Use of Iris Pigment Epithelium to Replace Retinal Pigment Epithelium in Age-Related Macular Degeneration: A Gene Expression Analysis

Objective  To determine the gene expression profiles of primary retinal pigment epithelium (RPE) and iris pigment epithelium (IPE) using microarrays.

Methods  Primary RPE and IPE from 6 human donor eyes were collected, and total RNA was isolated. Differences in gene expression were determined using a human genechip (human U95Av2 [12 600 probes]; Affymetrix Inc, Santa Clara, Calif).

Results  Hierarchical cluster analysis differentiated the gene expression profiles of RPE and IPE clusters into 2 distinct groups. A mean ± SD of 5308 ± 416 gene probes were expressed in RPE vs 6130 ± 205 in IPE. Sixty-eight genes were expressed only in RPE; 154 genes were expressed only in IPE. Twenty-two additional genes had greater than 3-fold increased expression in RPE vs IPE, and 147 genes had greater than 3-fold decreased expression in RPE vs IPE.

Conclusion  There are major differences in the gene expression profiles of primary RPE vs IPE.

Clinical Relevance  The different gene expression profiles of primary RPE vs IPE harvested from the same donor eyes infer that it may be difficult for IPE to replace all aspects of damaged RPE function in transplantation studies.

Replacement of diseased or damaged retinal pigment epithelium (RPE) has been an ongoing area of active investigation for the last 3 decades. Efforts in this area are fueled by the fact that RPE damage or dysfunction can lead to severe visual loss in many ocular disorders; therefore, RPE replacement may prove beneficial to many patients with untreatable diseases. For example, RPE replacement may be a potential treatment for Leber congenital amaurosis, which develops because of improper recycling of visual pigments by RPE from a defect in the RPE65 gene.¹ A defect in the MERTK gene in the Royal College of Surgeons rat model, which leads to improper phagocytosis of shed photoreceptor outer segments, was treated successfully by RPE transplantation, and homologous disease exists in humans.^2-23 Retinal pigment epithelium transplantation may also have a role in the management of age-related macular degeneration (AMD), which is the leading cause of blindness in patients older than 60 years in the Western world.²⁴ There are 2 ways in which RPE replacement may be a potential treatment for advanced AMD. In geographic atrophy, loss of RPE precedes choriocapillaris and photoreceptor loss, and RPE transplantation may prevent or reverse these changes.^25-28 In exudative AMD, the rationale for RPE transplantation is deceptively simple, as the native RPE is excised unavoidably with the choroidal neovascular complex during submacular surgery, and RPE removal leads to further choriocapillaris atrophy and photoreceptor loss.²⁹ Retinal pigment epithelium transplantation may prevent subsequent choriocapillaris atrophy and improve the visual prognosis.^30-34 Initial clinical trials of RPE transplantation in patients with exudative or nonexudative AMD have resulted in nonsignificant improvement in vision; however, graft survival is hampered by allograft rejection or by damage of Bruch's membrane at the time of submacular surgery.^25-27,35-45

Several major obstacles need to be overcome before successful RPE replacement in AMD, including the prevention of graft rejection by immune modulation or by tissue matching. Retinal pigment epithelium is immunogenic, and the immune privilege of the subretinal space is relative rather than absolute.^46-56 The outer blood-retinal barrier, which is ordinarily formed by tight junctions between adjacent RPE, is compromised in AMD. The inner blood-retinal barrier may be compromised in some eyes as well, because neovascularization can originate in the retina in eyes with retinal angiomatous proliferation.^57,58 Compromise of the blood-retinal barrier increases the probability that allogeneic RPE will undergo immune rejection in the subretinal space. Several techniques have been used to prevent immune rejection of transplanted cells, including translocation of autologous RPE suspensions or patches, transscleral biopsy techniques to harvest autologous peripheral autologous RPE before transplantation, and transplantation of iris pigment epithelium (IPE).^59-71

In principle, autologous IPE is an attractive candidate to replace diseased or damaged RPE because IPE and RPE share a common embryological origin. Iris pigment epithelium can be obtained by performing a peripheral iridectomy before subretinal transplantation, and the donor cell population can be expanded in vitro. In principle, iris biopsy can provide a ready source of donor cells and eliminate concerns about infectious disease transmission and immune rejection of the transplanted cells. Iris pigment epithelium transplantation has been performed in animal models of human disease and in patients with AMD, without dramatic improvement in vision.^35,59-74 Some aspects of IPE and RPE function have been compared in vitro and in vivo, but, to our knowledge, a detailed analysis of the gene expression profiles of IPE and RPE has not been performed. Data reported herein should help answer the question of whether IPE can replace human RPE effectively.

The development of microarray chips has made it technically feasible to compare the gene expression profiles of IPE and RPE harvested from the same human donor eyes. The objectives of this study were to compare the gene expression profiles of primary RPE vs IPE harvested from the same donor eyes and to determine the potential usefulness of replacing RPE with IPE in various disease states.

Primary RPE and IPE from the same human donors (aged 63, 71, 75, 76, 85, and 86 years) were obtained from the National Disease Research Interchange (Philadelphia, Pa), with cells harvested within 29 hours of death (Table 1). Because the study involved postmortem tissue without identification of individual patients, it was exempt from institutional review board approval. Primary IPE and RPE cells were prepared from the anterior and posterior poles of human cadaver eyes as previously described.^75,76 On receipt in the laboratory, eyes were cleaned of extraocular tissue. Iris pigment epithelium was separated from the stroma of the iris in the anterior segment of donor eye globes using forceps. Other anterior segment structures, vitreous, and retina were removed, leaving an eyecup with native RPE on the inner surface. For these studies, 500 000 primary RPE cells and 300 000 IPE cells were collected with trypsin from each pair of globes harvested from 6 human donors. The RPE and IPE cells were washed 3 times with cold Dulbecco phosphate-buffered saline and stored at –80°C before isolation of RNA in the DNA microarray study. Six pairs of eyes from human donors were used; given the expense and limited availability of human tissue, small sample sizes have been used in the past to generate important data on gene expression within human tissue.^77-79

Cells from RPE and IPE were stained using a pancytokeratin antibody to verify that all cells were of epithelial origin.^39,45 For this purpose, RPE or IPE cells on a microscopic glass slide were fixed with 4% paraformaldehyde for 30 minutes and washed with a phosphate-buffered saline solution. The cells were treated for 1 hour at room temperature with 3% bovine serum albumin (Sigma Chemical Co, St Louis, Mo) in phosphate-buffered saline solution to block nonspecific binding sites. The cells were then incubated at 37°C for 1 hour with a fluorescein isothiocyanate–conjugated monoclonal antipancytokeratin antibody to cytokeratins 5, 6, and 8 (Sigma Chemical Co). The cells were washed 3 times with a phosphate-buffered saline solution and were examined under a fluorescence microscope. An irrelevant isotypic IgG primary antibody (antihuman von Willebrand antibody; Sigma Chemical Co) coupled with a fluorescein isothiocyanate–conjugated secondary antibody was also used and showed no background staining. All harvested cells were positive for pancytokeratin, indicating that the cells were of epithelial origin.

Cells from RPE or IPE (approximately 3 × 10⁵ to 5 × 10⁵) were disrupted, and total RNA was isolated using a QIA shredder and an RNeasy Mini Kit (QIAGEN Inc, Valencia, Calif). Briefly, 600 μL of lysing buffer (RLT) was added to cells in a 1.5-mL microfuge tube, and cell lysate was loaded onto a QIA shredder column and centrifuged for 2 minutes at 13 000 rpm. The homogenized lysate was then mixed with 600 μL of 70% ethanol and was applied to an RNeasy mini spin column and was centrifuged for 15 seconds at 13 000 rpm. Next, 700 μL of buffer RW1 and buffer RPE was added and centrifuged sequentially for washing twice. Then, 60 μL of ribonuclease-free water was used to elute total RNA from the RNeasy mini spin column. All total RNA used in the experiments was pure as determined by the ratio of absorbance (A) at 260 vs 280 nm (A260/A280 ratio > 1.9). Total RNA was stored at −80°C for later use.

A T7-(dT)₂₄ oligomer, superscript reverse transcriptase II and DNA polymerase I (GIBCO BRL, Gaithersburg, Md) were used for first-strand and second-strand complementary DNA (cDNA) synthesis using total RNA as templates. Double-stranded cDNA was cleaned by Phase Lock Gels (Eppendorf AG, Hamburg, Germany) phenol-chloroform extraction and ethanol precipitation. Biotin-labeled antisense cRNA was produced by an in vitro transcription reaction (ENZO BioArray High-Yield RNA Transcript Labeling Kit; Affymetrix Inc, Santa Clara, Calif) and incubated with fragmentation buffer (Tris-acetate, potassium acetate, and magnesium acetate [Sigma Chemical Co] at 94°C for 35 minutes). Target hybridization, washing, staining, and scanning probe arrays were performed according to an Affymetrix GeneChip Expression Analysis Manual.⁸⁰

For quality control, the Affymetrix human U95Av2 microarray chip included 20 housekeeping gene probes to measure the consistency of the hybridization signals from the 3′, middle, and 5′ fragments of messenger RNA (mRNA) coding regions.⁸¹ Gene expression analyses, including global normalization and scaling, were performed using Affymetrix GCOS 1.2 and Array Assist 3.01 (Stratagene, La Jolla, Calif) software. For the purpose of this study, gene expression was considered present if the gene was detected using the Affymetrix GCOS 1.2 statistical algorithm⁷⁴ in at least 4 of 6 samples within a cell type (IPE or RPE) and the expression levels were at least 50 on densitometry. Gene expression was considered absent within a cell type (IPE or RPE) if the gene was undetected using the Affymetrix GCOS 1.2 statistical algorithm in at least 4 of 6 samples within a given cell type. To minimize ambiguity, we excluded any genes whose expression levels were marginally present or absent. Genes were considered differentially expressed if they were present in both RPE and IPE, the expression levels were at least 50 on densitometry, and the difference in expression level was greater than the 3-fold difference that was statistically significant (P<.01, t test).

Real-time polymerase chain reaction (PCR) was performed as follows: RPE and IPE were harvested from 6 additional donors (aged 55, 65, 71, 77, 81, and 83 years), different from those used to generate the microarray data. The RPE and IPE cells were harvested, and total RNA was prepared as already described. The LightCycler system (Roche Diagnostics, Welwyn Garden City, United Kingdom) was used for real-time quantitative reverse transcriptase–PCR. An RNA Amplification Kit SYBR Green I (Roche Molecular Biochemicals, Mannheim, Germany) was used to synthesize the first-strand cDNA and subsequent amplification using gene-specific primers. The PCR reaction solution contained 0.5 μg of total RNA, 6-mM magnesium chloride, and 0.5-μM of each primer (primer oligo sequences are available online at http://www.columbia.edu/~hc2002/RPEvsIPE_data_onWeb/). Other components in the reverse transcriptase–PCR master mix included buffer, enzyme, SYBR Green I, and deoxyribonucleotide triphosphate. For reverse transcription, the 20 μL of reaction capillaries were incubated at 55°C for 10 minutes, followed by incubation at 95°C for 30 seconds. Polymerase chain reaction was performed using an initial denaturation for 1 second at 95°C, followed by 45 cycles of denaturation for 1 second at 95°C, annealing for 10 seconds at 55°C, and extension for 13 seconds at 72°C in a programmable LightCycler. A melting curve analysis was performed by following the final cycle with incubation at 95°C for 1 second, at 65°C for 10 seconds, and then at a temperature transition rate of 20°C per second to reach 95°C. Negative control samples for the reverse transcriptase–PCR analysis, which contained all reaction components except RNA, were performed simultaneously to determine when the nonspecific exponential amplification cycle number was reached.

Quality was assessed using the hybridization signals from the 3′, middle, and 5′ fragments of mRNA of 20 housekeeping genes coded in the Affymetrix DNA chips.⁸¹ All 12 DNA chips passed quality control (data not shown).

Clustering analysis is a statistical technique to sort heterogeneous samples into several distinct clusters so that samples within each cluster are more closely related to one another than samples from different clusters.^82,83 Hierarchical cluster analysis yields a tree diagram, with the branches indicating the relationship of samples within the cluster to other samples within and outside the clusters.⁸⁴ Hierarchical cluster analysis demonstrated that the gene expression profiles of RPE and IPE cluster into 2 distinct groups with a discernable overlap in 1 sample (Figure 1).

Among 12 600 gene probes on the Affymetrix human U95Av2 microarray chip, a mean ± SD of 5308 ± 416 gene probes were expressed in 6 RPE samples, compared with 6130 ± 205 gene probes in 6 IPE samples. Of these, 4895 genes were expressed in all 6 samples of both cell types (Figure 2). Sixty-eight genes were expressed in RPE samples but absent from IPE samples based on the criteria already defined (ie, present in 4 of 6 samples) (Table 2). Reverse transcriptase–PCR in a few genes confirmed the microarray data. Of the 2500 most abundantly expressed genes in adult RPE, 438 genes were absent from the 2500 most abundantly expressed genes in IPE. Of the 2500 most abundantly expressed genes in IPE, 442 genes were absent from the 2500 most abundantly expressed genes in RPE. One hundred fifty-four genes were detected in IPE that were undetected in RPE. Other details and the complete lists of genes expressed in RPE and IPE are available at the URL provided in the “Real-Time Polymerase Chain Reaction” subsection of the “Methods” section.

After data normalization, there were 22 genes in RPE samples whose expression levels were greater than 3-fold higher compared with IPE samples, and 19 of these genes were contained within the 2500 most abundantly expressed genes in RPE (Table 3). There were 147 genes in IPE samples whose expression levels were greater than 3-fold lower compared with RPE samples, and 119 of these genes were contained within the 2500 most abundantly expressed genes in IPE (details are available at the URL provided in the “Real-Time Polymerase Chain Reaction” subsection of the “Methods” section). We compared the expression levels of genes that were associated with known RPE functions using a searchable database available at the Affymetrix Web site (http://www.affymetrix.com).⁸⁵ Retinal pigment epithelium function–related genes detected in RPE but not in IPE included 2 genes involved in phagocytosis (thrombospondin 1 and ras-related C3 botulinum toxin substrate 2), 1 gene important for vitamin A metabolism (retinol dehydrogenase 5), 1 gene involved in angiogenesis (angiopoietin 1), and 15 genes involved in cell adhesion (Table 4). Using similar search techniques for analyzing function-related genes expressed only in IPE, we identified 32 genes that were not expressed in RPE; these were involved in cell adhesion, extracellular matrix remodeling, phagocytosis, tight junction formation, and vitamin A metabolism (Table 5).

In the last decade, investigators pioneered the use of IPE as a replacement for RPE in retinal degenerations, including AMD.³⁵ This use of IPE is based on the common embryological origin of these 2 cell lines, the ready availability of autologous IPE via iris biopsy, and the need to replace RPE in various disease states. Application of IPE transplantation for treatment of tapetoretinal degenerations due to a known gene defect, such as Leber congenital amaurosis and RPE-dependent forms of retinitis pigmentosa, is likely to be unfruitful because autologous IPE and RPE would have the same genetic defect. The largest clinical application for autologous IPE transplantation may be in repair of age-related cell and tissue loss in AMD, in which transplanted IPE could replace native RPE removed during submacular surgery for exudative AMD or lost during the development of geographic atrophy in nonexudative AMD.

To date, a few laboratory and clinical studies have been performed to determine the ability of IPE to survive after subretinal transplantation and to perform RPE functions, including outer segment phagocytosis, recycling of visual pigment, and release of cytokines and other growth factors.⁶⁷ Previous authors have concluded that IPE can survive at least 6 months after subretinal transplantation, but proper interpretation of these results is confounded by difficulty in identifying transplanted cells unequivocally.^{65,66,73,86,87} Results of initial studies^68,88 suggested that subretinal or choroidal IPE transplantation may slow the rate of photoreceptor degeneration in the Royal College of Surgeons rat model for several months compared with untreated controls. However, the rescue effect of transplanted IPE cells was no better than that of sham surgery.⁶⁹

Iris pigment epithelium function in vitro and after subretinal transplantation in vivo has also been previously investigated. Iris pigment epithelium is capable of retinol metabolism,⁸⁹ and transplanted IPE can ingest outer segments.⁶⁹ The ability of cultured IPE to phagocytose latex beads comprises 76% of the activity of RPE.⁹⁰ Cultured IPE maintains melanogenesis for up to 5 passages in tissue culture.⁹¹ Iris pigment epithelium and RPE form monolayers on Descemet membrane^92,93 and exhibit similar growth on native and micropatterned human lens capsules.⁹⁴ Iris pigment epithelium can form tight junctions, raising the possibility that transplanted IPE could reestablish the blood-retinal barrier normally formed by RPE.⁶⁷

Cultured bovine IPE and RPE express retinol-binding protein and cellular retinaldehyde–binding protein 1 (Table 2, probe set 35887_at), suggesting that both cell types have some capacity for transporting and metabolizing retinol.⁹⁵ Semiquantitative reverse transcriptase–PCR demonstrates that bovine IPE and RPE express mRNA for cellular retinaldehyde–binding protein and 11-cis-dehydrogenase at similar levels, but IPE expresses lower levels of p63, the presumed RPE membrane receptor for retinoids.⁹⁶ Iris pigment epithelium and RPE synthesize and release a similar but unidentical pattern of cytokines and their receptors, but major differences exist in the release of insulin and tumor necrosis factor α.^97,98 Messenger RNA levels of vascular endothelial growth factor and vascular endothelial growth factor receptor 2 (FLK-1) were lower in IPE cells than in RPE cells, which may be an important functional difference because vascular endothelial growth factor may maintain the choriocapillaris and assist in the pathological development of choroidal neovascularization in vivo.^97,98

Only a few clinical studies have been performed to date on subretinal transplantation of IPE to replace surgically excised RPE in patients with exudative AMD. Autologous IPE transplantation has been performed in 35 patients after removal of subfoveal choroidal neovascular membranes, with no significant difference in visual acuity between patients who received transplants vs those who underwent choroidal neovascularization removal alone. Autologous IPE translocation after submacular membranectomy can preserve foveal function at a low level but does not improve visual acuity.^62,63 These poor functional results are consistent with the suboptimal attachment and survival of IPE and RPE on aged Bruch’s membrane.⁹⁹ Despite the lack of visual improvement, subretinal IPE transplantation in patients with AMD may prevent recurrence of subretinal neovascularization.^35,68

We observed major differences in the gene expression profiles of primary RPE vs IPE harvested from the same donor eyes, including the lack of expression in IPE of genes known to be critical for RPE function. For example, IPE does not express the gene for retinol dehydrogenase, whose gene product is necessary for recycling visual pigments. Recoverin is a visual cycle protein expressed in abundance in RPE but not in IPE, although its role in RPE function is unknown. Iris pigment epithelium does not express other major functional RPE genes, including angiopoietin 1, S-antigen, and a transcriptional regulator of the c-Fos promoter. Numerous cell adhesion genes and additional genes related to RPE phagocytosis, tight junction formation, and vitamin A metabolism are missing from IPE, including thrombospondin 1 and ras-related C3 botulinum toxin substrate 2 (Table 4).

For IPE to replace surgically excised or dysfunctional RPE, the transplanted IPE would probably need to develop an expression profile that closely resembles that of native RPE. Our results suggest that the native IPE gene expression profile may be a potential obstacle to successful subretinal transplantation. Because the microenvironment of cells influences their behavior and gene expression, we cannot exclude the possibility that the gene expression profile of IPE may change after subretinal transplantation to more closely resemble that of native RPE. However, our data suggest that the expression levels of many genes must change for IPE to resemble RPE. Some authors have suggested that transplanted IPE can serve as a potential reservoir for a single growth factor or cytokine and thereby rescue adjacent cells from the effects of progressive tapetoretinal degeneration.¹⁰⁰ For example, IPE that is induced to transcribe the BNDF gene protects against retinal damage due to N-methyl-D-aspartate–induced neuronal death and light toxicity.^101,102 Iris pigment epithelium that is genetically modified to express pigment epithelial–derived factor inhibits choroidal neovascularization in a rat model of laser-induced choroidal neovascularization and increases the survival of and preserves the rhodopsin expression of photoreceptor cells in the Royal College of Surgeons rat model.⁷⁰ For such applications, the striking difference in the gene expression profiles between RPE and IPE may be less of an obstacle to successful cell-based therapy. Additional studies, including determination of the gene expression profiles of IPE and RPE after subretinal transplantation, are needed to determine if the microenvironment of the subretinal space will have a marked effect on the gene expression profile of IPE.

Correspondence: Lucian V. Del Priore, MD, PhD, Department of Ophthalmology, Harkness Eye Institute, Columbia University, 635 W 165th St, New York, NY 10032 (ld394@columbia.edu).

Submitted for Publication: August 8, 2005; final revision received November 23, 2005; accepted December 12, 2005.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Robert L. Burch III Fund, the Macula Society, the Foundation Fighting Blindness, and the Commonwealth of Kentucky Research Challenge Trust Fund (Dr Kaplan) and by unrestricted funds from Research to Prevent Blindness. Dr Tezel is the recipient of a Research to Prevent Blindness Career Development Award.