2. Central Role of Inflammation and Parainflammation in AMD
2.1. Clinical Evidence of Inflammation
In the early and intermediate stages of macular degeneration, measurable photoreceptor dysfunction is reflected in abnormalities dark adaptation, visual field, photo stress, and electro-retinographic changes [
36,
37,
38,
39]. These disease-correlated changes in visual function demonstrate inflammation as the underlying pathology behind all stages of macular degeneration. Prolonged dark adaptation, which progressively worsens in step with stage of AMD, is caused by inflammatory visual cycle impairment and indicates a worsening of inflammation as AMD progresses [
40,
41]. In one study, patients with early AMD had qualitative visual changes and symptoms of distortion that could be detected and quantified by visual field analysis [
37]. These deficits of form recognition and sensitivity were found in areas of RPE atrophy not defined as geographic atrophy within this patient population, indicating asymptomatic vision loss in areas of overactive inflammation [
37]. Photo stress recovery time, a measure of visual pigment recycling time, was inversely correlated with visual acuity, and inflammation-induced prolongation of recovery time was directly correlated with the presence or absence of geographic atrophy and advancing age [
39]. In late AMD, inflammatory slowing of implicit time and amplitude reduction on electroretinograms (ERGs) of patients with geographic atrophy were seen in areas bordering fundus autofluorescence-defined geographic atrophy [
42]. Foveal ERG performed in fellow eyes of patients with wet AMD who did not have severe visual acuity changes demonstrated implicit time prolongation, indicating that fellow eyes were also undergoing inflammation [
38]. Visual cycle alteration represented the first indication of the central role of inflammation in AMD.
2.2. Anti-Oxidant Therapy for Early-Stage AMD
While clinical findings in AMD pointed to inflammation as the central driver of AMD progression, initial therapeutics for AMD were focused on anti-oxidation. To this day, the only consensus treatment for early/intermediate-stage AMD is the use of anti-oxidant therapies studied in a large National Eye Institute study called the Age-Related Eye Disease Study 1 and 2 (AREDS 1 and 2) [
43]. This study has produced over 28 reports and countless publications on the benefit of anti-oxidant therapy for the prevention of late AMD progression [
44]. While the conclusions from these studies demonstrated a reduction in the rate of development of large drusen, geographic atrophy, and exudative AMD, the modest reduction in rate of progression implicates oxidation as only a stimulator of inflammation rather than the main driver for AMD progression [
45,
46].
2.3. Oxidation-Induced Dysfunctional Parainflammation in Early AMD
Demographic, environmental, and genomic risk factors combined with electrophysiological and psychophysical studies definitively implicate retinal inflammation as the major underlying factor in the development of all stages of AMD [
47,
48,
49,
50]. The degree of dysfunctional inflammation determines the stage and clinical presentation [
51]. Multiple papers have described macular degeneration as a disease that is initiated by dysregulated parainflammation leading to the drusen stage of AMD then progressing to overt inflammation that triggers the late stage of AMD [
52,
53,
54,
55]. In 2008, Medzhitov postulated a parainflammatory state that lay between basal homeostatic conditions and true inflammation [
56]. This parainflammatory state is considered an adaptive immune response to low level tissue stress such as the age-related accumulation of oxidative byproducts [
57]. This acquired dysfunctional parainflammation that occurs with aging has been termed inflamaging and likely explains the early stages of AMD with drusen development and RPE changes [
58]. The development of late exudative and geographic atrophy stages, on the other hand, is an overt innate immune activation with end-stage pathology determined by macrophage polarization [
59].
Oxidative damage is the main tissue stress that stimulates parainflammation in the retina [
60] (
Figure 1(1)). The macula, which is exposed to photo, metabolic, phagocytic, and mitochondrial reactive oxygen species (ROS) production, is a site of tremendous oxidative stress. An association between blue light exposure (photo-oxidative light) and low anti-oxidant levels demonstrated an association with early and neovascular forms of AMD, implicating photo-oxidative light as a contributor to oxidative stress in AMD [
61].
Figure 1. Early/intermediate-stage AMD. (1) Photo-oxidative blue light (2) oxidizes retinal lipids to produces oxidative byproducts (MDA, CEP) and reactive oxygen species, (3) which activate microglial cells to secrete cytokines. These activated macrophages are hindered from becoming phagocytic due to the (4) upregulation of ST6 that produces (5) sTN on the surface of retinal cells and drusen that agonizes Siglecs to prevent phagocytosis of drusen. (6) Reactive oxygen species also cause photoreceptors to secrete CX3CL1, which promotes migration of microglia and macrophages to the retina. (7) Oxidized phospholipids stimulate RPE cells to produce MCP—1 (8), which recruits peripheral blood monocytes (PBMCs) to areas of phospholipid oxidation. (9) Upregulation of FUT5 on PBMCs localizes monocytes to areas that are secreting chemokines such as CX3CL1 that upregulate e-selectin on vascular endothelium. (10) Monocytes are found between RPE cells and Bruch’s membrane. (11) Microglial cells and macrophages are also found to be pigment-ladened in this sub-RPE space and at the RPE cell layer, which appears as RPE pigment clumping. (12) The overactivation of complement caused by the polymorphism-induced impaired function of CFH will produce complement pathway metabolites that accumulate in drusen and can activate microglial cells.
Photo-oxidative stress, which is focused and concentrated onto the macula by the ocular lens system [
62], produces oxidative byproducts such as malondialdehyde (MDA) [
63,
64], 2-(ω-carboxyethyl)pyrrole(CEP) [
65] (
Figure 1(2)), and oxidized phospholipids (
Figure 1(7)). CEP alone can activate microglia and initiate findings of early-stage AMD [
66]. Oxidized phospholipids stimulate RPE to upregulate MCP-1m a major chemoattractant to macrophages [
67] (
Figure 1(8)). After photo-oxidative stress, fractalkine upregulation from the photoreceptor segment and outer nuclear layers mediates the cross talk between photoreceptors and microglia expressing the CX3CR1 receptor for fractalkines. This release of fractalkine serves to attract and activate microglia towards the injured photoreceptors [
68] (
Figure 1(6,7)).
Photoreceptors are constantly being bombarded with oxidative-, metabolic-, and toxic-mediated stress. Over exposure to photo-oxidative blue/violet light can cause S-cone retinal degeneration and apoptosis [
62]. A fluorophore called lipofuscin is another source of ROS. Lipofuscin has been identified as Bis-retinoid N-retinyl-N-retinylidene ethanolamine (A2E), an autofluorescent toxic visual cycle metabolite that has been implicated in the progression of geographic atrophy [
69]. A2E cannot be enzymatically degraded, and accumulations in RPE cells and in healthy youthful eyes are cleared by microglia, but those in elderly eyes are cleared by microglia and peripheral blood-derived macrophages [
70]. When lipofuscin is exposed to blue/violet light, it produces more ROS, making it yet another source of oxidative stress [
71].
RPE cells, due to their role in autophagy of the photoreceptor outer segment (POS), produce metabolically and phagocytic-activity-derived ROS [
60]. RPE cells are also enriched with mitochondria to support their high metabolic activity, which in themselves are a major source of ROS [
72]. RPE cells are the main housekeepers of the retina and are the most phagocytically active cells in the whole human body [
73]. Because of the constant physiologic destruction of the POS from oxidative stress, the POS must be constantly shed and renewed. Through a process called heterophagy, RPE cells must ingest the shed POS and degrade them with their lysosomes [
74]. Autophagy, a self-eating phenomenon, regulates oxidative stress as well as homeostasis of proteins, lipids, nucleic acid, and mitochondria and protects the cell from oxidized lipids, misfolded proteins, damaged organelles, and other cellular waste or faulty cellular components and directs them to the cellular lysosome [
73]. Mitophagy, a specific form of autophagy, eliminates damaged or mutated mitochondria, which produce significant amounts of ROS [
75]. Heterophagy, autophagy, and mitophagy are RPE cell functions that counterbalance the extreme oxidative environment and enable these cells to function as the main anti-oxidant protector of the retina [
73,
76]. Age-related impairment in these anti-oxidative RPE functions initiate parainflammation and early macular degeneration [
77].
To enhance their anti-oxidative function, RPE cells also utilize nuclear factor erythroid 2 (Nrf2), a basic leucine transcription factor, to protect cells against oxidative stress by coordinating a transcriptional program that protects from oxidative cellular injury but allows cellular redox homeostasis [
60]. In a non-oxidative environment, NRF2 is sequestered and degraded in the cytosol by Kelch-like ECH- associated protein 1 (Keap1), which is a substrate adaptor protein to a ubiquitin ligase complex, which prevents NRF2 from translocating into the nucleus. Interaction between Keap1 and ROS releases Nrf2 and translocates it into the nucleus, binding to the anti-oxidant response element (ARE). This binding results in the secretion of anti-oxidant enzymes like catalase and superoxide dismutase as well as regulating glutathione and thioredoxin levels [
78]. An additional role for Nrf2 is to maintain mitochondria by controlling the expression of several mitochondrial enzymes that promote adenosine tri phosphate production (ATP) [
79].
With age, the production of Nrf2 mRNA and protein decline along with the anti-oxidant response [
80]. Aging also brings about decreased RPE phagocytosis, increased mitochondrial DNA damage, a decrease in mitophagy, reduced heterophagy, and dysfunctional autophagy [
73]. This age-related impairment leads to accumulation of oxidative byproducts such as malondialdehyde (MDA) [
63,
64], 2-(ω-carboxyethyl)pyrrole(CEP) [
65] (
Figure 1(2)), and oxidized LDLs [
81], which are stimulators of parainflammation and implicated in the initiation and progression of early AMD [
66].
2.4. Complement Pathway-Induced Parainflammation
Despite the strong association of complement factor polymorphisms with the development of AMD and the development of two FDA-approved treatments for geographic atrophy (Pegcetacoplan and Avacincaptad pegol), there is a consensus that complement does not fully account for disease development [
82,
83,
84,
85]. Its direct role in disease may be in the development of drusen, and it may play only an indirect role in the development of late-stage disease [
86] (
Figure 1(12)). Evidence to support this role in producing drusen is found in the proteomic analysis of drusen from retinas of patients with AMD, where a large proportion of patients had complement proteins 9 (C9) and 3(C3) in their drusen [
87]. C9 is a protein found in conjunction with the c5-9 complex or membrane attack complex. This finding agreed with earlier histopathologic analysis that corroborated the presence of multiple complement factors in drusen [
88]. A meta-analysis found that systemic complement overactivation was a feature associated with early/intermediate AMD rather than late-stage geographic atrophy [
89].
While no trials have been performed to determine the effect of complement factor depletion on the development of drusen, several trials have looked at the effect of complement C3(Pegcetacoplan) and C5 (Avacincaptad pegol) depletion and demonstrated a modest reduction in the growth rate of GA [
90,
91]. According to both pre-clinical and clinical studies, complement overactivation is important in initiating AMD by exacerbating parainflammatory overactivation and dysfunction, which enhances drusen formation. The inability of profound complement depletion to halt the progression of late-stage GA relegates complement as a minor player in the pathogenesis of late-stage AMD [
90,
91].
2.5. Early-Stage AMD Dysfunctional Parainflammation
At the cellular level, the early/intermediate stage is the accumulation of oxidized, metabolic, inflammatory debris that appears as yellow sub retinal deposits called drusen [
87]. The pigmentary clumping seen in AMD represents pigment-ladened microglia or macrophage migration towards the retina, Bruch’s membrane, and under surface of the RPE [
92] (
Figure 1(9–11)). The accumulation of proinflammatory oxidative byproducts CEP and MDA leads to phagocytic/cytokine secreting microglial polarization, activation of parainflammatory mechanisms, and, potentially, recruitment of peripheral blood macrophages to the subretinal space to compensate for this accumulation of subretinal toxic substances [
55] (
Figure 1(2,3,10)). With age-dependent RPE cell senescence and dysfunction of autophagy, parainflammatory activation of microglial cells accelerates, resulting in the recruitment of peripheral blood-derived macrophages into the subretinal space [
93].
2.6. Microglia’s Central Role in Dysfunctional Parainflammation
Microglia play a central role in modulating parainflammation [
94] (
Figure 2(1)). Microglial overactivation represents a common pathomechanism in a variety of retinal degenerative diseases and is often overactivated prior to the onset of overt retinal cell death [
95]. In the retina, microglia’s dynamic motility allows comprehensive surveillance coverage of the entire retina in a short time period [
96] (
Figure 2(5)). This motility allows microglia to interact with retinal neurons and macroglia and play a central role in retinal homeostatic maintenance and clearance of cellular and metabolic debris [
97].
Figure 2. Microglial parainflammatory regulation: Microglia remove toxic metabolites, apoptotic cells, and oxidative debris while maintaining the health of photoreceptors and retinal cells in a process called parainflammation. (1) Parainflammation is initiated by pattern recognition receptors found on (2) M0 microglia such as TLR, TREM2, CX3CR1, NLRP3, and CD36, which use these receptors to bind and sense toxins, apoptotic cells, damage, and pathogen-associated molecular patterns. Once microglia are activated, they polarize to the (3) M1 like state, (4) secrete proinflammatory cytokines, and (5) migrate to the areas of stressed RPE cells and photoreceptors. (6) Upon activation, neuropilin-2-PSA and E-selectin ligand 1-PSA are secreted by these activated microglia. (6) The PSA on these glycoproteins binds Siglecs on activated microglial cells (7) to polarize them into the M2c healing state (8), which releases growth factors to protect and regenerate the stressed photoreceptors and RPE cells. (9) This homeostatic maintenance function of microglial parainflammation, if not adequately modulated with sialic acid checkpoint regulation, will result in recruitment of peripheral blood monocytes and AMD disease progression. (10) A PSA-nanoparticle (PSA-NP) mimics E-selectin ligand 1 PSA and Neuropilin 2-PSA to (7) polarize M1 activated microglia into the M2c healing state.
The retinal environment is unique amongst all other CNS compartments in that it undergoes constant damaging stress, which in youth is maintained by RPE cells but with age and RPE cell senescence will lead to a constant microglial and parainflammatory activation state [
54] (
Figure 2(1)). To maintain this homeostatic retinal parainflammation, the regulators of microglial function have to be tightly controlled to determine the species of microglia that predominate. The different microglial states can be M0 sentinel and resting (
Figure 2(2)), M1 proinflammatory and phagocytic, (
Figure 2(3)), M2d anti-inflammatory and proangiogenic, M2a, b anti-inflammatory and profibrotic, and M2c anti-inflammatory, neuroprotective, and healing (
Figure 2(7,8)). If the balance between these polarization states tilts towards any other states besides M2c, pathology will ensue.
Activated microglia in the retina have been implicated in many retinal pathologies. In SD-OCTs, hyperreflective spots are visualized in AMD and diabetic retinopathy, which represent large amoeboid microglial cells, the activated microglial appearance, as opposed to the resting ramified microglial configuration [
98,
99]. Macrophages and microglia are bloated with engulfed membranous material near areas of drusen, suggesting that drusen components are restraining or checking phagocytic function [
100,
101]. While microglia can maintain retinal homeostasis during the first 5 to 6 decades of life, the onset of drusen and pigmentary changes signal the migration of microglia and the recruitment of peripheral blood-derived macrophages to Bruch’s membrane [
102]. This sequence of events indicates the requirement of peripheral blood monocyte recruitment and loss of checkpoint restraint of these phagocytic cell in the progression of the drusen stage to the late stages of AMD.
2.7. Peripheral Blood-Derived Macrophages’ Role in Transition to Late AMD
The major role of peripheral blood monocyte-derived macrophages in the progression of the early drusen stage to late-stage geographic atrophy and exudative AMD is evidenced by the increase in number of activated macrophages in the choroid and Bruch’s membrane as AMD progresses. The observation that the highest number of activated macrophages are found in eyes with choroidal neovascularization further supports this central role [
102].
The recruitment of peripheral blood-derived macrophages is a major function of retinal microglial cells when injury or toxic material overwhelm the microglia’s ability to phagocytose toxic oxidative byproducts, injured cells, or apoptotic cells [
103]. To maintain macular health, the microglia must phagocytose oxidative byproducts (oxidized phospholipids, malondialdehyde, carboxyethyl pyrrole), apoptotic cells, complement proteins, abnormal proteins (Amyloid B), and toxic metabolites (A2E). If microglia do not encounter appropriate checkpoint ligands in the form of sialic acid, then chemokine signaling will recruit peripheral blood macrophages [
104] (
Figure 2(9)).
Patho-mechanistically, the recruitment of peripheral blood monocytes in clinically evident macular degeneration is mediated by microglial chemokine signaling, which is responsible for the recruitment of monocytes to the choroid and the retina. The chemokine receptors CCR2 and CX3CR1 and their respective ligands CCL2 (monocyte chemotactic protein-1, MCP-1) and CX3CL1 (fractalkine) mediate this recruitment [
68,
105,
106].
While fractalkine-CX3CR1 signaling is proinflammatory, it is also critical for progesterone-mediated neuroprotection of the retina [
107]. This dual role of Fractalkine-CX3CR1 interaction in microglial proinflammatory activation and neuroprotective properties demonstrates the necessity of tight regulation of microglial cells. If the proinflammatory properties of fractalkine-CX3CR1-activated microglia could be checked, but they could maintain their neuroprotective properties, then polarizing macrophages to the resolution state would eliminate inflammation and attenuate phagocytosis while providing neuroprotection. Without appropriate checkpoint regulation of microglial cells, the inflammatory activation state of microglia will promote more photoreceptor degeneration rather than rescue of photoreceptors.
2.8. Macrophage Recruitment Indicator of Late-Stage AMD
The central role of macrophages in the pathogenesis of neovascular wet AMD is widely accepted [
108,
109,
110,
111], but the critical role of activated macrophages in the pathogenesis of geographic atrophy has been widely overlooked due to the focus on the complement pathway [
112]. Prior to the genomic association between complement factors and AMD, histopathologic evidence pointed to macrophage/mononuclear phagocyte/multinucleated giant cells as the central causative factor in the pathogenesis of GA and the main phagocytic cause of retinal cellular clearance that manifests as RPE and photoreceptor loss [
101,
113] (
Figure 3(5,6,9)).
Figure 3. Geographic atrophy: (1) Progressive accumulation of oxidative byproducts (CEP, MDA) and (2) chronic overactivated complement pathway chronically activate microglia (3), which secrete neuraminidase and desialylate photoreceptors and RPE cells. (4) This loss of sialylation prevents restoration of homeostasis. (5) Chronically overactivated phagocytosis and inflammation recruits peripheral blood macrophages by upregulation of the fucosyltransferase FUT5 that produces Lewis X glycosylation on monocytes to bind E-selectin and (6) localize monocytes to sites of inflammation (7), allowing for diapedesis of the monocyte across the blood–retinal barrier. (8) These monocytes polarize to M1 macrophages when they enter the retina and are not able to clear substances like lipofuscin and other undegradable substances. (9) The macrophages form multinucleated giant cells because they phagocytose structures that are undegradable. (10) Since there is no sialic acid or polysialic acid to polarize to the healing M2c state, the macrophages are unchecked and result in elimination (11) first of the RPE cells then the (12) photoreceptors. The unchecked macrophages are the main determinant of growth of geographic atrophy.
One of the earliest demonstrations of macrophages’ central role was on electron microscopic/histopathologic studies of postmortem retinal specimens of seven patients with geographic atrophy [
114]. In this study, mononuclear phagocyte series (MPS) cells (situated between the basement membrane of the RPE and inner collagenous layer of Bruch’s membrane) and multinucleated giant cells (derived from macrophage fusion created by inability to degrade constituents of Bruch’s membrane or RPE pigment) were found in or around areas of geographic atrophy [
114]. Interestingly, no phagocytic RPE cells, which at that time were thought to be the cause of cellular loss, were found at the border or within GA. This hallmark study implicated microglia and macrophages as the main phagocytic cells in the pathology of geographic atrophy as opposed to RPE cells, the predominant phagocytic cell in the normal retina [
73,
101]. Clinico-pathologic correlation demonstrated that both MPS and giant cells containing pigment are typically found in the outer edge of the geographic atrophy lesion or in areas of coarse pigmentary mottling, which is clinically predictive of the development of geographic atrophy [
115,
116]. The close apposition of activated macrophages/microglial cells at the leading edge of GA indicates that it is monocyte-derived macrophages that phagocytose RPE and photoreceptors that lead to the development and growth of geographic atrophy in AMD [
101].
The question that these findings raised in the 1990s was how the activation of macrophages resulted in the development of atrophic and neovascular phenotypes, which were considered two very different processes at the time. The answer was slowly revealed over the next two decades with the discovery of different macrophage/microglial polarization states [
117]. The potential for macrophages/microglia to polarize between a phagocytic proinflammatory and a vascular endothelial growth factor-producing angiogenic polarization state determine the development of geographic atrophy, exudative AMD, or both [
59].
2.9. Macrophage Polarization Determines Late-Stage AMD
The classically activated (M1) and alternatively activated (M2) binary description of macrophage polarization, based on biomarker expression and cytokine production, does not reflect the true character of these subtypes. A function-based description of macrophage polarization better characterizes their role in pathology [
59,
117,
118]. The M1 polarization state is characterized as the proinflammatory phagocytic state. The M2 state can be subdivided into four M2 subtypes. The M2 a, b subtype are the anti-inflammatory pro-fibrotic type, the M2d is the pro angiogenic phenotype, and the M2c is the anti-inflammatory and neuroprotective type [
117,
118,
119]. The M2c can also dedifferentiate myofibroblasts, so is considered anti-fibrotic [
120].
In vitro, M1 macrophages are predominantly neurotoxic with modest axon growth-promoting effect, in contrast to the M2 macrophages, which promote long-distance axon growth without neurotoxicity [
121]. In vivo studies in traumatic brain or spinal cord injury have characterized the time course and characteristics of M1/M2 polarization after injury [
121,
122].
M1-like macrophages release oxidative metabolites and proteases that kill neurons and glial cells [
121]. In contrast, M2-like cells facilitate tissue repair [
123]. In spinal cord injury models, increased M2c microglia expressed in the first week after injury correlated with better neurological outcome, indicating a healing neuroprotective function of M2c microglia/macrophages [
122]. A time course study comparing M1 versus M2 levels in this model show that M1 expression is upregulated for at least a month, while M2 levels diminish drastically 1 week post injury. The level of M2a and M2c upregulation within the first week correlated well with neurological recovery. The neurological recovery resulted from the neuroprotective properties of microglia polarized to the M2c or M2a state. These properties were the secretion of trophic factors such as brain-derived neurotrophic factor (BDNF) and glial-derived neurotrophic factor (GDNF) and the ability to perform controlled phagocytosis, which prevents necrosis of surrounding tissue [
124] (
Figure 2(8)).
While at first glance, macular degeneration may not appear to be a disease of acute CNS injury, the macrophage microglial behavior in the retina is consistent with what is seen in injury models of the CNS [
55,
125]. Eyes with advanced AMD had a higher ratio of M1 to M2 macrophages than age-matched normal autopsied eyes [
126]. This enhanced chronic expression of neurotoxic phagocytic M1 macrophage with the reduction in M2c neuroprotective macrophages results in unchecked photoreceptor, RPE cell degeneration, and phagocytosis (geographic atrophy) [
126]. The abundance of M1 polarized macrophages and adenosine polarize M1 to M2D VEGF-producing macrophages, resulting in the development of subretinal neovascularization [
127]. If the M2 a, b polarization predominates, then retinal fibrosis will develop (disciform scar) [
59]. Like CNS injury, the failure to polarize M1 macrophages towards the M2 c state will result in failure of functional recovery [
122] (
Figure 4).
Figure 4. Macrophage polarization determines AMD phenotype. The plasticity and different polarization states correlate with the clinical picture seen in late-stage macular degeneration. In geographic atrophy, RPE cells and photoreceptors are phagocytosed as a function of the M1 polarized phagocytic macrophages. Exudative AMD is neovascularization produced by overexpression of VEGF, the main cytokine secreted by the M2d polarization state. The sequelae of exudative AMD untreated with anti-VEGF therapies is a disciform fibrotic scar in the control of the pro-fibrotic M2 A, B state. With appropriate sialic acid signaling such as PSA, all polarization states can transform into the healing M2c state.
2.10. Sialylation Controls Microglial Activation and Macrophage Polarization in AMD
Non-genomic factors determine microglial activation and macrophage polarization in AMD. To determine which of these factors were important in AMD, Emilsson et al. performed a large proteo/genomic analysis of serum proteins in patients with macular degeneration [
128]. This group identified several serum proteins that were elevated differentially in early and late AMD. As expected, serum protein complement factor H related 1 (CFHR1) correlated with early and late AMD. Complement factor H related 5 (CFHR5) correlated only with late-stage AMD. A novel but important finding from this study was the differential correlation of a class of proteins called sialyltransferase with different stages of AMD [
128].
In this analysis, elevated serum protein levels of Alpha-N-acetylgalactosaminide alpha-2,6-sialyltransferase 1 (ST6GALNAC1/ST6) and Alpha-(1,3)-fucosyltransferase (FUT5) were highly correlated with patients with early-stage AMD [
128]. In late-stage AMD patients, only FUT5 was found to be highly correlated. The expression of FUT5 in both late and early AMD and the expression of ST6 in only early AMD point to alterations of sialic acid glycan expression as a major determinant in the progression to late-stage AMD. It also implicates the absence of sialic acid self-associated molecular patterns (sSAMPs) in late AMD, which bind sialic acid-binding immunoglobulin-like lectins (Siglecs) to resolve activated macrophages to the resolution M2c state [
129].
The loss of ST6 expression and the upregulation of FUT 5 in late AMD explain why microglia and macrophage activation is unchecked. When ST6 is expressed, the sialylated Tn antigen (sTn) rapidly sialylates the inflamed or oxidatively damaged glycocalyx of cells as well as the debris of these damaged cells [
130]. This sialylation of the debris found in drusen [
131] by sTN resolves activation of microglial cells, resulting in reduced phagocytosis, no recruitment of peripheral blood macrophages, no overt inflammation, and eventual accumulation of metabolic and inflammatory debris (
Figure 1(9)). This debris, combined with complement pathway-created proteins and lipid byproducts, will accumulate in the subretinal space and worsen drusen and pigmentary clumping [
88].
During the lates stage of AMD, ST6 is no longer upregulated; instead, FUT5 becomes the predominant sialyltransferase. FUT5 is a critical glycotransferase that produces Lewis x (3Gal
𝛽�1,4[Fuc
𝛼�1,3] GlcNAc-), sialyl Lewis x (sLe
x,1 NeuAc
𝛼�2, 3Gal
𝛽�1,4[Fuc
𝛼�1,3]GlcNAc-), Lewis a (Le
a, 3Gal
𝛽�1,3[Fuc
𝛼�1,4]GlcNAc), sialyl Lewis a (sLe
a, NeuAc
𝛼�2, 3Gal
𝛽�1,3[Fuc
𝛼�1,4]GlcNAc-), and Lewis b (Le
b, Fuc
𝛼�1,2Gal
𝛽�1,3[Fuc
𝛼�1,4]GlcNAc-). These glycans are the main binding determinants for selectins, in particular, e-selectin, which is responsible for localizing monocytes to areas of active inflammation [
132]. Without sialic acid SAMP checkpoint restraint on activated microglia and recruited macrophages, immune cell activation will overproduce cytokines such as VEGF and phagocytose both drusen and underlying RPE cells and overlying photoreceptors, resulting in geographic atrophy (
Figure 3(4,5)).
Clinical and proteomic findings in all stages of macular degeneration point to loss of microglia and macrophage sialic acid/Siglec immune checkpoint control as a major determinant of disease progression to late-stage geographic atrophy or CNV [
133,
134]. If appropriate sialic acid-mediated checkpoint control could be regained and homeostatic neuroprotective microglial function restored, late-stage disease could be halted and inflammation-impaired visual function could potentially be improved [
135].
The alteration of the glycocalyx, “the sugar coat” of retinal cells, which in healthy young retinas are decorated with sialic acid caps (SAMPs), which agonize Siglec checkpoint receptors, determines the activation and polarization state of microglia and macrophages. By either binding to complement factors such as CFH and properdin or to Siglec receptors on immune cells, these sialic acid caps serve as checkpoint ligands to resolve innate immune activation [
104]. This overlooked sialic acid sugar coating that determines the immune self is the master checkpoint regulator of immune cells and immune function [
136].
Even in a non-inflammatory environment, the sugar coating on immune and host cells determines the polarization state of microglia and macrophages [
137]. If there are enough sialic SAMP patterns, microglia are placed into the resolved state, which allows them to take on a neuroprotective and regenerative phenotype by producing cytokines such as BDNF or CTGF and selectively pruning neuronal dendrites and apoptotic cells [
118,
138]. In this sialic SAMP-checked state, microglia also can sense stressed, damaged, and sick cells or those undergoing apoptosis, which activate the microglia to the transient M1-like state to eliminate the damaged cell-released toxins and apoptotic cells [
139]. If there are adequate sialic SAMPs available to bind the microglial Siglecs, the activation is short-lived and will result in the repolarizing of these M1 microglia to the M2c-like state [
118].
Macrophage polarization determines how macular degeneration will clinically manifest, but it is the loss of cell surface sialic acid SAMP, a glycoimmune checkpoint ligand, that permits the disease to transition between the different polarization states [
59,
140]. Over time, overactivated complement and cumulative oxidative damage erodes the complex sialylated glycan structures of the cellular glycome [
141]. Loss of sialic acid glyco-immune checkpoint restraint on microglial cells permits M1 activation of the microglia and active recruitment of peripheral blood macrophages. FUT5-dependent production of Lewis antigens drives diapedesis of peripheral blood monocytes by binding e-selectin on the inflamed vascular endothelium [
132] (
Figure 3(5,7)). The recruitment of predominantly M1 polarized macrophages to the retina in the absence of appropriate sialic acid SAMPs results in unchecked inflammation and clinical progression of macular degeneration.
A potential therapy would be to agonize the Siglec and CFH checkpoint receptors with a sialic acid SAMP mimetic to attenuate complement amplification and repolarize activated microglia/macrophages to their non inflammatory, healing, and homeostatic form. This therapeutic strategy would normalize checkpoint control of the immune cell, in contrast to other anti-inflammatory strategies that deplete cytokines to inhibit inflammatory pathways [
59,
140].