Phosphorylation of p27KIP1 homologs KRP6 and 7 by SNF1-related protein kinase–1 links plant energy homeostasis and cell proliferation

Authors

  • Thomas Guérinier,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Laurine Millan,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Pierre Crozet,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Current affiliation:
    1. Instituto Gulbenkian de Ciência, Oeiras, Portugal
    Search for more papers by this author
  • Céline Oury,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
  • François Rey,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
  • Benoit Valot,

    1. Plateforme d'Analyse Protéomique de Paris Sud-Ouest, Unité Mixte de Recherche de Génétique Végétale Institut National de la Recherche Agronomique, Université Paris-Sud, Centre de Recherche de Biochimie et de Génétique Cellulaires, Institut National Agronomique Paris-Grignon, Gif-sur-Yvette, France
    Search for more papers by this author
  • Chantal Mathieu,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
  • Jean Vidal,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
  • Michael Hodges,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
  • Martine Thomas,

    1. Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author
  • Nathalie Glab

    Corresponding author
    • Institut de Biologie des Plantes, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8618, Bâtiment 630, Université Paris-Sud, Saclay Plant Sciences, Orsay Cedex, France
    Search for more papers by this author

For correspondence (e-mail nathalie.glab@u-psud.fr).

Summary

SNF1-related protein kinase–1 (SnRK1), the plant kinase homolog of mammalian AMP-activated protein kinase (AMPK), is a sensor that maintains cellular energy homeostasis via control of anabolism/catabolism balance. AMPK-dependent phosphorylation of p27KIP1 affects cell-cycle progression, autophagy and apoptosis. Here, we show that SnRK1 phosphorylates the Arabidopsis thaliana cyclin-dependent kinase inhibitor p27KIP1 homologs AtKRP6 and AtKRP7, thus extending the role of this kinase to regulation of cell-cycle progression. AtKRP6 and 7 were phosphorylated in vitro by a recombinant activated catalytic subunit of SnRK1 (AtSnRK1α1). Tandem mass spectrometry and site-specific mutagenesis identified Thr152 and Thr151 as the phosphorylated residues on AtKRP6- and AtKRP7, respectively. AtSnRK1 physically interacts with AtKRP6 in the nucleus of transformed BY–2 tobacco protoplasts, but, in contrast to mammals, the AtKRP6 Thr152 phosphorylation state alone did not modify its nuclear localization. Using a heterologous yeast system, consisting of a cdc28 yeast mutant complemented by A. thaliana CDKA;1, cell proliferation was shown to be abolished by AtKRP6WT and by the non-phosphorylatable form AtKRP6T152A, but not by the phosphorylation-mimetic form AtKRP6T152D. Moreover, A. thaliana SnRK1α1/KRP6 double over-expressor plants showed an attenuated AtKRP6-associated phenotype (strongly serrated leaves and inability to undergo callogenesis). Furthermore, this severe phenotype was not observed in AtKRP6T152D over-expressor plants. Overall, these results establish that the energy sensor AtSnRK1 plays a cardinal role in the control of cell proliferation in A. thaliana plants through inhibition of AtKRP6 biological function by phosphorylation.

Introduction

The control of energy balance is of paramount importance for living organisms. During stresses that decrease energy status, eukaryotic cells react by limiting anabolism and enhancing catabolism. This crucial function is performed by central regulators including members of the animal AMP-activated protein kinase (AMPK), yeast sucrose non-fermenting 1 (SNF1) and plant SNF1-related protein kinase–1 (SnRK1) protein kinase families (Hardie, 2007). Similar to its mammalian and yeast orthologs, SnRK1 has a combinatorial, trimeric structure comprising a catalytic (α) and two regulatory (β and γ) subunits encoded by several nuclear genes (Polge and Thomas, 2007). Protein kinase activity is totally dependent on phosphorylation of the catalytic α–subunit on a T–loop threonine (Thr172 of Arabidopsis thaliana SnRK1α1) by the Arabidopsis thaliana SnRK1 Activating Kinase upstream kinases (Shen et al., 2009; Crozet et al., 2010). Moreover, we have shown recently that activated SnRK1α1 applies negative feedback phosphorylation on these upstream kinases (Crozet et al., 2010). The activity of phosphorylated AMPK is also positively modulated by AMP (direct activation in animals, but not in yeast; Davies et al., 1989) and ADP (indirect activation by protection from dephosphorylation in yeast and animals; Mayer et al., 2011; Oakhill et al., 2011). In plants, understanding of SnRK1 metabolic control is still elusive. Trehalose-6–phosphate (T6P), an SnRK1 inhibitor, has been suggested to be a carbon gauge controlling in vivo SnRK1 activity, which increases with decreasing T6P levels (Zhang et al., 2009), while ADP, which increases with stress, may inhibit an SnRK1 phosphatase (Sugden et al., 1999). The cardinal role of the Arabidopsis thaliana SnRK1 in controlling metabolism, energy balance, growth and survival has been clearly established (Baena-González et al., 2007). During stress such as darkness and hypoxia, AtSnRK1 orchestrates the reprogramming of a plethora of catabolism-related (activation) and anabolism-related (inhibition) genes. The massive transcriptional reprogramming observed when AtSnRK1α1 activity is increased in A. thaliana protoplasts is in good agreement with the global function of AtSnRK1 in saving energy and bypassing the stress (Baena-González et al., 2007). One high energy-requiring function is cell division. Indeed, in transgenic potato tubers (Solanum tuberosum) with a decreased T6P level (and therefore a potential increase in SnRK1 activity), transcript profiling revealed down-regulation of some genes involved in cell proliferation and growth while cell cycle-specific inhibitors were up-regulated (Debast et al., 2011). Interestingly, in mammals, AMPK has been found to phosphorylate p27KIP1 on residues Thr170 and Thr197/198 (Liang et al., 2007; Short et al., 2008, 2010). p27KIP1 is a member of the Cip/Kip family of cyclin-dependent kinase inhibitors. These proteins were initially characterized as negative regulators of cyclin-dependent kinase (CDK)/cyclin complexes that drive cell-cycle progression. p27KIP1 interacts with most CDK/cyclin complexes, inactivating CDK2/cyclin E complexes while facilitating assembly and nuclear import of CDK4/6/cyclin D (Coqueret, 2003). p27KIP1, like other Cip/Kip proteins, is an intrinsically disordered protein, adopting specific conformations only after binding to target proteins (Yoon et al., 2012). Such conformational flexibility may explain why this protein has both negative and positive roles in the cell cycle and also non-cell cycle functions that are independent of CDK/cyclin complexes, with involvement in transcription, cell migration and even oncogenesis (Besson et al., 2008). In metabolically stressed mammalian cells, AMPK-dependent phosphorylation of p27KIP1 promotes its stabilization and cytoplasmic sequestration, thereby inducing cell-cycle arrest, autophagy or cell survival (Liang et al., 2007; Short et al., 2008, 2010).

In plants, of the two cyclin-dependent kinase inhibitor families, KRPs (Kip-related proteins) constitute a family of small proteins with seven members in A. thaliana: AtKRP1–7 (Torres Acosta et al., 2011). KRPs have the ability to bind both CDK and cyclin partners, and to inhibit in vitro the kinase activity of CDK/cyclin complexes via a conserved domain that is partially shared with their mammalian counterparts, while the rest of the protein diverges (De Veylder et al., 2001; Nakai et al., 2006). Information on other KRP functions is currently restricted to KRP2, which was recently shown to facilitate accumulation of a D–type cyclin, CYCD2;1, in the nucleus, like its mammalian homolog p27KIP1 (Sanz et al., 2011).

In order to highlight a potential link between energy homeostasis and cell proliferation in plants, we investigated whether SnRK1-mediated phosphorylation of KRP takes place in plants, and, if so, whether it modifies cell-cycle progression in A. thaliana. Here, we demonstrate the ability of AtSnRK1 to interact with and phosphorylate AtKRP6 and 7. Yeast experiments, together with analysis of A. thaliana plants over-expressing both AtKRP6 and AtSnRK1α1, suggest that SnRK1 contributes to the control of cell proliferation via phosphorylation of KRP6.

Results

Arabidopsis thaliana SnRK1α1 phosphorylates KRP6 and KRP7 in vitro

The A. thaliana genome encodes seven cyclin-dependent kinase inhibitor p27KIP1 homologs (AtKRP1–7) identified on the basis of the presence of a short domain that is partially shared with mammalian cyclin-dependent kinase inhibitors and is involved in the interaction with CDK/cyclin complexes. However, this domain is located at the C–terminus of plant proteins but the N–terminus of mammalian proteins (Figure 1a). Furthermore, although the amino acid similarity is restricted to this short domain, all of these proteins belong to the intrinsically disordered protein family. The folding index profile of AtKRP6 and AtKRP7 was determined using the FoldIndex© prediction tool (Prilusky et al., 2005). As shown in Figure 1(b), AtKRP6 and AtKRP7 have a highly negative folding index profile, similar to their animal counterpart p27KIP1. Indeed, using the same prediction tool, while SnRK1 and CDKA;1 were shown to contain only 29 and 33% of protein residues that are predicted to be disordered, this proportion reached 77, 78 and 91% for AtKRP6, AtKRP7 and p27KIP1, respectively.

Figure 1.

Arabidopsis thaliana KRP6 and KRP7 are intrinsically disordered proteins that contain SnRK1 consensus phosphorylation sites. (a) In silico analysis of AtKRP6 (At3g19150) and AtKRP7 (At1g49620) using the consensus target sequence [MLVFI]-X-[RKH]-X2/3-[ST]-X3-[MLVFI] revealed potential SnRK1 phosphorylation sites (ScanProsite; (de Castro et al., 2006)). The conserved Thr152/151 site (AtKRP6/AtKRP7) is located at the beginning of the CDK/cyclin interaction domain. A comparison with mammalian p27KIP1 shows that these sites are different from the known AMPK-dependent target sequences [Thr170 (Short et al., 2008); Thr197 (Short et al., 2010)]. In the rat protein sequence, Thr170 is located close to the NLS (nuclear localization signal) and Thr197 is the C–terminal residue. (b) The fold index profile of analysed proteins was determined using the FoldIndex© prediction tool (Prilusky et al., 2005; http://bip.weizmann.ac.il/fldbin/findex), with the window of analysis set at 21 amino acids.

An in silico analysis using Scan Prosite (de Castro et al., 2006) allowed detection of potential SnRK1 consensus phosphorylation sites in several AtKRP proteins. We focused on AtKRP6 and AtKRP7, the only family members that displayed a common putative phosphorylation site in the CDK/cyclin interaction domain (Thr152/Thr151 at the N–terminal end of the domain). Interestingly, AtKRP6 also contained a second SnRK1 consensus phosphorylation site at Ser91 (Figure 1a).

To investigate whether AtSnRK1 phosphorylates AtKRP6 and AtKRP7 in vitro, glutathione S–transferase (GST)-fused AtKRP6 and AtKRP7 proteins were produced in Escherichia coli and purified (Figure S1). They were used in a reconstituted medium containing recombinant GST–AtSnAK2 (required to phosphorylate and activate AtSnRK1α1; Shen et al., 2009; Crozet et al., 2010) and GST–AtSnRK1α1. In the absence of AtKRP, two phosphorylated proteins were observed corresponding to GST–AtSnRK1α1 (upper band, 83 kDa), which is phosphorylated by AtSnAK2, and AtSnAK2 (lower band, 68 kDa), which is both autophosphorylated and feedback-phosphorylated by activated AtSnRK1α1 (Figure 2b, lane 1) as previously shown (Crozet et al., 2010). Interestingly, when either AtKRP6 or AtKRP7 were present in the reaction mixture, autoradiography showed that both proteins incorporated γ32–phosphate from radiolabelled ATP (Figure 2b, lanes 3 and 5). This phosphorylation was dependent on activated AtSnRK1α1, as AtSnAK2 alone was unable to phosphorylate AtKRP6/7 (Figure 2b, lanes 2 and 4). A tandem mass spectrometry (MS/MS) analysis of in vitro phosphorylated AtKRP6/7 indicated that peptides containing the common consensus SnRK1 target site previously identified in silico (VRKTPT152AAEI in AtKRP6; MEKSPT151QAE in AtKRP7, Figure 1) were phosphorylated (Table 1). The AtKRP7 phosphorylation site was unambiguously identified as Thr151; however, doubt remained regarding AtKRP6 as the peptide contained two closely located threonine residues (Thr150 and Thr152). To resolve this ambiguity, a site-directed mutagenesis strategy was used and the following AtKRP6 mutated forms were produced: AtKRP6T150A, AtKRP6T150D, AtKRP6T152A and AtKRP6T152D (Figure S1). In the reconstituted medium, both the AtKRP6T150A and AtKRP6T150D mutant forms exhibited phosphorylation on Thr152, whereas AtKRP6T152A and AtKRP6T152D did not show Thr150 phosphorylation (Table 1). Thus, clearly, the AtSnRK1-dependent phosphorylation sites on AtKRP6 and AtKRP7 are at Thr152 and Thr151, respectively. To better understand the function of this post-translational modification, it was decided to focus only on AtKRP6.

Table 1. Identification of phosphorylated residues in GST–AtKRP7, GST–AtKRP6WT, GST–AtKRP6T150A, GST–AtKRP6T150D, GST–AtKRP6T152A and GST–AtKRP6T152D after phosphorylation by GST–AtSnRK1α1 in the presence of AtSnAK2
A. thaliana gene numberGene descriptionNumber of peptidesPercentage coverage of proteinPeptideNature/position of modified residue, m/z modification
  1. Bold letters indicate residues 150/152 or 151 for KRP6 and KRP7, respectively.

At1g49620KRP7952SPTQAELDDFFSAAERT3: +79.96633
At3g19150KRP6WT754KTPTAAEIEDLFSELESQDDKKKT4: +79.96633
At3g19150KRP6T150A1354KAPTAAEIEDLFSELESQDDKKKT4: +79.96633
At3g19150KRP6T150D1054KDPTAAEIEDLFSELESQDDKKKT4: +79.96633
At3g19150KRP6T152A954TPAAAEIEDLFSELESQDDKKKNone
At3g19150KRP6T152D854TPDAAEIEDLFSELESQDDKKKNone
Figure 2.

AtSnRK1 phosphorylates AtKRP6 and AtKRP7 in vitro. (a) SDS–PAGE (12% acrylamide) and Coomassie brilliant blue staining of GST–AtKRP6/7 phosphorylation assay proteins. (b) Autoradiography of the phosphorylation assay: 3 h incubation at 30°C with 2 μCi [γ–32P]ATP, 3 μg GST–AtSnRK1α1 purified extract (α1), 2 μg GST–AtSnAK2 purified extract (s2) and 2 μg GST–AtKRP6 (k6) or GST–AtKRP7 (k7) purified extract.

AtKRP6 physically interacts with AtSnRK1α1, and AtKRP6 phosphorylation status does not modify its nuclear localization

In order to investigate this post-translational modification in vivo, we first addressed the question of whether AtSnRK1α1 physically interacts in planta with AtKRP6. After transient expression in BY–2 tobacco protoplasts using bimolecular fluorescence complementation (BiFC), YFP fluorescence was detected in several protoplasts, demonstrating that AtSnRK1α1 and AtKRP6 are interacting partners in vivo. In addition, this interaction only occurred in the nucleus (Figure 3a,b), in accordance with the known subcellular localizations of AtSnRK1α1 and AtKRP6, which are nucleocytoplasmic and nuclear, respectively (Bird et al., 2007; Bitrián et al., 2011). In mammals, p27KIP1 phosphorylation by AMPK on Thr170 induces its cytoplasmic re-localization (Liang et al., 2007; Short et al., 2010). To investigate the effect of AtKRP6 phosphorylation on its subcellular localization, chimeric AtKRP6–GFP proteins (containing AtKRP6WT, AtKRP6T152A or AtKRP6T152D) were transiently expressed in A. thaliana protoplasts. Both AtKRP6T152A and AtKRP6T152D showed nuclear localization like the WT form. Thus, it appears that, under our conditions, Thr152 phosphorylation alone does not lead to re-localization of AtKRP6–GFP to the cytoplasm (Figure S2).

Figure 3.

AtKRP6 and AtSnRK1α1 interact in tobacco cell nuclei. Proteins fused to N- or C-terminal fragments of YFP were transiently expressed in BY–2 tobacco protoplasts. Yellow fluorescence was detected by confocal microscopy in the nuclei of protoplasts expressing AtKRP6–YFPN and AtSnRK1α1–YFPC (a) or AtKRP6–YFPN and YFPC–AtSnRK1α1 (b). No yellow fluorescence was detected in protoplasts expressing AtKRP6–YFPN and YFPC, AtKRP6–YFPN and AtSnRK2.6–YFPC (c) or AtKRP6–YFPN and AtSK2.3–YFPC, the combinations that were used as negative controls.

Phosphorylation-mimetic AtKRP6 no longer inhibits yeast cell division

To assess the biological function of AtKRP6 Thr152 phosphorylation on CDK/cyclin complexes, it was first necessary to investigate whether A. thaliana KRP6 was capable of altering cell division. To this end, a heterologous system in yeast was set up, taking advantage of the ability of A. thaliana CDKA;1 to functionally complement the absence of the unique yeast CDK CDC28. To perform more efficient complementation of the yeast cdc28 mutant, a mutated AtCDKA;1 variant, AtCDKA;1–AF, harbouring Thr14→Ala and Tyr15→Phe mutations to circumvent their potential inhibitory phosphorylation was generated (Porceddu et al., 1999). Mutated yeast cdc28 cells, which are unable to divide at a restrictive temperature (Figure 4b), were complemented by AtCDKA;1–AF, leading to yeast cell multiplication (Figure 4c). The complemented yeast cells were then transformed with either WT or mutated AtKRP6. Interestingly, AtKRP6WT and AtKRP6T152A appeared to block cell division (Figure 4e,g), thus confirming the inhibitory function of AtKRP6 on cell division. In contrast, the phosphorylation-mimetic form AtKRP6T152D did not alter the complementation process (Figure 4i), strongly suggesting that the phosphorylation event abolishes the AtKRP6 inhibitory effects on AtCDKA;1. In order to unravel the molecular mechanism involved, the interactions of the phosphorylation-mimetic form AtKRP6T152D with CDK and cyclin partners were assessed in a two-hybrid system. It had been established previously by a two-hybrid approach that AtKRP6 interacts with A. thaliana CDKA;1 (De Veylder et al., 2001). As shown in Figure 5(a), AtKRP6T152D and AtKRP6WT interact to a similar degree with AtCDKA;1–AF, the AtCDKA;1 variant. We then tested the interaction of AtKRP6WT and AtKRP6T152D with a D–type cyclin, the second partner in the CDK/cyclin complex, which was previously shown to be a true partner of AtKRP6 in planta (Van Leene et al., 2007). Interestingly, the phosphorylation-mimetic form AtKRP6T152D partially lost its ability to interact with its cyclin partner (Figure 5b). Taken together, these results suggest that the inability of the phosphorylation-mimetic form AtKRP6T152D to inhibit CDK kinase activity may be due to loss of its interaction with the cyclin partner of the CDK/cyclin complex.

Figure 4.

Phosphorylation-mimetic form AtKRP6T152D prevents AtKRP6-dependent CDKA/cyclin inhibition in a heterologous yeast complementation system.

The complementation of the S. cerevisiae cdc28-4 mutant with the wild type ScCDC28 was used as a positive control (a). The S. cerevisiae cdc28–4 mutant strain was complemented by the A. thaliana CDKA;1-AF protein allowing yeast cell multiplication (c) cdc28–4. The simultaneous presence of either AtKRP6WT (e) or AtKRP6T152A (g) inhibits AtCDKA;1 complementation but the presence of AtKRP6T152D (i) does not. The S. cerevisiae cdc28–4 mutant strain transformed with the various AtKRP6 forms alone is shown in (d), (f) and (h). A negative control (transformation with empty vectors) is shown in (b). Growth of yeast transformants is shown after 3 days of incubation at a restrictive temperature (35°C) after selection at permissive temperature (25°C) and serial dilution (OD600nm = 1, 0.05 and 0.01, left to right).

Figure 5.

AtKRP6T152D interacts with AtCDKA;1-AF but not with AtCycD3;1 in a two-hybrid system. The yeast PJ69 strain was transformed with AD+BD combinations as indicated. The AD (activating domain) and BD (binding domain) empty vectors pGAD424 and pGBT9 were used to generate the negative controls, and the full-length yeast GAL4 (GAL4) was used as a positive control. After serial dilutions (OD600nm = 2, 0.001 and 0.0001, left to right) and 3 days of incubation at 30°C, growth of double transformants is shown on synthetic medium (a) without leucine and tryptophan (-LW), without leucine, tryptophan and histidine (-LWH), and without leucine, tryptophan and histidine and supplemented with 5 mm 3–amino-triazol (-LWH+3AT5), and (b) without leucine and tryptophan (-LW), without leucine, tryptophan and histidine and supplemented with 10 mm 3–amino-triazol (-LWH+3AT10), and without leucine, tryptophan and histidine and supplemented with 20 mm 3–amino-triazol (-LWH+3AT20).

Double OE-AtSnRK1/AtKRP6WT and phosphorylation-mimetic OE-AtKRP6T152D plants display attenuated OE-AtKRP6-associated phenotypes

Finally, experiments were performed to clarify the function of AtKRP6 phosphorylation in A. thaliana. The phenotype of plants constitutively over-expressing AtKRP6 has been described previously (Zhou et al., 2002; Liu et al., 2008). Most of the KRP family members over-expressed in A. thaliana display several common features, including rosette size reduction, increased cell size, characteristic leaf serration and a reduced capacity of mesophyll cells to undergo callogenesis (Le Foll et al., 2008). Here, we first confirmed that AtKRP6 over-expression (OE-AtKRP6WT) induced the same pleiotropic and gradual developmental alterations (Figures 6 and 7a–g). The severity of the phenotype is in agreement with an increase in AtKRP6 transcript levels and also with higher mRNA levels for two D–type cyclins, AtCycD3.1 and AtCycD4.1 (Figure S3). Several phenotypic classes were determined according to leaf serration, namely wt–L for plants displaying a WT-like phenotype, and LS, MS, SS and HS for lightly, moderately, severely and highly serrated phenotypes respectively (Figure 6a,c,d, line OE-AtKRP6WT). This classification of the plant population was validated by analysing several plants from each class using two others parameters: the capacity of mesophyll cells to undergo callogenesis (Figure 6b and Figure S4) and the size of leaf epidermis cells (Figure 7a–g). The ability to produce calli was quantified by weighing the leaf explants to determine the fresh weight increase factor. This factor gradually decreased (Figure 6b and Figure S4) in accordance with the severity of the serration phenotype (Figure 6a), whereas the cell size dramatically increased (Figure 7 and Figure S5). Indeed, the proportion of cells with an area above 10 000 μm2, which is very low (1.5%) in the wt–L plants, reaches 45% in HS plants (Figure 7h and Figure S5). Based on the results mentioned above, we hypothesized that At SnRK1 phosphorylation of AtKRP6 on Thr152 interferes with AtKRP6 function in planta. To assess this, we generated double OE-(AtSnRK1α1/AtKRP6WT) plants (AtKRP6 transcript levels are given in Figure S3; OE-AtSnRK1α1 plants are described by Jossier et al., 2009). These plants (total n = 253) were classified according to their leaf serration phenotype as described above. We observed that the plants in each class of this population present the same parameters (callogenesis capacity and leaf epidermis cell area) as the plants of the corresponding class in the OE-KRP6WT population (Figure S4 for callogenesis; Figure 7h and Figure S5 for cell size). Interestingly, by comparing the distribution of the plants in the various classes for the two genotypes, we found that AtSnRK1α1 over-expression partially reversed the OE-AtKRP6 phenotype (Figure 6c,d). While the plant population over-expressing AtKRP6WT alone (total n = 343) showed variable severity of the expected developmental alterations with 12.7% classified as wt–L, 44.4% as intermediate phenotype (LS + MS), 21.4% SS and 21.4% HS, the population over-expressing AtKRP6WT together with AtSnRK1α1 displayed a shift towards the WT-like phenotype (21.1% wt–L, 42% LS+MS, 19.8% SS and 17.1% HS) (Figure 6c,d). This shift towards the WT-like phenotype is due to a slight decrease (2–4%) in the proportions of all the other classes. In OE-(AtSnRK1α1/AtKRP6WT) plants, the higher AtSnRK1α1 content allows more AtKRP6 to be phosphorylated, but not all, especially in plants containing the highest amount of AtKRP6. Consequently, OE-AtSnRK1α1 only partially reversed the OE-AtKRP6 effect. To try to get a stronger reversion, plants over-expressing the phosphorylation-mimetic form AtKRP6T152D were produced (total n = 319), and classified as above according to the serration criteria (Figure 6c,d). The same correlation was observed between the classification of plants as wt–L, LS etc. and callogenesis capacity/cell size parameters (Figure 7h and Figures S4 and S5). Remarkably, the proportions of HS and SS plants drastically decreased among the OE-AtKRP6T152D population (0.5% and 8.0%, respectively), while the proportion of wt–L plants reached 40.9% (a 3.2-fold increase compared to the wt–L class of OE-AtKRP6WT) (Figure 6c,d). This result shows a very clear effect on the OE-AtKRP6 phenotype of mimicking the AtKRP6 Thr152 phosphorylation status. Measuring the leaf rosette diameter confirmed this result. Indeed, the population of OE-AtKRP6T152D plants clearly shifted towards the large-rosette phenotype (38.5% of rosettes with a diameter of 40–49 mm; Figure 8). Collectively, these data suggest that AtSnRK1α1 phosphorylates AtKRP6 in planta, and this alters cell proliferation.

Figure 6.

OE-AtKRP6-associated phenotypes are attenuated in both an OE-AtSnRK1α1 genetic background and OE-AtKRP6T152D plants. (a) Classes of leaf serration phenotype for OE-AtKRP6WT plants were established according to severity (a), and named WT-like (wt–L), lightly serrated (LS), moderately serrated (MS), severely serrated (SS) or highly serrated (HS). (b) These classes correlated with the inability to undergo callogenesis (see also Figure S4). (c) Over-expression of AtKRP6WT in an OE-AtSnRK1α1 background (OE-(AtSnRK1α1/AtKRP6WT)) or over-expression of a phosphorylation-mimetic AtKRP6 form (OE-AtKRP6T152D) in a Col–0 background attenuated these phenotypes. Three sets of plants from each genetic background were classified. The percentages presented correspond to the mean proportion of each plant class among OE-AtKRP6WT, OE-AtSnRK1α1/AtKRP6WT and OE-AtKRP6T152D populations. Standard deviations are indicated.

(d) Graphic representation of the percentages shown in (c).

Figure 7.

The severity of serration phenotype correlates with an increase in leaf epidermis cell size.

(a–g) Leaf epidermis from (a) Col–0, (b) OE-AtSnRK1α1 plants, and OE-AtKRP6WT plants of the wt–L (c), LS (d), MS (e), SS (f) and HS (g) classes. (h) The area of 280 cells per plant class was measured for the three genetic backgrounds (OE-AtKRP6WT, OE-(AtSnRK1α1/AtKRP6WT) and OE-AtKRP6T152D). Cell areas were split into five size categories. The percentage of cells that belong to each size category was determined for each class in each genetic background (see Figure S5). Finally, the mean repartition of the total population (from the three genetic backgrounds) among the various phenotype classes is shown, together with those for Col–0 and OE-AtSnRK1α1 (OE-α1). wt–L, LS, MS, SS and HS correspond to phenotype classes WT-like, and lightly, moderately, severely and highly serrated plants.

Figure 8.

Arabidopsis thaliana plants over-expressing the AtKRP6T152D form display overall larger rosettes than OE-AtKRP6WT plants. Rosette diameters (mm) were measured in three sets of 3-week-old plant populations (OE-AtKRP6WT, diamonds, total n = 343; OE-(AtSnRK1α1/AtKRP6WT), squares, total n = 253; OE-AtKRP6T152D, triangles, total n = 319), and the distribution of the plant populations is represented. Numbers around the hexagon refer to the rosette diameter.

Discussion

In mammals, p27KIP1 has been shown recently to be phosphorylated on two residues by the energy sensor kinase AMPK. This results in a cytoplasmic re-localization in the case of Thr170 phosphorylation, which is located close to the nuclear localization sequence (Short et al., 2008), and an increase in stability via 14–3–3 binding for Thr198 phosphorylation on human p27KIP1 (Liang et al., 2007) and Thr197 phosphorylation on rat p27KIP1 (Short et al., 2010), culminating in apoptosis or the initiation of autophagy. Here we show that the recombinant plant p27KIP1 homologs AtKRP6 and AtKRP7 are phosphorylated in vitro on Thr152 and Thr151, respectively, by the recombinant catalytic subunit α1 of the AtSnRK1 complex. Such a post-translational modification of plant KRP proteins may highlight a link between energy sensing and cell proliferation in plants. The existence of such a link was recently suggested by the work of Skylar et al. (2011) while studying the effect of carbohydrate signalling on A. thaliana meristematic cell proliferation.

The expression of AtKRP6WT in mutated yeast cdc28 cells complemented with AtCDKA;1 led to proliferation arrest, confirming its function as a CDK/cyclin inhibitor. In contrast, the phosphorylation-mimetic form AtKRP6T152D did not affect cell proliferation, suggesting a biological function for this post-translational modification. This hypothesis is strengthened by the results obtained with A. thaliana OE lines. As expected, leaf development was highly perturbed (leaf serration) and callogenesis was impaired in OE-AtKRP6 plants, while reversion towards the WT-like phenotype was found in double OE-(AtSnRK1/AtKRP6WT) and was more clearly observed in phosphorylation-mimetic OE-AtKRP6T152D plants. These data suggest that AtSnRK1-dependent phosphorylation of AtKRP6 on Thr152 leads to inhibition of its function. Interestingly, the mechanism involved appears to differ between mammals and plants. While p27KIP1 phosphorylation by AMPK leads to its cytoplasmic sequestration, the phosphorylation-mimetic form AtKRP6T152D remained in the nucleus in our experiments. However, unlike in the mammalian system, the AtKRP6 phosphorylated residue is located in the CDK/cyclin complex binding domain, its phosphorylation may thus block KRP6 inhibition by preventing interaction with the complex. Indeed, two-hybrid experiments show that the phosphorylation-mimetic form AtKRP6T152D partially loses its ability to interact with its cyclin partner. AtKRP6, like its mammalian counterpart p27KIP1, belongs to the intrinsically disordered protein family, members of which display a flexible structure allowing adaptation to different targets. AtKRP6 Thr152 phosphorylation may trigger a similar signalling conduit controlling cell-cycle progression as described for p27KIP1 (Yoon et al., 2012).

SnRK1 involvement in cell division has been suggested previously by the observation of a dramatic decrease in yeast cell size after over-expression of RKIN (rye SnRK1) in a yeast snf1 mutant grown on minimal medium (Dickinson et al., 1999). Thus it is very tempting to hypothesize that the AtSnRK1α1/AtKRP6 signalling module may play an important regulatory role in actively dividing cell areas in plants. For instance, a clear asymmetric pattern of SnRK1 expression was previously reported in the plant apical meristem, with high expression in leaf primordia and low or undetectable expression in meristems (Pien et al., 2001). In this case, AtSnRK1 phosphorylation of AtKRP6 may favour cell division in leaf primordia, where stoichiometric adjustment between the two partners may allow fine control of cell division levels.

However, it has been clearly established that, under energy stresses such as darkness, hypoxia or inhibition of photosynthesis (Baena-González et al., 2007), AtSnRK1α1/2 is activated to reduce anabolism and increase catabolism, thereby resulting in energy saving. Under our conditions, this results in AtSnRK1-dependent AtKRP6 phosphorylation and thus its inability to bind and inhibit the CDK/cyclin complex. Therefore, we are faced with the paradoxical situation that the high energy-consuming process of cell division is favoured by stress.

As stated by Hardie et al. (2012): ‘It is not immediately apparent why a kinase activated by energy stress should be required for passage through mitosis. Perhaps mitosis is accelerated by AMPK in cells undergoing stress so that an orderly cell cycle arrest can occur in the ensuing G1 phase. However, it may be that this is an ancillary function of AMPK that is unrelated to its role as an energy sensor’. Thus, it appears that the present knowledge regarding AMPK function beyond metabolism is still very confusing. In the current quest to integrate underlying and novel findings, our results may be considered in different contexts.

One interpretation of our results is that the basal AtSnRK1 activity is unable to exhaustively phosphorylate the huge amounts of AtKRP6 present in OE-AtKRP6WT plants. This explains the inability to fulfil a complete cell cycle and the observed phenotypes. In this respect, increasing AtSnRK1 activity (OE-AtSnRK1α1/AtKRP6) allows a return to the normal situation, in which cells may properly divide. Thus, in the absence of stress, AtSnRK1 has a basal activity to ensure proper cell division. Thus, the ‘ancillary function’ mentioned by Hardie et al. may still operate in plants.

In this context, the stress sensing/response and the involvement in cell-cycle regulation may be two independent and unrelated functions of AtSnRK1 complexes. Indeed, multiple AtSnRK1 complexes with various αβγ subunit combinations are present in plants and thus may be involved in various functions (Polge and Thomas, 2007; Polge et al., 2008). It is thus possible that only one specific SnRK1 complex recognizes and phosphorylates AtKRP6/7.

Alternatively, cell proliferation may be the consequence of energy stress. In mammals, the SnRK1 homolog AMPK has been implicated in several aspects of cell growth control, including whether to enter either apoptosis or autophagy (Liang et al., 2007; Short et al., 2010; Hardie, 2011). In plants, meristematic areas, where cell division occurs, represent only a small proportion of the total tissues, and levels of energy expenditure are globally low and easily fed by the stress-activated catabolism programme. However, in cases of prolonged stress leading to fuel shortage, cells enter autophagy. Further work is required to investigate such rescue mechanisms in plants.

Experimental Procedures

Plant growth conditions

Seeds were sown directly on soil, and plants were grown in a culture chamber at 70% relative humidity with a light intensity of 180 μmol m−2 sec−1 and a day/night regime of 8 h at 21°C and 16 h at 18°C, respectively.

Plant transformation

Arabidopsis thaliana ecotype Columbia wild-type (Col–0) and OE-AtSnRK1α1 plants (Jossier et al., 2009) were transformed using the floral-dip method (Clough and Bent, 1998) using the pGREENII–0229 vector (Hellens et al., 2000) carrying a cassette containing AtKRP6WT or AtKRP6T152D ORFs under the control of the 35S CaMV promoter and the Nos terminator. After a 4-week selection period on sand supplied with glufosinate (7.5 mg L−1), plantlets were transferred to soil and grown in a culture chamber. For each genetic background, three sets of plants were analysed. For OE-AtKRP6WT, OE-(AtSnRK1α1/AtKRP6WT) and OE-AtKRP6T152D, the mean sizes of each set are n = 114, n = 84 and n = 106, respectively.

Protoplast preparation and transformation

BY–2 protoplasts were obtained and transiently co-transformed as described by Le Foll et al. (2008) using plasmids encoding AtKRP6–YFPN (AtKRP6 ORF cloned into the pUC-SPYNE vector; Walter et al., 2004) and AtSnRK1α1–YFPC (AtSnRK1α1 cloned into the pBiFP4 vector; Desprez et al., 2007) or YFPC–AtSnRK1α1 (AtSnRK1α1 cloned into the pBiFP3 vector; Desprez et al., 2007). Negative controls were obtained by co-transformation of plasmids encoding AtKRP6–YFPN and AtSK2.3–YFPC, AtSnRK2.6–YFPC (the AtSK2.3-p43YC and AtSnRK2.6-p43YC plasmids, respectively, kindly provided by S. Filleur, Université Paris 7, Institut des Sciences du Végétal Gif sur Yvette, France) or YFPC alone (pUC-SPYCE vector; Walter et al., 2004). Protoplasts were incubated for 24 h at 25°C in the dark before confocal microscopic observations. Arabidopsis thaliana protoplasts from wild-type Col–0 rosette leaves were produced and transformed as described by Yoo et al. (2007). Derivatives of pBi/smGFP (Jasinski et al., 2002), carrying AtKRP6WT, AtKRP6T152A or AtKRP6T152D ORFs, were used for transient expression. Protoplasts were incubated for 10 h in the dark before confocal microscopic observations.

In vitro culture

Leaves of 30-day-old plants grown in soil were in vitro cultured as described by Le Foll et al. (2008). Prior to observations, the plates were incubated for 21 days at 23°C at 70% relative humidity and a light intensity of 100 μmol m−2 sec−1 under a 16/8 h day/night regime. Callogenesis capacity was quantified by weighing the leaf explants before transfer to callogenesis medium and after the 21-day incubation period.

Microscopy/cell size observation

For confocal microscopy, observations were performed using a Leica-Microsystems TCS SP2 confocal microscope (Wetzlar, Germany) equipped with an argon laser. The 488 nm laser line was used for GFP and chlorophyll excitation, and the spectral detector was set between 500 and 535 nm and between 659 and 732 nm, respectively. For YFP excitation, the 415 nm laser line was used, and the spectral detector was set between 520 and 570 nm. For cell size observations, leaf tissues were fixed in an ethanol/acetic acid solution (3:1 v/v) and rinsed in 70% ethanol. Fixed leaf tissues were cleared in a hydrate chloral/glycerol/H2O solution (8:2:1 w/v/v) before observations with a AZ100 macroscope (Nikon, Tokyo, Japan). Cell areas were determined using ImageJ 1.62 free software (NIH, Bethesda, MD, USA) and an Intuos3 pen tablet (Wacom, Vancouver, WA, USA).

Transcript level analysis

First-strand cDNA was synthesized using 2 μg leaf total RNA extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and ImProm–II reverse transcriptase (Promega, Fichtburg, WI, USA), and used as template in PCR reactions with primers 5′-GAGGATCCATGAGCGAGAGAAAGCGAGAGC-3′ and 5′-GAGTCGACAAGTCGATCCCACTTGTAGCGACC-3′ for AtKRP6 (1.5 μl first-strand cDNA, 25 cycles with a 30 sec hybridization step at 49°C and 40 sec of elongation); 5′-GTAGAGAAGAACAGAGCAATTCGTTTC-3′ and 5′-CTGTAAACCGATGCGGTCCACTGGTAG-3′ for AtCycD3.1 (At4g34160) (1.5 μl first-strand cDNA, six cycles with a 30 sec hybridization step at 68°C, reducing by 1°C for each cycle, and 1 min elongation, and 20 cycles with a 30 sec hybridization step at 62°C and 1 min elongation); 5′-ATGGCAGAGGAGAATCTAGAACTGAG-3′ and 5′-GTGAACATAAGTCTGAGCCATCACTC-3′ for AtCycD4.1 (At5g65420) (1.5 μl first-strand cDNA, 25 cycles with a 30 sec hybridization step at 61°C and 1 min elongation); 5′-GCGAAGAAGATCAAGCTGTC-3′ and 5′-GGTTAGTTGAACCCTCCTTG-3′ for AtTCTP (translationally controlled tumour protein, At3g16640) used as a constitutive control (1.5 μl first-strand cDNA, 25 cycles with a 30 sec hybridization step at 60°C and 20 sec of elongation).

Expression and purification of recombinant proteins

The fusion constructs used to produce N–terminally GST-tagged A. thaliana SnAK2 and SnRK1α1 proteins have been described previously (Crozet et al., 2010). The coding sequences of KRP6 and KRP7 were reverse-transcribed from A. thaliana total RNA, and the resulting PCR-amplified cDNAs were inserted into the pDEST15 expression vector (Invitrogen, Carlsbad, CA, USA) as described for AtSnAK2 and AtSnRK1α1 (Crozet et al., 2010). KRP6 variant forms were generated by PCR site-directed mutagenesis as described by Reikofski and Tao (1992), and inserted into pDEST15. GST–AtSnAK2 and GST–AtSnRK1α1 recombinant proteins were produced in E. coli and purified by glutathione–Sepharose 4B affinity gel (Sigma, Saint-Louis, MO, USA) chromatography as described by Crozet et al. (2010) Slight modifications were applied for GST–AtKRP6/7 expression. Escherichia coli were cultured at 37°C and induced with 500 μm isopropyl thio-β–d–galactoside at OD600 nm = 0.9 for 1.5 h. Production of recombinant proteins was validated by Western blot analyses using 10 μl ml−1 anti-GST antibodies (Calbiochem, Merck-Millipore, Darmstadt, Germany) and 0.5 μl ml−1 horseradish peroxidase-coupled anti-mouse secondary antibodies (Pierce, Thermo Fisher Scientific, Waltham, MA, USA). Enhanced chemoluminescence (PerkinElmer, Waltham, MA, USA) was performed to detect GST-containing proteins.

Kinase activity assays

Kinase activity was measured in the presence of 2 μCi [γ–32P]ATP (500 μCi mmol−1) in a kinase activity buffer containing 0.1 m HEPES/NaOH, pH 7.3, 5 mm dithiothreitol, 10 mm MgCl2, 0.5 mm EGTA, 20 μm (for radioactive assays) or 1 mm (for mass spectrometry analyses) ATP, 3 μg recombinant GST–AtSnRK1α1, 2 μg recombinant GST–AtSnAK2 and 2 μg of either GST–AtKRP6WT, GST–AtKRP6T150A, GST–AtKRP6T150D, GST–AtKRP6T152A or GST–AtKRP6T152D purified extract, 1‰ v/v anti-protease mixture (P9599; Sigma); 1‰ v/v of each anti-phosphatase mixture (P2850 and P5726; Sigma), at 30°C in a total volume of 30 μl. Following a 3 h incubation at 30°C, the proteins were dissociated in Laemmli buffer and subjected to SDS–PAGE (12% acrylamide) as described by Laemmli (1970). Radioactive bands were revealed using a phosphor screen and Personal Molecular Imager FX system (Bio–Rad, Hercules, CA, USA).

Identification of phosphorylated residues by mass spectrometry

In-gel digestion was performed using the ProGest system (Genomic Solutions Ltd., Huntingdon, UK) according to a standard trypsin protocol and as described by Martin et al. (2006).

HPLC was performed on a NanoLC-Ultra system (Eksigent, Dublin, CA, USA) as described by Zhang et al. (2011) with slight modifications. The pre-column cartridge was a PepMap 100 C18 column (5 μm particle size, 300 μm internal diameter, 50 mm length; Dionex, Thermo Fisher Scientific, Waltham, MA, USA) and the separating column was a C18 Biosphere column (3 μm size particle, 75 μm internal diameter, 150 mm length; Nanoseparations, Nieuwkoop, The Netherlands).

Eluted peptides were analysed online using a QExactive mass spectrometer (Thermo Electron, Thermo Fisher Scientific, Waltham, MA, USA) with a nanoelectrospray interface. Ionization (1.8 kV ionization potential) was performed using stainless steel emitters (30 μm internal diameter; Thermo Electron, Thermo Fisher Scientific, Waltham, MA, USA). Peptide ions were analysed using Xcalibur 2.1 software (Thermo Fisher Scientific, Waltham, MA, USA) with the following data-dependent acquisition steps: (1) full MS scan (mass-to-charge ratio (m/z) 400–1400, resolution 70 000) and (2) MS/MS (normalized collision energy 30%, resolution 17 500). Step 2 was repeated for the five major ions detected in step 1. Dynamic exclusion was set to 40 sec.

A database search was performed using XTandem version 2011.12.01.1 (http://www.thegpm.org/TANDEM/). Enzymatic cleavage was set as a trypsin digestion with one possible mis-cleavage. Cys carboxyamidomethylation and Met oxidation were set to static and possible modifications, respectively. The precursor mass was 100 ppm and the fragment mass tolerance was 0.02 Da. A refinement search was added with similar parameters except that semi-trypsic peptide, possible N–terminal protein acetylation and phosphorylation of serine threonine or tyrosine were searched. The UniprotKB database (ftp://ftp.ebi.ac.uk/pub/databases/integr8/last_release/) restricted to yeast, the modified sequence of KRP proteins, and a contaminant database (trypsin, keratins, etc.) were used. Only peptides with an E–value less than 0.1 were reported. Identified proteins were filtered and grouped using X!TandemPipeline (http://pappso.inra.fr/bioinfo/xtandempipeline/) as described by Zhang et al. (2011). In cases of identification of mutated AtKRP6 and phosphorylation sites, similarity between the experimental and the theoretical MS/MS spectra was visually checked. To take redundancy into account, proteins with at least one peptide in common were grouped. This allowed grouping of proteins of similar function. Within each group, proteins with at least one specific peptide relative to other members of the group were reported as sub-groups.

Yeast functional complementation and yeast two-hybrid assay

AtKRP6WT, AtKRP6T152A and AtKRP6T152D ORFs were cloned into pNSG1 (vector made in our lab), a derivative of pGBT9 (Clontech, Takara, Tokyo, Japan) lacking the binding domain encoding sequence. The ORF of AtCDKA;1-AF, a dominant positive allele of AtCDKA;1 carrying two mutations (Thr14→Ala and Tyr15→Phe), was cloned into pNSG2 (vector made in our lab), a derivative of pGAD424 (Clontech, Takara, Tokyo, Japan) lacking the activation domain encoding sequence. The Saccharomyces cerevisiae CDC28 ORF cloned into YEp13 (New England Biolabs, Beverly, MA, USA) was used as a positive control. Plasmids were co-transformed into the S. cerevisiae cdc28–4 thermosensitive mutant strain Ely227 (Barral et al., 1995). Yeast two-hybrid analysis was performed as described by Jasinski et al. (2002). AtKRP6WT and AtKRP6T152D ORFs were cloned into pGBT9 (binding domain (BD) vector, Clontech, Takara, Tokyo, Japan) and AtCDKA;1-AF and AtCycD3;1 ORFs were cloned into pGAD424 (activation domain (AD) vector, Clontech, Takara, Tokyo, Japan). Various AD+BD combinations were co-transformed into the S. cerevisiae PJ69 strain (James et al., 1996).

Acknowledgements

We thank Elena Baena-Gonzales (Instituto Gulbenkian de Ciência, Lisboa, Portugal) for her technical expertise with A. thaliana protoplast experiments, Mélanie Héry, Alexis Bioy and Thibaut Di Méo for their contribution to construct preparation and the production of recombinant GST–AtSnRK1α1, and Sophie Filleur (Université Paris 7, Institut des Sciences du Végétal, Gif sur Yvette, France) for the gift of the two BiFC plasmids: AtSK2.3-p43YC and AtSnRK2.6-p43YC. This work was supported by the Centre National de la Recherche Scientifique and the Université Paris-Sud. T.G., L.M. and P.C. were supported by fellowships from the French Ministère de l'Enseignement Supérieur et de la Recherche. This work has benefited from the facilities and expertise of IMAGIF (Centre de Recherche de Gif; www.imagif.cnrs.fr).

Ancillary