Contribution of Copy Number Variation in the Regulation of Complement Activation Locus to Development of Age-Related Macular Degeneration

Abstract

Purpose.: To develop an assay for determining the number of copies of the genes encoding complement factor H related 3 (CFHR3) and 1 (CFHR1) and determine the contribution of copy number variation (CNV) at CFHR3 and CFHR1 to the development of age-related macular degeneration (AMD).

Methods.: A multiplex ligation-dependent probe amplification (MLPA) assay was developed to quantify the number of copies of CFHR3 and CFHR1 in humans. Subjects with (n = 252) and without (n = 249) AMD were genotyped using the assay, and the impact on AMD risk was evaluated.

Results.: The MLPA assay provided a consistent estimate of the number of copies of CFHR3 and CFHR1 in 500 of the 501 samples. Four different combinations of CNVs were observed with frequencies as follows: both CFHR3 and CFHR1 deletion (14%), CFHR3-only deletion (0.4%), CFHR1-only deletion (1.1%), and CFHR1 duplication (0.1%). Deletion of both copies of CFHR3 and CFHR1 decreased the odds of having AMD eightfold (95% CI 2–36) and always occurred on a protective haplotype, never on the risk haplotype tagged by the Y402H risk allele in CFH. The protection conferred by deletion of CFHR3 and CFHR1 could not be distinguished from the absence of the risk haplotype.

Conclusions.: Both deletions and duplications of genes in the regulation of complement activation locus segregated in Caucasians. Deletion of CFHR3 and CFHR1 protected against the development of AMD at least in part because the deletion tagged a protective haplotype and did not occur on the risk haplotype.

Age-related macular degeneration (AMD) is the leading cause of vision loss in elderly individuals in the developed world. ¹ Genetic variation in genes encoding proteins within the alternative pathway of complement plays a major role in the development of AMD. ^2–4 Reported loci include complement factor H (CFH), the tightly linked genes complement component 2 and factor B (C2/CFB), complement component 3 (C3), and factor I (CFI). ^5–12 In addition to variation in alternative pathway of complement genes, a locus called age-related maculopathy susceptibility 2 (ARMS2) contributing to AMD has been localized to a region on chromosome 10, region q26, containing a hypothetical gene (LOC387715) and the promoter region of the gene encoding high temperature requirement A1 (HTRA1). ^13,14  

The 300-kb regulation of complement activation (RCA) locus contains CFH and five ancestrally related genes that arise through duplication of CFH (Fig. 1). The genes are referred to as complement factor H related (CFHR) and are numbered one through five. There is a low rate of recombination across the RCA locus, resulting in extensive linkage disequilibrium (tendency of two polymorphic markers to be inherited together) across the region. The combination of large regions of DNA sequence similarity secondary to the duplications that gave rise to the CFHR genes and the linkage disequilibrium makes it difficult to distinguish the pathogenic sequence variation from the extensive polymorphic variation found at approximately 1 in 300 bp across the genome. 

Exons 2 through 9 of CFH are located within a 27-kb region of very high linkage disequilibrium. ⁵ There are four common haplotypes (ancestral segments of DNA that tend to be inherited as a unit or block) within this region. One of these haplotypes (risk [R]) that also contains the Y402H polymorphism markedly increases the risk of AMD. ^5–8 Two of the other three haplotypes (protective 1 and 2, or P1 and P2) appear to protect against the development of AMD, whereas the third appears to be neutral (N). ⁷ Statistical analyses looking for independent effects of these four haplotypes suggest that the impact on AMD risk could be explained by the risk haplotype alone. ⁵ Such conditional analyses provide a valid estimate of how to most efficiently explain the contribution of the haplotypes determined with a given set of variants to the risk of developing AMD. However, there is concern that statistical modeling can obscure underlying unique biological effects of the individual haplotypes. 

The demonstration that the P2 haplotype in CFH was in tight linkage disequilibrium with a downstream deletion of the CFHR3 and CFHR1 genes provided direct evidence that the protective haplotypes could have unique effects on AMD risk. ¹⁵ That is, although both the P1 and P2 haplotypes are distinct from the R haplotype (e.g., they do not carry the risk allele for Y402H), only the P2 haplotype contains the CFHR3 and CFHR1 deletion. CFHR3 and CFHR1 contain the cell surface and complement binding domains of CFH, but lack the amino terminal domains that inactivate the C3 convertase. Thus, the protective effect of the P2 haplotype carrying the deletion of CFHR3 and CFHR1 led to the hypothesis that their presence increases the risk of development of AMD by competing with the inhibitory activity of factor H that downregulates the alternative pathway of complement. ¹⁵  

Because of the extensive homology between the duplicated genomic regions containing the CFHR genes, unequal homologous recombination events would be expected to generate a variety of copy number variation patterns in the RCA locus. Indeed, such events have been observed to arise de novo in patients with rare complement-related disorders. ¹⁶ We hypothesized that copy number variation in the RCA locus beyond the combined deletion of CFHR3 and CFHR1 would segregate in the Caucasian population and contribute to the development of AMD. We also sought to determine whether the protection from AMD conferred by the combined deletion of CFHR3 and CFHR1 is restricted to the P2 haplotype and whether it has a greater effect in the presence of the R haplotype (tagged by Y402H). 

Thus, as a first step in a general study of contribution of copy number variation in the RCA locus on AMD, we developed a quantitative copy number (genotyping) assay for CFHR3 and CFHR1. To overcome the extensive homology between the CFHR genes, we used the multiplex ligation-dependent probe amplification (MLPA) technique (Fig. 1). ^17–19 The results showed that deletions of CFHR3 and CFHR1 individually and in combination segregated in the population as do duplications of CFHR1. The combined deletion of CFHR3 and CFHR1 was restricted to the P2 haplotype in CFH. However, the impact on AMD risk could not be distinquished from variation at Y402H. Thus, we cannot confirm that the presence or deletion of CFHR3 and CFHR1 contributes to AMD independent of the common AMD risk haplotype (R) in CFH. 

Materials and Methods

Subjects

The study was approved by the institutional review board of the Mayo Clinic and adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all subjects. The Mayo subjects were composed of the 501 Caucasian individuals described in Table 1. The ascertainment and characterization of the Mayo subjects has been reported. ²⁰ Briefly, diagnosis was determined by review of fundus photographs, as described elsewhere. ^5,21,22 All subjects with diagnosed AMD had large drusen (≥125 μm) with sufficient drusen area to fill a 700-μm circle, or more advanced findings. Control subjects had five or fewer hard drusen (<63 μm) without pigment changes or more advanced findings. Geographic atrophy and exudation were defined according to the Wisconsin age-related maculopathy grading system. ²³ The Mayo subjects have been graded multiple independent times by two individuals. ²⁰ Mayo subjects with both neovascular and primary geographic atrophy—namely, the development of geographic atrophy before the onset of exudation—were included in the analysis for each subtype. When a unique grade for each subject was required, the subjects graded “both” were added to the grade with a smaller number of subjects (geographic atrophy) to increase power. ^20,24  

Table 1.

 

Demographic and Clinical Features of the Mayo Subjects

Subjects	n	Age (Mean ± SD)	Male-Female Ratio
AMD subtotal	252	76.3 ± 9.3	0.47
Early AMD	147	74.4 ± 9.7	0.48
Exudation	69	79.1 ± 8.6	0.41
Geographic atrophy*	36	78.9 ± 7.4	0.57
Total	501	73.1 ± 9.0	0.65

*  Nine Mayo subjects who had both primary geographic atrophy and exudative AMD (category Both) were included in the geographic atrophy group as previously described. ²⁰

Multiplex Ligation Dependent Probe Amplification

Design of Oligonucleotides.

Oligonucleotides for MLPA were designed to bind specifically to either CFHR3 or CFHR1 (Fig. 1). Oligonucleotides binding to SS18 on chromosome 18 and to PDCD8 (AFIM1) on the X chromosome were designed and included as control specimens. The four oligonucleotide pairs consisted of a left and a right oligonucleotide, referred to herein as the left oligo and right oligo. The pairs were designed to have their ligation points specifically at a known sequence variation of the homologous genes as described in Figure 1. They were designed to have a GC content of 40% to 60%. The forward primer in the polymerase chain reaction (PCR) was labeled with a fluorescent 6-carboxy-fluorescine (FAM) marker at the 5′ end, to allow quantitative analysis with capillary electrophoresis. Quenching of the fluorescent FAM marker was avoided by redesigning the standard MLPA forward primer, ^17,19 and the sequence is shown in Figure 1. The oligonucleotides and the primers were obtained from Integrated DNA Technology (Coralville, IA; Table 2). 

Table 2.

 

Target Sequences of the Oligonucleotides Used in the MLPA Assay

Target Gene	Oligonucleotide Side	MLPA Oligonucleotides (5′-3′) Hybridization Sequences*	Product Length (bp)
SS18	Left oligo	CAGCAGCCACCTATGGGAATGATG	96
	Right oligo	P-GGTCAAGTTAACCAAGGCAATCATATGATG
CFHR3	Left oligo	CAAAAGCGCAGACCACAGTTACATGTAC	104
	Right oligo	P-GGAGAAAGGCTGGTCTCCTACTCCCAGATGCATC
CFHR1	Left oligo	CAACATTTCATGTGTAGAACGGGGCTGGTCCACC	112
	Right oligo	P-CCTCCCAAATGCAGGTCCACTGGTAAGTACAATGCT
PDCD8	Left oligo	CGCAAATACAACAGGTATCAGAACTGCTGGCCCCAGATTAAG	128
	Right oligo	P-CTTCAGATGGTGAACTCTGTGCACTTCCACCCATGCAGTCACCT

*  The left oligo begins with the forward primer sequence 5′-TGGGTCCCTAAGGGTTGGA-3′ followed by the sequence shown in the table. The 5′-end of the right oligo is phosphorylated (P) and the sequence shown in the table is followed by the complement of the reverse primer 5′- TCTAGATTGGATCTTGCTGGCAC-3′. Figure 1 and the methods provide additional details.

Assay Conditions for MLPA.

We performed the MLPA assay according to the manufacturers' instructions (MLPA EK kit; MRC-Holland, Amsterdam, The Netherlands). We used 50 ng DNA per sample at 10 ng/μL for each assay. Two control samples without deletions of CFHR3 and CFHR1, based on the MLPA assay, and a water (no DNA) control were included in each experiment. The 5-μL DNA samples were denatured in a thermocycler for 5 minutes at 98°C and cooled to 25°C. Three microliters of a master mix containing 1.5 μL of the MLPA buffer and 6 fM of both the left and right oligos for each target was added. The temperature was raised to 95°C for 1 minute then reduced to 60°C for 16 hours for hybridization. The temperature was reduced to 54°C, and a ligation mix containing 3 μL of the ligase-65 buffer A, 3 μL of the ligase-65 buffer B, 25 μL of distilled H₂O, and 1 μL of ligase-65 was added and the ligation allowed to proceed for 15 minutes at 54°C. The ligation was terminated by raising the temperature to 98°C for 5 minutes and the samples were cooled to 4°C. For amplification, 10 μL of the ligation product, 4 μL of the PCR buffer (SALSA; MRC-Holland), and 26 μL of distilled H₂O was combined and brought to 60°C. Ten microliters of a master mix containing 6 pM forward and 4 pM reverse primers, 10 mM dNTP, 2 μL of enzyme dilution buffer (SALSA; MRC-Holland), 4.5 μL of distilled H₂O, and 0.5 μL of polymerase was added (SALSA; MRC-Holland). Amplification conditions were 33 cycles at 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 60 seconds, followed by 20 minutes at 72°C and a hold at 4°C. 

Assay Conditions for Capillary Electrophoresis.

Two microliters of the amplified products were plated into a well with 18 μL of a mixture of 1 part formamide (Hi-Di; Applied Biosystems, Inc. [ABI] Foster City, CA) and 0.015 parts size standard (GeneScan 120 LIZ; ABI). The plate was spun down, denatured for 2 minutes at 94°C, and held at 4°C for 5 minutes before capillary electrophoresis on a DNA analyzer (model 3730; ABI) was performed. The injection rate was set at 1.6 kV for 10 seconds. The maximum threshold for fluorescence intensity on the DNA analyzer was 30,000 relative fluorescence units. If the detection threshold was crossed, the amplification process was repeated with 30 cycles using the same ligation product. If the fluorescent intensities were still too high after the reduced cycles, the product of the reduced cycles was diluted in distilled water (1:50), and 2 μL of the dilution was used for capillary electrophoresis. Similar results were obtained when repeating the PCR with fewer cycles and with dilution. 

Data Analysis.

Mapping software (GeneMapper ver. 4.0; ABI) was used to perform fragment sizing with the internal size standard (120 LIZ; ABI), automated peak calling, and peak normalization. ²⁵ The four peak areas of the four probes (SS18, CFHR3, CFHR1, and PDCD8) of each sample were exported for further data analysis. The peak areas were normalized by dividing each peak area by the sum of all four peak areas in each sample. Each normalized peak area was divided by the means of that peak area in the two control samples to standardize the peak areas to the control samples included on every run. The ratio for the probes CFHR3, CFHR1, and PDCD8 of every sample was generated by dividing the normalized and standardized peak area of the probe by the normalized and standardized peak area of the control probe SS18 in that sample. The probe of the X chromosome was used as a control for unintended interchange of samples. Each assay run was calculated separately because of possible interassay variations, as recommended. ¹⁷ The person performing the assay and the calculations was masked to the ophthalmic diagnosis of the subjects. 

Replication of MLPA Assay Results.

The MLPA assay was repeated on samples that had a CFHR3 or CFHR1 probe with a ratio of <0.70 or >1.30, based on prior work suggesting that a change of this magnitude could indicate copy number variation. ^18,19,25 If the repeated result did not confirm the initial ratio category (<0.70 or >1.30), MLPA was repeated twice more for a total of four independent MLPA assays. If the repeated sample proved to have a ratio of <0.70 or >1.30 of only CFHR3 or CFHR1, then the MLPA assay was performed six independent times and the result verified with a quantitative PCR assay (TaqMan; ABI). 

Quantitative PCR

Quantitative PCR was performed to independently confirm the copy number variation of selected samples. Specific genomic targets in CFHR3 and CFHR1 were amplified (TaqMan Gene Expression Master Mix; ABI) combined with the specific target gene assay mix consisting of a 0.25 mM MGB probe (TaqMan; ABI) and 0.9 mM of each PCR primer. Forward primer 5′-TGTTTTGCCAACGGACCTATTTAGT-3′ and reverse primer 5′-GCCCATTAATAGGAGCATTTATTTTGCT-3′ was used for CFHR3. Forward primer 5′-ACATCTCCAATTTAGATCCTTTGATTAACCA-3′ and reverse primer 5′-GCATTTTCTTAGTGAATAAGCAAAGATTTAAAAACA-3′ was used for CFHR1. RNase P was analyzed simultaneously in the same reaction mix as the endogenous control gene. Probes and primers (TaqMan) were obtained from ABI. The quantitative PCR was performed on a real-time PCR system (model 7900; ABI): 95°C for 10 minutes followed by 40 cycles of 15 seconds at 95°C and 60 seconds at 60°C. All reactions were performed in triplicate. The values obtained for CFHR3 and CHFR1 copy number were normalized to endogenous control gene RNase P and quantified relative to the copy number of control samples using the ^ΔΔCT method. ²⁶  

Genotyping

Single-nucleotide polymorphisms (SNPs) tagging common haplotypes across the RCA locus were genotyped as described previously. ^5,24,27  

Copy Number, SNP, and Haplotype Statistical Analyses

All SNPs and copy number assays were noted to be in Hardy-Weinberg equilibrium (P > 0.05). Single SNP analyses of genotype distributions were performed (SAS ver. 9.13; SAS Institute, Cary, NC) by using logistic regression assuming a log-additive genetic model where SNPs were coded as 0, 1, or 2 for the number of minor alleles. The copy number variation of CFHR3 or CFHR1 was coded as 0, 1, or 2 for the number of deletions of each gene (the single subject with a duplication was not included in these analyses). Fisher's exact tests were also performed on genotype distributions. Haplotype analyses on SNPs across the RCA locus and the occurrence of CFHR3 and CFHR1 copy number variation was performed using the haplo.stats packages in R. Interaction between rs1061170 (Y402H in CFH) and CFHR3 or CFHR1 copy number variation on AMD status was evaluated by fitting two log-additive models with logistic regression using two main-effect terms (genotype for each polymorphism), with and without their interaction terms, and assessed by a likelihood ratio test. The effect of polymorphisms on AMD subtypes was investigated in baseline-category logit models in which the control was considered baseline. Age is confounded with diagnosis (i.e., the cases are older than the controls and age is a risk for AMD), thus correction for age might have unpredictable effects; all analyses were performed with and without correction for age and sex. ²⁴ Nominal probabilities are reported. 

Results

Development of a New MLPA Assay

As noted in the introduction, the P2 haplotype in the CFH locus has been reported to frequently carry the combined CFHR3 and CFHR1 deletion. Using previously genotyped subjects, five subjects homozygous for P2 and five subjects heterozygous for P2 were selected for assay development. ^5,24,27 The CFHR3 and CFHR1 copy number in each of these 10 samples was evaluated with the quantitative PCR assay. The expected copy number was observed, with the subjects heterozygous for haplotype P2 carrying one copy of CFHR3 and CFHR1 and those homozygous for P2 having no amplification (no copies) of CFHR3 and CFHR1. 

These 10 samples, and 5 additional samples chosen at random without known haplotype information, were used to test the designed oligonucleotides for the MLPA assay (Fig. 1). The MLPA assay results were in complete agreement with the results of the quantitative PCR. Four of the other five samples had two copies of CFHR3 and CFHR1, whereas one sample had one copy of CFHR3 and CFHR1. The results were shown to be reproducible by repeating the assay five times. The reproducibility of the capillary electrophoresis was demonstrated by repeating this step twice. Examples of the raw data are shown in Figure 2. 

Figure 1.

 

MLPA strategy to quantify copy number variation at CFHR3 and CFHR1. The RCA locus is shown in (A). It comprises CFH and the five CFH related genes (CFHR1 to -5). (B, C) The MLPA assays. The oligonucleotide pairs for MLPA consist of left and right oligos. The left oligo contains the MLPA forward primer (FWP) at its 5′-end, followed by the complement of the target sequence (the gene to be assayed). The right oligo begins with the complement of its target sequence (which is immediately adjacent to the end of left oligo) and ends with the complement of the MLPA reverse primer (RVP). The right oligo is phosphorylated at its 5′-end to allow ligation to the left oligo after hybridization. An MLPA assay contains a genomic copy number control, in our case SS18 and an X-chromosome target. The MLPA procedure consists of four steps. First, the left and the right oligos are hybridized to the targets. This step is driven to completion and is the basis of the quantitative comparison to the controls. Second, the oligos hybridized to the target are ligated; a 1-base-pair mismatch at the ligation site is all that is required for specific ligation. Vertical bar: the target site at which ligation occurs. Third, the hybridized strands are denatured and the oligonucleotide pairs are amplified by polymerase chain reaction. Fourth, the amplification products are quantified using capillary electrophoresis. One oligonucleotide pair was designed to hybridize specifically to exon 3 of CFHR3 (B), and the other to exon 3 of CFHR1 (C). Capital letters in the wild-type sequence correspond to exons, whereas lowercase indicates introns.

Figure 2.

 

Example of an MLPA copy number assay. Raw fluorescence intensity (y-axis, relative fluorescence units) observed during capillary electrophoresis are illustrated. The first peak represents the amplicon of the control probe SS18 with a product size of 96 bp (x-axis, base pairs), followed by the other probes as illustrated. The ratio of peak area of CFHR3 and CFHR1 to SS18 determines the copy number. (A) A wild-type subject; (B) a subject heterozygous for deletion of both CFHR3 and CFHR1, (C) a subject homozygous for deletion of both CFHR3 and CFHR1. The X-chromosome marker (PDCD8) shows that the subjects in (B) and (C) are male.

The products amplified by each of the four amplicons in the MLPA assay were electrophoresed on agarose gels, and single bands of the correct product sizes were observed. No amplification was observed in subjects with homozygous deletions. The specificity of the target was confirmed using subjects with deletions of only CFHR3 or CFHR1. Note that samples from these subjects were also studied with quantitative PCR (TaqMan; ABI) and the same copy number was observed with both assays. 

Genotyping of 501 Subjects with MLPA Assay

Five-hundred one samples were chosen based on availability of prior genotyping at SNPs across the RCA locus ^24,27 and genotyped using the MLPA assay. As noted, previous work with MLPA assays had suggested that a change of greater than 30% in normalized peak area relative to the control was consistent with copy number variation. ^18,19,25 We observed 148 samples meeting these criteria and repeated the MLPA assay on all these samples (Table 3). Any samples not giving the same result (e.g., first run, <0.70, and second run, ≥0.70) were repeated two additional times. All samples with copy number variation other than combined deletion of both CFHR3 and CFHR1 were assayed two additional times and confirmed with the quantitative PCR. 

Table 3.

 

Path to Final Allele Count of Observed Copy Number Variants at CFHR3 and CFHR1

Initial Copy Number Polymorphism Estimate for CFHR3/CFHR1		Did First Repeat Confirm Initial Result?	Number of Assays (1–6) Confirming Initial Result	Number Out of Total (N = 501)	Final CFHR3 and CFHR1 Copy Number
No deletion of CFHR3/CFHR1	2/2	—	—	353	2/2
CFHR3/CFHR1 heterozygous deletion	1/1	Yes	2 of 2	99	1/1
		No	3 of 4	6	1/1
		No	2 of 6	1*	2/2
CFHR3/CFHR1 homozygous deletion	0/0	Yes	2 of 2	15	0/0
CFHR3 only heterozygous deletion	1/2	Yes	6 of 6	2	1/2
		Yes	5 of 6	2	1/2
		No	1 of 4	1	2/2
		No	1 of 4	3	1/1
CFHR1 only heterozygous deletion	2/1	Yes	6 of 6	5	2/1
		Yes	5 of 6	4	2/1
		Yes	4 of 6	1	2/1
		No	5 of 6	1	2/1
		No	1 of 4	1	2/2
CFHR3 only heterozygous duplication	3/2	No	1 of 4	1	2/2
CFHR1 only heterozygous duplication	2/3	Yes	3 of 6	1†	Excluded
		No	5 of 6	1	2/3
		No	2 of 6	1	2/2
		No	1 of 3	1‡	2/2
		No	1 of 4	2	2/2

*  Subject 459 had 1/1 suspect at initial run but became a 2/1 suspect after the second repeat and was repeated four more times.

†  Subject 48 was a 2/3 suspect but ended up being undeterminable and was excluded from subsequent analyses.

‡  Subject 1773 was a 2/3 suspect at the initial run but the assay was repeated only two more times.

Validation of Ratio Change Thresholds for Copy Number Variation in the MLPA Assay

Scatterplot diagrams comparing the first and second independent assays of CFHR3 and CFHR1 for the 148 samples with a ratio change of greater than 30% showed excellent agreement (Fig. 3). Based on this evidence and the replication using the quantitative PCR assays (TaqMan, ABI), the standard MLPA ratio criteria for heterozygous deletion (<0.70), homozygous deletion (<0.30), and heterozygous duplication (>1.30) were used throughout the study. 

Figure 3.

 

Scatterplot comparing the copy number estimates determined for the first and second MLPA assays on 148 subjects. Results are shown for (A) CFHR3 and (B) CFHR1. The data show excellent agreement between independent assays for deletions of either gene. The lack of agreement for no deletion (near 1.0/1.0) reflects the samples that were not replicated (Table 3).

MLPA Assay Reliability

Repeated samples and the previously described 15 test samples were used to calculate the interassay coefficient of variation (CV). The probes for CFHR3 and CFHR1 showed excellent consistency, except for homozygous deletions because of the small sample size, and supported our decision to run the assay once on samples unless copy number variation was suspected (Table 4). 

Table 4.

 

Reproducibility of the MLPA Assay

Final Allele Count	CFHR3				CFHR1
	Subjects (n)	Mean Ratio ± SD	Range	CV (%)	Subjects (n)	Mean Ratio ± SD	Range	CV (%)
2	45	1.01 ± 0.10	0.41–1.91	10.6	38	1.00 ± 0.10	0.21–2.20	10.6
1	118	0.49 ± 0.06	0.10–1.11	12.0	125	0.48 ± 0.06	0.21–1.07	12.3
0	4	0.06 ± 0.07	0.00–0.26	120.3	4	0.06 ± 0.07	0.00–0.26	115.6

The MLPA assay for one subject indicated an allele count of 2 for CFHR3, but a consistent result for CFHR1 could not be determined (Table 3). The PCR assay indicated an allele count of three for CFHR3 and four for CFHR1. Because of these discrepancies the sample was excluded from subsequent studies, resulting in 500 successfully genotyped samples. 

Novel Copy Number Observation

We observed novel copy number variation, including duplication of CFHR1 and deletions of either CFHR3 only or CFHR1 only, in 16 subjects (Table 5). Eleven (2.2%) subjects were heterozygous for deletion of CFHR1 only, and one subject (0.2%) had a duplication of CFHR1 only. Four (0.8%) subjects were heterozygous for deletion of CFHR3 only. All 16 samples were verified by PCR (Table 5). 

Table 5.

 

Subjects with Novel Copy Number Variation in CFHR3 or CFHR1

Subject ID	Final Allele Count		Demographics		Diagnosis		CFHR3		CFHR1
	CFHR3	CFHR1	Sex	Age	Case or Control	Subtype	MLPA Mean Ratio ± SD	PCR Copy Number (2−ΔΔCT)	MLPA Mean Ratio ± SD	PCR Copy Number (2−ΔΔCT)
6	2	1	F	66	AMD	Exudative	1.11 ± 0.10	1.87	0.58 ± 0.07	1.28
24	2	1	F	43	AMD	Early	1.24 ± 0.25	2.11	0.67 ± 0.18	1.32
840	2	1	M	68	AMD	Exudative	1.03 ± 0.15	1.78	0.55 ± 0.11	1.11
924	2	1	M	88	AMD	Exudative	0.93 ± 0.25	1.90	0.53 ± 0.16	1.12
945	2	1	F	63	Control		0.88 ± 0.24	1.86	0.50 ± 0.13	1.06
1022	2	1	M	65	Control		1.06 ± 0.08	1.69	0.58 ± 0.09	1.01
1052	2	1	M	66	Control		1.14 ± 0.21	1.84	0.60 ± 0.13	1.06
1817	2	1	F	61	Control		1.01 ± 0.08	1.77	0.57 ± 0.02	1.06
1872	2	1	F	65	Control		1.10 ± 0.12	1.74	0.58 ± 0.06	1.02
1977	2	1	F	71	Control		1.06 ± 0.09	1.84	0.57 ± 0.07	0.84
2007	2	1	M	79	Control		1.08 ± 0.10	1.80	0.57 ± 0.06	1.12
1137	1	2	F	69	Control		0.58 ± 0.09	1.14	1.13 ± 0.13	2.42
1734	1	2	F	78	AMD	Early	0.58 ± 0.16	1.08	1.05 ± 0.30	2.20
1804	1	2	M	65	Control		0.51 ± 0.06	1.18	1.03 ± 0.15	2.41
1835	1	2	M	85	AMD	Exudative	0.56 ± 0.20	1.06	1.05 v 0.36	1.89
929	2	3	F	59	AMD	Early	0.97 ± 0.15	1.76	1.46 ± 0.26	2.70

Genotype Association with AMD

The allele frequencies for deletion of CFHR3 and CFHR1 and their association with AMD are shown in Table 6. We observed a significant protective effect of deletions of CFHR3, CFHR1, and both CFHR3 and CFHR1 on risk of having AMD. The effect is statistically significant for the AMD subtypes early and exudative, but not for geographic atrophy, because of the small sample size. There is no statistically significant difference between the AMD subtypes when early AMD is compared with geographic atrophy and early AMD with exudation (Table 6). 

Table 6.

 

Genotype Distributions Of CFHR3, CFHR1, and Combined CFHR3 and CFHR1 Deletions, and Their Association with AMD

Gene	Minor Allele Frequency (%)	Final Allele Count	Number Subjects (%)	Control n (%)	AMD Cases n (%)	Early AMD n (%)	Number Geographic Atrophy AMD (%)†	Number Exudative AMD (%)†	OR for 2 Alleles (95% CI)	Log-Additive P ‡
CFHR3	14.2	0	15 (3.0)	13 (2.6)	2 (0.4)	2 (0.4)	0 (0.0)	0 (0.0)	7.69 (1.75–33.3)	1.11 × 10−5
		1	112 (22.4)	68 (13.6)	44 (8.8)	24 (4.8)	10 (2.0)	10 (2.0)
		2	373 (74.6)	168 (33.6)	205 (41.0)	121 (24.2)	26 (5.2)	58 (11.6)
		Total	500 (100.0)	249 (49.8)	251 (50.2)	147 (29.4)	36 (7.2)	68 (13.6)
CFHR1	14.9	0	15 (3.0)	13 (2.6)	2 (0.4)	2 (0.4)	0 (0.0)	0 (0.0)	8.33 (1.79–33.3)	1.46 × 10−5
		1	119 (23.8)	73 (14.6)	46 (9.2)	24 (4.8)	10 (2.0)	12 (2.4)
		2	365 (73)	163 (32.6)	202 (40.4)	120 (24.0)	26 (5.2)	56 (11.2)
		3	1 (0.2)	0 (0.0)	1 (0.2)	1 (0.2)	0 (0.0)	0 (0.0)
		Total	500 (100.0)	249 (49.8)	251 (50.2)	147 (29.4)	36 (7.2)	68 (13.6)
CFHR3 and CFHR1	14.3	0	15 (3.1)	13 (2.7)	2 (0.4)	2 (0.4)	0 (0.0)	0 (0.0)	8.33 (1.79–33.3)	9.86 × 10−6
		1	108 (22.3)	66 (13.6)	42 (8.7)	23 (4.8)	10 (2.1)	9 (1.9)
		2	361 (74.6)	161 (33.3)	200 (41.3)	119 (24.6)	26 (5.4)	55 (11.4)
		Total	484* (100.0)	240 (49.6)	244 (50.4)	144 (29.8)	36 (7.4)	64 (13.2)

*  The total for combined deletion of CFHR3 and CFHR1 is 484, because 16 samples had different copy number variations (Table 5).

†  Both early AMD (CFHR3, P = 0.0004; CFHR1, P = 0.0002) and exudative AMD (CFHR3, P = 0.0006; CFHR1, P = 0.002) were significantly different from the control, whereas geographic atrophy was not, because of small sample size (CFHR3, P = 0.2; CFHR1, P = 0.1). There is no statistically significant difference between early AMD and geographic atrophy (CHFR3, P = 0.3; CFHR1, P = 0.3) or early AMD and exudative AMD (CHFR3, P = 0.4; CFHR1, P = 0.8) for either CFHR3 or CFHR1, according to logistic regression.

‡   P for all cases versus controls. Fisher P-values were similar.

Deletions of CFHR3 Only or CFHR1 Only and AMD

The observation of subjects with deletion of either CFHR3 or CFHR1, but not both, enabled us to evaluate the independent effect of deleting either gene on risk of AMD. The minor allele frequencies for the CFHR3 only deletion was 0.4%, for the CFHR1 only deletion was 1.1%, and was 0.1% for the CFHR1 duplication. The majority of subjects with single gene deletions were controls, consistent with a possible protective effect (Table 5). Deletion of CFHR1 alone is not significantly different from subjects with deletions of both CFHR3 and CFHR1 (OR = 0.56 with 95% CI = 0.12–2.23). There was an insufficient number of subjects with the CFHR3-only deletion for us to perform a meaningful evaluation (OR = 0.99 with 95% CI = 0.07–13.73). A larger sample size is needed to determine if deletion of CFHR3 or CFHR1 alone alters the risk of developing AMD. 

Relationship between the P2 and R Haplotypes and AMD Risk

We also were interested in exploring the possibility that the combined deletion of CFHR3 and CFHR1 marking the P2 haplotype in CFH would have a different impact on the risk of having AMD in subjects with factor H^402Y compared with factor H^402H. The factor H^402H allele (a cytosine) tags the common risk (R) haplotype. There was no interaction between rs1061170 and the deletions of CFHR3 (P = 0.16) and CFHR1 (P = 0.15), suggesting that any independent effect the deletion might have would be independent of the Y402H polymorphism. 

However, the observation that combined deletion of CFHR3 and CFHR1 was not observed with the factor H^402H allele, raises the possibility that the protective effect of the deletion may be explained by the absence of the risk (R) haplotype. We explored this possibility by evaluating two-locus genotype frequencies (Table 7). The ORs for combined deletion of CFHR3 and CFHR1 was much lower (2.86) in the absence of the Y402H allele (C, rs1061170) than the single locus estimates obtained in Table 6 (∼eightfold). Thus, part of the effect of the combined deletion of CFHR3 and CFHR1 can be explained by the absence of the Y402H risk allele in these subjects. Logistic regression including Y402H and the combined deletion showed no effect of the deletion (P = 0.22) on AMD risk, whereas Y402H remained highly significant (P = 1.44 × 10⁻⁹). 

Table 7.

 

Two-Locus Genotype Frequencies and ORs for Combined Deletion of CFHR3 and CFHR1 Versus Y402H

Copies of CFHR3/CFHR1	Control (n) rs1061170			Case (n) rs1061170			OR (95% CI)
	TT	TC	CC	TT	TC	CC	TT	TC	CC
0	13	0	0	2	0	0	1.00	NA	NA
1	44	22	0	15	27	0	2.22 (0.53–5.22)	7.98 (1.94–54.59)	NA
2	50	76	35	22	97	81	2.86 (0.71–19.28)	8.30 (2.21–54.09)	15.04 (3.89–99.50)

*  Genotype counts do not add to 500 because of missing data points in some subjects. The OR corrected for age and sex for rs1061170 (Y402H) alone was 3.2 (2.1–5.1) for TT versus TC and 6.5 (3.8–11.1) for TT versus CC in this dataset.

Haplotype Studies

We had previously genotyped SNPs across the RCA locus and were able to determine the haplotypes on which the deletions of CFHR3 and/or CFHR1 were observed. Haplotype effects on AMD risk were similar to prior observations (Table 8). The combined deletion of CFHR3 and CFHR1 was always present on a haplotype most similar to P2. Notably SNP rs6677604 was present on all estimated haplotypes with the combined deletion of CFHR3 and CFHR1. Inspection of genotypes at rs6677604 showed the GG genotype in 325 (98%) of 331 subjects without the combined deletion, the GA genotype in 95 (100%) of 95 subjects heterozygous for the combined deletion, and the AA genotype in 14 (100%) of 14 subjects homozygous for the combined deletion. 

Table 8.

 

Common Haplotypes across the CFH Locus Observed in the 500 Successfully Genotyped Subjects

	Haplotype*	Controls	Cases	Hap-Score	P
R	CGTCTGAGGPP	0.275	0.405	4.729	0.000
P1	CATTCGAAGPP	0.214	0.108	−4.575	0.000
N	TGTTCGAGTPP	0.132	0.143	−0.132	0.895
R2	CGTCTGCGGPP	0.050	0.132	4.148	0.00005
P2	CGCTCAAGGDD	0.114	0.058	−3.100	0.002

*  The order of variants in the haplotypes in rs3753394, rs800292, rs3766404, rs1061170, rs2019724, rs6677604, rs419137, rs2284664, rs1065489, CFHR1 (P, present; D, deleted), and CFHR3 (P, present; D, deleted).

We observed that the single gene deletions of CFHR3 or CFHR1 occurred on different haplotypes. The four CFHR3-only deletions occurred on four different rare haplotypes, suggesting that the deletion events were independent. None of the four haplotypes carried the Y402H risk allele. The 11 CFHR1-only deletions occurred on six different estimated haplotypes, two of which carried the risk allele for Y402H. The other four haplotypes carried tag SNPs from more than one of the common haplotypes. Thus, the deletion of CFHR1 only appears to have arisen multiple times. 

Discussion

There is extensive copy number variation in the human genome and its potential role in contributing to disease has been gaining attention. ^28,29 Copy number variation is associated with several human diseases. ^16,30–38 The well established protective association between AMD and combined deletions of both CFHR3 and CFHR1 is of particular interest because of the potential implications for our understanding of the pathophysiology of AMD and the unknown function of the CFHR genes. To enable standard analyses of genotypes and haplotypes, it is necessary to have an accurate genotyping assay, in this case an accurate estimate of the number of copies of CFHR3 and CFHR1. Toward these goals, we developed a novel assay that accurately estimates the number of copies of both CFHR3 and CFHR1 and used the new assay to perform a genetic association study on 501 subjects with and without AMD. 

The RCA locus is a difficult region in which to perform genetic analysis because of the extensive homology between the CFH and the CFHR genes. This homology extends beyond coding sequences into the intronic and intergenic regions. ⁵ We chose to use the MLPA assay after inspecting the homology between CFHR3 and CFHR1 and the other genes in this region because of its specificity. Notably the MLPA assay has two high-fidelity steps, the hybridization of oligonucleotides and their ligation. Thus, only a single base-pair difference is necessary to distinguish between two locations in the genome. ^17,19 We demonstrated the specificity of the new assay using a variety of methods and confirmed the results with quantitative PCR. The advantages of the MLPA over quantitative PCR are the lower variability decreasing the number of replications necessary and the ability to simultaneously query at least 20 targets. ^17,19 Thus, the assay we described can be expanded to include other copy number variations or even SNPs in the RCA locus and beyond. The accuracy of the assay was 99% for combined deletions of CFHR3 and CFHR1 and is thus similar to standard genotyping assay call rates. The accuracy was lower for other copy number variants, suggesting opportunities for future improvement. 

The AMD case–control association study provided several unique observations. First, we observed novel copy number variants including single gene deletions of CFHR3 or CFHR1 and a duplication of CFHR1. Although the sample size was not large enough to enable proper statistical analysis, most of the CFHR1-only deletions were in the controls. If this observation holds true in larger studies, it would demonstrate that the as yet unknown biological effect of CFHR3 and CFHR1 could be accounted for by the absence of CFHR1 alone. Further, the presence of CFHR1 deletions on a proportion of the CFH risk (R) haplotypes carrying the Y402H polymorphism would decrease the overall AMD risk in subjects carrying both polymorphisms, assuming that the effects are independent of each other. 

It has been hypothesized that the combined deletion of CFHR3 and CFHR1 could protect against AMD by decreasing competition with CFH binding to cell surfaces or sites of active deposition of complement. ¹⁵ This hypothesis is an attractive one, and our case–control study enabled us to ask whether the protective effect of the combined deletion of CFHR3 and CFHR1 is independent of the CFH Y402H variant and the risk haplotype it tags. We found that the protection from combined deletion of CFHR3 and CFHR1 does not significantly interact with Y402H. This demonstrates that the combined deletion of CFHR3 and CFHR1 has the same effect on AMD risk in the presence of Y402 and 402H. 

We also observed that the deletions were not found on the same chromosome as the risk allele at Y402H. This finding suggests that part of the protective effect of the deletions could be the absence of the risk allele at Y402H (or the absence of the risk haplotype tagged by this risk allele). Two locus genotype frequencies (Table 7) confirmed this observation. Logistic regression analyses suggested that the deletions had no effect on AMD risk in our subjects. However, Hughes et al. ¹⁵ reported an independent effect of the deletions, suggesting that further studies are needed to resolve this important question. Thus, the effect of the deletions is at least in part a consequence of their tagging a protective haplotype (P2) and not occurring on the haplotype tagged by Y402H. Any residual independent effect of CFHR3 and CFHR1 could be mediated by competing with CFH binding to C3b, but our results suggest that such an activity would not be influenced by Y402H. It must be remembered that CFHR3 and CFHR1 could be in linkage disequilibrium with an as yet unknown functional variant and (if there is an independent effect) have no impact themselves on AMD risk. 

Of practical importance we observed that the CFH intronic SNP rs6677604 was an excellent proxy for the combined deletion of CFHR3 and CFHR1. Investigators can study the combined deletion of CFHR3 and CFHR1 with approximately 99% (434/440, see results) accuracy by the more cost-effective and widely available genotyping of SNP rs6677604. However, this approach would not detect the less common deletions of CFHR3 or CFHR1 alone, and the latter is sufficiently common to potentially have a meaningful impact on overall AMD risk. 

This study has limitations. Our estimates of the accuracy of genotyping are based on the discordance between two independent assays on the same sample. Additional work using at least three assay replicates is needed to determine the true false-positive and -negative rates for detecting copy number variation. The sample size of 501 subjects is not large enough to enable a study of the effects of less common copy number variation such as deletion of CFHR1 only. The results of all association studies, including this one, require replication in multiple independent groups of subjects. We encourage other investigators to study copy number variation in the regulation of complement activation locus on subjects with and without AMD. 

In summary, we have developed a new MLPA assay for determining copy number variation in CFHR3 and CFHR1. We determined that part, if not all, of the protection from AMD conferred by the combined deletion of CFHR3 and CFHR1 is mediated through decreasing the chance of having the CFH risk haplotype. Deletions of CFHR3 or CFHR1 alone and duplication of CFHR1 alone was observed. We speculate that additional copy number variation in other CFHR genes exists and could contribute to the risk of having AMD. 

Footnotes

 Supported by the Max Kade Foundation, New York, NY; Grant EY014467 from the National Eye Institute, Bethesda, MD; the Foundation Fighting Blindness, Owings Mills, MD; the American Health Assistance Foundation, Clarksburg, MD; unrestricted departmental grants from Research to Prevent Blindness, New York, NY; and the Mayo Foundation, Rochester, MN.

Footnotes

 Disclosure: K.E. Schmid-Kubista, None; N. Tosakulwong, None; Y. Wu, None; E. Ryu, None; L.A. Hecker, None; K.H. Baratz, None; W.L. Brown, None; A.O. Edwards, None

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

The authors thank Daniel Kluge for assistance with the capillary electrophoresis.