Complement Factor H Mutation W1206R Causes Retinal Thrombosis and Ischemic Retinopathy in Mice

Abstract

Single-nucleotide polymorphisms and rare mutations in factor H (FH; official name, CFH) are associated with age-related macular degeneration and atypical hemolytic uremic syndrome, a form of thrombotic microangiopathy. Mice with the FH W1206R mutation (FH^R/R) share features with human atypical hemolytic uremic syndrome. Herein, we report that FH^R/R mice exhibited retinal vascular occlusion and ischemia. Retinal fluorescein angiography demonstrated delayed perfusion and vascular leakage in FH^R/R mice. Optical coherence tomography imaging of FH^R/R mice showed retinal degeneration, edema, and detachment. Histologic analysis of FH^R/R mice revealed retinal thinning, vessel occlusion, as well as degeneration of photoreceptors and retinal pigment epithelium. Immunofluorescence showed albumin leakage from blood vessels into the neural retina, and electron microscopy demonstrated vascular endothelial cell irregularity with narrowing of retinal and choroidal vessels. Knockout of C6, a component of the membrane attack complex, prevented the aforementioned retinal phenotype in FH^R/R mice, consistent with membrane attack complex–mediated pathogenesis. Pharmacologic blockade of C5 also rescued retinas of FH^R/R mice. This FH^R/R mouse strain represents a model for retinal vascular occlusive disorders and ischemic retinopathy. The results suggest complement dysregulation can contribute to retinal vascular occlusion and that an anti-C5 antibody might be helpful for C5-mediated thrombotic retinal diseases.

Retinal vascular occlusive disorders collectively constitute the most common causes of visual disability in the middle-aged and elderly population.¹ Activation of the complement cascade, a component of the innate immune system, can promote retinal vascular occlusion, as seen in some patients with atypical hemolytic uremic syndrome (aHUS).2, 3, 4

Complement is activated via three pathways: classic, lectin, and alternative pathways. Factor H (FH) regulates the alternative pathway by inhibiting the activity of C3 convertase, C3bBb, both on the cell surface and in the fluid phase.5, 6, 7 FH contains 20 short consensus repeat domains; rare mutations in short consensus repeats 19 and 20 of human FH are associated with aHUS8, 9, 10 and age-related macular degeneration.11, 12 The C3 convertase activates the C5 convertase, which cleaves C5 into anaphylatoxin C5a and C5b. The latter promotes formation of the membrane attack complex (MAC), which contains C5b, C6, C7, C8, and C9. The MAC can cause lysis of microbes and human red blood cells as well as activation and injury of nucleated eukaryotic cells by generating holes in the cell membrane.

We previously generated a W1206R mutation in the mouse FH gene corresponding to the W1183R mutation in human FH found in familial aHUS.8, 9, 10 These mice developed thrombosis in multiple organs, including, as shown in vivo by fluorescein angiography (FA), the retinal vasculature.¹³ Herein, we tested the hypothesis that MAC formation is necessary for the thrombosis by determining whether FH W1206R mutation (FH^R/R)/C6 double-knockout mice are protected from the thrombotic phenotype of the eye. Detailed in vivo and histologic characterization of the retinal phenotype was studied. The results showed that FH^R/R mice developed retinal vessel occlusion, delayed and insufficient retinal perfusion, retinal detachment, inner retinal thinning, retinal pigment epithelium (RPE) degeneration, and basal RPE complement deposits. Knocking out C6, but not the receptor for anaphylatoxin C5a [C5a receptor 1 (C5aR1)] rescued the retinal phenotype of FH^R/R mice, suggesting that the MAC is involved in the pathogenesis. Anti-C5 antibody treatment also protected FH^R/R mice from developing the observed retinal phenotype. These findings suggest a causal relationship between FH dysfunction–induced MAC activation and retinal vascular thrombosis.

Materials and Methods

Animals

The generation of FH^R/R mice on a C57BL/6 background was described previously.¹³ C6^−/− mice on a C57BL/6 background were generated in house from C6 gene–targeted embryonic stem cells obtained from VELOCIGENE (VG16017; Tarrytown, NY).¹⁴ FH^R/R mice were crossed to C5aR1^−/−15 and C6^−/− mice to generate FH^R/R/C5aR1^−/− and FH^R/R/C6^−/− mice. All animals were screened for the Rd8 mutation, and results were negative. Age-matched littermates were used as controls, and both sexes were studied. Numbers of mice used in each experiment are shown in the corresponding figure legends. Experimental procedures were performed in accordance with the Association for Research in Vision and Ophthalmology statement for the use of animals in ophthalmology and vision research. All protocols were approved by the animal care review board of the University of Pennsylvania (Philadelphia, PA).

Fundus Imaging and Fluorescence Angiography

After anesthesia with a mixture of (in mg/kg body weight) 100 ketamine, 2 xylazine, and 2 acepromazine, bright-field imaging of the fundus was performed using a Micron III intraocular imager (Phoenix Research Labs, Pleasanton, CA). After fundus images were acquired, 100 μL fluorescein sodium (0.5 mg/L, AK-Fluor; Akorn Inc., Lake Forest, IL) was injected intraperitoneally, and a fluorescence image was captured immediately at 30-second intervals.

Morphologic Analysis

Enucleated eyes were immersion fixed in 2% paraformaldehyde/2% glutaraldehyde overnight. Then, eyecups were made by removing the cornea and lens before dehydration in increasing concentrations of ethanol, infiltrating overnight, and embedding the next day in plastic (JB4; Polysciences, Inc., Warrington, PA). For standard histology, plastic sections (3 μm thick) were cut in the sagittal plane and stained with 1% toluidine blude O and 1% sodium tetraborate decahydrate (Sigma-Aldrich, St. Louis, MO) for 5 seconds. Stained sections were then dried and imaged using bright-field microscopy (TE300; Nikon, Tokyo, Japan).

Immunofluorescence

After the globes were fixed in 4% paraformaldehyde, eyecups were generated by removing the anterior segment. The eyecups were infiltrated in 30% sucrose overnight and embedded in Tissue-Tek O.C.T. compound (Sakura Finetek, Torrance, CA). Immunofluorescence was performed on frozen sections (10 μm thick) with fluorescein isothiocyanate–conjugated C3 antibody (number 0855500; MP Biomedicals, Solon, OH), albumin (A90-134A; Bethyl, Montgomery, TX), and CD31 (ab76533; Abcam, Cambridge, MA). Control sections were treated identically with isotype-matched antibodies (ab37374; Abcam).

For immunolabeling of flat mounts, after enucleation, eyes were immersion fixed in 4% paraformaldehyde on ice for 20 minutes. Eyecups were then generated by circumferential incision at the pars plana, removing the cornea, iris, ciliary body, and lens. After blocking in 10% normal donkey serum in tris-buffered saline (TBS) plus 0.1% Triton X-100 (TBST), eyecups were incubated in 100 μL of primary antibody in 2% normal donkey serum in TBST at 4°C overnight. Eyecups were then washed three times by transferring the eyecup to a well with 200 μL of 1× TBST, at room temperature for 10 minutes. The secondary antibody was then applied by transferring the eyecup to a well with 100 μL of 2% normal donkey serum in 1× TBST containing secondary antibody (1:200 dilution) and incubating at room temperature for 2 to 3 hours. The eyecup was then washed three times by transferring to a well with 200 μL of 1× TBST, at room temperature for 10 minutes, then washed once by transfer of the eyecup to a well with 200 μL of 1× TBS, at room temperature for 10 minutes to remove the Triton X-100. The retina was then dissected from the RPE/choroid/sclera, flattened with four to six radial incisions, placed on a slide, and dried by wicking surrounding fluid with a Kimwipe (Kimberly-Clark, Irving, TX). Mounting medium was applied to coverslips, which were then inverted onto the slides. Antibodies used were as follows: rabbit anti-mouse CD31 (1:100 dilution; Abcam; ab28364), rat anti-mouse CD31 (1:100 dilution; Abcam; ab7388), BV421 rat anti-mouse CD41 (1:100 dilution; BD Horizon, Franklin Lakes, NJ; 562957), rabbit anti-human fibrinogen/fluorescein isothiocyanate (1:100 dilution; Dako, Carpinteria, CA; F0111), Cy3-Donkey Anti Rabbit IgG (H+L) (1:200 dilution; Jackson Immuno Research Laboratories, Inc., West Grove, PA; 711-165-152), Cy3-Donkey Anti Rat IgG (H+L) (1:200 dilution; Jackson Immuno Research Laboratories, Inc.; 712-165-153), and Alexa Fluor 488 Donkey Anti Rat IgG(H+L) (1:200 dilution; Invitrogen, Waltham, MA; A21208).

The sections were analyzed by fluorescence microscopy with identical exposure parameters (model TE300 microscope; Nikon) with ImagePro software version 1.8.0_112 (Media Cybernetics, Silver Spring, MD).

Spectral Domain OCT Imaging

Mice were anesthetized, and their pupils were dilated with 1% tropicamide (Bausch & Lomb, Inc., Bridgewater, NJ). Mice were placed in the Bioptigen (Durham, NC) AIM-RAS holder, and artificial tears were used throughout the procedure to maintain corneal clarity. Optical coherence tomography (OCT) images of the retina were acquired with the Envisu R2200-HR SD-OCT device (Bioptigen) with the reference arm path length set at 950 mm. Image acquisition software built into Envisu R2200-HR SD-OCT was used.

Electron Microscopy

Retinas (including RPE) fixed in 2% paraformaldehyde and 2% glutaraldehyde were postfixed in 1% osmium tetroxide and 0.1 mol/L sodium cacodylate buffer. Specimens were dehydrated and embedded in Epon (Ted Pella, Inc., Redding, CA). Ultrathin sections were cut, stained with uranyl acetate, and examined with a JEOL 1010 transmission electron microscope (Japan Electron Optics Laboratory Company, Mitaka, Tokyo).

Anti-C5 Antibody Treatment

FH^R/R mice were treated with an anti-C5 monoclonal antibody (mAb)¹⁶ intraperitoneally, 1 mg twice weekly, for 8 weeks starting at 4 weeks of age. The same concentration of MOPC, an isotype-matched IgG, was used as a control.

Statistical Analysis

The means ± SD were calculated for each comparison pair using a two-group t-test. Statistical analyses for quantification of C3 immunostaining were performed in GraphPad Prism 6.0 (GraphPad Software, Inc., San Diego, CA) by the one-way analysis of variance with a Tukey posttest comparing means. P < 0.05 was considered statistically significant for all analyses.

Results

FH^R/R but Not FH^R/R/C6^−/− Mice Develop Vascular Occlusion and White Spots Visible with in Vivo Retinal Imaging

Retinal imaging was performed on 8-week–old mice (Figure 1) to profile ophthalmic changes in FH^R/R, FH^R/R/C5aR1^−/−, FH^R/R/C6^−/−, and wild-type FH (FH^W/W) controls. FH^W/W mice had a normal retinal appearance. However, in FH^R/R and FH^R/R/C5aR1^−/− mice, occluded vessels appearing as white lines were observed (Figure 1). In addition, white ischemic patches were found in both FH^R/R and FH^R/R/C5aR1^−/− mice (Figure 1), with variable numbers of white spots (Figure 1). In contrast, FH^R/R/C6^−/− mice were rescued, exhibiting normal retinal appearance.

Figure 1.

Fundus images taken at 8 weeks of age. A–D: Representative normal fundus images of FH^W/W mice. E–H: FH^R/R mice show occluded vessels (arrows), white ischemic patches (arrowhead), and white spots. I–L: Similarly, FH^R/R/C5aR1^−/− mice show occluded vessels (arrows), ischemic patches (arrowheads), and white spots. M–P: However, FH^R/R/C6^−/− mice have normal fundus images. See Figure 2B for quantification.

FH^R/R Mice Show Delayed and Incomplete Retinal Perfusion

FA was performed at 8 weeks of age. The whole retina was perfused within 30 seconds in FH^W/W and FH^R/R/C6^−/− mice, and no fluorescein leakage was observed (Figure 2A). In contrast, both FH^R/R and FH^R/R/C5aR1^−/− mice exhibited delayed and/or incomplete retinal perfusion and fluorescein leakage around the optic nerve head. There was almost no perfusion at even 90 seconds after fluorescein injection in FH^R/R mice.

Figure 2.

A: Representative fundus images with fluorescein angiography (FA) at the indicated time points. Both FH^W/W and FH^R/R/C6^−/− mice show normal fundus and FA images. An FH^R/R mouse had almost no perfusion up until 90 seconds, whereas an FH^R/R/C5aR1^−/− mouse had delayed and incomplete retinal perfusion and fluorescence leakage around the optic nerve head. B: Quantification of abnormalities on fundus images and FA images. Abnormalities, including vessel occlusion, white spots, ischemic patches on fundus images, and perfusion delay and vascular leakage on FA images, are observed in FH^R/R and FH^R/R/C5aR1^−/− mice, but not in FH^W/W and FH^R/R/C6^−/− mice. Only one anti-C5 monoclonal antibody (mAb)–treated FH^R/R mouse shows vessel occlusion and white spots; the others exhibit no pathologic changes. Fractions represent ratios of abnormal mice/all experimental mice for each genotype. Fisher exact test reveals a significant association between genotype and likelihood of exhibiting all of the observed pathologic changes. Dotted lines separate different graphs. Bars within each graph are displayed in the same order as genotypes in the key. n = 17 FH^R/R mice (B); n = 14 FH^R/R/C5aR1^−/− mice (B); n = 8 FH^W/W mice (B); n = 15 FH^R/R/C6^−/− mice (B); n = 5 anti-C5 mAb–treated FH^R/R mouse (B). ^∗∗∗P < 0.001, ^∗∗∗∗P < 0.0001.

Quantification of abnormal phenotypes was performed using retinal and FA images (Figure 2B). None of the FH^W/W (n = 8) and FH^R/R/C6^−/− (n = 15) mice showed retinal abnormalities. In contrast, 82% of FH^R/R (n = 17) (Figure 2B) and 50% of FH^R/R/C5aR1^−/− (n = 14) mice had retinal vessel occlusion. A total of 88% of FH^R/R and 71% of FH^R/R/C5aR1^−/− mice had white spots. A total of 53% of FH^R/R and 43% of FH^R/R/C5aR1^−/− mice had ischemic patches. A total of 100% of FH^R/R (n = 8) and 89% of FH^R/R/C5aR1^−/− (n = 9) mice showed perfusion delay. A total of 100% of FH^R/R and 67% of FH^R/R/C5aR1^−/− mice showed vascular leakage. There were significant associations between genotype and likelihood of exhibiting the pathologic changes.

Pharmacologic Blockade of C5 Ameliorates Retinal Pathology in FH^R/R Mice

Because FH^R/R/C6^−/− mice were protected from retinal vascular occlusion, the therapeutic efficacy of a systemically administered anti-C5 antibody was tested (Figure 3, D–F, J–L, and P–R) compared with a mouse IgG1-κ monoclonal isotype control (Figure 3, A–C, G–I, and M–O). This anti-C5 antibody has been shown previously to block activation of the terminal complement cascade.¹⁶ Retinal images from a control antibody-treated FH^R/R mouse show hypopigmented spots (Figure 3, A and G) with leakage on FA (Figure 3M) at 4 weeks of age. After 4 weeks of treatment with the control antibody, this representative mouse showed pathologic progression that was indicated by ischemic patches (Figure 3B) and by hypopigmented spots (Figure 3H). There was also retinal vascular occlusion (Figure 3, C and I) leading to decreased retinal perfusion (Figure 3O) after 8 weeks of control antibody treatment. In comparison, hypopigmented spots (Figure 3D), ischemic patches (Figure 3J), and occluded vessels (Figure 3J) that were observed before anti-C5 treatment improved after 4 and 8 weeks of treatment, as demonstrated by imaging the same mice over time (Figure 3, E, F, K, and L). Note specifically that the vessel at the 5:30-o'clock position in Figure 3P reopened after anti-C5 treatment (Figure 3, Q and R); and leakage around the optic nerve head (Figure 3P) stopped after anti-C5 treatment (Figure 3, Q and R). Only one anti-C5 treated FH^R/R mouse (n = 5) (Figure 2B) showed vessel occlusion and white spots; the others exhibited no pathologic changes.

Figure 3.

Serial fundus images and fluorescein angiograms (FAs) with increasing age from representative mice before and after anti-C5 monoclonal antibody (mAb) treatment. A–C, G–I, and M–O: In a control antibody–treated mouse, white spots are found in both eyes (red arrowheads), with vascular leakage in the left eye (M and N). B, H, and N: After 4 weeks of treatment with control antibody, ischemic patches (black arrowheads in B) develop in the right eye and hypopigmented spots (red arrowheads in H) and vessel occlusion (arrow in H) are found in the left eye. After 8 weeks of treatment with control antibody, additional white spots and vessel occlusion (arrows in C and I) are seen in both eyes, with corresponding hypoperfusion on FA (O). Before anti-C5 mAb treatment (D and J), white spots (red arrowheads in D), ischemic patches (black arrowhead in J), and vessel occlusion (arrows in J) are observed, which then disappears after 4 weeks (E and K) and 8 weeks (F and L) of anti-C5 treatment. Insufficient retinal perfusion and leakage around the optic nerve head before treatment (P) improves after treatment with anti-C5 (Q and R).

Deposition of Fibrin and Platelets Is Found in FH^R/R Mice

To detect clots, immunolabeling of retinal flat mounts was performed with antifibrin and antiplatelet antibodies. Retinal vessels were labeled with vascular endothelial cell (VEC) marker anti-CD31 plus either antifibrin or antiplatelets. Retinas of FH^W/W mice appeared normal (Figure 4, A and C). However, strong staining for fibrin and platelets was observed within clots within the retinal vasculature in FH^R/R retinas (Figure 4, B and D).

Figure 4.

A and C: Minimal labeling is seen in FH^W/W mice. B and D: Fluorescence photomicrographs of retinal flat mounts showing fibrin (green; B) and platelet (green; D) staining in the retinal vasculature of FH^R/R mice. Vascular endothelial cells were labeled with anti-CD31 (red). Scale bars = 50 μm (A–D).

In Vivo OCT Imaging Shows Retinal Protection in FH^R/R/C6^−/− Mice

OCT imaging was performed to assess retinal morphology in vivo. No abnormalities were found in FH^W/W (Figure 5, A–E) and FH^R/R/C6^−/− (Figure 5, P–T) mice. In FH^R/R and FH^R/R/C5aR1^−/− mice, retinal detachment was observed (Figure 5, F, K, and M), as was outer nuclear layer (ONL) palisading in regions of disorganized photoreceptor inner/outer segments (Figure 5, F, G, and L), thinning of inner retinas (Figure 5, G and K) and of total retinas (Figure 5, H and M), occluded arteries (Figure 5H), as well as retinal edema (Figure 5, I, J, L, N, and O). No abnormalities were observed in anti-C5–treated FH^R/R mice (Figure 5, U–X).

Figure 5.

A: Optical coherence tomographic (OCT) images of retinas. A–E: Normal retinas of FH^W/W mice. F–O: In FH^R/R (F–J) and FH^R/R/C5aR1^−/− (K–O) mice, retinal detachment (white asterisks), outer nuclear layer (ONL) palisading in regions of disrupted photoreceptor inner/outer segment (IS/OS; red arrows), occluded vessels (yellow arrowhead), inner retinal thinning (single white arrows), total retinal thinning (double white arrows), and retinal edema (red arrowheads) are found. P–T: FH^R/R/C6^−/− mice show normal retinas. U–X: FH^R/R mice with anti-C5 monoclonal antibody (mAb) treatment have normal retinas. Y: Quantification of abnormalities on OCT images. Abnormalities, including vessel occlusion, retinal detachment, ONL palisading, inner retinal thinning, and retinal edema, are observed in FH^R/R and FH^R/R/C5aR1^−/− mice, but not in FH^W/W, FH^R/R/C6^−/−, and anti-C5 mAb–treated FH^R/R mice. Fractions represent ratios of abnormal mice/all experimental mice for each genotype. Fisher exact test reveals a significant association between genotype and likelihood of exhibiting all of the observed pathologic changes. Dotted lines separate different graphs. Bars within each graph are displayed in the same order as genotypes in the key. n = 4 FH^R/R mice (F–J), FH^R/R/C5aR1^−/− mice (K–O), FH^R/R/C6^−/− mice (P–T), and FH^W/W and FH^R/R/C6^−/− mice (Y); n = 5 anti-C5 mAb–treated FH^R/R mice (U–X) and FH^R/R and anti-C5 mAb–treated FH^R/R mice (Y); n = 6 FH^R/R/C5aR1^−/− mice (Y). ^∗P < 0.05, ^∗∗∗∗P < 0.0001. Scale bar = 50 μm (A–D, F–I, K–N, P–S, and U–X). INL, inner nuclear layer; RPE, retinal pigment epithelium.

Quantification of abnormalities in OCT images was performed (Figure 5Y). A total of 100% of both FH^R/R (n = 5) (Figure 5Y) and FH^R/R/C5aR1^−/− (n = 6) mice showed retinal detachment, vessel occlusion, and ONL palisading. A total of 80% of FH^R/R mice and 100% of FH^R/R/C5aR1^−/− mice showed inner retinal thinning. A total of 60% of FH^R/R and 50% of FH^R/R/C5aR1^−/− mice showed retinal edema. None of the FH^W/W (n = 4), FH^R/R/C6^−/− (n = 4), and anti-C5–treated (n = 5) mice had these abnormalities. There were significant associations between genotype and likelihood of exhibiting pathologic changes.

Retinal Ischemia and Degeneration in FH^R/R Mice

Morphologic analysis was performed at 8 weeks of age. Plastic sections of FH^W/W (Figure 6, A–D) and FH^R/R/C6^−/− (Figure 6, M–P) mice revealed normal retinal histology. In contrast, sections from FH^R/R (Figure 6, E–H) and FH^R/R/C5aR1^−/− (Figure 6, I–L) mice revealed occluded arteries and veins (Figure 6I), dilated veins (Figure 6, G, K, and L), retinal thinning (Figure 6, F, H, and L), ONL palisading (Figure 6, E, F, and J), vacuolar degeneration of photoreceptor inner/outer segments (Figure 6, E, G, J, and L), migration of photoreceptor nuclei toward the RPE (Figure 6, E, G, and I), migration of RPE cells toward the ONL (Figure 6, F and G), and vacuolar degeneration of the RPE (Figure 6, E and J). Among the five anti-C5–treated FH^R/R mice, one mouse showed retinal thinning (Figure 6S) and another one showed focal RPE vacuolization in the peripheral retina close to the ora serrata (Figure 6T). The other three exhibited normal retinas (Figure 6, Q and R). In comparison, control antibody–treated FH^R/R mice (n = 3) showed vein dilation (Figure 6, U and X) and retinal thinning (Figure 6V). One mouse showed minimal degeneration (Figure 6U), whereas another exhibited extensive total retinal thinning (Figure 6, W and X). One possibility for this observation is that the FH^R/R mutation may be associated with variable severity of retinal degeneration. This variability does not track with sex. However, the three control antibody–treated FH^R/R mice were the ones that survived the full 8 weeks of antibody treatment; another five only survived between 1 and 5 weeks after treatment initiation. Therefore, there may be selection bias in that the three healthiest control antibody–treated mice were analyzed, including a minimally affected one.

Figure 6.

Photomicrographs of plastic sections of mouse retinas. Each image is representative of its source eye, as pathologic severity does not noticeably change with retinal location. A–D: FH^W/W mice have normal retinas. E–L: In contrast, FH^R/R (E–H) and FH^R/R/C5aR1^−/− (I–L) mice exhibit palisading of the outer nuclear layer (ONL; E, F, and J), vacuolar degeneration of photoreceptor inner/outer segment (green arrowheads), migration of photoreceptor nuclei toward the retinal pigment epithelium (RPE; red arrowheads), migration of RPE cells toward the ONL (green arrows), vacuolar RPE degeneration (red arrows in E and J), dilated veins (black arrowheads in G, K, and L), occluded arteries and veins (yellow arrowheads), and retinal thinning (double black arrows in F, H, and L). M–P: FH^R/R/C6^−/− mice show normal retinas. Q–T: FH^R/R mice with anti-C5 monoclonal antibody (mAb) treatment have normal retinas, except for one mouse that shows focal thinning of the retina (black double arrows in S) and another mouse with focal RPE vacuolization in the peripheral retina (red arrows in T). U–X: Control (Ctrl) antibody–treated FH^R/R mice (W and X from same mouse) show vein dilation (black arrowheads in U and X) and retinal thinning (black double arrows in V). W and X: One mouse showed extensive total retinal thinning. Y: Quantification of abnormalities in images of plastic sections. Abnormalities, including vessel occlusion, inner retinal thinning, ONL thinning/palisading, and RPE degeneration, are observed in FH^R/R and FH^R/R/C5aR1^−/− mice, but not in FH^W/W and FH^R/R/C6^−/− mice. Only one anti-C5 mAb–treated FH^R/R mouse showed inner retinal thinning and RPE degeneration; the others exhibited no pathologic changes. Fractions represent ratios of abnormal mice/all experimental mice for each genotype. Fisher exact test revealed a significant association between genotype and likelihood of exhibiting all of the observed pathologic changes. Dotted lines separate different graphs. n = 4 FH^W/W mice (A–D), FH^R/R mice (E–H), FH^R/R/C5aR1^−/− mice (I–L), FH^R/R/C6^−/− mice (M–P), and FH^W/W and FH^R/R/C6^−/− mice (Y); n = 5 anti-C5 mAb–treated FH^R/R mice (Q–T), FH^R/R, FH^R/R/C5aR1^−/− (Y), and anti-C5 mAb–treated FH^R/R mice; n = 3 Ctrl antibody–treated FH^R/R mice (U–X). ^∗∗P < 0.01, ^∗∗∗P < 0.001, and ^∗∗∗∗P < 0.0001. Scale bar = 50 μm (A–X). GCL, ganglion cell layer; INL, inner nuclear layer.

Quantification showed that 100% of both FH^R/R (n = 5) (Figure 6Y) and FH^R/R/C5aR1^−/− (n = 5) (Figure 6Y) mice had vessel occlusion. A total of 100% of FH^R/R and 80% of FH^R/R/C5aR1^−/− mice had retinal thinning. A total of 80% of both FH^R/R and FH^R/R/C5aR1^−/− mice had ONL thinning/palisading and RPE degeneration. None of the FH^W/W (n = 4) and FH^R/R/C6^−/− (n = 4) mice showed these pathologies. Only one anti-C5–treated FH^R/R mouse exhibited inner retinal thinning and RPE degeneration (n = 5) (Figure 6Y). There were significant associations between genotype and likelihood of exhibiting pathologic changes.

Extravascular Albumin and C3 Leakage into the Neural Retinas of FH^R/R Mice

Albumin labeling was observed exclusively within the retinal vasculature in FH^W/W (Figure 7B), FH^R/R/C6^−/− (Figure 7P), and anti-C5–treated FH^R/R (Figure 7T) mice. In contrast, albumin labeling was seen outside of the vasculature in FH^R/R (Figure 7F), FH^R/R/C5aR1^−/− (Figure 7K), and control antibody–treated FH^R/R (Figure 7X) mice, indicating vascular leakage.

Figure 7.

Fluorescence photomicrographs showing C3, albumin, and CD31 localization. A–D: In FH^W/W mice, no C3 labeling is seen in the retina and retinal pigment epithelium (RPE). E–N: However, there is prominent retinal and sub-RPE C3 labeling in FH^R/R (E–I) and FH^R/R/C5aR1^−/− (J–N) mice. O–V: No retinal labeling and patchy sub-RPE C3 labeling are seen in FH^R/R/C6^−/− (O–R) and anti-C5 monoclonal antibody (mAb)–treated FH^R/R (S–V) mice. W–Z: Control (Ctrl) antibody–treated FH^R/R mice exhibit retinal C3 labeling similar to FH^R/R and FH^R/R/C5aR1^−/− mice, but patchy sub-RPE labeling. B, P, and T: Albumin labeling is exclusively within the retinal vasculature in FH^W/W (B), FH^R/R/C6^−/− (P), and anti-C5 mAb–treated FH^R/R (T) mice. F, I, K, and N: C3 also localizes to the basolateral side of the RPE (white arrowheads), and albumin localizes to both apical and basolateral sides of RPE cells (red arrowheads). F, K, and X: Extravascular albumin labeling is seen in FH^R/R (F), FH^R/R/C5aR1^−/− (K), and control antibody–treated FH^R/R (X) mice, indicating vascular leakage. N: Higher-magnification shows colabeling of C3 with endothelial cells (white arrow) and extravascular C3 labeling (red arrow). AA: Quantification of sub-RPE C3 signal intensity reveals that FH^R/R and FH^R/R/C5aR1^−/− mice have significantly higher signals than the other groups, whereas FH^W/W mice have a lower signal than all other groups. BB: Quantification of extravascular retinal albumin signal intensity shows that FH^R/R, FH^R/R/C5aR1^−/−, and control antibody–treated FH^R/R mice have significantly higher signals than FH^W/W and FH^R/R/C6^−/− mice. All anti-C5 mAb–treated FH^R/R mice, except one, and one FH^R/R mouse have intensities similar to FH^W/W and FH^R/R/C6^−/− mice. n = 4 (A–D, FH^W/W mice, E–I, FH^R/R mice, and O–R, FH^R/R/C6^−/− mice); n = 6 (J–N, FH^R/R/C5aR1^−/− mice); n = 5 (S–V, anti-C5 mAb–treated FH^R/R mice); n = 3 (W–Z, Ctrl antibody–treated FH^R/R mice). Scale bar = 50 μm (A–H, J–M, and O–Z). ^∗P < 0.05, ^∗∗P < 0.01, and ^∗∗∗∗P < 0.0001. INL, inner nuclear layer; NSR, neurosensory retina; ONL, outer nuclear layer.

Immunolabeling was performed with anti-C3 to test for complement deposition. There was no C3 labeling in FH^W/W mice (n = 4) (Figure 7A). However, there was prominent sub-RPE C3 labeling in FH^R/R (n = 4) (Figure 7E) and FH^R/R/C5aR1^−/− (n = 6) (Figure 7J) mice. Staining with CD31 antibody is shown (Figure 7, C, G, L, Q, U and Y). Merged channels are shown (Figure 7, D, H, I, M, N, R, V and Z). Higher-magnification images of sections colabeled with a CD31 antibody against VECs showed that C3 colocalized with retinal VECs (Figure 7N) and was also present outside vessels within the neurosensory retina (NSR) (Figure 7N). Patchy sub-RPE C3 labeling was still observed in FH^R/R/C6^−/− (n = 4) (Figure 7O) and anti-C5–treated FH^R/R (n = 5) (Figure 7S) mice. Surprisingly, control antibody–treated FH^R/R mice (n = 3) (Figure 7W) had minimal sub-RPE labeling. C3 labeling in the neural retina was similar to the albumin leakage patterns, consistent with C3 leaking into the neural retinas from the vasculature.

Pixel density analysis revealed a significantly higher sub-RPE C3 signal in FH^R/R and FH^R/R/C5aR1^−/− mice relative to the other groups (Figure 7AA). In comparison, FH^W/W mice had lower signal than all other groups. Notably, there was a reduction in sub-RPE C3 signal intensity, relative to FH^R/R mice, in both anti-C5–treated FH^R/R and control antibody–treated FH^R/R mice, suggesting a possible inhibition of sub-RPE C3 deposition conferred by antibody treatment in a nonepitope-specific manner. Regarding extravascular albumin in the NSR, the FH^R/R group (with the exception of one mouse) along with the FH^R/R/C5aR1^−/− and control antibody–treated FH^R/R groups exhibited higher albumin signal in the NSR than the other groups (Figure 7BB). Only one of five anti-C5–treated FH^R/R mice had a relatively high albumin signal.

Electron Microscopy Reveals Irregular VECs Associated with Constriction of Superficial Retinal and Choroidal Vasculature of FH^R/R Mice

To better determine the nature of the observed vascular occlusion, ultrastructural analysis was performed using electron microscopy. Relative to FH^W/W mice (n = 4) (Figure 8, A–C), FH^R/R mice (n = 5) (Figure 8, D–G) exhibited decreased lumen diameters of both choroidal capillary (Figure 8D) and retinal vessels (Figure 8, E–G) and irregularly contoured VECs (Figure 8, D–G).

Figure 8.

Electron micrographs showing choroidal (A and D) and superficial retinal vessels (B, C, E, F, and G). All sections were taken from near the optic nerve. Relative to FH^W/W mice (A–C), FH^R/R mice (D–G) exhibit enlarged vascular endothelial cells (VECs) (D), narrowed vessel lumens, irregularly contoured VECs, and enlarged pericytes in small arterioles (E–G). Arrows indicate VECs. n = 4 (A–C); n = 5 (D–G). Scale bar = 4 μm (A–G). CC, choroidal capillary; RPE, retinal pigment epithelium.

Discussion

Herein, we report a mouse model of retinal vascular occlusion and ischemic retinopathy. Key pathologic changes in eyes of FH^R/R mice include retinal vascular occlusion leading to atrophic thinning of the retina. These features are also found in human retinal vascular occlusive disorders. Moreover, although previous mouse models of retinal vascular disease have used photothermal damage and photodynamic activation of rose bengal to induce occlusion,¹⁷ this model uniquely features occlusion secondary to a mutation in a complement regulator gene.

The renal phenotypes of FH^R/R mice share features of human aHUS.¹³ The W1206R mutation in mouse FH is equivalent to the W1183R mutation in human FH found in some aHUS patients.8, 9, 10 This mutation impairs FH interaction with the host cell surface but does not affect the complement-regulating function of FH. Thus, one major consequence of this and other similar mutations in FH C-terminus is susceptibility of host cells, particularly kidney and retinal endothelial cells, to complement-mediated injury with increased risk of thrombotic vasculopathy. Glomerular thrombotic microangiopathy is the pathologic hallmark of aHUS.¹⁸ Interestingly, ocular involvement in both adult19, 20, 21 and pediatric patients20, 22, 23, 24, 25, 26 with aHUS has been reported. It is likely that most of the ocular findings in FH^R/R mice occur secondary to thrombotic microangiopathy after MAC-mediated VEC damage because these had extensive retinal nonperfusion and vascular leakage on FA whereas FH^R/R mice that also had deficiency in C6, a component of the MAC, had normal retinas. Terminal complement activation most likely damaged VEC and caused stenosis of vessels, as observed by electron microscopy, predisposing the vessel wall to thrombosis, which was observed in retinal flat mounts immunolabeled with antifibrin and antiplatelet antibodies.

Neutrophils or monocytes are activated by C5a via the C5aR1, which is highly expressed on these cells. Activation of the C5a/C5aR pathway causes release of tissue factor to participate in clot formation in both capillaries and large blood vessels.27, 28 Deletion of C5aR1 from FH^R/R mice prevented macrovessel thrombosis in several organs, such as the liver and spleen, but did not prevent thrombotic microangiopathy in the kidney.²⁹ Results presented herein suggested that the C5a/C5aR pathway is similarly not involved in the retinal pathologies of FH^R/R mice; instead, MAC component C6 was involved. Thus, it is likely that both the retinal phenotypes and kidney thrombotic microangiopathy are MAC dependent. It is possible that direct injury by MAC leads to endothelial swelling and narrowing of capillary lumens, as we have observed in the retinas of FH^R/R mice (Figure 8). Released hemoglobin from sheared and fragmented red blood cells passing through the narrowed lumen can scavenge nitric oxide, a major vasoprotective and platelet activation inhibitory molecule, further promoting the occurrence of retinal thrombosis.³⁰

Interestingly, C3 label in the retina colocalizes with extravascular albumin in FH^R/R and FH^R/R/C5aR1^−/− neural retinas, indicating leakage across a damaged blood-retinal barrier, consistent with the leakage shown by FA. In FH^R/R/C6^−/− and anti-C5 mAb–treated FH^R/R mice, probably because of the downstream blockade of MAC formation, there was less damage to VEC and no leakage of albumin or C3 into the NSR. However, C3 label on the basal RPE most likely occurs because of the loss of RPE membrane protection by FH. Treatment with anti-C5 mAb diminished this labeling. However, this effect of anti-C5 mAb is unlikely to be related to its MAC-inhibiting activity as the control antibody had a similar effect. Whether this represented a nonspecific anti-inflammatory effect similar to i.v. Ig treatment remains to be investigated.

In patients, Purtscher and Purtscher-like retinopathy are rare occlusive thromboembolic retinopathies. Purtscher retinopathy refers to a traumatic etiology, whereas Purtscher-like retinopathy has nontraumatic causes.³¹ Although the pathogeneses of Purtscher and Purtscher-like retinopathy are multifactorial, embolization of the retinal circulation has been proposed as the common cause of the ocular findings.32, 33, 34, 35, 36 Most patients with Purtscher-like retinopathy show cotton wool spots and intraretinal hemorrhages on funduscopic examinations, and there is evidence of occlusive thromboembolic retinopathy with nonperfused areas and fluorescein dye leakage from retinal arterioles, capillaries, venules, and the optic disk.31, 33, 37 Previous reports showed that select patients with Purtscher-like retinopathy were successfully treated with i.v. eculizumab,2, 32 a humanized monoclonal anti-C5 mAb that prevents activation of the terminal complement cascade, analogous to the protective effect of anti-C5 mAb in our FH^R/R mice.

These anti-C5 mAb therapy results indicate that C5 activation participates in the pathophysiology of vascular occlusion and ischemic retinopathy in FH^R/R mice. The finding that an anti-C5 mAb protects against retinal thrombosis and ischemia in this aHUS model highlights differences between this model and age-related macular degeneration. For age-related macular degeneration patients, i.v. anti-C5 therapy with eculizumab was not clinically effective. These findings also highlight the disparate retinal effects of the FH W1206R mutation and the age-related macular degeneration–associated Y402H mutation. The study supports the concept that anti-C5 antibodies may represent a valuable therapeutic approach, particularly in patients with Purtscher-like retinopathy in whom complement is the primary driving factor. It will be of interest to determine whether complement may play a role in additional patients with retinal vascular thrombosis.