Liposome–trimethyl chitosan nanoparticles codeliver insulin and siVEGF to treat corneal alkali burns by inhibiting ferroptosis

Abstract Alkali burns are potentially blinding corneal injuries. Due to the lack of available effective therapies, the prognosis is poor. Thus, effective treatment methods for corneal alkali burns are urgently needed. Codelivery nanoparticles (NPs) with characteristics such as high bioavailability and few side effects have been considered effective therapeutic agents for ocular diseases. In this study, we designed a new combination therapy using liposomes and trimethyl chitosan (TMC) for the codelivery of insulin (INS) and vascular endothelial growth factor small interfering RNA (siVEGF) to treat alkali‐burned corneas. We describe the preparation and characterization of siVEGF‐TMC‐INS‐liposome (siVEGF‐TIL), drug release characteristics, intraocular tracing, pharmacodynamics, and biosafety. We found that siVEGF‐TIL could inhibit oxidative stress, inflammation, and the expression of VEGF in vitro and effectively maintained corneal transparency, accelerated epithelialization, and inhibited corneal neovascularization (CNV) in vivo. Morever, we found that the therapeutic mechanism of siVEGF‐TIL is possibly relevant to the inhibition of the ferroptosis signaling pathway by metabolomic analysis. In general, siVEGF‐TIL NPs could be a safe and effective therapy for corneal alkali burn.


| INTRODUCTION
Corneal alkali burns are one of the most common emergencies in ophthalmology, accounting for 11.5%-22.1% of all ocular traumas. 1 As a result of corneal alkali injury, the ocular surface and anterior eye segment are extensively damaged, causing permanent vision impairment or even complete blindness. 2 It has been reported that corneal oxidative stress occurs immediately after alkali damage, which precedes the corneal inflammatory response. 3 During alkali burn-induced injury, excessive oxidative stress in the cornea, oxidative changes occur in cellular macromolecules, and lipid peroxidation occurs in the membrane, 4 leading to an antioxidant/pro-oxidant imbalance in corneal tissues. On the other hand, the activity of antioxidant enzymes is decreased, while the expression and activity of catalytic enzymes running at physiological levels or even increases, leading to an increase in reactive oxygen species (ROS) production and a decrease in ROS decomposition. 5 These factors can cause a high level of oxidative stress, eventually resulting in excessive intracorneal inflammation, scarring, and corneal neovascularization (CNV). 6 Similar to oxidative stress, CNV plays a critical role in the pathophysiology of corneal alkali burns. CNV increases vascular permeability, which exacerbates inflammation, chronic edema, lipid exudation, and corneal scarring, potentially resulting in permanent vision loss. 7 Currently, topical corticosteroids and nonsteroidal anti-inflammatory drugs (NSAIDs) remain the top priorities. However, these treatments can delay wound healing, and long-term use of corticosteroids can lead to increased intraocular pressure (IOP), cataracts, and an increased risk of infection. 8 Even though various other treatment options have been available in the clinic, such as amniotic membrane transplantation, their effectiveness has not been optimal in the past two decades. 9 Consequently, it is urgent to explore a more efficient and safe treatment for severe corneal alkali burns. Cruz-Cazarim et al. 13 showed that INS could treat dry eye syndrome and corneal injuries. Morever, a recent clinical study suggested that topical INS was an effective way to safely promote the healing of persistent epithelial defects in patients who were unresponsive to standard treatment. 14 However, the use of INS in the treatment of corneal alkali burns has been rarely reported. We hypothesize that the local application of INS may be a promising strategy for the treatment of corneal alkali burns due to its antioxidant capacity.
In addition to inhibiting oxidative stress, the treatment of CNV is also essential for corneal alkali burns. CNV can be effectively treated by inhibiting vascular endothelial growth factor (VEGF) and its receptors, which modulate angiogenesis. 15 Anti-VEGF antibodies are therefore used to treat CNV, either through topical or subconjunctival applications. 16 Unfortunately, anti-VEGF antibodies are generally limited due to their poor efficacy, side effects, and drug resistance. 7 RNA interference is a powerful approach to knocking down target genes. 17 VEGF small interfering RNA (siVEGF) reduces VEGF expression and CNV. [18][19][20][21] Accordingly, it is reasonable to speculate that siVEGF could enhance the therapeutic effects of INS on corneal alkali injury, and codelivery of INS and siVEGF may provide a new combination therapy for corneal alkali injury.
It is well-established that corneal physiological and physical barriers impair drug and siRNA penetration. To improve bioavailability, nanopharmaceuticals have been extensively developed to deliver siRNA or ocular drugs to treat ocular diseases. [22][23][24] A wealth of studies have suggested that INS-loaded liposomes could increase the bioavailability of INS. 25,26 However, the poor stability of liposomes leads to the rapid release of the encapsulated drugs, which impairs the therapeutic effects of drugs. Chitosan (CS) is a deacetylated pyran polysaccharide isolated from chitin that is biocompatible, nontoxic, and biodegradable and has been widely used to prepare nanocarriers such as micelles and nanoparticles (NPs). CS can also be used as a coating for liposomes to improve their stability in vitro and in vivo. 27 Furthermore, CS can form NPs and be loaded with negatively charged nucleic acids and have been considered promising carriers for gene delivery. 28 Morever, this encapsulation protects nucleic acids from host nucleases. 29 As a quaternary CS derivative, trimethyl chitosan (TMC) also possesses these properties, and it is preferred due to its high water solubility, ionic stability, and cationic density. 30 In this study, we developed a novel eye drop formulation based on liposomes and TMC to encapsulate and deliver INS and siVEGF.
We expect these NPs to enhance the treatment efficacy of corneal alkali burns through the cooperative effects of INS and siVEGF, as well as their enduring effects and high bioavailability. We also explored the potential mechanism of the siVEGF-TMC-INS-liposome (siVEGF-TIL) NPs in treating corneal alkali burn.
The mixture was ultrasonicated in a water bath to form w/o emulsion and then transferred into a 100 mL round-bottomed flask, which was subsequently evaporated under reduced pressure with a rotating speed of 50 rpm at 30 C for 3 h to remove the organic solvent. Afterward, 4 mL citric acid-Na 2 HPO 4 buffer (pH 5.6) was added to hydrate the films until a homogeneous dispersion and this mixture was transferred to a 10 mL EP tube. Then, 0.5 mL PFOB was added to the mixture, which was sonicated (55 W, four 3 min) with a sonicator (Sonics & Materials Inc.) in an ice bath, and then centrifugation was performed at 6000 rpm/min for 5 min. Supernatants were removed and sediments were collected and resuspended by phosphatebuffered saline (PBS; pH 7.4), then stored at 4 C for further use. Subsequently, an aliquot of INS-lip was mixed with the same volume of TMC (0.5 mg/mL) solution in PBS and then shaken and incubated at 4 C for 1 h to prepare TMC-INS-lip (TIL). Finally, siVEGF was loaded by electrostatic adsorption with an optimal ratio to obtain siVEGF-TMC-INS-lip (siVEGF-TIL). siVEGF-TMC-lip (siVEGF-TL) was made using the same protocol but without INS. Similarly, Empty-TMC-lip (TL) was also prepared using the same protocol but with the omission of INS and siVEGF.

| Characterization of NPs
The morphology NPs was observed by light microscope transmission electron microscope (TEM) (Hitachi H-7600). The particle size and zeta potential were measured using a laser particle size analyzer system (Nano, ZS90; Malvern Instrument Ltd   were calculated by the 2 ÀΔΔ C t method using GAPDH as a control.

| Cellular uptake of NPs
Each gene was analyzed in triplicate to reduce randomization error.

| CCK-8 assays
HCECs were plated into a 96-well plate at the density of 5 Â After being cultured for 24 h, the cells were washed thrice with PBS, and then freshly prepared CCK-8 solutions were added to each well.
The CCK-8 was used to detect cell viability in vitro according to the manufacturer's instructions for CCK-8. Absorbance at 450 nm was measured by a microplate reader (BioTek Instruments Inc.).

| In vitro inhibition of oxidative stress, inflammation, and neovascularization by NPs
To investigate the antioxidant stress, anti-neovascularization, and anti-inflammatory capacity in vitro, HCECs cells were exposed to immediately. [31][32][33] The depth of corneal injury was involving corneal epithelium and superficial stroma which was confirmed by H&E ( Figure S1).

| Clinical evaluations
After alkali burn, the SD rats were randomized into six groups (PBS,   siVEGF-TL, INS, INS-lip, TIL, siVEGF-TIL). Five microliters different reagents were dropped into the right eye twice a day respectively. No treatment for left eyes. To observe the degree of corneal opacity, corneal epithelial repair, and CNV, alkali-burned corneas were examined by portable slit lamps before and after fluorescein sodium staining every day and photographed on Days 1, 3, 7, and 14. The IOP was measured using a handheld tonometer (iLab tonometer; iCare). Corneal opacity was scored using a scale of 0-4 (Grade 0 = completely clear; Grade 1 = slightly hazy, iris and pupils easily visible; Grade 2 = slightly opaque, iris and pupils still detectable; Grade 3 = opaque, pupils hardly detectable; and Grade 4 = completely opaque with no view of the pupils). The corneal epithelial healing rate was calculated according to the following formula (k represents the corneal epithelial healing rate, S 0 represents the 0-day staining area, and S t represents the observed staining area): For CNV, the total corneal area and vessel area were manually selected with ImageJ. The CNV area was presented as the percentage with the following formula:

| Antioxidant stress and anti-inflammatory activity in vivo
Corneas of each group at 14 days were harvested and the levels of SOD, GSH, and MDA were quantified by commercial kits according to the instructions. Also, the corneal tissues were collected and homogenized with RIPA lysate, followed by centrifugation at 15,000 rpm for 15 min. Then, the levels of Glu, TNF-α, IL-6, and MMP-9 in the supernatant were detected by commercial ELISA kits according to the procedure provided by the manufacturer.

| Histological and immunohistochemical analysis
At 14 days after treatment with different reagents, the normal and alkali-burned corneas were enucleated for histological and immunohistochemical analysis, fixed in 10% buffered formalin, and successively dehydrated in a series of concentrations of ethanol and dimethylbenzene. Afterward, the treated tissues were fixed in paraffin, and tissue slices (thickness 8 mm) were stained with H&E. In addition, the level of CD31 in the corneal tissues was identified using IHC.

| Statistical analysis
Statistical analysis was performed by the GraphPad Prism 7 program.
Quantitative data were reported as mean ± standard deviation.
Two-group comparisons were conducted using a two-tailed Student's t-test. One-way analysis of variance followed by Tukey's multiple comparisons test was used for multigroup comparisons. p < 0.05 was considered statistically significant.   (p > 0.05), which indicated that neither TMC nor siVEGF affected the EE or DL of INS liposomes (Table S1). To confirm the siRNA binding capabilities of TIL, agarose gel electrophoresis was performed after mixing the TIL with siVEGF at different TIL/siRNA ratios. As shown in Figure 1f, the migration of siVEGF in the gel gradually slowed as TIL ratios increased.

| Preparation and characterization of NPs
Almost no free siVEGF could be detected at mass ratios above 5, demonstrating the complete binding of siVEGF by TIL conjugates. The capability of TIL to protect siRNA from nuclease degradation was verified by incubating siVEGF-TIL with RNase A for 30 min. As shown by agarose gel electrophoresis assays, the naked siVEGF RNase (À) group had a free RNA band, while the naked siVEGF RNase (+) group had no visible bands, indicating that siVEGF had been degraded in the presence of RNase.
Regarding siVEGF-TIL, both the RNase (+) and RNase (À) groups showed no apparent bands, but siVEGF-TIL NPs shook at 4 C for 2 h with or without RNase both could observe the bands, indicating that siVEGF could be released from the NPs and siVEGF-TIL could protect siVEGF from RNase degradation (Figure 1g).

| Sustained release of INS and siRNA in vitro
As shown in Figure 2a, (Figure 2b).
Overall, siVEGF-TIL and TIL showed ideal sustained release, which was conducive to maintaining concentrations of drugs and genes in the cornea and thus provided potent and prolonged therapeutic efficacy. 35 In addition, the sustained-release system can decrease the side effects of drugs on the cornea and significantly improve medication safety. 36 Furthermore, the sustained release of drugs can reduce dosing frequency, which is one way to enhance patient adherence. 37

| Efficient delivery of NPs in vitro and in vivo
Efficient intracellular uptake of NPs is required to improve the therapeutic efficacy of drugs. 38 Therefore, a CLSM was performed to examine the intracellular uptake of NPs in this study. As shown in 63.92 ± 5.36% for siVEGF FAM -TIL and 68.20 ± 5.90% for siVEGF FMA -Lipo2000 (Figures 4c and S2). Compared with siVEGF FMA -Lipo2000, siVEGF-TIL exhibited almost the same transfection efficiency. These results showed that siVEGF-TIL had adequate transfection efficiency in vitro.
To identify the efficiency of siVEGF-TIL in downregulating VEGF expression, qRT-PCR was performed. As shown in Figure 4d, qRT-PCR demonstrated that compared with that in the normal group,

| NPs improve the viability of H 2 O 2 -stimulated HCECs
As shown in Figure S3A     Alkalis saponify the fatty acids in cell membranes, which results in membrane disruption and dissolution; alkali quickly penetrates through the cornea into the deeper parts of the eye, and hyphema is present in the anterior chamber, followed by increased IOP. 40 On the other hand, early direct chemical injury can cause tissue shrinkage and disruption of the trabecular meshwork and outflow channels. Subsequent chronic inflammation may lead to synechiae and angle closure, which contribute to secondary increased IOP. 41 Figure 6g shows that the baseline IOP of the rats did not significantly differ among the groups. Statistically significant differences in IOP were first noted on the third day, from then, the median IOP was significantly increased in

| UHPLC-MS metabolomics analysis
The mechanism by which siVEGF inhibits VEGF expression and CNV is currently well understood. Briefly, siVEGF binds with RISC, causing the decomposition of the target mRNA to prevent it from being translated into a functional protein. However, the mechanism by which INS affects corneal alkali burn is unclear and was investigated in this study. 42 INS is an anabolic agent; therefore, we hypothesized that INS could treat alkali-burned corneas, through metabolic regulation.
Metabonomics is the accurate metabolomic analysis of dynamic metabolic changes in cells, tissues, and whole organisms. 43  Principal component analysis showed a trend in metabolites that were partially separated between the PBS group and INS group, indicating differences among them ( Figure S4A,B). To further determine the differences in metabolic profiles between the two groups, orthogonal projection to latent structure-discriminant analysis (OPLS-DA) score plots were constructed. As shown in Figure 7a Table S2). Based on KEGG analyses, 27 essential signaling pathways associated with these altered metabolites were identified, with 17 associated with significantly higher levels of glutamate in the PBS group than the other group (Table S3).
Glutamate is a nonessential amino acid that naturally occurs in the L-form and plays an important role in protein and carbohydrate metabolism, boosting resistance to hypoxemia, stimulating oxidation processes, preventing potential redox decreases, affecting glycolysis in tissues, and exerting hepatoprotective effects. 44 In addition, glutamate is a pivotal regulator of ferroptosis. 45 In this metabolomic analysis, the ferroptosis pathway was significantly enhanced in the PBS group compared with the INS group, and the differential abundance score was 1 (Figure 7f,g). Ferroptosis is closely associated with oxidative stress. Therefore, we hypothesized that INS and all the INS-loaded NPs in this study could treat alkali-burned corneas by decreasing glutamate levels and inhibiting the ferroptosis pathway.

| NPs may treat corneal alkali burn by inhibiting the ferroptosis pathway
Ferroptosis is a form of regulated cell death that is driven by peroxidative damage to polyunsaturated fatty acid-containing phospholipids in cellular membranes. Specifically, ferroptosis is induced by suppressing xCT and GPX4 activity and promoting the accumulation of ROS and a reduction in GSH. 46 Excessive levels of extracellular glutamate can impair or inhibit cysteine uptake via xCT, resulting in GSH depletion. 45 GSH depletion decreases GPX4 activity, and lipid peroxides can not be suppressed and metabolized, ultimately accelerating ferroptosis. 47 F I G U R E 8 NPs may treat corneal alkali burn by inhibiting the ferroptosis pathway. Glu content (a), and GSH concentration (b) in normal corneas and alkali-burned corneas that received different treatments (n = 3 per group). Results were presented as the mean ± SD. **p < 0.01; ***p < 0.001. Comparison between each group and the normal group ( # p < 0.05; ## p < 0.01; ### p < 0.001). (c) Representative Western blots showing xCT and GPX4 in normal corneas and alkali-burned corneas that received different treatments. The quantification of the Western blot assay for relative expression of xCT (d) and GPX4 (e). Results were presented as the mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001. Comparison between each group and the normal group ( # p < 0.05; ## p < 0.01; ### p < 0.001). n = 3per group. SOD activity (f) and MDA content (g) in normal corneas and alkali-burned corneas that received different treatments (n = 3 per group). Results were presented as the mean ± SD. **p < 0.01; ***p < 0.001. Comparison between each group and the normal group ( # p < 0.05; ## p < 0.01;

| In vivo inhibition of inflammation and neovascularization by NPs
Corneal alkali burn can lead to oxidative stress and severe inflammatory reactions, which can promote each other. As shown in F I G U R E 1 0 Legend on next page.
did not inhibit neovascularization in alkali-burned corneas. This may be because oxidative stress and inflammatory reactions first occur in alkali-burned corneas, and then they stimulate angiogenic factors and promote neovascularization. 6 Therefore, inhibiting VEGF without controlling oxidative stress and inflammation does not inhibit CNV. However, siVEGF-TIL treatment combines the ability of INS to inhibit oxidative stress and inflammation with the ability of siVEGF to inhibit neovascularization; morever, this treatment exhibits superior penetration and adsorption to enhance the bioavailability of drugs and genes, contributing to good therapeutic effects on corneal alkali burns.

| Biocompatibility of siVEGF-TIL in vivo
In vivo biocompatibility was assessed by corneal stimulation assessment in normal SD rat eyes treated with the different formulations, followed by corneal examination using a slit-lamp microscope (Figure 10a). After 30 days of the various treatments, no evidence of corneal opacity, CNV, inflammation, or congestion was found in any corneas. The integrity of the corneal epithelium was evaluated by fluorescein staining. The results showed that the corneal epithelium was intact. In addition, corneal anatomy was examined by H&E staining, and the results showed that the corneas in each group had a regular appearance, were closely and orderly arranged and lacked inflammatory cells or CNV ( Figure 10b). Morever, H&E staining of the major visceral organs (heart, liver, spleen, lung, and kidney) revealed that various reagents in this study treatment did not cause significant histological changes. As a result, siVEGF-TIL NPs have no obvious toxic effects and have excellent biocompatibility, paving the way for clinical applications (Figure 10c).

| CONCLUSION
To the best of our knowledge, this is the first study using a liposome-TMC nanosystem for the delivery of siVEGF/INS as a combination therapy to treat corneal alkali burns. siVEGF-TIL treatment showed significant effects in alleviating oxidative stress-induced HCEC damage (in vitro) and alkali injury in corneas (in vivo). Morever, siVEGF-TIL treatment had the ideal properties of NPs, including good biosafety profiles, lack of toxicity, facile preparation, adherence, and sustained release, suggesting that this strategy holds potential as a novel delivery platform for the cornea. Furthermore, the molecular mechanism of siVEGF-TIL treatment was revealed in this study. We found that corneal alkali burn was linked to the regulation of ferroptosis, which could be suppressed by INS. This is also the first report showing the effects of INS on ferroptosis. Notably, siVEGF-TIL could substantially inhibit both ferroptosis and CNV, eventually preventing alkali damage in corneas. siVEGF-TIL treatment is an up-and-coming therapeutic agent for future clinical applications in corneal damage. There were still many shortcomings in this study. Only male rats were used in this study because males were more susceptible than females to corneal alkali burn. This research did not compare siVEGF TIL NPs with existing treatments for corneal alkali burns (such as topical corticosteroids and NSAIDs). The absolute concentration of INS or siVEFG in the NPs was not detected, which was a limitation of this study regarding the further clinical translation of siVEGF-TIL treatment. cstc2021ycjh-bgzxm0064). The authors also would like to thank the technical support of the SHANGHAI BIOTREE BIOTECH CO., LTD.

CONFLICT OF INTEREST STATEMENT
The author declare no conflict of interest.

DATA AVAILABILITY STATEMENT
Data available on request from the authors.