Doxycycline Inhibition of Interleukin-1 in the Corneal Epithelium

Abstract

purpose. To evaluate the effect of doxycycline on the regulation of interleukin (IL)-1 expression and activity in human cultured corneal epithelium.

methods. Human corneal limbal epithelium (HLE) was cultured from explants prepared from limbal rings of donor corneas. Primary cultured limbal epithelial cells were treated with either 10 μg/ml lipopolysaccharide (LPS), LPS with 10 μg/ml doxycycline, or LPS with 0.1 mg/ml methylprednisolone (MP) for 24 hours. The intracellular and supernatant protein amounts of IL-1α, the precursor and mature forms of IL-1β, IL-1 receptor antagonist (IL-1 RA), and the intracellular level of IL-1β–converting enzyme (ICE) were measured with enzyme-linked immunosorbent assays (ELISAs). Western blot analysis was performed to evaluate IL-1 RA protein. mRNA steady state amounts were determined by RNase protection assay (RPA) for IL-1α, IL-1β, IL-1 RA, and ICE.

results. LPS increased the mRNA and protein amounts of intracellular and released IL-1α, mature IL-1β, and IL-1 RA. Doxycycline inhibited the LPS-induced IL-1β increase in the mRNA and protein amounts in the corneal epithelium and upregulated the expression of the anti-inflammatory IL-1 RA protein. In addition, doxycycline reduced the steady state level of the cellular ICE protein but did not affect the level of ICE transcripts. IL-1β secreted to the conditioned media of HLE was functionally active in inducing matrix metalloproteinase (MMP)-1 and MMP-3 in cultured corneal fibroblasts. Doxycycline significantly decreased IL-1β bioactivity in the supernatants from LPS-treated corneal epithelial cultures. These effects were comparable to those induced by the corticosteroid, MP.

conclusions. Doxycycline can suppress the steady state amounts of mRNA and protein of IL-β and decrease the bioactivity of this major inflammatory cytokine. These data may partially explain the clinically observed anti-inflammatory properties of doxycycline. The observation that doxycycline was equally potent as a corticosteroid, combined with the relative absence of adverse effects, makes it a potent drug for a wide spectrum of ocular surface inflammatory diseases.

Keratitis sicca, the corneal epithelial disease that develops in dry eye, is among the most common and problematic conditions faced by ophthalmologists. In mild cases, it is associated with symptoms of irritation, redness, and blurred vision. In the more severe forms, sight-threatening corneal problems may develop, such as filamentary keratitis, corneal epithelial erosions, corneal stromal vascularization, and ulceration. The exact mechanism by which keratitis sicca develops has not been established. Our group has reported that inflammation may be the primary factor causing this condition. ¹ The proinflammatory cytokine interleukin (IL)-1 has been identified as a factor that may play a key role in the initiation and perpetuation of this inflammation. We have observed that as tear clearance from the ocular surface decreases, the concentrations of both isoforms of the proinflammatory cytokine IL-1, IL-1α, ² and IL-1β (Solomon et al., unpublished results, 2000), increase in the tear fluid. The IL-1 gene family is a group of potent cytokines that function as major mediators of inflammation and immune response. ³ This family is composed of three forms: two proinflammatory forms, IL-1α and IL-1β, each having a precursor form, and an anti-inflammatory form, IL-1 receptor antagonist (IL-1 RA). 

Recent data suggest that the IL-1 cytokines play an important role in the regulation of inflammation and wound healing on the ocular surface. IL-1β was found in the epithelium, stroma, and endothelium of the cornea, at the mRNA and protein levels. ⁴ Type 1 receptor for IL-1 is expressed in stromal fibroblasts. ⁵ Both IL-1α and IL-1β have been found to modulate matrix metalloproteinase (MMP) expression by corneal stromal fibroblasts ⁶ and their own synthesis in keratocytes, ⁷ to regulate apoptosis of keratocytes in response to corneal epithelial wounding, ⁸ and to upregulate hepatocyte growth factor and keratocyte growth factor in corneal fibroblasts. ⁴ These findings make IL-1 a prime candidate for inducing ocular surface disease, especially the chronic subclinical ocular surface inflammation of dry eye. 

Traditional therapies for keratoconjunctivitis have consisted of artificial tears and aqueous-conserving therapies, such as punctal occlusion. Although these therapies transiently improve irritation symptoms, they are often ineffective in treating the severe complications of dry eye, such as recurrent corneal epithelial erosion and corneal stromal ulceration. Therapies targeting the underlying inflammatory environment of the ocular surface would represent a major improvement in the management of these conditions and would have a major clinical impact. Consistent with the concept that inflammation is a key feature in the pathophysiology of keratitis sicca is the finding that both aqueous tear deficiency and meibomian gland disease are effectively treated with the corticosteroid methylprednisolone (MP). ^{9 10} Unfortunately the long-term use of topical corticosteroids is limited by potential sight-threatening side effects, such as glaucoma and cataract. Therefore, there is a clinical need for nontoxic steroid-sparing anti-inflammatory therapies that target IL-1 expression in the corneal epithelium. 

Systemically administered tetracycline antibiotics have long been recognized as effective therapies for ocular surface inflammatory diseases. The semisynthetic tetracycline, doxycycline, has been reported to successfully treat the common dry eye condition acne rosacea, ¹¹ as well as recurrent corneal erosions ¹² and phlyctenular keratoconjunctivitis. ¹³ We hypothesized that one of the mechanisms of action of doxycycline in dry eye is the downregulation of the IL-1–mediated inflammatory cascade in the corneal epithelium. Therefore, the purpose of this study was to evaluate the effect of doxycycline on the regulation of IL-1 expression and activity in the human corneal epithelium. 

Methods

Reagents

Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), HEPES buffer, F12 (Ham’s), were from Life Technologies (Rockville, MD). Tissue culture plates were from Becton Dickinson (Franklin Lakes, NJ). Cholera toxin subunit A, epidermal growth factor (EGF), hydrocortisone, LPS (derived from Serratia marcescens), doxycycline, and MP were from Sigma (St. Louis, MO). IL-β precursor, IL-1β–converting enzyme (ICE), and IL-β mature enzyme-linked immunosorbent assay (ELISA) kits were from Cistron (Pine Brook, NJ); IL-1 α and IL-1 RA ELISA kits were from R&D systems (Minneapolis, MN); and MMP-1 and MMP-3 ELISA kits were from Oncogene Research Products of Calbiochem (Cambridge, MA). RNA lysis and RNase protection kits were from Ambion (Austin, TX). IL-1 RA was from Genzyme (Cambridge, MA). The BCA protein assay kit was from Pierce (Rockford, IL). 

Culture of Human Corneal Limbal Epithelium

Human corneal limbal epithelium (HLE) was cultured from explants of human donor corneoscleral rims, provided by the Florida Lions Eye Bank at the Bascom Palmer Eye Institute. Each corneoscleral rim was trimmed, the endothelial layer and iris remnants were removed, and the tissue was treated with dispase for 15 minutes. Each rim was dissected into 12 equal parts, which were applied to six-well plastic dishes and covered with a drop of FBS overnight. The explants were cultured in supplemented hormonal epithelial medium (SHEM) containing equal amounts of DMEM and Ham’s F12 medium, supplemented with 5% FBS, 0.5% dimethyl sulfoxide, 2 ng/ml EGF, 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenium, 0.5 μg/ml hydrocortisone, 30 ng/ml cholera toxin A, 50 μg/ml gentamicin, and 1.25 μg/ml amphotericin B. Cultures were incubated at 37°C under 95% humidity and 5% CO₂. The medium was changed every 2 days. Cultures were maintained for 10 to 14 days until confluence and then switched to the serum-free medium described above, without FBS, for 24 hours before the additions of treatments. 

To demonstrate the effect of doxycycline on the corneal epithelium and to compare it with that of a corticosteroid, primary cultures of HLE were treated with 10 μg/ml bacterial LPS alone or in combination with either 0.1 mg/ml MP or 10 μg/ml doxycycline. These treatments were maintained for 24 hours. 

After a 24-hour treatment, the culture supernatant was collected from each well, centrifuged, and stored in −80°C until assayed by ELISA. Cell lysis solution, containing 50 mM Tris-HCl (pH 7.6), 300 mM NaCl, and 0.5% Triton X-100, was added to the cells for 3 hours, and the cellular protein was collected, centrifuged, and stored in −80°C until assayed by ELISA. In parallel cultures, the cells were subjected to lysis buffer (Direct Protect; Ambion), and total RNA was isolated for further assessment by RNase protection assay (RPA). The ELISA and RPA were targeted at determining the protein and mRNA levels, respectively, of IL-1α, IL-1β, IL-1 RA, and ICE. 

RPA Template Construction

Partial cDNAs for human IL-1α, IL-1β, IL-1 RA, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were prepared by reverse transcription–polymerase chain reaction (RT-PCR). PolyA+ RNA was isolated from cultured human corneal epithelial cells using oligo-dT–coated beads (Oligotex Direct mRNA Isolation System; Qiagen, Valencia, CA), according to the manufacturer’s instructions. RT was performed using 200 ng mRNA as template and gene-specific primers (see Table 1 ) were prepared to human IL-1α (gene accession, X02531), IL-1β (gene accession, X02532), IL-1 RA (gene accession, M63099), and GAPDH (gene accession, NM 002046), according to the manufacturer’s instructions (Superscript II Reverse Transcription Kit; Life Technologies). The resultant first-strand cDNA was used for PCR (PCR Kit, Life Technologies) using a gene-specific upstream primer and the same downstream primer-2 used for RT. An aliquot of the initial PCR reaction (except for the GAPDH probe that required only a single round of PCR) was reamplified using the same upstream primer and a third gene-specific primer, downstream primer-1. The primers used in this second amplification were designed with 12 nucleotide additions (four copies of a trinucleotide repeat containing a single deoxyuracil in each repeat) at their 5′ ends, to facilitate rapid cloning of the amplimers (pAMP1 System; Life Technologies). 

RPA Probes

RPA requires hybridization of sense mRNA to complementary radiolabeled antisense RNA. Subsequently, double-stranded mRNA-radiolabeled antisense RNA hybrids are treated with RNase specific for single-stranded RNA. Protected hybrids can then be resolved and quantified using gel electrophoresis and autoradiography. Radiolabeled antisense RNA was transcribed using the a kit (Maxiscript T7; Ambion) and labeling with [α-³²P]CTP (800 Ci/mmol). Plasmids were digested at a unique BamHI site upstream of the cloned cDNAs. RNA probes were generated for IL-1α, IL-1β, and IL-1 RA, and GAPDH. The GAPDH message is a housekeeping gene that is found to be expressed 10 to 20 times more than the messages for the cytokines measured in the RPA. The GAPDH probe was therefore transcribed to yield a specific activity 10 times less than that of the cytokine probes, to allow simultaneous detection of protected cytokine probe fragments as well as GAPDH probe fragments, given the range of sensitivity provided by the x-ray film. After transcription, probes were DNase treated to remove template DNA, and unincorporated nucleotides were removed using RNA Quick Spin columns (Roche Molecular Biochemicals, Indianapolis, IN). A template set containing DNA templates for ICE and GAPDH RNA probes, was purchased from PharMingen (San Diego, CA). 

RNase Protection Assay for IL-1α, IL-1β, IL-1 RA, and ICE

RNase protection assays were performed (Direct Protect System; Ambion) as described by the manufacturer. Briefly, cultured human corneal epithelial cells were resuspended in lysis buffer at approximately 10⁷ cells/ml. The cell lysis buffer included concentrated guanidine thiocyanate, which rapidly solubilizes cells and also rapidly inactivates ribonucleases. Assays were performed using 50 μl of cell lysate, 10⁵ cpm of each cytokine probe (specific activity, 5 × 10 ⁵ cpm/μg) and 4 × 10⁴ cpm of the GAPDH probe (specific activity, 5 × 10 ⁴ cpm/μg) for each sample. Samples were allowed to hybridize overnight at 37°C. They were then treated for 30 minutes at 37°C with RNase solution, after which the RNase was inactivated with proteinase K. Protected RNA fragments were precipitated and separated on a 6% polyacrylamide urea-Tris-base, boric acid, EDTA (TBE) sequencing gel. 

RPAs for IL-1α, IL-1β, and IL-1 RA were repeated four times on primary cultures derived from four different donor corneas. RPAs for ICE were repeated three times on primary cultures from three donor corneas. Autoradiographs from these gels were scanned and then analyzed using image analysis software (Gel-Pro; Media Cybernetics, Silver Spring, MD). The digitized data for each band was plotted, and the area under the curve for each peak was calculated with statistical software (GraphPad Prism; GraphPad Software, San Diego, CA). The value for each cytokine band was divided by the corresponding value of the GAPDH band in the same lane to calculate the relative mRNA amount for each gene. Results are shown as means ± SEM of relative mRNA amounts from three or four experiments. 

Immunodetection of IL-1β Precursor, IL-1β Mature, IL-1α, IL-1 RA, and ICE

The conditioned media and cell lysates of corneal limbal epithelial cells from four independent primary cultures, derived from four different donor corneas were collected, centrifuged, and stored at− 80°C until assayed. The concentrations of IL-1β precursor and IL-1β mature in the cell lysates and in the supernatants and of ICE in cell lysates were measured by ELISAs according to the respective manufacturer’s protocol. The cellular protein concentration in cell lysates was measured with the BCA protein assay kit. 

Western Blot Analysis for IL-1 RA

To evaluate the expression of IL-1 RA protein in the conditioned medium and in the cells, we further incubated primary limbal epithelium with LPS, LPS and MP, LPS and doxycycline, or LPS with MP and doxycycline, using the same concentrations as described earlier. 

Cell lysates and conditioned media containing equal quantities of protein were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) using 4% to 15%, 0.75-mm thick polyacrylamide gel (Mini-ready; Bio-Rad, Hercules, CA) at a constant 200 V for 45 minutes, in a mini-protean electrophoresis apparatus (Bio-Rad). A positive control (human recombinant IL-1 RA; R&D) and prestained (7.5–203 kDa) molecular weight protein markers (Bio-Rad) were run simultaneously with the samples. Resolved proteins were transferred to nitrocellulose membranes (BioTrance NT, Ann Arbor, MI) using a minitank blot apparatus (Bio-Rad). Membranes were blocked in 3% fat-free milk for 45 minutes After a 1-hour incubation with polyclonal rabbit anti-human IL-1 RA antibody diluted in 1% bovine serum albumin, Tris-buffered saline, and 0.5% Tween 20, the membranes were incubated with IgG-horseradish peroxidase–conjugated goat anti-rabbit (Sigma) diluted 1:80,000 in 1% bovine serum albumin, Tris-buffered saline, and 0.5% Tween 20. Signals were detected with an immunodetection kit (Renaissance Enhanced Chemiluminescence [ECL]; NEN Life Science Products, Boston, MA), and then exposed to x-ray film (Eastman Kodak, Rochester, NY) from 30 seconds to 3 minutes. 

IL-1β Activity Assay

To evaluate the functional activity of IL-1β secreted by cultured HLE, we sought to develop a bioassay that is based on the induction of MMP-1 and MMP-3 in corneal fibroblasts by IL-1β. ⁶ IL-1β has been previously demonstrated to induce MMP-1 and MMP-3 secretion in human synovial fibroblasts, ¹⁴ endometrial stromal cells, ¹⁵ and fibrochondrocytes. ¹⁶ We therefore cultured early-passaged corneal fibroblasts in conditioned media that were collected from the HLE cultures. The resultant MMP-1 and -3 supernatant concentrations were measured. 

Corneal fibroblasts were cultured as previously described. ¹⁷ Briefly, the central corneas of donor eye bank eyes were isolated with a 6-mm trephine after removal of the epithelium and the endothelium with a cell scraper. Explant cultures were prepared in the same manner as described earlier for limbal epithelial culture, except that DMEM containing 10% FBS (D-FBS) was used. Cultures were incubated at 37°C under 95% humidity and 5% CO₂, and the medium was changed twice a week. Fibroblasts were subcultured with 0.05% trypsin and 0.85 mM EDTA in a calcium-free MEM at 80% to 90% confluence with 1:3 to 1:4 split for three passages. 

Third-passage fibroblasts were seeded in six-well tissue culture plates at a density of 2 × 10⁵ cells per well. After 5 days in culture, on confluence, cultures were switched to a serum-free medium containing DMEM supplemented with 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenium, 50 μg/ml gentamicin, and 1.25 μg/ml amphotericin B (D-ITS). Some cultures were maintained in D-FBS. After a 24-hour incubation in D-ITS, cultures were divided into two groups. 

The first group of cultures served as a positive control and were treated directly with human recombinant IL-1β with one of the following: D-ITS alone, recombinant human (rh)-mature IL-1β (10 ng/ml), rh-pre-IL-1β (10 ng/ml), and rh-pre-IL-1β (10 ng/ml) with matrix MMP-9 (1 μg/ml). 

The second group of cultures were treated with conditioned media (CM) derived from HLE cultures that had been treated as described earlier. These treatments included CM from HLE culture treated with medium alone (CM-SHEM), CM from HLE culture treated with LPS (CM-LPS), CM from HLE culture treated with LPS and doxycycline (CM-LPS + doxy), and CM from HLE culture treated with doxycycline (CM-doxy). To exclude the possibility that the drugs contained in the conditioned media (LPS and doxycycline) altered MMP secretion in corneal fibroblasts, two additional treatments were added: SHEM with LPS (10 μg/ml; SHEM-LPS) and SHEM with LPS and doxycycline (10 μg/ml each; SHEM-LPS + doxy). 

Cultures were incubated with one of these treatments for 24 hours, and thereafter their supernatants were collected for measurement of MMP-1 and MMP-3 concentrations by ELISAs. 

Measurement of ICE Activity

The specific activity of ICE in cell lysates was measured, as previously described, ¹⁸ by incubating lysates with the fluorogenic substrate AcTyr-Val-Ala-Asp-AMC (YVAD-AMC; Upstate Biotechnology, Lake Placid, NY) in buffer containing 100 mM HEPES, 10% sucrose, 0.1% NP40, and 10 mM dithiothreitol (pH 7.5) for 1 hour at 37°C. Fluorescence was determined with a fluorometer at 360 nm (excitation) and 530 nm (emission). 

Statistical Analysis

Results are expressed as mean ± SEM of at least three experiments performed on cultures derived from at least three different donor corneas, respectively. Statistical analysis was performed using Student’s t-test or one-way analysis of variance (ANOVA) where appropriate. P < 0.05 was considered significant. 

Results

Expression of the IL-1 Genes in the Human Corneal Epithelium

All three genes of the IL-1 family: IL-1α, IL-1β, and IL-1 RA were demonstrated at the RNA and protein levels in primary cultures of human corneal limbal epithelium (Fig. 1A , first lane; Figs. 2 and 3 5 and 6 ). Both precursor and mature forms of IL-1β were found in these cultured epithelial cells and their supernatants. Considerably higher concentrations of the mature IL-1β were found within the cells (Figs. 2A 2B) , whereas higher precursor IL-1β concentrations were found in the extracellular environment (Figs. 3A 3B) . 

Regulation of IL-1β by Endotoxin, Doxycycline, and MP

The IL-1β mRNA was expressed at very low levels in the nonactivated corneal epithelium. However, when stimulated with LPS, it was significantly upregulated (sevenfold; P = 0.037) when compared with its baseline level (Figs. 1A 1B) . The addition of either MP or doxycycline to the culture significantly inhibited the LPS-induced upregulation (P = 0.037) and reversed IL-1β expression almost to its basal level. 

The cellular protein concentration of pre-IL-1β was low, varying between 6.4 ± 6.9 pg/mg protein in cells cultured in medium alone and 23.3 ± 13.6 pg/mg protein in cells stimulated with LPS (Fig. 2A) . The concentration of pre-IL-1β was not affected by MP or doxycycline. 

Considerably higher levels of IL-1β mature protein were found within the cells (Fig. 2B) . A significant increase of IL-1β mature was observed after treatment with LPS (from 284.8 ± 50.8 to 550.6 ± 89.1 pg/mg protein; P = 0.035, ANOVA). This LPS-induced elevation was significantly inhibited by the addition of either MP (397.7 ± 38.1 pg/mg protein), or doxycycline (427 ± 47.05 pg/mg protein; P = 0.035, ANOVA). 

The supernatant concentrations of IL-1β precursor protein showed an increase from 13.24 ± 6.31 pg/mg protein in untreated cells to 42.17 ± 13.9 pg/mg protein in LPS-treated cells (Fig. 3A) . This change was not significant. The addition of either MP or doxycycline caused a small, nonsignificant decrease in the supernatant concentration of pre-IL-β. In contrast, the mature form of IL-β in culture supernatant significantly increased after LPS treatment (from 3.47 ± 1.42 to 11.29 ± 3.96 pg/mg protein), and decreased to baseline when either MP (4.71 ± 1.41pg/mg protein) or doxycycline (2.9 ± 0.61 pg/mg protein) was added to LPS-treated cultures (P = 0.026, ANOVA; Fig. 3B ). The IL-1β mature to precursor ratio decreased after the addition of either MP or doxycycline, indicating an inhibitory effect of these two drugs on the activation of IL-1β (Fig. 3C) . 

Functional Activity of IL-1β Secreted from Cultured Corneal Epithelium

Mature IL-1β, but not precursor IL-1β, significantly increased MMP-1 and MMP-3 secretion by corneal fibroblasts compared with the media control (D-ITS), showing that this is a feasible method for determining IL-1β bioactivity (Figs. 4A 4C ). Preincubation of precursor IL-1β with MMP-9, an MMP that is known to process precursor IL-1β to the mature biologically active form, ¹⁹ resulted in MMP-1 and MMP-3 levels comparable to those observed with mature IL-β. MMP-1 and -3 were detected in corneal fibroblast cultures treated with HLE-conditioned media (Figs. 4B 4D) . Significantly increased production of these MMPs was observed when the LPS-treated HLE-conditioned media were added. Treatment of the fibroblast cultures with conditioned media from LPS + doxycycline–treated HLE abrogated this LPS-induced increase (Figs. 4B 4D) . In contrast, direct treatment of the corneal fibroblasts with LPS resulted in negligible protein concentrations, excluding a direct effect of drug that remained in the HLE-conditioned media. 

Regulation of IL-1α by Endotoxin and MP

The mRNA of IL-1α showed upregulation after stimulation with LPS. This increase was partially reversed when either MP or doxycycline was added to LPS (Fig. 1B) . The changes in the mRNA amounts for IL-1α were not statistically significant, although they followed a trend similar to that of IL-1β. 

The intracellular IL-1α protein concentration significantly increased from 525 ± 151 to 1134 ± 155 pg/mg protein (P < 0.001, ANOVA) in LPS-treated cultures and remained high when either MP or doxycycline was added to the culture (Fig. 5A ). A similar trend was observed in the culture supernatants (Fig. 6A ). LPS increased IL-1α concentration from 3.4 ± 1.2 to 10.7 ± 2.7 pg/mg protein (P = 0.055), but neither MP nor doxycycline affected the concentration of this cytokine. The cellular concentration of IL-1 α protein in corneal epithelial cells was 10-fold higher than the concentration released into the culture medium. 

Effect of Doxycycline on IL-1 RA

IL-1 receptor antagonist was upregulated by LPS alone or with either MP or doxycycline at the mRNA and protein levels. LPS induced a 3.5-fold increase of mRNA expression (P = 0.017), which was partially inhibited when MP was added but was unchanged when doxycycline was added to LPS (Fig. 1B) . 

No differences were noted between the four culture groups in the concentrations of the intracellular IL-1 RA protein (Fig. 5B) . However, the concentrations of IL-1 RA in the cultures supernatants demonstrated a significant increase when either MP or doxycycline was added to LPS (862 ± 222, 2730 ± 333, and 2230 ± 229 pg/mg protein for control, LPS + MP, and LPS + doxycycline, respectively (P < 0.001; Fig. 6B ). 

Western blot analysis for IL-1 RA demonstrated a marked increase in the bands for culture supernatants when doxycycline was added to LPS. The combination of doxycycline and MP added to LPS demonstrated the highest IL-1 RA protein expression (Fig. 7 , top). The expression of the intracellular IL-1 RA was not changed by any of the treatments, when compared with medium alone (Fig. 7 , bottom). These data correspond with the ELISA data, showing significant changes in secreted IL-1 RA, whereas the intracellular expression was stable (Figs. 5B 6B) . 

At the supernatant level, the IL-1 RA-to-IL-1α ratio markedly decreased after the addition of LPS alone, but addition of either of these two drugs to LPS returned the ratio to that observed in the control cultures (Fig. 6C) . 

Effect of Doxycycline on ICE

The mRNA transcripts of ICE were demonstrated in nontreated limbal epithelial cells (Fig. 8) , but no changes were demonstrated among the different treatments. The active p20 subunit of ICE protein was demonstrated by ELISA in limbal epithelium cell lysates (Fig. 9) . Its concentration increased significantly after stimulation with LPS (from 6.4 ± 1.1 to 8.5 ± 0.6 pg/μg protein; P = 0.011, ANOVA). There was no significant change in the cellular concentration of ICE protein in the MP-treated group. Doxycycline, however, significantly reduced the intracellular protein concentration when compared with LPS alone (6.3 ± 1.8 pg/μg protein, P = 0.011, ANOVA). 

Although the active part of ICE (the p20 isoform) was readily measured by ELISA, we could not demonstrate ICE activity in our samples using the ICE-specific fluorogenic substrate (YVAD-AMC). A possible explanation is that this assay, which was previously described for detecting ICE in monocytes and macrophages ²⁰ and THP-1 myelomonocytic cells, ²¹ did not have sufficient sensitivity to detect this enzyme’s activity in HLE. 

Discussion

This study demonstrates the effects of doxycycline on the expression patterns of the IL-1 gene family in the human limbal epithelium in response to a strong inflammatory stimulus. The results of our study demonstrate a novel inhibitory effect of doxycycline on the expression of the inflammatory cytokine IL-1β, with a concomitant upregulation of the anti-inflammatory IL-1 RA (Table 2) . We further report on the expression of ICE (caspase-1) in the corneal epithelium, at both the mRNA and protein levels, and demonstrate an inhibitory effect of doxycycline on that enzyme as well. These effects are comparable to those of the corticosteroid MP. Our results imply that some of the clinically proven benefits of tetracycline compounds (tetracycline, doxycycline, and minocycline) in treating the ocular surface disease of dry eye may be mediated through their regulatory effects on the synthesis, processing, or release of IL-1. 

We have demonstrated that in nonstimulated corneal limbal epithelial cells, all three forms of IL-1 (IL-1α, IL-1β, and IL-1 RA) were expressed at low levels. A strong inflammatory stimulation with LPS upregulated all three forms. When MP was added to LPS, mRNA expression of all three forms was inhibited, with IL-1 RA inhibited to a lesser degree. When doxycycline was added as the anti-inflammatory agent, only the level of IL-1β RNA was decreased. Both doxycycline and the corticosteroid decreased the concentration of mature IL-1β protein and increased the concentration of the anti-inflammatory IL-1 RA protein in the supernatants of limbal epithelial cultures. 

Two main mechanisms have been identified for activating precursor IL-β into its mature form. The first involves ICE, an intracellular cysteine protease that is highly specific for the cleavage of IL-1β. ¹⁸ ICE has been reported to be upregulated by immunologic stimuli such as LPS and granulocyte-macrophage colony-stimulating factor (GM-CSF). ²² To the best of our knowledge, this is the first demonstration of ICE expression by the corneal epithelium. We demonstrated ICE at the mRNA level and found that this gene is not regulated by either MP or doxycycline. Others have reported that LPS induced only a moderate increase of ICE activity in monocytes and that transcriptional induction of the ICE gene does not play a role in LPS-induced ICE activation. ²³ Immunodetection of the active p20 fragment of ICE by ELISA in this study demonstrated the protein in nontreated cells, which was increased by LPS, moderately decreased by MP, and significantly decreased by doxycycline. It therefore appears that doxycycline may decrease IL-1β at the protein level through direct inhibition of ICE. 

A second mechanism for IL-1β activation involves the MMPs. MMP-9, -2, and -3 can process precursor IL-β into its active form. ¹⁹ Activated forms of these MMPs are found in the tear fluid of patients with delayed tear clearance. ² This suggests that these enzymes may activate precursor IL-1β in the extracellular environment as found in our conditioned media. Among these various MMPs, MMP-9 was observed to produce the most stable biologically active IL-1β. It is possible that doxycycline inhibits the conversion of precursor IL-1β produced by the corneal epithelium, into its mature form, by inhibiting proteolytic cleavage by MMP-9. MMP-9 is the principal gelatinase produced by the human corneal epithelium. ⁶ We have observed a more than 70% decrease in MMP-9 activity in doxycycline-treated corneal epithelial cultures. ²⁴ This finding is consistent with reports of doxycycline’s decreasing MMP-9 activity in nonocular tissues. For example, doxycycline was shown to inhibit MMP-1 and MMP-9 activities in inflamed gingival tissue of adult patients with periodontitis. ²⁵ Therefore, another possible mechanism of inhibition of IL-1β activation by doxycycline may be MMP-9–mediated. 

Within cells, the majority of IL-1β was in the mature form, whereas in the supernatant, the precursor form was higher. LPS stimulation increased the concentration of mature IL-1β. Consistent with this finding was an increase in IL-1 bioactivity in the supernatants of these stimulated cultures. This finding is similar to the normal human tear fluid, in which we have detected predominantly precursor IL-1β, with very little mature form. The release of IL-1β has been reported to occur during apoptosis or cell death. ²⁶ If the presence of IL-1β in the culture supernatant were due to release from dying cells, we would expect to find mainly mature IL-1β. Therefore, it is possible that the corneal epithelium in our culture systems, as well as in vivo, releases precursor IL-1β into the environment, where it is susceptible to proteolytic cleavage and activation. Indeed, treatment of our cultures with either MP or doxycycline appeared to inhibit activation of precursor IL-1β in the supernatant of LPS-activated cells. 

Tetracyclines have been used widely to treat several localized inflammatory diseases, such as chronic acne, periodontitis, and the dermatologic and ocular manifestations of rosacea. ^{11 27 28 29} One study ²⁷ reported dramatic improvement in the symptoms and signs of ocular rosacea in 111 patients treated with oral tetracycline or doxycycline. The exact mechanisms of doxycycline-induced suppression of the corneal epithelium IL-1β system are not clear. Tetracycline was found to inhibit the LPS-induced IL-1β secretion, but not the mRNA accumulation in human monocytes, suggesting a posttranscriptional block in cytokine production. ³⁰ In addition, tetracycline blocked LPS-stimulated tumor necrosis factor (TNF)-α secretion, by inducing retention of membrane-associated TNF-α in monocyte cell membranes, thus preventing this cytokine’s release into the culture media. ³¹ Because no transport mechanism has been identified for the IL-1β protein, this mechanism of inhibition is unlikely. 

The functional activity of IL-1β generated from our limbal epithelial cultures was tested on corneal fibroblasts, measuring their MMP-1 and -3 secretion. This bioassay provides an in vitro system that closely resembles the in vivo situation, wherein epithelial cells of the ocular surface produce inflammatory cytokines that regulate MMP secretion by corneal and conjunctival fibroblasts. ⁶ Using MMP-1 and -3 secretion as a measure of IL-1β activity in conditioned media from HLE cultures, we found significantly lower concentrations of these matrix-degrading enzymes when conditioned media from LPS + doxycycline-treated cultures were added. This result may be either IL-1β–mediated or a result of a direct effect of doxycycline on these enzymes. 

Intervention in one of the IL-1β regulatory mechanisms may provide important strategies for the treatment of various inflammatory processes on the ocular surface. The increased tear fluid concentrations of IL-1β in dry eye conditions may result from release of proinflammatory cytokines, especially IL-1β, from stressed or damaged ocular surface epithelium into the tear film. A low volume of tear fluid and its delayed clearance from the ocular surface may further increase the concentration of IL-1β. This in turn may induce recruitment and activation of ocular surface inflammatory cells. Increased activity of proteolytic enzymes (such as MMP-9) in the tear fluid of patients with delayed tear clearance may also process precursor-IL-1β to its mature form, further increasing the concentration of IL-1β in the tear fluid. This inflammatory cascade could be responsible for the signs and symptoms of ocular irritation so commonly encountered in dry eye. Pharmacologic intervention may be capable of breaking this vicious inflammatory cycle. We have demonstrated that doxycycline has the potential to reduce the protein levels of cellular and extracellular mature IL-1β at a level comparable with that of a potent corticosteroid. We therefore conclude that based on our in vitro model, the efficacy of doxycycline for treatment of keratitis sicca may be due to inhibition of the IL-1β system. 

 Supported in part by Public Health Service Research Grant EY11915 (SCP); Department of Health and Human Services, National Eye Institute Grant CA61038 (BLL); National Cancer Institute; an unrestricted Grant from Research to Prevent Blindness; and the Drs. David and Maureen Smith Ocular Surface and Tear Research Fund.

 Submitted for publication November 23, 1999; revised February 23, 2000; accepted March 17, 2000.

 Commercial relationships policy: N.

 

Table 1.

 

Primers Used in RNAse Protection Assay

cDNA	Size (bp)	Upstream Primer (5′–3′)	Downstream Primer-1 (5′–3′)	Downstream Primer-2 (5′–3′)
IL-1α	407	CAA GGA GAG CAT GGT GGT AGT AGC AAC CAA CG	GCA CTG GTT GGT CTT CAT CTT GGG C	TAG TGC CGT GAG TTT CCC AGA AGA AGA GGA GG
IL-1β	348	GCT ACG AAT CTC CGA CCA CCA CTA CAG C	CCT TGT ACA AAG GAC ATG GAG AAC ACC	CTT ATC ATC TTT CAA CAC GCA GGA CAG G
IL-1 RA	308	CCA TTC AGA GAC GAT CTG CCG ACC	GCT TGT CCT GCT TTC TGT TCT CGC	CTG TCT GAG CGG ATG AAG GCG AAG C
GAPDH	188	GAC ATC AAG AAG GTG GTG AAG CAG GC		CCA AAT TCG TTG TCA TAC CAG GAA ATG AGC

Figure 1.

 

(A) RNase protection assay of RNA extracted from primary cultured human corneal epithelial cells. Cells were treated with either 10 μg/ml bacterial LPS alone or in combination with either 0.1 mg/ml MP (LPS + MP) or 10 μg/ml doxycycline (LPS + Doxy). (B) mRNA amounts for IL-1α, IL-1β, and IL-1 RA corrected for the different amounts of GAPDH. Data are the mean ± SEM of four different experiments on primary cultures from four different donor corneas. A significant increase in the mRNA amounts of IL-1β was observed after treatment with LPS, with a subsequent significant decrease to the control level when either MP or doxycycline was added to LPS (*P = 0.037, ANOVA). Similar nonsignificant changes occurred in the IL-1α mRNA expression. The mRNA amounts of IL-1 RA were significantly increased by LPS (**P = 0.017, ANOVA) but were not markedly changed with the addition of MP or doxycycline.

Figure 2.

 

Cellular protein amounts of the precursor (A) and the mature (B) forms of IL-1β measured by ELISA in cell lysates of HLE collected after 24 hours in serum-free medium and stimulated with either bacterial LPS, LPS + MP, or LPS + doxycycline (LPS + Doxy). Protein amounts are expressed in picograms per milligram of total cellular protein assayed in corresponding cellular lysates. Data are the mean ± SEM from four independent experiments performed in HLE cultures from four different donor corneas. LPS induced a significant increase in the intracellular concentration of the mature IL-1β protein, with subsequent decrease when MP or doxycycline was added (*P = 0.035, ANOVA).

Figure 3.

 

Supernatant proteinamounts of the precursor (A) and the mature (B) forms of IL-1β measured by ELISA in conditioned media of HLE. Treatment conditions, expression of data, and number of experiments are as described in Figure 2 . LPS induced a significant increase in the extracellular concentration of the mature IL-1β protein, with subsequent decrease when MP or doxycycline was added. (*P = 0.026, ANOVA). The IL-1β mature-to-precursor ratio of protein concentrations (C) was markedly reduced after the addition of either MP or doxycycline.

Figure 4.

 

Functional activity of IL-1β measured by MMP-1 and -3 secretion from corneal fibroblasts treated with conditioned media from HLE. (A) MMP-1 secretion from corneal fibroblasts treated with serum-free medium (D-ITS), rh-mature IL-1β (mIL-1β), rh-precursor IL-1β (pIL-1β), and precursor IL-1β treated with MMP-9 (pIL-1β+ MMP-9). Both mIL-1β and pIL-1β + MMP-9 significantly increased MMP-1 protein expression, when compared with D-ITS alone (*P = 0.03, t-test). (B) MMP-1 secretion from corneal fibroblasts treated with conditioned media (CM) from HLE cultures that had been treated with SHEM alone (CM-SHEM), LPS (CM-LPS), LPS + doxycycline (CM-LPS + doxy), or doxycycline alone (CM-doxy). Corneal fibroblasts were also directly treated with medium containing LPS (SHEM-LPS) or LPS + doxycycline (SHEM-LPS + doxy). A significant increase in MMP-1 secretion was noted with CM-LPS compared with CM-SHEM, indicating that the IL-1β protein from HLE cultures was functionally active (*P = 0.01, t-test). (C) MMP-3 secretion from corneal fibroblasts treated as in (A). Both mIL-1β and pIL-1β + MMP-9 significantly increased MMP-3 protein expression, when compared with D-ITS alone (*P = 0.005, **P = 0.003, respectively). (D) MMP-3 secretion from corneal fibroblasts treated as in (B). Whereas CM-LPS significantly increased MMP-3 secretion compared with CM-SHEM (*P = 0.04), CM-doxy decreased MMP-3 secretion (**P = 0.01), both indicative of the activity of IL-1β protein derived from the HLE cultures. A significant decrease of MMP-3 after direct treatment with SHEM-LPS + doxy (***P = 0.03) is a result of the direct effect of doxycycline on MMP-3. Data are expressed as treatment means ± SEM from three independent experiments derived from conditioned media from three HLE cultures.

Figure 5.

 

Cellular protein amounts of IL-1α (A) and IL-1 RA (B) measured by ELISA in cell lysates of HLE. Treatment, stimulants, expression of data, and number of experiments are as described in Figure 2 . LPS, LPS + MP, and LPS + doxy induced a significant increase in the intracellular concentration of the mature IL-1β protein, when compared with control. (*P < 0.001, ANOVA).

Figure 6.

 

Supernatant protein amountsof IL-1α (A) and IL-1 RA (B) measured by ELISA in conditioned media of HLE. Treatment, stimulants, expression of data, and number of experiments are as described in Figure 2 . LPS, LPS + MP, and LPS + doxy induced a borderline significant increase in the extracellular concentration of the mature IL-1α protein, when compared with control (*P = 0.055, ANOVA). The IL-1 RA protein secretion was significantly increased by both LPS + Mp and LPS + doxy (**P < 0.001, ANOVA). The IL-1 RA to IL-1α ratio of protein concentrations (C) was markedly reduced after LPS treatment but returned to the control value after the addition of either MP or doxycycline.

Figure 7.

 

Western blot analysis of IL-1 RA in conditioned media (top) and in cell lysates (bottom) of HLE cells cultured in medium alone (Control), or treated with LPS, LPS + MP, LPS + doxycycline (LPS + doxy), or the combination of LPS with MP and doxycycline (LPS + MP + doxy). The positive control is rh-IL-1 RA. Samples were taken from the same cultures used for the ELISA analyses shown in Figures 5B and 6B . The data are representative of four experiments.

Figure 8.

 

RPA of primary cultured human corneal epithelial cells. Treatment conditions are described in Figure 1 . (B) mRNA amounts for ICE corrected for the different amounts of GAPDH. Data are the mean ± SEM of three experiments on primary cultures from three donor corneas. No significant changes were noted between the different treatments.

Figure 9.

 

Protein amounts of intracellular ICE measured by ELISA in cell lysates of HLE. Treatment, stimulants, and expression of data are described in Figure 2 . ICE protein amounts were significantly elevated after stimulation with LPS (*P = 0.011, ANOVA), slightly reduced after the addition of MP, and returned to the nonstimulated level after the addition of doxycycline. Data are from five independent experiments performed on HLE cultures from five donor corneas.

Table 2.

 

Significant Changes of mRNA and Protein Expression of the IL-1 System by Different Treatments in HLE Cultures

Treatment	IL-1α			Mature IL-1β			IL-1 RA			ICE
		mRNA	Protein	mRNA	Protein	mRNA	Protein	mRNA	Protein
Intracellular
LPS (compared with control)		↑	↑	↑	↑			↑
LPS+ MP (compared with LPS)			↓	↓
LPS+ doxy (compared with LPS)			↓	↓				↓
Supernatant
LPS (compared with control)		↑		↑
LPS + MP (compared with LPS)				↓		↑
LPS + doxy (compared with LPS)				↓		↑

 ↑ Significant increase; ↓ significant decrease.