The Potential of Tear Proteomics for Diagnosis and Management of Orbital Inflammatory Disorders including Graves’ ophthalmopathy

Abstract

Background: Orbital compartments harbor a variety of tissues that can be independently targeted in a plethora of disorders resulting in sight-threatening risks. Orbital inflammatory disorders (OID) including Graves’ ophthalmopathy, sarcoidosis, IgG4 disease, granulomatosis with polyangiitis, and nonspecific orbital inflammation constitute an important cause of pain, diplopia and vision loss. Physical examination, laboratory tests, imaging, and even biopsy are not always adequate to classify orbital inflammation which is frequently deemed “nonspecific”. Tear sampling and testing provide a potential “window” to the orbital disease process through a non-invasive technique that allows longitudinal sampling as the disease evolves.

Using PubMed/Medline, we identified potentially relevant articles on tear proteomics published in the English language between 1988 and 2021. Of 303 citations obtained, 225 contained empirical data on tear proteins, including 33 publications on inflammatory conditions, 15 in glaucoma, 15 in thyroid eye disease, 1 in sarcoidosis (75) and 2 in uveitis (77,78). Review articles were used to identify an additional 56 relevant articles through citation search. In this review, we provide a short introduction to the potential use of tears as a diagnostic fluid and tool to investigate the mechanism of ocular diseases. A general review of previous tear proteomics studies is also provided, with a focus on Graves' ophthalmopathy (GO), and a discussion of unmet needs in the diagnosis and treatment of orbital inflammatory disease (OID). The review concludes by pointing out current limitations of mass spectrometric analysis of tear proteins and summarizes future needs in the field.

1. Introduction

Diseases of the orbit and periorbital tissues manifest in a wide variety of clinical presentations. Space occupying lesions in the orbit include infections, inflammations, vascular malformations, and malignancies. Despite advances in imaging, physical examination and laboratory tests, a biopsy is often needed for diagnosis and to guide treatment. Unfortunately, the biopsy is too often read as non-specific or idiopathic inflammation, a term that gives minimal guidance to the patient or to the clinician.

Biological fluids like serum, saliva, cerebrospinal fluid, or tears can contain molecular signatures such as proteins, mRNA, lipids, or metabolites that are specific to disease and can serve as biomarkers. The term “biomarker” refers to an indicator that objectively measures and evaluates biological or pathogenic processes. A biomarker may be any type of specific molecular signature such as a protein, gene, mRNA, lipid, or a metabolite. They are detected using large-scale profiling technologies such as proteomics, genomics, transcriptomics, lipidomics, glycomics, or metabolomics. Biomarkers can aid diagnosis, predict prognosis, and monitor therapeutic responses to guide treatments.

Proteomics is the large-scale identification and expression profiling of proteins in biological systems. Proteomics relies heavily on advances in mass spectrometry instrumentation. Instrument advances in sensitivity, specificity, accuracy, and dynamic range are fast paced. Bench protocols, labeling reagents, liquid and gas phase separations upstream of the mass spectrometers are also improving at rapid rates. Proteomics experiments that would have been unfeasible a few years ago, might succeed today. Since mass spectrometry-based biomarker discovery techniques are often not high throughput, validation of candidates often involves targeted analytical techniques capable of measuring a smaller number of putative biomarkers, but in larger numbers of subjects. The goal ultimately is to have a more traditional clinical diagnostic or prognostic test for the biomarker that uses biological fluid obtained by a minimally invasive technique. These biomarkers may exhibit higher concentrations in more proximal fluids and, thus, are easier to detect than in other fluids like serum. Collection of proximal fluid from most tissues is invasive and less practical for general clinical use. However, the eye has a unique non-invasive proximal fluid, namely, tears that may be a source of biomarkers for orbital and other ocular diseases. Tears have a high concentration of proteins and proteomics is a logical choice for biomarker exploration in tears.

A short review of current mass spectrometric technologies for studying the tear proteome is additionally provided, as well as a discussion on the limitations of these methods and a section discussing the future directions of tear proteome research.

2. Tears as a diagnostic fluid

Tear fluid is a thin layer of extracellular fluid covering the corneal and conjunctival epithelia and provides a clear and protective surface. (1) Normal tear fluid volume is between 5 μL and 10 μL (2), the secretion rate is about 1.2 μL/min (3), and the tear film thickness is between 3 μm and 40 μm. (4) Tears are a complex mixture of proteins, lipids, mucins, water, and salts. The tear film can be divided into three layers, including an inner mucin layer, a middle aqueous layer, and an outer lipid layer. (5) In the aqueous middle layer, there are many proteins and the composition of proteins differs between the open and closed eye and between reflex and basal tears. (4)

Tears offer a biofluid which can be obtained non-invasively and which reflect ocular and systemic pathologies. For example, tears have specific alterations in macular degeneration (6), glaucoma (7), anterior uveitis (8), Sjogren’s syndrome (9), and graft versus host disease. (10) Tear proteins are also affected in systemic diseases which include Alzheimer’s disease (11), scleroderma, diabetes mellitus (12), multiple sclerosis (37), cancer (13), Parkinson’s disease (14) systemic sclerosis (38), sarcoidosis (75) and cystic fibrosis (39) among others. One key question is whether these tear biomarkers are specific to certain diseases.

However, biomarkers in tears could still be potentially used in diagnosis, in predicting prognosis and response to therapy, and clarifying pathogenesis. (15-18) Tear biomarkers may also enhance empirical results in clinical studies by linking changes in molecular and cellular pathways to clinical responses. (19) New drugs to better treat ocular diseases may be developed in response to measured changes in tear composition.

Modern proteomic analysis is capable of identifying numerous proteins in tears, and also demonstrating that the tear fluid proteome is closely related to ocular health. A comprehensive study has identified over 1500 tear proteins via proteomics (20), suggesting that, unlike the plasma proteome, the tear proteome can be effectively analyzed without prior depletion of major proteins. This has led to an increased interest in determining novel tear biomarkers for the identification of systemic diseases. (21) These major proteins are most likely secreted by the lacrimal glands, meibomian glands, and conjunctival goblet cells. The results from some of the proteomic studies summarized in this paper were used to compile an updated resource of 9782 nonredundant proteins in the human eye. (22)

The major tear proteins can be grouped into two categories: proteins produced by glandular secretion, and serum proteins that leak from the conjunctival capillaries. The major tear proteins lysozyme (LYZ), lactotransferrin, secretory immunoglobulin A (IgA), lipocalin, albumin, and lipophilin constitute about 80-90 % of the total amount of tear proteins.(22) This updated catalogue sheds light on the molecular makeup of previously undescribed proteomes within the human eye, including optic nerve, sclera, iris, and ciliary body, while adding additional proteins to previously characterized proteomes such as aqueous humor, lens, vitreous, retina, and retinal pigment epithelium/choroid.

3. Review of tear proteomics literature

We performed a PubMed/Medline search to identify all potentially relevant articles on tear proteomics published in the English language between Aug 1988 and Jan 2021. The study included all comprehensive descriptions and reviews of tear proteomics in relation to eye disorders and checked for duplication based on overlapping authorship, study description, number of participants, and participant characteristics. We combined multiple database-specific subject headings (such as Mesh terms). (Figure 1)

Figure 1:

Flow chart representing the Medline/PubMed search on tear proteomics in eye disorders.

Of 303 citations obtained, 33 articles on the role of tear fluid biomarkers for diagnosis of systemic diseases were found, including thyroid disease (Table 1) (23-35), cancer, diabetic retinopathy, multiple sclerosis, Parkinson disease (14), Alzheimer’s disease, systemic sclerosis and cystic fibrosis (Table 2/S) (36-39). A detailed description of tear-based studies of thyroid disease listed in Table 1 are discussed here. An additional 225 citations contained empirical data on tear proteomics in eye disorders including 33 inflammatory conditions plus GVHD (Graft versus host disease), 15 in glaucoma, 7 in ocular allergy including KC (Keratoconjunctivitis) and PUK (Peripheral Ulcerative Keratitis), as well as single publications on keratopathy, trachoma and aniridia (Table 3/S). Ocular surface disorder related articles comprised most of the citations, with 138 in dry eye disorders (Table 4/S). We excluded citations from non-English literature as well as letters and editorials. Review articles were used to identify 56 additional articles through an examination of their reference lists. (S= Supplement)

Table 1.

Representative thyroid eye disease studies based on tear proteomics

Author /References		Publication Year	Technique used in Analysis	Size of study Case/Control	-- Upregulated, -- Downregulated
Khalil et al.	(23)	1988	liquid chromatography	50/20	Tear protein profiling
	(24)	1989	(ELISA) and HPLC	69/28	Secretory IgA and lysozyme (Ratio)
Baker et al.	(25)	2006	Electrophoresis/MS	85/70	Zinc-alpha2-glycoprotein and lactoferrin
Ujhelyi et al.	(27)	2012	Multiplex bead array	72/24	IL-lbeta, IL-6, IL-13, IL-17A, IL-18, TNF-alpha, RANTES
Cai et al.	(28)	2013	ELISA assay	20/20	IL-7
Matheis et al.	(29)	2012	SELDI-TOF-MS/MALDI-TOF-MS/MS	45/15	Proline-rich protein 4 (PRP4), ss2-microglobulin lysozyme C, cystatin S
Huang et al.	(30)	2014	Multiplex bead analysis	24/16	IL-1beta and IL-17A, IL-6
Jiang L et al.	(31)	2015	SDS-PAGE/MS	25/25	lysozyme C and lactoferrin
Aass C et al.	(32)	2017	ELISA assays	21/21	LYZ, LACRT and AZGP1
	(33)	2016	2D LC-MS/MS	21/21	lysozyme C, lacritin, antileukoproteinase and zinc-
	(20)	2015	2D LC–MS/MS	21/21	alpha-2-glycoprotein1, periplakin, cystatin D (Cyst D), mammaglobin A and prelamin A/C
Kishazi E et al.	(34)	2018	TMT six plex™ and western blot	28/25	Alpha-1-antichymotrypsin, Cystatin-C, NAD(P)H dehydrogenase, Haptoglobin, Thioredoxin protein 5 Phospholipase A2, Signal transducer and activator of transcription 1-alpha, Protein ABHD14B, Retinal dehydrogenase 1, Alcohol dehydrogenase class-3
Song, R. H et al.	(35)	2020	High-throughput protein microarray technology	7/7	CD40, CD40 Ligand, GITR, IL-12p70, IL-1 beta, IL-2, IL-21, IL-6, MIP-3 alpha and TRANCE GM-CSF, IL-1 sRI and IL-13 CD40 and CD40 ligand (CD40L)

Abbreviations:

ELISA enzyme-linked immuno-sorbent assay, HPLC high pressure liquid chromatography, MS mass spectrometry, SELDI-TOF-MS surface-enhanced laser desorption ionization-time of flight-mass spectrometry, MALDI-TOF-MS/MS matrix-assisted laser desorption ionization-time of flight-tandem mass spectrometry, SDS-PAGE sodium dodecyl sulfate-polyacrylamide gel electrophoresis, 2D LC-MS/MS two-dimensional liquid chromatography-tandem mass spectrometry, TMT tandem mass tag, ESI-Q-TOF MS/MS electrosprav ionization-quadmpole-time of flight tandem mass spectrometry .

Description of Table 1: We summarize previous work studying tear proteome alterations in Graves' Ophthalmopathy (GO).

Early studies by Khalil et al. (23) demonstrated, using high performance liquid chromatography that a normal tear profile consists of five peaks and these may be augmented in GO. They subsequently found that secretory IgA, produced by plasma cells in the lacrimal gland, was elevated in patients with GO, suggesting that the lacrimal gland may be involved in the pathogenesis of GO. (24) Mathesis et al. (29) using gel electrophoresis followed by surface enhanced laser desorption ionization time-of-flight mass spectrometry (SELDI-TOF-MS) technology demonstrated that human tear samples contained up to 1000 peaks. They also identified six peptides that were differentially expressed in tears from GO versus healthy controls. An eye protective peptide, PRP, was negatively correlated with a clinical activity score. In contrast, Mathesis et al found lysozyme C, a possible inflammatory protein, was upregulated in GO. Baker et al. (25) used 1D-gel electrophoresis and mass spectrometry to identify an increase in the levels of zinc-alpha 2-glycoproteins (AZGP1) and lactoferrin in tears of patients with GO and healthy smokers suggesting a possible mechanism of smoking induced exacerbation of GO.

Okrojek et al. (26) demonstrated that a combination of chromatography and mass spectrometry (SELDI-TOF-MS) using a protein chip array, could be used to efficiently create tear protein profiles in patients with GO that suggested possible GO tear biomarkers. Aberrant cytokine expression plays a role in the pathogenesis of GO. Ujhely et al. (27) investigated tear cytokine profiles using a “multiplex bead array”. They found that compared to tears from healthy patients, tears from patients with GO had a significant increase in interleukin (IL-1beta, IL-6, IL-13, IL-17A, IL-18), TNF-α, and RANTES (regulated upon activation, normal T-cell expressed, and secreted). Tears from patients with Graves’ disease were not significantly different from either of the two prior groups. However, these patients had cytokine values varying between the healthy control and GO groups, suggesting that there is a molecular spectrum with Graves’ disease on one end and GO on the other. In another tear cytokine study, Cai et al. (28) identified differential expression of IL-7 in tears which correlated with orbital tissue immunohistochemistry from patients with inactive GO and healthy controls.

Although they did not have any active GO orbital biopsies, this study suggested disease activity of GO may be predicted by a tear biomarker. The highest levels of IL-7 in tears were in patients with inactive GO, followed by healthy controls, and lastly active GO. This was reconfirmed by Huang et al. (30) They further identified possible additional biomarkers of GO disease activity: IL-1β, IL-6, and IL-17A. Their levels increased with increasing clinical activity score. Song et al. (35) compared tear fluid from patients with active GO to healthy controls via high-throughput protein microarray technology, reaffirming the relative overexpression in GO patients of IL-1β and IL-6, but also CD40, CD40 Ligand (CD40L), GITR, IL-12p70, IL-2, IL-21, MIP-3 alpha and TRANCE.

Using modern liquid chromatography with tandem mass spectrometry (LC-MS/MS), Aass et al. (20, 33) identified 1526 proteins in the human tear proteome. They used stable isotope dimethyl labeling to compare tear fluid samples from patients with active GO to patients with GD. Mass difference of the dimethyl labels allowed relative protein quantification to determine which ones were up/downregulated in the two groups. They identified 1,212 proteins, 16 of which were significantly differentially expressed proteins between the active GO and GD groups. These included lacrimal gland proteins: lysozyme C (LYZ), lacritin (LACRT) and zinc-alpha-2 glycoprotein 1 (AZGP1), which were upregulated in tear fluid from patients with active GO. They then quantified the individual levels of LYZ, LACRT, and AZGP1 using ELISA in tears samples from active GO and GD, which suggested that these proteins were overexpressed in active GO and could be used as biomarkers alone or in combination for diagnosis or disease activity. (32) Later, Kishazi E et al. (34) used off-gel electrophoresis followed by LC-MS/MS to identify 712 proteins, of which 528 were in common with those identified by Aass et al. (32,33). They then compared tear fluids from patients with active GO and healthy controls, demonstrating 10 differentially expressed proteins. Three of these proteins were suggested as possible biomarkers by subsequent validation with Western blot and ELISA analysis: cystatin c (up-regulated), alpha-1 Antichymotrypsin (up-regulated) and retinal dehydrogenase (down-regulated). (33) (Table 5).

Table 5:

Upregulated and downregulated biomarkers in GO summarized here. (32, 33)

Upregulated Biomarkers	Downregulated Biomarkers
Caspase 14, Dermcidin, Antileukoproteinase (SLPI)	Periplakin
Procollagen-Lysine2-oxoglutarate5-dioxygenase 2	Cystatin D
Mesothelin, Apolipoprotein D (Apo D)	Mammaglobin A
Deleted in malignant brain tumors 1 (DMBT1)	Prelab A/C
Zymogen granule protein 16 homolog B(ZG16B)
Lysozyme C (LYZ), Glutathione peroxidase 3
Extracellular glycoprotein lacritin (LACRT)
Zinc-alpha-2-glycoprotein 1(AZGP1)

4. Overview of orbital inflammation/unmet needs

Inflammation of the orbit is not a specific diagnosis but a physiological end response to multiple disease processes. In addition, several systemic, immune-mediated diseases can involve the orbit (40). Entities that cause orbital inflammation include infections that may be bacterial, viral or fungal in etiology(49), systemic autoimmune diseases(46) such as Graves’ disease (41,47,48), sarcoidosis (42,75), systemic lupus, IgG4 disease (43,76), granulomatosis with polyangiitis (GPA) (44), neoplastic diseases, and as a diagnosis of exclusion, non-specific orbital inflammation (NSOI; idiopathic orbital inflammation, orbital pseudotumor). (44, 45) The differential diagnosis of orbital inflammation can be narrowed using epidemiologic data on age distribution, imaging studies (50), gene expression (51,52) and the probability of diseases.

Potential tear biomarkers in OID would allow monitoring of patients longitudinally, clarify disease pathogenesis, develop new therapeutic targets, predict response and optimal timing of therapy, as well as determine which patients will go on to develop sight threatening disease. Tear Proteins Calcium binding protein A4 (S100A4) and Prolactin Induced Protein (PIP) are Potential Biomarkers to predict the progression to severe TED (64). Thyroid eye disease (TED) is the most prevalent OID under study. (Table 5) A better understanding of the pathogenesis in other OIDs through proteomic analysis of tears may lead to additional therapies. With technological advancements, the ability to identify subtle differences in the tear proteome increases the utility of tear proteomic analysis.

5. Recent advances in proteomics will speed tear biomarker discovery

Proteomics is the large-scale study of proteins expressed in any biological compartment at a given time, using advanced separations and high throughput protein identification and quantification strategies. Large-scale untargeted (discovery-driven) proteomic analysis is considered vital in clinical analysis to identify and quantify potential biomarkers for earlier detection of diseases. (53) Mass spectrometry plays a critical role in these discovery-driven proteomic analyses and can also play a role in biomarker validation. Below we summarize four mass spectrometric based technologies that will play a significant role in advancing the use of tears as a diagnostic fluid: 1) data-dependent acquisition, 2) isobaric tagging and multiplexing, 3) targeted analysis, and 4) data-independent acquisition. (Figure 2)

Fig. 2. LC/MS methodologies used to analyze tear proteins.

Figure 2A Data-dependent acquisition MS survey scans detect peptides with specific m/z values that are then individually isolated and fragmented in the mass spectrometer to produce MS2 spectra in order of abundance. The precursor mass of each peptide and its fragment ions in the MS2 spectra are then used to query a protein database using a search engine to map individual peptides to specific proteins.

Data-Dependent Acquisition (PDA).

To date, the most common mode used to generate large scale proteomic results is data-dependent acquisition (DDA) (Fig. 2A). In DDA experiments, the mass spectrometer is configured to identify as many proteins as possible while analyzing peptides from a complex digest of proteins, created most commonly by trypsinization. Survey mass spectrometer scans (also called MS scans) are used to take snap shots of the m/z values of peptides as they emerge from reverse phase chromatographic columns. These MS scans are used by instrument control software in millisecond time scales to create pick lists of peptides, based on their intensities, so they can be identified using MS2 scans. MS2 scans are created by isolating specific peptides in the gas phase and fragmenting them along their backbones through collisions with gas molecules. The peptide mass determined from the MS scans and the fragment ions in subsequent MS2 scans are then used by software to identify each peptide by comparison to peptide masses and fragment ion masses calculated from sequences in protein databases. Identified peptides are then collated into sets matching particular proteins. Instrument control software also uses a feature called dynamic exclusion that, once analyzed, places a peptides’ m/z value on an exclusion list, so that MS2 scans for peptides from abundant proteins will not repeatedly initiate MS2 scans. This prevents peptides from abundant proteins masking the detection of minor proteins.

These DDA methods can provide not only protein identification, but also measure relative abundance differences of proteins across samples. Since the numbers MS2 spectra assigned to individual proteins is related to their abundance, summing the numbers of assigned MS2 spectra to each protein (spectral counting) can provide an abundance measure across samples. For example, this MS2 spectral counting technique was used in a study of tears from patients with dry eye and Meibomian gland dysfunction. (54) The intensity of peptide ions in survey MS scans can also be used as a measure of protein abundance, and this was recently used to assess differences in tear proteins from subjects wearing hard versus soft contact lenses. (55) These methods are often called label free quantitation. While they are simple to perform, since there is no need to tag peptides prior to analysis, they are less quantitative. This is due to changes in instrument response when analyzing long queues of samples. Tear digests analyzed independently without tagging and multiplexing also suffer from missing data. These missing data occur when MS2 scans for particular peptides are not reliably triggered in all runs. Thus, multiplexing of peptides following isotopic tagging before their analysis can provide a more accurate method to assess changes in the tear proteome.

Isobaric labeling and multiplexing.

Isobaric (same nominal mass) tagging is an extremely useful methodology that has the potential to greatly enhance biomarker detection in tears (Fig. 2B). In this strategy, a collection of chemical tags with nearly identical mass are used to label up to 16 different tear samples so that they can be mixed and analyzed simultaneously. (56) Due to the similar mass of the chemical tags, m/z features of peptides in MS scans are not increased. This allows for deeper sampling of peptides during MS2 scans and increases the numbers of identified proteins. When these multiplexed peptides with single m/z features are fragmented, they yield reporter ions with distinct masses and intensities. These are used to measure the differential abundance of proteins from which they are derived. The reporter ions are generated by cleavage of the isobaric chemical tag, either during the MS2 scan used to identify the peptide, or during a specialized MS3 scan that increases accuracy of the method. (56)

Figure 2B.

Isobarictagging and multiplexing: Tear samples from up to 16 different subjects are collected, tryptic digests prepared, each digest labeled with a different isobarictag, samples combined, and data-dependent acquisitionof MS2 data collected following a two-dimensional LC separation. Single m/z features for each combined peptide are then fragmented and low mass reporter ions are liberated, providingrelative abundance measures of each tear protein acrossthe multiplexed samples.

The generation of reporter ions with different masses is possible due to a distribution of heavy isotopes of carbon and nitrogen within the isobaric tag on either side of the molecule where it breaks to create the reporter ion. The simultaneous analysis of up to 16 samples avoids problems caused by changes in instrument response over time and avoids the missing data problem mentioned above during DDA analyses. However, perhaps the biggest advantage is the ability to devote longer instrument data collection times using two dimensions of chromatographic separation of peptides that increase the numbers of identified minor proteins. Using two-dimensional peptide separations requiring tens of hours (or even several days of instrument time) would be impractical if analyzing each sample individually. It is also possible to combine results from multiple isobaric tagging experiments by dedicating one or more of the isobaric tags to a common pooled standard of peptide digests. This common pooled standard is used to normalize reporter ion intensities between multiple runs. (57) For example, if two tags are dedicated to a common pool in each run, three multiplexed analyses with 16-plex reagents would allow the comparison of tear samples from 42 subjects. The sample requirement for a typical tear sample isobaric labeling experiment is 20 μg of tear protein digest, a requirement that is usually met with a standard collection using Schirmer strips. Thus, isobaric tagging is a valuable methodology to drive biomarker discovery in tears. (58-64)

Targeted mass spectrometry.

While DDA analyses and isobaric tagging are useful strategies to discover putative biomarkers in tears, targeted mass spectrometry can be utilized as an alternate means to confirm their differential abundance (Fig. 2C). The method can also be used to determine the absolute concentration of tear protein biomarkers. This is especially useful if an antibody against the protein is not available. A classical approach to targeted analysis is to define a list of peptides that give the most intense signal intensities when the protein of interest is digested, and then configure the mass spectrometer so that it specifically isolates these peptides and creates their MS2 fragment ion spectra. These types of analyses are called multiple reaction monitoring (MRM) and have historically been performed in triple quadruple instruments, where each peptide of interest has both its precursor m/z and the m/z values of one or more of its fragment ions specified. Since acquisition speeds during a targeted analysis are fast, it is possible to create lists for peptides from dozens of proteins of interest so that the appearance of their fragment ions can be monitored throughout the time course of their elution during an LC/MS analysis. The resultant chromatographic peaks from these fragment ions can then be integrated and used for protein quantification. Higher resolution hybrid quadrupole/time of flight (66) or quadrupole/orbitrap instruments can also be used for these types of analyses. (67) Because the signal of each peptide is essentially filtered twice, once for its precursor m/z and once for its specific fragment ion m/z, the methodology is more specific than when monitoring only precursor m/z values. Furthermore, it is possible to add heavy isotope labeled synthetic peptide standards of known concentration to digests, so the absolute concentration of a protein can be determined. For example, this technique with heavy labeled peptide standards was used to measure the absolute abundance of lactoferrin in tears. (65) The results of targeted analyses of tear proteins are greatly facilitated by the open-source tool Skyline, which automates the detection and integration of chromatographic peaks from targeted mass spectrometric results. (68) Skyline software also assists in the calculation of absolute protein concentrations when using heavy labeled peptide standards. (65) While not essential, having confirmed MS2 spectra for peptides of interest from prior DDA experiments is beneficial. This provides information to Skyline so that the most intense fragment ions can be monitored. Once targeted analysis methods for tear proteins are created in Skyline, they can be readily shared between laboratories using the Panoroma Public data repository. (69) Sharing DDA results from tear proteome studies will greatly assist creation of these targeted assays. An example of this type of data sharing can be found in the publication of the Guntermann et al., which studied tears (70).

Figure 2C.

Targeted analysis Tear proteins are sampled, trypsin digested, and isotope labeled synthetic tryptic peptide standards added at known concentration. LC/MS isthen performed by specifying the specific m/z values of both endogenous light and synthetic heavy isotope standards and specific fragment ions of each monitored. Skyline software is then used to integrate chromatographic peaks and calculatethe absolute abundance of each tear protein of interest.

Data-Independent Acquisition.

A final technology with great promise to speed biomarker discovery in tears is data-independent acquisition (DIA) (Fig. 2D), also sometimes known as SWATH (Sequential Windowed Acquisition of All Theoretical Ion Mass Spectra). While similar to DDA analyses described above, in that the method is designed to detect as many peptides as possible in a complex mixture, DIA does not require detection of peptide features to initiate MS2 scans. Instead, the mass spectrometer is configured to sequentially pass peptides in bins from several to tens of m/z range at a time until the entire m/z range of interest is sampled. This theoretically results in fragment ions from all peptides eluting during chromatographic separations being fragmented, and like the MRM targeted analysis described above, these fragment ions are repetitively sampled across peptide peaks as they elute during LC/MS analysis. The wider m/z window of peptide isolation results in multiple peptides being simultaneously fragmented to create MS2 spectra. Due to the resulting complexity of the fragment ions and lack of accurate peptide precursor masses, DIA results are not usually interrogated by searches against protein sequence databases, as used to process DDA results. However, armed with a comprehensive MS2 library of peptides from prior DDA analyses, it is possible to extract fragment ion spectra from DIA experiments to detect peptides of interest. The fragment ions are then used to generate chromatographic peaks that can be quantified. In this regard, DIA analyses are similar to targeted MRM peptide analyses and can provide superior quantitative results compared to label free DDA methods. While examples of DIA experiments of tear proteins are few, the methodology was recently used to compare tears sampled using capillary tubes versus Schirmer strips (71) and detect tear protein alterations in Bechet’s disease–associated uveitis. (72) The methodology will undoubtedly be used more commonly as MS2 spectral libraries of peptides from tear proteins become more widely available.

Figure 2D.

Data-independent acquisition Tear samples are trypsinized and separated by LC as in (C), except instead of specifying specific m/z values for peptides, all peptides are fragmented using sequential swaths of 3-20 m/z space that continuously cycle during the LC separation. In this example, the fragment ions of 4 peptides are relatedly sampled during their elution and used by Skyline software to create chromatograms that are integrated to determine the abundance of the 4 proteins of interest. These fragment ions are typically selected using MS2 spectrallibrariesof peptides found in shared databases.

6. Limitations of mass spectrometric analysis of tear proteins.

Even following the numerous tear proteomics studies summarized above, clinically useful biomarkers in orbital inflammatory diseases (OID) are rare. An ideal molecular biomarker would be minimally invasively obtained allowing for repeat testing, specific to the disease with high sensitivity, and inexpensively quantified. Large differences in tear protein concentrations limit the detection of minor proteins, such as many cytokines, using typical mass spectrometry approaches. These may be more successfully quantified using antibody-based assays. Many tear proteins are extensively post translationally modified, such as by glycosylation or proteolysis. “For example, tear proteins are known to be heavily glycosylated (81), and peptides containing these glycosylation sites would not be identified using standard searches of tandem MS results from tear protein digests”.

This may impact their ability to be detected by mass spectrometry. Mass spectrometry may be best suited for discovery and validation of biomarkers, as it may be more practical to develop antibody-based clinical tests due to the complexity and cost of mass spectrometry-based assays.

7. Conclusions and future needs.

Human tear proteomic studies have the potential to yield biomarkers to develop innovative clinical tests to diagnose and better treat orbital diseases. As tear sampling is a noninvasive and rapid method, tear-based tests hold promises for future diagnostic methods, and especially present opportunities to better manage inflammatory orbital diseases. Additionally, as a complex mixture, tears can yield not only protein biomarkers, but also biomarkers based on RNAs, lipids, and metabolites. These biomarkers could complement the conventional clinical tools accessible to ophthalmologists. Over the past decade, advances in mass spectroscopy have considerably improved our understanding of the tear proteome. Prior studies detecting and quantifying over 1,500 human tear proteins are now available. However, the overlap in the identities of tear proteins in these large datasets are still relatively poor (72), possibly due to variations in sample collection, preparation, instrumentation, protein databases used, and subjects studied. The variation in identified tear proteins between studies could also be due to the high abundance of a small number of tear proteins that create difficulties in reproducibly detecting minor proteins, especially in DDA datasets. This can be addressed with longer data collection times using two-dimensional separations of the tear protein digests, as well as by using newer, more sensitive, and faster scanning mass spectrometer platforms. (73-74) This will undoubtedly greatly expand the number of identified tear proteins, allow sets of identified tear proteins between labs to be more similar, and allow more comprehensive comparisons of differential protein abundance to identify biomarkers. The expansion of spectral libraries of peptides derived from tear proteins will also speed the creation of targeted assays of even minor tear proteins, as well as interpretation of results from comprehensive DIA analyses of tear protein digests. An increased knowledge of how the normal tear proteome differs between individuals of different age, gender, and genetic background are required, as well as changes caused by environmental factors, drug treatment, and even contact lens wear. Finally, the impact of sample collection methods on the tear proteome requires closer examination before the promise of clinically actionable tear biomarkers are realized. (79)

With a potential to redirect the diagnosis of idiopathic orbital inflammatory disorders, tear proteomics provide a new insight into the pathophysiology and therapeutics of various ocular inflammatory diseases. (78) Tears represent potential repositories for proteomic biomarkers discovery. Determining the molecular pathways enriched in OID helps in considering the drugs that are available and could readily target the pathways. A recent, promising clinical advance is the use of a monoclonal antibody to the insulin-like growth factor- (IGF)-1 receptor.(80)

Highlights:

• Description of tears as a diagnostic fluid to better manage eye disease.

• Extensive review of tear proteomics literature.

• Discussion of unmet needs in treatment of orbital inflammation and promise of tear proteomics.

• Overview of major proteomics methods used to study tears.

• Discussion of limitations and future of tear biomarker research