The rare C9 P167S risk variant for age-related macular degeneration increases polymerization of the terminal component of the complement cascade

Abstract

Age-related macular degeneration (AMD) is a complex neurodegenerative eye disease with behavioral and genetic etiology and is the leading cause of irreversible vision loss among elderly Caucasians. Functionally significant genetic variants in the alternative pathway of complement have been strongly linked to disease. More recently, a rare variant in the terminal pathway of complement has been associated with increased risk, Complement component 9 (C9) P167S. To assess the functional consequence of this variant, C9 levels were measured in two independent cohorts of AMD patients. In both cohorts, it was demonstrated that the P167S variant was associated with low C9 plasma levels. Further analysis showed that patients with advanced AMD had elevated sC5b-9 compared to those with non-advanced AMD, although this was not associated with the P167S polymorphism. Electron microscopy of membrane attack complexes (MACs) generated using recombinantly produced wild type or P167S C9 demonstrated identical MAC ring structures. In functional assays, the P167S variant displayed a higher propensity to polymerize and a small increase in its ability to induce hemolysis of sheep erythrocytes when added to C9-depleted serum. The demonstration that this C9 P167S AMD risk polymorphism displays increased polymerization and functional activity provides a rationale for the gene therapy trials of sCD59 to inhibit the terminal pathway of complement in AMD that are underway.

Figure 1

Plasma C9 levels in the North American cohort. (a) Plasma levels by P167S status irrespective of phenotype. The median C9 plasma levels were as follows: P167S, 14.7 μg/ml; no variant, 23.5 μg/ml; (b) C9 plasma levels by P167S variant and AAMD status. The median C9 plasma levels were as follows: AAMD with P167S, 16 μg/ml; non-AAMD with P167S, 13.9 μg/ml; AAMD without variant, 25 μg/ml and non-AAMD without variant, 21.9 μg/ml. Statistics shown include comparison of the median by a Mann–Whitney test (a) and Dunn’s multiple comparisons test (b). Interquartile range and median are shown by bars. Statistically significant results are indicated by (***) or (****). Defined as ^****P < 0.0001, ^***P < 0.001.

Figure 2

Plasma C9 levels in the UK cohort. (a) Plasma levels by P167S status irrespective of phenotype. The median C9 plasma levels were as follows: P167S, 19.7 μg/ml; no variant, 24.9 μg/ml; (b) C9 plasma levels by P167S variant and AAMD status. The median C9 plasma levels were as follows: AMD with P167S, 19.7 μg/ml; Control with P167S, 19.2 μg/ml; AMD without variant, 24.9 μg/ml; and control without variant, 25.2 μg/ml. Statistics shown include comparison of the median by a Mann–Whitney test (a) and Dunn’s multiple comparisons test (b). Interquartile range and median are shown by bars. Statistically significant results are indicated by (*). Defined as ^*P < 0.05.

Figure 3

Plasma sC5b-9 levels from North American cohort. (a) sC5b-9 plasma levels by AAMD status irrespective of genotype. The median C9 plasma levels were as follows: AAMD, 149.3 ng/ml; non-AAMD, 122.4 ng/ml. (b) sC5b-9 plasma by P167S variant and AAMD status. The median sC5b-9 plasma levels were as follows: AAMD with P167S, 146.7 ng/ml; non-AAMD with P167S, 116.5 ng/ml; AAMD without variant, 151.4 ng/ml; and non-AAMD without variant, 129.1 ng/ml. (c) Plasma sC5b-9 levels by P167S status irrespective of phenotype. The median sC5b-9 plasma levels were as follows: P167S, 149 ng/ml; no variant, 151.4 ng/ml. Statistics shown include comparison of the median by a Mann–Whitney test (a, c) and Dunn’s multiple comparisons test (b). Median with interquartile range is shown by bars. Statistically significant results are indicated by (*), (**) or ***). Defined as ^*P < 0.05, ^**P < 0.01 and ^***P < 0.001.

Figure 4

Expression and purification of recombinant C9 proteins. (a) Multiple independent transient transfections of CHO cell lines with the pDR2ΔEF1α vector containing WT (n = 6) or P167S variant (n = 6) C9 were performed. The C9 level in cell supernatants was measured, and there was a significantly lower C9 level of the P167S compared to WT (5 vs. 22.7 ng/ml, P < 0.0001). Mean with 95% confidence interval is shown by bars. (b) Stable transfections of recombinant WT and P167S CHO lines were produced, and C9 was purified by affinity chromatography and size exclusion prior to non-reduced SDS PAGE and Coomassie staining to assess purity. ^****P < 0.0001.

Figure 5

Negative staining EM of WT, P167S and 3:1 WT:P167S mix membrane attack complex formation. Representative negative stain images of MAC containing either WT, P167S or a 3:1 WT:P167S ratio mix of C9 assembled on DOPC:DOPE monolayers. Examples of oligomeric MAC complexes are indicated on each image (black arrow). Scale bar, 25 nm.

Figure 6

Hemolytic activity of recombinant C9 proteins. Sensitized SRBCs were incubated with C9-depleted serum reconstituted with recombinant C9 WT or P167S. Multiple independent assays were performed (n = 8) in triplicate and averaged. CH50 was 4.64% (or 2.62 μg/ml) for WT and 3.72% (or 2.05 μg/ml) for P167S. Statistically significant results are indicated by (*).

Figure 7

Lytic activity of WT and a 3:1 WT:P167S ratio. Guinea pig erythrocytes were incubated serially with C5b6, C7 and C8 followed by either WT C9 or a 3:1 ratio of WT/P167S. Multiple independent assays were performed (n = 5) in triplicate and averaged. No statistical significance was demonstrated.

Figure 8

Polymerization of recombinant C9 proteins. (a) Freshly purified WT and P167S C9 were incubated at 37°C 1 h in the presence or absence of EDTA. The amount of polymerized C9 was quantified by in house ELISA. The P167S variant polymerized to a greater extent than WT protein and was more resistant to EDTA. (b) Confirmation of polymerization was also assessed with western blotting of freshly purified WT and P167S, incubated at 37°C for 1 h in the presence or absence of EDTA. High molecular weight species was only detected with the P167S variant.

Figure 9

Surface plasmon resonance analysis of polymerization. Preparations of C9 were polished by size exchange chromatography to remove aggregates and flowed slowly (5 μl/min) at 37°C across a Biacore chip that had been immobilized with aE11 mAb specific for polymerized C9. Polymers of C9 formed by WT and P167S C9 were captured on the antibody. Solid line represents poly-C9 captured from the preparation of mutant P167S C9 and dashed line represents capture of poly-C9 in the WT preparation. The experiment was performed multiple times with different preparations of C9; a representative analysis is illustrated.

Figure 10

Localization of P167 on C9. (a) Structure of MAC and soluble C9. Only one of the 18 C9 in MAC is shown for clarity. All subunits are represented as surfaces and colored individually. The position of P167 (red) is highlighted on the surface of C9 (white). (b) Ribbon representation of soluble (top) and MAC inserted (bottom) C9. P167 side chain is shown as red spheres.