DIFFERENCES IN GENE EXPRESSION PROFILE OF CULTURED ADULT VERSUS IMMORTALIZED HUMAN RPE Lee Geng, Hui Cai and Lucian V. D

1. DIFFERENCES IN GENE EXPRESSION PROFILE OF CULTURED ADULT VERSUS IMMORTALIZED HUMAN RPE Lee Geng, Hui Cai and Lucian V. Del Priore Department of Ophthalmology, Columbia University, New York, New York Genes expressed in ARPE cells but not detected in adult RPE cells Purpose: Immortalized human RPE (cell line ARPE-19) are used widely to draw inferences about the behavior of adult RPE. We have used DNA microarray analysis to compare the gene expression profiles of these two cell types. A. Methods: Cultured primary RPE from five human donors (age: 48 - 80 years) and ARPE-19 cultured to confluence in five dishes were used for DNA microarray study. Total RNA was isolated using a Qiagen RNeasy Mini Kit. First and second strand cDNA were synthesized with a T7-(dT)24 oligomer for priming and double-stranded cDNA was cleaned with Phase Lock Gels-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) and incubated with fragmentation buffer (Tris-acetate, KOAc and MgOAcat; 94oC for 35 minutes). Target hybridization, washing, staining and scanning probe arrays were done following an Affymetrix GeneChip Expression Analysis Manual. Microarray data were treated with normalization, log transformation, statistic determination for “presence” or “absence” for RPE gene expression profile. Clustering analysis, including PCA was used with Affymetrix Microarray Suite 5.0, Genesis 1.30 software. B. Genes expressed in adult RPE cells but not detected in aRPE-19 cells Figure 2. A. a scatter plot of expression levels of about 6,000 genes in pRPE vs. aRPE-19 shows an incomplete overlap in the gene expression profiles of these two cells types. B. Figure shows the distribution of differentially (1.5-fold) expressed genes in pRPE and aRPE-19 cell types Results: ARPE Figure 3. The expression of 5,932 genes (out of 12,600 genes on microarray Human 95UA chip) was detected in ARPE-19 cells, in comparison to expression of only 4,849 genes in adult RPE cells from all 5 human donor eyes aRPE-19 express primary RPE A. B. Conclusions:There are some similarities but significant differences in the gene expression profile of cultured adult and immortalized ARPE cells, and it is important to note that some specific genes are only expressed in one of these two groups. These studies suggest caution should be exercised when generalizing results obtained from ARPE-19 to results that would be obtained with adult RPE. Figure 1. Principle Component Analysis (A) and hierarchic clustering analysis (B) demonstrate that the gene expression profile of the adult RPE (shown in blue color group) and ARPE-19 ( shown in red color group) cluster into two distinct groups with no discernable overlap. Supported by Research to Prevent Blindness, Robert L. Burch III Fund, and the Foundation Fighting Blindness

2. AGING OF BRUCH’S MEMBRANE DECREASES RETINAL PIGMENT EPITHELIUM (RPE) PHAGOCYTOSIS Reiko Koyama, Hui Caiand Lucian V. Del Priore Department of Ophthalmology, Columbia University, NY, New York Purpose: The earliest changes in age-related macular degeneration occur within Bruch’s membrane. Bruch’s membrane aging affects the attachment and survival of the overlying RPE.1-3 Herein we determine the effects of Bruch’s membrane aging on RPE phagocytosis ability. p<0.02 Methods: Explants of human Bruch’s membrane were prepared as previously described. Donor ages(young eyes: ages 33 - 44; older group: 73 -94). 1-3 Native RPE were removed by bathing the choroid-BM-RPE complex with 0.02 N ammonium hydroxide followed by washing in PBS. The choroid-BM complex was set on polytetrafluoroethylene membrane with the basal lamina of the RPE facing the membrane. 4% agarose was poured onto the choroid -BM- complex from choroidal side and the tissue was kept at 4oC to solidify the agarose. The polytetrafluoroethylene membrane was peeled away and 6 mm circular buttons were trephined from choroid-BM-gel complex. Buttons were placed on 4% agarose at 37oC in non-treated polystyrene wells of a 96 well plate (Figure1). 50,000 immortalized ARPE-19 were seeded onto wells containing Bruch’s membrane explants and bare control wells (plastic only) for 72 hours(5 wells for each age group). 1ul of fluorescent latex beads (3.6x105beads/ul) were added to each well for another 24 hours. ARPE-19 were passaged by trypsinization. Ingested beads were counted using a FACS Flow Cytometer. Data were generated with at least three independent experiments. p>0.05 % of RPE ingesting beads % of RPE ingesting beads Fig.3 Percentage of RPE cells that ingest beads was higher for RPE cultured onto Bruch’s membrane explant than on bare plastic wells (22.22±11.59% vs. 10.04±3.52%). Fig.4 Percentage of RPE cells which have the ability to ingest beads on younger Bruch’s membrane versus older Bruch’s membrane were similar (25.74±15.94% vs. 22.22±11.59%). p<0.0003 p<0.008 Results: Flourescent intensity/cell Fluorescent intensity/cell A. on younger Bruch’s membrane B. on older Bruch’s membrane cells not ingesting beads cells not ingesting beads cells ingesting beads cells ingesting beads cell counts cell counts Fig. 5. Comparison of the phagocytosis ability of RPE cells cultured on young Bruch’s membrane explant versus older BM. Average fluorescent intensity per cell (measure of capacity of phagocytosis per cell) on Bruch’s membrane was higher than on plastic (291.61±80.91 vs. 167.50±35.01). Fig. 6 Observation of the population of cells that had ingested beads. The average fluorescent intensity per cell, which is a measure of capacity of phagocytosis per cell in the younger Bruch’s membrane group was higher than on older Bruch’s membrane (312.60±83.80 vs. 267±71.08) fluorescent intensity fluorescent intensity Fig.2. Flow cytometry histogram demonstrating the distribution of fluorescent intensity. RPE cell population which ingest fluorescent beads (green zone) in young BM group (A) is larger than older BM group (B). Fig 1. Photo shows 6-mm circular buttons which were trephined from BM and placed into 96 well plate Conclusions:This study suggests that Bruch’s membrane promote RPE phagocytosis compared to bare plastic tissue culture wells, and aging of Bruch’s membrane reduces the ability of RPE to ingest beads. To our knowledge, this is the first demonstration that aging of Bruch’s membrane can modulate RPE phagocytosis ability. Further study is required to determine the implications of this age-dependent decrease in RPE phagocytosis in the pathogenesis of AMD. • References: • 1. Tezel TH, Del Priore LV, Kaplan HJ. Fate of Human Retinal Pigment Epithelial Cells Seeded onto Layers of Human Bruch’s Membrane. Invest Ophthalmol Vis Sci 1999;40:467-476. • 2. Tezel TH, Del Priore LV. Repopulation of Different Layers of Host Human Bruch’s Membrane by Retinal Pigment Epithelial Cell Grafts. Invest Ophthalmol Vis Sci 1999;40:767-774. • 3. Del Priore LV, Tezel TH. Reattachment Rate of Human Retinal Pigment Epithelium to Layers of Human Bruch’s Membrane. Arch Ophthalmol 116;335-341, 1998. • Supported by Research to Prevent Blindness, Robert L. Burch III Fund, and the Foundation Fighting Blindness Supported by Research to Prevent Blindness, Robert L. Burch III Fund, and the Foundation Fighting Blindness.

3. Microarray qRT-PCR Probe Set ID Gene Name Symbol old/young P value old/young 160025_at transforming growth factor, alpha TGFA 1.7 0.0031 up 1933_g_at ATP-binding cassette, sub-family C (CFTR/MRP) ABCC5 0.5 0.0058 n/c 31886_at 5'-nucleotidase, ecto (CD73) NT5E 1.7 0.0057 33377_at vitronectin (serum spreading factor) VTN 0.4 0.0049 down 348_at kinesin family member C1 KIFC1 1.6 0.0074 35016_at CD74 antigen (invariant polypeptide of MHC) CD74 1.5 0.0072 36197_at catilage GP-39 protein Y08374 0.4 0.0016 n/c 36310_at keratin, hair, acidic, 1 KRTHA1 1.5 0.0041 36916_at sialyltransferase 4C SIAT4C 1.6 0.0066 36924_r_at secretogranin II (chromogranin C) SCG2 0.4 0.0059 n/c 37393_at hairy and enhancer of split 1 HES1 0.5 0.0106 38489_at heparin-binding growth factor binding protein HBP17 2.4 0.0036 n/c 38957_at doublecortin and CaM kinase-like 1 DCAMKL1 0.4 0.0010 down 39171_at catenin, beta interacting protein 1 CTNNBIP1 0.5 0.0041 n/c 39375_g_at G-2 and S-phase expressed 1 GTSE1 1.5 0.0091 39771_at Rho-related BTB domain containing 1 RHOBTB1 0.5 0.0002 down 40257_at Homo sapiens clone 24649 mRNA sequence Al400011 1.9 0.0093 40641_at BTAF1 RNA polymerase II BTAF1 1.7 0.0105 41119_f_at Homo sapiens, clone IMAGE:4310637 W27452 1.7 0.0088 0.0091 41479_s_at RAD51 homolog C RAD51C 1.5 BRUCH'S MEMBRANE AGING ALTERS THE RPE EXPRESSION PROFILE OF PROLIFERATION, MIGRATION AND APOPTOSIS BUT NOT ANGIOGENESIS GENES Hui Cai, Lucian V. Del Priore Department of Ophthalmology, Columbia University, New York, New York A. B. PROLIFERATION GENES OVERALL EXPRESSION Purpose: Principal component analysis (PCA) is a technique used to determine global changes in gene expression in response to changing cellular conditions. We have used PCA to determine the gene expression pattern of the retinal pigment epithelium (RPE) in response to age-related changes within Bruch’s membrane (BM). YOUNG BM OLD BM PC#2 22.0% PC#2 18.3% Methods: Immortalized human ARPE-19 cells were seeded onto human BM (five young samples: donor age = 31- 47 yr and five older samples: donor age = 71 – 81 years) harvested from human eye bank eyes. ARPE-19 were harvested 72 hours after seeding onto human BM and total RNA was isolated using a Qiagen RNeasy Mini Kit. First and second strand cDNAs synthesis, Biotin-labeled antisense cRNA Target hybridization, washing, staining and scanning probe arrays were done following an Affymetrix GeneChip Expression Analysis Manual. RPE gene expression profile was analyzed with Affymetrix Miroarray Suite 5.0, SAM and Genesis 1.3 software. PC#3 12.4% PC#3 10.7% PC#1 27.4% PC#1 47.2% C. D. MIGRATION GENES ANGIOGENESIS GENES PC#2 21.0% PC#2 18.4% Results: PC#3 11.3% PC#3 15.5% PC#1 35/2% PC#1 33.1% Figure 2. Principal component analysis of gene expression. (A) The pattern of gene expression of human RPE seeded onto Bruch’s membrane explants from younger donors (blue) shows a tighter clustering than the gene expression profile of RPE seeded onto older donors (red) Bruch’s membrane (B) Cell proliferation and migration genes of RPE seeded onto young Bruch’s membrane (blue) show a tight clustering with spread in the expression profile of proliferation-related genes of human RPE seeded onto older Bruch’s membrane explants (red). (C) There was a similar pattern for apoptosis-related and genes . (D) There is no age-dependent alteration in the spread in the expression profile of angiogenesis genes. Figure 1. The expression of approximately 6,000 genes (out of 12,600 genes on microarray Human 95UA chip) was detected. Scatter plot of gene expression within RPE cultured onto 31 year-old vs 38 year-old Bruch’s membrane. More than 96% of genes are expressed consistently among all samples tested within the young age group (data not shown). The correlation co-efficient is 0.989 suggesting limited variation between these individuals. Table I. 20 genes and EST’s with the lowest p-values. Microarray data suggests that aging of Bruch’s membrane increases the expression level of numerous genes. We performed RT-PCR on several genes of interest, including up regulated genes transforming growth factor alpha, CD74 antigen, and heparin-binding growth factor binding protein, and down regulated genes that include the ATP-binding cassette, vitronectin, cartilage GP-39 protein, doublecortin and CaM kinase-like 1, and catenin. RT-PCR confirms the up regulation of TGF alpha and the down regulation of vitronectin, doublecortin and CaM kinase-like 1, and Rho-related BTB domain containing 1. Conclusions: Age-related changes within BM alone induce significant spreading of the gene expression profile of proliferation, apoptosis and cell migration genes, with no change in angiogenesis genes. These observations suggest some of cellular changes that develop within the RPE as a function of age, such as occur in age-related macular degeneration, may be the result of substrate-induced alterations in the behavior of the overlying RPE. Supported by Research to Prevent Blindness, Robert L. Burch III Fund, and the Foundation Fighting Blindness

4. A. B. young older young older BM donor ages RPE EXPRESSION OF VITRONECTIN AND ITS RECEPTOR ARE DOWNREGULATED WITH AGING OF HUMAN BRUCH’S MEMBRANE L. V. Del Priore, H. Cai, and T. H. Tezel Department of Ophthalmology, Columbia University, New York, New York Purpose: Previous studies shows that vitronectin is a major constituent of ocular drusen and vitronectin mRNA is synthesized in RPE cells. The purpose of this study is to determine if Bruch’s membrane aging alters the level of vitronectin mRNA and its receptor in the overlying RPE. Figure 2. DNA microarray data show vitronectin (VN) and its receptor alphaV transcript expression patterns. ARPE –19 cells were cultured on different aged Bruch’s membrane explants for 72 hours. Data show VN and its receptor subunit alphaV expression level decreases in RPE cells overlaying on aged Bruch’s membrane. Methods: DNA microarray and semi-quantitative RTPCR method were used for this study. Immortalized human ARPE-19 cells were seeded onto human Bruch’s membrane (five samples from donors age < 50 yr and five samples from donors age > 70 yr) harvested from eye bank eyes. ARPE-19 cells were harvested 72 hours after seeding onto human Bruch’s membrane and total RNA was isolated using a Qiagen RNeasy Mini Kit. First and second strand cDNAs synthesis, Biotin-labeled antisense cRNA target hybridization (on Affymetrix U95A chip), washing, staining and scanning probe arrays were done following an Affymetrix GeneChip Expression Analysis Manual. Real-time quantitative polymerase chain reaction (Roche Lightcycler) using samples generated from different experiments with different donor tissues were used to confirm and further study the genes expression patterns. Oligo primers were determined with LC Primer Design and RTPCR data were analyzed with Lightcycler3 Data Analysis software. GADPH VN A. B. Results: Figure 3. Real time semi-quantitative RT-PCR. The Bruch’s membrane samples were from different batches of donors than those shown above. (A) Quantitative RT-PCR was performed , after establishing standard curves with GAPDH house-keeping gene using serial dilutions of total RNA. (B) Vitronectin expression level is decreased in RPE cells seeded onto aged Bruch’s membrane (dashed lines in duplicates). (C) vitronectin receptor alphaV subunit mRNA in RPE cells is also decreased upon culturing on aged BM (dashed lines). C. VN RECEPTOR Figure 1. A. Plot of individual DNA microarray data on vitronectin (in red color) and its receptor subunit alphaV (in blue color) transcript expression levels. Data show a general decreased expression level trends for both VN and its receptor mRNA in RPE cells seeded onto aged Bruch’s membrane. B. Heat map shows VN and its receptor expression levels (high level in red color and low level in green) in RPE cells cultured on BM explant from different donor ages. Conclusions: Aging of Bruch’s membrane downregulates the expression profile of vitronectin mRNA and its receptor in human RPE. The vitronectin receptor may play an important role in phagocytosis of photoreceptor outer segments and vitronectin partially mediates RPE attachment to human Bruch’s membrane. These observations suggest some of the changes seen in age-related macular degeneration may be the result of substrate-induced alterations in the behavior of the overlying RPE. Supported by Research to Prevent Blindness, Robert L. Burch III Fund, and the Foundation Fighting Blindness.

5. r1 r2 a1 a2 F in F out ANATOMICAL SIMILARITIES BETWEEN STAGE 3 AND STAGE 4 MACULAR HOLES: IMPLICATIONS FOR TREATMENT Jon Wender, Tomohiro Iida, M.D.,Lucian V. Del Priore, M.D., Ph.D.Department of Ophthalmology, Columbia University, New York, NY ABSTRACT Purpose: To determine whether the observed anatomy of macular holes can be explained by a hydrodynamic model in which fluid flow through the hole is balanced by fluid pumping across the RPE. We use this model to draw conclusions about the possible role of vitreomacular traction in determining the morphology of macular holes and their resolution after vitreous surgery. Design: Cross sectional. Methods: Retrospective study in a clinical practice. The study included 42 eyes of 42 patients, each with a stage 3 or 4 macular hole (Gass classification). Macular holes were staged on the basis of clinical exam, Optical Coherence Tomography, and intraoperative findings. We measured the radius of the macular hole and the radius of the surrounding cuff of subretinal fluid from color or red free fundus photographs, and determined the relationship between these 2 variables. Results: The mean age of the patients was 68.0  7 years old (range 51-80). 25 patients had stage 3 macular holes and 17 patients with stage 4 macular holes. The radius of the neurosensory detachment radius was related to the square of the macular hole radius for stage 3 and stage 4 holes, with no significant difference between the stage 3 and stage 4 linear trend lines (p=0.999). There was no correlation between patient age and the area of the macular hole (r = 0.0645) or neurosensory detachment plus hole (r=0.156) over the range of age in this study (51-80 years). However, the area of the doughnut-shaped cuff of subretinal fluid increased with increasing patient age (p = 0.0493), thus suggesting an age-dependent decline in the pumping ability of the RPE. Conclusions: Our data is consistent with a hydrodynamic model in which macular hole anatomy is determined by a balance between fluid flow through the hole and fluid outflow across the RPE. Since Stage 3 and 4 macular holes exhibit a similar relationship between the size of the macular hole and the size of the cuff of subretinal fluid around the hole, simple relief of vitreomacular traction would not lead to resolution of the subretinal fluid cuff unless it is accompanied by a reduction in hole diameter due to approximation of wound edges. Results • Mean age: 68.0  7 years (range 51-80) • 25 patients with stage 3 macular holes • 17 patients with stage 4 macular holes • The neurosensory detachment radius was related to the square of the macular hole radius for stage 3 and stage 4 holes, with no significant difference between the stage 3 and stage 4 linear trend lines (p=0.999). • Similarly, the neurosensory detachment area was related to the square of the macular hole area for stage 3 and stage 4 holes, with no significant difference between the stage 3 and stage 4 linear trend lines. Mathematical Model of Macular Holes Our model relies on the fact that there is flow of fluid from the vitreous cavity, across the intact retina and RPE, into the choroid in the normal human eye. The development of a full thickness defect in the neurosensory retina will allow fluid from the vitreous cavity to flow through the hole and detach the retina from the RPE (Fig. 1). Fluid flow through the hole will create an enlarging neurosensory detachment. The neurosensory detachment around the hole will increase in size and ultimately be limited by the pumping ability of the underlying RPE. In equilibrium (i.e., when there is no further enlargement of the hole), the fluid flow into the hole (Fin) and fluid flow out of the hole through the RPE (Fout) will be equal (Fig. 1). For a Newtonian fluid, the rate of fluid flow into the hole is limited by the size of the macular hole itself. Mathematically, fluid flow through the macular hole (Fin) is inversely proportional to the resistance (R) to fluid flow through the hole. Thus, we write: (Eq. 1) Fin  1/R The resistance is proportional to the square of the area of the macular hole, where the area of the macular hole is given by r12. Thus, (Eq. 2) Fin =  / 1/2r14 = 2r14, where  is an arbitrary constant. The flow out (Fout) of the subretinal space is due to active pumping of fluid by the underlying RPE. If we assume that the pumping ability of the RPE is homogeneous (i.e., does not vary across the area of the neurosensory detachment), then outward flow will be directly proportional to the area of the RPE under the macular hole and the surrounding neurosensory detachment. Thus, we write: (Eq. 3) Fout = Kr22 where r2 is the radius of the surrounding neurosensory detachment (Fig. 1). A priori, it is not known if K is a constant or varies with patient age. We note this explicitly by writing: (Eq. 4) Fout = K(age) r22 In this model, the subretinal fluid cuff will increase in size until enough RPE is exposed to allow the flow out to balance the fluid inflow through the hole. In equilibrium, (Eq. 5) Fin = Fout (Eq. 6) 2 r14 = Kr22 (Eq. 7) r14 = (K/) r22 (Eq. 8) r12 = (K/) r2 Thus, the hydrodynamic model predicts that r12 would be proportional to r2; i.e., as the radius of the macular hole doubles, the radius of the neurosensory detachment would quadruple. Figure 4. Subretinal fluid cuff area (a2-a1) vs. macular hole area (a1) for stage 3 and stage 4 macular holes. There appears to be no significant difference between the 2nd order polynomial trend lines for stage 3 (y = 10-05x2 - 0.7656x + 258038, R2 = 0.6752) vs. stage 4 (y = 10-05x2 - 1.8772x + 426462, R2 = 0.8225) macular holes. Figure 2. Relationship between neurosensory detachment radius (r2) and macular hole radius squared (r12). Data fit to a linear regression model, with no significant difference between the stage 3 (y = 0.0042x + 220.04, R2 = 0.8098) and stage 4 (y = 0.0042x + 243.61, R2 = 0.8046) linear trend lines (p=0.999). FIGURE 1. (Top) Schematic diagram of fluid dynamics of macular hole with surrounding neurosensory detachment. r1 = radius of macular hole, r2 = radius of subretinal fluid cuff. (Bottom) Diagram illustrating fluid dynamics for macular holes. Fin and Fout represent the fluid flow into and out of the subretinal space, respectively. For a Newtonian fluid, Fin is inversely proportional to the resistance. Fout is proportional to the area of the underlying RPE. Figure 3. Relationship between neurosensory detachment area and macular hole area squared. Note that the data now fits a linear regression model with no significant difference noted between stage 3 (y = 10-05 + 275658, R2 = 0.7677) and stage 4 (y = 10-05x + 337788, R2 = 0.8691) linear trend lines (p=0.904). Conclusions Our data is consistent with a hydrodynamic model of macular hole anatomy in which fluid flow through the hole is balanced by the outflow of fluid across the RPE. Since Stage 3 and 4 macular holes exhibit a similar relationship between the size of the macular hole and the size of the cuff of subretinal fluid around the hole, simple relief of vitreomacular traction would not lead to resolution of the subretinal fluid cuff unless it is accompanied by a reduction in hole diameter due to approximation of wound edges. Methods Retrospective study in a clinical practice. The study included 42 eyes of 42 patients, each with a stage 3 or 4 macular hole (Gass classification). Macular holes were staged on the basis of clinical exam, Optical Coherence Tomography, and intraoperative findings. We measured the radius of the macular hole and the radius of the surrounding cuff of subretinal fluid from color or red free fundus photographs, and determined the relationship between these 2 variables.