Article pubs.acs.org/jpr
Global Metabonomic and Proteomic Analysis of Human Conjunctival Epithelial Cells (IOBA-NHC) in Response to Hyperosmotic Stress Liyan Chen,†,‡,¶ Jing Li,†,§,∥,¶ Tiannan Guo,⊥ Sujoy Ghosh,# Siew Kwan Koh,† Dechao Tian,†,▽ Liang Zhang,†,▽ Deyong Jia,† Roger W. Beuerman,†,§,○ Ruedi Aebersold,⊥,◆ Eric Chun Yong Chan,*,‡ and Lei Zhou*,†,§,○ †
Singapore Eye Research Institute, The Academia, 20 College Road, Discovery Tower Level 6, Singapore 169856, Singapore Department of Pharmacy, Faculty of Science, National University of Singapore, 18 Science Drive 4, Singapore 117543, Singapore § Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, 1E Kent Ridge Road, NUHS Tower Block Level 7, Singapore 119228, Singapore ∥ Department of Ophthalmology, Xin Hua Hospital, Shanghai Jiao Tong University School of Medicine, 1665 Kongjiang Road, Shanghai 200092, China ⊥ Institute of Molecular Systems Biology, Department of Biology, ETH Zurich, Auguste-Piccard-Hof 1, 8093 Zürich, Switzerland # Cardiovascular and Metabolic Disorders Program & Centre for Computational Biology, Duke-NUS Graduate Medical School, 8 College Road, Singapore 169857, Singapore ▽ Department of Statistics and Applied Probability, Faculty of Science, National University of Singapore, 6 Science Drive 2, Singapore 117546, Singapore ○ Neuroscience and Behavioral Disorders Program, Duke-NUS Graduate Medical School, 8 College Road, Singapore 169857, Singapore ◆ Faculty of Science, University of Zurich, Winterthurerstrasse 190, 8057 Zürich, Switzerland ‡
S Supporting Information *
ABSTRACT: “Dry eye” is a multifactorial inflammatory disease affecting the ocular surface. Tear hyperosmolarity in dry eye contributes to inflammation and cell damage. Recent research efforts on dry eye have been directed toward biomarker discovery for diagnosis, response to treatment, and disease mechanisms. This study employed a spontaneously immortalized normal human conjunctival cell line, IOBANHC, as a model to investigate hyperosmotic stress-induced changes of metabolites and proteins. Global and targeted metabonomic analyses as well as proteomic analysis were performed on IOBA-NHC cells incubated in serum-free media at 280 (control), 380, and 480 mOsm for 24 h. Twenty-one metabolites and seventy-six iTRAQ-identified proteins showed significant changes under at least one hyperosmotic stress treatment as compared with controls. SWATH-based proteomic analysis further confirmed the involvement of inflammatory pathways such as prostaglandin 2 synthesis in IOBA-NHC cells under hyperosmotic stress. This study is the first to identify glycerophosphocholine synthesis and O-linked β-Nacetylglucosamine glycosylation as key activated pathways in ocular surface cells under hyperosmotic stress. These findings extend the current knowledge in metabolite markers of dry eye and provide potential therapeutic targets for its treatment. KEYWORDS: dry eye, osmostic stress, hyperosmolarity, osmoprotectants, metabolomics, metabonomics, proteomics, iTRAQ, SWATH
■
INTRODUCTION
prevalence of dry eye is high and estimated among individuals over 40 years old in Caucasian populations,2−4 and 21−33% in tions.5−7 The prevalence of dry eye increases higher in females.2
“Dry eye” is a multifactorial disease of the tear film and ocular surface. It is characterized by symptoms of discomfort, disturbance in vision, and tear film instability.1 There are two mechanistic subtypes of dry eye: aqueous deficient and evaporative dry eye. Both subtypes coexist.1 In particular, for evaporative dry eye, tear hyperosmolarity plays a role in inducing ocular surface inflammation and cell damage.1 The © XXXX American Chemical Society
to be 11−17% predominantly Asian populawith age and is
Received: May 20, 2015
A
DOI: 10.1021/acs.jproteome.5b00443 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
suggesting that these pathways may be therapeutic targets in treating dry eye.
The disparity in global prevalence rates could be attributed to the use of different diagnostic criteria, with individual clinical signs demonstrating poor agreement with disease severity.8 A combination of clinical assessments including self-reported symptoms, corneal and conjunctival staining, tear break up time, tear volume by Schirmer’s test, and meibomian gland function are used for the diagnosis of dry eye and measurement of response to treatment.1 While the combination of such methods yields improved sensitivity and specificity,9 diagnosis remains time- and resource-consuming and highly dependent on the judgment of trained physicians.8 The lack of an objective assessment of treatment outcomes also hinders the development of targeted pharmacotherapies. Artificial tear formulations are available, but they merely provide symptomatic relief. In view of the gaps in clinical knowledge, current research efforts in dry eye are directed toward: (i) selective biomarker discovery to reliably diagnose dry eye and disease subtypes and (ii) elucidation of disease mechanism in animal and cell models. Several studies have been conducted to identify biomarkers of dry eye disease using proteomic approaches,10 including subsets of mild dry eye11 and contact lens-related dry eye.12 Our group has also studied the reproducibility of dry eye protein biomarkers in tears in geographically different populations.13 Hyperosmolarity is known to exert pleiotropic effects on cellular function on mammalian renal cells,14 including cell cycle arrest and increased apoptosis.15 Several in vitro studies have been performed to investigate the mechanisms associated with hyperosmolarity-induced cell death of ocular surface cells. Li et al. demonstrated the involvement of mitogen-activated protein kinase (MAPK) signaling pathway.16 Subsequent studies demonstrated that c-Jun N-terminal kinases (JNK) and extracellular signal-regulated kinase (ERK) pathways17 induce apoptosis in human corneal epithelial through a cytochrome c-mediated pathway18 dependent on transglutaminase.19 Collectively, these in vitro mechanistic studies confirmed the involvement of cellular signaling pathways in hyperosmolarity-induced cell death and suggested the potential dysregulation of signaling pathways and endogenous metabolites in dry eye. Most published work has focused on mRNA transcripts or proteins, analyzed by targeted approaches.16−21 Considerably less information is available on the regulation of metabolites in ocular surface cells when exposed to hyperosmolarity.22 We postulate that global (nontargeted) profiling of the metabolome and proteome could reveal novel and complementary information on pathways affected in dry eye. In addition to iTRAQ-based proteomics, SWATH-based proteomic analysis was performed to validate the findings on hyperosmolarityaffected pathways gleaned from iTRAQ data. Using fragment ion spectral libraries generated from standard data-dependent acquisitions, SWATH-MS-acquired data sets could be iteratively mined to quantify peptides of interest. The dataindependent nature of SWATH-MS acquisition captures both qualitative and quantitative information on peptides and with improved sensitivity over label-free quantitation.23 In this study, we investigated perturbations in a cell model of dry eye using global and targeted metabonomic analysis, iTRAQ-based proteomics, and SWATH-based proteomic analysis. Glycerophoshocholine was found to be the key osmoprotectant in human conjunctival epithelial cells. O-linked β-N-acetylglucosamine glycosylation, S-adenosylmethionine (AdoMet) metabolism, and prostaglandin synthesis were identified as key pathways activated under hyperosmotic stress,
■
EXPERIMENTAL METHODS
Cell Line
The nontransfected, spontaneously immortalized normal human conjunctival epithelial (IOBA-NHC) cell line was a gift from Dr. Yolanda Diebold at the University of Valladolid, Spain.24 Cells were grown to 80% confluence and adapted to serum-free media overnight. The cells were further cultured for 24 h in serum-free media at 280, 380, and 480 mOsm for the control, treatment T1, and treatment T2 respectively. Five replicates were prepared for global metabonomic analysis, and three replicates each were prepared for targeted metabonomics and iTRAQ- and SWATH-based proteomic analysis (Figure S1). Cell culture and sample preparation methods are provided in the Supporting Information. Global Metabonomic Analysis
Analysis was performed on a Prominence UFLC system (Shimadzu) coupled to a TripleTOF 5600 mass spectrometer (AB Sciex). A Waters T3 C18 2.1 × 100 mm, 3 μm column (Waters) was used for separation. Each sample was reconstituted in 50 μL of water, and 10 μL was injected for each analysis in positive and negative ionization mode. The mobile phase for positive ionization mode was 0.1% formic acid (FA) in water (eluent A) and 0.1% FA in acetonitrile (eluent B). The gradient profile was 2% B from 0 to 2 min, 15% B at 8 min, 50% B at 14 min, 90% B from 18 to 20.5 min, and 2% B from 21 to 27 min. The mobile phase for negative ionization mode was 5 mM ammonium formate and 0.05% FA in water (eluent A) and 5 mM ammonium formate and 0.05% FA in 9:1 acetonitrile/water (eluent B). The gradient profile was 2% B from 0 to 2 min, 15% B at 8 min, 50% B at 14 min, 100% B from 18 to 20.5 min, and 2% B from 21 to 27 min. The flow rate in both chromatographic methods was 0.3 mL/min. Acquisition of MS/MS spectra in both ionization modes was controlled by the information-dependent-acquisition (IDA) function of the Analyst TF software (AB Sciex). One TOF MS survey scan (150 ms) was followed by 25 MS/MS scans (50 ms each). The mass range TOF-MS and MS/MS were 50−1000 and 25−1000 m/z respectively. The following parameters were applied: dynamic background subtraction, charge monitoring to exclude multiply charged ions and isotopes, and dynamic exclusion of former target ions for 5 s. Collision energy (CE) was ramped from 20−30 V in MS/MS scans. Peak Finding and Alignment of Metabonomic Data
MarkerView (AB Sciex) was used for peak detection and alignment of raw chromatographic data obtained from acquisition in positive and negative ionization mode, respectively. Peak detection and alignment parameters are provided in Table S1. One peak table comprising the mass to charge ratio (m/z), retention time (RT), and integrated peak area of detected peaks in the analyzed samples was generated for each ionization mode. The following peak classes were excluded from subsequent analysis: (i) peaks attributed to isotopes, (ii) peaks with an average area that were less than 3 times the average area in blank injections, and (iii) peaks with a coefficient of variation greater than 80% among replicates. The samples in each peak table were normalized to their respective 90th percentile peak area sums. B
DOI: 10.1021/acs.jproteome.5b00443 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
Samples were reconstituted in 20 μL of dissolution buffer (iTRAQ kit). iTRAQ reagents 114, 115, and 116 were added to control and T1 and T2 samples, respectively. Each mixture was mixed for 2 h at 25 °C. Labeled samples were then pooled together, dried, and desalted on ultramicro spin columns (The Nest Group) prior to nano-LC−MS/MS analysis. Chromatographic separation of peptides was performed on an Ultimate 3000 nano-LC system (Dionex, Thermo Fisher Scientific). Samples were first loaded onto a 75 μm × 2 cm Acclaim PepMap 100 C18 trap column (Dionex, Thermo Fisher Scientific) at a flow rate of 5 μL/min. After a 5 min wash with 0.1% FA and 2% acetonitrile in water, the system was switched into line with a 75 μm × 15 cm Acclaim PepMap RSLC C18 analytical column (Dionex, Thermo Fisher Scientific). The mobile phase was 0.1% FA in water (eluent A) and 0.1% FA in 98:2 v/v of acetonitrile/water (eluent B). The gradient profile was 5% B at 0 min, 7% B at 5 min, 60% B at 97 min, and 95% B from 100 to 105 min. The flow rate was 300 nL/min. The analytical column was connected to a spray tip (New Objectives), and the spray tip was coupled to the nanospray interface of the 5600 TripleTOF mass spectrometer. Data were acquired via IDA mode using the Analyst TF 1.6 software (AB Sciex). Key parameter settings for the TripleTOF 5600 mass spectrometer were as follows: ionspray voltage floating (ISVF) 2500 V, curtain gas (CUR) 30, interface heater temperature (IHT) 125, ion source gas 1 (GS1) 12, and declustering potential (DP) 100 V. The IDA parameters are as follow: 0.25 s of TOF MS survey scan (m/z range: 350−1250) was followed by 30 product ion scans of 75 ms each (m/z range: 100−1800). The switching criteria was set to include only ions with m/z between 350 and 1250 with charge state of 2−5 and an abundance threshold of over 100 counts. Former target ions were excluded for 12 s. CE was controlled by the IDA rolling CE parameter script.
Statistical Analysis and Metabolite Identification
The Kruskal−Wallis test was used to identify statistically significant m/z-RT metabolite peaks in at least one group (control, T1, or T2). Subsequently, the Mann−Whitney U test was used to identify m/z-RT peaks with peak areas that were significantly different in treatments T1 or T2 when compared with controls. Metabolite fold-changes between treatment and control groups were calculated using the arithmetic mean. The m/z values of each marker peak were queried in the METLIN metabolite database25 to obtain its putative metabolite identifications. For this step, the batch search mode was adopted, with mass accuracy set to 12 ppm. The following adducts were applied to the m/z values of marker peaks obtained via positive ionization mode: [M+H]+, [M+NH4]+ [M + Na]+, [M+K]+, and [M−H2O+H]+. Similarly, [M−H]−, [M− H2O−H]−, [M+Cl]−, [M+FA−H]− were applied to m/z values obtained via the negative ionization mode. Metabolite identities were confirmed by retrieval of experimental MS/MS spectra and comparison with the respective entries in METLIN25 or Massbank26 spectra databases. Selected metabolites were verified with standards. Unresolved metabolites with the same molecular formula were reported as mixtures. Data from global Metabonomic analysis has been deposited to MetaboLights27 with the data set identifier MTBLS214 (http://www.ebi.ac.uk/metabolights/MTBLS214). Targeted Metabonomic Analysis
Carnitine and glycerophosphocholine standards were purchased from Sigma-Aldrich. Separation was performed on a Prominence UFLC system (Shimadzu) with a SeQuant ZIC HILIC 2.1 × 100 mm, 3.5 μm column (Merck). The mobile phase for was 5 mM ammonium formate and 0.05% FA in water (eluent A) and 5 mM ammonium formate and 0.05% FA in 9:1 acetonitrile/water (eluent B). The gradient profile was 80% B at 0 min, 40% B at 8 min, 20% B from 9 to 11 min, and 80% B from 11 to 15 min. The flow rate was 0.4 mL/min. Multiple-reaction monitoring (MRM) was performed on a QTrap 6500 mass spectrometer (AB Sciex). The optimized MRM parameters for each target compound are shown in Table S2. Metabolite levels in lysates were normalized to protein content.
Analysis of iTRAQ Data
ProteinPilot (AB Sciex, version 4.5) was used to analyze MS/ MS iTRAQ data output and searched against the UniProt database (September 2010 release, 40 516 proteins searched). Important settings in ProteinPilot were configured as follows: (i) sample type: iTRAQ 4plex (peptide labeled); (ii) cys alkylation: iodoacetamide; (iii) digestion: trypsin; (iv) instrument: TripleTOF 5600; (v) special factors: none; (vi) ID focus: biological modifications; (vii) search effort: thorough ID; and (viii) bias correction and background correction. 95% confidence level was used at the peptide level. False discovery rate (FDR) analysis in the ProteinPilot software was performed, and FDR
