The relationship between keratan sulfate glycosaminoglycan density and mechanical stiffening of CXL treatment

Abstract

The corneal stroma is primarily composed of collagen fibrils, proteoglycans, and glycosaminoglycans (GAGs). It is known that corneal crosslinking (CXL) treatment improves mechanical properties of the cornea. However, the influence of stromal composition on the strengthening effect of CXL procedure has not been thoroughly investigated. The primary objective of the present research was to characterize the effect of keratan sulfate (KS) GAGs on the efficacy of the CXL therapy. To this end, the CXL method was used to crosslink porcine corneal samples from which KS GAGs were enzymatically removed by keratanase II enzyme. Alcian blue staining was done to confirm the successful digestion of GAGs and uniaxial tensile experiments were performed for characterizing corneal mechanical properties. The influence of GAG removal and CXL treatment on resistance of corneal samples against enzymatic pepsin degradation was also quantified. It was found that removal of KS GAGs significantly softened corneal tensile properties (P < 0.05). Moreover, the CXL therapy significantly increased the tensile stiffness of GAG-depleted strips (P < 0.05). GAG-depleted corneal buttons were dissolved in the pepsin digestion solution significantly faster than control samples (P < 0.05). The CXL treatment significantly increased the time needed for complete pepsin digestion of GAG-depleted disks (P < 0.05). Based on these observations, we concluded that KS GAGs play a significant role in defining tensile properties and structural integrity of porcine cornea. Furthermore, the stiffening influence of the CXL treatment does not significantly depend on the density of corneal KS GAGs. The findings of the present study provided new information on the relation between corneal composition and CXL procedure mechanical effects.

Introduction

The cornea is the transparent tissue at the front of the eye. In addition to its unique optical properties, the cornea behaves as a protective layer for the internal delicate content of the eye. The cornea is composed of different layers but its mechanical properties are primarily because of its extracellular matrix, i.e., stroma. The stroma constitutes about 90% of corneal thickness and is composed of thin collagen fibrils that are embedded in a hydrated proteoglycan (PG) matrix. PGs play an important role in the assembly and maintenance of highly regular microstructure of collagen fibrils inside the stroma; which is necessary for corneal transparency. PGs are macromolecules composed of covalently attached glycosaminoglycan (GAG) chains to a core protein. They have distinct biological functions ranging from signal transduction and proliferation of cells to collagen fibril assembly and degradation (Iozzo, 1998). GAGs are sulfated carbohydrates that provide PGs with their unique properties. Keratan sulfate (KS), chondroitin/dermatan sulfate (CS/DS), and heparan sulfate (HS) are among different types of GAG side chains. GAGs are negatively charged, attract water into the tissue, generate an effective osmotic swelling pressure, and subsequently contribute to the ability of corneal extracellular matrix to resist external forces.

There are several eye diseases that affect the composition, microstructure, and biomechanics of the corneal stroma. For example, keratoconus is a progressive ectatic disorder in which the cornea bulges out in a cone shape. The exact etiology of the keratoconus is not known but it usually starts in the teenage age. In keratoconus, the cornea continuously loses its mechanical strength and acquires a cone shaped structure. It is now possible to intercept keratoconus progression at early stages using the corneal crosslinking (CXL) treatment. The CXL therapy uses UVA light and riboflavin solution to create crosslinks inside the corneal stroma in order to improve its mechanical properties. The CXL technique has been under intense investigation to improve its outcome since early 2000’s (Meek and Hayes, 2013; Spoerl et al., 1998; Wollensak et al., 2003a). The efficacy of CXL treatment in in vitro studies has often been determined by characterizing the increase in stiffness of healthy human donor and animal corneas. However, it is well documented that keratoconus causes significant changes to the composition and structure of the corneal stroma (Meek et al., 2005; Morishige et al., 2007). In particular, it has been reported that KS GAG content is significantly low in keratoconic corneas (Sawaguchi et al., 1991; Wollensak and Buddecke, 1990). In a recent work, we have shown that tensile strength significantly decreases after removing KS GAGs from corneal extracellular matrix (Hatami-Marbini, 2023). We have also characterized the effect of KS GAGs on corneal stromal microstructural properties such as the collagen fibril diameter and interfibrillar spacing (Hatami-Marbini and Emu, 2023). The present work extended these previous studies by investigating the effect of KS GAGs on the strengthening effect of the CXL treatment. To this end, we used the CXL procedure to crosslink GAG-depleted samples. We then determined the structural integrity and mechanical response of these samples by characterizing their resistance against pepsin enzymatic degradation and measuring their tensile properties.

2. Materials and Methods

2.1. Sample Preparation

Fresh porcine eye globes were collected from an abattoir and transported to the laboratory in a closed container on ice. After removing excess muscles and fat with a sharp scissors and rubbing off epithelial layer by the blunt edge of a scalpel, we excised corneal rings with about 2 mm scleral rim. The endothelial layer was then removed and corneal strips with a width of ~ 5 mm and length of ~ 15 mm were cut in the nasal temporal direction using a custom-made double bladed device. The corneal strips were randomly assigned to Control, Control CXL, Enzyme, and Enzyme CXL groups. In addition to corneal strips, 8 mm full thickness corneal discs were also prepared for biochemical analysis and pepsin digestion studies.

2.2. Enzyme treatment protocol

The samples were completely dried inside a desiccator and their dry weight was measured using a digital scale. In order to deplete GAGs, we placed dried corneal strips in individual containers filled by the enzyme solution, composed of 0.1 U/ml keratanase II and 100mM sodium acetate buffer, for 20 h at 37 °C and pH 6.0 under gentle agitation. After measuring their thickness using an ultrasonic pachymeter (DGH Technology Inc., PA), we washed all samples in 1× phosphate buffered saline (PBS) five times.

2.3. Corneal crosslinking (CXL) treatment

The conventional CXL protocol as described in our previous studies was used to crosslink corneal strips (Hatami-Marbini and Emu, 2022; Hatami-Marbini and Rahimi, 2015; Wollensak et al., 2003a). Briefly, samples were soaked 30 minutes in 20% dextran 0.1% riboflavin solution and were subjected to 3 mW/cm² UVA irradiation (370 nm) for 30 minutes such that they experienced a total energy of 5.4 J/cm². The desired irradiance was controlled using a UVA meter. During the irradiation period, riboflavin photosensitizer solution was dropped on samples at 5 minutes intervals to ensure sufficient supply of riboflavin. The corneal strips were washed in PBS after the CXL treatment.

2.4. Mechanical experiments

The uniaxial tensile tests were performed using an RSA-G2 testing machine (TA Instruments, MD). We used five samples for Control and Control CXL groups and ten samples for Enzyme and Enzyme CXL groups, Table 1. Sandpaper was used to avoid any possible slippage at grips. The loading gap between upper and lower mounting clamps was about 7 mm. After mounting the samples, the loading gap was reduced to create slack and remove any unwanted tension. Then, similar to our previous work (Hatami-Marbini and Jayaram, 2018a, b), a 20-30 mN preload was applied in order to straighten the strips and determine their initial length for strain calculation. The tensile strain was calculated from dividing the change in the length of samples by their initial length. Samples were stretched to 20% by applying a displacement rate of 3 mm/min. Unlike our previous studies (Hatami-Marbini, 2014; Hatami-Marbini and Rahimi, 2014), we did not use a bathing solution because experiments were quick without any significant sample dehydration (Hatami-Marbini and Jayaram, 2018a), which was confirmed here by comparing the thickness of strips before and after the mechanical tests. The initial thickness of strips was about 700 μm to prevent hydration effects on mechanical measurements (Hatami-Marbini and Etebu, 2013).

Table 1.

Number of samples used in each group of mechanical tests and pepsin digestion studies.

Techniques	Groups
	Control	Control CXL	Enzyme	Enzyme CXL
Mechanical tests	5	5	10	10
Pepsin digestion	5	5	5	5

2.6. Histology

Histological analyses were performed to determine successful removal of GAGs. Specimens were fixed in 10% buffered neutral formalin for about 24 hours and 5 μm thin sections were prepared from samples embedded in paraffin blocks. Thin sections were stained with Alcian blue (StatLab Medical Products) for 30 minutes at room temperature, washed, and counterstained with nuclear fast red (StatLab Medical Products) for 3 minutes. Specimen slides were examined with Aperio Brightfield 20x whole slide scanner. Aperio ImageScope and ImageJ softwares were used to process images.

2.7. Pepsin Digestion

Control, Control CXL, Enzyme, and Enzyme CXL groups were utilized in pepsin digestion study. Five corneal disks of size 8 mm were used in each group, Table 1. We used the same pepsin digestion methodology that was previously employed to determine corneal enzymatic resistance following the CXL procedure (Aldahlawi et al., 2016; O’Brart et al., 2018). Briefly, corneal samples were put in sealed transparent containers filled with 8 ml of pepsin digestion solution, i.e. 1 g ≥400 U/mg pepsin from porcine gastric mucosa (Sigma-Aldrich Co. LLC) in 10 mL 0.1 M hydrochloric acid at pH 1.2. A digital caliper was used to measure the diameter of anterior surface of corneal specimens for 13 days or until the tissue was distinguishable in pepsin digestion solution. It is noted that we did not use thickness measurements for assessing enzymatic digestion because of significant corneal swelling in pepsin solution.

2.8. Data Analysis

Statistical analyses were performed using SAS analytic tool. Shapiro–Wilk tests were done to check the normality of data. ANOVA and t-test were performed to find significant differences among sample groups. A P-value less than 0.05 was considered statistically significant.

3. Results

The tensile response of different groups is shown in Figure 1. This figure also shows histology images of a typical sample from Control and Enzyme groups. GAGs, stained in blue, were seen throughout the Control sample but keratanase II enzyme treatment caused a significant reduction of blue staining in the GAG-depleted sample, which confirmed successful removal of GAGs. The tensile stress and tangent modulus at 5%. 10%, 15%, and 20% strain were calculated from stress-strain curves, Table 2. Although the CXL procedure significantly increased the maximum tensile stress of control samples, GAG depletion significantly reduced their maximum tensile stress (P < 0.05), Figure 2. Furthermore, the CXL treatment of GAG-depleted samples increased their maximum tensile stress significantly (P < 0.05), Figure 2. No significant difference was observed between the maximum tensile stress of Enzyme CXL and Control samples.

Figure 1.

The tensile response of Control (n=5), Control CXL (n=5), Enzyme (n=10), and Enzyme CXL (n=10) strips. Control samples were tested immediately after dissection. Samples in the Enzyme group were immersed in enzyme solution for 20 h before mechanical testing measurements. Control CXL specimens were prepared by crosslinking corneal strips immediately after dissection. Enzyme CXL samples were subjected to enzyme solution before being crosslinked using the conventional CXL protocol. The symbols and vertical lines at each point represent the average and one standard deviation, respectively. The histological evaluation done by Alcian blue stain showed a significant reduction of GAG density in GAG-depleted samples.

Table 2.

Tensile stress and tangent modulus at 5%. 10%, 15%, and 20% strain and the average number of days taken to digest corneal buttons in different groups.

Strain (%)	Control		Control CXL		Enzyme		Enzyme CXL
	Stress (MPa)	Modulus (MPa)	Stress (MPa)	Modulus (MPa)	Stress (MPa)	Modulus (MPa)	Stress (MPa)	Modulus (MPa)
5	0.03±0.01	0.97±0.18	0.05±0.01	1.45±0.24	0.03±0.01	0.70±0.16	0.03±0.01	0.89±0.17
10	0.17±0.05	3.71±1.31	0.20±0.04	4.93±0.80	0.11±0.02	2.27±0.54	0.13±0.03	3.20±0.97
15	0.48±0.13	7.88±1.58	0.63±0.07	11.84±1.42	0.32±0.04	5.79±0.50	0.43±0.09	8.31±1.27
20	1.03±0.17	13.32±2.75	1.44±0.13	19.42±2.35	0.73±0.06	9.88±0.79	1.05±0.14	14.32±1.47
Average number of days for complete pepsin digestion	9		>13		6		8

Figure 2.

The effect of enzyme and crosslinking treatment on a) maximum tensinle stress and b) maximum tangent modulus of porcine cornal strips. The depletion of GAGs using keratanse II enzyme significantly reduced the tensile stress and tangent modulus (P < 0.05).The CXL therapy significantly increased tensile stress and modulus of both control and GAG-depleted samples (P < 0.05). The symbol * denotes that the difference between groups was signficant.

Figure 3 shows the diameter of corneal samples as a function of time during the pepsin digestion study. Except for CXL-treated Control samples, all other samples were completely digested. The average time taken for complete digestion of GAG-depleted samples, i.e. 6 days, was significantly shorter than the average time, i.e. 9 days, required for Control samples (P < 0.05). The CXL treatment of GAG-depleted disks significantly increased the average time needed for their complete digestion from 6 to 8 days (P < 0.05). The reduction in the average diameter of samples in Control and Enzyme CXL group did not significantly differ until day 7.

Figure 3.

The variation of sample diameter as a function of time in order to capture the resistance of samples from different groups to pepsin digestion. Only CXL-treated Control samples did not digest within 13 days. GAG-depleted samples took significantly less time to be completely digested. The CXL therapy of GAG-depleted samples significantly increased the time that it took for their complete digestion. Control and Enzyme CXL samples showed relatively similar response. The symbols and vertical error bars represent the average and standard deviation, respectively.

4. Discussion

The CXL treatment procedure slows down and even stops further progression of keratoconus by stabilizing the cornea over a long period of time (Raiskup et al., 2023). Since its introduction about three decades ago, significant research has been done to modify (and improve) the original CXL protocol (Spoerl et al., 1998; Wollensak et al., 2003a). The stiffening effect of CXL treatment has been commonly determined by crosslinking healthy corneal samples in vitro and measuring their mechanical properties. However, it is known that keratoconus causes significant variation in composition and microstructure of corneal extracellular matrix. The primary objective of the present study was to investigate the influence of CXL treatment on tensile properties of samples from which KS GAGs were enzymatically removed.

The results shown in Figure 1 characterized the nonlinear stress-strain relationship of control, crosslinked, and GAG-depleted samples. Figure 2 compares the maximum tensile stress and tangent modulus of different groups. First, it is noted that the CXL treatment significantly increased the tensile properties of porcine cornea, which is in agreement with previous reports. For example, the average tensile stress of Control CXL samples at 10% was 200 KPa, same as previous reports by Wollensak et al. and Hatami-Marbini and Jayaram (Hatami-Marbini and Jayaram, 2018a; Wollensak et al., 2003b) . The small variation observed among these studies is because of differences in experimental protocols and samples. Figure 2 also shows that removing KS GAGs significantly reduced corneal tensile properties, which agrees with our recent work (Hatami-Marbini, 2023). In particular, the average maximum tangent modulus of GAG-depleted and control samples was 9.88 MPa and 13.32 MPa, respectively; about 25% softening (the CXL treatment caused ~46% stiffening of control samples). KS is the predominant GAG in the corneal stroma where it appears as the side chain of lumican, keratocan, or mimecan PGs (Quantock et al., 2010). The high resolution electron microscopy studies suggested that KS GAGs form antiparallel duplexes acting as bridges between neighboring collagen fibrils (Lewis et al., 2010; Scott et al., 1981). These macromolecules contribute to the short-range stability of collagen fibril arrangement and their regular hexagonal organization (Knupp et al., 2009). The GAG side chains of PGs can be considered as a ladder-type duplex structure connecting adjacent collagen fibrils together, Figure 4. When the corneal stroma is subjected to tensile load, GAGs behave as mechanical bridges whose deformation affects the load bearing ability of collagen fibrils. In GAG-depleted samples, the density of GAGs significantly decreased, which was clearly captured as a marked reduction in blue color in Alcian blue analyses, Figure 1. Thus, lesser GAG duplexes existed in these strips, which explains why they became significantly soft. This reduction in tensile properties because of KS GAGs removal is similar to the effect of keratoconus on corneal mechanical properties (Andreassen et al., 1980; Funderburgh et al., 1990). It is noted that the findings of the present work suggest that GAG depletion may be used for generating an animal model for keratoconus. The efforts in the literature have so far been focused on using collagenase type 2 treatment in animal models to mimic the lower elastic modulus observed in keratoconus cornea (Qiao et al., 2018; Wei et al., 2023).

Figure 4.

A schematic showing possible interactions of collagen fibrils and GAGs in a) Control, b) Enzyme, c) Control CXL, and d) Enzyme CXL samples. In control samples, all KS GAS are present and form many duplexes resisting against external tensile force. In GAG-depleted specimens, the density of KS GAGs is lower causing a significant reduction of corneal tensile stiffness. The CXL treatment creates additional crosslinks, which are not dependent on KS GAGs, and stiffens both control and GAG-depleted strips.

The CXL procedure improves mechanical properties of keratoconus cornea and stops the progression of this disease. The crosslinking of corneal stroma is a molecular level phenomenon but its exact photochemical mechanisms are not fully known. It is assumed that the interaction between the substrate and excited riboflavin triplets initiates corneal crosslinking (Brummer et al., 2011; Kamaev et al., 2012). The oxygen is essential to drive the process as it modulates Type-I and Type-II photochemical kinetic mechanisms. CXL crosslinks are currently assumed to occur within and between PG core proteins and/or within and between collagen molecules (Hayes et al., 2013; Zhang et al., 2011). The creation of additional crosslinks, Figure 4, causes stiffening of tensile response of Control samples. Furthermore, crosslinking of GAG-depleted samples improved their tensile properties, which implies that depletion of KS GAGs did not interfere significantly with the strengthening effect of CXL treatment. In other words, despite significantly lower density of KS GAGs in GAG-depleted samples, the CXL procedure still caused significant improvement of their tensile properties. This suggests that additional crosslinks caused by CXL treatment, Figure 4, stiffened GAG-depleted samples.

The exact etiology of keratoconus is unknown but stromal protein digestion is believed to contribute in corneal thinning and reduction of its stiffness. Previous studies showed a significant decrease in mechanical properties is accompanied by a significant increase in the activity of protease enzymes in keratoconus cornea (Andreassen et al., 1980; Spoerl et al., 2004). Early studies on the CXL procedure concluded that this treatment improved corneal resistance to enzymatic digestion, which is important for stopping the progression of keratoconus (Aldahlawi et al., 2016; Spoerl et al., 2004). In the present work, we determined the enzymatic resistance of corneal samples to the pepsin degradation solution. Similar to previous studies, we found that the rate of enzymatic degradation of Control corneal samples increased significantly when they were subjected to CXL treatment. Furthermore, we observed that although GAG-depleted samples had significantly lesser resistance to pepsin enzymatic degradation, CXL therapy significantly enhanced their resistance against enzymatic degradation. These observations were in agreement with our mechanical measurement studies and suggested that the CXL procedure had significant stabilizing biochemical effects on corneal structure and mechanics.

In the present work, we did not consider the inhomogeneous microstructure, composition, and mechanical properties of the cornea (Hatami-Marbini, 2018; Jester et al., 2010). Furthermore, we only removed KS GAGs from corneal samples and we did not confirm whether all KS GAGs were removed; however, histological analyses using Alcian blue staining confirmed a significant reduction. In future studies, different types of enzymes can be used to digest other types of GAG. Furthermore, immunohistochemical staining with antibodies against KS can be used to obtain more detailed information about the distribution of GAGs (Daugaard et al., 1991). Despite these and other possible limitations, experimental measurements of the present work provided novel information for the role of KS GAGs in CXL treatment efficacy. We found that removing KS GAGs significantly reduced corneal tensile properties because the number of GAG bridges between neighboring collagen fibrils decreased significantly. Although keratoconus involves other microstructural effects, this ocular disease also causes a significant reduction of KS GAG density and corneal tensile stiffness (Andreassen et al., 1980; Funderburgh et al., 1990). The present work showed that the CXL treatment improved significantly the mechanical response and structural integrity of GAG-depleted samples, which agreed with stiffening effects of CXL on keratoconic cornea.

Highlights.

KS GAG depletion significantly softened corneal tensile properties.

The CXL treatment significantly stiffened GAG-depleted cornea.

The CXL enhanced the resistance of GAG-depleted cornea against enzymatic degradation.

The efficacy of CXL therapy is not significantly dependent on the KS GAG density.