| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
First published online March 4, 2005; 10.1105/tpc.104.030486 © 2005 American Society of Plant Biologists
Novel Functions of Plant Cyclin-Dependent Kinase Inhibitors, ICK1/KRP1, Can Act Non-Cell-Autonomously and Inhibit Entry into Mitosis
|
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
). Common cell cycle variants in both animals and plants are endocycles, in which cells replicate their DNA without undergoing a subsequent mitosis leading to polyploid cells (Edgar and Orr-Weaver, 2001). Endoreplication has been implicated in cell differentiation and cell growth, for instance, in the development of Drosophila melanogaster nurse cells, Medicago truncatula nodule cells, or Arabidopsis thaliana leaf hairs (trichomes) (Kondorosi et al., 2000; Edgar and Orr-Weaver, 2001; Schnittger and Hulskamp, 2002; Sugimoto-Shirasu and Roberts, 2003; Kondorosi and Kondorosi, 2004). In addition to cell type–specific settings, cell number and cell size are also coordinated on a supracellular level, maintaining tissue and organ growth in a highly predictable manner as well as taking environmental conditions into account (Day and Lawrence, 2000; Doonan, 2000; Potter and Xu, 2001).
The central convergence point of eukaryotic cell cycle control, where intrinsic and extrinsic cues are integrated, is a group of Ser/Thr kinases, CYCLIN DEPENDENT KINASEs (CDKs). Activated CDKs phosphorylate a plethora of proteins, resulting in the entry into a new round of DNA replication and the entry into mitosis, respectively. Recently, many putative CDK substrates have been identified (Ubersax et al., 2003). However, little is known about how CDK activity is modified for different cell cycle modes, in particular in endocycles.
One way of controlling CDK activity is mediated by CDK inhibitors (CKIs) that stochiometrically bind to CDKs and inhibit their kinase activity. In animals, two classes of inhibitors have been identified, the Inhibitor of CDK4 (INK4) class and the CDK Interacting Protein/CDK Inhibitor Protein (CIP/KIP) family.
The INK4 class comprises p15, p16, p18, and p19, which inhibit CDK4 but can also bind to CDK6. Inhibitors of the CIP/KIP family block cyclin D–, E–, and A–dependent kinases, but predominantly inhibit CDK2 activity (Pavletich, 1999; Sherr and Roberts, 1999). Besides a negative role in CDK regulation, CKIs have also been found to help assemble and stabilize a CDK4-cyclin D complex (Sherr and Roberts, 1999). It is not clear, however, whether these CDK-cyclin D-CKI complexes are active (Olashaw et al., 2004).
Putative CKIs have also been found in plants (Wang et al., 1998; De Veylder et al., 2001; Jasinski et al., 2002). In Arabidopsis, seven proteins were identified, which display homologies to the animal p27Kip1 protein and thus were named INHIBITORs/INTERACTORs OF CDK (ICKs) or KIP RELATED PROTEINS (KRPs) (Wang et al., 1998; De Veylder et al., 2001). The homology to p27Kip1, however, is restricted to 
30 amino acids in the C terminus, and information about plant CKIs is still very limited.
In yeast two-hybrid interaction assays, ICK1/KRP1 could bind to CDKA;1 and CYCLIN D3;1, and it has been demonstrated that ICK1/KRP1 can inhibit the histone phosphorylation activity of CDKA;1 in vitro (Wang et al., 1997, 1998). In several misexpression studies, it has been found that ICK/KRPs can block endoreplication and reduce cell numbers, leading to dwarfed plants in extreme cases (Wang et al., 2000; De Veylder et al., 2001; Zhou et al., 2002; Schnittger et al., 2003). All these results are consistent with the presumed function of ICK/KRPs as inhibitors of CDKs at the G1-S transition point.
Here, we show that ICK1/KRP1 can also function outside of a G1-phase. After misexpression of ICK1/KRP1, we observed that cells skipped mitosis and underwent endoreplication. Our data suggest that ICK1/KRP1 acts in a concentration-dependent manner and only blocks G1-S transition at high concentrations, whereas at low concentrations G2-M transition is blocked. In contrast with animal CKIs, we demonstrate that Arabidopsis ICK1/KRP1 can act in a non-cell-autonomous manner with possible implications for tissue organization and organ growth. In addition, we found that cells induced by ICK1/KRP1 to endoreplicate prematurely can still adopt a new cell fate in accordance with a developmental program. Moreover, we found that endoreplicated cells even can reenter mitosis, demonstrating the high degree of flexibility of plants cells during development.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
, 2002b, 2003). Trichomes are single-celled leaf hairs, which are initiated with a controlled distance to each other in the basal part of young and growing leaves. Archetypical for many differentiating cells, incipient trichomes exit a mitotic program and switch to an endoreplication mode. Concurrent with outgrowth and initiation of branches, trichomes undergo approximately four rounds of endoreplication, leading to a mature trichome with three to four branches and a DNA content of 32C (Marks, 1997; Hulskamp et al., 1999).
Specific misexpression of ICK1/KRP1 in trichomes using the GLABRA2 promoter (ProGL2) resulted in trichomes with reduced cell size, a decreased number of branches, and reduced endoreplication levels, and finally these trichomes underwent cell death (Figures 1A and 1B) (Schnittger et al., 2003). This effect was enhanced in plants expressing an N-terminally truncated protein version (amino acids 2 to 108 were deleted), named ICK1/KRP1109, providing in planta evidence for a previously identified negative regulatory domain in the N terminus of ICK1/KRP1 (Wang et al., 1998; Schnittger et al., 2003).
|
1200 µm2, on comparable leaves of ProGL2:ICK1/KRP1109 plants, trichome-neighboring cells encompassed a >10 times larger total surface area of 13500 µm2 (Table 1). Examining transgenic plants carrying the full-length ICK1/KRP1 misexpression construct, which showed a weaker trichome phenotype, we observed an enlargement of the trichome-neighboring cells to an average of 4800 µm2 (Table 1).
|
|
). Hence, we raised the hypothesis that because of a compromised and eventually dead trichome as a result of ICK1/KRP1 misexpression, the lateral inhibition is released and the trichome-neighboring cells start to develop into trichomes. To address this question, we used a very sensitive trichome-fate reporter construct, in which a fusion of a nuclear localization signal (nls) to the GREEN FLUORESCENT PROTEIN (GFP) and the ß-GLUCURONIDASE (GUS) is under the control of the GL2 promoter (ProGL2:nls:GFP:GUS). Analysis of this marker line, however, revealed no expression in cells surrounding the ICK1/KRP1-misexpressing trichomes, indicating that an initiated trichome developmental program is not responsible for the observed phenotype (Figures 3A and 3I).
|
; Kirik et al., 2001). In addition, similar to trichomes on ProGL2:ICK1/KRP1-expressing plants, cpr5 mutant trichomes have been reported to die. However, neither the trichome-neighboring cells in the gl3 nor in the cpr5 mutant displayed any significant difference to wild-type trichome-neighboring cells with respect to cell size and DNA content (Figures 1E, 1F, and 2, Table 1). Taken together, these data suggest that trichome-neighboring cell enlargement and increase in DNA content is due to ICK1/KRP1 misexpression in trichomes and is not a general feature of an altered trichome development.
Intercellular Localization of ICK1/KRP1
Based on the conclusion that the phenotype of trichome-neighboring cells is specific for ICK1/KRP1 misexpression, we reasoned two different scenarios by which ICK1/KRP1 could influence the cells surrounding a trichome. First, ICK1/KRP1 might act indirectly, and its expression in trichomes would induce a non-cell-autonomous response. Alternatively, given that plant cells are symplastically connected by plasmodesmata (Ding et al., 2003; Oparka, 2004), ICK1/KRP1 itself might move into the neighboring cells.
To test the localization and mobility of ICK1/KRP1, the yellow fluorescent protein (YFP) was fused to ICK1/KRP1 and ICK1/KRP1109, and misexpression lines using the GL2 promoter were generated. Homozygous lines were created, and based on mRNA expression strength, comparable lines were chosen as reference lines for further investigations (see Supplemental Figure 1A online). All data provided in the following was obtained from the same reference line. As a control, transgenic plants expressing a cell-autonomous version of the GFP with a localization signal for the endoplasmatic reticulum (ProGL2:GFP5ER) and plants expressing an untagged YFP protein (ProGL2:YFP) were created (Siemering et al., 1996; Haseloff et al., 1997; Crawford and Zambryski, 2000).
Plants expressing the fusion proteins were first analyzed with respect to their trichome phenotype. Plants carrying an N-terminal YFP fusion to ICK1/KRP1 (ProGL2:YFP:ICK1/KRP1) displayed smaller and under-branched trichomes, which eventually died, resembling the ICK1/KRP1-misexpression phenotype (Table 2; data not shown). The expression of ICK1/KRP1 with a C-terminal fusion (ProGL2:ICK1/KRP1:YFP) did not result in a phenotype, and transgenic plants were not further analyzed. For ICK1/KRP1109, plants misexpressing both N- and C-terminal fusion proteins with YFP resembled the phenotype of ProGL2:ICK1/KRP1109 plants (Table 2). Similarly to the expression of the unfused ICK1/KRP1, we recognized that expression of fusion proteins containing the N-terminally truncated ICK1/KRP1109 resulted in a stronger trichome phenotype than the expression of fusion protein with the ICK1/KRP1 full-length version (Table 2; data not shown). Thus, although fusions in the C terminus to the full-length ICK1/KRP1 seemed to interfere with protein action, we conclude that a fusion with YFP in the other three constructs did not result in an altered ICK1/KRP1 protein activity as judged by their trichome phenotypes.
|
; Zhou et al., 2003). The limitation in detection of the full-length CKI argues that this domain might regulate the stability of ICK1/KRP1 protein. Consistently, Zhou et al. (2003) recently reported that an N-terminally truncated version was present in much higher abundance than the full-length inhibitor. Next, the localization of the fusion proteins was analyzed by confocal laser scanning microscopy. As controls, we first analyzed the expression of two GL2 reporter lines in a wild-type background and in plants expressing ProGL2:ICK1/KRP1109. In the wild type, both GFP5ER and a nls:GFP:GUS fusion protein expressed from the GL2 promoter were only detected in trichomes and trichome precursor cells (Figures 3A and 3E; data not shown). In the F1 generation of the cross of the ProGL2:nls:GFP:GUS reporter line with the reference line expressing ProGL2:ICK1/KRP1109, the GFP signal was still restricted to trichomes and trichome precursor cells, indicating that trichome-specific expression of ICK1/KRP1109 did not alter the expression domain of the GL2 promoter (Figure 3I).
In contrast with the trichome-specific localization of the two GL2 promoter reporter lines, we could detect the ICK1/KRP1 fusion proteins also in cells around trichomes. In young leaves, ICK1/KRP1 fusion protein could be detected in many epidermal cells (Figures 3B and 3C). In older leaves, the full-length ICK1/KRP1 fused to YFP was predominantly found in one to two concentric rings around a trichome, and the truncated version ICK1/KRP1109 was detectable in three to four rings with decreasing intensity (Figures 3F and 3G). Also, we could detect a weak YFP signal in the nuclei of mesophyll cells, demonstrating that movement of ICK1/KRP1 fused to YFP is not restricted to epidermal cells but reflects rather a general feature of ICK1/KRP1-YFP fusion proteins (Figures 3K, arrowhead). Based on these localization patterns, it is conceivable that the unfused ICK1/KRP1 when expressed in trichomes will also enter the neighboring cells.
A morphological analysis of the trichome-neighboring cells revealed, however, that only plants expressing the N-terminally truncated ICK1/KRP1109 fused to YFP displayed a significant increase in trichome-neighboring cell size and DNA content with 2500 µm2 and 9.4C (Table 1, Figure 2; data not shown). Thus, in contrast with trichomes, the alterations of the trichome-neighboring cells were correlated with the protein size of the misexpressed ICK1/KRP1 protein (i.e., smaller proteins caused a more severe phenotype: ICK1/KRP1109 [10 kD] > ICK1/KRP1 [22 kD] > ICK1/KRP1109:YFP [37 kD] > YFP:ICK1/KRP1 [49 kD]).
To address the dynamics of the movement of ICK1/KRP1 and to test whether larger fusion proteins were less abundant in trichome-neighboring cells than smaller ICK1/KRP1 versions, we compared the fluorescence intensities of ICK1/KRP1-YFP fusions with that of free YFP. As previously reported, the YFP-related GFP is able to diffuse up to 16 cells wide in microprojectile bombardment experiments in Arabidopsis (Itaya et al., 2000). Consistently, in the generated transgenic plants expressing YFP without any localization signals from the GL2 promoter (ProGL2:YFP), YFP could be detected in trichomes and in neighboring cells (see Supplemental Figures 2A and 2B online). Determination of the fluorescence intensity of trichome-neighboring cell nuclei in comparison with trichome nuclei revealed for ICK1/KRP1109:YFP (37 kD) a similar ratio of 0.5 as for YFP (27 kD), whereas for the larger ICK1/KRP1 fusion (49 kD), a lower ratio of 0.2 was obtained (see Supplemental Figure 2C online). This is consistent with a reduced movement and, therefore, a lower concentration of increasingly larger fusion proteins in trichome-neighboring cells.
However, we could not exclude that the different ICK1/KRP1 protein versions have different molecular properties in trichome-neighboring cells versus trichomes (e.g., protein stability and/or nuclear import rate), which could influence the ratio of fluorescence intensities independent of protein size. To test more directly for a protein size–dependent movement, we created transgenic plants expressing another ICK/KRP fusion protein, in which the GUS protein was combined with YFP:ICK1/KRP1109; the size of this fusion protein is 105 kD. Expression of GUS:YFP:ICK1/KRP1109 from the GL2 promoter caused a significant reduction in trichome branch number similarly to the other ICK1/KRP1 protein versions, demonstrating the functionality of this fusion protein (Table 2). Confocal laser scanning microscopy revealed that GUS:YFP:ICK1/KRP1109 was restricted to trichomes (Figures 3D, 3H, and 3L), and no increase in trichome-neighboring cell size nor DNA content was observed (Figure 2,Table 1).
Taken together, we conclude that ICK1/KRP1 can act non-cell-autonomously and that the phenotype of the trichome-neighboring cells in the ICK1/KRP1 misexpression lines is due to a direct action of the CKI in the neighboring cells.
Intracellular Localization of ICK1/KRP1
In animals, the intracellular localization of the CKI p27Kip1 is strictly regulated and appears to be inherently connected with protein abundance and activity (Sherr and Roberts, 1999; Slingerland and Pagano, 2000). The general notion is that p27Kip1 exerts its inhibitory function in the nucleus and becomes degraded in the cytoplasm (Tomoda et al., 1999; Connor et al., 2003). The regulatory elements that mediate p27Kip1 localization are not conserved in plant CKIs; therefore, we were interested in the intracellular localization of ICK1/KRP1.
Whereas YFP expressed from the GL2 promoter could be detected in the nucleus and the cytoplasm, ICK1/KRP1 fused with YFP exhibited a nuclear localization (Figures 3B, 3F, and 3J; see Supplemental Figure 2B online). While this work was in progress, a similar intracellular localization of ICK1/KRP1 was reported by analyzing GFP fusions with ICK1/KRP1 (Zhou et al., 2003). Consistent with the report by Zhou and colleagues, we found that YFP fusions with the truncated ICK1/KRP1109 localized to the nucleus and the cytoplasm in trichomes (Figures 3C, 3G, and 3K; data not shown); a cytoplasmic localization was even more prominent for the GUS:YFP:ICK1/KRP1109 fusion protein (Figures 3D, 3H, and 3L).
In the trichome-neighboring cells, however, both N- and C-terminal YFP fusions with ICK1/KRP1109 could only be detected in the nucleus (Figure 3K; data not shown). On the one hand, this could indicate different cell type–dependent dynamics of the intracellular localization of ICK1/KRP1. On the other hand, it is very well possible that a cytoplasmic fraction of ICK1/KRP1109:YFP was below the detection limit because already in the much brighter stained trichomes the cytoplasmic fluorescence was weak (see Supplemental Figure 2C for a reduction of fluorescence intensities in trichome-neighboring cells).
Premature Endoreplication Does Not Interfere with the Adaptation of Cell-Specific Marker Gene Expression
In the wild type, the cells directly neighboring a trichome develop into morphologically distinct cells, called socket or support cells. Socket cells are rectangular versus the typically lobed pavement cells and are oriented in their longitudinal axis toward the trichome (Figure 1A). In addition, the expression of a few genes and enhancer trap lines has been found to discriminate socket cells from epidermal pavement cells (Molhoj et al., 2001; Vroemen et al., 2003).
Because the trichome-neighboring cells in the ICK1/KRP1-misexpressing plants were greatly enlarged and developed lobes (Figure 1B), we asked whether these cells still have socket cell fate. The analysis of two GAL4 enhancer trap lines from the Scott Poethig collection (http://enhancertraps.bio.upenn.edu/) marking trichome socket cells, 232 and 254, crossed into the reference line for ProGL2:ICK1/KRP1109 revealed expression in the cells surrounding a trichome (Figures 1G to 1I; data not shown; note also the increase in cell size and the enlarged neighboring-cell nuclei in line 254 expressing ProGL2:ICK1/KRP1109). In addition, most of the cells surrounding a trichome were still polarized toward the trichome (Figure 1B). Taken together, these data suggested that the trichome-neighboring cells in ICK1/KRP1-misexpressing plants have developed, at least to some degree, into socket cells.
Entry into an endoreplication cycle has been found to be associated with cell differentiation and the adoption of the special cell morphology occurring after cell fate specification (Nagl, 1976; Sugimoto-Shirasu and Roberts, 2003). Our data, however, implied that trichome-neighboring cells in the ICK1/KRP1-misexpressing plants become specified as socket cells independent and after the onset of an endoreplication program. To explore this hypothesis, we analyzed the cell division activity around incipient wild-type and ICK1/KRP1-misexpressing trichomes more closely. Figure 4 shows that in cells adjacent to young and growing wild-type trichomes, newly formed cell walls can be found, indicating a recent cell division (Figures 4A to 4C). By contrast, around young trichomes of ICK1/KRP1-misexpressing plants, we observed that the neighboring cells had already started to enlarge (Figures 4D to 4F). Consistent with this, we found in DAPI staining that nuclei of trichome-neighboring cells in ICK1/KRP1-misexpressing plants had already started to endoreplicate in contrast with wild-type leaves (Figures 4G and 4H).
|
). Consistent with this, the expression of the two socket cell markers used above only starts when the trichome is already three-branched and expanded (Figure 1G). Further evidence from the gl2 mutant supports an instruction of socket cells at a time point late during trichome development. In gl2 mutants, two classes of trichomes can be found: one class of expanding and even branching trichomes surrounded by socket cells, and the other class displays aborted trichomes, which had started to grow out but failed to expand and become arrested as young bulges (Koornneef, 1990; Rerie et al., 1994). In this latter class, stomata can be found to develop in direct contact with trichomes, suggesting that socket cells have not yet been specified. Finally, in the ICK1/KRP1-misexpressing plants, the socket cell marker became also expressed at later stages of trichome development (data not shown). Taken together, these findings suggest that in the ICK1/KRP1-misexpressing plants, endoreplication has started in the trichome-neighboring cells before these cells have been specified as socket cells; thus, we conclude that plant cells can be specified independent of an endoreplication program.
The Induction of Endocycles by ICK1/KRP1 Depends on the Cell Cycle Mode and the Developmental State
To test whether ICK1/KRP1 is generally a positive regulator of endoreplication in trichome-neighboring cells and its expression is always sufficient to promote endoreplication, we misexpressed ICK1/KRP1 at late stages of socket cell development. For that, ICK1/KRP1109 was cloned behind a UAS regulatory element and introduced into the GAL4 driver line 254 from the Scott Poethig collection by transformation (cf. Figure 1G) (http://enhancertraps.bio.upenn.edu/). Examining plants expressing pUAS:ICK1/KRP1109 in the GAL4 line 254 for a socket cell phenotype revealed neither an alteration in cell size nor in DNA content in comparison to line 254 itself or in wild-type plants (Figure 2, Table 1; see Supplemental Figure 1 online). This observation together with the finding that the trichome-neighboring cells will undergo a few cell division rounds when the GL2 promoter is already highly active (cf. Figures 3A and 4A to 4C) indicated that the induction of endocycles by ICK1/KRP1 depends on the developmental state and/or the cell cycle mode of the cells. This is also supported by the observation that in all transgenic lines generated expressing the various ICK1/KRP1 constructs in trichomes, we have never observed any indication for an increase of endoreplication levels in trichomes by ICK1/KRP1.
To test further whether induction of endocycles by ICK1/KRP1 depends on the cell cycle mode of the cells, we analyzed the effect of ICK1/KRP1 misexpression in other proliferating cells. For that, we made use of the observation that GL2 is also expressed during embryo development starting at heart stage and persisting until bent cotelydon stage (Lin and Schiefelbein, 2001; Costa and Dolan, 2003). Figures 5A and 5B show a torpedo stage embryo with the typical expression pattern of the GL2 promoter in approximately every second cell file in the embryonic epidermis. Expression of ICK1/KRP1 under the GL2 promoter did not alter this expression pattern, as revealed by the analysis of the GL2 promoter reporter line ProGL2:nls:GFP:GUS crossed into plants expressing ProGL2:ICK1/KRP1109 (Figures 5C and 5D). Similar to leaves, we found that ICK1/KRP1-YFP fusion proteins were found in almost all epidermal cells and also weaker in subepidermal cells, demonstrating that the movement of ICK1/KRP1 is not restricted to leaf cells (Figures 5E and 5F).
|
8 µm2 (Figure 5E).
Because of the experimental limitation of embryonic epidermis, we sought for another promoter active in dividing cells, yet not active in all mitotic cells to interfere as little as possible with plant fertility and viability. For this purpose, we used the promoter of the TOO MANY MOUTHS gene (ProTMM) (Nadeau and Sack, 2002a). TMM is expressed during early leaf development in cells of the stomatal lineage and some adjacent cells; many of these cells will undergo at least one more cell division during leaf development.
Transgenic plants misexpressing from the TMM promoter the N-terminally truncated ICK1/KRP1 version fused to YFP were generated. The phenotypes obtained were striking. Expression of ICK1/KRP1109 from the TMM promoter resulted in plants with smaller leaves with an increased degree of serration and displaying much fewer but greatly enlarged epidermal cells (Figures 6A to 6D). Also here, we found that the ICK1/KRP1-YFP fusion proteins moved between cells in the epidermis and also into the mesophyll layer, and it is likely that the strong leaf phenotype resulted from a large number of cells receiving ICK1/KRP1109 (data not shown).
|
Mode of ICK1/KRP1-Induced Endoreplication
From animals, it is known that a conversion of a mitotic cycle into an endocycle can be initiated from different phases of a mitotic cell cycle discriminating different endocycles. For instance, in the first endocycles of Drosophila nurse cells, a new G1-phase is initiated shortly after S-phase, whereas mammalian megakaryocytes progress through a G2-phase and switch to a G1-phase with the beginning of mitosis (Edgar and Orr-Weaver, 2001).
To determine how ICK1/KRP1-induced endocycles proceeded, we used a promoter reporter line for a mitotic cyclin, CYCLIN B1;2 (CYC B1;2, ProCYCB1;2:DB:GUS), which marks cells in a late G2-phase until M-phase of a cell division cycle (Schnittger et al., 2002a). Next, the number of ProCYCB1;2:DB:GUS-positive socket cells surrounding outgrowing but not yet maturated trichomes were compared in a wild-type background and in plants misexpressing ICK1/KRP1109 from the GL2 promoter. We found that wild-type as well as ProGL2:ICK1/KRP1109 plants displayed approximately the same proportion of stained cells adjacent to a trichome, 31% versus 35% (Table 3). Thus, endoreplicating trichome-neighboring cells in ICK1/KRP1 misexpressing plants still entered a G2-phase.
|
). Such a marker becomes degraded in late mitosis with the onset of APC/C activity, which degrades mitotic substrates as cyclins and securin and promotes by that exit from mitosis (Colon-Carmona et al., 1999). Trichome-neighboring cells in ICK1/KRP1-misexpressing plants, however, did not display a continuous staining of the DB:GUS marker, indicating a cyclic degradation of the marker.
Expression of ICK1/KRP1 Can Correct the siamese Mutant Phenotype, and Reduction versus Increase of Endoreplication Levels Correlate with ICK1/RKP1 Transcript Levels
The observation that ICK1/KRP1 could only induce endoreplication in cells with a mitotic cell cycle program and not in endoreplicating cells as trichomes or trichome-neighboring cells suggested that ICK1/KRP1 acts by blocking a mitotic activity while allowing S-phase entry rather than by actively promoting S-phase entry. This is also supported by the cyclic expression of a late G2 reporter.
It is not clear, however, why ICK1/KRP1 misexpression only in trichomes and not in proliferating cells appeared to interfere with S-phase entry. To test whether other developmental cues might be responsible for a differential response of trichome-neighboring cells versus trichomes with respect to S-phase entry, we made use of the siamese (sim) mutant. In sim mutant plants, trichomes undergo mitosis leading to clustered and multicellular trichomes with strongly reduced endoreplication levels; yet these multicellular trichomes display characteristics of typical trichomes with branch formation and papillae on the outer surface (Figure 7A) (Walker et al., 2000).
|
Next, we measured the DNA content of wild type–like sim mutant plants expressing ProGL2:YFP:ICK1/KRP1109. Although nuclei of these trichomes did not fully reach wild-type replication levels, both a quantitative and a qualitative increase in endoreplication levels was found. In sim mutants, 20% of the individual nuclei have a DNA content of 4C or less, and the average DNA content of all nuclei is 8C. By contrast, all of the trichome nuclei on plants expressing ProGL2:YFP:ICK1/KRP1109 in the sim mutant background had a DNA content of more than 4C, and the overall average DNA content was 13C (Figure 7C, line 15). These data showed that ICK1/KRP1 expression can at least partially rescue the sim mutant phenotype. Thus, also in a trichome environment, ICK1/KRP1 expression can induce endoreplication, suggesting that the difference between trichomes and trichome-neighboring cells is more directly associated with the execution of a mitotic program than with other developmental differences.
Furthermore, the spectrum of phenotypes obtained by expressing ICK1/KRP1 in sim mutant plants suggested that ICK1/KRP1 could act in a concentration-dependent manner. Semiquantitative RT-PCR of representative plants from the different phenotypical classes revealed that weak sim-like and wild type–like phenotypes were correlated with low expression strength of the ICK1/KRP1 construct, whereas an ICK1/KRP1-like phenotype was associated with higher expression levels of the construct (Figure 7D). Thus, our data suggest that ICK1/KRP1 supplies a mitosis-suppressing function that is compatible with an endoreplication program at a low concentration, whereas at higher levels of expression, ICK1/KPP1 blocks cell cycle progression completely.
Endoreplicated Trichome Socket Cells Reenter Mitosis
Along with maturation and differentiation, most of Arabidopsis leaf cells switch to an endoreplication cycle (Melaragno et al., 1993) (cf. Figures 6E, 6G, and 6I). Correspondingly, cell divisions become progressively restricted to the basal part of the leaf and finally stop completely (Donnelly et al., 1999).
Surprisingly, we observed that in very old leaves of ProGL2:ICK1/KRP1109 plants, the ProCYCB1;2:DB:GUS reporter was expressed again, that is in trichome socket cells, indicating that these cells again entered a G2-phase (Figure 8A). A comparison with wild-type plants carrying the ProCYCB1;2:DB:GUS transgene confirmed that in comparable stages on wild-type leaves, cell divisions have ceased with the exception of a few meristemoid cells at the leaf base. We determined the ratio of GUS-positive trichome-neighboring cells to total number of trichomes and obtained for leaves of ProGL2:ICK1/KRP1109 plants with a few meristemoid cells in a G2-phase a ratio of 0.024 and on somewhat older leaves without any other detectable cells in a G2-phase a ratio of 0.006 (Table 4). Analysis of these mature socket cells with scanning electron microscopy revealed new cell walls in very large cells (Figure 8B).
|
|
). This correlation suggested that only after the withdrawal from the ICK1/KRP1 regime trichome-neighboring cells entered mitosis.
The general notion is that cells, which have started an endoreplication program, are terminally differentiated and cannot reenter mitosis (Nagl, 1976; Melaragno et al., 1993; Edgar and Orr-Weaver, 2001). However, at the time when neighboring cells resumed cell division, all of them appeared to have undergone substantial endoreplication, suggesting that endoreplicated cells were able to reenter mitosis. To find further support for this possibility, we examined DAPI-stained leaves for the appearance of mitotic figures (Figures 8C to 8F). Figures 8D and 8F show two representative mitotic figures, most likely a metaphase (Figure 8D) and a late anaphase or telophase (Figure 8F) of trichome-neighboring cells in ICK1/KRP1-misexpressing plants. The comparison with similar mitotic stages of wild-type root meristem cells or young leaf cells, which are not polyploid (Figures 8C and 8E), revealed that mitotic figures obtained from ICK1/KRP1-misexpressing plants contained more DNA than dividing cells in the wild type (Figures 8D and 8F). This demonstrates that endoreplicated trichome-neighboring cells underwent mitosis.
As judged by the number of cell walls we identified with scanning electron microscopy, many neighboring cells reentered mitosis (Figure 8B). DAPI staining revealed that the most common nuclear type was an interphase nucleus, indicating that cell divisions did not result in abnormal mitoses or mitotic arrest but rather that mitosis of an endoreplicated cell proceeded without aberrations. Thus, we conclude that plant cells maintain the ability after going through endoreplication cycles to divide again, demonstrating a high degree of flexibility in plant development.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
CKIs as Multiple Cell Cycle Switches
Based on this study and previous experiments, we propose that CKIs have at least three functions in plants. First, ICK/KRPs might be important regulators involved in switching from a mitotic to an endoreplicating cell cycle mode in differentiating cells. As demonstrated by misexpression in trichome-neighboring cells, embryonic epidermis cells, and ProTMM-positive cells, ICK1/KRP1 is a very potent inhibitor of entry into mitosis, whereas it allows S-phase to proceed. Such an inhibitory function might be needed in cells determined to switch to an endoreplication cycle but still containing mitotic regulators. For instance, in Medicago, mRNA of a mitotic cyclin has been detected in the zone of nitrogen-fixing nodules, in which cells will enter an endoreplication cycle (Cebolla et al., 1999). Consistently, the ICK1/KRP1 mRNA was detected in Arabidopsis in mature leaves, in which cells often endoreplicate (Ormenese et al., 2004). Lastly, the rescue of sim mutant trichomes by ICK1/KRP1 expression argues for a function of CKIs in facilitating the switch to an endoreplication cycle. Intriguingly, SIM encodes a small protein with limited homology to ICK/KRPs (J. Larkin, personal communication).
Derived from our finding that ICK/KRPs can block entry into mitosis, we postulate a second function of ICK/KRPs in dividing cells by assisting to establish a G1-phase. Licensing of origins of replication in a G1-phase requires a low CDK activity (Stern and Nurse, 1996). One way to inactivate kinase activity after a preceding mitosis is the APC/C-dependent destruction of mitotic cyclins (Peters, 1998; Harper et al., 2002). In addition, it has been shown that in Drosophila a special CDK inihibitor, ROUGHEX (RUX), binds to and inactivates mitotic CDK complexes, helping to establish a G1-phase with low CDK activity (Foley et al., 1999; Foley and Sprenger, 2001). RUX is an essential gene in Drosophila, demonstrating that there is a high demand for this inhibitory activity. Recently, for the human CKIs p21Cip1 and p27Kip1 and for the RETINOBLASTOMA protein, a similar function in controlling mitotic exit by inactivating mitotic CDK activity was found (Chibazakura et al., 2004). A function of ICK/KRPs in contributing to a G1-phase could also explain the expression of ICK/KRPs in highly proliferating cells, an observation that is so far not understood and appears even contradictory to the previously described function of ICK/KRPs as inhibitors of cell proliferation (Breuil-Broyer et al., 2004; Ormenese et al., 2004). Additional hints for a function of ICK1/KRP1 in or after mitosis come from transcriptional profiling studies of an Arabidopsis cell culture that revealed an expression peak of ICK1/KRP1 mRNA in late G2/M-phase (Menges and Murray, 2002; Menges et al., 2003). Furthermore, genes expressed in late G2-phase and mitosis often contain mitosis-specific activator elements in their promoters, for instance, the promoter of CYCB1;2 shows five elements (Ito et al., 1998; Ito, 2000). In the promoter of ICK1/KRP1, at least eight mitosis-specific activator elements can be found, supporting an expression during mitosis. However, it remains to be seen how in this scenario ICK1/KRP1 is prevented from a premature inhibition of a mitotic CDK complex.
Finally, as shown in previous experiments, misexpression of ICK/KRPs can lead to cells with a reduced DNA content (De Veylder et al., 2001; Jasinski et al., 2002; Schnittger et al., 2003). Therefore, the third function of ICK/KRPs might be to terminate/assist to terminate mitotic as well as endoreplication cycles. This is supported by the analysis of ICK1/KRP1 transcript over time. In 5-week-old Arabidopsis leaves, in which presumably all cell cycle activity has ceased, an increased level of ICK1/KRP1 transcript in comparison with CDKA;1 was found (Wang et al., 1998).
Throwing the Switch
What determines which CKI function is executed? Why does an endoreplicating cell undergo an S-phase block, whereas a proliferating cell is preferentially blocked at mitosis? It is conceivable that ICK1/KRP1 could target different CDK complexes or that it has different affinities to the various cyclin-CDK complexes in endoreplicating trichomes versus mitotic cells. Also, additional components might be present in mitotic and endoreplicating cells, respectively. Misexpression of human p21Cip1, for instance, has led to endoreplication only if the RETINOBLASTOMA protein is absent (Niculescu et al., 1998). Also, the Drosophila inhibitor RUX was found upon misexpression to block mitosis and convert the 16th embryonic cycle into an endocycle. Earlier embryonic cycles, however, were only converted when cyclin E was also absent (Vidwans et al., 2002). Thus, ICK1/KRP1 could have a cell type–specific function depending on a specific set of cell cycle regulators.
All previous data, however, were obtained from misexpression studies using strong promoters, either the GL2 promoter or the 35S promoter of Cauliflower mosaic virus (CaMV35S), precluding any analysis of CDK function at weaker concentrations. In this study, we looked at ICK1/KRP1 moving from trichomes into their neighboring cells and the comparison of fluorescence intensities of YFP-tagged ICK/KRP proteins between trichomes and their neighboring cells revealed a more than twofold difference for ICK1/KRP1109:YFP to YFP:ICK1/KRP1 between the two cell types. In addition, the GL2 promoter appears to have weaker expression in young embryos than later in trichome or root development as judged by the strength of the in situ hybridization signal and fluorescence intensity of reporter genes (Lin and Schiefelbein, 2001; Costa and Dolan, 2003). Not much is known about the relative strength of the TMM promoter, but it is presumably weaker than the CaMV35S promoter. Thus, it is possible that CKIs act as concentration-dependent switches that block entry into S-phase only at high concentrations. This is substantiated by our finding that an ICK1/KRP1 misexpression–like phenotype was found in sim mutant plants with high levels of ICK1/KRP1 expression, whereas at lower expression levels, increased endoreplication levels in comparison with the sim mutant were found. Interestingly, the study of temperature-sensitive CDK alleles in yeast has suggested that for entry into mitosis, higher levels of CDK activity are required than for entry into S-phase (MacNeill et al., 1991; Ayscough et al., 1992). One deduction from the above is that if CKIs are involved in establishing endocycles, and thus are already expressed in endoreplicating cells (e.g., trichomes), the additional expression of ICK1/KRP1 might then reach a threshold concentration of CKI, resulting in a block of S-phase entry. This could explain why among the large number of transgenic plants generated expressing various ICK/KRP versions in trichomes we have only found plants with apparently reduced endoreplication levels in trichomes.
Of course, cell type–specific action and concentration dependency of CKIs are not mutually exclusive. Also, endocycles induced in trichome-neighboring cells differed from endocycles in wild-type trichomes because in trichomes neither CYCB1;1 nor CYCB1;2 promoter activity can be recognized (Schnittger et al., 2002a).
Remarkably, the CYCB1;1 promoter reporter indicating a G2-phase did not accumulate in endoreplicating trichome-neighboring cells. This reporter carries a DB, indicating that at least some activity of the APC/C remained even though CDK activity was presumably blocked. In animals and yeast, CDK activity has been found to be needed to phosphorylate the CDC20 class of APC/C-cofactors and by that activate the APC/CCDC20 (Shteinberg et al., 1999; Kramer et al., 2000). One possibility for ICK1/KRP1-misexpressing plants could be that only the affinity to certain substrates or only certain CDKs might be blocked by ICK1/KRP1, still permitting the activation of APC/CCDC20. One candidate for a CDK that cannot be blocked by ICK/KRPs are the plant-specific B-type CDKs (Joubes et al., 2000). B-type CDKs were not found to interact with ICK/KRPs in yeast two-hybrid studies but are highly expressed during mitosis (De Veylder et al., 2001; Porceddu et al., 2001; Zhou et al., 2002).
Alternatively, a different APC/C complex also could be involved because the CDC20-dependent APC/C is active only in late mitosis (Shteinberg et al., 1999; Kramer et al., 2000). In animals, with the end of mitosis and during a G1-phase of a following cell cycle, another APC/C is assembled containing the CDH1 cofactor class (Zachariae et al., 1998). Studies from Drosophila have revealed that the APC/CCDH1 is also active in the G2-phase and needs to be inactivated before mitosis to allow accumulating mitotic cyclins (Grosskortenhaus and Sprenger, 2002). In contrast with CDC20, phosphorylation has been found to inactivate CDH1 (Kotani et al., 1999; Kramer et al., 2000). Thus, in the case of ICK1/KRP1 misexpression, another possibility is that blocked CDK activity might result in an active ACP/CCDH1. Yet, it remains to be seen whether plant APC/C is similarly regulated by CDC20 and CDH1 homologs.
Non-Cell-Autonomous Action of CKIs
To our knowledge, CKIs have not been found to function in a non-cell-autonomous manner in animals. The finding that ICK1/KRP1 can move between cells adds another level of complexity to plant development and challenges cell cycle control on a tissue and organ level. One way for plants to keep CKIs in check might be the nuclear localization and the high instability of the proteins. In contrast with YFP expressed from the GL2 promoter, full-length ICK1/KRP1 protein could not be detected on protein gel blots. For the N-terminally truncated protein ICK1/KRP1109, a band of the expected size was found. Intriguingly, whereas the full-length ICK1/KRP1 was exclusively found in the nucleus, the truncated version was also located in the cytoplasm. Similar results were recently obtained by Zhou et al. (2003) analyzing roots of plants misexpressing ICK1/KRP1 from the CaMV35S promoter. One likely possibility is that ICK1/KRP1 becomes degraded in the cytoplasm and that for this degradation a motif in the N terminus of the protein is required. In animals, p27Kip1 abundance and localization is strictly regulated (Sherr and Roberts, 1999; Slingerland and Pagano, 2000). p27Kip1 exerts its inhibitory function in the nucleus, and in many experimental systems, p27Kip1 has been found to become degraded in the cytoplasm (Tomoda et al., 1999; Connor et al., 2003). Because no sequence similarities exist between their regulatory domains, it will be interesting to see whether ICK1/KRP1 is subject to similar regulatory pathways as p27Kip1 in animals.
Besides controlling CKIs within a cell, the non-cell-autonomous action of ICK1/KRP1 offers a possibility to link decisions on a cellular level with the supracellular division and growth pattern in organs. For instance, it has been found that starting from the leaf tip, epidermal cells enter an endocycle (Melaragno et al., 1993). CKIs could help to spread the entry into an endoreplication cycle. In addition, CKIs could be involved in linking developmental programs (e.g., trichomes with trichome-neighboring cells). In contrast with other epidermal cells, we found that the level of endoreplication in trichome-neighboring cells is quite constant around 4C to 8C. Of course, this could be a feature of socket cell fate. Alternatively, this could als
