Introduction

Age-related macular disease (AMD) is a complex, multifactorial disease and now the most common cause for legal blindness in industrialized countries [11, 36]. Prevalence will further increase given demographic developments in ageing populations. Current prophylactic and therapeutic options are limited. A better knowledge of the pathogenesis appears mandatory to develop efficient therapies in order to prevent central visual loss from AMD. In the human eye various age-related changes occur in the outer retina, the retinal pigment epithelium (RPE), Bruch’s membrane and the choroid. Several of these changes, which are influenced by a variety of initiating, promoting or inhibiting genetically and exogenous factors, are thought to contribute to the pathogenesis of AMD. The understanding of underlying molecular mechanisms has been expanding over the recent years and is reviewed herein. Novel insights have already led to new targets for intervention.

Oxidative damage

Several experimental and clinical findings indicate that oxidative mechanisms contribute to the disease process. Because of the high oxygen demand, life-long exposure to light and the presence of polyunsaturated fatty acids the retina is particularly prone to photo-oxidative damage [7] with subsequent formation of free radicals and peroxidation of adequate substrates. These mechanisms may become more important with age as there is a reduction in the local antioxidative enzymes in the RPE and a decrease in macular pigment density, which serves as a filter for short-wavelength light and with its two constituents, lutein and zeaxanthin, as an antioxidant. With increasing cumulative oxidative stress peroxidation of biomolecules including lipids could result in the formation of undegradable higher molecular polymers which, in turn, lead to accumulation of material in the lysosomal compartment of RPE cells, so-called lipofuscin (LF) granules. Lipofuscin may sensitize the RPE to light with subsequent aggravation of oxidative damage and contribute to impairment of pigment epithelial functions. The correlation between RPE lipofuscin accumulation and distribution of drusen and the fact that race, light iris colour and age are known risk factors for AMD further support the hypothesis of oxidative mechanisms in the pathogenesis of AMD. It has been argued that enhancing of the concentration of antioxidants in the outer retina could have a protecting effect against the oxidative damage. Interestingly, recent results from the Age-Related Eye Disease Study (AREDS) indeed demonstrate a prophylactic effect with regard to the conversion from early to late-stage AMD of high doses of antioxidants (vitamin C, vitamin E, beta-carotin) [1].

Lipofuscin formation and interaction with cellular function

Release of incompletely degraded material at the basal RPE cell side as an initial event implicates malfunctioning lysosomal function with age. An essential function of postmitotic RPE is the phagocytosis of constantly shed photoreceptor outer segment disks and degradation with subsequent release of completely degraded material at the basal cell side where it is normally cleared by the choriocapillaris. With age lipofuscin accumulates in the lysosomal compartment [12, 15, 21, 23, 49, 72]. Although the mechanisms of lipofuscinogenesis are incompletely understood, there is strong evidence that oxidative damage plays an important role, with antioxidant deficiency or prooxidant conditions being of importance [4, 7, 14, 61]. Electron microscopic studies have shown a stepwise conversion of lysosomal structures to LF granules. Apparently, once formed the RPE cell has no means to either degrade or transport LF material/granules into the extracellular space via exocytosis. Subsequently, these granules are trapped in the cytoplasm and may occupy large amounts of the cytoplasmic volume. Although lysosomal hydrolytic enzymes are capable of degrading most cellular macromolecules, aggregation of molecules and posttranslational modifications generates biomolecules that are no longer degradable by human enzymes.

Recent analyses of compounds in isolated human LF granules revealed various molecules with toxic properties including malondialdehyde (MDA), 4-hydroxynonenal (HNE) and advanced glycation endproducts (AGE) [67] as well as a Schiff base reaction product, N-retinylidene-N-retinylethanolamine (A2-E) [18]. A2-E represents a major fluorophore of LF in the RPE [18, 19]. Molecular mechanisms have been identified how A2-E interferes with normal lysosomal function upon reaching critical levels [9, 40, 65]. A2-E induces a striking inhibition of lysosomal degradation in human RPE cells [19] by inhibition of the ATP dependent lysosomal proton pump [9]. Furthermore, A2-E has phototoxic and detergent properties and is capable of inducing disintegration of various organelle membranes upon reaching a critical concentration [64–66]. In search for further potentially toxic compounds of LF besides A2-E, RPE human LF granules have been isolated and purified and a proteome-analysis has been performed. Interestingly, a broad spectrum of LF proteins showed specific posttranslational modifications including MDA, HNE and AGE [67]. This underscores the potential contribution of oxidative damage in lipofuscin biogenesis.

Further evidence for a pathophysiologic role of LF includes a similar topographic distribution of LF and drusen, accelerated accumulation of LF in monogenetic macular dystrophies such as Best or Stargardt disease [2, 17, 53, 70] and a striking deposition of A2-E in RPE cells in ABCR knockout mice with strong dependence on light exposure. Recently, additional animal models with RPE LF accumulation have been established, including P-8 strain senescence accelerated mice, cathepsin-D-deficient transgenic mice, canine ceroid lipofuscinoses, and rds −/− mice [13, 47, 48, 54]. Interestingly, besides LF accumulation some of these models also display other AMD-like phenotypic changes including sub-RPE deposits [13]. These models will be helpful to further understand pathogenetic mechanisms and to identify targets for intervention in order to prevent deleterious effects of excessive LF accumulations.

Information on LF accumulation in the RPE has been largely obtained from in vitro studies using fluorescence microscopy techniques [16, 23, 24, 71, 72] and in vivo from fundus spectro-photometric investigations [15]. More recently, with the advent of scanning laser ophthalmoscopy, it is now possible to record fundus autofluorescence (FAF) images and to study topographic variations in vivo when using appropriate excitation wavelengths and barrier filters [8, 51, 52, 62, 69]. The technique was initially introduced by Rückmann and co-workers [62]. We recently introduced another novel confocal scanning laser ophthalmoscope (cSLO, Heidelberg Retina Angiograph, HRA) for FAF imaging which directly generates digital images [8, 10, 39, 46]. Delori and co-workers have shown that the autofluorescence signal derives from RPE LF [15]. Using the cSLO technique we demonstrated impairment of neurosensory retinal function and subsequent development of atrophy and visual loss confined to areas with prior excessive LF accumulation in eyes with AMD. Therefore, these clinical findings are in accordance with the above hypotheses based on experimental results implicating LF accumulation in cellular dysfunction and macular degeneration. FAF imaging has also led to new, refined phenotypic classifications of AMD manifestations which may be helpful to identify genetic factors.

Bruch’s membrane, composition and biogenesis of drusen

Bruch’s membrane between the RPE and the choroid shows various morphological and biochemical alterations with age [22]. These include changes in elastic and collagenous fibres, diffuse thickening and accumulation of material, which are thought to result in reduced elasticity and increased resistance to diffusion. The earliest clinically visible abnormality in AMD is the accumulation of drusen in the extracellular in inner aspects of Bruch’s membrane. Biochemically the deposited material is a complex composition of lipids, glycoproteins and proteins [38, 59]. Especially the lipoidal compounds increase exponentially beyond 40 years of age, and the accumulation in the macular area is higher than in peripheral Bruch’s membrane [68]. The proportion of cholesterol esters of the lipid component—the major lipid species are phospholipids and fatty acids rather than cholesterol and cholesterol esters—argue for an intracellular and, therefore, pigment epithelium, as opposed to the extracellular, choroidal origin of the material [37]. Especially diffuse thickening and progressive deposition of material in Bruch’s membrane may result in a barrier to normal metabolic exchange between the RPE and the choriocapillaris [25, 55]. The reduction of the hydraulic conductivity of Bruch’s membrane may hamper the movement of water towards the choroid, thus causing it to accumulate in the sub-RPE space and leading to RPE detachment. All late-stage manifestations associated with severe visual loss in AMD are considered as processes reactive to these initial alterations in Bruch’s membrane and the RPE.

Role of immune response

Recent evidence suggests a role for local inflammation with complement activation and immune complex deposition in eyes with AMD [3, 6, 44, 56]. Cellular debris that is derived from compromised RPE cells becomes sequestered between the RPE basal lamina and Bruch’s membrane. Failure to eliminate the entrapped material generates a local pro-inflammatory signal that is sufficient to trigger subsequent events including local upregulation of cytokines, acute phase reactants, and other pro-inflammatory mediators, activation of the complement cascade, invasion of incipient drusen by processes of dendritic cells from the choroid and the induction of an immune response to exposed antigens in the sub-RPE space. Thus, extracellular debris and the chronic inflammatory stimulus constitute a “nucleation” site for drusen growth, which is supported by the findings that drusen contain a variety of inflammation-related proteins including CRP, vitronectin, α1-antichymotrypsin, amyloid P component and fibrinogen [6]. A number of these proteins appear to be produced locally by distressed RPE cells rather than circulating in the blood [5, 33, 58]. The identification and localisation of multiple complement activators (nuclear fragments, membrane bound vesicles, lipofuscin, cholesterol and microfibrillar debris) [20, 42] as well as terminal complement compounds in drusen support the conclusion that drusen compounds act as a trigger of the complement cascade, a basic physiological reaction to foreign cells, dead and sick cells or cell fragments.

Interestingly, these inflammatory responses appear to be analogous to processes observed in other age-related diseases such as Alzheimer’s disease or atherosclerosis, where accumulation of extracellular deposits elicits a local chronic low-grade inflammatory response [5, 31, 32, 44, 45, 56]. Furthermore the presence of IgG and complement C3 and C5 intermediates is strongly suggestive of the presence of immune complexes, and is consistent with the presence of circulating autoantibodies in patients with AMD [30, 60]. The possible contribution of the immune system is further underscored by findings in patients with membranoproliferative glomerulonephritis type II. Here, complement activation and immune complex deposition cause glomerular injury, and the ocular phenotype is characterized by drusen resembling those in AMD in ultrastructure and composition (including C5 and IgG deposition) [57], as well as secondary CNV [50].

A mouse model deficient either in monocyte chemoattractant Ccl-2 or its cognate receptor Ccr-2 has recently been shown to develop cardinal features of AMD including accumulation of RPE-lipofuscin and A2-E, drusen, atrophy and choroidal neovascularisation [3]. This intriguing model implicates macrophage dysfunction in the pathogenesis of AMD. The extracellular debris between RPE and Bruch’s membrane contains IgG and C5a (activated complement compound C5), which mediate choroidal macrophage infiltration. In Ccl-2 −/− and Ccr-2 −/− mice and in human eyes with AMD impaired macrophage recruitment may allow the accumulation of C5a and IgG, usually degraded by the macrophages. This accumulation may induce vascular endothelial growth factor (VEGF) production by RPE cells and thus possibly mediating the development of CNV [3].

Finally, drusen appear to be targets of invasion by processes of dendritic cells [33]. As part of the inflammatory response, dendritic cells originating from the choroid may extend through Bruch’s membrane into the extracellular space below the RPE and have a primary role in antigen capture and presentation to T-lymphocytes. In rodents, choroidal dendritic cells have been implicated in experimental models of ocular inflammation [26, 27, 43], but the exact role and timing of an immune response directed against one or more autoantigens in the sub-RPE space and its relationship to drusen biogenesis remains to be elucidated.

Choroidal neovascularization

CNV represents the most common cause for severe visual loss in AMD. As soon as new vessels originating from the choriocapillaris grow through Bruch’s membrane they expand in horizontal direction either between the inner Bruch’s membrane and the RPE cell layer (“occult” CNV by angiographic characterstics) or in a plane further anterior through the RPE cell layer (“classic” CNV). Subsequent dysfunction of the neurosensory retina results from various mechanisms including hyperpermeability and haemorrhages. Over time there is a reduction in the vascular component of the membrane with a concurrent increase in the proportion of fibrous tissue. These neovascular processes in AMD are thought to result from a local imbalance of growth factors inducing growth of blood vessels inwards from the choroid [73]. Maintenance of the physiological architecture of the choriocapillaris requires a well-regulated balance between various factors derived from the retinal pigment epithelial basolateral plasma membrane domain including pigment epithelial derived factor (PEDF), VEGF and angiopoetins 1 and 2. VEGF, overexpressed by stimuli including hypoxia, induces the formation of new blood vessels, while PEDF serves as an inhibiting factor. The effect of PEDF is apparent in the vitreous and the cornea, which do not contain blood vessels. In addition to VEGF and PEDF, recent results indicate that angiopoetins 1 and 2 with their receptor Tie-2 play a part in the formation of CNV [34]. Ang 1 functions as an agonist and promotes vascular integrity and maturation, inhibits apoptosis and reduces permeability. Ang 2 is an antagonist of Ang 1 and promotes VEGF-induced angiogenesis. The receptor Tie-2 as well as Ang 1 and 2 have been shown to be present in choroidal neovascular tissue [34]. There are also receptors present that are specific for active endothelium in CNV membranes including av-integrins and TGF-beta receptor CD 105.

Interestingly, with age there is a reduction in the amount of PEDF in the vitreous [35], while there is an increased level of VEGF in humans with AMD and CNV [28]. Thus the concept has been brought forward that the normal non-vascular nature of Bruch’s membrane is due to the suppression of inward growth of choroidal blood vessels by the RPE. The stimulus to change in growth factor production by RPE is unknown, but it is assumed that it may be due to lack of metabolic supply from plasma due to reduced diffusion of material through the thickened Bruch’s membrane, or to reduced oxygen supply consequent upon changes in the choroidal capillaries [55]. Recent studies suggest immune responses as a contributing cause for CNV development. A chronic low-grade inflammatory process in the sub-RPE space, as outlined above, may induce local damage to Bruch’s membrane which would allow new vessels to breach the sieve-like extracellular matrix. In addition, RPE overexpression of VEGF, stimulated by complement components and IgG combined with fragmentation of the Bruch’s membrane, provides an environment permissive for CNV [3].

A similar mechanism could account for retinal angiomatous proliferations (RAP) [74]. This particular phenotype may occur in the setting of early AMD and involves invasion of the outer retina by retinal blood vessels, which may precede or occur simultaneously with CNV. Just as growth factors may prevent vascularisation of Bruch’s membrane under physiologic conditions they may also account for the avascular nature of the outer retina. In the first instance this would involve growth factors secreted through the basolateral plasma membrane domain of the RPE cell and in the second through the apical domain. It is therefore speculated that RAP may be due to a change in the balance of growth factors arising from the apical domain rather than the basal lateral plasma membrane domain.

Geographic atrophy

Geographic atrophy represents the second most common cause for severe visual loss in AMD and has been interpreted as the natural evolution of the disease if neovascular events do not occur prior to immediate cell death. In geographic atrophy there is a loss of the outer neurosensory retina, the RPE and the choriocapillaris and the outer plexiform layer of the retina becomes adherent with Bruch’s membrane [29]. Typically, the atrophic areas have well-defined borders with increased pigmentation at the margin. The atrophy initially tends to develop in the perifoveal area, while the fovea may be spared until later during the clinical course remeniscent of other bull’s eye maculopathies. This pattern reflects the density of rod photoreceptor cells. Primary dysfunction and cell death of RPE cells is thought to occur initially followed by collateral loss of neighbouring photoreceptor cells and the choriocapillaris. Recent findings with FAF imaging using a confocal scanning laser ophthalmoscopy implicated excessive accumulation in RPE cells in the evolution of geographic atrophy [41]. Hereby, an excitation wavelength of 488 nm is generated by an argon laser or an optically pumped solid state laser [39, 46, 63]. Emission is recorded above 500 nm by inserting a barrier filter. The evidence that the signal originates from lipofuscin in the RPE is derived from the work of Delori and co-workers [15]. Areas with increased FAF signal and, thus, excessive RPE lipofuscin precede the development of new atrophic areas and the enlargement of preexisting atrophy [41]. Hereby, a high interindividual variation in the phenotypic patterns of abnormal fundus autofluorescence has been noted with an impact on the progression of disease [10].

Furthermore in patients with unilateral visual loss from AMD, abnormal fundus autofluorescence in the fellow eye is associated with an increased risk for the development of GA. If the presence of A2-E or increasing lipofuscin is important to genesis of geographic atrophy, targets to slow down accumulation of lipofuscin in RPE cells need to be identified for potential future prophylactic and therapeutic interventions.

Outlook

New insights have emerged in recent years that lead to a better understanding of the key pathogenetic pathways associated in various manifestations of AMD. Better knowledge of molecular mechanisms is a prerequisite for the development of new strategies for more successful therapeutic interventions than those available to date.