A Perspective of AMD Through the Eyes of Immunology

Abstract

Despite strong genetic associations, compelling human histological data and numerous hypotheses generated with supportive animal data, the mechanisms of inflammation or inflammatory control of cell health during progression of age-related macular degeneration arguably remain elusive. This perspective delivers a view that maintaining tissue health requires active immune cellular and tissue pathways, but when responses are perturbed or exaggerated, chronic inflammation is destructive. There are potential pathways and processes to enable understanding and determine how potential causative factors including altered cellular metabolism, senescence, oxidative stress disrupt tissue homeostasis are engaged. Establishing differences in the immune phenotype between normal aging and AMD, and how the inter-relatedness of these triggers contribute to pathobiology is integral for future therapeutic success.

The relentless and increasing burden of AMD is confounded, in common with other chronic, insidious neuro-inflammatory disorders by many factors. Genetic–environmental interactions, the aging cell, tissue, and the immune system are implicated, but causality has yet to be established.^1,2 Unquestionably, the anti-VEGF renaissance has revolutionized treatments for neovascular AMD and prevents (at least in the short term) acute visual loss for millions of individuals.³ However, we are still treating late in the disease course; therefore, unsurprisingly, atrophy ensues over time. Our focus needs to address and develop new approaches that halt the progressive loss of RPE and photoreceptors that characterize macular degeneration and understanding immune responses in AMD is a tractable target for such therapy. 

Although clinical and genetic data support the association of chronic low-grade inflammation in the pathogenesis of AMD, the underlying mechanisms responsible are not fully appreciated, and we still have no current inroads to therapeutically redress the ocular inflammatory environment and prevent progression. Genetic studies strongly imply the central role of immune responses and particularly innate immunity in pathogenesis and disease severity, not least the compelling association with complement regulation.⁴ Chronic accumulation of immune cells (infiltrating monocytes and tissue resident microglia) at either the subretinal or choroid RPE interface are associated with early, intermediate, and advanced forms of AMD and likely contribute to disease progression.^5–10 Similarly, increased numbers of activated choroidal macrophages and mast cells (MCs) are observed around atrophic lesions of human donor eyes, implicating complement-mediated MC degranulation.^11,12 Notwithstanding the idiosyncrasies of animal models not truly reflecting the human state, functional studies do indicate that the recruitment and accumulation of activated immune cells are critical for both neovascular and degenerative responses that characterize AMD.^5,13–17 Nonresolving, persistent low-grade inflammation is a major drive of many chronic, age-related diseases, including metabolic disorders, neurodegenerative diseases, and cancer. What unifies these conditions? The persistence of myeloid-derived cells that may amplify inflammation and tissue damage.¹⁸ 

Complicating our understanding is the lack of a clear sequence for the events and triggers driving initiation and progression of AMD from the very early stage. Drusen contain immunologically active deposits such as lipids, complement, and other immune activating components. However, is a druse the potential trigger of the initial inflammatory responses or do drusen simply accentuate inflammation? Additional inflammatory triggers include oxidative stress and secondary mediators (e.g., reactive oxidative species [ROS], cytokines, complement effectors, and C-reactive protein [CRP]), which will also dysregulate active pathways controlling immune homeostasis. Constitutive retinal immune regulation serves to preserve tissue and cellular function and is delivered via canonical hemopoietic (or yolk sac) derived immune cells (microglia), as well as nontraditional immune cells (Müller Glia, endothelium, RPE). Despite constant lifetime stress, the retina in many individuals continues to perform adequately late in life, which suggests to some degree that either inflammatory responses are essential and beneficial or intrinsic regulatory mechanisms are not perturbed. Thus, an alternative hypothesis may consider that heightened parainflammatory responses (both cellular and physiologic) required for sterile inflammation are also likely to act as metabolic sensors to regulate and maintain retinal health and homeostasis. Finally, an untapped area of research questioning is understanding how the retina/choroid responds to systemic challenges and determining the roles infection (or noninfective systemic immune activation) and immuno-senescence play; both can alter immune-surveillance or directly stimulate immune-competent cells in the retina to provoke or progress degenerative disorders.^19,20 

The Status Quo

Maintaining tissue health requires active protective inflammatory responses. However, when responses are heightened, or dysregulated, chronic inflammation is destructive.²¹ Inflammation is an adaptive response to tissue stress, regulated at the level of the tissue, but also by local (tissue-resident nonimmune and immune cells) and systemic immune pathways. When tissues are exposed to noxious stresses (environmental factors), cell-autonomous responses (e.g., inflammasome activation or autophagy) are initiated to repair the damage and promote survival to restore homeostasis. The tissue response to stress also elicits inflammatory signals and changes in the tissue microenvironment, which, monitored by the local immune network (tissue resident macrophages and the complement system), trigger transient inflammatory responses. Termed, para-inflammation, these responses are considered an important immunologic mechanism that serves to maintain tissue homeostasis and monitor for tissue malfunction.²² Varying magnitudes of tissue stress or malfunction are exhibited as a spectrum and accordingly parainflammatory responses are matched: they range from low level or basal housekeeping functions through to unchecked proinflammatory responses. Failure to restore the optimal homeostatic threshold results in excessive parainflammatory responses that manifest into chronic immune-mediated degenerative conditions.^23,24 

In the retina, immune activation and initial recruitment of monocyte-derived macrophages (in addition to microglia) are required to help process photoreceptor and RPE byproducts, thus controlling overt inflammation. Infiltrated monocytes and macrophages respond to the triggers and cues that the environment exposes them to. Hence, the current and useful paradigm of monocytes²⁵ and macrophage subsets²⁶ is worthy but does not fully assist our understanding of cell functionality or purpose, nor the kinetics or dynamic nature of cell influx, residency, or efflux within the tissue. The continued understanding through animal models is therefore useful to at least glean knowledge of fate, function, and turnover^27–29 (discussed further in The “Sick” Activated Microglia section). For example, oxidized lipoproteins and free radicals are major causes of tissue stress, serving as local triggers for para-inflammation and exhibited by the activation and increased motility of resident microglia. 

The sick activated microglia is a notion documented in dementia,³⁰ and potentially contributes to breakdown of the blood retinal barrier and promotion of inflammation. One example of para-inflammation in the eye is complement activation and accumulation of both microglia and infiltrated macrophages (mononuclear phagocytes) to the subretinal space, mast cell activation, and morphologic changes to the melanocytes in the choroid, as well as fibrosis. To prevent rapid progression and curtail inflammation, there are several levels of immune checkpoints, including innate immune regulators. As stated earlier, if drusen drive immune responses (as they contain immune activating components), they will target the coaligned RPE, microglia, and macrophages, driving consequent inflammatory cytokine and angiogenic mediators.^31–34 To date, the most tractable therapy given the genetic variant association is against complement.³⁵ 

Innate Immunity, Complement, and Gene Association

The most widely studied genetic risk factor for AMD is complement factor H (CFH).^36–39 CFH regulates complement activation in plasma, host cells, and tissue, in particular at sites of tissue inflammation supporting myeloid cell adhesion and migration facilitate phagocytosis of microbes and cellular debris, following injury, or during degeneration.⁴⁰ The genetic variant substitutes histidine for tyrosine (Y402H) in the cellular binding domain of CFH. The CFH(H402) variant does not regulate the alternative pathway (AP) of complement activation as efficiently and is associated with early and advanced AMD, suggesting CFH(H402) may drive disease onset.⁴ How does this relate to dysregulating immune homeostasis? 

Recent data offer new insight into the role of CFH in AMD that is as surprising as it is compelling, revealing a noncanonical role of CFH in AMD distinct from the complement pathway.⁴¹ Cx3cr1-deficient animals and TRE2 mice (a humanized transgenic mouse expressing the AMD-risk APOE2 isoform) model the age-dependent subretinal accumulation of both monocytes and resident microglia and the associated photoreceptor degeneration observed in human AMD.^5,16,42 Contrary to the expectation that perturbed complement regulation would augment disease, CFH genetic deficiency in these models instead inhibits the progressive cellular accumulation, supporting the similar age- and CFH-dependent increase reported in the choroid of CFH heterozygous vs homozygous mice.⁴³ Through iterative experiments, this elegant study demonstrates that CFH binding to CD11b perturbs the mechanism of CD47-mediated elimination of mononuclear phagocytes (MPs) required to maintain homeostasis in the subretinal space. Normally, during resolution of inflammation, MPs are eliminated through CD11b–CD47 interaction and signaling with its cognate ligand thrombospondin-1 (TSP-1), which is secreted by RPE cells. This finding may also explain previous reports describing increased and prolonged subretinal inflammation observed in TSP-1–deficient mice.^44–46 The clinical relevance of this mechanism was highlighted where the risk variant of AMD CFH(Y402H) binds more efficiently than the nonrisk isoform to CD47-CD11b, thereby inhibiting the immunosuppressive effects of TSP-1 (i.e., elimination of MPs from the subretinal space). These findings raise important questions and new avenues of investigation, as well as insights into disease etiology. Are functional polymorphisms related to CD11b or CD47 conferring similar risks for AMD? Does the association of the high-risk variant with the CD11b–CD47 complex share similarities in its complement-inhibiting activity? The work highlights that development of therapeutic agents directed toward cellular clearance strategies may be of benefit in preventing progression. 

CRP is the prototypical acute-phase reactant and an active regulator of the innate immune system. It is a serum biomarker for chronic inflammatory conditions, heart disease, and, more recently and arguably, AMD.⁴⁷ CRP has been identified in drusen⁴⁸ and the choroid. Among the multiple functions ascribed to CRP are activation of the classical complement pathway and inactivation of the AP.⁴⁹ However, CRP exists in at least two conformational and functionally distinct forms: the native pentameric CRP (pCRP) and a modified/monomeric CRP (mCRP).^50,51 The mCRP isoform induces proinflammatory responses in endothelial and innate immune cells,^52,53 is localized within retinal tissue, and exhibits binding affinity for CFH.⁵⁴ In relation to AMD CFH isoforms, binding affinity to mCRP of the non–risk-associated variant CFHY402 is stronger than the H402 variant,⁵⁵ and individuals who are homozygous for the latter show 2.5-fold higher CRP levels in the RPE–choroid layer compared with those homozygous for the nonrisk variant.⁵⁶ 

Mechanistic insights into how CRP contributes to the development of AMD demonstrate that mCRP upregulates IL-8 and CCL2 levels in RPE,⁵⁷ and CFH binding to mCRP suppresses this proinflammatory activity. CFH from patients carrying the “risk” CFHY402H polymorphism displays impaired binding to mCRP and is unable to restrain these proinflammatory effects of CRP. Understanding the structural basis of the CFH–mCRP complex will give further insight into resolution of inflammation and novel approaches to augment these effects in patients. An additional hypothesis for our consideration is that mCRP plays a role in the inflammatory response and tissue damage at the level of RPE. Data showing impairment of RPE functionality (increased paracellular permeability through the disruption of tight junction proteins) suggest that mCRP could induce RPE barrier breakdown and promote chronic inflammation.⁵⁸ 

Other genetic variants associated with AMD include CFI, CFB, and C3.^59–61 The complement hypothesis for the patho-etiology of AMD describes the notion that certain individuals carry a “complement hyperinflammatory phenotype,” resulting in uncontrolled AP activation in response to cellular damage and debris in the retina.^62,63 The polymorphism rs10490924 identified in the ARMS2 gene loci is highly associated with AMD and induces mRNA instability.⁶⁴ A recent study illuminates a role of ARMS2 as a complement activator, binding directly to apoptotic and necrotic cells via properdin complex and augmenting C3b surface opsonization for phagocytosis. Notably, ARMS2 protein appears absent in both blood monocytes and inducible pluripotent stem cells (iPS)--microglia, derived from patients homozygous for the ARMS2 AMD risk variant. One conclusion drawn is that ARMS2 is likely involved in complement-mediated clearance of cellular debris, and ARMS2 protein deficiency may therefore drive drusen formation.⁶⁵ 

The complement cascade terminates in the cell surface assembly of the membrane attack complex (MAC), promoting inflammation by causing aberrant signal transduction and cell lysis. Age-related accumulation of MAC is observed in the choroid from healthy individuals, with levels significantly elevated in individuals with the high-risk CFH(H402) variant.^66,67 Thus, alterations in complement regulatory mechanisms leading to increased MAC may accelerate choroidal atrophy in early stages of disease, representing a potential trigger for subsequent inflammatory responses and AMD progression.^68,69 As the critical interface between the choroid and outer retina, the basal aspect of RPE is also a recognized site for age-related increases in MAC deposition, with higher levels reported in donor tissues from patients with risk-associated AMD genotypes.^66,67,70 In vitro, MAC assembly on the surface of RPE elevates levels of IL-6, IL-8, and MCP-1 creating a proinflammatory environment.⁷¹ To maintain cellular integrity, intrinsic mechanisms enable the RPE to regulate formation and elimination of MAC through recycling of CD59 (MAC inhibitor), lysosome exocytosis, and endocytic pathways for MAC clearance.^72,73 Disruption of these critical homeostatic mechanisms compromises the protective responses, leading to mitochondrial damage and oxidative stress following sublytic MAC attack. Despite direct evidence this occurs in AMD, it seems plausible that oxidative stress may establish a self-propagating cycle (mitochondrial damage, inflammation, and dysfunction in the RPE). Thus, efforts to identify focused strategies to boost these innate protective mechanisms may help preserve RPE health and function.⁷³ 

The emerging relationship of complement and its contribution to cellular metabolism machinery has generated a conceptual framework referred to as the Complosome. Complement activation regulates key metabolic pathways and thus can impact fundamental cellular processes, such as survival, proliferation, and autophagy. Newly identified functions of intracellular complement include a key role in shaping metabolic reprogramming, which underlies T-cell effector differentiation and a role as a nexus for interactions with other effector systems, including the inflammasome and the Notch transcription-factor network.⁷⁴ Therefore, if such mechanisms translate to RPE or macrophages/microglia, this emphasizes the essential role of complement activation to maintain tissue health. 

Loss of Regulatory Control: Sick Cells and Altered Immune Responses

The “Sick” Activated Microglia

Yolk sac–derived microglia regulate tissue immune responses and are central players in many neuro-inflammatory disorders of the brain including multiple sclerosis and Alzheimer's disease (AD).^75,76 Recognized as a long-lived and self-renewing population, tissue resident microglia are distinct both in terms of origin and phenotype to that of infiltrating monocyte-derived macrophages that are recruited to the inflamed CNS from the circulation.^77–80 In the healthy CNS, microglial immune activity is restrained by dedicated immune inhibitory pathways that suppress undesirable inflammatory responses and tissue destruction that are often associated with immune activation.⁸¹ These homeostatic checkpoint mechanisms include direct interaction of microglia with neurons through receptor–ligand pairs (CD200-CD200R and CX3CL1-CX3CR1), soluble mediators such as TGF-β, or intracellular regulators.⁸² Similarly, the coordinated and divergent regulation of triggering receptor expressed on myeloid cells (TREM) signaling pathways (TREM1 versus TREM2) influence microglial inflammatory and homeostatic activity.⁸³ The complex functional roles of microglia are both beneficial and detrimental to disease pathogenesis, including phagocytosing/degrading toxic proteins (i.e., amyloid plaques in the brain) and promoting neurotoxicity through excessive inflammatory cytokine release. Alterations in the normal homeostatic functions such as surveillance, synaptic pruning, and plasticity may also contribute to excessive synapse loss and cognitive dysfunction in AD and other neurodegenerative diseases.³⁰ 

The contribution of systemic immunity, recruitment of monocytes, and tissue-resident microglia to AD onset and disease progression remains controversial. To expand on this further, an understanding of fate and function gives inroads to dissecting mechanisms. In addition to classification of monocyte and macrophage subsets, fate-mapping techniques now allow us to reliably determine function and distinguish in animal models between resident microglia versus infiltrating cells (whether newly recruited from bone marrow or replacement from choroidal populations).^16,28 One concept suggests that chronic proinflammatory responses of microglia are associated with disease escalation.^84,85 Microglial priming (responding to environmental or systemic cues) and the subsequent exaggerated response of these cells to a secondary systemic trigger highlight treatment approaches for neurodegenerative diseases that target systemic disease or block signaling pathways that mediate the CNS response to systemic inflammation.^20,86 The crucial role of the host microbiome in shaping and controlling the innate immune function of microglia must also be acknowledged. Altering the composition or eradication of gut microbes results in global defects in maturation, differentiation, and functional responses of microglia, again reminding us of the important inter-relationship between the periphery and CNS in both health and disease.^87,88 This may also apply to AMD, supported by emerging data indicating that the modified intestinal microbiome composition of patients may be associated with development of neovascular disease.⁸⁹ Collectively, these concepts could equally apply to retinal microglia and explain an early event and trigger toward AMD. 

However, there remains contention as to whether microglial function in neurodegenerative diseases is beneficial but insufficient or whether these cells are effective at early stages of disease but lose their efficacy? One aspect of potentially translating this to the retina and AMD is to consider current understanding in dementia. Extrapolating from other diseases reminds us that we must acknowledge the structural differences between brain and ocular tissue and that AMD occurs at the interface between a highly vascularized connective tissue, a layer of neuroepithelium, and a multilayered neural tissue. Nevertheless, increasing recognition of photoreceptor damage and reduced function is noted in AMD with loss of dark adaption.⁹⁰ Harnessing multi-omic technology and engaging in single cell RNA-seq analysis have identified a subpopulation of disease-associated microglia (DAM), which are found in the brains of AD (in both human and mouse), amyotrophic lateral sclerosis (ALS), and aged mice.⁹¹ These cells are spatially localized in proximity to amyloid plaques and exhibit markers and activated pathways (previously associated with AD risk factors) but were not specifically attributed to microglia. To become DAM, microglia undergo an initial TREM2-independent step of activation, followed by a second step that is TREM2 dependent. Consequently, microglia lose their homeostatic gene signature and end up by upregulating genes involved in lipid metabolism and phagocytosis, eliciting a protective phenotype. In another study, the TREM2-mediated APOE signaling pathway mediates a switch from homeostatic to DAM phenotype, and targeting this pathway restored the resting signature of microglia in ALS and AD mouse models.⁹² There is a tradeoff in homeostasis between the number of DAM cells with phagocytic activity and checkpoint mechanisms that keep them under tight control (e.g., CX3CR1 inhibitory signaling). Thus, checkpoint mechanisms, essential on one hand for function of microglia to ensure risk-free immune activation, may become negative factors when the threshold is altered, and a strong phagocytic activity is needed in aging or under detrimental genetic backgrounds. In this scenario, DAMs are not considered the primary disease effectors, but mediate the removal and clearance of the misfolded and aggregated proteins (drusen in AMD) that accumulate in neurodegenerative diseases and general aging induced damage.⁹³ Perhaps and counterintuitively, this may suggest that blocking microglia-specific checkpoints may provide a therapeutic approach to trigger the ability of resident microglia to combat neuro-inflammation. 

The dual functional roles of immune surveillance and tissue maintenance provided by microglia are equally important to the ocular environment. They are critical as regulators of immune health within the retina, acting as phagocytic sentinels to detect initial danger signals (DAMPs), and then rapidly activate to process both pathogens and dying cells. Equally, microglia secrete neurotrophic factors, which impact the physiology and survival of photoreceptors, contributing to the maintenance and integrity of the neuronal network and tissue function.⁹⁴ There are checkpoint regulators through ligand–receptor interactions that mediate tonic inhibitory signals maintaining the resting phenotype of resident microglia and regulate myeloid cell inflammatory and angiogenic responses. These include CD200 (expressed on neurons) and myeloid-expressing inhibitory CD200R⁹⁵ to maintain normal homeostatic control and limit immune-mediated damage and can be exploited therapeutically.⁹⁶ For example, RPE destruction in the model of laser-induced choroidal neovascular membranes (CNV) polarizes infiltrating myeloid cells toward a proangiogenic phenotype. Augmentation of the inhibitory CD200R signaling pathway or administration of Th2 cytokines could suppress macrophage activation or drive antiangiogenic function respectively.^32,97 Similarly, the CX3CR1 signaling pathway modulates cell activation and migration. With age, deficiency in the CX3CR1 receptor was considered linked to increased microglial activation, subretinal migration, and retinal degeneration,⁵ supporting GWAS data that Cx3cr1 polymorphisms are a potential risk factor for AMD. However, conflicting reports raised questions as to whether the accumulation of dysfunctional subretinal macrophages was directly attributable to retinal degeneration.^98–101 Determining genomic background differences between C57BL/6 substrains in these studies revealed the Crb1 mutation (rd8), now firmly recognized to impact and enhance degenerative phenotypes in ocular inflammatory models.^101,102 Thus, subsequent studies using rd8-ve transgenic strains demonstrate that deficiency of CX3CR1 (but also CCL2-CCR2) signaling pathways differentially affects the trafficking of microglia and macrophages with age in the retina but do not cause retinal degeneration per se.¹⁰³ However, Cx3CR1 signaling does play an important role in controlling retinal inflammation, as Cx3cr1^−/−mice (rd8-) are susceptible to oxidative stress-induced retinal inflammation and photoreceptor loss.¹⁰⁴ 

An Imbalance of Physiologic Regulation

Inflammasome Activation

The NLRP3 inflammasome complex is recognized as a sensor that monitors cellular stress through pattern recognition receptors (e.g., Toll-like receptors [TLRs]), activating inflammatory caspases.¹⁰⁵ Activation of NLRP3 (through either foreign or endogenous danger signals) drives caspase-1–mediated liberation of two proinflammatory cytokines, IL-1β and IL-18, and ultimately pyroptosis or apoptosis.¹⁰⁶ The NLRP3 response is almost certainly considered a protective response initially, providing a rapid response to danger to preserve tissue function and integrity. However, the corollary is that sustained inflammasome activation will also damage tissues, as seen in the pathogenesis of autoinflammatory disorders and implicated in AD, cancer, diabetes, and AMD.^33,107,108 

Understanding the regulatory mechanisms controlling inflammasome activation, and specifically how signals from multiple stimuli (cellular damage and stress) are integrated and processed, is essential to progress. To date, increasing evidence has highlighted how different triggers implicit for NLRP3 activation generate effector molecules that are implicated in AMD pathogenesis. Whole drusen extracts, as well as the complement component C1q, isolated from donor AMD eyes activates NLRP3 lead to the secretion of IL-1β and IL-18.³³ Moreover, carboxyethylpyrrole (CEP) proteins that accumulate in the retina in age and serve as a biomarker of AMD can also prime the macrophage inflammasome.¹⁰⁹ In other important studies, repetitive element-derived Alu RNA transcripts are recognized as an endogenous activator of the inflammasome, present in the RPE of patients with geographic atrophy (GA).¹⁰⁸ As noncanonical targets of DICER1-mediated enzymatic degradation, accumulation of these transcripts is associated with the loss of DICER1 expression (which might result from oxidative stress in the RPE) and functions as both priming and activating signals to stimulate NLRP3 signaling pathways.¹¹⁰ Human drusen extracts induce NLRP3-dependent IL-1β secretion from lipopolysaccharide-primed peripheral blood mononuclear cells. Inflammasome-mediated IL-1β release from RPE is supported by evidence of increased IL-1β mRNA in RPE donor eyes with GA,¹⁰⁸ release following lysosomal destabilization,¹¹¹ and activation of RPE with A2E (lipofuscin) components.¹¹² It has also been shown that the proinflammatory cytokine IL-17A can induce IL-1β from RPE via NLRP3 activation.¹¹³ 

Although blockade of the NLRP3 inflammasome appears an attractive concept and considered the next-generation therapeutic target for dry AMD,¹¹⁴ we must also consider that effectors perceived as deleterious may also have the capacity to exert protective effects within a tissue. The complexity of the underlying immunobiology should also remind us of the compelling but contrary evidence that altered immune responses (which increase with age) can also provide a capacity to maintain and preserve tissue function.²³ Evidence of NLRP3 activation and IL-18 upregulation in the RPE of human atrophic AMD donor eyes^108,111 supports a putative role for NLRP3 inflammasome in the development of atrophic disease. However, and despite controversy, it is conceivable that inflammasome activation may also exert protective effects through production of IL-18 to protect RPE and attenuate pathologic neovascularization.^115–117 The therapeutic adjuvant effects suggest IL-18 has capacity to modulate multiple pathways triggered from inherent inflammasome activation in AMD, and its expression is a mechanism to augment aspects of inflammatory responses in a bid to protect the tissue. 

A recent publication, both pertinent and timely to this perspective, now requires us to re-evaluate our interpretation of the role that NLRP3 plays as the driving force behind RPE dysfunction in nonexudative AMD.¹¹⁸ A robust and iterative study, it demonstrates NLRP3 is not expressed by human primary or human established RPE cell lines subjected to multiple inflammasome priming conditions or ex vivo macular RPE from AMD patients. Furthermore, it highlights the importance of validating and authenticating the specificity of commercial NLRP3 reagents. Thus, although these observations challenge current dogma, they do not fully negate the contribution that NLRP3 brings to AMD but instead suggest its influence more likely relates to the resident microglia and infiltrating immune cells.³³ Thus, further evidence is still required to characterize the presence, origin, and activation leading to secretion of the IL-1β and IL-18 effectors in AMD. Cognizant that NLRP3 is just one member of a wider NLR inflammasome family, including NLRP1, NLRC4, and AIM2 (all implicated in pathogenesis of ocular diseases), determining whether these complexes are also expressed by RPE and therefore capable of contributing to AMD pathogenesis will be important.¹¹⁹ Accepting that alternative pathways may in fact drive these effectors is further emphasized in a recent study showing the link between how NLRP1 functions as an innate immune sensor in the context of metabolic stress to produce IL-18, preventing obesity and metabolic syndrome.¹²⁰ 

Recent investigations of IL-33 in relation to neovascular ocular disease extends current understanding into the functional diversity of members of the IL-1 cytokine family. IL-33 is unique as it is active without caspase-1 cleavage and does not require inflammasome activation for secretion and bioactivity.¹²¹ Monocyte recruitment, contributing to photoreceptor loss in a mouse model of retinal degeneration,¹²² infers a pathogenic role of endogenous IL-33 and an a priori for neutralizing IL-33 to reduce myeloid cell accumulation as a possible intervention. However, in consideration of the emerging role of IL-33 in inflammatory disorders¹²³ and in the absence of progressive cell death, IL-33 also regulates tissue responses. Administration of recombinant IL-33 protects against fibrosis and CNV development.¹²⁴ IL-33 is released from activated Müller glia (without cell death), implicating inflammasome-mediated release and inactivation of IL-33.¹²⁵ As TLR-dependent upregulation of IL-33 by RPE does not influence cell viability, it is likely that this is an adaptive response to maintain homeostasis. Colocalization of IL-33 within both membrane-bound cytoplasmic vesicles and nuclear euchromatin suggests an interorganelle dynamic of IL-33 trafficking and release in the absence of cellular necrosis.¹²⁶ 

Autophagy

Despite the knowledge of risk immune response–related genotypes, these have not led to an understanding of how and when immune regulation is impaired. Whereas oxidative stress serves as a primary environmental factor that induces altered mitochondrial activity, impaired intracellular RPE processing pathways (autophagy, phagolysosome, and protein trafficking) and induction of RPE senescence^{35,104,109,127} are tractable pathways that can modulate and determine immune responses. Furthermore, oxidative stress elicits a proinflammatory response in RPE cells with increased IL-1β, IL-1R1, IL-18, and IRAK-1 gene expression. Increased levels of inflammatory cytokines are observed systemically and in the eyes of patients with AMD, and when normal RPE homeostasis is perturbed (e.g., when autophagy is impaired), the regulatory intracellular IL-1 receptor–associated kinase-M (IRAK-M) is inhibited, increasing cell susceptibility to IL-1β–mediated cytotoxicity.³¹ 

Impaired autophagy associated with age-related degenerative disorders is highlighted by studies in which pharmacologic or genetic manipulation of autophagy pathways can induce cellular and tissue degeneration in vitro and in vivo.¹²⁸ In RPE, this leads to RPE transcytosis and exocytosis and early signs of RPE degeneration.¹²⁹ Furthermore, human RPE cultured from AMD donors is functionally impaired, demonstrating increased susceptibility to oxidative stress, increased ROS production, and reduced mitochondrial activity.¹³⁰ Therefore, regulation of homeostatic mechanisms, whether at the level of the tissue (autophagy or inflammasome activation) or the local immune network (para-inflammation), involves precise and complex signaling cascades and negative feedback mechanisms,¹³¹ but may offer pathways for redressing ongoing degeneration. For example, the negative regulator, IRAK-M, expressed by immune and epithelial cells, serves as a key inhibitor for MyD88/NFκB-mediated inflammatory pathways, and low IRAK-M expression is associated with chronic inflammation, obesity, and metabolic syndrome or when autophagy is inhibited in RPE cells.^31,132–134 

Future Perspectives

By its nature, a perspective cannot address all aspects of inflammation that may equally underpin AMD development and progression. Unquestionably, the primary focus for AMD research has been innate immunity as the central driver for disease, but compelling evidence also implicates adaptive immune responses. Understanding whether adaptive pathways elicit proinflammatory or regulatory functions or represent bystander effects remains limited, in part due to the lack of functional intraocular lymphatics and tangible evidence of direct B- or T-cell involvement at sites of neovascular or geographic lesions. Notwithstanding, reports confirm the presence of T-cell subsets in human choroid,^135,136 as well as the potential for interplay between innate and adaptive immunity, including increased levels of C5a¹³⁷ and elevated levels of IL-22 and IL-17.¹³⁸ Furthermore, autoantibodies (AAbs) targeting proteins involved in autophagy, immunomodulation, and protection from oxidative stress and apoptosis are detected in AMD patient sera, including during early stage disease.¹³⁹ Tantalizingly, this offers potential biomarkers for diagnosis and prognosis of AMD, but we still do not appreciate or understand whether AAbs are causative or pathogenic or simply represent secondary products generated during AMD progression. One potential hypothesis is that AAbs present early in subsets of AMD patients are the pathogenic drivers of disease. An emerging concept is that autoantibodies may create further complement-mediated damage or activate innate cells to switch from protective parainflammatory to pathogenic responses, but this still warrants further investigation.^139–142 

Also requiring our consideration are the factors of aging, senescence, and bioenergetics, equally important and often nascent in our exploration of the disease process. The role of the Warburg effect in the pathogenesis of metabolic disorders, such as diabetes and atherosclerosis, and its contribution to inflammatory processes relevant for disease is recognized.^143,144 The Warburg effect rapidly provides ATP and enhances metabolic pathways to support the need for increased biosynthetic demands and rapid energy production. Extrapolating from observations that tumor cells undergo a bioenergetic switch (permissive for survival and proliferation) to aerobic glycolysis, we now appreciate that such a bioenergetic switch occurs in the aging and early AMD RPE. With age, there is increasing strain on mitochondrial function, autophagy, and mitophagy to maintain cellular and tissue health. A response for the good is to divert energy sources, the Warburg effect, to maintain function against the stress. Cell senescence may also drive immune-mediated degenerative disorders such as AMD. Mitochondrial dysfunction is seen in aging cells and senescent cells, and increasing evidence highlights that a decline in mitochondrial health plays a prominent role in the pathogenesis of AMD.¹⁴⁵ Cybrid models indicate that mitochondrial variants mediate not only energy production but also determine the cell's ability to switch energy source, in turn impacting signaling pathways and phenotype of immune activation.¹⁴⁶ Furthermore, decreased mitochondrial and glycolytic function in AMD donor RPE suggests a bioenergetic crisis contributes to AMD pathology.¹⁴⁷ 

In combination with big data and genetic determination of possible causality, unraveling the biological and molecular complexity underlying immune dysregulation also requires an understanding of the potential drivers of immune activation and at each inflection time point of disease progression. However, this is further compounded by how immune responses change with age. To this end, understanding the behavior of resident and infiltrating immune cells within tissues themselves is of growing importance. Appreciating that, although other immune-mediated ocular inflammatory animal models are not fully comparable (because the inciting disease mechanisms differ), we may still contrast the underlying cellular mechanisms and tissues responses. In the context of uveitis, the experimental autoimmune uveoretinitis (EAU) model¹⁴⁸ demonstrates how chronic remodeling of the tissue is associated with the persistence of inflammatory immune infiltrate.^149–153 Thus, two aspects of note that may display convergence of mechanisms to AMD: first, there is remodeling and vascular changes associated with microglial activation and macrophage infiltrate,^44,154 and second, there remains a persistent T-cell response, although assumed to be highly regulated in the chronic phases and be involved in both antibody and complement engagement.^155,156 

We now recognize that immune responses within tissues are localized and heterogenous, where only small regional subsets of immune cells adopt either inflammatory or protective phenotypes. Awareness of functional diversity (either beneficial and detrimental) exerted by microglia in tissue health and disease challenges dogma and means neurodegenerative diseases are considered the consequence of aberrations in the physiologic and homeostatic responses of these cells. Thus, targeting divergent microglial functions, thereby redressing homeostasis, will provide novel paradigms for therapies and biomarkers. Success will require us to rethink and develop new models of disease, including human iPS cells derived microglia and organoids to fully understand which findings may translate to provide microglia-targeted approaches. New molecular multi-omics techniques will enable us to further explore the concept that tissues (RPE, Müller cells, and microglia) are sick, meaning normal function and regulation are either perturbed or exaggerated. Thus, targeted therapeutic approaches through modulation of cellular bioenergetics, inflammasome activation, or autophagy pathways may serve to protect normal immunity and tissue health. Establishing what the differences in immune phenotype between normal aging and AMD are, and how the inter-relatedness of senescence and inflammation contributes to pathobiology, is integral to this approach. Altered epigenetic mechanisms and post-transcriptional control of microRNAs regulating gene expression are also known to impact the immune and tissue homeostasis in the retina.¹⁷ 

Acknowledgments

Disclosure: D.A. Copland, None; S. Theodoropoulou, None; J. Liu, None; A.D. Dick, AbbVie (C), Roche (C), Novartis (C), ARVO (S)