Cross-linked decellularized porcine corneal graft for treating fungal keratitis

This study aims to develop a cross-linked decellularized porcine corneal graft (cDPC) as a substitute for lamellar donor corneas and to evaluate the feasibility of using cDPC to treat fungal keratitis. The cDPC was prepared by decellularization, chemical crosslinking and γ-ray irradiation. Transparency, effectiveness of decellularization and biomechanical strength of cDPC were evaluated. The safety and efficacy of using cDPC to treat fungal keratitis were evaluated in the rabbit model. The transparency of cDPC was similar to that of a native porcine cornea (NPC), and no intact cells were observed in cDPC except for an insignificant amount of residual shrinking cellular nucleus. Compared to the NPC, the biomechanical strength of the cDPC was significantly increased. In the rabbit model of lamellar keratoplasty, the implanted cDPC reduced the incidence of corneal perforation, and also maintained transparency in majority. The results of this study suggest that the cDPC is capable of restoring the original transparency of cornea while effectively treating fungal keratitis. The cDPC is a highly promising ideal substitute for the donor human cornea.

collagenase 12 , etc.) processes, or any combination thereof. However, the decellularization process may reduce the biomechanical strength 13 and transparency of APCS.
The development of a decellularized porcine cornea (DPC) would be ideal as it would maintain normal biomechanical properties and transparency by conforming to the existing extracellular matrix (ECM). Bearing this in mind, we prepared DPC using a chemical crosslinking procedure (1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC)/N-Hydroxysulfosuccinimide (NHS) crosslinking) to produce a cross-linked decellularized porcine cornea (cDPC) without the introduction of exogenous toxic substances. EDC/NHS crosslinked DPCs have excellent biomechanical and biocompatibility properties while maintaining transparency [14][15][16] .
Here, we demonstrated the development of a decellularized xenogeneic cornea as a substitute for human corneal donor grafts. In this study, cDPC were prepared by decellularization with a gradient osmotic pressure process, followed by EDC/NHS crosslinking and γ-ray irradiation. The cDPCs were tested as donors in the rabbit fungal keratitis model and evaluated for safety and efficacy.

Results
General morphology of cDPC. In order to examine the general appearance of the cDPC prior to transplantation, we examined cDPC by light microscopy. As seen in Fig. 1, in terms of general morphology, the cDPC graft remained transparent after the graft preparation procedure (i.e., decellularization, crosslinking and γ-ray irradiation).

Light transmittance evaluation of transparency.
To obtain a quantitative measure of light transmittance (LT), we measure LT over a range from 280 nm to 780 nm using an ultraviolet spectrophotometer. There was no significant difference between the NPC and cDPC groups at any wavelength between these two different wavelengths (n = 10, p > 0.05) (Fig. 2). H&E staining evaluation of structural integrity. Nuclear staining indicated that, under 100× magnification, corneal epithelial cells, stromal cells, and its endothelial cells were distinctly visible in the NPC group. The stromal collagen fibers appeared to have a normal arrangement, and an abundance of blue-dyed cells with intact structures were observed in the stroma. However in the cDPC group, no intact epithelial or endothelial cells could be observed except for an insignificant amount of residual shrinking cellular nucleus, while the stromal collagen fibers appear to have a normal orderly arrangement as observed in the NPC group (Fig. 3).
Residual DNA Amounts following decellularization. In order to assess the effectiveness of decellularization, we measured the residual amount of DNA that remained on cDPC after processing, and compared it to that in the NPC. The DNA amount of the NPC group (469.97 ± 155.51 ng/mg, n = 6) was much higher than that in the cDPC group (82.16 ± 7.24 ng/mg, n = 6) (p < 0.05).
Determination of swelling rates. In order to determine the swelling rate of NPC and cDPC, we performed the assay as described in the Methods Section. Our results indicate that the highest swelling rates were Figure 1. Appearance of the cDPC graft after preparing procedures including decellularization, crosslinking and γ-irradiation. The transparence of the cDPC is indicated by the letter "F" which can be obviously observed through the graft.
observed during the first hour in both groups. Compared to the NPC group, the stabilized swelling rate of the cDPC group was significantly lower (p < 0.05) (Fig. 5).
Tensile strength testing. The mean tensile strength of the cDPC group was 4.1750 ± 0.7438 Mpa (n = 10) which is significantly larger than 3.4775 ± 0.1584 Mpa (n = 10) in the NPC group. (p < 0.05). In contrast, the   elastic modulus of the cDPC group was 5.75 ± 2.87 Mpa (n = 10), which was not significantly different from that of the NPC group (7.25 ± 2.75 Mpa, n = 10) (p > 0.05). The typical stress-strain curve of cDPC and NPC groups is shown in Fig. 6.
Cytotoxicity. The optical density (OD) values after 48 h and 72 h incubation are shown in Fig. 7. There was no significant difference in the OD values between the cDPC group (n = 9) and the NPC group (n = 9) (p > 0.05). The MTT (3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide) results showed that fibroblast L929 cells were able to proliferate in both cDPC and NPC groups, and there is no significant difference in cell proliferation between the two groups (p > 0.05) (Fig. 7).

Establishment and validation of a rabbit fungal infection model. By the fifth day post-injection
with Fusarium solani into the rabbit corneal matrix, the focus of the injection spread to the peripheral areas with visible satellite focuses, pseudopodia, but no empyema was visible in the anterior chamber (Fig. 8a). PAS staining (which stains fungal hyphae) revealed abundant curved and relatively thick mycelium in the superficial stroma consistent with a fungal infection (Fig. 8b).
Examination under slit lamp after fungal keratitis treatment. Lamellar keratoplasty was performed on both cDPC and NPC groups. As a control, another group was treated only with natamycin and was designated as the Natamycin group. Examinations under the slit lamp were performed after 1, 4 weeks, and 24 weeks post-surgery, respectively (Fig. 9).
Healing status and complications after fungal keratitis treatment. Twenty-four weeks after fungal keratitis treatment was applied, the number of rabbits without fungal infection was 11 (91.7%) in the cDPC group (n = 12), 8 (66.7%) in the NPC group (n = 12), and 9 (75%) in the Natamycin group, respectively (n = 12). The remaining animals had corneal perforations.  Long-term corneal transparency for fungal keratitis. The long-term results suggest that, in Natamycin group, corneal central opacity with significant neovascular growth into the central cornea can lead to severe vision loss. About 80% (7/9) had severe opacity through which no clear iris texture or pupil could be observed. The remaining animals had moderate opacity through which the pupil could be seen but the cornea was not clear.

Discussion
Here, we describe the development of a novel method for preparing cDPC by decellularization with a gradient osmotic pressure procedure, combined with EDC/NHS crosslinking and γ-ray irradiation. Our results demonstrate that the combination of decellularization, crosslinking and γ-ray irradiation effectively reduces the immunogenicity while maintaining the transparency in both in vitro and in vivo experiments, and provides better biomechanical strength, thermal stability, and biocompatibility than other methods. The most important performance parameter of the corneal graft is transparency. Transparency of the corneal stroma depends on retention of the orderly lamellar arrangement of the collagen structure 17 . In this study, we utilized a more moderate and progressive gradient concentration of glycerol in the decellularization process, which protects the three-dimensional structure of the collagen fibers by preventing their disruption that accompanies fast and intense dehydration. In glycerol preservation solutions of high concentrations, the cornea is in an environment that prevents dehydration and the cornea effectively maintains its original shape, thickness and curvature. This precaution ensures that the collagen fibers maintain an arrangement close to a freshly harvested cornea. Likewise, we use slow and uniform crosslinking of the cornea in order to preserve the morphology of the cornea and the structure of collagen fibers.  For reasons stated above, glycerol preservation is superior as it preserves the well-organized collagen fibers found in the fresh cornea and provides a transparent cDPC graft.
While 10kGy γ-ray irradiation disrupts collagen structure 18 , glycerol preservation effectively protects the glycosaminoglycan in the cornea and maintains the microstructure of the acellular matrix 19 and thus maintains corneal transparency. Four weeks after surgery, epithelialization and restoration of endothelial cell function were observed in the cDPC group and was accompanied by a gradual decrease of the water content in the space between the collagen fibers allowing gradual recovery of corneal transparency 20 . Twenty-four weeks after surgery, the cDPC is completely transparent.
As cellular components are considered the main antigens involved in immune rejection, decellularization is the best way to overcome immunogenicity of donor tissue. The decellularization process often increases the space separating the collagen fibers and disrupts the original three-dimensional structure of the extracellular matrix (ECM) 21 . Additionally, residual detergent can be cytotoxic affecting cell regeneration. Thus, it is critical to balance complete decellularization and maintenance of ultrastructure of corneal collagen fiber. Such considerations are key to the successful preparation of donor corneal substitutes. To accomplish this goal, we developed a novel The use of an osmotic pressure process with gradient glycerol concentrations effectively reduces the immunogenicity of the tissue as glycerol not only removes most of the cells, but also reduces the infiltration of inflammatory cells and reduces the incidence of rejection 22 . Glycerol disrupts cell membranes by destroying lipid components, eliminating or inactivating bacteria, fungi, and viruses 23 . At the same time, glycerol effectively protects the glycosaminoglycan in the cornea and maintains the structure of the extracellular matrix (ECM). Previous studies have reported that anhydrous glycerol preservation in a −78 °C freezer was a good method to preserve eligible tissues for deep anterior lamellar keratoplasty (DALK) 24 . Glycerol is useful as a hyperosmotic solution during osmotic treatment as it not only dissolves lipid membranes, removes membrane associated antigens, and soluble proteins and cells, but also changes the osmotic pressure and facilitates the removal of the ECM material. Our method of decellularization disrupts all the cell structure although a minimal amount of nuclear remains, which is less than the acquired amount (100 ng/mg) defined by the China Food and Drug Administration for the xenografts used in human tissues. Moreover, our compromise in complete cell removal leads to an optimal balance between the maintenance of ultrastructure of corneal collagen fibers and the maximal degree of decellularization.
The γ-ray irradiation is generally used as a method for sterilization of biological materials, but also effectively reduces the immunogenicity of the material 25 . γ-ray irradiation works by directly destroying DNA and causing indirect damage via reactive oxygen species (ROS) generation 26,27 . Recent studies have reported that the γ-irradiated sterile decellularized cornea can be effectively applied to lamellar keratoplasty [28][29][30] . The combined use of polyethylene glycol and γ-ray irradiation makes the maximal decelluarization effect, because neither porcine endogenous retroviral DNA sequence nor alpha-1,3 galactose epitope was detected after treatment 31 . Histological examination confirmed that cells were extensively removed and the DNA residual amount significantly decreased in cDPC, while the ultrastructure of the collagen fibers remained nearly unchanged and similar to that of the NPC group. These results demonstrate that the process of decelluarization did not disrupt the corneal stroma. Among  lamellar keratoplasty of cDPCs in rabbits, there was neither evidence of inflammatory cell infiltration, nor corneal vascularization.
Although decellularization of xenografts removes nearly all the cellular antigens and prevents immune rejection, the biomechanical strength and biostability of the material is often compromised. In order to improve the biomechanical strength and biostability of the decellularized porcine cornea, we employed a chemical process of crosslinking using EDC/NHS. In this crosslinking process, no new substances were introduced [32][33][34] , and the results of in vitro and in vivo cytotoxicity experiments have shown that the EDC/NHS cross-linked biosynthetic implants were not cytotoxic.
In order to verify that the crosslinking reaction occurred after EDC/NHS treatment, the thermal denaturation temperature of the collagen fibers was measured by DSC (Differential Scanning Calorimetry). A higher thermal denaturation temperature indicates a more stable material and more complete crosslinking 16 . Our results showed that the thermal stability of the cross-linked EDC/NHS sample was significantly higher than that of the un-crosslinked sample, which indicates that crosslinking effectively improved the physical properties of the cDPC.
Corneal swelling resulting from water imbibition affects its surface regularity, refractivity, surgical suture, cell migration, and tissue healing, while acellular tissue swells more easily than cellular tissue. However, our swelling experiments suggest that after crosslinking with EDC/NHS, the swelling rate of the cDPC was significantly lower than that in the NPC group. This suggests that crosslinking makes cDPC more resistant to edema than NPC.
The biomechanical properties of the cornea are directly related to the preservation and alignment of the matrix 35 . Many of the synthetic biomaterials previously reported as potential corneal substitutes failed to withstand interrupted sutures and stitch removal, and were not structurally strong enough. Previously reported acellular natural biomaterials also were not structurally sound after removal of cellular components 23 . The results of our study suggest that the cDPC underwent an increase in tensile strength that was significantly greater than the NPC group. However, we observed no significant difference in the elastic modulus between the cDPC and NPC groups. This difference may be caused by the crosslinking-induced connections formed between the collagen fibers. Such an effect, also explains the increased biomechanical strength of cDPC and the ability of the graft to withstand the surgical procedure.
Thus, our study suggest that EDC/NHS crosslinking is a crucially important step in the cDPC preparation process. However, excessive crosslinking could result in an increase in the rigidity and elastic modulus of the cDPC which may not comply with the clinical requirements. Therefore, the optimum EDC/NHS concentrations for crosslinking reactions should be further explored.
Xie et al. and others have established some animal models of fungal corneal infection 36 . We are the first to establish a rabbit model of fungal keratitis to evaluate the safety and efficacy of cDPCs in treating fungal keratitis. Our results demonstrate that the recurrent fungal infection rate in the cDPC group is significantly lower than that in the NPC group or the Natamycin group. Only one case of corneal perforation was observed in the cDPC group due to fungal infection recurrence. This resulted from primary disease recurrence after surgery of infectious keratitis, but was not a complication related to the keratoplasty. The NPC group had the lowest cure rate of fungal infections with four cases of perforation. The implantation of NPC without decelluarization led to early rejection due to intense local inflammation, eventually leading to corneal perforation. In the case of the Natamycin group, natamycin is a hydrophobic fat-soluble drug that has limited permeation in the cornea and has poor results for treating the fungal infection in superficial and deeper cornea. Eventually corneal fungal hyphae passed through the Descemet's membrane into the anterior chamber to cause perforation.
Our study also demonstrates that, in addition to controlling fungal keratitis, the corneal transparency of cDPC group was obviously improved compared to the other two groups. Corneal transparency is crucial for a good prognosis after keratoplasty and has a direct effect on patient regaining or maintaining normal visual function. It is important to note that while the Natamycin group had a higher cure rate of fungal infections than NPC group, but corneal transparency was very poor. Long-term observations showed that all the corneas of the Natamycin group developed vascular scars during healing that led to moderate to severe corneal opacity. This often results in loss of visual function and induction of neovascularization that seriously damages the immune privilege of the cornea. Due to high antigenicity, the grafts of the NPC group were rejected and led to moderate to severe opacity. In the cDPC group, the post-surgery cornea remained transparent due to very low antigenicity. Crucial to long-term survival of the cDPC grafts, proliferation of rabbit corneal stroma cells and restoration of corneal transparency occurred after 24 weeks.

Methods
Cornea graft preparation. Fresh porcine eyeballs were obtained from local abattoirs (Guangzhou Animal and Husbandry Corporation, Guangzhou, China) and rapidly cooled. The ocular globe was washed three times with phosphate buffered solution (PBS) containing antibiotics (penicillin, 60 mg/L; streptomycin, 100 mg/L; Invitrogen, Carlsbad, CA, USA). Corneas were divided randomly into two groups: native porcine cornea (NPC) group and cDPC group. The NPC group consisted of fresh porcine corneas without further treatments. The cDPC group consisted of porcine corneas prepared by decelluarization, crosslinking and irradiation processes, as described below.
The prepared ocular globe was immersed in a gradient glycerol solutions with sequential concentrations of 95%, 70%, 50%, and 90%. The ocular globe was placed in each of these glycerol solutions for four hours at 4 °C with shaking, in order to ensure decellularization. The ocular globes were then washed with a preservation solution (500 g/L glycerol, 10 g/L hyaluronic acid, buffered salt solution).
Crosslinking was achieved using a glycerol buffered solution containing NHS and EDC (at a ratio of 1:2). The ocular globes were treated with this solution for two hours and washed with the preservation solution described above.
ScieNtiFic REPORTS | 7: 9955 | DOI:10.1038/s41598-017-08207-3 The anterior lamellar stroma was cut into a graft (250 μm in thickness and 10 mm in diameter from the eyeball) with a microkeratome (AMADEUS, Ziemer Corporation, Switzerland), Grafts were sealed in sterile plastic containers containing the preservative solution, irradiated with γ rays at 25 KGy, and stored at −20 °C or less until being used. All procedures were conducted under sterile conditions. Light transmittance assay. The cDPC and NPC grafts were immobilized on black corneal holders and placed in a spectrophotometer (U-4100, Shimadzu Ltd., Kyoto, Japan) for measuring transmittance. Transmittance was measured in the wavelength range from 280 nm to 780 nm at a scanning interval of 10 nm 7 .
Differential scanning calorimetry. Differential scanning calorimetry (DSC Q2000, TA instruments Ltd., New Castle, USA) was used to obtain thermal properties of corneal grafts. Grafts of the cDPC and NPC groups were freeze-dried before determination. Then the samples of the cDPC and NPC grafts were heated at temperature ranging from room temperature to 200 °C (with an increasing rate of 10 °C /min) with a nitrogen flow rate of 50 mL/min, by setting an empty pan as a reference. Denaturation temperature (TD) was determined as the peak value of the corresponding endothermic phenomena. Tensile strength testing. The mechanical properties of the cDPC and NPC grafts were assessed via tensile testing (n = 10/group). To accomplish this purpose, grafts were cut into rectangular specimens with a gauge length of 9 mm and width of 2.5 mm, the thickness of samples was determined by a handheld thickness gauge (MZ2041, Jiangsu Mingzhu Testing Machinery Ltd., Jiangsu, China). The tensile strength was measured at a loading rate of 0.1mm/s by using an ElectroForce Test Instrument (ElectroForce 3200, Bose Corporation, USA) at the room temperature 38 . Force was applied until fracture occurred in the Test Instrument.
Cytotoxicity of cDPCs. The cDPC(n = 9) and NPC(n = 9) grafts were extracted at 37   Surgery. The 36 rabbits infected with Fusarium solani were divided into three groups randomly (n = 12 per group). The first group (cDPC group) was treated with cDPC lamellar keratoplasty. The second group (NPC group) was treated with NPC lamellar keratoplasty. The third group (Natamycin group) did not receive keratoplasty but was treated with natamycin eye drops (Shenyang Sinqi Pharmaceutical Corporation, Shenyang, China) four times per day. The rabbits in the cDPC group and the NPC group were anesthetized and sterilized. The eyelids were kept open with an eye speculum, and the upper and lower rectus muscles were pulled and fixed with a suture. In the central area of the rabbit cornea, a 7.5 mm trephine was used to prepare a 7.5 mm-diameter circular lamellar receipt bed. The diameter of this region was greater than the range of the infected ulcer focus and the depth of which was about 2/3 rd of the corneal thickness. After the infection focus was removed completely, a 7.75 mm-diameter circular graft with a thickness of about 250 μm was cut from each of the prepared cDPC and NPC groups (12 per group) by a corneal trephine, with the epithelial surface facing outward. 12-16 stitches of in situ interrupted suture were placed with the tightness incision of the suture perfectly aligned and the knot buried in the cornea. Ofloxacin ophthalmic ointment (North China Pharmaceutical Group Corporation, Shijiazhuang, China) was applied post-surgery. After surgery eyes were treated with eye drops for four weeks, tacrolimus solution (Senju Pharmaceutical Corporation, Osaka, Japan) and natamycin four times a day, while ofloxacin ophthalmic ointment twice a day.
Routine examination and histological analysis. Transplanted grafts were examined under a slit lamp at the 1 week, 4 weeks and 24 weeks after surgery. At 24 weeks post-surgery, all rabbits were sacrificed by euthanasia and eyeballs were collected for histopathological analysis.
Statistical analysis. Statistical analysis was performed using SPSS18.0 software. All data were expressed as Mean ± SD and analyzed for statistical differences using a one-way Analysis of Variance (ANOVA) test (p < 0.05 was considered a significant difference).
Data availability. The datasets generated and analysed during the current study are available from Dr. Wang on reasonable request.

Conclusion
In conclusion, the introduced novel method of preparing a decellularized corneal graft (cDPC) by decellularization, chemical crosslinking and γ irradiation produces the cDPC graft which has high transparency, good mechanical properties, long-term stability, low immunogenicity, and good biocompatibility. The study demonstrates that in a rabbit model of fungal keratitis, the cDPC can effectively control the infection while recovering the original transparency of the cornea. The cDPC may be an potential substitute for DALK in patients with fungal keratitis.