The retinal pigment epithelium (RPE) forms the outer blood-retinal barrier (oBRB) and is the prime target of early age-related macular degeneration (AMD). C-reactive protein (CRP), a serum biomarker for chronic inflammation and AMD, presents two different isoforms, monomeric (mCRP) and pentameric (pCRP), that may have a different effect on inflammation and barrier function in the RPE. The results reported in this study suggest that mCRP but not pCRP impairs RPE functionality by increasing paracellular permeability and disrupting the tight junction proteins ZO-1 and occludin in RPE cells. Additionally, we evaluated the effect of drugs commonly used in clinical settings on mCRP-induced barrier dysfunction. We found that a corticosteroid (methylprednisolone) and an anti-VEGF agent (bevacizumab) prevented mCRP-induced ARPE-19 barrier disruption and IL-8 production. Furthermore, bevacizumab was also able to revert mCRP-induced IL-8 increase after mCRP stimulation. In conclusion, the presence of mCRP within retinal tissue may lead to disruption of the oBRB, an effect that may be modified in the presence of corticosteroids or anti-VEGF drugs.
- blood-retinal barrier
- C-reactive protein
- macular degeneration
- tight junctions
age-related macular degeneration (AMD), a retinal degenerative disease, is the leading cause of irreversible vision loss among elderly populations in developed countries and remains a major public health problem (19, 32). The retinal pigment epithelium (RPE) monolayer is believed to be the primary target of early disease. AMD presents RPE cell abnormalities, disruption of the outer blood-retinal-barrier (oBRB), and degeneration of photoreceptors, which require a normally functioning RPE to survive (22, 37). Although the mechanisms of AMD are not yet clearly understood, several pathogenic pathways have been proposed, including RPE cell dysfunction, inflammation, and oxidative stress (6). Because the presence of larger and numerous drusen, the hallmark extracellular deposits of AMD, in the macula is a common risk factor for AMD (10), analysis of drusen components has aided the understanding of AMD pathogenesis. Components of the complement system and inflammatory proteins are present in drusen, providing further evidence that local inflammation and immune-mediated processes play a central role in AMD pathogenesis (5, 15).
C-reactive protein (CRP) is the prototypical acute-phase reactant and an active regulator of the innate immune system. CRP is a serum biomarker for chronic inflammation and AMD (27, 29, 34). Indeed, CRP has been identified in ocular drusen and other sub-RPE deposits (2, 26), as well as in the choroid. Notably, a common polymorphism in the gene encoding complement factor H (CFH) (Y402H) has been reported as a risk factor for AMD (7, 11, 12), and recent studies indicate that the AMD-associated CFH variant CFH-Y402H in the SCR7 domain of CFH binds more weakly to CRP than the wild-type protein (16).
In plasma, CRP typically exists as a cyclic, disk-shaped pentamer (pentameric CRP, pCRP) composed of five noncovalently linked subunits of 23 kDa (35). However, pCRP can undergo dissociation into its subunits, thereby acquiring distinct biological functionality (14). Oxidative stress and bioactive lipids from activated or damaged cells can dissociate pCRP into its 23-kDa subunits (8, 14, 21, 30) through a mechanism that is dependent on lysophosphatidylcholine exposure after phospholipase A2 activation (30). This alternative conformation of CRP, termed monomeric CRP (mCRP), has different antigenicity-expressing neoepitopes than pCRP and represents the tissue-based insoluble form of CRP. Unlike pCRP, mCRP displays a proinflammatory and prothrombotic phenotype in several cell types (17, 18, 25, 33). Recently, mCRP has been localized within retinal tissues (21), and we have recently showed that mCRP increases cytokine expression in ARPE-19 cells (23). However, its contribution to barrier breakdown remains unclear.
The oBRB is formed by tight junctions (TJs) between RPE cells that confer a high degree of control of solute and fluid permeability that govern the exchange of metabolites and waste products between the choriocapillaris and the retina to maintain an appropriate environment deemed necessary for adequate visual function. Changes in the integrity of the oBRB, and more concisely alterations in TJ protein levels such as zonula occludens-1 (ZO-1) and occludin, may affect the neurovascular unit (9).
Various families of drugs are used to treat retinal inflammatory diseases such as AMD. Wet AMD is currently treated with anti-VEGF therapy, which blocks the development of new blood vessels and leakage from abnormal vessels within the eye, and although considerable efforts have been done in recent years there is still no treatment for dry AMD. Corticosteroids are also used to treat intraocular inflammation because they inhibit leukocyte infiltration at the site of inflammation and interfere with mediators of inflammatory response. Immunosuppressive drugs and biological therapies such as tocilizumab, an antibody against the soluble and transmembrane receptor of IL-6, are also used to treat intraocular inflammation and could be potential candidates for the treatment of AMD because IL-6 has been positively related to progression of AMD (29).
In the present work, we studied the effect of CRP isoforms on barrier properties and the ability of various drugs to modulate mCRP-induced barrier disruption in RPE cells.
MATERIALS AND METHODS
High-purity human pCRP (Calbiochem) was stored in 10 mmol/l Tris·HCl and 140 mmol/l NaCl buffer (pH 8.0) containing 2 mmol/l CaCl2 to prevent spontaneous formation of mCRP from the native pentamer. Monomeric CRP was obtained by urea chelation from purified human CRP as previously described (25). Briefly, pCRP at 1 mg/ml was chelated with 10 mmol/l EDTA and incubated in 8.0 mol/l urea for 6 h at 37°C. Urea was removed via dialysis against low-ionic-strength TBS (0.01 mol/l Tris·HCl and 0.05 mol/l NaCl, pH 7.3). Monomeric CRP concentration was determined using a bicinchoninic acid protein assay. The filtered solution was stored at 4°C. Pentameric CRP was also dialyzed with TBS to remove sodium azide.
ARPE-19, a spontaneously arising human retinal pigment epithelium cell line, was obtained from the American Type Culture Collection. ARPE-19 cells were cultured in a 50:50 mixture of DMEM and Ham’s F-12 [PAA (now GE Healthcare)] supplemented with 10% fetal bovine serum (FBS, PAA), 2 mM l-glutamine (PAA), 100 U/ml penicillin (PAA), 0.1 mg/ml streptomycin (PAA), and 1 mM sodium pyruvate (Sigma-Aldrich) in a humidified incubator at 37°C in 5% CO2. Cells were passed every 4 to 5 days by trypsinization. Human primary retinal pigment epithelial (h-RPE) cells (Lonza) were maintained in Retinal Pigment Epithelial Cell Basal Medium (Lonza) containing supplements (l-glutamine, GA-1000, and bFGF, Lonza) and fed triweekly. h-RPE were passed weekly by trypsinization and used until passage 6.
For cytokine stimulation and immunofluorescence, cell suspensions of 1 × 105 cells/ml were seeded onto 24-well tissue culture plates and coverslides, respectively, and incubated overnight. Cells were then washed gently with PBS to remove cell debris and starving fresh culture medium with 0.6% FBS was introduced for 18 h. Cells were then stimulated with or without CRP isoforms or control buffer (mCRP buffer alone) for 24–48 h. To test the ability of various drugs to modulate the proinflammatory response induced by mCRP, cells were treated with methylprednisolone (1 × 10−6 to 1 × 10−8 M), cyclosporine A (10–50 ng/ml), tocilizumab (10–200 ng/ml), and bevacizumab (0.5–2.5 mg/ml) either simultaneously or after mCRP treatment.
Cell proliferation and cell viability.
The cell proliferation test was based on the ready-to-use cell proliferation reagent 4-[3-(4iodophenyl)-2-(4-nitrophenyl)-2H-5-tetrazolio]-1,3-benzene disulfonate (WST-1; Roche Diagnostics, Indianapolis, IN). ARPE-19 cells were seeded in 96-well tissue culture plates at a concentration of 1 × 104 cells/well in medium containing 10% FCS and, once the monolayer was formed, cells were cultured for 3 wk in medium containing 1% FBS, and the medium was changed every 3 days. Cells were then treated with various concentrations of CRP isoforms or control buffer (24–48 h). After stimulation, 10 μl of WST-1 reagent were added to the medium in each well. Cells were incubated in a humidified atmosphere at 37°C in 5% CO2/95% air for 2 h, the multititer plate was shaken thoroughly for 1 min, and absorbance was read at 450 nm. The background absorbance was measured in wells containing only the dye solution and culture medium. Cell proliferation data were obtained in duplicate from at least five experiments in separate 96-well plates. The mean optical density values corresponding to the untreated controls were taken as 100%. Results were expressed as percentage of the optical density of treated cells relative to that of untreated controls.
The paracellular permeability of ARPE-19 and h-RPE monolayers was assessed by measuring the passive permeation of FITC-dextran (40 kDa, Sigma-Aldrich) across confluent cells grown on filters. For monolayer culture, cells were seeded at 2 × 105 cells/cm2 onto Transwell filters (12 mm diameter, 0.4 μm pore size; Costar) and, once the monolayer was formed, cells were cultured for 3 wk in medium containing 1% FBS (ARPE-19 cells) or serum-free medium (h-RPE cells), and the medium was changed every 3 days.
At day 19, cells were treated with CRP isoforms or control buffer for 48 h. At day 21, 500 μg/ml FITC-dextran were added to the apical compartment of the chamber, and samples (200 μl) from the basal medium (lower chamber) were collected 90 min after addition of FITC-dextran. Absorbance of basal and apical medium samples was measured at 485 nm excitation and 528 nm emission in a microplate reader (SpectraMax Gemini; Molecular Devices). Each condition was assayed in triplicate and repeated in at least three independent experiments. The diffusion rate was expressed as a percentage and calculated as follows: (amount of dextran lower chamber) × 100/(amount of dextran upper chamber).
Measurement of transepithelial electrical resistance.
Transepithelial electrical resistance (TEER) was measured using a commercial electrical resistance system (Millicell; Millipore) in ARPE-19 and h-RPE monolayers grown on Transwell filters as described above. TEER values were calculated by subtracting the value of a blank (Transwell filter without cells) and expressed as resistance/area (Ω/cm2) relative to the resistance/area of the control at day 21. Measurements were repeated at least three times for each filter, and each experiment was repeated three times using two filters.
Distribution of TJs ZO-1 and occludin in ARPE-19 and h-RPE monolayers was examined by immunofluorescence. Cells grown on coverslides or on Transwell filters were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature (RT) and washed three times with PBS. Cells were then permeabilized with 0.2% Triton X-100 in PBS for 15 min and blocked twice with filtered 1% BSA. Cells were then incubated with primary antibody anti-ZO-1 or anti-occludin (Invitrogen) overnight at 4°C. After cells were washed three times with PBS, they were incubated with secondary antibody Alexa-Fluor anti-mouse 488 IgG for 1 h at RT. Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI). Controls were stained with secondary antibodies only. Stained cells were washed and covered with Prolong Gold antifade reagent (Life Technologies). Images of immunostained cells were recorded on a Zeiss Axio Vert fluorescence microscope equipped with an Apotome imaging system (Carl Zeiss AG).
Quantitative real-time PCR.
One microgram of RNA from ARPE-19 cell lysates extracted using an RNeasy isolation kit was reverse-transcribed with M-MuLV reverse transcriptase for 1 h at 42°C in a total volume of 25 μl. TPJ1 (ZO-1), and OCCLN (occludin) mRNA levels were analyzed via real-time PCR. Assays on demand (Applied Biosystems) were used for ZO-1 (Hs01551861m1), and occludin (Hs00170162m1). Human GAPDH was used as the endogenous control. Quantitative real-time PCR was performed on an ABI PRISM 7000 Detection System. The comparative CT method was used to determine the relative fold changes in gene expression.
Enzyme-linked immunosorbent assay.
Supernatants from CRP-stimulated ARPE-19 cells were collected and centrifuged (1,000 g for 5 min) to remove particulates and stored at −70°C until further analysis. IL-8 was measured using human ELISA development kits (Ready Set Go eBiosciences). Each stimulation experiment was repeated five times.
Results are expressed as means ± SE. Student’s t-test or ANOVA as appropriate were used to determine statistical significance between treatments. A value of P < 0.05 was considered significant. All calculations were performed using GraphPad Prism (GraphPad Software, San Diego, CA).
Monomeric but not pentameric CRP affects cell viability.
The effect of CRP isoforms on ARPE-19 cell viability was measured with the WST-1 assay. Exposure to mCRP for 24 or 48 h significantly decreased cell viability in a dose-dependent manner (Fig. 1) compared with untreated cells. At 24 h this effect was statistically significant only at a concentration of 50 μg/ml of mCRP (P < 0.05) and at 48 h at concentrations equal or higher than 10 μg/ml of mCRP (P < 0.05). In contrast, pCRP did not affect cell viability at any tested dose.
Monomeric CRP increases the paracellular permeability and reduces TEER of RPE cells.
To determine the effect of CRP isoforms on RPE barrier integrity, we assessed the effect of CRP on the paracellular permeability of ARPE-19 and h-RPE monolayers by measuring the transepithelial diffusion rate of FITC-dextran (40 kDa). As shown in Fig. 2A, ARPE-19 cells exposed to 5 and 10 μg/ml of mCRP showed significantly greater diffusion rate of FITC-dextran (35%) compared with untreated cells (P < 0.05). Conversely, addition of 10 μg/ml of pCRP did not show any effect on FITC-dextran diffusion rate. Similarly, in h-RPE cells, 10 μg/ml of mCRP showed a significantly greater diffusion rate of FITC-dextran compared with untreated h-RPE cells (Fig. 2B, P < 0.05). On the other hand, addition of pCRP did not show any effect on FITC-dextran diffusion rate in h-RPE cells.
We next determined the effect of CRP isoforms on TEER in both ARPE-19 and h-RPE cells. In line with our findings on paracellular permeability, mCRP significantly reduced TEER (P < 0.05) in both RPE cell lines, whereas pCRP had no effect (Fig. 2, C and D).
Monomeric CRP alters expression of ZO-1 and occludin in RPE cells.
We next evaluated the effect of CRP on the distribution of ZO-1 and occludin in RPE cells grown on Transwell inserts. Figure 3 shows that ZO-1 and occludin expression was continuous and regular in untreated RPE cells (both ARPE-19 and h-RPE cells). Expression was defined at the membrane of the cells, and it was especially intense at the cell-cell border. However, exposure of RPE cells to mCRP (10 μg/ml) for 24 h resulted in a markedly disturbed distribution of occludin, and especially ZO-1. The abnormal distribution of ZO-1 and occludin was manifested as fragmented staining and diffuse cytoplasmic distribution. In contrast, addition of 10 μg/ml of pCRP did not alter ZO-1 and occludin expression, and the distribution was similar to that observed in untreated samples.
The effect of CRP isoforms on ZO-1 and occludin expression in ARPE-19 cells was also tested at the RNA level by quantitative RT-PCR. As observed in Fig. 4, mCRP resulted in a significant decrease in ZO-1 (Fig. 4A) and occludin (Fig. 4B) mRNA expression (P < 0.05) at a concentration of 10 μg/ml. By contrast, pCRP did not modulate ZO-1 and occludin expression.
Pharmacological treatment can inhibit mCRP-induced cytokine expression.
We recently demonstrated that mCRP at doses equal to or greater than 5 μg/ml result in increases in IL-8 expression in ARPE-19 cells (23). Thus, we assessed the effect of various drugs currently used in the clinical setting on mCRP-induced IL-8 expression in ARPE-19 cells. Cells were simultaneously treated with mCRP (10 μg/ml) and the various drugs for 24 h, and levels of secreted IL-8 were determined by ELISA. As expected, IL-8 secretion was significantly elevated in ARPE-19 cells treated with mCRP (10 μg/ml). Noteworthy, different doses of methylprednisolone and bevacizumab prevented mCRP-induced IL-8 secretion (Fig. 5). However, neither cyclosporine A (10–100 ng/ml) nor tocilizumab (10–200 μg/ml) had any effect on mCRP-induced IL-8 secretion at any tested dose.
Bevacizumab but not methylprednisolone reverts IL-8 production after mCRP stimulation.
To test whether methylprednisolone and bevacizumab were also able to revert mCRP-induced cytokine expression, the drugs were added to the cell cultures 24 h after mCRP (10 μg/ml) treatment and then cultured for an additional 24 h. Although methylprednisolone prevented the mCRP-induced IL-8 increase (Fig. 5), none of the tested concentrations of methylprednisolone significantly reverted the inflammatory phenotype of ARPE-19 cells induced by the presence of mCRP (Fig. 6). Nevertheless, bevacizumab partially reverted up to 60% of the mCRP-induced IL-8 secretion at concentrations of 1.25 or 2.5 mg/ml, even after stimulating cells with mCRP for 24 h.
Methylprednisolone and bevacizumab prevent mCRP-induced barrier disruption.
Because methylprednisolone and bevacizumab prevented mCRP-induced IL-8 secretion in ARPE-19 cells, we evaluated whether these drugs were also able to prevent mCRP-induced barrier disruption. For this purpose, the paracellular permeability of ARPE-19 monolayers treated with mCRP (10 μg/ml) and the above mentioned drugs, was determined by measuring the transepithelial diffusion rate of FITC-dextran (40 kDa) through ARPE-19 monolayers (Fig. 7). Treatment with 1 × 10−7 M methylprednisolone or 2.5 mg/ml bevacizumab significantly reduced the increased diffusion rate of FITC-dextran across ARPE-19 monolayers induced by mCRP. Indeed, cells treated with these drugs and mCRP showed a similar diffusion rate compared with untreated cells. Interestingly, the addition of methylprednisolone or bevacizumab alone did not modify ARPE-19 basal permeability (P > 0.05).
Methylprednisolone and bevacizumab prevent ZO-1 and occludin disruption due to monomeric CRP.
Finally, we assessed the ability of methylprednisolone and bevacizumab to restore TJs in ARPE-19 cells treated with mCRP. As can be seen in Fig. 8, the diffuse expression of ZO-1 and occludin in the cytoplasm when cells were exposed to mCRP (10 μg/ml) was not observed with addition of the anti-VEGF antibody or corticosteroid. In fact, ZO-1 and occludin intensity and distribution in cells treated with mCRP and the two drugs were comparable to those of untreated cells.
In this study, we show that mCRP contributes to RPE disruption in vitro by promoting paracellular permeability and disorganizing TJ distribution in RPE cells. Moreover, we also demonstrate that methylprednisolone and bevacizumab can prevent mCRP-induced cytokine release and barrier disruption in ARPE-19 cells.
Several epidemiological studies have shown that high levels of circulating CRP are associated with AMD (28, 29, 34). Nevertheless, data regarding the contribution of CRP to ocular inflammation are scarce, and the pathophysiological importance of CRP in AMD is far from being fully understood. Clinically, the threshold of CRP plasma concentration associated with AMD is >2.5 μg/ml (28, 29, 34), whereas levels above 10 μg/ml are usually attributed to other causes such as acute infection or inflammation. Therefore, we used in our experiments concentrations associated with a risk for AMD.
The functional integrity of the oBRB, composed by the RPE, critical for the maintenance of the specialized environment for the neural retina, is dependent on the structures of TJs, including the transmembrane protein occludin, claudins, and junctional adhesion molecules. Zonula occludens (ZO-1, ZO-2, ZO-3) interact with transmembrane proteins on the cytoplasmic side of the cell membrane and serve to anchor them to the actin cytoskeleton (3). To investigate whether CRP isoforms could affect barrier integrity we studied the functional and structural integrity of cultured epithelium monolayers by measuring diffusion of FITC-dextran and distribution and expression of TJs. Exposure to mCRP but not pCRP significantly increased the diffusion rate of FITC-dextran compared with that of untreated cells, suggesting that mCRP could compromise the barrier function of the RPE monolayer. Notably, these mCRP concentrations were also able to disturb the distribution and expression of ZO-1 and occludin, which could explain the diffusion rate of FITC-dextran induced by mCRP. However, TJs are highly complex structures, and the relative contribution of the various proteins to the function of the RPE barrier remains unclear. Thus, in addition to the disorganization of occludin and ZO-1, mCRP might also modulate other TJ proteins or other pathways associated with RPE permeability. Interestingly, in a previous study, Kuhlmann and collaborators (20) showed that CRP could induce blood-brain barrier disruption by increasing permeability and disrupting TJ distribution.
Several therapeutic approaches, including synthetic glucocorticoids, immunosuppressants, and biologics that target inflammatory and angiogenic molecules such as anti-VEGF therapies, are currently employed to treat a variety of ocular diseases with an inflammatory basis. Herein we assessed the ability of various drugs to prevent mCRP-induced cytokine expression and barrier disruption in ARPE-19 cells. Interestingly, we observed that methylprednisolone, a synthetic glucocorticoid, and bevacizumab, an anti-VEGF, prevented the effects of mCRP, whereas cyclosporine A and tocilizumab had no effect. The increased IL-8 secretion induced by mCRP was prevented when methylprednisolone or bevacizumab were simultaneously added to the cultures. Similarly, the greater diffusion rate of FITC-dextran and TJ disorganization induced by mCRP was also prevented when cells were treated with these two drugs upon mCRP stimulation. Moreover, unlike methylprednisolone, bevacizumab was also able to revert the increased secretion of IL-8 when added 24 h after mCRP treatment.
How methylprednisolone and bevacizumab prevent mCRP-induced cytokine release and barrier disruption is not clear. First, the receptors that mediate mCRP activities have not been fully characterized. In human neutrophils, mCRP binds FcγRIII (CD16) (17); however, in endothelial cells and epithelial cells, functional blockade of CD16 showed only a slight attenuation of mCRP-induced activation (18, 24). Instead, it has been found that mCRP can activate endothelial cells through its interaction with lipid raft microdomains (13). Therefore, mCRP may also interact with ARPE-19 cells through lipid raft microdomains. Because glucocorticoids interact with lipid rafts (31, 36), it could then be possible that this interaction hampers the interplay of mCRP with ARPE-19 cells, further averting the proinflammatory effects of mCRP. A similar mechanism might occur with bevacizumab because VEGFR2 is present in lipid rafts (4). It appears unlikely that bevacizumab prevents mCRP-induced proinflammatory effects directly via VEGF inhibition, as mCRP does not seem to induce VEGF release in ARPE-19 cells (unpublished data). Given that mCRP effects are mediated by p38 MAPK and JNK pathways (24), it could be plausible that methylprednisolone and bevacizumab may play a role in these pathways that prevent mCRP proinflammatory activities.
The main limitation of our study is the use of ARPE-19 cells, a spontaneously arising RPE cell line. Although ARPE-19 cells exhibit barrier functions mediated by TJs and they secrete cytokines, they exhibit reduced transepithelial resistance and limited formation of the hexagonal shape, a distinctive characteristic of the RPE (1). Hence, special consideration needs to be taken with the applicability of the current results to an in vivo setting in which RPE cells could behave in a different manner. Nevertheless, despite these weaknesses, ARPE-19 cells are commonly used for studying oxidative stress and cell signaling in AMD because they exhibit features of aged RPE. Indeed, to overcome this limitation, some of our key experiments were also validated using human primary RPE cells, and therefore, the results reported herein appear valid as a first attempt to evaluate the role of mCRP over RPE cells in an in vitro setting.
In summary, the current work shows that mCRP contributes to oBRB disruption. Given that mCRP has been detected in drusen of patients with AMD (21), our results raise the interesting hypothesis that this molecule may play a role in the inflammatory responses of the oBRB. Thus, mCRP could eventually induce alterations in the RPE favoring a chronic inflammation state in the RPE-Bruch membrane complex. Such mCRP-mediated barrier breakdown could allow passage of inflammatory cells into the retina, contributing to chronic inflammation and accelerating tissue damage. Nevertheless, further research with AMD animal models is warranted to comprehensively describe the specific contribution of mCRP to the disease etiology, and to test the therapeutic potential of compounds that either prevent CRP dissociation or block mCRP proinflammatory activities such as corticosteroids or anti-VEGF therapies.
Support for this study was provided by Ministry of Science and Innovation of Spain, Instituto de Salud Carlos III, Fondo de Investigación Sanitaria Grant RD12/0034.
No conflicts of interest, financial or otherwise, are declared by the authors.
B.M., conceived and designed the research; B.M. and A.P.M. performed experiments; B.M. and A.P.M. analyzed data; B.M. and V.L. interpreted results of experiments; B.M. prepared figures; B.M. drafted manuscript; B.M., J.Z.-V., M.M., and J.M. edited and revised manuscript; B.M., A.P.M., V.L., J.Z.-V., M.M., A.A., and J.M. approved final version of manuscript.
- Copyright © 2017 the American Physiological Society