Article pubs.acs.org/jpr
Quantitative Proteomics Analysis of Vitreous Humor from Diabetic Retinopathy Patients Sirpa Loukovaara,† Helka Nurkkala,‡,⊥ Fitsum Tamene,‡,⊥ Erika Gucciardo,§ Xiaonan Liu,‡,⊥ Pauliina Repo,§ Kaisa Lehti,§ and Markku Varjosalo*,‡,⊥ †
Unit of Vitreoretinal Surgery, Department of Ophthalmology, University of Helsinki and Helsinki University Hospital, FI-00029 Helsinki, Finland ‡ Molecular Systems Biology Research Group, Institute of Biotechnology, University of Helsinki, FI-00014 Helsinki, Finland § Research Programs Unit, Genome-Scale Biology and Haartman Institute, Biomedicum Helsinki, University of Helsinki, FI-00014 Helsinki, Finland ⊥ Proteomics Unit, Institute of Biotechnology, University of Helsinki, FI-00014 Helsinki, Finland S Supporting Information *
ABSTRACT: Initial triggers for diabetic retinopathy (DR) are hyperglycemia-induced oxidative stress and advanced glycation end-products. The most pathological structural changes occur in retinal microvasculature, but the overall development of DR is multifactorial, with a complex interplay of microvascular, neurodegenerative, genetic/ epigenetic, immunological, and secondary inflammation-related factors. Although several individual factors and pathways have been associated with retinopathy, a systems level understanding of the disease is lacking. To address this, we performed mass spectrometry based label-free quantitative proteomics analysis of 138 vitreous humor samples from patients with nonproliferative DR or the more severe proliferative form of the disease. Additionally, we analyzed samples from anti-VEGF (vascular endothelial growth factor) (bevacizumab)-treated patients from both groups. In our study, we identified 2482 and quantified the abundancy of 1351 vitreous proteins. Of these, the abundancy of 230 proteins was significantly higher in proliferative retinopathy compared with nonproliferative retinopathy. This specific subset of proteins was linked to inflammation, complement, and coagulation cascade proteins, protease inhibitors, apolipoproteins, immunoglobulins, and cellular adhesion molecules, reflecting the multifactorial nature of the disease. The identification of the key molecules of the disease is critical for the development of new therapeutic molecules and for the new use of existing drugs. KEYWORDS: diabetic retinopathy, quantitative proteomics, label-free quantification, vitreous humor, complement system, coagulation cascade, apolipoproteins
■
INTRODUCTION Diabetic retinopathy (DR) is the most common eye complications of diabetes causing progressive damage of retinal microvessels characterized by microaneurysms, small intraretinal hemorrhages, capillary closures, and hard exudates.1,2 In the early, nonproliferative state of the disease, the microvascular structural changes are mild or nonexistent; the retinal capillaries are weakened causing microaneurysms to protrude from their walls. The microanuerysms may leak fluid into the retina, which leads to swelling of the macula causing diabetic macular edema (DME). These complications, if untreated, can eventually lead to ischemia and more advanced disease stateproliferative diabetic retinopathy (PDR). PDR is characterized by abnormal angiogenesis with endothelial progenitor cell activation and lymphatic-like differentiation, increased vascular permeability, and fibrotic responses.3,4 In untreated conditions, neovascularization and fibrosis formation will ultimately lead to vision loss due to tractional retinal detachment (TRD). Overall, the pathogenesis of DR is multifactorial, with a complex interplay © XXXX American Chemical Society
of microvascular, neurodegenerative, genetic/epigenetic, immunological, and secondary inflammation-related factors. DR remains the leading cause of vision loss among adults aged under 40 years.5 In 2013, it was estimated that over 381 million people globally were affected by diabetes, and approximately one-third of them develop some degree of DR. Current treatment modalities for DR, such as laser photocoagulation, intravitreal corticosteroids, intravitreal antivascular endothelial growth factor (VEGF) agents, and vitreo-retinal surgery, are applicable only at advanced stages of the disease, have large patient-to-patient variation, and are associated with significant adverse effects.6−8 Therefore, more systems level understanding of the underlying molecular pathomechanisms, novel prognostic indicators, and pharmacological treatment options of the different stages of the DR are required. Received: April 22, 2015
A
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
visual acuity, intraocular pressure, axial length, and biomicroscopy of anterior and posterior segment of the eye. Clinical surgical enrolment indications for diabetic vitrectomy were severe prolonged nonclearing vitreous hemorrhage (>2 months), active PDR with or without TRD, or DME with or without vitreous traction.23 The vitreous humor samples (up to 1000 μL) were collected before the conventional three port pars plana vitrectomy (20G or 23G, Constellation, Alcon Instruments, Inc., Fort Worth, TX, USA) without an infusion of artificial humor. Samples were collected by manual suction into a sterile syringe and transferred into sterile 1.5 mL microcentrifuge tubes, snap frozen, and stored at −70 °C until further analysis. Altogether, 164 vitreous humor samples were analyzed resulting in a total of 199 LC−MS analyses. For LC−MS reproducibility analysis, 35 samples were analyzed as two technical replicates. From the 164 individual samples, 26 were excluded as they were either originating from the same patient (both eyes operated) or contained a high amount of α- and βcrystallins.24 The remaining 138 samples from 138 individual patients consisted of 49 samples in the non-PDR group and 74 in the PDR group, and additional 10 non-PDR and five PDR samples from anti-VEGF (bevacizumab, Avastin; Genentech)treated patients.
The research on DR has been hindered by the lack of suitable animal models demonstrating all the vascular and neural complications associated with the advanced, proliferative stages of DR that occur in humans.9 Since direct access to the human retina is not possible, the vitreous humor samples obtained by vitreoretinal surgery are used to study the pathology.10−12 The role of the vitreous humor, containing soluble proteins and other macromolecules, is to maintain normal morphology of the eye, allow light to reach retina, and support homeostasis in the surrounding tissues.13 Unlike the continuously replenished fluid-like aqueous humor in the anterior chamber of the eye, the gel-like vitreous humor is more stagnant.14 Although the vitreous lacks its own vasculature, surrounding blood vessels from retina and ciliary body nourish the vitreous with hyaluronic acid, prealbumin, transferrin, glycoproteins, and, in low quantities, hundreds of other proteins.13,15,16 Recently, mass spectrometry (MS) based proteomics has provided a means for global proteome characterization of the human vitreous humor,13,15,17 and MS analyses have been used in the analysis of ocular fluids in different eye conditions including cataract,18 idiopathic epiretinal membranes,19 rhegmatogenous retinal detachment with proliferative vitreoretinopathy,20 and DR.10−12 Despite these efforts, we are still far from detailed understanding of the complex mechanisms involving vitreous proteins in various vitreoretinal eye diseases. Yet, the qualitative and quantitative analysis of the vitreous proteome alterations has great potential in identifying better diagnostic/prognostic markers and drug target candidates for improved therapeutic interventions. Since multiple vitreous-resident proteins will contribute to the pathogenic mechanisms in human DR,21,22 a global proteome-wide analysis is of high importance. In this study, we quantitatively mapped the vitreous proteome changes in the disease progression from non-PDR to PDR using liquid chromatography coupled to mass spectrometry (LC−MS). Our results provide: (1) the largest protein atlas of the DR vitreous proteome to date with 2482 identified and 1351 quantified intravitreal proteins to be used by further follow-up analyses; (2) the identity of proteins differing significantly between the non-PDR and PDR, providing a candidate target list for diagnostic, prognostic, and therapeutic approaches; (3) an extended view on the role of, for example, complement, coagulation, and kinin-kallikrein cascades as well as apolipoproteins in the progression of DR; and (4) insight on the effect of anti-VEGF treatment on the non-PDR and PDR vitreous proteomes.
■
Sample Preparation for the LC−MS
To clear the vitreous humor from cells or cell debris, the samples were centrifuged at 21 000g for 15 min at 4 °C. Average protein concentrations (μg/μL) of the vitreous samples were measured using bicinchoninic acid (BCA) protein assay kit (Pierce, Thermo Scientific, Waltham, MA USA) according to the manufacturer’s instructions. An aliquot of each vitreous sample was diluted in sterile water before the measurement, thus leading to sample dilutions between 1:30 and 1:50 in the BCA reaction mix that minimized any effects from blood (iron) or other interfering substances. For LC−MS analysis, an amount corresponding to 100 μg of total protein was taken. The individual samples were reduced with dithiothreitol, alkylated with iodoacetamide, followed by trypsin digestion with Sequencing grade Modified Trypsin (Promega, Madison, WI, USA) in the presence of 1 M urea. Tryptic peptides were purified with C18 microspin columns (Nest Group, Southborough, MA, USA). LC−MS Analysis
MS analysis was performed on Orbitrap Elite hybrid mass spectrometer (Thermo Scientific) coupled to EASY-nLC II system (Thermo Scientific) using the Xcalibur version 2.7.0 SP1 (Thermo Scientific). The tryptic peptide sample mixture was automatically loaded from autosampler into a C18-packed precolumn (EASY-Column 2 cm × 100 μm, 5 μm, 120 Å, Thermo Scientific) at a flow rate of 1 μL/min in 10 μL volume of buffer A (1% acetonitrile (ACN), and 0.1%, formic acid (FA), in HPLC grade water). Peptides were transferred onward to C18-packed analytical column (EASY-Column 10 cm × 75 μm, 3 μm, 120 Å, Thermo Scientific) and separated with 120 min linear gradient from 5 to 35% of buffer B (98% ACN and 0.1% FA in HPLC grade water) at the flow rate of 300 nL/min. This was followed by 5 min gradient from 35 to 80% of buffer B, 1 min gradient from 80 to 100% of B, and 9 min column wash with 100% B at the constant flow rate of 300 nL/min. The MS analysis was performed in data-dependent acquisition where one high resolution (120 000) FTMS full scan (m/z 300−1700) was followed by top20 CID-MS2 scans in ion trap
EXPERIMENTAL SECTION
Diabetic Subjects Enrolled and Vitreous Sample Collection
The study design was a prospective consecutive observational study. Diabetic patients were recruited to the study in the Unit of Vitreoretinal Diseases, Helsinki University Hospital, Helsinki, Finland during November 2010 to December 2012. This tertiary clinic serves the vitreoretinal patients with sightthreatening eye conditions from Southern Finland, covering roughly a district with 1.5 million inhabitants. The present study was conducted according to the principles of the Declaration of Helsinki and was approved by the Institutional Review Board and Ethical committee of Helsinki University Central Hospital. Signed informed consent was obtained from each participant before inclusion to the study. All patients underwent proper eye examination preoperatively by the recruiting surgeon. Eye examination included measurement of B
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
Figure 1. The label-free quantitative proteomics pipeline for analyzing the vitreous humor samples from patients with DR. (A) A cross-section of the human eye (left inset) and the illustration of the fundus of healthy, nonproliferative DR (non-PDR), and proliferative DR (PDR) eyes, respectively (right inset). (B) Schematic overview of the sample acquisition and the quantitative proteomics pipeline. Vitreous humor samples were collected via vitrectomy, cleared by centrifugation, and after trypsin digestion subjected to LC−MS/MS analysis. The label-free quantification of the peptides and the corresponding proteins was done using the Progenesis LC−MS analysis software, and the proteins were identified with SEQUEST database search followed by bioinformatic analysis of the quantitative differences of non-PDR and PDR vitreous proteomes.
filtered to a maximum false discovery rate (FDR) of 0.05. For protein identification, two unique peptides were required.
(energy 35). Maximum FTMS fill time was set to 200 ms with Full AGC target 1 000 000, and the maximum fill time for the ion trap was 100 ms with the MSn AGC target of 5000. Only the precursor ions with more than 500 ion counts were allowed for MSn. Preview mode was used to enable the high resolution in FTMS scan. Charge state rejection was enabled (charge state 1 was rejected) as well as dynamic exclusion. For the selected ions, dynamic exclusion time was set to 30 s.
Spectral Counting
In the comparison of DR patient samples with nondiabetic macular hole and macular pucker patient samples, spectral counts for each protein in each sample were extracted from the SEQUEST database search results and used in relative quantification of protein abundance differences.
MS1 Quantification and Protein Identification
Data Processing
The Progenesis LC−MS software (v4.1, Nonlinear Dynamics Limited, Tyne, UK) was used to obtain the MS1 intensities of peptides for label-free quantification. Thermo.raw files were used as input, and the reference run was selected automatically. Retention time was limited to 10−140 min excluding first and last 10 min of the recorded data. Only peptides with charge states from +2 to +7 were allowed. For protein identification, the MS2-scan data acquired from Progenesis LC−MS were searched against the human component of the UniProtKBdatabase (release 2013_05, 25725 entries) using SEQUEST search engine in Proteome Discoverer software (version 1.4, Thermo Scientific). Carbamidomethylation (+57.021464 Da) of cysteine residues was used as static modification, and oxidation (+15.994491 Da) of methionine was used as dynamic modification. Precursor mass tolerance and fragment mass tolerance were set to less than 15 ppm and 0.8 Da, respectively. Maximum of two missed cleavages were allowed. Results were
Statistical significant tests of abundance changes between nonPDR and PDR eyes, and between bevacizumab treated and untreated patients, were conducted using Student’s t test. An abundance change with q-value of 0.01 or less was considered as significant change. Q-values were used instead of conventional p-values to maximize the power of statistical test. For example, a p-value of 0.01 indicates that 1% of all tests are likely to be false significant abundance changes. On the other hand, a q-value of 0.01 indicates that only 1% of significant tests are likely to be false significant abundance changes. The hierarchical clustering of the identified proteins was constructed using Gene Cluster 3.0. The line graph and boxplots were produced using R version 3.1.1. Gene Ontology (GO) annotations were obtained from DAVID bioinformatics resources.25,26 The cellular locations of identified proteins, being either intracellular, transmembrane, or extracellular, were extracted from Phobius predictor.27 The peptides from C
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
Figure 2. Biological processes, molecular functions, and cellular localization of the 2482 identified proteins. (A) The GO-BP and (B) the GO-MF terms for the proteins identified from the DR vitreous obtained via DAVID bioinformatics resources. (C) Predicted cellular localization of the detected proteins show majority of the proteins being extracellular or transmembrane.
samples from patients with prior surgical complications (such as cataract complications) and dublicate samples from the same patient, we had a sample set of 138 unique samples. Out of these 138 samples, 59 were obtained from non-PDR and 79 from PDR eyes. Ten non-PDR patients and five PDR patients had received intravitreal bevacizumab prior to vitrectomy. The workflow used for our quantitative proteomics analysis of non-PDR and PDR vitreous humor patient samples is shown in Figure 1, panel B. The vitreous samples obtained with vitreoretinal surgery were cleared from possible insoluble cellular fractions using centrifugation, and protein concentrations were quantified using BCA. The vitreous protein concentrations were 4.7 ± 1.2 μg/μL for non-PDR samples and 5.1 ± 1.8 μg/μL for PDR (Supplementary Table S1). The proteins were digested into peptides and analyzed using LC− MS/MS. The corresponding ion intensities for the MS1 features were obtained and used for label-free quantification with Progenesis LC−MS software followed by peptide and protein identification using the SEQUEST search engine31 via the Proteome Discoverer software. The median alignment percentage for the 199 analyses was 69.2% (Figure 1B, right middle inset). To further validate the robustness of our analysis pipeline, on feature alignment and detection level, we performed analysis of 29 samples in technical replicates. The distribution of the MS1 feature alignment showed extremely high reproducibility (96.4%), and MS1 quantitation level correlation was 99.9% (Supplementary Table S2).
transmembrane proteins were analyzed separately by QARIPtool28 to study whether they are accessible for the extracellular proteinases or if they are transmembrane or intracellular. For obtaining the known protein−protein interactions, the PINA2 protein interaction database was used.29 Dot Blot Analysis
Five micrograms of total protein from 23 non-PDR and 36 PDR samples was dot blotted to nitrocellulose membrane using Bio-Rad 96-well dot blot (DB) system (Bio-Dot Microfiltration Apparatus, Bio-Rad) according manufacturers instructions. Twenty antibodies against the differentially expressed proteins between non-PDR and PDR samples were selected based on availability of high-quality antibodies suitable for Western blotting (Supplementary Table S6A). Anti-mouse (NA931, GE Healthcare) or anti-rabbit (P0448, Dako) was used as secondary antibodies, and signals were visualized by chemiluminescence using Amersham ECL Western Blotting analysis system (GE Healthcare, UK). The DBs were analyzed and quantified using Dot Blot Analyzer for ImageJ.30 For each of the proteins, the Spearman’s correlation was calculated between the corresponding MS1 and DB intensities.
■
RESULTS
Diabetic Retinopathy Vitreous Proteome Analysis Pipeline
To identify the soluble vitreous humor proteome that contributes to retinal microvascular changes in the human diabetic eye, we collected and analyzed a total of 164 vitreous humor samples (Figure 1). Detailed patient demographics are shown in Supplementary Table S1. After the exclusion of
Diabetic Retinopathy Vitreous Humor Proteome
In total, we identified 10 097 different peptides from 2482 unique proteins (Supplementary Table S3A,B). Out of the 138 D
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
Figure 3. The non-PDR and PDR vitreous proteomes show clear separation. (A) Hierarchical clustering of the 123 non-PDR and PDR vitreous proteomes shown in blue and red, respectively. Note: The color gradient illustrates the relative protein abundance measured as MS1 intensities. (B) The fold change differences of the 238 significantly differing (q ≤ 0.01) between non-PDR and PDR sample groups. Several enriched protein families and groups are color coded (right inset), and proteins linked to specific cellular functions are highlighted (yellow star sign). Two-fold level is shown in red dashed line.
hierarchical clustering of the 123 (bevacizumab-groups excluded) vitreous samples based on their protein content. The quantitative proteomes (n = 1351 quantifiable proteins) of the non-PDR and the PDR samples differ and result in clear separation of the two disease states (Figure 3, panel A, nonPDR shown in blue, PDR in red). The abundances of 238 proteins differed significantly between non-PDR and PDR groups (q-value ≤0.01), with 230 proteins being up-regulated in the PDR samples (Figure 3B and Supplementary Table S5). To validate the quantitative MS results, we performed a highthroughput DB analysis for 20 proteins, for which high quality antibodies were available (Supplementary Figures S1−4 and Supplementary Table S6A,B). The correlation with the MS1 quantification was high (median 0.51, p-values ranging from 2.01 × 10−8 to 7.50 × 10−2). The majority of the proteins (n = 136) that were more abundant in the PDR group had higher than two-fold enrichment, with 25 proteins having four-fold or higher increase. The eight proteins that were significantly more abundant in non-PDR group had moderate fold changes ranging from 1.7−3.1. When comparing the enriched BP-GO terms of these 238 proteins with terms of the 2482 proteins from the whole data set, several terms linked to wound healing and inflammatory processes were notably enriched in the smaller subset (Supplementary Figure 5). Additionally, we noticed that the proteins up-regulated in the PDR samples interestingly belonged to a limited number of enriched protein families and groups (Figure 3B). These included 15 complement components and factors (C1QB, CFAB, CFAD, CFAH, CFAI,
samples, we were able to quantify in total 1351 proteins with 6865 nonconflicting peptides, respectively (Supplementary Tables S4A,B). To get an initial understanding of the diabetic vitreous proteome, the proteins from the 123 DR patients (not treated with bevacizumab) were categorized according to their involvement in different biological processes using the PANTHER GO classifications via DAVID bioinformatic database. The GO “Biological Process” and GO “Molecular Function” terms and corresponding frequencies (≥1%) are shown in Figure 2, panels A and B. To assess the cellular localization of the identified 2482 vitreous proteins, we used Phobius27 to assign the identified intravitreal proteins to either secreted/extracellular, transmembrane, or intracellular location. Out of the 2482 proteins, 1986 (80%) were extracellular, 482 (19%) transmembrane, and 14 (1%) intracellular (Figure 2C). Because of the large number of identified proteinases (n = 80; Supplementary Table S3A) in the vitreous samples, it could be reasoned that some of the transmembrane proteins were detected due to shedding of their ectodomains from the cell surface by these proteinases.32,33 To analyze this, we used QARIP28 and Phobius27 to assign the detected peptides from transmembrane proteins to be either extracellular or intracellular. According to this analysis, the majority of the peptides (69%) from the transmembrane proteins were on the cell surface and thus could potentially be released into the vitreous by proteolytic shedding. Quantitative Proteome Dramatically Differs between the Two DR Disease States
To assess the qualitative and quantitative differences between the non-PDR and the PDR vitreous proteome, we performed E
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
Figure 4. Complement and coagulation cascade is up-regulated in the PDR. (A) Pathway analysis of the 238 proteins changing between non-PDR and PDR groups showed clear pathway enrichment to complement and coagulation cascades. The 29 of the PDR up-regulated proteins are highlighted in red, eight proteins quantified but not differing are shown in yellow, and three proteins identified but not quantified are shown in gray. (B) Interaction analysis using the public protein−protein interactions of the 238 proteins reveals the interconnected network of 97 proteins with 161 interactions.
CO2, CO3, CO4B, CO5, CO7, CO8A, CO8B, CO8G, CO9, and FHR1); nine protease inhibitors of the serpin family
(AACT, A1AT, A2AP, ANGT, ANT3, CBG, HEP2, IC1, and KAIN); nine apolipoproteins (APOA1, APOA2, APOA4, F
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
Figure 5. Bevacizumab treatment down-regulates several proteins in the PDR but had no medium- or long-term effect on the non-PDR vitreous. Intravitreal bevacizumab treatment lowers the abundance of 72 proteins in PDR. Proteins are classified based on their cellular functions. The groups include cell adhesion, growth factor (insulin) signaling, cilia proteins, ATPases, apolipoproteins, apoptosis, transport, immune response, crystallin, and peptidase activity. Proteins that were detected being up-regulated in PDR are highlighted in red.
APOB, APOC1, APOC2, APOC3, APOH, and APOM); eight ciliary proteins (ABLM1, AKAP4, CFA52, CROCC, DYHC2, FAT4, GELS, and TT21B); six immunoglobulins (IGHA1, IGHA2, IGHG2, IGHG3, IGHM, and MUCB); six collagens and fibrinogens (CO4A2, CO6A3, CORA1, FIBA, FIBB, and FIBG); five protease inhibitors of the interalpha-trypsin family (AMBP, ITIH1, ITIH2, ITIH3, and ITHI4); and four alpha glycoproteins (A1AG1, A1AG2, A1BG, and FETUA). When the PDR proteomes were compared to vitreous proteomes of patients suffering from nondiabetic sight-threatening eye diseases such as idiopathic macular hole (MH) or pucker (P), a clear difference was detected. Compared with diabetic vitreous, several apolipoproteins, fibrinogens, complement system, immunoglobulins, interalpha-trypsin inhibitor heavy chain proteins, and Serpins could not be detected in vitreous of patients with MH or P (Supplementary Figure S6 and Supplementary Table S7).
found in the vitreous of patients with DME and suggested to play important role in the DME pathogenesis.43 Additionally, markers of oxidative stress and reactive oxygen species (PRDX2, PRDX6, CATA ROMO1) were significantly upregulated in PDR eyes, confirming their involvement in the progression of DR.44−46 Hypoxia, altered extracellular matrix (ECM) components, and increased concentrations of ROS are all commonly associated with tissue damage or malfunction.47 Several important factors and regulators of coagulation (FA5, PROS, plasma kallikrein (KLKB1), prothrombin (THRB), hyaluronan-binding protein 2 (HABP2), CBPB2, FINC, plasminogen (PLMN), and kininogen (KNG1) were also highly enriched in our PDR samples. Moreover, serum albumin (ALBU), nitric oxide synthase (NOS1), and hypoxia upregulated protein-1 (HYOU1) were detected with higher abundance in the PDR samples than in non-PDR samples (Figure 3B and Supplementary Table S5).
Potential Prognostic Markers of PDR
Complement and Coagulation Pathway Proteins are Up-Regulated in the PDR Vitreous
Out of the herein identified novel proteins upregulated in PDR, two are extracellular matrix proteoclycans lumican (LUM) and keratocan (KERA), which contribute to collagen fibrillogenesis, tissue hydration, wound healing, inflammation, and innate immunity.34−38 In innate immune responses, these proteins act together with CD14 and regulate Toll-like receptor (TLR) 4 signaling.38 Furthermore, LUM and KERA bind and recruit chemokines to the site of inflammation, which is essential for the migration of cells from the vascular system to the site of tissue damage.39 Developing therapeutic agents that block LUM/KERA from binding to the TLR receptors may provide a more targeted approach to minimize retinal inflammation compared with currently used corticosteroids. We also detected other proteins, which have been previously suggested as biomarkers for type-1 and type-2 diabetes. These include α-2-macroglobulin (A2MG) and vitamin D binding protein (VTDB). Several studies have shown that A2MG levels are increased in the blood of diabetic patients (type-1 and -2) with complications such as DR.40−42 Similarly, VTDB has been
The previous protein group and GO analyses of the PDR upregulated proteins suggested high enrichment of complement factors as well as regulators of coagulation, and using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, a clear “Complement and coagulation cascades” pathway enrichment was also detected (p-value 1.4 × 10−37). This pathway is known to consist of highly interconnected coagulation and complement cascades as well as the kininkallikrein system. Altogether, 29 proteins enriched in PDR samples belonged to these three cascades, covering 45% of the 63 cascade components (Figure 4A). First, 20 complement cascade components were quantified (Supplementary Table S4A), of which 15 (C1Q, CFAB, CFAD, CFAH, CFAI, CO2, CO3, CO4, CO5, CO7, CO8A, CO8B, CO8G, CO9, and ICI) were found to be up-regulated in PDR samples (Figure 4A and Supplementary Table S5). Second, we quantified 15 components of coagulation pathways (Supplementary Table S4A), of which 12 (A1AT, A2MG, ANT3, G
DOI: 10.1021/acs.jproteome.5b00900 J. Proteome Res. XXXX, XXX, XXX−XXX
Article
Journal of Proteome Research
proteins from three nondiabetic individuals using a selection of prefractionation methods including liquid phase isoelectrical focusing and SDS gel electrophoresis followed by analysis with UPLC coupled to LTQ Orbitrap XL MS.15 Murthy et al. identified 1205 proteins from 10 vitreous samples of various origin using different prefractionation methods, such as OFFGEL fractionation, and analyzed the fractions using LTQ Orbitrap Velos.48 Compared to the previous studies, our sample set was almost 10-fold larger, which together with the next generation orbitrap MS instrumentation allowed the detection of significantly more vitreous proteins than was identified in total prior to our study. Although the identification of several of the proteins might be associated with the DR status of the analyzed sample, we identified 562 overlapping proteins detected in the three abovementioned studies. In comparison, our analysis is also the first to provide and utilize the MS based quantitative information from the diabetic vitreous proteomes. The quantitative data gathered in this study are as unbiased as possible, containing no prefractionation or depletion steps. The identified 2482 vitreous proteins represent not only the proteome of pathological condition, but also an in-depth human vitreous protein atlas. This is highlighted in the identified GO-biological processes, since the majority of the most abundant terms describe processes required for normal tissue functions (Figure 2A,B). The majority of the detected 2482 vitreous proteins are extracellular or transmembrane, corresponding well with the fact that most of the proteins expected to be present in the human vitreous are secreted from the neighboring tissues or are part of the ECM (Figure 2C). Several of the secreted proteins are expected to be cytokines and growth factors, of which we detect 53 (Supplementary Table S3A). The detection of cell surface receptors is most likely a partial result from their ectodomain shedding by the large number (n = 80) of proteases present in the vitreous (Supplementary Table S3A).
CBPB2, FA5, FA12, FIBA, FIBB, FIBG, HEP2, PROS, and THRB) were enriched in PDR samples (Figure 4A and Supplementary Table S5). This finding shows a clear increase in coagulation components during the progression of the diabetic eye disease from non-PDR to PDR. Third, all of the five kinin-kallikrein system proteins (A2AP, FA12, KLKB1, KNG1, and PLMN) were enriched in PDR vitreous (Figure 4A, Supplementary Tables S4A and S5). Out of these, the kininogen (KNG1) is further digested into bradykinin and kallidin peptides. To extend our analysis beyond the annotated pathways, we derived the protein−protein interactions for the up-regulated proteins from the PINA2 database. This analysis resulted in a network of 97 proteins with 161 interactions (Figure 4B), highlighting the interconnectivity of complement and coagulation processes. Additional links from these processes were to oxidative stress and apoptosis linked proteins. Effect of Anti-VEGF (Bevacizumab)-Treatment on DR Vitreous Proteomes
̈ non-PRD and PDR samples, we In addition to anti-VEGF naive also analyzed 15 samples from diabetic eyes treated with intravitreal bevacizumab-injections prior to vitrectomy (10 from non-PDR and five from PDR patients). Several intravitreal proteins showed a clear down-regulation postbevacizumab treatment, the effect of the inhibitor being more profound in the PDR samples. When comparing the bevacizumab-treated and untreated PDR samples, 72 proteins showed downregulation with bevacizumab treatment (q-value
