Subnormal Cytokine Profile in the Tear Fluid of Keratoconus Patients

Keratoconus, historically viewed as a non-inflammatory disease, is an ectatic corneal disorder associated with progressive thinning of the corneal stroma. Recently, a few inflammatory mediators have been reported to be elevated in the tear fluid of keratoconus patients. Consequently, we investigated a wide range of inflammation regulating cytokines in the tears and sera of keratoconus and control subjects. Interleukin (IL)-1β, IL-4, IL-6, IL-10, IL-12, IL-13, IL-17, interferon (IFN)-γ, chemokine C-C motif ligand 5 (CCL5) and tumor necrosis factor (TNF)-α were tested in tear samples and sera of keratoconus and control individuals by multiplex immuno-bead assays. Selected cytokines were further tested by standard ELISA on pooled tear samples. All cytokines in the sera were generally low, with no significant changes between keratoconus and control subjects. However, in tear fluids, clear differences were detected between the two groups. These differences include increased IL-6, and decreased IL-12, TNF-α, IFN-γ, IL-4, IL-13 and CCL5 in keratoconus compared to control tear fluids. The decreases in IL-12, TNF-α and CCL5 were statistically significant, while the IL-13 decrease was statistically significant in the severe keratoconus group only. IL-17 could not be detected by multiplex immuno-bead assay, but showed an increase in keratoconus by conventional ELISA on a limited number of pooled tear samples. Our findings confirm increased IL-6, but dispute earlier reports of increased TNF-α, and suggest a cytokine imbalance in keratoconus disrupting corneal homeostasis. Moreover, an increase in IL-17 suggests tissue degenerative processes at work, contributing to the thinning and weakening of the corneal connective tissue in keratoconus.


Introduction
Keratoconus is an ectatic corneal disease associated with progressive thinning of the corneal stroma, and a protruding, cone-shaped cornea that produces astigmatism and myopia. Keratoconus is a leading cause of corneal transplantation, affecting 1 in 2000 individuals with a mean age of onset at 15.4 years [1,2,3,4]. The disorder typically progresses until the third to fourth decade of life [3], and the factors that determine the progression or stabilization of the disease are not well characterized or understood.
Although the etiology of keratoconus is poorly understood, it is traditionally viewed as a non-inflammatory corneal thinning disease [1]. Accordingly, cellular infiltration and vascularization are not clinically apparent in keratoconus. However, keratoconus has been linked to atopy since the 1950s [5], with further support coming from later studies that reported elevated levels of immunoglobulin (Ig) E in the sera of keratoconus patients [6,7]. Recent studies have suggested pro-inflammatory factors as key to keratoconus pathogenesis based on their findings of elevated interleukin (IL)-6, tumor necrosis factor (TNF)-a and matrix metalloproteinase (MMP)-9 in the tear fluid of keratoconus patients [8,9]. Moreover, increased binding of IL-1 by keratoconus corneal fibroblasts has led another group to suggest a role for inflammation in the onset or progression of keratoconus as well [10].
Despite these initial findings of specific changes in inflammatory cytokines, there are no studies that have examined a range of cytokines to determine whether keratoconus is associated with an imbalance in the repertoire of cytokines that regulate inflammatory and immune responses driven by subsets of T-helper cells, T H 1, T H 2, and T H 17 in the corneal environment. To begin addressing this question, we quantified T H 1 cytokines (IL-12, IFNc and TNF-a), T H 2 cytokines (IL-4, IL-10 and IL-13), the T H 17 representative cytokine IL-17, and other inflammatory cytokines/ chemokines (IL-1b, IL-6, and RANTES or CCL5) in tear fluids and serum samples of keratoconus patients and control subjects. Although this initial study is based on a relatively small sample size, decreases in specific cytokines suggest down regulations of both pro-inflammatory and immunoprotective responses to play a role in this disease.

Demographics
A set of 18 keratoconus (KC) and 11 controls were used for the multiplex cytokine analyses on tear fluids and sera (Table 1). Two (11%) of the KC patients were graded as mild (steepest K,45D), 6 (33%) as moderate (45D #steepest K#52D), and 10 (56%) as severe (steepest K.52D). Using the Mann-Whitney test, there was a statistically significant (p = 0.02), 10 year difference in the mean age between the controls and the keratoconus subjects, who were on average 33 and 43 years old, respectively. Nine of the KC and four of the control subjects wore contact lenses. Due to their disease status, keratoconus patients normally wear hard contact lenses, while unaffected subjects wear soft contact lenses. Although matching by contact lens use would not per se remove this difference, to determine if contact lens use had an effect, irrespective of disease status, we fit a multivariate linear model to determine the effects of contact lens wear on tear fluid cytokine levels after adjusting for the age of the subjects. Levels of all cytokines tested were higher in the 13 contact lens users when compared to non-contact lens users, and while the differences were moderate for some of the cytokines, they were not statistically significant except for IL-4 (Table S1). The average time since diagnosis of keratoconus was 16614 (mean 6 SD) years ( Table 1). Three of the 18 keratoconus patients had a medical history including atopy. Tear fluid samples were collected between 9 am and 4 pm, with no statistically significant differences in collection times between control and keratoconus subjects. We also noted that the keratoconus group was ethnically more diverse than the control group. Our recent efforts are geared towards collecting tear samples from appropriately matched control subjects, and when applicable to collect tear fluids from the unaffected eye of keratoconus patients as well.
A second set of samples comprising 29 KC and 38 control samples were collected for performing individual ELISA assays for validation purposes (Table S2). In this set, the control subjects were matched more closely to the KC group for age and ethnicity. The mean 6 SD ages were 38610 and 40612 years, for KC and control subjects, respectively. Some of the KC subjects wore contact lenses, while the control subjects for this group were not contact lens users (Table S2). Of the keratoconus and control cases (Table S2), atopy was reported for six and four of the individuals, respectively.

Multiplex cytokine analysis of keratoconus and control subjects
To investigate the relationship between disease status and cytokine levels, we performed a linear multivariate correlation study of natural log transformed cytokine levels and disease, after correcting for age and contact lens use, as covariates. The results were reported as significant in Table 2, only if disease was a significant independent predictor of cytokine levels. All cytokines tested by multiplex immuno-bead assay, except IL-17, were detectable in tear samples. Table 2 shows median, 25 th and 75 th percentile values of tear fluid cytokines of all keratoconus and control subjects. Overall, we found IL-6 levels were 3 fold elevated (based on mean values) in keratoconus samples compared to control subjects (Table 2 and Figure 1), confirming a previous report of increased IL-6 in keratoconus [8]. The increase in IL-6 did not reach statistical significance due to the variation in levels between subjects. IL-1b was essentially unchanged, while the remaining cytokines showed decreased levels in our keratoconus samples. After subtracting the effects of age and contact lens use, the decreases in T H 1 related cytokines IL-12 (p = 0.03), TNF-a (p = 0.04) and CCL5 (p = 0.05) were statistically significant. The decrease in TNF-a seen here, contradicts an earlier observation of increased TNF-a in keratoconus. Statistical differences in cytokine levels were also analyzed after stratifying the keratoconus group into mild/moderate (steepest K#52D) and severe (steepest K.52 D) groups. No significant differences were evident when the mild to moderate disease was compared to the controls (analysis not shown). The comparison of the data from severe KC to that of controls is shown in Table 3. Despite the resulting decrease in sample size, IL-12, TNF-a and CCL5 showed even stronger effects in severe keratoconus. The IFN-c decrease in keratoconus compared to controls was significant by Mann-Whitney testing (p = 0.02, analysis not shown), but not by multivariate analysis, where the large variance of the measurements appeared to diminish its significance (Table 3). Of the T H 2-related cytokines, the decrease in IL-13 was statistically significant in severe keratoconus versus control subjects. The multiplex immuno-bead assay could not detect IL-17 in control or keratoconus tear samples. Atopy was not considered as a testable variable since it was detected in similar numbers in keratoconus and control individuals. We found no correlation (Methods) between the time of tear fluid collection and cytokine levels (not shown).
In the sera, there were no statistically significant differences between keratoconus patients and control subjects for the cytokines tested (Table 4). IL-10, IL-12 and IL-13 could not be detected in any of the sera tested, while IL-1b, IL-4, IL-6 and TNF-a were present at very low levels, and CCL5 was clearly detected. Thus, the cytokine changes detected in the tear film are likely a reflection of localized events in the eye and not systemic changes.

Measurements of selected cytokines in tear fluid by standard ELISA
The multiplex data implicate reductions in T H 1 (IL-12, TNF-a, IFN-c), and possibly T H 2 cytokines (IL-4 and IL-13) in keratoconus. To further explore these possibilities, we performed standard individual ELISAs for selected cytokines. In general, the conventional, standard ELISA (high sensitivity) is less sensitive than multiplex immuno-bead assays. Therefore, it was necessary to pool the tear samples from different KC and control subjects (Table S2) for these ELISA experiments. Based on the commercial availability of high sensitivity ELISA kits, we selected TNF-a and IFN-c as T H 1, IL-4 as T H 2 and IL-17 as T H 17 representative cytokines. We found the sensitivity of the TNF-a kit to be too low for TNF-a detection in the tear fluids by ELISA (data not shown). IFN-c was detectable and showed a decrease in keratoconus compared to pooled control samples but without statistical significance (p = 0.5, data not shown). We detected 14.26 8.6 pg/ml (mean 6 SD) of IL-4 in the control and 7.665 pg/ ml in the KC pool ( Figure 2); the IL-4 decrease in KC was statistically significant (p = 0.04), confirming the multiplex immuno-bead assay results (Figure 1, Tables 2 and 3). IL-17 was detectable in one control and 3 keratoconus pools in one experimental set. All negative values in that set, being below the level of detection, were set at the minimum detection level for this ELISA kit. Based on these limited data (Figure 3), there was a trend towards increased IL-17 (72.5659.2 pg/ml) in keratoconus tear fluids compared to control samples (18.567.8 pg/ml).

Discussion
In this study, we investigated cytokines in the sera and tear fluids of keratoconus and control individuals to determine if altered inflammatory response is a factor contributing to the etiology of keratoconus. Of the cytokines measured in the serum, most were at very low levels and none showed significant differences between patients and control individuals. This is consistent with the idea that keratoconus is not associated with major systemic inflammation. However, our results indicate perturbations in immune related homeostasis in the tear film and the corneal microenvironment of keratoconus patients.
The published data on actual baseline cytokine concentrations in tear fluids is limited, and quite variable. To compare our findings to those of others, we examined the literature and found only a few instances where tear fluids were assessed for multiple cytokines. Since only a limited amount of tear fluid is obtainable without stimulation, measuring multiple cytokines is challenging. Li and coworkers used a solid phase antibody protein array-based method to estimate the relative amounts of tear fluid cytokines in Sjögren's syndrome, without reporting actual levels of cytokines [11]. A recent study used an immuno-bead based multiplex system to obtain baseline levels of 30 different cytokines and chemokines [12]. Others have examined isolated cytokines by conventional sandwich ELISA and reported concentrations of IL-6, TNF-a and IL-10 in control and keratoconus subjects [8,13]. Our immunobead based assay, on average showed high levels of IL-4, IL-6, IL- 12, IFN-c, TNF-a and CCL5 in the tear samples of control subjects. Our baseline findings for IL-6 and TNF-a by multiplex assays were 5-10 times higher than the reported single ELISA measurements. However, our assay yielded baseline ranges for IL-10, IL-13, CCL5 and IL-6 that were within the range reported by the recent immuno-bead based study [12]. On the other hand, our estimates of TNF-a and IFN-c were higher, and that of IL-1b was lower than that reported in the earlier study [12]. Since our immuno-bead assay kit is different from that used in the other study, some of this difference may arise from different primary and secondary antibodies. In addition to the variation seen between different studies, the range for each cytokine within a given study, including ours, is very large. Some of this variability could be due to individual-to-individual variation in cytokine levels and the extent of tearing even though samples are collected without direct stimulation.
Comparing tear fluid cytokines in the control and keratoconus subjects, we noted an increase in IL-6 in keratoconus, in agreement with earlier reports [8,13]. In addition, we identified significant decreases in IL-12 and TNF-a. The latter observation disagrees with an earlier report of a small increase in TNF-a of keratoconus tear fluids. The possible reasons for this difference are, 1) use of different antibody-based assays, 2) actual TNF-a levels detected in the other study is very close to the lower limit of TNF-a detection by the conventional ELISA kit used in that study, and 3) differences in patient population between the two studies. IL-12 promotes the differentiation of T H 1 cells; its decrease in keratoconus is consistent with decreases in two signature T H 1 cytokines, IFN-c and TNF-a. In severe keratoconus, IL-12 and TNF-a decreases were more pronounced and this may play a role in increased IL-17 and associated tissue degenerative processes. Both IL-4 and IL-13 cytokines were also reduced in keratoconus, and the decrease in IL-4, as measured by conventional ELISA, was statistically significant. IL-4 is a key T H 2 cell differentiation cytokine, and both IL-4 and IL-13 play crucial roles in amplification of the T H 2 response through up regulation of STAT5, STAT6 and GATA3 transcription [14]. Thus, decreases in IL-4 and IL-13 suggest that T H 2 responses may be dampened in keratoconus. IL-4 is associated with allergic response and promotes the synthesis of IgE [15]. Based on the reported link between atopy and keratoconus, it is conceivable that IL-4 levels would be higher in the keratoconus group. However, the number of patients with a history of atopy in our samples was extremely low. Thus, separate from atopy, the decrease in IL-4 we see may Table 3. Tear-fluid cytokine concentrations in severe (K.52D) keratoconus and control subjects.  be pertinent to the pathophysiology of keratoconus itself. How do our tear fluid cytokine observations in keratoconus compare to other ocular surface diseases? A study reported increases in IL-1b, IL-2, IL-5, IL-6, IL-12, IL-13, in seasonal allergic conjunctivitis (SAC), vernal keratoconjunctivitis (VKC) and atopic keratoconjunctivitis (AKC), while IL-4, IFN-c and IL-10 levels were increased in SAC and VKC compared to controls [16]. These observations are certainly different from the cytokine profile we see in keratoconus. Thus, the broad decrease in inflammatory cytokines reported in our study may indeed be keratoconus disease-specific. Decreased IL-4 in keratoconus may have broad immune and non-immune consequences [17] that should be considered in the context of corneal thinning and abnormal stromal extracellular matrix (ECM). IL-4 and IL-4 receptor engagement through the JAK-STAT and the IRS1/2 pathways regulate cell proliferation and tissue homeostasis [17]. Dermal fibroblasts stimulated with IL-4 and IL-13 up regulate production of collagens type I and III [18]. IL-4 mediated induction of STAT-6 pathway in intrahepatic cells has been shown to induce collagen synthesis [19]. While its implication in hepatic disease is over-induction of STAT-6, increased collagen, and fibrosis, reduction of normal levels of IL-4 in the corneal environment may tip the balance toward poor stromal keratocyte cell survival, oxidative stress, altered collagen and ECM stability associated with keratoconus [20,21,22,23].
In vitro and in vivo studies indicate reciprocal interactions between IFN-c and IL-17, and between IL-4 and IL-17, and that these cytokines restrict undue amplification of the T H 17 response [24]. It is possible that in the pathogenesis of keratoconus, decrease and dysregulation of several cytokines and growth factors encourage aggravated T H 17 response, production of metalloproteinases and tissue damage. Furthermore, TGF-b and IL-6 levels regulate differentiation of Treg and T H 17 subsets of T cell [25]. TGF-b members are present in all ocular tissues [26,27,28], and, by immunohistochemistry we detected stronger immunostaining for total TGF-b in the epithelial layer (our unpublished observations) of keratoconus corneas. Moreover, IL-6 levels were increased in the tear fluid of keratoconus patients as indicated in this study and reported earlier [8]. Indeed, whether relative TGFb and IL-6 levels are altered sufficiently in keratoconus ocular tissues are not known at this time. On testing IL-17 by the multiplex immuno-bead assay, we could not detect it in the tear fluid. However, using an IL-17 ELISA kit, we detected higher levels of IL-17 in 3 of the five patient pools in one set of samples. An increasing number of studies link IL-17 to autoimmune and inflammatory diseases such as rheumatoid arthritis, lupus, asthma and psoriasis [29]. IL-17 is produced primarily by T H 17 subset T lymphocytes, while IL-17 receptors are found in a broad array of cell types including fibroblasts and myofibroblasts [30,31,32]. Multiple studies are beginning to link IL-17 mediated induction of fibroblasts and myofibroblasts and the production of tissue degrading proteases and cytokines [31,33,34,35]. Recent studies have also linked IL-17A to ocular pathogenesis of Sjögren's Syndrome [36]. Desiccating stress in dry eye disease appears to also involve increased MMP-9 and T H 17 inflammation [37,38,39]. In dry eye, IL-17 promoting cytokines have been shown to be produced by isolated T cells in the conjunctiva and the cornea [37]. In our hands quantitative reverse transcriptase polymerase chain reaction on total RNA from scraped corneal epithelium failed to detect IL-17A. Additional immunohistology of conjunctiva impressions will be needed to further address the source of IL-17 producing cells and the validity of an IL-17 mediated inflammation and tissue damage in keratoconus. Two recent studies performed proteomic analyses on the tear fluids of unaffected individuals and keratoconus patients. In one the authors reported down regulations in lactoferrin and immunoglobulin kappa chains [40]. Using a cytokine antibody array, the second study [41] found a decrease in TNF-a in agreement with our findings here. This study reported a slight increase in TGF-b2 and b3, which would conform to our speculation that conditions in keratoconus may favor aggravated Th17 response.
In summary, keratoconus is a complex heterogeneous disease in which its altered corneal structure and functions may be related to multiple factors. Our data suggest that instead of a simple increase in pro-inflammatory cytokines, there may be a complex imbalance between pro-inflammatory and anti-inflammatory cytokines, and possibly aggravated T H 17 response leading to altered epithelial and stromal functions. Clearly, it is important to confirm the observed cytokine changes in a larger patient population. To gain insights into the immunological processes that may contribute to these cytokine changes it is important to track down the cellular sources of these changes. While historically many studies have  Table S2 were pooled into 13 control and 10 keratoconus samples (Methods). Each ELISA value is an average of two measurements. The decrease in IL-4 in the keratoconus pool (7.665 pg/ml), compared to controls (14.268.6 pg/ml) was statistically significant (p#0.05, Mann-Whitney test). The horizontal bar indicates the mean value for each group. doi:10.1371/journal.pone.0016437.g002  focused on the cornea itself, the tear fluid changes implicate the conjunctiva and the lacrimal gland as having some paracrine effects in this disease.

Samples
All patients and normal subjects provided written informed consent according to a protocol used in the current study that was approved by the Johns Hopkins Institutional Review Board. Inclusion criterion for participation in the study was presence of keratoconus as determined by clinical examination and corneal modeling (Pentacam, Oculus, Lynnwood, WA). Participants completed a comprehensive questionnaire designed to collect information on quality of life, contact lens use, past medical history, and family history. For the multiplex cytokine analysis ( Table 1) we used 18 patients (61% males, mean age 6 standard deviation (SD) = 43613 years) and 11 control (45% males, mean age 6 SD = 33610 years) tear samples. Whole blood was collected for serum extraction.
Conventional ELISA (Table S2) was performed for selected cytokines on pooled samples from an additional 29 keratoconus patients (76% male, mean age = 38610) and 38 control individuals (52% male, mean age = 40612). These samples were pooled to generate 13 control and 10 keratoconus pools. The keratoconus samples were pooled based on disease status (mild, moderate and severe).
Severity of keratoconus was graded by the steepest keratometry (K) measurement with ,45 diopters (D) being mild, 45#K#52D being moderate, and severe .52D or not measurable. Ophthalmic exams consisted of best corrected visual acuity measurements, slit lamp biomicroscopy, and corneal topography using the Pentacam.

Tear Analysis
An aliquot of 5 to 8 ml of tear fluid was collected from each eye of participants, provided that the eye had not undergone previous surgery. Tear samples were obtained by capillary flow using micropipettes with care taken to minimize reactive tear production. Samples were frozen to 280uC within one hour of collection. Tear samples were diluted 1:25 in phosphate buffered saline (Quality Biological, Gaithersburg, MD) containing 1 mg/ml BSA to a final volume of 100 ml according to protocols supplied by the multiplex immuno-bead assay kit (Human Bio-Plex suspension array system, Biorad). The concentrations of cytokines (interleukin (IL)-1b, IL-4, IL-6, IL-10, IL-12, IL-13, interferon (IFN)-c, tumor necrosis factor (TNF)-a, and IL-17) and the chemokine (CC-motif) ligand 5 (CCL5) were measured using a Bio-Plex Luminex 100 (Biorad).

Serum Analysis
An aliquot of 10 ml blood was collected in a clot activating tube (Becton Dickinson, Franklin Lakes, NJ) and allowed to stand at room temperature one to two hours before spinning 15 minutes at 4uC. Serum aliquots were immediately transferred to 280uC for storage. Undiluted serum samples were assayed using the Bio-Plex Human Cytokine Assay kit, (Biorad, Hercules, CA) for IL-1b, IL-4, IL-6, IL-10, IL-12, IL-13, IL-17, IFN-c, CCL5, TNF-a as described for tear fluid samples.

Statistical Analysis
Chi-square tests were used to determine statistical significance for categorical variables between groups. Cytokine levels in keratoconus and control subjects were first compared using a 2tailed, Mann-Whitney nonparametric test (Prism; GraphPad Software, San Diego, CA.), with a p-value#0.05 considered to be statistically significant. To determine significant (p#0.05) associations between disease status and cytokine changes, we fit a multivariate linear model using natural log transformed cytokine levels, with age in years, and contact lens use as covariates. These multivariate analyses were performed using the R statistical software (http://www.R-project.org). We found no correlation between cytokine level and time of tear fluid collection (Spearmen correlation, Prism; GraphPad) and did not include collection time as a variable in the multivariate analysis.

Supporting Information
Table S1 Cytokine concentrations in contact lens users and non-users (DOC)