Author contributions: S.X., T.M.G., V.B., and Y.G.T.: conception and design, collection of data, data analysis and interpretation, manuscript writing, final approval of manuscript; H.X.: collection of data, final approval of manuscript; K.M.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLS EXPRESS 2009.
Transcriptional control of stem cell genes is a critical step in differentiation of embryonic stem cells and in reprogramming of somatic cells into stem cells. Here we report that Lsh, a regulator of repressive chromatin at retrotransposons, also plays an important role in silencing of stem cell-specific genes such as Oct4. We found that CpG methylation is gained during in vitro differentiation of several stem cell-specific genes (in 11 of 12 promoter regions) and thus appears to be a common epigenetic mark. Lsh depletion prevents complete silencing of stem cell gene expression and moreover promotes the maintenance of stem cell characteristics in culture. Lsh is required for establishment of DNA methylation patterns at stem cell genes during differentiation, in part by regulating access of Dnmt3b to its genomic targets. Our results indicate that Lsh is involved in the control of stem cell genes and suggest that Lsh is an important epigenetic modulator during early stem cell differentiation. STEM CELLS 2009;27:2691–2702
During development cells traverse from totipotent, to pluripotent, to terminally differentiated cells and gain a stable gene expression profile that is characteristic of the tissue-specific cell type. Considerable evidence suggests that epigenetic states can influence chromatin structure and ultimately the accessibility of transcription factors and the transcriptional machinery to their genomic targets . Thus epigenetic changes that are dependent on histone modifications and CpG methylation reflect progression of developmental stages and are thought to be characteristic of lineage-committed or undifferentiated cells.
Embryonic stem cells and embryonal carcinoma cells can be used as a model to study early events during differentiation since they are capable of differentiation into many distinct cell types in vitro . Several reports have identified unique epigenetic features at genes that are poised for transcription in ESCs but committed to activation or repression upon differentiation [3, 4]. In contrast, another subset of genes is subject to a different scheme of regulation being uniquely expressed in embryonal stem cells and later silenced in any terminally differentiated cells. Several of these “stem cell-specific” genes such as Oct4 and Nanog have been shown to be critical for stem cell function [5, 6]. Nuclei of terminally differentiated cells can be reprogrammed to a pluripotent state by applying distinct approaches such as nuclear transfer, cell fusion, exposure to nuclear extracts, or transduction with specific factors (among them Oct4) [2, 7–11]. An important feature for evaluating the efficiency of reprogramming has been the reactivation of stem cell-specific genes such as Oct4. Revealing the molecular mechanisms that are involved in repression of stem cell genes is important for better understanding of early events of cellular differentiation and epigenetic reprogramming.
DNA methylation is known to have profound effects on embryogenesis, influencing X-inactivation and genomic imprinting and playing a role in repression of retroviral elements [12–15]. The transcriptional activity of the Oct4 gene is tightly correlated with its CpG methylation levels. The Oct4 promoter is unmethylated at the blastula stage but gains CpG methylation at day 6.5 of gestation and remains methylated in somatic tissues . Mouse and human embryonal stem cells show DNA methylation at the Oct4 promoter region upon differentiation in culture, corresponding to transcriptional repression [17–19]. Methylated Oct4 transgenes are silent in vivo  and the cloning efficiency after nuclear transfer is enhanced when using hypomethylated donor nuclei [20, 21]. Moreover, the success of epigenetic reprogramming by transcription factors has been monitored in part by induced DNA hypomethylation of the Oct4 promoter . Thus cytosine methylation is thought to be a critical epigenetic modification that is involved in Oct4 expression during differentiation or reprogramming.
Several major DNA methyltransferases have been identified in mammals [22, 23]. Dnmt1 localizes at the replication fork and shows a preference for hemimethylated substrates over that of unmethylated substrate [22, 23]. Together with NP95 (UHFR1), a protein that also recognizes hemimethylated CpG sites and tethers Dnmt1 to chromatin, Dnmt1 preserves methylation patterns during cell division by specific methylation of hemimethylated CpG dinucleotides [24, 25]. Two other DNA methyltransferases, Dnmt3a and Dnmt3b, prefer unmethylated substrates in vitro and are thought to play a role predominantly in de novo methylation during embryonic development . Dnmt3L, related to Dnmt3a and Dnmt3b, lacks catalytic activity and functions as a coregulator for a subset of genomic imprinted sites and repeats [13, 22].
Lsh, a member of the SNF2 family of chromatin remodeling protein [27, 28], was previously demonstrated to be an important factor in setting DNA methylation patterns during mouse development at repeat sequences [29–32]. Lsh is crucial for de novo methylation of reporter plasmids containing retroviral sequences, and Lsh can interact directly with Dnmt3a and Dnmt3b or indirectly with Dnmt1 via Dnmt3b [32, 33]. In addition, Lsh plays a role in transcriptional silencing of the developmentally regulated Hox genes; the silencing is also accompanied by alterations in DNA methylation levels . Thus Lsh plays an important role in CpG methylation and loss of Lsh perturbs heterochromatin structure, leading to multiple physiologic defects [29, 35–37].
In the current study, we addressed the question whether DNA methylation mediated by Lsh plays a role in stem cell gene expression. We tested the idea of whether CpG methylation is a general epigenetic mark in silencing of stem cell genes during differentiation of pluripotent cells and whether Lsh is involved in this process. Furthermore, we determined whether Lsh plays a role in maintaining the stem cell phenotype and whether DNA methylation is critical for silencing of stem cell-specific genes in somatic tissue.
MATERIALS AND METHODS
Mouse ESCs (V6.4) were cultured on gelatin-coated dishes in Knockout Dulbecco's modified Eagle's medium (Invitrogen/GIBCO, Grand Island, NY, http://www.invitrogen.com) with 15% Knockout Serum Replacement and 1,000 U/ml ESGRO (Chemicon, Temecula, CA, http://www.chemicon.com). P19 mouse embryonal carcinoma (EC) cells were grown in α minimum essential medium with 10% fetal bovine serum. For differentiation, cells were placed with 1 μM all-trans retinoic acid (RA) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) in Petri dishes to allow for aggregation (P19 cells were kept as monolayers; shown in supporting information Fig. 2a). Cells were transfected with siRNA-Lsh oligonucleotides using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) before RA treatment. For the clonal growth assay limited dilution was performed. P19 cells were split into 96-well plates to achieve an approximate concentration of 5-10 cells per milliliter. Clonal growth was assessed after 10 days in culture. For alkaline phosphatase staining cells were fixed with 4% paraformaldehyde for 1 minute and stained using the Alkaline Phosphatase Detection kit (Chemicon). For intracellular staining (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com) cells were incubated with anti-Oct4 antibody (Santa Cruz Biotechnology Inc.) or affinity purified rabbit anti-Lsh antibody raised against recombinant Lsh  followed by secondary goat anti-mouse IgG2B-phycoerythrin (PE) and donkey anti-rabbit IgG-fluorescein isothiocyanate (FITC) (Santa Cruz Biotechnology Inc.). Cells were analyzed by fluorescence-activated cell sorting (FACS) for separate and dual PE/FITC staining. For immunohistochemistry staining the same antibodies were used, followed by secondary staining with biotinylated goat anti-mouse IgG or biotinylated goat anti-rabbit IgG (Santa Cruz Biotechnology Inc.). After incubation with peroxidase-conjugated streptavidin, cells were exposed to 3.3% diaminobenzidine for signal development.
Western Blot Analysis
Nuclear extracts were generated as previously described . Samples were separated on 4%--12% Tris-glycine SDS--polyacrylamide gel electrophoresis gels and blotted onto Immobilon P membrane (Millipore, Billerica, MA, http://www.millipore.com) and proteins detected using enhanced chemiluminescence detection reagents (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com). Antibodies used for Western analysis were affinity purified rabbit anti-Lsh antibody raised against recombinant protein , anti-Dnmt3b antibody (Alexis), and PCNA antibody (Santa Cruz Biotechnology Inc.).
Polymerase Chain Reaction Analysis
Total RNA was prepared using Trizol reagent (Invitrogen) and genomic DNA was eliminated with TURBO DNA-free Kit (Ambion, Austin, TX, http://www.ambion.com). About 1 μg of total RNA was reverse transcribed using iScript reverse transcriptase (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Omission of reverse transcriptase served as a negative control. cDNA was amplified using Platinum Polymerase Chain Reaction (PCR) SuperMix (Invitrogen). The PCR was performed as follows: 5 minutes at 94°C, 30 cycles of 60 seconds at 94°C, 60 seconds at 57°C–60°C, and 60 seconds at 72°C, followed by 1 cycle for 5 minutes at 72°C. For real-time PCR analysis, the MyiQ Single-Color Real-Time PCR machine (Bio-Rad) and Platinum SYBR Green Quantitative PCR (qPCR) SuperMix UDG (Invitrogen) were used: 1 cycle of 50°C for 2 minutes, 1 cycle of 95°C for 5 minutes, followed by 45 cycles of 95°C for 30 seconds, 57°C–60°C for 30 seconds, and 72°C for 30 seconds, finally followed by a melting curve analysis. A negative control without template was carried out for each PCR analysis. For quantification, standard titrations were performed for each template and primer set, and linear regression equation and the calculation for DNA amounts were established using Prism 3.0 software (GraphPad Software, San Diego, http://www.graphpad.com) and Microsoft Excel (Redmond, WA, http://www.microsoft.com). Chromatin immunoprecipitations (ChIPs) PCR conditions were as follows: 94°C for 4 minutes, 94°C for 1 minute, 55°C for 1 minute, 72°C for 1 minute (35 cycles), and 72°C for 7 minutes. All primers are listed in supporting information Table 2.
For ChIPs cells were cross-linked with 1% formaldehyde, lysed, and sonicated on ice to generate DNA fragments with an average length of 200-800 bp . After preclearing, 1% of each sample was saved as input fraction. Immunoprecipitation was performed using specific antibodies against the indicated proteins or IgG as control. After reversal of cross-linking, DNA was prepared for PCR analysis.
Methylated DNA Immunoprecipitation Assay
Genomic DNA was sonicated to produce fragments ranging in size from 300-1,000 bp. Five μg of fragmented DNA was used for a standard methylated DNA immunoprecipitation (MeDIP) assay  and precipitated with 10 μl monoclonal antibody against 5-methylcytidine (Eurogentec, Seraing, Belgium, http://www.eurogentec.be) followed by incubation with protein A agarose beads. DNA was recovered by phenol-chloroform extraction followed by ethanol precipitation. Primers (supporting information Table 2) were chosen within −1000-bp upstream of the transcriptional start sites and the amplicon sizes varied between 200-350 bp.
Genomic DNA was subjected to bisulphite treatment using CpGenome DNA modification kit (Chemicon). The PCR products were separated in agarose gels, purified, and subcloned as described .
Southern Blot Analysis
Digested genomic DNA was separated by 1.5% agarose gels. After transfer, Nytran plus membranes (Schleicher-Schuell, http://www.schleicher-schuell.com) were hybridized overnight at 42°C in hybridization buffer (Amersham) with a 32P-labeled probe for satellite or line 1 sequences  or a probe covering 8.5-kb upstream of the transcriptional start site of Oct4 (kindly provided by Dr. Matsui). The hybridized membranes were washed and visualized by autoradiography.
Silencing of Stem Cell-Specific Genes During Differentiation
To address the question whether Lsh-mediated DNA methylation plays a role during silencing of stem cell-specific genes we first used P19 cells, which are widely used as a well-established model for cellular differentiation. These EC cells are pluripotent with respect to their capacity to form embryonic chimeras . EC cells differentiate in culture into different lineages similar to ESCs, such that dimethyl sulfoxide treatment supports the differentiation into cardiac and skeletal muscle, aggregates with RA treatment lead to development of neuronal and glial tissue, and RA treatment in monolayers leads to epithelial tissues [40, 41]. We used EC cells as a model system for studying the molecular mechanism of transcriptional silencing during differentiation. Since Western analysis had shown Lsh expression in EC cells (data not shown) we first determined Lsh expression at a single-cell level using immunohistochemistry. Almost all of EC cells stained positive for Lsh, similar to Oct4 staining that was used as a control (supporting information Fig. 1). To determine the precise frequency of Lsh- and Oct4-expressing cells, we performed intracellular staining followed by FACS analysis (Fig. 1A). Approximately 86.3% of undifferentiated cells were double positive for Lsh (FITC) and Oct4 (PE), indicating that the majority of EC cells express Lsh and the pluripotency marker Oct4. This is consistent with a recent report of a meta-analysis of 38 transcriptome studies that found only 40 genes specifically expressed in human ESCs, among them was HELLS, the human homologue of Lsh, that therefore can serve as signature gene for ESCs . Next, EC cells were treated with RA and then examined for expression of the pluripotent genes Oct4 and Nanog by reverse-transcription PCR analysis (Fig. 1B). Both genes were expressed in undifferentiated cells (day 0) and suppressed after 4 and 8 days of RA treatment, whereas Nestin and Mash1 mRNA levels increase during culture, indicating a lineage commitment toward neuronal and glial cells. Likewise, RA treatment decreases transcription of Oct4 and Nanog genes in monolayer cultures (supporting information Fig. 2a). Using the Gene Expression Omnibus (GEO) expression database another 21 genes were selected for analysis that showed strict expression in preimplantation embryo and silencing in somatic tissues. Those genes, termed stem cell-specific genes in this study, were also found to be reduced after RA treatment in EC cells (Fig. 1B and supporting information Table 1). Real-time PCR analysis confirmed that complete repression of Oct4, Nanog, Gdf3, Dppa3, and Tdgf1 was established by day 4 after RA treatment (supporting information Fig. 2B). Thus the stem cell-specific genes that were investigated in this study showed expression in undifferentiated EC cells and became silenced upon differentiation.
Induction of CpG Methylation at Promoter Regions of Stem Cell-Specific Genes
To determine whether transcriptional silencing of stem cell-specific genes is associated with CpG methylation, we first performed Southern analysis of the Oct4 promoter region using the methylation-sensitive restriction enzyme HpaII. The Oct4 locus showed signs of hypomethylation in undifferentiated EC cells (Fig. 1C) but gained DNA methylation upon differentiation consistent with previous reports [16, 18, 19]. Minor satellite repeats (Fig. 1D) and Line1 elements (supporting information Fig. 3) served as negative controls since these regions maintain DNA methylation after RA treatment. Next, we used MeDIP analysis  to examine DNA methylation for a dozen selected stem cell-specific genes (among them Oct4 and Nanog) that were silenced during EC cell differentiation. At first, we examined several genomic sites that are expected to show a stable DNA methylation pattern during culture to validate the immunoprecipitated material. Repetitive elements such as minor and major satellites or intra-cisternal particle (IAP) show signs of DNA methylation in undifferentiated and in differentiated cells (Fig. 1E). Genomic imprinted sites such as the differentially methylated regions of Igf2R, two regions in the imprinting center upstream of H19 and KVDMR1 , are expected to maintain DNA methylation during differentiation and showed signs of DNA methylation before or after RA treatment (Fig. 1E). In contrast, stem cell-specific genes (altogether 11 of 12) were enriched after immunoprecipitation, indicating an acquisition of DNA methylation in the promoter regions during the differentiation process (among them Oct4, Nanog, Dppa2, Dppa3, Dppa4, Dppa5, Fbx15, Gdf3, Ndp52, Tdgf1, and Rex1). The promoter regions of seven housekeeping genes and unmethylated Nestin and Mash1 served as negative controls. Thus the stem cell-specific genes investigated in this study (with the exception of Sox2) showed a gain of DNA methylation as a stable epigenetic mark acquired upon differentiation.
Lsh Is Involved in Transcriptional Silencing of Stem Cell-Specific Genes
To evaluate the role of Lsh in silencing of stem cell-specific genes, we attempted to deplete Lsh protein by RNA interference. To study the short-term effects of Lsh depletion, we used a mixture containing four different siRNA-Lsh oligonucleotides. Treatment with 20 nM siRNA-Lsh oligonucleotide could effectively decrease Lsh protein levels by day 4 (Fig. 2A, 2B). The treatment reduced Lsh protein levels in the majority of cells and did not affect Oct4 gene expression (Fig. 2C, 2D). Moreover, the Lsh decrease did not affect the cell cycle compared with control siRNA-treated EC cells. In addition, retroviral expression (such as the intracisternal A particle) was not altered after pretreatment with siRNA-Lsh, consistent with the previous observations that Lsh was not required for maintenance of DNA methylation at repeat sequences  (supporting information Fig. 4). Next we examined the effect of RA treatment and Lsh depletion on the expression of 10 stem cell genes that had shown methylation gains during differentiation. Whereas in control cultures stem cell genes such as Oct4, Nanog, Gdf3, Dppa3, and Tdgf1 were fully silenced, siRNA-Lsh treatment resulted in incomplete repression (Fig. 2E and supporting information Fig. 5) and mRNA expression was maintained after differentiation, albeit at reduced levels compared with day-0 control cells. This indicated that Lsh is required for complete silencing of those examined stem cell genes. The partial repression despite Lsh depletion may be due to low Lsh levels that remained after siRNA-Lsh treatment or alternative repressive pathways (as discussed below).
Lsh Depletion Maintains Stemness of EC Cells
To investigate whether Lsh depletion inhibits the differentiation process and maintains stem cell properties we further characterized siRNA-Lsh-treated EC cells. First we determined whether expression of stem cell genes in siRNA-Lsh-treated EC cells was transient or continuous in culture. Extension to 20 days in culture after retinoic acid treatment did not change their ability to maintain expression of Oct4, Nanog, Dppa3, Tdgf1, and Gdf3 in Lsh-depleted EC cells (supporting information Fig. 6).
After that we monitored soft agar growth since loss of anchorage-independent growth is one of the hallmarks of differentiation . Whereas siRNA-Lsh-treated cells were able to grow in soft agar after RA treatment similar to undifferentiated EC cells (Fig. 3A), control siRNA-treated cells did not show anchorage-independent growth. In addition, the capacity for clonal growth, assessed by limiting dilution analysis, was increased in siRNA-Lsh-treated cells after RA treatment compared with control siRNA-treated cells (Fig. 3B). The Lsh-depleted EC cells that were selected from soft agar stained positive for alkaline phosphatase, a characteristic mark of stem cells that is typically lost in differentiated cells (Fig. 3C). In addition, siRNA-Lsh-treated cell lines derived from soft agar colonies after RA treatment continued expression of stem cell-specific genes (Fig. 3D). .
Taken together, Lsh-depleted EC cells partially escaped the silencing program and maintained stem cell characteristics of undifferentiated EC cells, pointing to an important role of Lsh during early cellular differentiation.
Reduced CpG Methylation at Stem Cell-Specific Genes in the Absence of Lsh
Next, we sought to resolve the question of whether Lsh depletion leads to reduced de novo DNA methylation at stem cell genes that in turn may be responsible for maintenance of stem cell characteristics. Bisulphite sequencing of a region about 350-bp upstream of the transcriptional start site of Oct4 revealed an increase in DNA methylation from 6.6% to 51.6% after RA-induced differentiation (Fig. 4A), whereas Lsh-depleted cells showed decreased CpG methylation levels (51.6% compared with 21.6%). This suggests that Lsh is required for normal DNA methylation levels at the Oct4 promoter region. Subsequently, we analyzed 11 more stem cell-specific genes for gain of CpG methylation using MeDIP (Fig. 4B). As controls, we used repeat sequences and imprinted sites that do not require Lsh for maintenance of DNA methylation [32, 35] and that showed signs of DNA methylation in siRNA-Lsh- and control siRNA-treated cells. In contrast, the stem cell-specific genes showed reduced DNA methylation at their promoter regions upon Lsh depletion. Housekeeping genes, serving as negative controls, showed no signs of DNA methylation during RA treatment and Lsh depletion. Taken together, the data suggest that Lsh is required to achieve normal DNA methylation level during differentiation at the selected stem cell-specific genes.
Lsh Depletion Interferes with Stem Cell Gene Silencing and Differentiation in ESCs
To further confirm Lsh's role in stem cell gene silencing we used genuine ESCs. Using immunohistochemistry we found Lsh and Oct4 expression in a majority of ESCs (supporting information Fig. 7). FACS analysis (Fig. 5A) exhibited about 96.7% cells that stained positive for Lsh (80.5% for Oct4). Next, ESCs were induced to differentiate after leukemia-inhibitory factor removal and RA treatment. ESCs showed a substantial downregulation of Lsh and Dnmt3b protein levels upon differentiation (Fig. 5B), supporting the idea that both proteins have a primary function in stem cells but not in somatic cells. Depletion of Lsh by siRNA-Lsh treatment resulted in barely detectable Lsh protein levels up to day 12 of culture as assessed by Western analysis (Fig. 5C, 5D). Using real-time PCR analysis, we examined the expression of 10 stem cell-specific genes after RA treatment. Whereas control siRNA-treated ESCs repressed stem cell genes within 4-8 days after RA treatment, Lsh-depleted ESCs only partially decreased Oct4, Nanog, Gdf3, Dppa3, Tdgf1, and other mRNAs (Fig. 5E). This indicated that Lsh is required for complete repression of the selected stem cell genes in ESCs similar to our observation in EC cells. Moreover, when RA-treated ESC cultures were stained for stem cell markers, siRNA-Lsh-treated cells partially maintained alkaline phosphatase expression, whereas control siRNA-treated ESCs failed to do so (Fig. 5F). Thus ESCs similar to EC cells exhibited an important role for Lsh in stem cell gene silencing and maintenance of ES stem cell characteristics.
Effect of Lsh Depletion in Lsh−/− Embryonic Tissue
Subsequently, we tested whether Lsh depletion can influence Oct4 silencing during embryonic development in vivo. Lsh−/− depletion in mice leads to perinatal death, reduced birth weight, renal abnormalities, hematopoietic deficiencies, and germ cell defects as well as mitotic problems [27, 29, 35–37, 44]. We examined Oct4 expression around day 8.5 of gestation by comparing Lsh−/− embryos with Lsh+/+ littermate controls. Enhanced Oct4 and Nanog expression was detected in five Lsh−/− embryos in comparison with littermate wild-type controls at day 8.5 of gestation (Fig. 6A), implying an in vivo role for Lsh in silencing of Oct4. However, Lsh−/− somatic tissue at day 18.5, such as brain, whole embryo, or MEFs, does not show sustained Oct4 expression (data not shown). These results leave open the possibility that either Lsh is dispensable for DNA methylation at the Oct4 gene during embryonic development or that alternative silencing mechanisms exist to compensate for loss of DNA methylation. To discriminate between the two options we performed bisulphite sequencing analysis of the Oct4 promoter region spanning about 1,700 bp using genomic DNA derived from Lsh−/− embryonic tissue compared with Lsh+/+ embryonic tissue (Fig. 6B). The methylation state of 25 CpG sites was greatly reduced in the absence of Lsh (82.8% compared with 10.4%). This confirms that Lsh regulates in vivo DNA methylation at the Oct4 promoter. Furthermore, it suggests that additional silencing mechanisms independent of DNA methylation participate in the transcriptional control of Oct4 gene expression.
Lsh Controls Association of Dnmt3b with Stem Cell Genes in ES Cells
To reveal more about the molecular mechanisms of how Lsh controls DNA methylation we performed ChIPs in ESCs. Lsh is specifically bound to the Oct4 and Nanog promoter regions, suggesting that Lsh is directly involved in the control of Oct4 and Nanog expression (Fig. 7A). Lsh as well as Dnmt3b were both already found associated with the stem cell genes in undifferentiated cells (Fig. 7A, 7B). Therefore RA treatment does not induce recruitment of Lsh/Dnmt3b to the Oct4 or Nanog promoter regions but must activate another mechanism that controls DNA methylation activity. ChIP analysis showed a gradual decline of promoter-bound Lsh and Dnmt3b over time during differentiation (Fig. 7C, 7D) that may be due in part to downregulation of both proteins after RA treatment (Fig. 5B). To address the question of whether Lsh is required for Dnmt3b recruitment, we performed Lsh depletion in ESCs by treatment with siRNA-Lsh (Fig. 7C, 7D). Lsh binding to Oct4 and Nanog was undetectable in siRNA-Lsh-treated cells in contrast to control siRNA-treated cells, confirming efficient reduction of Lsh protein levels after siRNA-Lsh treatment. Moreover, Dnmt3b association with either stem cell gene was undetectable after Lsh depletion (Fig. 7D). Similar results could be demonstrated in EC (P19) cells (supporting information Fig. 8). Taken together, the data suggest that Lsh is necessary but not sufficient to control Dnmt3b activity at a specific genomic site.
We report here that Lsh plays a role in silencing of stem cell genes and this is accompanied by acquisition of DNA methylation. We furthermore give evidence that DNA hypomethylation, although preceding reprogramming in somatic cells, is not sufficient for reactivation of Oct4 gene transcription. Finally, we demonstrate that Lsh depletion supports maintenance of the stem cell phenotype, suggesting that Lsh is involved as cofactor in the control of cellular differentiation.
Role of DNA Methylation in Stem Cell Gene Regulation
De novo methylation at the Oct4 gene and other stem cell genes is gained within 96 hours after RA treatment (Fig. 1C, 1E). In the same time span Lsh and Dnmt3b are dramatically downregulated (Fig. 5B), suggesting a critical role of these proteins before lineage commitment. The gain of CpG methylation is found at 11 of 12 examined stem cell-specific genes. These findings are consistent with a recent report showing DNA methylation differences at seven pluripotency associated genes when comparing ESCs with neuronal progenitor cells . Our findings support the notion that DNA methylation gain is not unique for Nanog or Oct4 gene regulation but is a frequent epigenetic mark associated with the investigated stem cell-specific genes. Since CpG methylation can be faithfully propagated over 50 cell generations it is considered a very stable epigenetic mark, thus contributing to stable repression of stem cell genes in somatic cells. Our data are consistent with the model that the first step in embryonic differentiation comprises silencing of stem cell genes before subsequent steps allow for tissue-specific gene expression.
Role of Lsh in Transcriptional Repression
Lsh is highly expressed in pluripotent cells, downregulated during embryogenesis, and shows reduced expression in adult tissues. The effect of Lsh on stem cell gene expression is tightly correlated with DNA methylation. Methylated promoters are usually inactive, and associated with hypoacetylated chromatin [12, 46]. Proteins that selectively bind to methylated CpG sites (such as MecP2) or Dnmts can recruit histone deacetylases  and thus may contribute to repression since histone acetylation plays an important role in the formation of a Pol II initiation complex . DNA methylation can also interact with other silencing pathways such as polycomb proteins [34, 45, 47]. In addition, Lsh has been reported to show repressive function on reporter gene expression separate from CpG methylation . This repressive activity was dependent on interaction of Lsh with Dnmts, but not on the catalytic activity of the methyltransferase. Although our results demonstrate an effect of Lsh on DNA methylation they do not exclude alternate possibilities of Lsh effects.
How Does Lsh Affect DNA Methylation at Stem Cell Genes?
Part of the Lsh effect may be due to stabilization of Dnmt3b association with genomic loci since Lsh depletion reduces Dnmt3b binding (Fig. 7D). Likewise, we have found Dnmt3b association at HoxA genes and at the PU.1 locus is dependent on Lsh [34, 44]. This may be in part by direct interaction between Lsh and Dnmt3b, or alternatively the presumed chromatin remodeling activity of Lsh might support Dnmt3b binding to DNA [32, 33]. Since Lsh and Dnmt3b binding at the Oct4 gene was observed before differentiation, Lsh is necessary but not sufficient to control DNA methylation. Furthermore, neither Lsh nor Dnmt3b is a rate-limiting step in DNA methylation. Dnmt3b activity might be regulated by post-translational modifications [48, 49] or an as yet unidentified cofactor for Dnmt3-mediated DNA methylation could be involved, similar to the recent discovery of NP95 supporting Dnmt1 activity [24, 25]. Another possibility is that simultaneous downregulation of the H3K4me3 mark at stem cell-specific genes may facilitate the intimate interaction of Dnmt3b with the nucleosome. Cocrystallization has recently shown that Dnmt3L, required for CpG methylation of retrotransposons and imprinted sites in germ cells, promotes interaction of Dnmt3a with its genomic target by binding to histone 3 tails that are unmethylated at the H3K4 residues [50, 51]. Although a role for Dnmt3L has not been addressed in ESCs a similar mechanism in regulating Dnmt3b activity may play a role for stem cell gene silencing. Finally, a new study has reported a high turnover of CpG methylation marks at an estrogen-responsive promoter region . Since our study addressed only the equilibrium of DNA methylation pattern we cannot exclude the possibility that Lsh additionally controls CpG methylation turnover rates.
Role for DNA Hypomethylation in Reactivation of Oct4
Oct4 promoter analysis has revealed proximal as well as distal enhancer elements [19, 53]. Multiple sites close to the transcriptional start site have been shown to undergo complete or partial CpG methylation (the HpaII sites around −1000, −700, and −260 bp and the HhaI site at −130 bp) during ESC differentiation [16, 18]. Using Xenopus oocytes it was demonstrated that DNA demethylation at proximal sites (−24, −166, −289) are critical for Oct4 promoter regulation, but not at distal sites (at −754 or −1148 bp) . Reprogramming of somatic cells is correlated to hypomethylation at proximal promoter sites. Although Lsh is involved in DNA methylation and silencing of Oct4 in vitro and in vivo, other pathways can ultimately maintain Oct4 silencing. Lsh deficiency in embryos delays Oct4 mRNA repression during development (Fig. 6), but nevertheless in Lsh−/− newborn tissues Oct4 mRNA is undetectable. The presence of repressors such as GCNF could in part be accountable for silencing. GCNF binds directly to the Oct4 promoter region and recruits corepressor complexes such as SMRT and N-CoR . On the other hand a lack of activating factors may play a role, too. Oct4 has been shown to be regulated by a self-reinforcing loop, since Oct4 can bind to its own promoter region and enhance its own expression . Moreover, Oct4 repression effects expression of other stem cell-specific genes such as Nanog, Sox2, Rex1, Tdgf1, or Dppa4, and Nanog and Sox2 in turn have been shown to associate with the Oct4 promoter and to regulate its expression as well [56, 57]. Thus a lack of Oct4 and other stem cell-specific genes may contribute to a failure to re-activate Oct4 in Lsh−/− tissues. In addition, compacted, inaccessible chromatin, the presence of polycomb proteins, or Pol II stalling could possibly be responsible. Thus DNA methylation may act as an additional safeguard to prevent aberrant Oct4 expression. Since hypomethylated genomes show improved cloning efficiency after somatic cell nuclear transfer , it is yet to be examined whether Lsh-depleted cells can improve nuclear reprogramming efficiency. To further unravel these molecular mechanisms will be challenging in the future and beneficial to advance our understanding of cellular differentiation as well as to improve techniques for cellular reprogramming and their applications to regenerative medicine.
We show here that Lsh is involved in DNA methylation and silencing of stem cell genes. This suggests that Lsh plays an important role in setting the gene expression pattern during cellular differentiation and has the potential to serve as a molecular target for epigenetic programming and reprogramming.
We thank Drs. Jonathan Keller, Steven Hou, and Nancy Colburn for critically reviewing of the manuscript. We thank Dr Takaaki Matsui for the generous gift of the mouse Oct4 promoter region. We are grateful to Jenny Mercardo and Jennifer Waters for excellent animal technical assistance.
This project has been funded in whole or part with federal funds from the National Cancer Institute (NCI), National Institutes of Health, under contract N01-C0-12400. NCI-Frederick is accredited by Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) International and follows the Public Health Service Policy for the Care and Use of Laboratory Animals in accordance with the “Guide for the Care and Use of Laboratory Animals.”58
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.