Loss of AHR delays termination of the regenerative response following acute colonic injury
To clarify the role of AHR in epithelial cell maintenance under steady-state conditions, we compared the number of mature epithelial subsets and proliferating cells between Vil-cre AHRfl/fl mice and their wildtype littermate controls. Epithelial-intrinsic loss of AHR at steady state did not impact the differentiation of colonic epithelial subsets, nor result in hyperproliferation (Supplemental Fig.S1), suggesting that AHR regulation of this process is dispensable, at least under the steady state conditions within our animal facility. The protective roles of AHR in epithelial barrier function have largely been demonstrated in the context of barrier perturbation by infection or following injury with DSS [10, 12, 16, 19, 22], indicating that AHR may have a prominent role in the reacquisition of colonic epithelial homeostasis post-injury; a process whose dysregulation could render epithelial cells susceptible to malignant transformation [2, 3]. To investigate the epithelial-intrinsic role for AHR during tissue regeneration, we characterized the response of Vil-cre AHRfl/fl and littermate-controls to an acute model of colonic epithelial injury (2% DSS) for 5 days.
Expression of the regenerative marker Sca-1 was evident in injured epithelial cells as early as day 4 post-DSS administration (Fig.1A) in line with previous reports [1]. Furthermore, epithelial cells within these regenerative foci exhibited marked tissue hyperplasia, alterations to crypt morphology and mucosal thickening characteristic of an epithelial wound repair response (Fig.1B). This was accompanied by a loss of differentiated cells such as mucus secreting goblet cells (Fig. 1C). In both WT and Vil-cre AHRfl/fl mice, peak epithelial Sca-1 expression was observed at day 12 post-DSS administration, with epithelial cells in Vil-cre AHRfl/fl having significantly higher Sca-1 expression compared to WT littermates (Fig.1D). By day 21, regenerative foci were resolved in WT control mice but persisted in Vil-cre AHRfl/fl mice even at D30 post-injury. This was accompanied by the persistent reduction of markers expressed by mature epithelial cells such as Muc2 and Krt20 (Fig.1E), indicating that an epithelial-intrinsic loss in AHR activity results in an impaired termination of the regenerative response.
Transcriptional profile of AHR deficient colon organoids recapitulates defective termination of the regenerative response
Intestinal organoid formation mimics the regenerative response in vivo, and similarly requires transient Yap1 activation [21]. To gain a deeper understanding of epithelial defects resulting from AHR loss of function, we compared transcriptome profiles of wildtype and AHR-deficient colon organoids grown under regenerative (WENR, Wnt3a supplementation) or differentiating conditions (d4 ENR, 4 days post-Wnt3a withdrawal (Supplemental Fig. 2A).
In both conditions, AHR KO organoids showed a pronounced alteration in their transcriptional profile in comparison to WT organoids. The number of differentially expressed genes (DEGs) were significantly higher in the d4 ENR conditions (2055, ¯ 1967 DEGs) than WENR conditions ( 622, ¯ 295 DEGs). Notably, 50-60% of DEGs identified in the WENR condition remained altered in the d4 ENR condition with many of these genes involved in processes such as cell migration/adhesion and metabolism (Supplemental Fig.S2B). Hallmark gene set enrichment analysis (GSEA) of DEGs identified in AHR KO vs WT organoid grown under WENR conditions showed a modest enrichment for pathways involved in wound repair and epithelial mesenchymal transition (EMT), coagulation or STAT-signalling (Fig. 2A). In contrast d4 ENR AHR KO organoids showed pronounced changes in several pathways (Fig. 2B). Pathways known to be enriched in mature colonic epithelium such as fatty-acid metabolism, peroxisome function, bile acid metabolism were negatively enriched in AHR KO epithelial cells, whereas pathways associated with stemness and injury/regeneration such as those involving Myc and E2F signalling (proliferation processes), unfolded protein response (ER stress) and epithelial-mesenchymal transition (EMT) were positively enriched (Fig. 2B). These findings highlight AHR involvement in the resolution of the regenerative program and acquisition of mature epithelial identity.
In order to identify regulatory networks influenced by AHR signalling that could account for these changes, we used the IPA Ingenuity software to pinpoint upstream transcriptional regulators predicted to be activated or inhibited in AHR KO organoids in either WENR or d4 ENR conditions. In WENR AHR KO organoids, genes regulated by tumor-suppressor p53 and b-catenin were found to be moderately enriched (positive z-score) (Fig. 2C; Supplemental table 1). In d4 ENR conditions, Yap1, along with other proto-oncogenes genes such as Myc, b-catenin and Foxm1, were predicted to be activated upstream of genes upregulated in AHR KO compared to WT organoids (Fig. 2C). In contrast, downregulated genes were identified as targets of key TFs involved in intestinal epithelial differentiation such as Cdx2, Hnf1a and Hnf4a (negative z-score) (Fig. 2C; Supplemental table 2). Yap1 is required for colonic epithelial regeneration both in vivo and in vitro, and its activation results in the transient acquisition of a “fetal-like” transcriptional signature which resolves upon exit from the regenerative program. Through comparison of our d4 ENR data to published datasets generated from fetal-spheroids or mature adult organoids [23] we found that the gene signature in d4 ENR AHR KO organoids bore more similarity to a fetal-spheroid signature (Fig. 2D) whereas the WT organoids exhibited a signature more similar to mature adult organoids (Fig. 2E) further illustrating the failure of AHR KO organoids to terminate the regenerative program.
Yap/fetal-like regenerative signature is retained in d4 ENR AHR KO organoids
AHR KO organoids grown in d4 ENR conditions also showed enrichment for a conserved Yap1 signature, expressing higher levels of canonical Yap1 transcriptional targets such as Ctgf and Cyr61 alongside other fetal-like markers associated with colonic epithelial regeneration compared to WT organoids (Fig.3A, B). This corresponded with higher surface expression of Sca-1 compared to WT organoids (Supplemental Fig.S2C). WNT signaling is required for the initiation of the Yap-dependent regenerative program [1, 24, 25]. In line with the role of AHR in restricting Wnt function, AHR KO organoids also showed increased expression of some canonical WNT targets such as, Lgr5 and Ascl2 and Wnt/Yap1 target Sox9 (Fig.3C). To determine whether the expression of Yap1 target genes can be reversed by AHR-signaling, we stimulated WT and AHR KO organoids with the high affinity ligand FICZ under WENR conditions. FICZ-stimulation under regenerative conditions caused the downregulation of Sca-1 (Fig. 3D) and Yap1 targets in an AHR-dependent manner (Fig. 3E). Extracellular matrix (ECM) components are key factors that regulate the wound repair process [1]. Yap1 functions as a mechano-sensor, and upon tissue damage increased deposition of Collagen I within the wound bed and consequently ECM-stiffness, leads to Yap activation in a Wnt and FAK/Src dependent manner. To test whether AHR KO organoids are more sensitive to mechano-activation, we grew cells in a matrix enriched for collagen I (60% collagen+ 40% Matrigel). WT and AHR KO organoids grown in d4 ENR conditions responded to mechanical stress by upregulation of Yap targets (i.e. Ctgf) at the expense of differentiation markers (i.e. Slc26a3) (Fig.3F). Notably, WT organoids grown in 60% collagen expressed these genes at levels similar to AHR KO organoids grown in 0% collagen which may indicate altered mechano-sensing in the absence of AHR. These changes were reversible in WT organoids upon activation of AHR-signaling with AHR agonist FICZ. Thus, loss of AHR signaling impaired termination of the regenerative state, characterized by sustained expression of YAP targets in differentiating conditions and that Yap1-mediated transcriptional regulation can be reversed upon activation of AHR signaling.
Loss of AHR causes prolonged chromatin accessibility to YAP/TEAD targets
AHR activation following FICZ-stimulation resulted in the downregulation of canonical Yap1 targets, suggesting a direct role for AHR in antagonizing Yap/Tead-dependent transcriptional activity. However, our RNA-seq analysis of AHR KO organoids in both WENR and d4 ENR state did not reveal changes in the expression of known regulators of Yap1 transcriptional activity, and total Yap1 protein levels were largely comparable with WT levels at d4 ENR (data not shown). Given that global re-wiring of transcriptional networks during intestinal organoid differentiation is preceded and accompanied by mass remodelling of the epigenome [26], we questioned whether AHR might antagonize Yap/Tead-mediated transcriptional activity through regulating chromatin accessibility to target genes. To address this, we compared ATAC-seq datasets generated from WT and AHR KO organoids grown in either WENR or d4 ENR conditions and characterized global changes in chromatin accessibility as cells transitioned from a regenerative to a differentiated state. We found that the epigenetic landscape of organoids grown in WENR conditions was comprised of largely open chromatin, whereas that accessibility declined under differentiating, d4 ENR conditions [27] (Supplemental Fig.S3A). Differentially accessible regions between AHR KO and WT organoids grown in d4 ENR conditions (1829, ¯415; Fig.4A) or WENR conditions (453, ¯723; Supplemental Fig.S3C) were primarily located in intronic and intergenic regions, suggesting that AHR may influence chromatin accessibility at distal enhancers (Supplemental Fig.S3B). Next, we integrated the d4 ENR ATAC-seq data (AHR KO vs. WT) with RNA-seq data generated from d4 ENR organoids (AHR KO vs. WT) to evaluate whether changes in accessibility correspond with the altered transcriptional profile of AHR KO organoids. Out of differentially accessible genes identified in AHR KO organoids, 22.48% corresponded with transcriptional changes observed in the RNA-seq data (Fig 4B; Supplemental table 3). Gene ontology analysis of targets identified to overlap between both datasets pinpointed pathways affected by loss of AHR function, corresponding with changes in chromatin accessibility in d4 ENR organoids (Fig 4B). Genes annotated to differentially accessible peaks overlapping with d4 ENR RNA-seq data were involved in pathways associated with cell motility, migration and tissue morphogenesis – pathways that are highly regulated by Yap/Tead transcription factor (TF) activity [28-30]. Indeed, we found that several Yap/Tead targets such as (e.g. Ctgf, Cyr61, Amotl2) and fetal-like genes (e.g. Clu, Tacstd2, Ly6a/Sca-1) fell into this category of overlapping genes, with significantly increased accessibility in differentiating AHR KO organoids (Fig.4C). This corresponded with the higher expression of these genes in d4 ENR conditions in comparison to WT controls (Fig.3B). In contrast, genes involved with cellular differentiation such as Slc26a3, Muc2, ChgA did not exhibit differences in chromatin accessibility compared to WT controls (Supplemental Fig.S3D). Thus, AHR may antagonize Yap1 transcriptional activity by restricting accessibility of its co-activator Tead to target genes.
To test this hypothesis, we used TOBIAS, a transcription factor (TF) footprinting tool [31] to predict differential TF occupancy on the genome of AHR KO vs WT organoids as they transitioned from WENR to d4 ENR conditions. In both WT and AHR KO organoids, the differential binding score/transcriptional footprint for Ap-1 and Tead, which drive chromatin opening and Yap1-mediated transcription, was notably decreased, whereas pro-differentiation TFs Hnf4a and Rxra, had an increased transcriptional footprint in d4 ENR organoids (Supplemental Fig.S4A; Supplemental table 4). While both WT and AHR KO organoids seemed to initially share predicted differential binding of TFs during WENR to d4 ENR transition, d4 ENR AHR KO organoids retained a higher differential binding score and footprint for Tead factors compared to d4 ENR WT controls (Fig.4D-F; Supplemental table 4). Differentially accessible sites in AHR KO organoids also exhibited higher scores for Ets-like family of TFs upon differentiation which was not evident in WT organoids (Fig.4D). As multiple TFs can bind similar motifs, we clustered TF motifs based on their archetype group [32], highlighting that many of the identified factors with an increased differential binding score in d4 ENR AHR KO organoids belong to the Tead and Ets-like family of TFs (Fig.4E-F).
In the WENR state, Cdx2 was predicted to have decreased binding in AHR KO organoids, whereas Ap-1 family transcription factors were predicted to have increased binding in WT organoids (Supplemental Fig.S4B, Supplemental Table 4). Collectively, these findings suggest that while both genotypes undergo similar trajectories during the differentiation process, loss of AHR perturbs the overall chromatin landscape, influencing the binding of key TFs that dictate the balance between regenerative and pro-differentiation programs.
AHR KO organoids show impaired differentiation
AHR has been shown to promote differentiation in a variety of cell types [33, 34] and our data indicate that loss of AHR in epithelial cells prolongs the regenerative program at the expense of differentiation (Fig.2A-E). To assess whether AHR targets specific epithelial cell subtypes or influences the differentiation program in general, we compared our d4 ENR transcriptomic datasets with a published dataset generated from single cell transcriptomic analysis of epithelial cells across the small intestine [35]. The DEGs from d4 ENR AHR KO organoids positively correlated with the transcriptional signature of intestinal stem and progenitor cells, and were negatively enriched for mature epithelial subsets, particularly enterocytes (Fig.5A). For example, AHR KO organoids expressed considerably lower levels of maturation markers typical for colonocytes (e.g. Slc26a3, Car4, Alpi), goblet cells (e.g. Clca3b, Muc2) and enteroendocrine cells (e.g. ChgA) (Fig. 5B, C). Gene ontology analysis (GO Biological process) also revealed that many downregulated genes in d4 AHR KO organoids are involved in key functions of mature intestinal cells such as absorption and digestion (Fig.5D, E). Intestinal organoids undergo significant metabolic rewiring as they transition from WENR state to differentiating ENR conditions, characterized by increasing metabolic reliance on oxidative phosphorylation at the expense of glycolysis [36]. To get an additional insight into how AHR regulates the differentiation process, metabolic features of stem cell and differentiating colonic organoids were assessed. WT organoids underwent a metabolic shift indicated by an increasing OXPHOS/Glycolysis ratio (Fig. 5F) and a concomitant decrease in the glycolytic index following the initiation of the differentiation process. In contrast, AHR KO organoids showed sustained preference for glycolysis for its energy demands (Fig.5F) and impaired ability to undergo a switch to increased OXPHOS reliance under differentiating conditions. This finding is complementary to the decreased enrichment of d4 ENR AHR KO organoids for genes associated with oxidative phosphorylation (Fig.2B).
AHR directly regulates key factors involved in the regulation of the regenerative response
AHR functions as a ligand-dependent transcription factor in mammalian cells, and we hypothesized that AHR may regulate the exit from the regenerative state and drive differentiation through transcriptional regulation of key factors involved in this process. To address this, we conducted CHIP-seq analysis of FICZ treated WT and AHR KO organoids to identify ligand-dependent AHR-specific targets under WENR conditions. Through this analysis, we identified 121695 peaks bound by AHR annotated to 18416 unique gene targets (Supplemental table 5). To focus on targets most likely to have a biologically relevant outcome in our system, we restricted our analysis to binding sites in epigenetically accessible regions of the genome that would identify AHR targets primed for activation or repression in the WENR state. We therefore overlapped our ATAC seq dataset from WT organoids grown under WENR conditions with our CHIP-seq data. Our analysis identified 3662 targets annotated to 2785 unique genes, with AHR binding occurring in regulatory elements within intragenic (43.8%) and intergenic (42.1%) regions similar to what has been described in other AHR-CHIP datasets (Supplemental Fig.S5A; Supplemental table 5) [37, 38]. HOMER motif analysis of these AHR-bound regions revealed the Ahr:Arnt motif as the most enriched, followed by the Ap-1 motif (Supplemental Fig.S5B). Next, we integrated this dataset with transcriptomic data generated from RNA-sequencing of WT and AHR KO organoids stimulated with FICZ (4-hours post-stimulation; WENR) to further refine our analysis to AHR-dependent targets with a transcriptional outcome. We found that 13.52% of AHR targets overlap with approximately 21.42% of differentially upregulated and 19.63% of downregulated genes (Fig 6A, Supplemental Table 6). Collectively we labelled AHR targets identified from the integration of these datasets as “active” AHR targets. Gene ontology analysis revealed enrichment for several pathways involved in epithelial differentiation and tissue/structural morphogenesis in genes activated upon AHR activation, whereas genes involved in the regulation of cell migration and biological adhesion processes were downregulated (Fig. 6B). Notably, we identified Cdx2, a key transcription factor involved in intestinal epithelial differentiation and specification, as an active AHR target, alongside other transcription factors known to amplify and/or crosstalk with Cdx2 in the intestine such as Cdx1, Hnf1a [39-41] and Rxra [20] (Fig. 5C; Supplemental Table 4). Conversely, Sox9 (a Wnt and Yap1-target gene) and some canonical targets of Yap/Tead such as Cyr61 and IL33 which have known roles in stem cell maintenance, cell migration and tumor invasiveness - were repressed by AHR (Fig. 6C). AHR bound to an annotated enhancer element downstream of the Cdx2 gene (OREG1865271; PAZAR) involved in positive regulation of Cdx2 (Fig. 6D, F). In contrast, AHR bound to the Sox9 promoter and repressed its expression (Fig. 6E, F). Cdx2/Hnf1a and Sox9 were predicted to be functionally inhibited and activated, respectively, in the IPA upstream regulator analysis (Figure 2B; Supplemental table 2) and our integrated CHIP-seq analysis validated these factors as key transcriptional targets of AHR. Thus, in addition to restricting chromatin accessibility to Yap/Tead targets, AHR orchestrates the termination of the regenerative state through direct transcriptional regulation of genes involved in intestinal epithelial differentiation, and repression of genes involved in stem cell and wound-repair associated programs.
AHR is required for the reacquisition of intestinal identity by optimal induction of Cdx2
Cdx2 is required for differentiation towards mature epithelial cell fates – especially enterocytes and Muc2 expressing goblet cells [40, 42-44]. Epithelial-specific deletion of Cdx2 results in the anteriorization of the intestine, causing the expression of gastric-like genes [45]. Given that Cdx2 is positively regulated by AHR (Fig.6D, Fig.2C), along with TFs such as Hnf1a and Rxra which synergize with Cdx2 to drive intestinal epithelial differentiation, we assessed whether loss of AHR corresponds with changes due to Cdx2 deficiency by comparing the transcriptional signature of d4 ENR AHR KO and Cdx2 KO organoids. We observed a strong positive correlation between both datasets, with the majority of genes up or downregulated in Cdx2 KO organoids corresponding with changes observed in the transcriptional signature of AHR KO organoids (Fig. 7A). In addition, the transcriptional signature of AHR KO organoids was negatively enriched for the expression of genes characteristic for the small intestine and traverse/proximal colon where functions such as absorption and generation of mucins for barrier protection occur [46]. Instead, AHR KO organoids exhibited positive enrichment for genes expressed in anterior regions of the GI tract such as oesophagus and stomach and for the sigmoid colon which represent regions with decreased Cdx2 activity (Fig. 7B) [47]. Furthermore, we observed increased expression of gastric genes and decreased expression of intestinal genes (Fig. 7C) in d4 ENR AHR KO organoids compared with WT controls. Expression of Cdx2 and Sox9 are inversely correlated during organ specification and colorectal cancer and Sox9 is a known repressor of Cdx2 in the intestine [42, 48, 49] . In line with this, regenerative foci retained in Vil-cre AHRfl/fl at d30 post-DSS exhibited decreased Cdx2 staining and increased Sox9 staining (Fig.7D, E). Collectively, these data demonstrate significant parallels between AHR KO and Cdx2 KO phenotypes and provide evidence for AHR as a key factor driving the re-establishment of intestinal identity.