Serine Protease Inhibitor A3K Protects Rabbit Corneal Endothelium From Barrier Function Disruption Induced by TNF-α

Abstract

Purpose.: To determine if a serine protease inhibitor A3K (SA3K) reduces TNF-α–induced declines in rabbit corneal endothelial junctional barrier integrity.

Methods.: New Zealand rabbit corneas were incubated ex vivo for 24 hours in Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS with or without TNF-α, in the presence or absence of SA3K at different concentrations. Corneal endothelial barrier function permeability was determined based on measurements of FITC-dextran tissue accumulation. Apical junctional complex (AJC) integrity was evaluated of zonula occludens-1 (ZO-1), vascular endothelial (VE)-cadherin, and filamentous actin (F-actin) and associated microtubules, as well as myosin light chain (MLC) by immunofluorescent staining, Western blot analysis, and/or RT-PCR.

Results.: TNF-α (20 ng/mL) increased corneal endothelial FITC-dextran permeability by 1.8-fold compared with the untreated control. SA3K (100–200 nM) dose dependently suppressed TNF-α–induced increases in permeability. SA3K nearly completely reversed TNF-α–induced disruptions of tight junctional ZO-1 and subjacent adherens junctions VE-cadherin integrity. Interestingly, SA3K reversed TNF-α–induced disruption of AJC linkage to the cytoskeletal F-actin array by restoring F-actin double-band structures. SA3K also attenuated TNF-α–induced microtubule disassembly. Furthermore, SA3K blocked increases in MLC phosphorylation status elicited by TNF-α.

Conclusions.: SA3K exposure markedly reduced TNF-α–induced disruption of barrier structure and function in the rabbit corneal endothelium by maintaining AJC integrity. These protective effects are due to suppression of MLC activation. SA3K may have, in vivo, a therapeutic potential to offset TNF-α–induced declines in endothelial barrier structural integrity and function.

Introduction

The corneal endothelium plays an essential role in preventing the stroma from imbibing excessive fluid resulting in declines in tissue transparency due to an increase in stromal thickness. This stromal dehydrating function is mediated through outward directed net osmotically coupled fluid transport from the stroma into the anterior chamber. Net fluid uptake into the stroma depends on the arithmetic difference between stromal fluid uptake through both the transendothelial and paracellular low resistance junctional pathways and cellular extrusion of osmolytes into the anterior chamber. A decline in endothelial ion transport activity and/or compromise of paracellular junctional integrity leading to an increase in permeability will increase stromal hydration and cause the stroma to swell and become less transparent. Therefore, maintenance of corneal deturgescence and its refractive power is critically dependent on preventing fluid leaks across the paracellular pathway from exceeding the ability of endothelial fluid transport activity to sustain a constant stromal hydration state. ¹  

Stromal fluid imbibition is caused by the physiochemical properties of the nondiffusible, negatively charged, hydrophilic molecules such as glycosaminoglycans in the stromal ground substance. This fluid leak across the paracellular pathway from the anterior chamber is modulated by changes in endothelial barrier function and net ion transport activity mediated by endothelial cell sodium, potassium, and adenosine triphosphatase (Na⁺/K⁺-ATPase) activity. Furthermore, endothelial barrier integrity is also essential for maintaining fluid pump activity since this property assures maintenance of transmembrane osmolyte transporter polarity (e.g., Na⁺/K⁺-ATPase) and local osmotic gradients needed for active fluid movement from the stroma into the anterior chamber. Therefore, the barrier integrity of the endothelium is indispensable for maintenance of stromal deturgescence and corneal transparency. ^2–4  

Corneal endothelial tight junctions are focally present around endothelial cells. These junctional (TJ) complexes are composed of transmembrane proteins such as claudin and occludin; membrane-associated proteins such as zonula occludens (ZO)-1, -2, -3; and actin filaments. ZO-1 plays an important role in maintaining the barrier function and serves as a TJ biomarker. Adherens junctions (AJs) are immediately subjacent to TJs and are essential for tethering forces generated between the adjacent cells. AJ and TJ composition are similar to one another. Both of them consist of a transmembrane protein (vascular endothelial cadherin, or VE-cadherin) and cytoplasmic linker proteins (catenin α, β, γ). The corneal endothelial and epithelial TJs are similar to one another since they both are composed of a thick actin cytoskeletal band at the apical junctional complex (AJC), which has been referred to as the perijunctional actomyosin ring (PAMR). The cytoplasmic domains of the transmembrane molecules in the TJs and AJs are structurally and functionally linked to the thick cytoskeletal cortical filamentous actin (F-actin) band via linker proteins. ^5–7 Microtubules are another key constituent of the endothelial cytoskeleton, which is important for endothelial integrity and barrier function. There is significant cross-talk between F-actin cytoskeleton and microtubule, ^8,9 which are presumed to resist the compression generated by actin–myosin contractility. 

When the corneal endothelium is subjected to allograft rejection, proinflammatory mediators, surgical trauma, or other diseases, barrier integrity disruption occurs leading to increases in stromal hydration. ^2,3,10–14 One of these mediators is TNF-α, and it plays a major role in corneal endothelial dysfunction. Its levels are elevated in the aqueous humor of humans afflicted with anterior uveitis and those undergoing corneal allograft rejection. ^15–18 In the rabbit corneal endothelium, Watsky et al. ¹⁹ showed that TNF-α can increase carboxyfluorescein permeability. Other studies have verified that TNF-α–induced barrier dysfunction occurs concomitant with AJC remodeling and microtubule disassembly. ^8,20 Furthermore, TNF-α exposure induces F-actin cytoskeletal contractility. ¹⁹ This response occurred as a consequence of myosin light chain (MLC) phosphorylation. ²¹ Recently, Shivanna et al. ²⁰ found that TNF-α disrupted corneal endothelial barrier integrity concomitant with transient activation of p38 mitogen-activated protein (MAP) kinase. This response to TNF-α can be suppressed by preexposure to a selective p38 MAP kinase inhibitor. Other signaling mechanisms controlling barrier integrity include protein kinase A (PKA) and the Rho/ROCK pathway, since increases in cAMP, ²² induced by paclitaxel ⁸ stabilized microtubules whereas the Y-27632 (a Rho kinase inhibitor) ²³ suppressed barrier integrity disruption in different cell or animal models. However, there are no available pharmacologic interventions to overcome in a clinical setting endothelial dysfunction. Therefore, novel therapeutic options are needed to reduce corneal endothelial dysfunction by reducing declines in barrier function that reduce corneal transparency. 

Serine protease inhibitors (SERPINS) are the largest super family of protease inhibitors involved in many critical biologic processes. Many of them can prevent TNF-α–induced barrier disruption. For example, 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF) prevents TNF-α–induced blood–brain barrier opening. ²⁴ Kallistatin can inhibit a number of biologic responses induced by TNF-α. ²⁵ Another serine protease inhibitor A3K (SERPINA3K, or SA3K) was first identified as a SERPIN family member that selectively inhibits tissue kallikrein. It is expressed at high levels in the liver, and lower levels in other tissues, such as the kidney, pancreas, and retina, ²⁶ but not in the cornea. ²⁷ Similar to many other serpins, SA3K is present at high levels in the serum. ²⁸ Recent studies demonstrated that SA3K inhibits neovascularization, ²⁹ inflammation, oxidative stress, ^30,31 and fibrosis ³² in the retina and cornea ³³ through the suppression of Wnt signaling pathway. ^32,34  

Furthermore, SA3K reduces retinal vascular leakage in an oxygen-induced retinopathy model and also prevents hypoxia-induced declines in tight junctions occludin expression. ³⁰ These findings prompted us to hypothesize that in the rabbit corneal endothelium SA3K also suppresses TNF-α–induced barrier dysfunction. 

Materials and Methods

Experimental Animals

New Zealand rabbits, weighing between 1.5 to 2.0 kg, were purchased from Shanghai Shilaike Laboratory Animal Co., Ltd. (Shanghai, China). Care, use, and treatment of all animals in this study were in strict agreement with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the guidelines in the Care and Use of Laboratory Animals set forth by the University of Xiamen (Fujian, China). Animals were housed in a temperature-, humidity-, and light-controlled room. Food and water were available ad libitum. 

Purification of SA3K

The SA3K/pET28 plasmid expressing SA3K was introduced into Escherichia coli strain BL21. The vector provides a signal peptide that enables the recombinant protein to enter the periplasmic space. The expression and purification followed the protocol recommended by GE Healthcare (Piscataway, NJ) with some modifications. Briefly, expression was induced by the addition of isoprupylthio-β-galactoside (IPTG) and carried out overnight at 25°C. Periplasmic proteins were released by digestion with lysozyme and separated from cells by centrifugation. SA3K was purified by passing through the His-Bind column (GE). The purity of recombinant SA3K was examined by SDS-PAGE. Endotoxin concentrations were monitored using a limulus amebocyte kit (BioWhittaker, Walkersville, MD). Activity of the purified protein was examined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay using primary human umbilical vein endothelial Cells (HUVEC). 

Materials and Antibodies

Pentobarbital sodium was from Abbott Laboratories (North Chicago, IL); recombinant human TNF-α (biologic activity ≥ 2 × 10⁷ units/mg, endotoxin free), FITC-dextran (3-5 kDa), MTT, and dimethyl sulfoxide (DMSO) were from Sigma-Aldrich (St. Louis, MO); Dulbecco's modified Eagle's medium nutrient (DMEM), fetal bovine serum (FBS), and Texas Red–X phalloidin were obtained from Invitrogen-Gibco (Carlsbad, CA). Rat polyclonal antibody to ZO-1 from Zymed-Invitrogen (Carlsbad, CA); mouse monoclonal antibodies to VE-cadherin, α-tubulin, and to MLC were from Sigma-Aldrich; mouse monoclonal antibodies to MLC phosphorylated Ser19 (p-MLC) from Cell Signaling Technology (Beverly, MA); Alexa Fluor 488-conjugated donkey anti-rat IgG, Alexa Fluor 546-conjugated donkey anti-mouse IgG were from Invitrogen-Gibco; horseradish peroxidase-conjugated goat antibodies to rabbit and mouse IgG were purchased from Merck (Whitehouse Station, NJ). Mounting medium with 4, 6-diamidino-2-phenylindole (DAPI) and BSA were from Vector Laboratories (Burlingame, CA); enhanced chemiluminescence (ECL) kit was obtained from GE Healthcare UK (Chalfont, UK). 

Ex Vivo Model of Corneal Endothelial Barrier Disruption

Corneal endothelial barrier disruption was evaluated with an ex vivo rabbit tissue model. Rabbits were euthanized by intravenous overdoses of pentobarbital sodium. Corneal epithelial cells were scraped mechanically and the eyes were enucleated. Under a dissecting microscope (Model SZ40; Olympus, Tokyo, Japan), the corneas were carefully excised 3 mm outside of limbal rim and the remaining iris, lens, and retina were carefully discarded. The freshly isolated corneal tissues, incubated in DMEM containing 10% FBS and an antibiotic mixture (penicillin 100 U/mL and streptomycin 100 ug/mL), were treated with TNF-α at different concentrations (0–50 ng/mL), in the presence or absence of SA3K (100 nM or 200 nM) at 37°C in a humidified atmosphere containing 5% CO₂ for 24 hours. Tissues without treatment were used as controls. 

Measurement of Permeability of Endothelium to FITC Dextran

The endothelial tight junctional permeability was evaluated based on measurements of FITC-dextran tissue accummulation. ^19,22 Briefly, corneas were treated ex vivo for 24 hours using the aforementioned endothelial cell barrier disruption model. They were washed in PBS three times and placed with their endothelial cells facing upwards in wells containing 500 μL PBS on a of 24-well plate. One hundred microliters of FITC-dextran in PBS, at a concentration of 1 mg/mL, were added to the medium bathing the endothelial cell surface and incubated at 37°C for 10 minutes. After removal of the FITC-dextran solution, the corneal tissues were then carefully transferred out of the wells in preparation for fluorescence intensity measurements. They were washed three times with PBS and then homogenized in 1 mL PBS. Subsequently, the medium was centrifuged at 9000g for 30 minutes at 4°C. The supernatants were collected for detecting FITC fluorescence intensity with the SpectraMax M2e Microplate Reader (Molecular Devices, Sunnyvale, CA) by excitation at 485 nm, and emission monitoring at 538 nm. 

Immunofluorescence Microscopy

The treated corneal tissues were fixed in PBS with 4% paraformaldehyde for 5 minutes at room temperature (RT) and in acetone for 3 minutes at −20°C, and four incisions were made in each cornea. After washing in TD buffer (PBS with 1% dimethyl sulfoxide [DMSO] and 1% Triton X-100), the tissues were incubated in 2% BSA diluted in PBS for 1 hour at RT to block nonspecific binding. For staining F-actin, the corneas were incubated overnight at 4°C with Texas Red–X phalloidin (1:100; Invitrogen-Gibco). For staining AJC and microtubules, tissues were incubated overnight at 4°C with the primary antibody for ZO-1 (1:100), VE-cadherin (1:100), and α-tubulin (1:100) in a mixture of 1% BSA in PBS. Subsequently, they were rinsed three times with TD buffer, and incubated with secondary antibodies (donkey anti-rat/mouse IgG conjugated Alexa Fluor 488/546, 1:1000 in a mixture of 1% BSA in PBS) for 1 hour at RT. Stained whole-mount cornea tissues were mounted endothelial side up on a slide and stained with DAPI. Tissues without primary antibody were used as negative control. Finally, the cornea tissues were visualized under a laser confocal microscope (Olympus Fluoview 1000; Olympus). 

Western Blot Analysis

The incubated corneal tissues were washed with PBS, and the endothelial layer was isolated mechanically. Equal protein amounts in the cell lysates were subjected to electrophoresis on 8% (ZO-1, 220 KDa), 10% (VE-cadherin, 125 KDa), or 12% (MLC and p-MLC, 18 KDa) SDS-PAGE and then electrophoretically transferred to a PVDF membrane. After blocking in 2% BSA for 1 hour at RT, the membranes were incubated with primary antibodies for ZO-1 (1:1000), VE-cadherin (1:1000), MLC (1:1000), p-MLC (1:500), and β-actin (1:10,000) as a loading control overnight at 4°C with gentle rocking. After three washes with Tris-buffered saline with 0.05% Tween-20 for 10 minutes each, the membranes were incubated with HRP-conjugated goat anti-rabbit or mouse IgG (1:10000) for 1 hour at RT. The blots were visualized using an ECL reagent. Band intensities were measured using Molecular Imager ChemiDoc XRS System (Bio-Rad, Hercules, CA) and analyzed with image analysis software (Quantity One; Bio-Rad). 

RT-PCR Analysis

Corneal endothelial layers were harvested as described above, and total RNA was isolated using Trizol (Invitrogen-Gibco) according to the manufacturer's instructions. mRNA was reverse transcribed to complementary DNA (cDNA) (Revert AidTM First Strand cDNA Synthesis Kit; Fermentas EU, St. Leon-Rot, Germany) at 25°C for 5 minutes, then first at 42°C for 2 hours, followed by 70°C for 5 minutes, finally cooled to 4°C. The PCR protocol was designed to maintain amplification in the exponential phase. Sequences of the PCR primers were as follows: ZO-1 sense, 5′-GTCTGCCATTACACGGTCCT-3′; antisense, 5′-GGTCTCTGCTGGCTTGTTC-3′ (307bp); VE-cadherin sense, 5′-AACGGTTCGTGAGGAACAAC-3′; antisense, 5′-GTGGTCAGACAGGGACAGGT-3′ (356bp); glyceraldehyde-3-phosphate dehydrogenase (GAPDH; internal control) sense, 5′-ACCACAGTCCACGCCATCAC-3′; and antisense, 5′-TCCACCACCCTGTTGCTGTA-3′ (456bp). PCR were performed at 94°C for 3 minutes, 94°C for 30 seconds, 55°C for 30 seconds, 68°C for 55 seconds, and 68°C for 7 minutes. The reaction mixture was finally cooled to 4°C. RT and PCR incubation were performed with a PCR system (GeneAmp 2400-R; Perkins-Elmer, Foster City, CA) and the products of amplification were fractionated by electrophoresis on a 2% agarose gel and stained with ethidium bromide (EB). Band intensities were measured using Image Acquisition and Analysis System (UVP, Cambridge, UK) and analyzed with image analysis software (VisionWorksLS; UVP). 

Statistical Analysis

Quantitative data are presented as means ± SE from three independent experiments. Differences were evaluated by ANOVA followed by Dunnett's multiple comparison test. A P value of less than 0.05 was considered statistically significant. 

Results

SA3K Protected Endothelium From Increased Permeability Caused by TNF-α

The effect of SA3K on barrier integrity in the presence of TNF-α was determined based on measurements of FITC-dextran tissue accumulation. Initially, corneas were exposed to either 1, 5, 10, 20, or 50 ng/mL TNF-α for 24 hours. With 1 ng/mL TNF-α, permeability was unchanged from the control condition. Figure 1A shows that beginning with 5 ng/mL TNF-α, the permeability increased in a concentration-dependent manner to plateau at a value 1.8-fold above the control and identical to that measured with 20 ng/ml. Accordingly, the effects of either 100 nM or 200 nM SA3K were determined on tissue permeability induced by 20 ng/mL TNF-α. Cotreatment with 100 nM SA3K attenuated TNF-α (20 ng/mL)–induced increases in corneal endothelial paracellular permeability. Interestingly, 200 nM SA3K completely reversed the increases in permeability induced by 20 ng/mL TNF-α, while 200 nM SA3K alone did not change FITC-dextran permeability (Fig. 1B). 

Figure 1

 

SA3K suppresses TNF-α–induced increases in tight junctional permeability: FITC-dextran (3-5 KDa) uptake was measured in de-epithelialized isolated rabbit endothelium (A). The endothelium was exposed to DMEM + 10% FBS medium with various concentrations (0, 1, 5, 10, 20, 50 ng/mL) of TNF-α for 24 hours. (B) Tissues were co-incubated with TNF-α (20 ng/mL) + SA3K (100 nM or 200 nM) or with just 200 nM SA3K alone. Data are means ± SE from three independent experiments. *P < 0.05, **P < 0.01 for the indicated comparisons (ANOVA followed by Dennett's test).

Protective Effects of SA3K Against TNF-α–induced Losses in TJ and AJ Integrity

The patterns of distribution were determined of ZO-1 and VE-cadherin within TJs and AJs with immunofluorescence microscopy (Figs. 2A, 3A, respectively). In control tissues, ZO-1 and VE-cadherin expression was evident in the TJ or AJ, forming a hexagonal pattern. Consistent with previous observations, ⁵ ZO-1 distribution at higher magnification revealed the presence of gaps, which appeared to occur uniformly at the Y-junctional area between three adjacent endothelial cells, demonstrating that the corneal endothelial ZO-1 was not circumferentially continuous around cell borders. In contrast, VE-cadherin was stained contiguously at the cell borders. Exposure to TNF-α for 24 hours dispersed ZO-1 and disrupted VE-cadherin distribution. These changes were inhibited by SA3K. We also confirmed the expression of ZO-1 and VE-cadherin in the corneal endothelium by Western blot analysis (Figs. 2B, 3B, respectively) and RT-PCR (Figs. 2C, 3C, respectively). Exposure of the tissues to TNF-α for 24 hours had no significant effects on ZO-1 and VE-cadherin abundance. 

Figure 2

 

Protection by SA3K of TJ structure disruption by TNF-α. Corneal endothelia were incubated with TNF-α (20 ng/mL) in the presence or absence of SA3K (100 nM, 200 nM) for 24 hours. TJ protein ZO-1 distribution was determined with immunofluorescence microscopy (A). In control tissues, ZO-1 intensely stained at the apical junctions of two adjacent cells. ZO-1 distribution at a higher magnification reveals gaps, which appear to occur uniformly at the Y-junctional area between three adjacent endothelial cells. Exposure to TNF-α resulted in ZO-1 distribution dispersion. On the other hand, SA3K dose-dependently inhibited this change. Negative effects of TNF-α on ZO-1 abundance and gene expression based on Western blot analysis (B) and RT-PCR (C).

Figure 3

 

SA3K protects against AJs structural alterations induced by TNF-α. Corneal endothelia were incubated with TNF-α (20 ng/mL) in the presence or absence of SA3K (100 nM, 200 nM) for 24 hours. AJ VE-cadherin protein localization was visualized by immunofluorescence microscopy. Under control conditions, VE-cadherin was continuously positive at the apical junctions between two adjacent cells. Exposure to TNF-α resulted in VE-cadherin dislocation and this effect was inhibited by SA3K (A). Negative effects of TNF-α on VE-cadherin protein and mRNA expression levels based on Western blot analysis (B) and RT-PCR (C).

Reversal by SA3K of TNF-α–Altered F-actin Cytoskeletal Distribution

F-actin cytoskeletal organization was visualized by immunostaining with Texas Red–X phalloidin (Invitrogen-Gibco) (Fig. 4). Its distribution shows that F-actin microfilaments are under control conditions localized to the apical cell borders and form a circumferential bundle. Each cell contained its own separate F-actin cortex along the periphery, producing the double-band appearance noted by Barry et al. ⁵ There was a broad connection between the double bands, especially at the Y-junctional regions of three adjacent cells. On the other hand, TNF-α caused apparent microfilament bundle fusion between cells and disappearance of its double-band appearance. TNF-α also caused the F-actin cytoskeletal organization to undergo rearrangement, which was almost completely reversed by SA3K cotreatment. 

Figure 4

 

SA3K blocks TNF-α–induced changes in F-actin distribution and cytoskeletal rearrangement. Corneal endothelia were treated with TNF-α (20 ng/mL), in the presence or absence of SA3K (100 nM, 200 nM) for 24 hours. Under control conditions, Texas Red–X phalloidin staining of cytoskeletal F-actin was continuous along the apical cell borders and formed a circumferential bundle. At a higher magnification, each cell individually contained an F-actin cortical ring producing a double-banded appearance and there was a diffuse connection between the double bands, especially at the Y-junctional regions of three adjacent cells. An apparent fusion of the microfilament bundles between cells with absence of the double-banded structure in TNF-α–treated corneal endothelia. The F-actin cytoskeletal organization was rearranged by TNF-α, but this was almost reversed by SA3K cotreatment.

SA3K Blockage of TNF-α–Induced Disassembly

Anti–α-tubulin antibody immunostaining revealed an intricate microtubule fibrillary structure (Fig. 5). Under control conditions, microtubules were evident throughout the cell cytoplasm, extending toward the cell periphery. In addition, microtubules were present in the nucleus. In contrast, TNF-α induced loss of fibrillary extensions and microtubule disassembly. SA3K cotreatment blocked TNF-α–induced microtubule changes. 

Figure 5

 

SA3K blocks microtubule disassembly induced by TNF-α. Corneal endothelia were incubated with TNF-α (20 ng/mL) in the presence or absence of SA3K (100 nM, 200 nM) for 24 hours. Microtubules were stained with anti–α-tubulin antibody. The pattern revealed an intricate fibrillary structure, extending toward the cell periphery throughout the whole cell. Treatment with TNF-α induced the loss of the fibrillar extensions and resulted in the disassembly of microtubules. TNF-α–induced disassembly of the microtubules was blocked by cotreatment with SA3K.

Suppression by SA3K of TNF-α–Induced MLC Phosphorylation

To determine how SA3K protects against TNF-α–induced disruption of AJC and microtubule organization, Western blot analysis determined if it changed MLC and p-MLC expression levels. Exposure to 20 ng/mL TNF-α for 24 hours increased MLC phosphorylation. The overall abundance of MLC was not affected by TNF-α in the absence or presence of SA3K (Fig. 6A) whereas cotreatment with SA3K suppressed TNF-α–induced rises in p-MLC expression (Fig. 6B). 

Figure 6

 

SA3K suppresses TNF-α–induced MLC phosphorylation. Corneal endothelia were treated with TNF-α (20 ng/mL) in the presence or absence of SA3K (100 nM, 200 nM) for 24 hours. MLC abundance was not affected by TNF-α in the presence or absence of SA3K based on Western blot analysis (A). Reversal by SA3K of TNF-α–induced increases in p-MLC abundance (B). Data are means ± SE from three independent experiments. *P < 0.05, **P < 0.01 for the indicated comparisons (ANOVA followed by Dunnett's test).

Discussion

The corneal endothelium is able to maintain constant stromal hydration despite very substantial age-related declines in cell density. Adult human eyes have an approximate cell density of 2500 cells/mm², which declines approximately 0.5% per year, Nevertheless, stromal hydration is maintained as long as the cell density is greater than 700 cells/mm. ^22,35 In clinical practice numerous cases are encountered involving declines in corneal transparency. They can be caused by endothelial decompensation resulting from allograft rejection, inflammation, and surgical trauma. Currently, there are no pharmacologic interventions to overcome endothelial dysfunction, leaving corneal transplantation as the only viable option. This limitation accounts for the approximately 40,000 procedures performed annually in the United States, but it is not a risk-free surgery. ³⁶ In this study, we found in the isolated rabbit corneal endothelium that SA3K can suppress TNF-α–induced increases in paracellular FITC-dextran permeability. This protective effect was accompanied by blockage of the disruptive effects of this proinflammatory cytokine on ZO-1 and VE-cadherin distribution. TNF-α also induced both F-cytoskeletal actin rearrangement and microtubule disassembly while with SA3K neither of these changes occurred. Furthermore, TNF-α elicited a dramatic increase in MLC phosphorylation, whereas SA3K significantly suppressed this rise. Our findings suggest that SA3K can protect the corneal endothelium from inflammatory mediator-induced declines in its dehydrating function. This finding suggests that decompensated endothelial cell function may be reversed leading to a decline in the need for corneal transplantation. 

SA3K Maintained Endothelium Barrier Integrity Disrupted by TNF-α

Under control conditions, ZO-1 distribution was not continuous around the endothelial cell membrane. There were gaps which preferentially appeared at the Y-junctions between three adjacent cells, which is consistent with the corneal endothelium being a leaky cell layer owing to the presence of electrically low resistance tight junctions. ³⁷ In contrast, VE-cadherin expression was continuous at the cell borders. TNF-α induced ZO-1 and VE-cadherin dislocation from the interfaces of neighboring endothelial cells accompanied by increases in FITC-dextran permeability. These changes are supportive of TNF-α–induced barrier function disruption. Furthermore, ZO-1 and VE-cadherin disruption from the borders of endothelial cells was not accompanied by a decrease in the total amount of ZO-1 and VE-cadherin detected by Western blot analysis and RT-PCR. The lack of a correlation between the total cellular expression of TJ-associated proteins and their junctional localization was previously described. In other epithelial cells or endothelial cells models, most studies revealed that the abundance of TJs or AJs did not change, whereas ZO-1, occludin, claudin, or VE-cadherin underwent dislocation. ^21,38–40 Nevertheless, there are a few reports indicating that barrier integrity disruption was accompanied by declines in ZO-1 expression. ^38,39 Our findings further confirmed that TJs and AJs structural integrity maintenance is more important for preserving barrier permeability than ZO-1 and VE-cadherin abundance. 

F-actin filament contraction and stress fiber formation in some other studies ⁴¹ is associated with TJ and AJ component redistribution, which contributed to barrier function disruption. Accordingly, we determined if such changes were associated with TNF-α–induced barrier function disruption. Indeed, our results show that TNF-α disrupted F-actin structure since apparent microfilament bundle fusion occurred between cells along with disappearance of the normal F-actin double-band appearance. Previous studies in the corneal endothelium demonstrated that TNF-α–induced barrier integrity loss occurred through microtubule disassembly. Furthermore, microtubule stabilization effectively blocked this disruptive response to this proinflammatory cytokine. ^8,9 We also showed in the rabbit corneal endothelium that SA3K protects against losses in barrier integrity through suppression of microtubule disassembly. 

Dependence of SA3K Endothelial Barrier Function Protection on MLC Phosphorylation Regulation

TNF-α induces barrier dysfunction through a variety of mechanisms including disruption of TJs proteins. Exposure of pulmonary microvessel endothelial monolayers to TNF-α–induced changes in AJs protein and F-actin cytoskeletal localization. These effects caused microtubule disassembly, and led to MLC phosphorylation through PKC-mediated myosin light chain kinase (MLCK) activation. ⁴² Similarly in other studies on the corneal endothelium, RhoA GTPase, ⁴³ p38 signal pathway activation were contributory to these changes. ²⁰ Furthermore, another change contributing to barrier function disruption included MLC phosphorylation by MLCK. Its phosphorylation elicited actomyosin contraction and formation of F-actin stress fibers. Accordingly, these responses contributed to barrier function disruption. ⁴ In accordance with the inhibitory effects of SA3K on vascularization, inflammation, fibrosis, and oxidative stress induced responses in the retina and cornea, it also reduced retinal vascular leakage in an oxygen-induced retinopathy model and also prevented hypoxia-induced declines in occludin ³⁰ through blocking Wnt signaling pathway activation. ^32,34 In this study, TNF-α induced p-MLC formation, while SA3K inhibited this effect. Additional studies on the corneal endothelium are needed to further clarify how SA3K protects against TNF-α–induced barrier function disruption. 

Speculations About How SA3K Protects Against Endothelial Barrier Function Disruption

Kallistatin competes with TNF-α binding to its cognate receptor in its heparin-binding domain. ²⁵ As SA3K is also a heparin-binding protein and it inhibits VEGF binding to its cognate receptor through competing with heparin binding, ²⁹ it is possible that SA3K protects against TNF-α endothelial barrier function disruption through the same mechanism. Recently, it was suggested that SA3K down-regulates the expression of proinflammatory factors such as TNF-α and Intercellular Adhesion Molecule 1 (ICAM-1). ²⁹ Further studies are needed to determine in vivo whether SA3K protects against declines in corneal endothelial barrier function through decreasing TNF-α expression. 

In conclusion, SA3K protects the rabbit corneal endothelium against TNF-α–induced disruption of barrier structure and function. SA3K maintains the integrity of AJC through suppression of MLC phosphorylation. These results suggest that SA3K application may provide a new approach for improving corneal endothelial function by suppressing declines in tight junctional barrier function in various disorders. 

Acknowledgments

The authors thank Peter S. Reinach from the State University of New York (SUNY) for his critical reading and careful revision of the manuscript. 

Supported by National Basic Research Program of China (Project 973) Grant 2011CB504606; National Natural Science Foundation of China (Beijing, China) Grants 81100638 and 30973249; and Science and Technology Project of Xiamen Grant 3502Z20124039. 

Disclosure: J. Hu, None; Z. Zhang, None; H. Xie, None; L. Chen, None; Y. Zhou, None; W. Chen, None; Z. Liu, None 

Footnotes

 JH and ZZ contributed equally to the work presented here and should therefore be regarded as equivalent authors.