgn.cornell.edu/). These two sequences were named SlKRP3 (SGN-U320533) and SlKRP4 (SGN-U318507) and assigned the GenBank database accession numbers FN794406 and FN794407, respectively.

Alignments of ICK/KRP primary sequences isolated from various plant species highlighted the presence of six conserved motifs (see the Supporting Information, Table S1 and Fig. S1). Motif 1 corresponds to the CDK/CYC binding motif at the protein C-terminus, which originally allowed the identification of the first plant CDK inhibitor sequence (Wang et al., 1997). This sequence is now used routinely as a marker for the identification of CDK inhibitors both in plants and animals. Motif 2 is found in all published ICK/KRP sequences or in databases, as well as in inhibitors of the SIM/EL2 family. A putative role as a D-type cyclin binding domain was proposed in SIM/EL2 proteins, (Peres et al., 2007). Similarly, motifs 3–6 are plant specific and of unknown function, but they only occur in a subset of ICK/KRPs (Fig. S1).

Phylogenic classification of KRPs

To investigate the relationships between known ICK/KRPs and the two newly isolated SlKRPs, a phylogenic tree was generated using the full-length sequences of ICK/KRPs from various species found in literature or in databases (Table S1 and Fig. 1). The ICK/KRPs fall into two distant subgroups that will be referred to as subgroup 1 and subgroup 2. Arabidopsis ICK1/KRP1, ICK6/KRP3, ICK7/KRP4 and ICK3/KRP5 lie within subgroup 1, together with tomato SlKRP1, SlKRP2 and SlKRP3. Subgroup 2 hosts ICK2/KRP2, ICK4/KRP6, ICK5/KRP7 from Arabidopsis and SlKRP4. Hence within subgroup 1, SlKRP1 and SlKRP3 share clear sequence homologies with ICK7/KRP4, while SlKRP2 is closer to ICK6/KRP3. The primary sequence of SlKRP4 is highly divergent from that of the three other tomato KRPs because it shares only 19% of identity with SlKRP1, SlKRP2 and SlKRP3, and is only slightly related to ICK1/KRP1 and ICK2/KRP2.

Figure 1.

 Phylogenetic tree of interactor of Cdc2 kinase/kip-related proteins (ICK/KRP). Sequence alignments were created using clustalw, and the tree was derived by minimum evolution analysis. The statistical reliability of the inferred tree topology was assessed by bootstrap test (1000 replicates). The tree was condensed with a cut-off value of 50. Conserved motifs present in each sequence, as referred to in the text, are represented on the right-hand side of the tree. According to the tree branches, ICK/KRPs are clustered in two subgroups, namely subgroup 1 proteins, tinted light grey, and subgroup 2 proteins, tinted dark grey. Accession numbers corresponding to the different ICK/KRP genes are given in the Supporting Information Table S1. At, Arabidopsis thaliana; Cp, Carica papaya; Cr, Chenopodium rubum; Ee, Euphorbia esula; Gm, Glycine max; Mt, Medicago truncatula; Nta, Nicotiana tabacum; Nto, Nicotiana tomentosiformis; Os, Oryza sativa; Ps, Pisum sativum; Pt, Populus trichocarpa; Sl, Solanum lycopersicum; Sb, Sorghum bicolor; Vt, Vitis vinifera; Zm, Zea mays.

Notably, among the different ICK/KRPs originating from monocotyledonous species to be found in databases and in the literature, not a single one belongs to subgroup 2, while more than half of ICK/KRP sequences from dicotyledonous species used to construct the phylogenetic tree fall into subgroup 2 (Fig. 1). While the two subgroups are composed of sequences from evolutionary distant species, branches inside each group tend to encompass proteins from evolutionary close plants. For example, maize ICK/KRPs are on the same subbranches as rice ICK/KRPs, and tobacco ICK/KRPs are close to tomato ICK/KRPs.

The present description of ICK/KRPs into two subgroups reveals a phylogenetic separation according to the presence of conserved motifs in the primary sequences of ICK/KRPs (Figs 1, S1). Indeed, ICK/KRPs belonging to subgroup 1 display the presence of nearly all of the six defined conserved motifs with some exceptions lacking motif 4, 5 or 6. By contrast, ICK/KRPs from subgroup 2 harbour only motifs 1 and 2 and display their own specific motifs (Fig. S1).

Subcellular localization analysis of SlKRP1 and SlKRP4

To address the subcellular localization of tomato ICK/KRPs and other candidate proteins, transient expression assays were performed in tomato leaf protoplasts and cultured cells derived from fruit pericarp of the SWEET100 variety, using various constructs encoding YFP fused in-frame to the N-terminus of the proteins tested under the control of the CaMV 35S promoter. As representative members of each subgroup of ICK/KRPs, the sub-cellular localization of SlKRP1 and SlKRP4 was tested first. The protein constructs YFP-KRP1 and YFP-KRP4 were both localized in the nuclei of tomato leaf protoplasts (Fig. S2), as expected for ICK/KRPs (Bird et al., 2007). Similar results were obtained in cultured cells. However, the use of these two biological materials caused technological difficulties, hampering the observations. Nuclei from leaf protoplasts tend to be overlaid by the surrounding plastids, thus greatly affecting the interpretation of results, and the efficiency in cell transformation using the SWEET100 tomato cultured cells was very low, thus leading to poor reproducibility.

We therefore decided to use onion epidermal cells as a model of choice for plant subcellular localization studies because of the presence of a unique large-sized cell layer. Onion epidermal cells are characterized by the presence of a large vacuole occupying most of the cell volume, squeezing the cytoplasm and nucleus to the outer perimeter of the cell. Nevertheless onion epidermal cells allow a good microscopic visualization of nuclei. After transient transformation of onion epidermal cells by biolistics using the corresponding constructs, YFP-KRP1 and YFP-KRP4 were both localized in the nucleus (Fig. 2a). Alternative fusion of YFP at the C-terminal end of SlKRPs gave similar results (data not shown). As shown in Fig. 2(a), the subnuclear distribution of YFP-KRP1 and YFP-KRP4 was different as YFP-KRP1 was localized reproducibly, according to a punctuate distribution similar to all ICK/KRPs belonging to the first phylogenetic subgroup (Fig. 1; Bird et al., 2007), while YFP-KRP4 was distributed uniformly all over the nucleoplasm (Fig. 2a).

Figure 2.

 Subcellular localization of SlKRP1 and SlKRP4. (a) Yellow fluorescent protein (YFP)-tagged SlKRP1 and SlKRP4 were transiently expressed in onion epidermal cells. Corresponding bright-field images are shown below. Arrows point to subnuclear punctuations. (b) Colocalization between YFP-tagged SlKRP1 or SlKRP4 and 4,6-diamidino-2-phenylindole (DAPI). The arrow points to a typical 2 μm nuclear body. Bar, 20 μm.

To investigate whether the particular localization of YFP-KRP1 is linked to DNA, onion epidermal cells were stained with DAPI following the transient transformation with YFP-KRP1 (Fig. 2b). We found YFP-KRP1 to localize in the vicinity to chromatin DNA, as well as in nuclear bodies of c. 2 μm in diameter physically unlinked to DNA. Conversely, YFP-KRP4 appeared to be uniformly distributed in the nucleus, while DNA had a heterogeneous distribution.

Functional analysis of protein motifs responsible for SlKRP1 subnuclear localization

To determine the primary sequence motifs in SlKRP1 responsible for nuclear localization, we generated different constructs corresponding to a deletion series of SlKRP1 variants (Fig. 3).

Figure 3.

 Schematic view of the series of SlKRP1 deletion mutants used for subcellular localization, bimolecular fluorescent complementation (BiFC) and two-hybrid experiments. Conserved motifs are shown as black boxes and numbered above. The sequence for SlKRP4 used in the SlKRP1Δ53–210:SlKRP4 chimeric construct is tinted grey.

Deleting the C-terminal part of SlKRP1 (construct referred to as SlKRP1Δ165–210) had no effect on the nuclear localization of the protein, while the SlKRP1Δ1–44 variant lacking the first N-terminal 44 amino acids was spread in both the nucleus and cytoplasm (Figs 4a, S3). To confirm whether the N-terminal end of SlKRP1 plays a part in nuclear localization, the SlKRP1Δ54–210 variant was generated. As shown in Fig. 4(a), the 53 aa-long N-terminal end of SlKRP1 was sufficient to locate the protein with a punctuate distribution in the nucleus.

Figure 4.

 Identification of domains and sequences affecting SlKRP1 subnuclear targeting. Onion epidermal cells were transformed with fluorescently tagged SlKRP1 variants and analysed by confocal laser scanning microscopy. (a) Subcellular localization of yellow fluorescent protein (YFP)-tagged SlKRP1. (b) Colocalization of cyan fluorescent protein (CFP)-tagged SlKRP1Δ36–44 and YFP-tagged SlKRP1Δ18–28. (c) YFP-KRP1Δ53–210:KRP4 chimeric protein comprising the N-terminal part of SlKRP1 (KRP1Δ54–210) followed by full-length SlKRP4. Arrows point to subnuclear punctuations. Bar, 20 μm.

The N-terminal part of SlKRP1 harbours the conserved motifs 6, 5 and 3 (Figs 1, S1). Alternative forms of the protein lacking one of these motifs (SlKRP1Δ1–4 lacking motif 6, SlKRP1Δ18–28 lacking motif 5 and SlKRP1Δ36–44 lacking motif 3) did not disturb the nuclear localization (Fig. 4a); neither did the concomitant deletion of motifs 6 and 5 or 6 and 3 (data not shown). However, the deletion of both motifs 5 and 3 (SlKRP1Δ18–28 Δ36–44) induced the reallocation of much of the protein outside of the nucleus, as revealed by the localization along the cytoplasmic stands shown in Fig. 4(a). Such a modification in sub-cellular localization of SlKRP1 was also observed when the deletion of motif 5 was combined with a mutation of the highly conserved Tyrosine36 to Alanine (Y36A) within motif 3 (SlKRP1Δ18–28 Y36A).

We then performed a coexpression assay of the variants of SlKRP1 lacking motif 5 on the one hand and motif 3 on the other using CFP-KRP1Δ36–44 and YFP-KRP1Δ18–28 fusions respectively. Both variants colocalized perfectly (Fig. 4b). We thus conclude that motifs 5 and 3 are together involved in the nuclear localization of SlKRP1 according to the same punctuate distribution within the nucleus.

As SlKRP1 and SlKRP4 display different subnuclear localization patterns, we fused the N-terminal part of SlKRP1 to the full-length sequence of SlKRP4 (YFP-KRP1Δ53–210:KRP4). This chimeric protein displayed a punctuate pattern of localization identical to that of native SlKRP1 (Figs 4c, S3), and thus differed from the uniform nuclear distribution of native SlKRP4 (Fig. 2a). Hence it demonstrated that not only the N-terminal part of SlKRP1 is necessary and sufficient to drive its nuclear localization, but also that it is associated with the punctuate distribution of the protein within the nucleus.

SlKRP1 contributes in housing SlCDKA1 and SlCYCD3;1 in the nucleus

Previously, we showed that SlCDKA;1 and SlCYCD3;1 interact with SlKRP1 and SlKRP2 (Bisbis et al., 2006). To investigate whether SlKRP3 and SlKRP4 share the same protein partnership, targeted yeast two-hybrid experiments were performed using the different cell cycle proteins that have been reported so far in tomato (Joubès et al., 1999; Joubès et al., 2000b, 2001). We showed that SlKRP3 and SlKRP4 interact specifically with SlCDKA;1 and SlCYCD3;1, alongside SlKRP1 and SlKRP2 (Table 1). However, differences in yeast growth as an indicator of the interaction strength were observed among the yeast transformants. Yeasts cotransformed with constructs harbouring SlKRP1 and SlCDKA;1 were able to grow on 40 mM of 3-AT, while growth of yeasts cotransformed with SlKRP2 and SlCDKA1, SlKRP3 and SlCDKA1 or SlKRP4 and SlCDKA1 could only be visualized on plates supplemented with 20 mM of 3-AT. Except for SlKRP3, growth was induced when all KRPs were coexpressed with SlCYCD3;1, despite the presence of high concentrations of 3-AT (40 mM). Yeasts cotransformed with SlKRP3 and CYCD3;1 could only grow under a 20 mM 3-AT pressure. We then used BiFC assays as an in vivo technique to confirm these results. Positive signals of interaction between YFPN-SlKRP1 and SlCDKA;1-YFPC and between YFPN-SlKRP1 and SlCYCD3;1-YFPC were found within the nucleus (Fig. 5a). Interestingly, SlCDKA;1 and SlCYCD3;1 were also found mainly in the nucleus when co-expressed, as to reconstitute a CDKA/CYCD3;1 complex (Fig. 5a), while on their own they both localized in the nucleus and the cytoplasm (Figs 5b, S3).

Table 1.   Analysis of putative interactions between tomato kip-related proteins (KRPs) and candidate cell cycle proteins as determined by yeast two-hybrid assays
 NoneKRP1KRP2KRP3KRP4RAS
  1. Yeasts were cotransformed with the combination of bait and prey vectors as indicated. Transformants were then cultured on SD-LTH for 36 h at 30°C. −, No specific growth; +, weak growth (in the presence of 20 mM of 3-AT (3-amino 1,2,4 triazol)); ++, moderate growth (in the presence of 40 mM of 3-AT); +++, strong growth (in the presence of 80 mM of 3-AT). RAS and RAF proteins were used as a positive control of interaction.

None
CDKA1+++++
CDKB1
CDKB2
CDKC
CycA1;1
CycA2;1
CycA3;1
CycB1;1
CycB1;1
CycD3;1+++++++++
CSN5A++++
RAF+++
Figure 5.

In cellulo interaction between SlKRP1 and SlCDKA;1 or SlCycD3;1. (a) Bimolecular fluorescent complementation (BiFC) analyses in onion epidermal cells of the SlKRP1-SlCDKA;1 (YFPN-KRP1 + CDKA1-YFPC), SlKRP1-SlCycD3;1 (YFPN-KRP1 + CYCD3;1-YFPC) and SlCDKA1-SlCycD3;1 (YFPN-CDKA1 + CYCD3;1-YFPC) interactions. (b) Localization of yellow fluorescent protein (YFP)-tagged SlCDKA1 and SlCycD3;1. Enlarged views (× 6.0) of the nucleus regions are provided below to show the localization of both proteins in the nucleus and cytoplasm. (c) Colocalization experiments of YFP-KRP1 with CFP-CDKA;1 or CFP-CycD3;1. The colocalization of YFP-CDKA;1 with CFP-CycD3;1 was shown as a positive control. Bar, 20 μm.

In Arabidopsis, it was reported that ICK1/KRP1 drives AtCDKA1 into the nucleus (Zhou et al., 2006). In coexpression experiments using SlKRP1 fused to YFP (YFP-KRP1) and SlCDKA;1 fused to CFP (CFP-CDKA;1) or SlCYCD3;1 fused to CFP (CFP-CYCD3;1), we showed that the signals associated to SlCDKA;1 and SlCYCD3;1 were clearly much more abundant within the nucleus than within the cytoplasm (Fig. 5c) compared with their cellular distribution when expressed alone (Fig. 5b). To confirm these results we used the SlKRP1Δ54–210 variant deleted for the C-terminal domain necessary for the interaction with CDK–CYC complexes as a negative control for coexpression experiments (Fig. S4). This allowed us to demonstrate that the interaction of SlKRP1 with either SlCDKA;1 or SlCYCD3;1 does contribute to concentrating both SlCDKA;1 and SlCYCD3;1 inside the nucleus, as the CFP-tagged CycD3;1 or CDKA1 retained their original localization (both in nucleus and cytoplasm), while YFP-KRP1Δ54–210 showed the expected exclusive nuclear localization.

A new domain of interaction between CDKA;1 and CYCD3;1 is functional for their translocation into the nucleus

Yeast two-hybrid assays were performed taking advantage of the deletion series of SlKRP1 variants generated (Fig. 3) to unravel putative functional domains necessary for the interaction of SlKRP1 with different candidate proteins. As expected, the C-terminal part of SlKRP1 (KRP1Δ1–145) is necessary for the interaction with both SlCDKA;1 and SlCYCD3;1 (Table 2) because the evolutionary conserved motif 1 at the C-terminus of KRPs is involved in the binding to CDK/CYC complexes in plants (Wang et al., 1998). We determined that SlKRP1 possesses a second site of interaction with SlCYCD3;1 because the variant KRP1Δ165–210 (lacking the C-terminal CDK/CYC binding motifs 1 and 2) still interacted with SlCYCD3;1 (Table 2). Even if this interaction seemed weaker (yeast growth was impaired over 20 mM of 3-AT selection pressure), it was still significant and the interaction was indeed confirmed using BiFC (Fig. 6a). This interaction domain is probably localized in the central part of the SlKRP1 sequence, between residues 45 and 164, because the interaction between SlCYCD3;1 and KRP1Δ1–44 Δ165–210 was still positive (Table 2) while the KRP1Δ54–210 variant (harbouring only the N-terminal end) could not interact with SlCYCD3;1 according to the yeast two-hybrid assay (Table 2) or BiFC despite repeated experiments (data not shown).

Table 2.   Functional analysis of structural domains in SlKRP1 putatively involved in the interactions with candidate proteins as determined by yeast two-hybrid assays
 NoneKRP1 Δ1–44KRP1 Δ165–210KRP1 Δ54–210KRP1 Δ1–145KRP1 Δ1–44 Δ165–210KRP1 Δ188–210KRP1 Δ169–174Control
  1. Yeasts were cotransformed with the combination of bait and prey vectors as indicated. Transformants were then cultured on SD-LTH for 36 h at 30°C. −, No specific growth; +, weak growth (in the presence of 20 mM of 3-AT (3-amino 1,2,4 triazol)); ++, moderate growth (in the presence of 40 mM of 3-AT); +++, strong growth (in the presence of 80 mM of 3-AT).

None
CDKA1++++
CycD3;1++++++++++
CSN5A+++
Control+++
Figure 6.

 Functional identification of a new motif of interaction between SlKRP1 and SlCDKA;1 and SlCycD3;1. (a) Bimolecular fluorescent complementation (BiFC) analyses in onion epidermal cells of the KRP1Δ165–210-SlCycD3;1 and KRP1Δ165–210-SlCDKA;1 interactions. (b) Colocalization experiments of YFP-KRP1Δ165–210 with CFP-CDKA;1 or CFP-CycD3;1. Bar, 20 μm.

Although the interaction between SlCDKA;1 and SlKRP1Δ165–210 was not observed in yeast two-hybrid assays (Table 2), the truncated SlKRP1Δ165–210 variant (YFPN-SlKRP1Δ165–210) lacking the classical CDK–CYC interaction motif was still able to interact with both SlCDKA;1-YFPC and SlCYCD3;1-YFPC according to the reconstituted YFP signal observed in BiFC experiments (Fig. 6a). When compared with the full-length SlKRP1 (Fig. 5c), SlKRP1Δ165–210 could also contribute to the concentration of both SlCDKA;1 and SlCYCD3;1 inside the nucleus, as demonstrated in coexpression assays (Fig. 6b). Together, these data suggest that a new motif present in the central part of the SlKRP1 sequence is involved in the binding to SlCYCD3;1 which appears sufficient to allow the reconstitution of a CDKA/CYCD3;1 complex to be imported into the nucleus.

Motif 2 in SlKRP1 is involved in the interaction with SlCSN5A

Le Foll et al. (2008) revealed that the COP9 signalosome subunit CSN5A interacts with either the full-length sequence of the tobacco KRP NtKIS1a or its spliced variant NtKIS1b (lacking motif 1). We confirmed that the tomato SlCSN5A protein interacts with all tomato KRPs using yeast two-hybrid assays (Table 1). Using BiFC, the relevance of this interaction was demonstrated in tomato leaf protoplast or onion epidermal cells (at least for SlKRP1 and SlCSN5A) (Figs 7a, S3). When compared with the uniform distribution of YFP-CSN5A in nucleus and cytoplasm (Fig. 7b), the interaction between SlKRP1 and SlCSN5A was exclusively localized within the nucleus (Fig. 7a, left panel). SlCSN5A could also interact with the truncated form of SlKRP1, KRP1Δ1–145, only harbouring the C-terminal motifs 1 and 2 (Table 2), according to a uniform cellular distribution (Fig. 7a, right panel), thus confirming the role of motifs 3 and 5 in nuclear localization. Similar results were obtained in coexpression experiments using YFP-SlKRP1 and CFP-SlCSN5A or YFP-KRP1Δ1–145 and CFP-CSN5A (Fig. 7c), showing the exclusive nuclear or uniform cellular localization, respectively.

Figure 7.

In cellulo interaction between SlKRP1 and SlCSN5A. (a) Bimolecular fluorescent complementation (BiFC) analyses in onion epidermal cells of the SlKRP1-SlCSN5A (YFPN-KRP1 + CSN5A-YFPC) and SlKRP1Δ1–145-SlCSN5A (YFPN-KRP1Δ1–145 + CSN5A-YFPC) interactions. (b) Subcellular localization of yellow fluorescent protein (YFP)-tagged SlCSN5A. An enlarged view (×6.0) of the nucleus region is provided on the right side to show the localization of CSN5A in the nucleus and cytoplasm. (c) Colocalization experiments of YFP-KRP1 with CFP-CSN5A or KRP1Δ1–145 with SlCSN5A. Bar, 20 μm.

In addition, SlCSN5A could also interact with the truncated variant of SlKRP1, KRP1Δ188–210 lacking the last 22 residues encompassing motif 1 (Table 2). Conversely, the KRP1Δ165–210 variant lacking motifs 1 and 2 could no longer interact with SlCSN5A (Table 2). As KRP1Δ1–145 and KRP1Δ188–210 only share the presence of motif 2, we suspected the interaction of SlCSN5A with SlKRP1 occurred via motif 2. To confirm this hypothesis, the interaction was tested between SlCSN5A and the KRP1Δ169–174 variant only lacking the seven amino acids constituting motif 2. As a result, the interaction ceased (Table 2) and despite many different repeated experiments we could never demonstrate this interaction using BiFC.

Discussion

The phylogenic separation of ICK/KRPs in two subgroups correlates with sub-nuclear behaviour

The ICK/KRP primary sequences identified so far share very little similarities (De Clercq & Inze, 2006). As a common feature, they all display a highly conserved C-terminal functional domain similar to the N-terminal domain present in mammalian CIP/KIP CDK inhibitors that is required for the interaction with components of CDK/CYC complexes (Wang et al., 1998; Zhou et al., 2003b, 2006). A careful analysis of ICK/KRP sequence alignments has allowed the identification of other conserved domains that are restricted to short tracks of amino acid residues (De Veylder et al., 2001; Wang et al., 2007) for which very little functional information is currently available (Zhou et al., 2006; Bird et al., 2007).

No clear phylogenetic relationships have emerged from previous analyses of ICK/KRP sequences (Barroco et al., 2006; Pettko-Szandtner et al., 2006), most probably owing to their wide sequence variability. We provide in Fig. 1(a) new phylogenetic tree showing a clear partitioning of ICK/KRPs into two subgroups: subgroup 1 includes ICK/KRPs with more than the two C-terminal conserved motifs 1 and 2, and subgroup 2 gathers those displaying only motifs 1 and 2. Interestingly this partitioning correlates with the subnuclear behaviour of ICK/KRPs. Clearly, all ICK/KRPs are nuclear localized, but their distribution can be either uniform or accords with a punctuate pattern as observed in Arabidopsis (Bird et al., 2007) and for SlKRP4 and SlKRP1, respectively (Fig. 3). The ICK/KRPs uniformly localized in nucleoplasm all belong to subgroup 2 (lacking motifs 3 to 6), and punctuate localized ICK/KRPs are all found within subgroup 1 (Fig. 1). According to Zhou et al. (2006), the motif responsible for the punctuate localization of ICK/KRPs lies within the N-terminal part of the protein and could be motif 3. However, when motif 3 is deleted in ICK6/KRP3, the punctuate localization still occurs, which implies the involvement of other motifs (Bird et al., 2007). We found an apparent redundancy of function between motifs 3 and 5, which could putatively explain why ICK6/KRP3 lacking motif 3 is still localized in the nucleus according to a punctuate distribution. These results suggest the occurrence of an ancient separation between the two main subgroups during evolution, and argue for two separate functional roles for ICK/KRPs. The meaning of this punctuate localization for ICK/KRPs belonging to subgroup 1in relation to their precise function or mode of action in plants remains to be fully understood.

Some of the conserved motifs found in ICK/KRPs from subgroup 1 are probably part of the same functional domains. This hypothesis is supported by the observation of relative constant distances between motifs: motif 2 is always close to motif 1; motifs 3, 5 and 6 are all localized in the N-terminal part of the protein (Fig. S1). In addition, there is no constant distance between motif 4 and the other groups of motifs, suggesting that motifs 1 and 2, motif 4, and motifs 6, 5 and 3 may correspond to three different functional domains. Interestingly, some ICK/KRPs of subgroup 1 only share some of the conserved motifs like ICK1/KRP1 harbouring motif 3 but not motif 5 which suggests a partial functional redundancy of conserved domains among members of subgroup 1.

SlKRP1 interaction with CDKA;1 and CYCD3;1

Using yeast two-hybrid experiments and a BiFC approach, we showed the direct interaction of SlKRP1 with both SlCDKA;1 and SlCYCD3;1 (Fig. 5a). These data confirm what is observed in Arabidopsis (Wang et al., 1998; Jakoby et al., 2006; Zhou et al., 2006). The interactions between ICK/KRP and D-type cyclins were also observed in two-hybrid targeted screens (Wang et al., 1998; Jakoby et al., 2006). Furthermore the in planta interaction was indirectly demonstrated by the simultaneous overexpression of AtCycD3;1 and NiKIS1a from tobacco, as a wild-type leaf phenotype was restored in the double overexpressing plants (Jasinski et al., 2002b; Schnittger et al., 2003; Zhou et al., 2003a). In addition, SlKRP1 is able to import SlCDKA;1 into the nucleus (Fig. 5c), in full agreement with previous data obtained for ICK1/KRP1 and CDKA from Arabidopsis (Jakoby et al., 2006; Zhou et al., 2006). However, we report here for the first time that SlKRP1 also helps in reallocating SlCYCD3;1 into the nucleus (Fig. 5c). The functional meaning of such an interaction is not fully understood, although Jasinski et al. (2002b) proposed that CycD3;1 and ICK/KRP operate as mutual antagonists.

Our results showed that SlKRP1 does not exclusively need the CDK/CYC interaction domain composed of motifs 1 and 2, localized between residues 165 and 210 (SlKRP1Δ165–210), to drive both SlCDKA1 and SlCYCD3;1 into the nucleus (Fig. 6a). A second motif of interaction with SlCYCD3;1, outside of the already known C-terminal domain of SlKRP1, seems to occur within the central part of SlKRP1 between residues 45 and 164 (referred to as KRP1Δ1–44 Δ165–210; Table 2). Such a motif was putatively reported to exist in ICK1/KRP1 as a truncated variant comprising the first 108 amino acids interacted with AtCYCD3;1 in a yeast two-hybrid experiment (Jakoby et al., 2006). Although SlCDKA1 does not interact with KRP1Δ165–210 lacking motifs 1 and 2 in yeast two-hybrid assays, the in cellulo interaction was demonstrated using BiFC (Fig. 6a). Interestingly, SlCDKA1 is preferentially housed in the nucleus when coexpressed with the full-length SlKRP1 or its truncated variant KRP1Δ165–210. To reconcile the absence of a detectable interaction in the yeast two-hybrid assay and the positive BiFC interaction between SlCDKA1 and KRP1Δ165–210, we hypothesize that the SlKRP1-dependent allocation of SlCDKA1 into the nucleus is mediated in plant cells by an intermediary such as SlCYCD3;1, which is also driven into the nucleus by SlKRP1Δ165–210 (compare Fig. 5 with Fig. 6).

Data obtained in animal cells indicate that p27KIP1 may act as a CDK/CYC complex assembly factor when expressed at low level (Labaer et al., 1997; Sherr & Roberts, 1999). As most of the transgenic plants overexpressing ICK/KRPs were generated using strong promoters like the CaMV 35S promoter, the relevance of such a role in plants has not yet been addressed. Interestingly, although ICK3/KRP5 is able to bind to CDKA/CYCD complexes, it does not display any significant inhibitory activity towards its targets in vitro (Nakai et al., 2006), and may argue for such a role as an assembly factor.

Does SlKRP1 interact with CSN5 for subsequent degradation via the SCF pathway?

The COP9 signalosome (CSN) is an evolutionarily conserved multisubunit protease with a central role in the ubiquitin–proteasome pathway, as it regulates the activity of Cullin–Ring Ligase (CRL) families of ubiquitin E3 complexes (Wei et al., 2008; Schweichheimer & Isono, 2010).

In animal cells, the post-translational regulation of p27KIP1 has been extensively studied. At the G0/G1 checkpoint the CSN subunit 5 (CSN5), also called JAB1, interacts with p27KIP1 to trigger its proteolytic degradation via the SCF pathway involving the Cullin1-containing SCF-type CRL (Tomoda et al., 1999; Yang et al., 2002). Here we demonstrate that SlKRP1 does interact with SlCSN5A (Fig. 7), thus confirming the observed interaction between NtKIS1a and NtCSN5 in tobacco (Le Foll et al., 2008). In addition, this interaction took place through motif 2 in SlKRP1 (Table 2). As motif 2 is present in every ICK/KRP sequence analysed so far, it argues for a universal mechanism for the signalosome-mediated degradation of ICK/KRPs via the SCF pathway, which has been largely documented these last years (Kim et al., 2008; Liu et al., 2008; Ren et al., 2008).

Motif 2 in ICK/KRPs is the only conserved motif also present in the second family of plant CDK inhibitors called the SIAMESE (SIM) proteins (Churchman et al., 2006), which are able to interact with both CDKA and D-type Cyclins. Peres et al. (2007) showed that the rice SIM protein OsEL2 deleted for this consensus motif, loses its ability to interact with OsCYCD5;3, suggesting a putative role for this motif as a D-type cyclin-binding domain. It would be interesting to analyse whether SIM proteins are also CSN5 interactors and investigate the interplay between CYCD3 and CSN5 at the level of this particular motif.

The functional relevance of the reallocation of CDK/CYC complexes to the nucleus in the presence of ICK/KRPs could be associated to the proteolytic degradation of KRPs to take place in the nucleus. In early male germ cells from A. thaliana, the degradation of ICK4/AtKRP6 and ICK5/AtKRP7 coincides with the expression within the nucleus of the F-box protein AtFBL17, belonging to the SCF E3 ubiquitin ligase complex (SCF(FBL17)) (Kim et al., 2008). Furthermore, AtFBL17 interacted physically with ICK4/AtKRP6 and ICK5/AtKRP7 using BiFC, and the signal of interaction was exclusively located within the nucleus.

Conclusion: towards a functional map-based model for plant ICK/KRPs

Despite the large amount of available data dealing with the biochemical activity and post-translational regulation of the animal p27KIP1 (Vervoorts & Luscher, 2008), very little is known about the functional and structural characterization of ICK/KRP counterparts. So far only the function associated to the conserved motif 1 in ICK/KRPs can be transposed from animal studies (Wang et al., 1997).

Based on the data described herein, we propose a functional map of ICK/KRPs belonging to subgroup 1 (Fig. 8), using SlKRP1 as a representative member. Motifs 3 and 5 within the N-terminal region allow the specific allocation within the nucleus and contribute to addressing the protein in nuclear bodies of a yet unidentified nature. A second region central to ICK/KRPs, encompassing motif 4, is still functionally uncharacterized, but is able to interact with Cyclin D3;1 and drive it into the nucleus. Finally the C-terminal part of ICK/KRPs encompasses two functionally important domains: motif 1 is the binding domain to CDK/CYC complexes, and motif 2 binds to CSN5A for the putative post-translational regulation of ICK/KRPs via the 26S proteasome degradation pathway. This study offers a frame for future characterization studies of structurally and functionally important domains in plant ICK/KRPs.

Figure 8.

 Schematic representation of conserved functional domains present in SlKRP1. The different conserved domains are indicated as grey-tinted boxes. The positions of the first and last amino acid residues for each domain are indicated above delineating brackets. The interaction of cyclin D3;1 with SlKRP1 central region encompassing motif 4 (amino acids 45 to 165) was experimentally determined while the interaction with cyclin-dependent kinase (CDK) was deduced from colocalization experiments shown in Fig. 6b (indicated as a dotted line). Vertical arrows point to proposed functional roles for the different domains as deduced from the partner interaction or protein localization studies.

Acknowledgements

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 express our deepest thanks to Dr Von Arnim (University of Tennessee, Knoxville, USA) and Dr Tsuyoshi Nakagawa (Shimane University, Japan) for the kind provision of vector constructs used in this study.

Ancillary