A detailed description of the methodology used along this research can be found in Figure 1.

Entire female porcine reproductive tracts with intact vasculature were donated by a slaughterhouse in Mercavalencia, Spain according to ISO 9001 quality management and subjected to whole organ decellularization. Uterine horns (n = 5) were attached to a peristaltic pump (Cole-Parmer Instruments, Fisher Scientific) and decellularized by perfusion of 0.1% sodium dodecyl sulfate (SDS) and 1% Triton X-100 through the uterine artery for 48 h, following a two-cycle protocol (Campo et al., 2017). Then, decellularized (DC) horns stored at −80°C were cut transversally into 1 mm thick ring-shaped disks. The endometrial fraction was isolated via microdissection under a stereomicroscope (SMZ800, Nikon) by cutting the luminal side of the inner circular myometrial layer, which appears as a dense line in both No-DC and DC horns (Figure 2; Campo et al., 2019). The remaining myometrial fraction was kept as a control to verify the correct isolation of pure endometrium and the presence of tissue-specific components during proteomic analysis. Endometrial tissue from No-DC uterine horns (n = 5) was also isolated via microdissection as a control for subsequent analyses. Isolated tissues were stored at −80°C.

Additional steps were performed to assure the removal of residual DNA and detergents. Endometrial and myometrial tissue stocks from uterus decellularization were thawed, weighed, and washed in cold PBS (10 mL/g tissue) for 30 min at 200–250 rpm. Then, tissues were incubated for 1 h in 5 μg/mL Dnase I solution (D5025, Sigma-Aldrich) at room temperature and washed again. Aliquots of washing media were stored to detect residual SDS in the DC endometrial tissue.

The presence of SDS in the washing media after endometrial isolation (n = 8 DC horn pieces) and subsequent washes (pool of total isolated endometrial tissue) was quantified. It was done by measuring absorbance when SDS reacts with Stains-All dye (Stains-All, Sigma Aldrich; Rupprecht et al., 2015). A calibration curve of 7 standards (0,01–0,1 mg SDS/ml) were made with serial dilutions in PBS. Stains-all was dissolved in N, N-dimethylformamide to 2.0 mg/ml and then diluted 1:20 in ultrapure distilled water to working solution. In the assay, 10 μl of all standards and samples were pipetted in a 96-well plate containing 140 μl 0.1X PBS. Afterward, 50 μl Stains-all working solution (1:20) was added and absorbance was immediately measured at 453 nm using a microplate reader (Spectra Max 190, bioNova Scientífica, S.L.). SDS concentration was calculated by adjusting to the calibration curve with a lineal fit. Total amount of SDS was calculated from initial volume and normalized to individual weight of DC endometrial tissue used per wash. Samples were done in triplicate.

For every histological and immunohistochemistry/fluorescence analysis, representative tissue sections were deparaffinized using xylene and rehydrated by decreasing concentrations of ethanol and distilled water. Heat-mediated antigen retrieval was performed in 10 mM Citrate Buffer 0.05% tween pH6.0 for 20 min in a 95°C water bath. Sections were permeabilized with 0.05% Tween 1X PBS, blocked with 3–10% bovine serum albumin (BSA) for 1 h at RT and incubated with primary antibodies in 1% BSA overnight at 4°C. For bright-field microscopy, DAB Substrate Kit was used according to the instructions and sections were counterstained by hematoxylin. For immunofluorescence, slides were incubated with an Alexa-Fluor 488 secondary antibody (1:500 dilution) and mounted with 6-diamidino-2-phenylindole (DAPI, P36931, Thermo-Fisher Scientific).

The absence of cellular components and nuclei was investigated by Hematoxylin and Eosin (H&E) staining or by DAPI. Collagen preservation was also evaluated by Masson’s Trichrome (MT) staining using standard protocols. Lastly, samples were immunoassayed for the presence of the alpha-gal epitope (α−Gal Epitope monoclonal antibody M86, ALX-801-090-1, Enzo Life Sciences, 1:5 dilution) by bright-field immunostaining.

Isolated DC endometrial tissue stock was flash-frozen in a mortar with liquid N₂, milled manually, and lyophilized (Lyoquest-85, Telstar, Valencia’s Polytechnic University) over 96 h at 20 Pa. The resulting endometrial lyophilized powder was digested and neutralized using a modified protocol (Brown et al., 2017). Briefly, 1% (w/v) lyophilized powder was suspended in 0.01 M HCl (H1758, Sigma-Aldrich) with 0.1% pepsin (P7000, Sigma-Aldrich) and digested for 48 h under agitation. The solution was left on ice and neutralized with 10% (v/v) 0.1 M NaOH (S8045, Sigma-Aldrich), 11.11% (v/v) 10X PBS (P5493, Sigma-Aldrich), and 1X PBS was used to reach the desire concentration. The resulting EndoECM solution was stored at −80°C.

This process was also performed with isolated DC myometrial and No-DC endometrial tissue stocks to create myometrial extracellular matrix (MyoECM) and No-DC endometrial matrix (No-DC Endo). MyoECM and No-DC Endo were used as controls for subsequent proteomic analyses.

The gelation kinetics of EndoECM (n = 3) were evaluated by turbidimetry. Absorbance at 405 nm for 100 μL of 3, 6, and 8 mg/mL EndoECM was measured every minute in a microplate reader (SpectraMAX 190, Molecular Devices) at 37°C. Absorbance was normalized with the following formula as described by Freytes et al. (2008):

Where NA is the normalized absorbance, A: absorbance t at given time, A₀: initial absorbance, and Amax: maximum absorbance.

Kinetic parameters (Lag time, Time to half gelation, Time to complete gelation, and Gelation rate) from the different concentrations were compared (Freytes et al., 2008). The Lag time (T_Lag) was defined as the intercept of the linear region of the gelation curve with 0% absorbance, the Time to half gelation (T_1/2) as the time to 50% absorbance, the Time to complete gelation (T₁) as the time to 100% absorbance, and the Gelation rate (S) as the slope of the linear region of the gelation curve. Statistical data analyzed respect to concentration of 3 mg/mL.

To test stability and sterility, the resulting EndoECM hydrogels were left in Dulbecco’s modified Eagle’s medium (DMEM/F12; Sigma-Aldrich) containing 10% fetal bovine serum (FBS) and 0.1% streptomycin/penicillin for 7 days under standard in vitro culture conditions (37°C, 5% CO₂).

The ultrastructure of 3, 6, and 8 mg/mL EndoECM, 8 mg/mL MyoECM, and 8 mg/mL No-DC Endo hydrogels was evaluated using scanning electron microscopy (SEM). Sample processing was performed in the proteomics facility of SCSIE University of Valencia. This proteomics laboratory is a member of Proteored, PRB3 and is supported by grant PT17/0019, of the PE I + D + i 2013–2016, funded by ISCIII and ERDF. Hydrogels were fixed in 2.5% glutaraldehyde in PBS (Sigma Aldrich, grade II, 25%) for 24 h, washed in PBS, and kept in PBS at 4°C. Then, hydrogels were treated with 2% osmium tetroxide for 2 h and dehydrated in a graded series of alcohol (30, 50, 70, 90, 100% ethanol) for 30 min per wash and kept in 100% ethanol overnight at 4°C. Hydrogels were washed 3 additional times in 100% ethanol for 30 min and critical point dried using a Autosamdri^® 814 Critical Point Dryer (Tousimis) with carbon dioxide (CO₂) at high pressure (1200 pound-force per square inch, psi) as the transitional medium and a maximum heating temperature of 40°C. Dried samples were coated with gold-palladium for 2 min using a SC7640 Sputter Coater (Quorum technologies) and imaged with a SEM FEG Hitachi S-4800 (SCSIE University of Valencia, Spain). To analyze fiber diameter, four measurements per three 30.0 k fields per sample were measured using ImageJ software (Schindelin et al., 2012).

The proteomic analysis was performed in the SCSIE proteomics facility of University of Valencia. 50 μg of EndoECM, MyoECM, and No-DC Endo (8 mg/mL) were loaded and resolved in a 1D SDS-PAGE gel. Every sample lane was sliced into seven fragments. Gel slides were digested using sequencing grade trypsin (Promega) at 37°C as described elsewhere (Shevchenko et al., 1996). 200 ng of trypsin were used for samples and digestion was performed at 37°C. Trypsin digestion was stopped with 10% trifluoroacetic acid (TFA) and the supernatant (SN) was removed, then the library gel slides were dehydrated with pure acetonitrile (ACN). The new peptide solutions were combined with the corresponding SN. The peptide mixtures were dried in a speed vacuum and resuspended in 2% ACN; 0.1% TFA. The final volume was between 6 and 25 μL.

Liquid chromatography and tandem mass spectrometry (LC–MS/MS) were performed. 5 μL of sample was loaded onto a trap column (NanoLC Column, 3 μ C18-CL, 350 μmx0.5 mm; Eksigen) and desalted with 0.1% TFA at 2 μL/min for 10 min. Peptides were then loaded onto an analytical column (LC Column, 3 μ C18-CL, 75 μm × 12 cm, Nikkyo) equilibrated in 5% ACN 0.1% FA (formic acid). Elution was carried out with a linear gradient of 5a40% B in A for 60 min. (A: 0.1% FA; B: ACN, 0.1% FA) at a flow rate of 300 nL/min. Peptides were analyzed in a mass spectrometer nanoESI qQTOF (5600 TripleTOF, ABSCIEX). Sample was ionized applying 2.8 kV to the spray emitter. Analysis was carried out in a data-dependent mode. Survey MS1 scans were acquired from 350–1,250 m/z for 250 ms. The quadrupole resolution was set to “UNIT” for MS2 experiments, which were acquired 100–1,500 m/z for 50 ms in “high sensitivity” mode. Following switch criteria were used: charge: 2+ to 5+; minimum intensity; and 70 counts per second (cps). Up to 50 ions were selected for fragmentation after each survey scan. Dynamic exclusion was set to 15 s. The system sensitivity was controlled with 2 fmol of 6 proteins (LC Packings).

ProteinPilot default parameters were used to generate peak list directly from 5600 TripleTof wiff files. The Paragon algorithm (Shilov et al., 2007) of ProteinPilot v 5.0 was used to search the UniprotMammals database (version 03-2018) with the following parameters: Trypsin specificity, (iodoacetamide) cys-alkylation, taxonomy not restricted, and the search effort set to through. The protein grouping was done by Pro group algorithm. The formation of protein groups was guided entirely by observed peptides only, which originated from the experimentally acquired spectra. Because of this, the grouping was guided by spectra. Unobserved regions of protein sequence played no role in explaining the data. Proteins showing unused score > 1.3 were identified with confidence ≥ 95%. Mass spectrometry information of all the fragments were combined for protein identification using the UniprotMammals database (SCSIE University of Valencia).

Filtered output files for each peptide were grouped according to the protein from which they were derived and their individual coverage (% cov) was determined as an indicator of protein abundance of relative quantitation analysis. Common contaminants were excluded following the exclusion criteria of Hodge et al. (2013) and based on their expression in target tissue according to The Human Protein Atlas database¹ (Uhlen et al., 2015). A list of peptides found in proteomic analysis in EndoECM (Supplementary Table 1), MyoECM (Supplementary Table 2), and No-DC Endo (Supplementary Table 3) can be found in Supplementary Material. Gene ontology (cellular component and molecular function) analysis of the detected proteins was performed through the PANTHER classification system (Mi et al., 2018) and refined according to processes related to ECM.

This study was approved by the Human Ethics Committee at the IVI Foundation (1706-FIVI-053-IC, Valencia, Spain). In vitro cytocompatibility studies were carried out with epithelial and stroma cells from human endometrial stem cell lines and primary human endometrial cells. Epithelial (ICE6) and stromal (ICE7) endometrial stem cell lines were obtained using Hoechst methodology and cloning efficiency (Clone ICE6 and Clone ICE7, Richmond, BC, Canada). Characterization, purity, and clonogenicity were previously reported by Cervelló et al. (2010, 2011). Passages 7 to 12 of ICE6 and ICE7 were used for experiments. Endometrial epithelial cells (EECs) and endometrial stromal cells (ESCs) were obtained from fresh endometrial biopsies from healthy oocyte donors (n = 12). Briefly, biopsies were mechanically and enzymatically disaggregated. Then, EECs and ESCs fractions were separated based on size, sedimentation, and membrane filtration (Simón et al., 1993; Cervelló et al., 2010, 2011). Fresh or first passage (P1) EECs and ESCs were used for experiments.

All mouse procedures were performed in accordance with Directive 2010/63/EU and the Ethics Committee for Animal Welfare of University of Valencia (A-1510673251016). Injections of 200 μL of 8 mg/mL EndoECM or No-DC Endo (control) were given in the dorsal subcutaneous space of female immunocompetent mice (C57BL/6; UCIM, SCSIE University of Valencia). Hydrogels remained inside the mice for 2 (n = 3 for EndoECM and n = 2 No-DC Endo), 7 (n = 1 EndoECM and n = 1 No-DC Endo), and 14 (n = 1 for EndoECM and n = 1 No-DC Endo) days before mice were euthanized. MT staining was used for quantitative assessment of cell infiltration and scaffold morphology. The macrophage response to implanted hydrogels at 2-, 7-, and 14-days post-surgery was characterized by immunolabeling of CD68 pan-macrophage marker (CD68 polyclonal antibody, ab125212, ABCAM, 1:100 dilution). Five x20 fields per sample were quantify using QuPath analysis software v0.2 (Bankhead et al., 2017).

Data were analyzed using RStudio^® software version 3.6.3 (RStudio Team, 2020) and presented as mean ± standard deviation (SD). All statistical analysis was done with a linear regression model to account for total variability (non-parametric analysis). Here, the value of each variable from the groups of study was estimated by the average difference with respected to a reference group. The P value was obtained from contrast hypotheses of the lineal model, indicating with 95% confidence that the difference between the groups is not zero and different without the need of multiple comparison tests. In all cases, P < 0.05 was considered significant.

Whole uterus decellularization was carried out using one or both uterine horns from healthy pigs and the entire endometrial fraction was isolated via microdissection (Figures 2, 3AIII). Decellularization efficiency was verified by H&E (Figures 3B1,2) and MT staining (Figures 3B3,4), which showed complete depletion of cellular material, while DAPI staining showed a total removal of cell nuclei (Figures 3B5,6). Blue coloration in MT staining in DC tissues indicates collagen conservation, a principal ECM component.

Immunostaining of the α-gal epitope, the major responsible for hyperacute rejection of pig xenograft organs in humans (Macher and Galili, 2008), showed a high abundance in luminal and glandular epithelium as well as in blood vessels from No-DC endometrial tissue; in contrast, there was no signal present throughout DC endometrium (Figures 3B7,8).

Quantification of SDS following endometrial isolation detected 158 ± 60.1 μg SDS/g wet tissue. After six 30 min washes with ice-cold PBS under agitation, this was reduced to 33.9 ± 12.4 μg SDS/g wet tissue, corresponding to a statistically significant reduction of 78.6% (P < 0.0001; Figure 3C).

After decellularization, isolated endometrial tissue was milled and lyophilized. Comparison of the DNA content of DC and No-DC showed a significant reduction of nuclear material (5.4 and 7.6% of DNA remained in DC with respect to No-DC wet endometrial tissue and lyophilized endometrial powder, respectively, P < 0.0001) and no DNA bands after electrophoretic analysis (Figure 4A). The final viscous solution, EndoECM, formed hydrogels after incubation at 37°C (Figure 3AIV).

Protein quantities and composition were investigated during every step in the production of the EndoECM (shown in Figure 4B). Analysis of total protein content showed a significant 80% decrease in global concentration (19.9 and 19.2% of wet endometrial tissue and lyophilized endometrial powder, respectively) with a significant enrichment of collagen (207 and 148% of wet endometrial tissue and lyophilized endometrial powder, respectively), indicating substantial removal of the cellular protein fraction. Moreover, preservation of elastin and GAGs (25 and 18% of wet endometrial tissue and lyophilized endometrial powder, respectively) was observed (P < 0.0001). In EndoECM, the effects of pepsin were apparent, showing an increase in total protein content to 42%, while the percentage of collagens was reduced to 61.3% and no detectable concentration of elastin was found (P < 0.0001). Content of GAGs was not affected by pepsin digestion (P < 0.0001).

The gelation kinetics of EndoECM hydrogels from different digestions were evaluated spectrophotometrically. All concentrations presented a sigmoidal curve (Figure 5) with a gelation rate (S) that increased with concentration (0.13 ± 0.03 min^–1 in 3 mg/mL, 0.22 ± 0.05 min^–1 in 6 mg/mL, and 0.20 ± 0.02 min^–1 in 8 mg/mL, P < 0.05; Table 1). Time to start of gelation (Lag time, T_Lag) and the time to 50% (T_1/2) and 100% gelation (T₁) was inversely related to hydrogel concentration (P < 0.05). Hydrogels formed completely after 20 min (20.50 ± 3.81, 14.70 ± 1.12, and 14.10 ± 1.86 in 3, 6, and 8 mg/mL, respectively; Table 1). Opacity and thickness of hydrogels was proportional to ECM concentration (Figure 5). EndoECM hydrogels remained intact and no bacterial growth was seen for 7 days in standard in vitro culture conditions, confirming long-term stability, and adequate sterility for culture and subcutaneous injection.

Endometrial extracellular matrix hydrogels present a homogenous, randomly interlocking fibrillar ultrastructure (Figure 6A), and no significant differences in fiber thickness were found among different concentrations (Supplementary Figures 1A,B). Non-decellularized endometrial powder, No-DC Endo, also formed hydrogels and SEM images showed remains of cellular components along the fibers, altering the ultrastructure (Figure 6A, arrows).

Endometrial extracellular matrix hydrogels and MyoECM, both hydrogels made from myometrial fractions, were predominantly composed of approximately 0.10-μm thick fibers, and no significant differences were found (Supplementary Figure 1C). This fiber size and D-period structure are characteristic of collagen fibrils (Veres and Lee, 2012).

To identify the matrisome, proteins were sorted depending on cellular or extracellular origin and classified according to Matrisome database (MatrisomeDB) into core matrisome proteins (collagens, ECM glycoproteins, and proteoglycans) and matrisome-associated proteins (ECM regulators, ECM-affiliated proteins, and secreted factors). Proteins not found in MatrisomeDB but belonging to the extracellular space were classified as others. Preliminary qualitative analysis showed that half of the No-DC Endo extracellular proteins did not appear in EndoECM. There were four extracellular proteins in EndoECM (dermatopontin, fibrinogen, azurocidin, and extracellular kinases) that did not appear in No-DC Endo (Supplementary Figure 2A and Supplementary Table 4).

Quantitative analysis showed that the ECM hydrogels designed in this study consisted almost entirely of ECM (91.8% in EndoECM and 93.3% in MyoECM), while No-DC Endo comprised 41.4% ECM. These enriched ECM components were principally collagens (83.7% in EndoECM and 82.3% in MyoECM; Figure 6B) and maintained their physiological ratios. Immunoreactive molecules in EndoECM decreased (7.5% cellular components and 0.7% immunoglobulins in EndoECM with respect to 41.1% and 16% in No-DC Endo), while no MHC antigens were detected. No proteoglycans were detected in both EndoECM and No-DC Endo hydrogels.

Moreover, EndoECM and MyoECM hydrogels showed a similar removal of cell immunoreactive components but presented differences in composition in both qualitative (Supplementary Figure 2B and Supplementary Table 4) and quantitative (Figure 6B) analysis.

The role of ECM proteins was identified using GO molecular function and refined according to those functions related to ECM (Supplementary Table 4) in Supplementary Material.

Statistical analysis showed no difference in cell growth between uncoated wells and coatings made from collagen, Matrigel, or EndoECM conditions. EndoECM did not induce a significant inhibition of cell growth compared to other groups (Figure 7A).

To test three-dimensional culture cytocompatibility, the matrix quality between collagen, Matrigel, and EndoECM at 3 mg/mL was evaluated. Two extra EndoECM concentrations, 6 and 8 mg/mL, were also included to test the effect of EndoECM concentration. Each cell type was first plated on top of the hydrogels (2.5D culture system; Figure 7B) and a significant increase in proliferation of endometrial stem cells in EndoECM was observed compared to standard matrices at 3 mg/mL. This was not observed in primary endometrial cells, where EECs grown in collagen showed significantly increased proliferation; in Matrigel, both EECs and ESCs showed increased proliferation compared to EndoECM. In ICE6, ICE7, and ESCs, the proliferation in EndoECM increased significantly with concentration (up to 8 mg/mL, P < 0.05).

Lastly, each cell type was encapsulated in EndoECM hydrogels to form a 500-μm thick hydrogel (3D culture system; Figure 7C). At 3 mg/mL, a significant increase in proliferation of ICE6, ICE7, and ESCs was observed in EndoECM compared to collagen. Improved proliferation was also observed in EndoECM in relation to Matrigel in endometrial stem cells, but no significant difference was found for EECs or ESCs. ESC proliferation significantly decreased as EndoECM hydrogel concentration increased (P < 0.0001 from 3 to 8 mg/mL).

Three-dimensional co-culture systems were performed using both stromal and epithelial cells from human endometrial stem cell lines (ICE6-7 constructs) or isolated human primary cells from endometrial biopsies (EECs-ESCs constructs). To slow the reduction of size of the ECM hydrogel, epithelial cells were seeded at day 0 (Method A) or at day 3 (Method B) after stromal cell encapsulation. The ICE6-7 and EECs-ESCs constructs were maintained for up to 10 days. This experimental design is shown in Figure 8A.

After 5 days, all constructs underwent remarkable remodeling, forming a compact round shape at day 7 through day 10 (Figure 8B). When initial cell concentration was high, usually in constructs made using method A, the hydrogel degradation was more aggressive, and constructs acquired a disk shape.

Viability assays showed that both ICE6-7 and EECs-ESCs constructs were viable for up to 10 days (around 90%, Supplementary Figure 3). Comparing Ki67 immunohistochemistry, around 33% of cells in the ICE6-7 constructs and 60% in the EECs-ESCs constructs were proliferative, and no statistical difference was found between methods A and B (Figure 9).

Comparing MT staining, ECM of the contracted constructs were found predominantly in collagens (Figure 10A) at a higher density than in unseeded EndoECM hydrogels (Figure 10B). However, EECs-ESCs constructs made using method B showed little to no blue staining. Both ICE6-7 and EECs-ESCs constructs had vimentin-positive cells surrounded by collagen fibers in a natural 3D shape, while E-cadherin positive epithelial cells formed a luminal-like monolayer (Figure 10A). No apico-basal polarization was seen.

The in vivo biocompatibility of EndoECM hydrogels was tested in an immunocompetent murine model. After 48 h, hydrogels appeared gelled and opaque in the subcutaneous tissue (Figure 11A, indicated by an arrow). MT staining showed encapsulation in No-DC Endo and a high infiltration of rounded cells with large nuclei, which corresponds to the morphology of inflammatory cells (Figures 11BIII,IV). There was a 4-fold increase (P < 0.0001) in cell infiltration in No-DC Endo (6243 ± 244 cells) compared to EndoECM (1599 ± 402 cells; Figures 11B,C). Further immunological evaluation of CD68 revealed early macrophage infiltration in both EndoECM and No-DC Endo hydrogels that was significantly higher in EndoECM (72.0 ± 15.0% and 19.2 ± 7.77% in EndoECM and No-DC Endo, respectively, P < 0.05) at 48 h (Figure 11D). These results provide a proof of concept for our study.

A decrease in volume of both hydrogels was observed after 14 days. Although cell infiltration remained unchanged, MT images of EndoECM samples showed a shift in prevalence from inflammatory-like cells toward spindle-shaped elongated fibroblast-like cells (Supplementary Figures 4A,B). Moreover, the infiltration of CD68 + cells was maintained in EndoECM (72.0 ± 15.0% and 59.5 ± 14.6% at 2 and 14 days, respectively; Supplementary Figures 4C, 5). In No-DC Endo samples, encapsulation disappeared at 14 days, with an increase in CD68 + cells (19.2 ± 7.77% and 35.4 ± 17.5% at 2 and 14 days; Supplementary Figures 4A,C, 5).