Inflammatory mechanisms of age-related macular degeneration

Introduction

Late age-related macular degeneration (AMD), specifically central geographic atrophy (GA) and choroidal neovascularization (CVN),¹ is the leading cause of irreversible vision loss in the elderly in the developed world.² AMD is categorized as either i) non-neovascular, or dry, when CNV is not present, or ii) neovascular, or wet, when CNV is present. Increasing age is the most significant risk factor for AMD development.³ In the United States, AMD is more common in Caucasians, and smoking is strongly associated. The hallmark of AMD clinically is the presence of drusen situated under the retinal pigment epithelium (RPE). Drusen size is categorized as small (≤62μm), medium (63–124μm), or large (≥125μm), with presence of large drusen being a significant predictor of progression to late AMD and associated vision loss.⁴ The Age-Related Eye Disease Studies (AREDS and AREDS2) have shown that high-dose antioxidant and zinc supplementation delay the progression of AMD in moderate- and high-risk patients.^{5, 6} Anti-vascular endothelial growth factor (VEGF) treatment is the mainstay of therapy for wet AMD.⁷ Currently, there is no treatment for GA.

While AMD pathogenesis is undoubtedly multifactorial, including the effects of aging and oxidative stress as well as genetic and environmental factors, significant evidence has emerged implicating inflammation and the immune system. The major role of the immune system is to identify and respond to physiologic insults, such as infection, malignancy, and tissue damage. Often this takes the form of robust inflammatory responses, such as those seen in various forms of uveitis, despite the potent down-regulatory immune environment within the eye.⁸ In AMD, immune dysregulation in the form of overt intraocular inflammation is not clinically apparent. It has been proposed that the role of inflammation is not always negative. The concept of parainflammation has been suggested to describe inflammatory responses to tissue stress that are intermediate between basal and robust inflammatory states and function in a reparative manner.^{9, 10} As AMD is not accompanied by an intense inflammatory reaction, it is possible that dysregulation of reparative parainflammatory mechanisms, in the context of the aging eye, lead to a low grade chronic inflammatory response and subsequent AMD pathology. This review will focus on the potential inflammatory mechanisms of AMD pathogenesis based on evidence from human studies.

Immunogenetics of AMD

Over the past 10 years, AMD has been associated with several genetic single nucleotide polymorphisms (SNP), many of which encode proteins involved in inflammatory cascades and the immune system (Table 1). Variants of several genes encoding proteins in the complement system, including complement factor H (CFH),^11–14 CFB, complement component 2 (C2),^{15, 16} C3,^17–19 and C5,²⁰ have been associated with AMD suggesting that dysregulation of the complement cascade may be involved in AMD pathogenesis. CFH functions to down regulate the alternative complement pathway.²¹ The CFH polymorphism resulting in tyrosine to histidine at position 402 (Y402H) may be associated with up to 50% of all AMD cases²² and has also been associated with sarcoidosis-related and other forms of posterior uveitis,^{23, 24} possibly suggesting a larger role for this polymorphism in ocular inflammation. CFB and C2 are activators of the alternative and classical complement pathways, respectively, and variants of these genes were found to be protective against development of AMD.^{15, 16, 25}

Table 1.

Immune-related gene variants associated with AMD

Gene	Chromosomal Location	Function	Variant Effect	References
CFH	1q32	Alternative pathway inhibitor	Harmful	11–14
CFB	6p21	Alternative pathway activator	Protective	15, 16, 25
C2	6p21	Classical pathway activator	Protective	14, 15, 22
C3	19p13	Complement component; multiple inflammatory functions	Harmful	17–19
C5	9q33-q34	Complement component; multiple inflammatory functions	Depends on variant	20
CX3CR1	3p21	Chemokine receptor found on retinal microglia	Conflicting/Harmful	27–31
CCR3	3p21	Chemokine receptor	Harmful	32
IL-8/CXCL8	4q13-q21	Chemokine for neutrophils; pro-angiogenic	Harmful	33, 34

CFH: Complement factor H; CFB: Complement factor B; C2: Complement component 2; C3: Complement component 3; C5: Complement component 5; IL: Interleukin

In addition to the complement system, polymorphisms in genes encoding chemokines and their receptors have been associated with AMD. Chemokine receptors are expressed on immune cells as well as other cell types, such as endothelial cells, and function to direct cells to sites of inflammation in response to ligation by their cognate chemokine. Genetic variants of the chemokine receptor CX3CR1, which is expressed on retinal microglia,²⁶ have been associated with AMD.^{27, 28} However, not all studies have confirmed this association^{29, 30} and different studies have identified distinct polymorphisms that may be associated with AMD.³¹ Polymorphisms in the genes encoding the chemokine receptor CCR3³² and chemokine CXCL8 (also known as IL-8),^{33, 34} which have been implicated in angiogenesis,^{35, 36} have also been associated with AMD. Therefore, AMD has been associated with genetic variants of various inflammatory molecules perhaps suggesting that several inflammatory pathways can lead to the same clinical disease.

The complement system in AMD

The complement system encompasses over 30 proteins that function in an inflammatory cascade resulting in enhanced chemotaxis, phagocytosis, agglutination, and target cell lysis via formation of the membrane attack complex (MAC). As discussed above, several complement gene variants have been associated with AMD suggesting a role for the complement system in AMD pathogenesis. Complement activation products, including C3a, C5a, as well as C5b-9 that comprise the MAC, have been identified in drusen from human eyes with AMD^37–42 and were found to be elevated in serum of AMD patients compared with age-matched controls.⁴³ Levels of these complement activation markers correlated with the presence of at least one copy of the CFH Y402H risk allele. In addition, dry AMD patients homozygous for the CFH variant had higher serum levels of interleukin (IL)-6, IL-18, and tumor necrosis factor (TNF)-α compared with those either heterozygous for or lacking the variant.⁴⁴ Given the function of CFH as an inhibitor of the complement cascade, these results suggest that the Y402H CFH variant may reduce its inhibitory function allowing deregulated activation of the complement system and subsequent inflammatory and cytolytic responses. In accord with this notion, a rare genetic variant of C3, which reduces its ability to bind CHF, has also been associated with AMD.^{45, 46}

The C5a receptor has been identified in human choroid⁴⁷ as well as on ARPE-19 cells.⁴⁸ Exposure of choriocapillaris endothelial cells to C5a in vitro lead to upregulation of ICAM-1, an adhesion molecule that binds activated leukocytes.⁴⁷ Exposure of ARPE-19 cells to C5a in vitro suppressed RPE-derived transforming growth factor (TGF)-β2, an important immunosuppressive molecule in the eye, and reduced RPE viability.⁴⁸ Furthermore, C5a increased the proliferative capacity of peripheral blood mononuclear cells (PBMC) and decreased the ability of RPE cells to suppress PBMC proliferation. In a separate study, C5a was shown to promote interleukin (IL)-22 and IL-17A secretion from human CD4 T cells isolated from patients with exudative AMD in a monocyte-dependent mechanism.⁴⁹ IL-17A has been shown to induce angiogenesis in human choroidal endothelial cells in vitro.⁵⁰ IL-17A can also induce pyroptosis and apoptosis in ARPE-19 cells.⁵¹ These results suggest that RPE: i) choroidal endothelial cell, and circulating immune cells are responsive to complement activation products, ii) RPE/choroid exposed to complement activation products can modulate interactions between monocytes and lymphocytes, and that iii) cytokines expressed by T cells exposed to complement activation products may promote neovascularization and cell death.

Recently, there have been a number of reports implicating exosomes in the pathogenesis of AMD. Exosomes are small membrane-bound vesicles of endocytic origin secreted by many cell types and thought to play a role in intercellular communication.⁵² Abundant oxidative stress within the RPE is a well-accepted component of AMD pathogenesis.⁵³ Oxidative stress results in production of reactive oxygen species (ROS), which can damage mitochondrial DNA (mtDNA) and other macromolecules. Mitochondrial damage has been illustrated in human AMD maculae.^{53, 54} In an in vitro model where mtDNA damage was induced in ARPE-19 cells by treatment with rotenone, a mitochondrial complex I inhibitor, increased autophagy as well as increased exosome release and chemokine secretion were observed.⁵⁵ Markers of autophagy and exosomes were also demonstrated in drusen of human AMD patients. In the in vitro model, secreted exosomes were coated with complement factor C3, which was able to bind CFH. The authors speculated that mutation of CFH might decrease the ability of CFH to bind and clear C3-coated exosomes, leading to enhanced drusen formation and AMD.

Inflammatory cytokines in AMD

A number of inflammatory cytokines have been found to be elevated either systemically in the serum or locally in the ocular tissue or fluids of patients with AMD. As mentioned above, systemic levels IL-6, IL-18, and TNF-α correlated with CFH haplotypes in AMD patients.⁴⁴ Furthermore, systemic IL-6 levels have been shown to correlate with the incidence⁵⁶ and progression⁵⁷ of AMD. IL-18 has recently been reported to play a protective role against development of CNV.^{58, 59} However, results are conflicting as to whether IL-18 functions in anti-angiogenic manner.⁶⁰ As discussed below, IL-18 may also be involved in RPE atrophy.⁶¹

IL-22 and IL-17A, cytokines produced by the Th17 subset of helper CD4 T cells, were found to be elevated in the serum of patients with exudative AMD.⁴⁹ Two SNPs within the IL-17A gene were recently reported to be associated with AMD as well as higher levels of IL-17A secretion from PBMC stimulated ex vivo.⁶² Elevated IL-17A and IL-17RC levels were also detected in AMD lesions compared with normal macular tissues.⁵¹ In vitro, IL-22 induced apoptosis of primary human RPE cells,⁶³ and IL-17A was shown to enhance production of inflammatory cytokines, including IL-6, IL-8 and CCL2, from ARPE-19 cells⁶⁴ as well as induce angiogenesis in human choroidal endothelial cells.⁵⁰

Several studies have employed multiplex protein screens to identify cytokines in ocular fluids of patient with neovascular AMD.^65–70 Table 2 lists select cytokines with direct involvement in inflammatory cascades. Consistent with the systemic findings discussed above, IL-6, which has pro-angiogenic properties, has been found in higher concentrations in aqueous fluid of eye with neovascular AMD compared to controls.^{68, 69} However, as with many observations from human samples, whether these cytokines are involved in the primary pathogenesis of AMD or are a result of the disease process remains to be determined.

Table 2.

Select inflammatory cytokines elevated in intraocular fluids of neovascular AMD patients

Cytokine	Function	References
IL-6	Acute phase reactant; pyrogenic; pro-angiogenic	68, 69
IL-8/CXCL8	Chemokine for neutrophils; pro-angiogenic	68, 69
MCP-1/CCL2	Chemokine for monocytes, memory T cells, and dendritic cells	67, 69, 70
ICAM-1	Adhesion molecule that facilitates transmigration of leukocytes	69, 70

IL: Interleukin; MCP: Monocyte chemotactic protein; ICAM: Intercellular adhesion molecule

Inflammasome activation in AMD

Recently, inflammatory signaling cascades within RPE cells have been implicated in AMD pathogenesis. The term inflammasome refers to multi-protein intracellular complexes formed in response to stimulation of pattern recognition receptors, such as NOD-like receptors (NLR), by microbes, toxins, or products of cellular stress.⁷¹ Inflammasome activation leads to production of active caspase-1, an enzyme capable of cleaving and activating several pro-peptides, including IL-18. Drusen isolated from human eyes with AMD were shown to activate the NLRP3 inflammasome in vitro.⁵⁸ In addition, RPE from human eyes with GA was shown to have increased mRNA levels of NLRP3 and IL-18 and increased protein levels of NLRP3 and caspase-1 compared to age-matched normal controls.⁷² The same group also reported increased levels of a non-coding RNA sequence known as Alu RNA as well as decreased levels of the RNase DICER1 in RPE from human eyes with GA.⁷³ DICER1 cleaves and inactivates Alu RNA; therefore, deficiency in DICER1 could result in increased levels of Alu RNA, which was shown to activate the NLRP3 inflammasome with subsequent elaboration of active IL-18.⁷² Furthermore, RPE from human eyes with GA were shown to have increased protein levels of the phosphorylated IRAK1 and IRAK4,⁷² which are involved in MyD88 signaling, as well as active caspase 8⁷⁴ compared to controls. Inhibiting caspase 8 in vitro prevented RPE cell death.⁷⁴ Taken together, these findings suggest that DICER1 deficiency may result in increased levels of Alu RNA, which activates the NLRP3 inflammasome resulting in production of active IL-18, which signals via MyD88 to elicit active caspase 8, which causes apoptosis of RPE cells. Why DICER levels are decreased in eyes with GA remains to be determined. Nonetheless, targeting this pathway may lead to novel therapeutic strategies for GA, which currently has no treatment.

Microglia and Macrophages in AMD

Innate immune cells, such as microglia and macrophages, are sentinels in the body’s response to tissue insult and function to initiate inflammatory responses, clear debris, and remodel tissue to a state of homeostasis. Microglia are specialized myeloid-derived cells that reside in the retina and central nervous system (CNS). In rodent models, microglia enter the CNS prior to closure of the blood-brain barrier, and their population is maintained through self-renewal in the absence of known stem cells.⁷⁵ Microglia cell bodies are typically situated in the inner retina in the normal human retina.²⁶ These cells can migrate to the subretinal space in response to inflammatory stimuli.¹⁰ Activated microglia have been found in the outer retina and subretinal space in eyes with AMD.^{26, 76} Retinal microglia express the chemokine receptor CX3CR1.²⁶ Genetic polymorphisms in the CX3CR1 gene have been associated with AMD and reduce the chemotactic ability of monocytes.²⁶ In murine models, deficiency of CXCR3 lead to subretinal accumulation of microglia with normal aging and exacerbated laser-induced CNV.

Several studies have identified the presence of macrophages in or around drusen, areas of geographic atrophy, and choroidal neovascular membranes on pathologic analysis of eyes from AMD patients.^{53, 77–82} However, it is unclear from these studies whether macrophages are involved in the development of these pathologic findings or accumulate as a consequence of the pathology. In support of a causative role for macrophages in wet AMD, macrophages from human CNV tissue were found to express the pro-angiogenic protein VEGF, suggesting that macrophages are involved in the development of CNV lesions.⁸³ Alternatively, macrophages from human CNV tissue have also been shown to express scavenger receptors for oxidized proteins, such as lectin-like oxidized low-density lipoprotein receptor-1 (LOX-1), suggesting that these cells may migrate to sites of oxidative damage to clear debris.^{84, 85} However, these conclusions are not mutually exclusive, and animal models of laser-induced CNV have also provided conflicting results with some studies suggesting macrophages contribute to CNV formation while others show a protective role for macrophages in preventing CNV.^86–88 Macrophages are a heterogeneous and can adopt a spectrum of phenotypes. For instance, M1 macrophages are proinflammatory, while M2 macrophages are relatively anti-inflammatory and may be pro-angiogenic in certain microenvironments.⁸⁹ Interestingly, M1 macrophages were more common in the eyes with GA AMD and M2 macrophages were more common in the neovascular AMD eyes.⁹⁰ Therefore, the polarization of macrophages within AMD lesions may in part determine the course of pathology. Further research is required to determine the precise role of microglia and macrophages in human AMD pathogenesis.

Adaptive immunity in AMD

Adaptive immunity refers to antigen-specific responses generated by B and T lymphocytes against peptides the body detects as foreign. Direct evidence implicating the adaptive immune system in AMD is sparse in humans. Several studies have reported the presence of anti-retinal antibodies in the serum of AMD patients.^91–93 Expression of major histocompatibility (MHC) class II molecules, which are required for presentation of antigen to CD4 T cells, is seen on RPE in human eyes with AMD (Figure 1). Furthermore, T cells from AMD patients have been shown to proliferate in response to retinal antigens (unpublished data from RBN). Therefore, the adaptive immune system is able to sense and respond to retinal antigens in patients with AMD. However, whether these responses are the cause or result of retinal damage is unknown.

Figure 1.

HLA-DR expression on RPE in eyes with (A) geographic atrophic and (B) neovascular AMD. Arrows indicate areas of strongly positive staining.

As people age, the immune system undergoes changes referred to as immunosenescence, a phenomenon perhaps most apparent in the T cell compartment where the ratio of memory to naïve T cells drastically increases, in part due to reduced thymic output of naïve T cells with age.⁹⁴ In addition, T cells undergo phenotypic changes with age, including down-regulation of CD28, a co-stimulatory molecule involved in the generation of antigen-specific responses, and upregulation of markers typically seen on natural killer (NK) cells, such as CD56, that may allow antigen non-specific reactivity. A case-control study of over one hundred AMD patients and controls reported an increased frequency of CD28⁻CD56⁺ T cells in the peripheral blood of AMD patients compared with controls.⁹⁵ Most of this difference appeared to be in the CD8 T cell compartment. These results suggest that heightened immunosenescent responses may play a role in AMD pathogenesis. Unfortunately, the control group was significantly younger than the AMD group confounding interpretation of these results.

As discussed above, CD4 T cells from AMD patients secreted IL-22 and IL-17A in a monocyte dependent manner in response to the complement component C5a,⁴⁹ which has been found in drusen and is also elevated in the serum of AMD patients. In addition, IL-17 induced angiogenesis in human choroidal endothelial cells and cell death in ARPE-19 cells in vitro.^{50, 51} Therefore, interplay between the innate (complement system and monocytes) and adaptive (T cells and their cytokines) may be involved in the complex pathogenesis of AMD.

Immunotherapy for AMD

There are several recently completed and ongoing clinical trials evaluating immune-based interventions for both dry and wet AMD. As discussed above, T cells may play a role in AMD pathogenesis. In patients with GA, subconjunctival sirolimus, an mTOR inhibitor that suppresses T cell responses, did not reduce GA area compared to fellow non-treated eyes.⁹⁶ In a randomized non-masked trial in patients with CNV, however, systemic sirolimus treatment was shown to significantly reduce the number of anti-VEGF intravitreal injections required to control CNV.⁹⁷ In the same study, systemic daclizumab, a monoclonal antibody directed against the alpha subunit (CD25) of the IL-2 receptor expressed on T cells, also reduced the number of anti-VEGF injections in patients with CNV. These results suggest that systemic inhibition of T cell responses may alter the clinical course of wet AMD.

TNF-α has been localized to human CNV lesions with the suggestion that it may promote angiogenesis.⁹⁸ Systemic treatment of neovascular AMD patients with the anti-TNF-α monoclonal antibody, infliximab, had no effect on the number of anti-VEGF injections required or visual acuity, but the power of the study was low.⁹⁷ A number a case reports describe the use of intravitreal anti-TNF-α agents in neovascular AMD patients refractory to anti-VEGF therapy; however, all reports describe significant intraocular inflammation as a complication.^99–101

Broad spectrum immunosuppressive medications have also been tested in neovascular AMD. Topical bromfenac, a non-steroidal anti-inflammatory drug (NSAID),¹⁰² and intravitreal triamcinolone,¹⁰³ have been tested in conjunction with anti-VEGF intravitreal injections in the treatment of neovascular AMD. While having no effect on visual acuity, both agents reduced the number of required anti-VEGF injections. Therefore, therapies directed against certain inflammatory molecules and pathways may alter the clinical course of AMD.

Conclusions

Several lines of evidence implicate various components of the immune system in AMD pathogenesis. However, the exact inflammatory mechanisms involved remain elusive. While vision loss attributable to AMD-associated CNV has been drastically reduced over the past decade with the introduction of intravitreal anti-VEGF therapies, there remains no effective treatment for the more common atrophic form of late AMD. Therapies targeting several components of the complement cascade are currently being tested.¹⁰⁴ Clinical trials are also underway investigating transplantation of stem cell-derived RPE into areas of GA in human patients.¹⁰⁵ This raises additional issues regarding inflammatory responses and their potential effect on photoreceptors and RPE. Beyond the inflammatory reaction induced by surgery, RPE derived from fetal embryonic stem cells will likely be seen as foreign by the host requiring systemic immunosuppression to prevent rejection.

Oral tolerance is another potential therapeutic avenue that may prove beneficial in the treatment of AMD. Induction of immunologic tolerance, meaning a specific dampening of immune responses toward specific antigens, has been accomplished by oral administration of antigens to humans with several inflammatory conditions,^106–108 including uveitis where patients fed S antigen were able to stably discontinue systemic immunosuppressive medications.¹⁰⁹ It is clear that further research is needed to elucidate the inflammatory mechanisms involved in AMD pathogenesis if novel targeted therapeutics are to be effective.