Figures
Abstract
Background
Increased intake of ω-3 long-chain polyunsaturated fatty acids (LCPUFAs) and use of peroxisome proliferator activator receptor (PPAR)-activating drugs are associated with attenuation of pathologic retinal angiogenesis. ω-3 LCPUFAs are endogenous agonists of PPARs. We postulated that DNA sequence variation in PPAR gamma (PPARG) co-activator 1 alpha (PPARGC1A), a gene encoding a co-activator of the LCPUFA-sensing PPARG-retinoid X receptor (RXR) transcription complex, may influence neovascularization (NV) in age-related macular degeneration (AMD).
Methods
We applied exact testing methods to examine distributions of DNA sequence variants in PPARGC1A for association with NV AMD and interaction of AMD-associated loci in genes of complement, lipid metabolism, and VEGF signaling systems. Our sample contained 1858 people from 3 elderly cohorts of western European ancestry. We concurrently investigated retinal gene expression profiles in 17-day-old neonatal mice on a 2% LCPUFA feeding paradigm to identify LCPUFA-regulated genes both associated with pathologic retinal angiogenesis and known to interact with PPARs or PPARGC1A.
Results
A DNA coding variant (rs3736265) and a 3'UTR-resident regulatory variant (rs3774923) in PPARGC1A were independently associated with NV AMD (exact P = 0.003, both SNPs). SNP-SNP interactions existed for NV AMD (P<0.005) with rs3736265 and a AMD-associated variant in complement factor B (CFB, rs512559). PPARGC1A influences activation of the AMD-associated complement component 3 (C3) promoter fragment and CFB influences activation and proteolysis of C3. We observed interaction (P≤0.003) of rs3736265 with a variant in vascular endothelial growth factor A (VEGFA, rs3025033), a key molecule in retinal angiogenesis. Another PPARGC1A coding variant (rs8192678) showed statistical interaction with a SNP in the VEGFA receptor fms-related tyrosine kinase 1 (FLT1, rs10507386; P≤0.003). C3 expression was down-regulated 2-fold in retinas of ω-3 LCPUFA-fed mice – these animals also showed 70% reduction in retinal NV (P≤0.001).
Conclusion
Ligands and co-activators of the ω-3 LCPUFA sensing PPAR-RXR axis may influence retinal angiogenesis in NV AMD via the complement and VEGF signaling systems. We have linked the co-activator of a lipid-sensing transcription factor (PPARG co-activator 1 alpha, PPARGC1A) to age-related macular degeneration (AMD) and AMD-associated genes.
Citation: SanGiovanni JP, Chen J, Sapieha P, Aderman CM, Stahl A, Clemons TE, et al. (2013) DNA Sequence Variants in PPARGC1A, a Gene Encoding a Coactivator of the ω-3 LCPUFA Sensing PPAR-RXR Transcription Complex, Are Associated with NV AMD and AMD-Associated Loci in Genes of Complement and VEGF Signaling Pathways. PLoS ONE 8(1): e53155. https://doi.org/10.1371/journal.pone.0053155
Editor: Lynette Kay Rogers, The Ohio State Unversity, United States of America
Received: June 18, 2012; Accepted: November 26, 2012; Published: January 15, 2013
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: Funding support for NEI-AMD was provided by the National Eye Institute (JPSG); EY017017 and Research to Prevent Blindness (RPB) Senior Investigator Award (LEHS). Additional support was provided by Children’s Hospital Ophthalmology Foundation and Charles H. Hood Foundation (JC), the Canadian Institues of Health Research (Institute of Nutrition, Metabolism and Diabetes) (PS) and the Deutsche Forschungsgemeinschaft, Freifrau von Nauendorf Stiftung and Deutsche Ophthalmologische Gesellschaft (AS). Mouse microarray studies were performed by the Molecular Genetics Core Facility at Children’s Hospital Boston supported by National Institutes of Health NIH-P50-NS40828, and NIH-P30-HD18655. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: Jing Chen acts as an editor at PLOS ONE. Dr. Clemons is an employee of the EMMES Corp., Rockville. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.
Introduction
Neovascular (NV) age-related macular degeneration (AMD) is a common sight-threatening disease in the elderly, accounting for more than 80% of all AMD-related vision loss in people of western European ancestry. [1] The cardinal lesions of NV AMD are proliferative growth of and exudation from vessels in the choriocapillaris, the major vascular network of the outer retina. [2], [3], [4] More than 2 million U.S. residents have advanced AMD. [5] Current treatments (intraocular injections with anti-angiogenic drugs) are a substantial financial burden on society, with direct annual medical costs reaching ∼570 million dollars. AMD-related outpatient services are incurred annually by ∼1.4 million people aged 65-and-older and contribute to ∼0.5 billion dollars in Medicare claims per year. [6] Less expensive and non-invasive treatment options for AMD are needed.
Nutrient-based approaches to AMD treatment have been focused on compounds demonstrating: 1) intake-dependent and –modifiable accretion to retinal cell types affected in AMD; and 2) biophysical and biochemical capacity to act on processes implicated in pathogenesis and pathophysiology. Large-scale human studies on AMD suggest a reduced likelihood of having or progressing to NV AMD among people reporting highest dietary intakes of omega-3 (ω-3) long-chain polyunsaturated fatty acids (LCPUFAs). [7], [8], [9], [10] ω-3 LCPUFAs act as key structural and signaling molecules in the retina. [11] Findings from work on in vivo model systems support the idea that increasing retinal tissue status of these nutrients protects against pathologic intraretinal [12], [13], [14] and choroidal [15], [16] neovascularization. The role of ω-3 LCPUFAs in cell survival and rescue is an emerging area of research, as these nutrients are precursors to families of potent neuroprotective autacoids. [17], [18], [19] Biosynthetic and cleavage enzymes, transporters, receptors, and transcriptional regulators that interact with ω-3 LCPUFAs, their precursors, metabolites, and targets are expressed in retinal areas manifesting neurodegenerative and angiogenic lesions of AMD (Institute of Human Genetics, University of Regensburg, http://www.retinacentral.org/. Accessed 2012 Nov 30). Among these are cognate lipid-sensing nuclear receptors of the peroxisome proliferator-activated receptor (PPAR) family. ω-3 LCPUFAs are endogenous PPAR agonists. [20].
PPARs act as master regulators of gene transcription and have been studied in the context of retinal vascular disease. In 2005 we first discussed the putative role of LCPUFA-PPAR relationships in the retina [11] and have recently demonstrated direct PPAR-gamma (PPARG)-mediated effects of dietary ω-3 LCPUFAs on retinal vessel formation in a oxygen-induced retinopathy (OIR) model of pathologic retinal angiogenesis. [13], [14] Use of synthetic PPAR agonists has been associated with lower likelihood and severity of pathologic retinal neovascularization. Troglitazone, a PPARG agonist, inhibited laser-induced choroidal neovascularization (a hallmark of NV AMD) in cynomoglus monkeys. [21] Fenofibrate, a PPAR activator, reduced the need for laser treatment for proliferative (neovascular) retinopathy in a large phase III clinical trial of 9795 people with type 2 diabetes. [22] The rationale for using PPAR ligands as therapeutic agents for NV AMD has been raised elsewhere. [23].
The PPARG-retinoid X receptor (PPAR-RXR) transcription complex is involved in ligand-activated transcription; this complex typically binds the AGGTCANAGGTCA DNA consensus sequence in peroxisome proliferator hormone response elements (PPREs) within the promoter region of target genes – when the PPAR is bound by a ligand (e.g. LCPUFAs or drugs), transcription is altered. The activity of PPARs is dependent on the shape of their ligand-binding domains and the physical interaction with co-activator and co-repressor proteins. PPAR agonists bind the PPARG-RXR transcription complex, causing a conformational shift that permits displacement of co-repressor proteins and a subsequent docking of co-activator proteins. We examined the possible influence of DNA variation in the PPARG co-activator 1 alpha gene (PPARGC1A), as an extension of the work on endogenous and pharmacologic PPAR agonists in pathologic retinal angiogenesis (discussed in the previous paragraph). PPARGC1A encodes a transcriptional regulator protein involved in constitutive activation of ω-3 LCPUFA-sensing PPARG-RXR complex target genes. PPARGC1A is a major PPARG co-activator. Our results suggest that multiple constituents (ligands and transcriptional co-activators) of the PPAR-RXR system may influence pathogenic processes implicated in NV AMD and offer promise for efficiently examining combined therapies for this blinding disease of public health significance.
Results
We examined DNA sequence variants in PPARGC1A for association with NV AMD in three independent U.S.-based cohorts of western European ancestry. All data are from large-scale projects designed to investigate the molecular genetics of AMD. The panel of 20 sequence variants we tested were taken from the ILLUMINA HumanCNV370v1 microarray and included 2 single nucleotide polymorphisms (SNPs) resident in exonic PPARGC1A regions and one resident in the 3′ untranslated region (UTR) of this gene. We applied a 4-phase approach to testing, first using our largest independent cohort as a ‘discovery’ sample, then examining the magnitude and direction of measures of NV AMD-PPARGC1A relationships in two other large-scale genotyping projects on AMD, and then combining measures of association with meta-analytic techniques – finally, testing these combined estimates with exact methods. After observing relationships of NV AMD with SNPs in exonic and regulatory regions of PPARGC1A from the meta-analysis, we examined interactions of these variants with established genetic loci for NV AMD resident in systems: 1) associated with AMD in other studies (complement, lipid metabolism, and vascular endothelial growth factor (VEGF) signaling systems); and, 2) responsive to ω-3 LCPUFA feeding in our animal models (complement cascade). Figure 1 is a schematic of putative relationships between PPARGC1A, an ω-3 LCPUFA activator of PPARs (docosahexaenoic acid, DHA), PPARs, and genes containing AMD-associated variants (symbols for these genes are colored red).
Genes associated with AMD in extant studies are shaded in red. DHA = docosahexaenoic acid (a PPAR agonist). Diagram was generated with Ingenuity Pathway Analysis® software. Full names for genes represented by symbols exist at http://www.ncbi.nlm.nih.gov/gene/. Direct effects of PPARGC1A are represented by blue arrows. Letters on the arrows represent the nature of evidence and are defined as follows: A = activation, E = expression, L = proteolysis, PP = protein-protein interaction, T = transcription.
In summary, we applied age-, sex-, and smoking-adjusted logistic regression models to analyze 843 people with NV AMD and 1032 of their elderly peers who were both AMD-free and ≥65 years-of-age. A cohort of 506 cases and 512 controls from a major university-based clinical center served as the discovery sample in phase 1. Our phase 2 replication samples consisted of 123+205 cases and 198+314 AMD-free elderly peers from 2 independent and geographically distinct university based research centers. Table 1 contains demographic characteristics of our cohorts. After examining single locus tests for between-cohort concordance, we computed combined age-, sex-, and smoking-adjusted estimates of NV AMD-SNP relationships with meta-regression, applying random effects models to account for sample heterogeneity (phase 3). We then used max(T) permutation (10,000 iterations) on the combined samples to derive exact P-values for sequence variants significant at P<0.005 in covariate-adjusted meta-analysis (phase 4). Two-locus epistatic (SNP-SNP) interactions of AMD-associated PPARGC1A variants were then conducted with established AMD risk loci of genes in the complement, lipoprotein metabolism, and VEGF signaling systems. These genes included complement factor H (CFH), complement component 3 (C3), complement factor B (CFB) complement component 2 (C2), LIPC hepatic lipase (LIPC), ATP-binding cassette transporter A1 (ABCA1), vascular endothelial growth factor A (VEGFA) and its receptors fms-related tyrosine kinase 1 (FLT1, VEGFR1) and kinase insert domain receptor (KDR, VEGFR2, FLK1)).
Single Locus Tests of PPARGC1A Variants for NV AMD
Findings from single locus tests on PPARGC1A are presented in Table 2. A relationship emerged for rs3736265 (odds ratio (OR) = 0.63, exact P = 0.005), a variant in exon 9 yielding a deleterious peptide transition (SIFT score 0.01) from threonine to methionine or lysine at amino acid 612. People carrying the minor allele (AA or AG) were ∼40% less likely than their peers (GG) to have NV AMD. This variant allele was predicted with HaploReg (Broad Institute. www.broadinstitute.org/mammals/haploreg/. Accessed 2012 Nov 30) to enhance affinity to consensus sequence in hypoxia inducible factor 1 beta (aryl hydrocarbon receptor nuclear translocator, ARNT) a heterodimeric transcription factor involved with in hypoxia-induced angiogenesis. rs3736265 is in proximity to a highly conserved RNA binding domain (RNA recognition motif overlapping amino acids 678–738) involved in post-transcriptional gene expression processes (e.g. mRNA and rRNA processing, RNA stability, and RNA export). We observed relationships of the PPARGC1A intronic variant rs3755862 with NV AMD (OR = 0.61, exact P = 0.002); this SNP is in nearly complete linkage disequilibrium (r2 = 0.96) with the rs3736265 coding variant. Our inferences on a PPARGC1A-NV AMD relationship were strengthened by the observation that the 3′UTR in PPARGC1A also contains a NV AMD-associated variant (rs3774923, OR = 0.58, exact P = 0.003). As with SNP rs3736265, people carrying one or two copies of the minor allele showed a 40% reduced likelihood of having NV AMD, relative to their peers who were homozygous for the major allele. We do not have reason to believe that the rs3736265 coding SNP and the rs3774923 UTR SNP are co-inherited, based on measures of linkage disequilibrium (r2 = 0.64 in our analytic sample and r2 = 0.55 within the CEU+TSI cohorts from the International HAPMAP Project). It is important to acknowledge that the 300K SNP chip used for genotyping did not permit dense mapping of PPARGC1A, and thus constrained our range of inferences. Our conclusion from single-locus tests on PPARGC1A variants is that co-activators of PPAR-mediated processes may be reasonably implicated in pathogenesis of NV AMD.
Interaction Testing of NV AMD-associated PPARGC1A SNPs with AMD-associated Variants
Extant relationships depicted in Figure 1 justified our decision to examine statistical interactions of AMD-associated PPARGC1A variants discussed in the section above with those in genes of complement, lipid metabolism, and VEGF signaling systems. In model systems, PPARGC1A has been shown to directly activate promoter elements of C3 [24] and LIPC. [25] C3 and LIPC carry AMD-associated sequence variants. [26], [27] Via co-activation of PPARG [28], PPARGC1A may alter expression of ABCA1 [29], [30], [31] and VEGFA. [32], [33] ABCA1 and VEGFA carry AMD-associated sequence variants. [26], [34], [35], [36] Via co-activation of PPARA [37], [38], [39], [40], PPARGC1A may alter expression of CFH [41], and C2. [42] These genes carry AMD-associated sequence variants. [43], [44] In our cohorts, PPARGC1A SNPs showed statistical interactions with CFB, C3, C2, VEGFA, FLT1, and KDR variants. FLT1 and KDR encode VEGFA receptors.
In examining interactions, we first considered single variant findings for AMD-associated SNPs in CFH, C3, CFB, C2, LIPC, and ABCA1 reported in independent studies. [26], [43], [45] In our cohorts, AMD-associated loci in the complement system genes, LIPC, and ABCA1 also existed (Table S1). VEGF is a key molecule implicated in pathologic retinal angiogenesis variants in VEGFA have been implicated in AMD [46], [47], [48] – although, relationships have not always been replicated. [49], [50], [51], [52] While NV AMD-VEGF relationships did not exist in our cohorts for the SNPs on our microarray feature set (this may have been due to sparse coverage of the gene on the testing panel), we saw value in examining possible interactions with PPARGC1A SNPs.
Table 3 contains results for NV AMD-related interactions of PPARGC1A SNPs significant at P<0.005. Most notable are findings for the PPARGC1A coding variant (rs3736265) with CFB (rs512559, P≤0.0046) and VEGFA (rs3025033, P≤0.0037). CFB influences activation and proteolysis of C3. Commentary exists in the Discussion on a PPARGC1A-VEGF interaction (P≤0.003) for another coding SNP in PPARGC1A (Gly482Ser, rs8192678) with a variant in VEGFA receptor FLT1 (rs10507386) that changes the binding motif of the RXR consensus sequence. Inferences on the interactions of ω-3 LCPUFA-sensing PPARG-RXR complex constituents with complement and angiogenesis pathway genes in NV AMD are further strengthened by findings from: 1) large-scale population-based studies demonstrating interactions of fish (a primary source of ω-3 LCPUFAs) intake with CFH gene variants in early [53], [54] and late AMD [54]; and 2) ω-3 LCPUFA related alterations in C3 expression and attenuation of pathologic retinal neovascularization in in vivo systems (discussed in the section below). Our conclusion from interaction tests on PPARGC1A variants is that PPAR-mediated processes may be reasonably implicated to influence complement and VEGF signaling systems in the pathogenesis of NV AMD. Additional work on this concept is necessary to make conclusive inferences.
Genome-wide Expression Profiling of Murine Retinal Response to LCPUFAs
To consider the role of LCPUFA-regulated genes both associated with pathologic retinal angiogenesis and known to interact with PPARs or PPARGC1A we compared retinal gene expression profiles from an Illumina Mouse-ref 6 microarray in 17-day-old (P17) mice on a diet of 2.0% total fatty acids from ω-6 LCPUFAs (arachidonic acid, C20∶4 ω-6) to those P17 mice on a 2.0% ω-3 LCPUFA diet (1% EPA +1% DHA). Both diets were fed from time of birth. We found a 2.0-fold reduction in retinal expression of the AMD-associated C3 gene among animals fed the 2.0% ω-3 LCPUFA diet (FDR (false discovery rate) ∼ 0.8%). Notably, identical LCPUFA exposures led to retinal tissue LCPUFA status changes and subsequent alterations in the severity of pathologic retinal angiogenesis within the OIR model. Animals receiving ω-3 LCPUFAs showed a ∼20% reduction in retinal vessel loss (P≤0.05) and a ∼70% reduction in pathologic retinal angiogenesis (P≤0.001) (Figure 2). [12] These findings strengthen inferences from work in humans supporting links on the influences of dietary (ω-3 LCPUFAs), pharmacologic (trolitazone, fenofibrate), and constitutive (PPARGC1A) PPAR activators as they may relate to the complement and VEGF signaling systems, and development of neovascular pathology of the retina.
C57 BL/6 mouse pups fed with ω-3 or ω-6 PUFA enhanced diet were exposed to oxygen-induced retinopathy. Retinas were flat mounted at postnatal day (P) 17 to visualize vasculature with contralateral retinas from the same mice isolated for gene array analysis. (a). Representative retina vasculature stained with isolectin B4 shows vaso-obliteration and pathologic neovascularization in ω-6 or ω-3 fed mice. (ω-6, n = 7 and ω-3, n = 8). Scale bar: 1mm. Quantification of (b) vaso-obliteration and (c) neovascularization in ω-6 or ω-3 fed mice. * P≤0.05, *** P≤0.001.
Discussion
We tested two highly conserved DNA sequence variants that code for changes in the peptide structure (rs3736265 and rs8192678) and one highly conserved SNP (rs3774923) in the regulatory 3′UTR of PPARGC1A; all were associated with NV AMD. An association of NV AMD with rs3736265 emerged from exact tests in age-, sex-, and smoking-adjusted models. Statistical interactions existed between rs3736265 and SNPs resident in AMD-associated loci of genes encoding key factors in complement and VEGF signaling systems. rs8192678 was associated with NV AMD through an interaction with a SNP in the VEGFA receptor FLT1. rs3774923 was related with NV AMD in single locus analyses.
Our findings extend evidence implicating endogenous (ω-3 LCPUFAs) and pharmacologic (troglitazone, fenofibrate) PPAR agonists/activators as protective agents against pathologic retinal angiogenesis to support the action of a constitutive PPAR co-activator protein (PPARGC1A) in a similar capacity. PPARGC1A may serve as a hub molecule influencing AMD pathogenesis and pathophysiology as it activates the glitazone target PPARG [28], [55], the fenofibrate target PPARA [37], [55], and C3 [24], LIPC [25], and VEGF [33] promoter fragments. Glitazones have been tested in model systems for their protective effects on retinal cell survival. [56] Troglitazone inhibited choroidal neovascularization in a in vivo primate model of NV AMD. [21] Fenofibrate, a synthetic PPAR activator, reduced the need for laser treatment for proliferative diabetic retinopathy. [22] Chemical structures for PPAR agonists exist at (Kanehisa Laboratories, www.genome.jp/kegg/pathway/map/map07222.html. Accessed 2012 Nov 30); those for RXR agonists and antagonists are presented at (Kanehisa Laboratories, www.genome.jp/kegg/pathway/map/map07223.html. Accessed 2012 Nov 30).
We did not observe single locus associations of NV AMD with the VEGFA, FLT1, or KDR variants present on our microarray test panel. However, interaction of PPARGC1A SNPs with those in each of these genes existed at P-values ≤0.005. Notable PPARGC1A-VEGF interactions emerged for coding SNPs in PPARGC1A: rs8192678 with a variant in FLT1 (rs10507386, P≤0.003); and rs3736265 with a variant in VEGFA (rs3025033, P≤0.004). PPARGC1A rs8192678 is resident in the conserved consensus sequence for the EVI1 transcription factor binding site. The FLT1 rs10507386 variant with which it shows statistical interaction also changes the binding motif of the RXR consensus sequence.
There is a link between PPARGC1A, VEGF, and the estrogen signaling system. Arany et al. demonstrated a PPARGC1A-dependent regulation of VEGF via coactivation by estrogen related receptor alpha (ESRRA) and binding to the VEGF promoter sequence. [33] Our microarray feature set did not contain ESRRA variants. We have conducted preliminary work on a network of genes encoding constituents of a signaling system with the capacity to impact VEGFA via PPARGC1A- and ESRRA-mediated processes. Variants in estrogen receptor 1 (ESR1, rs1999805, Pmeta ≤0.002), estrogen related receptor beta (ESRRB, rs2361290, Pmeta ≤0.009), and estrogen related receptor gamma (ESRRG, rs1984137 and 2820879, respective Pmeta ≤0.01, 0.007) showed weak relationships with NV AMD. These genes encode proteins that interact with PPARGC1 and ESRRA and may impact VEGF signaling directly (ESR1) or via PPARs (ESRRB/ESRRG). PPARGC1A-VEGF relationships are germane to the present study since a number of drugs acting on VEGF or VEGF signaling have been tested in large-scale trials for treatment of advanced AMD [57], [58]; these include: the anti-VEGF monoclonal antibody bevacizumab [59] (Avastin), the VEGFA antibody ranibizumab [59], [60] (Lucentis), and the anti-VEGF165 aptamer pegaptanib [60] (Macugen).
We observed single locus relationships of ABCA1 and LIPC SNPs on NV AMD, but no interactions with the PPARGC1A SNPs on our test panel. Because PPARGC1A has both the capacity to alter ABCA1 expression through its interaction with PPARG [30], [61] and to activate the LIPC promoter fragment [25], we believe this is a promising area for future work.
In conclusion, we propose that constituents of the ω-3 LCPUFA sensing PPAR-RXR axis have the capacity to act on processes impacting pathologic retinal angiogenesis via complement and VEGF signaling systems. A number of FDA-approved drugs targeting constituents of the axis now exist. Testing combinations of endogenous and pharmacologic PPAR agonists/activators ligands and compounds that influence PPAR co-activator proteins in pre-clinical studies on NV AMD may elucidate promising therapies for this complex blinding disease of public health significance.
Materials and Methods
Large-scale Genotyping Study in Elderly Humans
Data used for human genetic analyses in this report were obtained from the NEI Study of Age-Related Macular Degeneration (NEI-AMD) Database at the U.S. National Center for Biotechnology Information (NCBI) database of Genotypes and Phenotypes (dbGaP). NEI-AMD is a collaborative of researchers from the University of Michigan (Ann Arbor, MI), Mayo Clinic (Rochester, MN), University of Pennsylvania (Philadelphia, PA), and the Age-related Eye Disease Study (AREDS) group including National Eye Institute intramural investigators. Institutional review boards at each NEI-AMD study site reviewed and approved the study protocols. Each participant provided written informed consent in accordance with the Declaration of Helsinki.
Subjects and Study Design.
Our analytic sample contained, respectively 506, 123, 205 people with NV AMD, and 514, 198, 318 AMD-free people (age ≥65 years) from the University of Michigan, University of Pennsylvania, and The Mayo Clinic. Details on the NEI-AMD genome-wide association (GWA) study and links to peer-reviewed publications from the project exist at: http://www.ncbi.nlm.nih.gov/gap/?term=MMAP.
Outcome Ascertainment.
Experienced graders (ophthalmologists) classified outcomes according to AMD diagnosis in the worse eye. All participants had negative history of: 1) severe macular disease or vision loss onset prior to 40-years-of-age; 2) juvenile retinal degeneration, macular damage resulting from ocular trauma, retinal detachment, high myopia, chorioretinal infection/inflammatory disease, or choroidal dystrophy; and, 3) retinal insult that would render the fundus ungradable. Existence of neovascularization in at least one eye, according to diagnostic criteria established by the International Age-Related Maculopathy Epidemiological Study and the Modified Wisconsin Age-Related Maculopathy Grading System, was the basis for classifying people with NV AMD. In all cases of unilateral NV AMD, drusen or pigment changes also existed in the fellow eye. The likelihood of developing AMD increases 2-to-6 fold after age 75 and it was therefore essential to select our oldest AMD-free participants to minimize the potential for non-random misclassification (false negatives) in the youngest members of the control group. Our AMD-free comparison group was composed of people ≥65-years-of-age who had no large or intermediate drusen in either eye; these participants received examinations and gradings by the NEI-AMD study ophthalmologists. If small drusen or pigment changes were present in the AMD-free group, they were neither bilateral nor extensive (≤5).
Array-Based SNP Genotyping.
All NEI-AMD specimens were genotyped with DNA microarrays at the Johns Hopkins University Center for Inherited Disease Research (CIDR, Baltimore, MD, USA) using the ILLUMINA HumanCNV370v1 chip (SNP batch IDs at http://www.ncbi.nlm.nih.gov/SNP/snp_viewBatch.cgi?sbid=1047132) with the Illumina Infinium II assay protocol. The Illumina BeadStudio Genotyping module (version 3.2.32) was used with the combined intensity of 99% of the samples to assign allele cluster definitions. The threshold for genotype calls was a gencall score ≥0.25. Reproducibility of blind duplicate samples was 99.992%. All sequence variants analyzed for the current study passed process quality and analytic filters for missingness (<5%), minor allele frequency (>1%) and Hardy-Weinberg equilibrium (HWE P<1 x 10-6 in the AMD-free group).
Bioinformatics.
We used positional coordinates (±1000 base pairs) to analyze PPARGC1A, and AMD-associated variants in genes of the complement, lipid metabolism, and VEGF signaling systems. To permit a deeper inference on our findings, we used public-access databases to annotate AMD-associated variants for residence within exons, consensus sequences of highly conserved transcription factor binding sites, epigenetic marks in histone protein H3 (mono- and tri-methylation and acetylation), DNase I hypersensitivity regions, and CpG islands.
Statistical Analyses.
We used Plink (version 1.07, http://pngu.mgh.harvard.edu/purcell/plink/) and SAS (version 9.1, Cary, NC) software for data analysis, first examining the allelic distributions of SNPs in people with NV AMD (relative to the AMD-free comparison group) with age-, sex-, and smoking-adjusted logistic regression analyses. Genotype was coded using additive, dominant (grouping minor allele homozygotes with heterozygotes), and recessive (grouping major allele homozygotes with heterozygotes) models of inheritance to obtain odds ratios (ORs) for variants within the discovery cohort (University of Michigan). Additive, dominant, and recessive models were run in the two replication cohorts. Combined ORs for single locus tests were computed across cohorts with results from each of the three models using age-, sex-, and smoking-adjusted meta-regression. Combined estimates were only computed within a given model (e.g. results from the additive model in the discovery cohort were only combined with results from the additive models run on the replication cohort). Sample heterogeneity was assessed with Cochrane’s Q statistic and random effects models were applied when indicated.
For variants attaining significance in the meta-analysis, we applied exact tests on empirical distributions of P-values generated with a max(T) permutation procedure set to 10000 iterations. Permutation procedures permit the computation of significance levels from empirically derived distributions. Exact P-values yielded by the procedure have tractable properties in obviating constraints of small sample sizes, while providing a framework for correction for multiple testing, and controlling for population substructure. In our cohorts of unrelated individuals, we swapped data values with the assumption that individuals are interchangeable under the null – this permitted construction of a new dataset sampled under the null hypothesis. Through the permutation approach only the phenotype-genotype relationship is destroyed (patterns of linkage disequilibrium between sequence variants will be preserved under the observed and permuted samples). As permutation methods sustain the correlational structure between SNPs, the approach provides a less stringent correction for multiple testing than the Bonferroni test (which assumes all tests are independent). As such, the corrected P-value is the relevant construct, so it is usually sufficient to apply a much smaller number of tests; resulting P-values ≥0.05 were considered significant.
Tests of SNP × SNP interactions (allelic by allelic epistasis) were conducted for PPARGC1A with AMD-related SNPs in complement, lipid metabolism, and, VEGF signaling genes using Plink. The analytic models were based on allele dosage. For each SNP (e.g. SNP A and B) the model took the form of Y ∼ β0+ β1A+β2B+β3AB+e. The test for interaction was based on the coefficient β3.
LCPUFA Intake and Retinal Gene Expression in Mice
Our animal study adhered to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and was approved by the Children’s Hospital Boston Animal Care and Use Committee. C57BL/6J mice (stock number 000664, the Jackson Laboratory) were used for the study. Beginning at postnatal day 0 (P0), nursing mothers were fed diets enriched with either 2% ω-3 (eicosapentaenoate+docosahexaenoate) to or 2% ω-6 (arachidonate) LCPUFAs [12]. To induce vessel loss, and subsequent pathological neovascularization, nursing mothers and pups were exposed to 75% oxygen from P7 to P12 and returned to room air and sacrificed at P17. Retinas from each group were isolated and flash frozen using RNase-free techniques. Total RNA was extracted and prepared for Illumina microarray analysis using the Mouse-ref 6 chip (n = 3 biological replicates for each diet group). The chip contained ∼45,000 probe sets representing ∼34,000 genes. Microarray studies, from cDNA synthesis to raw data normalization were performed by the Molecular Genetics Core Facility at Children’s Hospital Boston. Briefly, total RNA (1 µg each) were reverse transcribed, followed by a single in vitro transcription amplification to incorporate biotin-labeled nucleotide, and subsequent hybridization and staining with strepatavidin-Cy3 according to the manufacturer’s instructions. Data were acquired using the Illumina BeadStudio software and analyzed for quality control, background analysis and normalization with rank invariant algorithm. Further analysis was performed using Significance Analysis of Microarray (SAM), Gene Set Enrichment Analysis (GSEA), and J-Express Pro 2.7 software. We profiled retinal gene expression with the 2% LCPUFA feeding paradigm identify LCPUFA-regulated genes both associated with pathologic NV and involved in PPAR-mediated processes.
Supporting Information
Table S1.
NV AMD-associated variants in complement pathway genes, LIPC, and ABCA1. Abbreviations: CHR, chromosome; SNP, single-nucleotide polymorphism. Full names for the genes listed in the ‘Gene Symbol’ column exist at: http://www.ncbi.nlm.nih.gov/gene. Combined estimates (ORmeta) were computed with age-, sex, and smoking-adjusted meta-regression – random effects models were applied in instances indicated by Cochrane’s Q statistic. All P values are 2-sided.
https://doi.org/10.1371/journal.pone.0053155.s001
(DOCX)
Acknowledgments
We thank the NEI-AMD participants and the NEI-AMD Research Groups for their valuable contributions to this research. The data used for the genetic analyses were obtained from the NEI Study of Age-Related Macular Degeneration (NEI-AMD) Database found at: http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000182.v2.p1. Accessed 2012 Nov 30.
Author Contributions
Conceived and designed the experiments: JPSG JC PS LEHS. Performed the experiments: JPSG JC PS CMA. Analyzed the data: JPSG JC CMA. Contributed reagents/materials/analysis tools: TEC EYC LEHS. Wrote the paper: JPSG JC PS CMA AS TEC EYC LEHS.
References
- 1. Congdon N, O’Colmain B, Klaver CC, Klein R, Munoz B, et al. (2004) Causes and prevalence of visual impairment among adults in the United States. Arch Ophthalmol 122: 477–485.
- 2.
Hageman GS, Gehrs K, Johnson LV, Anderson D (2008) Age-Related Macular Degeneration (AMD). In: Kolb H, editor. Webvision: The Organization of the Retina and Visual System. 2011/03/18 ed. Salt Lake City (UT). Available: http://www.ncbi.nlm.nih.gov/books/NBK27323/. Accessed 2012 Nov 30.
- 3. Bird AC (2010) Therapeutic targets in age-related macular disease. J Clin Invest 120: 3033–3041.
- 4. Ambati J, Ambati BK, Yoo SH, Ianchulev S, Adamis AP (2003) Age-related macular degeneration: etiology, pathogenesis, and therapeutic strategies. Surv Ophthalmol 48: 257–293.
- 5. Friedman DS, O'Colmain BJ, Munoz B, Tomany SC, McCarty C, et al. (2004) Prevalence of age-related macular degeneration in the United States. Arch Ophthalmol 122: 564–572.
- 6.
(2007) Prevent Blindness America. The economic impact of vision problems. Available: http://www.preventblindness.net/site/DocServer/Impact_of_Vision_Problems.pdf?docID=1321. Accessed 2012 Nov 30.
- 7. Chong EW, Kreis AJ, Wong TY, Simpson JA, Guymer RH (2008) Dietary omega-3 fatty acid and fish intake in the primary prevention of age-related macular degeneration: a systematic review and meta-analysis. Arch Ophthalmol 126: 826–833.
- 8. Weikel KA, Chiu CJ, Taylor A (2012) Nutritional modulation of age-related macular degeneration. Mol Aspects Med 33: 318–375.
- 9. SanGiovanni JP, Agron E, Clemons TE, Chew EY (2009) Omega-3 long-chain polyunsaturated fatty acid intake inversely associated with 12-year progression to advanced age-related macular degeneration. Arch Ophthalmol 127: 110–112.
- 10. SanGiovanni JP, Agron E, Meleth AD, Reed GF, Sperduto RD, et al. (2009) {omega}-3 Long-chain polyunsaturated fatty acid intake and 12-y incidence of neovascular age-related macular degeneration and central geographic atrophy: AREDS report 30, a prospective cohort study from the Age-Related Eye Disease Study. Am J Clin Nutr 90: 1601–1607.
- 11. SanGiovanni JP, Chew EY (2005) The role of omega-3 long-chain polyunsaturated fatty acids in health and disease of the retina. Prog Retin Eye Res 24: 87–138.
- 12. Connor KM, SanGiovanni JP, Lofqvist C, Aderman CM, Chen J, et al. (2007) Increased dietary intake of omega-3-polyunsaturated fatty acids reduces pathological retinal angiogenesis. Nat Med 13: 868–873.
- 13. Sapieha P, Stahl A, Chen J, Seaward MR, Willett KL, et al. (2011) 5-Lipoxygenase Metabolite 4-HDHA Is a Mediator of the Antiangiogenic Effect of ω-3 Polyunsaturated Fatty Acids. Science Translational Medicine 3: 69ra12.
- 14. Stahl A, Sapieha P, Connor KM, Sangiovanni JP, Chen J, et al. (2010) Short communication: PPAR gamma mediates a direct antiangiogenic effect of omega 3-PUFAs in proliferative retinopathy. Circ Res 107: 495–500.
- 15. Koto T, Nagai N, Mochimaru H, Kurihara T, Izumi-Nagai K, et al. (2007) Eicosapentaenoic acid is anti-inflammatory in preventing choroidal neovascularization in mice. Invest Ophthalmol Vis Sci 48: 4328–4334.
- 16. Sheets KG, Zhou Y, Ertel MK, Knott EJ, Regan CE Jr, et al. (2010) Neuroprotectin D1 attenuates laser-induced choroidal neovascularization in mouse. Mol Vis 16: 320–329.
- 17. Mukherjee PK, Marcheselli VL, Serhan CN, Bazan NG (2004) Neuroprotectin D1: a docosahexaenoic acid-derived docosatriene protects human retinal pigment epithelial cells from oxidative stress. Proc Natl Acad Sci U S A 101: 8491–8496.
- 18. Halapin NA, Bazan NG (2010) NPD1 induction of retinal pigment epithelial cell survival involves PI3K/Akt phosphorylation signaling. Neurochem Res 35: 1944–1947.
- 19. Bazan NG, Calandria JM, Serhan CN (2010) Rescue and repair during photoreceptor cell renewal mediated by docosahexaenoic acid-derived neuroprotectin D1. J Lipid Res 51: 2018–2031.
- 20. Vanden Heuvel JP (2012) Nutrigenomics and nutrigenetics of omega3 polyunsaturated fatty acids. Prog Mol Biol Transl Sci 108: 75–112.
- 21. Murata T, He S, Hangai M, Ishibashi T, Xi XP, et al. (2000) Peroxisome proliferator-activated receptor-gamma ligands inhibit choroidal neovascularization. Invest Ophthalmol Vis Sci 41: 2309–2317.
- 22. Keech AC, Mitchell P, Summanen PA, O’Day J, Davis TME, et al. (2007) Effect of fenofibrate on the need for laser treatment for diabetic retinopathy (FIELD study): a randomised controlled trial. The Lancet 370: 1687–1697.
- 23. Del V Cano M, Gehlbach PL (2008) PPAR-alpha Ligands as Potential Therapeutic Agents for Wet Age-Related Macular Degeneration. PPAR Res 2008: 821592.
- 24. Kressler D, Schreiber SN, Knutti D, Kralli A (2002) The PGC-1-related protein PERC is a selective coactivator of estrogen receptor alpha. J Biol Chem 277: 13918–13925.
- 25. Rufibach LE, Duncan SA, Battle M, Deeb SS (2006) Transcriptional regulation of the human hepatic lipase (LIPC) gene promoter. J Lipid Res 47: 1463–1477.
- 26. Neale BM, Fagerness J, Reynolds R, Sobrin L, Parker M, et al. (2010) Genome-wide association study of advanced age-related macular degeneration identifies a role of the hepatic lipase gene (LIPC). Proc Natl Acad Sci U S A 107: 7395–7400.
- 27. Maller JB, Fagerness JA, Reynolds RC, Neale BM, Daly MJ, et al. (2007) Variation in complement factor 3 is associated with risk of age-related macular degeneration. Nat Genet 39: 1200–1201.
- 28. Puigserver P, Adelmant G, Wu Z, Fan M, Xu J, et al. (1999) Activation of PPARgamma coactivator-1 through transcription factor docking. Science 286: 1368–1371.
- 29. Glass CK, Witztum JL (2001) Atherosclerosis. the road ahead. Cell 104: 503–516.
- 30. Son NH, Park TS, Yamashita H, Yokoyama M, Huggins LA, et al. (2007) Cardiomyocyte expression of PPARgamma leads to cardiac dysfunction in mice. J Clin Invest 117: 2791–2801.
- 31. Yano M, Matsumura T, Senokuchi T, Ishii N, Murata Y, et al. (2007) Statins activate peroxisome proliferator-activated receptor gamma through extracellular signal-regulated kinase 1/2 and p38 mitogen-activated protein kinase-dependent cyclooxygenase-2 expression in macrophages. Circ Res 100: 1442–1451.
- 32. Fauconnet S, Lascombe I, Chabannes E, Adessi GL, Desvergne B, et al. (2002) Differential regulation of vascular endothelial growth factor expression by peroxisome proliferator-activated receptors in bladder cancer cells. J Biol Chem 277: 23534–23543.
- 33. Arany Z, Foo SY, Ma Y, Ruas JL, Bommi-Reddy A, et al. (2008) HIF-independent regulation of VEGF and angiogenesis by the transcriptional coactivator PGC-1alpha. Nature 451: 1008–1012.
- 34. Cipriani V, Leung HT, Plagnol V, Bunce C, Khan JC, et al. (2012) Genome-wide association study of age-related macular degeneration identifies associated variants in the TNXB-FKBPL-NOTCH4 region of chromosome 6p21.3. Hum Mol Genet 21: 4138–4150.
- 35.
Smailhodzic D, Muether PS, Chen J, Kwestro A, Zhang AY, et al.. (2012) Cumulative Effect of Risk Alleles in CFH, ARMS2, and VEGFA on the Response to Ranibizumab Treatment in Age-Related Macular Degeneration. Ophthalmology.
- 36. Yu Y, Reynolds R, Fagerness J, Rosner B, Daly MJ, et al. (2011) Association of variants in the LIPC and ABCA1 genes with intermediate and large drusen and advanced age-related macular degeneration. Invest Ophthalmol Vis Sci 52: 4663–4670.
- 37. Barger PM, Browning AC, Garner AN, Kelly DP (2001) p38 mitogen-activated protein kinase activates peroxisome proliferator-activated receptor alpha: a potential role in the cardiac metabolic stress response. J Biol Chem 276: 44495–44501.
- 38. Lee Y, Yu X, Gonzales F, Mangelsdorf DJ, Wang MY, et al. (2002) PPAR alpha is necessary for the lipopenic action of hyperleptinemia on white adipose and liver tissue. Proc Natl Acad Sci U S A 99: 11848–11853.
- 39. McGill JK, Beal MF (2006) PGC-1alpha, a new therapeutic target in Huntington’s disease? Cell 127: 465–468.
- 40. Miura S, Kai Y, Ono M, Ezaki O (2003) Overexpression of peroxisome proliferator-activated receptor gamma coactivator-1alpha down-regulates GLUT4 mRNA in skeletal muscles. J Biol Chem 278: 31385–31390.
- 41. Anderson SP, Dunn C, Laughter A, Yoon L, Swanson C, et al. (2004) Overlapping transcriptional programs regulated by the nuclear receptors peroxisome proliferator-activated receptor alpha, retinoid X receptor, and liver X receptor in mouse liver. Mol Pharmacol 66: 1440–1452.
- 42. Leuenberger N, Pradervand S, Wahli W (2009) Sumoylated PPARalpha mediates sex-specific gene repression and protects the liver from estrogen-induced toxicity in mice. J Clin Invest 119: 3138–3148.
- 43. Chen W, Stambolian D, Edwards AO, Branham KE, Othman M, et al. (2010) Genetic variants near TIMP3 and high-density lipoprotein-associated loci influence susceptibility to age-related macular degeneration. Proc Natl Acad Sci U S A 107: 7401–7406.
- 44. Klein RJ, Zeiss C, Chew EY, Tsai JY, Sackler RS, et al. (2005) Complement factor H polymorphism in age-related macular degeneration. Science 308: 385–389.
- 45. Kopplin LJ, Igo RP Jr, Wang Y, Sivakumaran TA, Hagstrom SA, et al. (2010) Genome-wide association identifies SKIV2L and MYRIP as protective factors for age-related macular degeneration. Genes Immun 11: 609–621.
- 46. Churchill AJ, Carter JG, Lovell HC, Ramsden C, Turner SJ, et al. (2006) VEGF polymorphisms are associated with neovascular age-related macular degeneration. Hum Mol Genet 15: 2955–2961.
- 47. Almeida LN, Melilo-Carolino R, Veloso CE, Pereira PA, Miranda DM, et al. (2012) Homozygosity for the +674C>T polymorphism on VEGF gene is associated with age-related macular degeneration in a Brazilian cohort. Graefes Arch Clin Exp Ophthalmol 250: 185–189.
- 48. Galan A, Ferlin A, Caretti L, Buson G, Sato G, et al. (2010) Association of age-related macular degeneration with polymorphisms in vascular endothelial growth factor and its receptor. Ophthalmology 117: 1769–1774.
- 49. Brion M, Sanchez-Salorio M, Corton M, de la Fuente M, Pazos B, et al. (2011) Genetic association study of age-related macular degeneration in the Spanish population. Acta Ophthalmol 89: e12–22.
- 50. Qu Y, Dai H, Zhou F, Zhang X, Xu X, et al. (2011) Vascular endothelial growth factor gene polymorphisms and risk of neovascular age-related macular degeneration in a Chinese cohort. Ophthalmic Res 45: 142–148.
- 51. Mori K, Horie-Inoue K, Gehlbach PL, Takita H, Kabasawa S, et al. (2010) Phenotype and genotype characteristics of age-related macular degeneration in a Japanese population. Ophthalmology 117: 928–938.
- 52. Richardson AJ, Islam FM, Guymer RH, Cain M, Baird PN (2007) A tag-single nucleotide polymorphisms approach to the vascular endothelial growth factor-A gene in age-related macular degeneration. Mol Vis 13: 2148–2152.
- 53. Ho L, van Leeuwen R, Witteman JC, van Duijn CM, Uitterlinden AG, et al. (2011) Reducing the genetic risk of age-related macular degeneration with dietary antioxidants, zinc, and omega-3 fatty acids: the Rotterdam study. Arch Ophthalmol 129: 758–766.
- 54. Wang JJ, Rochtchina E, Smith W, Klein R, Klein BE, et al. (2009) Combined effects of complement factor H genotypes, fish consumption, and inflammatory markers on long-term risk for age-related macular degeneration in a cohort. Am J Epidemiol 169: 633–641.
- 55. Miura S, Kai Y, Kamei Y, Ezaki O (2008) Isoform-specific increases in murine skeletal muscle peroxisome proliferator-activated receptor-gamma coactivator-1alpha (PGC-1alpha) mRNA in response to beta2-adrenergic receptor activation and exercise. Endocrinology 149: 4527–4533.
- 56. Pershadsingh HA, Moore DM (2008) PPARgamma Agonists: Potential as Therapeutics for Neovascular Retinopathies. PPAR Res 2008: 164273.
- 57. Menon G, Walters G (2009) New paradigms in the treatment of wet AMD: the impact of anti-VEGF therapy. Eye (Lond) 23 Suppl 1S1–7.
- 58.
Vedula SS, Krzystolik MG (2008) Antiangiogenic therapy with anti-vascular endothelial growth factor modalities for neovascular age-related macular degeneration. Cochrane Database Syst Rev: CD005139.
- 59.
Subramanian ML, Ness S, Abedi G, Ahmed E, Daly M, et al.. (2009) Bevacizumab vs ranibizumab for age-related macular degeneration: early results of a prospective double-masked, randomized clinical trial. Am J Ophthalmol 148: 875–882 e871.
- 60. Takeda AL, Colquitt J, Clegg AJ, Jones J (2007) Pegaptanib and ranibizumab for neovascular age-related macular degeneration: a systematic review. Br J Ophthalmol 91: 1177–1182.
- 61. Chawla A, Boisvert WA, Lee CH, Laffitte BA, Barak Y, et al. (2001) A PPAR gamma-LXR-ABCA1 pathway in macrophages is involved in cholesterol efflux and atherogenesis. Mol Cell 7: 161–171.