Bovine and human endometrium-derived hydrogels support organoid culture from healthy and cancerous tissues.

Heifers (Bos taurus, Angus, age 12–18 mo) were housed in a fenced open paddock with unfettered access to drinking water, where they would graze on Australian native grasses supplemented with lucerne hay. The density in the paddock is generally one animal per 10 acres. Grossly and histopathologically normal female reproductive tracts ( n = 12, n = 4 per group) were collected from an abattoir and transported immediately to the laboratory in ice-cold Dulbecco’s modified Eagle’s medium (Sigma, USA). Upon arrival, the tissues were washed briefly with phosphate-buffered saline containing 1% penicillin/streptomycin and amphotericin to remove excess blood and reduce initial microbial load. The uterus was carefully removed, and the rest of tissues (broad ligament, fallopian tubes, and ovaries) were discarded. Next, the endometrium was surgically separated from the myometrium. The endometrium was then cut into 2-cm pieces and placed in Milli-Q water (containing 1% penicillin/streptomycin and amphotericin) under constant agitation on a magnetic stirrer for overnight at room temperature. Fresh Milli-Q water was replaced every 3- to 4-h interval. Each batch of endometrial tissues for decellularization contained representative samples from four different animals. After 24-h wash with Milli-Q water, each batch of endometrial tissues ( n = 4 pooled) was treated with the respective detergents for decellularization: protocol 1 (P1)—4% SDS (Sigma), protocol 2 (P2)—1% SDS, and protocol 3 (P3)—4% SDC solution (Sigma). Each of the protocols were performed under constant agitation (stir plate), with fresh solution changed every 3 to 4 h at room temperature until all the tissue pieces became translucent (usually after 4 to 5 d). After detergent treatment, tissues were extensively rinsed with water overnight, followed by 2,000 kU DNase (Sigma) in 1 M NaCl for 3 h at room temperature. Decellularized tissues were then washed in Milli-Q water for additional 2 d, with multiple water replacement to eliminate any cytotoxic residues of SDS/SDC, prior to lyophilization and milling into powders. Normal human premenopausal endometrium and endometrial cancer tissue samples were collected from patients undergoing surgery by following guidelines approved by the Institutional Human Research Ethics Committee at the University of Newcastle and the Hunter New England Human Research Ethics Committee ( SI Appendix , Table S1 ). Written consent was obtained from all donors. The human endometrium was processed similar to the bovine tissue and was decellularized using protocol 3 (4% SDC). Full experimental details can be found in SI Appendix , Materials and Methods .

Bovine endometrium bears many histological and biochemical similarities to the human endometrium ( 18 , 19 ). Additionally, whole bovine uterus is readily available from abattoirs. Therefore, we decided first to optimize our decellularization protocol using bovine uterine tissue ( Fig. 1 ). Our protocol is divided in five crucial steps: (1) surgical isolation of the endometrium, (2) removal of cellular and nuclear material (decellularization), (3) lyophilization or freeze drying, (4) low-temperature milling to make powder, and (5) digestion, pH neutralization, and gelation ( Fig. 1 A ). For decellularization, we used three different detergent treatments, namely 4% sodium dodecyl sulfate (SDS) (hereafter referred as P1), 1% SDS (P2), and 4% sodium deoxycholate (SDC) (P3), to find the best condition for removing intracellular content with minimal impact on the tissue integrity and morphology ( Fig. 1 B ). Detergent treatments led to white translucency of tissues while maintaining their gross appearance and size ( Fig. 1 A ). The decellularized tissues obtained from the three protocols were then evaluated using histology and scanning electron microscopy (SEM) to check for the decellularization efficiency before progressing to the next step ( Fig. 1 B , a–i). Examination of hematoxylin and eosin (H&E)-stained native and decellularized endometrial tissue sections revealed the loss of nuclear material (hematoxylin) with no apparent loss in tissue morphology ( Fig. 1 B , a–f). Decellularization protocol and assessment for endometrial tissue. ( A ) Overview of the key steps involved in endometrial hydrogel preparation from the bovine endometrium. ( B, a–f ) Histological images of native and decellularized endometrial tissue sections ( n = 4) from the three protocols to check for the decellularization efficiency. Yellow arrowheads denote the presence of endometrial glands in the stromal compartment, which were also present in the decellularized tissue samples. ( B, g–i ) SEM highlights remarkable preservation of both luminal (le) and glandular (eg) epithelial compartments in the decellularized tissues. ( B, j–o ) SEM imaging of the lyophilized endometrium of P1–P3 ( n = 3 biological replicates per group) showed a three-dimensional network of long ECM protein fibers without any intervening cellular material. ( C, a ) DNA quantification in native and decellularized endometrium ( n = 6 biological replicates per group) was normalized to the initial dry weight of each sample; **** P < 0.0001, unpaired t test. ( C, b ) Gel electrophoresis of DNA extracted from the fresh and decellularized endometrium for P1–P3. ( C, c ) Dry weight (gram) in native and decellularized endometrium tissue ( n = 4 biological replicates per group); * P < 0.05, unpaired Student’s t test. ( C, d ) SDS quantification showing P1 and P2 sample ( n = 3 biological replicates per group) after decellularization; data represent mean ± SEM. ( D ) Immunostaining for fibronectin and hydroxyproline in the native and decellularized endometrium tissue ( n = 4 biological replicates per group). White or black dotted lines are marking endometrial glands. s, stroma. Endometrial glands are normally surrounded by the stromal compartment, and these glands were also easily identifiable in the decellularized tissue samples (yellow arrowheads; Fig. 1 B , d–f). SEM analysis confirmed the remarkable preservation of both luminal (le) and glandular (eg) epithelial compartments in the decellularized tissues ( Fig. 1 B , g–i). Next, we lyophilized the decellularized endometrium ( Fig. 1 A and B ). SEM imaging of lyophilized endometrium revealed the presence of a three-dimensional network of long ECM protein fibers without any intervening cellular material ( Fig. 1 B , j–o). Expectedly, the amount of DNA in the fresh endometrium tissue was significantly reduced after decellularization ( Fig. 1 C a), which was further confirmed with gel electrophoresis methodology, revealing an extremely low content of residual DNA in the decellularized endometrium ( Fig. 1 C , b). The dry weight of the decellularized endometrium was significantly reduced compared with the fresh native tissue ( Fig. 1 C , c). Residual detergents used for decellularizing tissue or organs could have cytotoxic effects and compromise the functionality of biological scaffolds and hydrogels ( 20 ). We determined that residual SDS content in the P1 and P2 groups was very low, indicating a successful elimination of cytotoxic SDS in our samples following ( Fig. 1 C , d). To monitor whether our decellularized endometrium preserves essential ECM proteins, such as fibronectin and collagen/hydroxyproline, we performed immunostaining and confirmed the presence of fibronectin and hydroxyproline in the decellularized endometrium was comparable to the native tissue ( Fig. 1 D ). Collectively, these data confirmed that our optimized decellularization protocols preserve many key features of the native tissue. The different ECM components provide structural support and transmit functional signals to resident cells ( 11 ). Therefore, it is important to maintain the native ECM proteins after decellularization in order to replicate the native-tissue ECM microenvironment and signaling. Here, we explored the characteristics and distribution of specific ECM in decellularized bovine endometrium. Raman spectroscopy is an infrared spectroscopy technique that is commonly used to determine the vibrational modes of molecules. This allows Raman spectroscopy to obtain structural fingerprints of molecules that can be used for molecular identification ( 21 ). In combination with optical microscopy, confocal Raman microspectroscopy offers a submicrometer lateral resolution. Here, we exploited confocal Raman microspectroscopy to visualize the spatial distribution of different components in extensive detail and compared the spectral profiles between native and decellularized endometrium from the three different decellularization procedures (P1–P3; Fig. 2 A and B ). Confocal Raman images showed remarkable similarity in the spatial distribution of ECM proteins in native and decellularized tissues ( Fig. 2 A ). The Raman spectra across all samples detected prominent peaks at 1,004 cm −1 (Phe [phenylalanine]), 1,061 cm −1 (GAG [glycosaminoglycan]), 1,135 cm −1 (C–C Asy [asymmetric carbon–carbon stretching mode]), 1,298 cm −1 (amide III), and 1,450 cm −1 (collagen) ( Fig. 2 B ). Among three different decellularization protocols, we observed a higher degree of peak intensities for GAG and amide III, which represents the collagen secondary structures ( 22 ), in P3 compared with other groups ( Fig. 2 B ). Biochemical characterization of endometrium-derived ECM. ( A and B ) Confocal Raman microspectroscopic analysis of the native and decellularized endometrial tissue sections ( n = 4 replicates per group) is represented as ( A ) high-resolution confocal Raman maps and ( B ) corresponding Raman spectra of the tissue sections. Raman signatures of the corresponding components of phenylalanine (Phe), glycosaminoglycan (GAG), asymmetric carbon–carbon stretching mode (C–C Asy), amide III, and collagen are labeled, while spectral intensities across samples have been normalized to the C–C Asy. ( C – F ) Quantification of soluble collagen, insoluble collagen, hydroxyproline, and sulfated GAG (sGAG) in the native and decellularized endometrial tissue ( n = 3 replicates per group). Soluble and insoluble collagen, hydroxyproline, and sGAG contents in the native and decellularized endometrium were normalized to the initial dry weight of the sample; **** P < 0.00001; *** P < 0.0001; ** P < 0.001; * P < 0.05, unpaired Student’s t test. ( G ) Fourier transform infrared spectroscopy spectra highlighting transmittance peaks of collagen amide A (∼3,300 cm −1 ), amide B (∼3,100 cm −1 ), amide I (∼1,650 cm −1 ), amide II (∼1,550 cm −1 ), and amide III (∼1,200 cm −1 ) in decellularized tissues belonging to all three groups. Scale bars, 100 µm unless indicated otherwise. To validate our Raman microspectroscopy data, we analyzed selected ECM proteins, including collagens (both soluble and insoluble types), sulfated glycosaminoglycans (sGAG), and hydroxyproline, in native and decellularized endometrial tissues ( Fig. 2 C – F ). We observed significant enrichment of both soluble and insoluble forms of collagen in decellularized tissues compared with the native tissue ( Fig. 2 C and D ). Hydroxyproline is a major component of collagen and involved in collagen biosynthesis, stability, and strength ( 23 ). Hydroxyproline concentration was increased in decellularized tissues than native controls in the P3 group ( Fig. 2 E ). In P1 and P2 groups, hydroxyproline concentration was either unchanged or decreased relative to their respective controls ( Fig. 2 E ). The amount of sulfated glycosaminoglycans was relatively stable across the samples, except a slight decrease in decellularized tissue samples belonging to the P3 group ( Fig. 2 F ). To further characterize the molecular structure and detect the functional groups, such as amide bonds and sugars, present in the decellularized tissue samples, we utilized Fourier transform infrared spectroscopy. The infrared spectra clearly showed transmittance bands corresponding to amide A (∼3,300 cm −1 ), amide B (∼3,100 cm −1 ), amide I (∼1,650 cm −1 ), amide II (∼1,550 cm −1 ), and amide III (∼1,200 cm −1 ) stretching and bending modes, consistent with the presence of peptide backbone of proteins in decellularized tissues in all three samples ( Fig. 2 G ), suggesting that the chemical integrity of ECM proteins is not compromised by our decellularization protocols. Collectively, these data highlight the efficacy of our three different protocols in decellularizing the bovine endometrium. To prepare ECM hydrogels, lyophilized decellularized bovine endometrium was cryomilled into a fine powder to ensure uniform digestion during the gelation step ( Fig. 1 A ). We measured the diameter of particles in ECM powder by dynamic light scattering. The particle size distribution yields an average diameter of ∼420.2 ± 10.27 nm for P1 ECM, 568.3 ± 7.01 nm for P2 ECM, and 420.2 ± 3.45 nm for P3 ECM ( Fig. 3 A ). We next generated hydrogels from ECM powders representing three different protocols (P1–P3) at 10 mg/mL concentration ( Fig. 3 B ). All three hydrogels solidified at physiological pH and temperature ( Fig. 3 B ). SEM imaging of solidified ECM hydrogels revealed distinct interconnected fibrillary organization akin to collagen type 1 hydrogels ( Fig. 3 B ) ( 24 ). We assessed the gelation kinetics of three hydrogels by turbidimetric analysis. This technique measures the increase in turbidity, and thus absorbance, observed during the assembly of collagen fibrils ( 25 ). All three hydrogels underwent gelation after a lag phase ( Fig. 3 C ). Compared with P1 and P2 hydrogels, the P3 hydrogel became more turbid and reached 90% of gelation within 15 min ( Fig. 3 C ), suggesting that the method of decellularization influences gelation kinetics of hydrogels. Hydrogels are polymeric and hydrophilic materials that have three-dimensional structures with the ability to entrap a large amount of water ( 26 ). Next, we tested the stability of these hydrogels by examining their ability to absorb water and undergo degradation in the presence of a reducing agent, dithiothreitol ( Fig. 3 D and E ). The water uptake capacity of P3 hydrogels was lower than P1 and P2 hydrogels ( Fig. 3 D ). Consistently, the degradation resistance of P3 hydrogel was threefold higher than P1 and P2 hydrogels ( Fig. 3 E ). Mechanical properties of endometrium-derived ECM hydrogels. ( A ) Particle size analysis of the decellularized endometrial powder of P1–P3 ( n = 4 biological replicates per group) by dynamic light scattering. ( B ) Gross and SEM images of endometrium-derived ECM hydrogels from the three protocols at concentrations of 10 mg/mL. ( C ) Turbidimetric gelation kinetics of the decellularized endometrium ECM hydrogel ( n = 3 biological replicates per group) at a concentration 10 mg/mL generated using three different protocols (P1–P3). ( D and E ) Water uptake capacity ( D ) and degradation ( E ) analysis of the three hydrogels ( n = 3 biological replicates per group). ( F – J ) Rheological properties of endometrial ECM hydrogels. Storage modulus, loss modulus, complex viscosity, and oscillation stress of hydrogels and Matrigel ( n = 3 biological replicates per group); **** P < 0.00001; *** P < 0.0001; ** P < 0.001, one-way ANOVA for multiple group comparisons. ( K ) A representative gross image of the mouse uterine arm before and after decellularization. ( L , a–d ) A homemade simple device consisted of a tuberculin syringe, a cotton bud, and a pipette tip to develop tubular structures from hydrogels and soft materials. A gross image of agarose tube developed using this device. ( L , e–l ) Gross images of tubular structures developed from different materials using our device. Scale bars, 100 µm unless indicated otherwise. To investigate the mechanical properties of endometrial hydrogels, we performed rheological assessments using a parallel-plate rheometer. The storage modulus (G’; represents elastic behavior of material when deformed) and loss modulus (G’’; reflects viscous behavior of material when deformed) of hydrogels were measured with increasing temperature from pregelling (5 °C) to physiological temperature (37 °C) ( Fig. 3 F ). The P3 hydrogel had a higher storage and loss modulus than P1 and P2 hydrogels and Matrigel ( Fig. 3 F and G ), suggesting that P3 hydrogel has higher viscoelasticity and stronger mechanical strength relative to other gels. To determine which hydrogel has mechanical properties equivalent to the endometrial tissue, we compared storage modulus, complex viscosity (total resistance to flow when force is applied), and oscillation stress (reflects the strength of the hydrogel network) of our three endometrial hydrogels, Matrigel, and decellularized bovine endometrial tissues ( Fig. 3 H – J ). The assessments of these parameters revealed that the mechanical properties of P3 hydrogels are more closely related to the decellularized endometrial tissue ( Fig. 3 H – J ), suggesting that the P3 hydrogel might provide a more-natural and supportive environment for the growth of endometrial organoids. To develop a proof of concept for using endometrial hydrogels for regenerative treatments to repair tubular organs (e.g., uterus), we tested the potential of our hydrogels to develop tube-like structures to mimic the shape and size of a decellularized mouse uterus ( Fig. 3 K and L ). We used a simple setup of a tuberculin syringe and a sterile cotton bud to develop these tubal structures ( Fig. 3 L , a). We first used agarose to show that a tubal structure with a patent lumen can be easily developed using our device ( Fig. 3 L , b–d). We then developed similar structures using endometrial hydrogels and Matrigel ( Fig. 3 L , e–l). Consistent with their mechanical properties ( Fig. 3 F – J ), Matrigel and P1/P2 hydrogel were too soft to form a proper tube-like structure ( Fig. 3 L , e–g). Tubal structures made from these materials were solid, uneven, and looked rough from the outside ( Fig. 3 L , e–g). In comparison, the P3 hydrogel–derived tubes were smooth from the outside and grossly appeared similar to the decellularized mouse uterus ( Fig. 3 K and L , h). However, these tubes were unable to maintain a patent lumen ( Fig. 3 L , h). To further improve the strength and stability of P3 hydrogels, we used two noncytotoxic crosslinkers, genipin and N -(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), that are known to crosslink adjacent collagen microfibrils ( 27 ). The addition of genipin, EDC, and their combination led to a stable patent lumen formation in P3 hydrogel–derived tubal structures ( Fig. 3 L , i–l). Therefore, these tubes now appear architecturally comparable to the mouse uterus. A dark blue coloration typically appears in hydrogels after adding genipin ( 27 ). To decipher the influence of decellularization methodologies on the biochemical composition of decellularized tissues and subsequently on hydrogels, we performed an in-depth characterization of ECM proteins present in the decellularized bovine endometrium ( Fig. 4 A ). We used the 4plex iTRAQ labeling system containing four pooled samples from each protocol (P1–P3), allowing us to compare the three protocols in the same liquid chromatography–tandem mass spectrometry (LC-MS/MS) condition, thereby reducing experimental variation ( Fig. 4 A ) ( 28 ). Our LC-MS/MS analysis revealed that 439 proteins were differentially retained between P3 and P1/P2 groups ( Fig. 4 B and SI Appendix , Dataset S1 ). Next, we screened the total identified proteins with an existing ECM protein database (Matrisome) to facilitate identification of ECM and ECM-associated proteins ( 29 ). This step-by-step process allowed us to define the matrisome signature enriched in the decellularized endometrium matrix achieved from the three different methods (P1–P3) and to analyze differences in ECM protein levels between them. Matrisome analysis led to the identification of ECM proteins that were then sorted into matrisome subcategories within the core matrisome (ECM glycoproteins, collagens, and proteoglycans) and matrisome-affiliated proteins (ECM-affiliated proteins, ECM regulators, and secreted factors) ( Fig. 4 C ). Altogether, 54% ECM proteins from three treatment groups were detected as core matrisomal proteins, which contained 49% ECM glycoproteins, 46% collagens, and 5% proteoglycans in the decellularized endometrium ( Fig. 4 C ). The remaining 46% ECM proteins were classified as matrisome-affiliated proteins composed of 25% ECM-affiliated proteins, 44% ECM regulators, and 31% secreted factors ( Fig. 4 C ). Proteomic profiling of decellularized endometrium. ( A ) Schematic representation of the proteomics workflow. Proteins were extracted from the respective P1–P3 decellularized endometrium samples, digested and labeled with 4plex iTRAQ, and combined ( n = 4 pooled from each treatment group). To reduce sample complexity, pooled peptides were fractionated by high-pH fractionation prior to LC-MS/MS analysis. ( B ) Hierarchical clustering analysis of 1,141 proteins differentially expressed in decellularized endometrium ECM comparing between the three protocols P1–P3. Proteomic analysis detected 25 and 28 ECM proteins(shown in the box) that were upregulated in the P3 group than in the P1 and P2 groups, respectively. ( C ) Dot plot represents the ECM protein expression profile for the core matrisome (encompassing ECM glycoproteins, collagens, and proteoglycans) and matrisome-associated (ECM-affiliated proteins, ECM regulators, and secreted factors) achieved from the three different methods (P1–P3). The color of each circle represents protein expression level, and the circle size indicates the relative value of the readout measurement across all conditions. ( D – G ) Western blot validating the expression of laminin ( E ), fibronectin ( F ), and collagen I ( G ) in endometrium decellularized using three protocols ( n = 3 biological replicates per group). GAPDH expression used as a loading control across the samples. *** P < 0.0001; ** P < 0.001; * P < 0.05; one-way ANOVA for multiple group comparisons. Our comparative analysis of matrisome revealed differences in the composition of P1–P3 ( Fig. 4 B ), suggesting the method of decellularization has a major influence on the biochemical composition of decellularized endometrium. Compared with the P3 group, 25 and 28 ECM proteins were decreased in the P1 and P2 groups, respectively ( Fig. 4 C ). Many of these proteins (e.g., laminin subunit alpha 2 [LAMA2]) are the central regulators of the physiological functions of the endometrium ( 30 ). We validated the expression of three key ECM proteins (LAMA1 [laminin subunit alpha 1], fibronectin, and collagen 1) in P1–P3 samples using western blotting ( Fig. 4 D ). Interestingly, laminin was only detectable in the endometrium decellularized using 4% SDC (P3), but not with 1% (P2) or 4% (P1) SDS ( Fig. 4 D and E ). Laminins are a major component of Matrigel and are required for organoid development and growth ( 31 ). Fibronectin and collagen 1 protein expression were comparable in P2 and P3 but significantly decreased in the P1 group ( Fig. 4 D , F , and G ). Overall, these results suggested that an SDC-based decellularization protocol (P3) is more effective in preserving the matrisome of the native tissue than SDS-based protocols (P1 and P2). To confirm that ECM hydrogels support the growth of organoids, we developed organoids from both human and mouse cells in endometrial hydrogels (P1–P3) and Matrigel ( Fig. 5 A – D ). Similar to a previous report ( 6 ), mouse endometrial cells showed robust growth and developed spherical-shape organoids in Matrigel ( Fig. 5 A , a). We then plated an equal number of mouse endometrial cells in an identical culture medium in the ECM hydrogels ( Fig. 5 A , b–d). No organoid growth was observed in the P2 hydrogel ( Fig. 5 A , c). Organoid forming efficiency, organoid shape, and organoid size were comparable between P3 hydrogel and Matrigel ( Fig. 5 A , d and i). Interestingly, organoids formed in the P1 hydrogel were not spherical and appeared to have both luminal and glandular compartments ( Fig. 5 A , b and e–j and B ), which were akin to the mouse endometrial tissue ( SI Appendix , Fig. S1 ). To validate whether endometrial organoids developed in the P1 hydrogel have a distinct glandular compartment, we examined the expression of Foxa2, a known marker of uterine glands ( 32 ) ( Fig. 5 B ). Foxa2+ cells were interspersed with Foxa2− cells in the spherical-shaped mouse endometrial organoids grown in the Matrigel ( Fig. 5 B a-d). In contrast, Foxa2+ cells were restricted to the glandular projections coming out of the Foxa2− central body of the organoids developed in the P1 hydrogel ( Fig. 5 B , e–h). The number of Foxa2+ and Ki67+ (a marker of proliferating cells) cells was comparable in organoids grown in both Matrigel and P1 hydrogel ( Fig. 5 B , i and j). Endometrium ECM-derived hydrogels support human and mouse organoid cultures. ( A , a–d ) Bright-field images of mouse endometrial organoids in Matrigel and endometrial hydrogels (P1–P3). ( A , e–h ) Compared with round-shaped organoids in Matrigel ( A, e ), organoids in P1 hydrogel showed budding ( A, f ), tubular ( A, g ), and glandular features ( A, h ). ( A, i ) The organoid forming efficiency of mouse endometrial organoids in Matrigel and P1–P3 hydrogel; **** P < 0.00001, two-way ANOVA with multiple comparison test. ( A, j ) Percentages of organoid Nos. representing round, tubular, and glandular-shaped organoids in Matrigel versus P1; **** P < 0.00001, two-way ANOVA with multiple comparison test. ( B, a–h ) H&E and immunostaining for Foxa2, Ck8, and Ki67 of organoids cultured in Matrigel and P1 hydrogel. Percentages of Foxa2+ ( B, i ) and Ki67+ ( B, j ) positive cells in organoids grown in Matrigel versus P1 hydrogel. ( C , a ) Images of mouse endometrial organoid growth in P1 hydrogel supplemented with increased concentrations of laminin (L) (100P1:0L, 75P1:25L, 50P1:50L, 25P1:75L, and 0P1:100L). ( C , b and c ) Percentages of organoid formation efficiency and the number of round and branched-shaped organoids in P1 hydrogel containing different laminin concentrations; **** P < 0.00001; *** P < 0.0001, two-way ANOVA with multiple comparison test. ( D and E ) Bright-field images and organoid formation efficiency of human endometrial cancer cells, Ishikawa ( D, a–d and m ), normal endometrium ( D, e–h and n ), endometrial cancer ( D, i–l and o ), human colon ( E, a–e ), mouse colon ( E, f–j ), human lung ( E, k–o ), and mouse lung ( E, p–t ) organoids in P1–P3 hydrogel and Matrigel. **** P < 0.00001, *** P < 0.0001; ** P < 0.001, * P < 0.05, two-way ANOVA with multiple comparison test. ( F, a ) Gross images of normal human endometrium before and after decellularization. ( F, b and c ) Organoid formation efficiency in human endometrial P3 hydrogel versus Matrigel and human versus bovine endometrial P3 hydrogel. ( F, d ) Bright-field images of human endometrial organoids cultured in human endometrial P3 hydrogel and Matrigel. Scale bars, 100 µm unless indicated otherwise. Although organoids cultured in the P1 hydrogel closely mimic the architectural arrangement of endometrial epithelial cells of the native tissue, fewer organoids were developed in the P1 hydrogel than in Matrigel and P3 ( Fig. 5 A , i). Since our biochemical analysis revealed a lower laminin expression in the P1 hydrogel relative to P3 ( Fig. 4 D ), we tested whether external supplementation of laminin could improve the organoid forming efficiency of the P1 hydrogel ( Fig. 5 C , a–c). Therefore, we isolated laminin from the bovine endometrium and added it to the P1 hydrogel ( Fig. 5 C , a and SI Appendix , Fig. S2 ). Supplementation of laminin to the P1 hydrogel increased the number of organoids in a concentration-dependent manner ( Fig. 5 C , a and b). However, these organoids also progressively acquired a round morphology and become similar to organoids grown in Matrigel and P3 hydrogel ( Fig. 5 C c), suggesting laminin levels in ECM hydrogels are an important determinant of organoid shape and organoid forming efficiency. We then performed similar experiments using human cells to explore whether ECM hydrogels support human organoid cultures ( Fig. 5 D ). We cultured immortalized endometrial cancer cells (Ishikawa) and primary normal and cancerous patient-derived endometrial cells in ECM hydrogels and Matrigel ( Fig. 5 D ). No organoid growth was visible in the P1 and P2 hydrogels ( Fig. 5 D , a, b, e, f, i, and j). However, robust organoid development was present in both P3 and Matrigel ( Fig. 5 D , c, d, g, h, and k–o). Next, we independently tested the possibility of ECM hydrogels supporting the growth of organoids from other tubular organs, such as the human and mouse colon and lung ( Fig. 5 E ). Seeding of primary human and mouse colon and lung cells led to the robust development of organoids in the P3 hydrogel and Matrigel ( Fig. 5 E ). Compared with the human colon ( Fig. 5 E , a–e), mouse colon cells showed strong growth in all four culture conditions ( Fig. 5 E , f–j), suggesting species-specific differences in the growth pattern of organoids in endometrium-derived ECM hydrogels. Collectively, our findings showed that P3 hydrogel, one of three ECM hydrogels, is best for producing human and mouse organoids, and endometrium-derived hydrogels can also support the formation of organoids from other tubular organs. We next investigated whether ECM hydrogel derived from the normal human endometrium would also be applicable for culturing endometrial organoids ( Fig. 5 F ). We decellularized the human endometrium and developed hydrogels similar to the bovine endometrial-derived P3 hydrogel ( Figs. 1 and 5 F , a). Similar to the bovine P3 hydrogel, human P3 hydrogel supported the growth of normal human endometrial organoids, which was comparable to the growth observed in the Matrigel ( Fig. 5 F , b–d). These data support that both human and bovine P3 hydrogels can support the vigorous development of endometrial organoids. Studies in multiple organs have now firmly established that signals emanating from the ECM proteins drive the pathogenesis of neighboring epithelial cells ( 11 ). The matrisome of cancer tissues significantly differs from the matrisome of their normal counterparts ( 29 ). Given that Matrigel is extracted from a mouse tumor ( 17 ), we explored the molecular differences in organoids cultured in the normal (P3 hydrogel) and abnormal (Matrigel) ECM. Therefore, we compared the proteome of normal human endometrial organoids cultured in the bovine P3 hydrogel and Matrigel ( Fig. 6 A – E ). The proteome of human endometrial organoids cultured in the P3 hydrogel was more closely related to the proteome of the native patient endometrial tissue than the organoids grown in the Matrigel ( Fig. 6 A and B ). One thousand three hundred proteins were commonly identified between the tissue and organoids ( Fig. 6 A ). MS intensities of these proteins were normalized to the Z score ( Fig. 6 A and C ). We then ranked the proteins in the tissue, P3 hydrogel organoids, and Matrigel organoids based on the Z score ( Fig. 6 C and D ). This analysis revealed that the P3 hydrogel organoids are more proteomically similar to the native endometrial tissue than the Matrigel-grown organoids ( Fig. 6 C and D ). Ingenuity pathway analysis revealed that some of the major signaling pathways were either downregulated or missing in the organoids cultured in Matrigel relative to the P3 hydrogel organoids and the native tissue ( Fig. 6 E ). For example, the integrin-linked kinase pathway and RHOA signaling were detected in both the native tissue and P3 hydrogel organoids, but not in the Matrigel-grown organoids ( Fig. 6 E ). Both signaling pathways are the major regulators of endometrial functions and fertility, and their dysregulation leads to the pathogenesis of endometriosis and endometrial cancer ( 33 , 34 ). Our proteomic analysis of bovine and human P3 hydrogels confirmed comparable expression of the major ECM proteins ( Fig. 6 F ). In summary, these data support that endometrial-derived P3 hydrogel is superior to Matrigel in recapitulating the native endometrium tissue–like environment for more natural growth of endometrial organoids. Proteomic analysis of human organoids grown in endometrium-derived hydrogels and Matrigel. ( A ) Venn diagram highlighting the number of common and differentially expressed proteins in the native tissue versus organoids grown in the P3 hydrogel and Matrigel. ( B and C ) Heatmap and violin plot analysis of 1,300 proteins commonly identified between the three conditions. The MS intensities were normalized to Z score. ( D ) A line plot analysis displaying organoids cultured in the P3 hydrogel had a similar protein expression profile with the native human tissue from which they are derived than the organoid grown in the Matrigel. ( E ) Ingenuity pathway analysis highlighted the major signaling pathways were either downregulated or missing in the organoids cultured in Matrigel relative to the P3 hydrogel organoids and the native tissue. Z scores were used to predict activation (Z score ≥ 2; orange) or inhibition (Z score ≤ −2; blue) of each function. ( F ) Comparison of ECM proteins between human and bovine P3 hydrogels. Porous scaffolds are three-dimensional polymeric materials with interconnected pores that are widely used in the field of tissue engineering ( 35 ). We determined whether P3 hydrogel is biocompatible for fabricating porous scaffolds to support organoid growth and for future applications in regenerative therapies. To achieve this, we used the ice templating technique, where P3 hydrogel was progressively frozen so that biomaterial particles started concentrating around the growing ice crystals, and once the sample was fully frozen, ice crystals were removed by lyophilization, leaving a desired porous biomaterial ( Fig. 7 A ) ( 36 ). Using this technique, we can develop scaffolds of required shapes and sizes that can be stored at room temperature ( Fig. 7 B ). SEM and H&E staining confirmed a honeycomb-like porous appearance of these scaffolds ( Fig. 7 C – E ). Next, we seeded endometrial cancer cells (Ishikawa: a human endometrial epithelial cell line) onto P3 scaffolds and assessed their growth after 12 d of the culture period. Both grossly and histologically, the three-dimensional growth of endometrial cells was clearly appreciated throughout the P3 scaffold ( Fig. 7 F and G ). Immunolocalization of phospho-histone 3 (PH3), a nuclear mitotic marker ( 37 ), and cytokeratin 8 (CK8), a marker of epithelial cells ( 37 ), revealed the presence of PH3+ CK8+ proliferating epithelial cells in these scaffolds ( Fig. 7 J ), suggesting that endometrial cells are actively dividing to form these organoids. We also cultured primary human endometrial cells on these scaffolds to show that normal patient-derived cells are also able to grow on these scaffolds ( SI Appendix , Figs. S3 and S4 ). Overall, these data demonstrate the development of endometrium-derived ECM hydrogel-based scaffolds for growing endometrial cells for their potential applications in advancing high-throughput drug screening and tissue-engineering approaches for endometrial diseases. Generation of ECM hydrogel-derived porous scaffolds for culturing human endometrial cells. ( A ) Schematic workflow for fabricating porous scaffolds from bovine P3 hydrogel by using lyophilization procedure. ( B ) The gross view of the P3 scaffolds generated for 96-, 24-, and 12-well culture plates can be stored at room temperature. ( C – E ) SEM and H&E images confirmed the honeycomb-like porous appearance of scaffolds. ( F and G ) Gross images of scaffolds seeded with endometrial cells (Ishikawa). ( H and I ) Histology analysis showed that these scaffolds support the growth of organoids. ( J ) Immunolocalization of phospho-histone 3 (PH3) (a marker of proliferating cells) and cytokeratin 8 (CK8) (a marker of epithelial cells) detected the presence of proliferating epithelial cells in these scaffolds. Scale bars, 100 µm unless indicated otherwise.

The uterus is a central organ of the female reproductive tract and is essential for the reproduction and perpetuation of most mammals ( 38 ). The histoarchitecture of the uterus is relatively similar across species, from viviparous lizards to primates, and consists primarily of three major layers: an outer covering, the serosa; a middle muscular layer, the myometrium; and an inner mucosa layer, the endometrium ( 39 , 40 ). The endometrium is a complex layer that lies nearest to the uterine lumen and is further subdivided into epithelial and stromal compartments ( 41 ). The epithelium consists of luminal and glandular epithelium ( 41 ). The endometrium undergoes significant remodeling, primarily due to cyclical changes in ovarian hormones ( 42 ). In women, the endometrium undergoes massive expansion under the influence of estrogen during the proliferative phase of the menstrual cycle, and it acquires a more-secretory phenotype during the second half of the cycle due to the rising levels of progesterone to support the embryonic development ( 43 ). If embryonic implantation and subsequent pregnancy are not established, then the endometrium undergoes breakdown and sloughs off during menstruation ( 43 ). Endometrial repair ensues during menstruation, leading to the scarless repair of the endometrium and transition to the next menstrual cycle ( 43 ). The process of endometrial repair is proposed to be dependent on the endometrial stem/progenitor cells located toward the base of endometrial glands in the basal layer of the endometrium ( 43 , 44 ). In most organ systems, active Wnt signaling marks stem/progenitor cells that are responsible for their maintenance and repair ( 45 ). High Wnt signaling activity is observed in the uterine glandular epithelium, and the loss of Wnt signaling in uterus severely compromises its development and functions ( 41 , 46 ). Cell lineage–tracing studies in mice have identified Lgr5 and Axin2, two well-known Wnt signaling targets, as markers of a subset of glandular epithelial cells that act as endometrial stem/progenitor cells ( 44 , 47 ). Under high Wnt culture conditions, endometrial stem/progenitor cells form endometrial organoids ( 44 , 47 ). In the past few years, endometrial organoids have been developed from normal and abnormal human endometrium ( 1 , 48 ). These organoids replicate many endometrial physiological functions, including histological and secretory changes associated with cyclical hormonal variations, and have already provided unique insights into the pathogenesis of human endometrial diseases, such as endometriosis, adenomyosis, and endometrial cancer ( 49 ). To increase the clinical transferability of organoid technology, we need to optimize organoid derivation and expansion protocols ( 17 ). One of the current limitations of organoid technology is the frequent use of Matrigel for establishing and propagating patient-derived organoids ( 17 ). Matrigel is an ill-defined composition of ECM extracted from undifferentiated mouse tumor cells known to deposit an excessive amount of basement membrane proteins ( 16 , 17 ). Given there are significant differences in ECM composition between normal and tumor tissues, Matrigel is unable to mimic the native tissue-like environment in which epithelial cells typically reside ( 17 , 29 ). Therefore, both natural and synthetic hydrogels are urgently needed to optimally recapitulate the natural environment of epithelial cells. The matrisome of tissues is highly complex and consists of more than 1,000 ECM and ECM-related proteins; therefore, it is difficult to replicate this biological complexity in the synthetic hydrogels ( 29 ). This study has developed ECM hydrogels from the bovine and human endometrium. We showed that these hydrogels support the growth of both human and mouse endometrial organoids, which was comparable to their growth in Matrigel. More importantly, the proteome of organoids cultured in the ECM hydrogel was more similar to the native tissue from which they were derived than Matrigel-grown organoids. We chose to take this approach to develop endometrial hydrogels because animal ECM-based medical products are already approved by FDA and implanted in more than a million patients ( 13 ). Therefore, there is a well-established regulatory pathway and approved safety profile for the potential clinical translation of animal endometrium-derived ECM hydrogels ( 13 , 14 ). Decellularization of tissues is the crucial initial step required to preserve the complex ECM protein structure for its applicability in developing hydrogels and scaffolds ( 50 ). Ionic detergents (e.g., SDC are SDS) are commonly used for decellularizing tissues ( 50 ). SDS is effective in removing cellular and nuclear material in a shorter timeframe. However, SDS treatment negatively impacts ECM matrisome, causing loss of collagens, and GAGs, resulting in poor biomechanical integrity of the ECM ( 51 ). SDC has comparable decellularizing abilities to SDS and is better at preserving the biochemical and biomechanical properties of the ECM ( 51 ). Previous studies have relied mainly on SDS to decellularize tissues, including the intestine and endometrium ( 52 , 53 ). Unfortunately, hydrogels developed from the resulting decellularized tissues were unable to support the growth of organoids ( 53 , 54 ). This study systematically compared three different decellularization procedures on the endometrial tissue and showed that all three protocols preserved the general morphology, integrity of ECM, and histological structure. However, in-depth proteomic analysis showed significant differences in tissues decellularized using 1% SDS, 4% SDS, and 4% SDC. SDC disrupts cell membranes when used in decellularization procedures while retaining the native protein structure better than SDS ( 55 ). Consistently, we observed the retention of many key ECM proteins (e.g., laminins and collagens) that are required for organoid growth and development in hydrogels prepared using an SDC-based protocol compared with SDS-based protocols. Our assessment of the mechanical properties of hydrogels showed that the SDC-based protocol delivered a stronger hydrogel with superior viscoelastic features (storage and loss modulus) compared with SDS-based protocols. The preservation of the native ECM biomechanical cues is essential for maintaining the biocompatibility of hydrogels/scaffolds following recellularization and for the successful regeneration of intended tissues in patients after transplantation ( 50 ). Tubular organ regeneration and repair have advanced significantly, in part due to their unified structural characteristics and simpler three-dimensional microarchitecture than solid organs, like the liver ( 56 ). We demonstrated the ability of our SDC protocol-based hydrogel (P3) to generate tubular structures similar to the decellularized mouse uterus and showed its promise for tubular organ repair and regeneration. In summary, we have successfully developed endometrial ECM hydrogels as an alternative to Matrigel to improve organoid cultures to better reflect human endometrial biology and pathologies in a more-natural environment. These ECM hydrogels are relatively easy to prepare and more economical than Matrigel, which will allow for their broader and rapid adoption by others in the field.