AMs were obtained from the Gynaecology and Obstetrics Department at Cruces University Hospital (Barakaldo, Spain), their use in this study was approved by the Cruces Ethics Committee (CEIC code E20/05). All human tissues were handled in accordance with the tenets of the Declaration of Helsinki regarding medical research involving human subjects. The patient selection process followed the same criteria established for AM donation intended for ocular surface surgery. The inclusion criteria required participants to be between 18 and 42 years of age, seronegative for human immunodeficiency virus (HIV), hepatitis B and C, and syphilis, and to have experienced full-term pregnancies (38–42 weeks). Exclusion criteria ruled out individuals with gestational or fetal diseases.

Human placentas were managed in alignment with the principles outlined in the Declaration of Helsinki. Following the acquisition of informed consent, four human placentas from Caucasian women were collected following elective caesarean deliveries and promptly processed under sterile conditions. First, the placentas were rinsed with sterile saline solution to eliminate blood clots. The AM was then meticulously detached from the chorion using blunt dissection and thoroughly rinsed several times in a saline solution enriched with antibiotics and antifungal agents (penicillin 10,000 U/mL, streptomycin 50 mg/mL, and amphotericin B 2.5 mg/mL). Subsequently, each AM was transferred to a sterile container containing a 1:1 mixture of Dulbecco’s Modified Eagle’s Medium (DMEM) and glycerol (Gibco®, USA) and preserved at −80°C until required for use.

The protocol for extracting proteins from the AM was carried out as established previously (14).

Label-free protein identification based on liquid chromatography-mass spectrometry (LC–MS) was performed to assess the differences in the proteome of the AM tissue used for transplantation (directly in contact with the corneal epithelium) and the AME obtained by sonication and incorporated into the hydrogels studied. For AME, a previously described mild homogenization protocol was used (14). This method is based on a gentle approach designed to preserve the functionality and biological activity of the extracted proteins. In contrast, protein extraction from the AM tissue was obtained through Precellys (Bertin Technologies, Montagne-le-Bretonneux, France) homogenization of the AM extract in a buffer containing 7 M urea, 2 M thiourea and 4% CHAPS. The proteins obtained by the two methods were digested with trypsin (Sigma Aldrich, St. Louis, MO, USA) following the FASP protocol described previously (33). The trypsin was prepared in 50 mM ammonium bicarbonate and added to the protein at a trypsin: protein ratio of 1:10, incubating the mixture overnight (ON) at 37°C. The resulting peptides were dried in a RVC2 25 speedvac concentrator (Christ, Vienna, Austria), resuspended in 0.1% Formic Acid (FA) (Sigma Aldrich, St. Louis, MO, USA), desalted and again resuspended in 0.1% FA using C18 stage tips (Millipore, St. Louis, MO, USA). The peptides recovered were then loaded onto an EvoSep One (EvoSep, Odense, Denmark) chromatograph coupled on-line to a TIMS ToF Pro mass spectrometer (Bruker, Coventry, UK), using Parallel Accumulation Serial Fragmentation (PASEF) acquisition to rapidly achieve a highly sensitive analysis following the 30 SPD protocol (approx. 44 min runs) under default Evosep settings. Data-dependent acquisition (DDA) was used and the data was processed with MaxQuant software (Max Planck Institute of Biochemistry, Martinsried, Germany) using default parameters (34).

Searches were carried out against a database consisting of human protein entries from Uniprot/SWISSPROT [UniProt Consortium, European Bioinformatics Institute (EMBL-EBI), Hinxton, UK; Swiss Institute of Bioinformatics (SIB), Geneva, Switzerland; and Protein Information Resource (PIR), Georgetown University, Washington, DC, USA]. Carbamidomethylation of cysteines was considered as a fixed modification and the oxidation of methionines as a variable modification. Proteins passing a false detection rate (FDR) filter of <1% at the protein level were considered for further analysis. Spectral Count data was used to compare the different conditions, specifically using a modified spectral counting method named Normalized Spectral Abundance Factor (NSAF) (35). Briefly, protein spectral counts (the sum of all peptide identifications obtained for a given protein) were corrected by protein length, obtaining the Spectral Abundance Factor (SAF) for each protein. These SAF values were further normalized against the sum of all SAF values in a specific sample and expressed as a percentage of the total.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (36) partner repository with the dataset identifier PXD055969.

Gene Ontology (GO) enrichment analysis was carried out using the DAVID online tool (http://david.abcc.ncifcrf.gov/summary.jsp, Accessed 7 October 2020) (37). DAVID is a GO Term annotation and enrichment analysis tool used to highlight the most relevant GO terms associated with a given gene list.

Two hydrogels were synthesized, designated as hydrogel A (treatment 3) and B (treatment 4), each with distinct formulations but encapsulating the same AME proteins.

The release of AME proteins from the hydrogels was evaluated in a similar manner in both cases. Following the encapsulation of the lyophilized AM extract, 50 μL of hydrogel A and a fragment of hydrogel B, designed in a format suitable for application on the conjunctiva with dimensions of 0.5 × 0.5 × 0.1 cm, were placed in a release system developed by our group and previously reported (39).

Both hydrogels were exposed to the external environment while a buffering agent was introduced at the upper end, flowing uniformly over the surface of the material. Aliquots of this buffer containing the proteins released from the hydrogels were subsequently collected. The hydrogels gradually dissolved in the buffered solution, mimicking the action of tears.

Despite the stabilization of hydrogel B through physical interactions, the introduced liquid was able to dissolve the polymers comprising the matrix, ultimately leading to the complete dissolution of the hydrogel. The liquid used was an isotonic saline solution for ocular irrigation, known as Balanced Salt Solution (BSS; AJL Ophthalmic, Spain). Since the natural tear flow rate in the human eye is 2.2 μL per minute, this value was scaled up by a factor of 50–110 μL per minute to meet the experimental requirements. The total release of proteins was completed after 360 min, and samples were collected every 15 min to study the release kinetics. Protein quantification was performed indirectly using EZQ fluorescence with a BioTek Synergy Neo 2 multimode reader and GEO 5 3.1 software. Subsequently, the values were interpolated using a calibration curve in the range of 1–25 μg/mL. Finally, the data obtained were processed using GraphPad Prism software.

This study was carried out on male New Zealand White rabbits (n = 44, weight 2.5–2.7 kg), with all the animal procedures adhering to the guidelines set forth by the Association for Research in Vision and Ophthalmology (ARVO) regarding the Use of Animals in Ophthalmic and Vision Research. Approval for the experimental protocols was obtained from the Animal Experimentation Ethics Committee of the Basque Public Health System and the Institute for Health Research Biogipuzkoa (permit number: CEEA20/15), as well as the Provincial Council of Gipuzkoa (Cod. PRO-AE-SS-196), ensuring compliance with both European and national legislation.

An animal model of alkali corneal burn was established as described previously (40). Briefly, rabbits were sedated with a subcutaneous injection of Meloxicam (0.2 mg/kg: Boehringer Ingelheim, Spain) and an intramuscular injection of ketamine/xylazine (35/20 mg/kg: Halmein Pharmaceuticals, Germany/Carlier, Spain), followed by anaesthesia with Sevofluorane 3% (Abbvie, Spain) and topical anesthesia with 2–3 drops of Colircusi Anestésico Doble® (Alcon-Cusí, Spain). An alkali burn was induced in the right eye of each rabbit by placing an 8 mm strip of filter paper soaked in 1 M NaOH on the central cornea for 60 s under an ophthalmic surgical microscope. Subsequently, the eyes were rinsed with 200 mL saline solution for 1 min, and eye drops containing tobramycin and dexamethasone phosphate (Alcon Cusi, Barcelona, Spain) were administered post-surgery (p.s.) to each animal. The rabbits were then randomly divided into four groups (n = 11 per group) and they received one of the following treatments: group 1 (no experimental treatment), post-surgical eye drops alone; group 2, Amniotic Membrane Transplantation (AMT); group 3, dynamic hydrogel (hydrogel A); and group 4, physically cross-linked ocular insert hydrogel (hydrogel B). Both hydrogels were loaded with the AME and the contralateral uninjured eye (native) was used as the control.

After inducing the corneal lesion, hydrogels A and B were placed in the conjunctival sac following different approaches. Hydrogel A was administered to rabbits in Group 3 via a syringe in gel format, spread over the entire corneal surface by blinking. Hydrogel B was pre-hydrated for 60 s in a Balanced Salt Solution (BSS) before it was placed in the upper conjunctival sac of Group 4 rabbits, applying gentle pressure to enhance gel adhesion. Hydrogels were administered every 12 h over 7 days by an experienced ophthalmologist. After inducing the corneal wound, the AM attached to the cornea as a monolayer in Group 2 (41) was removed on the 7th day of follow-up following administration of 2–3 drops of topical anesthesia.

Corneal wound healing was monitored by capturing photographs (EOS 750D, Canon Inc., Japan) immediately post-surgery (day 0) and on days 7, 14, 21 and 28 p.s. The images were obtained under cobalt blue light post 0.1% fluorescein sodium staining and the area of the corneal epithelial lesion was quantified (mm²) three times using Image J software (National Institutes of Health (NIH), Bethesda, MD, USA). Technical replicates were employed for the analysis of the de-epithelialized area, with three measurements performed for each biological replicate. The biological replicates consisted of nine specimens per group.

Two rabbits were selected randomly from each group and sacrificed with an overdose of anesthesia on day 2 post-alkali burn (acute healing), while the remaining animals were sacrificed on day 28 (late healing). The corneas were harvested, fixed in 4% paraformaldehyde, preserved in 30% sucrose, embedded in OCT to obtain cryostat sections (7 μm). The sections were stained with hematoxylin and eosin (H&E) or Masson trichrome and random fields of the injured area and the adjacent tissue were analyzed. Images were obtained under a light microscope (Nikon Eclipse 80i, Japan), and the epithelial and stromal thicknesses were measured at three random locations in each field using ImageJ software. The presence or absence of an endothelial layer and inflammatory cell density were recorded.

The data were analyzed using SPSS 22 software (IBM-SPSS, USA). In the analysis of the in vivo studies, and given the sample size and the non-normal distribution of the variables, non-parametric tests were used (Mann–Whitney U, Kruskal Wallis and Chi-squared). Variables with a normal distribution, such as the stromal and epithelial thickness, and their relationship with endothelial damage, were studied using parametric tests (ANOVA and Student T-test). Significance was set at p < 0.05.

The proteins in the AM tissue (group 2) and the AME introduced into the hydrogels (groups 3 and 4) were analyzed, identifying a total of 1,689 proteins. Of these, 1,059 were common to both the tissue and the extract, with 288 proteins (17%) identified exclusively in the AM tissue and 342 proteins (20%) specific to the AME (Figure 2A and Supplementary File S1 for the complete list of proteins in these formulations). The difference in composition may be attributed to the different protein extraction protocols used in this study. For protein extraction from the AM, a gentle homogenization protocol was employed, based on a non-aggressive approach designed to preserve the functionality and biological activity of the extracted proteins. In contrast, protein extraction from the AM tissue was performed using a Precellys device and a buffer containing 7 M urea, 2 M thiourea, and 4% CHAPS. This latter protocol was designed to obtain a broader spectrum of proteins, including those that are insoluble or membrane-associated.

A GO analysis was carried out to characterize the functional events in which the proteins identified in both the AME and the AM tissue are involved (Figure 2B). These processes can be mainly grouped as: (i) Cellular activity and wound healing consisting of cell adhesion, extracellular matrix remodeling and protein synthesis; (ii) Immune responses, combining the GO terms of innate immune responses, complement activation (classical pathway), complement activation and inflammatory responses; and (iii) Platelet aggregation, comprising the GO terms of blood coagulation, platelet activation and homotypic cell–cell adhesion.

When evaluating the release kinetics of AM proteins from hydrogel A, it was observed that 34 ± 9% of the total protein content was rapidly released within the first 15 min (Figure 3A). Subsequently, the protein concentration in the collected BSS medium gradually decreased over time, eventually falling below the assay’s detection threshold after 60 min. At this point, the cumulative protein release reached approximately 60%, corresponding to a release duration equivalent to 2 days when considering the natural tear flow rate in the human eye. These findings suggest that AM proteins are retained within hydrogel A via weak physical interactions, distinct from the dynamic bonds observed in hydrogel B, which influence the hydrogels’ solubility in BSS during the experiment. The pronounced release of proteins during the first hour of the experiment (up to 30 ± 3 μg/mL, Figure 3B) is particularly significant for ocular regeneration. This initial burst release of AM proteins could potentially facilitate rapid activation of the healing process in the damaged area, while the subsequent sustained release would help maintain this therapeutic activity over an extended period.

Hydrogel B gradually dissolved into the serum, releasing both the AM proteins and the polymers constituting the hydrogel. The total amount of released proteins was quantified every 15 min, showing that 50% of the total protein content was released after 165 min, 80% after 225 min, and 100% after 265 min (Figure 3C). The maximum protein release occurred at 241 min (Figure 3D). This sustained release highlights the material’s ability to ensure a consistent flow of proteins into the animal’s eye.

In this study, 44 rabbits were distributed evenly into four groups and a corneal lesion was induced in each one with 1 M NaOH as indicated in the Methods. Of these, eight rabbits were sacrificed on day 2 p.s. and 36 rabbits on day 28 p.s. to perform histological analyses (see section 3.3). While one of the four study groups did not receive any experimental treatment, Group 2 received AMT applied directly to the eye, whereas Groups 3 and 4 received the AME delivered through hydrogels. The ointment-like consistency of hydrogel A, combined with the mucoadhesive properties of HA-SH, facilitated its uniform spread across the cornea. By contrast, the physical crosslinking in hydrogel B conferred this matrix with remarkable stability and good mucoadhesion.

Following the induction of corneal wounds, Hydrogel A was applied to Group 3 animals every 12 h for 7 days in the upper conjunctival sac. The hydrogels were safely administered via syringe and demonstrated biocompatibility in the lower fornix. Daily general observations indicated that 6 h post-application, the hydrogel was no longer visible on the ocular surface, having completely degraded. After 12 h, no traces of the hydrogel were detected.

Similarly, after inducing corneal wounds, Hydrogel B was applied to Group 4 animals every 12 h for 7 days in the upper conjunctival sac, with gentle pressure applied to enhance gel adhesion. Observations revealed that the hydrogels were safely retained and showed biocompatibility in the upper fornix. Daily evaluations of the films demonstrated that, while the films exhibited mild wear 6 h post-application, they did not degrade or dissolve entirely. By the 12-h observation, no hydrogel remnants were detected.

In both cases, no signs of irritation, redness, or swelling of the eyes or ocular structures were observed in any animal as a result of hydrogel application. The animals were monitored and assessed over the 7-day application period to detect any potential signs of pathology.

The progression of epithelial defects was monitored for 28 days using fluorescein staining (Figure 4A) and a reduction in the epithelial lesion area (mm²) was observed in all the treatment groups on days 7, 14, 21, and 28 p.s. relative to the day of intervention (p < 0.005, Wilcoxon test: Table 1). A high incidence of persistent epithelial defects and wound reopening was observed in 91.6% of corneas, probably reflecting the intensity of the alkaline burn and the fragility of the newly formed epithelial (42). There were statistically significant differences in the area of the epithelial defect between the groups on day 7 (p = 0.014, Kruskal-Wallis test), with significant differences between group 1 and group 4 (p = 0.02) and between group 2 and group 4 (p = 0.006, Mann–Whitney U test). However, there were no significant differences between the groups on days 14, 21 and 28 (Kruskal–Wallis test: Figures 4C–E). Hydrogel A facilitated the fastest healing, while eyes treated with hydrogel B exhibited delayed wound repair. Thus, the differences between the untreated eyes and those that received hydrogel B were only significant on day 7 p.s., with open wounds evident in all eyes (Figure 4B).

The proportion of completely healed corneas did not differ significantly between the untreated and treated eyes on the days studied (Figure 5). On day 14 p.s., the proportion of completely healed corneas was higher in group 3 (44.4%) and 4 (55.5%) than in the untreated eyes in group 1 (33.3%). The enhanced healing following these treatments, even after the treatment was discontinued on day 7 p.s., suggests that AM proteins may enhance the induction of wound closure in the early stages of healing.

The histological analysis focused on the stromal, epithelial and endothelial layers, as well as assessing the presence of inflammatory cells during the early and late healing phases (days 2 and 28 p.s., respectively). No significant differences were observed in the stromal and epithelial thickness between groups (ANOVA test), despite the apparent increase in epithelial thickness in the eyes of group 2 animals on the days analyzed (Table 2). Central corneal sections exhibited a thicker stroma and thinner epithelium in wound sections relative to the native corneas (Figure 6A). Masson’s trichrome staining highlighted severe alterations to the collagen staining that persisted from the early to the late healing phases (Figure 6B).

Abnormal epithelial regeneration was evident, with disrupted junctions between the epithelial layer and basement membrane observed (Figure 7). Fewer cells were present in the anterior stromal third on day 2 p.s. relative to day 28 p.s., with significant differences in the presence of inflammatory cells on day 28 p.s. between the eyes from group 2 animals relative to those from groups 1 and 3 (p = 0.01, Kruskal Wallis Test: Figure 8). Moreover, alterations to the endothelium were noted, with more extensive endothelial damage correlating with a significantly thinner epithelium (p = 0.001, Student T-test).