Purpose: To investigate the expression of proteases, proteolytic activity and cytokines in the tear film of people with keratoconus.
Methods: Basal tears from people with keratoconus, from individuals who had undergone corneal collagen cross-linking for the treatment of keratoconus, and from normal controls were collected using a capillary tube. Corneal curvature of each subject was mapped. The total protein in tears was estimated. Levels and activity of proteases in the tears were analysed using specific antibody arrays and activity assays.
Results: The total tear protein level was significantly reduced in keratoconus (4.1 ± 0.9 mg/ml) compared with normals (6.7 ± 1.4 mg/ml) (p < 0.0001) or subjects who had undergone corneal collagen cross-linking (5.7 ± 2.3 mg/ml) (p < 0.005). Significantly (p < 0.05) increased tear expression of matrix metalloproteinases (MMP) -1, -3, -7, -13, interleukins (IL) -4, -5, -6, -8 and tumour necrosis factor (TNF) -α, -β were evident in keratoconus. Tear IL-6 was the only cytokine significantly (p < 0.05) increased in tears of keratoconus subjects compared with the collagen cross-linked group. No significant difference in tear proteases were observed between the normal and the cross-linked groups, although the expression of TNF-α was significantly (p < 0.05) increased in the cross-linked group compared with the controls. Elevated gelatinolytic (87.5 ± 33.6 versus 45.8 ± 24.6 FIU, p < 0.0001) and collagenolytic (6.1 ± 3.2 versus 3.6 ± 2.0 FIU, p < 0.05) activities were observed in tears from keratoconus compared with normal subjects. The activity of tear gelatinases (69.6 ± 22.2 FIU) and collagenases (5.7 ± 3.3 FIU) in the collagen cross-linked group was not significantly different compared with either keratoconus or normals.
Conclusion: Tears of people with keratoconus had 1.9 times higher levels of proteolytic activity and over expression of several MMPs and cytokines compared with tears from controls. Further investigations are required to study the possible implications of these changes and whether they can be used to monitor disease progression or determine the success of corneal collagen cross-linking.
Keratoconus (KC) is a poorly understood degenerative disease of the cornea. The disease is characterized by progressive thinning of the cornea giving rise to a cone-shaped cornea instead of the normal spherical shape. The disease mostly affects people in the productive period of their lives and can cause mild to marked impairment of vision. Although considered not common, KC is not rare and is often under diagnosed. Screening for KC is crucial to avoid complications such as ectasia or progressive thinning of the cornea after laser assisted in situ keratomileusis (LASIK) (Flanagan & Binder 2003). Once diagnosed, vision can be partly restored using contact lenses. Corneal collagen cross-linking (CXL) using UVA/riboflavin is a technique used to slow or stop the progression of KC (Sommer et al. 2003).
The thinning and ectasia of a KC cornea is mainly attributed to the increased degradation of extracellular matrix (ECM), which is made up of 70% collagen (Collier 2001). Type I collagen is predominant in the corneal stroma, and type IV collagen is observed in the epithelial basement membrane and Descemet’s membrane (Nakayasu et al. 1986). These collagen fibrils are interwoven by mature elastic fibres, consisting of the protein elastin (Kamma-Lorger et al. 2010).
Proteases are enzymes that cleave or break down other proteins and can be involved in the degradation of ECM proteins or activation of cellular apoptosis (Ollivier et al. 2007). Matrix metalloproteinases (MMP) are a family of zinc-dependent endopeptidases that include gelatinases (MMP-2,-9), collagenases (MMP-1, -8, -13), stromelysins (MMP-3, -10) and matrilysins (MMP-7, -26). These MMPs are synthesized by corneal epithelial cells and stromal cells (Fini et al. 1998). Matrix metalloproteinases can be inhibited by tissue inhibitors of matrix metalloproteinse (TIMP), which include TIMP-1,-2,-3 and -4. Raised levels of MMP-1 and -9 have been reported in tears of vernal keratoconjunctivitis (Leonardi et al. 2003) and MMP-8 in the tears of atopic blepharoconjunctivitis (Maatta et al. 2008). Cathepsin S (CATS) exhibits elastinolytic properties (Alexander & Werb 1989; Taleb et al. 2005), which could be important in the maintenance of the ECM architecture of the cornea. The corneas of people with KC have increased levels of proteases (Seppala et al. 2006). These may contribute to the changes in corneal shape. Although there have been several reports of tear proteases in KC patients, for example increased levels of pro-MMP-9 (Lema & Duran 2005; Lema et al. 2009) or active MMP-1 (Pannebaker et al. 2010), the relative activity of MMP-9 and other proteases are unknown and have not been studied or reported previously.
Cytokines or inflammatory molecules play an important role in the regulation of ocular surface inflammation and immunological reactions. These cytokines are multipotent peptides expressed by the cells of the ocular surface (Nakamura et al. 1998). The stromal cells in KC express high levels of binding sites for interleukin (IL)-1 (Fabre et al. 1991). Levels of interleukin (IL) -4 and -5 have been shown to be increased in tears of people with a proliferative type of atopic keratoconjunctivitis (Uchio et al. 2000). Tears of people with dry eye syndrome have elevated levels of IL-6 and tumour necrosis factor (TNF) –α (Yoon et al. 2007). Whereas there has been a report of significantly decreased levels of IL-12, TNF-α, IL-13, CCL5 in tears of people with KC compared with controls (Jun et al. 2011). Others have shown a significant increase in the level of IL-6 or TNF-α in the tears from KC (Lema & Duran 2005).
In this study, we have examined the levels and activity of proteases in tears collected from normal subjects, people with KC and, individuals after corneal collagen cross-linking for the management of KC.
Materials and Methods
Ethics was approved by the Human Research Ethics Advisory Panel at the University of New South Wales. All procedures were conducted in accordance with the 2000 Declaration of Helsinki. Written informed consent was obtained from each subject prior to being enrolled in the study.
Three groups were included in the study, consisting of control subjects (C) who had not been diagnosed with KC, subjects with KC and subjects who had undergone a corneal collagen cross-linking procedure (CXL) to stabilize the progression of their KC. A total of 80 eyes (C = 28, KC = 32, CXL = 20) were used to study the total protein levels, and 60 eyes (C = 20, KC = 25, CXL = 15) were used to study the protease levels. The gelatinase and collagenase activities were studied on 49 eyes (C = 17, KC = 19, CXL = 13) and 42 eyes (C = 16, KC = 15, CXL = 11), respectively. These differences in subject numbers were because of the need to increase the volume of tear samples collected during the study. The demographic profile of the subjects is illustrated in Table 1. Tears were collected during single study visits, and no subject was in more than one group. All the subjects either gave a history of no or discontinued contact lens wear for at least 1 month before the study. No participant in this study gave a history of active allergy at the time of tear collection. The cross-linked subjects had a postoperative period ranging from 3 to 6 months. Subjects who had a history of any ocular surgeries or were under topical or systemic medication were excluded from the study.
Sterile thin glass micro capillary tubes (BLAUBRAND® intraMARK, Wertheim, Germany) were used to collect tears. A minimum of 10 μl of tears was collected using 6–7 micro capillary tubes from the inferior cul-de-sac of the eye by the same investigator for every subject. Tears were collected only until the orange band on the tube, 1.5 mm from the collecting end, which is equivalent to a volume of 1.67 μl. This procedure was followed to ensure a steady tear flow rate of <1μl/min (Fullard & Snyder 1990). The average time taken to collect 10 μl of tears was 20–25 min. Extreme care was taken not to touch the ocular surface to avoid reflex tearing. Following tear collection, the samples were centrifuged at 2700 g for 10 mins at 4°C and stored at −70°C until used for analysis (Sitaramamma et al. 1998).
The cornea was mapped using the Allegro Oculyzer (Wavelight GmbH, Erlangen, Germany) on all the subjects. The Allegro Oculyzer with the Pentacam technology examines the cornea three dimensionally using the Scheimpflug principle. The Belin/Ambrosio Enhanced Ectasia Display was used for KC screening (Ambrosio et al. 2011).
Total tear proteins
The total protein concentration of the tear samples was quantified by Bicinchoninic acid (BCA) protein assay (Smith et al. 1985). A bovine serum albumin (BSA) standard curve was obtained by preparing serial dilutions of BSA. The BCA solution (Pierce BCA kit; Thermo Scientific, Scoresby, Vic., Australia) was added to standards or tears, and the absorbance was measured at 562 nm using spectrophotometer (Tecan Spectrofluoro Plus; Tecan Group Ltd., Männedorf, Switzerland). The levels of total protein in individual tears were estimated against the standard BSA curve.
Proteases and cytokines in tears
The proteases or the cytokines in the tear film were measured using a RayBio® human custom G- series antibody array (Ray Biotech, Inc., Norcross, GA, USA). The array on a glass chip provides a highly sensitive approach to simultaneously detect multiple active forms of proteases or cytokines from individual samples. A preliminary cross-reactivity test was performed, and no cross-reactivity was found for the selected antibodies of proteases and cytokines. In brief, antibodies to the individual proteases, protease inhibitors or cytokines were immobilized onto glass chips in discrete spots. Then, tear samples were incubated with the antibodies on the glass chips for 2 hrs at room temperature and washed with phosphate buffer saline. A cocktail of biotinylated antibodies was added and incubated for further 2 hrs. The final step involved addition of labelled streptavidin before reading the fluorescence (excitation: 532 nm) using Axon GenePix® scanner (Molecular Devices, Inc., Sunnyvale, CA, USA). The signal intensities between the samples were studied using the Ray Bio® analysis tool. The ratio of fluorescence intensity to total protein was taken as the measure of expression of individual proteases and cytokines in each sample.
Tear samples were pooled based on similarities in the steepest keratometry reading of the corneas within each group. Fifty micro litre of pooled tear sample from each group was used in the analysis (recommended minimum volume of individual tear sample for the array). The C group had four tear samples, each sample obtained by pooling tears from five eyes. Keratoconus group had five samples, each obtained by pooling tears from five eyes. The CXL group had three samples, each obtained by pooling from five eyes. All the samples were masked before examination.
Activity of proteases
The gelatinase and collagenase activities in individual tear samples were examined using the Enzchek® Gelatinase/Collagenase Assay Kit (Invitrogen Australia Pty, Mount Waverley, Victoria, Australia). The protocol was modified in our laboratory to be used with tear samples. Individual tear samples were incubated with the substrate and reaction buffer (0.5 m Tris–HCL, 1.5 m NaCl, 50 mm CaCl2, 2 mm sodium azide, pH 7.6) at room temperature for 2 hrs. DQ™ (Invitrogen Australia Pty, Mount Waverley, Vic., Australia) gelatin (pig skin) and DQ™ collagen (Type IV from human placenta) were the substrates used to determine the active forms of gelatinases and collagenases, respectively. These substrates are heavily labelled with fluorescein and express highly fluorescent peptides when digested by gelatinases or collagenases. The intensity of the fluorescence (excitation: 485 nm, emission: 535 nm) was measured by a fluorescence micro plate reader (Tecan Spectrofluoro Plus). The ratio of fluorescence intensity to total protein was taken as the measure of tear gelatinolytic or collagenolytic activity.
The results obtained were expressed as mean ± standard deviation. Significant differences in total tear protein levels and protease activity between the groups were determined by univariate analysis followed by post hoc testing. The expression levels of individual tear proteases between the three groups were analysed using the Kruskal–Wallis nonparametric test, and Mann–Whitney test was used to confirm the interaction of proteases between the groups. The keratometry reading was correlated to total protein, protease levels and protease activity using linear regression. spss predictive analytics software, version 18 was used for all the analyses, and a p-value of <0.05 was considered significant.
Total tear protein concentration
Age- or gender-related differences were not statistically significant between the groups. The total tear protein concentration in C, KC and CXL is shown in Table 2. Keratoconus patients had a significantly lower level of total tear protein compared with the C (non-keratoconic) (p < 0.0001) and CXL groups (p < 0.05). The difference in total protein was not significant between C and CXL. The total tear protein levels were correlated with the keratometry reading, and the results are given in Fig. 2. A significant (p < 0.05) but not strong negative correlation (r = −0.28) was found between total tear protein and keratometry reading by simple regression analysis.
Table 2. Total protein levels, gelatinase and collagenase activities in tear samples.
* Significant levels comparing all the three groups, that is C, KC and CXL.
Total protein (mg/ml)
6.7 ± 1.4
4.1 ± 0.9
5.7 ± 2.3
Gelatinase activity (FIU/mg total protein)
45.8 ± 24.6
87.5 ± 33.6
69.6 ± 22.2
Collagenase activity (FIU/mg total protein)
3.6 ± 2.0
6.1 ± 3.2
5.7 ± 3.3
The proteases and their expression levels in C, KC and CXL are shown in Table 3. The proteases (measured as fluorescent intensity units (FIU) per mg total protein) significantly (p < 0.05) that were increased in KC patients when compared with C group were MMP-1, -3, -7, -13. The difference in protease levels observed in CXL was not significant compared with KC or C.
Table 3. Expression of proteases and cytokines in the tears of the three groups.
* Significant levels comparing all the three groups, that is KC versus both C and CXL. Results are expressed as mean ± standard deviation.
30.4 ± 5.4
45.6 ± 8.9
39.8 ± 13.0
37.1 ± 4.8
52 ± 8.2
55.1 ± 31.1
57.7 ± 4.4
150.6 ± 62.0
502.5 ± 679.0
17.8 ± 3.3
28.9 ± 4.4
22.7 ± 6.8
356.8 ± 569.3
377.3 ± 380.7
191.9 ± 180.7
349.1 ± 545.3
442.1 ± 474.5
217.5 ± 211.2
52.8 ± 3.9
86.7 ± 6.5
76.7 ± 20.7
3272.8 ± 553.3
4533.6 ± 1015.4
3288.8 ± 1429.6
3937.3 ± 2720.2
6576.5 ± 1793.0
3631.8 ± 1166.3
85.7 ± 6.8
133.6 ± 9.5
119.9 ± 59.2
51.7 ± 16.4
67.4 ± 14.2
57.7 ± 29.9
44.8 ± 6.9
66.3 ± 8.4
60.6 ± 14.6
46.7 ± 3.9
72.4 ± 4.5
62.3 ± 16.1
47.4 ± 3.1
81.1 ± 11.5
66 ± 21.3
66.7 ± 8.7
313.6 ± 232.2
109.7 ± 35.2
3072.5 ± 881.2
5132.5 ± 3211.1
2720.3 ± 1516.8
1168.0 ± 405.2
2893.4 ± 1758.5
1308.1 ± 694.4
55.1 ± 11.2
76 ± 10.1
65.1 ± 18.8
20.5 ± 2.1
30.6 ± 3.3
29.4 ± 13.4
54.1 ± 7.0
92.2 ± 8.5
80.4 ± 26.5
76.8 ± 5.5
130.9 ± 21.3
114.85 ± 30.3
6524.9 ± 570.4
8321.1 ± 698.9
7072.6 ± 2295.2
Levels of individual proteases or protease inhibitors were correlated to the keratometry reading (Table 4). Significant positive correlations were found. As the level of MMP-13, CATS, TIMP-1 or TIMP-2 increased in the tear film so did the level of corneal steepening (increase in keratometry reading). The ratios of MMPs to TIMPs, that is MMP-9: TIMP-1 and MMP-2: TIMP-2, were not significant between the three groups and did not correlate with keratometry.
Table 4. Correlations between tear proteases/cytokines and keratometry (D) of control (C) and keratoconus (KC) group.
CATS = Cathepsins S; MMP = matrix metalloproteinases; TIMP = tissue inhibitors of matrix metalloproteinse; TNF = tumour necrosis factor.
* Significant positive correlation showing increase in keratometry (D) is accompanied by an increase in the tear protease/cytokine levels.
y = 1.052x + 20.22
y = 168.1x−4041
y = 453.9x−16231
y = 4.633x−108.5
y = 0.977x−20.43
y = 3.692x−100.6
y = 201.35x + 2072.8
The cytokine expression levels in C, KC and CXL are shown in Table 3. Keratoconus patients had significantly (p < 0.05) increased tear levels of IL -4,-5,-6,-8 and TNF-α,-β compared with C subjects. Interleukin-6 (IL-6) was the only cytokine over expressed in KC when compared with CXL group (p < 0.05). TNF-α was the only cytokine increased in CXL compared with C (p < 0.05). There was a significant positive correlation between the levels of IL-1α, IL-10 and TNF-α and corneal topography (Table 4).
Proteolytic activity of tears
The activity of gelatinases and collagenases in the tear film of C, KC and CXL was examined, and the results are shown in Table 2. The tear film of KC subjects had significantly (p < 0.0001) higher gelatinolytic activity compared with C subjects. The activity of gelatinases in CXL was not significantly different compared with KC or C. Similar results were observed for the activity of collagenases. Collagenases were significantly (p < 0.05) more active in KC compared with C group but were not significantly more active when the CXL group was compared with KC and C groups. The activity of gelatinases and collagenases was maximum in the KC subjects followed by the CXL. There was no correlation between gelatinolytic or collagenolytic activity and corneal topography readings. The substrate specificity of MMPs exhibiting gelatinolytic or collagenolytic activity (Alexander & Werb 1989) is shown in Table 5.
Table 5. Substrate specificity of collagenases (MMP-1,-8,-13), gelatinases (MMP-2,-9), stromelysin (MMP-3) and matrilysin (MMP-7).
The role of proteases in KC, a topic widely discussed and speculated over many years, has been reviewed by the authors recently (Balasubramanian et al. 2010). In KC, the cornea is known to have increased gelatinolytic and collagenolytic activity (Kao et al. 1982; Rehany et al. 1982; Brown et al. 1993) and increased tear levels of pro-MMP-9 (Lema & Duran 2005) or active MMP1 (Pannebaker et al. 2010). However, there have been no reports on the activity of MMP-9 and other proteases in the tear film from subjects with KC. This study, for the first time, has examined the activities of gelatinases and collagenases in the tears of people with or without KC, and in people after corneal collagen cross-linking for the treatment of KC.
In this study, we have found increased expression of collagenases (MMP-1, MMP-13), stromelysin (MMP-3) and matrilysin (MMP-7) in tears of KC patients using the antibody array. The finding of increased levels of active MMP-1 concurs with the results of a previous study (Pannebaker et al. 2010), as does the lack of a significant increase in levels of TIMP-1 or TIMP-2. The levels of MMP-13, TIMP-1, -2 and CATS were positively correlated to the keratometry (Table 4). Cathepsins S had higher expression in KC patients compared with normal subjects, but this expression was not statistically significant. The proteolytic activities of MMP-1, -3, -7 and CATS might affect the stability of architecture of the cornea, because of their ability to degrade elastin (Shi et al. 1992; Li et al. 2000; Heinz et al. 2011), which is interwoven with collagen fibres in the cornea (Kamma-Lorger et al. 2010).
Our study indicates no difference in the level of active MMP-2 and -9 between the three groups. Whilst increased levels of pro-MMP-9 have been reported (Lema & Duran 2005), we used antibodies to the active MMP-9, which may explain the difference. These gelatinases are known for their ability to degrade denatured collagen. The collagenases are responsible for the ‘first insult’, causing the collagen to denature (Alexander & Werb 1989). Keratoconus corneas have a weak arrangement of collagen fibrils, and normal levels of gelatinases would still have a greater gelatinolytic activity in KC compared with a normal cornea with a healthy collagen. As KC is a slowly progressive disease, even minute levels and activity of proteases might add to the steepening effect on the cornea (Balasubramanian et al. 2010).
The tear proteolysis, that is gelatinolytic and collagenolytic activities, was significantly increased in KC patients compared with the C group. The protease expression/activities are not always continuous, and this could be the reason for certain protease levels (MMP-1,-2,-3,-7,-9) and the total proteolysis being inconsistent with the keratometry readings in KC.
There have been various studies with conflicting reports on the expression of various MMPs in KC corneas. Studies have shown increased levels of MMP-1 (Seppala et al. 2006) and MMP-13 (Mackiewicz et al. 2006) in KC corneas. Others have shown that normal and KC corneas showed no difference in the levels of MMP-2,-9 (Fini et al. 1992) and MMP-3,-10 (Saghizadeh et al. 2001). Increased levels of tear MMP-1, -13 shown in the current study supports the over expression of MMP-1, -13 observed in KC corneas, reported previously. (Mackiewicz et al. 2006; Seppala et al. 2006). Our data also supports the fact that there are no differences in the levels of MMP-2 or MMP-9 during KC, although we did see an increase in the level of MMP-3 in tears during KC.
Corneal collagen cross-linking (CXL) by combined riboflavin/UVA has a ‘freezing effect’ on the KC cornea (Sommer et al. 2003). Wolf et al. have reported decrease in corneal steepening in KC patients after CXL, but cases of keratopathy (Rodriguez-Ausin et al. 2011) and corneal melting (Labiris et al. 2011) after the procedure have also been reported. The complications and failure rates of CXL were 2.9% and 7.6%, respectively (Koller et al. 2009). To date, there have been no reports on the behaviour of tear proteases after CXL. In this study, the tears were collected from CXL group 3-6 months after CXL procedure. At this postoperative period, the epithelium and stroma are fully regenerated with a regular corneal surface in KC (Mazzotta et al. 2008). The levels of MMPs in CXL were intermediate between the KC and normal subjects. The positive correlation between proteases/cytokines (Table 4) and keratometry became insignificant when CXL group was included in the analysis. No difference was observed in the tear protease activity of CXL when compared with C or KC groups. Reduced protease levels and protease activity were observed in the CXL group compared with KC patients. Thus, CXL might positively affect the protease levels and activities in the KC cornea. Since tears from the subjects were not analysed before CXL procedure in this study, future studies may need to confirm the direct effect of cross-linking on tear proteases.
Keratoconus is defined as a noninflammatory disease of the cornea, but inflammatory molecules such as interleukins (IL) and TNF have been shown to be over-expressed in KC corneas (Fabre et al. 1991) and tear film of KC patients (Lema & Duran 2005). McMonnies has mentioned the possibilities that classifying KC as a noninflammatory condition might be inappropriate (McMonnies 2007). The present study found increased expression of IL-4,-5,-6 and TNF-α,-β in the tears of KC patients, and the levels of IL-1α,-10, TNF-α were positively correlated to keratometry. IL-6, in particular, was elevated in KC compared with CXL subjects. The expression of TNF-α was elevated in CXL compared with the control or C group.
As mentioned earlier, there is an active interplay between MMPs and ILs (Fig. 1). Matrix metalloproteinases-1,-2,-3-9 and IL-10 have an inhibitory effect on IL-1β (Ito et al. 1996) and CATS (Sendide et al. 2005), respectively (Fig. 1). This interaction could contribute to the insignificant levels of IL-1β and CATS in the present study. The increased expression of TNF-α in CXL compared with C might trigger the levels of MMPs and CATS leading to similar levels observed in a KC tear film as shown in Fig. 3. Unwinding this interplay would be crucial to establish the role of proteases in KC.
In conclusion, the expression and activity of proteases in the KC tear film appears to be profoundly altered. An in-depth analysis is essential to determine the use of tear proteases in early diagnosis of KC, monitoring the disease progression before and after cross-linking.
This study was supported by a scholarship from Brien Holden Vision Institute. The authors would like to thank Dr. Judith Flanagan for assistance in preparation of the manuscript. The authors would also like to extend their appreciation to Mrs. Ananthalakshmi for helping to recruit subjects for the study and Dr. Thomas Naduvilath for his invaluable guidance in statistical analysis.
Declaration of interest
This work was presented as a paper at the European Association for Vision and Eye Research Congress 2011 at Crete, Greece.