Dynamic Regulation of Histone H3 Methylated at Lysine 79 within a Tissue-specific Chromatin Domain*

Post-translational modifications of individual lysine residues of core histones can exert unique functional consequences. For example, methylation of histone H3 at lysine 79 (H3-meK79) has been implicated recently in gene silencing inSaccharomyces cerevisiae. However, the distribution and function of H3-meK79 in mammalian chromatin are not known. We found that H3-meK79 has a variable distribution within the murine β-globin locus in adult erythroid cells, being preferentially enriched at the active βmajor gene. By contrast, acetylated H3 and H4 and H3 methylated at lysine 4 were enriched both at βmajorand at the upstream locus control region. H3-meK79 was also enriched at the active cad gene, whereas the transcriptionally inactive loci necdin and MyoD1 contained very little H3-meK79. As the pattern of H3-meK79 at the β-globin locus differed between adult and embryonic erythroid cells, establishment and/or maintenance of H3-meK79 was developmentally dynamic. Genetic complementation analysis in null cells lacking the erythroid and megakaryocyte-specific transcription factor p45/NF-E2 showed that p45/NF-E2 preferentially establishes H3-meK79 at theβmajor promoter. These results support a model in which H3-meK79 is strongly enriched in mammalian chromatin at active genes but not uniformly throughout active chromatin domains. As H3-meK79 is highly regulated at the β-globin locus, we propose that the murine ortholog of Disruptor of Telomeric Silencing-1-like (mDOT1L) methyltransferase, which synthesizes H3-meK79, regulates β-globin transcription.

Post-translational modification of core histones in chromatin represents a common epigenetic mechanism that controls diverse nuclear processes. An almost endless number of histone modifications has been documented, and it has been hypothesized that a histone code exists in which unique combinations of histone modifications confer distinct functional consequences (1). Acetylation of core histones enhances factor access to the chromatin template by increasing accessibility of nucleosomal DNA (2,3). Increased accessibility does not appear to be accompanied by a major change in nucleosome structure in vitro (4,5). However, a 3-fold increase in acetylation strongly inhibits folding of chromatin fibers into higher order structures (6). Furthermore, the acetylated amino-terminal tails of core histones are recognized by bromodomains (7,8), which are often present in coactivators that mediate transcriptional activation. Although the structure of acetylated chromatin and the functional implications of acetylation have been studied for many years (9), considerably less is known about other histone modifications such as methylation.
The concept of a histone code is illustrated by methylation of histone H3 at amino acids 4 and 9. Histone H3 methylated at lysine 9 (H3-meK9), 1 but not at lysine 4 (H3-meK4), is selectively recognized by heterochromatin protein 1 (HP1) (10 -14). HP1 serves an important role in the assembly of repressive chromatin structures (15), thus linking a site-specific histone modification to heterochromatin formation. By contrast, H3-meK4 is preferentially enriched in transcriptionally active chromatin (16 -18), including transcriptional regulatory elements such as the murine ␤-globin locus control region (LCR) ( Fig. 1) (18,19). However, H3-meK4 is not uniformly present throughout the active ␤-globin locus in adult murine erythroid cells (18). The central portion of the locus, which contains the Ey and ␤H1 embryonic ␤-globin genes and is characterized by broad, low level acetylation (19 -21), also has low level H3-meK4. H3 dimethylated at Lys-4 is enriched within coding regions of transcribed genes in yeast, and this modification is established by the methyltransferase Set1 (22). A more recent analysis in yeast revealed that H3 dimethylated at Lys-4 was enriched at active and inactive genes, whereas H3 trimethylated at Lys-4 marked only active genes (23). The functional consequences of H3-meK4 include inhibiting interaction of the nucleosome remodeling and deacetylase repressor complex with the amino terminus of H3; H3-meK9 does not prevent nucleosome remodeling and deacetylase repressor complex binding (24,25).
In addition to acetylation and methylation of Lys-4 and Lys-9 within the amino terminus of histone H3, H3 is modified in the central globular domain by methylation at Lys-79 (26 -28). Lys-79 methylation is catalyzed by the histone methyltransferase disruptor of telomeric silencing 1 (DOT1) in yeast (26 -28) and DOT1-like protein (DOT1L) in humans (29). Methylation of H3 at Lys-79 requires Rad6-mediated ubiquitination of Lys-123 of H2B (H2B-K123) (30,31), analogous to the coupling of H2B ubiquitination and methylation of H3 at Lys-4 (32). DOT1 was identified by a genetic screen in yeast to isolate genes that disrupt telomeric silencing (33). DOT1 was also identified as a regulator of the pachytene checkpoint, which ensures proper chromosome segregation during the meiotic cell cycle (34). Mutation of Lys-79 and mutations that abrogate DOT1 catalytic activity impair telomeric silencing (26,27), indicating that DOT1 mediates telomeric silencing via methylation of Lys-79. Taken together with results that H3-meK79 is abundant in bulk yeast histones and is deficient from telomeres, it has been proposed that H3-meK79 excludes silent information regulator (SIR) complexes (27). SIR complexes assemble at telomeres and mediate heterochromatin formation (35). The model assumes that H3-meK79 excludes SIR proteins from the bulk of genomic DNA, thus yielding sufficient amounts of SIR proteins to assemble repressive complexes at heterochromatic sites.
The studies described above have led to a compelling model to explain the functional consequences of H3-meK79 in yeast. However, given the considerably larger size of mammalian genomes versus the yeast genome, it would be striking if all active chromatin regions require H3-meK79 to prevent SIRmediated heterochromatization. The distribution of H3-meK79 in mammalian chromatin has not been determined. Thus, we developed a quantitative chromatin immunoprecipitation (ChIP) assay to measure H3-meK79 levels at a mammalian chromatin domain, the erythroid-specific murine ␤-globin locus, and at other functionally distinct sites of mammalian chromatin. Although the distribution of H3-meK79 at these sites is consistent with the model in which H3-meK79 is enriched at active promoters, surprisingly, H3-meK79 levels were highly variable throughout the active ␤-globin domain in adult erythroid cells. As H3-meK79 levels were dynamically regulated by the hematopoietic activator p45/NF-E2 and differed between adult and embryonic erythroid cells, H3-meK79 is likely to be an important determinant of ␤-globin gene expression.
Chromatin Immunoprecipitation Assay-ChIP was performed as described (18,19,39). Erythroid maturation of MEL cells was induced by treatment of 0.5 ϫ 10 5 cells/ml with 1.5% Me 2 SO (Sigma) for 4 days. Aliquots of immunoprecipitated chromatin (1.5 l) were analyzed by real time PCR as described (18,19,39). Primers were designed by Primer Express R 1.0 software (PE Applied Biosystems) to amplify 50 -150-bp products. Samples from at least two independent immunoprecipitations were analyzed.
RT-PCR Analysis-Total RNA was prepared from the same CB3 and WT cells used for ChIP with Trizol (Invitrogen). cDNA was synthesized by annealing RNA (1 g) with 250 ng of a 5:1 mixture of random and oligo(dT) primers by heating at 68°C for 10 min. After denaturation, samples were incubated with reverse transcriptase (Superscript II, Invitrogen) combined with 10 mM dithiothreitol, RNasin (Promega), and 0.5 mM dNTPs for 1 h at 42°C. The reaction mixture was diluted to a final volume of 150 l and heat-inactivated at 95-99°C for 5 min. Quantitative real time RT-PCRs (25 l) contained 2.5 l of cDNA, 12.5 l of SYBR Green (Applied Biosystems), and the appropriate primers. Product accumulation was monitored by SYBR Green fluorescence. Relative expression levels were determined from a standard curve of serial dilutions of MEL cDNA samples. Forward and reverse primers for real time RT-PCR (5Ј-3Ј) are as follows: ␤-globin, CAGCCTCAGT-GAGCTCCACTG and GATCATATTGCCCAGGAGCC; and glyceraldehyde-3-phosphate dehydrogenase, GAAGGTACGGAGTCAACGGATTT and GAATTTGACCATGGGTGGAAT.

Quantitative Analysis of H3-meK79 Levels in Mammalian
Chromatin-A quantitative ChIP assay was developed to measure H3-meK79 levels in mammalian chromatin (Fig. 2). We initially analyzed the distribution of H3-meK79 within the erythroid-specific ␤-globin locus (Fig. 1) in Me 2 SO-induced MEL cells, which express high levels of the adult ␤-globin gene ␤major. The cross-linked chromatin fragments used for ChIP had an average size of 400 -500 bp ( Fig. 2A), and assays were conducted under linear conditions (Fig. 2B). Representative amplification curves for the detection of H3-meK79 at HS2 of the LCR and at the ␤major promoter are shown in Fig. 2C. Thermal dissociation of the PCR products revealed a homogenous dissociation curve (Fig. 2D), indicative of a single major PCR product for each primer set.
Analysis of H3-meK79 at functionally distinct sites of the ␤-globin locus (Fig. 3A) revealed strong enrichments at the active ␤major promoter and within the ␤major gene at intron 2 (Fig. 3B). By contrast, H3-meK79 levels were low at HS2, at the inactive ␤H1 promoter, and at the representative intergenic site IVR3. IVR3 does not contain known functionally relevant sequences. Acetylated H3 was high at HS2, at the ␤major promoter, and at ␤major intron 2. The fact that H3-meK79 was considerably lower at HS2 versus the ␤major promoter was surprising, because we showed previously (18 -20) that other histone modifications, notably acetylated H3, acetylated H4, and H3-meK4, are strongly enriched at both the HS2 and the ␤major promoter. Based on the fact that ϳ90% of yeast histone H3 is methylated at Lys-79, and based on the SIR localization model (27), we reasoned that H3-meK79 might be uniformly enriched throughout the active ␤-globin domain. Rather, H3-meK79 had a highly variable distribution, paralleling the distribution of acetylated H3, with the exception that H3-meK79 was low at HS2 (Fig. 3B).
To gain a broader perspective of how H3-meK79 is distributed in mammalian chromatin, we measured H3-meK79 at other active and inactive genes (Fig. 4). H3-meK79 levels were high at both the promoter and exon 2 of the broadly expressed cad gene. By contrast, upstream sequences, promoters, and exonic sequences from the inactive neuronal-specific necdin gene and the muscle-specific MyoD1 gene had very low levels of H3-meK79. Thus, taken together with the ␤-globin results of Fig. 3, H3-meK79 levels are highest at active genes.

FIG. 3. Acetylated histone H3 and H3-meK79 have distinct patterns at the ␤-globin locus of Me 2 SO-induced MEL cells.
A, organization of the murine ␤-globin locus. B, cross-linked chromatin was isolated from Me 2 SO-induced MEL cells. The relative levels of H3-meK79 and H3 acetylation were measured by quantitative ChIP analysis at various sites of the endogenous murine ␤-globin locus (mean Ϯ S.E., at least three independent experiments). Note that the first four graphs have a common y axis with the units indicated on the left, and the graph depicting ␤maj I2 data has a distinct y axis. Pro, promoter; I2, intron 2; PI, preimmune; acH3, acetylated H3.

FIG. 2. Quantitative ChIP analysis of H3-meK79 at the murine ␤-globin locus.
A, agarose gel analysis of sonicated, deproteinized chromatin used in the ChIP assay. Chromatin fragments smaller than 1 kb, averaging ϳ400 -500 bp, were visualized by ethidium bromide staining. B, standard curves of SYBR Green fluorescence signals obtained from dilutions of input DNA from induced MEL cells using HS2 and ␤major promoter specific primer pairs. C, representative amplification plot of H3-meK79 at HS2 and ␤major promoter in induced MEL cells; PI, preimmune. D, dissociation plot obtained from amplicon shown in C. The single peak indicates the generation of single amplicon for each primer pair.

Dynamic Regulation of H3-meK79 at the Endogenous Murine
␤-Globin Locus-We showed previously (20) that the central region of the murine ␤-globin locus, characterized by low level histone acetylation in adult erythroid cells, is highly acetylated in embryonic erythroid cells of a 10.5-day post-conception murine yolk sac. This result indicated that the acetylation pattern of the ␤-globin locus is developmentally dynamic. Accordingly, we hypothesized that establishment of broad low level acetylation at the central subdomain of the ␤-globin locus, which contains the embryonic Ey and ␤H1 genes, was required for silencing of the embryonic ␤-globin genes. Based on the low level H3-meK79 at the ␤H1 promoter and at IVR3, both residing within the central subdomain, we asked whether establishment and/or maintenance of H3-meK79 is developmentally dynamic. Rather than using yolk sac as in our previous analysis, we used primitive erythroid cells derived from murine ES cells via in vitro differentiation (Fig. 5A). This in vitro differentiation system recapitulates regulatory events occurring during normal erythropoiesis (41)(42)(43)(44). As large numbers of embryonic erythroid cell colonies, EryP, can be readily differentiated from ES cells, and it is difficult to obtain large numbers of primary embryonic erythroid cells from murine yolk sac, the ES cell system has major advantages for analyzing the chromatin structure of the ␤-globin locus during embryonic erythropoiesis. No ChIP analyses have been described using specific cell types differentiated from ES cells.

FIG. 5. In contrast to Me 2 SO-induced MEL cells, H3-meK79 is broadly distributed throughout the ␤-globin domain in primitive erythroid cells derived from murine embryonic stem cells.
A, EryP cells were generated via in vitro differentiation of murine ES cells. B, cross-linked chromatin was isolated from EryP, and the relative level of H3-meK79 was measured by quantitative ChIP analysis (mean Ϯ S.E. of two independent experiments). Pro, promoter; PI, preimmune. sis. H3-meK79 levels at HS2 were 2.8-fold greater in EryP versus induced MEL cells. However, H3-meK79 at HS2 in EryP was much lower than at the active ␤H1 promoter, analogous to the result in induced MEL cells, in which H3-meK79 was much lower at HS2 versus the active ␤major promoter.

Mechanistic Insights Arising from the H3-meK79 Pattern in
Mammalian Chromatin-The distribution of H3-meK79 in mammalian chromatin had not been measured previously. Our results on the distribution of H3-meK79 at functionally distinct sites of mammalian chromatin are summarized in Fig. 7. Results described herein support a model in which H3-meK79 is highly enriched in mammalian chromatin at active genes. Conversely, inactive genes are characterized by low level or no detectable H3-meK79. Because our results demonstrate that H3-meK79 is dynamically regulated, resembling the other functionally important modifications H3-meK4 and acetylated H3, we propose that H3-meK79 and, importantly, enzymes that establish H3-meK79, regulate ␤-globin expression. As multiple histone modifications of the ␤-globin locus are insensitive to transcriptional inhibition by the elongation inhibitor 5,6-dichloro-1-␤-D-ribofuranosyl-benzimidazole (data not shown), it is highly unlikely that dynamic changes in these modifications result from altered transcriptional activity. Rather, changes in histone modifications at the ␤-globin locus are likely to regulate transcription.
It is instructive to relate our results to the current model for the function of H3-meK79 derived from studies in yeast (27). Genetic and biochemical analysis provided evidence that DOT1 methylates H3 at Lys-79 and is required for telomeric silencing (26 -28). As summarized in the Introduction, it was proposed that H3-meK79 excludes SIR silencing complexes, which are critical for heterochromatin formation (27). Our results showing that inactive genes have low or undetectable levels of H3-meK79 are consistent with this model. However, the variable FIG. 6. p45/NF-E2-dependent establishment of H3-meK79 at the ␤major promoter but not at HS2. A, Western blot analysis of p45/NF-E2 expression in total cell lysates from CB3, CB3-WT21, and Me 2 SO-induced MEL cells. The arrows depict two p45/NF-E2 isoforms, which result from usage of distinct translation start sites. No functional differences have been ascribed to the two isoforms. B, real time RT-PCR analysis of ␤-globin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) RNA expression in CB3 and CB3-WT21 cells. C, relative levels of H3-meK79, H3-meK4, and H3 acetylation were measured by quantitative ChIP analysis at HS2, IVR3, ␤H1 promoter, ␤major promoter, and MyoD1 promoter in CB3 and CB3-WT21 (mean Ϯ S.E. of three independent experiments, except for H3-meK4 levels at HS2 in CB3-WT21, which represent duplicate values). Pro, promoter; PI, preimmune; acH3, acetylated H3. distribution of H3-meK79 throughout the mammalian ␤-globin locus in adult erythroid cells, being almost entirely absent from sites within the central subdomain and at HS2, was not predictable from the yeast model.
Why would the central subdomain and HS2 have such low levels of H3-meK79 in induced MEL cells? The central subdomain also has low levels of acetylated histones H3 and H4 (19,20) and H3-meK4 (18), is inaccessible to restriction endonucleases, and contains sites highly modified by CpG dinucleotide methylation (19). The histone acetylation pattern of MEL cells is physiologically relevant, because it is similar to that of normal adult erythroid cells of the 14.5-day post-conception murine fetal liver (20). It is unknown whether SIR proteins are components of the repressive chromatin structure of the central subdomain in erythroid cells. The low level H3-meK79 at the central subdomain might allow SIR proteins to localize to the central subdomain, thereby facilitating assembly of repressive complexes to establish and maintain the silent state of the embryonic ␤-globin genes. Alternatively, other factors might preclude the need to utilize H3-meK79 to exclude SIR proteins from the central subdomain.
Based on the recent cloning of human DOT1L, which catalyzes methylation of histone H3 at Lys-79 (29), and the dynamic regulation of H3-meK79 at the ␤-globin locus, we propose that mDOT1L is an important regulator of ␤-globin gene transcription. Whereas acetylated H3 and H4 and H3-meK4, which are hallmarks of active chromatin, are highly enriched at the LCR and the active promoters (18,20), H3-meK79 is uniquely high at active genes in induced MEL cells (Fig. 3). The variable distribution of H3-meK79 within the ␤-globin locus in MEL cells is consistent with an important regulatory function of H3-meK79 locally at the active adult ␤-globin genes. By contrast, H3-meK79 had a broader distribution throughout the locus in EryP, and notably the intergenic site IVR3 of the central subdomain had high levels of H3-meK79. The EryP results indicate that H3-meK79 does not reside exclusively at active promoters and genes.
In summary, we have described the first analysis of the distribution of H3-meK79 in mammalian chromatin. This analysis in adult and embryonic erythroid cells yielded patterns that were not predictable from prior studies in yeast. H3-meK79 was considerably more enriched at the ␤major promoter versus HS2, and H3-meK79 represents the only histone modification detected to date that is preferentially enriched at ␤major. Based on these findings, it will be important to determine how mDOT1L expression and activity are regulated during erythropoiesis and whether hematopoietic specific activators differentially recruit mDOT1L to the ␤-globin locus in embryonic and adult erythroid cells.