Elucidation of the Pathogenic Mechanism of Rhegmatogenous Retinal Detachment with Proliferative Vitreoretinopathy by Proteomic Analysis

Abstract

Purpose.: To understand the molecular mechanisms of rhegmatogenous retinal detachment (RRD) with proliferative vitreoretinopathy (PVR), the vitreous proteome in RRD patients with severe PVR (grade C or D) was investigated.

Methods.: The analysis of the vitreous proteome in RRD patients with PVR (n = 24) and donor samples (n = 8) was analyzed by one-dimensional (1D) SDS-PAGE and reverse-phase liquid chromatography tandem mass spectrometry (RP-LC–MS/MS). The data were analyzed using GeneGO MetaCore software. The research followed the tenets of the Declaration of Helsinki for the use of human subjects.

Results.: In total, 516 and 364 proteins were identified in the vitreous of RRD patients with PVR and donor samples, including 48 overlapping proteins. In the PVR vitreous samples, the levels of extracellular (EC) proteins were increased and the levels of cytoskeleton proteins were decreased. In the pathologic process of PVR, inflammation was identified as an important GeneGo network. Furthermore, the complement and coagulation cascade was the essential pathway. Among the interaction network, the key node proteins in this network were p53 and transcription factor E2F1, respectively.

Conclusions.: 1D–SDS-PAGE coupled with RP-LC–MS/MS is a valuable resource to aid in the characterization of the proteome of RRD patients with PVR. Inflammation is the important pathologic process of PVR, while complement and coagulation cascade was the crucial pathway. p53 and E2F1 may be the new targets for successful treatment of RRD with PVR.

Introduction

Proliferative vitreoretinopathy (PVR) essentially is an anomalous wound-healing response in the eye. ¹ This condition develops in 5% to 11% of cases of rhegmatogenous retinal detachment (RRD). ² It is one of the most common causes of failure to correct RRD, and it complicates as many as one-third of all surgical repairs. ^2,3 The management of this condition is complicated further by the fact that PVR can result in detachment of otherwise successfully reattached retinas or even cause new breaks, necessitating additional corrective surgeries. The most recent comprehensive review suggests that despite significant advancements in vitreoretinal microsurgical techniques, the incidence of PVR in primary RRD has failed to decline over the past 20 years. ⁴ Although anatomic success with the current treatment modalities has been reported in 60% to 80% of patients, the functional prognosis is disappointing, with only 40% to 80% of these cases recovering ambulatory vision. ² The absence of more effective treatment options is attributable in part to an incomplete understanding of the pathogenesis of PVR. 

Proteomic analysis is a powerful approach to determine the coordinated changes in protein levels in tissues and cells. ⁵ Recent proteomic studies of human vitreous samples have revealed many proteins in patients with vitreoretinal disease. ^6–9 Shitama et al. reported that in RRD and PVR samples, in comparison to proliferative diabetic retinopathy (PDR) and non-PDR samples, significantly higher levels of pigment epithelium–derived factor (PEDF), clusterin, transthyretin, and cathepsin D were detected. ¹⁰ In our previous study, the vitreous proteome of patients with PVR was investigated by two-dimensional nano-liquid chromatography coupled with tandem mass spectrometry (2D nano–LC-MS/MS). ¹¹ In total, 129, 97, and 137 proteins were identified in the vitreous of donors, patients with moderate PVR (grade B), and patients with severe PVR (grade C or D). Since 2D nano–LC-MS/MS also had some limitations in investigation of proteome, one-dimensional (1D)–SDS-PAGE combined with reverse-phase liquid chromatography tandem mass spectrometry (RP-LC–MS/MS) was used in our current research. This approach generally is accepted as an effective method of proteomic analysis. ¹² For the purpose of understanding the pathogenesis of RRD patients with PVR, we used this high-performance proteomic approach to analyze the vitreous humor. 

Materials and Methods

Sample Preparation

Normal human eyes without any known ocular diseases (n = 8) that had been donated for corneal transplant in accordance with the Standardized Rules for Development and Applications of Organ Transplants were obtained from the Eye Bank of Shanghai in China. The mean postmortem time was 4.5 ± 2.7 hours (range 2.3–12.0 hours). A total of 24 PVR patients with RRD (n = 24, 12 eyes of grade C and D) was enrolled in the study, including 14 patients from Shanghai Tenth People's Hospital and 10 patients from Affiliated Hospital of Nantong University. The following enrolled and exclusion criteria were applied. Patients with ocular trauma, age-related macular degeneration, diabetes mellitus, a history of ocular surgery, and other systemic diseases were excluded. Normal serum samples were collected from 20 normal, healthy, age-matched volunteers. The research was conducted in accordance with the tenets of the Declaration of Helsinki for the use of human subjects. Informed consent was obtained from all patients after a verbal and written explanation of the nature and possible consequences of the study had been provided. The ethics committee of the Tongji University School of Medicine approved the research protocol. The demographic data for the patients who provided the vitreous and serum samples are shown in the Table. 

The undiluted vitreous humor samples from RRD patients with PVR (PVR group) were collected using a method similar to the research of Yu et al. ¹¹ The harvested samples of vitreous humor were centrifuged for 15 minutes at 12,000 revolutions per minute (rpm) to separate the cell contents, and they were stored at −80°C until use. The protein concentrations for the samples were measured using a bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL). 

1D–SDS-PAGE and In-Gel Sample Digestion

Cold acetone (including 10% trichloroacetic acid [TCA] and 20 mM dithiothreitol [DTT]) was added to the mixed vitreous humor samples at a ratio of 5:1 (vol/vol) overnight at −20°C. The samples were centrifuged for 10 minutes at 12,000 rpm and washed with 1 mL 90% acetone three times. The supernatant was removed and the pellets were dried at room temperature. To avoid individual differences, the vitreous samples from the same group were mixed in equal volumes before the experiment. In the donor and PVR groups, 100 μL of each sample were mixed. 

After the supernatant was dissolved in SDS-PAGE loading buffer, the samples were boiled for 10 minutes. Then, the samples were loaded in a single lane on a 1 mm thick 10% polyacrylamide gel. The gel was visualized by silver staining after separation. Thereafter, the gel was cut into 10 slices and then cut into 1 mm³ gel particles for in-gel digestion. In-gel digestion was performed as follows: the gel particles were washed 3 times with deionized water and subsequently dehydrated with 100% acetonitrile (ACN) for 10 minutes. The particles were incubated with 10 mM DTT in 25 mM ammonium bicarbonate for 1 hour at 56°C for protein reduction. The resulting free thiol (-SH) groups subsequently were alkylated by incubating the samples with 55 mM iodoacetamide in 25 mM ammonium bicarbonate for 45 minutes in the dark. The gels were washed with 25 mM ammonium bicarbonate and a 50% ACN solution, and dehydrated with 100% ACN sequentially. The gel pieces were rehydrated with 10 ng/μL modified trypsin (Roche, Mannheim, Germany) in 25 mM ammonium bicarbonate. The gel pieces were incubated for 12 hours at 37°C for protein digestion. The supernatants were transferred to fresh tubes, and the remaining peptides were extracted by incubating the gel pieces twice with 50% ACN in 5% trifluoroacetic acid (TFA), followed by dehydration with 100% ACN. The extracts were combined and lyophilized for dryness, and the resulting peptides were used for mass spectrometric analysis. 

Nano–LC-MS/MS

The following procedures were used to determine the protein profiles of vitreous samples according to the research of Yu et al., ¹¹ with minor modification. The tryptic digests then were loaded onto a reverse phase trap column for enrichment at a flow rate of 10 μL/min. The trap column was connected sequentially in-line with an analytical C18 column, and the peptide mixtures were eluted into the QSTAR XL LC/MS/MS system (AB SCIEX, Framingham, MA) at a flow rate of 200 nL/min. The Agilent 1200 nanoLC system (Agilent Technologies, Santa Clara, CA) was used. A spray voltage of 2500 V was applied to a Tip nanospray emitter (New Objective, Woburn, MA) connected at the end of the analytical column. The MS/MS spectra were recorded using information-dependent acquisition and duty-cycle enhancement. 

Data Base Search and Analysis

The MS/MS spectra were searched against the human protein databases using MASCOT (available online at http://www.matrixscience.com) as reported previously. ¹¹ The search parameters were as follows: MS/MS spectra with a mass tolerance of 0.5 Da; modifications of methylation and oxidation were permitted. The MASCOT searching results were combined and filtered further as follows: (1) short matched peptides (five amino acids) were removed, (2) peptide assignments that were not the first rank in the matching list were removed, (3) peptides with a MASCOT score <15 were removed. 

Gene Ontology (GO) Enrichment/Depletion Analysis

To prepare an overview of our proteomic analysis of RRD with PVR, we categorized the specific proteins based on their GO assignments. GeneGo Metacore (Version 6.6; GeneGo, Carlsbad, CA) was used for the enrichment workflow analysis. For the enrichment/depletion analysis, a test dataset comprised of the identified proteins and a reference set of annotated proteins from the complete human proteome were needed. According to the instructions on the GOfact webpage, the custom GO annotation for the reference set (of whole IPI human dataset) was created by extracting the GO annotations with GOA (available online at http://www.ebi.ac.uk/GOA) according to their IPI IDs. The analysis was performed using R scripts with the hypergeometric test and FDR correction; the GO terms with P < 0.05 or P < 0.01 were selected as enriched/depleted or significantly enriched/depleted. The cellular component (CC), molecular function (MF), and biologic process (BP) of the selected proteins were annotated using the GO database. In our study, significant upregulation was defined as a peptide ratio of 3 or more and a peptide distance of 5 or more; significant down-regulation was defined as a peptide ratio of 0.5 or less and a peptide distance of −3 or less. Peptide ratio = PVR peptide hits/donor peptide hits, and peptide distance = PVR peptides hits − donor peptides hits. 

Pathway Analysis

ArrayTrack software (available online at http://www.fda.gov/ScienceResearch/BioinformaticsTools/Arraytrack/default.htm) was used for pathway analysis. ArrayTrack offers a simple query interface to retrieve information about human protein expression profiles and provides direct connections to related biologic pathways available from the Kyoto Encyclopedia of Genes and Genomes (KEGG). Based on the ArrayTrack manual, the IPI names of the differentially expressed proteins were converted to SWISS-PROT names using the ID convert tool before pathway analysis, and then entered into the Pathway Search panel. For the statistical analysis, a P value for pathway enrichment was obtained using Fisher's exact test, and a P < 0.05 was considered statistically significant. 

Statistical Analysis

Student t-tests, χ², and Fisher's exact tests were performed in the current study (SPSS 11.5; SPSS, Inc., Chicago, IL). A P value < 0.05 was considered statistically significant. 

Results

The Integrated Proteome

A total of 880 unique proteins derived from 2215 peptides was identified unambiguously by 2D nano–LC-MS/MS in donor and RRD with PVR vitreous samples. More proteins were detected in RRD with PVR (516 proteins from 409 genes) vitreous humor than in donor samples (364 proteins from 293 genes). Only 48 proteins were shared between the donor and RRD with PVR vitreous proteome (Fig. 1A). The molecular weight (MW) and isoelectric point (PI) are shown in Figures 1B and 1C, respectively. The gene database information for the PVR and donor samples is shown in Supplementary Table 1 (see Supplementary Material). 

Figure 1. 

 

Integrated analysis of proteome data from donor and PVR vitreous samples. (A) The diagram shows the proteins identified from the donor and PVR samples. The protein numbers are shown within the circles. (B) MW. (C) PI.

Comparison of the PVR and Donor Vitreous Proteomes

Eleven proteins were significantly upregulated in the RRD with PVR vitreous samples compared to the donor samples. The majority of these proteins were serum proteins, such as hemopexin, alpha2-HS–glycoprotein, alpha1B-glycoprotein, members of the serpin family, and complement components (Fig. 2A). However, many of the cytoskeleton (CK) proteins that were identified in the donor samples, such as actin family members and opticin precursors, were significantly down-regulated in the RRD with PVR samples (Fig. 2B). Furthermore, other CK proteins, such as tubulin, were undetected in the RRD with PVR vitreous humor. 

Figure 2. 

 

Comparison of the PVR and donor proteomes. (A) Eleven proteins were significantly upregulated in PVR compared to donor samples. (B) Five proteins were significantly down-regulated in PVR compared to donor samples.

GO Analysis

Based on the terms represented in the GO database, the differentially expressed proteins were divided into 3 categories: CC, MF, and BP. The top 10 CC, BP, and MF from the RRD with PVR samples are shown in Figure 3. The top 10 CC, BP, and MF from the donor samples are shown in Supplementary Figure 1 (see Supplementary Material). The top CCs in the RRD with PVR samples were extracellular components, whereas in the donor samples, the top CCs were cytoskeletal components. The top MF in the RRD with PVR samples was peptidase regulator activity, whereas in donor samples, the top MFs were cytoskeleton construction and metabolism. The top BPs in the RRD with PVR samples were detection of endogenous stimuli, protein activation cascade, and negative regulation of hydrolase activity, while in the donor samples the top BPs were protein polymerization, synaptic vesicle priming, and microtubule-based movement. 

Figure 3. 

 

The top 10 significant cellular components (left), biological processes (middle), and molecular functions (right) in the PVR samples. Bars represent the inverse log of the P value.

GeneGo Process Networks

The GeneGo process networks in the RRD with PVR samples were ranked in terms of the enrichment of the differentially expressed proteins (P value), and the top five networks were inflammation (kallikrein-kinin system [KKS]), inflammation (complement system), cell adhesion (integrin priming), inflammation (innate), and inflammatory response (Fig. 4A). The predominant network processes in the donor samples were cytoskeleton (regulation of cytoskeleton), inflammation (complement system), development (neurogenesis, axonal guidance), cytoskeleton (intermediate filaments), and cell adhesion (integrin-mediated cell-matrix adhesion, Fig. 4B). 

Figure 4. 

 

The top 10 of GeneGo process networks in PVR (A) and donor (B) samples. Bars represent the inverse log of the P value.

KEGG Pathways

In donor vitreous, the significant KEGG pathways were related to cytoskeleton regulation (pathogenic Escherichia coli infection, gap junction, phagosome) and metabolism (glycolysis/gluconeogenesis and the pentose phosphate pathway), while in RRD with PVR vitreous, the most significant pathways were the complement and coagulation cascades (hsa04610, P < 0.01), including coagulation cascade, KKS, and complement cascade. The KEGG pathways in the RRD with PVR and donor vitreous samples are listed Supplementary Table 2 (see Supplementary Material). Although the complement and coagulation cascade pathway also was identified in the donor vitreous samples, 7 proteins from this pathway were not detected in the RRD with PVR vitreous samples (A2 M, KNG1, CFH, CFI, F2, FGA, SERPIND1, Fig. 5). 

Figure 5. 

 

The complement and coagulation cascade pathways in PVR and donor samples. The coagulation cascade, KKS, and complement cascade were included. The proteins in blue were detected in PVR and donor samples, and the proteins in red were unique proteins in PVR. KNG, kininogen; SERPIN, serpin peptidase inhibitor; BF, complement factor B; HF, complement factor H; IF, complement factor I; A2M, alpha-2-macroglobulin; F2, thrombin; FG, fibrinogen alpha chain.

Interaction Networks

The proteins of the PVR proteome exist in interaction networks. The direct interaction network is shown in Figure 6. The top 6 key node proteins in this network were p53, E2F1, Thrombin, MMP-12, PKC-epsilon, and N-CoR. 

Figure 6. 

 

The direct interaction network in PVR vitreous proteome. p53, 37 edges; E2F1, 24 edges; Thrombin, 9 edges; MMP-12, 7 edges; PKC-epsilon, 7 edges; N-CoR, 7 edges.

Discussion

In our study, a systematic analysis to characterize the proteomes of RRD with PVR and normal vitreous was performed. There were 880 distinct proteins identified through 2D-nano–LC-MS/MS in the integrated donor and RRD with PVR proteome, which is significantly more than in previous reports of the vitreous proteome, and previous studies have been conducted using 2DE combined with mass spectrometry (MS). ^5,13,14 Furthermore, the number of proteins detected also was larger than in our own previous study. ¹¹ More proteins were detected in the RRD with PVR samples (516) than in the donor samples (364). This suggested that RRD with PVR is a complicated process involving the production of a significant number of new proteins in the vitreous humor. According to the GO database, the top CC, MF, and BP in the PVR samples were the extracellular (EC) component, peptidase regulator activity and protein activation cascade, respectively. This suggested that during the RRD with PVR process, the increase in the expression of EC proteins may be involved in transcriptional regulation and protein activation. There were 11 serum proteins that were significantly upregulated in the RRD with PVR samples, including hemopexin, alpha2-HS–glycoprotein, alpha1B-glycoprotein, members of the serpin family, and complement components, corresponding to the destruction of the blood–retina barrier. Many of the CK proteins identified in the donor samples, such as actin family members and opticin precursor, were significantly down-regulated, and other CK proteins, such as tubulin, were undetected in PVR vitreous humor. This finding indicated that in the RRD with PVR process, the normal CK proteins are destroyed. Therefore, our results suggested that normal CK proteins were down-regulated while the expression of extracellular proteins significantly increased in RRD with PVR vitreous. 

The top GeneGo process networks in the RRD with PVR samples were inflammation and cell adhesion. The most significant KEGG pathways in RRD with PVR were complement and coagulation cascades, as well as in the donor samples. However, 7 proteins were detected in the RRD with PVR samples that were not present in the donor samples. In the complement and coagulation cascades, the proteins undergo a series of cascade responses resulting in cell lysis, cell adhesion, migration, proliferation, fibrin degradation, and inflammation. Our results indicated that the complement and coagulation cascade pathways have important roles in the pathologic process of RRD with PVR. A previous study reported that activation of the complement system is involved in PVR. ¹⁵ The complement system is a principal constituent of humoral immune reactions, and is involved in opsonization, inflammation, lysis, and immune complex clearance. Our study found that high-abundance proteins, such as C4A and C4B, were upregulated significantly. In addition, complement factors B, H, and I also were detected in the RRD with PVR vitreous humor, which supported the study of Grisanti et al. ¹⁶ Cashman et al. found that mice injected with C3-expressing adenovirus exhibited significantly increased vascular permeability, endothelial cell proliferation and migration, RPE atrophy, loss of photoreceptor outer segments, reactive gliosis, and retinal detachment. ¹⁷  

In our study, high molecular weight kininogen (HK) was detected in the vitreous of RRD with PVR, which supported our previous study. HK is one of the important proteins of the complement and coagulation cascade. Kallikreins, which activate HK to HKa, can be regarded as anti-inflammatory and anti-oxidative agents that protect the brain against ischemic stroke-induced injuries. ¹⁸ It is worth noting that several human kallikreins can be used as biomarker in serum to aid in the diagnosis and monitoring of some cancers, such as ovarian, prostatic, and breast cancer. ^19,20 Chao et al. reported that kallikrein/kinin may serve as new drug targets for the prevention and treatment of heart failure, renal disease, and stroke in humans. ²¹ Therefore, kallikrein/kinin and anticomplement therapy may provide new treatment options for the prevention and treatment of PVR. 

An interesting finding in the study was that the interaction networks existed among the proteins of the PVR proteome. Among them, the key node proteins in this network were p53 (37 edges) and E2F1 (24 edges), which indicated that the 2 proteins had important roles in the disease. p53 is one of the tumor suppressor genes to be discovered among the human genome. It has a wide range of functions covering cell cycle control, DNA repair, apoptosis, genome integrity maintenance, metabolism, fertility, cellular reprogramming, and autophagy. While many studies focused on the importance of individual post-translational modifications, further explorations indicated a new layer of p53 coordination through the interplay of the modifications, which built up a complex “network.” ^22,23 In the absence of Sonic hedgehog, p53 induces apoptosis and inhibits retinal cell proliferation, cell-cycle exit, and differentiation in zebrafish. ²⁴ Many researchers had documented that p53 was involved in some cancers, such as breast cancer, ²⁵ head and neck squamous cell carcinoma, ²⁶ and some degenerative diseases, such as Parkinson's disease (PD) ²⁷ and Alzheimer's disease (AD). ²⁸ Some studies documented that p53 was related closely to the apoptosis of RPE cells. ^29,30 p53 also has a novel antioxidant function to the retinal ganglion (RG) cells. ³¹ Since RPE and RG cells were the essential cells of PVR, these studies indicated that p53 may be involved in the PVR process, which supported our current research. The activating E2f transcription factors (E2f1, E2f2, and E2f3) induce transcription and are viewed widely as essential positive cell cycle regulators. The E2F1 transcription factor is post translationally modified and stabilized in response to various forms of DNA damage to regulate the expression of cell cycle and pro-apoptotic genes. E2F1 also forms foci at DNA double-strand breaks (DSBs). Chen et al. found the new roles for E2F1 in the DNA damage response, which may contribute directly to DNA repair and genome maintenance. ³² E2F also is involved in the normal progression of the tumor suppressor gene, retinoblastoma gene (RB1). In the absence of normal RB1, genomic instability and chromosomal aberrations accumulate, leading to tumor initiation, progression, and ultimately metastasis. ³³ E2F1 deletion leads to increased mitochondrial number and function, increased body temperature in response to cold, and increased resistance to fatigue with exercise. Interestingly, we observed increased E2F1 protein levels in Duchenne muscular dystrophy (DMD) patients, suggesting that E2F1 might represent a promising target for the treatment of DMD. ³⁴ Indeed, the “cancer cell cycle” in Rb1 null cells is E2f-dependent. Absence of activating E2fs in flies or mammalian fibroblasts causes cell cycle arrest, but this block is alleviated by removing repressive E2f or the tumor suppressor p53, respectively. ³⁵ Future research will focus on the metabolisms of p53 and E2F1 in the process of RRD with PVR. 

However, there were some limitations in our study. The patients with RRD but without PVR were suitable for the control group. There were some reasons that could explain why they were not selected as the control group. Firstly, the patients with RRD but without PVR were few in our hospital. These patients could undergo the surgery in the primary hospital. Secondly, in the early stage, the acuity vision of the patients with RRD but without PVR was not decreased too low to be found in the Chinese people. The first time that the patients went to see a doctor was approximately 2 weeks later after the onset. At the moment, the majority of the patients with RRD appeared to have PVR. Thirdly, in China, most patients with RRD but without PVR underwent scleral buckling surgery instead of vitrectomy. The vitreous could not be obtained from these patients. Therefore, it was impossible to choose the patients with RRD but without PVR as the control group. 

Table. 

 

The Patient Demographics of the Vitreous and Serum Samples (mean ± SD)

Group	N	Age, y*	Sex (Male/Female)†
PVR C	12	60.0 ± 6.3	6/6
PVR D	12	59.1 ± 10.0	8/4
Donor eyes	8	58.2 ± 8.3	6/2
Normal group	20	56.2 ± 4.3	12/8
F		0.76	1.65
P		0.525	0.648

*  ANOVA.

†   χ ² test.

Supplementary Materials

pdf S1 

pdf S2 

xls S3 

Acknowledgments

We thank Hong Yu of Encode Genomics Co., Ltd., for his efforts in analysis of data using GeneGo MetacoreTM (Version 6.5), which was authorized by Bioinformatics Center, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. 

Footnotes

 Supported in whole or in part by National Nature Science Foundation Project (30901643), Phosphor Project (09QA1405100), Shanghai Science Committee Biology Department Pilot Project (10411964900), and The New Excellence Project of Shanghai Health Bureau (XYQ2011067).

Footnotes

²  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

Footnotes

 Disclosure: J. Yu, None; R. Peng, None; H. Chen, None; C. Cui, None; J. Ba, None