Vitreous humor proteome: unraveling the molecular mechanisms underlying proliferative and neovascular vitreoretinal diseases

Abstract

Proliferative diabetic retinopathy (PDR), proliferative vitreoretinopathy (PVR), and neovascular age-related macular degeneration (nAMD) are among the leading causes of blindness. Due to the multifactorial nature of these vitreoretinal diseases, omics approaches are essential for a deeper understanding of the pathophysiologic processes underlying the evolution to a proliferative or neovascular etiology, in which patients suffer from an abrupt loss of vision. For many years, it was thought that the function of the vitreous was merely structural, supporting and protecting the surrounding ocular tissues. Proteomics studies proved that vitreous is more complex and biologically active than initially thought, and its changes reflect the physiological and pathological state of the eye. The vitreous is the scenario of a complex interplay between inflammation, fibrosis, oxidative stress, neurodegeneration, and extracellular matrix remodeling. Vitreous proteome not only reflects the pathological events that occur in the retina, but the changes in the vitreous itself play a central role in the onset and progression of vitreoretinal diseases. Therefore, this review offers an overview of the studies on the vitreous proteome that could help to elucidate some of the pathological mechanisms underlying proliferative and/or neovascular vitreoretinal diseases and to find new potential pharmaceutical targets.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-022-04670-y.

Background: translational research in proliferative vitreoretinal diseases

Visual impairment and blindness severely affect the quality of life of the patients and are a significant burden for healthcare systems [1, 2]. Despite improvements achieved in the prevention and control of ocular diseases, the number of people with moderate-to-severe visual impairment is expected to increase due to the growth and aging of the world population [3, 4]. Diabetic retinopathy (DR) and age-related macular degeneration (AMD) are leading causes of visual impairment and blindness in middle-income and industrialized countries [5, 6]. Proliferative vitreoretinopathy (PVR) represents the most common cause of failure in retinal detachment (RD) surgery, accounting for 3.9–13.7% of all rhegmatogenous retinal detachment (RRD) cases [7, 8]. Despite the multiple treatments available, there is still progression to visual impairment and blindness in proliferative etiology [9], which reinforces the need for designing better strategies for the management of these diseases [10]. Considering their multifactorial nature, omics approaches provide a deeper understanding of the pathophysiological processes underlying these diseases and their evolution to a proliferative or neovascular etiology, in which patients experience significant vision loss. Ocular proteomics has emerged as a tool for discovering new biomarkers, which could help to unveil the pathophysiologic mechanisms underlying many ocular diseases, anticipate their progression, and predict the response to therapy [9, 11, 12].

Vitreous humor proteomics

Vitreous, also termed vitreous body or vitreous humor, is a translucent gel-like substance that fills the posterior cavity of the eye, surrounded by the retina, pars plana, and lens [13–15]. It is an avascular and virtually acellular connective tissue, composed mainly of water and a network of collagen fibrils surrounded by glycosaminoglycans, electrolytes, and soluble proteins [14, 16]. For many years, it was thought that vitreous function was merely structural, allowing to support and protect the surrounding ocular tissues from physical impact [14, 17, 18]. Nowadays, it is known that the vitreous has many other physiological functions. In addition to allowing the passage of light toward the retina, the vitreous contributes to eye growth and shape, serves as a barrier to biomolecules and cells, allows the repository and diffusion of the substances involved in the eye metabolism, and regulates oxygen levels within the eye [14, 18, 19]. In the last years, there has been a growing interest in the role of the vitreous proteome in different pathophysiological contexts, which is evidenced by the increase in the number of identified proteins from 545 to over 6538 in only 5 years [9, 20]. Vitreous proved to be extremely attractive from biological and analytical perspectives. The proteome and biochemical properties of the vitreous reflect the physiological and pathological conditions of the eye due to its close contact with the inner retina, lens, and ciliary body [11, 21]. Furthermore, aging-related changes in the vitreous, such as vitreous liquefaction and posterior vitreous detachment (PVD), are presumed to be underlying several retinal diseases [17, 19, 22]. Therefore, the outcome of vitreous proteome studies could help to reveal some of the pathological mechanisms underlying proliferative retinal diseases. This review aims to provide an overview of the reported changes in vitreous proteome in patients with PDR, neovascular AMD (nAMD), and PVR, and how these could be correlated with pathological events, such as inflammation, fibrosis, oxidative stress, neurodegeneration, and vitreous remodeling.

Characterization of the vitreous proteome in vitreoretinal diseases

Diabetic retinopathy

According to World Health Organization (WHO), about 146 million (34.6%) of 422 million adults with Diabetes Mellitus suffer from some form of DR [23], accounting for 1.1% of blindness among middle-aged adults [24]. The public health burden associated with DR will increase as a result of the substantial expansion of diabetes in low- and middle-income countries [25, 26]. The duration and type of diabetes, poor glycemic control, hypertension, and high levels of glycated hemoglobin A1c and triglycerides are the main risk factors for DR [27, 28]. The pathophysiology and treatment of DR, which have been reviewed by several authors [26, 29–32], are summarized in Fig. 1. In an early stage, microvascular changes occur in response to hyperglycemia, but increasing evidence suggests that neuroglial degeneration may precede microvascular changes [26, 29, 30, 33, 34]. As the severity of DR progresses, capillary nonperfusion and occlusion lead to retinal ischemia, which combined with an imbalance in the levels of inflammatory and pro-angiogenic cytokines can trigger intra-retinal and intravitreal neovascularization (NV) and, eventually, diabetic macular edema (DME) [26, 32, 33]. The association of NV, retinal gliosis, and fibrosis favors the formation and growth of fibrovascular proliferative tissue on the retinal surface. Tractional forces at the vitreoretinal interface due to the contraction of fibrovascular tissue can lead to tractional RD and, ultimately, vitreous hemorrhage [33, 35]. Furthermore, vitreoretinal structural and molecular changes are underlying the progression to PDR, since the production of advanced glycation-end products (AGEs) associated with hyperglycemia, among other processes, triggers abnormal crosslinks between collagen fibrils, causing the destabilization of vitreous structure [36–38].

Fig. 1.

Vitreous and retinal anatomy in pathophysiological events in diabetic retinopathy and proliferative diabetic retinopathy and current treatments for each clinical feature. RGC retinal ganglion cell

The characterization of the proteome of vitreous humor has contributed extensively to identify pathways involved in DR/PDR [37, 39–41] and DME [42], as well as potential disease biomarkers. Proteins that have been associated with these diseases using proteomic and multiplex ELISA approaches are listed in Supplementary Table 1.1. Overall, the first proteomic analyses, using two-dimensional electrophoresis (2DE) coupled with mass spectrometry, reported increased levels of plasma proteins, including apolipoproteins complement and coagulation proteins, and acute-phase proteins in DR [43–46] and PDR [44, 46–49] compared to non-diabetic controls, as well in DME compared to the non-DME group [50, 51]. Levels of pigment epithelium-derived factor (PEDF) and gamma-enolase (ENO2) were inconsistent among studies comparing PDR vitreous with non-diabetic controls [44, 46–49, 52]. In the case of DME, PEDF was consistently reported as augmented in proteomics studies [50, 53, 54], suggesting a protective effect on DME due to its role as an anti-angiogenic/neurotrophic factor [42]. Nevertheless, other authors reported lower levels of PEDF in DME [55, 56], implying that its protective effect depends on the pathological phases of DR [54]. Clusterin (CLU), inter-alpha-trypsin inhibitor heavy chain H2, and several retinol-binding proteins and crystallins were reported as underexpressed in PDR [48, 49, 52], but in DME, only CLU and crystallin S were found diminished [51]. Although these studies have provided some potential biomarkers of the progression to PDR, the complexity and wide dynamic range of human vitreous represent a challenge for quantitative analysis by 2DE [57, 58]. To overcome some of the limitations of gel-based techniques, proteomics techniques such as liquid chromatography coupled to mass spectrometry (LC–MS) and capillary electrophoresis coupled to mass spectrometry (CE–MS) have been used in recent years. In addition to the aforementioned proteins that were implicated in DR/PDR pathogenesis [53, 54, 59–66], these studies contributed to the identification of new potential protein biomarkers.

Besides contributing to the understanding of the role of carbonic anhydrase (CA) in the activation of the kallikrein–kinin system (KKS), Gao et al. reported lower levels of extracellular superoxide dismutase, neuroserpin, cell adhesion molecules (e.g., calsyntenin-1), and amyloid-forming proteins (e.g., amyloid-beta A4 protein [APP]) in PDR compared with diabetic patients [53, 54]. Kita et al. also investigated the role of the KKS in DME. Since no correlation was found between the levels of plasma kallikrein and vascular endothelial growth factor (VEGF), they demonstrated that KKS induces retinal thickening and vascular permeability in DM through a VEGF-independent mechanism. However, it was observed a positive correlation between kallikrein levels and several plasma proteins (e.g., complement components and acute-phase proteins) in DME, suggesting that KKS activation in the vitreous may result from plasma proteins influx after blood–retinal barrier (BRB) dysfunction [67]. Balaiya et al. also associated the levels of complement components and coagulation systems with KKS in PDR, whereas anti-angiogenic proteins, immune modulators, acute-phase proteins, proteases, and extracellular matrix (ECM) components were only detected in patients with epiretinal membranes (ERM) or macular hole [62].

Wang et al. identified 96 differentially expressed proteins in PDR vitreous compared to healthy donor samples, many of them related to the glycolytic process, visual perception, and hypoxia-inducible factor (HIF-1) pathways [60]. In another study, Gardner and Sundstrom proposed a new personalized medicine method for the prevention and treatment of early DR based on proteomics approaches. Proteins related to NRF2-mediated oxidative stress response and cell proliferation were found up-regulated in PDR vitreous, which suggests the activation of these processes. On the other hand, HIPPO signaling pathway and central nervous system development were predicted as inhibited [65]. In a more recent study, Schori et al. described increased levels of proteins involved in the regulation of cell adhesion, and ECM organization, including metalloproteinases (MMPs) and inhibitors, and proteoglycans in PDR vitreous. In turn, molecules associated with lysosomal activity/autophagy and biological processes in the nervous system, including developmental processes, ECM organization, and cell adhesion, were found downregulated. Interestingly, the reduction in the levels of some of these proteins, including APP, neuroserpin, acid ceramidase, and ceroid-lipofuscinosis neuronal protein 5, has been related to neurodegeneration [63].

Using a label-free quantitative method, Loukovaara et al. compared vitreous from patients with PDR, including those treated with bevacizumab. Overall, this study reported 1351 quantified vitreous proteins, describing higher levels of inflammatory mediators, cell adhesion molecules, oxidative stress markers, and ECM proteoglycans. In addition, the downregulation of 72 proteins was found after the administration of bevacizumab, including apolipoproteins, crystallins, immunoglobulins, insulin-like growth factor-binding proteins (IGFBPs), and proteins involved in cell adhesion and apoptosis [64]. More recently, Zou et al. studied the changes induced by ranibizumab in the vitreous proteome of patients with PDR and found that platelet degranulation and integrin cell surface interaction pathways are severely affected. Many complement and coagulation proteins, crystallins, and immunoglobulins were not detected in the vitreous after treatment. By contrast, metalloproteinase inhibitor 2 (TIMP2) was found up-regulated in response to ranibizumab, as well as several glycolytic proteins and antioxidant proteins (catalase, peroxiredoxin) [68]. Alternatively to proteomic assays, multiplex ELISA assays were used for the quantification of proteins like VEGF, whose intravitreal concentration varies in the range of picograms [69–71]. VEGF has been established as one of the key molecules in DR pathogenesis, and its levels in vitreous and plasma are correlated with the progression of PDR [72] and clinical features such as macular volume and central retinal thickness [73, 74]. Therefore, it is not surprising that anti-VEGF therapy was established as the first-line therapy for DME and PDR [75, 76]. Although anti-VEGF therapy displays some contraindications and requires further optimization to improve its long-term effectiveness, drugs, such as aflibercept, ranibizumab, and bevacizumab, have been shown to promote the regression of NV in DR patients [77–79] (reviewed in Refs. [75, 76]). Several studies found a positive correlation between VEGF intravitreal levels in PDR and levels of vascular cell adhesion protein-1 (VCAM-1) [80], alpha-crystallin B chain [81], MMPs [82, 83], oxidative stress markers [84–86], and proteins related to the renin-angiotensin system [82, 87]. Angiopoietins 1 and 2 (Ang-1 and Ang-2) were also implicated in the modulation of later stages of angiogenesis in DR/PDR [88]. They exert their effect by binding to a specific tyrosine kinase receptor (Tie-2), with Ang-1 acting as an activator, maintaining cell survival and vascular stability, and Ang-2 as an antagonist, promoting abnormal angiogenesis (reviewed in Refs. [88, 89]). Considering that, faricimab, a new drug that targets both Tie-2/Ang-2 and VEGF pathways, was recently approved for the treatment of DME, showing gains in visual acuity statistically superior to ranibizumab in treatment-naïve patients [90, 91]. Several authors reported higher levels of Ang-2 in the vitreous of patients with PDR [70, 92, 93]. Klaassen et al. found that the levels of Ang-2 and other pro-angiogenic mediators, such as PDGF and PlGF, were strongly correlated to the degree of fibrosis in PDR, whereas Ang-1 shows a strong correlation with NV [70]. The concentration of Ang-2 was twice that of Ang-1 in non-proliferative DR with DME, whereas only Ang-1 was increased in PDR [94]. Loukovaara et al. observed a significant correlation between Ang-2 levels and matrix metalloproteinase-9 (MMP9), VEGF, erythropoietin, and transforming growth factor (TGF-β) levels in vitreous from DR patients. Although Ang-1 were also increased, the Ang-1/Ang-2 ratio was found to be decreased, which is considered a critical control for a switch in inflammatory processes [95]. Apart from VEGF and Ang, other regulators of angiogenesis were found up-regulated in PDR vitreous compared to non-diabetic controls, including IGFBPs [70, 96], placental growth factor (PlGF) [70, 97], platelet-derived growth factor (PDGF) [70, 98–100], hepatocyte growth factor [70, 99, 101], and soluble VEGF receptor 1 (VEGFR-1) [93].

Other studies pointed to the role of chronic neuroinflammation in PDR through the quantitation of inflammatory mediators, adhesion molecules, and neurotrophic factors. The measurement of pro-inflammatory mediators could be relevant to assessing the use of alternatives (e.g., corticosteroids) to the standard treatment, especially in refractory DME and unresponsive cases to anti-VEGF therapy [102–104]. On the other hand, the measurement of neurotrophins opens new doors for the design of new therapies based on neuroprotection (see “Mechanisms of neurodegeneration and neuroprotection” section). Klaassen et al. found higher levels of intercellular adhesion molecule-1 (ICAM-1), tumor necrosis factor (TNF-α) receptors, and neurotrophic factor neuregulin-1 in PDR compared with non-diabetic patients with vitreous debris [70]. Other studies reported the up-regulation of cytokines in PDR vitreous, including interferon-gamma (INF-γ) [71], TNF-α [71, 99], granulocyte Colony-Stimulating Factor (CSF3) [71], interleukins 1 beta (IL1-β), 6 (IL-6), and 8 (IL-8) [69, 71], chemokines [e.g., C-X-C motif chemokine 10 (CXCL10)] [71], and adhesion molecules (e.g. ICAM-1, VCAM-1) [69, 80]. Downregulation of neurotrophins/anti-angiogenic molecules such as PEDF [93, 101] or nerve growth factor (NGF) [105] in PDR has been described, whereas levels of TNF-α, IL-8, neurotrophin-3 (NT-3), NGF, glial cell-derived neurotrophic factor (GDNF), and ciliary neurotrophic factor (CNTF) were higher in eyes with non-proliferative DR compared with active PDR [106]. By measuring the levels of inflammatory mediators at distinct stages of DR, Kovacs et al. found that IL-8 levels gradually increase with the progression of the disease to PDR [99]. Koskela et al. observed higher levels of IL-6 and IL-8, but not of adhesion molecules, in the vitreous compared to the plasma of patients with PDR, which suggests an active local production of these interleukins at the retinal level. Furthermore, they found higher levels of interleukin-10 (IL-10) and several adhesion molecules (ICAM-1, platelet endothelial cell adhesion molecule [PECAM-1], E-selectin, and VCAM-1) in PDR compared with controls [69]. Yoshida et al. detected significantly higher levels of monocyte chemoattractant protein-1 (MCP-1) and IL-6, but not IL-8, in patients with PDR after vitrectomy, suggesting a possible association between inflammation and post-operative macular edema. Also, levels of MCP-1 were higher in the vitreous of patients with DME compared to those without DME [107]. Deuchler et al. found a linear correlation between DR grade and the expression of ICAM-1, IL-8, and PlGF and between the average retinal thickness and the levels of ICAM-1 and IL-8. PDR has been associated with higher levels of ICAM-1, IL-8, and PlGF, whereas the levels of ICAM-1, interferon-gamma-inducible protein, PlGF, VEGF, and interleukins (IL-6 and IL-8) were higher in DME patients with PDR than those with non-proliferative DR [108]. More recently, Kuo et al. performed a systematic literature review to find reliable inflammatory biomarkers in DR and relate them to the disease progression. The activation of the inflammasome pathway as DR progresses is suggested by the increase of IL-1β and IL-18 [109]. Other studies also assessed several cytokines in vitreous during DME, but they have demonstrated inconsistent findings, as recently reviewed [110]. Higher levels of VEGF [55, 56, 71, 111–113], IL-6 [56, 71, 113, 114], IL-8 [71, 112, 114], and ICAM-1 [56, 111] MCP-1 [56, 107, 114], and acute-phase factors [115] have been associated to DME. Although these studies were included in the systematic review of Minaker et al., only VEGF levels were considered significantly higher in DME compared to controls, according to meta-analysis, while IL-8 failed the sensitivity analysis due to the lack of studies [110].

Age-related macular degeneration

AMD is a multifactor ocular disease that affects central vision and is characterized by a degeneration of photoreceptors and retinal pigment epithelium (RPE) in the macula [116, 117]. AMD represents the most common cause of blindness among the elderly in developed countries [118, 119]. According to WHO, about 195.6 million people suffered from AMD in 2020, of which about 10.4 million had moderate-to-severe vision impairment or blindness [23, 118]. Risk factors, such as age, smoking, high body mass index, and genetic factors, are strongly associated with the development of AMD [119–121].

In the early phase of AMD, yellowish deposits (drusen) are accumulated underneath the retina, accompanied by changes in RPE pigmentation, and thickening of Bruch’s membrane (Fig. 2) [116, 122–124]. AMD can be further classified as early or intermediate based on the area and size of the drusen in the macula, with the presence of larger soft drusen being associated with an increased risk of disease progression [123, 125, 126]. At more advanced stages, patients may develop geographic atrophy, which is characterized by progressive atrophy of RPE and overlying neurosensory retina and choriocapillaris [122, 127, 128]. Currently, the management of dry AMD is based on a regular follow-up evaluation, to detect signals of visual function deterioration and early NV. Changes in lifestyle and antioxidant micronutrient supplementation are also recommended to reduce the risk of developing AMD and the progression of early-to-late AMD [129–134]. Furthermore, several therapeutic strategies show the potential to prevent and delay the progression of dry AMD and restore vision [135, 136], including cell-based therapies [137, 138] and several therapeutic agents [139–143].

Fig. 2.

Vitreous and retinal anatomy in pathophysiological events in the dry age-related macular degeneration and neovascular age-related macular degeneration and current treatments for each clinical feature. BM Bruch’s membrane, RGC retinal ganglion cell, ROS reactive oxygen species, oxidative stress, RPE retinal pigment epithelium

Approximately 10–15% of all AMD patients develop nAMD, the main responsible for severe vision loss [124]. Until recently, the neovascular component was referred to as choroidal NV (CNV), but this nomenclature was replaced by macular NV (MNV), since NV does not always originate from the choroid [144, 145]. According to the Consensus on Neovascular Age-Related Macular Degeneration Nomenclature (CONAN) Study Group, nAMD includes type 1 MNV (occult CNV), type 2MNV (classic CNV), and type 3 MNV [145]. While MNV in type 1 arises from the choriocapillaris and proliferates into the sub-RPE space, contributing to the detachment of the fibrovascular pigment epithelium, type 2 refers to the proliferating of vessels above the RPE into the sub-neurosensory retinal space, where exudation or hemorrhage may occur [145, 146]. Type 3 refers to retinal angiomatous proliferation and is characterized by NV within the retinal vasculature, that extends posteriorly into the subretinal space, and eventually anastomose with the CNV [147, 148]. Laser treatments were the first treatments approved for nAMD, but currently, the first-line treatment has been the regular intravitreal injection of anti-VEGF agents (e.g., ranibizumab, aflibercept, and brolucizumab) [125, 149, 150]. Recently, the FDA-approved faricimab, a bispecific antibody that targets both VEGF and Ang-2 growth factors, showed a comparable but more durable therapeutic effect than aflibercept, which reduces a treatment burden in patients with nAMD [151]. The efficiency of biosimilar anti-angiogenic therapies based on sustained-release implants, encapsulated cell technology, or gene therapy has been also explored in several clinical trials [136, 152, 153].

Although complement cascade, inflammation, angiogenesis, and oxidative stress have been proposed as key pathways for the onset and progression of AMD [154, 155], a better understanding of pathophysiologic processes is crucial for the implementation of more effective treatments [117, 146]. Despite a few studies have focused on the characterization of vitreous proteomics in AMD, some potential disease biomarkers have been proposed (Supplementary Table 1.2) [63, 156, 157]. Using CE–MS, Koss et al. identified 19 putative biomarkers in the nAMD vitreous compared to idiopathic floaters. Changes in vitreous proteome reflected the up-regulation in nAMD of several proteins related to transport, immune responses, complement cascades, and protection against oxidative stress (e.g., glutathione peroxidase 3) [156]. Nobl et al. identified four biomarkers (opticin, CLU, PEDF, prostaglandin-H2 D-isomerase) by CE–MS and LC–MS by comparing vitreous collected from nAMD eyes with different degrees of CNV [157]. Using a label-free LC–MS/MS, Schori et al. found increased levels of lysosomal proteins (cathepsin Z and prosaposin) and inflammatory mediators (chitinase-3-like protein-1) in nAMD compared to ERM controls. Nonetheless, proteins associated with the same processes were found downregulated in dry AMD, suggesting that some of these pathways may be activated later in nAMD. Other proteins involved in the glycolytic process (phosphoglycerate mutase 2), oxidative stress (superoxide dismutase, glutathione reductase), and APP fibril formation (beta-2-microglobulin) were also found up-regulated in nAMD [63]. The role of VEGF in CNV, a hallmark of nAMD, has been strongly supported by its expression in choroidal neovascular membranes and by the clinical efficacy of anti-VEGF treatment [158, 159]. Our research group found lower levels of both VEGF-A and VEGF-B in nAMD compared to patients with retinal vein occlusion and PDR.[73]. Huber et al. observed increased intravitreal levels of VEGFR-1 and decreased levels of PEDF in nAMD vitreous, but the concentration of angiogenic proteins, such as VEGF and Ang-2, remained unchanged [93]. The levels of PEDF are in agreement with the other studies [160, 161], which can indicate that the reduced anti-angiogenic activity also contributes to CNV during AMD. Using a cytometric bead assay, Koss et al. found higher levels of MCP-1 but lower levels of IL-6 and VEGF in nAMD compared to DME vitreous [162]. An increase of inactive (pro-IL-1β) and active forms of IL-1β [163] and TGF-β [164] were also detected in the vitreous of nAMD patients. They also found that semaphorin 3A inhibits the CNV mediated by TGF-β1, as well as the VEGF and TGF-β responses [164]. Elevated levels of MMP9, interleukins (IL-8), PDGF receptor, and BCL-2 associated death promoter, among other molecules, were related to subretinal fluid (SRF) accumulation in AMD patients [165]. To determine the role of the complement activation in the progression of AMD, complement C3 (C3), and factors B and D were quantified in vitreous at distinct severity levels. Although there is no increase in these factors in the vitreous in parallel with the progression of AMD, the activation of the alternative complement pathway is suggested due to the higher levels of factor B fragments observed in more advanced stages of AMD [166].

Proliferative vitreoretinopathy

PVR is a major complication of RRD characterized by the growth and contraction of cellular membranes on the detached retina and within the vitreous cavity [8, 167]. Although it can occur in untreated eyes with RD, the incidence of PVR increases after surgery [7]. Extension and duration of RD, retinal tear size, presence of choroidal detachment and vitreous hemorrhage, chronic intraocular inflammation, and unsuccessful RRD repair are predisposing factors to the development of PVR [168–170]. The pathogenesis of PVR is summarized in Fig. 3. PVR progresses into three main wound-healing phases: inflammation, proliferation, and scar modulation [171, 172]. BRB breakdown leads to the influx of growth factors and inflammatory mediators, which increases the chemotactic and mitogenic activity in the vitreous and foments’ inflammatory recruitment and cellular proliferation [7, 167, 172]. The physical separation of the neurosensory retina and the underlying RPE creates an ischemic environment causing the death of the photoreceptors and neurons [8, 167]. Altogether, these events incite structural and cellular changes in surrounding RPE and glial cells by mechanisms not fully understood [173, 174]. One of these mechanisms is the epithelial–mesenchymal transition (EMT) of the RPE cells, in which cells lose their epithelial characteristics and gain the capacity of migrating and proliferating [8, 173]. In this early phase of PVR, vitreous haze, protein flare, and the presence of pigment clumps are manifested in the eye due to the multiplication of RPE in the vitreous [175]. The migration of proliferating cells, such as RPE and glial cells, into the vitreous cavity, their adhesion to the detached retina, and the uncontrolled production of ECM components lead to the formation of fibroproliferative membranes. The contraction of these membranes and the resulting tractional forces pull the retina into fixed folds [168, 171]. If not addressed promptly, these forces progressively create retinal wrinkles, tears, and tractional RD, which hinders surgical reattachment [7, 171, 172]. Scleral buckling, pars plana vitrectomy, and pneumatic retinopexy are commonly used for the management of RRD and early PVR (grade A) [176], whereas other surgical procedures (Fig. 3) are recommended at more advanced stages [8, 167]. Numerous drugs have been proposed as adjunctive in the treatment of PVR, including anti-inflammatory molecules [177, 178], anti-neoplastic/anti-proliferative agents [179, 180], anti-growth factor pathway inhibitors [181, 182], and anti-oxidants [183]. Several clinical trials (e.g., GUARD trial) [184, 185] have tested the efficiency of intravitreal injections of methotrexate during the post-operative period and an improvement in the retinal reattachment rate has been observed, suggesting a promising alternative for the management of advanced PVR [186].

Fig. 3.

Vitreous and retinal anatomy in pathophysiological events related to the progression from retinal detachment to proliferative vitreoretinopathy and current treatments for each clinical feature. The experimental treatments are marked with *. ECM extracellular matrix, iBRB inner blood-retinal barrier, oBRB outer blood-retinal barrier, RGC retinal ganglion cell, RPE retinal pigment epithelium

In recent years, different studies of vitreous proteome have been carried out to elucidate the mechanisms underlying the PVR pathogenesis. Shitama conducted one of the first proteomic approaches in PVR by studying the vitreous proteome in several vitreoretinal diseases. Increased levels of alpha-1-antitrypsin (AAT) and apolipoprotein A4 were found in PVR, while higher levels of PEDF were reported in RRD compared to PVR. Cathepsin D, transthyretin (TTR), and CLU were also found up-regulated in RRD and PVR compared to DR and PDR samples [46]. The increase of TTR levels in PVR vitreous, as well as complement C4, was confirmed by other authors using 2DE and ELISA [187]. By applying distinct proteomics approaches, kininogen 1 and insulin-like growth factor-binding protein 6 (IGFBP-6) were proposed as new biomarkers of PVR, and p53 and transcription factor E2F1 as potential therapeutic targets [188–190]. These studies showed an increase of alpha-2-HS-glycoprotein, alpha-1B-glycoprotein, serpin family members, and complement factors, suggesting that plasma proteins accumulate in vitreous as the PVR progresses [188, 189]. Although the increase of these components in the vitreous is not PVR specific, it can provide information on the integrity of BRB, and the degree of inflammation and wound healing [11, 170]. These studies also suggest that the complement and coagulation cascade plays an important role in PVR pathogenesis [188, 189]. Furthermore, ECM remodeling during PVR is suggested by the downregulation of several actin family members, tubulin, and opticin [188, 189]. Deregulation in the glycolysis/gluconeogenesis process was also implicated in PVR, since many proteins, some associated with the HIF-1 signaling pathway (e.g., glyceraldehyde-3-phosphate dehydrogenase, ENO1 or 2), were found downregulated in moderate and severe PVR [188]. Interestingly, several glycolytic enzymes were reported as up-regulated in RRD vitreous by our research group and Öhman et al. [191, 192], suggesting that in an early phase, retinal cells increase their metabolism to obtain more energy through glycolysis to compensate for the metabolic “stress” state of the retina [191].

Comprehensive multiplex and ELISA assays have analyzed the vitreous from patients with PVR. Among the factors found differently expressed in PVR are IFN-γ [193, 194], interleukins (e.g., IL-6, IL-8) [193–197], chemokines (e.g., CXCL10) [194, 195], colony-stimulating factors [193, 195], growth factors (e.g., VEGF, PDGF, TGF-β) [193, 194, 198, 199], and MMPs [196] (Supplementary Table 1.3). The results link these factors with PVR events, such as exacerbated wound healing, activation and proliferation of RPE and glial cells, ECM production, and even potential survival mechanisms [200, 201]. Banerjee et al. found a complex pattern of inflammatory mediators in PVR vitreous compared to other vitreoretinal disorders (e.g., PDR). The detection of IL-6, IL-10, TNF-α, IFN-γ, and CSF3, among others, in PVR vitreous reinforces the crucial role of inflammation in PVR. Curiously, VEGF was found at higher levels in PVR than in PDR, whereas fibroblast growth factor (FGF) was only detected in PVR [193]. The increase of VEGF levels in the vitreous and SRF from patients with PVR was confirmed by other authors [194, 198, 199, 202], suggesting a role in the pathogenesis of PVR. In another study, the increase in the levels of CSF3, interleukins (e.g., IL-6), and inflammatory chemokines (e.g., CXCL10) were more evident in PVR compared with ERM controls than in primary RD, showing an increase in inflammation with the progression to PVR [195]. Through the screening of 200 cytokines in the vitreous samples from patients at distinct stages of PVR, Roybal et al. found that 25 cytokines significantly up-regulated in PVR and 20 associated with advanced PVR [194]. The levels of cell adhesion molecules (ICAM-1, PECAM-1), growth factors (VEGF), and chemotactic factors (CXCL10, CCL15) increased gradually from the early to more severe stages of PVR. The up-regulation of cytokines associated with T-cell recruitment, fibrosis (e.g., PDGF), and mTOR activation (e.g., IL-6) was more evident in the early stages of PVR, whereas fibroblast markers (e.g., CXCL12), macrophage inflammatory proteins (e.g., CCL3), and stromal cell-derived factor 1 were more predominant in PVR-C [194].

Molecular mechanisms common to proliferative and neovascular vitreoretinal diseases

The proteins associated with DR/PDR, AMD, and RRD/PVR in diverse proteomics and multiplex studies were individually analyzed by STRING v11.5 [203], based on their protein interactions and functional enrichment, according to gene ontology for biological processes and pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome (Supplementary Table 2). These proteins and the related functional enrichment data were compared, as illustrated in Fig. 4, to set relationships and overlapping biological processes/pathways between their pathological mechanisms. Of these, biological processes, such as inflammation, fibrosis, oxidative stress, neurodegeneration, and vitreous remodeling, stand out in the functional analysis and, therefore, will be discussed below.

Fig. 4.

Proteins detected in proteomics and multiplex studies as differentially expressed in the vitreous of patients with non- and proliferative diabetic retinopathy (DR/PDR), age-related macular degeneration (AMD), and reghmatogeneous retinal detachment (RRD)/proliferative vitreoretinopathy are depicted in the Venn diagram. Protein–protein interaction network of the proteins found to be differentially expressed in more than one disease (black dashed line in Venn diagram) obtained with high confidence in STRING v11.5. Nodes corresponding to the top enriched terms of Gene Ontology and KEGG pathways are highlighted in different colors, as shown in the caption of the figure. Disconnected nodes were removed from the protein–protein interaction network

Mechanisms of inflammation

Nowadays, inflammation is known to play a central role in the development of proliferative pathologies, such as PDR [204–207], AMD [122, 208], and PVR [200, 209, 210]. Inflammatory events in PVR/PDR have been more strongly associated with the ischemic retinal environment resulting from BRB disruption [200, 206, 207, 209], whereas chronic inflammation in AMD may be linked to dysfunction of lipid homeostasis, in particular drusen formation [211]. Although the exact mechanisms have not been fully elucidated, pro-inflammatory cytokines, growth factors, ECM components, and proteolytic enzymes provide crosstalk between inflammation and other processes, including angiogenesis [212, 213], fibrosis [114, 201], oxidative stress [211], and retinal degeneration [214, 215].

The presence of these mediators in the vitreous can provide information on the integrity of BRB and the state of inflammation [11, 170]. As a matter of fact, the breakdown of the BRB and the massive influx of plasma proteins into the retina and vitreous cavity is a suitable predictor of the progression to a proliferative etiology [216, 217]. Increased intravitreal chemotactic and mitogenic activity stimulates the migration of macrophages and the proliferation of RPE and/or glial cells [217, 218]. Nevertheless, growing evidence indicates that changes in the immunosuppressive ocular microenvironment can precede changes in vascular permeability and BRB dysfunction [219]. Hyperglycemia-induced pathways [36, 220], hypoxic–ischemic conditions [221], and the activation of retinal glial cells, macrophages, and retinal neurons [222, 223] promote a pro-inflammatory environment that may disrupt the delicate balance of BRB. VEGF [224, 225], TNF-α [226], IL-1β [227], IGFBPs [228], ICAM-1 [111, 229], MMPs [230, 231], and the KKS [53, 67] induce the BRB breakdown through increased vascular and endothelial cells permeability, changes in tight junction proteins and cadherins, leukocyte recruitment, among other mechanisms. While most of these results are relevant to inner BRB dysfunction in hypoxic–ischemic and diabetic conditions, the outer BRB disruption in nAMD is more related to oxidative stress [221].

IL-1β was reported as the initiator of ocular inflammation [232] and, in combination with TNF-α, amplifies the inflammatory cascade through the modulation of VEGF, colony-stimulating Factors, MMPs, and adhesion molecules [232, 233]. Several studies suggest that IL-1β release is mediated by the activation of the NLRP3 inflammasome complex in response to the accumulation of drusen components and necrosis of adjacent RPE cells [234–236]. In turn, active IL-1β, TNF-α, and the induced pro-inflammatory mediators, such as IL-6 or IL-17, promote the recruitment of inflammatory cells, dysfunction, and proliferation of RPE, microglia, and Müller cells, activation of the complement cascade, and the production of ECM components. This leads to an inflammatory and post-fibrotic environment in the retina, which contributes to the degeneration of retinal neurons, RPE, and vascular endothelial cells [232, 237, 238].

IL-6 regulates the synthesis of acute-phase proteins and pro-inflammatory mediators, in addition to its role in hematopoiesis and the proliferation of fibroblast and glial cells [233, 239]. Acute-phase proteins are expressed by Müller cells in response to an inflammatory stimulus to restore homeostasis, but they can exert a protective effect or exacerbate tissue damage [240]. While AAT can have a protective effect [241], the presence of c-reactive protein, AAT, and serum amyloid in drusen has been associated with the activation of the complement cascade and immune cells, and neurodegeneration [121, 239, 242]. In its turn, molecules, such as ICAM-1 and VCAM-1, contribute to inflammation by mediating the adhesion of leukocytes to activated endothelial cells [225, 243], whereas other (e.g., PECAM1) mediates leukocyte transmigration [244]. Adhesion molecules are constitutively expressed by human retinal endothelial cells and are further increased in pathological conditions, which can predispose the retina to inflammation, and also lead to the death of endothelial cells and pericytes [245]. Vascular dysfunction favors a retinal ischemic environment, inducing the expression of VEGF and triggering the NV [204, 225]. Augmented levels of ICAM-1 in vitreous, vascular endothelium, and ERM have also been correlated with an increased risk of developing PVR [246, 247]. These correlations suggest that adhesion molecules could be common mediators between inflammation, fibrosis, and angiogenesis.

Activation of complement and coagulation cascades and fibrosis

Low levels of activation of complement and coagulation cascades are characteristics of the immune-privileged status that contributes to retinal homeostasis and integrity. Therefore, chronic activation of both cascades has been implicated in a variety of pathophysiological processes in the eye [248, 249]. Genetic variants in complement genes, such as C3, complement Factor H (CFH), and complement factor B, are considered as risk factors for AMD [166, 250]. CFH polymorphisms have been associated with sub-RPE accumulation of drusen and the extensive activation of the alternative pathway [250]. The activation of complement components like C3a and C5a, and their deposition in drusen, RPE, and choriocapillaris layers stimulate inflammatory responses, the recruitment of mononuclear phagocytes, and inhibit NV. On the other hand, the membrane attack complex has been related to the destruction of choroidal endothelium and the loss of pericytes [250–252]. Furthermore, the removal of retinal debris by the mononuclear phagocyte system could be compromised with aging, resulting in a sustained pro-inflammatory environment [253].

Although genetic studies have strongly supported the association of complement factors with AMD [250], a few studies have reported the quantification of these proteins in nAMD vitreous [63]. Changes in complement pathways also have a genetic component in PDR [254], but unlike nAMD, up-regulation of these components in PDR vitreous, was reported (see  “Diabetic retinopathy” section). Increased levels of both full-length C3 and its activated fragment and CFH were found in vitreous and serum samples of patients with PDR, suggesting the activation of the alternative pathway of complement [255]. Although their role in PDR is still unclear [255], activation of complement and coagulation cascades could be associated with increased leukostasis and vascular permeability, BRB breakdown, and microglial activation [53, 54, 60, 64]. Likewise, the complement activation was reported in RD/PVR [187–189, 256], as well as its involvement in pathological processes, such as increased vascular permeability, endothelial cell proliferation, migration, RPE atrophy, reactive gliosis, and loss of photoreceptor outer segments [189, 257]. Sweigard et al. demonstrated that the activation of the alternative complement pathway promotes early photoreceptor cell death during RD. By contrast, deficient levels of complement components were related to impaired signaling function in retinal layers, highlighting the relevance of this system to retinal homeostasis [257].

The role of the coagulation system in the development of PVR is well established and is evidenced by intraocular fibrin deposition [258]. The exposition of RPE cells to serum components, such as thrombin, fibrin, plasmin, or fibronectin upon BRB breakdown, triggers EMT in these cells [35, 201]. Increased levels of kininogen 1, Factor Xa, and thrombin activity, which were detected in the vitreous in PVR, can contribute to its pathogenesis through the increase of levels of PDGF, TGF, and pro-inflammatory cytokines [189, 190, 258, 259]. It has been suggested that these effects are exerted in RPE cells via NF-κB, by the activation of mediators of pro-inflammatory signaling, angiogenesis, and fibrosis [190, 258, 259]. In addition to coagulation components, intraocular levels of IL-6 [260–262], osteopontin [263], connective tissue growth factor (CCN2) [6, 264], insulin-like growth factors (IGFs) [262, 265], and other growth factors (e.g., TGF-β, PDGF) [70, 262, 266], were correlated to fibrosis. The role of inflammatory and fibrogenic factors in PVR has been recently reviewed [209, 210]. Although its role is less clear in DR and AMD, retinal fibrosis is also a unifying feature of neovascular retinal diseases, particularly at late stages [267]. The elucidation of these mechanisms has recently paved the way for the development of new antifibrotic therapies (e.g., TGF-β signaling antagonists) that can benefit patients by reducing ocular morbidity associated with ocular fibrosis [268]. Some of these potential therapeutics are described below.

TGF-β is a key molecule in the activation of the EMT and fibrotic process. In addition to its role in ECM remodeling, it is capable of inducing itself and other pro-fibrotic factors, interleukins, and MMPs [209, 269, 270]. Several therapeutic strategies have been devised to target TGF-β in fibrotic diseases, including the administration of neutralizing antibodies [266], TGF-β agonists [182, 269], or inhibitors of TGF-β-induced pathways, such as Smad (e.g., pirfenidone) [271, 272], Notch (e.g., RO4929097) [273], and RhoA/Rho-kinase (e.g., Y27632) [274], and TAK1 signaling (e.g., 5Z-7 oxozeaenol) [275]. Anti-fibrotic drugs, such as pirfenidone [276] or OM-101 [277], showed both efficacy and safety for the treatment of PVR in animal models by reducing significantly the fibrotic response [276], and the incidence of RD and PVR [277].

PDGF is a potent chemoattractant and mitogen for fibroblasts, glial cells, and RPE cells, and promotes EMT, cellular contraction, and collagen synthesis [278, 279]. The co-expression of PDGF and its receptor (PDGFR) in ERMs suggests that RPE cells gain the capacity of autocrine stimulation upon the loss of cell–cell contact [280, 281], which was also verified for IGFs [265, 282]. As with TGF-β, several therapeutic agents targeting PDGF and its receptor have been tested in animal models, including aptamers [283], RNA interference [284], and antibodies [262, 285]. Remarkably, the neutralization of intravitreal PDGFs in a rabbit experimental model did not reduce PVR, suggesting that PDGF could be a poor therapeutic target [285]. Nevertheless, non-PDGF growth factors are capable of activating the PDGFR, whereas PDGFs could act as a protective agent [262, 285, 286]. The mechanism of activation of PDGFR by non-PDGF growth factors is mediated by increased cellular levels of reactive oxygen species (ROS), leading to the activation of Src family kinases that promote phosphorylation and activation of PDGFR [287]. In light of these findings, targeting the activation of PDGFR with anti-oxidants (e.g., n-acetylcysteine) [183] or kinases inhibitors (e.g., rapamycin that targets mTOR pathway) [194, 288], could prevent PVR in patients undergoing retinal reattachment surgery or help to ameliorate the symptoms.

The role of VEGF in fibrosis is also intermediated by the binding to PDGFR [286, 289] and by its up-regulation in vitreous, SRF, and ERMs in vitreoretinal diseases associated with fibrosis [198, 202]. However, the contribution of VEGF to angio-fibrotic switch mechanisms in angiogenic diseases, such as PDR or nAMD, is still unclear [290–292]. The ratio between CCN2 and VEGF levels was found to be the strongest predictor of fibrosis [290, 291], which means that the angio-fibrotic switch may be accomplished by a reduction of intravitreal VEGF levels and a progressive increase of the CCN2 [291]. This fact could explain the increase in intraocular fibrosis observed after treatment with anti-VEGF drugs [290, 291, 293]. Therefore, controlling the fibrosis after the administration of anti-VEGF drugs in patients with nAMD and PDR could be beneficial to explore the feasibility of combining this treatment with anti-fibrotic therapy [293].

Oxidative stress

The high demand for energy and oxygen, the presence of high levels of polyunsaturated fatty acids, and constant exposure to light make the retina particularly susceptible to oxidative stress and lipid peroxidation [294, 295]. Impaired redox balance has been associated with several ophthalmologic disorders, including DR, AMD, and PVR, as recently reviewed by our research group [296]. The rate of production and accumulation of ROS increases with age, where chronic low-grade inflammation and hypoxia in the aged retina can be potential sources [297]. Also, the ability of the eye to counteract the effect of ROS declines with age [294, 297]. One of the protective mechanisms against oxidative stress of the retina and surrounding tissues is provided by the high antioxidant capacity of the vitreous. However, it also decreases with age, contributing to oxidative damage [295, 298, 299]. The activation of protective mechanisms is suggested by the increase of intravitreal antioxidant enzymes, such as glutathione peroxidase, catalase, and peroxiredoxins [296, 298].

Hyperglycemia stimulates oxidative stress in the ocular tissues, which culminates in changes in vitreous structure, Bruch’s membrane thickening, and loss of retinal and capillary cells [300, 301]. Multiple mechanisms in diabetes are responsible for mitochondrial dysfunction and subsequent ROS production, including the activation of protein kinase C, hexosamine, and polyol pathways and the increased production of AGEs and their receptors (RAGE), [302, 303]. On the other hand, polymorphisms in antioxidant enzyme genes, cigarette smoke, and exposure to sunlight, among other environmental factors, are considered risk factors for AMD [211, 299, 304]. A combination of light exposition, dysfunction in lipid homeostasis, and oxidative stress promotes lipid peroxidation, which is one of the pathological events in AMD [211]. It has been suggested that these mediators could cause geographic atrophy by promoting inflammatory responses and oxidative damage to cellular proteins, lipids, and DNA (particularly within mitochondria) [211, 305]. Lipid peroxidation and the consequent accumulation of lipofuscin contribute to lysosomal dysfunction in RPE cells, which are essential for removing waste products and maintaining neural retina nutrition and homeostasis [306–308]. Impaired autophagy contributes to the accumulation of intracellular lipofuscin and drusen in Bruch's membrane [309, 310], creating a physical barrier to the influx of oxygen and nutrients to the photoreceptors and the waste removal between RPE and choroid [311]. Metabolic dysregulation and hypoxia stimulate the production of growth factors and cytokines that compromises the integrity of BRB [221, 312], while lipofuscin accumulation mediates light-induced damages in RPE cells and adjacent photoreceptors [306, 313]. Furthermore, the increase of free radicals, oxidative stress byproducts, and drusen can generate chronic inflammation through the production of pro-inflammatory interleukins and adhesion molecules [304, 314].

It was also reported that oxidative stress plays a role in EMT, which contributes to the pathogenesis of PVR and AMD [304, 315]. A synergistic effect between TGF, macrophage migration inhibitory factor, and hydrogen peroxide induces EMT through the up-regulation of α-smooth muscle actin, vimentin, and fibronectin and downregulation of cadherins [315, 316]. Interestingly, RPE cells increase autophagic activity and secretion of exosomes in response to oxidative stress, aging, and in the presence of drusen [309, 317, 318], probably to remove damaged and toxic materials [319]. Although, initially, these mechanisms may contribute to retinal protection and survival, the dysfunction of lysosomal degradation contributes to drusen formation, whereas exosomes could carry proteins and miRNAs that promote NV and EMT in neighboring cells [309, 317–319].

Mechanisms of neurodegeneration and neuroprotection

The complex architecture and functionality of the retina and its high energy and oxygen requirements make it very susceptible to stress [320, 321]. As neurodegeneration depends on a balance between the levels of neurotoxic and neuroprotective factors, it has been suggested that neurotrophic factors, anti-oxidants, and anti-apoptotic therapy could mitigate the effects of secondary degenerative events and reduce the loss of neurons in the retina [322]. Therefore, neuroprotection is gaining interest to design therapies for preserving the photoreceptors and ganglion cells after retinal damage in DR [323–325], AMD [326, 327], and RD/PVR [8, 328, 329]. In this context, it is relevant to understand the endogenous protective mechanisms triggered in the eye, with RPE and Müller cells being the main ones responsible for the secretion of neurotrophic factors [330, 331].

In response to acute stress signals, microglia produce neurotrophic mediators, such as GDNF and NT3, that act directly in neurons. In turn, brain-derived neurotrophic factor (BDNF) and CNTF induce Müller cells secretion of secondary mediators (e.g., FGF, GDNF) that mediate the survival of photoreceptors [215, 332]. RPE cells also contribute to cell survival signaling by secreting a wide range of neurotrophic factors (e.g., PEDF, IGFs, and NGF), among others, including cytokines, angiogenic, and anti-angiogenic factors [333]. Neurotrophic factors are also expressed in the retina in response to mechanical injury, suggesting an important role in photoreceptor rescue. Several of these factors diminished the death of photoreceptors in experimental models of RD [328, 334]. However, the chronic activation of these mechanisms can initiate severe alterations in retinal integrity, and exacerbate neuronal death [215, 335]. Complement factors [255, 336], IL-6 [337, 338], TGF-β1 [339, 340], and VEGF [202, 341] can act as neurotrophic factors and have an important role in neuronal, glial, and vascular homeostasis. Nevertheless, due to their role in other pathological events (see  “Mechanisms of inflammation” and “Activation of complement and coagulation cascades and fibrosis” sections), they can eventually contribute to neurodegeneration.

Glial and RPE dysfunction, activation of Müller cells, and the loss of neurotrophic support are critical for the survival of photoreceptors and retinal ganglion cells [330, 331]. Activated Müller cells suffer morphological, biochemical, and physiological changes as a result of the dramatic increase in the expression of glial fibrillary acidic protein (GFAP) and vimentin [342]. The increase in the levels of these proteins in the RD vitreous was correlated with the severity of PVR, suggesting that reactive gliosis may play a key role in PVR formation [343]. Recently, Eastlake et al. described that proteins related to cytoskeleton, ECM, and ribosomes are up-regulated in gliosis compared to the normal retina, with Müller glia appearing to be the main source of vimentin, GFAP, polyubiquitin, and HSP90a [344]. Photoreceptor degeneration is greatly reduced in the absence of GFAP, vimentin, and other intermediate filaments, indicating that reactive gliosis impacts retinal damage [345, 346]. Taking this into account, the regulation of their expression in Müller cells might promote retinal regeneration after disease or injury [344].

Activated microglia also express retinal damage biomarkers such as ENO2 [347], and phagocytize toxic proteins like APP [330, 348]. ENO2 can act either as a neurotoxic or neuroprotector factor (e.g., against amyloid-β peptide toxicity), but upon neuron injury, it is expressed in M1-type microglia, promoting neurodegeneration [347, 349]. APP is a key component of drusen and its accumulation in the eye has been associated with neurodegeneration in AMD and glaucoma [350–352]. The increased phagocytic capacity of microglia and the expression of APP degrading enzymes could initially contribute to Aβ clearance. However, it eventually promotes neuroinflammation by the induction of pro-inflammatory cytokines and NLRP3 inflammasome [330, 348]. Likewise, it has been suggested that APP toxicity in neurons is mediated by its intralysosomal accumulation through macroautophagy, and consequent lysosomal membrane permeabilization [353]. On the other hand, the presence of αβ-crystallins in the drusen may reduce the aggregation of toxic proteins and augment autophagy-mediated clearance [354]. Crystallin mutant and knockout animal models exhibit inefficient lysosomal clearance, evidenced by the accumulation of lipofuscin-like material and increased susceptibility of RPE to oxidative stress and apoptosis [307, 355]. Moreover, several studies in animal models indicated that overexpression of heat shock proteins supports the survival of injured RPE, retinal ganglion cells, and photoreceptors [355–357].

AKT/mTOR pathway in RPE has also been associated with the impairment of autophagic flux and increased oxidative stress, glycolysis, and glycogen storage [358–360]. mTOR is activated in response to RPE stress, impairing autophagy and inducing dedifferentiation, hypertrophy, and metabolic reprogramming in RPE cells to promote its survival. However, the return to basal levels of mTOR activity is necessary to diminish the RPE glycolytic metabolism and to increase glucose supply to photoreceptors, indicating that chronic mTOR activation could lead to glucose deprivation and degeneration of photoreceptors [360, 361]. In turn, degenerative photoreceptors display low mTOR activity and high HIF-1 levels, as well as the activation of chaperone-mediated autophagy, suggesting that nutrient deprivation and prolonged starvation might underlie neurodegeneration.

Vitreous aging and ECM remodeling

With aging, the human vitreous suffers a progressive remodeling characterized by the loss of type IX collagen and short-range interactions with other ECM components, as well as the new synthesis and aggregation of collagen fibrils, which eventually leads to vitreous liquefaction and PVR [17, 362]. Although age-related ECM changes are frequent, they are intensified by pathological events, such as hyperglycemia, ocular inflammation, and oxidative stress [17, 22, 298, 362]. Therefore, vitreous liquefaction and PVD have been associated with various retinal pathologies [17, 19, 22]. ECM dynamics are tightly regulated by a balance between proteinases and their neutralizing substances [22, 363, 364], which suggests that the deregulation of these mechanisms might be underlying degenerative changes in vitreous and vitreoretinal diseases [365].

MMPs are the main proteases involved in the degradation of collagen and other ECM components [363, 366], although other proteases are also involved, including ADAMTS, cathepsins, and plasminogen [363, 364]. Under physiological conditions, MMPs are synthesized and secreted into the vitreous in their inactive form [22], but their expression and activity increase in tissue repair, inflammation, oxidative stress, or under-remodeling processes [366]. MMPs levels correlate with pathological features, such as SRF accumulation in nAMD [165], the duration and extent of RD [367], the grade of post-operative PVR [368], and PVD [369], and NV in DR [264]. ECM dynamics modulate a series of pathological features of these vitreoretinal diseases, like vascular permeability, NV [366, 370], inflammation [244, 371], and fibrosis [209, 372]. The degradation of ECM components by MMPs modulates these processes by providing a scaffolding via ECM–integrin binding that facilitates the cell adhesion and migration of immune cells and vascular endothelial cells [244, 363, 370]. Proteolysis via MMPs also changes the bioavailability of factors sequestered in ECM, including growth factors, chemoattractants, and other signaling molecules such as matricellular proteins or bioactive ECM fragments [363, 364, 370]. Besides that, MMPs process other bioactive molecules, including growth factors and other cytokines and cell surface molecules, which may promote their inactivation or potentiate their effects [244, 363, 371].

On the other hand, increased levels of metalloproteinase inhibitor 1 (TIMP1) were found in RRD/PVR [196, 367, 368] and PDR [63, 264], whereas TIMP2 only slightly increased in PDR [64]. The up-regulation of TIMP1 by pro-fibrotic factors (factor Xa, TGF-β) was associated with fibrosis, since their combined effect potentiates ECM production and prevents the destruction of the newly synthesized matrix [259, 368]. Nevertheless, it has been suggested that MMPs/TIMPs levels suffer concomitant changes during RRD/PVR, and that secretion of MMPs by RPE cells might assist in its migration into vitreous [209, 368]. In accordance, ERM removed from patients with PVR and PDR are particularly rich in structural (e.g., collagens) and adhesive ECM components (e.g., fibronectin) and matricellular proteins (e.g., tenascin, thrombospondin 1), but also contain MMPs/TIMPs [70, 373]. Decreased 72 kDa Type IV Collagenase (MMP2) and MMP9 activity were also associated with Bruch’s membrane thickening and accumulation of lipid-rich debris in AMD [374]. Conversely, higher levels of MMPs were associated with the progression of CNV, which was confirmed using an MMP2 and MMP9 knockout mouse model [375]. TIMP1 levels also correlated with the degree of NV, but its correlation with activated TGF-β2 levels was independent of the degree of DR, suggesting that it might have a more relevant role in angiogenesis than in fibrosis [264]. Furthermore, TIMP1 could result from the breakdown of BRB, in contrast with TIMP2 that is specifically secreted by intraocular tissues, which could explain its correlation with NV degree in PDR [376]. More recently, it was verified that TIMP2 and other protease inhibitors (e.g., SERPINA5) are up-regulated after treatment with ranibizumab [68]. TIMP2 seems to be more efficient at inhibiting angiogenesis than TIMP1, and it may mediate the effects of treatment with ranibizumab [68]. This can be related to the fact that TIMP2 inhibits growth and angiogenesis, not only by inhibiting MMPs but also via α3β1 integrin-mediated binding of TIMP2’s N-terminal domain to endothelial cells [377]. Although the role of ECM remodeling in proliferative diseases is not clear, it might be a unifying mechanism between vitreous degeneration and pathological events, such as fibrosis, inflammation, and NV. The structural integrity of the vitreous seems to be one of the decisive factors for the progression to a proliferating and/or neovascular etiology and, therefore, ECM molecules may be a potential therapeutic target.

Conclusions and future perspectives

During the progression of vitreoretinal diseases, the vitreous acts as a repository of the mediators involved in inflammation, fibrosis, oxidative stress, neurodegeneration, and remodeling of the ECM. Therefore, the study of vitreous proteome has shed light to elucidate some of the pathological mechanisms underlying these diseases and discover potential pharmaceutical targets. The characterization of the vitreous humor proteome has largely contributed to the recognition of pathways involved in PDR and to the identification of new therapeutic targets, in addition to VEGF. The lack of validation in a larger cohort of patients might explain the absence of reliable vitreous biomarkers. Proteomics studies focusing on nAMD and PVR are scarce, although they provided interesting evidence on the pathogenesis of these diseases. In light of proteomic studies, it would be beneficial to incorporate other therapeutic agents targeting fibrosis, neurodegeneration, oxidative stress, and vitreous degeneration in PVR, nAMD, and PDR.

However, the interpretation of proteomics results in the context of vitreoretinal pathologies could be challenging due to the high complexity of vitreous, as illustrated by the presence of several proteoforms, including isoforms and post-translational modified proteins, which have distinct or dual functions. In this context, proteomics studies should be complemented with functional proteomics studies, other -omics analyses (e.g., genomics and metabolomics), and in vitro and in vivo experiments. Genomics analysis could be relevant in nAMD, since this disorder has a significant genetic component. Cooperation between clinicians and basic researchers is essential to fill the gap between medical practice and basic science, contributing to discovering reliable pharmaceutical targets and finding correlations between biomarkers and disease progression. Unfortunately, vitreous is collected at later stages of vitreoretinal pathologies (e.g., ERM and vitreous hemorrhage), reducing dramatically the probability of finding reliable therapeutic targets for the earliest stages of the disease. Finally, it must be considered that invasive sampling hampers the use of the vitreous for diagnosis but, when obtained as part of clinical treatment, these biomarkers may be used for prognosis and/or to predict the response to treatment.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

2DE

Two-dimensional electrophoresis

AAT

Alpha-1-antitrypsin

AGEs

Advanced glycation end products

AMD

Age-related macular degeneration

Ang-1

Angiopoietin 1

Ang-2

Angiopoietin-2

APP

Amyloid-beta A4 protein

BDNF

Brain-derived neurotrophic factor

BRB

Blood-retinal barrier

C3

Complement C3

CA

Carbonic anhydrase

CCN2

Connective tissue growth factor

CE–MS

Capillary electrophoresis coupled to mass spectrometry

CFH

Complement factor H

CLU

Clusterin

CNTF

Ciliary neurotrophic factor

CNV

Choroidal neovascularization

CSF3

Granulocyte colony-stimulating factor

CXCL10

C-X-C motif chemokine 10

DME

Diabetic macular edema

DR

Diabetic retinopathy

ECM

Extracellular matrix

EMT

Epithelial-mesenchymal transition

ENO2

Gamma-enolase

ERM

Epiretinal membranes

FGF

Fibroblast growth factor

GDNF

Glial cell-derived neurotrophic factor

GFAP

Glial fibrillary acidic protein

HIF-1

Hypoxia-inducible factor

ICAM-1

Intercellular adhesion molecule 1

IGFBPs

Insulin-like growth factor-binding proteins

IGFs

Insulin-like growth factors

IL1-β

Interleukin-1 beta

IL-6

Interleukin-6

IL-8

Interleukin-8

IL-10

Nterleukin-10

INF-γ

Interferon-gamma

KKS

Kallikrein-Kinin system

LC–MS

Liquid chromatography coupled to mass spectrometry

MCP-1

Monocyte chemoattractant protein-1

MMP9

Matrix metalloproteinase 9

MMP2

72 KDa type IV collagenase

MMPs

Metalloproteinases

MNV

Macular neovascularization

nAMD

“Wet” or neovascular age-related macular degeneration

NGF

Nerve growth factor

NT3

Neurotrophin-3

NV

Neovascularization

PDGF

Platelet-derived growth factor

PDR

Proliferative diabetic retinopathy

PECAM-1

Platelet endothelial cell adhesion molecule

PEDF

Pigment epithelium-derived factor

PlGF

Placental growth factor

PVD

Posterior vitreous detachment

PVR

Proliferative vitreoretinopathy

RD

Retinal detachment

ROS

Reactive oxygen species

RPE

Retinal pigment epithelium

RRD

Rhegmatogenous retinal detachment

SRF

Subretinal fluid

TGF-β

Transforming growth factor

Tie-2

Tyrosine kinase receptor

TIMP-1

Metalloproteinase inhibitor 1

TIMP2

Metalloproteinase inhibitor 2

TNF-α

Tumor necrosis factor

TTR

Transthyretin

VCAM-1

Vascular cell adhesion protein-1

VEGF

Vascular endothelial growth factor

VEGFR-1

Vascular endothelial growth factor receptor 1

Author contributions

FMS: conceptualization, formal analysis, writing—original draft, and writing—review and editing. SC: conceptualization; writing—review and editing. JM: conceptualization; writing—review and editing. JPCS: conceptualization; writing—review and editing. AP: supervision; writing—review and editing. CTT: supervision, conceptualization, and writing—review and editing. LAPP: supervision, conceptualization, and writing—review and editing.

Funding

Fundação para a Ciência e a Tecnologia (Grant no. SFRH/BD/112526/2015, UIDB/00709/2020, UIDP/04378/2020, UIDP/00709/2020, UIDB/04378/2020); European Regional Development Fund (Grant no. Centro-01-0145-FEDER-000019-C4); Fondo Europeo de Desarrollo Regional (Grant no. PT17/0019/0001); Spanish Government (Grant no. SEV2017-0712); Portuguese Pharmacists Order (Grant no. BInov 2021)

Data availability

The data reviewed in this paper is available in the original papers from proteomics and multiplex studies. These revised papers include the references [43–48, 52, 54, 59, 60, 62–64, 68–71, 98] for DR/PDR, [63, 93, 156, 157] for AMD, and [46, 188–195, 256] for RRD/PVR. The reviewed data is displayed in the supplementary table.

Declarations

Conflict of interest

The authors declare that they have no competing interests.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.