Activation of the Alternative Complement Pathway in Vitreous is Controlled by Genetics in Age-Related Macular Degeneration

Abstract

Purpose.: To determine if the progression of age-related macular degeneration (AMD) is associated with complement activation in the eye.

Methods.: Immunohistochemistry and ELISAs were used to determine the distribution, concentration, and activation of the alternative pathway complement proteases factor B (FB) and factor D (FD) and the central complement protein C3 in genotyped human postmortem donor eyes graded as having no or minimal drusen (category 1; controls), large drusen (category 3), and large drusen with advanced AMD (category 4).

Results.: C3, FB, and FD were present in vitreous and Bruch's membrane choroid (BM/C) interface of the macula of eyes in all tested AMD severity categories (n = 100). C3, FB, and FD were predominantly located to the choroidal vasculature and Bruch's membrane and, together with the serum proteins transferrin and albumin, elevated in BM/C extracts of category 4 eyes (n = 23) compared with category 1 eyes (n = 24). A significant increase in FB activation was found only in vitreous of category 4 eyes (n = 23) compared with category 1 eyes (n = 25). Genetic variants of complement factor H (CFH), C3, C2, and FB associated with increased risk of AMD were correlated with alternative pathway complement activation in vitreous, but not with complement proteins in BM/C protein extracts

Conclusions.: Increased activation of the alternative complement pathway in vitreous was controlled by disease stage and genetic variation in the complement pathway, supporting a role for complement activation in AMD disease pathogenesis.

Introduction

Progressive loss of central vision due to age-related macular degeneration (AMD) occurs when aging, genetic predisposition, and environmental insults interact to trigger a succession of degenerative changes in the macula, beginning with accumulation of drusen and leading to geographic atrophy or choroidal neovascularization. ^1–3 Inhibition of vascular endothelial growth factor A has emerged as an effective therapeutic approach in patients with neovascular AMD. ^4–7 There are, however, no effective treatments that can arrest the progression of early AMD or prevent the development of atrophic changes in the macula. 

Recent genetic and pathophysiological studies have provided evidence that the progression of AMD may be due in part to an inflammatory state sustained by aberrant activity of the alternative pathway of complement. ^8–12 The initial step is proteolysis of complement C3, which plays a central role in the activation of all complement pathways (Fig. 1). ^13,14 While the scaffold protein complement component 3 (C3) and serine proteases factor B (FB) and factor D (FD) are required and sufficient for formation of the alternative pathway convertase, several complement proteins, including complement factor H (CFH), a major inhibitor of the alternative complement pathway, act as negative regulators of this enzyme complex. 

Figure 1. 

 

Alternative pathway amplification. (A) Central component C3 is cleaved by an active C3 convertase to generate C3a and C3b, which results in unfolding of the thioester domain (blue hexagon) and formation of an ester bond (red dot) with a cell surface; (B) C3b binds complement FB; (C) FB is subsequently cleaved to Ba and Bb by complement FD to form the C3bBb C3 convertase. This convertase can cleave additional C3 molecules and establish a positive feed-forward loop of C3 convertase formation and amplification of the alternative complement pathway (dotted line). Adapted with permission from Gros P. In self-defense. Nat Struct Mol Biol. 2011;18:401–402. Copyright 2011 Macmillan Publishers Ltd.

Common genetic variants in complement genes C3, CFH, and FB are associated with AMD risk and progression. ^{11,12,15–21} A single nucleotide polymorphism (SNP) in the promoter of HtrA serine peptidase 1 (HtrA1) in a chromosomal locus that also contains ARMS2 is also a major genetic risk factor for AMD. ²² Elevated plasma levels of some key proteins of the alternative pathway, including FD, FB, and activation fragments C3a, FBa, and FBb, have been reported in patients with AMD. ^12,23,24 To what extent complement proteins and activation products are elevated locally in the eye is unknown. 

In this study, we used a collection of well-characterized human donor eyes to correlate AMD pathology with complement genotypes and complement protein concentration. Our findings showed that AMD disease severity and complement genotypes are associated with complement activation in the eye. 

Methods

Human Ocular Tissues and Plasma Samples

Donor eyes were obtained from the National Disease Research Interchange (Philadelphia, PA). Each eye was snap-frozen in liquid nitrogen and stored at −80°C until needed for dissection, grading, and sample collection. Cryopreservation took place between 1.5 and 7.0 hours of death. Donor eyes (n = 100), were graded as having no or minimal drusen (category 1 [controls]), large drusen (category 3), and large drusen with advanced AMD (category 4) at the Cleveland Clinic according to the Minnesota Grading System for postmortem human eyes, based on the Age-Related Eye Disease Study grading system (Table 1). ²⁵ Grading was performed under a dissecting microscope after removal of the anterior segment, vitreous, and retina. Donor eyes with grade 2 disease were not identified. The eye dissection technique is described in Yuan et al., ²⁶ and further detailed in the online supplement of this publication. Briefly, each eye was transected behind the limbus to remove the anterior segment. The vitreous was then poured from the posterior eye cup into a 15-mL plastic tube and capped. The retina was cut at the optic nerve to free it from the globe. To eliminate remnants of the RPE, the inner wall of the posterior pole was filled with PBS and the complete RPE/Bruch's membrane was gently brushed with a soft artist bristle brush to systematically eliminate any remaining RPE. Drusen, when present on the surface of Bruch's membrane, were not dislodged by the brushing technique. With fine forceps, the choroid/Bruch's membrane lamina was stripped from the sclera at the level of the lamina fusca. After cutting the connection of the choroid at the optic nerve head, Bruch's membranes were dissected with attached choroidal layers (BM/C) from the macula and laid flat on a wax substratum (Fig. 2). Drusen were recognized using a Zeiss stereo-microscope (Figs. 2A–C). Circular discs (4-mm diameter) were punched from the macula area (Figs. 2D, 2E) and used for further analysis. We determined the tissue localization and concentration of complement precursor proteins C3, FB, and FD and their proteolytic products, as they are sufficient and required to initiate complement activation via the alternative pathway. ¹⁴ In addition, we analyzed protein C4 that is required for activation of the mannose-binding lectin and classical complement pathway. Eyes used for immunohistochemistry on retina, RPE, and BM/C were obtained from the Lyon's Eye Institute, Tampa, Florida. All tissue was acquired with consent of the donor or donor family in accordance with the principles outlined in the Declaration of Helsinki. Investigators identified the eyes only by eye bank number to ensure confidentiality. 

Figure 2. 

 

Dissection of macular samples for detection of complement proteins. Drusen is not visible in a category 1 donor eye (A), whereas extensive macular drusen are seen in a category 3 eye (arrows in [B] and close-up in [C]. A sampling of 4-mm trephine discs from the macular area (D, E). The illumination conditions and lens filters caused the melanin in (D, E) to appear green and not the usual brown/black appearance. I, Inferior; M, macula; N, nasal; ON, optic nerve head; S, superior; T, temporal. Scale bar in (C) = 100 μm; (E) = 4 mm (same magnification for [A, B, D, E]).

Table 1. 

 

Demographic Characteristics of the Donor Population and Ocular Status with Respect to AMD

Characteristic	AMD Category (Minnesota Grading System) 25
	Category 1*: No or Few Macula Drusen (n = 30†)	Category 3: Intermediate AMD (n = 45‡)	Category 4: Macular Drusen and End-Stage Wet or Dry AMD
			Geographic Atrophy (n = 5§)	End-Stage Neovascular AMD (n = 20§)
Male/Female, n	15/15	23/22	4/1	8/12
Age, y
Mean	85.2	86.3	89.2	83.5
Median	87	89	91	84
SD	8.2	7.5	8.9	7.2
Range	62–99	68–96	76–97	68–95

 Samples excluded from statistical analysis due to high albumin to transferrin ratios or sample handling:

*  Controls.

†  Category 1 BM/C samples: 6 females; Vitreous samples: 2 females and 3 males.

‡  Category 3 BM/C samples: 3 females and 2 males; Vitreous samples: 1 female and 1 male.

§  Category 4 BM/C samples: 2 males (1 with geographic atrophy and 1 with end-stage neovascular AMD); Vitreous samples: 2 males (1 with geographic atrophy and 1 with end-stage neovascular AMD).

Complement Protein Assays

Protein extraction of BM/C trephine discs and Western blot analysis were performed using standard techniques (see Supplementary Material for details). Treatment of BM/C with 1% SDS did not increase C3 proteins in the soluble fraction by Western blot analysis, confirming that all complement proteins were extracted from the BM/C preparation (see Supplementary Material and Supplementary Fig. S1). 

ELISA parameters are summarized in Supplementary Table S1 (see Supplementary Material and Supplementary Table S1). Complement protein standards were obtained commercially (Complement Technology, Inc., Tyler, TX). Test sample (BM/C lysates, vitreous fluid, and plasma) and standard curve dilutions were prepared in PBS supplemented with 0.5% bovine serum albumin (BSA; Equitech-Bio Inc., Kerrville, TX) and 0.5% Tween-20 (Sigma, Atlanta, GA). The capture antibodies for each assay (see Supplementary Material and Supplementary Table S1) were coated on 384-well high-bind ELISA plates (Greiner Bio-One, Monroe, NC) overnight at 4°C. Plates were washed 3 times with 0.05% Tween-20 in PBS, blocked for 1 to 2 hours with 0.5% BSA, and washed. Test samples and controls were added to the preblocked plates and incubated for 2 hours. After the unbound antigens were washed away, detection antibodies (biotinylated or horseradish peroxidase [HRP] conjugated) were added and incubated for 1 to 2 hours (see Supplementary Material and Supplementary Table S2). For biotinylated antibodies, streptavidin-HRP (Amersham Pharmacia Biotech, Piscataway, NJ) diluted 1/10,000 in 0.5% BSA/0.5% Tween-20 in PBS was added to the washed plates. Following 30-minute incubation and a final wash, tetramethyl benzidine (Kirkegaard & Perry Laboratories, Gaithersburg, MD) was added and the color was developed for 5 to 7 minutes. The reaction was stopped by adding 1 M phosphoric acid. The optical density was measured using a microplate reader (450 nm, 650 nm reference) and the sample concentrations were calculated from 4-parameter fits of the standard curves. The C3b ELISA (recognizes C3b, iC3b, and C3c) was performed as described by Katschke et al.²⁷ The ELISA analyses of C3a des Arg (hereafter named C3a) and C4a des Arg (hereafter named C4a) (BD Biosciences, San Diego, CA), as well as albumin, transferrin, and human IgG (Bethyl Laboratories, Montgomery, TX) were performed as per manufacturer's instructions. The minimum quantifiable concentrations for assays developed in-house are listed in the online supplement (see Supplementary Material and Supplementary Table S2). We acknowledge the limitations of analyzing complement proteins in tissues that potentially have been affected by postmortem processes.²⁵ To avoid overestimation of complement activation due to serum contamination, we excluded samples with an albumin:transferrin ratio greater than 13 and vitreous samples with albumin:transferrin ratio greater than 9 (see Supplementary Material and Supplementary Table S2) since these samples are more likely to have been contaminated with plasma proteins.²⁸ The sample analysis of complement proteins was done at 2 different time-points: phase 1 and phase 2 (see Supplementary Material and Supplementary Tables S3 and S4; see also under Statistical Analysis of the Materials and Methods section). 

Immunohistochemical Analysis and Drusen Evaluation

Frozen 5-μm BM/C sections were air-dried, fixed in acetone, air-dried again, and stored with desiccant at −80°C. For analysis, the sections were rehydrated in PBS. BM/C sections were pretreated with avidin followed by biotin according to the manufacturer's instructions (Vector Laboratories, Burlingame, CA), rinsed with 2 changes of tris buffered saline with Tween (TBST) for 5 minutes each, and blocked with 10% horse serum in 3% BSA/PBS for 30 minutes at room temperature. Sections were incubated with the relevant primary antibody: anti-FD and anti-FB (Genentech, Inc.), anti-C3c (clone A205, Quidel Corporation, San Diego, CA), anti-C4c (clone A703, Quidel Corporation) or isotype control antibodies (BD Pharmingen), rinsed with two changes of TBST for 5 minutes each, incubated with biotinylated horse anti-mouse secondary antibody at 2.5 μg/mL for 30 minutes at room temperature, and rinsed again with two changes of TBST for 5 minutes each. Sections were stained with Vector Red (Vector Laboratories, Burlingame, CA) per manufacturer's instructions, counterstained with Mayer's hematoxylin, rinsed in water, crystal mounted, and cover slipped. Chinese hamster ovary cell pellets (see Supplementary Material and Supplementary Fig. S2) and isotype-matched antibodies (results not shown) were used as control. The number of complement-positive and -negative drusen was quantified in immunohistochemically stained serial sections from BM/C. Twenty drusen were counted in at least four different microscopic fields in sections obtained from 21 to 25 donors. Eyes from a donor (67 years, female) diagnosed with category 3 AMD was fixed in Davidson's fixative (Electron Microscopy Sciences, Hatfield, PA) within 5 hours of death. After 24 hours, eyes were immersed in 70% ethanol and processed for paraffin embedding and sectioning. For immunohistochemical analysis, sections were de-paraffinized and hydrated in distilled water. Subsequently, sections were incubated in Dako Target Retrieval Solution (Dako, Carpinteria, CA) at 65°C with temperature raised to 99°C for 20 minutes followed by a 20-minute cool-down (∼74°C). Sections were incubated with a monoclonal antibody to FB (MCA2647, AbD Serotec, Raleigh, NC) or isotype control antibody (BD Pharmingen, San Diego, CA) diluted in blocking serum for 60 minutes at room temperature. 

Analysis of Single Nucleotide Polymorphisms

Genomic DNA (5–10 ng/mL) was isolated from frozen sclera using DNeasy Blood & Tissue Kit (QIAGEN, Germantown, MD). DNA samples were genotyped for 6 SNPs ^{15–18,20,29–31} : (1) HtrA 1 (5′) rs11200638 in the HtrA serine peptidase 1 gene on chromosome 10, change G > A; (2) CFH (Y402H) rs1061170 in the CFH gene on chromosome 1, change T > C; (3) C3 (R102G) rs2230199 in the complement component C3 gene on chromosome 19, change C > G; (5) FB (R32Q) rs641153 in the FB gene, change C > T; (6) C2 (E318D) rs9332739 in the C2 gene, change G > C. 

Genotyping was performed using the iPLEX platform (Sequenom, San Diego, CA), following the manufacturer's protocol. The genotyping procedure consisted of PCR amplification of DNA fragments containing the target SNPs, dephosphorylation of any unused deoxyribonucleotide triphosphates (dNTPs) with shrimp alkaline phosphatase, allele-specific single base primer extension, and detection of the primer extension products with matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Selected SNPs were sequence-confirmed by Polymorphic DNA Technologies (Alameda, CA). 

Statistical Analysis

A genetic risk score (gene score) for each sample with genotype data was calculated by summing the number of risk alleles carried by each individual weighted by the log of the odds ratio. The overall score was based on published odds ratios for SNPs in the complement pathway (CFH Y402H, C2, IVS10, CFB R32Q, C3 R102G; complement gene score) or combined SNPs in the complement proteins and in the HtrA1/ARMS2 promoter (rs10490924) for each donor ³² (multigene score). 

The associations among complement gene score, AMD category, and log complement protein concentration were evaluated by separate linear regression and ANOVA models. The sample set used for statistical analysis (BM/C n = 87; vitreous n = 91) did not include samples with high albumin-to-transferrin ratio. The sample analysis of complement proteins was done at two different time-points (phase 1 and phase 2, see Supplementary Material and Supplementary Tables S3 and S4). Potential differences in values caused by sample analysis at two different time points were corrected by including phase as an additional explanatory variable in each model. Additional analysis showed that interaction between phase and disease is insignificant (P value = 0.1383). A combined regression model was also used, with complement gene score, disease category, and phase as explanatory variables for log complement protein concentration. This allowed assessment of the complement protein complement gene score association while accounting for disease category effect and, similarly, the complement protein disease category association while accounting for complement gene score effect. The unique contribution of complement gene score and disease category to the variation in complement protein level were determined from their partial coefficient of determination. For evaluating disease category effect in both separate and combined models, all pairwise comparisons between disease categories were conducted using contrast t-test. Geometric mean ratio for each comparison was reported, along with its 95% confidence interval. The above analysis was repeated using all gene score in lieu of complement gene score to represent the genetic risk. 

All tests were conducted at the 5% significance level and the analysis was performed using SAS v9.2 (SAS Institute Inc., Cary, NC). 

Results

Increased Complement Activation in Vitreous of Advanced AMD

Complement proteins were present at the highest levels in the choroidal endothelial cells, followed by Bruch's membrane. Immunostaining specific for C3 and C4 was observed in the Bruch's membrane and choroidal endothelial cells of all samples. Figure 3 shows examples of rare drusen found in category 1 samples, as well as examples of more abundant drusen found in category 4 samples. Most drusen stained positively for C3 in 12 of 21 cases and for C4 in 3 of 21 cases. FD was present in BM/C of all category 1 donor eyes and donor eyes diagnosed with category 3 or 4 AMD (Fig. 3). Most drusen were positive for FD in 17 of 25 cases. FB was found in BM/C of all donor eyes, whereas most drusen were positive in 11 of 24 cases. Drusen analysis was performed in combined category 1, 3, and 4 samples. Immunohistochemical detection of FB in BM/C, RPE, and neural retina indicates predominant localization of FB to choroid vasculature, BM, and the basal side of RPE cells with no detectable expression in neural retina (Fig. 4 and see Supplementary Material and Supplementary Fig. S2). These results identify the choriocapillaris and Bruch's membrane, and not drusen, as the main structures in which alternative pathway complement proteins are localized at the BM/C interface of human postmortem eyes. 

Figure 3. 

 

Immunohistochemical staining of complement proteins in cross-sections of Bruch's membranes with attached choroidal layers (BM/C) obtained from the macula of category 1 and category 4 human donor eyes. Sections were counterstained with Mayer's hematoxylin. Red arrows, complement positive drusen; black arrows, complement negative drusen; asterisk, Bruch's membrane; arrowheads, choroidal endothelial cells.

Figure 4. 

 

Immunohistochemical staining for FB on a cross-section through the macula of a 67-year-old female donor with wet AMD diagnosed in the second eye. Red and pink staining reflects FB, brown represents melanin in RPE and choroid. Black arrow points to drusen accumulating between the RPE basement membrane and the BM. Inset shows detail of FB localization in RPE cells, BM, and choroid. Red arrow points to FB staining at the basal side of RPE cells, blue arrow points to FB staining on BM, and white asterisk indicates FB in the lumen of a choroidal vessel. INL, inner nuclear layer; ONL, outer nuclear layer; C, choroid; S, sclera. Scale bar = 100 μm.

Next we used sensitive and specific ELISAs to quantify the amount of complement proteins present in BM/C protein extracts obtained from category 1, category 3, and category 4 eyes. Total C3 proteins (C3, C3b, iC3b, C3c, and C3d fragments) in BM/C were significantly higher in category 4 AMD donor eyes compared with category 1 controls (P = 0.0237, Fig. 5A). The same was the case for C3a levels (P = 0.0007, Fig. 5B), which also showed a significant increase in concentrations in category 4 compared with category 3 disease (P = 0.0328), but not for C4a (Fig. 5C), reflecting activation primarily of the alternative- and not classical and mannose-binding lectin pathways ¹³ in BM/C preparations, in agreement with earlier reports on plasma complement concentrations in AMD. ²³ Combined (i)C3b and C3c levels (“activated C3”; see Supplementary Material and Supplementary Table S3) did not increase, potentially because these fragments are further degraded to C3d and C3dg, which are not recognized by the activated C3 ELISA. ²⁷ FB protein levels were significantly higher in BM/C of end-stage disease compared with category 1 disease (P = 0.0192; Fig. 5D). ²³ FBb, also examined in BM/C, was below detection limit in all samples. The concentration of FD was higher in category 4 compared with category 1 patients (Fig. 5E). Besides complement proteins, serum proteins albumin (P < 0.05, Fig. 5G) and transferrin (P < 0.05, Fig. 5H), but not total IgG (P > 0.05, Fig. 5I), were also significantly increased in category 4 compared to category 1 BM/C protein lysates, indicating a general increase of complement and noncomplement serum proteins in BM/C tissues of category 4 compared to category 1 eyes. 

Figure 5. 

 

Increased concentration of complement and other noncomplement serum components in protein extracts from BM/C of category 4 compared with category 3 and category 1 donor eyes. (A) Total C3 proteins; (B) C3a des Arg; (C) C4a des Arg, (D) factor B (FB); (E) factor D (FD); (F) Active C3; (G) albumin; (H) transferrin; (I) total IgG. Mean values are indicated by horizontal lines; dotted lines indicate lower limit of quantification; the upper dotted line in (B, C, F, I) represents the lower limit of quantification for phase 1 samples (gray symbols) or phase 2 samples in (H). P values from ANOVA and contrast t-test. The potential effect of “phase” on the complement protein levels was accounted for by including “phase” as a covariate in the statistical analysis. N.S., not statistically significant.

To evaluate the possibility that complement proteins were increased in other compartments of the eye, we collected vitreous from the same donor eyes used to prepare BM/C samples. Contrary to BM/C lysates, concentrations of total C3 proteins, FD, and FB were not elevated in vitreous of eyes with category 3 and 4 disease compared with category 1 eyes (see Supplementary Material and Supplementary Table S4). However, a significant increase in the ratio of FBb to FB (P = 0.0287; Fig. 6), a measure for alternative pathway complement activation determined for each individual, was found in eyes diagnosed with category 4 AMD as compared with category 1 eyes. 

Figure 6. 

 

Increased ratio of FBb to FB molar concentrations in vitreous of category 4 compared with category 1 eyes. P value from ANOVA and contrast t-test. FB, factor B.

At-Risk and Protective Allele Frequencies Correlate with Alternative Pathway Complement Activation in Vitreous

Analysis of control and AMD donor genotypes confirmed association of previously described risk and protective alleles with the incidence of AMD (Table 2). Statistically significant associations (P < 0.05) were observed for CFH and HtrA1 SNPs as well as similar allele frequencies and effect sizes of the C2, CFB, and C3 variants as previously reported. ^{11,15,17–20,22} Complement gene scores, calculated based on odds ratios for SNPs in the complement pathway (see Methods), were significantly higher in donors with category 3 and category 4 eyes compared with donors with category 1 eyes (P < 0.0001, Fig. 7A). Addition of the HtrA1/ARMS2 SNP to the analysis (Fig. 7B) improved the association of gene score with disease incidence, indicating that, in the studied population, polymorphisms in both complement and HtrA1/ARMS2 pathways predispose to category 3 and category 4 AMD. 

Figure 7. 

 

Increased gene scores in category 3 and category 4 donor eyes compared with category 1 donor eyes. Gene scores for complement genes (A) and multiple genes (B) were calculated based on odds ratios for SNPs in (CFH Y402H, C2, IVS10, CFB R32Q, C3 R102G) (A) or based on SNPs in the complement proteins and rs10490924 in the HtrA1/ARMS2 promoter region (B). *P < 0.0001 from ANOVA and contrast t-test.

Table 2. 

 

Odds Ratios for Risk and Protective SNPs in the Donor Population

SNP ID	Genetic Locus	Allele 1	Allele 2	Allele 1 Frequency		Chi-Square	P Value	Odds Ratio
				AMD	Control
rs11200638	HTRA1	A	G	0.286	0.117	6.68	0.0098	3.03
rs1061170	CFH	C	T	0.479	0.283	6.57	0.010	2.32
rs2230199	C3	G	C	0.236	0.200	0.31	0.58	1.23
rs641153	CFB	T	C	0.057	0.100	1.18	0.28	0.55
rs9332739	C2	C	G	0.007	0.033	1.95	0.16	0.21

Because SNPs in the complement proteins have been shown to affect complement activation, ³³ we next determined if increased concentration of complement proteins and activation products in BM/C correlated with AMD risk variants. We did not observe statistically significant association of AMD risk variants with increased complement activation in BM/C lysates. Levels of total C3, C3a, FB, and FD remained significantly different among disease categories when gene effect was accounted for (see Supplementary Material and Supplementary Table S5). In the vitreous, increases in the FBb:FB ratio in category 4 compared with category 1 eyes did not remain significant when gene effect was accounted for, indicating that increased FBb:FB could largely be attributed to genotype (see Supplementary Material and Supplementary Table S6). There was no evidence of association between complement levels and gene scores in the BM/C (see Supplementary Material and Supplementary Table S7), but vitreous protein levels of FBb and the ratio of FBb to FB were significantly positively correlated with complement gene scores (P < 0.05 and P < 0.0005, respectively, Figs. 8A, 8B). The correlation of the vitreous FBb:FB ratio with gene score remained significant when disease grade was also accounted for, suggesting that complement AMD risk variants result in intrinsic increases in local complement activation (see Supplementary Material and Supplementary Table S7). 

Figure 8. 

 

Partial regression of vitreous FBb and FBb:FB with gene score. Visualization of complement gene score effect on levels of FBb (A) and FBb:FB ratio (B) in vitreous in a multiple regression model. Partial regression plots “partials” out the phase and stage effect, showing the remaining association between complement protein and gene score.

Discussion

Using quantitative assays, this study demonstrated that complement activation in vitreous of well-characterized donor eyes is increased with disease progression in AMD. We further demonstrated that alternative pathway complement activation in the vitreous is dominated by genetic risk factors in complement proteins, whereas levels of complement and noncomplement serum proteins in BM/C extracts are independently associated with disease progression. 

Previous studies have shown that plasma levels of C3a and C3d, as well as FB, FBb, FBa, and FD, were elevated, whereas the regulator of the alternative complement pathway CFH was reduced in AMD compared with control cases, independent of genetic risk factors. ^12,23 In one of the studies, ¹² adjustment for genetic variants increased odds ratios for association of complement levels with disease incidence and showed a positive association between body mass index (a known risk factor for AMD) and plasma levels of C3, FB, CFH, iC3b, and C3a. In a separate study, inclusion of genetic variants in the complement genes CFH, FB, and C3 resulted in a positive association between risk allele frequency and systemic activation of the alternative complement pathway. ²⁴ This is in line with an effect of these mutants on complement activation in human serum. ³³ A recent study has found evidence for genetic association between an FD gene SNP and AMD and a significant increase in plasma FD concentration in AMD cases compared with controls. ³⁴ Thus, genetic and environmental factors that predispose to AMD also predispose to increased levels and increased activation of alternative pathway complement proteins in the circulation and in the eye. 

This raises the important question of whether local elevation of complement proteins and activation products in the eye may be a reflection of systemically elevated complement pathway activity. ^12,23,24,34 It has been demonstrated that complement C3, FD, and FB transcripts are absent in the neural retina of eyes and abundant in cells of the choroid and in peripheral organs, including the liver. ³⁵ Activated RPE cells, ³⁶ as well as microglia and Müller cells in the neural retina, ^37,38 can, in some circumstances, synthesize a wide range of complement proteins, but their contribution to complement activation in normal eyes and eyes affected by AMD is unknown. These results suggest that peripheral tissues are the main site for transcription and translation of complement proteins that could accumulate locally in the eye to drive alternative pathway complement activation. 

Although complement proteins, albumin and transferrin were higher in the BM/C of eyes with category 4 as compared with category 1 AMD, there was no significant difference in the amount of complement protein extracted from BM/C in macula with numerous drusen (category 3) compared with macula with few or no drusen (category 1). This is consistent with immunohistochemical findings that only a fraction of drusen stained positive for complement proteins, whereas Bruch's membrane and choriocapillaris of all specimens invariably stained positive. The choroidal microcapillaris aligning the Bruch's membrane may, in addition to the retinal vasculature, be a source of complement proteins released from the vasculature into the sub-RPE space, retina, and vitreous. The degree of extravasation of complement and other serum proteins could be dependent on the size of the proteins, explaining why a low-molecular-weight molecule like FD is relatively more abundant in vitreous than the higher molecular weight proteins FB and C3 (see Supplementary Material and Supplementary Fig. S3). Together, the combined increased presence of complement proteins and reduced activity of complement regulatory proteins, including CFH and membrane cofactor protein (CD46), ³⁹ may result in increased complement activation in the BM/C of eyes diagnosed with advanced AMD. 

The lack of association between complement levels and gene scores in the BM/C suggests two possibilities: either disease stage rather than AMD risk variants influence complement activation at the BM/C or the relatively small sample size did not allow for a statistically significant result. Further studies with larger numbers of donors need to be performed to explore if a correlation exists between complement activation in BM/C and genetic risk variants. 

In conclusion, this study for the first time demonstrates an interaction between genetic polymorphisms in complement pathway genes and alternative complement pathway activation in vitreous. These results support that AMD is associated with both systemic and local activation of complement with further implications for treatment of this multifactorial disease with inhibitors of the complement pathway. 

Supplementary Materials

pdf S1 

pdf S2 

Acknowledgments

We thank Somasekar Seshagiri for help with genotype analysis; Sheila Beddha, Linda Rangell, and Ronald Ferrando for immunohistochemical procedures; Jason Chinn for help with ELISA assays; Karen G. Shadrach for help with tissue preparation; and Floresita Puth and Julie Rae for help with tissues procurement. We thank the National Disease Research Interchange in Philadelphia and the Lyon's Eye Institute, Tampa, Florida for making donor tissues available for this research. 

Footnotes

 Supported by Genentech, Inc. Support for third-party writing assistance by Ivo Stoilov, MD, CMPP, of Envision Scientific Solutions, was provided by Genentech, Inc.

Footnotes

 Disclosure: K.M. Loyet, Genentech, Inc. (E); L.E. DeForge, Genentech, Inc. (E); K.J. Katschke Jr, Genentech, Inc. (E); L. Diehl, Genentech, Inc. (E); R.R. Graham, Genentech, Inc. (E); L. Pao, Genentech, Inc. (E); L. Sturgeon, Genentech, Inc. (E); S.-C. Lewin-Koh, Genentech, Inc. (E); J.G. Hollyfield, Genentech, Inc. (F, C, R); M. van Lookeren Campagne, Genentech, Inc. (E)