Attenuation of lysyl oxidase and collagen gene expression in keratoconus patient corneal epithelium corresponds to disease severity

Clinical study design and endpoints

This study was approved by the Narayana Nethralaya Institutional Review Board and performed as per the Indian Council for Medical Research (ICMR). In addition, institutional ethics guidelines were in accordance with the tenets of the Declaration of Helsinki. All patient samples were collected after obtaining informed, written consent as per institutional and ethics board guidelines. For minor patients, informed, written consent was obtained from their guardians or next-of-kin. This study has also been registered at ClinicalTrials.gov (Identifier: NCT01746823).

The study cohort was selected from patients who reported to the cornea clinic at Narayana Nethralaya, Bangalore. KC patients of Indian origin undergoing standard-of-care surgical intervention for the correction of visual defects were included in the study. In total, 80 KC patients undergoing topography guided photorefractive keratectomy (T-PRK) or corneal crosslinking were included for the collection of corneal epithelia. Additionally, 29 subjects with normal corneal topography undergoing photorefractive keratectomy (PRK) for the correction of refractive errors were used as controls for the corneal epithelia samples. Patients using contact lenses, any type of anti-inflammatory ocular or systemic medications (e.g., anti-allergic, anti-inflammatory drugs), or having undergone any prior surgical intervention (e.g., penetrating keratoplasty/corneal collagen crosslinking, cataract surgery, etc.) for either eye were excluded from the study. Patients with a recent (less than 3 months) allergic or infection history in either eye were excluded. All subjects also underwent a dry eye evaluation using Schirmer’s test and corneal staining. Subjects with concurrent symptoms of dry eye were excluded. Patients with systemic inflammatory or autoimmune diseases were also excluded. Tear samples were also collected from patients before surgery. Additionally, immunohistochemical (IHC) studies were performed on a separate group of archived tissue samples, as described below.

The experimental endpoints included analyses of the biomarkers LOX, MMP9, IL6, collagen I, and collagen IV in patient-derived samples. There were no safety endpoints, as there were no experimental interventions or any deviations from the standard of care.

Patient diagnosis and grading

Patient diagnosis for KC was done using retinoscopy, slit lamp biomicroscopy, and corneal refraction measurements. Corneal topographic images of all the patients were acquired with a Pentacam (OCULUS Optikgeräte GmbH, Germany) upon first presentation at the clinic and were used for the diagnosis and grading of KC patients [2,45] in this study. KC grades were determined from cumulative analyses of biomicroscopy data, KC features observed by slit lamp, spherical, and cylindrical refraction changes, mean central keratometry measurements, and corneal thickness, as defined by the Amsler–Krumeich classification [46,47]. However, grades 2 and 4 were combined as the severe KC group, as reported earlier [48]. No drug-based tests or analyses were performed during this study.

The KC subjects were graded based on methods reported in recent studies [4]. Mean keratometry was the primary classification variable. The mean keratometry reading was calculated as the average of the steep and flat axis keratometry readings. KC eyes with a mean keratometry reading of less than 48D were classified as grade 1. KC eyes with a mean keratometry reading of greater than 48D and less than 53D were classified as grade 2. KC eyes with a mean keratometry reading of greater than 53D were classified as grade 3, with grade 3 being the most severe.

Patient-derived corneal epithelium

Epithelial cells were collected using the epithelial debridement procedure routinely performed during ocular surface surgeries, such as corneal collagen crosslinking or T-PRK in KC patients. Control epithelial cells were collected from subjects undergoing PRK who demonstrated no corneal surface distortions and had no clinical signs of KC. The corneal epithelium study cohort had a total size of 142 eyes, with 52 control and 90 KC subjects (Grade 1 = 44, grade 2 = 29, and grade 3 = 17) with a mean age of 22.7±5.7 years. Surgical procedures for all patients and controls were performed under topical anesthesia with a proparacaine 0.5% ophthalmic solution. Debrided epithelial cells were immediately transferred to −80 °C for storage until processing for RNA extraction.

Isolation of RNA, cDNA synthesis, and real-time PCR

The isolation of mRNA and the analysis of gene expressions have been described before [49]. Total RNA was extracted from patient epithelia using a TRIZOL reagent according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA), followed by quantification and a quality assessment. In addition, cDNA was synthesized using a Superscript III (Life Technologies, Carlsbad, CA) cDNA conversion kit. Quantitative real-time PCR was performed as previously reported [50]. The quantitative real-time PCR cycle includes preincubation at 95 °C for 5 min, as well as 40 amplification cycles at 95 °C for 10 s, 60 °C for 15 s, and 72 °C for 30 s using a CFX Connect^TM real-time PCR detection system (Bio-Rad, Philadelphia, PA). The total levels of LOX, MMP9, IL6, COL IA1, and COL IVA1 were estimated after normalization to actin.

Patient tear collection

Tear samples were collected using capillary micropipette tubes, as described earlier [17]. The tears were collected from the exterior one-third of the lower fornix, taking care not to touch the conjunctiva or produce any reflex tearing. The tears were extracted from the micropipette tubes into sterile microfuge tubes followed by storage at −80 °C until analysis.

Lysyl oxidase activity assay

LOX activity in 20 µl tear samples from 30 control and 62 KC samples (grade 1 = 17, grade 2 = 17, grade 3 = 13, and grade 4 = 15) was measured using an Amplite™ Fluorimetric Lysyl Oxidase Assay Kit (AAT Bioquest Inc., Sunnyvale, CA) in a 96-well plate, as per the manufacturer’s instructions and as reported in the literature [51,52]. This kit and protocol were validated by the manufacturer using β-aminopropionitrile for the specific inhibition of LOX activity. Briefly, the fluorometric assay utilized a proprietary substrate that releases hydrogen peroxide using Amplite™ HRP substrate-coupled reactions to determine LOX activity. Enzyme activity was reported as relative fluorescence units (RFU/ml) from the mean of two independent readings. Blank wells without tear samples were used for determination of the baseline.

Immunohistochemical (IHC) studies

IHC studies have been reported in the literature for a variety of tissues, including KC corneas [49,53]. A cohort of 27 formalin-fixed, paraffin-embedded (FFPE) corneas from KC patients undergoing keratoplasty was obtained from the Narayana Nethralaya pathology department. In total, 15 unused donor corneas from individuals without clinical symptoms of KC were obtained from the Narayana Nethralaya Eye Bank to be used as normal controls for the IHC studies. All KC patient samples were classified as grade 4 KC. IHC experiments were performed using the Leica Novolink^TM Polymer Detection System (Leica Biosystems, Newcastle, UK, Cat # RE7159-K) as per the manufacturer’s protocols. Briefly, antigen retrieval was performed from 4-µm thick sections of FFPE tissue blocks in citrate buffer by boiling, followed by appropriate washing and blocking for endogenous peroxidase. Thereafter, primary antibodies for LOX (1:100 dilution; Merck Millipore, Billerca, MA; Cat # ABT112), collagen I (1:100 dilution; Novus biologic, Littleton, CO; Cat # NB600–408), and collagen IV (1:250 dilution; Abcam, Cambridge, MA; Cat # AB6586) were applied at 4 °C overnight, followed by washing and incubation with secondary antibodies as per the IHC kit directions. Counterstaining with hematoxylin was performed before the final substrate reaction step. Control and KC sample slides stained without a primary antibody were used as negative controls, and staining with an actin antibody was used as a positive control for the standardization of the IHC protocol. Photomicrographs were taken using a Leica Microsystems microscope and image acquisition software (DM 3000; LAS version 3.1.0). Protein expression levels were classified as high or low upon analysis of staining intensities (brown color) by two independent, blinded observers.

Statistical analysis

All data were compiled and grouped according to the stages of the diseases. The Friedman test for a comparison of group means was used. If the test was statistically significant, a post-hoc comparison of the group means was performed. The mean value of the individual groups was reported as mean ± standard error of mean (SEM). A linear regression analysis was performed among LOX, COL IA1, and COL IVA1. As the gene expression changed by a factor of 2 or more within the cohort, the expression data were converted to logarithmic values followed by a regression analysis. All statistical analyses were performed with MedCalc v12.5.0 (MedCalc Inc., Belgium) and GraphPad Prizm 6.0.

All continuous variables were assessed for normality of distribution. A one-way ANOVA was used to assess the difference between normal and KC grade means.

Clinical analysis of patient study cohort

The study cohort clinical characteristics are summarized in Table 1. KC patients were graded into four groups numbered 1–4, while controls were labeled as grade 0. As expected, the gradation reflected the concordant increase in the corneal curvature measured by keratometry; the mean K values were grade 0: 44.7+0.25, grade 1: 44.2+0.22, grade 2: 47.3+0.18, and grade 3: 51.5+0.39. The corneal refraction spherical measurements were grade 1: −1.1+0.35, grade 2: −2.14+0.55, and grade 3: −3.67+0.56. Consequently, visual acuity (logMAR) was reduced across the grades, as follows: grade 1: 0.11+0.02, grade 2: 0.17+0.03, and grade 3: 0.23+0.03. As a function of grade (Table 1), the best corrected distance acuity (BCDA), sphere, cylinder, spherical equivalent, K1, K2, Kmean, central corneal thickness (CCT), and thinnest regions were significantly different (p<0.001). The BCDA of the control was significantly different from all KC grades (p<0.05). Within the KC grades, grade 1 was similar to grade 2 (p>0.05), but it was significantly different from grade 3 (p<0.05). The cylinders of the control and all the KC grades were significantly different from each other (p<0.05). The sphere of the control was similar to those of grades 2 and 3 (p>0.05), but it was significantly different from that of grade 1 (p<0.05). The spheres of grades 1 and 2 were significantly different from that of grade 3 (p<0.05). The spherical equivalents of the control and of grade 1 were similar (p>0.05), while the spherical equivalent of grade 3 was significantly different from those of all the grades and the control (p<0.05). K1, K2, and Kmean were significantly different among all the grades and the control (p<0.05). The CCT of the control was significantly different from those of all the KC grades (p<0.05), and the CCTs of grades 1 and 2 were similar (p>0.05), but they differed from that of grade 3 (p<0.05). The thinnest region showed a trend similar to CCT. The data expectedly showed a strong correlation of the clinical grades with the corneal curvature and vision loss. Therefore, this data establishes the basis for the stratification of gene expression levels from the same group of controls and KC patients. The various gene expression and LOX activity values are summarised with their corresponding significance values across the different grades of KC in Table 2.

Lysyl oxidase mRNA expression is reduced in KC patient corneal epithelia and correlates with clinical grade of disease

We addressed the question of whether the LOX expression in the corneal epithelium undergoes any change in KC. Figure 1A demonstrates that the LOX expression is significantly reduced in KC corneal epithelial cells compared to controls. Furthermore, when we stratified the entire cohort based on the clinical grades of KC, the LOX expression values were grade 0: 2.12 ± 0.39 grade 1: 1.28 ± 0.27, grade 2: 0.89 ± 0.20, and grade 3: 0.6 ± 0.1. We clearly observed a reducing trend for LOX mRNA levels across the grades (p = 0.006), suggesting it may play a role in disease progression.

Reduction of the collagen gene expression in KC corneas shows a correlation with disease severity

As the LOX expression is reduced in KC, we investigated the levels of the collagen I (COL IA1) and collagen IV (COL IVA1) transcripts (Figure 2) in the same patient cohort. We observed that both collagen I and IV transcript levels were reduced in KC corneas compared to controls (Figure 2A). Although collagen I shows a reducing trend across grades 1–4 (grade 0: 24.2+5.5 grade 1: 14.3+5.73, grade 2: 12.8+4.58, and grade 3: 8.2+2.15), we did not observe a uniform reduction (p = 0.12) among the higher grades (Figure 2B). The collagen IV expression also reduced with the clinical grade of KC (p = 0.77) (grade 0: 6.77+2.16 grade 1: 6.5+3.48, grade 2: 4.22+1.42, and grade 3: 2.8+0.64; Figure 2B). However, the collagen IV expression demonstrated a significant reduction in grades 3 and 4, while grade 1 does not show a reduction compared to the control and grade 2 shows a modest reduction (Figure 2B).

The data suggests that along with the reduction in the LOX transcript levels, collagen expressions are also attenuated during KC disease progression. We conducted a statistical analysis using linear regression, as well as the correlation of the LOX expression with collagen I (Figure 2C) and collagen IV (Figure 2D) and of collagen I with collagen IV. The LOX expression demonstrated a strong positive correlation with collagen I (r = 0.35; p<0.0001). A strong positive correlation was also observed between LOX and collagen IV (r = 0.38; p<0.0001). Similarly, a strong positive correlation was also observed between collagen I and collagen IV (r = 0.72; p<0.0001).

KC corneal epithelium expresses high levels of MMP9

In the same cohort, the transcript levels of MMP9 were increased in the KC patients, while IL6 also showed a slight increase in KC patients over controls (Figure 3A). Figure 3B illustrates the differences in the gene expression of each MMP9 and IL6 marker when grouped according to the clinical disease severity of KC. MMP9 demonstrated a clear KC severity-dependent increase in expression (p = 0.007) (grade 0: 1.77+0.62 grade 1: 2.8+1.04, grade 2: 5.99+1.23, and grade 3: 6.2+1.69; Figure 3B). The IL6 expression, alternatively, showed a trend of increasing over the controls only in grades 2 and 3 (p = 0.98) (grade 0: 6.49+1.53 grade 1: 6.4+2.06, grade 2: 6.68+1.74, and grade 3: 7.4+1.48; Figure 3B).

Enzymatic activity of LOX in tears from KC patients is reduced compared to nonectatic controls

The corneal epithelium is known to secrete various factors in tears [54,55]. It has been reported that LOX enzymes are secreted by corneal epithelial cultures [41]. Therefore, we hypothesized that the human corneal epithelium may also secrete LOX, which could be analyzed in the tear film that bathes the cornea. We measured the total LOX activity in tear samples collected from the patients before their respective surgeries (Figure 4). Enzymatic activity was determined in 30 control and 62 KC patients using a commercially available assay kit that provides a functional readout in a fluorometric assay based on substrate oxidation. The total LOX activity was observed to be significantly reduced in the KC group (Figure 4A). There was a significant difference (p = 0.016) in the mean LOX activity between the control (6055 ± 16690 and KC (5039 ± 1928) groups. A reduction in LOX activity was also observed across KC grades (Figure 4B). When the control was compared with the grades of KC, there was a significant difference (p = 0.04) between the control and grade 3 (4753 ± 2021). However, the LOX activity of grades 1 (4956 ± 1619) and 2 (4753 ± 2021) were similar to the control (p>0.05). These data are reflective of the LOX transcript data (Figure 1B), suggesting the total LOX protein levels may be decreasing across different KC grades (Figure 4B).

Expression of LOX protein is reduced in corneal epithelium of KC patients

To determine whether the LOX protein levels were indeed reduced in KC cornea, we analyzed LOX levels by IHC in KC patient corneal tissues (Figure 5). A cohort of 27 KC and 15 healthy donor corneas in FFPE blocks were stained with a LOX antibody. Figure 5A–C are representative photomicrographs of corneas from three individual KC patients. Similarly, Figure 5D–F are photomicrographs of three individual healthy donor control corneas. The LOX expression was observed in the corneal epithelial layer, but not in the basement membrane or stroma (as indicated by arrows). We observed that of the 27 KC corneas, 70% demonstrated highly reduced levels of LOX, while only 20% of the control donor corneas showed reductions in LOX expression. There was considerably more variability in the expression levels in KC samples compared to the controls, as evident in the three representative individuals in Figure 5 (compare A and C). As controls for IHC, we performed staining with actin as a positive control and no primary antibody as a negative control (Figure 6). Actin staining (brown color) was of similar intensity in both the control and KC corneas, while the negative controls did not provide any background brown color.

Collagen IV is significantly reduced in the corneal basement membrane of KC patients, while the collagen I protein expression is slightly reduced in the corneal stroma

We next determined the levels of collagen I and collagen IV in the same cohort of KC and control corneal tissue samples (Figure 7 and Figure 8, respectively). Photomicrographs of the same patient samples represented in Figure 5 are shown in both Figure 7 and Figure 8. The collagen I expression was observed across the entire corneal stroma in all samples (as reported earlier [20]), but it was slightly higher in the healthy controls (Figure 7D–F) compared to the KC patients (Figure 7A–C), as indicated by the arrows. A reduced collagen I staining intensity was observed in 33% of the KC corneas compared to 13% of the healthy donor controls. It should be noted that we observed an increased intensity of collagen I staining near the corneal basement membrane, which is known [56] (Figure 7D, F). However, some of the control corneas do not show the same increased intensity near the basement membrane (Figure 7E). The donor corneas are obtained from subjects that expired of natural causes and are typically older in age, which may have led to a reduction in condensed collagen fibrils in some cases, as reported earlier [56].

We observed a clear expression of the collagen IV protein in the epithelial basement membrane, as reported previously [57,58] (indicated in the figure by the arrows). The collagen IV expression was significantly reduced or absent in 81% of the KC patient samples (Figure 8A–C) compared to 26% in controls (Figure 8D–F). These data mirror the gene expression levels observed in Figures 1 and Figure 2.

Conclusions

In summary, we demonstrate for the first time in corneal epithelial cells of a large patient cohort that mRNA expression levels of the corneal structure-related genes collagen I, Collagen IV, and LOX are reduced in KC and correlate with disease severity. Consequently, both LOX activity and protein levels are decreased in KC, suggesting a crucial role for this enzyme in disease pathogenesis. The corresponding reduction in collagen protein levels has direct implications on corneal function, and it may be used as a biomarker of disease progression. A reduction in the expression of corneal structure-related genes correlates strongly with a concomitant increase in MMP9 levels, suggesting a possible regulatory signaling loop between these genes. The data strongly suggest that the regulation of the expression of these genes in the corneal epithelium should be explored as a treatment target for KC.