|Home | About | Journals | Submit | Contact Us | Français|
CXXC finger protein 1 (CFP1) binds to unmethylated CpG dinucleotides and is required for embryogenesis. CFP1 is also a component of the Setd1A and Setd1B histone H3K4 methyltransferase complexes. Murine embryonic stem (ES) cells lacking CFP1 fail to differentiate, and exhibit a 70% reduction in global genomic cytosine methylation and a 50% reduction in DNA methyltransferase (DNMT1) protein and activity. This study investigated the underlying mechanism for reduced DNMT1 expression in CFP1-deficient ES cells. DNMT1 transcript levels were significantly elevated in ES cells lacking CFP1, despite the observed reduction in DNMT1 protein levels. To address the posttranscriptional mechanisms by which CFP1 regulates DNMT1 protein activity, pulse/chase analyses were carried out, demonstrating a modest reduction in DNMT1 protein half-life in CFP1-deficient ES cells. Additionally, global protein synthesis was decreased in ES cells lacking CFP1, contributing to a reduction in the synthesis of DNMT1 protein. ES cells lacking CFP1 were found to contain elevated levels of phosphorylated eIF2α, and an accompanying reduction in translation initiation as revealed by a lower level of polyribosomes. These results reveal a novel role for CFP1 in the regulation of translation initiation, and indicate that loss of CFP1 function leads to decreased DNMT1 protein synthesis and half-life.
Establishment of appropriate cytosine methylation patterns is crucial for embryonic development and survival. Mice deficient for DNA methyltransferase 1 (DNMT1), the maintenance DNMT, or the de novo methyltransferase DNMT3b die during embryogenesis (Li et al., 1992; Okano et al., 1999). Mice lacking the DNMT3a de novo DNMT die within 4 weeks of birth (Okano et al., 1999). In addition, aberrant cytosine methylation patterns observed in cancer cells result in abnormal silencing of genes involved in regulating cell cycle progression and cell survival (Jones and Baylin, 2002; Rhee et al., 2002).
DNMT1 mRNA is ubiquitously expressed and is coupled to the cell cycle, with peak expression occurring late in S phase and persisting through G2 phase (Robertson et al., 2000). A number of transcriptional (Rouleau et al., 1995; Bigey et al., 2000), posttranscriptional (Szyf et al., 1985; Liu et al., 1996; Detich et al., 2001), and posttranslational (Liu et al., 1996; Ding and Chaillet, 2002; Agoston et al., 2005) regulatory mechanisms ensure appropriate DNMT1 expression. Posttranslational regulatory mechanisms include allosteric activation involving interaction between the zinc-binding and catalytic domains within the DNMT1 protein that is stimulated by methylated DNA (Bacolla et al., 1999; Fatemi et al., 2001; Margot et al., 2003); protein–protein interactions resulting in inhibition of DNMT1 activity in vitro, as seen with retinoblastoma protein (Pradhan and Kim, 2002); protein–protein interactions resulting in stimulation of DNMT1 activity in vitro, as observed with p53 (Estève et al., 2005) or the viral oncoproteins adenovirus E1A and human papillomavirus E7 (Burgers et al., 2007); and protein–polymer interactions with ADP-ribose polymers that inhibit DNMT1 activity in vitro (Reale et al., 2005).
Mammalian CXXC finger protein 1 (CFP1), previously denoted CpG-binding protein, is an 88kDa transcriptional activator that is ubiquitously expressed and possesses a unique DNA-binding specificity for unmethylated CpG dinucleotides (Voo et al., 2000). Homologs of CFP1 are found in Saccharomyces cerevisiae, Saccharomyces pombe, Xenopus laevis, Drosophila melanogaster, Danio rerio, and Caenorhabditis elegans (Voo et al., 2000). CFP1 contains a highly conserved cysteine-rich CXXC domain that is necessary and sufficient for DNA binding (Voo et al., 2000; Lee et al., 2001). Interestingly, the CXXC domain of CFP1 is not present in homologs found in Caenorhabditis elegans or Saccharomyces cerevisiae, organisms that lack cytosine methylation (Voo et al., 2000), suggesting that the DNA-binding function of CFP1 evolved in organisms that utilize genomic cytosine methylation as an epigenetic mark. CFP1 additionally contains two plant homeodomain fingers, a domain found in many chromatin-associated proteins (Aasland et al., 1995; Voo et al., 2000; Shi et al., 2007).
Mouse embryos lacking the CXXC1 gene exhibit peri-implantation lethality between 4.5 and 6.5 days postcoitus (Carlone and Skalnik, 2001). This timing coincides with global epigenetic reprogramming that involves erasure of existing cytosine methylation followed by de novo establishment of cytosine methylation patterns (Reik et al., 2001; Li, 2002). CFP1-deficient embryonic stem (ES) cells fail to differentiate in vitro upon removal of leukemia inhibitory factor from the culture medium and exhibit a 70% reduction in global cytosine methylation (Carlone et al., 2005), while zebrafish treated with CFP1 antisense morpholino oligonucleotides exhibit decreased primitive hematopoiesis and a 60% reduction in global cytosine methylation (Young et al., 2006). The reduction in global cytosine methylation observed in CFP1-null ES cells occurs at repetitive elements, single copy genes, and imprinted loci (Carlone et al., 2005). Maintenance DNMT activity is reduced about twofold in the absence of CFP1, while de novo methylation activity is comparable to that observed in wild-type ES cells (Carlone et al., 2005). Taken together, these results indicate that CFP1 is required for maintenance of global cytosine methylation patterns that are essential for embryonic development.
CFP1 is additionally a component of the mammalian Setd1A and Setd1B histone H3K4 methyltransferase complexes (Lee and Skalnik, 2005; Lee et al., 2007). The yeast CFP1 homolog Spp1 is required for appropriate Set1-mediated trimethylation of histone H3K4 in S. cerevisiae and for all Set1-mediated histone H3K4 methylation in S. pombe (Roguev et al., 2003; Schneider et al., 2005). In contrast, murine ES cells lacking CFP1 carry altered patterns of histone modifications consistent with decreased 5-methylcytosine and decreased heterochromatin, including increased global histone H3K4me2 methylation and decreased histone H3K9me2 (Lee and Skalnik, 2005). In addition, an inappropriate increase in histone H3K4me3 is observed in ES cells lacking CFP1 upon induction of differentiation (Lee and Skalnik, 2005).
We previously demonstrated the importance of CFP1 in maintaining appropriate histone modifications and cytosine methylation throughout the genome (Carlone et al., 2005; Lee and Skalnik, 2005). Experiments were conducted to determine the mechanisms underlying decreased DNMT1 expression in the absence of CFP1. These studies reveal an important role for CFP1 in the posttranscriptional regulation of DNMT1 protein expression.
Murine ES and human embryonic kidney (HEK293) cell lines and culture conditions were described previously (Carlone et al., 2005).
Total RNA was isolated from wild-type and CFP1-null ES cell lines using TRIreagent (Molecular Research Center, Cincinnati, OH) per the manufacturer's recommendations. Twenty-five μg of RNA was separated by electrophoresis on a 1% formaldehyde agarose gel and transferred to a nylon membrane as previously described (Voo et al., 2000). DNMT1 and β-actin transcripts were detected by hybridization using a probe derived from murine DNMT1 cDNA (generously provided by Dr. J. Richard Chaillet, University of Pittsburgh, Pittsburgh, PA) and a β-actin riboprobe generated from pTRI-Actin-Mouse template using a MAXIscript in vitro transcription kit (Ambion, Austin, TX), respectively. Blot hybridization and washes were carried out as previously described.
For quantitative reverse transcription–polymerase chain reaction (qRT-PCR) analysis, total RNA was isolated from the following murine ES cell lines: wild-type, CFP1-null, CFP1-rescue, and CFP1-vector. The CFP1-rescue and CFP1-vector cell lines were generated by stably introducing a murine CFP1 cDNA expression construct or parental vector into CFP1-null ES cells, respectively, as previously described (Carlone et al., 2005). Five ng of total RNA was reverse transcribed using random primers and SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, CA) per the manufacturer's protocol. cDNA template generated from the reverse transcription reactions was combined with TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA) and TaqMan gene expression assays containing premixed primer/probe pairs for murine DNMT1 (Mm1151064_m1) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (murine endogenous control 4352932E). Quantitation of DNMT1 mRNA levels relative to GAPDH was carried out by real-time PCR analysis using universal cycling conditions (2min at 50°C, 10min at 95°C, followed by 40 cycles of 15s at 95°C and 1min at 60°C). Data were collected on a 7500 Real-Time PCR System from Applied Biosystems and compared by the ΔΔCT method using 7500 System SDS software from Applied Biosystems.
A fragment of murine DNMT1 cDNA (349–825bp) was amplified by PCR from murine DNMT1 cDNA and subcloned into the pGEX-4T-1 glutathione S-transferase (GST) fusion plasmid (Amersham–GE Healthcare–UK Limited Little Chalfont, Buckinghamshire, UK). The 43kDa GST fusion protein was expressed in Escherichia coli, affinity purified, and used as an antigen to produce rabbit antiserum (ProteinTech Group, Chicago, IL).
Whole-cell extracts derived from HEK-293 cells, wild-type ES cells, and DNMT1−/− ES cells (generously provided by Dr. En Li, Novartis Institutes for Biomedical Research, Cambridge, MA) were prepared in lysis buffer (100mM NaH2PO4, 100mM Tris-HCl [pH 8.0], and 8M urea), then separated on PAGEr-Gold precast 4–12% Tris-glycine gradient gels (Lonza Group, Basel, Switzerland) by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose (Amersham–GE Healthcare). Antisera generated against murine DNMT1, β-actin (Sigma, St. Louis, MO), GAPDH (Calbiochem, San Diego, CA), eIF2α (generous gift from Scott Kimball, Pennsylvania State University), and phospho-Ser51-eIF2α (Biosource, Carlsbad, CA) were used to probe membranes, followed by horseradish peroxidase–conjugated secondary antibody (Amersham–GE Healthcare). Proteins were detected using an ECL detection kit (Amersham–GE Healthcare).
One million wild-type murine ES cells or 1.3×106 CFP1-null ES cells were seeded in 10cm gelatin-coated tissue culture dishes and incubated for ~20h. Cells were rinsed twice with phosphate-buffered saline (PBS), twice with ES medium lacking methionine and cysteine, and then incubated for 15min in fresh medium lacking methionine and cysteine. For DNMT1 half-life determination, medium lacking methionine and cysteine was removed, and cells were labeled for 30min in ES medium containing 0.1mCi/mL EXPRE35S35S [35S] Protein Labeling Mix containing [35S]-methionine and [35S] cysteine (PerkinElmer, Waltham, MA). Labeling medium was removed, and the cells were incubated in medium containing excess nonradiolabeled methionine and cysteine, and then harvested in lysis buffer (0.25M NaCl, 5mM ethylenediaminetetraacetic acid, 50mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [pH 7.5], 0.1% NP-40, 0.5mM dithiothreitol, and 3% protease inhibitor cocktail [Sigma]) at 0, 3, 6, and 9h time points. For protein synthesis studies, cells were labeled for 10, 20, or 30min in ES medium containing 0.1mCi/mL [35S]-methionine and [35S]-cysteine, then rinsed in ice-cold PBS, and harvested at various times in lysis buffer (described above for Western blots).
After the in vivo labeling procedure, whole-cell extracts were normalized for protein concentration by the Bradford method (Bradford, 1976), and then used for trichloroacetic acid (TCA) precipitation to measure global protein synthesis, or immunoprecipitation of DNMT1 or the nucleosome remodeling factor BRG1 to measure either DNMT1 protein half-life or DNMT1 and BRG1 protein synthesis. For TCA precipitation, 40μg of whole-cell extract was mixed with 0.5μg bovine serum albumin and 20% TCA and incubated on ice for 30min. The reactions were spotted onto GF/C glass microfiber filters (Whatman, Florham Park, NJ) and washed extensively with 10% TCA followed by 100% ethanol. The filters were dried and subjected to liquid scintillation counting to measure [35S]-methionine and [35S]-cysteine incorporation into total protein. In parallel, 40μg of total protein was separated by SDS-PAGE and stained with Coomassie brilliant blue to serve as a loading control for the amount of protein used for TCA precipitation reactions. For DNMT1 and BRG1 immunoprecipitations, 200μg of whole-cell extract was precleared with protein G agarose (Upstate Cell Signaling Company, Lake Placid, NY), and then incubated with appropriate antibody (DNMT1—described above; BRG1—Santa Cruz Biotechnology, Santa Cruz, CA) and protein G agarose for 3h at 4°C. Beads were washed four times with lysis buffer and then boiled in 1× protein sample buffer and separated by SDS-PAGE. Gels were dried and incorporated radioactivity was quantitated by PhosphorImager analysis using a BioRad Personal Fx System (BioRad, Hercules, CA).
Wild-type and CFP1-null ES cells were grown to 50–60% confluency in 10cm gelatin-coated tissue culture dishes. Cells were washed twice with cold PBS (pH 7.4), and then lysed with ice-cold lysis solution containing 20mM Tris (pH 7.5), 10mM MgCl2, 100mM NaCl, 0.4% NP-40, and 10μg/mL cycloheximide. The extracts were passed through a 23-gauge needle for cell lysis, incubated for 10min on ice, and subjected to centrifugation at 10,000rpm for 10min at 4°C. The supernatant preparations were collected and layered onto 15–45% sucrose gradient containing 20mM Tris (pH 7.5), 10mM MgCl2, and 100mM NaCl, and subjected to ultracentrifugation for 2h at 40,000rpm in a Beckman SW-41 Ti rotor at 4°C. After centrifugation, the gradients were fractionated using a Biocomp gradient station. Fractions were monitored at A254 using a UV monitor. To analyze ribosomes dissociated into free 40S and 60S subunits, cell lysates were similarly prepared in the absence of cycloheximide treatment, and EDTA was added to 20mM to both the supernatant preparation and sucrose gradients. Fractionation was carried out with monitoring at A254. The monosome/polysome ratio was determined by measuring the area under each curve.
A polyclonal antibody directed against murine DNMT1 was generated, using as an antigen a region in the N-terminus of DNMT1 that is devoid of conserved domains and lacks similarity with other cytosine-5 methyltransferases (Fig. 1A). The antibody raised against murine DNMT1 specifically recognizes a protein of 190kDa in murine ES cells that is absent in DNMT1−/− ES cells (Fig. 1B) (Lei et al., 1996). This antiserum also cross-reacts with human DNMT1, as shown with whole-cell extract derived from HEK-293 cells (Fig. 1B). Consistent with an earlier report, Western blot analysis demonstrated a twofold decline in steady-state DNMT1 protein levels in CFP1-null ES cells compared to wild-type ES cells (Fig. 1C) (Carlone et al., 2005).
Because CFP1 is a transcriptional activator that specifically binds unmethylated CpG dinucleotides (Voo et al., 2000), we speculated that CFP1 activates the CpG island-containing DNMT1 promoter (Yoder et al., 1996). Surprisingly, however, Northern blot analysis demonstrated that the steady-state level of DNMT1 mRNA is elevated in CFP1-null ES cells compared to wild-type ES cells (Fig. 2A). Real-time RT-PCR analysis was performed to independently quantitate DNMT1 mRNA levels. These studies revealed that DNMT1 mRNA expression was elevated ~50% in CFP1-null and CFP1-vector ES cells compared to wild-type ES cells, and was restored to wild-type levels in CFP1-rescue ES cells (Fig. 2B). These findings indicate that decreased levels of DNMT1 protein in ES cells lacking CFP1 is not explained by a reduction in DNMT1 gene transcription.
Because DNMT1 transcript levels did not correspond with the observed decrease in DNMT1 protein level, the half-life of DNMT1 protein in CFP1-deficient ES cells was examined. DNMT1 protein half-life in wild-type and CFP1-null ES cells was determined. The half-life of DNMT1 protein in wild-type and CFP1-null ES cells was 7.0 and 5.8h, respectively, while the half-life of DNMT1 protein in CFP1-rescue and CFP1-vector ES cells was 9.9 and 5.6h, respectively (Fig. 2C, D). These data revealed a 17% decrease in DNMT1 protein half-life in CFP1-null ES cells compared to wild-type ES cells, and a 43% decrease in CFP1-vector ES cells compared with CFP1-rescue ES cells. However, these trends did not reach statistical significance.
Because the modest decrease in DNMT1 protein half-life in CFP1-deficient ES cells cannot fully account for the observed 50% decrease in DNMT1 protein levels, we turned our attention to translational regulation of DNMT1 expression. Before specifically analyzing DNMT1 translation, global protein synthesis in wild-type and CFP1-null ES cells was compared. Even though wild-type and CFP1-null ES cells similarly took up [35S] methionine and [35S] cysteine from the culture medium (Fig. 3A), CFP1-null ES cells incorporated significantly less [35S] methionine and [35S] cysteine into total protein at each time point, as demonstrated by TCA precipitation of whole-cell extracts (Fig. 3B). An independent assessment of global translation in wild-type and CFP1-null ES cells is shown in Figure 3C. Forty μg of whole-cell extract prepared from cells pulsed with [35S] methionine and [35S] cysteine was separated by SDS-PAGE and stained with Coomassie brilliant blue to demonstrate equivalent amounts of total protein in wild-type and CFP1-null ES cell samples. PhosphorImager analysis of this gel revealed reduced [35S] methionine and [35S] cysteine incorporation into the total protein pool at each time point in the absence of CFP1 (Fig. 3C). These data reveal that CFP1-null ES cells exhibit a 15% reduction in global protein synthesis compared to wild-type ES cells. This finding implicates CFP1, a critical epigenetic regulator, in the regulation of global protein synthesis.
An mRNA molecule can be translated concurrently by multiple ribosomes. Differences in the ratio of monosomes and free ribosomal subunits to polysomes can provide a measure of translation initiation. The mechanism of decreased global protein synthesis observed in CFP1-null ES cells was investigated by analyzing the polysome profile in wild-type ES cells and CFP1-null ES cells. Lysates prepared from these ES cells were subjected to sucrose gradient centrifugation, and absorbance at 254nm was recorded for each gradient. A representative profile from CFP1-null ES cells shows increased levels of free 40S and 60S ribosomal subunits, accompanied by reduced polysomes, compared to wild-type ES cells (Fig. 4A). The polysome to monosome ratio in wild-type ES cells was 2.2, compared to 1.6 in CFP1-null cells. These results indicate that the underlying reduction in global translation in CFP1-null cells is due to a defect in translation initiation.
Defects in ribosome processing and assembly have been reported to alter translation initiation (Foiani et al., 1991; Martin-Marcos et al., 2007). Given the role of CFP1 in genomic cytosine methylation, we reasoned that CFP1 may be important for rDNA expression or for nucleotide modification of rRNA that contributes to ribosome assembly and function (Lawrence and Pikaard, 2004; Sergiev et al., 2007). To begin to address whether there was altered assembly of the 40S or 60S ribosomal subunits in CFP1-null cells, we carried out a sucrose gradient analysis using lysates prepared and analyzed in the presence of EDTA, which depletes Mg2+ required for ribosome coupling. We found no significant difference in the relative levels of 60S and 40S ribosomes, with a 60S to 40S subunit ratio of 2.3 in wild-type and 2.0 in CFP1-null cells (Fig. 4A, lower panels). These results argue against the model that a selective defect in the assembly of the 40S or 60S subunits explains reduced translation initiation in CFP-null cells.
An important mechanism regulating translation initiation involves phosphorylation of the α-subunit of eIF2 at serine-51. Phosphorylation of eIF2α reduces global translation by lowering the levels of eIF2-GTP that are required for delivery of initiator Met-tRNAiMET to the translational machinery (Holcik and Sonenberg, 2005; Wek et al., 2006; Wek and Cavener, 2007). Phosphorylation of eIF2α is induced by a range of stress conditions, including nutrition deficiency, viral infection, heme deficiency, and endoplasmic reticulum stress. Previous studies have noted that genetic changes alter eIF2α phosphorylation. For example, mutations that alter upstream regulators of the eIF2 kinases or mutations in the subunits of eIF2B, the factor mediating exchange of eIF2-GDP to eIF2-GTP, alter eIF2α phosphorylation (Hinnebusch, 2005). We directly addressed the idea that CFP1 deficiency enhances eIF2α phosphorylation by Western blot using antibodies specific to the phosphorylated version of this translation factor. There was a clear increase of eIF2α phosphorylation in CFP1-null cells compared to wild type (Fig. 4B). These results support the idea that one reason for reduced global translation in CFP1-deficient cells is that this mutation triggers increased eIF2α phosphorylation.
While phosphorylation of eIF2α reduces global protein synthesis, translation of specific mRNAs can have a range of sensitivities to lowered eIF2-GTP. For example, translation of Csk mRNA is preferentially sensitive to eIF2α phosphorylation, with significantly reduced Csk protein synthesis compared to β-actin mRNA translation (Liang et al., 2008). By comparison, translation of mRNA encoding ATF4, a transcriptional activator of stress responsive genes, is preferentially increased in response to eIF2α phosphorylation (Harding et al., 2000; Vattem and Wek, 2004). These studies indicate that translation of specific mRNAs can be preferentially enhanced or inhibited in response to eIF2α phosphorylation and a dampening of general translation initiation.
To specifically determine the rate of DNMT1 protein synthesis in ES cells lacking CFP1, DNMT1 protein was immunoprecipitated in wild-type and CFP1-null ES cells that were pulsed with [35S] methionine and [35S] cysteine for the indicated time points (Fig. 5). DNMT1 protein isolated from CFP1-null ES cells contained less radiolabel compared to that isolated from wild-type ES cells, indicating reduced de novo synthesis of DNMT1 protein in the absence of CFP1 (Fig. 5A). The rate of DNMT1 synthesis in the presence and absence of CFP1 was determined by calculating the slope of lines generated by linear regression. The rate of [35S] incorporation into DNMT1 protein, as calculated from multiple independent experiments, was decreased 23% for CFP1-null ES cells compared to wild-type ES cells (Fig. 5B). As noted earlier, this decrease in DNMT1 protein synthesis occurs despite DNMT1 mRNA being significantly increased in CFP1-null ES cells compared to wild-type cells (Fig. 2). A similar analysis of BRG1, a chromatin remodeling protein that has unaffected steady-state protein levels in CFP1-null ES cells (Khavari et al., 2003; Lee and Skalnik, 2005), revealed a 13% decrease in synthesis in CFP1-null ES cells (Fig. 5C), which was similar to the observed 15% reduction in global protein synthesis. These results indicate that synthesis of DNMT1 protein is preferentially reduced in CFP1-null ES cells, which exhibit a dampening of global translation initiation and elevated eIF2α phosphorylation.
CFP1 is critical for both early embryogenesis and postgastrulation development (Carlone and Skalnik, 2001; Young et al., 2006), appropriate global histone modifications (Lee and Skalnik, 2005), and cytosine methylation (Carlone et al., 2005; Young et al., 2006). The importance of cytosine methylation and DNMT1 expression in normal development and survival is well established (Li et al., 1992; Okano et al., 1999; Jackson-Grusby et al., 2001; Biniszkiewicz et al., 2002). DNMT1 protein levels are decreased twofold in the absence of CFP1 (Fig. 1) (Carlone et al., 2005). The findings in this report demonstrate that the decrease in steady-state DNMT1 protein levels in CFP1-deficient ES cells is caused by a combinatorial mechanism of reduced protein half-life and reduced translation.
The half-life of DNMT1 protein, which is a target for polyubiquitination and proteasomal degradation (Agoston et al., 2005; Ghoshal et al., 2005), is reduced from 7 to 5.8h in murine ES cells lacking CFP1. Importantly, the observed 17% decrease in DNMT1 protein half-life is not sufficient to explain the twofold reduction in steady-state DNMT1 protein levels observed in CFP1-deficient ES cells. Additional studies revealed that the loss of CFP1 leads to a 15% reduction in global protein synthesis. Further, CFP1-null ES cells contain reduced levels of polysomes, and increased levels of free ribosomal subunits and phosphorylated eIF2α. These data reveal a defect in translation initiation in ES cells lacking CFP1. Additionally, DNMT1 protein synthesis is reduced by 23% in CFP1-deficient ES cells. This is an underestimate of the magnitude of the DNMT1 translation defect, because these cells contain a 1.5-fold elevated level of DNMT1 mRNA. These findings suggest compensatory mechanisms for induction of DNMT1 mRNA to ensure that DNMT1 protein expression does not fall below 50%, despite the reduced rate of DNMT1 protein synthesis.
Epigenetic regulation of transcription is not limited to RNA polymerase II genes, as the involvement of cytosine methylation and particularly DNMT1 activity in rRNA gene transcription has been the focus of several recent publications. Human colorectal carcinoma (HCT116) cells lacking DNMT1 exhibit disrupted nucleolar architecture that results in the scattering of rRNA gene clusters throughout the nucleus (Espada et al., 2004, 2007). DNMT1 also co-localizes in the nucleolus with nucleolin, an RNA-binding protein involved in ribosome biogenesis (Majumder et al., 2006). Therefore, decreased DNMT1 protein expression in CFP1-null ES cells may participate in a feedback mechanism to directly inhibit global translation through disruption of nucleolar function. It will be of interest to determine if nuclear architecture is disrupted in CFP1-null ES cells and if rRNA genes are aberrantly expressed.
S. cerevisiae strains lacking Set1 are deficient in rDNA and telomeric silencing due to decreased histone H3K4me3 (Briggs et al., 2001; Bryk et al., 2002; Fingerman et al., 2005). It is counterintuitive that loss of a transcriptional activating mark, histone H3K4me3, results in reduced transcriptional silencing. A proposed indirect mechanism is that loss of histone H3K4 methylation at euchromatic regions leads to promiscuous binding of the Sir proteins and thus relief of the silencing effects imposed by Sir protein binding at telomeres, mating-type loci, and rDNA (van Leeuwen et al., 2002). CFP1 is required for appropriately controlled histone H3K4 methylation by the mammalian Setd1A and Setd1B complexes (Lee and Skalnik, 2005; Lee et al., 2007), because CFP1-null ES cells display increased histone H3K4 methylation. Increased histone H3K4 methylation at rDNA promoters could lead to increased rDNA silencing if the transcriptional regulatory mechanism is conserved from S. cerevisiae to mouse. Analysis of histone modifications and cytosine methylation at rDNA promoters in CFP1-null ES cells will provide insight into the role of CFP1 in the transcriptional regulation of rRNA genes.
This work was supported by the Riley Children's Foundation, the Lilly Endowment, an Indiana University School of Medicine core utilization grant, and National Science Foundation Grants MCB-0344870 and MCB-0641851 (D.G.S.). J.S.B. was supported by a pre-doctoral fellowship from National Institutes of Health Grant T32 CA111198. We thank Dr. J. Richard Chaillet for providing the murine DNMT1 cDNA expression construct, and Dr. En Li for providing the DNMT1−/− J1 ES cell line. We would also like to thank Drs. Jeong-Heon Lee, Kristin Chun, and Ann Roman for helpful discussions.
No competing financial interests exist.