Effect of ultraviolet-A-induced corneal collagen cross-linking in the management of Staphylococcus epidermidis keratitis: In vitro study

Abstract

PURPOSE:

Bacterial keratitis remains a significant cause of corneal morbidity and visual impairment. Increasing resistance to antimicrobial therapy has prompted interest in alternative or adjunctive treatments, including corneal collagen cross-linking (CXL) with ultraviolet-A (UV-A) and riboflavin. The aim of the study is to develop an in vitro human corneal infection model and to evaluate the efficacy of UV-A-induced corneal collagen CXL as an adjuvant treatment for Staphylococcus epidermidis keratitis.

METHODS:

Three human donor corneas were dissected into quadrants and assigned to four experimental groups: (1) control (noninoculated and nontreated), (2) inoculated only, (3) inoculated and postinfection cross-linked, and (4) precross-linked and subsequently inoculated. All infected samples were exposed to S. epidermidis and the CXL procedure was applied either 24 h prior to or 24 h following bacterial inoculation. Corneal tissues were processed and analyzed using hematoxylin and eosin staining and immunohistochemistry.

RESULTS:

The postinfection CXL group demonstrated a significant reduction in bacterial load compared to the inoculated-only and precross-linked groups. Moreover, there was limited bacterial spread, and posterior stromal architecture was better preserved in the treated corneas.

CONCLUSION:

The study successfully established a human in vitro model of bacterial keratitis. UV-A-induced corneal collagen CXL shows promise as an effective adjunctive therapy for S. epidermidis keratitis by mitigating bacterial invasion and preserving tissue integrity.

INTRODUCTION

Infectious keratitis is a leading cause of blindness in the world, with bacterial pathogens being the most common culprits. Among these, Staphylococcus epidermidis, a coagulase-negative staphylococcus, is frequently implicated, especially in individuals who wear contact lenses and those with ocular surface disorders.[1,2] Standard treatment typically involves intensive topical antibiotics. However, the emergence of multidrug-resistant bacterial strains has posed significant therapeutic challenges.[3,4]

A number of recent studies highlighted the increasing resistance to fluoroquinolones, aminoglycosides, and cephalosporins, particularly in developing regions where antibiotic misuse is prevalent.[5,6] This has stimulated interest in alternative or adjunctive therapies, such as photoactivated chromophores for infectious keratitis, including corneal collagen cross-linking (PACK-CXL). Developed initially to halt the progression of keratoconus, CXL uses ultraviolet-A (UV-A) irradiation in combination with riboflavin to induce stromal stiffening and generate reactive oxygen species (ROS) that may exert antimicrobial effects.[7,8]

The effectiveness of CXL as a treatment for infectious keratitis has been the subject of numerous in vitro and animal investigations. CXL has been investigated as an antimicrobial treatment for a number of infections, such as Candida albicans, methicillin-resistant Staphylococcus aureus, Acanthamoeba, Fusarium solani, and Aspergillus fumigatus. According to the studies, UV and riboflavin work better together to reduce the quantity of bacteria than either UV or riboflavin alone.[9,10,11,12] They came to the conclusion that bacterial keratitis might 1 day be treated with riboflavin/UVA.

The antibacterial effectiveness of PACK-CXL has been examined in a number of clinical investigations. A study conducted in 2000 was the first to look at how CXL treatment affected noninfectious corneal ulceration and melting. Four individuals with corneal ulcers and corneal melting received CXL treatment in that trial. In three out of four patients, the study showed that CXL was successful in stopping the corneal melting. The remaining patient’s increasing keratitis was brought on by an immunological response rather than a chronic active fungal infection.[13,14]

A systematic review and meta-analysis reported 12 published studies involving 104 eyes treated with corneal collagen CXL, with 85% of cases showing cessation of corneal melting.[15]

A number of evidence supports PACK-CXL as a potential adjunctive treatment for early-to-moderate microbial keratitis, with studies reporting clinical improvements in ulcer size, infiltration depth, and epithelial healing.[14,16] However, there remains a need for standardized protocols and comparative studies to clarify its role in conjunction with conventional therapy.

The current study aims to investigate the efficacy of UV-A-induced corneal collagen CXL against S. epidermidis using an in vitro human cornea. Through examining tissue morphology, bacterial presence, and structural integrity, it seeks to expand the understanding of PACK-CXL’s utility in managing resistant infectious keratitis.

METHODS

Sample collection and ethical approval

Three healthy postmortem human corneal buttons that were inappropriate for transplantation were obtained from the Manchester Eye Bank (UK). Incomplete donor histories or insufficient endothelial cell counts resulted in the exclusion of the tissues. The study received ethical approval and informed consent from donors from the UK National Health Research Committee. Until used, the corneas were stored at 34°C in Eagle’s Minimal Essential Medium (MEM) supplemented with 2% fetal bovine serum (FBS), antibiotics, and antifungal agents. In addition, acquiring corneas from donors is a very long complicated process and requires a lot of effort and paperwork with limited access to resources. Despite the small sample size, we believe the consistent histological patterns across all duplicates support the robustness of our findings.

Organ culture and preparation

DMEM, which was supplemented with 10% FBS, glutamine, penicillin-streptomycin, and hydrocortisone, was used to incubate the corneal buttons. Before dissection and experimentation, each cornea was cultured in a 6-well plate at 37°C with 5% CO₂ for 72 h. A single cornea was used in the first set, while two corneas were used in the second. Two experimental sets were conducted.

Experimental design

A sterile surgical blade and forceps were used to dissect the corneal buttons into quarters in an empty Petri dish [Figure 1]. After dissection, the corneas were transferred to a new 6-well plate that contained 2 ml of fresh culture medium [Figure 2].

Figure 1.

Human corneal buttons are dissected into the quadrants for experimental classification

Figure 2.

The corneas were inserted into a new well plate that contained 2 ml of fresh culture medium

Following the dissection of the corneal buttons, the quarters were assigned to one of the four regimes set out below:

1. Control-infection-free, untreated

2. Inoculated and Pre-CXL-treated with CXL prior to infection

3. Only inoculated-infected without CXL

4. Inoculated and Post-CXL-infected and later treated with CXL.

Bacterial inoculation

A clinical strain of S. epidermidis (M28) was cultivated overnight in BHI broth at 37°C and adjusted to 4 × 10⁸ CFU/mL. Two linear scratches were applied to the corneal epithelium to imitate trauma. Each group received a 10 μL aliquot of bacterial suspension, except for the control. All samples were cultured for 24 h at 37°C after inoculation.

Procedure for corneal collagen cross-linking

The Dresden protocol was used to perform CXL.[7] The corneal sector was installed on an artificial anterior chamber (SD Healthcare, Irlam, Manchester, UK) with a retainer and locking ring to hold it in place [Figure 3].

Figure 3.

Illustrates the artificial anterior chamber. The cornea sector was affixed to the artificial anterior chamber prior to the cross-linking procedure

The CXL procedure was conducted without epithelial removal. An isotonic riboflavin solution (0.1% [w/w] riboflavin, 20% [w/w] dextran T500; both from Sigma-Aldrich) was applied to the corneal surface at 5-min intervals over 30 min. The cornea was subjected to UV-A irradiation using a medical electrical UV-A light emitter (370 nm, irradiance of 3 mW/cm²) (VEGA, C.S.O. srl, Florence, Italy) [Figure 4]. This process delivered a cumulative UV-A irradiation dose of 5.4 J/cm² to the corneal surface over a duration of 30 min, accompanied by the application of isotonic riboflavin at 5-min intervals. Subsequent to this procedure, the corneas underwent multiple washes with sterile phosphate-buffered saline (PBS, Sigma, Aldrich) and were then placed back into the 6-well culture plate, accompanied by 2 ml of fresh organ culture medium per well. The samples were incubated at 37°C for 24 h.

Figure 4.

(a and b) H and E staining of the control (noninfected) corneal quadrant showing normal epithelium and intact stromal structure (10x magnification and 20x magnification)

Tissue fixation and embedding

Samples were fixed in 4% paraformaldehyde and stored in PBS-sucrose solution at 4°C, 48 h postinoculation. Tissues were embedded in OCT compound, snap-frozen in liquid nitrogen, and subsequently stored at −40°C.

Analysis of histology and immunohistochemistry

Corneal cryosections, measuring 20 μm in thickness, were prepared using a cryostat and subsequently mounted on Superfrost^® Plus slides.

Hematoxylin and eosin staining

Staining is employed to assess general morphology and cellular architecture. Sections underwent rehydration, staining, dehydration via graded ethanol, clearing, and were subsequently mounted with DPX. Images were obtained using a Zeiss microscope.

Periodic acid-Schiff staining

Periodic acid-Sciff staining utilized for the visualization of bacterial clusters. Sections underwent treatment with periodic acid, Schiff’s reagent, and acid-alcohol, followed by mounting.

Immunohistochemistry

Immunohistochemistry (IHC) staining employed antibodies against cytokeratin 12 (epithelium) and CD34 (stromal cells), with nuclear counterstaining using DAPI. Fluorescent secondary antibodies (Alexa Fluor® 488 and DyLight® 550) were used. Sections were analyzed with a B × 60 fluorescence microscope and imaged using Evolution camera software.

RESULTS

The results demonstrated distinct histological and structural differences among the experimental groups. All observations represent qualitative histological findings; no quantitative microbiological endpoints were assessed.

Histological analysis using hematoxylin and eosin (H and E) staining provided insights into tissue preservation, inflammatory infiltration, and epithelial integrity. Periodic acid-Schiff (PAS) and IHC staining further confirmed the presence of bacteria and epithelial/stromal alterations. In the control group, corneal architecture was well-preserved with intact epithelium and a compact stromal matrix. No signs of inflammation or damage were observed [Figure 4a and b].

In contrast, the infected corneal sections without treatment showed extensive epithelial erosion, significant stromal infiltration, and disorganized tissue structure, indicating bacterial colonization and damage [Figure 5a and b].

Figure 5.

(a and b) H and E staining of the infected cornea showing epithelial disruption and inflammatory infiltration throughout the stroma. 10x magnification for a and b. Showing the invasion of bacteria (arrowed)

In the group treated with CXL postinfection, corneal tissue showed improved structural integrity, reduced bacterial spread, and better preservation of epithelial layers. Inflammatory cells were less prominent, and the stromal collagen appeared more organized [Figure 6a and b]. The images of the PAS-stained corneal sections from the Control group are shown to be healthy, intact and free from bacterial invasion [Figure 7a and b].

Figure 6.

(a and b) H and E staining of the cornea treated with cross-linking after infection. Note partial epithelial regeneration and reduced stromal inflammation. 10x magnification for a and b. bacteria through the stroma reduced after CXL treatment.(arrowed)

Figure 7.

(a and b) PAS-stained corneal section of a Control cornea showing the corneal epithelial and stromal layers at 4x magnification

The section of the cornea stained with PAS from the infected group, which was inoculated with S. epidermidis, is shown in Figure 8a and b. An extensive bacterial population infiltrates the corneal epithelial and stromal layers 48 h after inoculation.

Figure 8.

(a and b) Sections of an inoculation cornea stained with Periodic acid-Schiff show that the cornea is filled with clusters of bacteria, as shown by the arrows

Figure 9a and b present the images of the stained sections of the inoculated and crosslinked cornea. The images indicate a significant decrease in bacterial load following UV/CXL treatment when compared to the inoculated cornea that did not receive CXL treatment.

Figure 9.

(a and b) It shows a reduction in bacterial invasion (arrowed) after CXL treatment compared to the Inoculated cornea. 10x magnification for a and 4x for b

Comparing the immunostained corneal sections to the sections stained with H and E and PAS indicated comparable outcomes. Two antibodies – cytokeratin 12 and CD34 – were used for immunostaining. In order to distinguish between the corneal epithelial cells (green) and the stromal tissues (magenta), the cytokeratin 12 antibody was used. The blue cell nuclei were visualized using DAPI.

Using K12 and DAPI staining, Figure 10a shows an image of the control cornea’s undamaged, infection-free corneal section. Bacterial infection is not evident in the cornea’s epithelial layer or its intact structure. An image of the CD34 and DAPI stained corneal section from the uninfected, untreated control cornea is shown in Figure 10b. In this image, you can see that the cornea is intact and free of bacteria.

Figure 10.

(a and b) Illustrate a Periodic acid-Schiff-stained corneal section of an inoculated and crosslinked cornea, demonstrating a decrease in bacterial load (indicated by arrows) following cross-linking treatment

Structure damage caused by S. epidermidis invasion is shown in the K12 and DAPI stained corneal section [Figure 11a]. The bacterial infection caused significant damage to both the epithelial layer and the stromal tissue, as seen by the staining of the corneal section with CD34 and DAPI [Figure 11b]. Although stromal keratocytes are not evident, intact epithelial cells can be seen [Figure 11a].

Figure 11.

(a) We can see a magnified image of the control cornea’s immunostained segment (×4), where the green color represents the epithelial layer and the blue colour represents the cell nuclei. The image shows a cornea free of bacteria, as well as the epithelial layer (marked by an arrow) and the intact corneal structure. (b) The Control cornea (×4) immunostained with CD34 (magenta) and DAPI (blue) illustrates stromal tissue and cell nuclei, respectively; the image shows that there is no bacterial infection in the corneal epithelium and stromal layers

In Figure 12a, staining the cornea with K12 and DAPI demonstrates that the spread of bacteria decreased following CXL treatment. In Figure 12b, the bacterial invasion was reduced after UV/CXL therapy, as evidenced by the corneal segment stained with CD34 and DAPI.

Figure 12.

(a) Damage to the corneal structure caused by bacterial invasion is seen in this magnified image of an immunostained corneal section from the inoculation cornea (×4). The green colour represents the epithelial layer, while the blue colour represents cell nuclei; (b) At a magnification of × 4, this image shows the infected cornea’s immunostained corneal section. With the use of DAPI, the stromal tissue is magenta-colored, and the cell nuclei are blue-colored. The epithelial layer and the stroma are both severely damaged because of the bacterial invasion

DISCUSSION

The current study aims to gain an initial understanding of how PACK-CXL might affect corneal tissue exposed to S. epidermidis under controlled in vitro conditions. In the treated corneas, we noted a decrease in the amount of bacterial material together with clearer preservation of epithelial and stromal structures. These changes point toward a possible tissue-level response induced by the CXL process, one that could influence bacterial spread or stromal degradation. However, these observations remain qualitative, and they should be interpreted with caution as they did not include quantitative microbiological testing. Earlier work has examined the antimicrobial potential of photoactivated riboflavin. One study demonstrated a significant reduction in bacterial growth using CXL against a range of pathogens, including S. epidermidis.[17] Another study reported improved clinical outcomes in patients with infectious keratitis treated with PACK-CXL.[18] Our results fall broadly within the patterns described in those studies, although our data cannot confirm antimicrobial activity since we did not perform CFU counts or direct therapeutic comparisons.

More recent studies suggested that PACK-CXL might have value as an adjunctive option. One study concluded that PACK-CXL is a potential adjuvant therapy for infectious keratitis, particularly in early or moderate stages.[19] Another study highlighted the safety and efficacy of high-fluence PACK-CXL protocols in treating resistant microbial keratitis.[20] Moreover, rapid clinical resolution of infiltration and a reduction in ulcer size within 14 days have been reported following PACK-CXL.[14,16]

The primary mechanism for CXL’s antibacterial effect is the production of ROS in response to UV-A exposure to riboflavin. These ROS can disrupt bacterial membranes and DNA while also strengthening the corneal stroma through CXL, thereby inhibiting further microbial invasion.[8,21]

In terms of CXL therapy protocol, it was suggested that the effectiveness of CXL might be improved by increasing the stromal concentration of riboflavin through epithelium removal.[22,23] However, the proponents of CXL without epithelial removal claim that the surgery is less painful, has a lower risk of postoperative complications, and shortens the recovery period after treatment.[24,25] In our study, the procedure was done without removing the epithelium. The stiffening effect observed in all crosslinked corneas indicates that the riboflavin was able to disperse throughout the cornea, which is why it was successful. Both the effectiveness and safety of the CXL treatment depend on riboflavin’s diffusion through the corneal layers and its saturation of the superficial stroma.[25] We used of PAS staining revealed a reduction in bacterial clusters in posttreatment samples, and IHC confirmed the preservation of epithelial and stromal markers (K12 and CD34), supporting the effectiveness of CXL.

Recent research has increasingly concentrated on enhancing PACK-CXL parameters, particularly through high-fluence and accelerated methods that aim to increase the efficiency of antimicrobial drugs while inflicting minimal damage to the stroma. It was found that a 10 J/cm² high-fluence PACK-CXL method was safe and effective for treating infectious keratitis in patients.[26] High-fluence accelerated PACK-CXL may eliminate bacteria in an ex vivo porcine corneal model activated by UV-A or green light. The data suggested that altering the fluence and duration of irradiation significantly affects microorganism eradication and corneal integrity.[20] These evolving findings provide background to our study but remain outside the scope of what can be concluded from our qualitative observations.

This study has several limitations. The sample size was small, and the in-vitro model cannot replicate immune responses, tear-film interactions, or antibiotic penetration seen in vivo. Moreover, the absence of quantitative microbiological assays and the lack of antibiotic-only or UV-only control groups. This could mean that the patterns we observed should be viewed as exploratory and mechanistic rather than indicative of clinical efficacy.

CONCLUSION

The histological patterns we observed after treatment suggest that CXL can alter both bacterial distribution and stromal appearance, but these findings remain qualitative and should not be interpreted as proof of antimicrobial efficacy. Because the study did not include CFU measurements or comparison groups such as antibiotic-only or UV-only arms, the changes we report cannot be attributed to CXL alone.

The current model helps clarify possible mechanisms; however, studies that incorporate quantitative microbiological methods and direct treatment comparisons will be essential to confirm whether the tissue-level changes seen here translate into meaningful antimicrobial or therapeutic effects in vivo.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.