advertisement

Editors Selection IGR 13-2

Comments

Comment by Jonathan Crowston & Yeni Yucel & Balwantray Chauhan & Keith Martin & Art Neufeld & Lan Wang & Chris Leung & James Morgan & Ian Trounce on:

Cellular mechanisms that are common to the pathogenesis of neuronal death in various neurodegenerative disorders have long been sought. This study provides evidence derived from a rat ocular hypertension model that at least superficially points to a potential link between the causes of neuronal attrition in Alzheimer's disease and retinal ganglion cell loss in glaucoma. In this interesting study the authors demonstrated that β amyloid, a major constituent of senile plaques in Alzheimer's Disease, co-localized with apoptotic retinal ganglion cells and induced RGC death in ocular hypertensive rats. Three treatments that prevent aggregation of β amyloid through different pathways all reduced RGC apoptosis, with the largest protection afforded when all three treatments were given together.

Although the addition of exogenous β amyloid induced RGC death in this model, it is still a matter of debate as to whether β amyloid aggregation per-se is the cause or rather the result of neuronal injury in Alzheimer's disease. Amyloidogenic processing of the amyloid precursor protein (APP), a synaptic neurotrophic molecule, to β amyloid occurs in response to oxidative stress in the retina (Xiong et al. 2007). Retinal ganglion cell loss as such may result either from the loss of protective APP or from the toxic effects of amyloid β.

It is still a matter of debate as to whether beta amyloid aggregation per-se is the cause or rather the result of neuronal injury in Alzheimer's disease

1. Reduced APP was observed with time in the OHT rats in this study and so primacy among these changes still needs to be determined Regardless, these data add compelling evidence in addition to prior reports (McKinnon et al. 2002) that abnormal processing of the amyloid precursor protein does occur in ocular hypertensive rats. Guo and colleagues demonstrated RGC apoptosis in vivo using annexin-V-fitc and confocal scanning laser ophthalmoscopy. This technology has the significant advantage of allowing the investigator to demonstrate longitudinal changes in incident RGC apoptosis in a single animal. Co-labelling RGCs with DiI was also performed in order to determine the absolute numbers of RGCs and thus calculate % apoptosis.
2. It is not clear whether DiI retrograde labelled RGCs were also imaged in vivo in the same sitting as the annexin-V or from retinal flat mounts in separate animals.
3. The inclusion of data showing the changes in absolute RGC number would also have been helpful in order to show conclusively that treatment actually resulted in the long-term preservation of retinal ganglion cells. Although unlikely, it is possible that treatment per se interfered with annexin-V binding and so gave falsely lower levels of apoptosis.
4. A further potential concern is the high percentage of apoptosis in the OHT model. In the control group at around four weeks apoptosis rates peaked at around 15%. Although the length of time a cell is annexin-V positive varies according to cell type, duration of positive labeling is commonly in the region of 12 to 24 hours. Even shorter times have been suggested in data from prior manuscripts by this group. If RGCs are annexin-V positive for 12 hours then with a 15% apoptosis rate one might expect around 30% of RGCs to die in a 24-hour period. Such death rates are extremely high and not compatible with the time frame over which RGC attrition has been reported elsewhere for the rat OHT mode. Expressing cell death as % apoptosis without knowing the number of remaining retinal ganglion cells limits interpretation of this information.

   Expressing cell death as % apoptosis without knowing the number of remaining retinal ganglion cells limits interpretation of this information

5. The relevance of these findings to human glaucoma also needs to be established. β amyloid accumulation has still not been convincingly demonstrated in the retinas of human glaucoma patients.
6. In addition it is still unclear as to whether there is an increased prevalence of glaucoma in Alzheimer's patients and whether a predisposition to abnormal APP processing in the central nervous system is a risk factor for developing open angle glaucoma. Discerning the links between these neurodegenerative diseases is set to provide an exciting research focus in glaucoma.
   In vivo imaging of retinal ganglion cells (RGCs) offers a new paradigm to study glaucoma neuroprotection. In the study by Liet al., targeting the pathway of amyloid-β (Ab) formation was shown to reduce glaucomatous RGC apoptosis. The authors highlight the potential of using the technology of in vivo RGC imaging - DARC (Detection of apoptosing retinal cells) to monitor the treatment response. As illustrated in figure 4 of the article, Abab significantly reduced apoptosis at weeks 3, 8 and 16 whereas Congo R and b-secretase inhibitor were less effective at weeks 8 and 16.
   This novel finding is of great scientific value as it opens a new window to longitudinally study the effects of neuroprotective treatment.
7. The result may be made more convincing if images from the same eyes had been shown in each treatment group. For example, it is apparent that figures 4D (Abab treatment at week 3) and 4E (Abab treatment at week 8) were imaged from the same eye whereas figure 4F (Abab treatment at week 16) was imaged in a different eye as revealed by the different pattern of retinal vasculature. Using the same eyes at each time point in the longitudinal analysis is important to control for individual differences in response to treatment.
8. It would be informative to know whether media opacity (cataract, vitreous hemorrhage, retinal detachment, etc.) developed during the course of study secondary to repeated intravitreal injection (annexin 5 and Ab therapy) and its effect on imaging and cell counting.
9. As previously demonstrated by the authors, RGC apoptosis increased with the integral IOP in the experimental ocular hyper-tension model (Cordeiro MF et al., PNAS 2004;101:13352-6). It would be important to indicate in the study that the integral IOP was comparable between the control and each of the treatment groups.
10. My main issue with the paper is that there are no data provided on the intraocular pressure (IOP) profiles of the groups. The authors state that 'the IOP of both eyes in each rat was measured at regular intervals with a Tonopen XL, and the integral IOP was calculated', but these data are not shown. I could not even see a statement that the IOP profiles were similar between different groups. Was the effect on apoptosis demonstrated simply due to an effect on IOP? I suspect not, but it is impossible to tell from the data as presented. The strength of protection therefore cannot be assessed.
11. Secondly, although the DARC method has obvious attractions in terms of the ability to conduct longitudinal studies, I am still uncertain about how the results obtained correlate with true ganglion cell survival. I have found this information difficult to determine from previous publications as well as the current work. The outcome data presented in the current study relate to apoptosis rates as determined by annexin 5 labelling. It is certainly conceivable that the Alzheimer's treatments assessed simply reduced this labelling rather than protecting the ganglion cells. Without RGC body or axonal survival data it is impossible to be sure. The only RGC count information presented was as a denominator for the '% RGC apoptosis / total RGC count' in Figure 5, and true cell survival therefore cannot be determined from the data shown.
12. The premise of the paper - that patients with AD have RGC loss associated with glaucomatous changes - is attractive, but remains contentious. Curcio investigated this possibility in 1993 and concluded that this was not the case on the basis of careful histological studies. A balanced review of the literature should refer to her work which discussed some of the discrepancies with Blanks' data.
13. Many of the points of debate related to Guo's study relate to the techniques used to generate the glaucoma model and quantify programmed cell death. Although the authors provide some details of the model little is given the changes in IOP following the induction of glaucoma. The peak and average pressures should at the very least be included. Since the Morrison model was used it is important to confirm that an episcleral ring was inserted at the time of the hypertonic saline injection.
14. Another concern is that it is not always clear when the data is based on DARC or counts from histological analysis. Some of the findings are intriguing; firstly that the apoptosis count is elevated in control eyes (Figure 2) in the absence of amyloid peptide. If these data are based on retrograde labelling of RGCs, is it possible that this reflect the potentially damaging long-term effects of retrograde labelling in the superior colliculus. It is difficult in later figures to determine precisely the rate of cell loss. In Figure 5C there appears to be an 8% apoptosis rate relative to control, but this is in eyes that did not have any IOP elevation. Does this suggest that cells can demonstrate annexin labelling and not commit to cell death? It would be helpful to discuss the possibility that not all cells that are annexin positive commit to cell death. In this respect it is important to provide independent measures of the apoptosis rate (for example based on caspase labelling). Comment should also be made on the high rate of annexin labelling- given the short period of time during which a cell might be annexin positive prior to cell death- this implies a high rate of cell loss (e.g., when the rate is 15%) with substantial loss of the retinal ganglion cell population within a few weeks of the induction of experimental glaucoma.

    The DARC technique is ingenious, but requires further clarification

15. The DARC technique is ingenious, but requires further clarification. For a healthy RGC population, images taken a short time after the induction of ocular hypertension would give some idea of the rate of cell death. However, with increasing time from the event, this becomes more problematic because of the difficulty in determining the size of the underlying RGC population (since this will influence the number of cells that will be annexin labelled at any particular rate of cell death). For example, in Figure 4, panel F the optic disc appears to be extensively cupped and it is possible that extensive RGC loss has accounted for the reduced level of annexin labelling. The possibility should also be considered that the injection of several agents to prevent the accumulation of β amyloid could be toxic and induce a rapid phase of cell death- if this occurred before the time at which measurements were made (3 weeks) then the low rate of annexin V labelling would be attributed to a protective rather than damaging effect.
16. A sizeable area of the retina was sampled (40%), which is certainly a larger area of retina than has been used by most investigators over the last 10 years. The calculations performed for the data represented by the y-axes of many of the figures are not clearly indicated. Thus, a y-axis with '% RGC apoptosis compared to control' is difficult to understand. Nevertheless, when I wrote down equations that I thought might produce this kind of data and did the calculations based on assumed raw numbers for RGC counts, I came up with numbers that seemed to fit the data being presented in Figures 5 and 6. However, based on the ~250 apoptotic RGCs seen in control eyes (Figure 2B, 0 nmol Ab), I could not come up with the numbers in Figure 2A. Perhaps the y-axis in Figure 2A is incorrectly labeled. In any event, I would be interested in seeing the equations, and some raw data, that were used to generate the line and bar graphs in this paper.
17. Based on the reported data, the DARC method of detecting apoptotic RGCs must be very accurate and highly sensitive. My laboratory has used retrograde labeling of RGCs with Fluorogold and computer-assisted counting of RGCs in retinal flatmounts. We sampled about 25% of the retina and always compared the experimental eye to the contralateral control eye of the same animal. With these methods, we were able to distinguish 20% loss of RGCs from 10% loss from 0% loss with statistical significance. The DARC method appears to be able to show apoptosis differences, based on total number of RGCs, between 15%, 6% and 1% (Figure 5B, no statistics given) and to distinguish 75% from 85% reduction in RGC apoptosis (Figure 6). Retrograde labeling would not be able to show such small differences.
18. Finally, I am wondering if intravitreal injection of Ab caused displaced amacrine cells, which account for 40-50% of the neurons in the ganglion cell layer of the rat, to undergo apoptosis. Are RGCs selectively damaged by Ab. If both RGCs and displaced amacrine cells are affected by Ab, then the observations of Ab toxicity (Figure 2) most likely applies to both neuronal types.
19. Assessment of apoptotic cells is the single outcome measure of all experiments. The fact that this is based on annexin-V¹ is a major limitation, because Annexin-V, the sole assay, significantly interferes with the processes and the measured experimental outcomes. Raw data is not available other than ratios and what is presented in photographs.
20. Specificity of annexin positive cells for apoptotic retinal ganglion cells
    Microglia and macrophages are activated both in the glaucoma model used and also following intravitreal β-amyloid injection.^2-4 However, both microglia and macrophages also undergo apoptotic cell death.⁵ It is not possible to differentiate an apoptotic retinal ganglion cell from a apoptotic microglial cell by using a X16 magnification as the authors indicate they did. A host of available markers for microglia including Cd11b, are available for double-labeling experiments to determine how many profiles are apoptotic microglia or microglia containing apoptotic cell debris.
21. Confirmation of apoptosis using independent markers
    During apoptosis, a phosphatidylserine (PS) site moves from the inner cell membrane surface to the outer surface where it signals phagocytosis by microglia.^6-9 Annexin-V binds to this very same site, and thus interferes with the normal phagocytosis and clearance of degenerating apoptotic cells. This delayed clearance of apoptotic cells may overestimate the actual number of apoptotic cells at a given time point. Thus, independent and readily available methodologies such as TUNEL or caspase activation are needed to confirm the number of apoptotic cells without annexin-V injection. Furthermore, the PS site plays a major role in β-amyloid toxicity and the microglial response following β-amyloid injection. Annexin binding to the PS site will alter these processes and the measured outcome, ie. The numbers of apoptotic cells.^10,11 Thus, estimating the number of apoptotic cells without annexin-V injection by independent methodologies becomes even more important.
22. Evidence that annexin-positive cells are only RGCs
    In two previous studies^3,4 from the UK and not cited by the authors, β amyloid was injected into rats. They were not able to find RGC apoptosis, however did find photoreceptor apoptosis by TUNEL assay and atrophy of RGCs. It would be important to discuss the discrepancies between the results of previous studies with the authors' work.
23. Adequate controls
    In experimental eye injury and in β-amyloid toxicity, the authors used the fellow eye as the controls. The fellow eye is not adequate as it has been shown to undergo changes following unilateral β-amyloid injection^3,4 and unilateral optic nerve injury.¹²
24. Applicability of Dark Agouti rat strain
    The Dark Agouti rat strain has had its main use in immunology due to its particular susceptibility to autoimmune injury. For example, it is a rare strain in which experimental allergic encephalitis can be induced only with CNS tissue and without adjuvant injection.¹³ It is not clear whether results in this strain can be generalized to other rat strains. In the Dark Agouti strain, it would be particularly important to address microglial activation as in point #1.
    Guo et al. provide intriguing evidence implicating amyloid-β (Aβ) in the pathogenesis of experimental glaucoma. They show that intravirtreal injection of Aβ leads to death of retinal ganglion cells (RGCs), that targeting Aβ with a variety of agents, including an antibody to Aβ (Aβab) - the most potent of these neuroprotectants - leads to decreased RGC loss in the model and finally that combining these various agents (triple-combination therapy) provides the most effective neuroprotection strategy. Hence in a matter of six printed pages the authors have not only revealed a novel pathway of RGC that suggests a close link between glaucoma and Alzheimer's Disease typified by plaques containing Aβ, but also tested and provided the ideal combination for a new method of glaucoma treatment.
25. The authors use different methodologies to derive their conclusions and each depends on changes that occur in RGCs. In the first set of experiments, the authors show co-localization of annexin-5 (a marker of apoptosis) and Aβ to cells in the RGC layer. There is actually no evidence that these are really RGCs, hence statements on the unequivocal Aβ deposition in dying RGCs is far from substantiated. There is no quantitative evidence showing the degree of co-localisation of Aβ and annexin-5 in RGCs, or demonstration of Aβ deposition in naïve animals as negative controls. Further, the dramatic loss of inner plexiform layer thickness in the sections in Figure 1 shows that the damage created by the model used in these experiments is not confined to the RGC layer. There is a possibility of other cells types in the RGC layer such as microglia and macrophages in addition to displaced amacrine cells which normally comprise up to 50% of the cell population in the RGC layer. A more convincing methodology to show that these changes occur exclusively in RGCs would have been useful.
26. The authors used a technique developed in their laboratory called Detection of Apoptosing Retinal Cells (DARC), presumably an acknowledgment that this imaging method detects annexin-5 in cells besides RGC, yet, the authors continue to refer to RGCs in findings from DARC. The DARC images in Figs 4 and 6 show highly segmental annexin-5 labelling and is not in agreement with the distribution of RGC loss found by others who have used this model of glaucoma. Further, there are dramatic variations in the size of positive cells, with some being larger than second order arterioles.
27. Finally, the authors introduce the concept of 'peak apoptosis' as applied to glaucoma. The rate of RGC loss in human glaucoma is not known, however, it is possible at least in some patients, RGC loss occurs linearly and that at any one point in time a comparable number of RGCs are undergoing apoptosis, perhaps even in end-stage damage. The concept of peak apoptosis suggests that there is an insult or perhaps a series of closely separated events that cause RGC loss. The evidence for this model is not obvious and the relationship between any number of cells undergoing apoptosis and actual RGC counts remains elusive.

• ← Previous
• Next →