Discovery of tear biomarkers in children with chronic non-infectious anterior uveitis: a pilot study

Nine affected eyes of seven children (two oligoarticular and one polyarticular RF negative JIA-U and 4 I-CAU) were evaluated (Table 1). LC-MS/MS/SPS-MS3 was used to determine proteomic profiles from tears collected using Schirmer strips. We quantified 1804 unique proteins, following a QC step that filtered proteins based on expression levels, and 1605 proteins had a present call in half of the samples. These 1605 proteins were used in subsequent analysis.

We first performed a targeted analysis for cytokines and chemokines reported in AqH (Additional file 1) [8‐11]. We detected CD14, S100 calcium binding protein (S100) A8/A9, serum amyloid A (SAA1), latency-associated peptide (LAP3), TTR, and macrophage migration inhibitory factor (MIF) in all nine tear samples. On further comparison, CD14, S100A8, and SAA1 had higher expression in children with JIA-U compared to I-CAU, while S100A9, LAP3, TTR, and MIF had lower expression in children with JIA-U (Fig. 1).

We identified 29 unique proteins with significant differences between JIA-U and I-CAU. We then performed hierarchical clustering analysis using these differentially expressed proteins (Fig. 2). Green, black, and red reflect high, average, and low expression, respectively, in JIA-U compared to I-CAU. As shown in Fig. 2, these proteins could differentiate between samples obtained from children with JIA-U and I-CAU. Fourteen proteins were expressed at a higher level in JIA-U, and 15 were expressed at a lower level in JIA-U compared to I-CAU. Proteins with biological relevance were SAA1, metalloproteinase inhibitor 1 (TIMP1), beta-hexosaminidase subunit beta (HEXB), and serine/threonine-protein kinase/endoribonuclease IRE1 (ERN1), which were increased in JIA-U. Dnaj homolog subfamily B member 1 (DNAJB1) was decreased in JIA-U compared to I-CAU. The relevance of the other proteins is unknown (Fig. 2). These observations suggest that tear biomarker profiles could be useful in distinguishing subtypes of anterior uveitis.

Gene ontology annotation using DAVID categorized pathways of molecular functions, cellular components, and biological processes. Ontology revealed 15 out of 29 altered proteins in pathways related to extracellular exosomes (p = 6.4 x 10^–6) in both groups.

This uncontrolled pilot study included only a small number of children as a proof-of-principle that tears can identify proteins found in AqH. A limitation is that the sample size was too small to adjust for the potential confounders of treatment effect and level of disease activity, so these results should be considered as hypothesis-generating (Table 3). Differences in protein, cytokine, and chemokine levels have been reported in studies that examined AqH of children with uveitis and tears of adults with uveitis compared to controls without uveitis. However, it is important to examine the tear profile of children with JIA without uveitis and pediatric healthy controls without ocular disease since the controls in AqH studies had congenital cataracts and glaucoma. Serial tear collection with longitudinal follow-up is ongoing in children with JIA, other forms of uveitis, and pediatric controls to replicate and extend these findings to a larger group of study participants. These future studies will allow further comparisons by disease group, uveitis activity, and treatment response. A small number of children were included in our study, and it is not a reflection of the number of children diagnosed with uveitis in our practice. This was a pilot study to determine if we could detect proteins in tears that were previously detected in AqH and to refine our processes and methods. Not all children contributed tears from both eyes, as only the eyes with uveitis were included. We were not able to include all samples in our analysis due to issues with protein recovery which may be related to sample collection. Also, analyses did not adjust for correlation between the eyes of patients with bilateral disease. A strength of this study is that it is a pediatric study using state-of-the art techniques and demonstrates the feasibility of tear sample collection in this population.

Children with JIA-U and I-CAU were screened and enrolled at the Emory Eye Center during their routine ophthalmology clinic visits from October 2015 to March 2017. Inclusion criteria were (1) a diagnosis of chronic anterior uveitis diagnosed by a uveitis fellowship-trained ophthalmologist, with or without JIA by the International League of Associations for Rheumatology classification in children with arthritis [37] and (2) 5 years of age or older.

Data collected by medical record review included date of birth, sex, self-reported race/ethnicity, JIA category, uve- itis characteristics (onset date, diagnosis date, laterality, ocular complications), anterior chamber (AC) cell score per standardization of uveitis nomenclature (SUN) cri- teria [38], and ANA status. Use of topical and systemic medications was reviewed at time of tear collection. Data from the ophthalmic exam were recorded at time closest to tear collection.

Ophthalmologists or trained study staff collected tear samples from children using Schirmer strips which are routinely used for dry eye evaluation. We included eyes with a history of uveitis only. After local anesthesia (topical 0.5% proparacaine hydrochloride, Bausch & Lomb, Roches- ter, NY, USA) was administered, residual anesthetic fluid was removed from the conjunctival cul-de-sac while avoiding corneal contact. A Schirmer strip was placed into the temporal inferior fornix of each eye (approximately 6 mm from the lateral canthus), avoiding the corneal surface. Reflex tearing was avoided as much as possible. The eye was closed for 5 min. Alternatively, the strip was removed when fully saturated (maximum 5 min). The Schirmer strip was placed in an Eppendorf micro centrifuge tube on ice and stored at − 80 °C until processing.

Each strip was soaked and vortexed in 500 uL of urea lysis buffer (8 M urea, 100 mM NaH2PO4, pH 8.5), including 5 uL (100× stock) HALT protease and phosphatase inhibitor cocktail (Pierce). Protein supernatants were transferred to 1.5-mL Eppendorf tubes, and proteins were extracted as previously published [39]. An aliquot equivalent to 10 μg of total protein was removed from each sample and combined to obtain two global internal standards (GIS) used later for tandem mass tag (TMT) labeling (total was split into two aliquots of 40 μg of total protein). For each sample, 40 μg of total protein was processed.

The samples were randomized over four TMT 10-plex batches. In each batch, the GIS samples took up channels 1 and 10 (TMT-126 and TMT-131, respectively). Labeling was performed according to the manufacturer’s protocol, and cleanup was performed according to a previously published method [39].

The protocol for electrostatic repulsion interaction chro- matography (ERLIC) fractionation was adapted from a published method [40].

Liquid chromatography-tandem mass spectrometry with MS3 selective precursor selection (LC-MS/MS/SPS-MS3) was used to identify proteins in patients with JIA-U (n = 3) and I-CAU (n = 4). LC-MS/MS/SPS-MS3 was adapted from a published procedure [39, 41].

MS/MS spectra were searched against a Uniprot curated human database (downloaded on 4/15/2015 with 90,300 sequences) with Proteome Discoverer 2.1 (ThermoFisher Scientific, San Jose, CA, USA). Search parameters and protein quantitation using MS3 reporter ions were previ- ously reported [39]. Ratio of sample over the GIS of nor- malized channel abundances were used for comparison across all samples.

Descriptive statistics were performed using frequencies, percentages, medians, and IQR, as appropriate.

Network analysis explored associations among proteins involved in different biological processes to find the most significant shared pathways using DAVID 6.7 (the Database for Annotation, Visualization and Integration Discovery) [42, 43]. The input data for the network analyses consisted of the significantly expressed proteins represented on the heatmap analysis.