Transgenic Mice Expressing Variants of Complement Factor H Develop AMD-like Retinal Findings

Abstract

Purpose.: Complement factor H (Cfh) is a key regulator of the alternative complement pathway. A Cfh variant (Y402H) increases the risk for AMD. The purpose of this study was to develop a pathophysiologically relevant animal model of AMD based on this genetic risk factor.

Methods.: The authors generated chimeric Cfh transgenic mouse lines using two constructs consisting of the human CFH sequence for SCR6–8 (with either 402Y or 402H), flanked by the mouse sequence for SCR1–5 and SCR9–20. They tested the expression of the transgenic mRNA and protein molecules and examined the mice at 12 to 14 months of age for clinical and histologic retinal changes.

Results.: Nuclease protection assay and qRT-PCR analysis demonstrated transgenic mRNA expression in the liver and in the posterior segment of the eye. Western blot analysis showed that the transgenic proteins are present in the circulation at levels comparable to those of mouse Cfh. The chimeric proteins were found to be functional, as demonstrated by their ability to restore physiological serum levels of complement component C3 in Cfh KO mice. Clinical examination showed subretinal drusen-like deposits. Histology demonstrated an accumulation of subretinal cells that stained with a macrophage/microglia marker. Basal laminar deposits, long-spaced collagen, and increased numbers of lipofuscin granules were seen on electron microscopy. Immunohistochemistry showed a thicker sub-RPE band of C3d staining.

Conclusions.: Chimeric Cfh proteins led to AMD-like characteristics in mice. This may represent a good model for studying the role of complement and other components of the immune system in early AMD.

Approximately 1.8 million Americans have advanced age-related macular degeneration (AMD). ¹ Despite recent advances in the treatment of choroidal neovascularization (CNV), clinical outcomes remain suboptimal. Furthermore, there is no effective therapy for atrophic AMD. 

The early stages of AMD are hallmarked by an accumulation of debris (altered proteins and lipids) generated from oxidative damage to photoreceptors, RPE, or choriocapillaris. ^{2 –5} Recent literature suggests that an immune reaction to this debris could potentially promote more tissue damage, neovascular response, or both. ⁶ Multiple lines of evidence have strengthened the hypothesis that the immune system is intimately involved in the disease process of AMD. ^{6 –11} Histopathological studies have shown the presence of inflammatory cells in retinal lesions and anti-retinal antibodies in the sera of AMD patients. ¹² In addition, patients have increased systemic levels of the inflammatory markers C-reactive protein (CRP for the human protein, Crp for the mouse protein) and IL-6. ¹³ Still, it is debated whether the immune system plays a role in initiating or exacerbating the disease or whether these findings are a response to the retinal/RPE/sub-RPE damage that occurs in AMD. ¹⁴ The strongest evidence to date for a link between inflammation and AMD comes from genetic studies demonstrating a strong association between some variants of complement factor H (CFH for the human protein, Cfh for the mouse or chimeric protein) and the risk for AMD. 

CFH is a key regulator of the alternative complement pathway. A single nucleotide polymorphism that leads to a histidine (H) instead of a tyrosine (Y) at amino acid position 402 of CFH (Y402H) significantly increases the risk for AMD. ^{8 –11} Although still a controversial subject, the 402H variant may also make AMD patients resistant to therapy with anti-VEGF agents ^15,16 and less responsive to antioxidant prophylaxis. ^17,18 Our knowledge about the mechanisms explaining this increase in risk for disease and decrease in susceptibility to therapy is in its infant stages. 

Despite the obvious anatomic differences with humans, mice can develop subretinal lesions reminiscent of the basal laminar deposits, drusen, and CNV seen in AMD. ^{19 –22} However, there is a paucity of animal models invoking mechanisms identified by genetic studies to be of potential relevance in AMD. Because at least 50% of AMD cases are associated with the “at risk” H402 variant of CFH, an animal model based on this variant would be of great help. We describe here for the first time the generation and characterization of such a model: chimeric Cfh transgenic mice. 

Materials and Methods

Animals

C57BL/6 mice (B6) used for breeding with the Cfh transgenic mice were obtained from a UT Southwestern Medical Center breeding core facility (Wakeland Laboratory). The Cfh KO mice were developed by Marina Botto and collaborators. ²³ Animals were handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the National Institutes of Health Guidelines on Laboratory Animal Welfare. 

Design of Chimeric Cfh Constructs and Generation of Cfh Transgenic Mice

We generated two chimeric (part mouse/part human) Cfh constructs (Fig. 1). CFH consists of 20 short consensus repeats (SCRs). Our constructs consisted of mouse SCR1 to SCR5, followed by human SCR6 to SCR8, followed by mouse SCR9 to SCR20. One of the constructs codes for a tyrosine in amino acid position 402 (Y-Tg-Cfh), and the second one codes for a histidine in this position (H-Tg-Cfh). The plasmids were generated by cloning the coding chimeric Cfh DNA sequence into the MluI-ClaI restriction sites of the pLiv11 vector (gift from John Taylor ²⁴ ), which contains the constitutive human apoE gene promoter. The 13.7-kb transgenic sequences were released by SalI digestion, isolated with columns (Elutip-D; Whatman, Piscataway, NJ), and injected into fertilized C57Bl/6 eggs by the University of Texas Southwestern Transgenic Technology Center. 

Figure 1.

 

Chimeric Cfh transgenic constructs. Two chimeric Tg-Cfh constructs differing only in nucleotide 1204 (corresponding to amino acid 402) of the Cfh coding sequence were generated.

The resultant transgenic founders were crossed to B6 mice three times. Each time, only lines in which the litters showed close to 50% of pups positive for the transgene were kept for further breeding. The progeny of the third cross to B6 mice was used for breeding to Cfh-knockout mice (mCfhKO) to obtain CfhTg/mCfhKO mice. 

Genotyping of Mice

Transgenic founder mice were identified by PCR using tail DNA. The wild-type mouse cfh gene fragment was amplified using sense 5′-GTGCTTTCGTGACTCCTAC-3′ and anti-sense 5′-GGTATAAACAACCTTTGCACC-3′ primers, resulting in a 168-bp product. Chimeric cfh was amplified using sense 5′-GCTTCTATCCTGTAACTGGATCA-3′ and anti-sense 5′-GGAACATGTTTTGACACGGATGCATCTGGGAG-3′, resulting in a 458-bp product. For the cfh-KO sequence, sense 5′-GTAAAGGTCCTCCTCCAAGAG-3′ and anti-sense 5′-GGGGATCGGCAATAAAAAGAC-3′ primers resulted in a 0.5-kb product. 

Quantitative RT-PCR

Livers and posterior segment eyecups were collected from 8- to 10-week-old F3 transgenic mice on a C57BL/6 background. Total RNA was extracted from individual posterior eyecups or from 100 mg liver tissue using reagent (Trizol; Invitrogen, Grand Island, NY) according to the manufacturer's instructions. cDNA was synthesized from total RNA (SuperScript VILO cDNA Synthesis Kit; Invitrogen). Singlet qPCR reactions were run in triplicate (iCycler; Bio-Rad Laboratories, Hercules, CA) at 50°C for 2 minutes and 95°C for 2 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute (EXPRESSSYBR GreenER qPCR SuperMix Universal kit; Invitrogen). Each reaction contained 62.5 ng cDNA, 200 nM each primer, and 10 μL qPCR super mix in 25 μL total volume. The primers used were as follows: for chimeric cfh, forward 5′-TGGATCCCTGTTCCAAGATG-3′ and reverse 5′-TCTGAGGCATGGTACTGCTG-3′; for mouse GAPDH, forward 5′-ACTCACGGCAAATTCAACGGC-3′ and reverse 5′-ATCACAAACATGGGGGCATCG-3′. Fold changes in chimeric Cfh mRNA expression in the transgenic tissues were normalized to an endogenous reference gene, GAPDH, and expressed relative to the corresponding tissues in B6 control samples using the quantitative 2^-ΔΔC(t) method. ²⁵  

Nuclease Protection Assay

Thirty micrograms total liver RNA was hybridized in a 25-μL mixture containing 40 mM Tris-HCl (pH 7.4), 0.6 M NaCl, 4 mM EDTA, 40% (vol/vol) formamide, 40 mM DTT, and approximately 60,000 cpm P32-labeled riboprobe. The riboprobe was generated with uridine 5′-[α-P32] triphosphate (MP Biomedicals, Inc., Irvine, CA), Tg-Cfh plasmid, and T7 RNA polymerase using a transcription kit (Maxiscript T7; Ambion, Austin, TX) according to the instructions provided. After hybridization and S1 nuclease digestion, the protected transgenic (279 nucleotides) and wild-type (156 nucleotides) fragments were detected by electrophoresis on 8% polyacrylamide gel containing 8 M urea and analyzed with a phosphoimager (Typhoon 9410; GE Healthcare, Piscataway, NJ). The resultant bands were normalized for their uridine monophosphate contents. 

Western Blot Analysis

Serum was obtained by submandibular blood draws. Two microliters serum was mixed with 3 μL PBS and 10 μL sample buffer (β-mercaptoethanol/Laemmli sample buffer; 1:20), and the total 15 μL was loaded on a 7.5% Tris-HCl precast gel (Bio-Rad Laboratories). Electrophoresis was run for 45 minutes with 1× Tris/glycine/SDS running buffer at 180 V. The proteins were transferred to PVDF membrane (Hybond-P; Amersham, Arlington Heights, IL) in 1× Tris/glycine transfer buffer for 1 hour at a maximum of 400 mA (with cooling). The blot was blocked by overnight incubation at 4°C using 0.1% (vol/vol) Tween 20 and 5% (wt/vol) nonfat dried milk in Tris-buffered saline (TBS). After washing the blot with an excess of TBS-T (consisting of 0.2% [vol/vol] Tween 20, 0.2% nonfat dried milk in 1× TBS) for 2 × 10 minutes, the membrane was incubated at room temperature (RT) for 1 hour with affinity-purified polyclonal rabbit anti–human Cfh primary antibodies ²⁶ diluted 1:150 in blocking buffer. Three different primary antibodies were used. The anti–H antibody, developed to detect only human 402H-Cfh, detected our chimeric H-Tg-Cfh molecules but not Y-Tg-Cfh or mouse Cfh molecules. The anti–Y antibody detected only Y-Tg-Cfh. The anti–H/Y antibody detected both the 402H and 402Y variants with the same affinity ²⁶ but did not detect mouse Cfh. The blots were washed 3 × 5 minutes in excess TBS-T and were incubated at RT for 1 hour with goat anti–rabbit HRP conjugated secondary antibody (Southern Biotechnology, Birmingham, AL) diluted 1:4000 in blocking buffer. After a wash of 3 × 5 minutes in TBS-T, the chimeric protein was detected using reagents (Amersham ECL Plus; GE Healthcare UK Limited, Buchinghamshire, UK) with an advanced imaging system CCD camera (ChemiDoc XRS; Bio-Rad Laboratories). The image was captured live by exposing the membrane to high-sensitivity light (Chemi Hi; Bio-Rad Laboratories) for 120 seconds and was analyzed using image and analysis software (Quantity One; Bio-Rad Laboratories). 

Complement Factor 3 ELISA

The complement factor 3 (C3) concentration in mouse serum was measured using an immunoperoxidase assay kit (GenWay Biotech, Inc., San Diego, CA). Serum samples were diluted 1:20,000 for the assay. Transgenic mice that had been crossed twice to mCfhKO mice and were shown by genotyping to be deficient in mCfh were used for these experiments. 

Fundus Photography

We examined the central retinas (radius of 3–4 disc diameters around the disc) of our founder transgenic mice at 12 months of age by placing hydroxypropyl methylcellulose ophthalmic (Goniosol; Novartis, Basel, Switzerland) and a coverslip on their corneas and using an operating microscope (Zeiss, Thornwood, NY). We examined and photographed the eyes of the CfhTg/mCfhKO mice using a mouse fundus camera (Micron III; Toshiba, Irvine, CA) that allowed us to examine the entire retina, including the peripheral retina up to the ora serrata, in all directions. 

Histopathology and Electron Microscopy

Mice were anesthetized with 115 mg/kg ketamine and 5.6 mg/kg xylazine and were then perfused through the heart with 1% glutaraldehyde and 2% paraformaldehyde in PBS (pH 7.4). Fixed eyes were removed and sectioned behind the limbus, and eyecups were immersed in 2% glutaraldehyde and 2% paraformaldehyde in 100 mM cacodylate buffer (pH 7.4) overnight at 4°C. Eyecups were dehydrated in an ethanol series to 100%, embedded in epoxy resin (Poly/Bed 812; Polysciences, Inc., Warrington, PA). For light microscopy, 1-μm sections were stained with 1% toluidine blue. For electron microscopy, 70-nm thin sections were cut on an ultramicrotome (Leica E; Leica, Wetzlar, Germany) and placed on 100 mesh, polyvinyl formal resin (Formvar; Monsanto Chemical, St. Louis, MO)–coated, copper grids. Sections were stained with 2% aqueous uranyl acetate and lead citrate and viewed in a transmission electron microscope (1200EX II; JEOL, Tokyo, Japan), operated at 120 kV, equipped with a megapixel side-mount CCD camera (Sis Morada 11; Olympus). 

Immunohistochemistry

Eyes from transgenic and age-matched B6 mice at 12 to 14 months of age were enucleated and immediately placed in 10% Carson's formalin. After 48 hours they were embedded in paraffin, and 7-μm sections were cut for immunohistochemistry. After deparaffinization and hydration in xylene and a series of graded dilution of ethanol, sections were heated in 10 mM sodium citrate in 95°C water bath for 15 minutes twice for antigen retrieval. Nonspecific binding sites on the sections were blocked by incubation in 10% normal serum matching the secondary antibody at RT for 1 hour. The sections were incubated overnight at 4°C with primary antibody (rabbit anti–Iba1, 1:500 [Wako Chemicals, Richmond, VA] or goat anti–mouse C3d, 1:50 [R&D Systems Inc., Minneapolis, MN]). The anti–Iba1 antibody is specific for macrophages/microglia. ²⁷ After washing 2 × 15 minutes in phosphate-buffered saline with 0.01% Tween-20 (PBST), the sections were incubated for 1.5 hours at RT with a 1:200 dilution of secondary antibody (goat anti–rabbit IgG labeled with either AlexaFluor488 or AlexaFluor568 was used for Iba-1, and AlexaFluor555-labeled rabbit anti–goat IgG was used for C3d; Invitrogen). After washing 2 × 15 minutes in PBST, the sections were mounted in gel-mount antifade medium, and immunolabeling was detected on a confocal microscope (LSM 510; Zeiss). 

Flat Mount Preparation and Immunostaining

Eyes from CfhTg/mCfhKO and age-matched B6 mice were enucleated and placed in 4% paraformaldehyde for 2 hours at RT. After washing 2 × 10 minutes in PBS, the anterior segment was cut out, and the retina was removed from the eyecup. Each posterior eyecup was flattened (radial cuts). Tissues were blocked in 5% BSA containing 0.3% Triton X-100 for 2 hours at RT, followed by incubation with primary antibodies (in a 1:5 dilution of blocking buffer in PBS) overnight at 4°C. The primary antibodies used were rabbit anti–Iba1 alone (1:500) or a combination of anti–Iba1 plus rat anti–mouse F4/80 (1:10 of AbD clone CI:A3–1; Serotec, Raleigh, NC). The next day the flat mounts were washed 3 times in PBS and incubated with secondary antibodies (1:200) at RT for 2 hours (goat anti–rabbit AlexaFluor 488 was used for anti–Iba1, and goat anti–rat AlexaFluor 568 was used for the anti–F4/80). After three washes in PBS, the flat mounts were coverslipped and mounted with anti-fade reagent (Prolong Gold; Invitrogen) and were photographed using fluorescence microscopes (AxioObserver and StereoDiscovery; Zeiss). 

Results

Chimeric Cfh mRNA Expression in the Liver and Posterior Eye Segment of Transgenic Mice

We identified 10 lines of Y-Tg-Cfh and six lines of H-Tg-Cfh transgenic mice. These were tested by a nuclease protection assay (Fig. 2A) using a probe with a sequence complementary to a fragment of chimeric Cfh crossing the junction of mouse and human sequences. Different fragments of mRNA were protected from digestion in chimeric mRNA compared with native mouse mRNA molecules. This allowed us to determine the relative levels of chimeric compared with native mouse mRNA. Results varied between lines (Fig. 2B). Transgenic lines expressing chimeric Cfh mRNA in moderate levels (>50% of the mouse native Cfh mRNA: Y-TgL2, Y-TgL3, Y-TgL6, Y-TgL7, H-TgL1) or high levels (similar to mouse native Cfh mRNA: Y-TgL1, H-TgL2, H-TgL3, H-TgL4, H-TgL5) were further bred to B6 mice three times. Lines expressing very low levels of chimeric mRNA (Y-TgL4, Y-TgL5, Y-TgL8, Y-TgL9, Y-TgL10, H-TgL6) were discarded. Furthermore, lines consistently transmitting the transgene to >50% of the progeny (H-TgL5, suspect for multiple transgene insertions) or showing high fluctuation in the level of protein expression (Y-TgL2, Y-TgL6, Y-TgL7, H-TgL4) between transgene-positive littermates or between different generations were eliminated. Thus, we chose lines Y-TgL1, Y-TgL3, H-TgL1, H-TgL2, and H-TgL3 for further work. 

Figure 2.

 

Chimeric Cfh mRNA is expressed at levels similar to those of mouse Cfh mRNA in the transgenic liver and is also expressed in the posterior segments of the eyes. (A) Nuclease protection assay on liver tissue of transgenic or wild-type mice showing bands corresponding to chimeric and mouse Cfh mRNA expression. The results were confirmed in a separate experiment. (B) Ratio of transgenic mRNA to native mouse Cfh mRNA in each animal in (A) after correcting for the content of radioactive nucleotides in each fragment. (C) qRT-PCR was performed on either liver samples or posterior segments of eyes of H transgenic mice versus Y transgenic mice versus B6 mice. The bars are the result of the analysis by the 2^-ΔΔC(t) method and represent the level of expression of chimeric Cfh mRNA normalized to GADPH expression. Each bar represents the average of four mice. The experiment was reproduced twice, each time in duplicate.

It is still unclear whether the relevant CFH in human AMD proceeds from the liver or from the RPE. Since ApoE is normally expressed in human RPE and retina, ²⁸ we wanted to determine whether our chimeric Cfh construct was being expressed in the eyes of our transgenic mice. RT-PCR analysis was performed on liver samples and posterior eyecups of Y-TgL1, H-TgL3, and B6 mice. Figure 2C demonstrates that the level of TgCfh expression in the liver was similar in both transgenic mouse lines. This was also the case for the eyecups. Liver samples seemed to have higher levels of chimeric mRNA than eye samples. However, we could not control for the relative composition of posterior eyecups (retina vs. RPE vs. choroid vs. sclerae). Thus, although we can say that our transgenic mice express chimeric Cfh mRNA in the posterior eye segment, it is impossible to compare the level of chimeric Cfh expression between RPE cells and liver cells. 

Expression of Stable Chimeric Cfh Proteins

Western blot analysis was performed using three affinity-purified polyclonal antibodies ²⁶ that recognize human CFH molecules in the SCR6–8 region with great specificity. The anti–H antibody (developed to recognize the human 402H-CFH variant) detected H-Tg-Cfh protein but not the Y-Tg-Cfh protein (Fig. 3). The anti–Y antibody detected Y-Tg-Cfh but not H-Tg-Cfh, despite the fact that the two variants differ in only one amino acid. Figure 4A shows a Western blot analysis using an anti–H/Y antibody that recognizes the 402H and 402Y variants with similar affinity. ²⁶ The serum chimeric Cfh levels (Fig. 4B, table) for lines H-TgL3 and Y-TgL1 (also for H-TgL2; data not shown) were 200 to 210 μg/mL. This was comparable to reported Cfh levels in rats (238 ± 21 μg/mL) ²⁹ and humans (110–615 μg/mL). ³⁰ The expression level was 25% lower for line H-TgL1 and was somewhat variable in line Y-TgL3. 

Figure 3.

 

Chimeric Cfh protein molecules are expressed in serum and conserve the expected structure in the area of the mutation. Western blot using highly purified polyclonal antibodies specific for the 402Y variant of human CFH (top) or the 402H variant of human CFH (bottom). The transgenic mice are labeled with Y or H, depending on the type of chimeric construct they express. The letter is followed by the number of the original founder from which they proceed. The transgenic mice in this figure are the product of breeding to mouse Cfh KO mice (mCfhKO) and thus lack mouse Cfh. In addition, these antibodies do not recognize mouse Cfh (see lack of band on B6 control). The experiment was reproduced multiple times with transgenic mice on a wild-type background and transgenic mice on a Cfh KO background.

Figure 4.

 

Chimeric Cfh molecules were expressed in the sera of our transgenic mice on a mCfh KO background. The level of transgenic Cfh expression in our transgenic mice was similar to that of native mouse Cfh in B6 mice. (A) Western blot using a highly purified polyclonal antibody that recognizes both variants of human Cfh (hCFH) with similar affinity. The antibody does not recognize mouse Cfh. Commercial pure hCFH (a mixture of H and Y variants) was loaded in amounts of 280 and 420 ng. (B) Known amounts of hCFH in (A) were used to generate a standard curve to estimate the concentration of TgCfh in the serum of our transgenic mice. Results are representative of three separate experiments. (C) Western blot using a rabbit polyclonal antibody (M300 antibody) that recognized the amino terminal of the mouse Cfh molecules and could not distinguish between mouse Cfh and our transgenic Cfh molecules. This antibody could not detect purified human Cfh.

We crossed our selected lines (Y-TgL1, Y-TgL3, H-TgL1, H-TgL2, H-TgL3) with Cfh KO mice and were able to obtain Cfh transgenic mice on a homozygous Cfh KO background (CfhTg/mCfhKO) for all them. Western blot analysis was performed in our founder mice and in every subsequent generation including, as shown in Figures 3 and 4, after crossing to Cfh KO mice. Although there was some variation between lines regarding the level of protein expression, for each of our selected lines the level of expression remained stable compared with the founder mice with subsequent progeny. 

Restoration of Physiological Serum C3 Levels after Transgenic Cfh Expression

To assess the functionality of the chimeric Cfh proteins, we used transgenic mice deficient in mCfh. The only Cfh molecules in these mice are chimeric. Mice deficient in mCfh have an uncontrolled activation of fluid-phase C3 because of the “tick-over” pathway, in which spontaneous hydrolysis of C3 leads to the formation of C3 convertase, generation of C3b, and further activation of the complement cascade. Cfh blocks this process by inhibiting the formation and causing the inactivation of C3b. In the absence of Cfh, activation of the complement cascade leads to severe depletion of plasma C3 and other complement components. We would only be able to observe a rescue of plasma C3 levels in the CfhTg/mCfhKO mice if our chimeric Cfh molecules were functional in vivo. As seen in Figure 5, the level of C3 in B6 mice was approximately 600 μg/mL but was undetectable in Cfh KO mice. However, C3 returned to normal levels in the presence of the Cfh transgene. Moderate levels of chimeric Cfh expression (H-TgL1 line and some Y-TgL3 mice) led to significant, yet incomplete, C3 rescue. Thus, the chimeric Cfh proteins are functional in vivo, and in several lines they are expressed at levels that allow for full functional effect in preventing spontaneous complement consumption. 

Figure 5.

 

Chimeric Cfh molecules were functional in vivo, blocking the spontaneous activation and depletion of C3 in the serum. ELISA demonstrated that the serum levels of C3 in our transgenic mice were normal even in the absence of mouse Cfh. Note that C3 was completely depleted in mCfhKO mice. Serum samples at a 1:20,000 dilution were tested in duplicate. Results were reproduced in two separate experiments. Each of the four transgenic lines is represented using a different shade. The two H-CfhTg/mCfhKO lines are shown with cross-hatching.

Drusen-like Subretinal Yellow Deposits and Retinal Morphology of the Cfh Transgenic Mice

We examined the central retina (3–4 disc diameters around the disc) and peripheral retina of the CfhTg/mCfhKO mice at approximately 12 to 13 months of age and compared them with age-matched B6 mice. The transgenic mice had more subretinal, yellow, drusen-like deposits in the central retina than age-matched B6 and mCfhKO mice (Figs. 6A–D). However, peripheral retinal examination demonstrated that both the transgenic mice and the age-matched B6 mice had many similar yellow spots in the far retinal periphery (Figs. 6E–H). 

Figure 6.

 

Transgenic mice develop “drusen-like” subretinal yellow deposits in clinical examination by 12 months of age. Fundus images were obtained using a mouse fundus camera. Photographs were taken from B6 mice and a mCfhKO mouse (left) and from CfhTg/mCfhKO mice (right). Images correspond to (A) central retina of B6 mouse; (B, D) central retinas of H-TgL3/mCfhKO and H-TgL2/mCfhKO mice, respectively; (C) central retina of mCfhKO mouse; (E, G) peripheral retinas of B6 mice; (F) peripheral retina of a Y-TgL1/mCfhKO mouse; and (H) peripheral retina of a H-TgL2/mCfhKO mouse.

Histopathologic examination of the central retina was performed in B6 mice and CfhTg/mCfhKO mice at 14 months of age. Toluidine blue staining of plastic sections (Figs. 7A–C) revealed that the retinal structure in our CfhTg/mCfhKO mice was undisturbed. We consistently documented the presence of large subretinal cells in our transgenic mice (Figs. 7F–H). These cells had the morphologic appearance of macrophages. In the B6 mice (Figs. 7D–E), we could only find one cell at the photoreceptor-RPE interface in all the tissues examined, and it was not phenotypically similar to the cells seen in the transgenic mice (Fig. 7E). Furthermore, the CfhTg/mCfhKO mice had some areas showing a subtle increase in subretinal material (Figs. 7C, 7H [bracket], 7I [bracket]). 

Figure 7.

 

Transgenic mice accumulate subretinal infiltrating cells in the central retina. Toluidine blue–stained retinal sections of 14-month-old mice. (A–C) Full-thickness sections (32 × magnification) of (A) B6, (B) Y-TgL3/mCfhKO, and (C) H-TgL3/mCfhKO mice demonstrating that there is no significant degeneration of the retina detectable at the light microscopy level in the transgenic mice. (D–I) Toluidine blue–stained sections (63 × magnification) of the central retinas of (D, E) B6, (F) Y-TgL1/mCfhKO, and (G–I) H-TgL3/mCfhKO mice. (E) Only one infiltrating subretinal cell was seen in three B6 mice examined by histopathology. Note the dense nucleus and minimal agranular cytoplasm. On the other hand, all six transgenic mice examined by histology (three Y-Tg mice on a Cfh KO background and three H-Tg mice on a Cfh KO background) exhibited multiple subretinal cells with a phenotype suggestive of activated macrophages (F–H; asterisks). This phenotype is characterized by increased cytoplasm, increased cytoplasmic granules, and more irregular nuclear chromatin. The transgenic mouse specimens also had areas of thicker subretinal debris (H, I; brackets).

To study the subretinal cells further, we prepared RPE-choroid flat mounts from CfhTg/mCfhKO mice and B6 mice and performed immunostaining (Fig. 8) using an anti–Iba1 antibody (microglia/macrophage marker). This allowed us to detect Iba1+ cells over the RPE. The distribution of these cells was similar to that of the yellow spots in clinical examination. In other words, there were more of these cells in the posterior pole of CfhTg/mCfhKO mice than in age-matched B6 mice (Figs. 8A–F). Again, many Iba1+ cells were seen in the far periphery of both CfhTg/mCfhKO and B6 mice (Fig. 8A). Iba1 staining of paraffin sections confirmed the location of these cells as being in the subretinal space (Figs. 8G, 8H). Double immunostaining of RPE-choroid flat mounts from CfhTg/mCfhKO mice using anti–Iba1 and anti–F4:80 antibodies revealed that these subretinal cells were also positive for the macrophage marker F4:80 (Fig. 9). 

Figure 8.

 

Subretinal cells in the transgenic mice are macrophages/microglia. (A–F) RPE-choroid flat mounts stained with a rabbit anti–Iba1 primary antibody, followed by an AlexaFluor488-labeled goat anti–rabbit IgG antibody. (A) Very low magnification view of the entire flat mount of an H-TgL2/mCfhKO eye showing the distribution of cells. Bright spots are individual macrophages. (B, C) 20 × and 100 × magnification views of the macrophages in an H-TgL3/mCfhKO eye. (D, E) 10 × views of the central retina, showing a 1.3 × 1.7-mm rectangle centered on the nerve of a B6 eye and an H-TgL3/mCfhKO eye, respectively. Note the greater number of macrophages in the transgenic central retina. These rectangular views were generated by overlapping four or five images at 10 × magnification using the optic nerve (the large green spot in the center of the images) as a guide. (D) Saved views of the inferior retina did not overlap perfectly, accounting for the narrow white defect seen. White circles: artifacts caused by small debris on the sections. (F) Cells found in similar 1.3 × 1.7-mm rectangles of central retinas of two B6 and three CfhTg/mCfhKO mice were counted. (G) Paraffin-embedded retinal section (H-TgL3 mouse in B6 background is shown) stained with the same anti–Iba1 primary antibody. (H) Same image as (G), but analyzed with imaging software. This allows the visualization of the RPE and retina layers while also showing the staining of the subretinal cells. Brighter colors represent positive staining.

Figure 9.

 

Subretinal cells are F4:80 positive. RPE-choroid flat mounts were double stained with anti–Iba1 (A, green channel) and anti–F4:80 (B, Texas Red channel) antibodies. Y-TgL2/mCfhKO mouse is shown. Scale bars, 20 μm.

To determine the state of activation of the macrophages in transgenic mice, we isolated resting peritoneal cells from B6 mice compared with transgenic mice and plated them to enrich for macrophages. Using flow cytometry we found that there was no increase in the expression of major histocompatibility complex class II molecules in the transgenic/mCfhKO mice compared with B6 mice (data not shown). This suggested that the transgenic mice did not have a widespread activation of macrophages. 

Basal Laminar Deposits, Long-Spaced Collagen, and C3 Deposition in the Cfh Transgenic Mice

Ultrastructural analysis with a transmission electron microscope (Fig. 10) of 13- to 14-month-old CfhTg/mCfhKO mice showed thickening of the RPE basal infoldings/Bruch's membrane complex in the CfhTg/mCfhKO mice compared with B6 and mCfhKO mice (Figs. 10A–C; brackets). Lipofuscin-like granules were frequently seen within RPE cells in all CfhTg/mCfhKO mice (Figs. 10D, 10E, 10G, 10I; asterisks). These granules (less electron dense than melanin granules) were only occasionally seen in B6 mice. Finally, we were surprised to consistently find large patches of basal laminar deposits (electron-dense material) under the RPE in the CfhTg/mCfhKO mice (Figs. 10C–I) but not in age-matched B6 or mCfhKO mice. Long-spaced (wide-banded) collagen (Figs. 10D, 10G, 10H; arrowheads) was also documented in many of these areas in CfhTg/mCfhKO but not in B6 mice. The absence of basal laminar deposits and wide-banded collagen in our B6 and mCfhKO mice was consistent with findings by several other groups in B6 mice younger than 16 months of age and on a normal diet ^{31 –33} and in 2-year-old mCfhKO mice. ³⁴  

Figure 10.

 

Transgenic mice have thicker basal RPE infoldings and accumulate basal laminar deposits and long-spaced collagen. Electron microscopy images at 25,000 × magnification (except H, which was taken at 75,000×) of the RPE/Bruch's membrane of 13- to 14-month-old (A) B6, (B) mCfhKO, (C) H-TgL1 in B6 background, (D) Y-TgL1/mCfhKO, (E, F) Y-TgL3/mCfhKO, and (G–I) H-TgL3/mCfhKO mice. Note the absence of basal laminar deposits in B6 and mCfhKO mice. Transgenic mice, both on the B6 and the mCfhKO background, demonstrate basal laminar deposits of variable electron density between the RPE cell plasma membrane and the rest of Bruch's membrane. The transgenic mice also have increased thickness of basal RPE infoldings (A–C, brackets) and RPE granules suggestive of lipofuscin accumulation (D, E, G, I; asterisks). Long-spaced collagen (< and >) was not seen in B6 or KO mice but was easily and frequently found in CfhTg/mCfhKO mice.

We performed immunostaining in our paraffin-embedded tissue looking for C3d deposition. A band of C3d-positive staining was detected in the region of Bruch's membrane in both the transgenic mice and the B6 controls. The intensity of staining was similar in both groups of animals, but the band appeared to be more irregular and approximately 30% thicker in the transgenic mice than in controls (Fig. 11). This was the case both in Cfh transgenic mice that also expressed native mCfh (Fig. 11C) and in CfhTg/mCfhKO mice (Fig. 11D). 

Figure 11.

 

The sub-RPE band of C3d deposition is thicker in transgenic mice. Immunostaining for C3d on a B6 retina (A) and a transgenic retina (B). Sections obtained from three B6 eyes and three transgenic eyes on a B6 background were stained with an anti–C3d antibody followed by an AlexaFluor555-labeled rabbit anti–goat IgG secondary antibody. The area of the sub-RPE band of C3d staining was measured in 24 images for each group of mice using the ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html) software. Average ± SE for the area of the bands (in standard area units) in B6 versus transgenic samples was calculated (C). All images were captured at 100 × magnification. The same experiment was repeated comparing B6 mice, Y-TgL1/mCfhKO, and H-TgL2/mCfhKO mice (two mice in each group), and the results are summarized in (D). Differences between transgenic mice and B6 mice were statistically significant. *P < 0.01. **P < 0.05.

Discussion

A mouse model of AMD based on genetic variants that have been associated with the disease in humans would help us improve our understanding of the early pathogenetic events. We have generated chimeric Cfh transgenic mice with the aim to develop such a model. The goal is to reconstitute the important interactions among Cfh, Crp, glycosaminoglycans (GAGs), and C3 in the retina. Our chimeric Cfh molecules should recognize mouse cell surfaces (GAGs and sialic acid) through the mouse SCR9–20 and should recognize mouse C3b through its mouse SCR1–5 and SCR9–20. One of the main areas of potential interaction of Cfh with Crp localizes to SCR7, which is also where the Y402H variant position is located. The human SCR6–8 fragment should allow the chimeric molecules to interact with human CRP once the mice are crossed with human CRP transgenic mice. The SCR6–8 fragment is 181 amino acids long, and there is 55% identity and 70% homology with mouse SCR6–8. It should be noted that the regions linking SCR5 to SCR6 (the 4-amino acid linker TLKP) and SCR8 to SCR9 (the 3-amino acid linker IKS) are identical in mice and humans, which should allow for a nondisruptive swap of mouse and human SCR6–8 sequences. Recent work by Sjoberg et al. ³⁵ showed that a human SCR6–8 peptide behaved similarly to the full human CFH molecule with regard to the interaction with CRP and other molecules. This has been confirmed by recent crystallography studies. ³⁶  

Because the main source of Cfh is the liver, we chose to express the Cfh transgenes using a promoter of human ApoE which is constitutively expressed in the liver. RPE cells seem to express mRNA for Cfh. ¹¹ Yet it is unknown whether the protein is produced or secreted. ³⁷ Interestingly, to our advantage, the ApoE promoter is functional both in the liver and in the RPE. We have shown that our transgenic mice express the chimeric Cfh mRNA molecules in both the liver and the posterior segments of the eyes, similar to the expression pattern of endogenous Cfh. Furthermore, nuclease protection assay has shown that the level of transgenic chimeric mRNA expression in the liver is similar to the level of mouse endogenous Cfh mRNA expression. 

Western blot analysis was particularly important in our chimeric Cfh transgenic mice because successful expression of the mRNA does not confirm that the chimeric protein is produced or is stable. We showed that our mice maintain serum levels of transgenic Cfh proteins that are comparable to the levels of native Cfh reported in rats and humans. Furthermore, using a C3 ELISA, we have demonstrated that the chimeric proteins are functional in vivo and are present at levels that completely prevent the spontaneous activation and depletion of C3 seen in mCfhKO mice. 

By 12 months of age, our transgenic mice develop clinically visible drusen-like subretinal deposits. The only histopathologic findings that seem to correlate in terms of size, frequency, and location (deep to the retina) of these lesions are large subretinal cells that we have identified by immunohistochemistry as Iba1⁺, F4:80⁺ macrophages. These cells are more common in the central retina of CfhTg/mCfhKO mice than of B6 mice, but they are also present in the peripheral retinas of both transgenic and B6 mice. Similar cells have been described in other models of retinal degeneration. ^{38 –40} Although the significance of this finding is still unknown, it is interesting for several reasons. First, macrophages may be important players in the pathogenesis of AMD. Some studies suggest that they have a protective role in AMD, ^22,41,42 and others suggest that they promote disease progression. ^{42 –44} Second, subretinal or choroidal macrophages are seen in some recent mouse models of AMD in which the authors have manipulated genes directly involved in the migration of macrophages and microglial cells (e.g., Ccl2 and Cx3cr1). ^21,22 Instead, our model is based on the manipulation of the gene expressing Cfh, which is primarily involved in the regulation of complement activation and has been identified by many epidemiologic studies to be associated with human AMD. It is unclear whether one of the downstream consequences of Cfh manipulation may be the regulation of macrophage functions. This may occur indirectly because of, for example, increased accumulation of subretinal debris (Figs. 7H, 7I) leading to increased macrophage activation or differentiation. Finally, the size of these macrophages is approximately 15 μm, similar to the size of retinal vessels in mice, and their cytoplasmic content may give them a yellow appearance. It is interesting that their size would correspond to tiny “drusen” seen in AMD patients. Therefore, it is possible that we are often seeing these cells in AMD patients and not realizing it. The idea that “not all macrophages are made equal” and under different conditions may have opposing roles in AMD is very attractive. Our model may allow us to explore this further. 

We were particularly intrigued by the ultrastructural findings in our transgenic mice. Electron microscopy revealed that our transgenic mice accumulated basal laminar deposits and long-spaced collagen under the RPE cells; both findings are associated with early AMD. ^45,46 It should be noted that some experts believe that basal laminar deposits (along with the more external basal linear deposits) lead to soft drusen, pigment epithelial detachments, and choroidal neovascularization. ⁴⁵ There was also greater accumulation of lipofuscin granules in the Cfh transgenic mice than in the B6 mice. Finally, staining with anti–C3d antibodies revealed that the area of complement activation involving Bruch's membrane was thicker in the Cfh transgenic mice than in the B6 mice. 

This Cfh transgenic model demonstrates that a substitution of mouse Cfh domains SCR6–8 by the human sequence can lead to early AMD-like characteristics even in mice. The finding of basal laminar deposits, long-spaced collagen, and increased lipofuscin granules in these mice is interesting and not predictable. Even assuming an important role for the complement system in human AMD and in mouse retinal physiology, it is not clear to what extent other complement regulatory proteins in the mouse (e.g., Crry) may compensate for abnormalities in Cfh. Our model suggests that Cfh is important and not completely replaceable by other complement regulatory proteins in the regulation of complement activation in the mouse retina. A second issue is that, in contrast to mutations associated with retinal dystrophies, the Cfh variants are associated only with increased susceptibility to AMD in humans and do not directly cause AMD. Thus, we would be surprised to see spontaneous choroidal neovascularization or geographic atrophy develop in CfhTg/mCfhKO mice. However, we can envision how our chimeric variants could lead to the subtle changes, mostly visible on electron microscopy, in these mice. In a human scenario, these changes could predispose affected persons to further progression toward AMD. 

The next question was how we could link mechanistically the Cfh variants to the ultrastructural changes we were observing. In our mouse model, the chimeric Cfh molecules could function well in serum, preventing the spontaneous depletion of complement components (in contrast to the Cfh KO mouse model) and allowing the complement cascade to be available to act in the tissues. Thus, the mechanistic explanation should involve an impaired ability of the chimeric Cfh molecules to regulate complement activation at the of tissue surface level. This may be due to an altered affinity to Crp, GAGs, or both. 

Recent studies suggest that CRP plays an anti-inflammatory role in normal homeostasis. ⁴⁷ It may also play an anti-inflammatory role in the retinas of healthy persons by opsonizing subretinal debris and helping macrophages clear this debris in a noninflammatory fashion. This process may depend on the ability of CFH to bind CRP, preventing it from fully activating the last steps of the complement cascade. The “at-risk variant” of CFH appears to have a reduced affinity for CRP. ^{35,48 –50} Patients homozygous for this variant will still experience binding of CRP to the subretinal debris. However, given that their CFH molecules are less able to bind CRP, the complement-mediated proinflammatory role of CRP may override the opsonization-mediated anti-inflammatory role of CRP. This would result in a vicious circle of further accumulation of debris, further accumulation of CRP, further activation of complement, and further tissue damage. In support of this hypothesis, Johnson et al. ³⁷ recently demonstrated that persons homozygous for the at-risk variant of CFH have elevated levels of CRP in the choroid. Although this seems like an appealing hypothesis, it should be noted that Bíró et al. ⁵¹ and Hakobyan et al. ⁵² argue that CRP-CFH interactions are an artifact of in vitro testing and that the interaction requires the presence of denatured CRP. Refuting this, most recently Okemefuna et al. ^50,53 documented CFH-CRP interactions occurring in solution and at physiological CFH, CRP, Na+, and Ca++ concentrations using analytical ultracentrifugation, surface plasmon resonance, and synchrotron x-ray scattering. They claim that their conditions avoided the occurrence of denatured CRP. In their studies they also found that the interaction of CRP with the different fragments of factor H was strongest with the 402H variant of SCR6–8, followed by the 402Y variant of SCR6–8, followed by SCR16–20. Both the authors and the editorial comments to these articles ^50,53,54 discuss the notion that these findings may provide a mechanism for the known association of the CFH variants with AMD. Still, it is difficult to prove that in vivo CRP-CFH interactions are indeed relevant. 

Some groups suggest that the Y402H variant may instead affect the binding of CFH to GAGs and sialic acid and thus limit the ability of CFH to function in tissues. Of note, both CRP and GAGs bind CFH in at least two different binding sites. The correct orientation of the CFH molecule that, though bound to a tissue surface, would allow it to maximize the inactivation of C3 may depend on the combination of the CFH-CRP and the CFH-GAGs interactions at these multiple CFH-binding sites. In this context it was not surprising that both constructs (Y-Tg-Cfh and H-Tg-Cfh) led to similar clinical and histologic findings. The affinity between mouse Crp (or GAGs) and both of the chimeric Cfh variants is probably significantly altered compared with mouse Cfh. 

Finally, it is interesting that, although less striking, the ultrastructural changes we observed in the CfhTg/mCfhKO mice were also present in CfhTg mice that were still expressing mouse Cfh. The multiple binding sites model outlined could explain how chimeric Cfh molecules may compete with the mouse Cfh molecules from their binding sites to GAGs and sialic acid (mouse SCRs 9–20). The chimeric molecules, however, may then be unable to achieve the proper secondary interaction with either mouse Crp or GAGs because of the human (different from mouse) SCR6–8 sequence. This may prevent the bound chimeric Cfh molecules from orienting themselves properly (or potentially prevent them from undergoing changes in configuration) to maximize inhibitory interactions with complement fragments. 

Clearly, there is still much to be learned about the roles of CFH and CRP in AMD. It is known that Crp in mice does not have the acute-phase reactant properties characteristic of CRP in humans. We will cross our Cfh-transgenic mice with hCRP-transgenic mice expressing human CRP (these mice demonstrate CRP functions similar to those of CRP in humans). We will investigate whether the resultant mice have a stronger or earlier AMD-like phenotype and whether the phenotype is more prominent on the H-Tg lines. Whatever the results, this model may help us explore the relevance of Cfh-Crp and Cfh-GAGs interactions in vivo. 

Footnotes

 Supported by the Disease Oriented Clinical Scholars program at the UT Southwestern Medical Center, the Hawn Foundation, the Charles Y. C. Pak Foundation, and an unrestricted grant from Research to Prevent Blindness, Inc.

Footnotes

 Disclosure: R.L. Ufret-Vincenty, None; B. Aredo, None; X. Liu, None; A. McMahon, None; P.W. Chen, None; H. Sun, None; J.Y. Niederkorn, None; W. Kedzierski, None

The authors thank Marina Botto for providing the Cfh KO mice, Elizabeth Mayhew for help with the paraffin sections, Katherine Luby-Phelps and Abhijit Bugde for help with the fluorescence microscope, and Tom Januszewski for help with the electron microscope.