The Drusenlike Phenotype in Aging Ccl2-Knockout Mice Is Caused by an Accelerated Accumulation of Swollen Autofluorescent Subretinal Macrophages

Abstract

Purpose.: Drusen, which are defined clinically as yellowish white spots in the outer retina, are cardinal features of age-related macular degeneration (AMD). Ccl2-knockout (Ccl2 ^−/−) mice have been reported to develop drusen and phenotypic features similar to AMD, including an increased susceptibility to choroidal neovascularization (CNV). This study was conducted to investigate the nature of the drusenlike lesions in vivo and further evaluate the Ccl2 ^−/− mouse as a model of AMD.

Methods.: The eyes of 2- to 25-month-old Ccl2 ^−/− and C57Bl/6 mice were examined in vivo by autofluorescence scanning laser ophthalmoscopy (AF-SLO) and electroretinography, and the extent of laser-induced CNV was measured by fluorescein fundus angiography. The retinal morphology was also assessed by immunohistochemistry and quantitative histologic and ultrastructural morphometry.

Results.: The drusenlike lesions of Ccl2 ^−/− mice comprised accelerated accumulation of swollen CD68⁺, F4/80⁺ macrophages in the subretinal space that were apparent as autofluorescent foci on AF-SLO. These macrophages contained pigment granules and phagosomes with outer segment and lipofuscin inclusions that may account for their autofluorescence. Only age-related retinal pigment epithelium (RPE) damage, photoreceptor loss, and sub-RPE deposits were observed but, despite the accelerated accumulation of macrophages, we identified no spontaneous development of CNV in the senescent mice and found a reduced susceptibility to laser-induced CNV in the Ccl2 ^−/− mice.

Conclusions.: These findings suggest that the lack of Ccl2 leads to a monocyte/macrophage-trafficking defect during aging and to an impaired recruitment of these cells to sites of laser injury. Other, previously described features of Ccl2 ^−/− mice that are similar to AMD may be the result of aging alone.

Age-related macular degeneration (AMD) is the commonest cause of vision loss in the elderly population in industrialized countries. ¹ A clinical hallmark of AMD is the appearance of drusen, which are opaque, yellowish white spots visible in the fundus. ² Histologically, drusen are extracellular deposits between the basal lamina of the retinal pigment epithelium (RPE) and inner collagenous layer of Bruch's membrane (BM). Drusen have a complex protein/lipid composition, that includes immunoglobulins, activated complement components, and complement regulators, ^3,4 as well as lipids, intracellular proteins, and cytosolic stress response proteins. ⁵  

Targeted disruption of the gene-encoding monocyte chemoattractant protein 1 (MCP-1), also known as CC-cytokine ligand 2 (Ccl2), or its receptor CC-cytokine receptor 2 (Ccr2) in mice leads to defects in monocyte recruitment to sites of inflammation. ^6–8 Recently, it has been reported that both Ccl2 and Ccr2-knockout mice develop drusen and other features of AMD, such as the accumulation of lipofuscin in RPE cells, progressive outer retinal degeneration, and geographic/RPE atrophy. A higher incidence of spontaneous development of choroidal neovascularization (CNV) was also reported in a proportion of senescent mice (4/15 mice aged >18 months). ⁹  

It has therefore been suggested, that chemokines and their receptors, including CCL2 and CCR2, may be involved in the etiology of AMD. Although genetic data indicate that polymorphisms in the cytokine receptor CX3CR1 are associated with an increased risk of developing AMD, ^10,11 an increased risk for AMD has not been associated with CCL2 or CCR2. ¹² Recently, it has also been shown that CX3CR1-positive microglia can be found ectopically in the outer retina and in the subretinal space of patients with AMD in close proximity to drusen and to CNV. The CX3CR1 T280M-polymorphism leads to an impaired chemotactic response of monocytes to CCL2 in the presence of bound CX3CL1 in vitro. ¹¹ Furthermore, with age, Cx3cr1-deficient mice show a progressive accumulation of microglia cells in the subretinal space. This accumulation leads to a drusenlike appearance in Cx3Cr1-knockout mice and is accompanied by retinal degeneration, suggesting that CX3CR1 signaling in microglia may play a role in the development of AMD. ^11,13  

The absence to date of a genetic association between CCL2 or CCR2 and AMD therefore prompted us to characterize the drusen development in Ccl2-knockout mice in more detail and to reevaluate them as a model for AMD. 

Material and Methods

Animals

We used Ccl2-knockout (Ccl2 ^−/−) mice on a C57Bl/6 background kept as a homozygous line (a kind gift from Kath Else at the University of Manchester from the original stock maintained by Barrett Rollins). ⁶ Age-matched control animals (C57Bl/6 6JOla Hsd; Harlan UK Ltd., Blackthorn, UK) were kept in the same animal room and 12-hour light–dark cycles. For in vivo procedures, the mice were anesthetized by a single intraperitoneal (IP) injection of a mixture of medetomidine hydrochloride (1 mg/ kg body weight; Domitor; Pfizer Animal Health, New York, NY), and ketamine (60 mg/kg body weight) in water. Whenever necessary, the pupils were dilated with 1 drop of 1% tropicamide. The animal experiments were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Autofluorescence Scanning Laser Ophthalmoscopy (AF-SLO)

Autofluorescence imaging was performed using a scanning laser ophthalmoscope (HRA2; Heidelberg Engineering, Heidelberg, Germany) as previously described. ^14–16 With a 55° angle lens, projection images of 30 frames per fundus were taken after positioning the optic disc in the center of the image and focusing on the outer retina. The number of autofluorescent spots per fundus image was determined for each eye and the mean number of autofluorescent spots per fundus image ± SD of the left and the right eye for each animal was calculated. 

Histopathology

Paraffin-embedded histology was performed as described before. ¹⁷ Briefly, eyes were enucleated and fixed for at least 18 hours in Serras fixative and were embedded in paraffin after dehydration in 70% and 100% isopropanol. Sections were cut at 5 μm and stained with hematoxylin and eosin (H&E). Images were obtained from the central and peripheral regions of sections containing the optic nerve. Rows of photoreceptors were counted in three columns per image from at least two images per animal, and the mean number of photoreceptor rows ±SD per section was calculated for the central and the peripheral retina of each animal. 

For ultrastructural analysis, eyes were enucleated and fixed in 3% glutaraldehyde and 1% paraformaldehyde in 0.08 M sodium cacodylate-HCl (pH 7.4) for at least 30 hours at 4°C. The cornea and lens were removed and the eye cups oriented and postfixed in 1% aqueous osmium tetroxide for 2 hours, dehydrated by passage through ascending ethanol series (50%–100%) and propylene oxide, and infiltrated overnight with a 1:1 mixture of propylene oxide. After a further 8 hours in full resin, the eyes were embedded in fresh resin and incubated overnight at 60°C. Semithin (0.7 μm) and ultrathin (70 nm) sections were cut in the inferior–superior axis passing through the optic nerve head with a microtome (Ultracut S; Leica, Wetzlar, Germany). Semithin sections were stained with a 1% mixture of toluidine blue-borax in 50% ethanol, and ultrathin sections sequentially contrasted with saturated ethanolic uranyl acetate and lead citrate for imaging in a transmission electron microscope (TEM model 1010; JEOL, Tokyo Japan) operating at 80 kV. Images were captured with a CCD camera calibrated against a 2160-lines/mm grating (Gatan Orius; Agar Scientific, Stansted, UK) in digital micrograph format and exported as .tiff files into Image J for quantification (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). The thickness of BM, defined from the basal lamina of the RPE cell to the basal lamina of the choroidal vessel, was measured in a uniform, randomized way to prevent a potential bias in the observer's choice. The average BM thickness in each eye was obtained from single images taken at a standard magnification of 5000 positioned approximately 1 mm on each side of the optic nerve. After the calibration of Image J, BM was rotated to the horizontal or vertical, and a grid was superimposed on the image. Ten measurements were taken at points where grid lines passed across BM to yield an average value. Finally, the average measurements for superior and inferior were added, and a mean value ± SD was obtained for BM thickness in each animal. 

Immunohistochemistry

The eyes were enucleated, and the cornea and lens were removed during fixation in PBS/4% paraformaldehyde for 1 hour. The specimens were cryoprotected with PBS/20% sucrose overnight at 4°C before the eyes were embedded in OCT compound (Tissue Tek; Sakura Finetek, Thatcham, UK). Twelve-micrometer-thick retinal sagittal sections were air dried for at least 1 hour before rehydration of the sections with PBS. For the CD68 immunohistochemistry, nonspecific binding sites on the sections were blocked for 1 hour with PBS/1%BSA (Sigma Aldrich, Steinheim, Germany)/5% nonspecific goat serum (AbD Serotec, Kidlington, UK) including 0.3% Triton X-100 for permeabilization before being incubated with a 1:200 dilution (0.5 μg/mL) of rat anti-mouse CD68:biotin antibody (MCA1957B, clone FA-11; AbD Serotec) in blocking solution overnight at 4°C. After four to five washes (5–10 minutes each) in PBS at room temperature (RT) the sections were incubated with 1:500 AlexaFluor 546 nm–conjugated streptavidin (#S11225; Invitrogen-Molecular Probes, Leiden, The Netherlands) for 2 hours at RT, washed again four to five times in PBS, and mounted with fluorescence mounting medium containing Hoechst 33342 (Dako, Cambridgeshire, UK). Since F4/80 staining was generally very weak, we used blocking and amplification reagents from the catalyzed signal amplification (CSA) system (K1500; Dako) according to the manufacturer's instruction, together with the primary rat anti-mouse F4/80:biotin antibody (1:500, 0.2 μg/mL; MCA497B, clone CI:A3–1; AbD Serotec) and the AlexaFluor 546 nm–conjugated streptavidin (1:500, S11225; Invitrogen-Molecular) as the final detecting antibody. Eyes for RPE/choroidal flatmounts were fixed for 1 hour in PBS/4% paraformaldehyde, dissected, and directly mounted on glass slides before images were obtained with a laser scanning microscope (LSM 510; Carl Zeiss Microimaging, Jena, Germany). 

Electroretinography and Choroidal Neovascularization Analysis

Standard scotopic Ganzfeld ERGs were recorded from dark-adapted mice (16 hours) from both eyes simultaneously (Espion E²; Diagnosys LLC, Cambridge, UK). All procedures for recording were performed under dim red light. Single-flash recordings were obtained at increasing light intensities from 0.001 to 10 cd/m². Ten responses per intensity level were averaged with an interstimulus interval of 30 (0.001–1 cd/m²) and 60 (3, 5, and 10 cd/m²) seconds. A- and b-wave amplitudes were analyzed (Espion software; Diagnosys LLC). 

Laser-induced CNV and quantification with fundus fluorescein angiography (FFA) were performed as described before. ¹⁸ Three laser lesions per eye were delivered two to three disc diameters from the papillae with a slit lamp–mounted diode laser system (wavelength 680 nm; laser settings: 210-mW power, 100-ms duration, and 100-μm spot diameter; Keeler, Ltd., Windsor, UK). After 2 and 5 weeks, in vivo FFA was performed and images from the early (90 seconds after fluorescein injection) and late (7 minutes) phases were obtained (Kowa Genesis, Tokyo, Japan) small animal fundus camera with appropriate filters. The pixel area of CNV-associated hyperfluorescence was quantified for each lesion using image-analysis software (Image Pro Plus; Media Cybernetics, Silver Spring, MD) and the proportionate increase between the early- and late-phase was calculated providing us with a measure for the degree of vascular permeability. 

Results

Accelerated Accumulation of Autofluorescent Foci in AF-SLO Fundus Images

Senescent Ccl2 ^−/− mice were evaluated by normal funduscopy, which confirmed the drusenlike phenotype previously described. ⁹ To characterize further the ocular phenotype during aging and learn more about potential pathologic changes and the nature of the drusenlike structures in vivo, we examined Ccl2 ^−/− mice (n = 11) and aged-matched C57Bl/6 wild-type mice (n = 9) with AF-SLO at different ages (2–3, 5–7, 16–18, and 20–25 months). This confocal imaging technique enables us to detect changes in the normal distribution of fluorophores (e.g., lipofuscin) inside the retina and the RPE, which are often associated with pathologic processes and provides some localization information, because it does not detect sub-RPE changes in pigmented animals, because the 488-nm laser light gets absorbed by the pigment. ¹⁴  

In wild-type C57Bl/6 mice, we observed a consistent age-related increase in the number of autofluorescent spots inside the retina up to 2 years (Fig. 1A, top). In Ccl2 ^−/− mice, the number of such autofluorescent spots also increased steadily with age but was consistently higher than in age-matched control mice (Fig. 1A, lower row). Quantitative analysis for each genotype confirmed a significant positive correlation of age and number of autofluorescent spots per fundus image for both genotypes (Pearson correlation C57Bl/6: r = 0.813, P = 0.002, n = 11; Ccl2 ^−/−: r = 0.806, P = 0.009, n = 9) and revealed a significantly higher number of autofluorescent spots in 16- to 25-month-old Ccl2 ^−/− mice compared with young Ccl2 ^−/− (2–7 months), young C57Bl/6 (2–7 months), and old C57Bl/6 (16–25 months) mice (P < 0.01, one-way ANOVA with Bonferroni post hoc test). This result indicates that the accumulation of autofluorescent spots inside the retina is a normal age-related process that is accelerated in Ccl2 ^−/− mice. 

Figure 1.

 

Accelerated accumulation of autofluorescent spots in Ccl2 ^−/− mice with age. AF-SLO projection images of 30 frames taken with a 55° angle lens. (A) Representative fundus images obtained by AF-SLO imaging from both genotypes at 2 to 3, 5 to 7, 16 to 18, and 20 to 25 months. Whereas in both C57Bl/6 and Ccl2 ^−/− mice the number of autofluorescent spots were increased with age, the process was accelerated and more pronounced in the Ccl2 ^−/− mice. (B) Quantification of autofluorescent spots in C57Bl/6 (■) and Ccl2 ^−/− mice (♢) per AF-SLO images. The number of autofluorescent spots per AF-SLO image is shown as the mean number of spots ± SD in the right and the left eyes of each animal. For both genotypes, the mean number of spots correlated significantly with age, indicating an age-related increase in autofluorescent spots as a normal process (Pearson correlation C57Bl/6: r = 0.813, P = 0.002, n = 11; Ccl2 ^−/−: r = 0.806, P = 0.009, n = 9). One-way ANOVA with Bonferroni post hoc test revealed a significantly higher number of autofluorescent spots in 16- to 25-month-old Ccl2 ^−/− mice compared with aged-matched wild-type C57Bl/6 mice and compared with the young (2–7 months) groups of both genotypes (P < 0.01).

Similar Age-Related Loss of Photoreceptors in Wild-Type C57Bl/6 and Ccl2-Knockout Mice

Histologically the retinas of Ccl2 ^−/− mice and aged-matched wild-type control animals appeared similar in both the central and peripheral regions (Fig. 2), except for occasional large macrophage-like cells in the Ccl2 ^−/− mice. We observed a similar age-related loss of outer nuclear layer (ONL) cells in the central and peripheral retina of both genotypes (Figs. 2A, 2B), which was significantly different between young and old animals of both genotypes, but not between aged-matched wild-type and Ccl2 ^−/− mice (Figs. 2C, 2D). Similarly, we observed higher amplitude ERG recordings in young compared with 10- to 12-month-old mice in both strains (one-way ANOVA with Bonferroni post hoc test; Fig. 3; P < 0.05), but no difference between aged-matched wild-type and Ccl2 ^−/− mice (Fig. 3) indicating that the lack of Ccl2 ^−/− does not enhance the normal age-related loss of retinal function. 

Figure 2.

 

Similar age-related photoreceptor loss in wild-type (WT) and Ccl2-knockout (Ccl2 ^−/−) mice. Comparison of central (A) and peripheral (B) retinal histology of hematoxylin-eosin–stained sagittal sections (5 μm) of WT (C57Bl/6) and Ccl2 ^−/− mice at the ages shown. No obvious histologic differences between age-matched WT and Ccl2 ^−/− mice were found, except the observation of some subretinal macrophage-like cells in old Ccl2 ^−/− mice (arrow). When young versus old WT and young versus old Ccl2 ^−/− mice were compared, thinning of the ONL with age was obvious in both genotypes. RGC, retinal ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. Original magnification, ×400. (C, D) Morphometric analysis of the mean number of photoreceptor rows per section in the central (C) and peripheral (D) retina in all animal groups: young WT (n = 5), young Ccl2 ^−/− (n = 5), old WT (n = 5), and old Ccl2 ^−/− (n = 6) mice. A significant reduction in rows of photoreceptors with age was found in the central and the peripheral retina in both genotypes, but no significant difference was found between aged-matched C57Bl/6 control and Ccl2 ^−/− animals. Kruskal-Wallis test (P = 0.004) with subsequent pair-wise comparison of all groups using the Mann-Whitney U test. *Statistically significant based on the Mann-Whitney U test (P < 0.05).

Figure 3.

 

Similar age-related reduction of ERG amplitude in both genotypes. Mean a-wave (A) and b-wave (B) amplitudes ± SD per group of animals: C57Bl/6, 2–3 months, n = 5 (■); Ccl2 ^−/−, 2–3 months, n = 6 (□); C57Bl/6, 10–12 months, n = 6 (▴); Ccl2 ^−/−, 10–12 months, n = 6 (▵). Amplitudes were obtained from dark-adapted mice by using the standard scotopic Ganzfeld ERG at different light intensities (0.001–10 cd/m²). Note the half logarithmic scale for light intensities. Solid lines: C57Bl/6; dashed lines: Ccl2 ^−/−. *Significant difference in amplitude compared with young C57Bl/6 mice (P < 0.05, one-way ANOVA with Bonferroni post hoc test). Note that there are no significant differences between the age-matched groups.

Increased Number of SubretinalMacrophage-like Cells

To determine the number of macrophage-like cells in the subretinal space and to evaluate RPE damage in a quantitative manner, we examined toluidine blue-borax–stained semithin histologic sections from young (2–3 months) and old (20–24 months) Ccl2 ^−/− and C57Bl/6 wild-type mice (Figs. 4A–E). This confirmed that significantly more macrophages were present in the subretinal space of 20- to 24-month old Ccl2 ^−/− mice (n = 10) compared with aged-matched control animals (n = 7; Figs. 4C, 4D vs. 4E). In young (2–3 months) mice of both genotypes (MCP-1: n = 5, C57Bl/6: n = 5; Figs. 4A, 4B) these cells were not observed (one-way ANOVA with Bonferroni post hoc test, P < 0.01). Furthermore, for both genotypes a significant and positive correlation between age and the number of macrophage-like cells in the subretinal space was found (Fig. 4F; Ccl2 ^−/− Pearson correlation r = 0.748, P < 0.01, n = 15 and C57Bl/6 mice Pearson correlation r = 0.890, P < 0.01, n = 12). Of note, the macrophages in aged MCP-1 mice were bigger and more swollen than those in old C57Bl/6 mice (Fig. 4C vs. 4G; arrows). Between 30% and 50% of these subretinal macrophages in both genotypes showed a prominent accumulation of pigment granules suggesting an uptake of RPE material and thus potentially a diseased RPE. To evaluate the pathologic effect quantitatively, we counted cellular events such as cell lysis, pyknosis, swelling, thinning, and thickening of RPE cells from the superior to the inferior end of the retina in each section and calculated the mean sum of RPE alterations per section from three sections per animal, which represents a measure for RPE damage according to a scheme by Hollyfield et al. ¹⁹ We compared all four groups (2–3-month-old C57Bl/6 [n = 5], 2–3-month-old Ccl2 ^−/− [n = 5], 20–24-month-old C57Bl/6 [n = 7], and 20–24-month-old Ccl2 ^−/− [n = 10] mice) and found a significant increase in RPE damage per section with age, but no significant difference between Ccl2 ^−/− mice and aged-matched control animals (one-way ANOVA with Bonferroni post hoc test; P < 0.01). Furthermore, we analyzed the correlation between age and RPE damage per section, combining both genotypes, and found a significant, positive correlation between age and RPE changes (Pearson correlation r = 0.585, P = 0.001, n = 27). 

Figure 4.

 

Ccl2-knockout (Ccl2 ^−/−) mice did not develop abnormal RPE changes with age, but showed an accumulation of bloated macrophages in the subretinal space. Toluidine blue-borax–stained semithin histologic sections of the outer retina, RPE, and choroid from young (2–3 months) and old (20–24 months) Ccl2 ^−/− and C57Bl/6 wild-type mice (A–E). No obvious histologic differences between young C57Bl/6 wild-type (A) and Ccl2 ^−/− (B) mice were observed. In old C57Bl/6 (C, D) and in old Ccl2 ^−/− (E) mice, photoreceptor morphology was also normal, but macrophage-like cells were observed in the subretinal space. These cells were bigger and swollen in Ccl2 ^−/− mice (C, D versus E, white arrows). Swelling/lysis of RPE cells and pyknosis (indicated by the darker staining) of RPE cells were observed in senescent animals of both genotypes, but are only shown as examples in C57Bl/6 wild-type mice (D, arrowheads). Note the higher pigmentation of the choroid in old C57Bl/6 and Ccl2 ^−/− mice compared with young mice (C, D, E versus A, B). Scale bar, 100 μm. Quantitative analysis of the number of macrophage-like cells in the subretinal space (F). The mean number of macrophages per individual animal is depicted [C57Bl/6, (■); Ccl2 ^−/− mice (♢)] and shows a significant positive correlation between age and number of macrophages in the subretinal space per section, which was much more pronounced in Ccl2 ^−/− mice. (G) Quantitative analysis of alterations in the RPE. The RPE damage count (sum of RPE areas showing cell lysis, pyknosis, swelling, thinning, or thickening of RPE cells per retinal section) for individual animals is shown (Ccl2 ^−/−, solid line; C57Bl/6, dashed line), revealing a similar accumulation of RPE alterations with age in both genotypes and an overall positive correlation of this accumulation with age (Pearson correlation r = 0.585, P = 0.001, n = 27).

Thus, although the accumulation of macrophages in the subretinal space is accelerated and more pronounced in Ccl2 ^−/− mice, age-related RPE changes occur to a similar extent in both Ccl2 ^−/− and wild-type C57Bl/6 mice. 

To confirm that the cells in the subretinal space are of the monocyte/macrophage lineage, we used CD68, a marker for activated and phagocytically active macrophages, and the pan macrophage marker F4/80 and found that the subretinal cells were positive for both of these markers (Figs. 5A–E). ²⁰ We also analyzed the autofluorescence in these cells at 488 nm in sections (red [CD68 stained]) versus green [488-nm autofluorescence]; Fig. 5B) and on RPE flatmounts (546-nm [red] and 488 nm [green]; Figs. 5F, 5G) and found that these cells contained autofluorescent material. We measured a mean diameter of macrophages on the apical side of the RPE of 19.83 ± 6.73 μm and (using the optic disc as a reference) 17.99 ± 4.90 μm for autofluorescent spots from AF-SLO images (Table 1). Thus, the regular pattern, size, and location of the cells indicate that the autofluorescent spots in the AF-SLO fundus images in senescent Ccl2 ^−/− mice were subretinal macrophages containing autofluorescent material. 

Figure 5.

 

CD68/ED1 and F4/80 immunohistochemistry staining showed that cells in the subretinal space were macrophages that were phagocytically active and contained autofluorescent material. (A–E) Comparison of H&E histology (A, C) and CD68/ED1 (B, D: CD68, AlexaFluor 546 nm; red) and F4/80 (E: F4/80, AlexaFluor 546 nm; red) immunohistochemistry of sagittal retinal sections from 20- to 22-month-old Ccl2 ^−/− mice. (A) H&E sections of 20- to 22-month-old Ccl2 ^−/− mice showed oval cells in the subretinal space on the apical side of the RPE that were relatively regularly spaced. (B) CD68 immunohistochemistry confirmed that these cells are activated (phagocytically active) macrophages, which had already been suggested by their shape. The macrophages contained autofluorescent material: compare the green autofluorescent channel (right, 488 nm) with the CD68-positive signal (center, red channel, 546 nm) and the yellow staining in the overlay (left). (C) Higher magnification of (A). (D) View on the subretinal space from a second old MCP-1 mouse. CD68-positive macrophages are regularly spaced on top of the RPE. (E) Immunohistochemsitry for the pan macrophage marker F4/80 further confirms that the observed subretinal cells are macrophages/monocyte lineage. Note the same location of F4/80- and CD68-positive cells in the subretinal space. Counterstaining in (B) (D), and (E) with Hoechst 33342 (blue). (F, G) Retinal pigment epithelium flatmount from an old (22 months) Ccl2 ^−/− mouse counterstained with Hoechst 33342 (blue). Autofluorescence was detected in both the red (546 nm) and the green (488 nm) channel. (F) Localization of macrophages on the apical site of the RPE flatmount indicated a regular pattern and spacing of these cells in the subretinal space. Fluorescence derives from inside the cells. Cho, choroidal vasculature; Scl, sclera. Scale bar: (A, B, D, E, F) 50 μm; (C, G) 10 μm.

Table 1.

 

Size Comparison of Macrophages in RPE Flatmounts and Autofluorescent Spots in AF-SLO Fundus Images of Old Ccl2 ^−/− Mice

Macrophage Diameter (μm)	Calculated Spot Size (μm)
30.60	17.99
27.79	23.99
21.33	17.99
18.19	23.99
11.73	11.99
12.10	11.99
17.07
19.83 ± 6.73	17.99 ± 4.90

 We used the size of the optic disc in confocal and AF-SLO images to calibrate the AF-SLO fundus images and thus determined the mean size ± SD of the autofluorescent spots to be 17.99 ± 4.90 μm. This shows that these autofluorescent spots have a similar size as macrophages on the apical site of RPE of senescent Ccl2 ^−/− mice, which have a mean diameter of 19.83 ± 6.73 μm based on confocal imaging.

To evaluate the subretinal macrophages and the nature of the autofluorescent material in more detail, we performed ultrastructural analysis with TEM (Fig. 6). Bloated subretinal macrophages were seen only in the senescent Ccl2 ^−/− mice (Fig. 6B) and were located between the apical surface of the RPE and the photoreceptor outer segments. They contained numerous pigment granules (Fig. 6C), phagosomes, and lipofuscin inclusions (Figs. 6C, 6D, arrows), but only rarely showed phagocytosis of outer segment material (Fig. 6E, arrowhead). Although the appearance of these cells was usually suggestive of macrophages (Fig. 6C), some seemed to have accumulated a lot of cellular debris and showed a high degree of vacuolization (Fig. 6F, showing an extreme example, see also Fig. 4E). 

Figure 6.

 

Subretinal macrophages (SrMC) in old Ccl2 ^−/− mice contained pigment granules, phagosomes, lipofuscin, and outer segment (OS) material. TEM images obtained from young (2–3 months) and old (20–24 months) C57Bl/6 and Ccl2 ^−/− mice. (A) Representative image for young mice of both genotypes. (B) TEM image from an old Ccl2 ^−/− mouse showing a typical bloated macrophage in the subretinal space and irregular infoldings on the basal side of the RPE. (C) Higher magnification of a bloated subretinal macrophage containing pigment granules and phagosomes with lipofuscin (arrows). (D, E) Content of bloated subretinal macrophages (SrM). (D) Phagosomes with lipofuscin inclusions (arrows) were very common. (E) Occasionally outer segment material was observed inside SrMs (arrowhead). (F) Extreme example of a vacuolized cell in the subretinal space, presumably an SrM. Note the healthy underlying RPE. OS, outer segments; RPE, retinal pigment epithelium; BM, Bruch's membrane; Ch, choroidal vasculature. Scale bar: (A, B, F) 5 μm; (C) 1 μm; (D, E) 0.5 μm.

Normal Age-Related Changes in BM

We also evaluated the integrity of the RPE in young (2–3 months) and old (20–24 months) wild-type C57Bl/6 and Ccl2 ^−/− mice at an ultrastructural level and focused in particular on age-related changes at the basal side of the cells and the underlying BM (Fig. 7). We observed normal outer segment phagocytosis by the RPE in senescent mice of both genotypes, indicating that the RPE was functional and to a certain extent healthy. Nevertheless, with age we found dramatic changes on the basal side of the RPE cells, which were very similar between genotypes (Figs. 7A–H). In young animals, normal basal infoldings in RPE cells (Figs. 7A, 7B; higher magnification, 7E, 7F) and a normal BM with a thickness of approximately 0.34 μm was observed (Table 2). In contrast, aged animals of both genotypes showed wide areas of large basal invaginations filled with amorphous sub-RPE deposits (Figs. 7C, 7D and 7G, 7H). There were no obvious differences in these deposits between wild-type and MCP-1 mice, and no drusenlike structures were identified in either of the two genotypes on the basal side of the RPE. Mean BM thickness for senescent C57Bl/6 mice was 0.71 ± 0.26 μm and for senescent Ccl2 ^−/− mice, it was 0.87 ± 0.25 μm (Table 2). These values are both approximately twice those from young animals, indicating that the major cause of BM thickening in both genotypes is aging. This conclusion was further supported by the fact that we did not identify significant differences between aged-matched wild-type and Ccl2 ^−/− mice (Table 2, one-way ANOVA with Bonferroni post hoc test P < 0.01). Furthermore, we identified a significant correlation between age and BM thickness in each genotype independently (Fig. 7I; C57Bl/6: Pearson correlation r = 0.753, P < 0.01, n = 12; Ccl2 ^−/−: Pearson correlation r = 0.832, P < 0.01, n = 15) further supporting the conclusion that normal aging and not the Ccl2 genotype is the major cause of the increase in sub-RPE deposits and BM thickness. 

Figure 7.

 

Age-related ultrastructural changes in the RPE and BM in Ccl2 ^{− / −} and wild-type mice. (A–H) TEM images of RPE, BM, and choroidal interface at two different magnifications, (A, B, E, F) Young mice: normal basal infoldings of RPE cells and normal appearance of BM in both genotypes. (C, D, G, H) Senescent mice: In both genotypes areas of irregular and big basal invaginations in RPE cells with amorphous sub-RPE deposits were observed. Scale bars: (A–D): 2 μm; (E–H): 1 μm. (I) Quantitative assessment of BM thickness with age for C57Bl/6 (■) and Ccl2 ^−/− (♢) mice. BM thickness is shown for individual animals as a mean ± SD of measures taken in the superior and inferior hemisphere. R ² values and the respective equations for the linear regression between age and BM thickness are shown for each genotype. There was a significant correlation between age and BM thickness in both genotypes. C57Bl/6 (Pearson correlation r = 0.753, P < 0.01, n = 12), Ccl2 ^−/− (Pearson correlation r = 0.832, P < 0.01, n = 15), but no significant difference between aged-matched Ccl2 ^−/− versus C57Bl/6 mice (one-way ANOVA with Bonferroni post hoc test; for details see Table 2). Ch, choroidal vasculature.

Table 2.

 

Quantitative Ultrastructural Analysis of BM Thickness

	2–3 Month C57B1/6 (1)	2–3 Month Ccl2 −/− (2)	20–24 Month C57B1/6 (3)	20–24 Month Ccl2 −/− (4)
BM thickness ± SD (μm)	0.34 ± 0.07	0.34 ± 0.05	0.71 ± 0.26	0.87 ± 0.25
n (animal)	5	5	7	9
One-way ANOVA with	1 vs. 2: P = 1.0	2 vs. 1: P = 1.0	3 vs. 1: P = 0.032	4 vs. 1: P = 0.003
Bonferroni post hoc test	1 vs. 3: P = 0.032	2 vs. 3: P = 0.029	3 vs. 2: P = 0.029	4 vs. 2: P = 0.031
(overall P<0.001)	1 vs. 4: P = 0.001	2 vs. 4: P = 0.001	3 vs. 4: P = 0.819	4 vs. 3: P = 0.819

 Mean BM thickness ± SD per group of animals according to genotype and age. Data for the 14-month-old Ccl2 ^−/− animal as shown in Figure 7 are excluded.

Reduced Laser–Induced CNV Lesion Size in Ccl2 ^−/− Mice

In a previous study, a higher susceptibility to spontaneous CNV development has been described in Ccl2 ^−/− mice. ⁹ Four of 15 Ccl2 ^−/− mice older than 18 months have been reported to develop CNV with clear evidence of vascular leakage and subsequent morphologic alterations at the ultrastructural level. In this study, however, we did not observe indications for a spontaneous CNV development in 11 senescent Ccl2 ^−/− mice between 16 and 25 months of age. We used fundus fluorescein angiography (n = 5, 16 months) and lectin-stained RPE and retinal flatmounts (n = 6, 23–25 months) to specifically study CNV development and did not observe any signs of CNV (data not shown). Furthermore, our morphologic analyses at the histologic and ultrastructural levels also did not reveal any CNV-associated morphologic alterations in a further 15 senescent Ccl2 ^−/− mice age 20 to 25 months (Figs. 2, 4, 6, and 7). To evaluate the susceptibility of Ccl2 ^−/− mice to CNV, we therefore analyzed the neovascular response after laser photocoagulation by using fundus fluorescein angiography. We observed a 45% to 64% reduction in the area of hyperfluorescence in Ccl2 ^−/− mice in both the early and the late phases of the angiography at 2 and 5 weeks after lasering (Fig. 8), which indicated that the lack of the CC-cytokine ligand 2/MCP-1 led to a reduced choroidal neovascular response after laser injury. 

Figure 8.

 

In vivo fundus fluorescein angiography 2 and 5 weeks after photocoagulation/laser induced choroidal neovascularization (CNV) revealed reduced vessel leakiness in CNV lesions in eyes of Ccl2 ^−/− mice. Representative images of early (90 seconds)- and late (7 minutes)-phase fundus fluorescein angiography (A, B). The images show CNV lesions from C57Bl/6 wild-type and Ccl2 ^−/− mice at 2 (A) or 5 (B) weeks after lasering. In Ccl2 ^−/− mice, the area of hyperfluorescence was smaller compared to that in wild-type animals. Thus, a reduced neovascular response in Ccl2 ^−/− mice was indicated. Quantitative analysis of CNV lesion size in vivo at 2 (C) and 5 (D) weeks after lasering confirmed the qualitative result obtained from the angiograms (A, B). *Significant reduction in CNV lesion size in Ccl2 ^−/− mice compared with that in C57Bl/6 mice (two-tailed t-test *P < 0.05; **P < 0.01). Hyperfluorescent area is given in arbitrary units.

Discussion

In this study we demonstrated that the drusenlike phenotype in Ccl2-knockout (Ccl2 ^−/−) mice is not caused by sub-RPE deposits, but is due to the accelerated age-related accumulation of bloated macrophages in the subretinal space. These cells contain autofluorescent material, in particular lipofuscin, and they almost certainly account for the autofluorescent spots in AF-SLO images. We also observed a similar, but not as pronounced accumulation of macrophages during aging in normal mice and thereby confirmed recent findings by Xu et al., ²¹ who showed an age-dependent accumulation of subretinal microglia. Together with our findings, these results indicate that the accumulation of macrophages/microglia in the subretinal space is a normal age-related process that is accelerated in Ccl2 ^−/− mice. Therefore, MCP-1/CCL2–CCR2 signaling is not only essential in the recruitment of monocytes to sites of inflammation, ⁶ but also plays a role in the trafficking of microglia/macrophages during aging. In the absence of MCP-1, macrophages seem to be less capable of leaving the subretinal space after uptake of cellular debris, and continue to degrade it to autofluorescent material, including lipofuscin, and finally vacuolize. A similar process has been reported in the development of foam cells in atherosclerotic plaque. ²²  

Further support for the role of cytokines in maintaining macrophage migration and mobility during aging comes from recent findings in another cytokine receptor mouse model, the Cx3cr1-knockout, which shows a very similar accelerated accumulation of subretinal microglia/macrophages. ¹¹ Lipid-bloated microglia are the origin of the drusenlike appearance in Cx3cr1-deficient mice. ¹³ It has been suggested that these cells can leave the subretinal space via two routes, the choroidal or the inner retinal circulation. CD68-positive microglia/macrophages double labeled with rhodopsin antibodies have been observed in the inner retina and in the choroid, suggesting that microglia may invade the subretinal space, clear up cellular debris such as outer segment material and then leave the subretinal space either via the inner retinal vasculature or the choroid. ²³ Thus, CCL2 or CX3CR1 deficiencies both seem to lead to a similar accumulation of cells from the monocyte/microglia/macrophage lineage in the subretinal space with age causing a macroscopically similar drusenlike phenotype. However, there seem to be distinct differences between these cytokine pathways. Whereas in bloated subretinal microglia of the Cx3cr1-knockout model, multiple lipid droplets were observed, ¹³ this distinct ultrastructural feature was not prominent in bloated subretinal cells of Ccl2 ^−/− mice. We observed lipofuscin accumulation and in 30% to 50% of the cells, pigment granules as the most prominent intracellular components. This raises the possibility that different subtypes of F4/80- and CD68-positive phagocytically active cells are accumulating in the two knockout models. Further support for the different roles of the MCP-1/CCR2 signaling and the CX3CR1-mediated signaling in the eye comes from the different responses to laser-induced choroidal neovascularization in the respective knockout models. In this study we observed a reduction of CNV lesion size and thus a reduced angiogenic response in Ccl2 ^−/− mice at 2 and 5 weeks after photocoagulation. Furthermore, we did not observe any spontaneous development of CNV in senescent Ccl2 ^−/− mice. These observations were in contrast to the higher susceptibility of aged Ccl2 ^−/− mice to spontaneous CNV, ⁹ but were consistent with previous observations in the Ccr2-knockout mouse line, which also showed a reduced response to laser-induced CNV, suggesting that CCL2 acts via CCR2 during laser-induced CNV and promotes neovascular responses. ²⁴ In contrast to these findings, Cx3cr1-knockout mice showed an exacerbated response to laser-induced CNV 2 weeks after photocoagulation, highlighting the different roles of the two monocyte recruiting signaling pathways during the CNV response. ¹¹ CCL2 actively recruits a subset of proinflammatory blood monocytes (CX3CR1^low, CCR2-positive, and Gr1-positive) to sites of inflammation and appears to be important for the angiogenic response during laser-induced CNV. CX3CR1 dependent signaling on the other hand, seems to have an antagonistic role by recruiting a second subset of antiangiogenic monocytes (CX3CR1^high, CCR2-negative, and Gr1-negative) to noninflamed tissue. ²³ These CX3CR1-positive monocytes are also thought to have homeostatic roles and to be involved in establishing the resident microglia populations in the eye with a turnover rate of approximately 6 months, which may, after activation, also be recruited to sites of laser injury. ^11,25  

In contrast to a previous report on Ccl2 ^−/− mice, ⁹ our study did not reveal an outer retinal degeneration in Ccl2 ^−/− compared with wild-type mice. Instead, we observed a normal age-related decline in the number of photoreceptor nuclei and ERG function, confirming that Ccl2 ^−/− mice do not show a higher level of photoreceptor degeneration than do wild-type mice. A similar observation was made during our quantitative histologic analysis of the RPE, which revealed an indistinguishable age-related increase in RPE damage in both genotypes and therefore no increased RPE atrophy in Ccl2 ^−/− mice. We applied an RPE pathology grading system developed for the CEP-aduct–immunized mouse model, in which oxidation-induced inflammation leads to an RPE pathology similar to AMD. ¹⁹ In our study, we found that the level of age-related damage observed in both, C57Bl/6 wild-type and Ccl2 ^−/− mice would only be graded as minor pathology, further supporting the conclusion that there is no major RPE atrophy in Ccl2 ^−/− mice, but rather normal and similar age-related changes in both genotypes. 

In ultrastructural histologic analyses of BM, we observed an age-related accumulation of an amorphous, electron-dense material in interdigitations on the basal site of the RPE that appeared to be similar to the basal laminar deposits found in aged human eyes with early age-related macular degeneration. ²⁶ However, the findings did not differ significantly between Ccl2 ^−/− and wild-type mice, confirming that the accumulation of electron dense material is a normal age-related process. Sub-RPE deposits similar to those in 20- to 25-month-old mice in this study have also been described in collagen XVIII/endostatin-knockout mice at the age of 16 months. Thus, one might speculate that ColXVII/ endostatin or its altered production may play a role in this age-related matrix deposition. ²⁷  

Overall, these findings indicate that most of the previously described hallmark features of age-related macular degeneration in Ccl2 and Ccr2-knockout mice ⁹ can be explained by normal aging and thus argue against the usefulness of these mouse models as animal models for evaluating treatments for AMD. Nevertheless, our data emphasize the importance of the MCP-1-CCR2 signaling not only for the extent of choroidal neovascularization, a major complication in AMD, ²⁸ but also for the maintenance of normal monocyte/macrophage trafficking during aging. 

Footnotes

 Supported by Moorfields Eye Hospital/Institute of Ophthalmology NIHR BMRC; The Royal College of Surgeons of Edinburgh; The Health Foundation; Royal Blind and the Scottish National Institution for the War Blinded.

Footnotes

 Disclosure: U.F.O. Luhmann, None; S. Robbie, None; P.M.G. Munro, None; S.E. Barker, None; Y. Duran, None; V. Luong, None; F.W. Fitzke, None; J.W.B. Bainbridge, None; R.R. Ali, None; R.E. MacLaren, None

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.