Corneal molecular and cellular biology for the refractive surgeon: the critical role of the epithelial basement membrane

Abstract

Purpose:

Provide an overview of the recent advances involving the corneal molecular and cellular biology processes involved at the wound healing response after surface ablation surgery.

Methods:

Literature review

Results:

The corneal wound healing response is a complex cascade of events that impacts the predictability and stability of keratorefractive surgical procedures such as photorefractive keratectomy and LASIK. The generation and persistence of corneal myofibroblasts—contractile cells with reduced transparency—arise from the interaction of cytokines and growth factors such as TGF-βand IL-1 produced by epithelial and stromal cells in response to the corneal injury. Myofibroblasts, and the opaque extracellular matrix they secrete into the stroma, disturb the precise distribution and spacing of collagen fibers related to corneal transparency lead to the development of vision-limiting corneal opacity (haze). The intact epithelial basement membrane has a pivotal role as a structure that regulates corneal epithelial-stromal interactions. Thus, defective regeneration of the epithelial basement membrane after surgery, trauma, or infection leads to the development of stromal haze. The apoptotic process following laser stromal ablation, which is proportional to the level of attempted correction, leads to an early decrease in anterior keratocyte density and diminished contribution of these non-epithelial cells of components such as perlecan and nidogen-2 required for normal regeneration of the epithelial basement membrane. Haze persists until late repair of the defective epithelial basement membrane.

Conclusions:

Defective regeneration of the epithelial basement membrane has a critical role in determining whether a cornea heals late haze after photorefractive keratectomy or with scaring at the flap edge in LASIK.

PRECIS

The variable nature of the corneal wound healing response impacts the predictability and stability of surface ablation surgeries such as photorefractive keratectomy. The epithelial basement membrane plays a critical role in this important reparatory process.

INTRODUCTION

The corneal wound healing response is a complex cascade of events that functions to repair corneal injuries—such as trauma, infection or injury produced by surgery—and reestablish normal tissue function.¹ Considerable effort has been devoted to understanding and manipulating these processes that underlie and impact the predictability and stability of keratorefractive surgical procedures..^{1, 2}

Prior studies have detailed the interaction of cytokines and growth factors produced by epithelial cells, lacrimal glands, and bone marrow-derived cells, that are triggered in response to corneal injury and modulate stromal keratocyte “activation” and the development of haze-associated myofibroblasts.³ These complex cellular interactions are involved in extracellular matrix reorganization and stromal remodeling that will define whether corneal repair will be achieved at the expense of vision-limiting corneal opacity (haze). It has recently been shown that regeneration of an intact epithelial basement membrane, strategically placed between epithelium and stroma and acting as a critical structure that regulates the function of transforming growth factor beta (TGF-β) and other cytokines, plays a pivotal role in determining the regenerative versus the fibrotic character of corneal repair.^4-6

This review will highlight key corneal molecular and cellular biological processes involved in the wound healing response after photorefractive keratectomy (PRK) and, to an extent, laser in situ keratomileusis (LASIK), and detail important insights gained since our last review of this topic published in the Journal of Refractive Surgery.⁷ Special attention will be given to the structure, function and regeneration of epithelial basement membrane (BM) and to recent observations that revealed that defective regeneration of the epithelial BM is the primary disorder associated with development of haze after surface ablation procedures such as PRK.

Understanding corneal haze

Corneal stromal opacity, or scarring, also referred to as haze (the term haze was coined in the 18^th century probably as a derivative of hazy, a term used along with fog and mist to express “opaqueness of the atmosphere”; lack of distinctness or clarity), is opacity noted immediately beneath the epithelium as a result of the wound healing process after corneal surgery or injury. The corneal wound healing response after PRK is typically more intense in the central cornea than after LASIK for similar levels of correction due to differences in the extent of corneal epithelial and epithelial basement membrane injury and the ensuing wound healing response. Therefore, complications related to wound healing, such as haze and regression, tend to be more significant after PRK and other surface ablation surgeries.⁸ PRK surgery requires extensive epithelial removal, including removal of the central corneal epithelial BM, and subsequent photoablation of the anterior corneal stromal tissue to reshape corneal curvature and correct refractive errors. In normal LASIK surgery, the corneal epithelial injury is limited to the edges of the hinged corneal flap that is created and lifted before the laser stromal sculpting.⁸

The generation and persistence of corneal myofibroblasts, which are contractile cells associated with reduced transparency that is attributable to decreased intracellular crystallin production in these cells and their production of disordered extracellular matrix, have been demonstrated to be the primary cellular contributors to the development of corneal haze.⁹ TGF-β and other cytokines derived from the overlying epithelium penetrate into the stroma at increased levels when there are defects (ultrastructural and functional) in the epithelial BM and thereby modulate myofibroblast generation from progenitor cells.

There are two different types of corneal haze observed after PRK. The most common type is a clinically insignificant haze that occurs after nearly all PRK surgeries, including those with perfect clinical outcomes. This mild transitory opacity is noted in the first weeks to months after surgery and is attributable to corneal fibroblast generation from keratocytes and altered crystallin production in these activated cells. The second type of haze, also known as late haze, is clinically significant (with symptoms including halos, glare, regression of the surgical effect, and decreased visual acuity) and affects corneas with excellent outcomes during the early post-operative period. Late haze typically develops at around two to three months after surgery, with a peak of incidence at three to six months after surgery in humans (Figure 1A). Late haze is attributable to the development of myofibroblasts from both keratocyte- and bone marrow-derived precursor cells and the opaque extracellular matrix these cells secrete into the stroma. The incidence of late haze, which has been remarkably diminished by the clinical use of mitomycin C on the ablated stroma after PRK, is proportional to the level of attempted correction (depth of the ablation), being very rare after corrections of less than five to six diopters with most modern excimer laser systems. Smoothness of the stromal surface, the time necessary to repair the epithelial defect, ultra-violet light exposure, decreased tear film production, laser ablation characteristics and genetic influences have been also described as risk factors for haze formation after PRK.^{7, 10}

Figure 1.

Slit-lamp photographs of late corneal haze after refractive surgery. (A) Central corneal haze after high-correction PRK ablation. (B) Circumferential corneal haze in the flap edge of LASIK treatment. Magnification 20X.

Another important clinical feature of severe late corneal haze after PRK is its response to topical corticosteroid treatment. It has been previously reported that a small percent (10-15%)⁷ of corneas that develop late haze after PRK are corticosteroid-sensitive and will have a marked response to an intensive trial (for example, eight times a day for one week with 1% prednisolone acetate) with a decrease in haze and improvement in symptoms. In these cases, corticosteroids should be gradually tapered over months so haze does not regenerate. Unfortunately, the other 85-90% of corneas that develop late haze after PRK are non-responders (corticosteroid-resistant) and will not benefit from topical corticosteroid treatment. Because our inability to clinically predict whether a cornea that develops haze will be responsive or nonresponsive to corticosteroids, it is our practice to always initiate a one to two week trial of corticosteroids when late haze occurs, with subsequent treatment dictated by the response to this treatment. The underlying biology accounting for these varying responses to corticosteroids remains uncertain, but it is our working hypothesis that the difference is related to whether the dominant precursor cells that generated the myofibroblasts in a particular cornea were keratocytes (steroid unresponsive) or bone marrow-derived fibrocytes (steroid responsive).^{11, 12}

Refractive surgeons use mitomycin C (MMC), an antibiotic classified as an ankylating agent, to prophylactically prevent haze formation after surface ablation procedures.⁸ The rationale for using MMC relies on its ability to block DNA replication, RNA transcription and protein synthesis,¹³ thereby inhibiting the proliferation of keratocytes and the development of myofibroblasts from precursor cells.⁸ The consequent diminished cellular repopulation of the anterior stroma after MMC treatment, leading to an early decrease in keratocyte density and decreased collagen production,¹⁴ has raised concerns about long-term effects, although none have materialized over the past 15 years. The observation that there is less tendency for spontaneous resolution of “breakthrough haze” (haze generation despite administration of MMC after PRK)—even three or more years after surgery—is important evidence that diminished keratocyte density and function compromises the ability of the cornea to restore transparency.

Conversely, LASIK has a low incidence of late corneal haze.¹⁵ This lamellar procedure entails the creation of an epithelial-stromal hinged flap that preserves the integrity of the central corneal epithelium and epithelial BM—that limits epithelial-stromal interactions responsible for driving myofibroblast generation, except at the edge of the flap where the epithelial BM is damaged and myofibroblast-related haze is commonly noted. Thus, in LASIK the most significant healing response and resultant scarring develops at the flap margin, where the incisional breaks in the epithelial BM allow direct contact of epithelium-derived cytokines such as TGF-β and the stromal cells. This process leads to peripheral myofibroblasts formation, clinically recognized as a circumferential haze at the edge of the flap (Figure 1B). However, central interface haze has been reported after LASIK under certain circumstances—including after diffuse lamellar keratitis, with retention of epithelial debris in the interface, and the production of donut-shaped¹ or thin flaps. The “thin-flap femtosecond laser LASIK” technique involves the creation of thinner flaps (usually 90-100 μm rather than 100-120 μm) designed to minimize the biomechanical injury to the anterior corneal stroma and, therefore, reduce the risk of postoperative ectasia. Following this thin-flap LASIK approach there is a higher incidence of central corneal haze.^{10, 16} This greater haze formation is likely related to epithelial BM injury attributable to the proximity of the femtosecond laser photodisruption process to this critical regulatory structure.¹⁰

The opacity in late stromal haze is a consequence of decreased transparency of myofibroblasts themselves due to diminished crystallin protein production (decreased production of proteins such as transketolase and aldehyde dehydrogenase 1) compared to keratocytes, in addition to the large quantities of disorganized extracellular matrix with differing collagen composition these cells excrete.⁹ This altered matrix disturbs the very precise distribution and spacing of collagen fibers related to corneal transparency.⁸

Myofibroblasts in the cornea can be derived from both keratocyte-derived cells and bone marrow-derived cells^14-17 that migrate into the corneal stroma from the limbus. The latter cells are attracted by pro-inflammatory cytokines and chemokines released by the injured epithelium and up-regulated in keratocytes.^{11, 17} Recent studies used chimeric mice with bone marrow-derived cells labeled with green fluorescent protein (GFP+) to conclusively demonstrate that corneal myofibroblasts can be derived from either bone marrow-derived or keratocyte-derived precursor cells (Figure 2).^16,17 It is hypothesized that the dominant precursor cell in a particular cornea is determined by the type and/or extent of injury, genetic factors, and even by juxtacrine interactions between the two cell types. No differences in function between myofibroblasts derived from keratocytes and myofibroblasts derived from bone marrow-derived cells have been discovered. It is likely, however, that mature myofibroblasts retain some intrinsic features of their precursor cells, such as steroid sensitivity.

Figure 2.

Chimeric (bone marrow transplant from green fluorescent protein [GFP+] mouse) mouse cornea with haze at 1 month after irregular PTK (method used to produce haze in mice). In the overlay (A), a high concordance can be seen between the red stain for alpha-smooth muscle actin marker for myofibroblasts (B) and the green stain for GFP+ (C) in several cells in the anterior stroma (arrows). Blue is DAPI staining of nuclei. Magnification: 800X. Reprinted by permission from Barbosa et al. Corneal myofibroblast generation from bone marrow-derived cells. Exp. Eye Res., 2010; 91:92-6.

Cellular biology of corneal repair

Apoptosis of anterior stromal keratocytes is the first detectable event following epithelial injury and occurs almost immediately after any epithelium injury.¹⁸ Wilson et al.¹⁹ first reported the morphologic changes of cell shrinkage, chromatin condensation and fragmentation, and cellular blebbing with the formation of membrane bound structures containing intracellular organelles called apoptotic bodies in superficial keratocytes beneath the site of corneal epithelial injury. The keratocyte apoptosis process can be demonstrated by classical morphologic cellular changes seen with transmission electron microscopy and with the terminal deoxyribonucleotidyl transferase-mediated dTUP-digoxigenin nick end label (TUNEL) assay, which detects fragmented ends of the DNA strands produced in the cell during the apoptotic process (Figure 3).

Figure 3.

Central cornea of rabbit eyes assayed for apoptotic cells by terminal deoxyribonucleotidyl transferase-mediated dUTP nick end labeling (TUNEL). (A) Rabbit eye 4 hours after −4.5D PRK ablation. (B) Rabbit eye 4 hours after −9.0D PRK ablation. Arrowheads indicate corneal surface in each panel. Arrows indicate some of the apoptotic keratocytes. Magnification: 200X.

Apoptosis is a programmed, involutional form of cell death that produces minimal collateral damage or inflammation in the surrounding tissue.²⁰ Pro-apoptotic cytokines, including interleukin-1 (IL-1) and soluble Fas ligand,²¹ are produced by epithelial cells and comprise a surveillance system, ready to bind to and activate receptors on keratocytes once the integrity of the epithelium and BM is compromised. The localization and extent of keratocytes apoptosis correlate to the type of corneal epithelial injury and, consequently, influence all the later events of the wound healing process. This explains, at least in part, the biological and clinical differences noted between low and high PRK corrections with regards to epithelial hyperplasia, treatment regression and haze generation.

The remaining keratocytes surrounding the zone of apoptosis in the peripheral and posterior stroma begin to proliferate around 12 to 24 hours after corneal injury. This mitotic process, which can be detected by bromodeoxyuridine incorporation or immunocytochemical staining for a mitosis-specific antigen ki-67,²² begins to repopulate the anterior stroma and gives rise to keratocyte-derived precursors to myofibroblasts that begin a developmental program that can lead to mature myofibroblasts.¹⁸ Simultaneously, bone marrow-derived precursors are drawn into the central superficial stroma from the limbal blood vessels.^16,17 Whether these myofibroblast precursors continue to develop into mature haze-associated myofibroblasts or undergo apoptosis themselves is determined by the localized stromal concentration of TGF-β and platelet-derived growth factor (PDGF) that penetrate from the overlying epithelium. Complete regeneration of the epithelial BM leads to a decreased stromal concentration of these growth factors and apoptosis of the myofibroblast precursors. Conversely, persistent structural and functional defects in the epithelial BM results in ongoing penetration of TGF-β⁹ and PDGF²³ and full development of mature myofibroblasts.²⁴

TGF-β and IL-1 have antagonistic opposing effects on corneal myofibroblast generation and persistence.²⁵ TGF-β is a critical driver of myofibroblast development from precursor cells and persistence of these cells, and also blocks IL-1-stimulated myofibroblast apoptosis when TGF-β is present at sufficiently high concentration. TGF-β is produced in high levels by epithelium but also becomes expressed in corneal stromal cells, including invading monocytes and corneal fibroblasts, after corneal injury, although this stromal expression seems to fade rapidly as the wound healing response progresses. Conversely, IL-1, unopposed by TGF-β, acts as a promoter of myofibroblast apoptosis. IL-1 seems to act through both paracrine and autocrine mechanisms. In the former case, IL-1 produced by corneal fibroblasts, keratocytes and inflammatory cells triggers myofibroblast apoptosis when TGF-β levels decline to sufficiently low levels in the anterior stroma. In the latter case, IL-1 expressed by myofibroblasts themselves triggers myofibroblast apoptosis (autocrine suicide) when TGF-β levels fall. Thus, it is believed that modulation of myofibroblast differentiation, viability and programmed cell death probably depend on the balance of TGF-β and IL-1 levels in the anterior stroma and these levels are in turn determined by whether or not there is functional regeneration of the epithelial BM after surgery, infection or injury.^{25, 26}

The intact epithelial BM structure acts as a barrier between corneal epithelium and stroma, preventing persistent penetration of epithelium-derived TGF-β and PDGF into anterior stroma at sufficient levels to drive mature myofibroblast development and maintain the viability of these cells once they are generated. Thus, in every cornea that has PRK the process of myofibroblast generation is initiated and a competition of sorts ensues between the processes leading to full development of opacity generating myofibroblasts and the apoptotic death of these cells and repopulation of the anterior stroma with normal keratocytes. Which process dominates in a particular cornea is determined by whether or not the epithelial BM fully regenerates. Even when severe late haze occurs following PRK, it usually resolves spontaneously over a period of one to five years after surgery.⁷ In these cases, the epithelial BM slowly recovers its structure and function, epithelium-derived TGF-β and PDGF levels fall to the point that TGF-β/IL-1 balance reverses, myofibroblasts undergo apoptosis and corneal transparency is restored by keratocytes repopulating the anterior stroma and remodeling the disordered matrix laid down by the myofibroblasts. This process can be clinically appreciated as transparent areas termed “lacunae” beginning to appear in an area of confluent corneal haze (Figure 4).

Figure 4.

Lacunae developing in the previously confluent haze of a human cornea at 2 years after high correction PRK performed without mitomycin C. Magnification: 30X.

Epithelial basement membrane: the master regulator of the stromal wound healing process

The epithelial BM is a highly specialized extracellular matrix approximately 50-100 nm in thickness that is assembled from four primary components: laminins, collagens, heparan sulfate proteoglycans (such as perlecans) and nidogens,^{27, 28} although the precise composition is tissue-specific. At the transmission electron microscopic (TEM) level the intact epithelial BM is noted to have two prominent layers—termed the lamina densa and lamina lucida (Figura 5).²⁹ TEM has been used to assess epithelial BM ultrastructure in rabbit corneas that underwent high and low correction PRK treatments with or without the development of late stromal haze, respectively.²⁴ The analysis revealed abnormal regeneration (absence of distinguishable lamina lucida and lamina densa layers) of the epithelial BM with disorganized extracellular matrix and high myofibroblast density immediately beneath the epithelium in corneas with severe haze at one month after −9.0 D PRK. Conversely, ultrastructural analysis showed complete regeneration of the epithelial BM lamina lucida and lamina densa, indistinguishable from unwounded controls, in corneas that had low −4.5 D PRK, with the normal organized pattern of the extracellular matrix and no myofibroblasts detectible in these latter corneas without haze (Figure 5). Interestingly, small areas of haze sometimes develop in corneas that have lower correction PRK. These areas also lack normally regenerated epithelial BM when they are analyzed in rabbits with TEM.³⁰ This defective regeneration of epithelial BM correlates with the light microscope observations of Netto and coworkers⁵ that myofibroblasts expressing α-smooth muscle actin (α-SMA+) were commonly noted just beneath disruptions in the epithelial BM of corneas with haze at one month after PRK.

Figure 5.

Transmission electron microscopy of sections from untreated and treated central corneas of rabbits. (A) A representative image from an untreated-control cornea. Regular pattern of lamina lucida and lamina densa (arrows) is clearly seen between the epithelium (e) and stroma (s). (B) Representative image of rabbit cornea at 1 month after −4.5D PRK with a regenerated epithelial basement membrane, presenting ultrastructure of lamina lucida and lamina densa (arrows) very similar to the control cornea in (A). (C) Representative image of rabbit cornea at 1 month after −9.0D PRK showing no lamina lucida-like or lamina densa-like structures suggestive of regenerated epithelial basement membrane. Also note the disorganized extracellular matrix (d) and myofibroblasts (m) with large amounts of rough endoplasmic reticulum in the stroma beneath the epithelium. Magnification: 30,000X.

Previous dogma has been that the epithelial cells themselves exclusively produce epithelial BM components. However, recent work has shown that non-epithelial cells, such as the stromal keratocytes, contribute several components of the epithelial BM that are critical in normal regeneration.^{30, 31} Thus, our laboratory has recently reported up-regulated production of both perlecan and nidogen-2, key components of the lamina lucida and lamina densa, in anterior to mid-stromal keratocytes after epithelial scrape injury to human corneas (Figure 6).³³ The level of production of perlecan and nidogen-2 by stromal cells during the wound healing response to PRK may be a critical determinate of whether or not the epithelial BM fully regenerates and, therefore, whether or not the cornea develops myofibroblasts and haze. Thus, in corneas with high correction PRK, more anterior keratocytes undergo apoptosis,²⁰ and, therefore, less perlecan and nidogen-2 are contributed by stromal cells to regenerate the epithelial BM.

Figure 6.

Immunohistochemistry for nidogen-2 in human corneas. (A) Nidogen-2 protein was noted in the epithelium (e) and epithelial basement membrane, as well as at low levels in keratocytes (arrows) in unwounded control corneas. (B) In a cornea at 30 minutes after epithelial scrape, nidogen-2 protein was up-regulated in stromal keratocytes (arrows), including anterior stromal keratocytes in the early stages of apoptosis detected with the TUNEL assay (not shown). Magnification: 400X. Reprinted by permission from Torricelli et al. Exp Eye Res., 2015;134:33-8.

In conclusion, the epithelial BM and its regeneration are critical determinants of whether the cornea heals with transparency or opacity after PRK. Further studies will likely lead to pharmacological agents that can promote epithelial BM regeneration and inhibit stromal haze after PRK without resorting to altering the cellularity of the stroma with mitomycin-C.