Abstract
More recently novel biomarkers of disease have been sought. The official National Institutes of Health (NIH) definition is “a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention” (2). Considerable effort has gone into identifying serum markers specific for cancer, but to date such approaches tend to be restricted to the monitoring of therapy; e.g., CA15.3 and CA27.29 for breast cancer (3). In autoimmune conditions measurement of autoantibodies can provide an indicator of disease; e.g., first-degree relatives of people with type 1 diabetes who are tested against a panel of relevant autoantigens to identify those at risk of immune-mediated pancreatic failure (4). In thyroid-associated ophthalmopathy (TAO), the focus of this editorial, the presence of thyroid-stimulating antibodies combined with the absence of thyroid peroxidise antibodies can pinpoint patients with Graves' disease most likely to develop TAO (5).
Biomarker discovery has also used the many and varied methods that comprise proteomics; i.e., the identification, characterization, and quantification of proteins. As the sequencing of the human genome neared completion, scientists turned their attention to the development of new techniques to dissect the complexity of proteins generated by variations in transcription and posttranslational modifications. The DNA sequence can predict the primary structure of a protein, i.e., the type and order of the amino acids, but is not able to predict novel proteins produced by alternative splicing. Once thought to be at best of minor importance and at worst an in vitro artefact, alternative splicing is now recognized as being central in producing the diversity required for normal biological function (6). There are more than 200 types of posttranslational modifications ranging from glycosylation, through sulphation to phosphorylation, and new ones are still yet to be discovered (7,8). About 5% of the proteome comprises enzymes that perform posttranslational modifications at distinct amino acid side chains or peptide linkages; this impacts protein activity, localization, and interaction with other cellular components including other proteins, nucleic acids, lipids, and cofactors. These modifications are vital. For example, TSH functions as a thyroid-stimulating agonist only when appropriately glycosylated. If not, hypothyroidism develops (9). It is apparent, therefore, that seemingly minor alterations in a mature protein can contribute to major disease processes. Technologies able to pinpoint such changes are important if the search for new diagnostics is to be successful.
The gold standard is two-dimensional (2D) gel electrophoresis but subsequent procedures are often necessary. The initial step separates proteins sampled from sera or biopsy material according to their molecular weight and iso-electric point. Ideally, the 2D gel images alone would uncover differences between disease and control samples. After separation on the 2D gel, the protein spots are stained by various methods. The distinct protein spots to be identified are then excised from the gel and digested with a protease, such as trypsin, which cleaves C-terminals to arginine or lysine residues. This results in peptides having basic amino acids at the C terminus. Subsequently, their characterization is usually based on mass spectrometry (MS) and includes matrix-assisted laser desorption/ionization (MALDI) with flight tube and drift region to measure the time-of-flight (TOF). These peptide masses form a fingerprint that is compared to theoretically expected tryptic peptide masses for each protein entry in a database. The criterion for unique identification of a protein is related to the number of peptides that match within 10%–20% coverage of the tryptic peptides of the whole protein (10). Unfortunately this approach is labor intensive and time consuming and requires a relatively large sample for the analysis and so is unlikely to be applicable in the clinical setting and certainly not for screening purposes. High-throughput techniques that require only a small amount of biological material are being sought.
In an article in this month's issue of Thyroid, a group led by George Kahaly in Mainz, Germany, has applied proteomic profiling of tears to identify biomarkers of potential diagnostic and pathophysiological importance in TAO (11). Their challenge was to identify the subclass of patients, mainly derived from those with Graves' disease, who are most likely to develop TAO. In this disorder tissue remodeling in the orbit causes proptosis and the attendant signs and symptoms, such as upper eyelid retraction, edema, and erythema of the periorbital tissues and conjunctivae. The most severe cases of TAO, about 5%, also experience intense pain, inflammation, and sight-threatening corneal ulceration or compressive optic neuropathy, which can cause blindness (12). The study by Matheis et al. (11) used surface-enhanced laser desorption/ionization time-of-flight (SELDI-TOF) mass spectrometry. SELDI-TOF is a type of MALDI analysis that uses a surface designed to capture proteins according to specific properties (e.g., electrostatic interaction). It is thus able to identify proteins and peptides from a mixture (13). SELDI-TOF is particularly useful for proteins of low molecular weight, but since it is not a tandem method (TOF/TOF), protein identification may be less precise. Therefore, to account for this, Matheis et al. (11) also applied MALDI-TOF mass spectrometry in the second phase of the study.
The main attraction of SELDI-TOF is that it requires very little biological material and its rapid high-throughput technology allows analysis of large numbers of samples. As noted, the surface used to bind the proteins has unique properties that provide a separation process. For example, proteins bound by metal affinity (nickel binds proteins containing histidine with high affinity) identified a panel of three biomarkers for breast cancer with 93% sensitivity and 91% specificity (3). However, even these markers are recommended only for monitoring therapy, not for diagnosis. In the Matheis et al. (11) study, it would have been interesting to know if the surface was chosen based on prior optimization and also to have more details of its features. This information would help to predict some of the characteristics of the tear proteins that lead to their separation and provided insights into their nature.
Data from studies using SELDI-TOF have been reported in two ways to date. One is to provide a signature of protein peaks that identifies a particular disease group but without knowing the precise proteins involved. This approach has been used with some success in, for instance, ovarian cancer. Another is to use SELDI-TOF to analyze large numbers of samples and to follow this with another liquid chromatography method such as MALDI-TOF to provide identification of any proteins associated with disease—as in the study by Matheis et al. (11). Analysis of a larger number of samples, maybe comparing data obtained using different surfaces, might lead to the identification of a TAO tear signature that would be more amenable to clinical applications; the second approach might provide insight into pathogenetic mechanisms.
Lacrimal gland enlargement and reduced tear production have been reported in patients with TAO. Lacrimal gland involvement in TAO (14,15) may result from targeting of acinar cells in the gland by TSH or thyroid hormone (16,17), and there appears to be altered composition and impaired cytokine balance in TAO (18,19). Proteomic analysis of tear films suggest that low molecular weight biomarkers between 3000 and 20,000 Da are down-regulated in TAO, with only a few proteins being over-expressed (20). In the current article (11), proline-rich protein 4 (PRP4), nasopharyngeal carcinoma–associated proline-rich protein 4, and β-microglobulin were down-regulated and lysozyme C was up-regulated in tears from the patient group compared with controls. The reported change in cystatin S may reflect the beneficial effect of steroid treatment in TAO. These observed differences could simply reflect the normal variation in tear composition, although a recent study reported very little day-to-day variation in tear peptides and proteins below 20 kDa in normal donors analyzed by MALDI-TOF mass spectrometry (21). This is reassuring especially since Matheis et al. (11) focused on proteins of low molecular weight.
However, there may be other considerations to take into account. Tear films are suitable for proteomic analysis because they contain a high concentration of proteins (about 8 μg/μL). Tears can be collected by glass microcapillary tubes or Schirmer strips with/without local anesthesia. The study in Thyroid (11) used Schirmer strips with local anesthesia, a method that gives more physiological tears since it measures basic tear secretion. If the test is performed without anesthetic, as in some previous studies (18), the strip collects both reflex and secreted tears and may thus yield aberrant results. The collected tears in the Matheis article were stored at −80°C but without mention of protease inhibitor addition. Clearly any inhibitors used would have to be highly specific (remembering that trypsin digestion is a required step in MALDI-TOF, for example) but comparisons of “native versus inhibited” samples could shed light on the proteolytic processes that have been recognized to be a part of disease pathogenesis.
The article by Matheis et al. (11) attempted to analyze the SELDI-TOF data from 45 TAO patients in groups defined by clinical activity score, differing treatment modalities and smoking status, etc. Much larger numbers of both patients and well-defined controls are required to obtain meaningful subset analyses. Of interest, one of the down-regulated proteins in TAO was PRP4, which has been reported to have a protective, antimicrobial function in the eye (22). PRP4 is also present in saliva, which would have provided a control sample in the patients whose tears were analyzed to determine whether the observed down-regulation is systemic or reflects a local and possibly disease-associated effect in TAO.
In conclusion, Matheis et al. (11) have applied a state-of-the-art method to analyze tears, with the aim of increasing our understanding of disease pathogenesis and identifying biomarkers useful in the diagnosis and management of TAO. The results of this preliminary study are promising and illustrate the need for larger, multicenter prospective studies to provide sufficient data for the establishment of a tear TAO signature. The attraction of the methodology across many disciplines, as a tool for diagnostic developments, will facilitate improvements that should transform a sophisticated scientific strategy into a clinical reality.