Distinctive signatures of histone methylation in transcribed coding and noncoding human beta-globin sequences

Abstract

The establishment of epigenetic marks, such as methylation on histone tails, is mechanistically linked to RNA polymerase II within active genes. To explore the interplay between these modifications in transcribed noncoding as well as coding sequences, we analyzed epigenetic modification and chromatin structure at high resolution across 300 kb of human chromosome 11, including the beta-globin locus which is extensively transcribed in intergenic regions. Monomethylated H3K4, K9, and K36 were broadly distributed, while hypermethylated forms appeared to different extents across the region in a manner reflecting transcriptional activity. The trimethylation of H3K4 and H3K9 correlated within the most highly transcribed sequences. The H3K36me3 mark was more broadly detected in transcribed coding and noncoding sequences, suggesting that K36me3 is a stable mark on sequences transcribed at any level. Most epigenetic and chromatin structural features did not undergo transitions at the presumed borders of the globin domain where the insulator factor CTCF interacts, raising questions about the function of the borders.

FIG. 1.

Transcript levels within and flanking the human β-globin locus. (A) Arrowheads represent individual globin genes in the human β-globin and indicate the direction of transcription. Vertical black arrows represent the LCR and locus flanking DNase I hypersensitive sites. Odorant receptor genes in flanking sequences are represented by black squares. Gaps in the sequence diagram are denoted by double-hashed lines representing gaps of 21 kb (left) or 40 kb (right). Vertical bars below the diagram represent the locations of the TaqMan probes. (B) Total RNA from K562 cells was isolated, and transcript levels at specific regions were assayed by RT-qPCR and compared to a genomic DNA control of known concentration by the comparative C_T method (n = 2). Probes within γ-globin coding sequences do not distinguish between the Gγ and Aγ genes but probes up- and downstream and between these genes amplify unique sequences. Black bars, with reverse transcriptase; gray bars, without reverse transcriptase. Controls include necdin (brain specific) and ζ-globin (unlinked α-like embryonic globin). (C) The data of panel B are presented at an expanded scale. Error bars indicate standard errors of the means (n = 3).

FIG. 2.

RNA polymerase II association at transcribed genic and nongenic β-globin sequences. ChIP was performed using antibodies specific to the pol II N-terminal region (N-20; Santa Cruz), the initiation/elongation Ser5P CTD form of pol II (H14; Covance), and the elongation/termination Ser2P CTD form of pol II (H5; Covance). Values obtained by real-time qPCR were normalized to input by using the comparative C_T method and may be rescaled, i.e., divided or multiplied by a factor, as indicated, to accommodate individual antibody efficiencies and allow for direct comparison of the patterns of pol II localization. The results are the averages of three RNA preparations.

FIG. 3.

Histone methylation patterns in and flanking the human β-globin locus. ChIP was performed using antibodies specific to the different methylated forms of histone H3K4, K9, and K36 (see Table S2 in the supplemental material for antibody sources). (A) Histone H3K4 mono-, di-, and trimethylation. Values were obtained by real-time qPCR and, in some instances, were rescaled (see the legend for Fig. 2) to allow for direct comparison of histone methylation patterns obtained with different antibodies. The results are averages of three different chromatin preparations. No Ab, no antibody. For additional control reactions where the scale has not been changed, including error bars, see Fig. S2 in the supplemental material. (B) Histone H3K36 mono-, di-, and trimethylation. (C) Histone H3K9 mono-, di-, and trimethylation. The diagram at the bottom is not drawn to scale.

FIG. 4.

H3K4, K9, and K36 trimethylation patterns in the human β-globin locus. The me3 patterns for H3K4, K36, and K9 from Fig. 3A, B, and C are plotted together and in some instances were rescaled (see the legend to Fig. 2) to allow for direct comparison of the methylation patterns obtained with different antibodies. ChIP was performed using antibodies to H3. All values are normalized to input using the comparative C_T method (n = 3), and the final values obtained were multiplied by 3 in order to compare the pattern of H3 distribution to those of histone trimethylation. For additional control reactions where the scale has not been changed, including error bars, see Fig. S2 in the supplemental material.

FIG. 5.

Limits of subdomains of H3 hyperacetylation and K4 dimethylation and CTCF interactions in the human globin locus. (A) ChIP was performed using antibodies to diacetylated H3 (K9 and K14), and the results were compared to the pattern for H3K4me2 from Fig. 3A. Error bars represent standard errors of the means (SEM) (n = 3). No Ab, no antibody. (B) ChIP was performed using antibodies to CTCF. No Ab, no antibody. Error bars represent SEM (n = 3).

FIG. 6.

Analysis of the human β-globin chromatin domain. (A) DNase I sensitivity was analyzed using qPCR with amplicons across the globin gene region as shown in Fig. 1A. The x axis indicates increasing concentrations of DNase I (20, 40, 80, and 160 U), and the y axis indicates the average level of DNase I sensitivity observed at each of the amplicons compared to a mock-treated DNA control (n = 3). The sites most sensitive to DNase I are labeled. (B) The data for digestion with 80 U of DNase I are plotted to reveal the differences in sensitivity among other sites in the globin locus and flanking regions. The horizontal red line divides the sites analyzed into two arbitrary groups based on sensitivity.