Effects of Glycyrrhizin Treatment on Diabetic Cornea

Abstract

Purpose: To test how glycyrrhizin (GLY) affects mouse corneal epithelial cells (MCEC) and the diabetic murine cornea.

Methods: Viability of MCEC grown under normal or high glucose (HG) with/without GLY was tested by an MTT assay. In addition, C57BL/6 mice were injected with streptozotocin and a subset of control and diabetic mice received GLY in their drinking water. mRNA and protein levels of proinflammatory and oxidative stress molecules were tested by reverse transcription–polymerase chain reaction (RT-PCR) in both models. Ex vivo studies using human diabetic versus control corneas analyzed proinflammatory and oxidative stress markers using RT-PCR and enzyme-linked immunosorbent assay.

Results: GLY protected against loss of cell viability induced by HG and significantly reduced HMGB1, IL-1β, TLR2, TLR4, NLRP3, COX2, SOD2, HO-1, GPX2, and GR1. In vivo, corneas of GLY-treated diabetic mice showed significantly decreased mRNA expression for CXCL2, iNOS, and all molecules listed above; GLY also lowered HMGB1 and IL-1β proteins (in vitro and in vivo). Ex vivo studies using diabetic human corneas revealed elevated mRNA levels of inflammatory and oxidative stress molecules (as listed above for in vivo) versus normal age-matched controls. Protein levels for HMGB1 and IL-1β also were elevated in diabetic human versus control corneas.

Conclusions: The data provide evidence that GLY treatment attenuates inflammation and oxidative stress in vitro in MCEC and in vivo in the cornea of diabetic mice. Ex vivo data support the similarities of proinflammatory and oxidative stress data in mouse compared to human, suggesting that GLY treatment would have relevancy to patient care.

Introduction

Diabetes mellitus (DM) is one of the most serious health problems in the world.¹ The National Diabetes Statistics Report from Centers for Disease Control and Prevention provided information that in 2018, DM affected ∼34.2 million people in the United States.² The prevalence of DM significantly increased from 4.2% in the age groups of 18–44 years, to 17.5% in 45–64-year olds and to 26.8% in ≥65-year age group.² In addition, over $237 billion in direct medical costs and $90 billion in reduced productivity were caused by DM as reported by the American Diabetes Association.³

Complications associated with DM include nephropathy, neuropathy, retinopathy, cardiovascular disease, stroke, and peripheral artery disease.⁴ Patients with both Type-I and Type-II diabetes develop proliferative diabetic retinopathy (PDR) as disease progresses. Approximately 40% of type-I and more than 50% of type-II diabetics lose their vision within 5 years of the onset of PDR.¹ While retinal damage caused by DM is widely characterized and remains the primary concern of physicians, other ocular tissues¹ and the anterior segment (cornea, conjunctiva, and lacrimal glands) are also affected.¹ Corneal complications occur in about 70% of examined diabetic patients.^1,5,6 Clinically observed diabetes-related corneal alterations include a thickened, more fragile cornea, with epithelial defects, and other pathologies, as well as well as other changes, including reduction in sensitivity.⁷ Chronic hyperglycemia promotes pathological pathways, including oxidative stress, mitochondrial dysfunction,¹ and advanced glycation end (AGE)^1,8 products and inflammation.¹ Although numerous treatments, including antioxidants, topical anti-inflammatory medications, autologous serum, and artificial tears, are currently being used as therapy and prophylaxis for DM-induced eye complications, more mechanistic studies are crucial for the development of improved treatment methods.¹

In this regard, glycyrrhizin (GLY) is a saponin that is extracted from licorice root with a great number of pharmacological effects⁹ and has been shown to be effective in animal models of sepsis,¹⁰ colitis,¹¹ lung,¹² brain injury,¹³ and keratitis.^14–16 It is also used clinically to treat patients with chronic hepatitis.¹⁴ Regarding keratitis, our laboratory has shown that GLY reduces protein levels of HMGB1, a classic alarmin, and is protective against corneal disease induced by an ocular clinical isolate (KEI 1025) or a cytotoxic strain (ATCC 19660) of Pseudomonas aeruginosa.^14–16 Previous studies have reported that GLY also improved retinal permeability—neuronal and vascular outcome (eg, neuroprotective and lowered blood glucose levels) in streptozotocin (STZ)-treated mice when given in drinking water over 6 months.¹⁷ Our current work is focused to examine the effects of GLY in vitro using mouse corneal epithelium cornea (MCEC) grown in high or normal glucose (NG) and in vivo in diabetic mouse cornea. In both models tested, GLY was effective in reducing inflammation and oxidative stress induced by diabetes. We also tested similar molecules in normal and diabetic human corneas. Our study revealed similarly increased inflammatory and oxidative stress markers in diabetic human cornea compared to nondiabetic age-matched controls.

Methods

Tissue culture and treatments

Cultured MCEC (derived from C57BL/6 mice) were grown as described before.¹⁸ We incubated cells either in 5 mM (normal) or 25 mM (high) glucose for 3 days.¹⁹ To evaluate the effects of GLY, a subset of cells treated with normal or high glucose (HG) was incubated with 1 mM GLY (Sigma-Aldrich, St. Louis, MO) for 24 h. All samples were collected on day 3 for further experiments.

MTT assay

An MTT 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (ThermoFisher Scientific, Grand Island, NY) assay was used to evaluate the effects of GLY and HG on cell viability. The assay was performed according to the manufacturer's protocol as reported before.²⁰ Briefly, 5,000 MCEC were cultured in 96-well plates overnight and then treated as described above (tissue culture and treatments). Then, 5 mg/mL MTT reagent was added to each well, incubated at 37°C for 4 h, and media removed. Dimethyl sulfoxide (DMSO) was added (50 μL/well) to dissolve the metabolic product formazan and the plate was placed on a shaker at 150 rpm for 5 min. Optical density was read at 540 nm using a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA).

Mice

Eight-week-old female C57BL/6 mice were purchased from the Jackson Laboratory (Bar Harbor, ME) and housed in accordance with the National Institutes of Health guidelines. They were humanely treated and in compliance with both the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Institutional Animal Care and Use Committee of Wayne State University (IACUC 18-08-0772). C57BL/6 mice were made diabetic by 60 mg/kg injections of STZ dissolved in citrate buffer for up to 5 consecutive days.¹⁷ Control mice received citrate buffer only. Glucose measurements were done biweekly, with glucose levels >250 mg/dL considered diabetic. Mice were not fasted before blood glucose measurements, and glucose measurement was taken on blood samples obtained through tail vein, with samples measured by a hand-held measurement device. A subset of the control and diabetic mice was then treated with GLY in their drinking water (150 mg/kg/day) as previously described.¹⁷ Mice were maintained on the drinking water for 6 months and water consumption was measured weekly for the first month, and then twice a week to ensure that mice were consuming the correct dose of drug.¹⁷

Human corneas

Human postmortem corneas were obtained from Eversight (Ann Arbor, MI). Diabetic donors were 60–75 years of age and nondiabetic donors were 55–75 years of age (Table 1). The corneas were enucleated within 6–8 h after death and were COVID 19 negative. Of the total 8 corneas, 4 corneas were purchased and 4 corneas were a kind gift from Dr. Renu Kowluru. Each cornea was cut into 4 quarters and stored at −80°C until further use. Four corneas (2 normal and 2 diabetic corneas) were processed for reverse transcription–polymerase chain reaction (RT-PCR) and the remaining 4 were processed for enzyme-linked immunosorbent assay (ELISA).

Table 1.

Summary of Donors' Age, Gender, and Cause of Death

Case number	Age	Gender	Cause of death
Diabetic patients
1	59	F	Gastrointestinal Bleeding
2	61	M	Myocardial Infarction
3	61	M	Myocardial Infarction
4	75	F	Hypoxia
Control patients
1	60	F	Myocardial Infarction
2	71	F	Pulmonary Hypertension
3	72	F	End-Stage Renal Disease
4	73	M	Myocardial Infarction

F, female; M, male.

Reverse transcription–polymerase chain reaction

For all studies, in vitro (MCEC) and in vivo (mouse cornea) and ex vivo (human cornea), total RNA was isolated (RNA STAT-60; Tel-Test, Friendswood, TX) as per the manufacturer's instructions reported before.²⁰ One μg of each RNA sample was reverse transcribed using Moloney-murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen, Carlsbad, CA) to produce a cDNA template PCR. cDNA products were diluted 1:20 with DEPC-treated water and a 2 μL aliquot of diluted cDNA was used for the RT-PCR. SYBR green/fluorescein PCR master mix (Bio-Rad Laboratories, Richmond, CA) and primer concentrations of 10 μM were used in a total 10 μL volume. After a preprogrammed hot start cycle (3 min at 95°C), the parameters used for PCR amplification were as follows: 15 s at 95°C and 60 s at 60°C with the cycles repeated 45 times. Levels of high-mobility group box 1 (HMGB1), toll-like receptor (TLR) 2, TLR4, interleukin (IL) 1β, chemokine (C-X-C) ligand 2 (CXCL2), nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing (NLRP)-3, glutathione peroxidase 1 (GPX1), glutathione peroxidase 2 (GPX2), glutathione reductase 1 (GR1), superoxide dismutase 2 (SOD2), heme oxygenase 1 (HO1), inducible nitric oxide synthase (iNOS), and cyclooxygenase (COX)2 were tested by real-time RT-PCR (CFX Connect real-time PCR detection system; Bio-Rad Laboratories). For in vitro studies, 3 separate experiments were performed and the samples were run in triplicates. For in vivo studies, 5 mice/group (repeated once) were used and samples were run in triplicate. For ex vivo studies, 2 corneas/group (each cut into 4 quarters) were used and run in triplicate. Since we obtained only 2 corneas for each group, RT-PCR data obtained from the 2 normal and diabetic corneas are presented as a scatter plot graph as [NC1 and NC2 (normal) and DC1 and DC2 (diabetic)]. The fold differences in gene expression were calculated relative to control and normalized to housekeeping genes β-actin (mouse) and GAPDH (human) and expressed as the relative mRNA concentration ± SEM. Primer pair sequences used are shown in Tables 2 (mouse) and 3 (human).

Table 2.

Nucleotide Sequence of Specific Primers Used for Polymerase Chain Reaction Amplification (Mouse)

Gene	Nucleotide sequence	Primer	GenBank
β -Actin	5′-GAT TAC TGC TCT GGC TCC TAG C-3′	F	NM_007393.3
	5′-GAC TCA TCG TAC TCC TGC TTG C-3′	R
Hmgb1	5′-TGG CAA AGG CTG ACA AGG CTC-3′	F	NM_010439.3
	5′-GGA TGC TCG CCT TTG ATT TTG G-3′	R
Tlr2	5′-CTC CTG AAG CTG TTG CGT TAC-3′	F	NM_011905.3
	5′-TAC TTT ACC CAG CTC GCT CAC TAC-3′	R
Tlr4	5′-CCT GAC ACC AGG AAG CTT GAA-3′	F	NM_021297.2
	5′-TCT GAT CCA TGC ATT GGT AGG T-3′	R
Cxcl2 (Mip2)	5′-TGT CAA TGC CTG AAG ACC CTG CC-3′	F	NM_009140.2
	5′-AAC TTT TTG ACC GCC CTT GAG AGT GG-3′	R
IL-1β	5′-TTC GAG GCA CAA GGC ACA AC-3′	F	NM_008361.3
	5′-TTC ACT GGC GAG CTC AGG TA-3′	R
Nlrp3	5′-TGC CTG TTC TTC CAG ACT GGT GA-3′	F	NM_145827.3T
	5′-CAC AGC ACC CTC ATG CCC GG-3′	R
Gpx1	5′- CTC ACC CGC TCT TTA CCT TCC T-3′	F	NM_008160.6
	5′-ACA CCG GAG ACC AAA TGA TGT ACT -3′	R
Gpx2	5′-GTG GCG TCA CTC TGA GGA ACA-3′	F	NM_030667
	5′-CAG TTC TCC TGA TGT CCG AAC TG-3′	R
Gr1	5′-CCA CGG CTA TGC AAC ATT CG-3′	F	NM_010344.4
	5′- GAT CTG GCT CTC GTG AGG AA-3′	R
Sod2	5′-GCG GTC GTGTAA ACC TCA AT-3′	F	NM_013671
	5′-CCA GAG CCT CGT GGT ACT TC-3′	R
Ho1	5′-CAC GCA TAT ACC CGC TAC CT-3′	F	NM_010442
	5′-CCA GAG TGT TCA TTC GAG C-3′	R
Cox2	5′-GCA GTT CCA GTA TCA GAA CCG CAT TG-3′	F	NM_011198.3
	GAG TGA GTC CAT GTT CCA GGA GGA TG-3′	R
iNos	5′-TCC TCA CTG GGA CAG CAC AGA ATG-3′	F	NM_010927.3
	5′-GTG TCA TGC AAA ATC TCT CCA CTG CC-3′	R

Primer F, forward; Primer R, reverse.

Enzyme-linked immunosorbent assay

For in vitro studies, MCEC were treated with HG and GLY as described in the cell culture and treatment section. Cell pellets (n = 3/group) were harvested in 500 μL of PBS containing 0.1% Tween 20 and protease inhibitors. For in vivo studies, after 6 months of diabetes, individual corneas (n = 10/group) from normal mice, normal mice treated with GLY, diabetic mice, and diabetic mice treated with GLY were harvested in 500 μL of PBS containing 0.1% Tween 20 and protease inhibitors. For ex vivo studies, normal and diabetic corneas (n = 2/group) were harvested in 1,000 μL of PBS containing 0.1% Tween 20 and protease inhibitors. HMGB1 (Chondrex, Inc., Redmond, WA) and IL-1β (R&D Systems, Inc., Minneapolis, MN) were detected by ELISA as per the manufacturers' protocol. Since we obtained only 2 corneas for each group, ELISA data obtained from the 2 normal and diabetic corneas are represented in the scatter plot graph as [NC1 and NC2 (normal) and DC1 and DC2 (diabetic)].

Statistics

A one-way ANOVA followed by the Bonferroni's multiple comparison test was performed to analyze cell viability, blood glucose levels, body weights, RT-PCR, and ELISA for in vitro and in vivo experiments. For human corneas, a 2-tailed, unpaired t test with Welch's correction was used for analyzing data from RT-PCR and ELISA. Data were considered significant at P < 0.05. All experiments were repeated at least once to ensure reproducibility and data are shown as mean ± SEM.

Results

GLY prevents the loss of cell viability

MCEC grown in HG (25 mM) showed significant decrease in cell viability (P < 0.05) compared to cells grown in NG (5 mM) (Fig. 1). GLY (1 mM)-supplemented cells grown in HG showed significantly greater cell viability (P < 0.05) versus cells in HG without GLY. No significant differences in cell viability were observed in cells under NG with or without GLY.

FIG. 1.

GLY and MCEC viability. Cells were treated with HG or NG for 3 days and then a subset was treated with 1 mM GLY. Cell viability is significantly decreased when cells are treated with HG. This is reversed by GLY treatment. Statistical analysis of data used a one-way ANOVA followed by Bonferroni's multiple comparison test and are expressed as mean ± SEM of triplicate experiments. *P < 0.05. GLY, glycyrrhizin; MCEC, mouse corneal epithelial cells; NG, normal glucose; HG, high glucose.

GLY reduces inflammation and oxidative stress

mRNA expression levels of proinflammatory molecules in cells grown in NG, NG with GLY (NG+GLY), HG, and HG with GLY treatment (HG+GLY) are presented in Fig. 2A–E. Cells treated with high versus NG had significantly elevated mRNA expression levels for HMGB1 (P < 0.001), IL-1β (P < 0.001), TLR2 (P < 0.001), TLR4 (P < 0.05), and NLRP3 (P < 0.001). Cells treated with HG and GLY versus no GLY had reduced expression levels of HMGB1 (P < 0.001), IL-1β (P < 0.001), TLR2 (P < 0.01), TLR4 (P < 0.05), and NLRP3 (P < 0.001). No difference between cells grown in NG with or without GLY was seen. Levels of oxidative stress markers: COX2 (P < 0.001), HO-1 (P < 0.001), SOD2 (P < 0.001), GPX1 (P < 0.01), GPX2 (P < 0.001), and GR1 (P < 0.001) were significantly higher in cells treated with high versus NG (Fig. 2F–K). Cells treated with HG in the presence versus absence of GLY showed significantly lower expression of COX2 (P < 0.001), HO-1 (P < 0.001), SOD2 (P < 0.001), GPX1 (P < 0.001), GPX2 (P < 0.001), and GR1 (P < 0.001). No difference between cells grown in NG with or without GLY was observed. Figure 3A–B show significantly elevated HMGB1 (P < 0.001) and IL-1β (P < 0.001) protein levels in MCEC treated with high versus NG. Cells treated with HG and GLY versus no GLY showed significantly decreased HMGB1 (P < 0.001) and IL-1β (P < 0.01) protein levels. No differences were seen in cells treated with NG with or without GLY for either protein.

FIG. 2.

Effects of high glucose and GLY on inflammation and oxidative stress. MCEC were treated with HG or NG for 3 days and then a subset was treated with 1 mM GLY. RT-PCR shows significantly increased mRNA expression for HMGB1 (A), IL-1β (B), TLR2 (C), TLR4 (D), NLRP3 (E), COX2 (F), HO-1 (G), SOD2 (H), GPX1 (I), GPX2 (J), and GR1 (K) in MCEC treated with HG compared to NG. No difference between cells grown in NG ± GLY was seen. GLY treatment reduced the mRNA levels in cells treated with HG for all molecules tested. Data were analyzed using one-way ANOVA followed by Bonferroni's multiple comparison test (n = 3/group) and expressed as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. RT-PCR. reverse transcription–polymerase chain reaction.

FIG. 3.

Protein levels of HMGB1 and IL-1β. MCEC were treated with HG or NG for 3 days and then a subset was treated with 1 mM GLY. ELISA showed increased levels of HMGB1 (A) and IL-1β (B) in cells treated with HG compared to NG. No differences in the 2 proteins were observed in controls ± GLY treatment. GLY treatment reduced the proteins levels of HMGB1 (A) and IL-1β in cells treated with HG. Data were analyzed using one-way ANOVA followed by Bonferroni's multiple comparison test (n = 3/group). Data are expressed as the mean ± SEM. **P < 0.01, ***P < 0.001.

GLY reduces blood glucose levels, body weights unaffected

Figure 4A shows that mice made diabetic with STZ injections had significantly increased blood glucose levels (P < 0.001) compared to control mice. Diabetic mice with GLY supplemented in their drinking water versus no supplementation had a significant reduction in blood glucose levels (P < 0.01). GLY supplementation did not affect blood glucose levels of control mice. Diabetic mice had significantly lower (P < 0.001) body weights than the controls (Fig. 4B). GLY supplementation did not affect body weights of control or diabetic mice.

FIG. 4.

Blood glucose and body weights of normal and diabetic mice. (A). Blood glucose levels of diabetic mice (STZ) are significantly higher than normal mice (CTL). GLY significantly reduced the effects of STZ on blood glucose levels. GLY had no effect on blood glucose levels of control mice. (B). Body weights of diabetic mice are lower than control mice and GLY had no effect on body weight for either group. Data were analyzed using one-way ANOVA followed by Bonferroni's multiple comparison test (n = 10/group) and expressed as the mean ± SEM. **P < 0.01, ***P < 0.001.

GLY protects mouse cornea

Figure 5A–F provides evidence that the corneas of STZ-treated diabetic versus control mice had significantly higher expression of HMGB1 (P < 0.001), IL-1β (P < 0.05), TLR2 (P < 0.01), TLR4 (P < 0.05), CXCL2 (P < 0.001), and NLRP3 (P < 0.01). GLY-treated versus no GLY-treated diabetic mice exhibited significantly lower expression of HMGB1 (P < 0.001), IL-1β (P < 0.05), TLR2 (P < 0.01), TLR4 (P < 0.05), CXCL2 (P < 0.001), and NLRP3 (P < 0.001). No significant difference between mRNA levels in control mice with or without GLY was observed. Figure 5G–M shows the cornea of diabetic versus control mice had significantly elevated mRNA expression levels for COX2 (P < 0.001), iNOS (P < 0.001), HO-1 (P < 0.01), SOD2 (P < 0.01), GPX1 (P < 0.01), GPX2 (P < 0.001), and GR1 (P < 0.05). The cornea of diabetic mice with GLY versus without GLY in their drinking water showed significantly lower expression of COX2 (P < 0.001), iNOS (P < 0.01), HO-1 (P < 0.05), SOD2 (P < 0.001), GPX2 (P < 0.001), and GR1 (P < 0.05). No significant difference between mRNA levels in control mice with or without GLY was seen. Higher levels of HMGB1 were observed when comparing corneas from controls with STZ-treated diabetic mice. GLY treatment versus no treatment lowered protein levels of HMGB1 (P < 0.001) and IL-1β (P < 0.001) in the cornea of diabetic mice when compared to controls (Fig. 6A, B). No differences in the 2 proteins were observed in controls with or without GLY treatment.

FIG. 5.

GLY effects on proinflammatory and oxidative stress molecules in diabetic mouse cornea. RT-PCR shows significantly increased mRNA expression for HMGB1 (A), IL-1β (B), TLR2 (C), TLR4 (D), CXCL2 (E), NLRP3 (F), COX2 (G), iNOS (H), HO-1 (I), SOD2 (J), GPX1 (K), GPX2 (L), and GR1 (M) in corneas of diabetic (STZ) compared to control (CTL) mice. Except for GPX1, all other molecules were lowered in GLY-supplemented diabetic mice. GLY had no effect on expression levels of any molecule in control mice. Data were analyzed using one-way ANOVA followed by Bonferroni's multiple comparison test (n = 5/group) and expressed as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

FIG. 6.

GLY reduces HMGB1 and IL-1β protein levels in diabetic mouse cornea. ELISA showed significantly increased HMGB1 (A) and IL-1β (B) levels in the corneas of STZ compared to CTL mice. GLY treatment significantly reduced levels of HMGB1 and IL-1 β in diabetic mouse cornea. No differences in the 2 proteins were observed in controls with or without GLY treatment. Data were analyzed using one-way ANOVA followed by Bonferroni's multiple comparison test (n = 10/group) and expressed as the mean ± SEM. *P < 0.05, ***P < 0.001.

Proinflammatory and oxidative stress markers in diabetic human cornea

Figure 7A–F shows that diabetic human corneas had significantly higher expression of HMGB1 (P < 0.001), IL-1β (P < 0.001), TLR2 (P < 0.001), TLR4 (P < 0.001), CXCL2 (P < 0.001), and NLRP3 (P < 0.001) compared to age-matched controls. Figure 7G–M indicate significantly elevated mRNA expression levels for COX2 (P < 0.001), iNOS (P < 0.001), SOD2 (P < 0.001), HO1 (P < 0.001), GPX1 (P < 0.001), GPX2 (P < 0.001), and GR1 (P < 0.001) in diabetic versus normal corneas. Data (Supplementary Fig. S1) shows fold changes. Protein levels of HMGB1 (P < 0.01) and IL-1β (P < 0.001) were both significantly increased when comparing diabetic to nondiabetic groups (Fig. 8A, B).

FIG. 7.

Diabetes induces inflammation and oxidative stress in human diabetic cornea. RT-PCR shows significantly increased mRNA expression for HMGB1 (A), IL-1β (B), TLR2 (C), TLR4 (D), CXCL2 (E), NLRP3 (F), COX2 (G), iNOS (H), HO-1 (I), SOD2 (J), GPX1 (K), GPX2 (L), and GR1 (M) in diabetic (DC) compared to normal cornea (NC). Data from individual normal corneas (NC1 and NC2) and diabetic corneas (DC1 and DC2) are shown. Data were analyzed using an unpaired t test with Welch's correction (n = 2/group) corneas (C). Data are expressed as the mean ± SEM. ***P < 0.001.

FIG. 8.

Diabetes induces inflammation in human cornea. ELISA showed significantly increased levels of HMGB1 (A, C) and IL-1β (B, D) in human diabetic cornea (DC) compared to age-matched nondiabetic controls (NC). Data from individual normal corneas (NC1 and NC2) and diabetic corneas (DC1 and DC2) are shown in (A, B), while the averaged values for normal and diabetic corneas are represented by a bar graph in (C, D). Data were analyzed using an unpaired t test with Welch's correction (n = 2/group) corneas (C). Data are expressed as the mean ± SEM. **P < 0.01, ***P < 0.001.

Discussion

Studies have provided evidence that hyperglycemia, a noninfectious inflammation, triggers danger associated molecular patterns (DAMPs).²¹ HMGB1 is a prototypic DAMP and its ability to amplify inflammation has been well demonstrated.²² Upon extracellular release, HMGB1 mediates inflammatory responses that involve engagement of extracellular receptors, particularly advanced glycation end products and toll-like receptors (TLR)-2, TLR-4, and TLR-9.²¹ Hyperglycemia induces release of HMGB1.²³ In this regard, previous clinical work has demonstrated increased amounts of HMGB1 in type 1 diabetic versus nondiabetic patients²⁴ and increased levels of the DAMP and stress markers in the vitreous compartment of PDR patients.²⁵ Consistent with these findings, our data showed elevated HMGB1 levels in both cultured corneal epithelial cells and in vivo in a mouse diabetic model, as well as ex vivo in diabetic human corneas. High HMGB1 levels also have been associated with hyperglycemia in several other cell types, including pigment epithelial²⁶ pericytes²⁷ and retinal endothelial,²⁸ glial,²⁹ and retinal ganglion cells.³⁰ Other innate immune markers: IL-1β,³¹ TLR2,³² TLR4,^32,33 and NLRP3³⁴ also contribute to inflammation in diabetes. Complimenting these studies, our data show increased mRNA levels of all of those markers as well as CXCL2, a chemokine that induces PMN infiltration,³⁵ as well as protein levels of IL-1β in cells treated with HG. In this regard, our studies are in agreement with other work in different cell types treated with HG that have shown increased TLR levels (retinal ganglion cells),³² increased IL-1β expression (retinal endothelial cells),³¹ and elevated NLRP3 and proinflammatory cytokines (retinal endothelial cells).^34,35 To suppress HMGB1 downstream signaling, recent work has focused on developing strategies to block HMGB1 activation and interaction with its receptors in a variety of different inflammatory models.²¹ Our data in cornea are consistent with findings in other model systems suggesting that blocking HMGB1²¹ is a novel potential therapeutic for preventing inflammatory damage. Approaches to do this include HMGB1 siRNA,³⁶ anti-HMGB1 antibodies,³⁷ and molecules that block HMGB1 release from cells.²¹ GLY directly binds to HMGB1, thereby preventing its mitogenic and chemoattractant activities.³⁸ For example, in the diabetic retina, our collaborators provided evidence that amplification of inflammation by HMGB1 was reduced by GLY downregulating HMGB1, with decreased reduction in oxidative stress markers.¹⁷ In this study, we show that GLY also inhibits the deleterious effects of diabetes in the cornea by lowering levels of molecules such as HMGB1 and IL-1β both in vitro and in vivo. Our data are consistent with other studies showing that in a hepatocyte model of acute liver failure, as well as in diabetic retina,¹⁷ HMGB1 reduction by GLY reduced levels of the inflammatory cytokine IL-1β.¹⁷ Our work also showed that GLY lowered corneal levels of TLR2, TLR4, CXCL2, and NLRP3 under diabetic conditions both in vitro and in vivo. In this regard, GLY has been shown to attenuate levels of innate immune markers similar to what we report (TLR-2, TLR-4,³⁹ CXCL2,⁴⁰ and NLRP3⁴¹) in other disease models.

GLY also was interrogated for its antioxidative properties. Antioxidant enzymes such as GPX1, GPX2, GR1, and SOD2 defend cells against free radicals by metabolizing them to harmless by-products.⁴² Increased ROS production and disruption of homeostasis between pro-oxidant and anti-oxidant enzymes have been shown to contribute to oxidative stress in conditions such as dry eye disease.⁴² Previous work has implicated ROS in enhanced levels of antioxidant genes linked to the activation of transcription factors such as nuclear factor erythroid 2-related factor 2 (Nrf2),^43,44 nuclear factor κB (NF-κB),^43,45 and p53.⁴³ Our data indicate increased mRNA levels of GPX1, GPX2, GR1, SOD2, HO-1 and COX-2 under hyperglycemic/diabetic conditions in vitro and in vivo. Furthermore, GLY reduced mRNA levels of all these molecules under diabetic conditions in vitro. In vivo, GLY decreased mRNA levels of all molecules, except GPX1 under diabetic conditions. In this regard, GLY has been shown to affect the expression levels of Nrf2,⁴⁶ a key regulator of antioxidant genes, including GPX2 and GPX4, but not GPX1 in mice.⁴⁷ In contrast, GLY has been shown to increase levels of GPX, GSH, Nrf2, and HO-1 in hepatocyte cultures and in an in vivo model of acute liver failure.⁴⁶ GLY also increased levels of HO-1 in vivo in diabetic rats.⁴⁸

GLY increased cell viability of corneal epithelial cells cultured under HG conditions. These data are consistent with previous work illustrating the protective effects of GLY on bone marrow stromal cell viability when cells were cultured under an HG environment.⁴⁸

Since early corneal neuronal changes have been reported in diabetic (type II) subjects (assessed by measuring corneal sensitivity) before clinical evidence of diabetic retinopathy,⁴⁹ we assessed this in our in vivo model. We tested corneal sensitivity in 6-month-old normal and diabetic mice using a Cochet-Bonnet esthesiometer, but found nonsignificant differences between the corneas of diabetic and normal mice corneas (data not shown), suggesting GLY would have no effect on this measurement, hence we did not test it.

Because of the disparate nature of mice versus humans, we tested diabetic corneas from humans to check levels of innate immune and oxidative stress markers to determine if they were the same. All the markers tested were comparable between the human and mouse (Table 4), suggesting that the mouse model (cornea) is an appropriate system to use for diabetes work and to test the effects of GLY. Future studies will determine efficacy of GLY in human corneal epithelial cells grown to mimic a stratified squamous nonkeratinized epithelium, characteristic of corneal surface cells (eg, 3D cultures). In other work, we found that results from 3D versus 2D culture were similar.²⁰

Table 4.

Comparison of Proinflammatory Indicators and Oxidative Stress Markers Between Mouse and Human Cornea Samples

Marker	Mouse	Human	Similar (Y/N)
HMGB1	√	√	Y
IL-1β	√	√	Y
TLR2	√	√	Y
TLR4	√	√	Y
CXCL2	√	√	Y
NLRP3	√	√	Y
COX2	√	√	Y
iNOS	√	√	Y
HO-1	√	√	Y
SOD2	√	√	Y
GPX1	√	√	Y
GPX2	√	√	Y
GR1	√	√	Y

Table 3.

Nucleotide Sequence of Specific Primers Used for Polymerase Chain Reaction Amplification (Human)

Gene	Nucleotide sequence	Primer	Primer
GAPDH	5′-GGA GCG AGA TCC CTC CAA AAT-3′	F	NM_002046.7
	5′-GGC TGT TGT CAT ACT TCT CAT GG-3′	R
HMGB1	5′-TGG CCA AGG AAT CCA GCA GTT-3′	F	NM_001313893
	5′-CTC CTC CCG ACA AGT TTG CAC-3′	R
TLR2	5′-GGC CAG CAA ATT ACC TGT GTG-3′	F	NM_003264.3
	5′-AGG CGG ACA TCC TGA ACC T-3′	R
TLR4	5′-CAG AGT TTC CTG CAA TGG ATC A-3′	F	NM_138554.4
	5′-GCT TAT CTG AAG GTG TTG CAC AT-3′	R
CXCL2	5′-AGC TTG TCT CAA CCC CGC ATC-3′	F	NM_002089.4
	5′-TTA GGC GCA ATC CAG GTG GC-3′	R
IL-1β	5′-TTC GAG GCA CAA GGC ACA AC-3′	F	NM_000576.2
	5′-TTC ACT GGC GAG CTC AGG TA-3′	R
NLRP3	5′-GAT CTT CGC TGC GAT CAA CA-3′	F	NM_004895.4
	5′-GGG ATT CGA AAC ACG TGC ATT A-3′	R
GPX1	5′-TCC GGG ACT ACA CCC AGA TGA-3′	F	NM_000581.4
	5′-CTT GGC GTT CTC CTG ATG CC-3′	R
GPX2	5′-TTC GCT CTG AGG CAC AAC CA-3′	F	NM_002083.4
	5′-CAC CCC CAG GAC GGA CAT AC-3′	R
GR1	5′-CAT CTA TGC AGT TGG GGA TGT-3′	F	NM_000637.5
	5′-TGA GTC CCA CTG TCC CAA TAG-3′	R
SOD2	5′-GCT CCG GTT TTG GGG TAT CTG-3′	F	NM_000636.4
	5′-GCG TTG ATG TGA GGT TCC AG-3′	R
HO1	5′-TGA CCC ATG ACA CCA AGG AC-3′	F	NM_002133.3
	5′-AGT GTA AGG ACC CAT CGG AGA-3′	R
COX2	5′- TAA GCG AGG GCC AGC TTT CA-3′	F	NM_000963.4
	5′- AAG GCG CAG TTT ACG CTG TC-3′	R
iNOS	5′- TGC AGA CAC GTG CGT TAC TCC-3′	F	NM_000625.4
	5′- GGT AGC CAG CAT AGC GGA TG-3′	R

Drinking water was supplemented with GLY to treat the deleterious effects of diabetes on mouse cornea. The data showed GLY protected the diabetic mouse cornea against inflammation and oxidative stress and provide proof of principle to guide further therapeutic testing. GLY could also be administered topically in the diabetic model, as studies using GLY by this route have been previously successfully used in our laboratory to treat mouse corneas infected with P. aeruginosa.¹⁴ In clinical studies, GLY has been administered topically to treat blepharitis and dry eye disease.^50,51 GLY also has effectiveness clinically, as use of an ophthalmic solution at a concentration of 5% had no ill effect in healthy individuals as well as patients with blepharitis.⁵⁰ Future studies in our laboratory will be directed at using a topical application of GLY on diabetic corneas to minimize oxidative and inflammatory damage.

In conclusion, we found that GLY is protective in the diabetic mouse cornea and that the model shares numerous similarities with diabetic effects in humans. Our data also suggest that GLY acts through inhibiting HMGB1 amplification of inflammation, the specifics of which will be examined in future studies.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by NIH grants P30EY04068, R01EY016058, R01EY028442, the Detroit Medical Center Foundation (DMCF) and Research to Prevent Blindness.

Supplementary Material

Supplementary Figure S1