| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
First published online September 14, 2007; 10.1105/tpc.107.054130 The Plant Cell 19:2822-2838 (2007) © 2007 American Society of Plant Biologists
INCURVATA2 Encodes the Catalytic Subunit of DNA Polymerase
|
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
of Arabidopsis thaliana. The strong icu2-2 and icu2-3 insertional alleles caused fully penetrant zygotic lethality when homozygous and incompletely penetrant gametophytic lethality, probably because of loss of DNA polymerase activity. The weak icu2-1 allele carried a point mutation and caused early flowering, leaf incurvature, and homeotic transformations of sepals into carpels and of petals into stamens. Further genetic analyses indicated that ICU2 interacts with TERMINAL FLOWER2, the ortholog of HETEROCHROMATIN PROTEIN1 of animals and yeasts, and with the Polycomb group (PcG) gene CURLY LEAF. Another PcG gene, EMBRYONIC FLOWER2, was found to be epistatic to ICU2. Quantitative RT-PCR analyses indicated that a number of regulatory genes were derepressed in the icu2-1 mutant, including genes associated with flowering time, floral meristem, and floral organ identity. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
). Loss of such cellular memory causes diverse developmental aberrations and diseases, including cancer (Rountree et al., 2001; Gil et al., 2005).
Certain small RNAs are involved in initiating the heterochromatinization of specific chromosome regions (Grewal and Moazed, 2003; Zilberman et al., 2003; Craig, 2004; Lippman and Martienssen, 2004; Mathieu and Bender, 2004). After initiation, protein complexes are recruited that introduce epigenetic marks in histones. Methylation of H3K9 (the Lys-9 residue of the N-terminal tail of the H3 histone) is the most widely studied of these epigenetic marks and seems to be an essential element in the heterochromatinization process (Soppe et al., 2002; Elgin and Grewal, 2003; Craig, 2004). Methylated H3K9 is recognized by the Heterochromatin protein 1 (HP1) of Drosophila melanogaster (Bannister et al., 2001) and its orthologs, Switching 6 (Swi6) of Schizosaccharomyces pombe and the human HP1 (Lachner et al., 2001). After its binding to methylated histones, HP1 mediates the recruitment of other proteins to assemble multiprotein complexes with histone methyltransferase activity (Fransz and de Jong, 2002; Craig, 2004), which repress gene expression through the formation of higher-order chromatin structures (Grewal and Elgin, 2002). During S phase, these structures need to be removed to allow DNA replication and then quickly reassembled to maintain the gene expression states.
The Polycomb group (PcG) of repressors includes crucial components of cellular memory mechanisms that rely on the inheritance of modifications in specific histone tails. They were first discovered in D. melanogaster and found to be responsible for the maintenance of early determined transcription patterns throughout the life cycle of the fly. Loss-of-function mutations in PcG genes result in ectopic expression of homeotic genes (Paro et al., 1998; Francis and Kingston, 2001; Cunliffe, 2003). PcG genes encode components of multiprotein complexes that catalyze methylation of the N-terminal tail of the H3 and H4 histones (Peterson and Laniel, 2004).
Chromatin-mediated gene repression systems exist in plants (He and Amasino, 2005), where an HP1 homolog and several PcG genes have been described. The only protein of Arabidopsis thaliana with overall sequence similarity to HP1 is TERMINAL FLOWER2 (TFL2; also known as LIKE HETEROCHROMATIN PROTEIN1; Larsson et al., 1998; Gaudin et al., 2001; Turck et al., 2007). TFL2 acts as a repressor of genes involved in flowering time, floral organ identity, meiosis, and seed maturation (Nakahigashi et al., 2005). TFL2 is a euchromatic protein that mainly associates with regions evenly distributed along euchromatin, each covering one or two genes marked by the trimethylated H3K27 repressive mark (Turck et al., 2007).
The PcG genes identified in Arabidopsis include, among others, CURLY LEAF (CLF; Goodrich et al., 1997), which represses the floral organ identity gene AGAMOUS (AG; Yanofsky et al., 1990) throughout plant development, VERNALIZATION2 (VRN2; Gendall et al., 2001), which is required for the maintenance of the repression of FLOWERING LOCUS C (FLC; Bastow et al., 2004) in response to vernalization, and EMBRYONIC FLOWER2 (EMF2; Yoshida et al., 2001), which is required to maintain vegetative development and represses flower development (Moon et al., 2003). It has been proposed that different PcG complexes exist in plants, which would include common components, such as EMF2, and specific components, such as CLF and VRN2 (Otte and Kwaks, 2003; Reyes and Grossniklaus, 2003; Chanvivattana et al., 2004). Loss-of-function alleles of the CLF and EMF2 PcG genes share a pleiotropic phenotype that include leaf curling, early flowering, and the misexpression of many genes, some of which encode transcription factors (reviewed in Reyes, 2006). Some of these phenotypic traits are also caused by mutations in genes that encode putative chromatin remodeling factors, such as TFL2, EARLY BOLTING IN SHORT DAYS (Piñeiro et al., 2003), and SPLAYED (Wagner and Meyerowitz, 2002), and by the silencing of BRAHMA (Farrona et al., 2004).
We have already studied at a preliminary level the incurvata2-1 (icu2-1) mutant, which displayed mildly incurved leaves and early flowering. It was identified in a study of 152 Arabidopsis mutants with abnormally shaped leaves, which belong to the Arabidopsis Information Service Form Mutants collection, 13 of which had been isolated by G. Röbbelen and displayed curled, involute leaves, a phenotype that we named Incurvata (Serrano-Cartagena et al., 1999). We found the icu mutations to belong to five complementation groups, one of which, ICU2, was represented by a single recessive allele, icu2-1, which caused the ectopic derepression of several floral organ identity genes in the leaves. In addition, the icu2-1 clf double mutants displayed a synergistic phenotype. Taken together, our results suggested a functional relationship between the ICU2 gene and PcG-mediated gene repression (Serrano-Cartagena et al., 2000).
Here, we report the positional cloning of the ICU2 gene, which was found to encode the putative catalytic subunit of DNA polymerase , and the isolation of lethal icu2 alleles. The morphological and molecular phenotypes caused by the icu2 mutations, the molecular nature of the icu2-1 allele, and its genetic interactions with mutations in TFL2 and several PcG genes suggest a role for the replication machinery in epigenetic inheritance and plant chromatin packaging. Our results suggest the existence in plants of a cellular memory mechanism that involves DNA polymerase , as proposed by previous authors for the fission yeast (Nakayama et al., 2000, 2001; Ahmed et al., 2001; Vermaak et al., 2003).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
). The icu2-1 mutant was early flowering, bolting 24.43 ± 2.10 d after sowing (the wild-type Enkheim-2 [En-2] bolted 37.80 ± 2.90 d after sowing) and produced less vegetative leaves (9.36 ± 0.81) than En-2 (12.40 ± 0.52).
|
). Further linkage analyses allowed us to define a 50-kb candidate region (Figure 2A
), which encompassed 15 annotated genes. All the putative transcription units within the candidate region were sequenced in icu2-1/icu2-1 and En-2 plants. A missense mutation was found in the At5g67100 gene (Figures 2A and 2B), which encodes the catalytic subunit of the DNA polymerase of Arabidopsis.
|
|
|
T transition in the 24th exon of At5g67100, in nucleotide position 6762 of the transcription unit (numbering from the initiation codon), which substituted Arg by Cys in amino acid position 1273 (Figure 2B), a residue that is outside the predicted polymerase catalytic domains of the enzyme (see the II, III, and V domains in Figure 2B; Wong et al., 1998) and other conserved regions. The promoter of this gene includes the E2F binding site motif TTTCCCGC, which is characteristic of cell cycle–regulated genes (Fry and Farnham, 1999; Chaboute et al., 2000).
Proteins interacting with HP1 in D. melanogaster share the PXVXL consensus sequence, named the MOD1-interacting region (MIR) domain (Murzina et al., 1999; Smothers and Henikoff, 2000). A similar pentameric sequence (PFVQV) is present in Swi7, the DNA polymerase of S. pombe, and is required for its binding to the HP1 homolog Swi6 (Nakayama et al., 2001). This sequence is also shared by other DNA polymerase proteins, including ICU2 (PHVQV; Figure 2B) and those of D. melanogaster (PHVQV) and mammals (PHVHV). A computer-based prediction of protein structure suggested that the amino acid change caused by the icu2-1 mutation (Figures 2C and 2D) would reduce or abolish access of some proteins to the MIR domain of ICU2.
The icu2-2 and icu2-3 Recessive Alleles Cause Early Lethality
A search for insertional alleles of ICU2 in public collections allowed us to find the GK-731A12 line in the GABI-Kat database (Li et al., 2003; Rosso et al., 2003), which was confirmed as carrying a T-DNA (Sulr) insertion in position 1075 bp (numbering from the initiation codon) in the first exon of the ICU2 gene, and named icu2-2 (Figures 2A and 2B). An enhancer trap line (N175161) that had been generated by J. Clarke and M. Bevan was confirmed as carrying a Ds (Kanr) transposable element in the first exon of the ICU2 gene and named icu2-3 (Figures 2A and 2B). The presence of Ds- or T-DNA–tagged read-through transcripts was tested by nonquantitative RT-PCR in ICU2/icu2-2 and ICU2/icu2-3 heterozygotes, with only the icu2-3 chimeric transcript being detected.
The ICU2/icu2-2 and ICU2/icu2-3 heterozygotes were phenotypically wild-type. No homozygous mutants were found in their progeny, which suggested recessive lethality. We did not find icu2-1/icu2-2 or icu2-2/icu2-3 heterozygous plants in the progeny of icu2-1/icu2-1 x ICU2/icu2-2 or ICU2/icu2-2 x ICU2/icu2-3 crosses, respectively. On the contrary, the icu2-1/icu2-3 heterozygotes were viable and displayed a morphological phenotype indistinguishable from that of the icu2-1/icu2-1 homozygotes (Figure 3B). Moreover, the icu2-2 allele was not transmitted to the viable progeny of reciprocal icu2-1/icu2-1 x ICU2/icu2-2 crosses (Table 1 ).
|
The icu2-3 allele also caused gametic lethality with incomplete penetrance, although weaker than that of icu2-2. This conclusion was reached from analyses of the progeny of selfed ICU2/icu2-3 plants and those of ICU2/ICU2 x ICU2/icu2-3 reciprocal crosses. The corresponding siliques displayed a higher proportion of normal seeds, together with aborted or unfertilized ovules as well as brown and shrivelled seeds that seemed to have arrested their development in a later stage of maturation. The proportion of ICU2/icu2-3 plants was in all cases higher than that of ICU2/icu2-2, further indicating that the deleterious effects of icu2-2 are stronger than those of icu2-3. Similar conclusions were reached from the analysis of reciprocal icu2-1/icu2-1 x ICU2/icu2-2 and icu2-1/icu2-1 x ICU2/icu2-3 crosses. Although gametogenesis seemed to be affected also by the weak icu2-1 allele, the aborted or unfertilized ovules found in the siliques of nonmanipulated, naturally selfed icu2-1/icu2-1 plants were absent from icu2-1/icu2-1 siliques pollinated by hand with icu2-1 pollen. This suggests that the early lethality observed in icu2-1/icu2-1 plants is due to insufficient pollen quantity rather than poor pollen quality. Taken together, these results indicate that the icu2-2 and icu2-3 alleles, but not icu2-1, affect both ovule and pollen formation.
The Phenotype of the icu2-1 Mutant Is Largely Due to the Derepression of AG and FT
Given that the early flowering phenotype of the tfl2 mutants is suppressed (Kotake et al., 2003) by mutations at the flowering promoter gene FLOWERING LOCUS T (FT; Kobayashi et al., 1999), and due to the resemblance between the phenotypes of tfl2-2, icu2-1, and clf-2, we crossed the ft-1 mutant to either icu2-1 or clf-2. Given that the double mutants displayed the leaf morphology of their icu2-1 or clf-2 parentals and late flowering, such as that of ft-1 (Figures 3F and 3G), we concluded that the early flowering of icu2-1 requires FT. We already found that the severe loss-of-function ag-1 mutation (Chen and Meyerowitz, 1999) suppressed the mutant phenotype of icu2-1/icu2-1 leaves (Figure 3H), where not only AG, but also APETALA1 (AP1; Mandel et al., 1992), AP3 (Jack et al., 1992), and PISTILLATA (PI; Bowman et al., 1989) were ectopically derepressed (Serrano-Cartagena et al., 2000). We obtained in this work double mutant combinations of icu2-1 with the strong loss-of-function ap1-1 (Bowman et al., 1993), ap3-4 (Jack et al., 1992; Irish and Yamamoto, 1995), and pi-1 (Bowman et al., 1991) mutations and found that both leaf morphology and flowering time of all the double mutants were similar to those of the icu2-1 single mutant. These results indicate that both AG and FT are epistatic to ICU2 and that their derepressions (see below) cause leaf incurvature and early flowering, respectively, in the icu2-1 mutant; the results also show that FT is repressed not only by TFL2 but also by CLF and ICU2.
ICU2 Genetically Interacts with CLF, TFL2, EMF2, FAS1, and FAS2
Ectopic expression of AG and other MADS box genes is known to be caused by loss of function of TFL2, CLF, or EMF2, among others (reviewed in Schubert et al., 2005). The mutants affected in these genes share several phenotypic traits with icu2-1, including leaf curvature and early flowering. Therefore, we decided to analyze the genetic interactions between icu2, tfl2, clf, and emf2. We obtained double mutants involving icu2-1 (Figure 4B
) and the null clf-2 (Goodrich et al., 1997) and tfl2-2 (Kotake et al., 2003) mutations (Figures 4C and 4D). The phenotypes of the icu2-1 clf-2, icu2-1 tfl2-2, and clf-2 tfl2-2 double mutants (Figures 4E to 4G) were very similar and much more severe than those of either parental, displaying strongly incurved and narrow leaves arranged in an extremely small rosette. These synergistic genetic interactions suggested that the ICU2, CLF, and TFL2 genes are functionally related.
|
; Chen et al., 1997), is extremely early flowering, and develops a tiny rosette composed of small, narrow, and involute leaves (Yoshida et al., 2001; Figure 4H). Its phenotype is similar to those of the icu2-1 clf-2, icu2-1 tfl2-2, and clf-2 tfl2-2 double mutants. We found emf2-5 to be epistatic to icu2-1, as deduced from a comparison of the icu2-1/icu2-1;emf2-5/emf2-5 plants (Figure 4I) with their ICU2/-;emf2-5/emf2-5 siblings, which were phenotypically indistinguishable in the F2 progeny (427 plants) of an emf2-5/emf2-5 x icu2-1/icu2-1 cross. Taken together with the previously described epistasis of emf2 to clf (Chanvivattana et al., 2004), our results suggested a functional relationship among ICU2, CLF, TFL2, and EMF2 genes.
We also crossed the icu2-1 mutant to viable mutants defective in Chromatin Assembling Factor-1 (CAF-1) function, which is required for histone deposition in Arabidopsis (Schönrock et al., 2006). CAF-1 is a heterotrimeric complex that includes the CAC1, CAC2, and CAC3 proteins in yeast, p150, p60, and p48 in mammals, and FAS1, FAS2, and MSI1 in plants, loss of function of which results in transcription of some silenced genes. Mammalian CAF-1 is necessary for coupling chromatin assembly with DNA replication (Hoek and Stillman, 2003), and its p150 subunit is known to interact with HP1 (Murzina et al., 1999; Quivy et al., 2004). The fas1 and fas2 mutants of Arabidopsis are characterized by fasciated and flat stems, disrupted phyllotaxy, altered floral organ structure and number, leaf serration, and inhibition of root elongation (Leyser and Furner, 1992; Figures 5B and 5E
). We crossed icu2-1 to the fas mutants and found that the icu2-1 fas1-1 double mutant displayed a synergistic phenotype consisting of small, narrow, and involute leaves (Figures 5C, 5D, and 5F), similar to those of the icu2-1 clf-2 and icu2-1 tfl2-2 double mutants obtained in this work. In addition, the icu2-1 fas1-1 floral organs had much more severe alterations than those of either parental (Figure 5F), displaying disrupted phyllotaxy and strongly curly siliques, which only contained aborted or unfertilized ovules. In addition, we performed an icu2-1/icu2-1 x fas2-1/fas2-1 cross, finding in its F2 progeny no icu2-1/icu2-1;fas2-1/fas2-1 plants, as well as a reduced frequency of the icu2-1/icu2-1;FAS2/fas2-1 and ICU2/icu2-1;fas2-1/fas2-1 genotypes (expected, 2/15 each; observed, 1/100 each), which were phenotypically indistinguishable to their icu2-1/icu2-1;FAS2/FAS2 and ICU2/ICU2;fas2-1/fas2-1 siblings, respectively.
|
10-fold more expression in shoot apices (where meristematic cells are more abundant) than in roots, flower buds, open flowers, and siliques, which, in turn, displayed 10-fold more expression than the whole rosettes that were used as a reference. The lowest levels of ICU2 activity were found in leaves and stems. The icu2-1 point mutation did not significantly change the transcriptional activity of the ICU2 gene in rosettes or in shoot apices (Figure 6B). The same conclusion can also be reached by examining the expression profiles of ICU2 obtained from Genevestigator (http://www.genevestigator.ethz.ch; Zimmermann et al., 2004).
|
). In this work, we used QRT-PCR to analyze the expression of these genes and of other genes, which included several members of the MADS box family (SEPALLATA3 [SEP3; Pelaz et al., 2000], CAULIFLOWER [CAL; Kempin et al., 1995], FRUITFULL [FUL; Gu et al., 1998], FLC [Michaels and Amasino, 1999], and SUPRESSOR OF OVEREXPRESSION OF CONSTANS1 [SOC1; Lee et al., 2000]), as well as meristem identity and flowering time genes (FT, LEAFY [LFY; Weigel et al., 1992], and WUSCHEL [WUS; Laux et al., 1996]). We found strongly upregulated the AG, AP1, AP3, PI, SEP3, CAL, FUL, SOC1, and FT genes both in excised leaves and whole rosettes of the icu2-1 mutant (Figures 5C and 5D), as well as LFY and, particularly, WUS only in whole rosettes (Figure 5D). To ascertain whether derepression of the AG, AP1, AP3, or PI floral organ identity genes might in turn be the cause of the derepression of the rest of the studied genes, we obtained the icu2-1 ag-1, icu2-1 ap1-1, icu2-1 ap3-4, and icu2-1 pi-1 double mutants and found no significant differences between their QRT-PCR analyses and those of the icu2-1 single mutant. These results indicated that derepression of SEP3, SEP1, SEP2, CAL, FUL, SOC1, FT, LFY, and WUS is not an indirect consequence of AG, AP1, AP3, or PI deregulation in the icu2-1 mutant. Given the pivotal contribution of AG and FT derepression to the phenotype of the icu2-1, clf-2, and tfl2-2 mutants, and due to the important role of SEP3 and WUS in flower organogenesis and meristem function, respectively, we decided to quantify the expression of these four genes in the icu2-1, clf-2, tfl2-2, and emf2-5 single mutants as well as in the icu2-1 tfl2-2, icu2-1 clf-2, and clf-2 tfl2-2 double mutants. We found the AG, SEP3, FT, and WUS genes to be upregulated (Figure 5E). AG and FT transcript levels were similar in the double mutants and the emf2-5 single mutant and clearly higher than those observed in the icu2-1, clf-2, and tfl2-2 single mutants. The highest WUS transcript levels were observed in the icu2-1 single mutant and the icu2-1 tfl2-2 and icu2-1 clf-2 double mutants, suggesting a role for ICU2 in the repression of this meristematic gene.
The AG Locus Is Enriched in the H3ac Active Mark in the icu2-1 Mutant
The genetic interactions among ICU2, CLF, TFL2, EMF2, FAS1, and FAS2 described here and the derepression of regulatory genes that we observed in the icu2-1 mutant suggested a role for ICU2 in epigenetic inheritance and plant chromatin packaging. In addition, it is known that expression of AG and STM, which are direct CLF targets, depends on their histone methylation pattern (Schubert et al., 2006). This prompted us to perform a chromatin immunoprecipitation (ChIP) analysis to test whether the icu2-1 mutation affects epigenetic marks in AG, which is derepressed in the icu2-1 mutant.
We tested the dimethylation of Lys-9 of histone H3 (H3K9me2), which in Arabidopsis is localized within silent heterochromatin (Lindroth et al., 2004), and the trimethylaltion of Lys-27 of H3 (H3K27me3), which is preferentially associated with open euchromatin regions (reviewed in Henikoff, 2005). These two epigenetic marks were not significantly changed in the promoter and second intron of the AG gene (Figure 7
). We also tested H3 acetylation and found a noticeable enrichment of H3ac in the AG promoter in the icu2-1 mutant (Figure 7). To determine whether or not these histone modifications were associated with changes in DNA methylation, we analyzed by bisulfite genomic sequencing the AG promoter and found that cytosine methylation was not significantly different in the icu2-1 mutant (0.38%; 15 clones sequenced) and the En-2 wild type (1.32%; 10 clones).
|
|
) was crossed to icu2-1/icu2-1 and En-2 plants. No GUS staining was observed in the leaves of the F2 plants carrying the transgene in a wild-type background (Figure 9A
), whereas patches of GUS-stained cells were frequent in the pAG-I:GUS;icu2-1/icu2-1 leaves (Figures 9B to 9D). This patchy ectopic AG expression was displayed by mesophyll, epidermal, and vascular cells of icu2-1/icu2-1 leaves. The expression of pAG-I:GUS was also studied in a tfl2-2/tfl2-2 mutant background, with GUS staining only being found in vascular tissues.
|
), which contains the whole coding sequence and the promoter of TFL2. All wild-type leaf cells displayed a nuclear GFP signal (Figure 9E), as reported previously (Gaudin et al., 2001; Kotake et al., 2003). A similar GFP subcellular localization was observed in many transgenic icu2-1/icu2-1 leaves (Figure 9F), which displayed in addition patches of cells in which the signal was both nuclear and cytoplasmic (Figure 9G) and extremely high (Figure 9H). Although the size and distribution of the patches showing ectopic AG expression were very similar to those showing perturbation of TFL2 subcellular distribution, we have no proof of their colocalization. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
Subunit, ß, 
, 
, and 
; Kornberg and Baker, 1991), to which at least another 10 have been added in the last years (Hubscher et al., 2002). The ß and polymerases have been associated with repair and the and with replication, all of them in the nucleus, whereas the polymerase is required for mitochondrial DNA replication. The Arabidopsis gene encoding DNA polymerase is the only plant DNA polymerase gene that has been studied at a mutational level so far, with all its described mutant alleles causing gametic or embryonic lethality (Jenik et al., 2005; Ronceret et al., 2005).
We positionally cloned ICU2 (At5g67100), a single-copy gene encoding the putative catalytic subunit of DNA polymerase , and identified and characterized three icu2 mutant alleles. DNA polymerase is an essential piece in the cell cycle, where it is required for the initiation of replication; in all eukaryotes, it consists of a catalytic core, two primases, and the so-called B-subunit (of 170, 50, 60 and 70 to 75 kD, respectively; Uchiyama and Wang, 2004). These proteins are encoded in Arabidopsis by the At5g67100, At5g41880, At1g67320, and At1g67630 genes, respectively. Consistent with its important role in the initiation of DNA replication, transcription of the gene encoding the DNA polymerase catalytic subunit is known to be cell cycle regulated in D. melanogaster, S. pombe, and human cells (Wahl et al., 1988; Hirose et al., 1991; Bouvier et al., 1992). In plants, its activity has been seen to peak before the S phase in rice (Oryza sativa) and maize (Zea mays; Yokoi et al., 1997; Gómez Roig and Vázquez-Ramos, 2003). As expected, the ICU2 promoter included the consensus sequence characterizing cell cycle–regulated genes, and its expression was higher in the tissues showing higher cell division rates, such as the shoot apex.
As expected from the essential role played by replication machinery elements, including DNA polymerase , the strong icu2-2 and icu2-3 insertional alleles caused fully penetrant zygotic lethality when homozygous and incompletely penetrant gametophytic lethality, probably because of loss of DNA polymerase activity. The early lethality of these alleles is likely to be due to their inability to contribute to the initiation of replication in gametogenesis or early postfertilization. Although no phenotypic evidence of perturbation in DNA replication was observed in the icu2-1 mutant, we cannot exclude that it also lacks some degree of polymerase activity. The phenotypic effects of the icu2-2 allele were the most severe, and the lethality of the icu2-1/icu2-2 genotype versus the viability of icu2-1/icu2-3 suggested that icu2-2 might not only be an extreme loss-of-function allele but also antimorphic. Although our icu2 alleles might shed light on plant DNA replication, further molecular research on their effects on the polymerase activity of ICU2 will be needed, which is beyond the scope of this work.
Genetic and Physical Interactions between ICU2 and TFL2
Like many other DNA polymerases, the structure of the catalytic DNA polymerase subunit has been compared with a right hand and divided into three canonical domains termed palm, fingers, and thumb (Wang et al., 1997; Rothwell and Waksman, 2005). In Saccharomyces cerevisiae, single mutations in the palm region affect replication fidelity and genomic stability (Limsirichaikul et al., 2003; Niimi et al., 2004), whereas those damaging the fingers reduce processability or the replication rate (Ogawa et al., 2003). In S. pombe, only a few weak alleles of Swi7, the gene encoding DNA polymerase , have been isolated, and null alleles are recessive lethal (Singh and Klar, 1993). The swi7-H4 allele carries a mutation in the conserved box VI and uncouples mitosis and DNA synthesis (Murakami and Okayama, 1995). The swi7-1 allele causes a Gly-to-Glu change in position 1116 of the thumb domain (12 residues upstream to the MIR domain) and perturbs mating-type switching. These results suggest a role for DNA replication in the organization of cellular differentiation programs (Singh and Klar, 1993). The icu2-1 mutation described here affects the thumb, changing the three-dimensional structure of the ICU2 protein, which probably affects accessibility to its MIR domain.
Both the presence of a MIR domain in ICU2 and the genetic interaction described here between the icu2-1 and tfl2-2 mutations pointed to the physical interaction of ICU2 and TFL2. In addition, a computer-based prediction of protein structure suggested that the icu2-1 mutation collapses the ICU2 thumb, which in turn could reduce or abolish the access of TFL2 to the MIR domain. We confirmed this genetic evidence by demonstrating that TFL2 binds to wild-type ICU2, a protein–protein interaction that is likely to be impaired by the amino acid substitution caused by the icu2-1 mutation.
We found in the icu2-1 mutant an abnormal subcellular distribution of TFL2, which was restricted to the nucleus in most leaf cells, but appeared in both the nucleus and the cytoplasm in other cells, which were often clustered. This suggested that reduced or abolished interaction between ICU2 and TFL2 perturbs the nucleocytoplasmic distribution of TFL2. However, we cannot rule out the possibility of position effect variegation, due to the characteristics of the genomic region flanking the gTFL2:GFP transgene, as the cause of the patchy redistribution and increase of the TFL2:GFP protein.
A Link between DNA Replication and Chromatin-Mediated Gene Repression in the Plant Kingdom
Several lines of evidence indicate a link between the DNA replication machinery and cellular memory (McNairn and Gilbert, 2003; Vermaak et al., 2003; Craig, 2004). On the one hand, mutations in the origin recognition complexes affect the repression state of different genes in yeast and perturb position effect variegation in D. melanogaster (Pak et al., 1997). On the other hand, some chromatin-associated proteins, such as HP1 or those of the CAF complexes, are known to physically interact with elements of the replication machinery, such as proliferating cell nuclear antigen (Zhang et al., 2000; Ehrenhofer-Murray, 2004; Sarraf and Stancheva, 2004). In addition, the DNA polymerase Swi7 binds to the HP1 homolog Swi6 in S. pombe, and the swi7-1 and swi7-H4 mutants, which lack such interaction, display deregulation of the mating-type region and an abnormal chromosomal localization of Swi6. These latter observations indicate that some epigenetic marks are replicated simultaneously to DNA with the participation of the DNA replication complexes, at least in the fission yeast (Ahmed et al., 2001; Nakayama et al., 2001).
We hypothesize that the ICU2 DNA polymerase plays a role in epigenetic inheritance in Arabidopsis by facilitating the interaction of TFL2 with histones, which results in propagation of epigenetic marks. In the icu2-1 mutant, this reheterochromatinization polymerase-mediated process would be impaired, making unstable the repression of many genes. Given that the icu2-1 mutation does not strongly affect viability or fertility, the interaction between ICU2 and TFL2 is probably not essential in Arabidopsis or not completely abolished in the icu2-1 mutant, since the MIR domain remains intact. In fact, in S. pombe mutants lacking Swi6–Swi7 interaction, DNA replication activity is almost normal and only the silencing of specific chromosomal regions is perturbed.
Chromatin-mediated gene repression is reduced by the icu2-1 mutation but not completely abolished, as suggested by the morphological aberrations and gene deregulation levels of the icu2-1 and tfl2-2 mutants, which are weaker than those of the icu2-1 tfl2-2 double mutant. The weak phenotype of the tfl2-2 mutant may be due to redundancy with other genes of the chromodomain family. In the icu2-1 mutant, chromatin-mediated cellular memory might fail for some genes in some cells and their progeny as a consequence of impaired TFL2–ICU2 interaction, which in turn would reduce or abolish TFL2–histone interaction, making the repression of TFL2-targeted genes unstable. This clonal behavior might explain the patchy pattern of epidermal cell size, AG ectopic expression, and TFL2 intracellular distribution found in the leaves of the icu2-1 mutant.
The dose level of HP1 is known to affect position effect variegation in D. melanogaster (Ayyanathan et al., 2003), which is consistent with a dynamic structure of silent chromatin, as proposed in the site exposure model (Ahmad and Henikoff, 2002). In the icu2-1 mutant, silent chromatin might lack epigenetic marks after replication longer than in the wild type, allowing misexpression of some genes. It is worth mentioning that ectopic expression of AG does not show a patchy pattern in the tfl2-2 and clf-2 mutants, in which it is uniform, which correlates with their not patchy morphological phenotype, consisting of leaf epidermal cells of reduced size (Larsson et al., 1998; Serrano-Cartagena et al., 2000). The patchy phenotype caused by the icu2-1 mutation might be a useful trait for analyzing the mechanism by which the epigenetic marks are transmitted through mitosis after DNA replication.
Other Genetic Interactions of ICU2
We found the emf2-5 mutation to be epistatic to icu2-1, clf-2, and tfl2-2. In addition, the morphological and molecular phenotypes of the icu2-1 tfl2-2, icu2-1 clf-2, and tfl2-2 clf-2 double mutants are very similar to those of the emf2-5 single mutant. Taken together, these results suggest that EMF2 acts in the same pathway or plays a more general role than that of ICU2, TFL2, and CLF. It is worth noting that the phenotypes of weak emf2 alleles are reminiscent of those of strong clf alleles (Chanvivattana et al., 2004), whereas those of the weak clf-9 mutation and icu2-1 are very similar. It seems that all these genes affect with different strengths the same biological process, which would be specifically related to the repression of FT and AG, among others.
The phenotype of the clf-2 mutant is weaker than those of the icu2-1 clf-2 and tfl2 clf-2 double mutants, suggesting that other PcG proteins or complexes participate in the repression mechanism, as has been proposed based on the partial redundancy found between CLF and SWINGER (Chanvivattana et al., 2004). The synergistic phenotype of the icu2-1/icu2-1;fas1-1/fas1-1 double mutants, together with the lethality of the icu2-1/icu2-1;fas2-1/fas2-1 genotype, and the reduced frequency of the icu2-1/icu2-1;FAS2/fas2-1 and ICU2/icu2-1;fas2-1/fas2-1 plants clearly indicate a functional relationship between ICU2 and the CAF-1 histone deposition complex. These observations reinforce the notion of a functional relationship between chromatin remodeling machinery and the ICU2 DNA polymerase .
Several Genes Derepressed by the icu2-1 Mutation Are TFL2 Targets
Quantitative RT-PCR analyses in the icu2-1 mutant allowed us to identify the upregulation of several closely related MADS box genes of the MIKC group (Parenicova et al., 2003) mainly responsible for floral organ identity, including SEP1, SEP2, SEP3, AG, AP1, AP3, PI, CAL, and FUL. In addition, the flowering promoter gene FT was found upregulated, while the flowering repressor gene FLC was downregulated. Derepression of these genes could be an indirect effect of the derepression of the direct targets of ICU2, as suggested by the demonstration of the regulation of SEP3 and FUL by FT (Teper-Bamnolker and Samach, 2005).
Some of the genes that we found derepressed in the icu2-1 mutant have been found upregulated in a microarray analysis performed on the tfl2-3 mutant (AG, AP3, FLC, PI, and SEP3; Nakahigashi et al., 2005), and some are targeted by the TFL2 protein and enriched in the H3K27me3 repressive mark in the wild type (AG, AP3, FLC, and LFY; Turck et al., 2007).
Another gene found derepressed in the icu2-1 mutant, WUS, is known to be regulated by the FAS1 and FAS2 chromatin assembling factors, which are related with chromatin packaging and DNA replication (Kaya et al., 2001). This suggests that WUS, an important meristematic gene, might also be regulated by chromatin-mediated gene silencing. Indeed, WUS is overexpressed as a consequence of loss of function of the GCN5 gene, which encodes a histone acetyltransferase in Arabidopsis (Bertrand et al., 2003). Unexpectedly, we did not find the phenotypic traits characteristic of WUS overexpression, perhaps because the overexpression of other genes suppressed its effects. Our results bring to mind those obtained by other authors on the repression of homeotic genes in Drosophila and mammals and that of mating-type genes in yeasts and reinforce previous results indicating that floral transition, the major developmental switch in plant life, is controlled mainly by chromatin-mediated gene silencing (reviewed in He and Amasino, 2005; Reyes, 2006).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
) and its corresponding wild type, En-2 (N1138), as well as the ag-1 (NW25; in the Ler genetic background), tfl2-2 (N3797; Col), clf-2 (N290; Ler), ft-1 (N56; Ler), fas1-1 (N265; En-2), and fas2-1 (N266; Ler) lines, were provided by the Nottingham Arabidopsis Stock Centre. The icu2-2 (in the Col genetic background), icu2-3 (Ler) and emf2-5 (Ler) mutants, and the pAG-I:GUS (No-0) and gTFL2:GFP (Col) transgenic lines were provided by GABI-Kat (http://www.mpiz-koeln.mpg.de/GABI-Kat/), Jonathan Clarke, Nobumasa Yoshida, Justin Goodrich, and Koji Goto, respectively.
Plants were grown in sterile (in 150-mm diameter Petri plates containing 100 mL of Murashige and Skoog agar medium, including 1% sucrose) or nonsterile (in pots containing a 2:2:1 mixture of perlite, vermiculite, and sphagnum moss) conditions at 20 ± 1°C, 60 to 70% relative humidity, and continuous illumination of 7000 lx, as described by Ponce et al. (1998). Sterile cultures were performed at a density of 100 regularly spaced seeds per plate in Conviron TC16 tissue culture chambers, and the plants were either collected to be studied or transferred to soil 21 d after sowing. When required, culture media were supplemented with 25 µg/mL of kanamycin or 10 µg/mL of sulfadiazine.
Sequencing
The synthetic oligonucleotides used for cycle sequencing (Table 2
) were bought from Applied Biosystems or from Sigma-Genosys. The sequences of the wild-type En-2 and mutant alleles of the ICU2 gene were obtained from both strands of the products of PCR amplifications of overlapping segments of the At5g67100 gene, whose wild-type Col-0 sequence was already available. Sequencing reactions were performed with ABI PRISM BigDye Terminator Cycle Sequencing kits in 5-µL reaction volumes, and sequencing electrophoreses were performed on an ABI PRISM 3100 Genetic Analyzer as described by Pérez-Pérez et al. (2004).
For bisulfite sequencing, genomic DNA was extracted from 21-d-old plants and bisulfite treated using the EZ DNA Methylation-Gold kit (Zymo Research; D5005) and the following reaction conditions: 10 min at 98°C, 10 min at 53°C, eight cycles consisting of 6 min at 53°C, and 30 min at 37°C, and a final step of 4°C. A 2-µL aliquot of bisulfite-treated DNA was used for each PCR amplification. Primer design (Table 2) and PCR conditions were similar to those previously described (Clark et al., 1994). PCR products were cloned using the TOPO TA cloning kit (Invitrogen). Ten to fifteen individual clones were sequenced for each experiment.
Complementation of the icu2-1 Mutation
A 9467-bp DNA molecule, encompassing the whole ICU2 transcription unit, was PCR amplified in 50-µL reaction mixes using K21H1 transformation-competent artificial chromosome as a template and the primers shown in Table 2, whose 5' tails included NotI and SalI restriction sites. The Expand Long Template PCR system (Roche) was used following the indications of the manufacturer and a polymerization time of 11 min. The PCR product obtained was digested with NotI and SalI and cloned into the pGreenII0179 vector (Hellens et al., 2000). Competent Escherichia coli DH5 cells were transformed, and transformants were isolated on Luria-Bertani plates supplemented with 25 µg/mL of kanamycin, 0.25 mg/mL of X-Gal, and 0.04 mg/mL of isopropylthio-ß-galactoside and tested by PCR for the presence of the construct. Plasmid DNA was isolated from positive clones and used to transform competent Agrobacterium tumefaciens C58C1 cells carrying the pSoup helper plasmid (Hellens et al., 2000). Agrobacterium transformants were selected on Luria-Bertani medium supplemented with 25 µg/mL of kanamycin and 5 µg/mL of tetracycline. Liquid cultures of the positives clones were used for in planta transformations of Arabidopsis thaliana wild-type and mutant plants (Bechtold and Pelletier, 1998). For the isolation of transformant plants, T2 seeds were sown in agar plates supplemented with 40 µg/mL of hygromycin. The presence of the transgene in the putative transformants was verified by PCR.
GUS Staining and GFP Visualization
Homozygotes for either the gTFL2:GFP or pAG-I:GUS transgene were crossed to mutant icu2-1/icu2-1 and wild-type En-2 plants and their F1 allowed to selfpollinate. F2 plants carrying the transgene were selected in 25 µg/mL of kanamycin-supplemented plates. For GUS analyses, 21-d-old plants carrying the pAG-I:GUS transgene in wild-type and icu2-1/icu2-1 backgrounds were collected and treated with 90% acetone for 15 min on ice. Samples were then washed with 100 mM sodium phosphate buffer, pH 7.0, containing 3 mM K4Fe(CN)6, 3 mM K3Fe(CN)6, and 0.1% Triton X-100 for 2 min on ice and then covered with 1 mM 5-bromo-4-chloro-3-indolyl ß-D-glucuronide in the same buffer and incubated overnight at 37°C. After staining, leaves were cleared in an ethanol series. For GFP visualization, vegetative leaves of the 4th and 5th nodes were excised from 21-d-old plants carrying the gTFL2:GFP construct, mounted in water on slides, and directly observed by confocal laser scanning microscopy in a Leica TCS-NT microscope equipped with FITC/TRITC filters.
Real-Time QRT-PCR
QRT-PCR amplifications and measurements were performed as described by Pérez-Pérez et al. (2004) in an ABI PRISM 7000 sequence detection system. For each of the genes under study, a primer pair was designed (Table 2) to amplify a product of 100 bp. To avoid amplification of genomic DNA, the 5' and 3' ends from each pair matched the sequences of two adjacent exons.
Plants were collected 21 d after sowing, frozen in liquid N2, and ground in RNase-free conditions. RNA was extracted with the Qiagen RNeasy plant mini kit, and finally resuspended in 88 µL of RNase-free water and treated with DNaseI at 37°C for 30 min. RNA was then precipitated and resuspended in 40 µL of RNase-free water. Three to five micrograms of the RNA solution obtained were reverse transcribed using SuperScript II reverse transcriptase (Gibco BRL) following the manufacturer's instructions. One microliter of the resulting cDNA solution was used as template in a 25-µL QRT-PCR reaction mix, which included 12.5 µL of the SYBR-Green PCR master mix (Applied Biosystems) and 0.4 µM of primers. Relative quantification of gene expression was performed using the 2–
CT or comparative CT method (Livak and Schmittgen, 2001). Reactions were made in triplicate, and the expression levels were normalized using the CT values obtained for the OTC housekeeping gene (Quesada et al., 1999), which was used as an internal reference gene (Cnops et al., 2004; Pérez-Pérez et al., 2004).
Pull-Down Assays
TFL2 cDNA was amplified from 5-d-old seedling mRNA using the Superscript III one-step RT system (Invitrogen) and specific primers spanning from the ATG to the stop codons. The cDNA was cloned in the Gateway-compatible vector D221 (Invitrogen). Afterwards, the cDNA was cloned
and Interacts with Genes Involved in Chromatin-Mediated Cellular Memory in Arabidopsis thaliana
