Proteomic Biomarkers of Retinal Inflammation in Diabetic Retinopathy

Diabetic retinopathy (DR), a sight-threatening neurovasculopathy, is the leading cause of irreversible blindness in the developed world. DR arises as the result of prolonged hyperglycemia and is characterized by leaky retinal vasculature, retinal ischemia, retinal inflammation, angiogenesis, and neovascularization. The number of DR patients is growing with an increase in the elderly population, and therapeutic approaches are limited, therefore, new therapies to prevent retinal injury and enhance repair are a critical unmet need. Besides vascular endothelial growth factor (VEGF)-induced vascular proliferation, several other mechanisms are important in the pathogenesis of diabetic retinopathy, including vascular inflammation. Thus, combining anti-VEGF therapy with other new therapies targeting these pathophysiological pathways of DR may further optimize treatment outcomes. Technological advancements have allowed for high-throughput proteomic studies examining biofluids such as aqueous humor, vitreous humor, tear, and serum. Many DR biomarkers have been identified, especially proteins involved in retinal inflammatory processes. This review attempts to summarize the proteomic biomarkers of DR-associated retinal inflammation identified over the last several years.


Introduction
Diabetic retinopathy (DR) is the leading cause of irreversible blindness in the developed world and contributes to the majority of blindness in working-age adults, affecting more than four million individuals [1][2][3]. The prevalence of this sight-threatening neurovasculopathy is expected to increase with longer life expectancies and the growing elderly population [1,4,5]. DR is the most common microvascular complication arising from diabetes, and diabetic patients have a twenty-five-fold greater risk of blindness than non-diabetic individuals. Since DR arises as a result of extended exposure to hyperglycemia, it occurs in almost all type I diabetics and the majority of type II diabetics, thereby nearly tripling their treatment costs [6][7][8][9].
DR is characterized by leaky retinal vasculature, retinal ischemia, angiogenesis, and retinal inflammation. These pathologies manifest clinically as cotton-wool spots, exudates, small tortuous veins, aneurysms, and areas of hemorrhage, which may result in reduced acuity, loss of color sensitivity, and problematic night vision [1,10,11]. While the underlying cause of DR is prolonged hyperglycemia,

Proteomic Changes in Biofluids Associated with Diabetic Retinopathy
Increased vascular permeability and elevated levels of inflammatory proteins in serum and ocular biofluids (VH, AH, and tear) are the primary characteristics of retinal inflammation associated with DR [11,15,[95][96][97][98][99][100][101]. While AH and tear are distally located and interact indirectly with the retina, both serum and VH have direct contact with the retina. On the other hand, AH and VH sample collection requires highly invasive procedures, whereas serum and tear samples can be collected using relatively non-invasive methods. The circulatory proteins in these biofluids can not only provide clues about DR pathogenesis, but can also serve as biomarkers, with diagnostic or prognostic potential [102][103][104]. Furthermore, proteomic changes in biofluids could offer a means of more personalized medicine. For example, plasma kallikrein, a central component of the pro-inflammatory kallikrein kinin system, has been shown to be elevated in the VH of DR patients [15]. The kallikrein kinin system works independently of VEGF regulation, and standard anti-VEGF intravitreal injection might not be effective in these subjects [11,15]. Several studies examining proteomic changes associated with DR using biofluids are discussed below and are summarized in Figure 1.

Proteomic Changes in Biofluids Associated with Diabetic Retinopathy
Increased vascular permeability and elevated levels of inflammatory proteins in serum and ocular biofluids (VH, AH, and tear) are the primary characteristics of retinal inflammation associated with DR [11,15,[95][96][97][98][99][100][101]. While AH and tear are distally located and interact indirectly with the retina, both serum and VH have direct contact with the retina. On the other hand, AH and VH sample collection requires highly invasive procedures, whereas serum and tear samples can be collected using relatively non-invasive methods. The circulatory proteins in these biofluids can not only provide clues about DR pathogenesis, but can also serve as biomarkers, with diagnostic or prognostic potential [102][103][104]. Furthermore, proteomic changes in biofluids could offer a means of more personalized medicine. For example, plasma kallikrein, a central component of the pro-inflammatory kallikrein kinin system, has been shown to be elevated in the VH of DR patients [15]. The kallikrein kinin system works independently of VEGF regulation, and standard anti-VEGF intravitreal injection might not be effective in these subjects [11,15]. Several studies examining proteomic changes associated with DR using biofluids are discussed below and are summarized in Figure 1.

Serum
Diabetes is a systemic disease therefore, biomarkers found in circulation may not only be indicators of the disease, but also of the progression of specific diabetic complications. As DR is primarily a microvascular complication, proteomic changes involved in DR pathology may be evident in the systemic circulation. Serum is one of the most accessible and easily obtained biofluids, allowing for sample collection from both DR patients and healthy controls. The ease of sample collection relative to other relevant fluids like VH or AH provides sufficient power to detect potential proteomic differences in patients with various clinical stages of DR including non-proliferative DR (NPDR), PDR, and DME [105].

Serum
Diabetes is a systemic disease therefore, biomarkers found in circulation may not only be indicators of the disease, but also of the progression of specific diabetic complications. As DR is primarily a microvascular complication, proteomic changes involved in DR pathology may be evident in the systemic circulation. Serum is one of the most accessible and easily obtained biofluids, allowing for sample collection from both DR patients and healthy controls. The ease of sample collection relative to other relevant fluids like VH or AH provides sufficient power to detect potential proteomic differences in patients with various clinical stages of DR including non-proliferative DR (NPDR), PDR, and DME [105].
In another study, the inflammation-regulating proteins α2-HS-glycoprotein (AHSG), α1-acid glycoprotein (AGP), apolipoprotein A-1 (APOA1), and haptoglobin (HP) were found to be differentially expressed in the serum from patients with NPDR and PDR as compared to healthy controls [10]. The serum levels of α-HS-glycoprotein were increased in PDR patients, whereas the levels of AGP and APOA1 were decreased relative to healthy controls [10]. The downregulation of AGP was not expected, since α1-glycoprotein is involved in pro-inflammatory response [10,107,108]. In a separate study, the inflammatory response protein azurocidin (AZU1) was identified as being elevated in the serum of diabetic patients, especially in patients with diabetic complications including retinopathy [109]. This protein is released by neutrophils in response to inflammatory stimuli on vascular endothelium and is known to play a role in vascular permeability within the retina [110,111].

Vitreous Humor
The vitreous humor is the gelatinous component of the posterior segment of the eye that gives the eye its spherical shape. In addition to its role in structural support, the transparent nature of the vitreous body aids in light transmission to the retina. Due to the proximal location of the vitreous body to the retina, pathological events in the retina can be monitored through VH examination [4,7,13,16,22,56,101,112,113]. In fact, due to its avascular nature, much of the protein content of the VH comes from the retina itself [81,[113][114][115]. VH is often obtained from DR patients during pars plana vitrectomy, in which the entire vitreous body is removed and replaced due to hemorrhage [2,5,22]. Vitreous can also be obtained by a needle biopsy, which is less invasive and can be conducted in the clinic rather than the operating room [22]. VH is obtained from individuals undergoing treatment for a pathological state such as proliferative vitreoretinopathy, rhegmatogenous retinal detachment, idiopathic macular hole (IMH), or epiretinal membrane (ERM) [4,22,63,81,116,117]. In addition, post-mortem VH is obtained from organ donors.
Several studies have also used VH proteomics to examine the effects of anti-VEGF intravitreal injections on disease progression [3,33,35]. Although Loukovaara et al. observed an increased presence of complement, coagulation, and other inflammatory proteins in the VH of DR patients, they were unable to identify a significant effect of the anti-VEGF agent bevacizumab on these proteins [3]. Interestingly, Wei et al. observed an increase in complement factors, coagulation factors, apolipoproteins, and immunoglobulins following intravitreal injection, while photocoagulation was able to reduce levels of the pro-inflammatory protein osteopontin (SPP1) [35]. Zou et al. compared the VH proteomes of DR patients treated with the anti-VEGF agent ranibizumab and found an expected decrease in VEGF levels as well as a decrease in acute inflammatory response, platelet degranulation, and complement activation proteins [33,34].

Aqueous Humor
Aqueous humor is the fluid in the anterior chamber of the eye; it is produced by the ciliary body epithelium [127,128]. AH is an integral component in many ocular health functions, including nutrient and oxygen supply, removal of metabolic waste, ocular immunity, and ocular shape and refraction [129]. The major constituents of AH are proteins, water, and electrolytes. Although proteins in AH are present in relatively low concentrations compared to blood serum, they are vital in the maintenance of anterior segment homeostasis [65,[128][129][130][131][132][133][134][135][136]. Previous studies [130][131][132][133]137], including one of our own [138], have shown significant alteration in several proteins in the AH obtained from glaucomatous eyes. Proteins in the systemic circulation may cross through fenestrated capillaries into the ciliary body, the production center of AH [13,65,139].
In a recent study examining the AH proteomic profile of PDR patients, LC-MS/MS analysis identified 10 proteins associated with PDR [13]. These proteins were involved in a number of biological processes including inflammation and included apolipoprotein A-I (APOA1), apolipoprotein A-II (APOA2), apolipoprotein A-IV (APOA4), and α-1-acid glycoprotein 1 (ORM1) [13]. This finding evidenced a retinal inflammatory response in individuals with PDR in which pro-inflammatory cytokines entered the AH via either the vasculature or the VH [13]. Another proteomic analysis compared the AH of patients with DR to that of patients without diabetes undergoing cataract surgery [65]. Following 2D-DIGE and MALDI-TOF MS, this study identified eleven differentially expressed proteins, three of which are associated with inflammation: apolipoprotein A-I (APOA1), selenoprotein P (SELENOP), and cystathionine β-synthase (CBS). Selenoprotein P plays an important role in maintaining oxidative balance, and was strongly downregulated in the AH of DR patients [140]. Cystathione β-synthase was highly upregulated in DR patients; this enzyme is responsible for the synthesis of hydrogen sulfide and is linked to inflammation and cellular apoptosis [141]. Chiang et al. conducted a study focusing on the proteomic differences of AH from diabetic patients with and without DR [65]. Inflammation-related protein, apolipoprotein A-I, and a number of other proteins related to angiogenesis, structural remodeling, and oxidative stress were identified [65]. Similar to the findings in VH, overall protein concentrations were higher in the AH from DR patients as compared to patients without DR [65].

Tears
Tears are an aqueous solution of proteins, lipids, and other components, and their proteome has been shown to respond to insults such as cataract surgery [17,68,[142][143][144]. Tear sampling can provide pathological information in retinal disease and is the least invasive method of biofluid collection, offering a potential method for diagnosis and pre-screening [17,68,69]. The tear proteome of DR patients has become the special focus of several recent studies since the first study by Herber et al. in 2000 [145].
Csosz et al. created a methodology for identifying DR biomarkers using MS analysis of pooled tear samples from healthy controls and diabetic patients at various stages of retinopathy (no DR, NPDR, and PDR) [17,104]. A general decrease in protein content was observed with DR onset, perhaps resulting from defective tear formation or more diluted tears [17]. Furthermore, several proteins were identified as being differentially expressed, including the inflammation-related protein lactotransferrin (LTF) which was significantly upregulated in DR [17]. Lactrotransferrin, along with the other five candidate biomarkers identified by this study (lipocalin 1 (LCN1), lacritin (LACRT), lysozyme C (LYZ), lipophilin A (SCGB1D1), and immunoglobulin λ chain (IGLC1)), were used to develop a machine learning model for diagnosis based on both the combined expression of these and other proteins and clinical images of the retina [17,68,69,104]. In a separate study, β-2-microglobulin (B2M) evidenced a decreased expression in DR patients relative to both healthy controls and diabetic patients without DR [104,146]. In addition to proteomic analysis, tear samples have also been used to examine glycomic differences between DR patients and healthy controls [59,104].

Post-Translational Modifications as Biomarkers of Diabetic Retinopathy
High mass accuracy instruments have enabled the identification of post-translational modifications (PTMs) in proteins [147,148]. Recent studies have examined the role of protein acetylation and glycosylation in DR pathogenesis [58,59]. MS proteomic approaches have shown that hyperglycemia induces histone acetylation in diabetic rat retinas, which correlates with increased retinal expression of the pro-inflammatory proteins ICAM-1, iNOS, and VEGF [58]. These changes in both histone acetylation and pro-inflammatory cytokine expression were significantly inhibited by a drug used in DR treatment (minocycline) as well as by a histone acetyltransferase inhibitor (garcinol) and histone deacetylase agonists (theophylline and resveratrol) [58]. The correlation between histone acetylation and the expression of pro-inflammatory cytokines suggests that such PTMs and epigenetic regulation may play a role in DR development and progression.
In another study, the inflammation-related proteins antithrombin III (ATIII), clusterin (CLU/APOJ), osteopontin (OPN), and vitronectin (VTN) were found to be present in truncated forms as indicated by unexpectedly low molecular weights obtained via SDS-PAGE/LC-MS/MS. Similarly, a glycomic study on tear proteins found that one O-linked glycan and five N-linked glycans evidenced differences between DR patients and healthy controls [59]. The specific proteins that were glycosylated by these polysaccharides, however, were not identified by the study [59,104].

Limitations
The primary limitations of large-scale proteomic analysis of VH and AH is the difficulty in obtaining adequate numbers and volumes of samples, as well as the lack of true controls [63]. Obtaining vitreous and aqueous humor samples requires highly invasive procedures that should only be undertaken when necessary for the patient's health [4,61]. In this way, it is impossible to obtain AH and VH samples from completely healthy living human eyes [4,61]. For these reasons, much recent focus has been turned towards proteomic screening of tear and serum, both of which can be obtained through non-invasive or minimally invasive procedures.
Different pathological states may cause changes in the total protein content, as retinal vascular leakage is known to increase the amount of proteins present in the VH and the AH by introducing non-native proteins [2,4,5,22,39,57,65,81,149]. Furthermore, highly abundant proteins like serum albumin, hemoglobin, and crystallins can mask the detection of lower abundance proteins [3,16,20,22,57,63,70,149]. Depletion of the most abundant proteins using column chromatography can ameliorate some of these issues and enrich proteins that are less abundant.

Future Directions
Proteomic studies examining the ocular biofluids of diabetic patients and diabetic animal models have made an important contribution to the advancement in our understanding of DR pathogenesis. As high-throughput protein analysis techniques have continued to improve, these studies have great future potential for both biomarker discovery and identification of novel therapeutic targets. It is now becoming possible to accurately characterize the proteomes of very small biological samples, making the use of AH and VH from both patients and diabetic animal models more feasible. Greater dynamic range makes it easier to reliably detect very low abundance proteins, which is particularly important for the use of serum samples in the study of DR [150]. Technological improvements have also expanded our ability to detect post-translational modifications of proteins, making it possible to evaluate differences in glycosylation, phosphorylation, and acetylation states, providing more functional context to proteomic differences between healthy and disease states [151].
Lastly, it is important that these wide-scale proteomic analyses are not limited to the identification of diagnostic biomarkers and therapeutic targets. Proteomic and other "-omic" analyses are also valuable tools to screen the effects of novel therapeutic agents. For example, phlorizin, a naturally occurring compound found in fruit trees and known to have anti-inflammatory and antioxidant effects, was used to treat C57BLKS/J db/db mice [9,152,153]. An iTRAQ proteomic analysis of retinal tissue showed changes in the expression of proteins involved in inflammation, apoptosis, and oxidative stress, which corresponded to functionally validated decreases in retinal cell apoptosis and injury [9]. In conclusion, high-throughput proteomic analyses offer a promising future for the discovery of diagnostic, prognostic, and theraputic monitoring biomarkers, as well as for the development of new therapies for this sight-threatening disease.