The nuclear periphery of embryonic stem cells is a transcriptionally permissive and repressive compartment

Abstract

Chromatin adapts a distinct structure and epigenetic state in embryonic stem cells (ESCs), but how chromatin is three-dimensionally organized within the ESC nucleus is poorly understood. Because nuclear location can influence gene expression, we examined the nuclear distributions of chromatin with key epigenetic marks in ESC nuclei. We focused on chromatin at the nuclear periphery, a compartment that represses some but not all associated genes and accumulates facultative heterochromatin in differentiated cells. Using a quantitative, cytological approach, we measured the nuclear distributions of genes in undifferentiated mouse ESCs according to epigenetic state and transcriptional activity. We found that trimethyl histone H3 lysine 27 (H3K27-Me(3)), which marks repressed gene promoters, is enriched at the ESC nuclear periphery. In addition, this compartment contains 10-15% of chromatin with active epigenetic marks and hundreds of transcription sites. Surprisingly, comparisons with differentiated cell types revealed similar nuclear distributions of active chromatin. By contrast, H3K27-Me(3) was less concentrated at the nuclear peripheries of differentiated cells. These findings demonstrate that the nuclear periphery is an epigenetically dynamic compartment that might be distinctly marked in pluripotent ESCs. In addition, our data indicate that the nuclear peripheries of multiple cell types can contain a significant fraction of both active and repressed genes.

Fig. 1.

The ESC nuclear periphery contains epigenetic signatures of both repressed and active genes. (A-D) Immunodetection of H3K4-Me₃ (red in A, white in B) or H3K36-Me₃ (red in C, white in D)_, found on active genes, indicates hundreds of foci throughout ESC nuclei, including signals (arrows) associated with the nuclear lamina (anti-lamin B1, green). (E,F) H3K27-Me₃ (red in E, white in F)_, found on repressed promoters, is concentrated at the nuclear periphery and in some regions of the nuclear interior. Arrows indicate a partial rim of H3K27-Me₃ at the nuclear periphery. Nuclei are counterstained with DAPI (blue). Single confocal optical sections are shown. (G) For Erosion analyses, ESCs were stained with anti-lamin B1 (green, left) and DAPI (blue, left) to mark the nuclear edge and invaginations of the nuclear envelope (arrow). 3D image masks with nested `shells' that follow the contour of the nuclear edge were generated by the Erosion program (red-white, right), which erodes a nuclear edge mask at 200-nm intervals. Optical X-Y sections (top) and corresponding sections of X-Z projections (bottom) are shown. Scale bars: 1 >m. (H) Erosion analyses of signal density (relative signal intensity per shell volume) indicate that H3K27-Me₃ (red) is most concentrated at the ESC nuclear periphery, indicated by the peak lamin B1 signal (shell 0) (n=108). H3K4-Me₃ (light green) and H3K36-Me₃ (dark green) are more concentrated in the nuclear interior (n=113 and 105, respectively). For comparison, the density distribution of bulk chromatin (blue) was measured from Sytox green counterstaining, which, unlike DAPI, does not preferentially highlight mouse chromocenters. (I) Measurements of the proportions of signals in each nuclear shell show 15% of H3K4-Me₃ and H3K36-Me₃ chromatin within 400 nm of the lamin B1 peak. Relative shell volumes are shown in supplementary material Fig. S4. Lamin B1 relative density is plotted (arbitrary units); axis showing density values is hidden. Error bars are s.e.m. and include variation in shell volume in all density plots.

Fig. 2.

Transcribing MMU14 loci associate closely with the nuclear periphery. (A) Map of genes (black bars, left) and gene `deserts' in the MMU14 region under study. Sce1, Slain1, Rbm26 and Ndfip2 are expressed in E14 ESCs (Shopland et al., 2006). BACs used for FISH probes are indicated on the right. (B,C) Two-color FISH detects Sce1 and Slain1 (green) and Rbm26 and Ndfip2 (red) near the lamina (blue) of ESCs. The cell in B shows three of these four loci (arrows) contacting the lamina (inset, white). Loci from the same chromosome are separated by approximately 1 >m. An X-Y optical section and sections from X-Z projections are shown. For the cell in C, a maximum X-Z projection and single X-Z and Y-Z sections are shown. (D) 3D distributions of Sce1and Slain1 (green), Rbm26 and Ndfip2 (red), and LMNB1 (black) were measured by Erosion for 97 ESCs. Both active loci are most often 200-400 nm from the peak lamin signal. (E) MMU14 gene positions (green) contrast with that of active Sox2 (red) on MMU3 (n=93). (F,G) MMU14 gene positions (green) are similar to those of two MMU14 gene deserts (red; n=97 and 99, respectively). In D-G, mean shell volumes relative to the total nuclear volume are shown (gray, arbitrary units). (H,I) RNA-FISH detects Rbm26 and Ndfip2 transcripts (red) adjacent to the nuclear lamina (anti-LMNB1, green) in two ESC nuclei. Single X-Y, X-Z and Y-Z sections of both FISH signals in each cell are shown as indicated. Scale bars: 1 >m. (J) Erosion analysis of 93 ESCs shows a peak of Rbm26 and Ndfip2 transcript signal 400 nm from the lamina, similar to the locus probed by DNA-FISH (D). Error bars, s.e.m.

Fig. 3.

The ESC nuclear periphery contains multiple transcription sites. (A-F) Three confocal sections of an ESC stained with anti-POLII-Ser2-P antibody (red in A,C,E; white in B,D,F) indicate sites of active transcription throughout the nucleus. Examples of transcriptions sites at the nuclear lamina (green) are indicated by arrows. Optical sections shown are separated by 800 nm in the Z-axis. Scale bar: 1 >m. (G) Erosion analysis of 102 ESCs indicates a peak density of POLII-Ser2-P (red) 1.0 >m from the nuclear lamina (black). (H) 16% of these transcription sites are within 400 nm of the nuclear lamina. Total chromatin distributions based on Sytox green stain (blue) are shown for comparison. Relative shell volumes are shown in supplementary material Fig. S4. Error bars, s.e.m.

Fig. 4.

ESCs chromatin distributions are distinct from NPCs and NIH-3T3 fibroblasts. (A) NPCs, (B) ESCs and (C) NIH-3T3 cells stained with the Sytox green, a DNA dye that does not highlight centromeric heterochromatin. Fewer dense concentrations of DNA are found in ESC nuclei, particularly at the nuclear periphery, compared with the other cell types. (D,E) Immunostaining with anti-H3K4-Me₃ antibody (red in D, white in E) shows signal throughout the NPC nucleus (blue, DAPI), including multiple foci associated with the nuclear lamina (green). A similar distribution was detected in NIH-3T3 cells (H,I). (F,G,J,K) Unlike ESCs, H3K27-Me₃ (red in F and J, white in G and K) is not concentrated at the nuclear periphery of NPCs (F,G) or NIH-3T3 cells (J,K). Single confocal optical sections are shown. Scale bars: 1 >m. (L-O) Erosion measurements indicate similar distributions of H3K27-Me₃ and H3K4-Me₃ in NPCs (n=105 and 106 cells, respectively) (L,M) and in NIH-3T3 cells (n=106 and 91 cells, respectively) (N,O), contrasting with the specific enrichment of H3K27-Me₃ at the ESC nuclear periphery (Fig. 1J,K). Total DNA distributions (blue) are based on Sytox green staining. Relative shell volumes are shown in supplementary material Fig. S4. Error bars, s.e.m.