Antioxidant Effects of the Prenylated Flavonoid, Xanthohumol, on Corneal Epithelial Cells in Experimental Dry Eye Disease

Elevated levels of oxidative stress in the corneal epithelium contribute to the progression of dry eye disease pathology. Previous studies have shown that antioxidant therapeutic intervention is a promising avenue to reduce disease burden and slow disease progression. In this study, we evaluated the pharmacological efficacy of Xanthohumol in preclinical models for dry eye disease. Xanthohumol is a naturally occurring prenylated chalconoid that promotes the transcription of phase II antioxidant enzymes. Xanthohumol exerted a dose-response in preventing tertbutylhydroxide-induced loss of cell viability in human corneal epithelial (HCE-T) cells and resulted in a significant increase in expression of nuclear factor erythroid 2-related factor 2 (Nrf2), the master regulator of the endogenous antioxidant system. Xanthohumol-encapsulating poly(lactic-coglycolic acid) nanoparticles (PLGA NP) were cytoprotective against oxidative stress in vitro, and significantly reduced corneal fluorescein staining in the mouse desiccating stress/ scopolamine model for dry eye disease in vivo by reducing oxidative stress-associated DNA damage in corneal epithelial cells. PLGA NP represent a safe and efficacious drug delivery vehicle for hydrophobic small molecules to the ocular surface. Optimization of NP-based antioxidant formulations with the goal to minimize instillation frequency may represent future therapeutic options for dry eye disease and related ocular surface disease.


Introduction
Dry eye disease presents in various clinical manifestations that pose a substantial burden on the affected individual and society as a whole. Existing pharmacologic management for dry eye disease targets T cell-mediated inflammatory pathways and is associated with limited efficacy and adverse effects in up to 25% of patients [1][2][3], highlighting an urgent unmet clinical need for novel efficacious and well-tolerated therapeutics.
Previous studies have implicated the generation of Reactive Oxygen Species (ROS) and the ensuing elevated levels of cellular oxidative stress as a key contributor to the pathophysiology of dry eye disease (reviewed in [4]). Specifically, elevated levels of oxidative stress have been identified in patients with dry eye disease [5,6], while hyperosmolar conditions cause oxidative stress in cultured corneal epithelial cells [7]. We have recently shown significant oxidative DNA damage in the corneal epithelium of mice exposed to dry eye inducing conditions of desiccating environment with scopolamine [8]. Similarly, lacrimal gland dysfunction as a result of mitochondrial oxidative stress produces an ocular phenotype reminiscent of dry eye disease in mice [9,10].
Notably, a mitochondrially-targeted antioxidant, SkQ1 (Visomitin) exerts anti-inflammatory effects in human conjunctival epithelial cells in vitro [11], and has shown therapeutic benefit in US Phase 2 clinical trials following approval in Russia in 2011 [12], providing proof-of-concept evidence supporting the development of therapeutic approaches using antioxidants to treat dry eye disease.
Major challenges associated with dry eye disease management are poor patient satisfaction and compliance with dosing regimens [13]. Therefore, one important drug development consideration for topical ophthalmic formulations is to enhance ocular surface retention times that minimize the number of instillations.
In this study, we evaluated the anti-oxidative and anti-inflammatory properties of Xanthohumol in preclinical models for dry eye disease. Xanthohumol is a naturally occurring prenylated chalconoid that is abundantly present in Humulus lupulus, the hops plant. Xanthohumol promotes the transcription of phase II antioxidant enzymes [14], by stimulating the dissociation of Kelch-like ECH-associated protein 1 (Keap1) from Nuclear factor erythroid 2-related factor 2 (Nrf2), the master regulator of the endogenous antioxidant response. Keap1 is the main negative regulator of Nrf2 targeting it for ubiquitylation and degradation. The dissociation of Keap1 from Nrf2 results in nuclear translocation of Nrf2 and subsequent activation of gene expression driven by the antioxidant response element. In addition, Xanthohumol exhibits direct ROS scavenging activity due to its chalconoid structure [15].
Xanthohumol was selected based on the rationale that exploiting its dual mechanism of boosting the endogenous antioxidant response by relieving Keap1 suppression of Nrf2 translocation and direct ROS scavenging may be advantageous over antioxidants with only direct ROS scavenging activity.
The objectives of this study were to determine the cytoprotective effects of Xanthohumol in human corneal epithelial cells in vitro, and in the mouse desiccating stress/ scopolamine model for dry eye disease in vivo, using both non-formulated and poly(lactic-co-glycolic acid) nanoparticle (PLGA NP)-encapsulating Xanthohumol.
Xanthohumol resulted in a dose-dependent protection against oxidative stress, as evident by a right-shift in the IC50 curves for tBHP in the MTT assay ( Figure 2A). Similarly, Xanthohumol caused a right-shift in the EC50 curves for tBHP in the LDH assay ( Figure 2B). Specifically, the IC50 for tBHP in the MTT assay was 15.2 ± 0.5 µM in the control condition. 1 µM and 5 µM Xanthohumol resulted in a statistically significant increase in the EC50 values for tBHP to 25.6 ± 3.2 µM (P < 0.05, n = 4) and 33.3 ± 3.4 µM (P < 0.01, n = 4; Figure 2C), respectively.

Xanthohumol elicits significant increase in Nrf2 protein levels in human corneal epithelial cells
Xanthohumol is a well-known activator of the endogenous antioxidant system that acts by stimulating the dissociation of Keap1 from Nrf2. In order to demonstrate the ability of Xanthohumol to elicit this effect in corneal epithelial cells, we performed a time course analysis of Nrf2 protein levels after exposure to Xanthohumol in HCE-T cells.
Together with results from the cell viability assays presented in Figure 2, our data suggest that Xanthohumol can exert antioxidant effects in human corneal epithelial cells.

Xanthohumol-encapsulating PLGA NP are cytoprotective against oxidative stress in HCE-T cells
We next generated Xanthohumol-encapsulating PLGA NP using an 85:15 ratio of poly-lactic and poly-glycolic acid, based on previously established release parameters [16]. Nanoparticle formulations were resuspended in saline and their properties analyzed by Dynamic Light Scattering using a ZetaSizer (Malvern Pananalytical Inc., Westborough, MA, USA). Encapsulation efficiency of Xanthohumol was 68.8% (data not shown).
Empty and Xanthohumol-encapsulating PLGA NP were similar in size and size distribution averaging ~200 nm ( Figure 4; Table 1). Similarly, the polydispersity index was below 0.05 for both PLGA NP formulations, suggesting a unimodal size distribution and absence of aggregation ( Table  1). The surface charge of PLGA NPs was negative, in line with previous observations [16] (Table 1).  To assess the cytotoxicity of PLGA NP and release of Xanthohumol, we performed cell viability assays in HCE-T cells analogous to the experiments described above. HCE-T cells were seeded in 96 well plates and incubated with increasing amounts of empty and Xanthohumol-encapsulating PLGA NP for 48 h. The concentration of Xanthohumol represents the total amount of Xanthohumol present in the NP applied to the cells. In the control condition, cells were exposed to an equivalent amount (milligrams) of empty PLGA NP. Increasing concentrations of Xanthohumol-encapsulating PLGA NP exerted a dose-dependent toxicity as evident by a decrease in MTT absorbance (n = 3 -5, P < 0.001; Figure 5A) and a concomitant increase in LDH release (n = 3 -5; P < 0.001; Figure 5B). In contrast, increasing amounts (matching the NP amount of each Xn NP dose) of empty PLGA NP did not exert any cytotoxicity ( Figure 5A, B). Differences were statistically evaluated by Two-Way ANOVA with Šídák's multiple comparisons test, indicating that concentration of 10 µM or higher resulted in statistically significant cytotoxicity in HCE-T cells. Next, we tested the ability of Xanthohumol-encapsulating NP to protect HCE-T cells from exogenously-applied oxidative stress. We incubated HCE-T cells with either empty or Xanthohumolencapsulating (5 µM) PLGA NP for 20 h, prior to exposing HCE-T cells to a dose-range of tBHP (25 -125 µM) for 5 h. Xanthohumol-encapsulating PLGA NP caused a statistically significant shift in the dose-response to tBHP (n = 3, P < 0.01; Figure 6A), with IC50 values for tBHP increasing from 16.6 µM (interquartile range: 14.1 µM -18.9 µM) to 21.2 µM (interquartile range: 17.9 µM to 24.1 µM). Similarly, EC50 for tBHP values derived from the LDH release assay increased significantly from 17.9 µM to 22.4 µM (n = 3, P < 0.01; Figure 6B). Based on these data, we have identified a safe dose of Xanthohumol-encapsulating PLGA NP in HCE-T cells and confirm that Xanthohumol delivered via PLGA NP can exert antioxidative effects in human corneal epithelial cells. In the next set of experiments, we tested the efficacy of Xanthohumolencapsulating PLGA NP in a preclinical dry eye disease model.

Xanthohumol-encapsulating PLGA NP are cytoprotective against oxidative stress in HCE-T cells
We used the mouse desiccating stress/ scopolamine model to test the efficacy of Xanthohumolencapsulating PLGA NP. Mice were exposed to SiccaSystem ® cages for a period of 15 days without intervention. Subsequently, mice were treated twice daily (8 am and 6pm) by topical instillation of either empty PLGA NP, Xanthohumol-encapsulating PLGA NP or cyclosporine. In this study, we did not include a separate vehicle control group, as we have previously determined that empty PLGA NP do not exert any cytoprotective effects compared with 0.9% saline solution (data not shown).
First, we quantified tear volumes, at baseline, before start of topical treatments on day 15 and at the end of the study on day 26. We observed a statistically significant reduction of tear volumes on Day 15 suggestive of successful induction of dry eye disease pathology (from 4.7 ± 0.3 mm to 2.0 ± 0.1 mm, n = 60 eyes, P < 0.001). Two Way ANOVA analysis revealed a statistically significant effect of time (P < 0.001), but not treatment (P = 0.29), and tear volumes showed a similar statistically significant increase of tear volumes from day 15 to day 26 (P < 0.05 for all treatment groups using Tukey's multiple comparisons test; Figure 7A). Effect sizes for each treatment, determined by calculating the difference between tear volume and day 15 and day 26, did also not differ between treatment groups (Kruskal-Wallis ANOVA, P = 0.86; Figure 7B). In order to determine possible effects on corneal damage, we performed corneal fluorescein staining, again before start of topical treatments on day 15 and at the end of the study on day 26 ( Figure 8A). Corneal fluorescein staining was quantified by determining the fluorescence intensity of fluorescein on the cornea. Empty PLGA NP did not significantly affect corneal fluorescein staining (P = 0.21; Figure 8B). In contrast, Xanthohumol-encapsulating PLGA NP (P < 0.05) and cyclosporine (P < 0.01) treatment resulted in a statistically significant reduction of corneal fluorescein staining. We have previously shown that the desiccating stress/ scopolamine model results in a significant amount of oxidative DNA damage that can be prevented by antioxidant treatment [17]. In order to  determine the efficacy of Xanthohumol-encapsulating PLGA NP, we stained corneal sections for 8hydroxy-2' deoxyguanosine (8-OHdG) and quantified immunoreactivity in corneal epithelial cells. Empty PLGA NP-treated eyes showed significant nuclear 8-OHdG immunoreactivity; in contrast, Xanthohumol-encapsulating PLGA NP showed a visible reduction in 8-OHdG intensity (Fig. 9A). Quantification revealed a statistically significant reduction in 8-OHdG staining by 49.3 ± 7.3% (n = 9 -10 per group; P < 0.01; Fig. 9B). This marked reduction in oxidative stress-associated DNA damage in corneal epithelial cells was not associated with marked changes in the histopathological properties of the cornea (Table 2). Specifically, epithelial and stromal thickness were not significantly affected by Xanthohumolencapsulating PLGA NP treatment.

Discussion
Our data provide strong in vitro and in vivo evidence that the natural compound, Xanthohumol, can exert cytoprotective and antioxidative effects in preclinical models for dry eye disease. Specifically, Xanthohumol and Xanthohumol-encapsulating PLGA NP were cytoprotective against oxidative stress injury in human corneal epithelial cells. Furthermore, Xanthohumol-encapsulating PLGA NP delivered topically reduced severity of corneal fluorescein staining and 8-OHdG labeling in the cornea, suggestive of reduced corneal damage and corneal oxidative DNA damage, respectively.
Previous studies have implicated increased cellular levels of oxidative stress in ocular surface disease. For example, lacrimal gland dysfunction can cause hyperosmolarity of the tear film [18], eliciting the generation of oxidative stress in human corneal epithelial cells [7]. Reactive Oxygen Species can activate nuclear factor-B (NF-B) [19], which regulates both the endogenous antioxidant system, but also pro-inflammatory signaling through toll-like receptor 4 [20]. In our previous studies, we have shown that exposure to the desiccating stress/ scopolamine model for dry eye disease causes significant increases in oxidative stress-mediated corneal damage [17], extending previous reports of apoptosis and damage to the corneal epithelium [19]. Therefore, the desiccating stress/ scopolamine model for dry eye disease is a useful model to investigate the effects of antioxidants and antioxidant formulations on the ocular surface.
We used HCE-T cells as in vitro model to determine toxicity and efficacy of Xanthohumol and Xanthohumol-encapsulating PLGA NP (Figures 1 -3 and 5 -6). While HCE-T cells are widely used, especially as they form a stratified epithelium with barrier properties and a characteristic morphology [21]; however, HCE-T cells also display genomic abnormalities suggestive of some genetic drift [22], which must be considered when interpreting in vitro findings derived from this cell line. Our mouse model for dry eye disease is based on a well-established paradigm that employs lowhumidity air flow and concurrent scopolamine administration to induce dry eye disease in wild-type mice [23]. We have previously refined the model and its quantitative readouts used to assess dry eye disease severity for determining the efficacy of novel anti-dry eye disease therapeutics, including antioxidants [17]. The magnitude of changes, as well as the response of the positive control, ophthalmic cyclosporine (Restasis), were similar to those reported by us and others previously for this model [17,23,24].
Exposure to the desiccating stress environment with concomitant scopolamine administration resulted in a statistically significant reduction of tear volumes (~60%), showing successful induction of ocular surface disease ( Figure 7A). In this study, all groups showed a statistically significant increase in tear volumes at the end of the 10-day treatment period, however, no statistically significant differences between vehicle, Xanthohumol and cyclosporine-treated eyes were observed ( Figure 7B). This suggests that the increase is primarily caused by lubrication of the cornea and tissues of the ocular surface, rather than due to a direct pharmacological effect. Here it may be important to note that tear volume measurements from mice are notoriously challenging and are easily confounded by physiological and environmental factors.
To determine the pharmacological efficacy of Xanthohumol in vivo, we used a PLGA NP-based formulation. PLGA NP are well-tolerated, biodegradable and approved by The United States Food and Drug Administration. Release from PLGA NP occurs as NP degrade and is governed, in part, by the ratio of poly-lactic and poly-glycolic acid parameters [25]. For this first proof-of-concept study, we used a ratio of 85:15 (poly-lactic:poly-glycolic acid), based on predicted release properties for the NP published by us and others previously [16,26].
One shortcoming of the current study is that PLGA (85:15) NP are negatively charged (Table 1). It is generally assumed that cationic NPs exhibit enhanced retention times on negatively charged ocular tissues, such as the cornea and the conjunctiva [25,27]. Therefore, we opted to administer Xanthohumol-encapsulating PLGA NP twice daily, matching the instillation frequency of cyclosporine. A detailed quantitative analysis of retention times of Xanthohumol-encapsulating PLGA NP on the ocular surface is beyond the scope of this article, which to our knowledge provides the first preclinical proof-of-concept supporting the use of Xanthohumol for ocular surface disease. Future studies will address modifications to PLGA NP formulations to include co-polymers such as chitosan or Eudragit RL100. For example, the latter, a copolymer of ethyl acrylate, methyl methacrylate and a low content of methacrylic acid ester with quaternary ammonium groups, has been successfully used for encapsulating cyclosporine with enhanced properties for topical delivery [28]. Nonetheless, Xanthohumol-encapsulating PLGA NP showed similar efficacy when compared against 0.05% ophthalmic cyclosporine emulsion (Restasis; Figure 8), which is the current standard of care for patients with moderate to severe dry eye disease in the United States [2,29].
Xanthohumol is generally considered to exert its cytoprotective effects through both stimulating the dissociation of Keap1 from Nrf2 and direct ROS scavenging activity [14,15]. Typically, scavenging of ROS results in the diminishing activation of the phase II antioxidant system [30], reducing the endogenous antioxidant potential as cellular levels of oxidative stress fall. Given the potent activation of Nrf2 in HCE-T cells in the absence of oxidative stress (Figure 3), Xanthohumol may be particularly well-suited for encapsulation in PLGA NP. In a previous study, we quantified the efficacy of three-times daily administration of the potent superoxide dismutase mimetic, manganese(III) tetrakis(1-methyl-4-pyridyl) porphyrin (Mn-TM-2-PyP). Intriguingly, Xanthohumolencapsulating PLGA NP had a much larger effect on 8-OHdG labeling in the cornea, reducing density of immunolabel by ~50% (Figure 9), compared with an ~25% reduction elicited by Mn-TM-2-PyP [17]. Given that the antioxidant potential of Mn-TM-2-PyP is significantly greater than that of Xanthohumol ( [17,31]), this finding may suggest that Xanthohumol-encapsulating PLGA NP are not only able to be retained at the ocular surface for a prolonged period of time despite their negative surface charge, but also achieve sustained activation of the endogenous antioxidant system.

Test articles, antibodies and chemicals
Xanthohumol was purchased from Cayman Chemicals (Ann Arbor, MI, USA) and dissolved in dimethyl sulfoxide at a concentration of 100 µM (Millipore Sigma, St. Louis, MO, USA) for in vitro experiments. Cyclosporine A for transporter assays was USP grade (99% purity) from Cayman Chemical Company (Ann Arbor, MI, USA). Ophthalmic cyclosporine emulsion was pharmaceutical grade, Restasis ® (0.05% cyclosporine; Allergan Plc., Irvine, CA, USA).
Unless otherwise specified, analytical grade reagents were obtained from Millipore Sigma (St. Louis, MO, USA).

Cell viability assays
To determine the cytoprotective effects of Xanthohumol against chemically-induced oxidative stress, we conducted 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) uptake and lactate dehydrogenase (LDH) release assays, essentially as previously described by us in detail [17]. In brief, supernatants (50 µL) were collected and LDH assays performed. Cells were incubated with MTT dye for 1.5 h and subsequently lysed in dimethyl sulfoxide. Data were normalized to the baseline control condition and expressed as fold-change.

Quantitative immunoblotting
Immunoblotting on HCE-T cell lysates was performed as described by us previously [31].

Generation and characterization of PLGA NP
PLGA NPs were prepared using an oil-in-water single emulsion technique, essentially as described previously [16]. Briefly, 50 mg of PLGA (85:15; Durect Corp., Birmingham, AL) were dissolved in 1 ml dichloromethane and slowly added to ice-cold polyvinyl alcohol (1% w/v, 10 ml), while vigorously vortexing. The resultant suspension was emulsified by probe sonication and diluted with 100 ml ice-cold PVA. The organic solvent was allowed to evaporate with constant stirring for 3 h at 23°C and the resulting PLGA nanoparticles were isolated by centrifugation (25,000 × g for 20 min at 4 °C) and washed three times with deionized water. PLGA NP were resuspended in sucrose (10 ml of 5 mg/ml sucrose in deionized water) and lyophilized. PLGA NP were stored at −80 °C until use. Xanthohumol-encapsulating PLGA NP were synthesized by dissolving 5 mg Xanthohumol in the initial organic phase. PLGA NP properties were determined by dynamic light scattering using a ZetaSizer analyzer (Malvern Pananalytical Inc., Westborough, MA, USA).

Desiccating stress/ scopolamine model for experimental dry eye disease
All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the European Commission Directive 86/609/EEC for animal experiments, using protocols approved and monitored by the Animal Experiment Board of Finland. C57BL/6JRj mice were purchased from Janvier Labs (Le Genest-Sainte-Isle, France). Mice were housed at a constant temperature (22 ± 1 °C) and in a light-controlled environment (lights on from 7 AM to 7 PM) with ad libitum access to food and water. Male mice (9 weeks of age) were used for experiments.
Dry eye disease-like pathology was induced by exposure to a combination of desiccating stress in SiccaSystem ® cages (K&P Scientific LLC, Forest Park, IL, USA) and transdermal administration of scopolamine (Scopoderm ® ; Glaxo Smith Kline, Middlesex, UK), as described by us previously [17,24]. In this study, mice were exposed to desiccating stress/ scopolamine for a total of 26 days; test articles were administered by twice daily (8 am and 5 pm) topical instillation (10 µl) into both eyes starting on day 16 for a period of ten days.
Corneal fluorescein staining measurements were performed on day 16 and on day 26, as described by us previously [17,24]. Animals were randomized and assigned to treatment groups based on the corneal fluorescein score on day 16.

Data Analysis and Statistics
All data were analyzed with the investigator blinded for treatment group. Data are presented as mean ± standard error of mean (SEM) or as median ± interquartile range or 25 th /75 th percentile. Data were analyzed using paired or unpaired Student's t-test, Wilcoxon signed rank test, One-Way analysis of variance (ANOVA) or Kruskal-Wallis ANOVA, or Two-Way ANOVA. Differences between groups were subsequently determined using either Tukey's, Dunn's, or Sidak's multiple comparisons tests as appropriate. Differences were considered statistically significant at the P < 0.05 level.

Conclusions
Xanthohumol was cytoprotective against oxidative stress injury in human corneal epithelial cells. Xanthohumol-encapsulating PLGA NP significantly improved dry eye disease pathology in the mouse desiccating stress/ scopolamine model. PLGA NP represent a safe and efficacious drug delivery vehicle for hydrophobic small molecules to the ocular surface. Future studies will optimize Xanthohumol NP-based formulations with the goal to minimize instillation frequency.