An experimental pig model with outer retinal degeneration induced by temporary intravitreal loading of N-methyl-N-nitrosourea during vitrectomy

We aimed to develop an outer retinal degeneration pig model induced by temporary intravitreal loading of N-methyl-N-nitrosourea (MNU) during vitrectomy. In a preliminary experiment involving 5 mini-pig cases to determine the appropriate concentration of MNU, the vitreous cavity of each eye was filled with 4, 8, 10, 12, or 16 mg/mL MNU for 10 min, which was then replaced with a balanced salt solution. Multimodal examinations including spectral-domain optical coherence tomography (OCT) images and full-field electroretinography (ffERG) were obtained at baseline and week 2, week 6, and week 12. The retinal degeneration was classified according to the amplitudes of a dark adaptive (DA) 10.0 a-wave amplitude. The degree of moderate retinal degeneration was defined as DA 10.0 a-wave amplitude ≥ 10% and < 60% of baseline amplitude. The degree of severe degeneration was defined as DA 10.0 a-wave amplitude < 10% of baseline amplitude, noise, or flat signal. Hematoxylin and eosin staining and immunohistochemistry were performed at week 12. The main experiments were conducted first with 10 cases of 5 mg/mL and later with 13 cases of 10 mg/mL. In the preliminary experiment, degree of outer retinal degeneration increased with MNU concentration. Use of 4, 8, 10, 12, and 16 mg/mL MNU showed no, moderate, severe, severe, and atrophic changes, respectively. In the main experiments, there were 9 cases of moderate retinal degeneration and 1 case of severe degeneration in 5 mg/mL MNU group. Two cases of moderate degeneration and 11 of severe degeneration were recorded in 10 mg/mL group. Mean thickness of total retina, inner nuclear layer, and outer nuclear layer decreased at week 2 in both groups. The mean amplitudes on ffERG decreased at week 2. The ffERG and OCT findings did not change from week 2 to week 6 or week 12. The results of staining supported those of ffERG and OCT. Temporal MNU loading in a vitrectomized pig-eye model induced customized outer retinal degeneration with changing the concentration of MNU.

Fundus infrared reflectance and spectral-domain optical coherence tomography. The 4 mg/mL MNU case showed that the 3 nuclear layers of ganglion cell layer (GCL), ONL, and INL were well maintained at week 2. The EZ and RPE layers were also intact in the 4 mg/mL MNU case. In the 8 mg/mL MNU case, the boundary of INL and ONL became indistinct at week 2, and the EZ became obscure. Both the 10 mg/mL and 12 mg/mL MNU cases showed indistinguishable EZ and ONL at week 2. However, the 12 mg/mL MNU case showed a decrease of INL thickness more than the 10 mg/mL MNU case at week 2 compared with baseline. These trends remained stable until weeks 6 and 12 in the 4 mg/mL, 8 mg/mL, 10 mg/mL and 12 mg/mL MNU cases. In the 16 mg/mL MNU case, the layers became atrophic and the focal retina melted, causing focal retinal detachment at week 2 ( Fig. 2, Supplementary Fig. 2).
Hematoxylin and eosin and immunohistochemistry. The 4 mg/mL MNU case showed that all INL, GCL, and nerve fiber layer (NFL) were intact. Photoreceptors were nearly intact in H&E staining (Fig. 3). For Immunohistochemistry, Rhodopsin staining, and PNA staining showed intact cone and rod cells. RPE65 staining showed intact RPE and co-staining of photoreceptors. NeuN staining showed intact ganglion cells and co-staining of photoreceptors. PKC-α staining showed intact bipolar cells, and GFAP staining showed nearly normal Müller cells (Fig. 4).
The density of photoreceptors decreased in H&E staining of the 8 mg/mL MNU sample (Fig. 3). The densities of ONL and INL were also reduced, while GCL was relatively intact. Rhodopsin staining and PNA staining showed few remnant rods and cones, respectively. RPE65 and PKC-α staining showed nearly intact RPE cells and bipolar cells. NeuN staining showed intact GCL but co-staining of few photoreceptors. GFAP staining showed mildly increased staining (Fig. 4).
The 10 mg/mL MNU case revealed few ONL cells among mainly INL cells in H&E staining (Fig. 3). Rhodopsin and PNA staining showed rare rod or cone cells with co-staining of RPE cells. RPE 65, PKC-α, and NeuN staining showed intact RPE cells, bipolar cells, and ganglion cells, respectively. No co-staining of photoreceptors was evident in NeuN staining. GFAP staining showed increased staining (Fig. 4).
The 12 mg/mL MNU case revealed no ONL cells among mainly INL cells in H&E staining (Fig. 3). Rhodopsin and PNA staining showed no rod or cone cells with co-staining of RPE cells. RPE 65, PKC-α, and NeuN staining showed intact RPE cells, bipolar cells, and ganglion cells, respectively. No co-staining of photoreceptors was evident in NeuN staining. GFAP staining showed increased staining (Fig. 4).
The 16 mg/mL MNU case revealed proliferative vitreoretinopathy and retinal detachment, and all layers were extremely thin in H&E staining (Fig. 3). In immunohistochemistry, RPE staining showed nearly intact RPE cells (Fig. 4).
Efficacy experiment results of the 5 mg/mL MNU group (n = 10). Full-field electroretinography and multifocal electroretinography. The 5 mg/mL MNU group included 9 cases of moderate degeneration and 1 case of severe degeneration. The mfERG of moderate degeneration cases produced severely decreased amplitudes and focal noise signals in the ffERG field. The delay of b-wave in DA 0.01 ffERG was not statistically significant respectively, in the 8 mg/mL MNU case. The amplitudes in scotopic and photopic ffERG decreased at week 2, but no signal was flat or appeared as noise, especially in photopic ffERG. (d) mfERG in the 8 mg/mL MNU case showed focal noise signal and signal flattening at week 2. (e) The DA 10.0 a-wave amplitude at baseline and week 2 was 38.6 µV and 6.21 µV, respectively, in a 10 mg/mL MNU case. All signals of ffERG showed nearly flat down signals or noise signals at week 2. (f) The mfERG in a 10 mg/mL MNU case showed only noise signals at week 2. (g) The DA 10.0 a-wave amplitude at baseline and week 2 was 61.3 µV and 5.67 µV, respectively, in a 12 mg/mL MNU case. All signals of ffERG showed nearly flat down signals or noise signals at week 2. (h) The mfERG in a 12 mg/mL MNU case showed only noise signals at week 2. (i,j) Both ffERGs and mfERGs in a 16 mg/mL MNU case showed noise signals at week 2. ffERG full field electroretinogram, MNU N-methyl-N-nitrosourea, DA dark adaptive, mfERG multifocal electroretinogram. www.nature.com/scientificreports/ at week 2 and at week 6. The mean amplitude of b-waves in the DA 0.01 ffERG showed a significant decrease at week 2, but no difference was evident between weeks 2 and 6. The mean implicit time in DA 10.0 ffERG showed a temporal delay at week 2. The mean amplitude in the DA 10.0 ffERG decreased at week 2 and then remained low from week 2 to week 6. The mean implicit time in LA 3.0 ffERG was not changed at weeks 2 and 6. The mean amplitude in LA 3.0 ffERG showed a decrease at week 2 and remained low at week 6. The mean amplitude of LA 3.0 flicker showed a decrease at week 2 and remained low at week 6 ( Fig. 5, Table 1).
Fundus infrared reflectance and spectral-domain optical coherence tomography. SD-OCT images at week 2 showed intact GCL and RPE. The ellipsoid zone on 55-degree SD-OCT fields was barely distinguishable at week 2 ( Fig. 6), and outer plexiform layer thickness decreased. From week 2 to week 6, subsequent changes were minimal and similar among layers. Mean thicknesses of INL, ONL, and TRL showed significant decrease at week 2. From week 2 to week 6, changes were minimal (Fig. 7, Table 2).
Hematoxylin and eosin and immunohistochemistry. In H&E staining of moderate degeneration cases (n = 9), the INL, GCL, and NFL were nearly intact (Fig. 8). Rhodopsin staining showed severely decreased rod cells, and PNA staining showed focal intact cone cells. RPE65 staining showed intact RPE and rare co-staining of photoreceptors. NeuN staining showed intact ganglion cells and co-staining of few photoreceptors. PKC-α staining showed intact bipolar cells, and GFAP staining showed nearly normal Müller cells. No apoptotic cells were evident in Tunel staining (Fig. 9).    www.nature.com/scientificreports/ 2 to week 12. The amplitude of LA 3.0 flicker showed a decrease at week 2 and remained decreased from week 2 to week 12. In the mfERG of all cases, only noise signals were observed at each follow-up period (Fig. 5, Table 3).

Fundus infrared reflectance and spectral-domain optical coherence tomography.
On SD-OCT, the EZ was indistinguishable at week 2. The thicknesses of INL, ONL, and TRL appeared to decrease at week 2 ( Fig. 6). Further losses of INL and ONL were not detected at week 6 or 12. In addition, there was decrease in mean thicknesses of INL, ONL, and TRL. From week 2 to week 12, each mean thickness showed no subsequent changes (Fig. 7, Table 2).
Hematoxylin and eosin and immunohistochemistry. In H&E staining of severe degeneration cases (n = 11), ONL cells were rarely observed among mainly INL (Fig. 8). Rhodopsin and PNA staining showed no rod or cone cells with co-staining of RPE cells. RPE 65, PKC-α, and NeuN staining showed intact RPE cells, bipolar cells, and ganglion cells, respectively. There was no co-staining of photoreceptors in the NeuN stain. GFAP staining showed increased staining, especially in the outer retina, and Tunel staining showed no stained lesions (Fig. 9).

Discussion
Our study used a mini-pig model to evaluate the effect of MNU on photoreceptors because the size, vascular anatomy, vascular function, and immunology of the pig eye are comparable to those of a human eye [25][26][27][28] .
Although the pig retina does not have a fovea, a so-called visual streak of high cone density (rod:cone ratio  33 . Because of its similarity with human, the pig retina might be considered as enough model for glaucoma studies. One study showed that chronically increased intraocular pressure could induce retinal ganglion cell death significantly in the mid-peripheral and peripheral retina, in which the temporal quadrants were mainly damaged 34 . Although rodent eyes could serve as an alternative to pig retinas in spite of their small size because mice experience similar retinal vascular development and retinal histological structures, the rodent retina is predominantly composed of rod cells (rod:cone ratio = approximately 98:2) 35 . A rabbit's eyeball size is similar to that of humans, but rabbit retinal vascular development differs from that of the human eye (melanotic vs. holangiotic) 36,37 . Numerous studies have employed pigs for large-animal models in the study of retinal diseases and therapeutic interventions 38,39 .
We analyzed the effectiveness of temporal MNU tamponade (exposure) with vitrectomy to induce unilateral diffuse outer retinal degeneration in mini-pig. In a previous experiment, globally diffuse outer retinal degeneration using intravitreal MNU injection was not obtained even after vitrectomy. Instead, localized outer retinal degeneration was found in the gravitational direction after intravitreal MNU injection in vitrectomized eyes 23 . The water solubility of MNU is 14.4 g/L (14.4 mg/mL) at 24 °C, but a small amount of intravitreal MNU injection (0.05 mL) appeared to be localized on the retinal surface by gravity. Because MNU is sensitive to moisture and light and hydrolyzes in water with a half-life of 1.2 h at pH 7 at 20 °C [40][41][42] , most effects on the retina were thought to take place rapidly. This may explain why small intravitreal MNU injections have a spatially localized effect on the retina. To avoid this problem, we temporarily filled a whole vitreous cavity with MNU solution diluted in BSS. The MNU tamponade was performed for 10 min, which was sufficient for retina whitening (Supplementary Video 1). After 10 min, as much as possible of the MNU solution was removed by 2 or 3 cycles of air-fluid exchange, and BSS tamponade was performed in the vitreous cavity at the end of surgery.
Degree of degeneration varied according to MNU concentration. In a preliminary experiment, degree of outer retinal degeneration of the retina increased with MNU concentration. Temporal tamponade of 5 mg/mL MNU and 10 mg/mL MNU during vitrectomy was effective to induce respective moderate and severe outer retinal degeneration without retinal atrophic change. Zulliger et al. showed retinal changes to be MNU concentrationdependent, a relationship that was most obvious with the ONL 43 . Other reports showed a dose-dependent response over the threshold dose as a carcinogen 44,45 .
Two moderate retinal degeneration cases in the 10 mg MNU group showed severe outer retinal change in most of the area except the far peripheral retina, where the retinal structure was intact. That may explain why moderate retinal degeneration cases presented diminished amplitudes in ffERG despite ONL layer loss in the 55-degree SD-OCT field. Remnant peripheral vitreous fluid may block adequate exposure of MNU to the far peripheral retina, which could be predicted after MNU solution tamponade during vitrectomy. After 10 min of MNU solution tamponade, the color of the degenerated retina changed to white; this color change was not evident in the non-degenerated far periphery during vitrectomy (Supplementary Video 1). This suggests that posterior vitreous detachment at the far periphery is important to ensure sufficient application of MNU to the whole retinal area.
Most degenerative changes were observed at week 2, and no further progress was observed during the study period. Our immunochemistry results showed rare apoptotic cells in Tunnel stain at week 6 or week 12. One study using intraperitoneal MNU injection reported that MNU induces peak photoreceptor cell loss 3 days www.nature.com/scientificreports/ post-injection during a 7-day follow-up period 46 . Numerous bone marrow-derived cells migrated to the retina and differentiated into immunoreactive microglia in response to photoreceptor degeneration within 7 days of MNU injection 47 . In the apoptosis cascade, down-regulation of Bcl-2 and up-regulation of Bax were seen at 12 h, and caspase-family activity peaked 72 h after MNU treatment 48 . Our study has some limitations. First, only 6 weeks of follow-up were conducted in the 5 mg/mL MNU groups, and the degree of outer retinal degeneration could worsen after 6 weeks. There were no significant changes from week 2 to week 6 in the 5 mg/mL MNU group. No significant changes were also observed from week 2 to week 12 in the 10 mg/mL MNU group. Furthermore, there were rare apoptotic cells of Tunnel stain in both groups at week 6 or 12. Because no evidence of retinal degeneration progress was observed in the preliminary study, the relatively short follow-up period for the 5 mg/mL MNU group compared with the 10 mg/ mL group did not appear to affect the results significantly. Second, the unstable chemical properties of MNU in water may have compromised the reliability and consistency of the results.
In conclusion, temporal exposure of MNU induced outer retinal degeneration in vitrectomized pig eyes. The degree of retinal degeneration was modulated by adjusting MNU concentration. The MNU-induced pig model used in this study is a candidate for further research.

Materials and methods
Animals. All procedures adhered to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. Approval for this study was obtained from the Institutional Animal Care and Use Committee of Korea University College of Medicine (KOREA-2018-0002-C1).
In the preliminary experiment, a total of 5 eyes from 5 female mini-pigs (MICROPIG, APURES Co., Ltd, Pyeongtaek-si, Korea) were used to determine degree of retinal change according to MNU concentration (4, 8, 10, 12, and 16 mg/mL). Each concentration was tested in 1 mini-pig. To identify morphological changes in the retina, fundus infrared reflectance (IR) images, 55-degree spectral-domain optical coherence tomography (OCT), full-field electroretinography (ffERG), and multifocal ERG (mfERG) were performed at baseline and 2, 6, and 12 weeks after surgery. Histological examinations using hematoxylin and eosin (H&E) and immunohistochemistry staining were performed on selected mini-pig eyes 12 weeks after surgery. In the second efficacy experiments, which were based on the first experiment, the effect of 5 mg/mL MNU (n = 10) was evaluated in the mini-pigs following the protocol used in the preliminary experiment, with the exception of week 12 follow-up examinations. Further study with 10 mg/mL MNU (n = 13) followed the same protocol with week 12 follow-up examinations.
Vitrectomy with MNU loading. The mini-pigs were anesthetized by intravenous injection of alfaxalone (1 mg/kg; Alfaxan, Vetoquinol, West Sussex, UK) into the marginal auricular vein following premedication, which comprised a subcutaneous injection of atropine (0.05 mg/kg), intramuscular injection of xylazine (1 mg/ kg; Rompun, Bayer Corp., Shawnee Mission, KA, USA), and azaperone (4 mg/kg; Stresnil, Mallinckrodt Veterinary Inc, Indianapolis, IN, USA). After general anesthesia, the eye was irrigated with 5% povidone-iodine and draped for surgery. Three-port, 23-gauge vitrectomy (Associate; DORC, Zuidland, Netherlands) was performed with an indirect lens (Oculus BIOM ready, Oculus Surgical, Inc., FL, USA). Three ports were prepared by inserting trocar cannulas into the sclera 3 mm from the limbus at ventromedial, ventrolateral, and dorsomedial sides. The vitreous was removed using a vitreous cutter, while continually supplying balanced salt solution (BSS; Alcon, Fort Worth, TX). After core vitrectomy, posterior vitreous detachment was induced gently to avoid an iatrogenic retinal break. Both peripheral vitrectomy and lensectomy were performed. Air-fluid exchange was Figure 5. The ffERG results of the 5 mg/mL MNU group (n = 10) and the 10 mg/mL MNU group (n = 13). (a-g) The ffERG of the 5 mg/mL MNU group. (a,b) The 5 mg/mL MNU group showed delayed implicit time of b-waves in dark adaptive (DA) 0.01 ERG at weeks 2 and and 6, but the change was not statistically significant. The 5 mg/mL MNU group showed significant decrease of b-wave in dark adaptive (DA) 0.01 ERG from baseline to week 2 (p = 0.028). (c,d) In DA 10.0 ERG, the 5 mg/mL MNU group showed delay of a-and b-waves at week 2 (p = 0.028, 0.018, respectively) and appeared to recover somewhat at week 6, but the amplitudes of both a-and b-waves decreased at week 2 (p = 0.005, 0.005, respectively) and remained unchanged from week 2 to week 6.
(e,f) In light adaptive (LA) 3.0 ERG, the 5 mg/mL MNU group showed decrease in a-and b-wave amplitudes at week 2 (p = 0.018, 0.018, respectively) and remained unchanged at week 6, but the implicit time of both aand b-waves did not change during follow-up periods. (g) The flicker amplitude in the 5 mg/mL MNU group showed a decrease in amplitude at week 2 (p = 0.028), but further decrease was not statistically significant. (h-n) The ffERG results of the 10 mg/mL MNU group. (h,i) The b implicit time in the DA 0.01 ERG did not show a significant change in the 10 mg/mL MNU group, but the amplitude was nearly flattened at week 2 (p = 0.008) and showed no recovery or variation after. (j,k) The b implicit time in the DA 10 ERG show a significant change in the 10 mg/mL MNU group (p = 0.008, 0.008, respectively), and the amplitude nearly flattened at week 2 (p = 0.008, 0.008, respectively) and showed no recovery or variation after that point. (l,m) In LA 3.0 ERG, the 10 mg/mL MNU group showed delay of a-and b-waves at week 2 (p = 0.008, 0.008, respectively), and showed decreased amplitudes of the a-and b wave at week 2 (p = 0.008, 0.021, respectively), and there was no significant change at week 6 or 12 compared with the results of week 2. (n) The flicker amplitude in the 10 mg/mL MNU group showed a decrease in amplitude at week 2 (p = 0.008) and then remained decreased. ffERG full field electroretinogram, MNU N-methyl-N-nitrosourea, DA dark adaptive, LA light adaptive. Full-field electroretinography and multifocal electroretinography. The ffERG protocol was based on the international standard for electroretinography from the International Society for Clinical Electrophysiology of Vision (ISCEV) 49,50 . The mini-pigs were anesthetized as described above and dark-adapted for 30 min, after which the pupils were dilated. Light stimulation and ffERG signal recording were performed with a commercial system (RETIcom; Ronald Consult, Germany), using a contact lens electrode with a built-in light resource (Kooijman/Damhof ERG lens; Medical Workshop BV, Netherlands). The reference and ground electrodes were subdermal platinum needle electrodes. Reference electrode was placed in the skin near the lateral canthus of the eyes, and a ground electrode was placed on the forehead between the two eyes. The mfERG protocol was based on the ISCEV standard 50,51 . Light stimulation at a distance of 26 cm from a 19-inch liquid crystal display monitor and mfERG signal recording used a commercial system (RETIpot; Ronald Consult, Germany). The contact lens electrode used the same instruments as in ffERG. The retinal degeneration was classified according to the amplitudes of a dark adaptive (DA) 10.0 a-wave amplitude. The degree of moderate outer retinal degeneration was defined as DA 10.0 a-wave amplitude between ≥ 10% and < 60% of baseline amplitude. The degree of severe degeneration was defined as DA 10.0 a-wave amplitude < 10% of baseline amplitude including a flat signal or noise with no a or b wave on DA 0.01, 10, light adaptive (LA) 3.0 ffERG.

Fundus infrared reflectance and spectral-domain optical coherence tomography. A-scan
biometry (SW-1000, Suoer, China) was obtained to measure the axial length of the eyeball at baseline. Images were obtained with 55-degree fundus infrared reflectance (IR) and spectral-domain optical coherence tomography (SD-OCT) images using the Spectralis OCT system (Heidelberg Engineering GmbH, Heidelberg, Germany). Vertical and horizontal line scans, as well as raster scans (33 B-scans over a 16.5 × 16.5-mm area in a 55-degree image), were performed at high resolution (1536 A-scans per B-scan, lateral resolution = 10 µm/pixel in 55-degree image). Up to 100 images were averaged in automatic real-time mode to obtain a high-quality mean image. Total retinal layer (TRL) thickness was measured along a horizontal line perpendicular to the retinal layers in cross-sectional images. Inner nuclear layer (INL) thickness and outer nuclear layer (ONL) thickness were measured at 10 different areas, along with a horizontal visual streak 2 mm dorsal to the optic disc (Fig. 10). TRL thickness was defined as the distance between the inner margin of the internal limiting membrane to the inner margin of the retinal pigment epithelium (RPE) layer. INL thickness was defined as the distance between the outer margin of the inner plexiform layer to the inner margin of the outer plexiform layer. ONL thickness was defined as the distance between the outer margin of the outer plexiform layer to the inner margin of the ellipsoid zone (EZ) 52 . www.nature.com/scientificreports/  Immunohistochemistry. Tissue sections were deparaffinized, rehydrated, and microwave-heated in antigen retrieval buffer (tris-ethylenediaminetetraacetic acid solution, pH 9.0). Sections were then blocked with 4% horse serum in phosphate-buffered saline with Tween, followed by incubation with each primary antibody at     Table 4. Summary of immunohistochemistry for pig retinal degeneration model. PKC-α Protein kinase-α, GFAP Glial fibrillary acidic protein, NeuN Neuron-specific nuclear protein, PNA peanut agglutin, RPE retinal pigment epithelium, IgG Immunoglobulin G, ELISA Enzyme-linked immunosorbent assay.

Sequence of antibody (primary/secondary) Dilution Industry Targeted cells or layers
Anti-PKC-α