Research Resource: Nuclear Receptor Atlas of Human Retinal Pigment Epithelial Cells: Potential Relevance to Age-Related Macular Degeneration

This study is a comprehensive assessment of nuclear receptor expression in three widely used human retinal pigment epithelial cell model systems.

Abstract

Retinal pigment epithelial (RPE) cells play a vital role in retinal physiology by forming the outer blood–retina barrier and supporting photoreceptor function. Retinopathies including age-related macular degeneration (AMD) involve physiological and pathological changes in the epithelium, severely impairing the retina and effecting vision. Nuclear receptors (NRs), including peroxisome proliferator-activated receptor and liver X receptor, have been identified as key regulators of physiological pathways such as lipid metabolic dysregulation and inflammation, pathways that may also be involved in development of AMD. However, the expression levels of NRs in RPE cells have yet to be systematically surveyed. Furthermore, cell culture lines are widely used to study the biology of RPE cells, without knowledge of the differences or similarities in NR expression and activity between these in vitro models and in vivo RPE. Using quantitative real-time PCR, we assessed the expression patterns of all 48 members of the NR family plus aryl hydrocarbon receptor and aryl hydrocarbon receptor nuclear translocator in human RPE cells. We profiled freshly isolated cells from donor eyes (in vivo), a spontaneously arising human cell line (in vitro), and primary cell culture lines (in vitro) to determine the extent to which NR expression in the cultured cell lines reflects that of in vivo. To evaluate the validity of using cell culture models for investigating NR receptor biology, we determined transcriptional activity and target gene expression of several moderately and highly expressed NRs in vitro. Finally, we identified a subset of NRs that may play an important role in pathobiology of AMD.

The human retinal pigment epithelium (RPE) is a monolayer of pigmented cells within the eye that play a critical role in retinal physiology and the visual cycle. Akin to epithelial cells found in other organs, RPE cells are polarized, characterized by an asymmetrical distribution of plasma membrane proteins and preferential release of secreted proteins into apical and basal compartments (1). Tightly sandwiched between the overlying neurosensory retina on its apical side, facing the light sensitive photoreceptor outer segments, and the underlying choroidal blood supply on its basal side, facing an extracellular matrix called Bruch's membrane, the RPE cell layer forms the outer blood retinal barrier (Fig. 1). The functions of RPE cells are complex and absolutely essential for visual function, as they provide support to the photoreceptors in multiple ways, serving as a conduit allowing delivery of nutrients from the outer blood supply and transporting ions, water, and metabolic end products from the sub-retinal space back to the blood. They play a necessary role in daily phagocytosis and digestion of shed photoreceptor outer segments, during which essential substances including retinaldehyde are recycled and returned to the photoreceptors to rebuild light-sensitive outer segments, required for phototransduction and the visual cycle (2). They also secrete a variety of growth factors and immunosuppressive factors that help maintain the structural integrity of the choriocapillary endothelium and photoreceptors, and help establish the immune privilege of the eye (3). Given the multi-functional characteristics of these cells, it is easy to see that a failure of any of these functions could lead to degeneration of the overlying retina, loss of visual function, and blindness, as seen in age-related macular degeneration (AMD).

Fig. 1.

Cross-section through human macula. The RPE (black arrows), a single epithelial cell layer, sits below the retina, providing daily nutritional support. The retina and RPE together execute the visual cycle. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; IS, photoreceptor inner segments; OS, photoreceptor outer segments.

AMD is a late-onset progressive degeneration of the retina and the major cause of vision loss in developed countries, effecting approximately 30% of people older than 50 yr old (4, 5). Though the etiology of AMD remains largely unknown, countless studies have pointed to genetic, systemic health, and environmental factors giving rise to multi-pathogenic paradigms associated with progression and potentially, initiation of this disease (6–8). At the center of the disease lay metabolic changes and degeneration of the RPE, affected in all three clinical subtypes of AMD (8). “Early AMD” is characterized by the accumulation of lipid- and protein-rich extracellular basal deposits secreted in part by the RPE (9), “geographic atrophy” is characterized by large areas of RPE atrophy, and “late or wet/exudative” AMD is characterized by endothelial invasion and pathological neovascularization through the overlying RPE (10, 11). This neovascularization is partially stimulated by factors secreted by the RPE. Given that RPE cell function is compromised in all forms of AMD, it is clear that improving our understanding of the cell biology of these cells is necessary to not only tease out potential pathogenic and signaling pathways involved in initiation and progression of this disease but also to lead the way to develop effective treatments. Furthermore, increasing our knowledge of the functions and pathways executed by RPE cells will also enhance our understanding of other retinal diseases in which RPE cells are a primary or secondary target including retinal dystrophies such as retinitis pigmentosa (12), fundus flavimaculatus (Stargardts disease) (13), Stargardt-like macular degeneration (14), Best's vitelliform macular dystrophy (15), and cone-rod dystrophies (16).

Nuclear receptors (NRs) are a superfamily of ligand-dependent transcription factors. They serve a critical role as sensors and effectors that translate endocrine and metabolic cues including fat-soluble hormones, vitamins, and dietary lipids into gene expression programs that regulate various functions including metabolism, development, cellular differentiation, proliferation, and apoptosis (17, 18). As such, NR dysregulation has been shown to promote disease states such as cancer, obesity, and inflammation. Additionally, NRs are well-validated therapeutic targets as illustrated by the extensive clinical use of NR agonists and antagonists. Endocrine therapies including tamoxifen, an estrogen receptor (ER) α selective ER modulator, and flutamide, an androgen receptor antagonist, are used to treat breast and prostate cancer, respectively. Treatments for type II diabetes include the use of thiazolidinediones, synthetic agonists of peroxisome proliferator-activated receptor (PPAR) γ (19). Glucocorticoid receptor (GR) agonists such as dexamethasone have been successfully used to reduce the inflammatory response (20). Menopausal symptoms can be alleviated by hormone replacement therapy and contraceptives targeting ER and progesterone receptor (PR) (21).

Similarly, select NRs have been shown to play a role in RPE physiology by several groups to date. Activation of GR has been shown to have mitogenic properties in RPE cells causing cellular proliferation in cases of proliferative vitreoretinopathy after retinal reattachment surgery (22). In vivo murine models of ERβ knockout have been associated with altered murine matrix metalloprotease-2 activity, increased collagen production, and sub-RPE deposit formation as seen in human dry AMD (23, 24) and ERα polymorphisms are associated with wet AMD (25, 26). N-retinyle-din-N-retinylethanolamin (A2E), a major metabolic fluorophore of lipofuscin in RPE cells, is a known ligand for retinoic acid receptor (RAR), through which it has been shown to up-regulate vascular endothelial growth factor expression in RPE cells (27). Because A2E accumulates with both age and pathologically in Stargardts disease in RPE cells, it may phenotypically alter the RPE cells through increased RAR activity (28) and predispose the RPE cell environment to choroidal neovascularization, as seen in wet AMD. To date there are no known NR therapies for AMD. However, in light of the fact that there is significant overlap between pathogenic pathways in AMD and other diseases currently under treatment in the clinic with validated NR drugs, it is imperative to search for the potential relevance of NR drugs in AMD therapy.

Despite the aforementioned reports, no one has investigated the eye as a potential secondary endocrine organ with endocrine-like functions. Furthermore, there are no complete expression studies of all human NRs in the RPE in which standardized methodologies have been used to collect the data. The goal of developing this RPE cell specific NR atlas was to create a comprehensive baseline of NR expression to elucidate the biochemical and physiological pathways under normal conditions as well as identify potential pathways that may contribute to pathogenic changes seen in retinal diseases. Herein we report the systematic profiling of NRs in three model systems of human RPE cells commonly used in research laboratories across the world and highlight differences in relative expression levels of these receptors. We validate activity and target gene expression of select NRs in vitro and, finally, identify NRs that may be relevant to the biology of AMD.

Results and Discussion

Human RPE samples

We used three human RPE cell models: a spontaneously arising human RPE cell line (ARPE19), primary cell lines derived from adult donors, and RPE cells isolated from freshly obtained adult donor tissue (Fig. 2). We chose to compare these cell models because they are most commonly used in research labs, each with inherent advantages and disadvantages. For example, ARPE19 cells are easy to culture, form a hexagonal cobblestone layer, and exhibit morphological polarization under certain culture conditions yet lack pigmentation and lose some RPE cell-specific markers (29). Early passage human primary RPE cells are cultured from human donor eyes, which are often limited in availability, generally used only until passage 10, and therefore have a finite lifespan. However, they retain normal physiological functions, including polarization, phagocytosis, and the ability to transport retinoids (30). RPE cells freshly isolated from human donor eyes for biochemistry and molecular biology are ideal, as they have not been subjected to culturing conditions and potential de-differentiation. However, as mentioned earlier, accessibility to donor tissue obtained within a short postmortem time is limited and due to the small amount of obtainable tissue per eye there are constraints on number of studies that can be conducted. Furthermore, there is variability in the biology of the samples obtained reflective of normal heterogeneity within the donor population. Our goal in using these three RPE samples was first to determine their NR expression profiles and second to determine the validity of using the more experimentally accessible in vitro cell culture models to pursue focused studies on NR related signaling pathways.

Fig. 2.

Characterization of RPE samples. A, ARPE19 cells grown to confluence develop tight junctions-green zona occludens immunoreactivity is visible around the perimeter of the cells. Nuclei are stained blue with Hoechst. B, Three biological samples of each RPE cell model type were used. The models included ARPE19, primary RPE cells up to passage 9, and RPE isolated fresh from donor eyes in less than 5 h postmortem. Age, gender, and CFH genotype are noted. C, To determine CFH genotype, PCR products were digested (+) or undigested (−) by restriction enzyme NlaIII. In NlaIII digested samples, bands size of 110, 57, and 52 bp are present in heterozygotes, 58 and 52 bp in homozygotes, and 110 in wild type. P, number of cell passages; TTP, time to preservation; WT, wild type; Het, heterozygote; Homo, homozygote.

In our in vitro cell culture systems, cells were grown for more than 3 wk to postconfluence developing tight junctions as demonstrated by zona occludens immunopositive borders (Fig. 2A, ARPE19 cells), before RNA isolation. In adherence to the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines, RNA quality was evaluated using Agilent bioanalyzer and the RNA integrity of all cell culture samples and freshly isolated RPE cells were 10 and greater than 7.5, respectively (Materials and Methods; Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org/) (31). Details of each of three biological samples for each RPE model used in this study are presented in Fig. 2B and include age and gender of donor from which cells were derived, passage number of the in vitro cell model systems, and complement factor H (CFH) genotype (Fig. 2C). CFH genotype was determined because polymorphisms in the CFH gene (CFH Y402H) are associated with increased risk for progression to AMD (32–35) in 50% of carriers. Among our nine samples only one, a primary cell line isolated from a normal human donor, was genotyped with the ‘most at risk' CFH (402HH) haplotype (Fig. 2B).

The atlas of RPE NRs

We profiled the expression of 48 members of the NR superfamily of proteins, two isoform variants of the PPARβ/δ and PPARγ, as well as aryl hydrocarbon receptor (AhR) and its obligate binding partner aryl hydrocarbon receptor nuclear translocator (ARNT) owing to their similar mechanisms of action to NRs. The relative standard curve method of real-time quantitative RT-PCR (qPCR) was used to survey NR expression levels in these RPE model systems relative to a universal human cDNA standard representing a broad range of human tissues. NR expression levels were grouped based on their expression level relative to the housekeeping gene β-actin. Receptors with mean threshold cycle (Ct) values of 34 or greater were called undetectable or absent, the rest were assigned to groups of low (ratio NR/β-actin <0.1), medium (0.1 ≤ NR/β-actin ≤ 1), and high (NR/β-actin >1) expression. Overall, β-actin Ct value was consistent between RT-PCR plates and within biological replicates of the three RPE model systems with coefficient of variations of 0.0025, 0.0039, and 0.0135 for ARPE19, primary cells, and freshly isolated RPE cells, respectively. Distribution of NR expression in these three RPE samples is depicted in pie charts (Fig. 3) and accompanying tables (Fig. 4).

Fig. 3.

Distribution of NR in RPE samples. The levels of NR expression in the three RPE cell models are indicated by the pie charts, and their names are shown in the tables to the right. Normalized NR mRNA-expression levels were defined as Absent if the Ct value was ≥34, Low if the level was below 0.1 arbitrary units, Moderate if the level was between 0.1 and 1.0, and High if the level was >1.0 arbitrary units.

Fig. 4.

qPCR data of all NRs presented as means ± sem of the three RPE model systems.

Analysis of these data revealed that 32 NRs are expressed in ARPE19 cells, of which six were expressed at high levels, 19 at moderate levels, and seven at low levels (Fig. 3). These included adopted orphan NRs that bind to low-affinity dietary lipids including polyunsaturated fatty acids and oxysterols [i.e., all isoforms of PPAR, except PPARγ2, which is expressed mainly in adipose tissue (36), retinoid X receptors (RXRs) α and β, and liver X receptor (LXR) β], traditional endocrine receptors [thyroid hormone receptor (TR) α, GR, mineralcorticoid receptor, and RARγ], orphan receptors with repressive activity (chicken ovalbumin upstream promoter- transcription factor I, II, III, testicular receptor 2 and 4, REV-ErBα, β), orphan receptors exhibiting activating and repressing activity [estrogen-related receptor (ERR) α and retinoic acid-related orphan receptor (ROR) α], as well as toxin activated AhR and ARNT. Primary RPE cells exhibited a NR expression pattern very similar to that seen in ARPE19 cells (Fig. 3). Specifically, a total of 36 NRs were detectable in our samples with three NRs expressed at high (AhR, RARα and β), 18 at moderate levels, and 15 at low levels. A noteworthy difference between the cell culture models is the moderate expression of Drosophila tailess homologue (TLX) in primary cells, which was absent or undetectable in ARPE19 cells (Fig. 3 and Table 1).

Table 1.

Differentially expressed genes in three models of human RPE

Gene	ARPE19	Primary	Fresh RPE
AR	+	+	−
PPARγ1	+	+	−
VDR	+	+	−
ERα	−	−	+
ERRβ	−	−	+
ERRγ	−	−	+
HNF4α	−	−	+
NOR1	−	−	+
PR	−	−	+
RORβ	−	−	+
RXRγ	−	−	+
TLX	−	+	+
TRβ	−	+	+
LRH-1	−	+	−

The largest number of NRs expressed at high levels is seen in RPE cells, isolated from human donor eyes, a total of 38 with 30 detectable at high levels, eight moderately expressed, and no NRs present at low levels (Fig. 3). As expected, all forms of the RARs (α, β, γ), RXRs (α, β, γ), and RORs (α, β, γ) are expressed at high levels in freshly isolated RPE, highlighting the specialized and critical role of retinoids in the visual cycle. Other representatives of highly expressed receptors include PPARα and β/δ, LXRβ, ERRα and γ, AhR, ARNT, and TRα and β. Several receptors are expressed at a moderate level including LXRα, endocrine steroid receptor PR, and hepatocyte nuclear factor 4γ. Thirteen NRs were absent or undetectable in our freshly derived human RPE samples including ERβ, pregnane X receptor, steroidogenic factor 1, and short heterodimer partner. Overall, several of the NR receptors expressed at a high level in the donor RPE were also expressed at a high or moderate level in primary cultures of human RPE including PPARβ/δ, testicular receptor 2 and 4, ARNT, RXRα, β, and RARγ. NR expressed at high levels in all RPE samples included AhR and RARα and β. Details of the relative expression levels of the NR between the three cell model systems are shown in Fig. 4.

A select number of genes were differentially expressed among the three RPE cell models (Table 1). Three genes expressed in the two cell culture lines but absent from fresh RPE were the androgen receptor, PPARγ1, and vitamin D receptor. Eight were absent from both cell culture lines but present in fresh RPE: ERα, ERRβ and γ, hepatocyte nuclear factor 4α, neuron-derived orphan A receptor 1, PR and RORβ and RXRγ. TLX and TRβ were identified in both the primary cell lines and fresh RPE but absent from ARPE19. Only one gene, liver receptor homolog 1, was present in the primary cell line yet absent from ARPE19 cells and freshly isolated human RPE cells. These results reflect an overlap in NR expression between freshly isolated RPE cells and cell culture lines with the differences detected potentially attributed to de-differentiation of RPE cells in culture.

NRs relevant to AMD pathogenesis

The primary purpose of developing the NR atlas in RPE cells was to identify the expression of NRs in physiologically ‘normal' cells. An important inference from this data set would be identification of NRs that may play a role in the physiology and ultimately pathology of retinal diseases such as AMD. Current hypotheses surrounding the pathobiology of AMD reflect its multifactorial nature (7, 37) and include the following: dysregulated lipid and cholesterol efflux and accumulation (38), fibrosis and altered extracellular matrix production and degradation (39), mitochondrial dysfunction (40, 41), choroidal nonperfusion or hyperperfusion and growth factor–stimulated angiogenesis (42–44), excessive lipofuscin accumulation/faulty A2E regulation (27, 45, 46), and oxidative injury and apoptosis (47–55). With this in mind, based on a literature search, we assembled NRs expressed in donor RPE tissue into functional groups with relevance to AMD pathology (Fig. 5). Available gene ontology (GO) terms are also presented. We selected the following groups: 1) NRs with retinoic acid binding activity (GO:0003708); 2) NRs involved in lipid metabolism (GO:0006629) (56–67); 3) NRs implicated in the regulation of immune responses, defined as processes that modulate the frequency, rate, or extent of an immune system process (GO:0002682); 4) NRs involved in inflammatory responses, defined as processes that are part of the immediate defensive reaction to injury or infection including vasodilation, extravasation, of plasma into intercellular spaces and accumulation of white blood cells and macrophages incited by chemical or physical agents (LXR, NURR1, GR, and AhR) (68–71); 5) NRs involved in remodeling of extracellular matrix (PPAR, AhR, RXR, and pregnane X receptor) (72–76); 6) NRs involved in apoptosis (testicular receptors 2 and 4, LXR, RAR, RXR) (77–80); 7) angiogenesis (TLX, chicken ovalbumin upstream promoter-transcription factor II) (81, 82); and 8) lipofuscin/A2E regulation (RAR, RXR) (27). The results of this analysis are shown in Fig. 5. As discussed above many of the NRs are highly expressed in freshly isolated human RPE (Fig. 5, A–C). From examining panel D in Fig. 5, it becomes further evident that many NRs involved in retinoic acid binding are all highly expressed in human RPE. Additionally, NRs within the other ‘AMD relevant' functional groups tend to be highly expressed, with a clearly skewed distribution (regulation of immune system processes, lipid metabolism, inflammation, apoptosis, lipofuscin/A2E). Table 2 lists different aspects of AMD pathobiology associated with available GO terms and local putative/pathogenic ligands for the NRs in each group.

Fig. 5.

Distribution of functional groups of NRs by expression levels in donor RPE tissue. A, Average expression of freshly isolated human RPE samples (n = 3, log10 scale). The order of NRs matches that in clustering diagram on panel B. Color scheme is the same as in Fig. 3, with red, blue, and purple representing absent, medium, and high expressing NRs, respectively. Many of the NRs are highly expressed in the freshly isolated human RPE (purple diamonds) B, Hierarchical clustering of NR expression levels in RPE samples. C, Distribution of clustered NRs among by expression bins. D, Distribution of NRs within select functional groups assembled based on literature search and GO terms along the expression axis.

Table 2.

Physiologically relevant NR in pathogenesis of AMD

Functional group	Nuclear receptors	AMD pathobiology	Putative pathogenic ligands
Retinoic acid binding activity, lipofuscin/A2E regulation	RXR, RAR	Lipofuscin accumulation, A2E regulation	Retinoids
Lipid metabolic processes	PPAR, LXR, GR, HNF4, RXR, TR, CTF	Cholesterol efflux, lipoprotein assembly	Dietary fatty acids, oxidized low-density lipoprotein, oxysterols, cholesterol
Regulation of immune system processes	AhR, ARNT, LXR, ROR, RAR	Complement activation, complement accumulation	Toxins, oxidized lipids, retinoids
Inflammation	AhR, ARNT, GR, CTF, LXR, NURR1, RXR	Inflammation, mitochondrial dysfunction	Toxins, oxidized lipids, steroids, retinoids
Extracellular matrix	AhR, PPAR, RXR	Extracellular matrix dysregulation, Bruch's membrane changes	Toxins, dietary lipids, retinoids
Apoptosis	AhR, LXR, RAR, RXR, TR3, TR4	Apoptosis	Toxins, dietary lipids, retinoids, thyroid hormone
Angiogenesis	TLX, CTF	Neovascularization, VEGF up-regulation	Unknown

Stimulated activation of select NRs and target gene expression in ARPE19 cells

We sought to demonstrate the functional activity of NRs with moderate to high expression levels and potential roles in RPE dysfunction based on known AMD pathology and GO categorization as elaborated above. Accordingly, we assessed the transcriptional activity of a select number of these intracellular receptors including AhR, GR, PPARα and β/δ, in response to known agonists using luciferase-based reporter and target gene qPCR assays (Fig. 6). We confirmed protein expression of these receptors in ARPE19 cells (Fig. 6, A, D, G, and J). Stimulation of ARPE19 cells with 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), a known environmental pollutant and component of cigarette smoke, increased the transcriptional activity (AhR-tk-Luciferase) of AhR and the mRNA expression of the canonical AhR target genes and members of the cytochrome p450 family, Cyp1A1, Cyp1B1, and CYP2S1 (Fig. 6, B and C). GR exhibits enhanced transcriptional activity using the MMTV-Luciferase reporter in ARPE19 cells treated with the synthetic GR ligand dexamethasone (Fig. 6E). Furthermore, thrombomodulin (THBD), glucocorticoid-induced leucine zipper (Gilz), and the sodium channel-non–voltage-gated 1 alpha (ENaCα) mRNA levels, endogenous targets for GR, were up-regulated with dexamethasone treatment (Fig. 6F). Similarly, PPARα and β/δ receptor signaling was confirmed by pharmacological treatment of ARPE19 cells using GW7647 and GW0747 respectively, eliciting increased PPAR transcriptional activity (DR1-Luc) and elevated expression of the PPAR α target genes carnitine palmitoyltransferase 1A (CPT1a), adipose differentiation related protein (ADFP), and fatty acid synthase (FASN) and PPARβ/δ target genes acyl-coA synthetase (ASC) and FASN (Fig. 6, H, I, K, and L). These results support the use of ARPE19 cells as a valid model system for further investigation of the physiology of NRs that are also expressed at moderate or high levels in vivo.

Fig. 6.

Stimulated activation of NR receptors and expression of target genes. A, D, G, J, Western blots of ARPE19 cell extracts showing protein expression of the NRs, AhR, GR, PPARα and β/γ. B, E, H, K, Increased activity of these NRs was seen after transfection of ARPE19 cells with a luciferase reporter and treating the cells with known ligands for the receptors. C, F, I, L, Using qPCR, a significant increase in expression (greater than 2-fold relative mRNA level) of receptor specific target genes CYP1A1, CYP1B1, and CYP2S1 for AhR after treatment with TCCD, thrombomodulin, GILZ, and ENaCa for GR after treatment with dexamethasone, FASN, ADFP, and CPT1a for PPARα after treatment with GW7647 and ACS, and FASN for PPARβ/γ after treatment with GW742 was seen. Error bars, sd.

Conclusions

Describing the pattern of gene transcription is a crucial first step in the path to investigating their function. Here, we report on a systematic quantitative expression atlas of all known NR genes in three human RPE models. RPE cells are of great interest in retinal diseases including AMD. We found by far most NRs were expressed in all three RPE cell models, yet the expression levels were different, with many expressed at high levels in freshly isolated RPE cells compared with the in vitro models. Furthermore, a select few were absent in the cell culture systems or from freshly isolated human RPE cells. Our results support the validity of using cell culture models to initiate investigation of the biology of the NRs after careful consideration of differences in mRNA expression levels between the cell culture line to be used and RPE in vivo. Furthermore, confirmation of protein expression levels of the NR of interest would provide additional support for future studies using these in vitro RPE cell models. Overall, however, the widespread expression of NRs in RPE cells provide support for the premise that the eye shares fundamental signaling pathways with other endocrine organs and exhibits endocrine-like functions.

A limitation of our study is the small sample size that prohibits sub-stratifying the donors of the primary cells and the fresh RPE cells based on factors such as dietary habits and genetics, which may effect the overall NR expression profile. For example, the expression of NR involved in lipid metabolism may be different in individuals who consume high-fat diets enriched n-6 poly-unsaturated fatty acid compared with a low-fat diet enriched n-3 fatty acid. Additionally, several genes have been associated with risk of AMD progression including CFH. Given our sample size in this study, it would be difficult to determine whether the patients' genetic status could effect NR expression. These factors must be considered in future studies of larger cohorts as they may shed light on differences in NR expression of groups at an increased vs. decreased risk for developing AMD. It is important to note, however, that even within our cohorts we did not observe a difference in NR expression profiles between samples with different CFH haplotypes.

Our study also gave us an opportunity to identify NRs that may play a role in AMD pathogenesis. An interesting finding was the absence of PPARγ in our freshly isolated human RPE samples but presence in the cell culture lines. Though reports directly linking PPARγ dysfunction with AMD pathology are limited, several recent investigations, using ARPE19 and/or primary cells for in vitro and mouse models for in vivo studies, have speculated on a role of PPARγ in AMD (83–87). This potential association has surfaced due to PPARγ's known participation in various mechanisms and biologic pathways related to lipid regulation, immune modulation, and oxidant/antioxidant pathways, pathways also associated with the pathogenesis of AMD. These studies have primarily focused on aspects of ocular angiogenesis, specifically CNV and fibrosis (87–89), as well as immune regulation (90) and oxidative injury (91). In contrast to our findings in aged human RPE, PPARγ message has been reported in human fetal RPE cells (91) and protein detected via immunohistochemistry in human retinal samples [age not reported (83)]. This disparity may be due in part to differences in the ages of the donor samples used in each of the studies, heterogeneity within sample populations, and/or unknown confounders. Nevertheless, these findings further highlight the necessity to thoroughly evaluate and confirm the expression of the NR target of interest in human samples before embarking on extensive in vitro studies.

We hope these data will serve as a framework for identifying mechanistic pathways associated with retinal diseases such as AMD. Though treatment options for the wet form of this disease are currently available and to some extent successful, there hasn't been any breakthrough in identification of drugs that directly target sub-RPE deposit accumulation, known as drusen, found in 85–90% of total AMD patients living with this burden. Therefore, it is critical to further our understanding of the molecular and biological mechanisms that contribute to drusen formation. Given the degree of cross-talk between some NRs which may have dual roles, and the complexity of AMD, these data should provide a unique resource to further explore the role of receptors relevant to the initiation and progression of this disease and perhaps take advantage of synthetic agonist/antagonists of NRs currently used to treat asthma, diabetes, atherosclerosis, and cancer, if applicable. Ongoing studies include targeted investigation of the biology of select NRs that may play a role in AMD as well as developing a NR atlas for RPE cells isolated from human donor eyes with clinically confirmed AMD and age-matched controls to further tease out differences in NR expression with disease.

Materials and Methods

Human RPE cell model systems

Three human RPE model systems were used in this study, including a spontaneously arising human cell line (ARPE19 cells), primary cell lines derived from adult donors (passages 6–9 were used), and freshly isolated RPE cells from donor eyes (Fig. 2). ARPE19 cells were cultured in DMEM/F12 medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Sigma, St. Louis, MO). For primary RPE cell cultures, human donor eyes were collected (North Carolina Organ Donor and Eye Bank, Inc., Winston-Salem, NC) less than 6 h after death and were cultured within 24 h of death in accordance with the provisions of the Declaration of Helsinki for research involving human tissue. The human donor eyes we used in this study were nondiabetic without ophthalmic history and were grossly normal. RPE cells for culture studies were harvested from eyes as previously described (92). Cells were grown in Eagle's minimal essential medium (MEM; Invitrogen) with 10% fetal bovine serum (Hyclone Laboratories, Logan, UT) and with 100 U/ml penicillin and 100 g/ml streptomycin sulfate (Invitrogen). All cell cultures were maintained at 37 C in a humidified environment containing 5% CO₂. In our cell culture systems, cells were grown to post confluence for more than 3 weeks, developing tight junctions as demonstrated with zona occludens-1 immunocytochemsitry before RNA isolation, in an attempt to mimic the postmitotic state of in vivo RPE cells. Three biological replicates were used for all experiments.

For fresh isolation of RPE cells, human eyes from three donors (45–66 yr of age) were obtained from the North Carolina Eye Bank stored in preservative (RNAlater; Ambion, Austin, TX), with an average death-to-procurement time of less than 5 h. Eyes were stored at 4 C in RNAlater for no more than 48 h, at which point the RPE was isolated. For RPE isolation, the anterior segment, iris, lens, and vitreous of each eye were carefully extracted. Using forceps the retina was gently teased away from the underlying RPE. After removal of the retina, RPE cells free of choroidal contamination were collected as follows: RPE cells were carefully brushed/rubbed away from Bruch's membrane using a smooth glass pipet within in the eye cup containing 200–300 μl of RNAlater avoiding the ora serata to avoid contamination with the underlying choroidal blood supply. Cells from each eye were recovered in a 2-ml microcentrifuge tube with a wide-bore transfer pipette and then spun down at 5000 rpm for 5 min at 4 C.

RNA isolation and cDNA preparation

Total RNA was isolated from all RPE as well as three retinal samples using RNeasy kit (Qiagen, Valencia, CA). All samples were treated with DNase I using DNA-free kit (Ambion) and RNA quality was determined. In accordance with MIQE guidelines (31) RNA quality was assessed using nanodrop and agilent bioanalyzer (Supplemental Fig. 1). Only samples with RNA integrity number of greater than 7.5 were used in this study. cDNA was reverse transcribed from 1 μg total RNA using iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). The resultant cDNA was diluted 1:25 for the qPCR. Universal human tissue cDNA prepared from a mixture of total RNAs collected from adult normal human tissues, representing a broad range of expressed genes was obtained from Clontech (Mountain View, CA). Because the NR PNR is found exclusively in retinal photoreceptors, we used human retinal cDNA that includes the photoreceptor cell layer as our reference for gene expression and standard curve formation. PNR was not detectable in our human RPE samples confirming purity of RPE layer with little or no contamination with retinal photoreceptors.

CFH genotyping assessment

RPE cDNA was amplified by using ReadyMix Taq PCR reagent (Sigma). cDNA (500 ng) was used as template DNA in 50-μl PCR reaction. PCR condition was 94 C for 30 sec, 55 C for 30 sec, and 72 C for 20 sec, 30 cycles, and holding at 4 C. The PCR product is 127 bp. Twenty microliters of PCR amplification products were digested by restriction enzyme NlaIII at 37 C for 4 h followed by heat-inactivation of the enzyme at 67 C for 20 min. PAGE was used to determine CFH genotype. Precast PAGE gels (Bio-Rad) were used. After electrophoresis, the gels were stained with SYBR Gold nucleic acid gel stain (Invitrogen) for 40 min and then visualized under UV light.

Primer validation and qPCR assay

NR qPCR primer sequences obtained from the NURSA website and from Harvard primer bank are listed in Supplemental Table 1. Primer oligonucleotides were purchased from Integrated DNA Technologies at 25 nm scale with standard desalting. The primer/probe sets for the NRs were validated, according to previous recommendations, in ARPE19 and primary cells based on a single peak in the dissociation curve, slope, and R² value (>0.95) of the standard curve plot of Ct value vs. cDNA quantity (93). Briefly, NR specific qPCR primers were validated for optimal PCR efficiency (90–110%) using a template titration assay of universal human tissue cDNA (Clontech) representing a broad range of expressed genes or human retinal cDNA. Two-fold serial dilutions of cDNA spanning 12 ng through 0.002 ng of the reference cDNA were used as template for standard curves to measure PCR efficiency of each NR qPCR primer set (final primer concentration 150 nm) using iQ SYBRGreen Supermix obtained from Bio-Rad and CFX iCycler (Bio-Rad Icycler CFX96 Thermo1000).

With these validated NR primers, qPCR was used to determine the NR expression levels in the three RPE model systems. Results were analyzed using the relative absolute standard curve method in which all the sample Cts are calculated from the reference cDNA standard curve and then normalized to the housekeeping gene, β-actin. qPCR assays were run in 96-well plates with 26 μl final volume per well. Batches of 4–8 plates were pipetted at a time and run on the iCycler on the same day. The qPCR plate set-up included triplicate samples of a standard curve of the human reference cDNA and triplicates of three biological replicates of the ARPE19, primary and donor RPE. All samples for each NR were run on the same plate, and the housekeeping gene, β-actin, was run on each set of four plates. Several housekeeping genes were tested including 18S, 36B4, cyclophilin, GAPDH with β-actin yielding the highest and most consistent expression levels in our cell types (coefficient of variations across plates in ARPE19 cells = 0.0025, Primary cells = 0.0039, and fresh isolated RPE cells = 0.0135). Additionally control samples included no reverse transcriptase control to confirm the absence of genomic amplification and no template control to confirm absence of primer dimer formation.

Electrophoresis and immunoblotting

Whole cell extracts of ARPE19 cells were prepared using RIPA buffer [50 mm Tris-HCl (pH 7.5), 150 mm NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.05% SDS, 1 mm EDTA, protease inhibitors (Sigma)]. Total protein concentrations were determined by Bradford Assay (Bio-Rad). Proteins (25 μg per well) were separated by 10% SDS-PAGE, transferred onto nitrocellulose membranes, and detected using the following antibodies: AhR (Abcam, Cambridge, MA), GR (Santa Cruz, Biotechnology Inc, Santa Cruz, CA), PPARα (Santa Cruz); PPARβ/δ, (Santa Cruz). Appropriate secondary antibodies were obtained from Jackson Immunoresearch (West Grove, PA).

Transcriptional activation assays and target gene expression validation

We assessed the transcriptional activity of select endogenous NRs in response to known agonists using luciferase-based reporter and target gene qPCR assays. ARPE-19 cells, passage 25–35, were plated in 24-well plates at a density of 50,000 cells per well. Lipofectin (Invitrogen, Carlsbad, CA)-mediated transfection was performed the following day using a total of 3 μg DNA per triplicate including 2 μg reporter plasmid. After 24 h, the transfected ARPE19 cells were treated with a known pharmacological and/or natural ligand for these receptors (AhR-TCDD at 0.1, 1, and 10 nm; GR, dexamethasone at 1, 100, 1000 nm; PPARα, GW7647 at 0.2, 1, and 5 μm; PPARβ/δ, GW0742 at 0.2, 1, and 5 μm) at the indicated doses for 24 h. Luciferase (reporter) and β-galactosidase (transfection normalization) activities were measured using a Perkin-Elmer Fusion Instrument. Concomitantly ARPE19 cells were treated with these same compounds, and RNA was isolated and cDNA produced. Expressions of NR target genes was determined as discussed earlier. List of reporters, ligands, and primer sequences of NR target genes are presented in Supplemental Table 2.

Binning of the expression data and hierarchical clustering

Receptors for which the Ct values were greater than 34 in two or more biological replicates per cell type were considered absent. Among the nonabsent receptors the binning was done based on the ratio of the standard quantity of the receptor mRNA to that of housekeeping control (β-actin), as determined by the standard curve method. Receptors with the ratio of receptor/housekeeping <0.1 were considered to be expressed at a “Low” level, those with this ratio between 0.1 and 1 were considered to be present at “Medium” levels, and those with the ratio above 1 were considered as “Highly” expressed. Hierarchical clustering was performed by Ward algorithm.