CREB-binding protein (CBP) and its homolog p300 possess histone acetyltransferase activity and function as key transcriptional co-activators in the regulation of gene expression that controls differentiation and development in animals. However, the role of CBP/p300-like genes in plants has not yet been elucidated. Here, we show that Arabidopsis CBP/p300-like genes promote flowering by affecting the expression of a major floral repressor FLOWERING LOCUS C (FLC). Although animal CBP and p300 generally function as co-activators, Arabidopsis CBP/p300-like proteins are required for the negative regulation of FLC. This CBP/p300-mediated FLC repression may involve reversible protein acetylation independent of histone modification within FLC chromatin.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Accurate control of differentiation and development in higher eukaryotes largely depends on the transcriptional regulation of spatial and temporal gene expression in response to developmental or environmental signals. Transcriptional gene regulation is a complex process and requires the orchestration of many transcription factors and multi-functional co-activators/co-repressors (Spiegelman and Heinrich, 2004). CREB-binding protein (CBP) and p300 are well-known transcriptional co-activators in animals and are closely related in their sequences and functions (Kalkhoven, 2004; Ogryzko et al., 1996). CBP/p300 consist of multiple structural domains through which they interact with a wide spectrum of nuclear proteins, including basic transcription factors such as TATA box-binding proteins and TFIIB, as well as various other transcription factors (Goodman and Smolik, 2000; Kalkhoven, 2004). Thus, CBP/p300 function as scaffolds connecting the transcriptional activators to the basic transcriptional machinery (Goodman and Smolik, 2000; Kalkhoven, 2004).
CBP/p300-like genes are also found in plants (Pandey et al., 2002). Of these gene products, the Arabidopsis CBP/p300-like protein HAC1 was demonstrated to possess HAT activity (Bordoli et al., 2001) and implicated in the transcriptional activation of a heat-shock-inducible gene in a protoplast system (Bharti et al., 2004). However, the relevance of such observations to the biological roles of HAC1 has not been addressed.
In plants, flowering is a major switch from the vegetative to the reproductive phase of development. Because flowering is closely associated with plant reproductive strategy, different plant species have distinct flowering times that are optimized to ensure reproductive success. Flowering time is regulated by the convergence of multiple endogenous and environmental signals, including developmental status, hormone signaling, light period and quality, and vernalization (reviewed in Boss et al., 2004; Simpson and Dean, 2002). Molecular genetics studies have identified major floral regulatory pathways in Arabidopsis, namely the photoperiod pathway, the gibberellin (GA)-dependent pathway, and the FLOWERING LOCUS C (FLC)-dependent pathway. The photoperiod pathway mediates the effect of day length on flowering. In Arabidopsis, flowering is promoted by long days (LD) but repressed by short days (SD). The GA-dependent pathway functions as a default floral promotion pathway in non-inductive SD in Arabidopsis.
FLC is a central floral repressor in Arabidopsis and acts as a convergence point of multiple floral regulatory pathways. FRIGIDA (FRI) functions as a transcriptional activator of FLC in winter-annual Arabidopsis. However, this positive effect of FRI on FLC is antagonized by a long-term cold treatment known as vernalization. Genes in the autonomous pathway negatively regulate FLC expression in summer-annual Arabidopsis (Noh and Noh, 2006). In these plants, mutations in the autonomous pathway genes result in increased FLC expression and late flowering, similar to winter-annual Arabidopsis. Studies using an FRI-containing winter-annual Arabidopsis have identified numerous factors required for the elevated expression of FLC. Many of these factors resemble proteins that are involved in chromatin modifications in other organisms (reviewed in He and Amasino, 2005; Noh and Noh, 2006). However, the molecular mechanisms of FLC regulation exerted by the genetically identified FLC repressors and activators are yet to be elucidated.
Here, we report that Arabidopsis CBP/p300-like proteins control flowering in the autonomous floral regulatory pathway by negatively affecting the expression of a major floral repressor FLC. Further, we show that this Arabidopsis CBP/p300-mediated FLC regulation might involve reversible protein acetylation independent of histone modification within FLC chromatin.
Identification of T-DNA insertional mutants of the Arabidopsis CBP/p300-like genes
The Arabidopsis genome encodes five CBP/p300-like proteins, namely HAC1, HAC2, HAC4, HAC5 and HAC12 (Figure 1) (Pandey et al., 2002). The domain composition and organization are conserved in these five proteins, although the HAC2 protein lacks the N-terminal TAZ-type ZnF domain. However, the domain organization and composition differ between animal CBP/p300 and Arabidopsis HATs of the CBP family (HACs). For example, the CREB transcription factor binding domain and the bromodomain that is known to bind to the acetylated histone tails (Dhalluin et al., 1999) are not detected in HACs; this suggests that the biochemical roles of plant and animal CBP/p300 might be different.
To elucidate the biological roles of the CBP/p300-like proteins in plants, we obtained T-DNA insertion lines of these CBP/p300-like genes from the SALK T-DNA collection. The T-DNA insertion site in each line was defined by sequencing the PCR product obtained using a T-DNA border primer and a gene-specific primer in order to confirm the presence of T-DNA. As shown in Figure 1, the T-DNAs were inserted within the coding regions of the HAC genes in all alleles, except for the hac2-1 allele in which the T-DNA was inserted 80 bp upstream from the start codon. In case of the hac12-2 allele, the T-DNA was inserted 8 bp upstream from the stop codon. RT-PCR analyses confirmed the absence of full-length messages in the hac1-1, hac1-2, hac4-1, hac4-2, hac5-1, hac5-2 and hac12-1 mutants (data not shown).
Mutations in HAC1 cause late flowering
Phenotypes of the hac1-1, hac1-2, hac4-1, hac4-2, hac5-1, hac5-2 and hac12-1 homozygous mutants were examined throughout their development. The only single mutant with an observable defect was hac1. Both hac1-1 and hac1-2 flowered later than the wild-type (wt) in LD (16 h light/8 h dark) as well as SD (8 h light/16 h dark) (Figure 2a–c). The delayed flowering was not due to a delayed leaf initiation rate (Supplementary Figure S1) but was due to delayed transition of the shoot apical meristem (SAM) from the vegetative to the reproductive phase as characterized by a higher number of rosette leaves at the onset of flowering (Figure 2b,c). The photoperiod-independent late-flowering phenotypes of hac1 suggested that HAC1 might play a role in the FLC-dependent pathway. The late flowering of hac1 was also effectively suppressed by vernalization (wt and hac1-1 flowered with 12.5 ± 0.9 and 13.2 ± 0.8 rosette leaves, respectively, in LD after 50 days of vernalization), similar to other late-flowering mutants in the FLC-dependent pathway (e.g. Michaels and Amasino, 2001).
In order to address the molecular mechanism underlying the late flowering of hac1, we first examined the expression levels of various floral pathway genes. These included FLC (a floral repressor in the autonomous and vernalization pathways), CONSTANS (CO; a floral activator in the photoperiod pathway), and FT and SUPPRESSOR OF OVEREXPRESSION OF CO1 (SOC1; downstream floral integrators). The FLC mRNA levels were increased in hac1-1 and hac1-2, whereas there was no detectable difference in the CO mRNA levels between wt and hac1. Consistent with these observations, the FT and SOC1 mRNA levels were decreased in the mutants (Figure 2d). Furthermore, the late-flowering phenotype of hac1 was completely suppressed by an flc null mutation (Figure 2e,f), thereby demonstrating that HAC1 controls flowering by negatively regulating FLC expression.
Arabidopsis CBP/p300-like genes have redundant functions in flowering time regulation
Although morphological phenotypes were not detected among the hac5 or hac12 single mutants (Figure 3c), the conserved domain organizations (Figure 1) and high sequence similarities among HAC1, HAC5 and HAC12 (Supplementary Figure S2) suggest the possibility of functional redundancy among these genes. In order to test this possibility, we generated double mutants between the three genes (hac1-1 hac5-1, hac1-1 hac12-1 and hac5-1 hac12-1) and examined their phenotypes. Additional loss of HAC5 activity or, to a lesser extent, loss of HAC12 activity in the hac1-1 background dramatically delayed flowering compared to that in wt or the hac1 single mutants (Figure 3a). However, the hac5-1hac12-1 double mutants flowered slightly later than wt but earlier than the hac1 single mutants (Figure 3a; wt, hac5-1 hac12-1 and hac1-1 flowered with 10.1 ± 0.7, 13 ± 1.3, and 16.1 ± 1.4 rosette leaves, respectively).
For the following two reasons, it was unlikely that the strong late-flowering phenotypes of hac1-1 hac5-1 and hac1-1 hac12-1 were caused by mutations independent of the lesions in HAC1, HAC5 and HAC12. Firstly, the two independent mutant alleles of HAC1 (hac1-1 and hac1-2) showed consistent late-flowering phenotypes, and this phenotype was enhanced by mutations in either HAC5 or HAC12, two homologous genes of HAC1. Secondly, it is very unlikely that both hac5-1 and hac12-1 alleles carry the same-type of mutations that enhance the late-flowering phenotype of hac1-1 in regions outside of HAC5 and HAC12, respectively. Taken together, these results suggest that (i) HAC1, HAC5 and HAC12 are functionally redundant in the regulation of flowering, (ii) HAC1 plays a dominant role over the other functionally redundant HAC genes, and (iii) HAC5 has a more significant role than HAC12.
Consistent with the flowering phenotypes, the FLC mRNA levels in the hac1 hac5 and hac1 hac12 double mutants were considerably increased compared to those in the hac1 single mutants (Figure 3b). The Arabidopsis genome has five genes encoding MADS box proteins that are highly related to FLC (MAF1–MAF5; Parenicova et al., 2003). Previous studies showed that a portion of these genes are co-regulated with FLC (Deal et al., 2005; He et al., 2004). Therefore, we examined whether HACs are also involved in the transcriptional regulation of these FLC paralogs. Among these genes, the expression of MAF3, MAF4 and MAF5 was also weakly but significantly elevated in the hac1 single mutants and in the hac1 hac5 and hac1 hac12 double mutants (Figure 3b).
Multiple roles of HACs in Arabidopsis development
In addition to differences in flowering time, the hac1 hac5 and hac1 hac12 double mutants also displayed multiple developmental and morphological defects that were not observed in any of the single mutants. The plant size of the double mutants was smaller than that of the wt, particularly at younger stages (Figure 3c). In addition, hac1 hac5 mutants developed round cotyledons and rosette leaves that were not observed in hac1 hac12, which suggests that HAC genes have distinct functions with respect to certain developmental programs. Similarly, human CBP and p300 are not completely redundant in their functions but possess unique roles (Kalkhoven, 2004; Kawasaki et al., 1998).
Flowers of the two double mutants displayed a gradient of morphological aberrancy. In a majority of the double mutant flowers, the pistils were enlarged and elongated, resulting in their protrusion from the perianth. This protrusion was not because the sepals and petals of the double mutant flowers were smaller than those of wt flowers but because the pistils of the double mutant flowers were longer than those of wt flowers (Figure 3d). In severe cases, the hac1 hac5 double mutant flowers contained twisted and enlarged pistils, underdeveloped petals and shorter filaments than those of wt flowers, with aberrantly developed pollen grains (Figure 3d). However, the numbers of each floral organ were not changed even in severe cases. Therefore, our data show that the functionally redundant Arabidopsis CBP/p300-like genes are required for the control of a variety of developmental processes including flowering.
Expression patterns of HACs
To gain a deeper insight into the functional redundancy among HAC genes, we examined their expression patterns by RT-PCR using RNAs obtained from various tissues such as roots, leaves, leaves with SAMs, flowers and seedlings. All five HAC genes were expressed in all the tissues examined; however, the expression patterns varied slightly (Figure 4a). A more careful examination of the spatial expression patterns of HAC1, HAC5 and HAC12 by using transgenic plants carrying promoter:GUS fusion constructs revealed GUS-staining patterns that were comparable to the RT-PCR results (Figure 4b). In general, all three genes showed similar expression patterns, and there was a good agreement between the gene expression sites and the organs that displayed morphological abnormalities in the mutants. The expression patterns of the three genes at the seedling stage were similar to the spatial expression pattern of FLC (Figure 4b), which was consistent with their function as FLC repressors. On the other hand, the absence of FLC expression in the anthers and stigmas, where HAC genes are highly expressed, and the high level of FLC expression in the styles, where the expression of HAC genes is undetectable (Figure 4b), suggest the possibility of FLC repression by HAC genes in these floral tissues. It was observed that the green fluorescent protein (GFP) fused with the full-length HAC1 protein was localized in the nucleus when transiently expressed in onion epidermal cells (Figure 4c). This suggests the role of HACs as transcriptional co-factors.
No reciprocal transcriptional regulation occurs between HAC genes and other FLC regulators
HAC-mediated FLC repression might involve reversible protein acetylation independent of histone acetylation within FLC chromatin
It is unlikely that HACs repress FLC transcription through histone acetylation of FLC chromatin because this process is correlated with the transcriptional activation rather than repression of FLC (Ausín et al., 2004; He et al., 2003; Noh et al., 2004a,b), and this is consistent with the general concept of the relationship between histone acetylation and transcriptional activation (Eberharter and Becker, 2000; Kalkhoven, 2004). In fact, chromatin immunoprecipitation (ChIP) assays with antibodies specific to penta-acetylated histone H4 or tetra-acetylated histone H3 revealed no detectable differences in the acetylation levels of the H3 and H4 histones within FLC chromatin in the hac mutants when compared with that in the wt (Figure 6b).
Therefore, what is the mechanism by which HACs repress FLC transcription? A growing body of evidence has shown that certain HATs, including CBP/p300, are also capable of acetylating non-histone transcription-related proteins and regulate their function (Bereshchenko et al., 2002; Gu and Roeder, 1997; Guidez et al., 2005; Kalkhoven, 2004; Korzus et al., 2004; Yuan et al., 2005). It is possible that such post-translational acetylation is involved in the HAC-mediated FLC repression. As the activity of HATs on either histones or non-histone proteins is dynamically counterbalanced by the opposing activity of histone deacetylases (HDACs) in order to attain appropriate transcription levels (Korzus et al., 2004; Martínez-Balbás et al., 2000; Yuan et al., 2005), we studied whether the loss of HAT activity in the hac mutants is compensated by the elimination of HDAC activity, and whether the hac mutations increase FLC expression by causing hyperacetylation of a target protein other than histones within FLC chromatin. When hac1 and hac1 hac12 were treated with the HDAC inhibitor trichostatin A (TSA), the elevated FLC transcript levels in the mutants decreased (Figure 6c), although the histone acetylation levels within FLC chromatin increased (Figure 6d). hac1 hac12 rather than hac1 hac5 double mutants were used for the TSA treatment of seedlings because of the severe sterility of hac1 hac5 mutant plants.
This result indicates that HDAC activity is required for the activation of FLC transcription as well as for the deacetylation of histones within FLC chromatin that has been implicated in the repression of FLC transcription (Ausín et al., 2004; He et al., 2003; Noh et al., 2004a,b). Thus, FLC transcription appears to be affected by the balance between HAC-mediated HAT activity and HDAC activity for substrates other than histones, and it may be suggested that another layer of protein acetylation/deacetylation independent of histone modifications is required for the regulation of FLC transcription. Alternatively, HACs might control FLC expression through a modification of histones within the chromatin of so far unidentified FLC regulators. In future studies, the identification of HAC-interacting proteins or genome-wide expression analyses using the hac mutants might be useful in elucidating the biochemical mechanisms of the HAC-mediated FLC regulation.
CBP and p300 are well-known transcriptional co-factors that are involved in multiple aspects of animal development (reviewed in Goodman and Smolik, 2000; Kalkhoven, 2004). Although the HAT activity (Bordoli et al., 2001) and transcriptional co-activator function in a protoplast system (Bharti et al., 2004) of the plant CBP/p300-like protein HAC1 have been demonstrated, the in vivo roles of plant CBPs have not yet been addressed. In this study, we demonstrated the in vivo roles of Arabidopsis CBP family genes using a reverse genetics approach. We found that three (HAC1, HAC5 and HAC12) of the five Arabidopsis CBP/p300-like genes have redundant roles in the promotion of flowering through the repression of mRNA expression of the central floral repressor FLC (Figure 3a,b). These genes also have other roles in Arabidopsis development as evidenced from the formation of smaller round leaves in hac1 hac5 (Figure 3c) and the abnormal flower development and reduced fertility in hac1 hac5 and hac1 hac12 (Figure 3d). Thus, our study reports multiple in vivo roles of plant CBPs in Arabidopsis development. One of their prominent roles is to promote flowering through the repression of FLC.
Eight members (LD, FCA, FPA, FY, FLD, FVE, FLK and REF6) of the autonomous floral regulatory pathway that promote flowering through FLC repression have been identified using forward genetics approaches. In this study, we report the ninth functionally redundant member of the autonomous floral regulatory pathway by using a reverse genetics approach. We found that HAC1 specifically represses FLC (Figure 2d,f), and this repressive role is shared by two other members of the Arabidopsis CBP/p300-like family (Figure 3a,b). Based on previous studies using forward molecular genetics approaches, we believe that the mild late-flowering phenotype of the hac1 single mutants (Figures 2 and 3a) might prevent the identification of HAC1 as an FLC repressor by these methods. Further, it might not have been possible to determine the roles of HAC5 and HAC12 as autonomous pathway floral regulators by using forward genetics approaches because their single mutants display normal flowering phenotypes (Figure 3a). Therefore, our study suggests that there might be more unidentified members in the autonomous pathway as well as in other floral regulatory pathways, and that the identification of such novel floral regulators by approaches other than forward genetics could lead to a better understanding of flowering time regulation.
In eukaryotic cells, histone acetylation and deacetylation have generally been implicated in the positive and negative regulation of transcription, respectively. Accordingly, HATs and HDACs have been identified as components of transcriptional activator and repressor complexes, respectively (reviewed in Lee and Young, 2000). This positive relationship between histone acetylation and transcription was also found in FLC transcription (Ausín et al., 2004; He et al., 2003; Noh et al., 2004a,b). Therefore, it was predicted that HATs and HDACs might play activator and repressor roles, respectively, in the transcriptional regulation of FLC.
Contrary to our expectations, we found that the CBP/p300-like HATs (HAC1, HAC5 and HAC12) function as transcriptional repressors of FLC (Figures 2d and 3b). One of the explanations for this reverse role of HACs is that CBP/p300 might activate the transcription of other FLC repressors that in turn repress FLC transcription. However, no change was observed in the mRNA expression of the eight previously identified FLC repressors (Figure 5b) and the known FLC activators (Figure 5c) in the hac1 single mutants and the hac1 hac12 and hac1 hac5 double mutants. Therefore, HACs might directly affect FLC transcription or be required for the regulation of so far unidentified FLC regulators.
Although the mechanism underlying HAC-mediated FLC repression remains unclear, our data suggest that HATs and HDACs might be involved in at least two different modes of transcriptional regulation of FLC. Firstly, the pharmacological inhibition of HDAC activities reduced FLC expression (Figure 6c) but induced the hyperacetylation of histones within FLC chromatin (Figure 6d). Secondly, the loss of HAT activities in the hac mutants caused increased transcript levels of FLC (Figure 3b) independently of histone acetylation within FLC chromatin (Figure 6b).
It is possible that HACs affect FLC expression by affecting the expression of unidentified FLC repressors. However, an aspect of CBP/p300 function suggests another possible mechanism by which FLC repression is directly mediated by HACs. It has been reported that HATs acetylate non-histone nuclear proteins as well as histones, resulting in either increased or decreased DNA–protein or protein–protein interactions for the non-histone proteins (Bereshchenko et al., 2002; Gu and Roeder, 1997; Guidez et al., 2005; Kalkhoven, 2004; Korzus et al., 2004; Yuan et al., 2005). Thus, it is possible that HACs might affect FLC transcription by modifying the activity of other FLC transcriptional regulators.
Our data show that the open configuration of FLC chromatin induced by increased histone acetylation is not sufficient for the transcriptional activation of FLC (Figure 6c,d). These results could be explained by the hypothesis that the reduced activity of HDAC might increase the pool of acetylated active FLC repressors as well as acetylated histones. In addition, the open configuration of FLC chromatin induced by histone acetylation might increase the accessibility of DNA to the acetylated active repressors as well as to the activators. Alternatively, the inhibition of HDAC activity might decrease the pool of functionally active deacetylated FLC transcriptional activators.
Although previous studies on FLD (He et al., 2003), FVE (Ausín et al., 2004) and REF6 (Noh et al., 2004a,b) showed positive relationships between histone acetylation and the expression level of FLC mRNA, biochemical evidence recently reported shows that amine oxidases homologous to FLD are actually demethylases specific for di-methylated histone H3 lysine 4 (Shi et al., 2004), and that the Jumonji C omains that REF6 has possess demethylase activities specific for histone H3 lysine 36 and/or lysine 9 (Tsukada et al., 2006; Whetstine et al., 2006; Yamane et al., 2006). The biochemical function of FVE is not yet clear. Therefore, the significance of the hyperacetylation of histones at the FLC locus observed in fld, ref6 and fve mutants is yet to be clarified, and it would be premature to claim that the hyperacetylation of histones is indeed a cause for the higher level of expression of FLC mRNA in those mutants. It is possible that the hyperacetylation is an indirect effect of altered methylation levels of histones at the FLC locus or an event accompanying the transcriptional activation of FLC. Our data in Figure 6 are consistent with the possibility that the hyperacetylation is not a direct cause of the transcriptional activation of FLC. We propose that HACs might contribute to the regulation of FLC transcription by acetylating another FLC transcriptional activator or repressor to alter its DNA-binding affinity or its ability to recruit other transcriptional co-regulators.
The flowering times were measured as the number of rosette and cauline leaves (LN) formed by the primary meristem, and the data are presented as means ± SD for at least 12 plants for each genotype.
Total RNA was isolated from 10-day-old or 7-leaf-stage plants using the TRI reagent (Sigma) according to the manufacturer's instructions. Reverse transcription was performed using Superscript II (Invitrogen Life Technologies, Carlsbad, CA, USA) followed by quantitative PCR of the first-strand DNA with ExTaq polymerase (TaKaRa Bio, Otsu, Shiga, Japan). Sequences of the RT-PCR primers used to study the expression of the HAC genes are provided in Supplementary Table S1. Sequences of the RT-PCR primers used to study the expression of the flowering genes other than HAC genes are available on request.
ChIP was performed as described previously (Noh et al., 2004a,b) with a few modifications. To purify DNA from immunoprecipitated complexes, the QIAquick Spin Column (Qiagen, Hilden, Germany) was used instead of the phenol/chloroform extraction method. The details and sequences of the primers that were used to amplify different FLC regions in the ChIP assays have been described previously (He et al., 2003; Sung and Amasino, 2004).
Histochemical GUS staining
In order to create the HAC promoter:GUS fusion constructs, DNA fragments containing the regions from −1187 to −1, −4370 to −1 and −3070 to −1 of HAC1, HAC5 and HAC12, respectively, were cloned into the pPZP211-GUS vector (Noh and Amasino, 2003). Each DNA construct was introduced into Agrobacterium tumefaciens strain ABI, followed by transformation of Col wt plants (Clough and Bent, 1998). Histochemical GUS staining was performed on the transgenic plants as previously described (Noh et al., 2001). Primers used for DNA construction are listed in Supplementary Table S2. The seedlings shown in Figure 4(b) were grown for 8 days under LD before staining.
Subcellular localization assay
The 5076 bp HAC1 cDNA was obtained by RT-PCR using HAC1GFP-F and HAC1GFP-R (Supplementary Table S2) as primers. The cDNA was cloned into the JJ461 binary vector in the region between the CaMV 35S promoter and GFP, thereby creating a C-terminal translational fusion of GFP to HAC1. For the transient expression of HAC1:GFP, the plasmid DNA was introduced into onion epidermal cells by DNA-coated gold particle bombardment. After incubation at 25°C for 1 day, the cells were observed using a confocal microscope (LSM 510 META; Carl Zeiss, Jena, Germany).
We are grateful to J.-S. Jeon and J.-I. Cho for providing the JJ461 vector, as well as their technical advice on GFP localization, and to SIGnAL for providing the knockout pools that contained the hac1, 2, 4, 5, 12, flk1 and fy-3 alleles. The work carried out in B.N.’s lab was supported by the BioGreen21 Program (20050401-034-753-145-03-00) from the RDA and by a grant from the MOST/KOSEF to the EB-NCRC (R15-2003-012-01001-0). The work in Y.-S.N.’s lab was supported by the Global Research Laboratory Program of MOST/KOSEF, by the Plant Diversity Research Center, by the BioGreen21 Program (20050401-034-753-145-02-00) of the RDA, by the BK21 Program, and by a KRF grant (MOEHRD, Basic Research Promotion Fund, KRF-2005-C00023).