All chemicals were purchased from Sigma-Aldrich (Poole, UK) or Fisher Scientific (Loughborough, UK). Fatty acid and lipid standards were from Nu-Chek-Pre Inc. (Elysian, MN) and Sigma-Aldrich, respectively. Silica gel G plates were from Merck KGaA (Darmstadt, Germany).

CD55 knockout (CD55^−/−) mice were provided by Prof Wenchao Song (University of Philadelphia, Philadelphia, PA) and back-crossed onto C57BL/6 for nine generations. ApoE^−/− mice were originally provided by J. Breslow (Rockefeller University, New York, NY). The strain background of these original mice was 71% C57BL/6 and 29% 129. The apoE^−/− mice were crossed with CD55^−/− mice to generate apoE^−/−/CD55^−/− double knockouts along with apoE^−/− single knockouts; these sex-, strain-, and age-matched littermates provided the appropriate controls. Mice were genotyped by polymerase chain reaction with the use of genomic DNA extracted from tail tips.

Male mice aged 8 weeks were fed a high-fat diet, containing 21% (wt/wt) pork lard and supplemented with 0.15% (wt/wt) cholesterol (Special Diet Services, Witham, UK), for 12 weeks. Animals were housed in a specific pathogen-free environment. Some mice were deprived of food for 16 hours overnight to obtain baseline levels of various parameters. All studies and protocols were approved by the institutional ethics review committee and the United Kingdom Home Office and conformed to the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, revised 1996).

Mice were anesthetized by intraperitoneal injection of sodium pentobarbitone and weighed before exsanguination by arterial perfusion via the abdominal aorta with PBS at a constant pressure of 100 mmHg, with outflow through the incised jugular veins. Brachiocephalic arteries were removed with a piece of the aortic arch and the stump of the right subclavian artery still attached to aid orientation during processing, immediately embedded in optimum cutting temperature compound (RA Lamb Ltd, Eastbourne, UK), and snap-frozen in liquid N₂.

Serial transverse 7-μm sections were cut along the brachiocephalic artery, starting from the proximal end. Sections were stained with Miller's Elastic/Van Gieson (Sigma-Aldrich). Macrophages and smooth muscle cells were identified with anti-murine macrophage mAb (diluted 1:100; F4/80; Serotec, Oxford, UK) and anti–α-smooth muscle actin mAb (diluted 1:100; clone α−1-A4; Sigma-Aldrich), respectively. Sections were fixed in ice-cold acetone and blocked with either avidin/biotin blocking kit (Vector Laboratories, Peterborough, UK) followed by 10% goat serum or Mouse-on-Mouse kit (Vector Laboratories). Blocked sections were incubated with appropriate biotinylated secondary antibodies: goat anti-rat Ig (Vector Laboratories; 3.5 μg/mL in 10% mouse serum) or anti-mouse Ig diluted as directed (Mouse-on-Mouse kit). Staining was developed with Fluorescein-Avidin D (diluted 1:200 in 2% bovine serum albumin in PBS; Vector Laboratories), and cell nuclei were counterstained with DAPI (Sigma-Aldrich).

Immunostaining for C activation used either rat anti-mouse C3b/iC3b mAb clone 2/11 (5 μg/mL; Hycult Biotech) or affinity-purified rabbit anti-rat/mouse C9 generated in house, proven reactive with MAC in mouse tissues (2 μg/mL).²³ For C3 and C9 staining, sections were fixed in acetone at 4°C, blocked in 2% bovine serum albumin in PBS, and, after staining with primary antibody, developed with either Alexa Fluor 488–labeled goat anti-rat (20 μg/mL; Invitrogen, Carlsbad, CA) or Alexa Fluor 594–labeled goat anti-rabbit IgG (20 μg/mL; Molecular Probes, Eugene, OR) respectively. Nuclei were counterstained with DAPI.

Negative controls included replacement of primary antibody with IgG isotype control. Staining was expressed as the percentage of lesion area staining positive, assessed by computerized image analysis (Image ProPlus 4.0; Media Cybernetics, Carlsbad, CA).

Five sections were taken per mouse at the same relative positions along the brachiocephalic artery and were assessed for the presence of plaque with the use of an established method.²⁵ Plaque area was calculated with image analysis as above.

Mice were sacrificed between 9 AM and 11 AM, blood (1 mL) was collected into tubes with or without EDTA, and serum or plasma was separated by centrifugation. Triglyceride and cholesterol levels were measured at the Clinical Biochemistry Laboratories, University Hospital Cardiff, on an Aeroset automated analyzer (Abbott Diagnostics, Berkshire, UK). For NEFAs, lipids were extracted and separated by one-dimensional thin-layer chromatography on 10 × 10-cm silica gel G plates, double developed with toluene/hexane/formic acid (140:60:1, v/v/v) for the entire plate, followed by hexane/diethyl ether/formic acid (60:40:1, v/v/v) to half height. Plates were sprayed with 0.05% (wt/v) 8-anilino-4-naphthosulphonic acid in methanol and viewed under UV light to show lipids. Free fatty acids were scraped from the plate and were identified and quantified by gas chromatography.

C3adesArg was measured in a sandwich ELISA with the use of a pair of anti-mouse C3a mAbs, one unlabeled as capture (0.2 μg/mL), the other biotinylated as detection (0.5 μg/mL), and recombinant mouse C3a (100 to 0.78 ng/mL) as standards (all from BD Pharmingen, San Diego, CA). Appropriately diluted plasma samples were included. The assay was developed with streptavidin-peroxidase (1:5000; Jackson ImmunoResearch, West Grove, PA).

Serum C3 levels were measured by ELISA essentially as previously described,²⁷ except that rat anti-mouse C3 (2 μg/mL; clone 11H9; Hycult Biotech) was used as the capture antibody. A standard curve of known concentrations, starting from 0.5 μg/mL, was produced with the use of purified mouse C3 (a kind gift from Dr Claire Harris; Cardiff University).

Percentage of body fat was measured by dual-energy X-ray absorptiometry scanning of whole animals with the use of a PIXImus scanner (Lunar Corp, Madison, WI) with small animal software.

Data are expressed as mean ± SEM, and significance was tested by two-tailed unpaired Student's t-test (GraphPad Prism software, version 3.0; GraphPad Software Inc., San Diego, CA), with significance assumed at P < 0.05.

Matched apoE^−/− and apoE^−/−/CD55^−/− mice were sacrificed at 20 weeks of age after 12 weeks on a high-fat diet, and the extent of atherosclerosis was assessed in the brachiocephalic arteries, a known site of predilection for plaque development.²⁸ Plaque cross-sectional area, assessed at multiple sites along the vessel, was reduced threefold in apoE^−/−/CD55^−/− mice compared with apoE^−/− controls (54.7 ± 11.2 × 10³ μm² versus 155.2 ± 16.8 × 10³ μm²; P < 0.001; Figure 1, A–C). Plaque stage and complexity were further explored by measuring smooth muscle cell content and macrophage infiltration; smooth muscle cells as a proportion of total cell number in apoE^−/−/CD55^−/− plaques were significantly lower than in apoE^−/− plaques (9.3% ± 2.0% versus 18.0% ± 2.8%; P < 0.05; Figure 1, D–F). Plaque macrophage content was similar in the groups (19.4% ± 6.5% versus 18.8% ± 4.2%; Figure 1, G–I).

To address whether CD55 deficiency influenced local C activation, plaques were stained for C3 fragments and MAC. C3 fragment deposition was assessed with mAb 3/26, a neoepitope-specific mAb that specifically detects C3b, iC3b, and C3c in tissues, whereas MAC was detected with affinity-purified anti-rat/mouse C9. Percentages of plaque area stained for C3b/iC3b/C3c and MAC were twofold reduced in plaques from apoE^−/−/CD55^−/− mice compared with apoE^−/− controls (C3b/iC3b/C3c: 28.0% ± 8.1% versus 57.3% ± 7.7%; P < 0.05; Figure 1, J–L; MAC: 17.9% ± 3.5% versus 30.5% ± 4.0%; P < 0.05; Figure 1, M–O).

To test whether the absence of CD55 affected lipid handling, lipid levels were measured in apoE^−/−/CD55^−/− mice and in apoE^−/− controls at 8 weeks old on normal diet and at 20 weeks old after 12 weeks on a high-fat diet. At 8 weeks, triglyceride levels were markedly reduced in the apoE^−/−/CD55^−/− mice than in the apoE^−/− controls (2.1 ± 0.1 mmol/L versus 5.1 ± 0.6 mmol/L; P < 0.01; Figure 2A); cholesterol levels were not significantly different between these groups (16.1 ± 1.4 mmol/L versus 18.2 ± 0.8 mmol/L; Figure 2B). After 12 weeks of fat feeding, triglyceride levels were little changed, as expected in the apoE^−/− model,²⁹ and remained significantly lower in the apoE^−/−/CD55^−/− mice compared with the apoE^−/− controls (1.7 ± 0.2 mmol/L versus 3.6 ± 0.5 mmol/L; P < 0.01; Figure 2A). Cholesterol levels were increased after 12 weeks of fat feeding in both groups but were significantly lower in the apoE^−/−/CD55^−/− mice than in the apoE^−/− controls (29.0 ± 1.8 mmol/L versus 38.3 ± 2.5 mmol/L; P < 0.01; Figure 2B). Plasma levels of NEFAs were measured in mice before and after being fed the high-fat diet. Significant increases in NEFA concentrations were seen in both groups after 12 weeks of a high-fat diet (apoE^−/−/CD55^−/−: 0.153 ± 0.016 mg/mL versus 0.333 ± 0.041 mg/mL before and after the high-fat diet, respectively; P < 0.001; apoE^−/−: 0.121 ± 0.006 mg/mL versus 0.505 ± 0.046 before and after the high-fat diet, respectively; P < 0.001; Figure 2C). Although no significant differences were observed in NEFA levels between apoE^−/−CD55^−/− and apoE^−/− mice before fat feeding, levels were significantly lower in apoE^−/−/CD55^−/− mice than in apoE^−/− controls after 12 weeks on a high-fat diet (0.333 ± 0.041 mg/mL versus 0.505 ± 0.046 mg/mL; apoE^−/−/CD55^−/− and apoE^−/−, respectively; P < 0.01; Figure 2C).

C3adesArg, also known as ASP, is a stable product of C3 activation and a potent adipokine that stimulates uptake of triglycerides and NEFAs, enhances triglyceride synthesis and storage, and inhibits triglyceride lipolysis in adipose tissue.^30–32 Circulating nonfasting C3 and its activation product C3adesArg (Figure 2, D and E, respectively) were measured before and after the high-fat feeding in apoE^−/−/CD55^−/− and apoE^−/− mice. C3 levels were significantly reduced in both apoE^−/−/CD55^−/− and apoE^−/− mice after 12 weeks of high-fat feeding compared with before the high-fat feeding values (apoE^−/−/CD55^−/−: 0.42 ± 0.03 mg/mL versus 0.3 ± 0.04 mg/mL; P < 0.01, respectively; apoE^−/− 0.48 ± 0.05 mg/mL versus 0.22 ± 0.05 mg/mL; P < 0.05, respectively). C3 levels were not significantly different between apoE^−/−/CD55^−/− and apoE^−/− mice at either time point (Figure 2D). Plasma C3adesArg levels were markedly lower after being deprived of food overnight compared with not being deprived of food regardless of age, diet, and genotype of the mice (compare Figure 2, E with F). Fasting C3a levels were similar in all groups of mice (Figure 2F); in contrast, nonfasting C3adesArg levels were significantly higher in apoE^−/−/CD55^−/− mice than in apoE^−/− mice but only on the high-fat diet (Figure 2E; 1.58 ± 0.14 μg/mL versus 1.00 ± 0.16 μg/mL; P < 0.01). The data show that high-fat feeding in apoE^−/−/CD55^−/− mice is associated with increased C3 turnover and higher circulating levels of C3adesArg compared with apoE^−/− controls.

To test whether the observed changes in circulating lipid levels and C3 activation in high fat–fed apoE^−/−/CD55^−/− mice affected fat storage, body weight and composition were compared between the groups. The high-fat diet caused significant increases in body weight for both groups, without any significant differences between apoE^−/−/CD55^−/− mice and apoE^−/− controls (apoE^−/−/CD55^−/−: 27.32 ± 0.6 g versus 38.16 ± 2.12 g; P < 0.0001; apoE^−/−: 30.64 ± 0.39 g versus 41.38 ± 1.17 g; P < 0.0001).

Both adipose mass and percentage of body fat were measured with dual-energy X-ray absorptiometry (Figure 3, A and B, respectively). After 12 weeks of high-fat feeding CD55 deficiency was associated with increased adipose tissue mass and percentage of body fat, significant in the latter case, in apoE^−/−/CD55^−/− mice compared with apoE^−/− controls (Figure 3B; before high-fat diet: 17.43% ± 0.64%; 17.8% ± 1.11% apoE^−/−/CD55^−/− versus apoE^−/−, respectively; after high-fat diet: 28.63% ± 4.18%; 19.54% ± 1.77% apoE^−/−/CD55^−/− versus apoE^−/−, respectively; P < 0.05).