Anti-inflammatory therapy for diabetic retinopathy

Abstract

Diabetic retinopathy (DR) is one of the most common complications of diabetes. This devastating disease is a leading cause of blindness in people of working age in industrialized countries and affects the daily lives of millions of people. Despite tight glycemic control, blood pressure control, and lipid-lowering therapy, the number of DR patients keeps growing and therapeutic approaches are limited. Moreover, there are significant limitations and side-effects for the current therapies. Thus, there is a great need for development of new strategies for prevention and treatment of DR. Studies have shown that DR has prominent features of chronic, subclinical inflammation. This review will focus on the role of inflammation in DR and summarize the progress of studies of anti-inflammatory strategies for DR.

According to the International Diabetes Federation (IDF-Atlas, 4th ed., 2009), diabetes currently affects nearly 285 million people worldwide and this number is expected to reach 438 million by 2030. Diabetic retinopathy (DR) is one of the most common complications of diabetes and is a leading cause of blindness in people of working age in industrialized countries [1]. Approximately 25% of type 1 diabetic patients may have signs of retinopathy after 5 years of diabetes, increasing to 60% after 10 years. Eventually, after 25 years, almost all (97%) type 1 diabetics will develop retinopathy [2, 3]. Type 2 diabetic patients may already have background retinopathy at time of diagnosis and over 60% will develop some form of retinopathy after 20 years [4]. In the USA, according to the Wisconsin Epidemiologic study of Diabetic Retinopathy (WESDR), DR causes 12,000 –24,000 new cases of blindness with costs of more than $500 million each year [5]. The number of patients with DR was 5.8 million in 2005. However this number will triple to 17.7 million in 2050 [5]. In China, it is estimated that 9.2 million Chinese people living in rural areas have DR and 1.2 million of them have vision-threatening retinopathy [6]. With changes in life style and increases in life span along with the global prevalence of diabetes, it is expected that DR will have a growing impact on large populations.

Tremendous efforts have been made to identify mechanisms and develop therapies for DR. These studies have led to the recognition of hyperglycemia, hypertension and dyslipidemia as major risk factors for DR. Consequently, tight glycemic control, blood pressure control, and lipid-lowering therapy have shown proven benefits in reducing the incidence and progression of DR [6]. Progress in developing new approaches for ocular therapy is relatively slow. Laser photocoagulation and vitrectomy remain as the two conventional approaches to treat sight-threatening conditions such as macular edema and proliferative diabetic retinopathy [7]. Anti-vascular endothelial growth factor (VEGF) therapy represents a recent breakthrough since clinical trials have shown beneficial effects of VEGF blockers (Pegaptanib, Ranibizumab, Bevacizumab) in reducing macular edema and causing neovascular regression particularly when combined with laser photocoagulation [6–8]. In spite of this progress, DR remains a major clinical challenge and the number of patients keeps growing. It is difficult to achieve tight glycemic control throughout the course of the disease. Moreover, DR can develop even after tight control is initiated due a phenomenon called “metabolic memory” in which the retinal endothelial cells manifest high glucose-induced biochemical alterations long after the initial insult [9, 10]. Laser photocoagulation and anti-VEGF therapy are not always effective and laser therapy can damage retinal neurons which may result in decreases in central vision and vision acuity and impaired night vision [6, 7]. Anti-VEGF therapy requires repeated treatment and may impair neuronal and vascular survival function [11]. Thus, there is a great need to develop new therapeutic approaches for this devastating disease.

The clinical progression of DR follows a pattern characteristic of ischemic retinopathy [2, 12]. The disease begins with biochemical and cellular alterations that are not clinically evident. Vascular alterations during this period include alterations in blood flow, death of retinal pericytes (perivascular contractile cells), basement membrane thickening and subtle increases in vascular permeability. As the disease progresses, obvious alterations in the vascular structure can be seen upon ophthalmoscopic examination. These include nonperfused vessels, microaneurysms, dot/blot hemorrhages, cotton-wool spots, venous beading, vascular loops, and significant vascular leakage. In many patients, the retinopathy progresses to proliferative DR, in which the new vessel walls are weak and allow the blood to leak out of the vessels, resulting in vitreous hemorrhage and subsequent retinal detachment. The mechanisms by which elevated blood glucose causes tissue injury and disease progression in the retina are not fully understood. However, studies have shown that DR is a multi-factorial disease involving multiple pathways, including aldose reductase pathway, oxidative stress, activation of protein kinase C, and formation of advanced glycation end products (AGEs). These pathways lead to retinal pathological changes by causing osmotic vascular damage, inducing cell dysfunction and apoptosis through activation of mitogen-activated protein kinases (MAPKs) and oxidation of intracellular components, inducing production of angiogenic cytokines, and breakdown of the vascular junction proteins. [13–18]. In addition to these pathways, recent studies indicate that inflammation is a critical player in the pathogenesis of DR and serves as a positive feedback mediator of multiple pathways.

Diabetic retinopathy has features of chronic inflammatory disease

DR has not been recognized as an inflammatory disease until recently because the retina has been considered as a tissue with immune privilege. However, recent studies in patient samples and animal models have shown that DR has features of chronic, subclinical inflammation.

Inflammation

Inflammation is the body’s defense against pathogens and also a critical step in wound healing. This process involves multiple mediators such as pro-inflammatory cytokines, chemokines and adhesion molecules that initiate the interaction between leukocytes and the endothelium and guide directional leukocyte migration towards infected or injured tissue. Pro-inflammatory cytokines (such as tumor necrosis factor (TNF)-α, interleukins) and chemokines (such as CCL2, CCL5) released from infected/injured tissue activate the endothelium to increase expression of adhesion molecules (such as E-selectin, intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule (VCAM)-1) and chemokines. The adhesion molecules make the endothelium “sticky”, causing the leukocytes to roll along the endothelium until they meet glycosamino-glycan-immobilized chemokines. After activation by a chemokine, leukocytes transition from selectin-mediated rolling to integrin-mediated firm attachment. The attached leukocytes then transmigrate through the endothelium, moving to interstitial inflammatory sites by following chemotactic gradients [19]. Whereas acute inflammation is beneficial, chronic and dysregulated inflammation can result in undesirable effects and is critically involved in a variety of diseases such as such as atherosclerosis, rheumatoid arthritis, multiple sclerosis, and asthma [19].

DR and inflammatory molecules

Typical clinical signs of DR such as edema and neovascularization are hallmarks of inflammation and suggest a possible role of inflammation in the pathology. The potential role of inflammatory mediators in DR has been supported by extensive evidence in the past two decades. TNF-α is a pro-inflammatory cytokine which is recognized as an initiator or inflammatory reactions. High TNF-α levels have been observed in vitreous, serum, and ocular fibrovascular membranes from patients with DR and in retinas from rodent model of diabetes mellitus [20–22]. TNF-α gene polymorphism has been shown to be associated with increased susceptibility to the disease [23]. Similarly, interleukin-1β (IL-1β) and its downstream signaling molecule caspase 1 are high in vitreous and retinas from diabetic patients and rats [21, 24, 25]. Interleukin-6 (IL-6) levels in the vitreous are significantly correlated with the severity and progress of DR [26, 27]. Chemokines such as CCL2, CCL5, CXCL8, CXCL10, CXCL12 are also upregulated in vitreous samples from DR patients [28–30]. Levels of TNF-α, IL-1β, CCL5 and CXCL12 are also significantly increased in the serum of DR patients [21, 30], suggesting that systematic inflammatory reactions may contribute to this disease. In addition to increases in these inflammatory mediators, increases in adhesion molecules such as ICAM-1 and VCAM-1 have been found to be related to the progression of DR [30, 31]. ICAM-1 immunoreactivity is high in the diabetic vasculature of humans and rodents and the K469E polymorphism of the ICAM-1 confers susceptibility to proliferative DR [30, 32–34].

DR and inflammatory cells

As stated above, a typical sign of inflammation is recruitment of leukocytes via inflammatory cytokine-induced activation of the vascular endothelium. Leukocytes, the cell class responsible for innate immunity (neutrophils, monocytes, eosinophils, basophils, and mast cells) and adaptive immunity (T lymphocytes, B lymphocytes, and natural killer cells), effectively protect the host from infectious pathogens but also induce tissue injury in certain diseases [35]. Among these cells, neutrophils and monocytes are very likely to induce microvascular injury because they have high probability to attach to the vessel wall and block blood flow due to their large population and size. Furthermore, they produce high levels of reactive oxygen species (ROS), which can injure vascular cells. The evaluation of the action of leukocytes in retinas of DR patients has been limited by the limited availability of human retinal tissue. Lutty and colleagures reported the very interesting observation that numbers of neutrophils are significantly elevated in both retinal and choroidal vessels from diabetic patients. The increase in neutrophils was found to be correlated with upregulation of ICAM-1 immunoreactivity in the vessels [33]. Monkey represents an excellent DR model that mimics the clinical aspects of human DR. In a study of retinas from diabetic monkeys, Lutty’s group provided further evidence that neutrophil accumulation is associated with capillary closure, further implicating an involvement of neutrophils in this disease [36]. Diabetic rodent models share some characteristics of human DR such as increased retinal vessel permeability, vessel closure and neuronal death. Although rodents do not develop retinal neovascularization as seen in human diabetes, rodents are widely used to study mechanisms of DR owing to accessibility and possibilities for genetic manipulation. Schroder et al. conducted a detailed morphological study in diabetic rats and provided the first evidence that attachment of monocytes and neutrophils is associated with capillary occlusion, dropout and intraretinal neovascularization [37]. Subsequent studies further demonstrated a cumulative and sustained increase in leukocyte adherence to the retinal vasculature (leukostasis) along with the progression of DR [38]. The increase in leukostasis may be attributable to diabetes-induced increases in ICAM-1 and integrins in endothelial cells and leukocytes, respectively [38–40]. Blockade of ICAM-1 or deletion of CD18 (ICAM-1 receptor subunit on leukocytes) prevents diabetes-induced leukostasis. In addition to leukocytes in the circulation, microglial cells in the retina are likely to play role in DR. The microglial cell is a resident macrophage. Microglia are scattered throughout the retina and form a network with potential immunoeffector cells [41]. In DR they are rapidly activated to release inflammatory cytokines such as TNF-α [42], indicating a local immune reaction in the disease. The role of microglia in DR is further supported by a recent study which showed that A2A adenosine receptor agonist effectively blocks TNF-α production in microglia and markedly decreases hyperglycemia-induced retinal cell death (Ibrahim et. al., 2011). Whereas multiple mechanisms may be involved in diabetes-induced microglial activation, studies indicate that the formation of amadori-glycated proteins has an essential role in microglial activation and this process involves ROS-mediated activation of extracellular signal-regulated kinase and p38MAPK (Ibrahim et. al., 2011).

Overall, this clinical and pre-clinical evidence demonstrates the inflammatory nature of DR and suggests involvement of low-grade chronic inflammation in this disease.

Role of inflammation in the pathogenesis of diabetic retinopathy

DR is clinically classified into two stages: non-proliferative DR and proliferative DR [43]. Non-proliferative DR is an early stage of DR in which vision loss is mainly caused by macular edema as the consequence of increased vascular permeability due to blood-retina barrier breakdown. Proliferative DR is a more advanced stage of DR which is characterized by neovascularization on the retinal surface elicited by hypoxia after vascular function is impaired by capillary occlusion, non-perfusion and degeneration. Vision loss in proliferative DR can be also caused by macular edema as occurs in non-proliferative DR. However, a unique feature of proliferative DR is that severe vision loss or even blindness may be caused by bleeding, hemorrhage and subsequent retinal detachment because of the newly formed fragile vessels [43]. Since inflammation is found as a common feature of DR, significant efforts have been made to understand how inflammation is involved in the pathogenesis of the disease. These studies have disclosed important roles of inflammatory cytokines and leukocytes in vascular leakage, vessel occlusion and degeneration, pathological neovascularization (Fig. 1), and neuronal death.

Figure 1. Role of inflammation in vascular disease in diabetic retinopathy.

During inflammation, pro-inflammatory cytokines and chemokines can induce BRB breakdown and vascular cell death, resulting vessel leakage and degeneration. Alternatively, such changes can be induced by inflammatory cytokine, chemokine and adhesion molecule-mediated leukocyte recruitment. The interaction between leukocytes and vascular cells induce BRB breakdown, vascular cell death and subsequent vessel leakage and capillary degeneration. In addition, leukocytes release MMPs and angiogenic factors to induce pathological neovascularization.

Abbreviations: BRB = blood-retinal barrier, ICAM-1 = intercellular adhesion molecule 1, IL-1β = interleukin-1β, MMP = matrix metallopeptidase, TNF-α = tumor necrosis factor alpha, VEGF = vascular endothelial growth factor.

Inflammation and vessel leakage

The microvascular endothelium forms an effective barrier to control the movement of blood fluid and proteins across the vessel wall [44]. This barrier is formed by tight junctions and adhesion junctions that join endothelial cells to each other. The tight junction structure consists from a number of transmembrane proteins including occluding claudins, junction adhesion molecules, zonula occludens (ZO), and several regulatory proteins [45]. Adherens junctions are formed by vascular endothelial (VE)-cadherin and catenins [44]. Signaling molecules that induce disorganization/redistribution of junction proteins in the retinal endothelium may lead to breakdown of the blood-retinal barrier (BRB), resulting in an abnormal extravasation of blood components and retinal edema. Endothelial cells can respond to most pro-inflammatory cytokines and chemokines to initiate inflammatory reactions. However, signaling events induced by these cytokines/chemokines may exert multiple functions other than induction of leukocyte adhesion molecule expression. TNF-α is known to cause significant retinal endothelial permeability within a few hours by a nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) and protein kinase C (PKC)ζ-mediated downregulation of both claudin-5 and ZO-1 expression in the absence of any effects on apoptosis [18]. Consistent with the in vitro data, intravitreal injection of TNF-α leads to increased retinal vascular permeability which is prevented by PKCζ inhibitor [18]. Blockade of TNF-α with Etanercept, a soluble TNF-α receptor, inhibits NF-κB activation and prevents BRB breakdown in the diabetic rat model [20]. Similar increases in endothelial permeability are found when endothelial cells are activated by other cytokines such as IL-1β, IL-6, CCL2 or CXCL8 although the mechanisms may be distinct. CCL2 induces endothelial barrier breakdown via PKCα/ζ-mediated phosphorylation and translocation of occludin, claudin-5, ZO-1 and ZO-2 [46]. In contrast, CXCL8 causes endothelial permeability by inducing VE-cadherin internalization through phosphatidylinositol 3-kinases/Rac/p21 activated kinase signaling axis [47]. However, it should be noted that the in vivo effect of inflammatory molecules in inducing BRB breakdown may be more complicated than their direct effects in altering junction proteins in endothelial cells since many cytokines, such as TNF-α and IL-1β, can also cause cell death when chronically applied. Additionally, they may work in a synergistic or additive manner in DR since multiple inflammatory molecules are simultaneously increased. More importantly, leukocytes recruited in inflammation apparently have a key role in exaggerating the vascular leakage. VEGF is a potent vascular permeability factor that can directly stimulate a breakdown of endothelial permeability barrier function. However, its in vivo effect on retinal vessel permeability also involves leukocytes because blocking ICAM-1 not only reduces leukostasis but also blocks VEGF-induced vascular leakage [48]. The key role of leukostasis in breakdown of the blood-retinal barrier is also demonstrated in diabetic rat and mouse models [38, 39]. The permeability-inducing actions of leukostasis may be attributed to their effects in inducing junction protein alteration, releasing cytokines or increasing ROS, as well as inducing vessel occlusion and injury [38, 39, 49].

Inflammation and vessel closure

Retinal vessel occlusion and degeneration is a typical feature of DR and is also a cause of neovascularization. Mechanisms leading to capillary degeneration may involve inflammatory cytokine-induced endothelial cell death since inflammatory cytokines as TNF-α and IL-1β are also known to increase caspase 3 activity and potently induce endothelial cell apoptosis [18, 50]. The apoptotic effect of inflammatory cytokines may even be exaggerated in the presence of hyperglycemia [50]. However, this mechanism does not explain why the appearance of non-perfused or regressed vessels in DR or in inflammatory cytokine-treated retinas is random instead of coordinated since vascular cells in adjacent vessels should be equally exposed to inflammatory cytokines. Thus, other mechanisms may be operative in this process. Considering the large cell volume and high rigidity of leukocytes, it is likely that leukostasis in DR may cause capillary occlusion. Supporting this possibility, analysis of diabetic retinal vessels reveals that capillary nonperfusion is closely associated with the presence of leukocytes [36, 37]. By studies using real-time analysis of leukostasis and capillary perfusion, Miyamoto et. al. provided further evidence of the temporal and spatial association between leukostasis and capillary nonperfusion [38]. Their study showed that transient leukocyte attachment to the vessel wall correlates with capillary nonperfusion and capillary reperfusion can occur when leukocytes detach and move on. However, some non-perfused vessels remain permanently closed after the leukocyte detachment [38], suggesting that leukocytes may cause vascular cell death during leukostasis. Further studies indicate that the level of Fas is increased in the retinal vascular endothelial cells during diabetes while its ligand Fas Ligand (FasL) is increased on neutrophils [51]. This change enhances the ability of neutrophils to induce endothelial cell death via Fas/FasL-mediated apoptosis [51]. Blockade of this pathway by anti-Fas neutralizing antibody reduces endothelial cell injury and apoptosis without altering leukostasis [51]. Together, these studies indicate that leukostasis has a key role in causing retinal vascular occlusion and injury in diabetes. Consistent with this possibility, blocking ICAM-1/CD18-mediated leukostasis significantly reduces retinal capillary occlusion and degeneration [39].

Inflammation and pathological neovascularization

Chronic inflammation is known to play a key role in angiogenesis in many diseases including rheumatoid arthritis and cancer where upregulated inflammatory cytokines in the tissue induce the infiltration of leukocytes. Leukocytes, in turn, actively enhance the formation of new vasculature by releasing angiogenic factors such as VEGF, angiopoietin 1, basic fibroblast growth factor, platelet-derived growth factor and transforming growth factor (TGF)-β. These agents promote recruitment of endothelial progenitor cells and enhance the survival, migration, and proliferation of pre-existing endothelial cells. They also induce matrix metallopeptidase (MMP) activity which promotes degradation of extracellular matrix [52]. A role of inflammation in proliferative DR has been suggested given that a number of inflammatory cytokines are known to increase in the disease. Nevertheless, direct investigation of inflammation in proliferative DR is hindered due to the lack of a small animal model that develops reproducible proliferative DR as occurs in human. Therefore, studies of retinal neovascularization are carried out in rodent models of oxygen-induced retinopathy by inducing retinal vessel obliteration with hyperoxia. These studies have shown that inflammatory genes are upregulated at the onset of the hypoxia as well as during the period of neovascularization [53, 54]. TNF-α levels are highly increased and blockade of TNF-α by genetic gene deletion or TNF-α receptor fusion protein (Etanercept) prominently attenuates neovascularization [55, 56]. Immunolabeling study reveals that monocyte-lineage cells are the major source of TNF-α [55], suggesting that these cells may be involved in pathological neovascularization. This possibility is supported by the finding of Ishida et al. that monocytes/macrophages are present in the neovascular tufts that grow into the vitreous and that depletion of cells of the monocyte lineage with clodronate-liposomes leads to the suppression of pathological but not physiological retinal angiogenesis [57]. However, it should be noted that some monocyte lineage cells present in oxygen-induced retinopathy are not pathogenic but instead, are necessary for removing neovascular tufts. For example, CCL2 deficiency reduces monocyte/microglia recruitment and delays neovascular tuft regression but does not prevent neovascularization [58]. Thus, although there is no doubt that inflammation is a critical player in pathological neovascularization, careful attention must be given when selecting inflammatory molecules as a therapeutic target.

Inflammation and retinal neuronal death

Inflammatory cytokines such as TNF-α and IL-1β are also proapoptotic proteins that are known to cause neuron death in many diseases [59, 60]. These cytokines, together with ROS and excessive nitric oxide (NO) released by leukocytes during inflammation, are key players in retinal neuron cell death [59, 61–64]. Blockade of leukocyte recruitment by ICAM-1 inhibition or deleting CD18 significantly suppresses leukocyte accumulation and retinal ganglion cell death in N-Methyl-D-aspartic acid-induced retinal neuropathy [65]. Presumably, similar mechanisms may be operative in DR induced neuron injury. Neuronal cell injury was first suggested as an early feature of diabetic retinopathy in the early 1960’s but has only recently been widely recognized. Barber et al. reported in 1998 that retinal thickness and ganglion cell numbers are reduced while neural cell apoptosis is significantly increased in a Streptozotocin-diabetic rat model [66]. However progress in this field has been very slow. A search of PubMed for “DR” in combination with “neural cell death” or “neuron death” or “ganglion cell death” produced only 75 papers about this topic. Among these papers, one study showed that salicylate-based anti-inflammatory drugs attenuate diabetes-induced loss of ganglion cells [67]. However, there is no direct evidence to demonstrate the impact of inflammatory cytokines and leukocytes on neuron death in DR. More investigations are needed to fully address this issue.

Mechanisms underlying retinal inflammation in diabetic retinopathy

Tremendous efforts have been made to understand how hyperglycemia causes inflammatory reactions because of the fundamental role of hyperglycemia in DR. These studies have demonstrated links between hyperglycemia-induced oxidative stress, dysregulation of nitric oxide synthase (NOS), formation of AGEs and production of inflammatory molecules and leukostasis in DR. In addition, recent studies also show that two other risk factors for DR, hypertension and dyslipidemia, independently link to inflammation and/or exaggerate the inflammatory reactions induced by hyperglycemia.

Oxidative stress

Oxidative stress happens when ROS are over produced or when endogenous anti-oxidant systems are impaired. Mitochondria have long been recognized as a key source of ROS formation during diabetes [2]. Mitochondria can generate ROS by leak of electrons to molecular oxygen at electron transport chain (ETC) complexes I, II and III [68]. In diabetes, the metabolism of glucose-derived pyruvate through the ETC complexes is increased because of high glucose concentration within cells, resulting in superoxide overproduction by mitochondria [69]. This pathway not only produces more ROS by itself but also initiates other pathways leading to a breakdown in the balance between pro-oxidant and endogenous anti-oxidant systems, such as increases in glucose flux through the aldose reductase pathway, formation of AGEs, and activation of PKC [70]. All of these changes can lead to oxidative stress by decreasing activities of anti-oxidant enzymes [71, 72] or further activating the ROS generating machinery inside the cells [73–77]. This mechanism has been thought as a dominant cause of oxidative stress in diabetes. However, data showing increased activity of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase in diabetic patients and animals and high glucose-treated endothelial cells suggest that NADPH oxidase is also an important and early source of ROS [78–81]. NADPH oxidase is the major source of ROS in the cardiovascular system. It consists of a NOX catalytic subunit, 4 phox subunits (p22phox, p40phox, p47phox and p67phox) and the low molecular weight G protein Rac [82]. Although NOX and p22phox form a flavocytochrome that contains all the catalytic machinery needed for superoxide production, the other subunits are required for regulation of enzyme activation. The p47phox and p67phox subunits are essential for activity. Assembly of the complex is initiated upon phosphorylation of p47phox, which is regulated by many intracellular signaling kinases such as PKC, MAPKs and Akt [83–87]. The p67phox subunit mediates direct binding of the complex with activated Rac, which initiates the electron transfer reaction that produces superoxide [88]. These regulatory mechanisms contribute to the unique feature of NADPH oxidase-derived ROS as signaling mediator in many cellular responses. However when these mechanisms are activated because of diabetes, they can cause NADPH oxidase to overproduce ROS. It is unclear, whether mitochondrial ETC or NADPH oxidase initiate the excessive ROS formation in diabetes. However, studies in other models have suggested a positive reciprocal regulation between them [89, 90]. Overall, diabetes-induced imbalance of the pro-oxidation and anti-oxidation results in sustained oxidative stress in the body including retina.

ROS are important intracellular signaling molecules in the inflammatory cascade. ROS play a key role in inflammatory gene expression by activation of transcription factors such as NF-κB, Signal Transducers and Activators of Transcription proteins (STAT) and activator protein 1 (AP-1) [91]. Among these factors, NF-κB is the key transcription factor in most inflammatory reactions. NF-κB is a transcription factor that is ubiquitously expressed and which regulates numerous genes related to inflammation and immune responses [92]. It exists in homo- or heterodimer forms that bind to κB-binding sites in gene promoters to regulate gene transcription at least partly by interacting with transcriptional coactivators such as p300/CBP. NF-κB also cooperates with other transcription factors such as the constitutive transcription factor specificity protein 1 and the inducible transcription factors, STATs, cAMP response element-binding, and AP-1, to achieve stimulus- and promoter-specific transcriptional regulation [93]. The NF-κB dimer is sequestered in the cytosol in an inactive form when it binds to inhibitor of κB (IκB) proteins. These IκB proteins mask its nuclear translocation signal and inhibit its DNA binding activity [93, 94]. When IκB is ubiquitinated and degraded following phosphorylation by IκB kinases, it releases NF-κB resulting in NF-κB activation [94]. NF-κB and many kinases involved in its direct or indirect activation are redox sensitive molecules [95], allowing it to serve as the downstream target of ROS and mediate ROS-induced inflammatory gene expression.

In physiological conditions, the production and elimination of ROS are precisely regulated and will not produce unnecessary inflammation in the body. However, diabetes-induced sustained oxidative stress is a major cause of retinal inflammation. The activity of NF-κB is increased in high glucose treated retinal endothelial cells, pericytes or glial cells, and in diabetic retinas from patients or animal models [96–101]. This activation is significantly blocked by inhibiting redox systems by blockade of NADPH oxidase or anti-oxidant treatment [100, 101], indicating a cause-and-effect relationship. Activation of NF-κB has an essential role in diabetes-induced retinal inflammation in that inhibition of NF-κB blocks both high glucose and diabetes-induced production of inflammatory molecules in retinal cells and retinal tissue, respectively [96–98]. Correlated with decreased cytokine formation, leukostasis is also attenuated by inhibition of NF-κB [96]. Similarly, studies in animal and tissue culture models have demonstrated the involvement of NADPH oxidase in diabetes-induced inflammation and breakdown of the BRB [34, 102]. The results showed that diabetes-induced increases in retinal ROS, VEGF expression and vascular permeability are accompanied by increases in the NADPH oxidase catalytic subunit NOX2 [34, 102, 103]. The experiments further demonstrated that deleting the NOX2 gene or inhibiting NADPH oxidase prevented diabetes-induced increases in retinal permeability and ICAM-1 expression and leukostasis, indicating that the NOX2 NADPH oxidase is critically involved in the pathology of DR [34, 102, 104].

NOS dysregulation

In addition to oxidative stress, diabetes-induced vascular inflammation is very closely related to nitric oxide (NO) which is an important second messenger that regulates many physiological and pathological events, including vascular dilation and vascular inflammation. Studies indicate that NO has a biphasic role in vascular and inflammatory diseases depending on the specific source and the amount produced [105]. The constitutively expressed NOSs, endothelial NOS (eNOS) and neuronal NOS (nNOS), are Ca2+-dependent and regulated to produce low levels of NO. NO from nNOS is involved in neural signaling and is also expressed in smooth muscle cells where it has a key role in regulating vascular responses to tissue hypoxia. NO from eNOS maintains blood flow and prevents platelet aggregation and leukostasis [106]. In fact, deleting eNOS results in elevated expression of inducible NOS (iNOS) (an NF-κB mediated inflammatory molecule) in retina which is associated with accelerated development and increased severity of DR [107]. In contrast with eNOS and nNOS, iNOS is Ca2+-independent, constitutively active and produces large amounts of NO [105]. It is not expressed in normal retinas, but is induced in retinal glial and microglial cells during inflammatory conditions. including DR. NO from iNOS plays a role in causing tissue damage and inflammation. Inhibition of iNOS by inhibitor or gene deletion prevents ICAM-1 expression, leukostasis, and vascular permeability in the diabetic retina [108]. One possible reasons for the opposite function between eNOS and iNOS is that NO produced at low levels by eNOS mainly exerts its effects directly by interactions with metal-containing proteins to maintain the normal vascular function. In contrast, high levels of NO produced by iNOS can act indirectly via the generation of reactive nitrogen oxide species (RNOS). RNOS are formed through the interaction of NO with molecular oxygen or superoxide. RNOS is strong pro-inflammatory molecule and can cause tissue damage due to lipid peroxidation, DNA damage, oxidation of thiols and nitration of tyrosine residues [109, 110].

The NOS pathway serves to link multiple pathways. On the one hand, ROS generated from NADPH oxidase or mitochondrial oxidase can rapidly react with NO to form RNOS and reduce bioavailable NO which reduces vessel dilation and increases leukocyte adhesion and platelet aggregation, leading to inflammation [105, 111]. On the other hand, NOS can be an important source of ROS due to “NOS uncoupling” in which the enzyme generates superoxide rather than NO when its substrate L-arginine is limited by increased arginase activity [105] or when the cofactor tetrahydrabiopterin is oxidized. NOS uncoupling can contribute to disease progression by multiple mechanisms. Uncoupling of eNOS can contribute to ischemia/hypoxia and tissue damage by reducing EC-dependent vasorelaxation and inducing platelet aggregation and leukostasis and also by impairing function in endothelial progenitor cells [105]. nNOS uncoupling could exacerbate the vascular damage by altering smooth muscle cell responses to hypoxia [105]. Uncoupling of iNOS could increase formation of inflammatory mediators by increasing NF-κB transcriptional activity due to decreased nitrosylation since NO is anti-inflammatory molecule which dampens inflammatory reactions by causing S-nitrosylation of the NF-κB subunits p65 and p50 [105, 112, 113]. This reduces their DNA binding which in turn decreases transcription of iNOS and other inflammatory mediators [112, 113]. Consistently, inhibition of arginase is shown to increase bioavailable NO, reduce superoxide formation, and block inflammatory reactions during retinal inflammation [104].

AGEs formation

The impact of hyperglycemia in diabetes-induced retinal inflammation is not only attributed to the direct effect of high glucose but also a consequence of the byproducts of hyperglycemia, particular AGEs. AGEs are formed by nonenzymatic glycation and oxidation of amino groups of proteins, lipids, and DNA [114]. This process is accelerated in diabetes in the presence of high glucose and oxidative stress [14, 114, 115]. In the diabetic retina, AGEs levels are prominently increased and AGE immunoreactivity is localized in vitreous, internal limiting membrane, and retinal vasculature [114]. The effect of AGEs is mostly mediated by the receptor for advanced glycation endproducts (RAGE), a multiligand signal transduction receptor of the immunoglobulin superfamily that has been identified in numerous cells including Muller cells, endothelial cells, and neurons in the retina [114]. Mediated by RAGE, AGEs can activate many downstream signaling molecules including NF-κB and ROS to induce inflammatory reactions such as ICAM-1 expression and leukostasis in retina and retinal cells [115, 116]. Unlike other inflammatory molecules, RAGE-mediated NF-κB activation overrides the endogenous autoregulatory negative feedback loop and shows a prolonged time course [115]. This feature together with positive regulation of RAGE expression by NF-κB allows AGEs to elicit sustained inflammation reactions in DR. The role of AGEs-RAGE pathway in DR has been directly addressed by using soluble RAGE to block AGEs activity in diabetic wild type mice and transgenic mice with RAGE overexpression under the control of VEGF receptor 2 promoter [117]. Consistent with the notion that RAGE is an inflammatory mediator, normal glycemic RAGE transgenic mice display significant high levels of ICAM-1 expression, leukostasis and VEGF expression as compared with wild type mice. Interestingly, the above changes were prominently augmented in diabetic mice. However these increases were normalized to basal level by soluble RAGE treatment, indicating the AGEs-RAGE pathway is a key mediator of diabetes-induced retinal inflammation.

Hypertension

In addition to hyperglycemia, hypertension is recognized as the second key factor associated with the development of DR. Although the mechanisms by which hypertension increases the risk of DR are not fully understood, hypertension can certainly contribute to the pathogenesis of DR by inducing inflammatory reactions. The spontaneously hypertensive rats (SHR) exhibit a significant increase in monocyte/microglial number, ICAM-1 and VEGF expression, and NF-κB protein level in retina in both prehypertensive and hypertensive conditions as compared with normotensive control rats [118]. These increases are further boosted when the SHR rats were made diabetic with Streptozotocin [118], suggesting that inflammation is a common target of both hyperglycemia and hypertension. Mechanisms of hypertension-induced retinal inflammation are not completely clear. However, the over activation of renin-angiotensin system (RAS) may have a critical role.

The RAS plays a key role in the development and pathophysiology of hypertension by producing angiotensin II (Ang II), a well-known vasoconstrictive peptide [119]. However, it is now known that the function of Ang II goes far beyond its impact on systolic blood pressure. Ang II, by binding to the Ang II type 1 receptor (AT1R), potently induces NADPH oxidase activation, ROS formation, and activation of a varieties of kinases, phophatase and transcription factors [120]. Ang II-induced activation of these signaling pathways, especially ROS-modulated NF-κB activation, is critically involved in vascular inflammation by producing inflammatory cytokines and recruiting leukocytes [119, 120]. Further, the Ang II pathway interacts with other pathways to amplify the inflammatory cascades. Ang II signaling enhances AGEs formation by accelerating the process of nonenzymatic glycation and oxidation meanwhile inhibiting glyoxalase system-mediated degradation of AGE precursor [119–121]. Ang II also induces dysregulation of NOS pathways [120]. In retina, extensive studies have been performed to understand whether the RAS has a role in retinal inflammation and injury. These studies have identified that all of the components of RAS as well as Ang II receptors are expressed in retina [122, 123], suggesting an involvement of the RAS in the retinal pathophysiology. This possibility has been assessed by blockade of RAS in disease models or by intravitreal or systemic delivery of Ang II to healthy animals. In animal models of DR, levels of angiotensinogen and AT1R are significantly increased in the retina [96]. Blocking AT1R prevents NF-κB activation, ICAM-1 expression and leukostasis typically seen in DR [96]. In SHR, an AT1R antagonist (Losartan), also completely blocks hypertension-induced retinal inflammation [124]. In the diabetic and hypertensive transgenic (Ren-2) rats that display an enhanced extra-renal RAS, an AT1R antagonist (Candesartan) attenuates diabetes-induced retinal vascular inflammatory aspects including leukostasis, NOS dysregulation, and AGEs formation [121]. In contrast, Ang II treatment induces retinal inflammatory reactions including IL-6 production, ROS formation and leukostasis in healthy animals [125, 126]. These studies indicate the RAS is necessary and sufficient for the production of retinal inflammation in diseases such as diabetes and hypertension.

In addition to the RAS, vascular stretch and activation of endothelin may also have roles in hypertension-induced retinal inflammation [127, 128]. However, more evidence is needed to address this possibility.

Dyslipidemia

Dyslipidemia is another risk factor of DR. Although it is well-known that dyslipidemia is critically involved in macrovascular inflammation in atherosclerosis, its specific role in DR is largely unclear. A recent study shows that basal VCAM-1 expression is increased in retinal vessels in hyperlipidemic mice [129], suggesting that hyperlipidemia is cause of retinal vascular inflammation. However, it is unknown whether or not this change will result in leukostasis because the basal level of other inflammatory molecules is not increased in hyperlipidemic mice. In addition, hyperlipidemia does not further increase diabetes-induced expression of inflammatory molecules in these mice. Further investigation is needed to determine the role of dyslipidemia in retinal inflammation during DR.

Anti-inflammation pathway

In addition to the above pro-inflammatory pathways, diabetes may exaggerate retinal inflammation by impairing endogenous anti-inflammatory mechanisms such as the proliferator-activated receptor (PPAR)γ pathway [101]. PPARγ is a member of a ligand-activated nuclear receptor superfamily and is involved in many biological processes such as adipogenesis, glucose metabolism, and angiogenesis [130]. Synthetic PPARγ agonists such as pioglitazone and rosiglitazone reduce biomarkers of inflammation through mechanisms involving inhibition of NF-κB activation [131]. The PPARγ pathway is operative in DR given that PPARγ agonist blocks retinal inflammation and vascular leakage in diabetic rats whereas inhibition of PPARγ exaggerates signs of DR [132]. Studies also show that the PPARγ pathway is impaired in DR in a mechanism involving NADPH oxidase [101].

In summary, risk factors of DR such as hyperglycemia, hypertension and dyslipidemia are all involved in retinal inflammation in DR through a variety of mechanisms including oxidative stress, NOS dysregulation, AGE formation, RAS activation, and inhibition of endogenous anti-inflammatory pathways (Fig. 2). However, it should be noted that inflammation is not an endpoint of the above inflammatory mechanisms. Inflammation, in turn, further activates these inflammatory mechanisms and serves as a key mediator in the positive feedback loop of inflammatory cascades (Fig. 2). For example, oxidative stress is not only a cause of inflammation but also a consequence of inflammation. During inflammation, many inflammatory cytokines induce oxidative stress and employ the NADPH oxidase pathway to induce expression of other inflammatory molecules such as CCL2 [133]. Leukocytes also release massive amounts of ROS in response to inflammatory stimuli. Consequently, inhibition of inflammatory cytokines has been shown to prevent oxidative stress and retinal pathology in experimental models [126]. As a key factor in the diabetes-induced inflammatory cascade and one of the major causes of retinal damage, inflammation serves as a promising therapeutic target for the treatment of DR.

Figure 2. Mechanisms of diabetes-induced retinal inflammation.

Hyperglycemia induces oxidative stress, NOS dysregulation, and AGEs formation. These changes can lead to activation of pro-inflammatory transcription factors such as NF-κB, AP-1 and STATs and inhibit anti-inflammatory pathways, resulting in retinal inflammatory reactions. Retinal inflammation can also be induced by hypertension and dyslipidemia. Retinal inflammation is enhanced by the reciprocal positive regulation among oxidative stress, NOS dysregulation, and AGEs formation.

Abbreviations: AGEs = advanced glycation end products, AP-1 = activator protein 1, NADPH = nicotinamide adenine dinucleotide phosphate, NF-κB = nuclear factor kappa-light-chain-enhancer of activated B cells, ROS = reactive oxygen species, STATs = signal transducers and activators of transcription proteins.

New approaches for diabetic retinopathy: Can we stop inflammation

Since inflammation is identified as a critical mechanism for DR, significant effort has been made devoted to the development of new strategies for treatment or prevention of DR by targeting inflammation. Clinical and pre-clinical trials have tested a variety of anti-inflammatory therapeutics (Table 1).

Table 1.

Effect of anti-inflammatory therapy on diabetic retinopathy (DR)

Drugs	Target	Anti-inflammatory mechanisms	Effect on DR pathology
Steroids	Glucocorticoid receptor	Repress pro-inflammatory transcription factors [135]	Prevent ICAM-1 expression, leukostasis and vascular leakage [136] Improve visual acuity, decrease macular thickness [138, 139]
NSAID (e.g. Aspirin, Salicylate)	COX	Inhibit production of pro-inflammatory prostaglandins	Prevent ICAM-1 expression, leukostasis, vascular leakage, capillary cell apoptosis and vessel degeneration [20, 67, 143, 146–148, 150] Reduce incidence of DR and development of retinal microaneurysms [151–153]
Etanercept, Infliximab	TNF-α	Block TNF-α-induced inflammation	Prevent retinal inflammation, retina cell injury and vessel leakage [20, 156] Improve visual acuity and reduce macular thickness [157]
SAR 1118	LFA-1	Block leukocyte recruitment	Reduce leukostasis and vessel leakage [158]
Anti-CD49a neutralizing antibody	VLA-4	Block leukocyte recruitment	Reduce levels of VEGF and TNF-α protein, leukostasis and vessel leakage [159]
soluble RAGE, LR-90	AGEs	Block AGEs-induced inflammation	Improve neuronal dysfunction and prevent vessel degeneration [160, 161]
Vitamins C and E	Oxidative stress	Anti-oxidative stress	Reduce vessel degeneration [162] Reverse some of the changes in the retinal vessels [163]
Apocynin, Statin	NAPDH oxidase	Block NADPH oxidase activity and reduces oxidative stress	Prevent oxidative stress, retinal inflammation, vessel leakage and neovascularization [34, 102, 103]
Captopril, Telmisartan, Talsartan, Olmesartan, Candesartan, Enalapril	RAS	Block RAS-mediated inflammation	Prevent oxidative stress, inflammation, and vascular damage [96, 165, 167] Reduce the risk and the progression of retinopathy [168–170]

Abbreviations: AGEs = advanced glycation end products, COX = cyclooxygenase, DR = diabetic retinopathy, ICAM-1 = intercellular adhesion molecule 1, LFA-1 = leukocyte function associated antigen-1, NADPH = nicotinamide adenine dinucleotide phosphate, RAS = renin-angiotensin system, TNF-α = tumor necrosis factor alpha, VLA-4 = very late antigen-4.

Steroids

Corticosteroids have been used to treat ocular inflammatory conditions for more than half a century, but their application to DR was not investigated until recently [134]. Corticosteroids are known to exert anti-inflammatory actions by repressing a number of pro-inflammatory transcription factors including NF-κB and AP-1 via the direct interaction between glucocorticoid receptor and transcription factors or indirect mechanisms [135]. Oral steroids are excluded from DR therapy due to tremendous systematic side effects including exacerbation of diabetes [134]. However, intravitreal delivery of steroids has been shown to be beneficial in both pre-clinical and clinical studies. In a rat model of DR, intravitreal injection of dexamethasone suppressed up-regulation of ICAM-1, leukostasis, and prevented retinal vascular leakage [136]. Dexamethasone is not suitable for a clinical trial owing to its three-hour half-life [134]. Triamcinolone acetonide (TA), with a longer intravitreal half life (18.6 days) [137], has been tested in several clinical trials. In 2001, Jonas et al. reported the first clinical study to treat a diabetic patient with macular edema by an intravitreal injection of crystalline cortisone. Although laser coagulation failed in stopping cystoid macular edema in this patient, TA efficiently improved visual acuity [138]. A multicenter randomized clinical trial with 840 eyes to compare the effect of TA with focal/grid laser in treating diabetic macular edema was reported by the DR Clinical Research Network [139]. These studies have shown that TA (4 mg) improves visual acuity better than laser coagulation after 4 months of treatment. However, at 2 and 3 years, the mean visual acuity in TA-treated group was less than that in the laser-treated group. The potent acute effect of TA vs the longer but slower effect of laser photocoagulation suggests that combination therapy may be better. This concept has been confirmed by small randomized clinical trials which showed that TA in combination with laser coagulation improved visual acuity and decreased central macular thickness more as compared with laser photocoagulation alone [139]. These studies suggest that anti-inflammatory therapy with steroids could be a therapeutic choice for DR. However, it should also be noted that intravitreal steroid therapy increases the risk of glaucoma and cataract [139].

Nonsteroidal anti-inflammatory drug (NSAID)

NSAIDs are a group of compounds, such as aspirin and salicylate, that exert anti-inflammatory activity by inhibiting cyclooxygenase (COX) enzyme-mediated eicosanoid formation [140]. The COX enzymes are targets by NSAIDs [141]. Among the three isoforms of COX enzymes, COX-1 and COX-2 have been studied most extensively. COX-1 is expressed constitutively in many cell types and is important for homeostatic processes [141]. In contrast, COX-2 is induced mainly at sites of inflammation and produces predominately pro-inflammatory prostaglandins (PGE) such as PGE2 and PGF2α [141]. The promoter region of COX-2 contains NF-κB binding sites and COX-2 expression is under the control of NF-κB activity during inflammation [142]. Since iNOS is also an NF-κB -driven molecule, iNOS and COX-2 are often concurrently expressed in inflammation. Products from iNOS, particularly RNOS, positively regulate COX2 expression and activity [142]. COX-2 expression and its products including PGE2 are significantly increased in retinas from diabetic animals and in high glucose-treated retinal cells [143, 144]. Inhibition of COX-2 but not COX-1 reduces production of PGE2 from diabetic rat retinas [145], suggesting the dominant role of COX-2 in DR. COX-2 inhibition also blocks diabetes-induced increases retinal inflammatory reactions such as increases in ICAM-1 expression and leukostasis [20, 143, 146]. Given that COX-2 has a role in retinal inflammation, NSAIDs may have therapeutic potential in the pathology of DR.

Pre-clinical studies have demonstrated that aspirin treatment prevents capillary cell apoptosis and vessel degeneration in the diabetic dog and rat [67, 147, 148]. Similar effects are observed in diabetic rats treated with two other salicylate-based drugs, sodium salicylate and sulfasalazine [67]. In addition to salicylates, specific COX-2 inhibitors also reduce vascular leakage, capillary cell apoptosis and vessel degeneration [20, 149, 150]. In the clinic, the beneficial effect of NSAIDs has been realized in a study showing reduced incidence of DR in patients taking salicylates for the treatment of rheumatoid arthritis [151]. This finding encouraged the launch of two clinical trials: the Early Treatment DR Study and the Dipyridamole Aspirin Microangiopathy of Diabetes Study [152, 153]. These studies showed that although treatment of patients with advanced DR with low dose of aspirin does not have any benefits, the development of retinal microaneurysms is significantly attenuated in patients with early stage of DR when treated with high dose of aspirin (900 mg/day). Clinical trial of specific COX-2 inhibitors has been discouraged given that systemic COX-2 inhibitors increase incidence of heart attacks and strokes [140]. However, in preclinical studies, topical administration of COX-2 inhibitor was shown to reduce signs of DR similar to systematic application of COX-2 inhibitor [149, 150]. It would be of importance to investigate the therapeutic benefits of topical COX-2 inhibitor in future clinical trials.

Blocking inflammatory molecules

In addition to the above blockers of general inflammation, studies have been performed to determine whether targeting specific inflammatory molecules can be beneficial for DR. As a key player in many inflammatory reactions, the pro-inflammatory cytokine TNF-α is an attractive target for many inflammatory diseases including rheumatoid arthritis [154]. By binding to its cell surface receptor (TNF-α receptor 1 and 2), TNF-α induces ROS formation, NF-κB activation and iNOS expression in inflammatory cells and rapidly upregulates the expression of adhesion molecules such as E-selectin, VCAM-1 and ICAM-1 at the endothelial surface, resulting in leukocyte recruitment [155]. TNF-α is implicated in DR since its level is significantly increased in DR [20–22]. The effect of anti-TNF-α therapy has been studied in diabetic animal models and a few clinical cases. Subcutaneous administration of TNF-α trap (a soluble TNF-α receptor/Fc construct, Etanercept) significantly blocks retinal inflammation, retina cell injury and vessel leakage in diabetic rats [20, 156]. There is no clinic trial report so far. However in one clinical study with 4 patients, Infliximab, a TNF-α neutralizing antibody, was shown to improve visual acuity and reduce macular thickness in patients who failed to improve in response to laser photocoagulation treatment [157].

Since leukostasis is critically involved in the pathogenesis of DR, pre-clinical studies have been performed to determine whether blocking the interaction between leukocytes and endothelial cells will have beneficial effect in diabetic animal models. Leukocyte function associated antigen-1 (LFA-1, an integrin) expressed in leukocytes is important for leukocyte-endothelial cell interaction by binding to ICAM. Topical delivery of SAR 1118, a small antagonist of LFA-1, dose-dependently reduced leukostasis and retinal vascular leakage in diabetic rat model [158]. In addition to ICAM-1, interaction between very late antigen-4 (VLA-4) in leukocytes and VCAM-1 in endothelial cells is also involved in leukostasis. VLA-4 is formed by integrin alpha 4 (CD49d) and integrin beta 1. Blocking VLA-4 by anti-CD49a neutralizing antibody significantly attenuated the diabetes-induced leukostasis and vascular leakage [159]. Moreover, increases in NF-κB activity and levels of VEGF and TNF-α protein were also reduced by anti-CD49a neutralizing antibody [159], indicating that leukocyte recruitment plays a positive feedback role in the inflammatory reactions in DR.

In addition to these specific inflammatory molecules, the AGEs/RAGE pathway is also an emerging target for DR given that it is an important mechanism leading to retinal inflammation. Blocking the RAGE axis with soluble RAGE improves neuronal dysfunction and limits the development of acellular capillaries and pericyte ghosts in diabetic mice [160]. A similar beneficial effect was also observed by inhibition of AGEs formation with LR-90, a new advanced glycation inhibitor, in a diabetic rat model [161].

Targeting oxidative stress

As a key mediator in inflammation, oxidative stress serves as an important target for anti-inflammatory therapy. A number of studies have been performed to determine whether blocking oxidative stress can be an approach to treat DR. In diabetic rats, supplement of anti-oxidants such as vitamins C and E attenuates the development of acellular capillaries and decreases the number of pericyte ghosts [162]. The effect of anti-oxidants is further boosted when a more comprehensive mixture of anti-oxidant diet is applied [162]. Clinical studies also show that high doses of vitamin E reverse some of the changes in the retinal vessels of diabetic patients [163]. However, it was also noticed that the non-selective anti-oxidants did not achieve the expected beneficial effects predicted by the pre-clinical studies and in some clinical studies the anti-oxidant therapy did not show any beneficial effect [164]. Thus, it is necessary to target specific dysregulated oxidative stress generating mechanisms and to avoid interfering with the physiological roles of reactive oxygen species. Regarding this issue, blockade of NADPH oxidase may provide better output given that NAPDH oxidase activity is dysregulated in DR and NADPH oxidase has a key role in the inflammatory cascade. This notion has been supported by several in vitro and in vivo studies. In retina and retinal cells, NADPH oxidase blockers are shown to prevent CCL2 production induced by different inflammatory stimuli [133]. In high glucose-treated retinal cells, NADPH oxidase inhibition significantly inhibits oxidative stress, NF-κB activation, RNOS formation, and inflammatory reactions [100, 102]. In diabetic animals or models of ischemic retinopathy, blockade of diabetes-induced increases in NADPH oxidase activity prevents vascular leakage and pathological neovascularization, respectively [34, 102, 103]. Moreover, these studies also reveal that statin treatment effectively blocks up-regulation of NADPH oxidase activity in DR [34], suggesting that the beneficial of statin treatment is at least partly due to its activity in blocking NADPH oxidase.

Blocking RAS

The RAS serves as a promising target for DR since this system is involved in both diabetes and hypertension-induced retinal inflammation, and is a pathway that crosstalks with multiple other pathways including oxidative stress and AGEs. Specific blockade of the RAS has been shown to prevent oxidative stress, inflammation, and vascular damage in diabetic animal models [165–167]. Several clinical trials have been performed to evaluate whether intervention of RAS can reduce the risk or slow the progression of DR. In a clinical trial involving 285 type 1 diabetic patients with normotensive and normoalbuminuria, blockade RAS with Losartan (an AT1R blocker) or Enalapril (angiotensin converting enzyme inhibitor) significantly reduced the progression of retinopathy by 70% and 65% respectively [168]. However, such dramatic benefit is not observed in the DIabetic REtinopathy Candesartan Trials which enrolled 3326 patients with type 1 diabetes and 1905 patients with type 2 diabetes to assess whether Candesartan (AT1R blocker) could reduce the incidence and progression of retinopathy in a five year period [169, 170]. All of the type 1 diabetic patients recruited in the study were normotensive and normoalbuminuric. Patients without diabetic retinopathy were assigned to treatment to study whether Candesartan can reduce the incidence of DR. Patients with existing DR were assigned to treatment to study whether Candesartan can slow down the progression of DR. This study showed that Candesartan treatment reduced the incidence of DR but did not prevent progression of DR [169]. For a trial in type 2 diabetic patients, patients recruited were normoalbuminuric, normotensive, or treated hypertensive people with mild to moderately severe DR. This study also showed that Candesartan treatment did not prevent progression of DR but that it significantly induced the regression of DR even after adjustment for baseline characteristics and variations in systolic blood pressure [170]. Overall, blocking AT1R has a beneficial effect for DR. However further trials are needed to resolve the discrepancy between the above trials and to fully evaluate the beneficial effect of RAS blockade before its clinical usefulness is fully understood.

Conclusion

Accumulated evidence has demonstrated that retinal inflammation plays a critical role in the pathogenesis of DR. Although there is no pathogen in the diabetic retina, diabetes apparently hijacks mechanisms of inflammation such as oxidative stress formation, dysregulation of the NOS pathway, and NF-κB activation. These changes together with diabetes-induced AGEs formation and impaired endogenous anti-inflammatory pathways lead to chronic inflammatory reactions in the retina by persistently inducing expression of inflammatory cytokines, chemokines, and recruiting leukocytes. The chronic inflammation is responsible for the pathogenesis of DR by causing neurovascular damage and ischemic neovascularization. The discovery of inflammation as a key player in DR has opened a new avenue for the development of new treatments and strategies for this devastating disease.

Future perspective

The mechanism by which the inflammatory cascade is initiated and maintained in DR is currently unknown as are the individual roles of inflammatory molecules in the different stages of DR. On the one hand, the inflammatory molecules have redundant functions given that many of them cause activation of similar downstream targets such as NF-κB and several chemokines that share the same receptor. On the other hand, many inflammatory molecules are indispensible for the positive feedback loop during inflammation since retinal inflammation has been shown to be attenuated by blocking TNF-α, iNOS or integrins. These mysteries have led to challenges when considering any of the inflammatory molecules as a drug target since blockade of any one of them may be effective if the molecule is indispensible. However, the blockade can also be ineffective or even worse if compensation happens or goes in an unwanted direction. These uncertainties may be resolved by understanding the changes that occur in the inflammatory cascade after blocking an individual inflammatory molecule and selecting several candidates for combination therapy. Alternatively, drugs targeting several mechanisms, such as statin and calcium dobesilate, may be more efficient than those that target a single mechanism. Related to this notion, the use of RAS blockers also represents a possible future clinical therapeutic approach for DR given that the RAS regulates multiple key factors of inflammation including oxidative stress, AGEs formation and hypertension. A subset of Ang II receptor blockers which possess partial PPARγ agonist activity (e.g. Telmisartan and Irbesartan) may be more powerful since they will block pro-inflammatory pathways meanwhile enhancing anti-inflammatory pathways. Few studies have been performed to understand the endogenous anti-inflammation mechanisms in DR. It would be helpful to identify such mechanisms as PPARγ in DR in order to correct or enhance their anti-inflammatory activity. Finally, more studies are needed to understand the role of inflammation in neuronal damage in DR.

Executive summary.

• DR is a vision threatening disease affects a large population globally. There is a growing need to develop new therapeutic approaches.

• DR has many features of inflammation, such as increases in inflammatory molecules, leukocyte recruitment, and activation of local immune cells.

• Inflammation plays a critical role in the development of the disease by causing vessel leakage, closure, pathological neovascularization, and neural death.

• Diabetes induces inflammatory reactions by many mechanisms including, oxidative stress NF-κB activation, NOS dysregulation, AGEs formation, hypertension, dyslipidemia and impaired anti-inflammatory pathways.

• Anti-inflammatory therapy has proven to be beneficial in both clinical and pre-clinical studies.

• Better understanding of inflammatory cascades and endogenous anti-inflammatory mechanisms may lead to better therapies in the future.

Abbreviations

AGEs

advanced glycation end products

Ang II

angiotensin II

AP-1

activator protein 1

AT1R

Ang II type 1 receptor

BRB

blood-retinal barrier

COX

cyclooxygenase

DR

diabetic retinopathy

eNOS

endothelial NOS

ETC

electron transport chain

ICAM-1

intercellular adhesion molecule 1

LFA-1

leukocyte function associated antigen-1

iNOS

inducible NOS

IκB

inhibitor of κB

IL-1β

interleukin-1β

IL-6

interleukin-6

MAPKs

mitogen-activated protein kinases

MMP

matrix metallopeptidase

NADPH

nicotinamide adenine dinucleotide phosphate

NF-κB

nuclear factor kappa-light-chain-enhancer of activated B cells

nNOS

neuronal NOS

NO

nitric oxide

NOS

nitric oxide synthase

NSAID

nonsteroidal anti-inflammatory drug

PGE

prostaglandin

PPARγ

proliferator-activated receptor gamma

RAGE

receptor for advanced glycation endproducts

RAS

renin-angiotensin system

RNOS

reactive nitrogen oxide species

ROS

reactive oxygen species

SHR

spontaneously hypertensive rats

STAT

signal transducers and activators of transcription protein

TGF-β

transforming growth factor beta

TNF-α

tumor necrosis factor alpha

VCAM-1

vascular cell adhesion molecule 1

VE-cadherin

vascular endothelial-cadherin

VEGF

vascular endothelial growth factor

VLA-4

very late antigen-4

ZO

zonula occludens