Tear proteases and protease inhibitors: potential biomarkers and disease drivers in ocular surface disease

Abstract

Tears are highly concentrated in proteins relative to other biofluids, and a notable fraction of tear proteins are proteases and protease inhibitors. These components are present in a delicate equilibrium that maintains ocular surface homeostasis in response to physiological and temporal cues. Dysregulation of the activity of protease and protease inhibitors in tears occurs in ocular surface diseases including dry eye and infection, and ocular surface conditions including wound healing following refractive surgery and contact lens wear. Measurement of these changes can provide general information regarding ocular surface health and, increasingly, has the potential to give specific clues regarding disease diagnosis and guidance for treatment. Here we review three major categories of tear proteases (matrix metalloproteinases, cathepsins, plasminogen activators) and their endogenous inhibitors (tissue inhibitor of metalloproteinases, cystatins, plasminogen activator inhibitors), and the changes in these factors associated with dry eye, infection and allergy, refractive surgery, and contact lenses. We highlight suggestions for development of these and other protease/protease inhibitor biomarkers in this promising field.

1. Tears as a Source of Biomarkers

1.1. Tear Production

The ocular surface is covered by a thin fluid named the tear film, which provides the avascular cornea with nutrients, promotes oxygen uptake, provides lubrication for comfort, provides light refraction to the eye, and protects against infection.¹ Closest to the ocular surface is a hydrophilic glycocalyx, consisting of cell associated mucins and galectins produced by corneal and conjunctival epithelial cells.^{2, 3} The mucoaqueous layer contains fluid, proteins and electrolytes secreted primarily by main and accessory lacrimal glands as well as soluble mucins.¹ Covering but integrated with the mucoaqeuos layer, there is a thin layer of lipids secreted from the meibomian glands in the eyelid.⁴ The complex composition of this biofluid is critical to its vital role in ocular surface homeostasis.

The lacrimal gland producing the aqueous component is the principal source of tear proteins. Secretion from the lacrimal gland contributes constitutively to produce basal tears, and provides regulated secretion of proteins and fluid in response to stimulation by corneal sensory nerve fibers, so-called reflex tears.⁵ The composition and protein content differs between basal and reflex tears.^{6, 7} The lipid content of reflex tears is reduced,⁸ while osmolality falls within the same range in both basal and reflex tears.⁹ Other physiological factors that affect tear composition include prolonged closure of the eyelids, diurnal and seasonal variations.^10–13

The total protein concentration of basal tears is high, ranging from 6–11 mg/mL in humans.¹⁴ Although the bulk of the tear protein concentration is comprised of just a few proteins (lysozyme, lactoferrin, secretory IgA, serum albumin, lipocalin and lipophilin), close to 1800 other proteins have been identified at concentrations ranging from pg/mL to mg/mL levels.¹⁵ A notable fraction of tear proteins are proteases and protease inhibitors, present in a delicate equilibrium which is affected by both local and systemic conditions.¹⁶ Proteases, also known as peptidases or proteinases, are enzymes that break down other proteins through enzymatic cleavage of peptide bonds.¹⁷ Based on the location of their preferred sites of action within the cleaved target, proteases can be further classified into serine, cysteine, threonine, aspartic, glutamic, asparagine and metallo- proteases. In this review, we sought to analyze what had been reported in the clinical literature on the role of three physiologically important families of tear proteases and their inhibitors in ocular surface disease: matrix metalloproteinases (MMPs) and tissue inhibitor of metalloproteinases (TIMPs); cathepsins and cystatins; and plasminogen activator (PA) and plasminogen activator inhibitor (PAI). These proteins are intimately involved with ocular surface physiology.^18–20 Dissociating events associated with their chronic dysregulation from the natural fluctuations in their activity that occur during physiological processes remains a central challenge. We have further highlighted those with the greatest biomarker potential for specific ocular surface diseases versus those more reflective of generalized inflammatory processes.

For this review, we utilized the MEDLINE/PUBMED database from 1957–2019 with general keywords and disease terms as follows in different combinations: tears; tear film; lacrimal fluid; tear fluid; protease; protease inhibitor; MMP; matrix metalloproteinase; tissue inhibitor of metalloproteinase; TIMP; cathepsin; cystatin; stefin; plasminogen activator; plasminogen activator inhibitor; dry eye; contact lens; wound healing; LASIK; PRK; refractive surgery; allergic eye; keratitis; conjunctivitis; keratoconus; and ocular Graft-versus-Host Disease. Our analysis revealed a shortage of review articles in our chosen focus area, although we identified several excellent reviews focused on roles of specific proteases in certain diseases, including MMP-9 in dry eye syndrome and MMPs in keratoconus.^21–23 Review articles focusing on tear biomarkers of ocular and systematic diseases were also identified; however their focus on proteases was limited.^24–27 In this review, we have not included articles utilizing in vitro systems or animal models, unless these same articles also included a clinical study component. We have also not included articles which highlight other types of tear proteases as putative biomarkers because of a relative lack of critical mass in the literature.

Biofluid samples or so-called “liquid biopsies” are a popular source in the quest for new biomarkers for both local and systemic diseases, as they can be assessed relatively non-invasively, and follow up samples may be obtained to monitor disease progression or responsiveness to treatment. Tears are less complex with regards to the number of different proteins found relative to serum or plasma. In 2014, the plasma protein database contained 10,546 proteins detected in serum/plasma, of which 3784 were reported in two or more studies.²⁸ Tears, in contrast, have 1800 identified species.¹⁵ Tear proteins are also more concentrated (6–11 mg/ml)¹⁴ relative to whole saliva (1 mg/mL).²⁹ Saliva sampling is further complicated by oral flora, which produce proteases and other proteins that may be unrelated to underlying disease.^{30, 31} Although, our focus is the relationship between tear proteases, their inhibitors and ocular surface diseases, other studies have revealed that tear composition can shed insights into systemic disease.^{26, 32–35}

Tears may be collected using different techniques, including collection with blunt capillaries held in the lateral tear meniscus or with a tear collection strip with or without topical anesthesia for basal and reflex tears, respectively. Less commonly-used methods include collection with polyester rods, or flush techniques where tears are collected after the addition of an eye drop. ^{36, 37} Collection using glass capillaries is slow and it can be difficult to obtain sufficient sample volume in subjects with low tear production. Furthermore, patients may feel uncomfortable having a pointed object close to the eye. It is also important to avoid contact with the eyelids and conjunctiva to avoid potential contamination, and reflex tearing (if basal tears are desired). Tear collection using collection strips is generally well accepted by patients, but contamination with epithelial cells may occur.³⁸ Furthermore, recovery from the strip may vary by protein and may also be affected by the elution method used.³⁹ It has long been known that the tear collection method impacts composition and the relative abundance of specific proteins.^40–42 One recent example of this was reported in a study comparing collection methods for the analysis of ocular mucins. Collection with glass capillaries, generally considered to represent basal tears, yielded the highest relative concentration of MUC16. In contrast, Schirmer’s strip tear collection without anesthesia, generally considered as reflex tears, yielded the highest MUC5AC concentration.⁴³ A review by Rentka el al. discusses the advantages and disadvantages of the most commonly-used tear collection methods based on the subsequent analysis.⁴⁴ Unfortunately, there are insufficient studies on the impact of the collection methods on the activity and abundance of the specific proteases and inhibitors discussed in the specific ocular surface conditions of interest here. As the collection method and post-collection processing of the sample may impact the tear composition and results, standardized methodologies would be desirable comparison across studies. As specific proteins are affected differently, such standards may still need to differ depending on the protein of interest.

1.2. Major proteases and protease inhibitors in tears

Historically, proteases in tears were identified based on their enzymatic activity towards specific substrates.⁴⁵ With the development of antibody-based detection methods including Western blotting and ELISA, and later the development of high throughput mass spectrometry techniques, the identification and quantification of proteases in tears is now often based on detection of protein structure and peptide sequence. The activity of secreted proteases in biofluids is tightly regulated by four principal mechanisms: 1. Gene expression; 2. Environmental factors including pH; 3. Zymogen activation (many proteases are synthesized in inactive pro-enzyme forms requiring post-translational activation); and 4. Inhibition by endogenous inhibitors.⁴⁶ Therefore, it may be important to measure the quantity and activity of the protease, and/or the quantity of potential inhibitors, if known, to understand aspects of dysregulation. As an example, in studying the cysteine protease, cathepsin S, and its endogenous inhibitor cystatin C, we have shown that the enzymatic activity of cathepsin S is highly increased in the tears of Sjögren’s Syndrome (SS) patients compared to patients with other autoimmune diseases and patients with other forms of dry eye.^{47, 48} Interestingly, this elevation in activity may be primarily due to the decreased abundance of its natural inhibitor, cystatin C, in tears of this SS patient population.⁴⁷ Findings on the enzymatic activity of cathepsin S in tears of SS patients, however, when viewed in isolation would appear to be inconsistent with findings by others that elevations in tear cathepsin S protein in SS patients have not been seen in proteomics studies.^{49, 50}

1.2.1. Matrix Metalloproteinases (MMPs) and Tissue Inhibitor of Metalloproteinases (TIMPs)

MMPs are a family of neutral zinc proteases involved in angiogenesis, inflammation, wound repair and tissue remodeling,⁵¹ that traditionally have been divided into gelatinases (MMP-2,-9), collagenases (MMP-1, -8, -13), stromelysins (MMP-3, -10– 11) and matrilysins (MMP-7, -26).^{52, 53} They are synthesized as pro-enzymes and activated by other proteases through a process called the cysteine switch.⁵⁴ These proteases include: other MMPs, for example MMP-3 can activate proMMP-1, proMMP-8, and proMMP-13;^55–57 plasmin, which activates proMMP-1, proMMP-3, proMMP-9, and proMMP-13;^{58, 59} and cathepsins, for example, Cathepsin K activates MMP-9.⁶⁰ MMPs are secreted or attached to the outer cell membrane. MMP primary activity is thus against proteins within secretory pathways, membrane proteins, or proteins in the extracellular space/body fluids.⁴⁶ MMP activity is further modulated by endogenous inhibitors called tissue inhibitor of metalloproteinases (TIMPs).^{61, 62} The most recognized role of MMPs is in ECM degradation and turnover, however MMPs have non-matrix substrates and can modulate the ECM microenvironment and stimulate cell proliferation through direct regulation of cytokines, growth factors and other proteases.^{63, 64} MMP-1, MMP-2, MMP-3 and MMP-9 all participate in repair of the corneal epithelium and stroma.^65–67 Increased tear MMP activity was first described in patients with failed corneal grafts, but has later been shown in many ocular surface conditions.⁶⁸

TIMPs are the endogenous inhibitors of MMPs. The four homologous TIMPs (TIMP1, -2, -3 and -4) can inhibit all MMPs, albeit with different efficacy.⁴⁶ The inhibition is upheld by interaction with the MMP active site. However, TIMPs can regulate ECM turnover through indirect mechanisms, including stimulation through tyrosine kinase and MAP kinase pathways and promotion of cell apoptosis.⁶⁹ TIMP binding to MMPs does not always result in inhibition, as MMP-2-TIMP-2 interaction results in activation of the enzyme. Although, MMPs have been extensively studied in tears, TIMP and MMP interactions have not been as well-investigated.⁷⁰

Another regulator of MMP-9 is neutrophil gelatinase-associated lipocalin (NGAL), also called lipocalin-2, that forms a complex with MMP-9 and preserves its activity by protecting it from degradation.⁷¹ Although, NGAL has been shown to be present in excess of MMP-9 in tears, and its levels in tears are subject to diurnal variations, it has not been extensively studied in the disease conditions discussed here.¹⁰

Levels of MMPs in tears are most commonly measured using either single or multiplex ELISA approaches. However, there are inconsistencies in the ELISA assays used as some are directed towards the pro-form, while others are directed to the active form or to both.^72–75 MMP activity is traditionally measured using gelatinase zymography for MMP-2 and MMP-9, or collagenase zymography for MMP-1, MMP-8 and MMP-13. Although zymography only provides semi-quantitative activity measurements, these assays provide additional information about the forms of MMP present.⁷⁶ For instance, for MMP-9 assayed by zymography, the pro-form can be observed at 92 kD, the active form at 85 kD and the MMP-9/NGAL complex at 135 kD.¹⁰ Reverse zymography can also be used to study the endogenous activity of TIMPs.⁷⁷

1.2.2. Cathepsins and Cystatins

Cathepsins are a diverse group of proteases which include serine cathepsins (A and G), aspartic cathepsins (D and E), and lysosomal cysteine cathepsins (B, C, F, H, K, L, O, S, V, X, W).⁷⁸ Initially considered as lysosomal proteases, cathepsins are increasingly recognized as key players in the extracellular space. The cysteine cathepsins have collagenase and elastase activity and participate in ECM remodeling.⁷⁹ These cathepsins can also serve as signaling molecules through activation of protease-activated receptors and processing of cytokines and chemokines.⁷⁹ One of the first studies of cathepsins in tears identified cathepsin C- and cathepsin B-like activity in healthy human tears.⁴⁵ However, their role in the tears and ocular surface has not been systematically investigated beyond cathepsin S.^{19, 47, 48, 80–83}

Cystatins. Cystatins have long been known as constituents of human tears and are thought to mediate the activity of cysteine proteases.⁸⁴ They are categorized into two subgroups: Type I are cytosolic while Type II are found intracellularly and extracellularly.⁸⁵ Cystatins are associated with cell proliferation, cell migration and cell differentiation.⁸⁶ Normally, cystatins of the type II category, especially cystatin C, B, S, SN, and SA, are detected in human tears.¹⁶

1.2.3. Plasminogen Activators (PAs) and Plasminogen Activator Inhibitors (PAI)

Plasminogen activators (PAs) are serine proteases that convert plasminogen into plasmin, a serine protease identified in the mammalian cornea in 1967.^{87, 88} Ocular PAs prevent fibrin deposition following inflammation,^{88, 89} stimulate leucocyte migration and activation of collagenases,^{90, 91} maintain aqueous humor fluidity;⁹² and regulate angiogenesis.⁹¹ Urokinase-type PA (uPA) and tissue-type PA (tPA) are the main types, with both present in human tears.^{87, 93} Each PA has different roles, with tPA involved in fibrinolytic processes and uPA involved in cell migration and tissue remodeling.^94–96 The lacrimal gland, conjunctival blood vessels and corneal epithelium are sources of tPA, while uPA originates from corneal and conjunctival epithelia.^{89, 97} uPA and tPA have different activity levels in tears from healthy individuals, with tPA activity higher than uPA activity.⁹³

Plasminogen activator inhibitors (PAI) are serpin-class anti-proteases that inhibit both uPA and tPA. PAI-1 mainly inhibits tPA, while PAI-2 is an effective inhibitor of uPA. PAI-1 and PAI-2 may be present in human tears when either cornea or conjunctiva are affected by disease.⁹⁸

2. Tear Proteases and Protease Inhibitors in Ocular Surface Diseases

2.1. Dry Eye Disease

Dry eye disease (DED) is a multifactorial disease resulting in discomfort, visual disturbance, and tear film instability and is associated with an increase in tear osmolarity and ocular surface inflammation.⁹⁹ Dryness is a downstream consequence of this inflammatory cycle, which is associated with increased production of many proteases in ocular surface tissues and tears.¹⁰⁰ Balance between protease and protease inhibitors plays an important role in ocular surface homeostasis in DED.¹⁰¹

DED patients can be stratified into those with aqueous tear-deficient dry eye (ADDE) or evaporative dry eye (EDE). Aqueous tear deficiency is caused by insufficient lacrimal gland tear secretion secondary to lacrimal gland destruction or dysfunction, leading to tear hyperosmolarity that can stimulate proinflammatory cascades.⁹⁹ Lacrimal gland dysfunction in ADDE leads to production of proinflammatory mediators which are delivered to the ocular surface and into tears.⁹⁹ Aqueous tear deficiency can be subdivided into 2 categories, Sjögren’s syndrome (SS)-associated and non-SS associated ADDE. SS is a systemic autoimmune disease that leads to dryness of many mucosal surfaces, including the eyes and mouth. A majority of SS patients are women while the spectrum of clinical symptoms includes ADDE, interstitial lung disease, renal failure, cryoglobulinaemic vasculitis, infections due to compromised mucosal surfaces, cardiovascular disease and development of B-cell lymphoma.^{102, 103} Non-SS ADDE is associated with obstruction of the lacrimal gland duct or with age-related DED and occurs in otherwise healthy individuals without autoimmune disease.⁹⁹

Evaporative DED is characterized by premature evaporation of the tear film in the setting of normal lacrimal gland secretion. Blepharitis is a primary cause of EDE, and can also be categorized into 2 subclasses, anterior and posterior blepharitis.¹⁰⁴ Meibomian gland dysfunction (MGD), often used as a synonym for posterior blepharitis, is associated with altered lipid composition in the meibum, and leads to reduced nonpolar lipids in tears and increased tear evaporation.¹⁰⁵ These categories are summarized in Figure 1.

Figure 1. Dry eyes disease classification.

Dry eyes diseases (DED) are categorized into evaporative dry eye (EDE) and aqueous deficient dry eye (ADDE). EDE is mainly caused by meibomian gland dysfunction (MGD)/blepharitis. ADDE is primarily based on defective secretion by the lacrimal gland (LG). ADDE is further subdivided into Sjögren’s syndrome (SS) related and non-Sjögren’s syndrome related (non-SS). Image created with BioRender.

The diagnostic criteria for both ADDE and EDE are complicated: symptoms overlap and diagnosis is based on patient symptoms obtained from questionnaires such as the Ocular Surface Disease Index (OSDI) and objective tests such as tear film break-up time (TBUT), corneal fluorescein staining and tear secretion tests.¹⁰⁶ Frequently, objective tests do not correlate with patient-reported symptoms, complicating workup of DED.¹⁰⁷ Proteases and their inhibitors have the potential for serving as biomarkers for DED because of the correlation of protease dysregulation in the tear film with DED symptoms and pathology.¹⁰⁸

2.1.1. MMPs and TIMPs

Changes in MMP levels have been extensively studied in DED. In a study of 46 newly diagnosed DED patients with OSDI levels >20 and exhibiting at least one of several additional indicators of disease, compared to 18 healthy controls, tear film MMP-9 activity was increased, with levels correlated with DED severity as defined by the DEWS criteria.¹⁰⁹ Tears from 9 SS ADDE patients also exhibited increased MMP-9 abundance and activity in conjunction with accumulation of the proinflammatory form of IL-1 (IL-1α and mature IL-1β) and decreased inactive IL-1β, relative to 10 healthy control subjects. ¹¹⁰ This finding linked elevated MMP-9 to activation of IL-1β. Levels of MMP-9, MMP-3 (an activator of MMP-9) and TIMP-1 were increased in tears of 15 ocular rosacea patients who met clinical criteria for EDE and MGD, relative to 8 healthy controls.¹¹¹ The concurrent elevation of TIMP-1 in these ocular rosacea patient tears suggested MMP-9 activation in this case might be related to MMP-3 accumulation.¹¹¹ Another study from this group in 13 ocular rosacea patients meeting clinical criteria for EDE and MGD versus 13 healthy controls showed that elevated MMP-9 activity in ocular rosacea patients tears, as measured by gelatin zymography, was correlated with high tear IL-1α, an effector of MMP-9 production.¹¹² Levels of pro-MMP-9 measured by ELISA were significantly elevated in tears of 12 blepharitis patients with diagnosis based on clinical symptoms and medical history compared to tears of 18 healthy controls, although the criteria for DED diagnosis in this study were not as well-defined as in other studies.¹¹³

The InflammaDry (Rapid Pathogen Screening, Sarasota, FL) is an immunoassay that measures expression of pro-MMP-9 and MMP-9 in human tears with a detection threshold of 40 ng/ml.²² If total MMP-9 expression is ≥ 40 ng/ml, the kit shows a positive result with 2 lines (one blue and the other one red). Schargus et al directly compared MMP-9 concentrations in tears measured by ELISA versus InflammaDry in 20 elderly patients aged ≥ 60 years and previously undiagnosed with DED.¹¹⁴ Of these patients, most met clinical criteria for mild to moderate DED based on OSDI, Schirmer’s test values, corneal staining and TBUT.¹¹⁴ Only 3 of the participants showed positive values for MMP-9 by InflammaDry (> 40 ng/mL) while the other 17 participants with negative InflammaDry test values showed mean ELISA MMP-9 value of 7.1 ng/ml.¹¹⁴ InflammaDry results in this study were correlated with tear MMP-9 levels measured by ELISA, with no false positives detected, suggesting that the InflammaDry test results were accurate in this small cohort of human tears.¹¹⁴ However, the test distinguished only DED patients with high MMP-9 levels in this study.

In a prospective multicenter clinical trial conducted in 143 DED patients meeting clinical criteria of OSDI >7, Schirmer’s test <10 mm, TBUT <10 sec and detectable keratoconjunctival staining versus 63 healthy controls, this assay showed 85% sensitivity and 94% specificity.¹¹⁵ In addition, a positive MMP-9 signal measured by the InflammaDry was correlated with DED in 47 patients meeting 3 out of 4 of the following clinical diagnostic criteria: OSDI>12 (P = 0.001), TBUT≤5 (P < 0.013), Schirmer’s test ≤10 (P < 0.001) and corneal staining >1 (P = 0.007), relative to 54 healthy controls.¹¹⁶ This study also linked meibomian gland obstruction and increased pathological features of meibomian gland secretion with a positive MMP-9 signal with the InflammaDry.¹¹⁶ While analysis of MMP-9 levels appears to have diagnostic utility in DED workup, particularly in identifying patients with significant underlying inflammation, this test is unable to distinguish between different DED states.

2.1.2. Cathepsins and Cystatins

Cathepsin S is globally involved in initiating inflammatory responses. It is an established effector of lysosomal protein degradation and ECM degradation, functions essential for cellular homeostasis and differentiation.¹¹⁷ In addition, cathepsin S mediates major histocompatibility complex (MHC) class II-antigen presentation by cleavage of invariant chain.^{118, 119}

In two studies utilizing two independent cohorts of primary and secondary SS patients, respectively, tear film cathepsin S activity was significantly higher in SS patients relative to those with non-SS DED, other autoimmune diseases without DED, or healthy controls.^{47, 48} Increased tear film cathepsin S activity may affect the tear quality in SS patients since it promotes the degradation of other tear proteins such as cystatin C, lactoferrin, and secretory IgA.⁴⁷ Although cathepsin S may discriminate between SS and non-SS ADDE, the possibility of utilizing it to distinguish between non-SS ADDE and EDE has not been evaluated.

Cystatin C is an endogenous cathepsin S inhibitor which interferes with antigen processing by regulating cathepsin S, and which also interacts with numerous inflammatory mediators to regulate inflammatory processes at the molecular level.⁸⁵ Decreased tear cystatin C, as measured by ELISA, was found in SS patient tears in one of the two studies evaluating identifying increased cathepsin S activity, which compared 33 primary and secondary SS patients diagnosed based on clinical signs and symptoms and Ro/SSA autoantigen seropositivity, to 35 patients with non-SS ADDE and EDE, 33 with rheumatoid arthritis, 31 with other autoimmune diseases, and 24 healthy controls.⁴⁷ The ratio of cathepsin S to cystatin C contributing to elevated enzymatic activity of cathepsin S may reflect the severity of ocular surface inflammation in SS, more so than actual cathepsin S protein.⁴⁷

2.2. Infection and Allergy

The tear film and conjunctiva are the eye’s physical barrier against insults such as allergens, chemicals and infectious agents. The conjunctiva is the most immunologically-active component of the ocular surface and consists of two layers: epithelium and substantia propria. The epithelium is stratified with several layers of epithelial cells, and is invested with goblet cells, Langerhans cells and lymphocytes. The substantia propria also contains multiple immune cell including mast cells, eosinophil and basophils. Upon exposure to irritants, the eyelids and tear film provide a first line of defense through the blink reflex and increased tear flow. The conjunctiva contributes anti-microbial and anti-infectious factors to the ocular surface and tears, while inflammatory activators are also released to regulate immune responses.¹²⁰ Both corneal (keratitis) and conjunctival (conjunctivitis) inflammation are caused by allergic or infectious triggers and in both cases, tear proteases are altered.

2.2.1. MMPs and TIMPs

Changes in MMPs are linked to conjunctivitis and keratitis, due to their role in remodeling of ECM and their modulation of inflammation. Vernal keratoconjunctivitis (VKC), a chronic allergic eye disease, leads to conjunctival tissue remodeling due to excessive ECM deposition, subepithelial fibrosis, chronic inflammatory cell infiltration and epithelial thickening and corneal erosion.⁷³ A small study utilized gelatin zymography to evaluate MMPs in tears of 6 individuals with active VKC, 14 with active allergic conjunctivitis and 6 healthy non-allergic volunteers. Tears of VKC patients contained both pro- and active MMP-2 and MMP-9, while healthy patient tears contained only pro-MMPs and only a subset of patients with allergic conjunctivitis showed active MMPs in tears.¹²¹ In a study of 16 patients with active VKC who were free of treatment for 3 days, increased MMP-1 and MMP-9 levels in tears were detected by ELISA in conjunction with increased TIMP-1, relative to 10 age-matched healthy control individuals. However, the increase of MMP-1 and MMP-9 exceeded TIMP-1 increases, causing elevated MMP-1/TIMP-1 and MMP-9/TIMP-1 ratios.⁷³ In another study of 24 patients with active VKC, VKC patient tears showed a higher frequency of detection of MMP-1, MMP-2, MMP-3, MMP-8, MMP-9 and MMP-10 compared to tears of 12 healthy control individuals.¹²² Further, in a study of 26 patients with atopic blepharoconjunctivitis versus 26 healthy control individuals, MMP-8 concentration was significantly elevated in tears of patients with atopic blepharoconjunctivitis.¹²³ This protein was mainly found localized in inflammatory cells within in the conjunctiva.¹²³ MMP-8 protein concentration and activity was increased in tears of 26 consecutive patients with non-allergic eosinophilic conjunctivitis compared to 26 asymptomatic controls. This elevation was also correlated with increased conjunctival neutrophils (ρ=0.66, P=0.0002).¹²⁴

Herpes stromal keratitis (HSK) is a blinding disease caused by herpes simplex or zoster virus infection. Tissue injury and vision loss results from host immune responses, rather than viral infection. The balance between MMPs and TIMPs may have a role in necrotizing HSK. A small study by Smith et al. measuring tear MMPs by zymography in 4 HSK patients and 10 healthy controls showed in the HSK patient tears that MMP-9 activity was ~4.5-fold higher.¹²⁵

A study evaluating MMP-9 activity in tears of 25 patients with fungal keratitis, 7 patients with bacterial keratitis and 3 with viral keratitis relative to 10 healthy control subjects found that 72% of fungal keratitis patients had detectable tear MMP-9 activity, 71.4% of patients with bacterial keratitis had detectable tear MMP-9 activity and 33.3% of patients with viral keratitis had detectable tear MMP-9 activity compared to tears from healthy controls.¹²⁶

2.2.2. Cathepsins and Cystatins

Analysis of reflex tears of patients with Fusarium keratitis found differential expression of cystatins. In this study, multiple samples were collected from 35 male individuals aged 20–60 years at early (within 7 days), intermediate (7–14 days) and late (after 14 days) stages of infection, relative to tears from 20 healthy individuals. In each case, 5 samples from each category were subjected to 2D-difference gel electrophoresis (2D-DIGE) and subsequent proteomics identification of species of interest. Tear cystatin SA was significantly reduced in early to late stages of infection in patients with Fusarium keratitis.¹²⁷ In another study utilizing pooled reflex tear samples from 16 Fusarium keratitis patients and 16 Aspergillus keratitis patients, versus 24 healthy control individuals, analyzed by two-dimensional electrophoresis and MALDI-TOF mass spectrometry, several cystatin precursors were decreased in tears from both Aspergillus and Fusarium keratitis patient tears.¹²⁸

2.2.3. PAs and PAIs

Changes in levels of PAs are linked to keratitis and conjunctivitis. 10 of 12 patients with superficial keratitis showed elevated uPA activity in tears relative to activity levels in 18 healthy control subjects.¹²⁹ In a study of 20 patients with active VKC compared to 19 age-matched healthy controls, tear uPA and tPA levels were significantly increased in VKC patient tears, although PAI-1 was undetectable in tears of both populations. Tear plasminogen activity in a subset of 10 VKC patients (4.7%±3.4%) was also greater than an equivalent number of tear samples selected from healthy controls (0.65%±0.83%), while immunohistochemistry evaluations showed that uPA and uPA receptors (uPAR) were elevated in conjunctival epithelium and stroma of VKC patients relative to levels in healthy controls. Overall, tear tPA levels were correlated with the severity of corneal injury (ρ=0.466, P < 0.04) in VKC.¹³⁰

2.2. Refractive Corneal Surgery

Corneal wound healing involves multiple steps such as inflammation, myofibroblast differentiation, ECM deposition and scar formation. The corneal epithelium, stroma and endothelium each may be affected and react differently to injury.^131–133 In deep corneal injuries where basement membrane is disrupted and wounding extends into the stroma, a fibrotic response can be activated which can lead to corneal scarring and opacification. Maintaining the appropriate balance of growth factors, cytokines, proteases and ECM signaling is essential to restore corneal integrity and maintain homeostasis. With increasing numbers of refractive surgeries performed, corneal wound healing has become a clinically-significant issue. Thus, we included this topic in the list of ocular surface conditions linked to protease dysregulation. Laser in situ keratomileusis (LASIK) is the most commonly performed vision correction surgery in the United States. Although generally successful, LASIK can generate complications including abnormal wound healing, flap detachment, and ectasia at a frequency of ~2%.¹³⁴ Photorefractive keratectomy (PRK) is another popular laser refractive surgery option. PRK differs from LASIK by ablating the corneal surface without creating a flap.¹³⁵ A main complication of PRK is corneal subepithelial haze, and PRK patients are more likely to suffer from corneal light scattering than LASIK patients.¹³⁶

2.3.1. MMPs and TIMPs

During corneal wound healing, many MMPs are upregulated to regulate cell migration by facilitating ECM degradation and modifying cell adhesion. Due to their intensive involvement, their utility as potential biomarkers for surgical wound healing has been proposed. A study comparing tear inflammatory mediators in 62 myopic patients undergoing treatment with orthokeratology (OK) contact lens or LASIK treatment found that of the 30 patients who had undergone LASIK, significant elevations of MMP-9 as measured by ELISA were detected in tears relative to levels in 30 healthy control subjects at a mean window of 10.8 months post-LASIK.¹³⁷ In the affected eyes of 12 patients with LASIK ectasia, significant elevations in tear MMP-9 and decreases in TIMP-1 as measured by ELISA were found, compared to tears from the affected eyes of 25 patients with uncomplicated LASIK surgery or the tears from eyes of 25 healthy controls.¹³⁸ The relationship between elevated MMP activity and LASIK complications may be complicated by individual patient predisposition to ectasia. For example, patients with keratoconus where MMP dysregulation is linked to disease pathogenesis are at higher risk for development of post-LASIK ectasia.¹³⁹

DED associated with LASIK surgery is one of its most common complications, attributable to neurotrophic effects, damage to goblet cells, altered corneal shape, and tear film instability following surgery.¹⁴⁰ The TFOS DEWSII report also identified surgical induced iatrogenic dry eye as an entity.¹⁴¹ A study evaluated multiple aspects of ocular surface integrity following bilateral LASIK surgery in a patient population that was rigorously prescreened and excluded from surgery if indications for DED and corneal abnormalities were detected. In these LASIK patients, the OSDI score was significantly increased and TBUT decreased in conjunction with a 10-fold increase in tear MMP-9 levels measured by sandwich ELISA, compared to age-matched healthy controls.¹⁴² Study participant tears were also assayed using the InflammaDry test in parallel. Despite measurement of significantly elevated MMP-9 levels by ELISA, only half of the patients showed a positive signal with InflammaDry. The InflammaDry assay may thus have a threshold of detection too high to detect clinically-significant increases in MMP-9 post-LASIK.

MMPs also appear to participate in corneal wound healing after PRK.^143–145 Tears of 24 patients and 21 healthy control subjects were collected one day post-PRK for evaluation of MMP-8, and in parallel, tears of 13 patients and 10 healthy control subjects were collected one day post-PRK for analysis of MMP-8-activating membrane type-1-MMP.¹⁴⁶ In post-PRK patients, tear fluid flow-corrected active and latent MMP-8, together with MMP-8-activating membrane type-1-MMP (MT1-MMP) levels were significantly elevated. These results suggest an increase in MMP-8 synthesis and activation in tear fluid in response to PRK. However, a study limitation was that MT1-MMP and MMP-8 were not measured simultaneously in the tear fluid of the same patients, and the role of TIMPs were not elucidated. Finally, a six-month long follow-up of tear MMP-9 and chondroitin sulfate levels (which regulate MMP expression and activity) in a single 20-year old patient post-PRK utilizing microcapillary tube collection showed that MMP-9 activity was elevated 1-day post-surgery and decreased after day 4.¹⁴⁷ At the same time, a form of chondroitin sulfate dramatically increased within a month, decreased from months 1–3 and increased again after 3 months. While the study was limited to a single patient, these temporal changes suggest complex interactions between MMPs and other factors during sustained corneal wound healing. With global significance of MMP upregulation in LASIK and PRK remaining unclear, MMP expression profiles may still serve as indicators for treatment response and identify patients at risk for complications such as ectasia.

2.3.2. Cathepsins and Cystatins

Zhou et al. used iTRAQ quantitative proteomics to assess tear protein changes after LASIK surgery. Tears were collected from the affected and contralateral eyes of 22 patients at 1 week and 3 months after the procedure. Of 824 proteins quantified, the cystatin SN precursor increased by two-fold, while cathepsin B was downregulated by 0.64-fold one week following surgery.¹⁴⁸ Further exploration of these changes, including the actual activity of the affected cathepsin B protein, under these conditions is warranted.

2.3.3. PAs and PAIs

PA activity is altered in individuals with corneal ulcers, epithelial defects, recurrent erosions and chemical burns.¹²⁹ PA activity results in generation of plasmin, an effector protease, involved in fibrinolysis, ECM degradation, cell migration, tissue remodeling, wound healing, angiogenesis and inflammation.¹⁴⁹ In a study of 16 patients with myopia treated with PRK, analysis of pre- and post-operative tears revealed a significant increase in flow-corrected plasmin excretion rate in tears (median 40.0 μIU/min range 13.3 to 222.8 μIU/min vs. median 1.6 to 41.5 μIU/min) and decrease in tear plasmin activity (median 0.65 IU/l range 0.6 to 1.49 IU/I vs. median 1.29 IU/I range 0.6 to 6.9 IU/I) in the first 2 days post-PRK. ¹⁵⁰ Both factors were restored to baseline by day 7. In a study comprising 86 patients undergoing PRK for correction of myopia at two different ablation depths, who were randomly assigned to three different treatment groups post-PRK, Brart et al. evaluated the effects of acute treatment with the plasmin inhibitor, aprontinin versus sustained treatment with 0.1% fluorometholone or no treatment for up to a year post-surgery. No beneficial effects on refractive outcomes were seen with aprontinin and a greater haze occurred in treatments correcting more than 6.00 diopters.¹⁵¹ In a study of 42 patients experiencing bilateral PRK surgery with an interval of 1–2 weeks between surgeries and monitored for uPA activity in tears collected with glass capillaries, uPA activity was decreased immediately post-surgery, increased dramatically on day 3, and returned to baseline by day 5.¹⁵² However, in patients with abnormal corneal healing and haze development, uPA activity remained low through the third postoperative day.¹⁵² Many studies have reported on similar uPA activity patterns in rabbits after PRK, and explored the relationship between tear uPA and corneal haze development in rabbit models.^{153, 154}

PAI levels may also be affected by PRK and LASIK. Levels of PAI-1 and PAI-2 in tears were assessed before and after surgeries in 46 eyes of 38 patients receiving PRK (8 patients with bilateral PRK) and 13 eyes of 8 patients with LASIK (5 patients with bilateral LASIK).¹⁵⁵ Only PAI-2 was upregulated in patient tears immediately post-PRK and LASIK, while returning to basal levels within a day. However, no correlation was observed between corneal haze and increased PAI-2 levels in this study.

Proteases including MMPs and PAs, as well as protease inhibitors such as cystatins are globally associated with wound healing and tissue remodeling processes. Abnormalities in tears are reported post-LASIK, with both TBUT and basal tear secretion decreasing after surgery. Surgical injuries to the epithelium increase tear cytokines and tear hyperosmolarity, which in turn trigger stress-activated kinases, other proinflammatory cytokines, MMPs and ultimately induce ocular surface inflammation.¹⁵⁶ DED post-LASIK may drive some elevations in tear MMP-9. However, another study found similar tear MMP-9 levels in patients with LASIK-induced DED and asymptomatic patients post-LASIK,¹⁴² suggesting that some elevated tear MMP-9 may also be due to prolonged corneal wound healing associated with LASIK.

2.4. Contact Lenses

Contact lenses (CL) contribute greatly to risk of ocular infection, DED and other ocular complications.^157–159 In a large study of 393 CL wearers, 213 glasses wearers and 287 individuals without refractive correction, DED incidence as determined by a self-reported questionnaire was significantly higher in the CL group than in glasses wearers and those without refractive correction.¹⁶⁰ A cross-sectional study of 730 subjects utilizing a self-reported questionnaire identified the most common reasons for discontinuing CL wear as CL discomfort and dryness.¹⁶¹ Symptoms of lens-related dryness result from increased tear film thinning by evaporation or tear film dispersion, resulting in increased tear film osmolarity.¹⁶² Increased osmolarity associated with CL affects the expression of inflammatory mediators in the ocular surface.¹⁶³ In addition, microbial keratitis, frequently associated with CL wear, can alter tear proteins.¹⁵⁸ A large clinical multi-site study of 500 full-time hydrogel CL wearers showed that 55% of CL wearers displayed corneal staining while 8% of this group had severe ocular surface damage.¹⁶⁴ Corneal damage from CL can also alter the tear proteome. Some of the changes in tear protease and protease inhibitor composition in CL wearers are discussed below, although the significant incidence of DED and infection in CL wearers means that many CL wearers conform to patterns already described.

2.4.1. MMPs and TIMPs

As noted already, MMP-9 elevations are linked to DED, infection and other ocular surface conditions. In a study of 20 myopic patients wearing rigid gas-permeable CL and with good CL tolerance versus 20 myopic non-CL wearers, increased tear MMP-9 as measured by ELISA was detected in capillary tears in CL wearers relative to those without CL wear.¹⁶⁵ Orthokeratology (OK) uses reverse geometry CL to temporarily correct corneal curvature in myopia by having the patient wear the CL overnight.¹⁶⁶ In 32 patients utilizing OK for 12 months, higher tear MMP-9 as measured by ELISA was seen compared to equal numbers of healthy age- and sex-matched individuals without CL wear.¹⁶⁷ The increased MMP-9 was associated with the degree of myopia to be corrected and also the presence of corneal fluorescein staining, suggesting that it may participate in the tissue redistribution and epithelial cell migration induced by OK.¹⁶⁷ The mechanisms associated with corneal erosion in CL wear and the possible relationship(s) to changes in MMP-9 levels or its activity are still unclear.^{159, 168, 169} Interpretation of data is complicated by diurnal variations in MMP-9 levels¹⁰ and inclusion of study cohorts with both extended wear versus daily wear CL.¹⁷⁰ However, increased MMP-9/TIMP-1 ratios may predispose CL-wearing individuals to corneal erosions.¹⁷⁰

2.4.2. Cathepsins and Cystatins

Several studies have shown cystatin abnormalities in tears of CL wearers. In a study of 13 soft CL wearers, 13 rigid CL wearers and 13 individuals without CL, cystatin SN analyzed from tears collected on Schirmer’s strips by ProteinChip SELDI-TOF (surface-enhanced laser desorption and ionization in time of flight mass spectrometry) was upregulated in rigid gas permeable CL wearers while, it was downregulated in soft CL wearers after 1 year of use relative to individuals without CL.¹⁶² In a study of 8 rigid gas permeable CL wearers, 9 soft CL wearers and 11 individuals without CL, Cystatin D, measured from tears collected with Schirmer’s strips, was decreased in both rigid gas permeable and soft CL wearers with 1 year of use compared to individuals without CL.¹⁷¹ It also remained decreased even after subjects stopped using CL for 4–5 days. In this study, subjects wearing rigid gas permeable CL had higher tear cystatin C than soft CL wearers.¹⁷¹ Collectively, these data suggest that soft CL wearers may have decreased cystatin inhibitor capacity in tears relative to those wearing rigid CL.

2.5. Other Ocular Surface Diseases

Keratoconus is a corneal ectasia associated with progressive thinning of the corneal stroma linked to degradation of corneal collagen.¹⁷² Keratoconus leads to irregular astigmatism due to loss of corneal shape from increased curvature of the cornea, resulting in loss of visual acuity.¹⁷² The cause of keratoconus is largely unknown, but it is of multifactorial etiology.^173–179 Dysregulation of MMPs is of great interest in keratoconus pathophysiology. In a study of 34 subjects with diseases of the anterior segment (7 patients with keratoconus with atopy, 7 keratoconus patients without atopy, 7 patients with non-herpetic keratitis, 4 patients with herpetic keratitis, 4 patients with SS ADDE and 5 patients with non-SS ADDE) and 11 healthy control subjects, Smith et al¹²⁵ showed that detectable activity levels of MMP-9 and an unidentified MMP of MW 116K measured by zymography were present in the tear film of all study groups. Furthermore, MMP-9 activity levels were elevated in all patient groups relative to healthy controls except keratoconus patients without atopy. Another study reported increased gelatinase (19 patients) and collagenase (15 patients) in tears of untreated keratoconus patients compared to healthy controls (17 and 16 healthy controls, respectively). However, the gelatinase (13 patients) and collagenase activity (11 patients) in patient receiving collagen cross-linking treatments were not significantly different from either untreated keratoconus patients or controls.¹⁸⁰ Recent studies have focused on measurement of levels of pro-MMP and active-MMP and/or total MMP protein levels in tears, rather than actual protease activity. Using a proteomics approach in a study population of 20 healthy controls who were CL wearers, 18 subjects with keratoconus wearing CLs, and six subjects with keratoconus (non-lens wearers), Pannebaker et al. found increased levels of MMP-1 in keratoconus patient tears compared to healthy controls, with the increase independent of CL wearing.¹⁸¹ Similarly, in the same study discussed above reporting collagenase and gelatinase activity in keratoconus tears, Balasubramanian found that tear levels of MMP-1 were increased in keratoconus patients using an antibody array approach¹⁸⁰, while these levels were reduced with collagen cross-linking treatment.

Studies have shown mixed results regarding whether other MMPs such as MMP-3 and MMP-7 are affected in tears of keratoconus patients.^180–182 MMP-9 changes are the best characterized in keratoconus, with elevated levels confirmed by several independent studies using activity and antibody based approaches.^{125, 183–185} However, neither the Pannebaker nor Balasubramian studies referenced above detected differences in MMP-9 tear levels between keratoconus patients and healthy controls.^{180, 181} Several reviews have been published specifically on MMPs in keratoconus.^{23, 186} Beyond MMPs, in a study of 36 keratoconus patients and 18 healthy controls, Balasubramanian et al showed a threefold increase in the abundance of tear cathepsin B and a two-fold decrease in cystatin S and cystatin SN using proteomics analysis of capillary tears of keratoconus patients compared to healthy controls.¹⁸⁷ Additional reviews in the keratoconus area reflect interest in other proteases/protease inhibitors that are potentially dysregulated.^{23, 186, 188, 189} Clinically, the diagnostic utility of protease dysregulation in keratoconus may have greatest value by identifying patients at risk of post-LASIK ectasia prior to their development of the characteristic thinning and curvature of the cornea.

Graft-versus-host disease (GVHD) is a systemic complication that can be a major cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (allo-HSCT).¹⁹⁰ GVHD can affect the skin, gastrointestinal system, liver, lung, oral mucosa, and eyes.¹⁹¹ 40% - 60% of allo-HSCT recipients and 90% of patient with chronic GVHD develop ocular GVHD.¹⁹⁰ Donor-derived CD4⁺ and CD8⁺ T cells are primary effectors of ocular GVHD pathogenesis.¹⁹² DED symptoms are hallmarks of ocular GVHD, which is associated with inflammatory damage and ECM degradation of the tissues of the ocular surface.¹⁹² Therefore, the proteases and protease inhibitors which are found in DED patient tears are also altered in ocular GVHD patient tears. A study of 20 allo-HSCT patients identified 8 with ocular GVHD based on Schirmer’s test values ≤ 5 mm and use of treatment for keratoconjunctivitis sicca. In tears of these patients, MMP-9 levels measured by ELISA were increased compared to levels in non-ocular GVHD patients and healthy controls.¹⁹³ Another study from Arafat. et al found that MMP-9 and MMP-8 levels were increased and TIMP-1 decreased, as measured by multiplex ELISA, in tears of 14 ocular GVHD (defined based on OSDI >23, Schirmer’s test ≤10 mm, and corneal fluorescein staining ≥4) compared to tears of healthy controls.¹⁹⁴ A study comparing tears of 10 patients with ocular GVHD (defined based on the severity of their ocular impairment, threat to visual acuity, need for treatment, and undefined parameters measuring corneal staining, Schirmer’s test, OSDI) versus GVHD patients without ocular symptoms found significant decreases in cystatin S and cystatin SN in tears of ocular GVHD patients.¹⁹⁵

3. Discussion

Tear biomarkers have potential use in diagnostics, prognostics, pharmacodynamic response, patient stratification and risk evaluation. Proteases and protease inhibitors are attractive candidates as potential tear biomarkers, due to their abundance in tears, sensitivity for detection in assays and relationship to disease pathophysiology. The aqueous component of the tear film is rich in proteases and protease inhibitors. These proteins are produced actively by tissues contributing to aqueous tears to maintain ocular surface homeostasis, and in response to challenges like dryness, infection and wounding. Proteases in tears have potent actions on the cornea including activation of signaling pathways, cleavage of proinflammatory cytokines and other tear proteins, and remodeling of the underlying stroma. These actions occur physiologically, but when they are prolonged or dysregulated, they may reflect disease pathogenesis. In turn, the proteases produced in corneal and conjunctival tissue in response to wounding and inflammation can be detected in tears as indications of homeostatic and/or pathogenic processes on the ocular surface. The relationship between tear composition and ocular surface integrity is interdependent. Tear protease dysfunction can disrupt ocular surface integrity, expose and sensitize pain-sensing nociceptors, and induce immune responses through T-cell recruitment to the ocular surface. Conversely, proteins upregulated in the cornea in ocular surface disease may directly contribute to alterations in tear composition. Figure 2A and B demonstrate the intricate interrelationships between tear composition and corneal integrity by illustrating the changes associated with two ocular surface conditions, DED and corneal wound healing, respectively.

Figure 2. Proteases and protease inhibitors in the tear film and the cycle of homeostasis/pathogenesis.

Proteases and protease inhibitors maintain ocular surface and tear film homeostasis. A. Dry Eye Disease (DED). Various conditions induce elevated tear osmolarity, production of proinflammatory cytokines and proteases in tears. High tear osmolarity and cathepsin S may increase the expression of proinflammatory cytokines and MMP-9 in corneal epithelium. Elevated tear cytokines may further upregulate proinflammatory cytokines, cathepsin S and MMP-9 which (1) cleave epithelial junctional complexes, accelerate apical corneal epithelial detachment, expose subapical epithelium and; (2) activate sensing nociceptors. Inflammatory cytokines and MMP-9 facilitate the activation of dendritic cells and their migration to the lymph nodes (afferent arm). In lymph nodes, dendritic cells present antigen to CD4⁺ T cells, which can be recruited to the ocular surface (efferent arm). Cytokines produced by CD4⁺ T cells feedback to affect epithelial cell survival and differentiation. B. Corneal Wound Healing. Corneal conditions such as wounding through refractive surgery give rise to alterations in tear composition. MMP-2, MMP-9 and uPA are upregulated in corneal epithelium and contribute to epithelial migration by (1) cleaving epithelial junctional complexes and (2) degradation of ECM. Cytokines in the epithelium also facilitate wound healing by inducing apoptosis in epithelial cells. IL-1α and TGF-β secreted by the epithelium activate fibroblasts (keratocytes) and trans-differentiation of myofibroblasts. (3) Myofibroblasts secret ECM and MMPs, including MMP-1 and MMP-3, which further facilitate stromal healing and ECM remodeling. (4) Plasmin and PAs modulate normal wound healing processes through ECM remodeling, fibrin degradation and pro-MMP activation. Corneal wounding is accompanied by inflammation, upregulation of proinflammatory cytokines and immune cell recruitment leading to increased tear uPA, MMP-8 and MMP-9. Image created with BioRender.

Table 1 summarizes the common changes in the families of proteases and protease inhibitors of focus in this review in tears across multiple ocular surface diseases, refractive surgery or CL use. As highlighted, a general trend in upregulated tear MMPs is observed across conditions sharing an inflammatory component. MMP-9, in particular, is elevated in DED, infectious conditions, allergic eyes, after refractive surgery and in CL users. Pflugfelder’s group also reported increased tear MMP-9 in patients with dysfunctional tear syndrome.¹⁰⁹ Therefore, MMP-9 abundance and/or activity alone is not a stand-alone diagnostic tool due to its lack of specificity in ocular surface diseases. However, findings of elevated MMP activities or their prolonged expression in tears may serve as a general tool to indicate delayed or complicated wound healing, identify unresolved infectious processes, reinforce a DED diagnosis, or identify at-risk patients prior to LASIK.

Table 1:

Changes in tear proteases and protease inhibitors associated with ocular surface diseases and conditions.

Diseases/Conditions	Proteases/ Protease Inhibitors
	Common		Possibly Distinct
	Protein	Activity	Protein	Activity
Aqueous tear-Deficient Dry Eye Disease	MMP-9115,116↗	MMP-9109↗
Sjögren’s Syndrome Aqueous tear-Deficient Dry Eye Disease	MMP-9110 ↗	MMP-9110↗	Cystatin C47↘	Cathepsin S47,48↗
Evaporative Dry Eye Disease	MMP-9111,113,116↗ MMP-3111↗ TIMP-1111 ↗	MMP-9112↗
Herpes stromal keratitis		MMP-9125↗
Bacterial Keratitis		MMP-9126↗
Fungal Keratitis	Cystatin S 128↘ Cystatin SN128 ↘ Cystatin SA127↘	MMP-9126 ↗
Vernal keratoconjunctivitis	MMP-173 ↗ MMP-973,122↗ TIMP-173↗	MMP-2121↗ MMP-9121↗	uPA130 ↗ tPA130 ↗	Plasminogen130↗
Atopic Blepharoconjunctivitis	MMP-8123↗	MMP-8123↗
Non-allergic Eosinophilic Conjunctivitis	MMP-8124↗	MMP-8124↗
Laser In Situ Keratomileusis Surgery	MMP-9137↗ PAI-2158↗↘ Cystatin SN148↗ In LASIK ectasia patients MMP-9138↗ TIMP-1138↘ In LASIK induced iatrogenic dry eye patients MMP-9142↗		Cathepsin B148↘
Photorefractive Keratectomy Surgery	MMP-8146↗ PAI-2155 ↗↘	MMP-9147↗↘	MT1-MMP146↗	uPA152↘→ (Corneal haze)
Contact lens	MMP-9165,167↗ Cystatin SN162↗ (Rigid CL) Cystatin SN162↘ (Soft CL) Cystatin D171↘ Cystatin C171↗ (Rigid > Soft CL)
Keratoconus	MMP-1181↗ MMP-9183–185↗ Cystatin S187↘ Cystatin SN187↘	MMP-9125↗	Cathepsin B187↗
Graft-versus-host disease	MMP-8194↗ MMP-9193,194↗ TIMP-1194↘ Cystatin S195↘ Cystatin SN195↘

Factors which may change more specifically in particular disease states may have greater utility as potential diagnostic biomarkers and to guide clinicians toward a particular treatment course. Proteases and protease inhibitors that may have more specificity in their association with individual ocular surface diseases or conditions are also highlighted in Table 1. The particular combination of CTSS and its endogenous inhibitor, cystatin C, shows the most distinctive change in protease and inhibitor composition seen thus far in the study of DED. However, more work is needed to evaluate whether this marker can be specific in distinguishing SS patients from those with EDE and non-SS ADDE. Likewise, uPA and tPA appear to be consistently and significantly elevated in VKC, but there is a lack of information on the corresponding PAIs that may be important in modulating this activity, and on PA/PAI ratios. Decreased Cathepsin B protein showed specificity in LASIK patients in correlating with rapidity of recovery from the procedure in a single study which utilized proteomics analysis, but without any corresponding evaluation of enzyme activity levels which would be a critical next step. Decreased uPA activity has repeatedly been associated with corneal haze, but better rigor in the association of its activity with PAI inhibitor levels may be needed to fully understand the ramifications and potential usefulness of this change. It is finally important to note that changes in MMPs have been explored in tears for several decades, and the apparent initial specificity seen with some of the “unique” protease/protease inhibitors highlighted may not be upheld if they are more extensively explored in tears of patients with different ocular disease states. Still, like MMPs, they may be further developed as powerful tools to investigate the extent of inflammation, identify delayed wound healing, and/or aid in predicting prognosis or guiding treatment.

We have mentioned throughout the review the general challenges that must also be addressed in advancing any of these protease/protease inhibitor biomarkers to clinical use. To the individual limitations discussed above, we must add adoption of a uniform tear collection modality for a particular protease of interest, and the concurrent assay of protease activity and abundance and/or its protease and protease inhibitor abundance. As well, in many disease areas, more rigorous patient inclusion and exclusion criteria will improve comparison of results across studies. Study inclusion criteria vary the most in the DED literature in studies evaluating ocular surface conditions including a component of DED (ocular GVHD, DED associated with CL wear, DED associated with LASIK) and are less variable for conditions such as keratoconus and ocular infection.

Both general and specific protease/protease inhibitor combinations are potentially powerful and easily accessible pharmacodynamic indicators of response to therapies. These can thus be of significant value in improving treatments for ocular surface diseases. Some studies have followed MMP-9 in human tears to monitor ocular surface disease in response to treatment, as a general inflammatory marker. MMP-9 in tears of 16 MGD patients with posterior blepharitis tears was reduced after 1% of topical azithromycin for 4 weeks, retaining a 70% reduction compared to pretreatment even after the drug was stopped for 4 weeks.¹⁹⁶ MMP-9 in DED patients tear was also reduced with 0.05% cyclosporine treatment.¹⁹⁷ Also, MMP-9 levels in tears were used to evaluate the efficacy of succinylated collagen bandage lenses on several corneal healing conditions including corneal ulcers, recurrent corneal erosions, DED, and corneal lesions in a limited patient cohort. ¹⁹⁸ After succinylated collagen bandage lenses treatment, patients in all groups showed significant reduction of pain, redness, irritation, foreign body sensation, and increased watering. Tear MMP-2 and MMP-9 were reduced after 7 days of treatment and further decreased to minimal levels after 21 days of treatment.¹⁹⁸ Finally, a large study of 94 keratoconus patients found that tear levels of MMP-9 were decreased in parallel with improved corneal curvature after 6 months of treatment with 0.05% cyclosporine.¹⁸⁵ These results collectively illustrate the potential for using tear MMP-9 as a general pharmacodynamic indicator of ocular surface inflammation. This utility may be similarly employed for other protease/protease inhibitor combinations in Table 1.

In conclusion, some changes in protease and protease inhibitor composition, such as MMP-9 and its effectors, already provide important information on ocular surface recovery from injury and/or its state of inflammation. Other changes in proteases and inhibitors such as cathepsin S/cystatin C or uPA/PAI may eventually represent more specific biomarkers which can improve patient stratification into disease subgroups, as in DED, or reflect a higher risk of adverse outcomes following PRK, respectively. Continued research in this area in expanded patient cohorts is critical in bringing forward the best candidates for clinical use, as well as developing accessible clinical tests for rapid measurement of protease/protease inhibitor levels and/or protease activity for biomarkers of interest.

Abbreviations

ADDE

Aqueous tear-deficient Dry Eye

APC

Antigen Presenting Cell

CL

Contact Lens

DED

Dry eye disease

ECM

Extracellular Matrix

EDE

Evaporative Dry Eye

GVHD

Graft versus host disease

HSK

Herpes Stromal Keratitis

LASIK

Laser in Situ Keratomileusis

MGD

Meibomian gland disease

MMP

Matrix Metalloproteinase

NGAL

Neutrophil Gelatinase-Associated Lipocalin

Non-SS

Non Sjögren’s Syndrome

OK

Orthokeratology

OSDI

Ocular Surface Disease Index

PA

Plasminogen activator

PAI

Plasminogen Activator Inhibitor

PRK

Photorefractive keratectomy

PUK

Peripheral ulcerative keratitis

SCBL

Succinylated collagen bandage lenses

SS

Sjögren’s Syndrome

TIMP

Tissue Inhibitor of Metalloproteinases

TBUT

Tear Break-up Time

tPA

tissue-type PA

uPA

Urokinase-type PA

uPAR

Urokinase-type PA Receptor

VKC

Vernal keratoconjunctivitis