Tissue extracellular matrix hydrogels as alternatives to Matrigel for culturing gastrointestinal organoids

Characterization of decellularized GI tissue-derived hydrogels

To develop a 3D matrix that better recapitulates the biochemical ECM composition of native GI tissues and ultimately replaces the use of Matrigel in organoid culture, we prepared ECM-based hydrogels derived from decellularized pig stomach and small intestinal tissues (Fig. 1a). First, we optimized decellularization protocols by adjusting the type (non-ionic versus ionic) and treatment time of detergents for decellularization (Supplementary Fig. 1). Using our optimized decellularization protocol (Protocol 1), which is based on a non-ionic detergent (Triton X-100)¹⁷, cellular components were completely removed from stomach and intestine tissues, whereas major ECM components (e.g., glycosaminoglycans; GAG) were preserved^17,18 (Fig. 1b, c). Another decellularization protocol (Protocol 2) that uses ionic detergents (e.g., sodium deoxycholate; SDC) and has been widely used for cell removal^19,20 effectively removed cellular components but impaired ECM preservation as indicated by a decrease in the GAG content (Supplementary Fig. 1a–c). Decellularized stomach-derived ECM (termed “SEM”) and decellularized intestine-derived ECM (termed “IEM”) were lyophilized, solubilized, and then induced to form 3D hydrogels at physiological pH and temperature via the kinetics of collagen fibril assembly (Supplementary Fig. 2). Both SEM and IEM hydrogels possessed a nanofibrous ultrastructure composed of interconnected ECM fibrils similar to collagen type I hydrogels (Fig. 1d)^20,21. The hydrogels prepared with our decellularization protocol exhibited a higher elastic modulus (storage modulus measured at 1 Hz frequency) (1.6–3.3-fold) than those prepared using an ionic detergent (Supplementary Fig. 1d), indicating superior ECM preservation with our protocol. The storage modulus of hydrogel implies the ability of the hydrogel to store deformation energy, and the storage modulus is proportional to elastic energy²². In other words, higher storage modulus indicates higher elasticity and stronger mechanical property of hydrogel. Thus, ECM hydrogels with higher storage modulus may be more appropriate for the formation, growth, and long-term maintenance of organoids. As our optimized decellularization protocol allows a larger amount of ECM preservation and retains the stability of GI tissue-derived ECM hydrogels, the resultant hydrogels are more suitable for GI organoid culture.

We verified the safety of decellularized GI tissue-derived ECM hydrogels. First, we found that endotoxin levels of SEM and IEM were at 0.344 ± 0.007 EU/ml and 0.225 ± 0.016 EU/ml, respectively (Supplementary Fig. 3a). This may indicate that our protocol to decellularize GI tissues and prepare ECM hydrogels can exclude contamination of SEM and IEM hydrogels from pathogens in the GI tissues. Given that the endotoxin level acceptable for implantable medical devices is 0.5 EU/ml in the guideline of the Food and Drug Administration (FDA)²³, our decellularized GI tissue-derived ECM hydrogels would be able to serve biomaterials with clinical feasibility. Next, we checked the immunogenicity of decellularized GI tissue-derived ECM hydrogels. Under the in vitro condition, the secretion of inflammatory cytokine tumor necrosis factor-α (TNF-α) from RAW 264.7 macrophages 3 and 6 h after exposure to SEM and IEM hydrogels was negligible, which was similar to that from cells with no treatment as a negative control (Supplementary Fig. 3b). Hematoxylin and eosin (H&E) staining of the tissue samples retrieved 1, 4, and 7 days after injection SEM and IEM hydrogels into subcutaneous space of mice revealed that neither tissue necrosis nor tissue damage was observed in the tissues injected with SEM and IEM hydrogels (Supplementary Fig. 3c). Toluidine staining also confirmed that excessive infiltration of inflammatory cells was not observed at the injection sites (Supplementary Fig. 3c). These results verified that decellularized GI tissue-derived ECM hydrogels are highly biocompatible and safe materials that can be considered for clinical applications. Decellularization of tissues allows manufacturing natural ECM-based materials with low immunogenicity by removing antigenic cellular components. Decellularized tissue-derived materials free from xenogeneic pathogens can be prepared through decellularization with a series of detergent treatments and sterilization processes (e.g., ethanol, ethylene oxide, ultraviolet (UV) irradiation, and supercritical carbon dioxide). Actually, Matrigel or Matrigel-derived products have never been approved by the FDA, while several commercialized products are based on decellularized tissue ECM (e.g., AlloDerm^®, Meso BioMatrix^®, and SynerGraft^®, etc.) have been approved for clinical use by the FDA.

Intensive proteomic analysis using mass spectrometry (MS) revealed the presence of stomach- and intestine-specific matrisome and non-matrisome components preserved in the SEM and IEM, which were distinct from Matrigel proteome components (Fig. 2a and Supplementary Fig. 4a, b). More core matrisome and matrisome-associated proteins were detected in SEM and IEM compared to Matrigel (Supplementary Fig. 4c). In total matrisome proteins, collagen subtypes and proteoglycans constituted about 67% and 13% in SEM and about 51% and 26% in IEM, respectively, whereas the portion of glycoproteins was about 17% in SEM and about 19% in IEM (Fig. 2c–e, h). In contrast, the majority of matrisome proteins in Matrigel were glycoproteins, which accounted for more than 96% of the total matrisome (Fig. 2b). Collagens are the most abundant structural elements of GI tissue ECM²⁴, and proteoglycans play important roles in the regulation of signaling pathways involved in GI epithelial cell proliferation and tissue homeostasis^25,26,27. Thus, SEM and IEM that contain a large quantity and variety of collagen subtypes and proteoglycans more closely reflect the in vivo GI microenvironment compared to Matrigel with only 0.4% collagen and 1% proteoglycans (Fig. 2b–d). Collagen Type VI (COL6A1, COL6A2, COL6A3, and COL6A5) and proteoglycans, such as decorin (DCN), were identified as the top 10 most abundant matrisome proteins in SEM and IEM, whereas glycoproteins, including laminin-111 (LAMA1, LAMB1, and LAMC1) and fibrinogen (FGA, FGB, and FGG), were the top 10 most abundant matrisome proteins in Matrigel (Fig. 2b–d).

Interestingly, the proteins significantly enriched in native stomach or intestine tissues were detected in much greater quantities in each tissue-specific ECM hydrogel compared to Matrigel (Supplementary Fig. 4d). Whereas Matrigel and IEM did not contain proteins known to be enriched in the native stomach, five non-matrisome proteins specific to the native stomach were identified only in SEM. Similarly, 5 matrisome proteins and 22 non-matrisome proteins enriched in the native intestine were detected only in IEM, verifying that IEM contains substantially more native intestinal proteins than other matrices. Although the exact roles of non-matrisome proteins have been relatively unexplored, several studies have demonstrated that non-matrisome proteins in some tissues are associated with cell growth and development^28,29. Thus, we postulate that non-matrisome components in SEM and IEM hydrogels help the formation and development of GI organoids. In gene ontology biological process (GOBP) analysis of non-matrisome proteins in Matrigel, SEM, and IEM, the top ten representative GOBP terms indicated that the non-matrisome proteins of SEM and IEM were involved in cellular component biogenesis, cytoskeleton organization, and structural development, whereas those of Matrigel were more related to translation and RNA processing and metabolic process (Supplementary Fig. 5). Collectively, we speculate that SEM and IEM derived from decellularized GI tissues may serve as alternative matrices to Matrigel for GI organoid culture by better reconstituting the native GI-like microenvironment.

To evaluate the variation of decellularized GI tissue-derived matrices in the ECM compositions, we compared the samples under two conditions: (1) comparison of multiple sample batches derived from different locations in the same donor tissue (batch variation—porcine A1–A3), (2) comparison of multiple samples derived from different donor tissues (donor variation—porcine A–C). The compositions of matrisome proteins are quite similar in three sample batches of SEM or IEM derived from different locations in the same donor tissue (Fig. 2e, h). Interestingly, there was little variation in the profiles of matrisome proteins in three sample batches of SEM or IEM derived from different donor tissues (Fig. 2e, h). Moreover, the lists of the top 10 matrisome proteins with the highest expression almost overlapped between different batches from the same donor tissue or the tissue samples from different donors (eight proteins in SEM samples and nine proteins in IEM samples) (Fig. 2e, h). More than 100 matrisome proteins were overlapped in three sample batches of SEM or IEM derived from different donor tissues (SEM: 102 and IEM: 114) (Fig. 2f, i). GOBP analysis revealed that these overlapped matrisome proteins are mainly involved in the organization of ECM and extracellular structure (Fig. 2g, j). Principal component analysis (PCA) plot and heatmap of Pearson’s correlation coefficients confirmed that SEM and IEM samples derived from different tissue batches or different donor batches have high similarity in terms of protein compositions and profiles (Fig. 2k–n). Together, these proteomic analytical data demonstrate that decellularized GI tissue-derived matrices have a relatively low batch-to-batch variation in comparison to Matrigel.

To further investigate the variation in matrisome proteins and non-matrisome proteins in SEM and IEM samples, we compared the matrisome proteins and non-matrisome proteins detected in the SEM and IEM that are also expressed in native stomach and intestine tissues, respectively (Supplementary Fig. 6). Interestingly, among proteins highly enriched in the native stomach, one matrisome protein (ANXA10) and 4 non-matrisome proteins (CHIA, CLIC6, GHRL, PGC) were commonly detected in all batches of SEM samples, indicating that there is no significant variation in matrisome and non-matrisome proteins in the SEM from different tissue locations (porcine A1–A3) and donors (porcine A–C) (Supplementary Fig. 6a). In the case of proteins highly enriched in the native intestine, five matrisome proteins (ITLN2, LGALS2, LGALS4, MUC2, TINAG) were detected in all batches of IEM samples and one matrisome protein (REG4) was detected in two IEM sample batches (porcine A2, C). Total 19 non-matrisome proteins (ACTG2, ALDOB, AOC1, APOA4, ARHGAP45, CASP7, EPS8L2, FABP6, GSTA1, KLC4, KRT20, KRT8, LASP1, MYH14, OLFM4, RPS6KA1, SELENBP1, SMTN, and SRI) were commonly detected in all batches of IEM samples (Supplementary Fig. 6b). Other non-matrisome proteins GSDMB and JCHAIN were detected in four (porcine A2, A3, B, C) and three batches (porcine A2, B, C) of IEM samples, respectively. Two non-matrisome proteins (BCL2L15, HSD17B11) were detected in two IEM batches (porcine B, C). Therefore, most of matrisome proteins (5/6) and non-matrisome proteins (19/23) enriched in the native intestine were overlapped in all batches of IEM samples, demonstrating again that the variation in matrisome and non-matrisome proteins in the IEM is insignificant between tissue locations and donors. Moreover, these proteins commonly detected in different batches of SEM and IEM have involved in GI tissue-specific functions such as digestion, intestinal absorption, and gastric acid secretion (Supplementary Fig. 6c, d), indicating that GI tissue-derived ECM hydrogels can provide GI organoids with native GI-like microenvironments in a highly reproducible manner.

Additionally, we analyzed ECM compositions in decellularized tissues from different parts of the GI tract. We decellularized the esophagus, an upper part of the GI tract, and compared ECM profiles of decellularized esophagus-derived ECM (EEM) with those of SEM and IEM. The types and compositions of matrisome proteins in EEM were slightly different from those in SEM and IEM (Supplementary Fig. 7a). Three matrisome proteins (COL6A1, COL6A2, COL6A3) among the most abundant top 10 matrisome proteins in EEM were overlapped with those in SEM and IEM (Supplementary Fig. 7b). Despite such a slight discrepancy, PCA plot confirmed that overall ECM profiles in different parts of the GI tract (IEM, SEM, and EEM) were close each other compared to Matrigel with quite different ECM components (Supplementary Fig. 7c). Together, we conclude that ECM profiles and compositions in decellularized tissues are varying depending on the location in the GI tract, but ECM hydrogels derived from the GI tract have high similarity in comparison to materials from non-GI tract such as Matrigel.

Utility of GI tissue-derived ECM hydrogels as a GI organoid culture matrix

To demonstrate the potential of decellularized GI tissue-derived ECM hydrogels (SEM and IEM) to substitute for Matrigel in organoid culture, we compared GI organoids generated in SEM and IEM hydrogels with those generated in Matrigel in the following four aspects: (i) organoid morphology, (ii) organoid formation efficiency, (iii) organ-specific gene expression level, and iv) organoid functionality. Before using the hydrogels in organoid culture, the rheological properties of SEM and IEM hydrogels were characterized at various concentrations to confirm their mechanical stability for organoid culture. Consistently higher storage modulus (G′) than the loss modulus (G″) in all tested concentrations indicated the stable formation of cross-linked ECM networks in the hydrogels (Supplementary Fig. 8a, b).

First, we identified each optimal concentration of SEM and IEM hydrogels that effectively supported the development of GI organoids. When mouse gastric glands were seeded in SEM hydrogels at different ECM concentrations (1, 3, 5, 7 mg ml⁻¹), gastric organoids formed in all tested concentrations (Fig. 3a), and the highest formation efficiency was observed at 5 mg ml⁻¹ (Fig. 3b). Quantitative real-time polymerase chain reaction (qPCR) demonstrated that the gene expression of a stem cell marker (Lgr5) was similar in gastric organoids grown in SEM hydrogels at concentrations of 3 and 5 mg ml⁻¹ and organoids grown in Matrigel. When we checked the expression levels of gastric epithelial cell markers (chief cell marker Pgc and parietal cell markers Atp4a, Atp4b) in the organoids grown in 5 mg ml⁻¹ SEM hydrogels, the expression of Pgc and Atp4a was comparable to that of Matrigel organoids and the expression of Atp4b was higher in the organoids grown in 5 mg ml⁻¹ SEM hydrogels than in the organoids in Matrigel (Fig. 3c). In general, stemness (Lgr5 and Axin2 for gastric stem cells) and gastric differentiation markers (Muc6 for mucous neck cells, Gif for parietal cells, and Pgc and Pga5 for chief cells) of gastric organoids grown in 5 mg ml⁻¹ SEM hydrogel were expressed at similar levels to those grown in Matrigel (Fig. 3d). Considering gastric organoid formation efficiency and gene expression levels, we selected 5 mg ml⁻¹ SEM hydrogel for further studies.

Similar experiments were conducted to optimize the concentration of IEM hydrogel used in intestinal organoid culture. All mouse intestinal crypts embedded in IEM hydrogels with 1–5 mg ml⁻¹ concentrations developed into intestinal organoids (Fig. 3g), although the formation efficiency was slightly lower than that of Matrigel (Fig. 3h). The expression levels of stem cell marker (Lgr5) and goblet cell marker (Muc2) in intestinal organoids encapsulated in IEM hydrogels were similar to or higher than those embedded in Matrigel (Fig. 3i), and the highest expression was observed at 2 mg ml⁻¹. All other intestinal differentiation markers (Lyz1 for Paneth cells, Chga for enteroendocrine cells, and Vil1 for enterocytes) of intestinal organoids grown in 2 mg ml⁻¹ IEM hydrogel were expressed at comparable levels to those grown in Matrigel (Fig. 3j). Given these results, we selected 2 mg ml⁻¹ IEM hydrogel for further studies in intestinal organoid culture. Although increased expression of stemness markers (Lgr5, Axin2) was observed in the intestinal organoids cultured in 2 mg ml⁻¹ IEM hydrogel, as proliferation and differentiation may occur simultaneously at the early phase of intestinal organoid development, we speculate that differentiated cells may not be significantly less in intestinal organoids in IEM hydrogel compared to those in Matrigel. Actually, a previous study using decellularized matrices for intestinal organoid culture also showed a simultaneous increase in both stem cell markers and differentiated cell markers in the organoids grown in ECM hydrogel²⁰, which is consistent with our current finding. Overall, GI tissue-specific stemness and differentiation signatures of the organoids in GI tissue-derived ECM hydrogels seem to be comparable to those of the organoids in Matrigel.

We further compared phenotypes and functions of organoids produced in optimized ECM hydrogels with those of organoids grown in Matrigel. Gastric organoids in SEM hydrogel (5 mg ml⁻¹) grew comparably to organoids in Matrigel in size and morphology (Supplementary Fig. 9a). Although the projected area of gastric organoids grown in SEM hydrogel was slightly smaller than that of organoids grown in Matrigel, the variation of size distribution and morphology was lower in SEM organoids compared to Matrigel organoids (Supplementary Fig. 9c, e). The coefficient of variation (CV) in size of gastric organoids was 66.7% in SEM hydrogel and 74.5% in Matrigel. Moreover, gastric organoids in both matrices underwent similar organotypic development and contained heterogeneous populations of KI67-positive proliferative cells and MUC5AC-, HK-, and CHGA-positive gastric cells (Fig. 3e). The areas stained positively for proliferation marker (SOX9) and major differentiation marker (MUC5AC) in the gastric organoids grown in SEM hydrogel were not statistically different from those in the organoids cultured in Matrigel (Supplementary Fig. 10a). These data indicate that the development of cellular compositions in the gastric organoids cultured in SEM hydrogel is generally similar to that in the Matrigel organoids.

The generation of functional parietal cells, one of the most important cells in the function of the stomach, is still challenging in the current culture system of gastric organoids. This limitation of the current gastric organoid culture system may be solved through recapitulation of gastric microenvironments. For example, a previous study demonstrated the co-culture of fundic organoids with stomach mesenchymal cells to maintain the function of parietal cells in gastric organoids³⁰. In our case, fundic organoids containing parietal cells could be developed in our gastric organoid populations because we used whole stomach tissues including fundus region for the generation of gastric organoids. The presence of parietal cells in gastric organoids was confirmed by the expression of parietal cell marker genes (Atp4a, Atp4b) and protein (HK) (Fig. 3c, e). We compared the gastric function of organoids grown in SEM hydrogel and Matrigel by evaluating acid secretion using acridine orange dye. The degree of acid accumulation in organoids was similar in both groups (Fig. 3f). We also compared the gene expression of three major parietal cell markers (Gif, Atp4a, and Atp4b) in gastric organoids cultured in SEM hydrogel with that of stomach and intestine tissues (Supplementary Fig. 11a). RNA sequencing analysis revealed that the expression of those parietal cell markers in SEM organoids was significantly higher than that in intestine tissue, while significantly lower than in stomach tissue. This result suggests that gastric organoids cultured in SEM hydrogel contain parietal cells, but lacks parietal cell differentiation in comparison to native stomach tissue. We examined the cellular ultrastructure of gastric organoids cultured in SEM hydrogel using serial-section scanning electron microscopy to clarify the presence of parietal cells in gastric organoids (Supplementary Fig. 11b). Gastric parietal cells are characterized by a large number of mitochondria and intracellular canaliculi associated with acid secretion³¹. We found that some cells in the gastric organoids grown in SEM hydrogel have such characteristic ultrastructure of parietal cells, which directly confirmed the existence of gastric parietal cells in SEM organoids. The function of parietal cells in gastric organoids cultured in SEM hydrogel was also validated by observing acid secretion from gastric organoids treated with histamine (Supplementary Fig. 11c). However, the portion of parietal cells is much less in SEM organoids than in gastric gland tissue, and thus additional improvements would be required to further enrich parietal cells critical for the functionality of gastric organoids.

Likewise, intestinal crypts seeded in IEM hydrogel at the optimized concentration (2 mg ml⁻¹) developed into intestinal organoids with similar shape and size as observed in Matrigel (Supplementary Fig. 9b). The variation of size distribution was lower in IEM organoids than in Matrigel organoids (CV: 35.1% in IEM hydrogel and 47.7% in Matrigel) (Supplementary Fig. 9d). The variation in morphology of intestinal organoids looks less in IEM hydrogel compared to Matrigel (Supplementary Fig. 9f). Additionally, intestinal organoids grown in IEM hydrogel retained intestinal epithelial cells, including absorptive cells and three types of secretory cells, in a similar manner as intestinal organoids grown in Matrigel (Fig. 3k). The areas stained positively for major differentiation marker proteins (MUC2, LYZ) in the intestinal organoids grown in IEM hydrogel were not significantly different from those in the organoids cultured in Matrigel (Supplementary Fig. 10b). We also compared the cystic fibrosis conductance regulator (CFTR) function of intestinal organoids cultured in IEM hydrogel and Matrigel using a forskolin-induced swelling assay (Fig. 3l, m). Intestinal organoids from both groups rapidly enlarged at similar rates upon forskolin treatment, suggesting that intestinal organoids produced in IEM hydrogel also possess the ability to regulate luminal fluid secretion. Taken together, our results confirm the utility of SEM and IEM hydrogels as GI organoid culture platforms that enable the production and differentiation of GI organoids with structural and functional features comparable to organoids grown in Matrigel.

As mucus secretion is one of the key functions in GI tissues, generation of mucus layer is required for studying GI physiology in disease models and interactions between host and gut microbiota³². MUC5AC and MUC2 immunostaining confirmed mucus secretion into the lumen of GI organoids cultured in SEM and IEM hydrogels (Fig. 3e, k), but the mucus layer covering the organoids was not observed due to an insufficient number of goblet cells. Previous studies reported several methods to generate a mucus layer in GI organoids by promoting differentiation of goblet cells through depriving culture medium of Wnt3A and R-spondin1 proteins or using culture devices such as microfluidic chip^33,34. Thus, a combination of GI tissue-derived ECM hydrogels with additional strategies of adjusting signaling factors or culture conditions would be required to produce functional GI organoids covered by a mucus layer.

To check the differential effect of subepithelial ECM and mucus ECM on organoid development, we prepared ECM hydrogels derived from decellularized gastric mucosa and intestinal mucosa, and compared mechanical properties and organoid formation capacity of decellularized GI mucosa-derived ECM hydrogels with those of ECM hydrogels prepared from decellularized whole stomach and intestine containing subepithelial ECM. When prepared at the same concentrations (5 mg ml⁻¹ SEM and 2 mg ml⁻¹ IEM), the elastic modulus of SEM and IEM derived from decellularized mucosal layers was significantly lower than that of SEM and IEM from decellularized whole tissues (Supplementary Fig. 12a–d). Due to poor physical and mechanical properties, mucosa-derived SEM and IEM did not support GI organoid culture. As they could not sustain the developing GI organoids, the organoids often grew attached to the bottom of the culture plate and underwent morphological deformation (Supplementary Fig. 12e, g). In particular, intestinal organoids did not retain their spherical shapes in the mucosa-derived IEM hydrogel. Consequently, the GI organoid formation efficiency was significantly lower in the mucosa-derived SEM and IEM hydrogels than in the whole tissue-derived SEM and IEM hydrogels (Supplementary Fig. 12f, h). These data demonstrate that ECM hydrogels prepared with whole GI tissues can serve as more effective culture matrices likely due to more abundant subepithelial ECM components than ECM hydrogels derived from other parts of GI tissues without subepithelial ECM.

Transcriptome profiles of GI organoids generated in GI tissue-derived ECM hydrogels

To characterize the GI organoids produced in ECM hydrogels at a transcriptomic level, we performed an RNA-sequencing analysis of GI organoids grown in GI tissue-derived ECM hydrogels in comparison to GI organoids grown in conventional Matrigel and to native GI tissues. Many differentially expressed genes (DEGs) were identified between the organoids grown in SEM and IEM hydrogels and those grown in Matrigel (Fig. 4a). Using a false-discovery rate (FDR) cutoff of <0.1 and a fold-change > 2, we identified 590 DEGs (312 upregulated genes and 278 downregulated genes) and 270 DEGs (213 upregulated genes and 57 downregulated genes) in SEM hydrogel-cultured gastric organoids and IEM hydrogel-cultured intestinal organoids, respectively. Gene ontology (GO) terms related to ECM and cell proliferation were significantly upregulated in both gastric and intestinal organoids grown in each tissue-specific hydrogel (SEM or IEM) (Fig. 4b). In comparison to Matrigel organoids, genes related to positive regulation of the developmental process, hormone activity, and calcium ion binding were also upregulated in SEM organoids, and genes related to response to wounding and cytokine activity were upregulated in IEM organoids. When the selection criterion was strengthened to genes that were upregulated at least fourfold in GI tissue-derived hydrogel groups compared to Matrigel group, most of the selected genes were involved in the extracellular region, the ECM, or the regulation of cell proliferation (Fig. 4c, d and Supplementary Fig. 13). For the GO categories related to the extracellular environment (e.g., extracellular structure organization and ECM binding) and cell–matrix interaction (e.g., cell–matrix adhesion), we compared the gene expression of GI organoids cultured in SEM or IEM with that of native GI tissues (Fig. 4e). Although many genes were downregulated in organoids compared to native tissues, the number and level of upregulated genes were much higher in organoids cultured in SEM or IEM hydrogels than in those grown in Matrigel. As the expression of ECM-related genes plays an important role in ECM remodeling and organization, which are critical for cell differentiation and organ regeneration^35,36, our results suggest that GI tissue-specific hydrogels provide added benefit to GI organoid development compared to Matrigel. Based on the RNA sequencing data, we compared the expression of enteroendocrine cell subtype markers in intestinal organoids cultured in IEM hydrogel and Matrigel (Supplementary Fig. 14). Enteroendocrine cells regulate intestinal activities by producing hormones, and their subtypes are classified according to their hormone products: L cells (Gcg, Pyy), I cells (Cck), K cells (Gip), N cells (Nts), S cells (Sct), EC cells (Tph1), X cells (Ghrl), G cells (Gast), and D cells (Sst)^37,38. The expression levels of enteroendocrine cell subtype markers were overall similar between IEM organoids and Matrigel organoids, except for Gip and Gast with higher expression in IEM organoids and Ghrl with higher expression in Matrigel organoids. These results may verify that IEM hydrogel could generate intestinal organoids containing enteroendocrine cells responsible for hormone production, which would be suitable for preclinical drug screening and translational applications.

Given that GI tissue-derived ECM hydrogels are collagen-rich matrices and collagen culture condition is known to activate Yes-associated protein (YAP) signaling³⁹, we examined YAP signaling in the organoids cultured in IEM hydrogel and Matrigel. Immunocytochemical staining to show YAP expression and distribution in the intestinal organoids and quantification of the nuclear/cytoplasmic ratio of YAP signals revealed that YAP nuclear/cytoplasmic ratio was higher in the intestinal organoids cultured in IEM hydrogel than in the organoids cultured in Matrigel (Supplementary Fig. 15a, b), indicating YAP dephosphorylation and activation under IEM condition. RNA sequencing analysis showed that the significantly higher expression of several YAP target genes, such as Ctgf, Ankrd1, Ereg, Cyr61, Anxa1, and Ly6a, was observed in the IEM organoids than in the Matrigel organoids (Supplementary Fig. 15c, d). Organoid culture in 3D collagen type I hydrogel was previously reported to induce YAP/TAZ activation in intestinal organoids and conversion of cell fate to fetal-like states with regenerative capacity^39,40. Yui et al. demonstrated that Wnt/collagen-based culture condition allows intestinal organoids to have genetic profiles similar to those observed during tissue regeneration. Therefore, our data from YAP analysis may suggest that IEM hydrogel comprising abundant collagens is more effective for YAP signaling activation than Matrigel and confers regenerative epithelium-like properties to intestinal organoids.

Importantly, GI organoids cultured in SEM and IEM hydrogels exhibited expression of several important genes related to GI epithelial homeostasis that more closely resembled the expression profiles of the native tissues than did organoids are grown in Matrigel (Fig. 4f, g). The expression of specific genes encoding core matrisome proteins (Pxdn and Nid1) was similar or slightly lower in gastric organoids grown in SEM hydrogel compared to the native stomach (Fig. 4f). In SEM organoids, the stomach epithelial stem cell marker Msi1, the acid-secreting parietal cell marker Dbn1, and the enteroendocrine cell marker Chgb were expressed at levels similar to the native stomach tissues. Neuregulin-1 (Nrg1) is known to promote stem cell proliferation and epithelial regeneration⁴¹ and was expressed in SEM organoids at levels as high as those in native stomach tissues. Both SEM organoids and native stomach tissues expressed Nrg1 at significantly higher levels than Matrigel organoids. The expression of genes involved in hormone activity, such as the gastric hormone gene Pyy and the ghrelin signaling pathway-related gene Cpt1c, was also similar in SEM organoids and native stomach tissues. In intestinal organoids grown in IEM hydrogel, we found that the expression levels of core matrisome protein-encoding genes (Col4a2, Nid1, and Lama3), cytoskeleton-related genes (Flna, Gsn, and Tuba1a), and an intestinal epithelial gene involved in thiamin uptake (Tm4sf4) were comparable to those in native intestinal tissues (Fig. 4g). Compared with Matrigel organoids, IEM organoids showed significantly increased expression of genes related to wound healing, inflammation, and immune response (Procr, Mcpt2, Icam1, Cxcl10, Cxcl16, and Timp3), which were expressed at levels similar to those of native intestinal tissues. The expression of these genes at levels similar to the tissue may help to maintain intestinal barrier homeostasis under physiological conditions^42,43. Therefore, organoid culture using IEM hydrogel may preserve the intrinsic ability of intestinal organoids to cope with injury and to maintain tissue homeostasis. Overall, our results demonstrate that GI tissue-derived ECM hydrogels contribute to producing more mature and functional GI organoids in comparison to conventional Matrigel and enable the generation of GI organoids which recapitulate the cellular compositions and functional properties of adult GI tissues.

Tissue-specific and age-related effects of GI tissue-derived ECM hydrogel on GI organoid development

To test the hypothesis that stomach- or intestine-specific ECM cues present in SEM or IEM hydrogels provide favorable in vivo-like microenvironments for GI organoid development, we explored the tissue-specific effects of ECM hydrogels in organoid culture. Decellularized tissue-derived ECM hydrogels were prepared from six tissue types (stomach, intestine, skin, lymph, heart, and muscle) and were tested as culture matrices for GI organoids (Fig. 5a–f and Supplementary Fig. 16). Live/Dead staining indicated that the viability of GI organoids was relatively lower in non-GI tissue ECM hydrogels (i.e., skin, lymph, heart, and muscle) than in GI ECM hydrogels (SEM and IEM) (Supplementary Fig. 16). Most cells were highly viable in GI organoids cultured in SEM and IEM hydrogels. In contrast, a larger number of dead cells was observed in GI organoids cultured in ECM hydrogels derived from decellularized skin, lymph, heart, and muscle tissues. This result clearly indicates that GI tissue-derived ECM hydrogels can provide the most favorable microenvironments for the viability and growth of GI organoids. Interestingly, gastric and intestinal organoids developed well in IEM and SEM hydrogels, respectively. The size and morphology of organoids and organoid formation efficiency were not substantially different even when they were cultured in their counterpart ECM hydrogels (Fig. 5a, b, d, e, and Supplementary Fig. 16). Combined with our observation that many important matrisome proteins were commonly present in both SEM and IEM hydrogels (Fig. 2c–e, h and Supplementary Fig. 4c), these results suggest that SEM and IEM provide quite equivalent ECM environments for GI organoids. In contrast, other types of tissue-derived ECM hydrogels (skin, lymph, heart, and muscle) did not support the development of GI organoids, and the structure of the formed organoids was immature (Supplementary Fig. 16). For example, GI organoids cultured in decellularized skin-derived ECM (SkEM) hydrogel exhibited irregular and disorganized morphology, and the generated organoids were much smaller in size and less proliferative than those grown in GI ECM hydrogels (Fig. 5a, d). Additionally, the organoid formation efficiency in SkEM hydrogels was lower than in SEM or IEM hydrogels (Fig. 5b, e), and the expression of a stemness marker was significantly lower in SkEM hydrogel organoids (Fig. 5c, f). The relative proteomic quantitation between SkEM and GI-tissue matrisomes (Fig. 5g, h) clearly demonstrated the tissue-specific effects of GI ECM hydrogels on organoid development. The quantity of the 56–59 core matrisome proteins that are critical for GI development and function was much greater in SEM and IEM hydrogels than in SkEM hydrogels. For example, the level of collagen type VI (COL6A3, COL6A5, and COL6A6) was higher in GI tissue-derived ECM than in SkEM. Collagen type VI is a major component of the mechanical microenvironment of GI tissue that regulates epithelial cell-fibronectin assembly^44,45. Although some glycoproteins, such as fibronectin (FN1) and laminin (LAMA5, LAMB1, LAMB2, and LAMC1), were detected in SkEM, the amounts were lower than those in SEM and IEM. These glycoproteins are known to be essential for GI epithelial cell proliferation and differentiation^36,46,47,48. Collectively, these results validate the importance of the tissue origin of ECM hydrogels in organoid culture.

Next, we investigated the age-related effects of tissue ECM on GI organoid culture. ECM remodeling constantly occurs throughout the lifetime of an organism, and changes in ECM composition and properties during aging affect cell proliferation and differentiation⁴⁹. Thus, we compared the performance of SEM and IEM hydrogels derived from pigs of different ages (2-month-old piglet weighing approximately 20 kg versus 6-month-old adult pig weighing approximately 100–120 kg) for GI organoid culture. When we checked organoid formation in adult pig-derived ECM hydrogels and piglet-derived ECM hydrogels, there was no significant difference in the organoid formation efficiency between each ECM hydrogel for both gastric and intestinal organoids (Fig. 5i, k). Interestingly, the expression of stem cell markers (Lgr5, Axin2, and Olfm4) was higher in GI organoids grown in piglet-derived ECM hydrogels than in organoids grown in adult pig-derived ECM hydrogels (Fig. 5j, l). To determine which components in piglet and adult pig ECM affected the differential gene expression profiles of GI organoids, we performed a proteomic analysis of the matrisome in each condition (Fig. 5m, n). When compared with adult pig SEM and IEM, the quantities of 48 components [e.g., fibrillin2 (FBN2)] and 35 components [e.g., fibronectin 1 (FN1) and tenascin C (TNC)] were greater in piglet SEM and IEM, respectively (Fig. 5m, n). Fibrillin2 and tenascin C increase epithelial proliferation⁵⁰, and fibronectin is important for stem cell survival³⁶ and tissue regeneration⁵¹. Therefore, we inferred that these proteins, which are abundant in the piglet matrisome, support the growth potential of GI organoids. These findings provide important clues regarding the factors that should be considered to produce high-quality organoids and to design ECM-based hydrogels for organoid culture.

Additionally, we verified whether highly upregulated ECM proteins in piglet IEM hydrogel (FN and TNC) are really responsible for the better ability of piglet IEM to support intestinal organoid culture. We investigated whether the addition of FN and TNC into the adult-pig IEM hydrogel improves the quality of intestinal organoids. Interestingly, the addition of FN alone into adult-pig IEM hydrogel enhanced the expression of stem cell markers (Lgr5 and Axin2) in intestinal organoids, whereas the addition of TNC alone or both FN and TNC did not increase stem cell marker expression (Supplementary Fig. 17). These results demonstrate that FN may be a more crucial factor involved in the improved ability of piglet ECM for supporting organoid culture. Here, we tested only two proteins of piglet ECM and limited concentrations (25 and 50 μg ml⁻¹), but future studies to systemically identify more potent ECM molecules and optimal concentrations would be able to facilitate the development of chemically defined hydrogels for GI organoid culture.

The practicality of GI tissue ECM hydrogels for organoid culture and transplantation

So far, we have shown that GI tissue-derived ECM hydrogels have highly desirable features. Next, we sought to determine the practical applicability of these hydrogels. First, we evaluated the long-term expansion and passage of GI organoids in SEM and IEM hydrogels compared to those in Matrigel. The subculture of GI organoids in SEM and IEM hydrogels enabled stable growth and expansion of organoids persistent up to more than passages 6–8 (Fig. 6a, c). Overall, we did not observe any difference in the turnover rate and splitting ratio of passage between the organoids grown in ECM hydrogels and Matrigel. qPCR analysis of stemness markers (Lgr5, Axin2, and Olfm4) indicated that GI organoids maintained self-renewal ability after long-term expansion (days 33–45) in GI tissue-derived ECM hydrogels at comparable levels to organoids subcultured in Matrigel (Fig. 6b, d). The colon organoids-derived mouse large intestine also grew well in IEM hydrogel as similar to Matrigel. We assume that IEM could be widely used for culturing other types of GI tissue organoids with similarity to small intestinal organoids (e.g., colon organoids) (Supplementary Fig. 18a). SEM and IEM hydrogels not only supported the growth of primary tissue-derived GI organoids but also supported the growth of human pluripotent stem cell (hPSC)-derived GI organoids or GI tumoroids (Supplementary Fig. 18b–g). In Live/Dead staining to check cell viability of hPSC-derived intestinal organoids cultured in IEM hydrogel and Matrigel, the organoids grown in both matrices showed no significant difference in Live⁺ and Dead⁺ areas with more than 80% viability (Supplementary Fig. 18c), indicating that the IEM hydrogel did not provoke significant impairment in the viability of human intestinal organoids. Immunocytochemical staining of human-induced pluripotent stem cell (hiPSC)-derived intestinal organoids at days 29 and 51 indicated development of human organoids expressing intestinal stem cell marker SOX9 and differentiation markers MUC2 and CHGA in IEM hydrogel (Supplementary Fig. 18d, e). These results prove their versatile utility for culturing diverse types of organoids. In the future, the applicability of ECM hydrogels needs to be fully validated for the long-term culture and development of human GI organoids.

The long-term storage potential of reconstituted ECM is important for the convenience and commercialization of hydrogel matrices. We, therefore, checked the long-term stability and bioactivity of ECM hydrogels after refrigeration or freezing. We tested SEM and IEM pre-gel solutions stored in a deep-freezer (−80 °C for 6 months) or a refrigerator (4 °C for 1 month) for GI organoid culture (Fig. 6e, h). The organoid formation efficiency in SEM and IEM hydrogels that had been frozen for 1–6 months or refrigerated for 1–4 weeks was not different from that of freshly prepared hydrogels (Fig. 6f, i). The morphology and size of gastric and intestinal organoids grown in stored SEM and IEM hydrogels were similar to those grown in fresh hydrogels (Fig. 6g, j). Finally, gene expression levels of self-renewal and differentiation markers in GI organoids were not significantly different in long-term stored ECM hydrogels and fresh ECM hydrogels, which were similar to or higher than those in Matrigel organoids (Supplementary Fig. 19).

More interestingly, ECM hydrogels could be cryopreserved together with organoids, enabling ready-to-use applications. When organoids frozen in IEM hydrogel or Matrigel were thawed, the organoids frozen in IEM hydrogel were highly viable and grew normally, but those frozen in Matrigel exhibited significantly lower viability (Supplementary Fig. 20a, b). We then quantified the live- or dead-staining areas of organoids and confirmed that GI tissue-derived ECM hydrogel protected organoids from cellular damage caused by cryopreservation better than Matrigel (Supplementary Fig. 20c). After thawing, IEM organoids preserved their original morphology before cryopreservation, whereas Matrigel organoids exhibited an unhealthy morphology (Supplementary Fig. 20d). Likewise, the expression level of the apoptotic marker Caspase-3 in intestinal organoids after thawing significantly increased in only Matrigel organoids (Supplementary Fig. 20e). Immunocytochemical staining 2 days after thawing intestinal organoid-laden IEM hydrogel or Matrigel indicated that the percentage of active (cleaved) caspase-3⁺ area in the whole organoid was much higher (4.8-fold) in Matrigel organoids (13.67%) than in IEM organoids (2.85%) (Supplementary Fig. 20f, g). These data suggest again that IEM hydrogel is more effective than Matrigel to reduce apoptotic signaling resulting from caspase-3 activation during freeze–thaw cycles. Previous studies demonstrated that supplementation of ECM proteins such as collagen peptides⁵² and hyaluronan^53,54 to the freezing medium or entrapment of cells with collagen scaffolds^55,56 could reduce damage and stress during the cryopreservation process. Thus, the protective effects of GI tissue-derived ECM hydrogels during cryopreservation of the organoids observed in our study may be owing to the presence of various matrisome proteins abundant in GI tissue ECM hydrogels that Matrigel does not contain (e.g., collagen subtypes, and hyaluronan). In addition to excellent storage potential, ECM hydrogels may be compatible with a wide variety of culture systems. For example, we successfully integrated GI tissue-derived ECM hydrogels with microfluidic devices stacked on a single rocker for high-throughput dynamic culture⁵⁷ (Supplementary Fig. 21). We anticipate that ECM-integrated microfluidic platforms will allow for mass production of GI organoids for drug screening, toxicological evaluation, and disease modeling.

Finally, we tested the feasibility of GI tissue-derived ECM hydrogels for in vivo applications. Matrigel and cell therapeutics produced using Matrigel cannot be used in human clinical applications due to safety issues arising from the tumor-derived origin of Matrigel^11,58. In contrast, ECM hydrogels derived from decellularized tissues with removal of cells containing antigenic genetic motifs can be used as safe scaffolds for clinical cell therapy with a low risk of immunological rejection and inflammatory responses^59,60. To test whether ECM hydrogels aid efficient engraftment of transplanted GI organoids, we injected gastric or intestinal organoids encapsulated in their respective tissue-derived ECM hydrogels into the stomach or small intestine with acute epithelial injury using acetic acid. We first verified that the treatment of acetic acid causes epithelial damage and inflammatory reactions in the stomach and intestine tissues. H&E staining clearly showed the damaged regions in the gastric tissue treated with acetic acid and the collapse of the organization and polarization of gastric glands, which are indicative of gastric ulcer (Supplementary Fig. 22a). Likewise, the structures of villi and crypt were severely impaired in the intestinal tissue treated with acetic acid (Supplementary Fig. 22b). Tissue morphologies of mouse stomach and intestine injury models induced by acetic acid treatment were completely distinct from those of intact stomach and intestine tissues. In immunofluorescence staining for immune cell marker, F4/80⁺ macrophages were rarely observed in normal gastric and intestinal tissues, while F4/80⁺ macrophage infiltration was clearly observed in the stomach and intestine tissues of the injured models (Supplementary Fig. 22c, d). In particular, massive infiltration of F4/80⁺ macrophages was detected in the injured intestine tissue (Supplementary Fig. 22d). Next, we transplanted only SEM and IEM hydrogels to confirm negligible inflammatory responses following administration of SEM and IEM hydrogels. When SEM hydrogel labeled with a fluorescent dye (TAMRA-SE) was injected into the submucosa region of mouse stomach tissue, the hydrogel decomposed rapidly, and the fluorescence signals of the SEM hydrogel decreased after administration (Supplementary Fig. 23a). Infiltrated macrophages were not observed in the injection sites 1 and 4 days after SEM hydrogel injection. In the case of IEM hydrogel, the fluorescence signals of the IEM hydrogel distributed widely over the villi structures 4 h after injection into the intestinal lumen (Supplementary Fig. 23b). Similar to SEM hydrogel, macrophage infiltration was barely observed in the injected sites 1 and 4 days (Supplementary Fig. 23b) Taken together, these data demonstrate that SEM and IEM hydrogels can serve as biocompatible organoid carriers for transplantation.

For organoid instillation, GI tissue-derived hydrogels were diluted in culture medium (1:20) as described previously^61,62. We utilized fluorescently labeled GI ECM hydrogels and organoids to show specific replacement of damaged tissues by transplanted GI organoids. To trace organoid growth in vivo, we used enhanced green fluorescent protein (EGFP)⁺ organoids derived from EGFP transgenic mice (Fig. 6k and Supplementary Fig. 24a, b). Injection of gastric organoids using TAMRA-labeled SEM hydrogel allowed efficient administration of organoids into the submucosa region of the stomach with ulcers (day 0, Fig. 6l). We found that 4-, 7-, and 28-days post-transplantation EGFP⁺ gastric organoids were successfully engrafted into host tissues and EGFP signals of organoids were specifically detected throughout the injured sites (Fig. 6m and Supplementary Fig. 24c). Likewise, injected EGFP⁺ intestinal organoids with TAMRA-labeled IEM hydrogel were also engrafted into the injured intestinal tissues 4 and 7 days after transplantation (Fig. 6m and Supplementary Fig. 24d). Since diluted ECM hydrogels were applied for organoid transplantation, TAMRA fluorescent signals of SEM and IEM disappeared probably due to the rapid degradation in vivo. These data suggest that SEM and IEM hydrogels could mediate efficient engraftment of GI organoids into the injured regions for epithelium reconstruction and support the use of tissue ECM hydrogels as an effective organoid carrier for GI tissue regeneration. A long-term transplantation study would be required to check-in vivo behaviors and regenerative efficacy of transplanted GI organoids in injured GI tissues.