Abstract
Age-related macular degeneration (AMD) is a major cause of visual impairment that affects the central retina. Genome wide association studies and candidate gene screens have identified members of the complement pathway as contributing to the risk of AMD. In this review, we discuss the complement system, its importance in retinal development and normal physiology, how its dysregulation may contribute to disease, and how it might be targeted to prevent damage to the aging choriocapillaris in AMD.
Introduction
Over the last decade, a new appreciation for the physiology and pathology of the complement system in the human retina has emerged, motivated largely by genome wide association studies of age-related macular degeneration (AMD) that have strongly implicated this pathway in disease risk. In this review, we will briefly discuss the complement system, evidence for its role in normal retinal physiology, and how its dysregulation due to genetic polymorphisms affect risk of macular disease. Finally, we discuss how this knowledge might be used to treat patients with macular degeneration.
The Complement System
The vertebrate complement system is a component of the innate immune system made up of a battery of dozens of activators and inhibitors, many of which are circulating serine proteases that become activated through a sequential cascade of proteolytic and conformational changes.
The complement system is evolutionarily very old, long predating immunoglobulins, with orthologs of the terminal, lectin and alternative pathway members present in proto-vertebrate chordates (e.g. Amphioxus) and echinoderms. In these older phyla, it is induced by bacterial and viral (1) products and activates non-self-recognizing, phagocytic immune cells (2). The use of a system that is always poised to recognize foreign particles is an advantage across phylogeny, and the complement system appears to be among the earliest and most widespread mechanisms to provide this defense against microorganisms.
Whereas invertebrates and proto-vertebrates possess the basics of a complement system, mammals including mice and humans have a complex complement system with over 50 circulating and cell surface proteins. The principal functions of the complement system include (1) removal of immune (antigen-antibody) complexes, (2) labeling (opsonization) of foreign antigens for enhanced removal by professional phagocytes, (3) recruitment and activation of nearby leukocytes, and (4) direct cytolysis of invading microorganisms.
In mammals, complement can be activated by several distinct mechanisms that (if unchecked) converge on the terminal pathway and result in generation of the membrane attack complex (Fig. 1). The first described, but phylogenetically most recent pathway, is the antibody dependent classical pathway. One molecule of IgM or multiple molecules of IgG (3) are recognized by a large, hexameric protein complex C1q, resulting in stepwise proteolytic cleavage of C1r, C1s, C4 and C2. Activated C2 and C4 form a C3 convertase, promoting downstream events in the pathway. The lectin pathway is similar to the classical pathway, converging at the activation of C2 and C4, but is instead initiated by mannose binding lectin (a structurally, genetically and functionally similar paralog of C1q) (4), followed by activation of MASP1 and MASP2 (paralogs of C1r and C1s), and converging with the classical pathway at the cleavage of C4 and C2. The alternative pathway relies on a constitutive, low level spontaneous dissociation of C3 in the absence of a distinct stimulus; its association with factor B; and the typical quenching of this complement activating species on inhibiting surfaces (i.e. host cell surfaces and extracellular matrix) by the action of CFH and other inhibitory proteins. When deposited on a non-inhibiting surface, activated C3 and B dimers are allowed to undergo amplification and propagation of the terminal pathway (5). In the terminal complement pathway, C5 is cleaved by a C5 convertase and the subsequent steps of activation are shared, regardless of the mode of initiation. Cleavage of C5 and its assembly with C6, C7, C8 and multiple membrane spanning C9 molecules results in the formation of the MAC pore, also referred to as the C5b-9 complex or the terminal complement complex.
Simplified summary of complement activation pathways. Genes with described polymorphisms associated with AMD are indicated by asterisks. While there are numerous fluid phase and cell surface complement inhibitors, deposition of the membrane attack complex (MAC) on the choriocapillaris is a consequence of normal aging but is much more striking in AMD and in patients with the high risk CFH genotype.
Other triggers for initiation do not strictly follow these patterns. For example, C1q can bind directly to carbohydrate moieties present on pathogen surfaces (6), advanced glycation endproducts (7,8), and apoptotic blebs (9), all independent of immunoglobulins. C-reactive protein can also activate complement along the classical pathway, although its effects on the terminal pathway are complex (10).
The presence of abundant, omnipresent circulating proteins which are constantly poised to be triggered through multiple mechanisms to activate and serve as powerful proinflammatory and lytic effectors requires a battery of soluble and cell surface defense mechanisms to prevent initiation and escalation of injury. These include proteins that are present on the cell surface, that circulate, and/or that associate with the extracellular matrix or cellular glycocalyx (11,12).
Role of Complement in Retinal Development and Physiology
While predominantly viewed as a system for host defense, it is perhaps not surprising that a meticulously engineered series of molecules with specific recognition, catalysis, and signaling properties has been co-opted in other areas of physiology.
Most of what is known about the role of complement genes and proteins in retinal development and normal physiology is derived from mouse models. It is clear that, for many primary diseases of the retina, mice and humans are genetically and anatomically sufficiently similar that the former can be used to model human retinal degeneration (e.g. in connecting cilium proteins and cytosolic enzymes (13,14). For some diseases affecting the retina, mice may be spared the effects of mutations that are pathogenic in humans. For example, human photoreceptor cells possess anatomical features (termed calyceal processes) harboring proteins that cause Usher syndrome when mutated (15). Mice deficient for Usher syndrome genes do have profound hearing loss, but are largely spared retinal degeneration. Most notably for macular disease, while mice do exhibit some central enrichment of retinal ganglion cells (16), they lack a well-formed macula lutea or a cone enriched fovea centralis.
With respect to the complement system, there is good evidence for conservation of most members between human and mouse (17) (although with some molecular distinctions (18)), and is likely representative of the situation in human. In this context, it is interesting that developing mouse retina requires C1q and C3 for proper synapse pruning in the inner neural retina as well as other parts of the CNS (reviewed in (19)). Recognizing illegitimate axons and synapses for removal by microglia is therefore subserved by the same mechanisms as removal of cell debris and pathogens. Rod photoreceptor outer segments of mice lacking the complement factor H gene show profound ultrastructural abnormalities, with the normally smooth, cylindrical profiles gnarled and bent, as well as a reduction in the rod electrophysiological response to light (20). While we are not aware that these phenotypes have been directly studied in humans with C1q, C3 or CFH deficiencies, lack of comment on these features may be largely due to the necessary medical focus on other severe, life threatening complications of these immune disorders (21), making evaluation of subtle retinal synatptogenesis phenotypes less likely to be prioritized.
Evidence from mouse models shows a role for the complement system in the development and maintenance of the mammalian retina. When the complement system becomes dysregulated, there is evidence for a significant role in age-related eyes diseases, discussed in the next section.
Complement in Eye Disease
The link between complement and AMD is now well established and we herein present a very brief summary of the relevant literature supporting this. Complement dysregulation has also been described in various other ophthalmic diseases including uveitis (22–24), diabetic retinopathy (1,25), retinal detachment (2,26) and glaucoma (3,27); these exciting associations are outside the scope of this short review.
Evidence for the Role of Complement in AMD
AMD is a degenerative condition affecting the macula in patients aged over 55 years. It is characterized by the appearance of clinically-visible, extracellular deposits (termed “drusen”) and, in some cases, subsequent progressive visual loss from either retinal atrophy or the development of choroidal neovascularization (CNV). The earliest evidence to implicate the complement system in AMD pathogenesis came from immunohistochemical and proteomic reports that drusen contain various complement proteins (4,28–30). Subsequently, a common CFH polymorphism (Tyr402His) was identified by genome wide association studies to be a major risk factor for disease (5,31–34). The observations that Y402H is (a) a coding variant with plausible—and, later, demonstrable—effects on its function; (b) had a large effect size (originally described as >4x risk in heterozygous and >6x risk in homozygous individuals) and (c) was present in a high enough fraction of the population that population based risk could be assessed; all contribute to the believability of CFH (and by extension the complement system) as a major contributor to AMD. AMD is a multifactorial, complex disease, with known environmental influences, and it was therefore surprising for one locus to have such a major impact on disease risk. Since that time, a number of complement-related genetic variants have been associated with AMD, including C2/CFB, C3, C7, C9, CFI and SERPING1 ((6,35,36)). Some of these appear to be associated with specific AMD phenotypes (7,8,37–40)) and certain rare, highly penetrant genetic variants in complement genes appear to be particularly important in familial AMD (9,10,41–43). Furthermore, the CFH Y402H polymorphism may predict response to treatment with AREDS vitamin supplements, zinc and anti-VEGF therapy, although published results are conflicting (11,12,44). In animal studies, phenotypes purported to resemble AMD have been described in mice administered C3 (13–45) as well as mice deficient in the regulatory complement component CD46 (15,46), although like other rodent models of macular disease (16,47) their similarity to human AMD is limited.
Several studies have shown altered systemic levels of complement in association with either AMD or AMD-associated genetic variants, albeit with some inconsistent findings (17,48). Systemic complement may conceivably contribute to retinal pathology, however locally produced complement seems more important. In a study of liver transplant patients Khandhadia et al found AMD risk to be associated with recipient rather than donor CFH Y402H genotype, even though the systemic circulating CFH protein allotype matched donor liver genotype (18,49).
Mechanistic Impact of Complement Activation in AMD
Despite these associations between complement and AMD, the mechanisms by which complement causes pathology in this condition remain to be fully elucidated. Chronic low-grade intraocular complement activation in patients carrying genetic risk variants and exposed to environmental triggers may lead to progressive retinal damage (19,50). Interestingly Cfh-/- mice show less retinal inflammation and age-related photoreceptor loss when kept in a pathogen-free environment, highlighting the importance of genetic/environmental interaction (20,51). Potential triggers that have been shown to activate complement include cigarette smoke (26654980, (21,52), oxidative stress (22–24,53) and amyloid beta (54–56).
While the functional relevance of most AMD-associated genetic variants is still largely unknown, considerable progress has been made regarding the non-synonymous CFH Y402H polymorphism (57). The minor allele encodes a histidine at residue 402 of the CFH molecule, which consequently has altered affinity for CRP (58), malondialdehyde (59), zinc (60) and glycosaminoglycans in Bruch’s membrane (61,62). All of these changes may reduce the ability of CFH to regulate complement activity as well as increasing oxidative stress. Indeed, there is increased MAC deposition at the choriocapillaris in eyes homozygous for the Y402H polymorphism (63).
MAC formation results in C5a release, with one C5a anaphylatoxin molecule released for each MAC complex. C5a itself has potent biological effects, that include inducing VEGF expression by the RPE (64), ICAM-1 upregulation on the choriocapillaris (65), as well as recruitment and activation of leucocytes (66). In vitro, direct exposure to MAC causes death of choroidal endothelial cells in addition to upregulated production of pro-angiogenic factors, potentially participating in the formation of choroidal neovascular membranes (67). Increased MAC formation over many years may therefore lead to loss of the choriocapillaris, which then fails to adequately remove debris allowing drusen formation in early AMD. In support of this hypothesis, histologic measurements of eyes homozygous for the CFH Y402H allele showed choroids that are almost 24% thinner than those with a low-risk genotype (68). Furthermore, MAC deposition selectively accumulates in the aging human choriocapillaris but not the capillary beds of other organs, providing a possible explanation for the local vascular loss observed in AMD (69–72). The biological basis for this tissue specific susceptibility is not yet understood.
Complement-Based Experimental Therapies for AMD
Current treatment for AMD consists of AREDS vitamins for dry AMD, which according to recent work may slow disease progression by modulating complement activity (73) or suppressing complement-mediated endothelial cell activation (74), and intravitreal anti-VEGF antibodies for neovascular AMD. Directly targeting the complement system promises to provide exciting new therapeutic options with the potential to slow AMD progression before the development of RPE atrophy or CNV (75). In murine models, protection against laser-induced CNV was initially described by Bora et al., who showed that genetic and toxicologic suppression of complement limits the severity of the neovascular complex (76,77). Moreover, it has been reported following administration of agents inhibiting complement activation (78,79), as well as the soluble complement regulators CFH (80) and CD59 (81). These studies demonstrate that, at least in a wound healing model of CNV, complement activation contributes to the severity of the lesion.
Since the strongest evidence in AMD comes from the alternative complement pathway this has been the area of greatest focus and interest. Complement inhibitors investigated in human trials include POT-4 (Clinicaltrials.gov number NCT00473928), ARC1905 (Clinicaltrials.gov number NCT00950638) Eculizumab (Clinicaltrials.gov number NCT00935883), and Lampalizumab. Results from systemic administration of Eculizumab, a C5 inhibitor, have been disappointing. This drug did not improve retinal drusen in patients with C3 glomerulonephropathy (82) or decrease the rate of GA growth in the COMPLETE trial (83). This may be in part due to the size of eculizumab (148kDa), which may have limited access to the major site of complement deposition in AMD, the abluminal choriocapillaris (Fig. 2). While the choriocapillaris is a relatively permeable vascular bed, it nevertheless is normally impermeable to molecules much greater than 40kDa (84). This suggests that smaller inhibitors and/or alternative delivery methods will be essential to protect macular cells form MAC injury. Results from a phase 2 trial (MAHALO) have been presented at ophthalmology meetings and suggest that lampalizumab, an antigen-binding fragment (Fab) of a humanized, monoclonal antibody directed against complement factor D, may result in reduction of progression in some eyes with GA, and that this effect may be more pronounced based on CFI genotype (Holz et al., Euretina meeting 2014, http://www.euretina.org/london2014/programme/free-papers-details.asp?id=3439&day=0). While these findings are encouraging, at the time of the preparation of this manuscript the full results have not yet been published in the peer reviewed literature.
Age-related macular degeneration. Top panel, fundus appearance of a 65-year-old patient with AMD. Note the presence of drusen and pigment abnormalities in the macula. Lower panel, localization of the membrane attack complex (green) to domains surrounding the human choriocapillaris in an eye with AMD. Choroidal endothelial cells are labeled with UEA-I lectin (red) and nuclei appear blue (DAPI). The retinal pigment epithelium autofluoresces yellow in the preparation.
In summary, the complement system is an ancient, critical facet of the mammalian immune system, with additional roles in development normal physiology. Like many other consequences of aging, and especially in individuals with a genetic predisposition to weak regulation, its over activity can contribute to macular degeneration. A better understanding of this system and how it participates in disease is coming to light.
Acknowledgements
The authors wish to thank Mr. Miles Flamme-Wiese for technical support, and the Iowa Lions Eye Bank and the tissue donors and their families who contribute so selflessly to the advancement of science and medicine.
Conflict of Interest statement. RFM-none; ANW-none; EHS-none; AJL- consultant for Gyroscope therapeutics and clinical investigator on Roche's lampalizumab trials.
Funding
NIH grants EY-024605(RFM), EY-026547(EHS), Complement UK (AJL), Rosetrees Trust (AJL), and the Elmer and Sylvia Sramek Charitable Foundation.
References
Author notes
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors.
