Ubiquitin Ligase Components Cullin4 and DDB1 Are Essential for DNA Methylation in Neurospora crassa*

DNA methylation and H3K9 trimethylation are involved in gene silencing and heterochromatin assembly in mammals and fungi. In the filamentous fungus Neurospora crassa, it has been demonstrated that H3K9 trimethylation catalyzed by histone methyltransferase DIM-5 is essential for DNA methylation. Trimethylated H3K9 is recognized by HP1, which then recruits the DNA methyltransferase DIM-2 to methylate the DNA. Here, we show that in Neurospora, ubiquitin ligase components Cullin4 and DDB1 are essential for DNA methylation. These proteins regulate DNA methylation through their effects on the trimethylation of histone H3K9. In addition, we showed that the E3 ligase activity of the Cul4-based ubiquitin ligase is required for its function in histone H3K9 trimethylation in Neurospora. Furthermore, we demonstrated that Cul4 and DDB1 are associated with the histone methyltransferase DIM-5 protein in vivo. Together, these results suggest a mechanism for DNA methylation control that may be applicable in other eukaryotic organisms.

Cytosine methylation is the most common form of DNA modification in eukaryotes. Although absent in the model organisms, such as Saccharomyces cerevisiae and Caenorhabditis elegans, and observed rarely in Drosophila melanogaster, cytosine methylation is ubiquitous in most eukaryotes, including the filamentous fungus Neurospora crassa (1,2). DNA methylation occurs primarily on CpG dinucleotides in mammals but is observed on other sites in fungi and plants (symmetric or asymmetric) (3). In plants and filamentous fungi, cytosine methylation is found mostly in transposons, repeats, and selected cellular genes associated with transposon control and gene silencing (4). However, very recent studies have demonstrated that DNA methylation in plants is not restricted to transposon control and gene silencing (5,6). DNA methylation of mammalian genomes plays important roles in development, X chromosome inactivation, gene imprinting, and repression of retrotransposons (7). Cytosine methylation is essential in mammals, and defects in DNA methylation can result in severe human diseases, including Angelman syndrome and Prader Willi syn-drome (8). Hypermethylation of tumor-suppressor gene promoters and hypomethylation of the genome are hallmarks of cancerous cells (9,10).
The RNA interference pathway directs de novo DNA methylation in plants (11,12). In animals, the involvement of non-coding RNA in silencing and imprinting of Igf2r indicates that RNA functions in de novo DNA methylation (13). It was recently found that Piwi-interacting RNA directs DNA methylation in mammalian germ cells (14). However, the RNA interference machinery is dispensable for DNA methylation in Neurospora (15), indicating that RNA-directed DNA methylation is not universal.
The histone H3K9 3 methyltransferase DIM-5 (Table 1) is required for all known DNA methylation in Neurospora (16,17). Similarly, histone methylation also directs DNA methylation in plants and mammals (18,19). Methylated H3K9 can be recognized by the heterochromatin protein 1, resulting in the recruitment of HP1 (20,21). In Neurospora, HP1 (Table 1) serves as an adaptor between methylated H3K9 and the DNA methyltransferase DIM-2 (Table 1) (22)(23)(24). The demethylation of H3K9 also controls genic DNA methylation in Arabidopsis thaliana (25), and unmethylated H3K4 is involved in de novo DNA methylation in mammals (26). Together, these studies demonstrate that modifications of histone H3 control DNA methylation in fungi, plants, and mammals. In N. crassa, short TpA-rich segments (27) or synthesized (TAAA) n or (TTAA) n segments (28) can trigger de novo DNA methylation, which suggests that A:T-rich sequences may be DNA methylation signals. However, how these signals are recognized and relayed to H3K9 trimethylation is not understood.
In fission yeast, Cullin4 and Rik1 are required for H3K9 methylation. Rik1 associates with Cul4 to form an E3 ubiquitin ligase that regulates the localization of H3K9 methyltransferase Clr4 to heterochromatic regions, contributing to heterochromatin assembly and maintenance (29). Rik1 protein is not found in other eukaryotes, but a homologous protein, DDB1, is evolutionarily conserved in eukaryotes. The Cul4based ubiquitin ligases share an architecture consisting of Cul4, DDB1 (Table 1), Rbx1, and a variable substrate-recruiting subunit named DCAF to target different substrates for ubiquitination (30 -33). Ubiquitination is involved in cell cycle regulation, DNA replication licensing, DNA repair, and gene expression processes (34,35). Intriguingly, the Cul4-based ubiquitin ligases may also influence histone modification in multicellular organisms (33,36). These results suggest that Cul4-DDB1 ubiquitin ligases play important roles in epigenetic control in higher eukaryotes. However, it is not known whether the Cul4-based ubiquitin ligase is required for DNA methylation in eukaryotic organisms.
In this study, we show that the Neurospora ubiquitin ligase components Cul4 and DDB1 are essential for DNA methylation. In addition, our results suggest that the E3 ligase activity of the Cul4-based ubiquitin ligase is required for its function in histone H3K9 trimethylation in Neurospora. Importantly, both Cul4 and DDB1 associated with the histone methyltransferase DIM-5. Together, these results suggest a mechanism for DNA methylation control that may be applicable in other eukaryotic organisms.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-87-3 (bd a) was used as the wild-type strain in this study. The bd ku70 RIP strain generated previously (37) was used as the host strain for introducing the different hph knock-out cassettes. The newly created knockout mutants in the bd ku70 RIP host strain include the ddb1 KO , cul4 KO , dim-5 KO , dim-2 KO , and hpo KO strains. The 301-6 (bd, his-3, A) and the dim-5 KO (bd, his-3) were used as the host strain for the his-3-targeting constructs.
Generation of ddb1, cul4, dim-5, dim-2, and hpo Knock-out Mutants by Gene Replacement-We used the knock-out procedures described by Colot et al. (39) to delete open reading frames (ORFs) of genes knocked out in this study. For cul4, ddb1, dim-5, dim-2, and hpo gene knockouts, the entire ORF knock-out cassette of each gene was created by PCR. The PCR fragment containing the hph gene replacement cassette was introduced into the bd, ku70 RIP strain by electroporation. The transformants with hph at the targeted gene locus were passaged on minimal slants with hygromycin for at least five generations, and then conidia were plated at different dilutions on plates containing hygromycin to obtain the homokaryon knock-out strains of these genes. PCR analyses confirmed the sequences of the homokaryon knock-out strains.
DNA Methylation Analysis-DNA methylation was analyzed by restriction enzymes DpnII and BfuCI (Sau3AI), followed by PCR. Genomic DNA (200 ng) was digested with DpnII or BfuCI (Sau3AI) in a 50-l volume. Control samples had no enzyme. The PCR primers of ku70, -, and ⌿63 regions are listed in Table 2. PCR was performed using 1 l of the digested sample as a template in 50 l, with cycles of 5 min at 94°C, followed by 30 -35 cycles at 94°C (30 s), 53°C (30 s), and 72°C (1 min). PCR products were resolved by electrophoresis on 2% agarose gels. Each experiment was independently performed at least three times.
ChIP Analysis-The ChIP assay protocol used here was the same as previously described (40). Briefly, the tissues were fixed in minimal medium containing 1% formaldehyde for 10 min at room temperature. The chromatin immunoprecipitation was performed using 5 l of antibody to trimethylated H3 Lys 9 (07-442; Upstate Biotechnology). After DNA extraction, pellets were washed with 70% EtOH and resuspended in 10 l of double-distilled H 2 O, and 0.7 l of DNA solution was used for PCR. The primers for ku70, -, ⌿63, and hH4 regions are listed in    ). Different PCR cycles were tested to ensure that DNA amplification was within the exponential amplification range. PCR products were resolved by electrophoresis on 2% agarose gels. Each experiment was independently performed at least three times, and immunoprecipitation with anti-Myc antibody was used as the negative control.
Creation of Co-transformation Strains-The expression vector, pqa-Myc-His, was constructed previously (41). An inducible qa-2 promoter controls the expression of the fusion protein.
To make the his-3-targeting construct, a PCR fragment containing the entire ORF and the 3Ј-untranslated region (from ddb1, cul4, or dim-5 genes) was cloned into pqa-Myc-His vector. To create the pqa-3FLAG vector, the 5Myc-His 6 sequence on pqa-Myc-His was replaced with the 3FLAG sequence. The ORF/3Ј-untranslated region fragment was cloned into the pqa-3FLAG vector. The Myc-tagged constructs were transformed into the 301-6 (bd, his-3, A) strain at the his-3 locus. Western blot analyses using a monoclonal c-Myc antibody (9E10, Santa Cruz Biotechnology) were performed to identify the positive transformants. To study the interaction of Myc-tagged protein/ FLAG-tagged protein, the FLAG-tagged construct was introduced into the transformants expressing various Myc-tagged proteins by co-transformation with pBT6 (containing the benomyl resistance gene, obtained from the Fungal Genetic Stock Center). Second round Western blot analyses using a monoclonal FLAG antibody (F3165-5MG, Sigma) were performed to identify the positive co-transformants. In positive co-transformants, immunoprecipitation using c-Myc antibody or FLAG antibody was performed to test for an interaction between Myc-tagged protein and FLAG-tagged protein. Protein extraction, quantification, Western blot analysis, and immunoprecipitation assays were performed as previously described (38,42).
Creation of the cul4K863R Knock-in Strains-To create the knock-in cassette with the neddylation lysine residue mutation of the cul4 gene, the plasmid pqa-3FLAG-Cul4 was used as the template for in vitro mutagenesis to mutate the Cul4 Lys 863 to Arg. To generate the knock-in cassette, the hph gene is inserted downstream from the cul4 3Ј-untranslated region and flanked by the entire cul4 gene (with the mutation) and 1-kb genomic DNA downstream from cul4 (see Fig. 6B). The cassette was then transformed into the bd ku70 RIP strain to select for hph-resistant transformants. Afterward, the homokaryotic cul4K863R knock-in strains were obtained by microconidia purification and confirmed by DNA sequencing.
Generation of Antiserum against DDB1-GST-DDB1 (containing DDB1 amino acids 1056 -1159) fusion protein was expressed in BL21 cells, and the soluble recombinant protein was purified and used as the antigen to generate rabbit polyclonal antiserum as described previously (41).

Repeat-induced Point Mutation (RIP)-triggered Cytosine
Methylation of ku70 RIP Allele-RIP is a genome defense process that detects DNA duplications during meiosis and induces the DNA sequence alteration from G:C to A:T in Neurospora (44). In addition, cytosine methylation is typically associated with the sequence targeted by RIP (45). To investigate DNA methylation in Neurospora, we used a previously created ku70 RIP (NCU08290) strain (37,46). DNA sequencing of the endogenous ku70 locus in this strain showed that the endogenous ku70 open reading frame was mutated by many G:C to A:T mutations, resulting in multiple premature stop codons (Fig. 1A). To test the DNA methylation in the ku70 RIP allele, genomic DNA of the ku70 RIP and a wild-type strain was digested with the isoschizomers DpnII or BfuCI (Sau3AI). Each restriction enzyme digests unmethylated GATC sites, but only DpnII cuts at sites when the cytosine is methylated. As shown in Fig. 1A, the wild-type ku70 ORF includes six DpnII/BfuCI sites.
As shown in Fig. 1A, three pairs of primers were designed for PCR of the ku70 ORF region; each pair of primers covers a 0.6 -0.9-kb fragment with at least one DpnII-BfuCI site. The genomic DNA of the wild-type and ku70 RIP strains was digested with DpnII or BfuCI, and fragments were amplified by PCR. As shown in Fig. 1B, no PCR products were detected for ku70 gene of the wild-type DNA after digestion with DpnII or BfuCI, indicating that the ku70 locus was unmethylated. A PCR control of a gene fragment (NUC03881) without the DpnII site showed that the DNA samples were not degraded after digestion (Fig.  1B, bottom). For the ku70 RIP strain, however, the digestion of ku70 RIP allele showed a different result; no PCR products were detected in the DpnII digestion sample, whereas amplification of BfuCI-digested DNA resulted in amounts of products similar to those of untreated samples (genomic DNA, Input lane), indicating that cytosines in the GATC sites are methylated in the ku70 RIP allele. These data indicate that RIP not only altered the DNA sequence of the ku70 gene, but also triggered its methylation.
The ku70 RIP strain exhibits normal growth development and the same circadian rhythms as a wild-type strain (37); however, mutation of the Neurospora ku70 gene causes loss of nonhomologous recombination, and thus the mutant strain has a nearly 100% homologous recombination rate (39,46). Thus, the ku70 RIP strain was used as the host strain for generating the gene knock-out mutant with a hygromycin resistance gene (hph) cassette replacement in this study.
Knock-out of the Neurospora cul4, ddb1, dim-2, hpo, and dim-5 Genes-Using the Cul4 sequence of fission yeast as a query, we used BLAST to search the Neurospora genome; we found that the hypothetical protein NCU00272 is a homolog of Cul4. Alignment with other Cul4 homologs showed that the Neurospora Cul4 is more similar to Cul4 of higher eukaryotes (Arabidopsis, human, Drosophila) than to the protein from fis-Cul4 and DDB1 Are Essential for DNA Methylation FEBRUARY 12, 2010 • VOLUME 285 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4357 sion yeast (Fig. 2, A and B). The most conserved sequences reside in the N-terminal extension sequence (␤1), helix2 (H2), and helix5 (H5) of the first cullin repeat (Fig. 2C). These regions are required for the interaction with DDB1 protein (31).
To test directly whether the Neurospora Cul4 and DDB1 are required for cytosine methylation, we deleted the Neurospora cul4 gene and ddb1 gene, respectively, by replacing their ORFs with a hygromycin resistance gene (hph) cassette in the ku70 RIP background strain. Several independent homokaryotic cul4 and ddb1 knock-out strains were obtained by microconidia purification. PCR analyses confirmed the integration of the knock-out cassette at the endogenous cul4 locus or ddb1 locus, and no cul4 or ddb1 ORF signals were detected in these knock-out mutants. In addition, we also created homokaryotic dim-2, dim-5, and hpo knock-out strains. The disruption of these genes was previously shown to abolish all DNA methylation in Neurospora.
Neurospora Cul4 and DDB1 Are Required for DNA Methylation-To test the roles of Cul4 and DDB1 in DNA methylation, genomic DNA of the wild-type strain and the ku70 RIP , cul4 KO , ddb1 KO , dim-2 KO , dim-5 KO , and hpo KO mutants was digested with BfuCI (Sau3AI) or DpnII. Digested and undigested DNA fragments were used as templates for PCR. As shown in Fig. 3A, BfuCI failed to cut ku70 RIP genomic DNA of the ku70 RIP strain. In contrast, no PCR products were detected in the cul4 KO and ddb1 KO mutants, indicating the loss of DNA methylation at the ku70 RIP locus. Similarly, loss of DNA methylation was observed in the dim-2 KO , dim-5 KO , and hpo KO mutants. It was previously shown that most of methylated regions are relics of transposons in Neurospora, which were inactivated by RIP. To confirm the role of Cul4 and DDB1, we also examined the methylation status of several representative methylated regions, including the relics of RIP (and 63). Compared with the hypermethylation of DNA in the wild-type and ku70 RIP strains, theregion (Fig. 3B) and 63 region (Fig. 3C) showed complete loss of DNA methylation in cul4 KO , ddb1 KO , dim-2 KO , dim-5 KO , and hpo KO mutants. As listed in Table 3, all previously examined cytosine methylation regions in Neurospora were lost in the cul4 KO or ddb1 KO strains, indicating that Cul4 and DDB1, like DIM-2, DIM-5, and HP1, play an essential role for DNA methylation in Neurospora.
Deletion of cul4 or ddb1 Causes Growth and Developmental Defects-To investigate the roles of Cul4 and DDB1 in Neurospora development, we compared the growth and developmental phenotypes among these knock-out mutants. In addition to the loss of cytosine methylation at the ku70 RIP , -, and 63 loci, the cul4 KO and ddb1 KO strains also exhibited a variety of developmental abnormalities. When inoculated on plates, ddb1 and cul4 mutants showed the same growth patterns as dim-5 and hpo mutants; these strains exhibited dense, cauliflower-like growth patterns with abnormal hyphae and asexual spores (Fig.  4A). In contrast, the growth pattern of the dim-2 knock-out strain was nearly the same as that of the wild-type strain (Fig.  4A). When their growth rates were compared by race tube assays, the cul4 KO , ddb1 KO , dim-5 KO , and hpo KO mutants exhibited slower growth rates than those of the wild-type and ku70 RIP strains (Fig. 4B), whereas the growth rate of the dim-2 KO mutant was similar to that of the wild-type strain (Fig.  4B). The growth phenotypes of our dim-2 KO , dim-5 KO , and hpo KO mutants are the same as those described by previous studies (16,23,24). Therefore, although each of these proteins is essential for DNA methylation, their phenotypes fall into two categories. These results suggest that Cul4, DDB1, DIM-5, and HP-1 participate in another process in addition to their essential role in DNA methylation, whereas DIM-2 is only involved in DNA methylation in Neurospora. Our results are consistent with previous results (24) showing that the loss of DNA methylation in dim-2 mutants did not result in noticeable growth and developmental defects in Neurospora. DDB1 and Cul4 Are Required for Trimethylation of Histone H3-Trimethylation at Lys 9 of histone H3 is a mark for DNA methylation in Neurospora crassa (17). Based on the phenotypes of the knock-out mutants described above, we hypothesized that DDB1 and Cul4 might affect methylation of histone H3K9 like DIM-5. To test the role of DDB1 and Cul4 in H3K9 trimethylation, we examined H3K9 trimethylation levels by chromatin immunoprecipitation (ChIP). Chromatin samples were immunoprecipitated with antibody against trimethylated Lys 9 of histone H3 and analyzed by PCR with probes targeted to methylated DNA regions. As shown in Fig. 5A, trimethylated H3K9 was associated with the methylated ku70 RIP region in the ku70 RIP strain, whereas no trimethylated H3K9 was detected in the unmethylated ku70 region of the wild-type strain, indicating a correlation between levels of DNA and histone methylation. The deletion of dim-2 did not reduce levels of trimethylated H3K9 at demethylated ku70 RIP region (Fig. 5A), but the deletion of histone methyltransferase gene dim-5 caused a complete loss of trimethylated H3K9 at this DNA region (Fig. 5A). Similarly, trimethylated H3K9 at the ku70 RIP region was abolished in both cul4 KO and ddb1 KO strains (Fig. 5A). These data demonstrated that Cul4 and DDB1 are required for trimethylation of H3K9 in Neurospora and further confirmed that trimethylation of H3K9 precedes DNA methylation.
We also examined the levels of trimethylated H3K9 atand 63 regions in the wild type, ku70 RIP strain, and knock-out mutants. ChIP analysis showed that trimethylated H3K9 in the ku70 RIP and dim-2 KO strains was observed at wild-type levels (Fig. 5, B and C). In contrast, the loss of H3K9 trimethylation was observed in the cul4 KO , ddb1 KO , and dim-5 KO strains (Fig.  5, B and C). Taken together, these data indicate that Neurospora  FEBRUARY 12, 2010 • VOLUME 285 • NUMBER 7 DDB1 and Cul4 are required for trimethylation of histone H3K9 at methylated DNA regions.

Cul4 and DDB1 Are Essential for DNA Methylation
To investigate the global effect of Cul4 and DDB1 on trimethylated H3K9 in Neurospora genome, we performed Western blot analyses of histones in the wild-type, cul4 KO , ddb1 KO , and dim-5 KO strains using antibodies against histone H3 or trimethylated H3K9. As shown in Fig. 5D, all strains had similar levels of histone H3. However, in contrast to the robust trimethylated H3K9 signal in the wild-type strain, very little trimethylated H3K9 was detected in the cul4 KO , ddb1 KO , and dim-5 KO strains. Together, these results strongly suggest that Cul4 and DDB1 are required for most if not all H3K9 trimethylation in Neurospora.
The Requirement of Cul4 Ubiquitin Ligase Activity in H3K9 Trimethylation-The cullin proteins can be modified by Nedd8 on a conserved lysine residue residing in the C-terminal domain, which stimulates their ubiquitin E3 activity (48). To determine whether the ubiquitin ligase activity of the Cul4 complex is required for H3K9 trimethylation, we created a knock-in cul4 mutant in which the predicted neddylation site Lys 863 is changed to Arg. Two FLAG-tagged cul4 constructs, which contained the Cul4 ORF or Cul4 ORF with Lys 863 3 Arg mutation, were transformed into a wild-type strain. The expression of the FLAG-Cul4 or FLAG-Cul4K863R in these two strains was confirmed by Western blot analysis using the FLAG antibody. As shown in Fig. 6A, two bands of FLAG-Cul4 protein were detected in the wild-type FLAG-Cul4 transformant, whereas only one band of FLAG-Cul4K863R was detected in the wild type FLAG-Cul4K863R transformant, confirming that the Lys 863 is indeed the neddylation site.
To obtain a cul4K863R mutant, a knock-in cassette with the K863R mutation was made (Fig. 6B). This cassette was then transformed into a bd ku70 RIP strain, and hph-resistant transformants were selected. PCR analysis confirmed the integration of the knock-in cassette at the endogenous cul4 locus. The homokaryotic strain with the cul4K836R mutation (in bd ku70 RIP background) was obtained by microconidia purification. DNA sequencing of the endogenous cul4 gene confirmed the mutated cul4K863R (codon AAG for Lys changed to codon CGG for Arg) in these strains. As shown in Fig. 6C, although the cul4K863R strains grew slightly faster than the cul4 KO strain, they exhibited the cauliflower phenotype on plates that are very similar to those of the cul4 KO strains.
ChIP assays were performed to examine H3K9 trimethylation levels in the cul4K863R mutant. As shown in Fig. 6D, the H3K9 trimethylation were abolished at ku70 RIP , -, and 63 regions in the cul4K863R strain, similar to those of the cul4 KO strain. This result strongly suggests that the activity of Cul4based E3 ubiquitin ligase is required for histone H3K9 trimethylation.
DDB1 and Cul4 Interact with DIM-5 in Neurospora-Our results suggest that Cul4 is a part of an ubiquitin ligase complex that mediates the histone H3K9 trimethylation and DNA methylation. To confirm this, we made a construct in which the Cul4 ORF under the control of the QA-inducible promoter (49). Five copies of the c-Myc epitope and six histidine residues (41) were inserted at the N terminus of the Cul4 ORF to facilitate the detection of the expression of Myc-Cul4 using a c-Myc monoclonal antibody (9E10). The construct was transformed into the wild-type strain and into the cul4 KO strain. As shown in Fig. 7A, Myc-His-Cul4 was expressed in the wild-type strain and the cul4 KO strains in the presence of QA. When the cul4 KO qa-Myc-His-Cul4 strain was grown on a minimal plate containing QA, the growth rate and the production of conidia and aerial hyphae were very similar to those of the wild-type strain and different from those of the cul4 KO strain (Fig. 7B). The DNA methylation in the cul4 KO qa-Myc-His-Cul4 strain was also similar to that of the wild-type strain (Fig. 7C). These data indicate that Myc-Cul4 functions as the endogenous Cul4 protein in Neurospora.
Our ChIP results and Western blot analyses of trimethylated H3K9 revealed that DDB1 or Cul4 are required for histone H3K9 trimethylation. To understand how DDB1 and Cul4 mediate H3K9 trimethylation and DNA methylation in Neurospora, we used immunoprecipitation to examine the interactions of various proteins. As the key components of Cul4-based ubiquitin ligase, DDB1 and Cul4 form a complex and recruit different adaptor proteins, DCAFs, to recruit protein substrates. Using immunoprecipitation, we found that FLAG-Cul4 and Myc-DDB1 interact in Neurospora (Fig. 8A). These data demonstrated that two key components of ubiquitin ligase indeed interact with each other in vivo. We then examined whether histone methyltransferase DIM-5 physically interacts with Cul4 protein in Neurospora. To evaluate the Cul4/ DIM-5 interaction in vivo, we introduced a construct (pqaMyc-His-DIM-5) to allow expression of Myctagged DIM-5 (under the control of a quinic acid-inducible promoter) into a transformant expressing FLAG-tagged Cul4 by co-transformation with pBT6 (containing the benomyl resistance gene, obtained from the Fungal Genetic Stock Center). Western blot analyses confirmed the expression of Myc-DIM-5 and FLAG-Cul4 in extracts from the co-transformants and the transformant with FLAG-Cul4 only.
Immunoprecipitation using a c-Myc monoclonal antibody showed that Myc-DIM-5 specifically interacted with FLAG-Cul4 in Neurospora (Fig. 8B). By Western blot analysis with anti-FLAG antibody, two FLAG-Cul4 bands were detected in total extract. The minor band migrated ϳ10 kDa slower than the major band, suggesting that it is the neddylated FLAG-Cul4 (Fig. 8B). As shown in Fig. 8B, Myc-DIM-5 precipitated most of neddylated Cul4 and little unneddylated Cul4. Because the neddylation of Cul4 is required for its function in H3K9 trimethylation, this result suggests  that Cul4-based E3 activity may regulate the function of DIM-5 in H3K9 trimethylation.
We then asked whether DDB1 can also interact with the histone methyltransferase DIM-5 in Neurospora. In transformants that express both FLAG-tagged DDB1 and Myc-tagged DIM-5, we can also detect the interaction between these two proteins.
To further confirm this interaction, we transformed qa-Myc-His-DIM-5 into the dim-5 KO strain. As shown in Fig. 8C, when the dim-5 KO qa-Myc-His-DIM-5 strain was grown on a minimal plate containing QA, the growth rate and the production of conidia and aerial hyphae were very similar to those of the wildtype strain and different from those of the dim-5 KO strain, demonstrating that Myc-DIM-5 can complement the function of the endogenous DIM-5 protein in the dim-5 KO strain. An immunoprecipitation assay showed that the endogenous DDB1 protein (detected by a polyclonal antibody specific for Neurospora DDB1) was present in the c-Myc immunoprecipitates, confirming that DDB1 and Myc-DIM-5 form a complex in vivo (Fig. 8D). Together, these results suggest that DDB1 and Cul4 form an E3 ligase complex that recruits DIM-5 to methylate histone H3K9, which then triggers the DNA methylation in this region.

DISCUSSION
Recent studies demonstrated that three proteins, DIM-2 (24), DIM-5 (16), and HP1 (23), are essential for DNA methylation in Neurospora. Although Cul4 and Rik1 in fission yeast regulate the localization of H3K9 methyltransferase Clr4 to heterochromatic regions (29), it is not clear whether they are also involved in DNA methylation in multicellular organisms. In our study, we showed that DDB1 and Cul4 are essential for DNA methylation in the Neurospora. Based on the comparison of the phenotypes of cul4 KO and ddb1 KO mutants with those of dim-2 KO , dim-5 KO , and hpo KO , we propose that DDB1 and Cul4 proteins function in an early step of DNA methylation by recruiting DIM-5 and regulating H3K9 trimethylation. To determine the roles of the cul4 and ddb1 gene products, we demonstrated that DDB1 and Cul4 interact with DIM-5 to form a complex for H3K9 trimethylation and then trigger the DNA methylation on these chromatin regions.
DDB1, Cul4, and DIM-5 Form a Complex in Neurospora-Rik1 plays a key role in heterochromatin formation in fission yeast; it recruits histone methyltransferase Clr4 for H3K9 methylation on heterochromatic regions (50). Although Neurospora, like the higher eukaryotes, does not have a rik1 gene, we showed that the Neurospora DDB1, which is related to Rik1, and its partner Cul4 are essential for DNA methylation through effects on the function of histone H3K9 methyltransferase DIM-5. We showed that Cul4 and DDB1 form a complex that interacts with DIM-5 (Fig. 8, B and D). These results suggest that Cul4 and DDB1 recruit the histone methyltransferase DIM-5 to chromatin to methylate histone H3K9. Following the trimethylation of H3K9, the HP1-DIM-2 complex will recognize these marks and trigger the DNA methylation (22).
The Requirement of Ubiquitin Ligase Activity of Cul4 in H3K9 Trimethylation-To determine if the ubiquitin ligase activity of Cul4 is involved in H3K9 trimethylation, we created the cul4K863R knock-in mutant in which Cul4 neddylation is abolished. This mutant exhibited trimethylated H3K9 (H3K9me3) defects similar to those of the cul4 KO strain, indicating that the ubiquitin ligase activity of Cul4 is required in H3K9 trimethylation. Further supporting this conclusion, we found that DIM-5 preferentially interacts with functionally active neddylated Cul4 species, suggesting that the activity of Cul4 ubiquitin ligase is important for the recruitment of DIM-5 to chromatin. Although how the Cul4-based E3 ligase activity is involved in this process is not clear, several lines of evidence suggest that ubiquitin ligase activity is critical for epigenetic regulation of chromatin structure (33, 36, 51). As a response to DNA damage, Cul4-DDB1-ROC1 ubiquitin ligase mediates the H3 and H4 ubiquitylation that weakens the interaction between histones and DNA and facilitates the recruitment of repair proteins to damaged DNA (36). Cul4-DDB1 associates with histone H3 methylated at Lys 4 , Lys 9 , and Lys 27 and regulates H3 methylation in mammalian cells (33). Thus, it is likely that the ubiquitination of the chromatin-associated protein may facilitate the recruitment of DIM-5. In Neurospora, Honda and Selker (22) found that a direct interaction between DIM-2 and HP1 is required for DNA methylation and that a stable DIM-2-HP1 complex can bind to the H3K9me3 mark to methylate cytosines. Whether the components of Cul4-DDB1 ubiquitin ligase directly interact with the DIM-2-HP1 complex is not clear. To determine the precise function of the E3 ligase activity in this process, the substrate(s) and DCAF homologs must be identified in the future. DDB1 and Cul4 are highly conserved in eukaryotes. In mammalian cul4 or ddb1 knockdown cells, the trimethylation of H3K4, H3K9, and H3K27 is reduced, and Cul4-DDB1 appears to specifically associate with trimethylated H3K9 and H3K27 (33). The histone H3K9 trimethylation directs all known DNA methylation in Neurospora (16,17). Similarly, histone methyl-ation also directs DNA methylation in plants and mammals (18,19). Furthermore, the demethylation of H3K9 also controls genic DNA methylation in A. thaliana (25), and unmethylated   H3K9. A, immunoprecipitation assays reveal an association between Myc-DDB1 and FLAG-Cul4. Wild-type (wt) strain and wild-type with expressing Myc-DDB1 strain were used as negative controls. The FLAG antibody was used for immunoprecipitation (IP), followed by Western blot analyses using the FLAG antibody and c-Myc antibody. B, immunoprecipitation assays reveal an association between Myc-DIM-5 and FLAG-Cul4. Myc-DIM-5 interacts with neddylated rather than unneddylated FLAG-Cul4. Wild-type strain and wildtype with expressing FLAG-Cul4 strain were used as negative controls. The extracts from wild-type, wild-type FLAG-Cul4, and wild-type Myc-DIM-5/ FLAG-Cul4 co-transformation strains were immunoprecipitated using c-Myc antibody, followed by Western blot analyses using the FLAG antibody and c-Myc antibody. C, Myc-tagged DIM-5 can complement the growth and developmental defects of the dim-5 KO strain (30°C, 32 h). D, Myc-DIM-5 binds to the endogenous DDB1, as shown by immunoprecipitation using c-Myc antibody in the dim-5 KO Myc-DIM-5 strain. The wild-type and wild-type with expressing Myc-tagged MCB (the regulatory subunit of cAMP-dependent protein kinase) strain were used as negative controls. The specificity of DDB1 antibody was indicated by the ddb1 KO strain.
H3K4 is involved in de novo DNA methylation in mammals (26). Together, these studies demonstrate that modifications of histone H3 control DNA methylation in fungi, plants, and mammals. These results raise the possibility that ubiquitin ligase components DDB1 and Cul4 might also be important for DNA methylation in mammals and plants as our data demonstrated in Neurospora.