Establishment of novel organoid cultures with regenerative features
To develop a novel organoid system recapitulating the process of intestinal epithelial regeneration after injury, we focused on identifying factors that support long-term expansion of intestinal organoids expressing an injury-associated regenerative signature. Clu, a specific marker of revival stem cells for intestinal regeneration upon damage to intestinal stem cells,23 as well as regenerative markers specifically expressed in the injury-associated repairing epithelium,21,22,23,24 including Sca1, Anxa1 and Reg3b, were used as major markers for screening. A panel of cytokines and small molecules targeting signaling and epigenetic pathways that govern stem cell expansion were screened. More than 30 primary hits were found to increase the expression levels of screening markers. Next, different combinations of these candidates were tested to explore their potential to further increase the expression of these markers (Supplementary information, Fig. S1a). To this end, a new culturing condition including 8 components (8C; LDN193189, GSK-3 Inhibitor XV, Pexmetinib, VPA, EPZ6438, EGF, R-Spondin 1 conditioned medium, and bFGF) was found to effectively upregulate the expression of these markers in intestinal organoid cultures (Fig. 1a). Notably, organoids cultured in the 8C condition grew faster than conventional intestinal organoids cultured in the presence of EGF, Noggin and R-Spondin 1 (collectively, the ENR condition1) and more robust expansion of Lgr5-GFP+ cells was also observed (Supplementary information, Fig. S1b–d). Freshly isolated Lgr5-GFP+ cells had a higher organoid-forming efficiency than Lgr5-GFP– cells under the 8C condition (Supplementary information, Fig. S1e–g). Organoids cultured in the 8C condition showed the presence of multiple differentiated intestinal lineages, including goblet (Muc2), enteroendocrine (Chga), and Paneth (Lyz1) cells (Supplementary information, Fig. S1h, i). In addition, organoids cultured in the 8C condition also showed genome stability after more than 20 passages (Supplementary information, Fig. S1j).
Next, we analyzed the regenerative features of organoids in our new culturing medium. Importantly, compared with organoids cultured in the ENR condition, the 8C condition significantly promoted the elongation of budding domains with a larger organoid size and more complex crypt–villus structures, which were morphologically similar to those of the hyperplastic crypts of injury-associated repairing epithelium21,22,24 (Fig. 1b; Supplementary information, Fig. S1c). Furthermore, immunofluorescence and flow cytometry analysis showed that key regeneration-associated markers, including SCA1, ANXA1, REG3b, and CLU, were highly expressed in the 8C condition in comparison with the ENR condition (Fig. 1c–e; Supplementary information, Fig. S1k). To functionally examine the regenerative capacity of organoids cultured in the 8C condition, we used an in vitro irradiation model,30 and found that transfer of irradiated ENR-cultured organoids into the 8C condition can maintain the growth of organoids without significantly altering the expression of intestinal lineage markers, in contrast to a rapid collapse of ENR-cultured organoids after irradiation (Supplementary information, Fig. S2a–c). These data suggest that organoids cultured in the 8C condition may acquire the regenerative features of in vivo injury-associated hyperplastic intestinal epithelium.
To further explore the regenerative features of organoids from the 8C condition at the transcriptome level, we performed RNA sequencing (RNA-seq) analysis to examine the expression of regeneration-associated genes. Gene set enrichment analysis (GSEA) showed that reported regeneration-associated genes, including those involved in YAP signaling and a fetal intestinal signature, were significantly enriched in the 8C condition in comparison with the ENR condition (Supplementary information, Fig. S3a). In addition, a previously reported colitis-associated regenerative epithelial signature22 was also highly enriched in the 8C condition (Supplementary information, Fig. S3a). To further verify the hyperplastic signature of organoids, we defined a comprehensive injury-associated regenerative signature underlying regeneration of the repairing epithelium after injury induced by irradiation,23 parasitic helminth infection21 and dextran sulfate sodium (DSS) treatment24 (Supplementary information, Table S1). Importantly, the 8C condition was associated with significant enrichment of the injury-associated regenerative signature, which resembled the phenotype of hyperplastic crypts upon damage in vivo (Fig. 1f; Supplementary information, Fig. S3b). Consistent with this result, correlation and dimension reduction analysis of the expression of injury-associated regenerative signature genes revealed that organoids cultured in the 8C condition resembled hyperplastic crypts from injury-associated repairing epithelium in different injury models more closely than those cultured in the ENR condition (Supplementary information, Fig. S3c, d). Collectively, these results indicate that organoids cultured in the 8C condition have a hyperplastic phenotype mimicking that of hyperplastic crypts upon damage in vivo.21,24,31 Therefore, we designated these organoids as hyperplastic intestinal organoids (Hyper-organoids).
scRNA-seq analysis of hyper-organoids revealed their unique lineage composition
Next, we analyzed the lineage composition of Hyper-organoids using single-cell RNA-seq (scRNA-seq), with organoids cultured in the ENR condition used as controls. Unsupervised clustering of the merged datasets across different organoids revealed 17 distinct epithelial clusters (Fig. 2a, b; Supplementary information, Fig. S4a). In comparison with ENR-organoids, Hyper-organoids showed significant reductions in the abundance of Paneth cells (PCs) and enterocytes, as well as a slight increase in the number of enteroendocrine cells, similar to injury-associated lineage dynamics observed in vivo21,22 (Fig. 2c; Supplementary information, Fig. S4b).
Using reported intestinal stem cell markers, clusters 1, 2, and 12 from the Hyper-organoids, as well as cluster 3 from the ENR-organoids, were identified as Lgr5+ intestinal stem cells (Fig. 2a–c; Supplementary information, Fig. S4a). Notably, despite shared expression of intestinal stem cell markers (Lgr5, Cd44, and Ascl2), there was a clear separation of the Lgr5+ stem cell populations as well as transit amplifying (TA) cells between Hyper-organoids and ENR-organoids (Fig. 2a; Supplementary information, Fig. S4c). To determine the differences of Lgr5+ stem cells between Hyper- and ENR-organoids, we integrated the sequencing data of Lgr5+ stem cells from different types of organoids as well as homeostatic and injury-associated intestinal epithelium21,23 (Supplementary information, Fig. S5a). We found that there were Lgr5-high and Lgr5-low subsets in ENR-organoids, while Lgr5-low subset rarely appeared in Hyper-organoids (Supplementary information, Fig. S5e). Although Lgr5+ population in the Hyper-organoids showed closer transcriptomic relationship with Lgr5-low subset from the ENR-organoids when compared to Lgr5-high subset in the ENR-organoids, the global transcriptomic features of Lgr5+ cells in the Hyper-organoids are distinct from that in the ENR-organoids (Supplementary information, Fig. S5f, g). Interestingly, we found that the global gene profiles of Lgr5+ stem cells in Hyper-organoids resembled that of Lgr5+ stem cells from injury-associated repairing intestinal epithelium (termed as injury-responsive Lgr5+ stem cells), and were significantly different from that of Lgr5+ stem cells in homeostatic epithelium and ENR-organoids (termed as homeostatic Lgr5+ stem cells) (Supplementary information, Fig. S5a–c). Differential gene analysis further showed that injury-responsive Lgr5+ stem cells enriched several key regenerative genes, such as Reg3b, Reg3g, and Sca1,24 and downregulated the expression of stem cell marker Olfm4, which was mainly expressed under homeostatic conditions32 (Supplementary information, Fig. S5c, d).
In addition to injury-responsive Lgr5+ stem cells, we also found that cluster 17, which contained representative marker genes (Clu, Anxa1 and Sca1) of revival stem cells in an irradiation model,23 was uniquely presented in Hyper-organoids (Fig. 2b; Supplementary information, Fig. S4d–f). Notably, the transcriptomic profiles of cluster 17 resembled that of revival stem cells (referred to as SSC2c in in vivo irradiation model23) from irradiated crypts (Supplementary information, Fig. S4e). Furthermore, cluster 17 was also enriched in fetal gene signatures, which represent the primitive molecular features of injury-responsive stem cells in vivo (Supplementary information, Fig. S4e, g).22,23 Taken together, Hyper-organoids contain Lgr5+ and Clu+ stem cell populations that resemble the injury responsive stem cells from the gut epithelium upon damage in vivo.
To further investigate the functionality of Lgr5+ regenerative stem cells in Hyper-organoids, we first purified Lgr5-GFP+ cells from organoids by FACS sorting and tested their self-renewal abilities at the single-cell level. We found that Lgr5+ cells from Hyper-organoids had a significantly higher organoid-forming efficiency than those from ENR-organoids (Supplementary information, Fig. S6a). We further traced the progenies of Lgr5-GFP+ cells by labeling them with tdTomato expression using a tamoxifen-induced CreERT2 system (Fig. 2d), which showed that progenies of Lgr5+ stem cells in Hyper-organoids accounted for more than 20% the total cell population after tamoxifen treatment (Fig. 2e, f; Supplementary information, Fig. S6b). Moreover, the expression of differentiated intestinal markers and regenerative markers co-expressed with tdTomato was also observed (Supplementary information, Fig. S6c, d), suggesting that Lgr5+ cells give rise to other progenies in the Hyper-organoids. Directed differentiation of hyper-organoids by modulating the Wnt and Notch signaling pathways significantly increased the expression of differentiated marker genes for different intestinal lineages, including PCs, enterocytes, enteroendocrine cells, and goblet cells (GCs) (Supplementary information, Fig. S6e). Notably, in an in vitro irradiation model, progenies of injury-responsive Lgr5+ stem cells showed significant enhanced survival ability whereas those of homeostatic Lgr5+ stem cells died completely (Supplementary information, Fig. S2d, e). Collectively, these data suggested that Hyper-organoids contained functional Lgr5+ regenerative stem cells.
VPA and EPZ6438 were critical for regulating hyperplastic features of intestinal organoids
To mechanistically explore the acquirement of regenerative features in Hyper-organoids, we first analyzed the effect of removing individual factors from the 8C condition. Removal of LDN193189 and R-Spondin 1 conditioned medium resulted in growth retardation of organoids. EGF and GSK3iXV were necessary for the continuous passage of organoids, and the addition of bFGF was beneficial for organoid growth (data not shown). Notably, we also found that GSK3iXV, Pexmetinib, VPA, and EPZ6438 were required to maintain the expression of regenerative markers after passaging (Supplementary information, Fig. S7a). Importantly, simultaneous removal of VPA and EPZ6438 led to rapidly and significantly decreased injury-associated regenerative signatures in the organoid within 7 days, whereas omission of GSK3iXV or Pexmetinib did not have this effect (Fig. 3a). Accordingly, VPA and EPZ6438 could be the major drivers of regenerative features in the Hyper-organoids.
We further performed GSEA analysis to analyze the relationship between regenerative features and VPA/EPZ6438 treatment. Downregulation of the reported regeneration-associated signatures in Hyper-organoids was observed upon the removal of VPA and EPZ6438 (Supplementary information, Fig. S7b). In addition, removal of VPA and EPZ6438 also significantly reduced the percentage of fetal gene-expressing cells, and downregulated fetal gene expression (Fig. 3c; Supplementary information, Fig. S7b, f), suggesting that VPA and EPZ6438 are required by Hyper-organoids to maintain a primitive state. Moreover, the dependence of the expression of injury-associated regenerative signature on these two compounds was also supported by dimension reduction analysis, which showed that removal of VPA and EPZ6438 made Hyper-organoids more closely resemble crypts from the homeostatic epithelium (Supplementary information, Fig. S7c). Consistent with these analyses, organoids cultured in the 8C minus VPA/ERZ6438 condition showed reduced complexity of crypt–villus structures and gradually lost Lgr5-GFP expression during serial passages (Supplementary information, Fig. S7d–f).
To further investigate the time window of the induction of the regeneration signature, we performed transcriptional analysis of organoids at different time points during the early stage of converting ENR-organoids into the 8C condition with and without VPA/EPZ6438. We found that representative genes of the regeneration signature were upregulated at day 2 after transferring ENR-organoids to the 8C condition, and were further increased from day 6 to day 8 (Supplementary information, Fig. S7g). In contrast, the upregulation of regenerative markers was significantly reduced upon the removal of VPA/EPZ6438 (Supplementary information, Fig. S7g), suggesting that these two small molecules induced the regeneration signatures at the early stage of 8C treatment. Importantly, we found that addition of VPA and EPZ6438 in ENR condition was sufficient to promote hyperplastic phenotypes of intestinal organoids (Supplementary information, Fig. S8a–d). Collectively, these results suggest that VPA and EPZ6438 are important for regulating the regenerative features of Hyper-organoids.
Next, we examined the effects of VPA and EPZ6438 on the stem cell composition of Hyper-organoids. Unsupervised clustering showed that Hyper-organoids without VPA/EPZ6438 treatment were transcriptomically clustered together with ENR-organoids, especially the stem cell clusters (Fig. 3b, d). In addition, removal of VPA/EPZ6438 led to the conversion of injury-responsive Lgr5+ stem cells back to homeostatic Lgr5+ stem cells (Fig. 3b, d), and also significantly decreased the expression of the revival stem cell signature in Hyper-organoids (Fig. 3e). These results reinforce the notion that VPA and EPZ6438 are critical for the development of injury-responsive stem cell populations in Hyper-organoids.
To further investigate the effect of VPA/EPZ6438 treatment on the induction of injury-responsive stem cells in Hyper-organoids, we reconstructed the regenerative trajectory by performing single-cell analysis at different time points in Hyper-organoids with and without VPA/EPZ6438. Interestingly, cells from Hyper-organoids were mainly enriched along the route starting from PCs, in which GCs and TA cells progressed toward Lgr5+ stem cells (Supplementary information, Fig. S9a). Notably, trajectory reconstruction of PC, GC, TA, Lgr5+ stem cells, and revSC-like cells from Hyper-organoids with and without VPA/EPZ6438 revealed two diverse branches (Fig. 3f). VPA/EPZ6438 treatment drove one branch to form injury-responsive Lgr5+ stem cells and Clu+ revival-like stem cell population, representing the successful branch (Fig. 3f; Supplementary information, Fig. S9a, c). Interestingly, we found that co-expression of PC marker Lyz1 and stem cell markers were dominant in the early pseudo-stage of the route, suggesting that regenerative stem cells potentially originated from PCs in the process of regeneration (Supplementary information, Fig. S9a, b, d). Collectively, our findings support a VPA/EPZ6438-mediated regenerative trajectory towards the generation of injury-responsive stem cells in Hyper-organoids.
Regulation of YAP signaling is important for regenerative responses driven by VPA and EPZ6438 in hyper-organoids
VPA and EPZ6438 are reported to epigenetically regulate global histone modifications,33,34 and we indeed found that their combination led to globally reduced H3K27 trimethylation and upregulated H3K27 acetylation in Hyper-organoids (Supplementary information, Fig. S10a). To further investigate the effects of these epigenetic changes on transcriptional regulation of regeneration in vitro, we analyzed differential gene expression between Hyper-organoids and those cultured under the 8C minus VPA/EPZ6438 condition. Notably, genes that are targets of YAP signaling were upregulated in Hyper-organoids (Supplementary information, Fig. S10b, c). In consistent with these observations, we also found reduced levels of H3K27me3 at YAP target genes in Hyper-organoids as compared to that in organoids cultured in the absence of VPA/EPZ6438 (Fig. 4f, g), revealing a potential epigenetic mechanism underlying YAP-dependent intestinal regeneration.
We further explored the roles of Yap signaling in VPA/EPZ6438-driven regenerative response using single-cell sequencing data. Notably, we observed enrichment of YAP signaling along the regenerative trajectory of Hyper-organoids, which was absent in the failed branch without VPA/EPZ6438 treatment (Fig. 4a–c). Consistent with this observation, the upregulation of the YAP transcriptional program in Hyper-organoids, especially in stem cell clusters, was dependent on the addition of VPA and EPZ6438 (Fig. 4d, e). Using SCENIC, we further analyzed the effects of VPA/EPZ6438 treatment on the gene regulatory network in Hyper-organoids. By comparing Hyper-organoids with ENR-organoids and Hyper-organoids without VPA/EPZ6438 treatment, we identified the gene regulatory network specifically active in Hyper-organoids, which includes one top transcriptional factor Tead2 that can form complex with YAP/TAZ19 (Supplementary information, Fig. S11a). Importantly, the enrichment of Tead2 regulon was observed in Hyper-organoids, especially in injury-responsive Lgr5+ stem cells, which was dependent on the treatment of VPA and EPZ6438 (Supplementary information, Fig. S11b–d). Moreover, organoid-forming efficiency was significantly impaired by the knockdown of Tead2 in Hyper-organoids (Supplementary information, Fig. S11e), suggesting the requirement of Tead2 regulon in maintaining the self-renewal ability of Hyper-organoids. Collectively, these data suggest that Yap signaling is important for regulating the molecular features of regeneration in Hyper-organoids, which is driven by VPA/EPZ6438 treatment.
YAP signaling is crucial for intestinal regeneration.23,32 Therefore, we examined whether YAP also regulates regenerative responses in Hyper-organoids. We observed a marked arrest of Hyper-organoid growth after exposure to Verteporfin (VP), a potent inhibitor of YAP35 (Supplementary information, Fig. S10d). Moreover, the addition of LPA, a YAP signaling agonist that functions by inhibiting LATS1/2,36 partially compensated for the absence of VPA and EPZ6438 in establishing regenerative features in Hyper-organoids (Supplementary information, Fig. S10d).
Interestingly, in addition to YAP signaling, the Wnt signaling pathway also showed significant enrichment in the successful branch during the regenerative process as indicated by KEGG analysis (Supplementary information, Fig. S10e). Moreover, the expression of regenerative markers was significantly reduced in Hyper-organoids after treatment with Wnt/β-catenin signaling inhibitor XAV939 (Supplementary information, Fig. S10f). Therefore, other signaling pathways like Wnt could also be important for driving the regenerative responses in Hyper-organoids.
VPA and EPZ6438 promote intestinal and colonic regeneration following damage in vivo
Since VPA and EPZ6438 were critical for the activation of the injury-associated regenerative signature and phenotype in Hyper-organoids in vitro, we explored whether combined treatment with VPA and EPZ6438 promoted injury-associated intestinal epithelial regeneration in vivo. Following administration of VPA and EPZ6438 in an in vivo irradiation model,23,32 we found that the growth of elongated crypts was markedly enhanced, and that the proliferation of crypts was slightly increased (Fig. 5a–c), indicating that combined VPA/EPZ6438 treatment promoted a regenerative response in the intestinal epithelium upon damage in vivo. In support of these in vivo results, when evaluated in an in vitro irradiation model,37 irradiated ENR-cultured organoids transferred into the ENR condition with the addition of VPA and EPZ6438 supported the survival of intestinal organoids after irradiation. (Supplementary information, Fig. S8e, f).
To gain more insight into the molecular traits of epithelial regeneration after VPA/EPZ6438 treatment, we analyzed the transcriptome of primary crypts after irradiation. Notably, upregulation of representative marker genes of YAP gene signature was observed in a time-dependent manner after VPA/EPZ6438 treatment (Fig. 5d), suggesting promotion of YAP-dependent intestinal generation in irradiated crypts after VPA/EPZ6438 treatment. Interestingly, we also found a significant downregulation of H3K27me3 enrichment in the Yap target genes after VPA/EPZ6438 treatment (Fig. 5e), suggesting epigenetic regulation of YAP signaling underlines the VPA/EPZ6438-driven regenerative response in vivo. Collectively, these data suggest that combined VPA/EPZ6438 treatment promoted intestinal epithelium regeneration upon injury.
We further investigated the effects of VPA/EPZ6438 treatment in promoting injury-associated epithelial regeneration in a DSS-induced colitis model.38,39 We found that combined VPA/EPZ6438 treatment could decrease disease activity index, and attenuate weight loss and colon length reduce (Supplementary information, Fig. S12a–d). In addition, VPA and EPZ6438 treatment markedly reduced pathological changes, with a well-preserved mucosal architecture and small foci of crypt loss, leading to a decreased histological score (Supplementary information, Fig. S12e). Collectively, these data suggest that VPA and EPZ6438 treatment alleviated symptoms of DSS-treated mice.