A New Therapeutic Option for Reversing the Deficits in Dark Adaptation Associated with Age-Related Macular Degeneration (AMD)

Abstract

The earliest functional marker in age-related macular degeneration (AMD) is the delayed recovery of rod photoreceptor sensitivity following a bright flash. Underlying mechanism is thought to be reduced levels of retinoids in the retinal pigment epithelium (RPE) compromising the rate of transfer of 11-cis retinal to the photoreceptor for rhodopsin regeneration. Normally, retinoids are lost due to photo-oxidation in the photoreceptor cell and inefficient processing of outer segment discs by the RPE but this loss is compensated for by delivery of plasma retinol across Bruch’s membrane. Ageing of Bruch’s membrane is associated with a 10-fold decrease in capacity for transport that is further exaggerated in AMD. We had previously shown that saponins can remove deposits from Bruch’s membrane resulting in improved transport. As a proof-of-principle we have undertaken a pilot study with six AMD patients on oral saponin supplementation for 2 months (200 mg saponins/day) to assess the possibility of improving the transport across Bruch’s membrane. Saponin supplementation improved the rate of recovery in rod sensitivity following a bright flash in all AMD subjects (p < 0.005. paired t-test), indicative of improved delivery of retinol across Bruch’s membrane. The saponin intervention provides a new approach to slow, halt, or reverse the progression of AMD.

1. Introduction

Age-related macular degeneration (AMD) is the single largest cause of untreatable blindness in the elderly population [1]. In the European Union (EU), the number of affected individuals has been estimated as 67 million, with the figure predicted to increase to 77 million by 2050 [2].

Clinically, AMD is divided into early, intermediate, and late stages. Early AMD is characterised by the presence of drusen and pigmentary abnormalities of the fundus such as hypo- or hyperpigmentation. Although the presence of drusen remains a hallmark of AMD, it is important to realise that they are also present in normal elderly eyes. It is the size, number, and degree of confluency of druse that determines the risk for AMD pathology [3]. Progression to the late form results in geographic atrophy of the RPE followed by photoreceptor degeneration, also known as ‘dry’ AMD. The late phase is also associated with secondary complications of pigment epithelial detachments (PEDs, accounting for 12–20% of patients) and neovascular episodes (comprising 10–20% patients), the latter being designated as ‘wet’ AMD.

Epidemiological studies have shown AMD to be a highly complex, multi-factorial disease that has both a genetic disposition and considerable gene-environmental interactions [4, 5]. The diverse genetic associations with AMD best reflect the complex nature of the disease process. These genetic variations include several members of the complement pathway, components in lipid metabolism (cholesterylester transfer protein (CETP), hepatic lipase (LIPC), apolipoprotein E), and heat shock serine protease (Htra1), amongst others [6, 7, 8, 9]). This complexity is further enlarged by the additional association of dietary, environmental, and cardiovascular risk factors.

The complexity of underlying mechanisms involved makes it very difficult to nominate suitable targets for intervention. Current efforts at intervention have addressed the late-stage complications of inflammatory and neovascular processes with some degree of short-time success but since the underlying instigators have not been targeted, the underlying degenerative phase will continue, leading to the death of RPE and photoreceptors.

We have adopted an alternative approach for understanding the causative mechanisms in AMD. Since age is the highest risk factor in AMD, it would appear that some component of the normal ageing process is accelerated in this disease leading to the pathological outcomes. Based on an assessment of the earliest functional alteration in AMD (namely, slowed rod dark-adaptation), we have nominated the transport system in Bruch’s membrane as a specific target for intervention and the reasoning for this approach is discussed below.

2. Delayed recovery of rod sensitivity following a bright flash

The earliest functional marker in AMD is the delayed recovery of rod photoreceptor sensitivity following a bright flash [10, 11, 12]. Briefly, when a subject is dark-adapted for about 30 min, their retinal sensitivity to light is very high, given as the dark-adapted threshold (Figure 1, point A). Following a bright light flash that bleaches 95–100% rhodopsin, their retinal sensitivity is reduced, denoted by point B (Figure 1). Recovery in sensitivity begins with cones reaching a plateau in about 12–20 min. This is followed by recovery of rod sensitivity reaching the dark-adapted threshold in 30–40 min. The switch from cone to rod recovery is known as the cone-rod break (CRB). The initial rate of rod recovery is known as the S2 gradient and in this 30-year-old subject, it was determined as 2.0 dB/min. In AMD subjects, the S2 gradients are much lower, reaching dark-adapted thresholds much later (curve C, Figure 1) [13, 14]. In some AMD patients, the rod component is absent within the period of experimentation and may require hours to days to reach the dark-adapted threshold (curve D, Figure 1) [15, 16].

Figure 1.

Representative dark-adaptation curves for control and patients with AMD. In control subjects, there is an early recovery in cone sensitivity followed by recovery in rod sensitivity, reaching the dark-adapted threshold in about 30 min. In AMD, the cone-rod break (CRB) is usually delayed with slowed recovery in sensitivity (C). In some AMD patients, rod recovery may be absent within the period of observation (D).

Molecular mechanisms underlying the recovery in rod sensitivity are described briefly (Figure 2). Following a light flash, rhodopsin is converted to the apoprotein opsin and all-trans retinal (AT-RL), with retinal sensitivities falling from threshold A to B. Because of the toxic nature of AT-RL, it is converted to all-trans retinol and transported to the RPE to enter the vitamin A cycle [17]. After several enzymatic steps, the retinol ester is converted to 11-cis-retinal esters and stored in the RPE retinoid pool. After rhodopsin bleaching, 11-cis retinal is released from the RPE retinoid pool for delivery to photoreceptors and the rate of this transfer is related to the S2 gradient [18, 19]. It should be noted that recovery is dependent on the release of stored 11-cis retinal from the retinoid pool in the RPE and not on the reconversion of released all-trans retinal to 11-cis retinal.

Figure 2.

Rapid deployment of 11-cis retinal from RPE stores for regeneration of rhodopsin and recovery in rod photoreceptor sensitivity following a bright flash. See text for an explanation of the rod recovery process. R-RBP, retinol-retinol binding protein; R-RBP-TTR. Retinol-retinol binding protein-transthyretin complex.

Abnormalities in the sequence of events described above following a light flash offer plausible mechanisms for reduced S2 gradients in AMD. In some AMD patients, the presence of deposits between the RPE and receptors known as reticular pseudodrusen (RPD) is likely to impede the rapid transport of 11-cis retinal from the RPE to photoreceptors, reducing the S2 gradient [12, 15, 16]. Mutations in the members of the enzymatic and transport machinery of the vitamin A cycle are known to result in slowed S2 gradients but none have yet been reported in AMD [20]. Finally, the level of retinoids in the RPE pool is an important factor for determining the initial rate of transfer of 11-cis retinal to the photoreceptor. A reduction in the level of retinoids is expected to reduce S2 gradients.

If the vitamin A cycle was 100% efficient and all released AT-RL (from rhodopsin) was converted to 11-cis retinal, then there would be no change in the retinoid pool in the RPE. However, several factors are known that can modify the retinoid pool (Figure 3). First, released AT-RL can react with phosphatidylethanolamine to form N-retinylidene-phosphatidylethanolamine (NRPE), which then reacts with a second molecule of AT-RL generating a variety of toxic bis-retinoids, compromising the retinoid pool [21, 22]. Second, the highly oxidative environment within photoreceptor outer segments leads to peroxidation of lipid such as polyunsaturated fatty acids (PUFA), resulting in lipid aggregates, lipid-protein complexes, and protein cross-link formation [23, 24]. These damaged outer segment discs are phagocytosed by the RPE and inefficient processing and extrusion onto Bruch’s membrane lead to further loss of retinoids. Within the RPE, the bis-retinoids undergo further modification, resulting in the production of A2E, the fluorophore of lipofuscin. Some of the A2E is voided onto Bruch’s membrane and the extent of retinoid loss can be judged by the age-related exponential increase in A2E content of the membrane [25].

Figure 3.

Mechanisms underlying the reduction in the retinoid pool of the RPE in normal ageing and AMD. In the photoreceptor outer segment, the major loss of retinoids is due to the conversion of AT-RL to bis-retinoids. Oxidatively damaged outer segment discs cannot be digested effectively in the RPE and the degradation products are either stored as lipofuscin particles or extruded onto Bruch’s membrane constituting a secondary loss of retinoids. PUFA, polyunsaturated fatty acids; A2E, the fluorophore of lipofuscin.

Thus, to maintain the retinoid pool in the RPE, this loss must be compensated for by delivery of plasma retinol across Bruch’s membrane. Retinol is transported across Bruch’s membrane as a trimeric complex of retinol, retinol binding protein, and transthyretin of aggregate molecular weight (MW) of 75 kDa [26].

3. Ageing and loss of transport across Bruch’s membrane

Normal ageing of Bruch’s membrane is associated with gross structural alterations due to the accumulation of lipid-rich debris, damaged and trapped transitory protein species, and extensive cross-linking of collagenous fibres, increasing its thickness and therefore reducing the driving diffusional gradients across the membrane [24, 27, 28].

The hydraulic and diffusional transport functions of human donor Bruch’s membrane have been assessed as a function of age. Diffusional processes were assessed using the albumin molecule (MW 64 kDa) of similar molecular weight to the carrier for retinol (MW 75 kDa) and other carrier proteins for vitamins, anti-oxidants, and metals. In normal ageing, transport functions across Bruch’s membrane showed an exponential decline with a half-life of 16–18 years, i.e., transport capacity was halved for every 16–18 years of life (Figure 4) [29, 30, 31]. In the elderly, transport capacities were reduced by nearly 10-fold. Such a reduction may compromise the delivery of retinol leading to a slight reduction in S2 gradients observed in the elderly [32, 33]. In the advanced ageing scenario of AMD, the diffusional transport across Bruch’s membrane is further reduced (compared to age-matched controls) and may reflect the reduced transport of retinol and subsequent deficits in S2 gradients [30].

Figure 4.

Age-related reduction in the transport properties of human Bruch’s membrane. Diffusional status of albumin across Bruch’s membrane declines exponentially with age with transport capacities being halved for every 16–18 years of life (after Lee et al., 2015, Ref. [29]).

The reduction in transport across Bruch’s membrane will not only affect the movement of retinol but will impact on all other carrier-mediated metabolites such as essential lipids, vitamins, anti-oxidants, and metals involved in the enzymatic protection machinery for dealing with oxidants and free radicals. The outward removal of toxic metabolites will also be affected, leading to their accumulation in Bruch’s membrane and triggering inflammatory mediators and thereby undermining the survival of the photoreceptor-RPE complex.

4. Regeneration of Bruch’s membrane

As mentioned earlier, ageing of Bruch’s membrane is associated with accumulation of lipid-rich debris that essentially forms a barrier to the transport of metabolites. Reactive lipid fragments, bis-retinoid products, and free radicals lead to adducted and denatured proteins and cross-linking of the collagenous framework of the membrane. In elderly Bruch’s membrane, nearly 50% of the collagen was found to be in a damaged state [24]. Normally, these ageing alterations are dealt with by continuous coupled processes of synthesis and degradation of the matrix, the latter process mediated by proteolytic enzymes called the matrix metalloproteinases (MMPs). These enzymes are released by the RPE in a latent form and following activation, can degrade most components of the matrix. Pro-matrix metalloproteinase 2 (Pro-MMP2) is the homeostatic enzyme in the system. For activation, proMMP2 and tissue inhibitor of metalloproteinase 2 (TIMP2) diffuse through Bruch’s membrane to form a trimeric complex with matrix metalloproteinase 14 (MMP14) (membrane bound) on the basal surface of the RPE (proMMP2-TIMP2-MMP14). A second molecule of MMP14 then cleaves the inhibitory peptide on pro-MMP2 to release activated-matrix metalloproteinase 2 (MMP2) [34]. This system works well in the young but deteriorates with age, leading to a reduction in the level of activated MMP2 [35]. This is due most likely to slowed diffusional processes in ageing Bruch’s membrane that restrict the formation of the trimeric complex for activation of proMMP2. In addition, released activated-MMP2 is most likely to be constrained and trapped within the grossly altered matrix of Bruch’s membrane. In the advanced ageing of AMD, the level of activated-MMP2 in Bruch’s membrane was about 50% of that in age-matched controls [36].

5. Potential target for intervention in AMD

From the above, it is obvious that any intervention must address the diminishing transport functions of Bruch’s membrane. One possibility for overcoming the deficit in transport is to increase the diffusional gradient of metabolites across the membrane. Since oxidative damage in the RPE and photoreceptors is the primary driver of ageing changes, anti-oxidant and vitamin supplement regimes have been devised as a possible interventionist measure to reduce oxidative stress and hopefully slow the progression of AMD [37]. Thus, the Age-Related Eye Disease Study (AREDS) dietary supplementation cocktail was devised (vitamin C (500 mg), vitamin E (400 international units (IU)), beta-carotene (15 mg), zinc oxide (80 mg), and cupric oxide (2 mg)). Despite the wide use of AREDS supplements for over 20 years, controversy remains as to its usefulness since it does not prevent legal blindness in advanced AMD [38]. It has been pointed out that the earlier reported decrease in progression was related to the occurrence of neovascularisation rather than slowing the progression of dry AMD [39].

Given the fact that the diffusion of metabolites across Bruch’s membrane is reduced by nearly 10-fold in the elderly, and more so in AMD, one must question the likely effectiveness of such dietary supplementation. The major problem with supplementation therapies is that they do not address transport in the opposite direction across Bruch’s membrane, i.e., the removal of toxic metabolites that are the likely triggers of neovascular and inflammatory episodes.

The alternative to supplementation therapies is to improve the bidirectional transport pathways across Bruch’s membrane. This would require the destabilisation and dispersal of the lipid-rich debris and the removal of normal and damaged proteinaceous deposits. Such a strategy would also release trapped activated MMP enzymes that could participate in hydrolysing the altered collagenous components. The expected improvement in intra-membrane mobility would favour greater activation of pro-MMP2, kick-starting the normal rejuvenation machinery in Bruch’s membrane. Essentially, we would need to reverse the ageing process, i.e., elevate the decaying transport curves shown in Figure 4 so that they no longer crossed the failure threshold within the lifetime of an individual.

6. Saponin-mediated improvement in the transport properties of Bruch’s membrane

In vitro, surfactant molecules such as sodium dodecyl sulphate (SDS) can remove lipids and denatured, damaged, and aggregated proteins from Bruch’s membrane and have been used to isolate proteins for SDS-PAGE (polyacrylamide gel electrophoresis). However, this class of surfactants is extremely toxic and cannot be considered for clinical intervention. Saponins, on the other hand, are naturally occurring amphipathic molecules, with the major source being the ginseng plant (Panax ginseng) and marine sea cucumbers (Apostichopus japonicus). In the Far East, saponin extracts have been used as tonics for thousands of years to promote general well-being without any side effects. Structurally, saponins from the ginseng plant have a 4-membered ring in the hydrophobic domain, whereas sea cucumber saponins have a 5-membered ring (Figure 5). The incorporation of different chemical groups at the R1, R2, R3, and R4 attachment sites (sugars, hydroxyl, hydrogen, and alkane chains) gives rise to a myriad of compounds and over 400 have been characterised [40, 41, 42]. The amphipathic nature of saponins (containing hydrophilic and hydrophobic domains) allows them to readily partition into lipid structures such as liposomes, micelles, membranes, or lipoidal deposits and assist dispersal [43, 44].

Figure 5.

Molecular structures of saponins derived from the ginseng plant and sea cucumbers. Saponins from ginseng have a 4-membered aglycone ring, whereas those from sea cucumbers have a 5-membered ring. The ring structures impart hydrophobic properties to the molecules. The several attachment sites containing primarily various glycosidic units represent the hydrophilic domain. The amphipathic nature of these molecules allows the solubilisation of lipoidal and proteinaceous deposits.

Using isolated human donor Bruch’s membrane preparations, we have shown that a single perfusion with saponins results in the release of various classes of lipid deposits, and attached and trapped proteinaceous material [29]. The released protein component also contained active-MMP2 enzymes. Release of this debris was associated with improvements in the transport of albumin-sized molecules across Bruch’s membrane (Figure 6A). A dose-response curve for saponin-mediated improvement in hydraulic conductivity of Bruch’s membrane showed hyperbolic saturation kinetics with the maximum improvement being 2-fold over basal levels (Figure 6B). With the in vitro work, saponins were capable of removing lipoidal and deposited proteinaceous debris from Bruch’s membrane. However, they did not target the cross-linked collagenous fibres of the matrix. In vivo, improved diffusional status of Bruch’s membrane would allow greater mobility of proMMP2 and TIMP2 for interaction with MMP14 on the RPE basal membrane, resulting in increased levels of activated-MMP2 to target the damaged collagenous network, leading to greater improvement in transport.

Figure 6.

Saponin-mediated improvement in the transport properties of human Bruch’s membrane. (A): Incubation of Bruch’s membrane with saponins improved the diffusional transport of albumin 2-fold (p < 0.001), open circles. Control; filled circles, saponin treated. (B): Dose-response curve for saponin-mediated improvement in hydraulic transport across Bruch’s membrane. After Lee et al., 2015, Ref. [29].

7. Saponins and dark adaptation

Saponin-mediated improvement in transport across Bruch’s membrane could provide an interventionist strategy in AMD. Improved transport of retinol would augment the retinoid pool in the RPE, leading to improvement in S2 gradients. To assess this possibility, we have undertaken a pilot study with six AMD patients.

The dosage of saponins to use in this preliminary study was estimated as follows. Based on the bioavailability of saponins as 1.12 ± 1.6%, the half-life of a given saponin (Rb1) in plasma of 19.5 h, and the dose-response curve of Figure 6B, we calculated that a daily oral dose of 200 mg saponins should achieve a 0.5–1.0-fold improvement in transport across Bruch’s membrane in our subjects over a period of 2–5 months [29, 45, 46, 47, 48]. This dosage is similar to those of the saponin products marketed in the Far East for general health. Rather than using a single saponin species, we decided to use a wide mixture so that the varied nature of deposits in Bruch’s membrane could be targeted. Thus, the final mixture of 200 mg saponins per day contained 50 mg of ginseng saponins and 150 mg of sea cucumber saponins.

Dark-adaptation curves were obtained using a Humphrey Field Analyser (HFA) (Series II, Model 750i, Zeiss). Use of the HFA to obtain dark-adaptation curves has previously been described [32, 33, 49]. In clinics, the HFA is normally used to monitor the progression of glaucoma. For this purpose, the viewing globe in the HFA has a background illumination of 10 candelas per square meter (cd m⁻²) to suppress rod function, allowing the monitoring of cone responses. To obtain dark-adaptation curves (i.e., both rod and cone responses), we could have temporarily removed the background illumination. However, restoring background illumination takes a long time for stabilisation of light levels and this would interfere with the normal operation of the glaucoma clinic. The alternative to switching off the background illumination was to ask our AMD subjects to wear light-tight goggles incorporating 3.0 neutral density filters. The highest light intensity of the test spot offered by the HFA was 3185 cd/m². With the 3.0 ND goggles, this energy was reduced to 3.185 cd/m², providing a sufficient test range to obtain rod and cone dark-adaptation curves, as shown in Figure 1.

Altogether, seven test spots were utilised to measure sensitivity, three on the vertical meridian going through the fovea (yellow cross) and two on either side of the meridian (Figure 7). The seven spot locations were chosen to (a) allow random presentation of the test spots within the bleached area and (b) to obtain an average sensitivity change over the bleached region.

Figure 7.

Fundus bleaching region and location of test spots. The bleached area in the upper hemisphere enclosed by the dashed rectangle subtended a width of 10.4° and a height of 6.6°.

Following dilation, subjects were dark-adapted for 30 min. Dark-adapted thresholds were measured several times so as to familiarise the patient with the operational procedure of the HFA. After obtaining the dark-adapted thresholds, the goggles were removed and the patient was moved to an adjacent table and positioned in front of a box housing the bleaching apparatus. This box contained a red limit-emitting diode (LED) for fixation, a coupled-charge device (CCD) camera for monitoring fixation, and a flash unit (Speedlight SB-800, Tokyo, Japan). The patient was asked to look at the fixation light, and once confirmed, a flash of white light of integrated intensity 7.53 log scotopic Troland seconds (log sc td) was delivered inducing a calculated bleach of ~98% [50]. The bleached rectangular area subtended a width of 10.4° and a height of 6.6° and was located in the superior macula, just under the arcades (Figure 7). After the bleach, neutral density goggles were replaced and the patient was moved to the HFA to undergo repeated threshold assessments every 2 min for a period of 30–40 min.

At each time point, data from the seven fundal locations were averaged to produce a mean ± SD and plotted to obtain a dark-adaptation curve. Both rod and cone portions were fitted separately to an exponential function of the type dB = a + (b*exp.(−t/k)) where a, b, and k are constants and t is the time (in minutes) after flash using the add-in Solver module of the Microsoft EXCEL spreadsheet (EXCEL Microsoft, Redmond, WA, USA) [13]. The intersection of the rod and cone curves was taken as the CRB time (min). Differentiation of the above equation with respect to time was used to calculate the gradient of the rod fit at time of the CRB and designated as the initial S2 gradient. In the absence of a rod portion within the 30–40 min of testing, the S2 gradient was designated as zero, S2 = 0.

In 10 control subjects, the S2 gradient was determined as 1.98 ± 0.16 dB/min, values equivalent to 0.2 log cd m⁻², similar to those reported in the literature [18, 51, 52, 53]. The reproducibility of the dark-adaptation technique was confirmed following repeated assessments in a control subject over a period of 6 months, showing little variation in S2 gradients, over the three visits (Figure 8).

Figure 8.

Reproducibility of the dark-adaptation procedure. A control subject was tested three times over a period of 6 months and demonstrated little variability in S2 gradients.

Dark-adaptation curves for all the AMD patients are shown in Figure 9. On the basal visit, the six patients in the present study had S2 gradients of 0.6 ± 0.38 dB/min, considerably reduced compared to controls. One of the patients (subject F) did not show the presence of a rod component on the basal visit, i.e., S2 = 0. After basal determination of S2 gradients, all subjects were started on the saponin supplementation for a period of 2-months. Dark-adaptation curves were repeated on these patients after 2-months and the five patients who showed the presence of a rod component at the basal visit displayed improved S2 gradients following saponin supplementation.

Figure 9.

Plots of dark-adaptation curves at basal visit and after 2 months on the saponin supplementation for all subjects in the study. All patients showed improved S2 gradients after saponin treatment. The horizontal dashed lines represent the pre-flash dark-adapted threshold.

Changes in S2 gradients following saponin supplementation are given in Figure 10. In this figure, basal S2 gradients are plotted against gradients after 2 months on saponins. The diagonal line in the figure is the line of no change. AMD patients on the saponin treatment showed significant improvements in their S2 gradients (p < 0.005, paired t-test). Patient F who lacked the rod component on the basal visit now demonstrated its presence after saponin intervention, with a S2 gradient of 0.43 dB/min.

Figure 10.

Effect of oral supplementation of saponins over a period of 2 months on rod S2 gradients in AMD subjects. The oral supplementation regime showed improved S2 gradients in all patients (p < 0.005, paired t-test).

8. Conclusions

In this preliminary study, saponin intervention (200 mg/day) for 2 months considerably improved S2 gradients in AMD patients, enhancing the rate of recovery in rod photoreceptor sensitivity following a bright flash. We and others had hypothesised that the delayed recovery in rod sensitivity following a bleach was most likely due to reduced transport of retinol across Bruch’s membrane, compromising the retinoid pool in the RPE. Saponins have been shown to remove lipoidal and proteinaceous debris from Bruch’s membrane improving its transport functions. This most likely underlies the improved transport of retinol leading to the observed improvement in S2 gradients in these AMD patients.

The proposed improvement in retinol transport across Bruch’s membrane has implications for all other carrier-mediated metabolites and waste products. Thus, transport of vitamins, anti-oxidants, and essential metals that support the anti-oxidant defence machinery will be increased serving to maintain the health of the RPE-photoreceptor complex. Furthermore, improved transport across Bruch’s membrane is expected to remove the accumulated toxic metabolites that are the triggers of neovascular and inflammatory processes. The current study has provided evidence for considering saponins as a potential new avenue for intervention in AMD. Further studies are now required to assess (a) if further improvements in S2 gradients can be elicited with prolonged saponin supplementation, (b) the stability of the improvement following withdrawal of the medication, (c) determining effects on the rate of progression of the disease, and (d) monitoring changes in clinical parameters such as drusen load. Given our current understanding of mechanisms underlying AMD, the saponin intervention is expected to slow, arrest, or reverse the progression of the disease.