Similarity of mRNA Phenotypes of Morphologically Normal Macular and Peripheral Retinal Pigment Epithelial Cells in Older Human Eyes

Abstract

purpose. To determine the expression profiles of morphologically normal human retinal pigment epithelial (RPE) cells that originate from the macula and periphery.

methods. Morphologically normal RPE cells from 15 human globes from donors aged 52 to 82 years old were laser capture microdissected. Total RNA from 5000 cells was SMART amplified, [33]P-labeled, and hybridized to a cDNA array containing 4325 known genes. Expression profiles were analyzed by hierarchical cluster analysis, Prediction Analysis of Microarrays (PAM), and Significance Analysis for Microarrays (SAM). Differentially expressed genes were evaluated further by real time RT-PCR.

results. The overall expression profiles of RPE cells from the macula and periphery were similar. Unsupervised and supervised hierarchical cluster analysis showed that patient genotype was a stronger separating factor than topographical location. SAM analysis identified 11 genes that were underexpressed by macular RPE cells. The expression patterns of these 11 genes were confirmed by real time RT-PCR, with 5 genes reaching statistical significance.

conclusions. Whereas the overall expression profiles were similar between cells from the macula and periphery, subtle differential expression of five genes could contribute to RPE phenotypic differences based on topographic location.

The retinal pigment epithelium (RPE) forms a selective barrier between the neurosensory retina and choriocapillaris. It has a wide range of functions including phagocytosis and recycling of photoreceptor material, isomerization of visual pigment, quenching of oxidative and photo-oxidative stress, and maintenance of the underlying Bruch’s membrane. ¹ The RPE has subtle, but distinct phenotypic differences based on topographical location. For example, macular RPE cells are more columnar with higher melanin content than peripherally located cells. The RPE has a differential response to chronological aging based on topographical location. Macular RPE cells undergo a preferential morphologic deterioration and apoptotic cell loss compared to peripherally located cells. ^{2 3 4 5} Whereas variations in the expression of a number of genes by the RPE due to topographical location have been previously reported, ^{6 7 8 9} a comprehensive understanding of the molecular events that distinguish RPE cells by topography, however, is relatively unknown. This information could give insights into the molecular events that predispose macular RPE cells to morphologic and cell survival changes with aging. 

A hypothesis that the mRNA phenotype of the RPE varies by topographical location was made. The onset of microarray technology has resulted in a rapid and more comprehensive molecular characterization than previously for a number of diseases. ^{10 11 12} Technical issues such as difficulty in dissecting macular from peripheral RPE cells, the inherent heterogeneous phenotype of RPE cells within a macula, particularly in elderly eyes, ² and the resultant small amount of tissue available for molecular studies, have been impediments that prevent determining accurate mRNA phenotypes. The development of laser capture microdissection allows for the removal of pure cell populations from small regions of tissue, such as the macula, for microarray studies. ¹³ Since little is known about the expression profiles of RPE cells in vivo, the expression profile of morphologically normal RPE cells microdissected from the macula and periphery of human donor globes was characterized. 

Materials and Methods

Tissue Processing

Fifteen globes from donors aged 52 to 82 years were obtained from NDRI (Philadelphia, PA) within 7 hours of death, with an average death to sectioning time of 37 hours (Table 1) . Based on the report by Johnston et al., ¹⁴ which found that premorbid conditions, such as rapidity of death, were the greatest influencing factor in RNA quality, globes were used where the donors had been on life support for <24 hours. To avoid introducing donor-related bias, one eye from a donor was used for study. Three independent observers verified that there was no evidence of drusen or other fundus abnormalities by examination with the dissecting microscope. Using RNase free conditions, 6 × 6 mm calottes of the macula and nasal equatorial-anterior periphery were obtained for this study. Each calotte was cryoprotected using the technique of Barthel and Raymond ¹⁵ with slight modification. Briefly, calottes were progressively infiltrated with sucrose by 10-minute incubations at 4°C in PBS containing 10%, 12.5%, 15%, and 20% sucrose (w/v). Calottes were then infiltrated in a 2:1 sucrose 20% (w/v):OCT compound (VWR International, Bridgeport, NJ) mixture for 30 minutes, embedded in fresh 2:1 sucrose 20% (w/v):OCT mixture, and frozen by immersion in isopentane (Fisher-Aldrich Chemical Co., Inc., Milwaukee, WI) chilled with dry ice. All tissue blocks were stored at −80°C for later use. 

Tissue Sectioning and Staining

Tissue blocks were sectioned on a cryotome (Leica Microsystems, Inc., Bannockburn, IL) at 7 μm thickness. For macular sections, only sections that included the optic nerve were used for this study. Individual sections were fixed in 70% ethanol for 30 seconds, and then stained with Hematoxylin and Eosin Y (Fisher Scientific, Inc., Pittsburgh, PA) each for 15 seconds. Sections were used immediately for laser capture microdissection. 

Laser Capture Microdissection

Sample sections from each sample were examined before laser capture microdissection, and three observers confirmed that RPE morphologic abnormalities, drusen, or basal deposits were not observed. Cells of interest were dissected with an Arcturus PixCell II laser capture microdissector (Arcturus Engineering, Inc., Mountain View, CA) using transfer film (Cap-Sure TF-100; Arcturus Engineering) according to our previously published protocol. ¹⁶ Morphologically normal RPE cells attached to nonthickened Bruch’s membrane were defined using the criteria established by Curcio et al. and Sarks. ^{17 18} Normal macular RPE were defined as having regular cuboidal–columnar cell shape, homogeneous melanin pigmentation, and a height estimated at 10 to 15 μm ¹⁹ using the 7.5 and 15 μm spot size of the laser aiming beam. Peripheral RPE typically have a cuboidal cell shape and a lower cell height. Normal peripheral RPE were defined as having a cuboidal cell shape, homogenous melanin pigmentation, and a height of ≥7.5 μm. In addition, cells were included as normal only if they were overlying a nonthickened Bruch’s membrane. For this age group, Okubo et al. ^{18 20} have reported a thickness of 3 to 5 μm, so we defined a nonthickened Bruch’s membrane as <¼ RPE cell height and not associated with drusen or other Bruch’s membrane abnormality. After dissection, the transfer cap was inspected with the microscope for contaminating tissue, which verified a cleavage plane at the RPE–Bruch’s membrane junction, before being placed in 200 μL denaturing buffer that contained 4 M guanidine isothiocyanate, 0.02 M sodium citrate, 0.5% sarcosyl, and 2 μL β-mercaptoethanol (14.5 M; Qiagen Inc, Valencia, CA). 

RNA Extraction

Total RNA was extracted from laser captured RPE cells using the RNeasy Mini-kit (Qiagen Inc.) according to the manufacturer’s recommendations. RNA was treated with DNase I (Qiagen, Inc.) during RNA purification. A sample of RPE cells was obtained from a peripheral calotte that was not used for microarray analysis from each donor which showed preserved 28S and 18S rRNA bands. Before synthesizing probe, RNA quality was assessed by the expression of GAPDH from 100 cells using real time RTPCR with primers designed at the 5′ end of the gene. 

Probe Synthesis

Probe was prepared from total RNA of 5000 laser captured RPE cells using the Super SMART PCR cDNA synthesis kit (BD Biosciences Clontech, Palo Alto, CA) according to the manufacturer’s recommendations. Total RNA was reverse transcribed, and column-purified first strand cDNA was PCR amplified. The PCR cycle number was optimized using a small aliquot (5 μL) after 15 cycles and every three cycles thereafter, with 1.2% agarose/ethidium bromide gel electrophoresis. Typically, 23 cycles produced ds cDNA that remained in the exponential phase of amplification that produced a smear from 0.5 to 5 kb. The cDNA was column purified with a NucleoSpin Extraction kit (BD Biosciences Clontech) and labeled using the BD Atlas SMART probe amplification kit (BD Biosciences Clontech) in the presence of 50 μCi [α-³³P]dATP, 1 μg random hexamers, and 2 units Klenow fragment at 50°C for 30 minutes. The probe was purified by passage through a Bio-Spin 6 chromatography column (BioRad Laboratories, Hercules, CA). 

Microarray Analysis

The labeled cDNA was denatured and hybridized to the cDNA GeneFilter “Named genes” human array (4325 genes; Invitrogen/Research Genetics, Inc., Huntsville, AL) using the manufacturer’s protocol. This array contains genes with known function that are an insert DNA from a sequence-verified IMAGE/LLNL clone from the 3′ end of the gene. Arrays were exposed for 3 days to a high density phosphorimager screen (BioRad Laboratories) and scanned at 50 μm resolution in a phosphorimager instrument (FX Pro-Plus; BioRad Laboratories). 

Image and Statistical Analysis

The TIFF images acquired from the phosphorimager were imported into the image analysis software (Pathways 3; Invitrogen/Research Genetics, Inc.). This software aligns the images, quantifies a signal intensity for each gene, and normalizes the different hybridization signals on the basis of the 75% average signal intensity of the entire array. To allow statistical comparison of the arrays, the gene expression signals were scaled according to the method of Tusher et al., and our previously published protocol. ^{21 22}  

Hierarchical cluster analysis of macular and peripheral RPE cells was determined with cluster analysis and visualized with TreeView. ²³ A class prediction from gene expression profiling based on the “nearest prototype (centroid) classifier” approach, was performed with the Prediction Analysis of Microarrays (PAM) software, to identify gene subsets that best characterize each class, i.e., macula and periphery. ²⁴ This program uses soft thresholding rather than screening, and focuses on misclassification error. ²⁴ The Significance Analysis of Microarrays method (SAM, version 1.12) was used to determine individual gene expression differences by topographical distribution. ²¹ SAM calculates a significance score for each gene based on the gene expression change relative to the SD of repeated values. 

Real Time Reverse Transcription–Polymerase Chain Reaction (RT-PCR)

An aliquot (1 ng) from the same ds cDNA used for microarray analysis was assayed using the LightCycler apparatus (Roche Diagnostics, Nutley, NJ). The primer sequences used in this study were designed using Primer 3 (Whitehead Institute/MIT, Cambridge, MA) or the LightCycler Probe Design software, and sequences were verified using NCBI Unigene (www.ncbi.nlm.hih.gov/) (Table 2) . The standard curve consisted of PCR products for the gene of interest using serial dilutions of 1 pg–10⁻⁶ pg. Thermo cycling of each reaction was performed in a final volume of 20 μL containing SYBR Green PCR Master Mix (10 μL; Qiagen, Inc.), Primer A and B (10 μM each), and 2 μL template DNA in a concentration of 2.5 mM MgCl₂. The cDNA was denatured at 95°C for 15 minutes followed by PCR settings of 94°C for 15 seconds, Tm-5°C for 20 seconds, and 72°C for x seconds where x = PCR product length (bp/20). PCR products were quantified using the second derivate maximum values calculated by the Light-Cycler analysis software. Negative controls without template were produced for each run. Expression levels of all genes were normalized to GAPDH mRNA levels. All PCR products were checked by melting point analysis. For each sample, the experiment was repeated once, and the average expression of each sample was used to calculate the expression ratio. The Wilcoxon signed rank test was used to compare the differential gene expression between macular and peripheral RPE. P < 0.05 was considered significant. 

Results

Morphology of RPE and Bruch’s Membrane

Figure 1 shows the typical appearance of the donor globes. Gross examination showed no evidence of pathology such as drusen, RPE pigmentary changes, or hemorrhage in any of the specimens. Histopathologic evaluation showed that the RPE and Bruch’s membrane of cells selected for microdissection were homogeneous without obvious abnormality. As seen in Figure 2 , cells had a typical cuboidal shape with dense melanin pigment, and Bruch’s membrane was nonthickened without drusen or basal deposits. 

Assessment of RNA Quality

Several parameters were used to assess the RNA quality in addition to GAPDH expression using real time RT-PCR (as described in Methods). The distribution of the cDNA after SMART amplification for each sample showed a product size distribution from 0.5 to 5 kb, as recommended by the manufacturer which was similar to the distribution from ARPE-19 cells which had a 28S/18S = 2.0. The normalized, corrected sum and mean signal for each array plotted against death to enucleation or dissection time showed a positively sloped regression line suggestive of no signal degradation. The expression of two housekeeping genes (GAPDH and β-actin) and three genes known to be expressed by the RPE at low to moderate copy level (tissue inhibitor of metalloproteinase 3 [TIMP3], arrestin-3, and retinoic acid receptor responder 2) plotted against the death to dissection time showed a positively sloped regression line that also suggests against signal degeneration (Fig. 3) . Similar results were seen with death to enucleation time for each gene (data not shown). 

Expression Profiles of Macular and Peripheral RPE

Table 3 lists the 50 most highly expressed genes by macular and peripheral RPE cells. A majority (84%) of genes were expressed by both macular and peripheral cells, and the remaining 16% of genes were expressed by both cells within the top 75 most highly expressed genes. Functional categories of genes that were expressed by both macular and peripheral cells include protein synthesis, processing, catabolism (26%), cell proliferation/survival (24%), cytoskeleton/differentiation (19%), and intracellular trafficking (12%). Several other highly expressed genes highlight the multiple functions of the RPE such as D-dopachrome tautomerase (melanin biosynthesis), carbonic anhydrase VA (fluid regulation), phosphodiesterase 6H, cGMP-specific, cone, gamma (visual transduction), glutathione peroxidase 3 (plasma; oxidative stress defense); cytochrome c oxidase subunit VIIc and surfeit I (respiratory chain/energy production). 

Unsupervised cluster analysis showed that the expression profiles of macular and peripheral RPE cells overall, were similar. As shown in Figure 4 , the transcriptomes were separated into two main branches consisting of eight donors in one branch, and seven donors in the other branch. Each macula and peripheral expression profile from a donor segregated within the same branch, indicating that donor genotype was a stronger influencing factor than topographical location, age, or gender. 

A supervised cluster analysis strategy was also performed to determine whether a smaller gene set identified differences in expression profiles. Altered cell morphology and reduced cell number occur preferentially to macular RPE cells with aging. ^{2 3 4 5} Hierarchical cluster analysis using 533 genes related to cell differentiation, and 370 genes related to cell cycle/proliferation/apoptosis, both showed no separation based on topographical location. 

Fifty-six genes that are involved in the main oxidative stress defense systems are represented on this array. The normalized average expression values are listed as seen in Table 4 . The most highly expressed antioxidant genes include superoxide dismutase (SOD), and genes from the glutathione antioxidant system including glutathione-S-transferase (GST), glutathione peroxidase (GPX), γ-glutamyltransferase-like activity 1, peroxiredoxin, and glutathione synthesis. A complete set of isoforms for several of these genes were represented on the array which allows a characterization of the relevant isoenzymes expressed by the RPE. For example, SOD1 and SOD2, which detoxifies superoxide, had higher expression than SOD3. GPX 3, GST A4, M1, M4, O1 and peroxiredoxin-1 were the most highly expressed isoforms. Supervised cluster analysis using this gene set also showed no difference between RPE cells derived from the macula or periphery. 

Currently, no single method has been established to determine significant gene expression profiles. Since hierarchical cluster analysis did not distinguish macular from peripheral RPE cells, an alternative approach to determine the gene set that best characterized each class (i.e., macula and periphery) with a “nearest shrunken centroid” strategy using the Predictive Analysis of Microarrays software. The purpose of this approach is to identify a group of genes that accurately classify categories based on gene expression. A number of thresholds were investigated. A threshold (delta = 2.5) that minimized the error, identified fragile histidine triad gene, zinc finger protein 74, UV radiation resistance associated gene, cysteine-rich, angiogenic inducer, 61, N-acetylneuraminic acid phosphate synthase, sialic acid synthase, aldehyde dehydrogenase 6, cytochrome c oxidase subunit VIII, and glutathione S-transferase M1 genes that were predictive of macular cells. This gene set predicted macular versus peripheral RPE cells in 29 of 30 arrays. However, the wide range of probabilities from 0.5 to 0.9 suggests that distinct separation is minimal. 

Gene Expression Differences between Macular and Peripheral RPE

Individual gene expression differences between macula and periphery were assessed by SAM analysis. With a false discovery rate of 9%, two-class unpaired SAM analysis identified 11 genes that had significant differential expression (Table 5) . The fold-differences in expression between macular and peripheral cells were small with a maximum of 4.2-fold. All genes were downregulated by macular RPE cells, and have functions related to apoptosis, differentiation, oxidative stress, metabolism, and matrix regulation. 

Real Time RT-PCR Confirmation

To validate the microarray results, real time RT-PCR was performed on macular and peripheral samples from 10 eyes for differentially expressed genes identified by SAM analysis. Table 5 shows that the reduced expression pattern by macular RPE was confirmed for all 11 genes on all the samples while 5 genes reached statistical significance. 

Discussion

We emphasize that this study examines topographical differences in gene expression by morphologically normal, native RPE cells, and not age-related or age-related macular degeneration changes. While the number of globes in this study constitutes the largest microarray analysis on native RPE reported thus far, the sample size is relatively small. The FDR <10% was used as a guide to estimate adequate sample size since this approach has been advocated for power calculations in microarray studies. ²⁵ Hwang et al. ²⁶ reported that homogeneous samples from an entire population reduce the sample size required to achieve statistical significance for microarray studies. We suggest that laser capture microdissection reduces the sample size because a uniform cell population produces smaller expression variability than heterogeneous cell populations. 

The ability to assess RNA quality and a sample quantity that misrepresents the intended cell population are potential limitations of laser capture microdissection. Donors were chosen who had limited premortem illness since the rapidity of death is a greater influence on RNA quality than postmortem factors. ¹⁴ The arrays were selected because they contain 700 to 1000 BP cDNA inserts from the 3′ end of the gene, the region which is most resistant to RNA degradation. We found similar expression of GAPDH on the arrays and by real-time RT-PCR using primers designed from the 5′ end, a wide size distribution of amplified cDNA products, and no signal degradation of total arrays or individually selected genes when plotted against postmortem factors. While these analyses do not quantify the degree of RNA degradation, they do indicate that this set of globes had similar quality RNA. Amplification can potentially induce bias. Seth et al. however, showed that the relative expression of low, medium, and high abundance genes was retained after SMART amplification independent of transcript abundance, coding region, and PCR product size. ²⁷ Amplification can allow a reproducible microarray signal from as few as 10 cells. Since several studies have demonstrated significant variability in gene expression across RPE cells in vivo, ^{6 8 9} we were reluctant to reduce the cell number to this level over concern of obtaining an expression profile that misrepresents the intended cell population. By using 5000 cells for our analysis, we acknowledge that the expression profiles identified in this study may not be generalizable to all RPE cells or alternatively, the cell population studied may not represent cells that are vulnerable to preferential aging changes within the macula or periphery. 

These data provide insights into the genes that maintain RPE homeostasis. The most highly abundant genes were common to RPE cells regardless of topographical location, and illustrate the wide functional diversity of the RPE cell. Of the most highly expressed genes, 26% are involved in protein synthesis and degradation. Sharon et al., ²⁸ using SAGE analysis of native RPE from a single donor, found that 30% of the most abundant tags were related to protein synthesis and degradation, possibly due to the RPE’s high phagocytic activity. Buraczynska et al. ²⁹ characterized a native RPE library of over 1100 genes and expressed sequence tags (ESTs) from two donors. One hundred and sixty-seven genes from this data set were identified on the array, and all were expressed by RPE cells, regardless of location. While a complete comparison with these libraries is difficult due to differences in technique, the present results along with these previous studies, help to define the mRNA phenotype of native RPE cells. 

The RPE is exposed to significant oxidative stress due to its high metabolic activity, high oxygen fluxes from the choriocapillaris, and marked light exposure, and as a result, has a significant antioxidant defense system. A complete isoform set for several important antioxidant enzymes were evaluated in this study. The high expression suggests that SOD1 and SOD2, GPX3, GST A4, M1, M4, O1, and peroxiredoxin 1 are relevant isoenzymes in the RPE cell’s oxidative defense system. While high catalase expression has been previously reported, ³⁰ our low catalase expression is more consistent with reports of an age-related decline by the RPE. ^{30 31} The RPE is susceptible to oxidative damage by lipid peroxidation products from phagocytosed outer segments. ³² Lipid peroxidation products are detoxified not by SOD and catalase, but instead by the GST alpha class and GPXs. ^{33 34 35 36} The high expression of GSTA4 and GPX3 suggests that these isoforms could be relevant lipid peroxidation detoxifiers for the RPE. 

Our hierarchical cluster analysis found that the overall expression profiles between macular and peripheral RPE were similar, and suggested that the genotype was a stronger factor than topography. Alternatively, race, gender, or exposure to similar environmental conditions such as UV exposure, diet, associated co-morbidities, or medications, could have influenced the gene expression profiles. Oleksiak et al. ³⁷ showed that, despite considerable expression profile variability among individuals within the same population, many important expression differences were small. The authors reasoned that genes relevant to a phenotype are tightly regulated and have constant expression, so that small expression changes will produce biologically important differences. ³⁷ This supposition has particular relevance when determining topographically-related changes since we would expect to find small expression changes by regional location. In the present study, the SAM program identified eleven genes with < 4.2-fold expression differences. Due to this small expression difference, statistically validated real time RT-PCR results were used to determine genes that may distinguish macular from peripheral cells. While the real time PCR experiments validated the expression pattern of all 11 genes tested, only five genes were statistically validated. Besides the relatively small expression differences between topographical location, the variable expression of GAPDH which was used for normalization, could be a confounding factor. Recent studies have suggested that there is no housekeeping gene that is optimally normalizes gene expression, and that using total RNA may be a more accurate method to normalize gene expression. ^{38 39} The small starting material from laser captured material prevents accurate measurement of total RNA from samples. While no statistical difference between macular and peripheral RPE GAPDH expression by the arrays and RT-PCR (data not shown) was found, the variability of GAPDH expression across donors could have influenced the differential expression results for real time RT-PCR validation experiments. Regardless, these results demonstrate the importance of statistical assessment for RT-PCR validation studies, the need for finding the optimal method for normalizing expression, and suggest that the genes with statistically significant differential expression are worth exploration to determine a role in distinguishing macular from peripheral RPE. The translation of an mRNA into a protein can be highly variable, and post-translational protein modifications are important changes that influence aging. Ultimately, these gene expression changes need to be correlated with their respective protein levels, and more importantly, how these changes influence their biological function. 

The macula experiences more apoptotic RPE cell loss than the periphery with aging. ^{3 4 5} The SAM and RT-PCR analysis revealed that the cell cycle gene c-KIT was underexpressed by macular cells. c-KIT has been linked to bcl-2 upregulation, and when downregulated, makes cells susceptible to apoptosis. ⁴⁰ Likewise, since cysteine-rich, angiogenic inducer 61 is a cytokine that promotes cell proliferation, it’s under-expression by macular cells would also support preferential cell loss in the macula. ⁴¹  

Decreased macular GSTM1 expression was revealed by SAM and RT-PCR analyses. The GST Mu class is highly polymorphic. For example, 50% of the white population lacks GSTM1 activity due to two GSTM1 null alleles. ⁴² Patients with the GSTM1, but not the P1, T1, or Z1 null genotype, do not neutralize photo-oxidative stress, and are highly susceptible to solar keratosis, ⁴³ or if they smoke, are at increased risk for atherosclerosis from an impaired ability to detoxify tobacco smoke, and have increased oxidatively-induced DNA damage within atherosclerotic lesions. ^{44 45} It is possible that reduced GSTM1 activity, whether from genetic susceptibility or age-related decline, could make macular RPE susceptible over time, to either tobacco related or photo-oxidative stress, which coincidently, are risk factors for age-related macular degeneration. ^{46 47}  

Compared to peripheral cells, Watzke et al. ² observed a preferential degeneration in macular RPE cell morphology with aging. Our analysis identified reduced expression of aldehyde dehydrogenase 6 (ALD6). ALD6 is an essential enzyme in the synthesis retinoic acid, which has an established role in epithelial cell differentiation. ⁴⁸ The reduced macular expression of ALD6 could support preferential aging related morphologic changes to macular RPE cells via alterations in the retinoic acid pathway. 

Genes were examined that, in general, contribute to the overall homeostasis of a number of cell types. This approach may be justified since to date, little is known about the overall RPE mRNA phenotype. A shortcoming of this strategy, of course, is that RPE specific genes were not examined. Currently, one available human RPE specific library contains 1100 nonredundant genes. ²⁹ We are currently evaluating in detail, the expression signature of macular RPE cells using an expanded RPE specific array. Whereas the functional effects of these differentially expressed genes remain unexplored, the results of this study provide a foundation for studying differences in macular RPE cells as a function of topography. 

 Supported in part by NIH/EY 14055 (JTH); the Michael Panitch Macular Degeneration Research Fund; gifts from Aleda Wright, and Rick and Sandy Forsythe; and an unrestricted award from the Research to Prevent Blindness (RPB) to the Wilmer Eye Institute. JTH is the recipient of a Lew R. Wasserman Merit Award from the RPB.

 Submitted for publication February 18, 2004; revised April 29, 2004; accepted June 1, 2004.

 Disclosure: K. Ishibashi, None; J. Tian, None; J.T. Handa, None

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

 

Table 1.

 

Donor Eyes Used for Microarray Analysis

Donor No.	Age (y)	Gender	Race	D–E (h)	D–D (h)	Cause of Death
32502	82	M	W	2:40	33	Pneumonia
42202	74	M	W	4:40	29	Pneumonia
42602	82	M	W	3:00	34	Lung carcinoma
52202	76	M	W	2:10	39	Heart failure
60502	74	M	W	3:50	26	GI cancer
70302	73	M	W	3:25	41	Stroke
71802	82	M	W	3:00	40	Skin melanoma
72302	74	F	W	3:35	33	Lung carcinoma
72702	71	M	W	3:45	31	Pneumonia
81302	82	F	W	2:25	41	Emphysema
81402	74	M	W	3:10	41	Emphysema
82002	52	M	W	6:50	40	Chondrosarcoma
082102a	69	M	W	3:55	40	Leukemia
082102b	57	F	W	5:45	29	Ovarian cancer
102602	77	M	W	2:40	39	Myocardial infarction

 D–E, Death to enucleation time, D–D, Death to section time.

Table 2.

 

PCR Primers and Conditions for RT-PCR

Gene		Sequence	Location	Size (BP)	Cycles	Tm (deg)
Fragile histidine triad gene	F	GAT GAA GTG GCC GAT TTG TT	519–538	150	45	83.5
	R	AGC CTT CCT GGG AAG AAC AT	655–674
Aldehyde dehydrogenase 6	F	CAA CAT GCG GAT TGC C	1261–1276	265	50	82.5
	R	TTC ACC TAG TTC TCT GCC A	1507–1525
c-KIT (Stem cell factor receptor)	F	TTC TTT CAA CTT GCA TCC AAC TCC	3024–3047	199	40	80
	R	TAC CTC CCT CTC TTT TTC CAA ATC	3199–3222
Protein kinase C, alpha	F	GTG GCA AAG GAG CAG AGA AC	1883–1902	151	45	81
	R	TGT AAG ATG GGG TGC ACA AA	2014–2033
Nuclear rec subfamily 4, Group A-1	F	CCT GCC AAT CTC CTC ACT TC	1395–1414	154	40	85.5
	R	CGG AGA GCA GGT CGT AGA AC	1529–1548
Cysteine-rich, angiogenic inducer, 61	F	TGG AGC TTG TGG AGT TGA TG	1478–1497	172	45	76.5
	R	TGC CCT CCC ATT TAC TTT TG	1630–1649
UV radiation resistance assoc gene	F	GCT GTA CTG TGG AGC AAG CA	2055–2074	162	45	82
	R	TCC TTC GGG AGA ACT CTT CA	2097–2216
Sialic acid synthase	F	ACA AGA CCT GGA AGG GGA GT	795–814	185	40	87
	R	TTC CGG AAT TTT CAC TTT GG	960–979
Glutathione S-transferase M1	F	TGA AGC CTC AGC TAC CCA CT	902–921	191	40	85.5
	R	AAC CAG TCA ATG CTG CTC CT	1073–1092
Zinc finger protein 74	F	GTC CTC CCT GGC TCC TAG AT	1660–1679	213	40	84
	R	TCC CCT TGT CAA CAA CTT CC	1753–1872
Cytochrome C oxidase subunit VIII	F	GCC AAG ATC CAT TCG TTG C	111–129	154	40	86
	R	AGA ACG GAC CCC TTC ACt CT	245–264
Glyceraldehyde-3-phosphate dehydrogenase	F	GAG TCA ACG GAT TTG GTC GT	95–114	185	40	82
	R	GAC AAG CTT CCC GTT CTC AG	260–279

 F, Forward; R, Reverse; Tm, annealing temperature.

Figure 1.

 

Gross pathologic examination shows no evidence of pathology such as drusen, RPE pigmentary changes, or hemorrhage in a 69-year-old male. None of the globes in this study showed any gross pathologic abnormalities. White spots superior to macula are light reflection artifact.

Figure 2.

 

Morphologically normal macular RPE cells overlying unthickened Bruch’s membrane that were laser capture microdissected. Cells from both the macula and periphery had a typical cuboidal shape with dense melanin pigment. Top panel: cells before laser microdissection; middle panel: the section after dissection with an area that is absent of RPE cells; and bottom panel: microdissected RPE cells that are adherent to the transfer cap. Bar = 10 μm.

Figure 3.

 

Gene expression versus death to dissection (D–D) time.

Table 3.

 

Fifty Most Highly Expressed Genes from Macular and Peripheral RPE

Macula					Periphery
GenBank Acc. No.	Unigene Cluster ID	Gene Name	GenBank Acc. No.	Unigene Cluster ID	Gene Name
R55188	Hs. 143288	Human pre-T/NK cell associated protein (3B3) mRNA, 3′ end	R55188	Hs. 143288	Human pre-T/NK cell associated protein (3B3) mRNA, 3′ end
AA459292	Hs. 352096	CDC28 protein kinase 1	AA459292	Hs. 352096	CDC28 protein kinase 1
AA772066	Hs. 21492	Phosphatidylinositol (4,5) bisphosphate 5-phosphatase, A	AA772066	Hs. 21492	Phosphatidylinositol (4,5) bisphosphate 5-phosphatase, A
AA397824	Hs. 180015	D-Dopachrome tautomerase	AA629719	Hs. 3462	Cytochrome c oxidase subunit VIIc
AA430675	Hs. 8047	Fanconi anemia, complementation group G	AA778392	Hs. 185055	BENE protein
AA778392	Hs. 185055	BENE protein	AA397824	Hs. 180015	D-Dopachrome tautomerase
AA629719	Hs. 3462	Cytochrome c oxidase subunit VIIc	AA430675	Hs. 8047	Fanconi anemia, complementation group G
AA699469	Hs. 177446	Carbonic anhydrase VA, mitochondrial	AA699469	Hs. 177446	Carbonic anhydrase VA, mitochondrial
AA707922	Hs. 54471	Phosphodiesterase 6H, cGMP-specific, cone, gamma	AA707922	Hs. 54471	Phosphodiesterase 6H, cGMP-specific, cone, gamma
N92646	Hs. 182378	Colony stimulating factor 2 receptor, alpha, low-affinity (granulocyte-macrophage)	AA872397	Hs. 113987	Lectin, galactoside-binding, soluble, 2 (galectin 2)
AA629923	Hs. 227823	pM5 protein	AA629923	Hs. 227823	pM5 protein
AA872397	Hs. 113987	Lectin, galactoside-binding, soluble, 2 (galectin 2)	AA644657	Hs. 181165	Major histocompatibility complex, class I, A
AA487197	Hs. 84285	Ubiquitin-conjugating enzyme E21 (homologous to yeast UBC9)	AA292536	Hs. 147049	Cut (Drosophila)-like 1 (CCAAT displacement protein)
AA644657	Hs. 181165	Major histocompatibility complex, class I, A	AA427934	Hs. 138860	Rho GTPase activating protein 1
AA281635	Hs. 315463	Suppression of tumorigenicity 16 (melanoma differentiation)	AA043228	Hs. 194662	Calponin 3, acidic
AA419177	Hs. 184601	Solute carrier family 7 (cationic amino acid transporter, y+ system), member 5	N92646	Hs. 182378	Colony stimulating factor 2 receptor, alpha, low-affinity (granulocyte-macrophage)
AA427934	Hs. 138860	Rho GTPase activating protein 1	AA664180	Hs. 336920	Glutathione peroxidase 3 (plasma)
AA669055	Hs. 73931	Major histocompatibility complex, class II, DQ beta 1	AA487197	Hs. 84285	Ubiquitin-conjugating enzyme E21 (homologous to yeast UBC9)
AA046690	Hs. 149436	Kinesin family member 5B	AA046690	Hs. 149436	Kinesin family member 5B
AA664180	Hs. 336920	Glutathione peroxidase 3 (plasma)	AA464755	Hs. 183805	Ankyrin 1, erythrocytic
AA043228	Hs. 194662	Calponin 3, acidic	R27585	Hs. 82159	Proteasome (prosome, macropain) subunit, alpha type, 1
AA292536	Hs. 147049	Cut (Drosophila)-like 1 (CCAAT displacement protein)	AA281784	Hs. 162808	Phosphoinositide-3-kinase, catalytic, delta polypeptide
AA281784	Hs. 162808	Phosphoinositide-3-kinase, catalytic, delta polypeptide	AA461527	Hs. 198767	COP9 (constitutive photomorphogenic, Arabidopsis, homolog) subunit 5
AA461527	Hs. 198767	COP9 (constitutive photomorphogenic, Arabidopsis, homolog) subunit 5	AA812973	Hs. 73072	Chaperonin containing TCP1, subunit 6B (zeta 2)
AA682851	Hs. 75841	Endoplasmic reticulum lumenal protein	AA669055	Hs. 73931	Major histocompatibility complex, class II, DQ beta 1
AA812973	Hs. 73072	Chaperonin containing TCP1, subunit 6B (zeta 2)	AA872001	Hs. 118796	Annexin A6
N78621	Hs. 5344	Adaptor-related protein complex 1, gamma 1 subunit	AA243439	Hs. 24297	Multiple endocrine neoplasia 1
AA253430	Hs. 91161	Prefoldin 4	AA281635	Hs. 315463	Suppression of tumorigenicity 16 (melanoma differentiation)
R27585	Hs. 82159	Proteasome (prosome, macropain) subunit, alpha type, 1	AA411440	Hs. 155191	Villin 2 (ezrin)
AA293218	Hs.693	Cleavage stimulation factor, 3′ pre-RNA, subunit 2, 64kD	AA608988	Hs.2051	Testis specific protein, Y-linked
AA464755	Hs. 183805	Ankyrin 1, erythrocytic	AA416952	Hs.279607	Calpastatin
AA427899	Hs. 179661	Tubulin, beta polypeptide	AA419177	Hs. 184601	Solute carrier family 7 (cationic amino acid transporter, y+ system), member 5
AA243439	Hs. 24297	Multiple endocrine neoplasia I	AA134555	Hs.78185	GT198, complete ORF
AA397813	Hs. 83758	CDC28 protein kinase 2	T81764	Hs. 172405	Cell division cycle 27
N30302	Hs.83147	Guanine nucleotide binding protein-like 1	AA284528	Hs.241561	Protease, serine, 2 (trypsin 2)
AA709271	Hs. 177691	Neural cell adhesion molecule 2	AA682851	Hs. 75841	Endoplasmic reticulum lumenal protein
AA458507	Hs. 2175	Colony stimulating factor 3 receptor (granulocyte)	N78621	Hs.5344	Adaptor-related protein complex 1, gamma 1 subunit
AA425299	Hs.184276	Solute carrier family 9 (sodium/hydrogen exchanger), isoform 3 regulatory factor 1	AA436564	Hs. 117078	c-mer Proto-oncogene tyrosine kinase
AA436564	Hs.117078	c-mer Proto-oncogene tyrosine kinase	AA709271	Hs. 177691	Neural cell adhesion molecule
AA434404	Hs.74519	Primase, polypeptide 2A (58kD)	N71628	Hs.192861	Spi-B transcription factor (Spi-1/PU.1 related)
AA098896	Hs. 110849	Estrogen-related receptor alpha	AA098896	Hs.110849	Estrogen-related receptor alpha
AA872001	Hs.118796	Annexin A6	AA699560	Hs.3196	Surfeit 1
AA411440	Hs. 155191	Villin 2 (ezrin)	AA397813	Hs.83758	CDC28 protein kinase 2
AA608988	Hs.2051	Testis specific protein, Y-linked	N29376	Hs.153837	Myeloid cell nuclear differentiation antigen
AA609655	Hs.112743	Synaptonemal complex protein 1	AA405800	Hs.89466	Dodecenoyl-Coenzyme A delta isomerase (3,2 trans-enoyl-Coenzyme A isomerase)
AA699560	Hs.3196	Surfeit 1	AA609655	Hs.112743	Synaptonemal complex protein 1
AA521431	Hs.75721	Profilin 1	AA757170	Hs.178215	Vertebrate LIN7 homolog 1, Tax interaction protein 33
AA604492	Hs.83958	Transducin-like enhancer of split 4, homolog of Drosophila E (sp1)	AA434404	Hs. 74519	Primase, polypeptide 2A (58kD)
AA046523	Hs.29463	Centrin, EF-hand protein, 3 (CDC31 yeast homolog)	AA670347	Hs.247551	Glucosidase, beta; acid, pseudogene
AA405800	Hs.89466	Dodecenoyl-Coenzyme A delta isomerase (3,2 trans-enoyl-Coenzyme A isomerase	AA458507	Hs.2175	Colony stimulating factor 3 receptor (granulocyte)

 Order is determined by the normalized, mean expression. Genes expressed by both macular and peripheral RPE is indicated in italics.

Figure 4.

 

Unsupervised cluster analysis of normalized, scaled arrays. The transcriptomes were separated into two main branches (above) consisting of eight donors in one branch, and seven donors in the other branch. Each macular and peripheral expression profile from a donor segregated within the same branch (links below) indicates that donor genotype was a stronger influencing factor than topographical location.

Table 4.

 

Relative Expression of Antioxidant Genes

GenBank Acc. No.	UniGene Cluster ID	Gene Name	Macular Expression	Peripheral Expression
R52548	Hs.75428	Superoxide dismutase 1	8.4	8.3
AA488084	Hs.372783	Superoxide dismutase 2	10.3	10.2
AA454160	Hs.2420	Superoxide dismutase 3	0.29	0.29
N30404	Hs.5002	Copper chaperone for SOD	0.069	0.069
H15685	Hs.395771	Catalase	0.32	0.31
Glutathione System Genes
AA485362	Hs.76686	Glutathione peroxidase 1	2.5	2.6
AA135152	Hs.2704	Glutathione peroxidase 2	0.26	0.26
AA664180	Hs.386793	Glutathione peroxidase 3	14.3	14.2
AA454856	Hs.2706	Glutathione peroxidase 4	0.37	0.36
AA777289	Hs.193974	Glutathione reductase	0.22	0.22
AA463458	Hs.82327	Glutathione synthetase	1.18	1.27
T73468	Hs.378199	Glutathione S-transferase A2	0.14	0.14
N30096	Hs.102484	Glutathione S-transferase A3	0.50	0.52
AA152347	Hs. 169907	Glutathione S-transferase A4	4.15	4.38
AA290738	Hs.301961	Glutathione S-transferase M1	6.3	8.8
AA142971	Hs.279837	Glutathione S-transferase M2	0.050	0.050
R63065	Hs.2006	Glutathione S-transferase M3	0.16	0.16
AA486570	Hs. 348387	Glutathione S-transferase M4	5.9	5.8
AA056232	Hs.75652	Glutathione S-transferase M5	0.060	0.060
R33642	Hs.226795	Glutathione S-transferase pi	0.19	0.19
H99813	Hs.77490	Glutathione S-transferase theta 1	0.098	0.095
AA490208	Hs.1581	Glutathione S-transferase theta 2	0.20	0.20
AA428334	Hs.26403	Glutathione S-transferase zeta 1	0.14	0.13
AA441895	Hs. 11465	Glutathione-S-transferase omega 1	7.1	7.8
AA495936	Hs.389700	Microsomal glutathione S-transferase 1	0.061	0.060
W73474	Hs.81874	Microsomal glutathione S-transferase 2	0.035	0.035
AA291163	Hs.28988	Glutaredoxin	0.26	0.27
AA775803	Hs.180909	Peroxiredoxin 1	3.7	3.6
H68845	Hs.432121	Peroxiredoxin 2	0.60	0.60
H19203	Hs.75454	Peroxiredoxin 3	0.38	0.37
AA459663	Hs.83383	Peroxiredoxin 4	0.082	0.082
Glutathione Synthesis Genes
H72018	Hs.446503	Gamma-glutamyltransferase 1	0.19	0.18
AA150687	Hs.1675	Gamma-glutamyltransferase-like activity 1	2.1	2.1
R87497	Hs.80206	Glucose-6-phosphate dehydrogenase	0.41	0.41
H56069	Hs.151393	Glutamate-cysteine ligase, catalytic	0.11	0.11
W96179	Hs.89709	Glutamate-cysteine ligase, regulatory	0.32	0.31
AA679907	Hs.5337	Isocitrate dehydrogenase 2 (NADP+), mito	0.76	0.87
AA459380	Hs.75253	Isocitrate dehydrogenase 3 (NAD+) gamma	0.083	0.080
Thioredoxin System
AA431967	Hs.432922	Thioredoxin	0.20	0.19
AA464849	Hs.13046	Thioredoxin reductase 1	0.32	0.32
AA078976	Hs.429366	Thioredoxin-like, 32kD	0.042	0.041
Miscellaneous Antioxidant Genes
AA192419	Hs.81029	Biliverdin reductase A	0.28	0.28
N76927	Hs.76289	Biliverdin reductase B	0.075	0.073
T71606	Hs.202833	Heme oxygenase 1	0.077	0.074
AA626370	Hs.284279	Heme oxygenase 2	0.28	0.27
AA872383	Hs.433205	Metallothionein 1E	0.52	0.55
H53340	Hs.433391	Metallothionein 1G	0.41	0.41
H77597	Hs.2667	Metallothionein 1H	0.31	0.30
N80129	Hs.380778	Metallothionein 1L	0.42	0.41
AA488081	Hs.124027	Selenophosphate Synthetase	0.11	0.10
AA070226	Hs.275775	Selenoprotein P, plasma, 1	0.066	0.063
AA283629	Hs. 14231	Selenoprotein W, 1	0.54	0.52

 Predominant isoform is indicated in Italics.

Table 5.

 

Differential Gene Expression between Macular and Peripheral RPE Sorted by SAM with an FDR of 9%, and Real Time RT-PCR

GenBank Accession Number	Unigene Cluster ID	Gene Name	Signal* (Arb units)	Fold Change (P/M) Array	Fold Change (P/M) RT-PCR (P value)	Function
AA2565123	Hs.77252	Fragile histidine triad gene	0.61	4.20	1.63 (0.776)	Cell cycle
AA455235	Hs.75746	Aldehyde dehydrogenase 6	0.24	2.40	53.4 (0.0037)	Retinoic acid synthesis
AA629838	Hs.3057	Zinc finger protein 74	8.5	1.20	3.16 (0.972)	RNA metabolism
N24824	Hs.81665	c-KIT (Stem cell factor receptor)	0.17	1.72	2.56 (0.0029)	Cell cycle
AA490501	Hs.13137	UV radiation resistance associated gene	6.0	1.21	1.02 (0.46)	Apoptosis
AA421701	Hs.274424	Sialic acid synthase	5.3	1.20	10.1 (0.011)	Metabolism
AA862813	Hs.81097	Cytochrome c oxidase subunit VIII	4.9	1.13	1.20 (0.46)	Metabolism
AA777187	Hs.8867	Cysteine-rich, angiogenic inducer, 61	5.3	1.21	2.15 (0.0146)	Matrix regulation
AA290738	Hs.301961	Glutathione S-transferase M1	6.1	1.40	2.45 (0.0356)	Oxidative stress defense
AA029890	Hs.169449	Protein kinase C, alpha	0.33	1.44	2.18 (0.345)	Signal transduction
N94487	Hs.1119	Nuclear receptor subfamily 4, group A-1	1.0	1.25	2.8 (0.49)	Apoptosis, differentiation

*  Median value of all samples (macula + periphery) from microarray analysis.

The authors thank NDRI for the donor eyes.