Search
Close this search box.

A MTA2-SATB2 chromatin complex restrains colonic plasticity toward small intestine by retaining HNF4A at colonic chromatin – Nature Communications

MTA2-containing NuRD associates with SATB2 and regulates colonic gene expression

SATB2 and its homolog SATB1 have been proposed as chromatin hubs that orchestrate protein-protein and protein-DNA interactions25,26. Reasoning that some of the SATB2-associated factors may regulate colonic plasticity, we purified protein complexes that contain SATB2 from murine colonic glands (Fig. 1a and Figure. S1a). Two independent AP-MS experiments identified a total of 628 proteins with a false discovery rate (FDR) < 1%. Of these, 78 proteins were significantly enriched in both samples (SATB2 AP-MS signal intensity and MS count over IgG controls > 2-fold), with SATB2 itself being the most enriched (Fig. 1b, Supplementary Data 1).

Fig. 1: MTA2 and the NuRD complex associate with SATB2 and regulate colonic transcription.
figure 1

a Candidate SATB2-associated proteins were identified from mouse colonic glands by affinity purification (AP) with an anti-SATB2 antibody followed by Mass Spectrometry (MS). b Of the 78 proteins enriched in both AP-MS experiments, the top 40 (highlighted as colored balls) contained many histones, matrix proteins, and chromatin remodeling factors. c Co-IP demonstrated interactions of SATB2 with MTA2, and MTA2 with CHD4, a core member of the NuRD complex. Three independent experiments were repeated with similar results. d CRISPR-CAS9 and gRNA were used to successfully disrupt Mta2 from cultured mouse colonic organoids, as shown by immunofluorescence staining. BF: bright field. ECAD: E-cadherin. Two independent experiments were repeated with similar results. Scale bar = 50 μm. e Immunoblot quantification showed significant reduction of seven SATB2-associated chromatin factors after CRISPR-mediated deletion in colonic organoids. Two independent CRISPR experiments and two controls were shown. Mean ± S.D. Source data are provided as a Source Data file. RNA-sequencing showed that disrupting NuRD members (Chd4, Mta2 and Gatad2a) but not SWI/SNF factors (Smarca5, Smarca4 and Smarcd2) or Ctbp2 caused transcriptomic shifts of colonic organoids toward that of Satb2 knockout, as illustrated by Principal Component Analysis (PCA, f) and Gene Set Enrichment Analysis (GSEA, g, h). NES: normalized enrichment score. P value was calculated by a phenotype-based permutation test and adjusted by Benjamini-Hochberg method. g Source data are provided as a Source Data file.

The top 40 candidate SATB2-associated proteins included an abundance of histones, nuclear matrix proteins, and chromatin remodeling factors, consistent with the proposed role of the SATB family as chromatin organizers (Fig. 1b and Figure. S1b). Four members of the NuRD complex, including CHD4, MTA2, RBBP4, and GATAD2A, were among the top 40 interactors, suggesting association of the NuRD complex with SATB2 (Fig. 1b). Using co-immunoprecipitation (co-IP), we observed interaction of both SATB2 with MTA2, and MTA2 with the NuRD core subunit CHD4 (Fig. 1c and Figure. S1c). We also validated the interaction of SATB2 with SMARCD2 and SMARCA4, two members of the SWI/SNF chromatin remodeling complex identified in AP-MS (Figure. S1c).

To evaluate the functional importance of candidate SATB2-associated factors in colonic transcription, we used CRISPR-CAS9 in murine colonic organoids to disrupt nine chromatin remodeling genes whose protein products were enriched in our AP-MS analysis (Figure. S1d); of these, seven achieved deletion efficiencies of 80-95% by immunoblot analysis (Fig. 1, e and Figure. S1e). RNA-sequencing indicated that deletion of Chd4, Mta2, or Gatad2a, but not the other factors, significantly altered colonic transcriptomes toward that of Satb2 knockout organoids (Fig. 1f–h and Figs. S1F and 1g, Supplementary Data 2). These data suggest that the NuRD complex interacts with SATB2 and is functionally important in regulating colonic transcription.

Activation of lipid absorptive genes in MTA2-deficient colonocytes

The colonic mucosa is a regenerative epithelium, with 4- to 7-day cycles of self-renewal powered by LGR5+ intestinal stem cells (ISCs) in the crypts of Lieberkuhn27. The colonic ISCs produce progenitors (transient amplifying cells) which give rise predominantly to absorptive colonocytes and secretory goblet cells (Fig. 2a). Immunohistochemistry revealed prominent MTA2 expression in upper, but not lower colonic glands, and in scattered sub-epithelial cells (Figs. 2b, c and Figure. S2a). In contrast, the NuRD subunits CHD4 and GATAD2A were present throughout the colonic epithelium, similar to SATB2 (Fig. 2b). The majority of MTA2+ cells (68.0 ± 7.6%) were CA1+ colonocytes and conversely, nearly all CA1+ mature colonocytes were MTA2+ (Figs. 2d and g). A minority subpopulation of MTA2+ cells were goblet cells (6.5 ± 3.1%), recognized by Alcian blue staining (Figs. 2f, g). LGR5+ colonic stem cells did not express MTA2 (Figs. 2e and g). Thus, an MTA2-containing NuRD complex is enriched in terminally differentiated colonocytes on the luminal surface.

Fig. 2: MTA2 expression is enriched in colonocytes.
figure 2

a Diagram of colonic epithelium and the cell lineages. Mature colonocytes are concentrated in the upper glands of mucosal surface. TA cells: transient amplifying cells. b Immunofluorescence staining showed prominent MTA2 expression in the upper glands of adult mouse colon whereas SATB2 and two other members of NuRD, CHD4 and GATAD2A, were present throughout the epithelium. Two independent experiments were repeated with similar results. Scale bar = 100 μm. MTA2 was expressed in Ecadherin+CA1+ mature colonocytes (c, d), but not in LGR5+ colonic stem cells (e). Seven independent experiments were repeated with similar results. Scale bar = 100 μm. f Histology stain for MTA2 and Alcian Blue stain for goblet cells showed few goblet cells expressing MTA2 (f). DAB: 3,3’-diaminobenzidine. Seven independent experiments were repeated with similar results. Scale bar = 25 μm. g MTA2 was present mostly in colonocytes. n = 7 independent experiments with 3 mice. Source data are provided as a Source Data file.

Intestinal mucosa specific Mta2 gene deletion in 2-month old Villin-CreER; Mta2f/f mice (Fig. 3a, hereafter referred to as Mta2cKO), led to near complete absence of MTA2 (Fig. 3b). One month after Tamoxifen treatment, Mta2cKO mice showed no overt changes in colonic histology or cell proliferation (Figure. S3a). RNA-sequencing of colonic glands revealed 200 up-regulated and 68 down-regulated genes (log2 fold change [LFC] > 1, adjusted p [padj] <0.05) (Fig. 3c, Supplementary Data 3) in Mta2cKO vs. control colon. Mta2cKO colon was enriched for functional pathways and gene sets in fat digestion and absorption, thiamine metabolism, and chemokine signaling (Fig. 3d-f). Transporters for amino acids, carbohydrates, bile salts and vitamins also exhibited a trend toward up-regulation in Mta2cKO colon (Figure. S3b). In contrast, no pathway was significantly enriched among the down-regulated genes (P < 0.001) (Fig. 3d). Immunohistochemistry showed expression of FABP6 and MTTP, two lipid transport proteins, in the upper glands of Mta2cKO colon (Figs. 3g, h). Alkaline phosphatase, a small intestine brush border enzyme, was also activated and localized to surface colonocytes (Fig. 3g). Consistent with these molecular changes, BODIPY staining showed lipid accumulation in ileal villi and the upper glands of Mta2cKO proximal colon, but not in control colon (Fig. 3i). Thus, MTA2 loss in colonocytes activated many small intestinal genes, particularly those involved in lipid transport and metabolism.

Fig. 3: Activation of lipid transport and metabolism genes in adult mouse colon after MTA2 loss.
figure 3

Mta2 was deleted from 2-month old VilCreER;Mta2f/f (Mta2cKO) mice by applying tamoxifen (a), leading to near complete absence of MTA2 in colonic epithelium (b). Three independent experiments were repeated with similar results. Scale bar = 100 μm (b). RNA-seq of control and Mta2cKO colonic glands identified 200 up-regulated and 68 down-regulated genes ([LFC] > 1, adjusted p [padj] <0.05) (c, volcano plot). P value calculated by Wald test and adjusted by Benjamini-Hochberg method. Although no molecular pathways were significantly enriched among the down-regulated cohort, genes involved in lipid absorption, transport, and metabolism were prominently enriched among the up-regulated cohort, as illustrated by KEGG pathway gene set enrichment analysis (d, e) and in the heatmap representation (f). P value calculated by a phenotype-based permutation test and adjusted by Benjamini-Hochberg method. d Source data are provided as a Source Data file. g, h Histology and immunofluorescence staining showed activation of lipid transport proteins FABP6 and MTTP and small intestine brush border enzyme Alkaline Phosphatase in the surface colonocytes of Mta2cKO colon. Three independent experiments were repeated with similar results. Scale bar = 100 μm. i BODIPY stain revealed presence of lipid accumulation in villi of ileum and surface glands of Mta2cKO proximal colon, but not control colon. Two independent experiments were repeated with similar results. Scale bar = 100 μm.

MTA2 retains HNF4A on colonic enhancers and prevents HNF4A from activating small intestine chromatin

MTA2 is part of the NuRD complex, which has been proposed to suppress alternative transcriptional programs in several tissues by direct binding and suppression of target genes28. To investigate how MTA2 modulates small intestine gene expression in the colon, we mapped genome-wide MTA2 binding by chromatin immunoprecipitation sequencing (ChIP-seq). Duplicate MTA2 ChIP data from colonic epithelia yielded highly concordant data with 23,557 peaks (q < 1 ×10−3, using input DNA and Mta2cKO as controls) (Figure. S4a). Colonic MTA2 binding occurred at promoters (49.1%, <2 kb from transcription start sites (TSSs)) and distal elements (50.9%, introns and intergenic regions) (Figure. S4b). Genes near MTA2 binding sites (<50 kb) were highly enriched for the colonic but not the small intestine signatures (Fig. 4a, Supplementary Data 4). For instance, small intestine genes activated in Mta2 null colon, such as Lgals2, Npc1l1, Abcg8, and Pla2g2a (Figs. 3c and f), lacked nearby MTA2 binding (Figure. S4c). These data indicate that MTA2 does not directly bind and suppress small intestine genes.

Fig. 4: MTA2 retains HNF4A on colonic enhancers and prevents HNF4A from activating small intestine chromatin.
figure 4

a Tissue enrichment scores of genes near MTA2 binding peaks (MACS2, q < 0.001, distance <50 kb) in colon showed a predominant colonic signature. The P value was calculated using Enrichr tool which relies on Fisher’s exact test and adjusted by Benjamini-Hochberg method. b Top 5 DNA binding motifs in MTA2 distal binding sites ranked by P value. The P value calculated using HOMER tool which used cumulative binomial distributions to calculate motif enrichment. c Venn diagram showing the overlap of MTA2 and HNF4A genomic binding in the colon. d DNA binding profiles of HNF4A sites that were either reduced (left panel, 2,065 sites) or gained (right panel, 4379 sites) after MTA2 loss in the colon. Corresponding ATAC signals in the colon or ileum are shown. Plots are 6-kb windows centered on each HNF4A binding site. e HNF4A mRNA levels were comparable in Mta2cKO vs. control colon. Mean ± S.D. n = 4 independent control mice and n = 5 independent Mta2cKO mice. Adjusted p value by Wald test corrected for multiple testing with Benjamini and Hochberg method. Source data are provided as a Source Data file. f Combined RNA-seq and HNF4A ChIP plot showed that after MTA2 deletion, the increase (up) or decrease (down) of gene expression was strongly associated with gain or loss of HNF4A binding. Adjusted p value calculated by Wald Test from DESeq2. g Genome Browser tracks of MTA2 and HNF4A binding and RNA-seq at genomic loci of two small intestine lipid transport genes. Gain of HNF4A binding at these loci (highlighted) correlated with transcriptional activation.

Using HOMER analysis, we identified the DNA-binding motif of the intestinal transcription factor HNF4A as the top enriched motif at distal MTA2 binding sites (Fig. 4b). HNF4A and its homolog HNF4G are expressed in both small and large intestines, and shown to be important in activating enterocyte gene transcription29. Thus, we evaluated whether MTA2 could regulate HNF4A in the colon. Indeed, 87.2% of the MTA2 binding sites on colonic chromatin overlapped with binding of HNF4A (Fig. 4c). HNF4A expression was unchanged after MTA2 loss but HNF4A binding was depleted at 2,065 sites and acquired at an additional 4379 sites in Mta2cKO colon (log2FC > 1.0, q < 0.01) (Figs. 4d, e and S4d). About 80% of depleted sites (1639 of 2065) and nearly all gained sites (4233 of 4379) were in distal elements (Figure. S4e), indicating that MTA2 regulates HNF4A binding primarily at distal enhancers. Consistent with this notion, the depleted and gained HNF4A sites corresponded to areas of open chromatin enriched in the colon and ileum, respectively (Fig. 4d). Moreover, loss and gain of HNF4A binding in Mta2cKO colon were strongly associated with down-regulation of colonic and up-regulation of ileal genes, respectively (Figs. 4f, g and S4F, S4g). Thus, MTA2 deletion led to HNF4A loss on colonic enhancers and its relocation to small intestine enhancers, triggering activation of small intestine genes in the colon. These data suggest that MTA2 retains HNF4A binding on colonic chromatin and prevents HNF4A from activating small intestine genes.

Both SATB2 and MTA2 co-localize with HNF4A on colonic chromatin but SATB2 restrains HNF4A more strongly than MTA2

Both SATB2 and MTA2 can regulate colonic plasticity and our findings indicate that they interact physically (Fig. 1a-c). Structural studies of SATB1/2 proteins have identified five functional domains: a N-terminal ubiquitin-like domain (ULD) that mediates oligomerization, a CUT-like domain (CUTL) and two CUT domains (CUT1 and CUT2) that are critical for DNA binding, and a c-terminal HOX domain30 (Fig. 5a). Although HOX domains often serve as a primary DNA binding domain, this is not the case for SATB1/231. The primary function of the HOX domain in SATB1/2 is unclear. We generated five SATB2 mutant proteins, each lacking one of the five functional domains (Fig. 5a). Co-IP studies with the mutant SATB2 proteins revealed that MTA2 interacts with SATB2 primarily via the HOX domain (Fig. 5b).

Fig. 5: SATB2 and MTA2 co-bind HNF4A on colonic chromatin but SATB2 retains HNF4A more strongly than MTA2.
figure 5

a, b We generated 5 mutant SATB2 proteins (M1-5) with each lacking one of the 5 functional domains. a Co-IP of the SATB2 mutants and MTA2 showed that the SATB2 HOX domain was required for SATB2 interaction with MTA2; without the HOX domain, the interaction was abrogated. M: mutation form. b Mean ± S.D. n = 6. All the different gels/blots were derived from the same experiment and were processed in parallel. Source data are provided as a Source Data file. Overlap of SATB2, MTA2, and HNF4A distal genomic binding sites in colonic tissues as shown in the Venn diagram (c) and the DNA binding profiles (d). Peaks were ranked by descending MTA2 occupancy. e DNA binding profiles of HNF4A sites that were either reduced (left panel, 5835 sites) or gained (right panel, 8531 sites) in Satb2cKO colon. In comparison, HNF4A loss or gain at these sites were modest in Mta2cKO colon, but nonetheless statistically significant by Kolmogorov-Smirnov test (K-S D values shown in the density plots). Peaks centered on HNF4A binding sites in 6 kb windows. f Genome Browser tracks of HNF4A binding at genomic loci of the small intestine genes Fabp6, Bcl2l15, and Sis and the colonic genes Car1 and Tspan33 in Mta2cKO and Satb2cKO colon. SATB2 can more strongly influence HNF4A binding than MTA2.

The physical interaction of MTA2 and SATB2 suggests that both proteins might co-localize with HNF4A on colonic chromatin. Because both MTA2 and SATB2 regulate HNF4A primarily at distal genomic sites (this study and reference 21), we assessed distal co-localization of the three factors. Alignment of MTA2 peaks with published SATB2 ChIP data showed 44.5% co-occupancy at distal elements, whereas 36.8% of distal MTA2 sites were co-bound by both SATB2 and HNF4A (Figs. 5c, d). Despite this extensive co-localization, SATB2 loss in the colon activated more small intestine genes than MTA2 loss, and larger numbers of HNF4A binding sites were lost and gained in Satb2cKO (Villin-CreER; Satb2f/f) than in Mta2cKO colon. The lost and gained sites correspond to sites of open chromatin in the colon and ileum, respectively (Figs.5e, f, and 6a).

Fig. 6: SATB2 regulates genomic binding of both MTA2 and HNF4A.
figure 6

a DNA binding profiles of HNF4A sites that were either reduced (C2, 5,835 sites) or gained (C3, 8,351 sites) in Satb2cKO colon. Corresponding MTA2 binding profiles and ATAC profiles in wild-type colon and ileum were shown. MTA2 binding loss and gain paralleled that of HNF4A in Satb2cKO colon, indicating that both MTA2 and HNF4A genomic binding were regulated by SATB2. Plots are 6-kb windows centered on each MTA2 binding site. b Genome Browser tracks showed coordinated shifts in MTA2 and HNF4A binding, ATAC and mRNA profiles at genomic loci of three colonic genes (Car1, Sult1a1, and Aqp4) and two small intestine genes (Fabp6 and Bcl2l15) before and after Satb2 deletion.

In Satb2-null colon, large numbers of MTA2 genomic binding sites shifted in parallel with those of HNF4A (Figs. 6a, b). In contrast, many fewer SATB2 binding sites were depleted or gained in Mta2-null colon (Fig. 7a) and these alterations were not associated with dysregulation of colonic or ileal genes (Figs. 7b, c). Thus, SATB2 regulates genomic binding of MTA2, but not vice versa. Altogether, these data imply that both MTA2 and SATB2 restrain HNF4A at colonic chromatin, but SATB2 regulates MTA2 and more robustly retains HNF4A binding.

Fig. 7: SATB2 genomic binding is affected minimally by Mta2 loss.
figure 7

a DNA binding profiles of SATB2 that were unchanged (C1, 23,177 sites), reduced (C2, 1951 sites), or gained (C3, 700 sites) in MTA2cKO colon. Plots are 6-kb windows centered on each HNF4A binding site. b Predictions of enhancer regulatory functions by BETA (Binding and Expression Target Analysis) indicate that loss or gain of SATB2 binding in Mta2cKO colon was not associated with transcriptional changes of colonic or ileal genes. Plots depict the cumulative score of regulatory potential for every gene based on enhancer distances from the TSS. Black lines represent the background of unaltered genes, and p values denote the significance of positive or negative associations relative to the background. The P value calculated using BETA (Binding and Expression Target Analysis) which used one tail KS-test. c Genome Browser tracks showed that MTA2 regulated HNF4A genomic binding and transcriptional activation of two small intestine genes (Fabp6 and Mttp) without affecting SATB2 binding.

Small intestine gene activation in both Mta2cKO and Satb2cKO colon depends on HNF4A

Our chromatin mapping data implicated HNF4A as a key mediator of colon-small intestine plasticity. Gain of HNF4A binding on small intestine chromatin is tightly associated with transcriptional activation of small intestine genes in Mta2cKO or Satb2cKO colon (this study and reference 21). If this association is causal, then removing HNF4A should block colonic transcriptional plasticity. To evaluate this hypothesis, we differentiated murine colonic organoids into colonoids enriched for CA1+ colonocytes (Figs. 8a and b). We used CRISPR-Cas9 to delete Hnf4a from either Mta2cKO or Satb2cKO colonoids, achieving deletion efficiencies > 90% (Figs. 8c and e). qPCR analysis of representative small intestine genes showed that their activation was strongly attenuated in both mutants (Figs. 8d and f). Thus, small intestinal gene activation in both Mta2cKO and Satb2cKO colon depends on Hnf4a.

Fig. 8: Activation of small intestine genes in Mta2cKO and Satb2cKO colon depends on HNF4A.
figure 8

a, b Bright field and immunofluorescence pictures of differentiated (DE) colonic organoids. CA1 staining showed enrichment of colonocytes in these organoids. BF: bright field. Three independent experiments were repeated with similar results. Scale bar = 100 μm. c Immunoblot and quantification showed that HNF4A levels were comparable in wild-type (WT) vs. Satb2cKO colonoids. CRISPR reduced HNF4A in Satb2/Hnf4a double knockout colonoids to less than 5% of control levels. Mean ± S.D. n = 3. P value by Unpaired t-test, two-tailed. All the different gels/blots were derived from the same experiment and were processed in parallel. d QPCR showed that loss of HNF4A blocked small intestine gene activation in Satb2 mutant colonoids. Mean ± S.D. n = 3 independent mice. P value by Unpaired t-test, two-tailed. e Immunoblot and quantification showed that HNF4A levels were comparable in wild-type (WT) vs. Mta2cKO colonoids. CRISPR reduced HNF4A in Mta2cKO colonoids to less than 2% of control levels. Mean ± S.D. n = 3 independent mice. P value by Unpaired t-test, two-tailed. All the different gels/blots were derived from the same experiment and were processed in parallel. f Loss of HNF4A attenuated small intestine gene activation in Mta2 mutant colonoids. Satb2 mRNA levels were not altered by Hnf4a deletion. Mean ± S.D. n = 3 independent mice. P value by Unpaired t-test, two-tailed.

Increasing HNF4A dosage in the colon activates small intestine gene transcription

Given that colonic enhancers are occupied by HNF4A whereas the small intestine enhancers are primed but lack HNF4A binding in colon, we reasoned that “excess” HNF4A, provided by ectopic expression, should engage small intestine enhancers in colon and activate transcription. To test this hypothesis, we over-expressed HNF4A in cultured mouse colonic organoids. RNA-seq showed 89 up-regulated and 7 down-regulated genes (log2 fold-change >1, p < 0.05, Figs. 9a, f). Gene sets characteristic of small intestine functions, such as cholesterol metabolism, fat and protein digestion and absorption, and retinol metabolism, were enriched among the up-regulated genes (Fig. 9b). We next over-expressed HNF4A in 5 independent human colonic organoid lines. RNA-seq studies revealed 90 up-regulated and only 2 down-regulated transcripts (log2 fold-change >1, p < 0.05, Figs. 9c, d, f). Small intestine functional pathways, including fat, protein and carbohydrate digestion and absorption, were activated (Figs. 9e and f). Thus, loss- and gain-of-function studies implicate HNF4A as a mediator of small intestine gene activity in colonic plasticity, conserved across species.

Fig. 9: An increase in the level of HNF4A in the colon activates the transcription of small intestine genes.
figure 9

RNA-sequencing showed over-expression of HNF4A (Hnf4aOE) in mouse colonic organoids led to predominant up-regulation of small intestine genes (a, volcano plot) enriched for pathways characteristic of small intestine functions as shown in the GSEA plot. P value calculated by (a) Wald test or (b) phenotype-based permutation test and adjusted by Benjamini-Hochberg method (a and b). c Principal component analysis (PCA) of human colonic organoid transcriptomes from control and HNF4A over-expression (Hnf4aOE) samples. Numbers denote human organoid lines used. RNA-sequencing showed predominant up-regulation of small intestine genes (d, volcano plot) in human colonic organoids over-expressing HNF4A. The up-regulated genes were enriched for small intestine functional pathway (e). P value calculated by (d) Wald test or (e) phenotype-based permutation test and adjusted by Benjamini-Hochberg method (d and e). f Heatmaps of all differentially expressed genes (LFC > 1, P < 0.05) in Hnf4aOE murine colonic organoids (left panel) or human organoids (right panel). P value by Wald test and adjusted by Benjamini-Hochberg method.

HDAC activity is required for small intestine gene activation in MTA2cKO colon

We hypothesized that Mta2 loss may lead to compositional and/or conformational changes of the SATB2-NuRD complex, resulting in altered HNF4A binding at colonic enhancers and activation of small intestine genes. Indeed, AP-MS of the SATB2 complex from Mta2cKO colon showed 71 enriched and 25 depleted proteins, compared with wild-type colon (signal intensity >2-fold or <2-fold in mutant vs. control samples, Supplementary Data 5). Immunoblots of colonoids showed less of the NuRD core subunit HDAC2 (P = 0.018) in Mta2 mutants whereas other core subunits, CHD4 and HDAC1, were no different (Fig. 10a). Co-IP studies, however, revealed stronger interactions of SATB2 with both HDAC1 and HDAC2, but not with CHD4, after Mta2 loss (Fig. 10b). Treatment of Mta2 mutant colonoids with HDAC1/2 inhibitors 4-phenylbutyric acid (4PBA) and SAHA strongly attenuated expression of small intestine genes, including Abcg8, Lgals2, Pla2g2a and Slc43a1 (Fig. 10c), suggesting that enhanced HDAC1/2 activities near SATB2 may weaken HNF4A binding and drive small intestine gene activation. The active enhancer mark H3K27ac was not reduced at genomic sites depleted of HNF4A in Mta2cKO colon, indicating that H3K27ac is not a primary target of HDAC1/2 at colonic enhancers (Figure. S5).

Fig. 10: HDAC1/2 interact with SATB2 more strongly in MTA2cKO vs. control colonoids, and HDAC activity is required for small intestine gene activation in MTA2cKO colon.
figure 10

a immunoblots and quantification of SATB2 and core NuRD subunits showed a slight decrease of HDAC2, but no change in HDAC1 or CHD4 in MTA2cKO vs. control colonoids. Mean ± S.D. n = 3 independent samples. P value by unpaired t test with Welch correction, adjusted by FDR (1%). All the different gels/blots were derived from the same experiment and were processed in parallel. b Co-IP and quantification showed stronger interaction of HDAC1/2 with SATB2, but not CHD4 with SATB2 in MTA2cKO vs. control colonoids. Mean ± S.D. n = 3 independent samples. P value by unpaired t-test with Welch correction, adjusted by FDR (1%). All the different gels/blots were derived from the same experiment and were processed in parallel. c QPCR showed treatment with the HDAC inhibitors 4PBA and SAHA attenuated small intestine gene activation in MTA2cKO organoids. Mean ± S.D. n = 3. P value by Unpaired t-test. d A proposed model of colonocyte plasticity regulation in which MTA2 and SATB2 form a chromatin complex at colonic chromatin to retain HNF4A. SATB2 restrains HNF4A more tightly than MTA2. MTA2 loss leads to a modest depletion of HNF4A on colonic and gain on small intestine (SI) chromatin, and modest down- and up-regulation of colonic and small intestine (SI) genes. In contrast, SATB2 loss results in the untethering of large numbers of HNF4A and consequently transcriptomic shift from colon to small intestine. Yellow bar denotes primed small intestinal enhancers in colon.

Collectively, our data support a model in which chromatin factors MTA2-NuRD and SATB2 form a complex to retain HNF4A at colonic chromatin and the degree of plasticity relates to the amount of HNF4A released from that retention. If relatively little HNF4A is liberated, as with MTA2 loss, then the increase in small intestine gene activity is modest; if more HNF4A is released, as occurs with SATB2 loss, then a larger transcriptomic shift ensues, with overt phenotypic tissue conversion (Fig. 10d).