Chapter 25 Inflammatory Response and Mediators in Retinal Injury
Retinal injury
Ischemia–hypoxia
The retina consumes oxygen more rapidly than any other tissue in the body in part to fuel the constant renewal and replacement of photoreceptor outer segments.1 The presence of abundant mitochondria in the photoreceptor inner segments is evidence of this high energy demand.2 In addition, the choroid receives a high rate of blood flow to supply oxygen and glucose to the retinal pigment epithelium (RPE).3 Inadequate blood supply decreases oxygen delivery to the retina (hypoxia) and may result in ischemic injury. The subsequent ischemia can induce neuronal apoptosis and neovascularization. Classical examples of the retinal diseases associated with hypoxia and ischemia are diabetic retinopathy and retinal vein occlusion.
Inflammation often occurs in hypoxic-ischemic conditions. Diabetic retinopathy is characterized by retinal vascular nonperfusion and leakage. In an experimental rat model of diabetes mellitus (DM) induced by streptozotocin, retinovascular leukostasis develops within 1 week, followed by a progressive breakdown of the blood–retinal barrier. After 1 week, areas of nonperfusion and reperfusion develop, and extravascular leukocytes are observed in the retina.4 On a microscopic level in human diabetic retinopathy, the inflammation is demonstrated by vascular dilatation, exudation and transudation of fluids, and leukocyte (macrophages, neutrophils, and T lymphocytes) infiltration in the retina (Fig. 25.1).5–7 Retinal vein occlusion is another common condition associated with hypoxia and ischemia. In a histopathological study of 29 cases with central vein occlusion, lymphocytes and macrophages were observed at the occlusion site in 47% of cases (Fig. 25.2).8
Further evidence of the involvement of inflammatory processes in diabetic retinopathy and retinal vein occlusion is based on the upregulation of several inflammatory genes and the detection of inflammatory cytokines and chemokines under hypoxic-ischemic conditions. Tumor necrosis factor (TNF)-α, interleukin (IL)-1β, monocyte chemotactic protein 1 (MCP-1/CCL2), and macrophage inflammatory protein (MIP)/CCL3 transcripts have been detected in the ischemic-hypoxic retina. These proinflammatory cytokines, particularly TNF-α and IL-1β, are thought to play a major role in the breakdown of the blood–retinal barrier and the degeneration of retinal capillaries.9 CCL2 and RANTES (regulated upon activation, normal T-cell expressed and secreted)/CCL5 are significantly elevated in sera of patients with severe nonproliferative diabetic retinopathy compared with those who have less severe retinopathy.7 Increased C-reactive protein (CRP), IL-6, and TNF-α, and especially intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule-1, and E-selectin are associated with nephropathy, retinopathy, and cardiovascular disease in both type 1 and type 2 diabetes.10 In proliferative diabetic retinopathy, vitreous cytokine levels of IL-6, IL-8, and CCL2 are strongly correlated with elevated VEGF.11 Patients with central retinal vein occlusion have significantly higher levels of aqueous and vitreous VEGF and IL-6 versus control subjects with nonischemic ocular disease.12,13
Type 1 DM is now considered a common autoimmune disorder with multiple genetic susceptibility loci as well as environmental risk factors. Recently, a genomewide association study of type 1 DM was combined in a meta-analysis with two other studies to examine a total of 7514 cases and 9045 reference samples. Several genetic regions, including those containing the immunoregulatory cytokine genes IL-10, IL-19, and IL-20, showed significant associations with type 1 DM, suggesting possible functional relevance of these genes to the disease pathogenesis.14 The Wellcome Trust Case Consortium genomewide association study of type 1 diabetic families identified a genetic susceptibility region which is close to the inflammatory genes Toll-like receptor (TLR)7 and TLR8.15 In another study, a custom panel of 1536 single-nucleotide polymorphisms (SNPs) located in 193 candidate genes was used to genotype 437 African Americans with type 1 DM, 128 with and 309 without severe diabetic retinopathy. The results indicated an association between genes involved in inflammation, such as ANGPT1 (angiopoietin 1), FLT1 (fms-related tyrosine kinase 1), PLA2G4A (cytosolic phospholipase A2), OLR1 (oxidized low-density lipoprotein lectin-like receptor 1), and PPARG (peroxisome proliferator-activated receptor-γ), and the progression of diabetic retinopathy.16 In contrast, a study of 315 patients and 335 age- and gender-matched control subjects showed no significant genetic association between central retinal vein occlusion and variants in many inflammation-related genes, including TNF-α, IL-1β, IL-6, IL-8, IL-10, and CCL2.17
The concept that diabetic retinopathy is a persistent low-grade inflammatory response to ischemia-induced retinal neovascularization and damage has led to the clinical application of anti-inflammatory treatments such as corticosteroids and anti-TNF-α agents.5,18 In addition to being a potent inflammatory molecule, TNF-α can also promote apoptosis and loss of retinal vascular pericytes and endothelial cells,19,20 increase retinal endothelial permeability,21 and activate microglia22 in experimental diabetic rats. Treatments that target factors related to retinal–blood barrier breakdown have also been used in retinal vein occlusion.
As corticosteroids have anti-inflammatory, antiangiogenic, and antivascular permeability properties, the administration of intravitreal corticosteroids (triamcinolone) has shown promising results for diabetic macular edema in a number of randomized clinical trials.9,23 However, in a carefully designed, prospective, randomized trial, the Diabetic Retinopathy Clinical Research Network showed that focal or grid laser photocoagulation was more effective and had fewer side-effects than intravitreal triamcinolone treatment of diabetic macular edema.24 Corticosteroids have also been used to treat both central and branch retinal vein occlusions.25 Case series have suggested that intravitreal injection of triamcinolone may be useful for the treatment of macular edema in patients with branch retinal vein occlusion.26 Nonetheless, due to a high rate of adverse events, the use of this treatment was not supported by the Standard care versus Corticosteroid for Retinal vein occlusion (SCORE) study, a randomized trial which included 411 patients with branch retinal vein occlusion and vision loss from macular edema.27 In another study of 437 patients with central retinal vein occlusion and 830 with branch vein occlusion, the longer-term delivery of intravitreal corticosteroids through a dexamethasone implant was reported to be associated with a shorter time to achieve a gain in visual acuity.28 Elevation of intraocular pressure and cataract progression are among the potential significant adverse effects for patients receiving intravitreal corticosteroids. Further development of noncorticosteroid agents that target the inflammatory components underlying retinovascular disease pathogenesis may provide future therapeutic options.
Oxidative stress
Oxidative processes occur through the removal of electrons from molecules. In biologic systems, energy is released when lipids, proteins, and carbohydrates are oxidized to form carbon dioxide and water. Oxidative reactions may also result in the production of reactive oxygen intermediates (ROIs), such as free radicals, hydrogen peroxide, and singlet oxygen. ROIs can damage other molecules and increase under conditions of irradiation, aging, reperfusion, inflammation, increased partial pressure of oxygen, cigarette smoke, and air pollution.29 The biologic mechanisms to prevent the detrimental effects of ROIs include cellular compartmentalization, repair of DNA and proteins, and neutralization by antioxidant compounds. The retina is an ideal environment for the generation of ROIs for several reasons: (1) high oxygen consumption; (2) high levels of cumulative irradiation; (3) RPE phagocytosis, which is an oxidative stress that produces ROIs; (4) high levels of polyunsaturated fatty acids in the photoreceptor outer-segment membranes; and (5) abundant photosensitizers in the neurosensory retina and RPE.29
Oxidative stress results when there is an imbalance between pro-oxidants and antioxidants, leading to molecular damage and/or a disruption of redox signaling.30 Inflammation and oxidative stress are closely interrelated: inflammatory cells can generate ROIs, and oxidative stress can induce inflammation through NF-κβ-mediated inflammatory gene expression.30 Oxidative stress plays a role in the pathogenesis of several retinal disorders, including AMD and retinopathy of prematurity (ROP).
Multiple changes are associated with the aging eye, including thickening of Bruch’s membrane, accumulation of lipofuscin in the RPE, and loss of parafoveal rods. In a model outlined by Curcio et al.31 and others, the RPE and Bruch’s membrane are modified or damaged by oxidative stress and enzymatic processes over time. The resultant damage slows the normal flow of materials such as lipids, protein, and lipofuscin from the outer retina and RPE through Bruch’s membrane. However, simple accumulation of material does not lead to AMD. The retained materials, including lipoproteins, may be modified by oxidative stress and become stimuli for inflammation.31 This model of Bruch’s membrane lipoprotein accumulation and modification in a chronic proinflammatory environment is similar to the current understanding of the pathogenesis of atherosclerosis.
Oxidative stress and inflammation in AMD
Several studies have provided evidence that oxidative stress is involved in AMD pathogenesis. Crabb et al. performed a proteomic analysis of drusen from normal and AMD donor eyes.32 They identified multiple proteins modified by oxidation, including cross-linked molecules of carboxyethyl pyrrole (CEP) protein adducts, tissue metalloproteinase inhibitor 3, and vitronectin. Carboxymethyl lysine, an advanced glycation end product produced through oxidation of carbohydrate, was also isolated. These oxidation-related products were identified more frequently in drusen from AMD eyes. In addition, drusen in AMD donor eyes were also more likely to contain crystallins, which are proteins upregulated during cellular stress such as oxidation. In another proteomics study, Yuan et al. isolated the Bruch’s membrane–choroid complex from postmortem AMD and control eyes.33 They found elevated levels of retinoid-processing proteins in the early and midstage AMD eyes. These retinoids are highly susceptible to oxidative damage, can accumulate in RPE lipofuscin granules, and activate complement. They also found significant elevation of an oxidation product-related receptor in eyes with advanced dry AMD. In the study by Yuan et al.,33 approximately 60% of the elevated proteins were involved in immune response or host defense, including complement factors C5 and C7, α-crystallin A, α-crystallin B, and major histocompatibility complex (MHC) class II molecule DRα. In contrast, several proteins classified as damage-associated molecular patterns (DAMPs) were present at similar levels in AMD and control eyes. Damaged cells release DAMPs which can bind receptors in the complement system, Toll-like receptors, and receptors for advanced glycation end products. These data suggest that both normal and AMD eyes express proteins in response to aging-associated damage; however, in AMD eyes the response to injury may evolve into a chronic proinflammatory environment.
Numerous histopathologic studies of AMD eyes have documented the presence of macrophages and multinucleated giant cells, mainly associated with vascular channels and breaks in Bruch’s membrane (Fig. 25.3).34–39 It is unclear whether the macrophages contribute to the development of AMD lesions or if they accumulate as an adaptive response to injury associated with AMD. Some evidence suggests the relationship between oxidative stress and inflammation in AMD may be mediated in part by macrophages similar to the pathogenesis of atherosclerotic plaques. Oxidation of lipoproteins exposes phosphorylcholine (PC), which is also present on the membranes of apoptotic cells. PC can be bound by CRP, an opsonic molecule that activates the classical complement pathway and enhances phagocytosis via phagocytic complement receptors.40,41 In this manner, oxidative damage to the retina may lead to macrophage infiltration and activation via the innate immune response. An immunohistochemical study of surgically excised AMD-associated choroidal neovascular (CNV) membranes provides in vivo evidence to support this hypothesis.42 Oxidized lipoproteins were detected in RPE cells as well as on macrophages associated with the CNV membranes. Scavenger receptors for oxidized lipoproteins were also present on RPE cells, macrophages, and some endothelial cells in the CNV membranes, although the type and amount of scavenger receptor differed for each cell type. These data suggest macrophages accumulate in AMD lesions in order to phagocytize oxidized lipoproteins via scavenger receptors.
Hollyfield et al. have tied together earlier in vitro experiments and in vivo observations by producing AMD-like lesions in mice exposed to an oxidative product.43,44 They immunized mice with CEP adducts formed by the covalent interaction with an oxidation fragment of docosahexanoic acid (DHA), and found higher levels of antibodies against CEP in these mice. The authors hypothesized that the mouse immune system responds to the CEP antibodies by depositing complement below the RPE. In support of this idea, they detected complement C3d in Bruch’s membrane and on RPE cells in the CEP-immunized mice but not controls. Furthermore, there was a close relationship between the level of CEP-specific antibodies and the severity of RPE pathology. An intact immune response was required for CEP-induced RPE changes since Rag-deficient mice which lack mature T and B cells did not develop RPE pathology after CEP immunization. Monocytes and macrophages were observed in the interphotoreceptor matrix. Hollyfield et al. suggested the macrophages probably did not initiate the pathology since many lesions were observed without associated macrophages. They hypothesized that macrophages could be attracted by the release of cytokines from lysed cells and participate in debris removal, as supported by the presence of melanin-laden macrophages in the RPE lesions.
Microglia and AMD
Microglia are specialized tissue macrophages which constitute the main resident immune cells in the neuroretina. In concert with the RPE cells, choroidal macrophages, and dendritic cells, the microglia play an important role in maintaining normal retinal immune function as well as supporting the surrounding cells composing the neuroretina. In their quiescent state, microglia are inconspicuous stellate cells that produce anti-inflammatory IL-10 in response to RPE-derived cytokines. Microglia can rapidly transform into enlarged, activated microglia in response to microenvironment signals of infection, neuronal injury, ischemia, or oxidative stress.45–49 Quiescent and activated microglia secrete polypeptide neurotrophic factors, including brain-derived neurotrophic factor, ciliary neurotrophic factor (CNTF), nerve growth factor, neurotrophin-3, and basic fibroblast growth factor which influence neuron physiology and survival.47
The sub-RPE space is immunologically protected by the outer blood–retinal barrier and normally devoid of immune cells. It is hypothesized that as lipofuscin and oxidized material accumulate in the RPE and Bruch’s membrane of the aging eye, the RPE may be stimulated to produce inflammatory cytokines and chemokines which activate microglia and stimulate sub-RPE migration.48,50 Under conditions of chronic inflammation and activation, advanced glycation end products and oxidized lipids bind microglia and promote further inflammatory response. The microglia may have a detrimental effect in the neuroretina as they perpetuate the inflammation and injury cycle by secreting molecules that kill neighboring cells. Ma et al. cocultured activated retinal microglia with primary cultures of mouse RPE cells and measured lower levels of junctional proteins and RPE-65 as well as prominent structural disorganization and increased proliferation of the RPE cells in comparison to controls.51 They also found increased RPE expression and secretion of multiple proinflammatory cytokines, including IL-1β, TNF-α, IL-6, and granulocyte–macrophage colony-stimulating factor; the anti-inflammatory cytokine IL-10 was not markedly changed. Coculture with activated microglia also fostered a proangiogenic environment with increased mRNA levels of the matrix metalloproteinases MMP1 and MMP9 as well as protein levels of MMP1, MMP2, MMP9, and VEGF. Finally, Ma et al. evaluated the in vivo effects of subretinal activated microglia by transplanting them into the subretinal space of adult wild-type mice. When the animals were euthanized and examined 4 days after the subretinal injections, there were large and prominent choroidal neovascular membranes in these mice.
Both human and mouse in vivo studies provide evidence of microglial migration in conditions of aging or stress. Gupta et al. examined three AMD eyes with geographic atrophy and two normal donor eyes.52 They found quiescent microglia in the inner layers of the normal retinas; in contrast, the microglia were balloon-shaped and present in the inner and outer nuclear layers, often in association with degenerate rods in the AMD eyes. The activated microglia were also observed in the subretinal space of the AMD eyes; their rhodopsin-positive cytoplasmic inclusions were evidence of their role in phagocytizing photoreceptor debris. The findings suggest microglial migration occurs in AMD.
Xu et al. studied the mouse retina and observed that the numbers of subretinal microglia increased with age.50 They also noted an age-dependent accumulation of lipofuscin in subretinal microglia, possibly related to phagocytosis of photoreceptor outer segments. Shen et al. reported that naloxone, a potent microglial inhibitor, ameliorated AMD-like retinal lesions in Ccl2/Cx3cr1 double deficient with rd8 background mice via suppression of proinflammatory molecules of nitric oxide (NO), TNF-α, and IL-β.53 Under experimentally induced photo-oxidative stress in rats exposed to blue light, Rutar et al. used microarray expression analyses to show that light-induced injury induced differential expression of complement-related genes, including upregulation of complement activators and receptors.54 They also detected complement C3-expressing microglia in the outer nuclear layer and in the subretinal spaces adjacent to areas of injury.
Other inflammatory-related molecules and pathways
The inflammatory response to oxidative stress involves many other molecules and pathways which may represent future therapeutic targests. Neuroprotectin D1 (NPD1) is an endogenous anti-inflammatory mediator which is derived from the essential omega-3 polyunsaturated fatty acid, DHA.55,56 Multiple studies have shown the RPE generates NPD1 in response to oxidative stress.57–60 NPD1 promotes RPE survival by upregulating antiapoptotic genes and downregulating antiapoptotic genes57,58,61 and also inhibits cytokine-mediated proinflammatory gene induction.57,62
The PPARs are transcription factors which are activated by fatty acids. They exist in three forms, PPAR-γ, -α, and -δ (also known as -β, NUC-1, or FAAR), which differ in tissue distribution and ligand specificity.63 PPAR-γ is expressed by normal human RPE and upregulated in response to oxidative stress, including in AMD lesions.64 PPAR-γ activation may protect against oxidative injury through upregulation of antioxidative enzymes,65,66 downregulation of chemokines such as CXCR4 and CCL2,67,68 and modulation of microglial activation.69 Pharmacologic agonists for both PPAR-α (fibrates) and -γ (thiazolidinediones) have been investigated as potential treatments for ocular neovascularization through inhibition of the VEGF pathway63,70–72; however, due to the overlap between many components of the pathways of angiogenesis and inflammation, further clinical studies are warranted.73
Nuclear erythroid-2-related factor (NRF2) is another transcription factor which plays an important role in the cellular response to oxidative stress and may be regulated by PPAR-γ.68 NRF2 regulates antioxidant gene expression and complement activation to prevent further inflammation-mediated oxidative stress.74 To date, most studies have focused on the role of NRF2 in lung disease (modulation of macrophage activation in response to cigarette smoke)75 and neuroinflammation (downmodulation of activated microglia).76,77
Heat shock proteins (Hsps) prevent harmful protein aggregation by serving as molecular chaperones that bind damaged intracellular proteins and assist in the processes of repair or removal and degradation.78 Hsps participate in chaperone-mediated autophagy which is an oxidative stress-activated lysosomal pathway that results in degradation of cytoplasmic material and organelles. Some Hsps also have cytoprotective effects through modulation of apoptosis and stabilization of cytoskeletal proteins.79,80 Age-related decreases in Hsp activity and autophagy may contribute to the accumulation of aggregations of oxidized proteins and lipids (lipofuscin) in RPE cells.81–83 Oxidative stress is known to increase the expression of many of the Hsps, especially Hsp27 (also known as HspB1) and Hsp 70.80,84,85
