HTRA1 Promoter Polymorphism in Wet Age-Related Macular Degeneration

See allHide authors and affiliations

Science  10 Nov 2006:
Vol. 314, Issue 5801, pp. 989-992
DOI: 10.1126/science.1133807


Age-related macular degeneration (AMD), the most common cause of irreversible vision loss in individuals aged older than 50 years, is classified as either wet (neovascular) or dry (nonneovascular). Inherited variation in the complement factor H gene is a major risk factor for drusen in dry AMD. Here we report that a single-nucleotide polymorphism in the promoter region of HTRA1, a serine protease gene on chromosome 10q26, is a major genetic risk factor for wet AMD. A whole-genome association mapping strategy was applied to a Chinese population, yielding a P value of <10–11. Individuals with the risk-associated genotype were estimated to have a likelihood of developing wet AMD 10 times that of individuals with the wild-type genotype.

Age-related macular degeneration (AMD) is the leading cause of vision loss and blindness among older individuals in the United States and throughout the developed world. It has a complex etiology involving genetic and environmental factors. AMD is broadly classified as either dry (nonneovascular) or wet (neovascular). The dry form is more common and accounts for about 85 to 90% of patients with AMD; it does not typically result in blindness. The primary clinical sign of dry AMD is the presence of soft drusen with indistinct margins (extracellular protein deposits) between the retinal pigment epithelium (RPE) and Bruch's membrane. The large accumulation of these drusen is associated with central geographic atrophy (CGA) and results in blurred central vision. About 10% of AMD patients have the wet form, in which new blood vessels form and break beneath the retina [choroidal neovascularization (CNV)]. This leakage causes permanent damage to surrounding retinal tissue, distorting and destroying central vision. The factors that underlie why some individuals develop the more aggressive wet form of AMD, while others have the slowly progressing dry type, are not yet understood.

Complement factor H (CFH) has been suggested to mediate drusen formation (1). In our previous whole-genome association study, in which the presence of large drusen was the primary phenotype under investigation, the CFH Y402H variant, in which Tyr402 is replaced by His, was shown to be a major genetic risk factor (2). More recently, it has been reported that the highest odds ratio for CFH Y402H was seen for cases with AMD grade 4 (i.e., the presence of CGA) in comparison with AMD grade 1 controls (3). An association between AMD and CFH Y402H, as well as other intronic CFH variants, has been demonstrated for more than ten different Caucasian populations (47).

In this report, we present results of investigations aimed at identifying novel genetic variant(s) that predispose individuals to the wet, neovascular AMD phenotype. Unlike our previous studies which were conducted in a Caucasian cohort (2), this investigation focused on patients of Asian descent. There were several reasons for this decision. First, epidemiological observations indicate that neovascular AMD is more prevalent among Asians than Caucasians (810), and the soft indistinct drusen that are characteristic of dry AMD are rarely seen in Asian individuals (1113). Consistent with these clinical observations, the CFH Y402H variant has been shown to occur less frequently in individuals of Japanese and Chinese ancestry (<5%) than in Caucasians (>35%) (1416).

From a cohort of Southeast Asians in Hong Kong, we identified 96 patients previously diagnosed with wet AMD and 130 age-matched control individuals who were AMD-free (17, 18). Retinal fundus photographs were examined from each of the 226 study participants. Indocyanine Green dye angiography was performed to exclude cases with polypoidal choroidal vasculopathy and to verify that CNV (AMD grade 5) was present in all cases (fig. S1). The AMD cases and controls had a mean age of 74 (19). Other characteristics of the study population are summarized in table S1.

We conducted a whole-genome association study on this Asian cohort to scan for single nucleotide polymorphisms (SNPs), using previously described genotyping and data-quality surveillance procedures (2). Of the 97,824 autosomal SNPs that were informative and passed our quality-control checks (19), rs10490924 was the only polymorphism that showed a significant association with AMD when we used the Bonferroni criteria (Table 1). The allele-frequency chi-square test yielded a P value of 4.1 × 10–12 (Fig. 1A and Table 1). The odds ratio was 11.1 (95% confidence interval [CI] 4.83 to 25.69) for those carrying two copies of the risk allele when compared with wild-type homozygotes, but was indistinguishable from unity, 1.7 (95% CI 0.75 to 3.68), for those having a single risk allele. The risk homozygote accounted for 86% of the population attributable risk (PAR), although this number may be artificially inflated, because the risk allele was carried by more than half (∼55%) of the AMD cohort (Table 1). When likelihood ratio tests were adjusted for gender and smoking status or when genomic control methods were applied to control for population stratification, there was little change in significance levels (19).

Fig. 1.

Genome-wide distribution of P values and characteristics of the associated genomic region. (A) Distribution of P values for the SNPs in a whole-genome association study of AMD. P values are plotted as the –log10(P) with the SNPs in chromosomal order along the x axis. The dashed horizontal line indicates the Bonferroni adjusted threshold for significant association at the 0.05 level. (B) A schematic of the genes in the 4-gamete (19) region on chromosome 10q26, as well as the location of SNPs genotyped by microarrays (+) and identified through sequencing (|). SNP rs10490924 is labeled “8,” and rs11200638 is marked with an asterisk. Above are the gene conservation data obtained from the University of California Santa Cruz (UCSC) Golden Path database for the 4-gamete region. The data show the degree of evolutionary conservation among 17 species by using the multiz alignment; an increase in the height of the bar indicates an increase in the level of conservation.

Table 1.

Association, odds ratios, and PAR for AMD in a Chinese population. Odds ratio and PAR compare the likelihood of AMD in individuals with the listed genotype of risk allele versus those homozygous for the wild-type allele.

Heterozygous risk versus wild-type homozygotesHomozygous risk versus wild-type homozygotes
SNP (alleles)Risk alleleAllelic χ2 nominal POdds ratio (95% CI)PAR (%) (95%CI)Odds ratio (95%CI)PAR (%) (95% CI)
rs10490924(G/T) T 4.08 × 10-12 1.66 (0.75-3.68) 29 (0-63) 11.14 (4.83-25.69) 86 (69-94)
rs11200638(G/A) A 8.24 × 10-12 1.60 (0.71-3.61) 27 (0-61) 10.00 (4.38-22.82) 84 (66-93)

SNP rs10490924 resides between two genes on chromosome 10q26 (Fig. 1B): PLEKHA1 encoding a pleckstrin homology domain–containing protein (GenBank ID 59338) and HTRA1 encoding a heat shock serine protease also known as PRSS11 (GenBank ID 5654). The low sequence homology across species in the intergenic region containing rs10490924 indicates that it is not evolutionarily conserved (Fig. 1B). Chromosome 10q26 has been linked to AMD in many independent family studies, and this linkage region was previously narrowed to SNP rs10490924 (5). SNP rs10490924 was originally thought to result in a protein coding change in the hypothetical locus LOC387715 (20, 21). Because only a single cDNA sequence was found in placental tissue, LOC387715 was subsequently removed from the GenBank database. Thus, it is reasonable to hypothesize that SNP rs10490924 is a surrogate marker that is correlated, or is in linkage disequilibrium (LD), with the putative AMD disease–causing variant in the vicinity. Unfortunately, haplotype analyses using our genotype data or data from the International HapMap Project were unsuccessful in identifying where the functional site resides (19).

We therefore sequenced the entire local genomic region, including promoters, exons, and intron-exon junctions of both PLEKHA1 and HTRA1, in search of the functional variant. We sequenced 50 cases that were homozygous for the risk allele and 38 controls that were homozygous for the wild-type allele on the basis of the genotypes of the marker SNP rs10490924. Of the 43 SNPs or insertion/deletion polymorphisms identified (Fig. 1B and table S6), one SNP (rs11200638), located 512 base pairs (bp) upstream of the HTRA1 putative transcriptional start site and 6096 bp downstream of SNP rs10490924, exhibited a complete LD pattern with SNP rs10490924. Genotyping of the entire cohort revealed that SNP rs11200638 occurred at frequencies similar to those for SNP rs10490924 (P = 8.2 × 10–12 for the allele-association chisquare test), and the two SNPs were almost in complete LD (D' > 0.99) (Table 1).

Computational analysis (19) of the HTRA1 promoter sequence predicted that SNP rs11200638 resides within putative binding sites for the transcription factors adaptor-related protein complex 2α (AP2α) and serum response factor (SRF). This DNA segment, containing the wild-type allele, is part of a CpG island and closely matches the consensus response sequences of these two transcription factors (fig. S5). The presence of the risk allele was predicted to alter the affinity of AP2α and SRF for the HTRA1 promoter.

To verify that the predicted transcription factors bind to the HTRA1 promoter in cultured human cells, we performed chromatin immunoprecipitation (ChIP) followed by quantitative real-time polymerase chain reaction (qPCR) analyses. Lysates were prepared from growing human cervical carcinoma cells (HeLaS3) heterozygous at rs11200638. ChIP was conducted with rabbit polyclonal antibodies against AP2α or SRF. qPCR tests of the ChIP DNA samples confirmed that both AP2α and SRF bind upstream of the HTRA1 gene (Fig. 2 and fig. S5).

Fig. 2.

Evaluation of transcription factor binding to the HTRA1 promoter. (A) AP2α (solid line), (B) SRF (solid line), and (A and B) normal rabbit IgG (dashed line) ChIP DNA were prepared from HeLaS3 cells and analyzed by qPCR to assess the binding of AP2α and SRF to the putative binding sites upstream of HTRA-1. Positive and negative control promoters were also tested (19). The log (ΔRn) (fluorescence intensity over the background) is plotted against the PCR cycle number. The ΔΔCt values (fold increases of transcription calculated relative to reference PCR reactions) are shown in parentheses. Three replicate batches of the ChIP DNA samples were prepared, each of which was tested by qPCR at least twice.

To investigate the influence of SNP rs11200638 on the HTRA1 promoter, human ARPE19 (retinal pigment epithelium) and HeLaS3 cells were transiently transfected with a luciferase reporter plasmid driven by the HTRA1 promoter, harboring either the wild-type (GG) or the risk homozygote (AA) genotype. Preliminary results showed a persistent trend of higher luciferase expressions with the AA compared with the GG genotype (fig. S7). Other factors such as epigenetic or environmental conditions also appeared to play a role for the different genotypes in response to different transcription factors. Further rigorous investigations to address this issue are warranted.

These results, together with the preliminary in vivo observations (22), suggest that the sequence change associated with SNP rs11200638 enhances the transcription of HTRA1 in individuals homozygous for the risk allele.

The HTRA1 gene encodes a heat shock serine protease that is activated by cellular stress (23). It is expressed in the mouse and human retina (24) (fig. S4), and its expression in human fibroblasts increases with advancing age (25). Drusen from the eyes of Caucasian patients with wet AMD stain positive for HTRA1 in immunohistochemistry experiments (22). This suggests that the two processes, neovascularization and drusen formation, may not be completely independent, rather, they may converge.

In summary, we have shown that the functional SNP rs11200638 in the promoter region of HTRA1 is significantly associated with wet AMD in an Asian population. This association has been extended to a Caucasian population (22). To dissect genetic factors important to the complex phenotype of AMD, the power of trans racial gene mapping was exploited by using a population with a more precisely defined phenotype. On the basis of these and earlier genetic findings, a general model for the pathogenesis of AMD can be proposed. Geographic atrophy with large drusen and neovascularization jointly define AMD. Two major genes, CFH and HTRA1, in two different biological pathways, each affect the risk for a distinct component of the AMD phenotype: CFH influences the drusen that characterize dry AMD, whereas HTRA1 influences CNV, the hallmark of the wet disease type. These two processes can be combined, which leads to the composite phenotypes that are seen in some cases of AMD.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Tables S1 to S8

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article