Motif distribution and DNA methylation underlie distinct Cdx2 binding during development and homeostasis

Abstract

Transcription factors guide tissue development by binding to developmental stage-specific targets and establishing an appropriate enhancer landscape. In turn, DNA and chromatin modifications direct the genomic binding of transcription factors. However, how transcription factors navigate chromatin features to selectively bind a small subset of all the possible genomic target loci remains poorly understood. Here we show that Cdx2-a lineage defining transcription factor that binds distinct targets in developing versus adult intestinal epithelial cells-has a preferential affinity for a non-canonical CpG-containing motif in vivo. A higher frequency of this motif at embryonic Cdx2 targets and methylated state of the CpG during development enables selective Cdx2 binding and activation of developmental enhancers and genes. In adult cells, demethylation at these enhancers prevents ectopic Cdx2 binding, instead directing Cdx2 to its canonical motif without a CpG. This shift in Cdx2 binding facilitates Ctcf and Hnf4 recruitment, establishing super-enhancers during development and homeostatic enhancers in adult cells, respectively. Induced DNA methylation in adult mouse epithelium or cultured cells recruits Cdx2 to developmental targets, promoting corecruitment of partner transcription factors. Thus, Cdx2's differential CpG motif preferences enable it to navigate distinct DNA methylation profiles, activating genes specific to appropriate developmental stages.

Fig. 1. Evolution of Cdx2 binding across gene promoters and enhancers supports developmental and homeostatic functions in intestinal epithelium.

a Heatmap showing Cdx2 CUT&RUN signal at its developmental and adult-specific binding sites in embryonic (E12.5), fetal (E16.5), and adult epithelial (villus) cells; Dev 1–3 and Adult 1–3 subgroups represent 2-fold change (q < 0.01) in Cdx2 signal among the three timepoints. b Pie charts showing percentages of Cdx2 binding sites shown in (a) (Dev 1, 2 and Adult 1, 2) at promoters (TSS −2 kb to +1 kb) and distal enhancers. c Gain, retention, and loss of Cdx2 binding at promoters (top panel) and distal enhancers (bottom panel) during development from E12.5 into adult epithelial cells; numbers outside the boxes and inside the dotted boxes represent gains and losses of Cdx2 binding at each of the timepoints, respectively. d Genome browser tracks showing progressive loss or recruitment of Cdx2 through epithelial development at Sox4 and Fabp2 gene promoters, respectively.

Fig. 2. Heightened presence of CpG containing Cdx2 motif at its developmental binding sites.

a Heatmap showing relative prevalence of TF motifs at developmental (Dev 1–3) and adult (Adult 1–3) cell-specific Cdx2 binding sites as determined by HOMER. b Line plot showing number of Cdx2 binding loci containing Cdx2 motifs within an increasing distance (on X-axis) from center of the Cdx2 peak. c Line plot showing abundance of canonical and CpG-containing Cdx2 motifs at developmental (shades of blue) and adult (shades of red) cell-specific Cdx2 binding sites within an increasing distance (on X-axis) from center of the Cdx2 peak. d Density plots showing relative prevalence of CpG-containing motif near the center of developmental Cdx2 peaks (Dev 1–3) in comparison with the adult cell-specific Cdx2 peaks (Adult 1–3); canonical Cdx2 motif shows widespread presence near Adult 1–3 sites in contrast with the Dev 1–3 loci. e Higher percent of regions with development-specific Cdx2 binding contain CpG-containing motif as compared to regions with adult cell-specific Cdx2 binding near the center of Cdx2 peak (within ±50 bp); beyond 50 bp of the peak center, canonical motif is more prevalent. f Heatmap showing DNA methylation at Cdx2 binding loci (as in Fig. 1a) in early endoderm (E6.5), and embryonic (E12.5), fetal (E16.5), and adult epithelial (villus) cells; Cdx2 binding in embryonic epithelium is associated with early loss of DNA methylation (blue dotted box), while sites first bound by Cdx2 in the fetal epithelium show higher methylation levels (red dotted box). g Fractional methylation change at CpG (4th base position) within the Cdx2 motif (For Dev1–3 and Adult 1–3, as in Fig. 1a) shows relatively early reduction in methylation at loci with developmental binding of Cdx2 (blue arrowhead) compared to binding in adult epithelium (red arrowhead) (p < 2.2e⁻¹⁶, right-sided Wilcoxon test). The CpG remains hypomethylated in adult villus at both developmental and adult loci. The horizontal bars within the box plots represent medians, the tops and bottoms of the boxes represent the 75th and 25th percentiles, respectively, and the upper and lower whiskers extend to the maximum and minimum values, respectively. The outliers are indicated by dots. h A model for DNA methylation-based Cdx2 binding at developmental target loci; recruitment at adult target loci is driven by DNA methylation independent binding. While relative abundance of the CpG-containing motif and methylation at the CpG within the Cdx2 motif during development allows its early recruitment to these sites, demethylation at these loci protects them from ectopic Cdx2 binding in adult cells. Source data are provided as a Source Data file.

Fig. 3. Cdx2 facilitates establishment of adult homeostatic super-enhancers by directing Ctcf recruitment.

a Percentage of developmental (blue) or adult (red) cell-specific Cdx2 bound loci (as in Fig. 1a) with designated number of other Cdx2 peaks nearby (±50 kb); drawing in the inset represents isolated Cdx2 peaks (not within 50 kb of each other) in developing cells, while adult cells show multiple neighboring peaks in the same area. Large number of (60%) of dynamic Cdx2 binding events are solitary in developing epithelium (blue arrowhead) as opposed to only 30% in adult cells (red arrowhead). b Representative genomic tracks showing clustered binding of Cdx2 that grows through development near gene Krt19. c boxplot showing number of neighboring (±50 kb) dynamic Cdx2 bound enhancers (sites in Fig. 1a) surrounding 7806 developmental (blue) or 8901 adult (red) cell-specific Cdx2 binding loci (p < 0.001, right-sided t-test). The horizontal bars within the box plots represent medians, the tops and bottoms of the boxes represent the 75th and 25th percentiles, respectively, and the upper and lower whiskers extend to the maximum and minimum values, respectively. The outliers are indicated by dots. d Percentage of 1327 developmental (E12.5 and E14.5) and 2373 adult super-enhancers that have dynamic Cdx2 binding during epithelial maturation. e Genome browser view of Cdx2, Ctcf, H3K27ac, and ATAC-seq signals at adult-specific Cdx2 cluster at a super-enhancers near Mcu13 gene; black arrowheads indicate the super-enhancer borders with Cdx2 and Ctcf cobinding. Plot on the right shows gain in expression of Muc13 through development. The horizontal bars represent medians, each dot represents individual values (n = 2 for E12.5 and E16.5, n = 3 for Villus) (p = 0.003, right-sided Wilcoxon test) (f) Heatmap showing Cdx2 and Ctcf signals at super-enhancers in E12.5, E16.5 and adult villus. Super-enhancers boundaries are shown with black arrowheads. g Heatmap showing Ctcf signal at developmental and adult-specific Cdx2 binding sites (as in Fig. 1a) in E12.5, E16.5, and adult villus cells; the regions are displayed in decreasing order of Ctcf signal. 19% and 15% loci with development-specific Cdx2 binding show Ctcf occupancy at E12.5 and adult villus samples, respectively. 4% and 53% loci with adult-specific Cdx2 binding show Ctcf binding in E12.5 and adult villus samples, respectively. Source data are provided as a Source Data file.

Fig. 4. Loss of PRC2 activity causes gain of DNA methylation at enhancers leading to Cdx2 recruitment.

a Experimental schematic showing 5 intraperitoneal injections of tamoxifen (TAM) on consecutive days cause deletion of Eed across the intestinal epithelium in adult mouse (2–6 month old); epithelial cells are collected 4 days after the last injection (experimental day 9). b Genome browser view showing gain of CpG methylation in Eed^−/− epithelium at multiple enhancer loci near Lasp1 gene that are identified as DMRs; promoter region remains unmethylated. Pie chart showing percentages of DNA methylation gains at promoters and enhancers. c Density plot showing DNA methylation at promoter (top panel) and non-promoter (bottom panel) linked UMRs and LMRs; LMRs (mostly representing enhancers) have the most prevalent gain of methylation in Eed^−/− cells. d Heatmap showing genomic states identified by CHROMHMM based on TF binding and histone modifications corresponding to promoter, enhancer, and repressed chromatin; Cdx2 occupied enhancers (H3K4me1+ and H3K27ac+) particularly show gain in DNA methylation in Eed^−/− cells, while promoters (H3K4me3+) and repressed chromatin regions (H3K27me3+ or H3K9me3+) show relatively stable unmethylated and methylated status, respectively. The horizontal bars within the box plots represent medians, the tops and bottoms of the boxes represent the 75th and 25th percentiles, respectively, and the upper and lower whiskers extend to the maximum and minimum values, respectively. The outliers are indicated by dots. e DNA methylation profiles at development and adult specific Cdx2 binding sites (Dev 1–3 and Adult 1–3, respectively) showing the gain of methylation upon loss of PRC2 activity (Eed^−/−). f Profile plot of Cdx2 signal show ectopic recruitment of Cdx2 at developmental loci (Dev 1) in adult Eed^−/− cells in concordance with the gain of DNA methylation in (e). g Genomic tracks showing Abcc4 gene locus with CpG-containing Cdx2 motif bound by the TF only during development, where gain of methylation at the Cdx2 binding locus in Eed^−/− leads to recruitment of the TF; dot plot at the bottom shows methylated (solid black dots) or unmethylated (white dots) status of the CpG in Cdx2 motif (4th position) in 8 independent sequencing reads from WT and Eed^−/− WGBS data. Source data are provided as a Source Data file.

Fig. 5. Altered Cdx2 binding in adult cells revives developmental Ctcf recruitment.

a Heatmap showing significant (q < 0.05, 1.5×) gain or reduction of Cdx2 binding in Eed^−/− cells as compared to the native villus cells (WT). b Profile plots showing Cdx2 signal at loci with gain and loss of Cdx2 binding in Eed^−/− cells from embryonic (E12.5), fetal (E16.5), adult villus, and Eed^−/− epithelium. c Profile plots showing DNA methylation signal at loci with gain or reduction of Cdx2 binding in Eed^−/− cells from embryonic (E6.5 and E12.5), fetal (E16.5), adult villus, and Eed^−/− epithelium. d Fractional methylation change at CpG at the 4th position within Cdx2 motif shows gain of methylation in Eed^−/− cells at the sites (panel a) with altered Cdx2 binding. The horizontal bars within the box plots represent medians, the tops and bottoms of the boxes represent the 75th and 25th percentiles, respectively, and the upper and lower whiskers extend to the maximum and minimum values, respectively. The outliers are indicated by dots. e A model showing DNA methylation-based Cdx2 binding at developmental target loci; Cdx2 is recruited to developmental sites upon gain of methylation of CpG (at 4th position) within the Cdx2 motif in adult Eed^−/− cells. f Heatmap showing chromatin accessibility dynamics at sites with gain or reduction of Cdx2 binding in Eed^−/− cells as compared to WT cells (as in a); chromatin accessibility at developmental Cdx2 target loci is lost with lack of Cdx2 binding in adult cells (as seen in a) and rebinding of Cdx2 to such loci in Eed^−/− cells reinduces open chromatin formation. g Genomic tracks showing recruitment of Cdx2 to developmental site in Eed^−/− cells accompanied by gain of chromatin accessibility and increase in DNA methylation at the Cdx2 binding locus; dot plot at the bottom shows methylated (solid black dots) or unmethylated (white dots) status of the CpG in Cdx2 motif (4th position) in 6 independent sequencing reads from WT and Eed^−/− WGBS data. h Heatmap showing alterations in active enhancer mark H3K27ac at the sites with Cdx2 gain or reduction in Eed^−/− cells as compared to WT cells (as in a); enhancer activity from developmental Cdx2 target loci is lost with lack of Cdx2 binding in adult cells (as seen in a) and rebinding of Cdx2 to such loci in Eed^−/− cells reactivates these enhancers. Source data are provided as a Source Data file.

Fig. 6. Perturbed Cdx2 binding resulting from DNA methylation changes leads to altered cobinding of TFs.

a Heatmap showing Ctcf signal in adult villus and Eed^−/− epithelium at loci with gain or reduction of Cdx2 binding (as in Fig. 5a). b Profile plots showing Ctcf signal from embryonic (E12.5), fetal (E16.5), adult villus, and Eed^−/− epithelium at the loci with gain or reduction of Cdx2 binding in Eed^−/− cells. c Genomic tracks showing corecruitment of Ctcf with Cdx2 in adult Eed^−/− cells reviving the early embryonic binding at E12.5; a nearby locus shows stable binding of Ctcf independent of Cdx2 occupancy. d Heatmap showing Hnf4a signal in adult villus and Eed^−/− epithelium at loci with gain or reduction of Cdx2 binding (as in Fig. 5a). e Heatmaps showing signal of various histone modifications in adult epithelial cells at sites with Cdx2 gain or reduction in Eed^−/− cells (as in Fig. 5a); regions in each cluster are arranged in decreasing order of the repressive H3K27me3 modification signal.

Fig. 7. Direct inhibition of DNA methylation causes loss of Cdx2 binding and remethylation at binding loci allows its recruitment.

a Experimental schematic showing treatment of HCT116 cells with DNA methylation inhibitor GSK-3484862 or DMSO (control) for 6 days in order to eliminate CpG methylation, followed by 3 days of recovery without treatment to allow remethylation. b Fluorescence micrographs showing loss of DNA methylation in cells upon treatment with inhibitor GSK-3484862 for 6 days, (4 independent experiments). c Representative genomic track showing fractional methylation of CpGs within ACTA1 gene body in HCT116 cells. Lollipop plots show 14 consecutive CpGs within 10 random representative amplicon sequences (from > 100× coverage on nanopore amplicon bisulfite sequencing) and their methylation status upon DMSO or GSK-3484862 treatment and 3 days of recovery; average methylation of all sequenced amplicons (>100× coverage) is represented for all treatment conditions in the right panel. d Heatmap showing loss of CDX2 binding at loci with CpG-containing CDX2 motif in HCT116 cells treated with DNA methylation inhibitor as compared to DMSO (control); upon removal of the inhibitor, CDX2 is recruited back to these loci within 3 days. e Representative genomic tracks showing loss of CDX2 binding upon elimination of DNA methylation in HCT116 cells through treatment with GSK-3484862 as opposed to the stable binding upon treatment with DMSO (control); removing the inhibitor from the cell culture media causes recovery of the CDX2 binding. Lollipop plots (right panel) represent methylation level at the CpG within CDX2 motif in HCT116 cells (WGBS, 7 reads from 2 replicates) and cells treated with DMSO or GSK-3484862 treated and after recovery (average methylation from >900 reads, nanopore amplicon bisulfite sequencing). f Box and whisker plots showing significant loss of CDX2 binding at 471 loci with CpG-containing CDX2 binding motif upon treatment with DNA methylation inhibitor and recovery of this binding upon removal of the inhibitor (right-sided Wilcoxon test; ns, p > 0.05); there is no significant change in CDX2 binding at 500 randomly chosen loci without the CpG-containing motif (g) (right-sided Wilcoxon test; ns, p > 0.05). The horizontal bars within the box plots represent medians, the tops and bottoms of the boxes represent the 75th and 25th percentiles, respectively, and the upper and lower whiskers extend to the maximum and minimum values, respectively. The outliers are indicated by dots. Source data are provided as a Source Data file.