Chromatin analysis of adult pluripotent stem cells reveals a unique stemness maintenance strategy

Abstract

Many highly regenerative organisms maintain adult pluripotent stem cells throughout their life, but how the long-term maintenance of pluripotency is accomplished is unclear. To decipher the regulatory logic of adult pluripotent stem cells, we analyzed the chromatin organization of stem cell genes in the planarian Schmidtea mediterranea. We identify a special chromatin state of stem cell genes, which is distinct from that of tissue-specific genes and resembles constitutive genes. Where tissue-specific promoters have detectable transcription factor binding sites, the promoters of stem cell-specific genes instead have sequence features that broadly decrease nucleosome binding affinity. This genic organization makes pluripotency-related gene expression the default state in these cells, which is maintained by the activity of chromatin remodelers ISWI and SNF2 in the stem cells.

Fig. 1.. Tissue isolations identify tissue-specific and constitutive genes.

(A) Schematic of the experimental design. (B and C) Heatmaps displaying normalized RNAseq data (B) of known tissue markers and (C) of all genes belonging to the specific clusters defined for each cell isolation. Z scores based on the transcripts per million (TPM), representing the relative expression levels of each gene across the different cell isolations normalized to the mean expression of the gene, are shown. (D and E) Heatmaps displaying (D) H3K4me3 CUT&Tag and (E) ATACseq. Z scores based on the reads per kilobase per million (RPKM) of the genes of the different tissue-specific clusters defined by RNAseq are shown. The gene order for (C) to (E) is identical. The RPKM for (D) and (E) was computed over a 2-kb region centered on the TSS. (F) Location of the H3K4me3 CUT&Tag peaks detected in each tissue. (G) Location of the ATACseq peaks detected in each tissue. Exons and introns are determined based on the mapping of poly-adenylated transcripts. The promoter is defined as the 2 kb upstream of the TSS. Transposable elements (TEs) were identified by RepeatMasker. (H) Analysis of the chromatin read coverage relative to the gene expression level. For each tissue, three gene categories were defined: highly expressed (>30 TPM), expressed (>10 TPM and <30 TPM), and weakly expressed (<10 TPM). Boxplots for the RNAseq (in TPM, in green), H3K4me3 CUT&Tag (in RPKM over the 2 kb centered on the TSS, in red), and ATACseq (in RPKM over the 2 kb centered on the TSS, in blue) are shown from left to right. Statistical significance was determined using a Wilcoxon test (P value with Bonferroni correction: *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001).

Fig. 2.. Chromatin organization of constitutive genes is notably distinct from that of tissue-specific genes.

(A) Metaplots showing distribution of ATAC and H3K4me3 reads over the TSS (±1 kb) region of tissue-specific and constitutive genes in brain, epidermis, intestine, and neoblast (in RPKM). Gene counts used for this analysis are listed in table S5. (B) Boxplots comparing the distribution of log₂(ATAC/H3K4me3) for tissue-specific and constitutive genes in brain, epidermis, intestine, and neoblast. Statistical significance was determined using a Wilcoxon test. P values are as indicated in the panel. (C) Schematic of the nucleosome occupancy. (D to F) Metaplots depicting nucleosome occupancy signal computed with nucleoATAC (83) centered on TSSs of the tissue-specific (colored lines) or constitutive (black lines) genes. (G) Putative Initiator (Inr) consensus sequence detected over the TSS (±40 nt). (H) Percentage of genes containing this putative Inr motif and TATA box over the TSS (±40 nt) in the different gene clusters (as determined by HOMER).

Fig. 3.. ATAC peaks reveal transcription factor motifs for tissue-specific genes, but not for neoblast genes.

(A) Z score of the frequency of transcription factor (TF) motifs (see table S2) within the region 2 kb upstream the TSS as identified by FIMO (92). (B) Metaplots depicting chromatin accessibility over several motifs of interest from (A) (±500 b) in each of the tissues. (C) Enriched motifs in ATACseq peaks localized in the promoter regions of intestinal and epidermal genes. No significantly enriched motifs were found for neoblasts or constitutive genes. (D) Aggregated A, T, C, and G content of the region around the TSS (±500 b) in the different gene clusters. (E to I) Metaplots of the A, T, C, and G content in the region around the TSS (±500 b) in the genes specific to brain (E), epidermis (F), intestine (G), neoblast (H), or constitutive genes (I). (J) Analysis of nonamer enrichment in the 500 nt upstream the TSS. (K) Analysis of T stretch frequency in the 500 nt upstream the TSS. Statistical significance is shown for the neoblast and constitutive sets relative to the tissue-specific sets as determined by Wilcoxon test (P value with Bonferroni correction: ns: not significant, **P ≤ 0.01, ***P ≤ 0.001).

Fig. 4.. Genomic A compartment is not enriched in promoters of stem cell or constitutive genes.

(A) Circos plot of the S. mediterranea genome organized per chromosome, showing A/B compartments (A in blue; B in red), gene density (light green for low, and dark green for high gene density), TE density (light purple for low, and dark purple for high TE density), and positions of the constitutive genes. (B) Comparison of the chromatin organization between the A and B compartment (A in blue; B in red). (C) Distribution of neoblast-specific genes and constitutive genes between the A and B compartments. (D) Enrichment in TE annotation per family over A/B compartment compared to the whole-genome percentage (set to 1.0). TE families are separated into retrotransposons and DNA transposons. Shading indicates the primary compartment for each family. Statistical significance is determined by Wilcoxon test (P value with Bonferroni correction: ns: not significant, *P ≤ 0.05).

Fig. 5.. Chromatin remodelers ISWI and SNF2 regulate stem cell gene expression.

(A) Heatmap of RNA expression levels of planarian chromatin remodeler genes. (B and C) Gene expression changes (qPCR) in neoblast genes and constitutive genes upon knockdown of chromatin remodelers snf2a (B) and iswib (C). Statistical significance is determined using a Student’s t test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.005). (D) Fluorescent in situ hybridization of smedwi-1 transcript, showing the number of neoblasts upon control RNAi treatment, or knockdown of snf2a or iswib. Scale bar, 50 μm. (E) Quantification of (D). (F) Illustration of the automated quantification of colony size. Different colonies as identified by the image analysis are marked in different colors. (G and H) Number of cells per expanding neoblast colony (G) and total neoblast density per animal (H) at 7 days (7d) after irradiation [snf2a(RNAi)] or 10 days (10d) after irradiation [iswia(RNAi) and iswib(RNAi)] compared to controls. Statistical significance is determined by Wilcoxon test (P value with Bonferroni correction: ns: not significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001).

Fig. 6.. Model of neoblast gene regulation.

Tissue-specific genes (bottom, green) are transcribed in differentiated cells (right) where they are marked by trimethylation of H3K4 (red). These genes are regulated by binding of tissue-specific TFs at a region of increased chromatin accessibility just upstream of the transcriptional start site (TSS). In the neoblasts (adult pluripotent stem cells, left), tissue-specific TFs are not present and the tissue-specific genes therefore remain silent. Neoblasts express a set of stem cell–specific genes (middle, yellow), and these are again marked by methylation of H3K4me3 (red). In contrast to the tissue-specific genes, activation of the neoblast genes does not depend on specific TFs, and no accessible regions upstream of the TSS are detected. Instead, the stem cell–specific promoters contain increased numbers of T-tracts, which decrease the affinity for nucleosomes. In the presence of chromatin remodelers ISWI and SNF2, this allows for the dynamic removal of weakly bound nucleosomes (blue) to provide access for polymerase II (Pol II) and activation of transcription. In the differentiated cells (right), the levels of these chromatin remodelers are much reduced and the neoblast genes remain silent. Constitutive genes (top, gray) are expressed in all cell types. These genes show promoter features similar to the neoblast genes but do not rely on the same chromatin remodelers for their expression.