Immunization of mice with CEP-MSA provides a valuable model to study dry AMD from an immunological perspective, helping to dissect the immune system's role in the development of disease. Studies presented here link the AMD-like histopathological changes with the presence of macrophages in the outer retina during early stages of disease, suggesting that macrophages are involved in the underlying pathology. Notably, our data suggest that BALB/c mice tend to be more sensitive to immunization with CEP-MSA than B6 mice by having a greater magnitude and earlier significant difference of inflammatory cells in the IPM when compared to age-matched controls. In addition, dry AMD-like pathology, such as RPE cell hypertrophy, vacuolization of RPE and ROS, and RPE cell pyknosis, is also found at greater magnitude and arises earlier following immunization in the BALB/c mice. We also show that old CEP-immunized BALB/c mice develop photoreceptor cell loss. This suggests that future studies using this model would benefit from a more rapid and amplified immunopathological effect in BALB/c mice than in B6 mice, yielding results as early as 40–100 days postimmunization.
Even if BALB/c mice show an earlier significant response to immunization with CEP-MSA than age-matched B6 mice, it is important to stress that CEP-immunized mice contain larger numbers of macrophages in the IPM and AMD-like pathology than naïve controls in both strains. Furthermore, our data showed no statistically significant differences between the two strains through time. This suggests that any age-related changes seen in immunized mice observed during the early stages of disease are of comparable magnitude regardless of strain, but that the higher number of macrophages present in the IPM of BALB/c mice makes it technically easier for quantification of disease onset. In other words, the reason that there seems to be no early differences in naïve versus CEP-MSA mice on the B6 background is because the actual number of IPM-infiltrating cells is too low at that point to achieve statistical significance.
Differences between these two strains could also be attributed to background-specific (genetic and/or immune) mechanisms or to the reduced melanin levels in BALB/c mice. B6 mice are prone to develop T helper type 1 (Th1) responses, whereas BALB/c are Th2-prone. On the other hand, it has been shown that melanin in the RPE provides protection from light damage [30
]. By showing that albino B6 (Tyr−/−
) mice have comparable inflammatory cell numbers in the IPM and AMD-like pathology with WT B6 mice in a much less robust form than BALB/c mice, the possibility that light damage largely contributes to pathology is less likely. Indeed, it has been previously shown that B6 (Tyr−/−
) are not vulnerable to light damage [31
]. Therefore, we believe that at least one major reason for the observed kinetic and quantitative differences is the number of inflammatory cells in the outer retinas of BALB/c mice. Whether there are significant differences in endogenous CEP levels in the retinas of these mice, inherent differences in RPE function and/or local oxidative damage responses in the retina, or the particular contribution of specific adaptive immunity pathways, is an aspect under current investigation in our laboratory.
This work also describes in detail the differences in subretinal macrophages between these two widely used mouse strains. While many macrophage-like cells are present in the subretinal space of young naïve BALB/c mice, we have not been able to successfully identify these cells based on surface marker expression. The true nature of these baseline retinal macrophages in BALB/c mice remains unknown. Importantly, we only found subretinal CD11b+/F4/80+/CD68+ macrophages in CEP-MSA-immunized but not naive mice of either strain. A previous study has shown the presence of these macrophage-like cells in WT B6 mice but only after 20 months of age [32
]. Because CEP-MSA immunization leads to the presence of these macrophages in younger mice, we believe that this is additional confirmation of the validity of our model in accelerating an endogenous aging-related process. Thus, the CEP model provides an ideal setting to study different subpopulations of retinal macrophages.
The controversial role suggested for macrophages in AMD stems primarily from the use of gene knockout mice as well as an acute model for choroidal neovascularization (CNV) that has been widely (and successfully) used to mimic wet AMD. For instance, the assertion that macrophages are antiangiogenic comes mostly from studies using laser-induced CNV [33
], which is actually an acute wound healing response, not a chronic pathological state progressing from a previously established disease state, such as human wet AMD. A further complication involves the two different forms of AMD: macrophages could have different roles in dry versus wet AMD. It is important to stress that the laser-induced CNV model is a completely different system from our CEP model of dry AMD, and findings in one model will not necessarily be directly comparable to the other.
The initial evidence linking macrophages with AMD came from the analysis of mice deficient in macrophage chemokine signaling components (Ccl
mice) which show retinal defects similar to AMD with advanced age (2-year-old mice or older), including spontaneous CNV and “drusen” formation [29
]. However, subsequent work by Luhmann et al. (2009) [32
] revealed that these findings were in fact an artifact due subretinal macrophage accumulation and that any AMD-like pathology in Ccl
mice was most likely due to aging alone. An additional problem with the knockout mice mentioned previously and their use as AMD models is the fact that these strains were found to include a known mutation (rd8) that by itself results in retinal degeneration [34
]. Therefore most, if not all, the previously published papers using these strains must be reevaluated in that context.
However, there is still acceptable evidence associating macrophages with AMD. For example, young macrophages inhibit CNV in the laser-induced model of wet AMD, but their antiangiogenic potential is reduced with age as they switch to an M2 phenotype [33
]. More recently, it has been shown that microglia can induce RPE cells to produce proinflammatory cytokines and chemokines [36
]. However, information is lacking to clarify the pathological role of macrophages at different stages of the AMD disease process, particularly at the time of onset of dry AMD before the transition to CNV. The presence of subretinal CD11b+/F4/80+/CD68+ macrophages in CEP-MSA immunized mice we show here is similarly reported in a recent paper by a different group [37
]. In addition, we showed that these macrophages were M1 polarized. This suggests a strong causal link between the M1 macrophages and outer retinal lesions.
In the original publication of our model, it was suggested that macrophages were present as a result of tissue damage and were not likely to cause disease [22
]. The rationale for this conclusion was the fact that many lesions occurred in the absence of these cells. However, that original paper did not go into detail on the characterization of these cells. Missing from the first study and addressed in this paper are three key parameters that now lead to the interpretation that there is a causal relationship between M1 macrophages and dry AMD-like pathology: (i) kinetics and magnitude (quantification) of macrophage infiltration into the IPM relative to lesion development; (ii) activation status of the observed macrophages; (iii) how are the cells being recruited? This current study provides evidence for the first time that the early involvement of M1 macrophages occurs in animals that are predisposed to develop retinal lesions. We also provide the mechanism for recruitment of these cells, as Ccl2 is elevated in retinas of CEP-immunized mice, and its receptor, Ccr2, is required for macrophage infiltration into the IPM.
While we cannot completely rule out at this time that the M1 macrophages present in the IPM of CEP immunized mice are actually microglia migrating from the inner retina, it is likely that these cells come from the blood because of the systemic nature of our immunization protocol; retinal microglia are present at their normal inner retina location in Rag-deficient mice that do not develop lesions upon CEP-MSA immunization [22
]. Furthermore, this model relies on the endogenous accumulation of CEP adducts in the outer retina, which should occur at equivalent rates in immunized versus naïve mice, allowing resident microglia an equal access to the CEP antigen. A more definitive distinction of the original source of these cells awaits the development of microglia-specific and/or macrophage-specific markers. Regardless, our work confirms the critical role of bone-marrow-derived macrophages in the development of retinal degeneration and provides an excellent platform to further characterize this process.
As mentioned previously, we are aware that both Ccr
mice develop AMD-like pathology with age [29
], even though the recent work by Luhmann et al. [32
] has challenged this notion, at least for Ccl
mice. A major difference between these other studies and ours is that our model allows us to focus on the evaluation of relatively young animals following immunization with CEP-MSA, in contrast to the retinal lesions described previously that develop in the older knockout animals; we analyzed mice before 12 months of age, the naïve Ccr
mice develop retinal pathology after 18–20 months. Therefore, it would be difficult to make a direct comparison with our study, but it provides the opportunity to explore new mechanisms that link immunity to AMD. Ccr
mice do not lack ocular macrophages, just defective (or delayed) age-related recruitment (to the choroid). In fact, as shown by Luhmann et al. 2009 [32
], “old” Ccl
mice (which closely resemble the Ccr2 macrophage phenotype) have increased macrophage recruitment to the subretinal space (the same area in which we observe macrophage infiltration in CEP-MSA mice) when compared to wild type, showing that a defective Ccl2/Ccr2 axis does not necessarily, by itself, preclude retinal infiltration of macrophages. While there is certainly a possibility that the observed pathology in CEP-immunized WT mice may not be due to macrophages, we think the Ccr
data in this paper answers that question: if macrophages did not play a detrimental role in our model (if the retinal lesions in CEP-MSA mice were macrophage independent) then we should have observed some pathology in the immunized Ccr
mice, which we did not. Because the Ccr
mice still develop CEP antibodies similar to WT (indicative of an effective adaptive immune response), the M1 phenotype of subretinal macrophages as well as the temporal relationship between macrophage infiltration and retinal lesions (macrophage recruitment precedes lesion development), we believe that our interpretation that macrophages are detrimental in our model is justified.
It is tempting to hypothesize that there could be two different populations of macrophages involved in the AMD disease process: one being early “harmful” M1 and the other being the “protective” late M2, which in turn may contribute to CNV (once disease has progressed sufficiently). We believe that our data is representative of the role of early M1 macrophages and provides a nice platform to study early events in the development of AMD. This does not exclude the idea that later cellular involvement may include M2 macrophages that could be important for resolution of disease, suggested in the published studies looking at aged Ccr2/Ccl2 knockout mice [29
]. In fact, it was recently shown in a retinal neuropathy injury model that IL-10-producing (M2) macrophages have a protective role [38
]. The balance between M1 and M2 at different ages may actually dictate the damaged versus repaired tissue status of the retina. To support this notion, a recent paper analyzing human AMD eyes showed that AMD correlated with increased M1/M2 ratios, whereas normal aging eyes had more M2 macrophages [39
]. In the context of the retina, CEP tilts the balance toward the M1 pathway for its role in inflammation-induced GA.
The enhanced presence of proinflammatory macrophages in our model offers new opportunities to investigate their role and function in AMD pathogenesis, as well as the immunological signals and inflammatory agents behind their activation and recruitment to the outer retina, a tissue historically thought of as an immunosuppressive environment. We believe that innovative immunotherapies that target the low-grade inflammatory responses at the early stages of our model can yield further promising information on the immune mechanisms that take place in response to oxidative damage in the retina.