Basement Membrane-Dependent Modification of Phenotype and Gene Expression in Human Retinal Pigment Epithelial ARPE-19 Cells

Abstract

purpose. To use porcine lens capsule (PLC) as basement membrane for ARPE-19 cells and to characterize its effects on cell differentiation and gene expression.

methods. Postconfluent cultures of ARPE-19 cells were established on either porous polyester filters or PLC membranes and characterized by electron microscopy, immunocytochemistry, and transepithelial electrical resistance measurements. Metabolic activity was assessed by measuring phagocytosis of rod outer segments. mRNA populations of ARPE-19 cells grown on polyester and PLC membranes were compared by suppressive subtractive hybridization. Differentially regulated messages were subsequently identified by DNA sequencing and their altered expression confirmed by Northern or virtual Northern blot analysis.

results. Culture of ARPE-19 cells on PLC membrane induced the formation of apical microvilli and the ability to phagocytose rod outer segments. These culture conditions also led to enhanced junctional distribution of ZO-1 and occludin, the formation of polarized membrane domains, and a significant increase in transepithelial resistance. Gene expression was significantly altered by growth on PLC membranes and 29 differentially expressed transcripts were identified.

conclusions. Culture of ARPE-19 cells on PLC membranes resulted in a more differentiated phenotype and in expression of a specific set of transcripts encoding protein products that may affect epithelial differentiation, polarity and survival.

Retinal pigment epithelial (RPE) cells perform a crucial and multifunctional role in retinal photoreceptor homeostasis and in maintaining the overall health of the retina. Situated between the neural retina and the vascular choriocapillaris, the RPE constitutes the posterior part of the blood–retinal barrier which is instrumental in controlling the passage of molecules and cells to and from the retinal parenchyma. Altered function of this critical tissue, whether through inherited or acquired ocular disease, can have a devastating effect on vision. Because of the importance of this structure, the unique properties of RPE cells have attracted considerable attention, particularly as RPE cells are likely to play a central role in the pathogenesis of various retinal diseases. Although much has been learned from in vivo studies, the limitations imposed by whole animal experimentation have led to the development of techniques to isolate and culture adult RPE cells from many different species. Despite this, mammalian RPE culture models generally fail to preserve many of the phenotypic characteristics which are exhibited by RPE in vivo. Such properties include the ability of RPE cells to produce functional tight junctions, to form apical and basolateral membrane domains through cell polarization, and to phagocytose rod outer segments effectively. One of the most promising human models is currently based on the long-term growth of primary RPE in a specialized media, ^{1 2} although the precise mechanism by which extended culture and soluble media supplements lead to a specialized RPE phenotype is unclear. 

In an effort to simplify in vitro RPE cell biology experiments, various immortalized cell lines have been created from several species, including rat ^{3 4} and human. ^{5 6} These cell lines retain many, but far from all, in vivo characteristics of RPE and it is unclear whether they are irreversibly de-differentiated, or still capable of re-differentiating when presented with appropriate cues. After transplantation of immortalized human RPE cells into the Royal College of Surgeons (RCS) rat, the grafted cells re-pigment with a characteristic apical distribution of pigment granules. ⁷ Such observations strongly suggest that even immortalized RPE cells can re-differentiate with respect to generation of pigment granules when placed within the retinal environment and are capable of effective cell polarization. It has also been reported that only those grafted RPE cells that had contacted Bruch’s membrane survived and were able to re-differentiate. This is not unexpected as RPE cells are anchorage-dependent cells and require engagement of cell surface integrins within the extracellular matrix to prevent apoptosis (anoikis). ⁸ Moreover, it is highly probable that cells also require a specific combination of integrin signaling along with soluble signals to remain fully differentiated. Indeed, it has recently been shown that molecular integrity of the basement membrane is necessary for RPE function in vivo. ⁹ A number of previous studies have used various extracellular matrix support media to determine alterations in RPE phenotype associated with these matrices. However, these studies have been limited largely to simple morphometric analyses. ^{10 11 12} The studies presented here demonstrated that the immortal human RPE cell line, ARPE-19, given an appropriate extracellular matrix in vitro, is capable of developing phenotypic characteristics associated with human RPE in vivo. Specifically, ARPE-19 cells are capable of displaying a more differentiated phenotype than seen with most other RPE in vitro models, and show alterations in gene expression consistent with the development of a differentiated RPE phenotype. These studies have been driven by the need to understand key extracellular matrix cues and to find a suitable extracellular matrix that will allow successful RPE transplantation strategies to be developed for the treatment of age-related maculopathy. 

Materials and Methods

Preparation of Porcine Lens Capsule Membrane Tissue Culture Inserts

Porcine lens capsules (PLCs) were obtained from fresh porcine eyes (Fresh Tissue Supplies, Ltd., West Sussex, UK). After surgical exposure of the anterior surface of the lens, a vectis loop was used to disinsert the zonules. The anterior capsule was cut circumferentially along the lens equator. Transwell polyester tissue culture filters (6 mm diameter; Costar, Cambridge, MA) were pre-prepared by removing the original membrane and coating the bottom edge with cyanoacrylate glue. The support was then gently glued onto the central anterior lens capsule. Once a complete and tight seal was established between lens capsule and Transwell support, the remainder of the lens was gently teased away from the capsule. The prepared PLC membrane inserts were then washed thoroughly with Ca/Mg-free Hanks Balanced Salt Solution (HBSS) in the presence of 0.5 mg/mL streptomycin and 500 U/mL penicillin (Sigma Chemical Company Ltd., Poole, UK) and left for 24 hours in a sterile environment. PLC membranes were then washed and incubated in Dulbecco’s Modified Eagle’s Medium (DMEM)-F12 growth medium for 48 hours to check sterility before seeding of ARPE-19 cells. 

ARPE-19 Cell Culture

ARPE-19 (ATCC CRL-2302), a spontaneously immortalized human RPE cell line, ⁵ was cultured at 37°C under 5% CO₂ in a 1:1 mixture of DMEM and Nutrient Mixture F12 Medium supplemented with 10% fetal calf serum (FCS). ARPE-19 cells were seeded at a density of 1.5 × 10⁵ cells/cm² onto PLC membrane inserts or unmodified 6 mm diameter Transwell polyester filters (0.4 μm pore size) and allowed to grow to confluence. At this point they were cultured for a further 6 weeks and then analyzed. 

Immunocytochemistry and Confocal Microscopy

ARPE-19 cell monolayers, grown for 6 weeks postconfluence on either PLC membrane inserts or on Transwell polyester filters, were fixed in 3.7% paraformaldeyde in PBS and extracted with 0.2% Triton X-100. Alternatively cells were fixed and extracted using methanol (−20°C). After washing (PBS three times for 5 minutes) and blocking (10% FCS or goat serum in PBS), cells were reacted with primary antibodies against ZO-1, occludin (both from Zymed, San Francisco, CA), Na/K-ATPase β (Abcam, Cambridge, UK) or monocarboxylate transporter 3 (MCT3, generously provided by Nancy Philp, Thomas Jefferson University, Philadelphia) at room temperature for 1 to 2 hours. The cells were washed and blocked before incubation with matching Cy3-, or FITC-conjugated goat secondary antibodies (Jackson Immuno-Research Laboratories, West Grove, PA). Finally, immunostained preparations were mounted using Vectashield (Vector Laboratories, Burlingame, CA) and examined by confocal laser-scanning microscopy on an Axioplan 2 microscope equipped with LSM 510 (Carl Zeiss, Inc., Thornwood, NY). Series of overlapping 0.8 μm sections spanning the entire cell thickness were recorded and projections generated as described in the figure legends. 

Transepithelial Resistance Measurement

ARPE-19 cells were grown to confluence on PLC membrane inserts and Transwell polyester filters. Subsequently changes in transepithelial resistance (TER) across the monolayers were monitored over a 6-week period using STX-2 chopstick electrodes connected to an EVOM epithelial voltohmmeter (World Precision Instruments, Herts, UK). Net TER values were calculated by subtracting the mean resistance determined for ten PLC culture cups or plastic filters in the absence of ARPE-19 cells, from the value recorded for each monolayer grown on PLC membrane inserts or plastic filter, respectively. Final resistance-area products (in Ω.cm²) were obtained by multiplication with the effective growth area. 

Electron Microscopy

ARPE-19 cells were grown on either PLC membranes or polyester filters for 6 weeks postconfluence. Monolayers were fixed in 2.5% glutaraldehyde, 0.1 M sodium cacodylate, 1 mg/mL CaCl₂, pH 7.4, for 1 hour, processed using standard methods, and finally embedded in Araldite resin. Thick (1 μm) sections were cut and stained with toluidine blue for screening by light microscopy. Thin sections of regions of interest were cut on a Reichert OM4 microtome (Leica) and counterstained with saturated uranyl acetate and 0.3% lead citrate before viewing on a Jeol electron microscope (Model 1200, Ex II; Welwyn Garden City, UK). Specimens for scanning electron microscopy were additionally critical-point dried, sputter coated with gold, and viewed on a Jeol scanning electron microscope. 

Phagocytosis Assay

Porcine rod outer segments (ROS) were obtained from porcine retinas as previously described. ¹³ Isolated ROS were stored at a concentration of ca. 2 × 10⁶ ROS/mL at −70°C in 10 mM Na-phosphate, pH 7.2, 0.1 M NaCl, and 2.5% sucrose. Just before use, ROS were thawed and labeled with Alexa fluor 555 carboxylic acid succinimidyl ester (Molecular Probes, Eugene, OR) according to the manufacturer’s instructions. ROS were then washed and resuspended in cell culture medium before addition to ARPE-19 monolayers. The Alexa fluor 555-labeled ROS were added (4 × 10⁴/mL) to 6 weeks postconfluent monolayers of ARPE-19 cells cultured on either PLC membrane inserts or Transwell polyester filters. After a 3-hour incubation period at 37°C under 5% CO₂, the cultures were gently washed three times with prewarmed PBS and fixed by the addition of 2% paraformaldehyde for 5 minutes. The RPE cell membranes were then stained with biotin and fluorescein-labeled streptavidin before evaluation of ROS phagocytosis by confocal microscopy using a Radiance 2000 (BioRad, Hercules, CA) mounted on a Axiomat microscope (Carl Zeiss, Inc.). Sections were taken through the plane of the cell nuclei to ensure recording of internalized (i.e., phagocytosed) ROS rather than adherent material. The amount of phagocytosed ROS was quantified as the integrated optical density of captured 8-bit images using Metamorph (Universal Imaging Corp., Downingtown, PA). Objects were thresholded based on size and intensity to include fully internalized ROS (green), but to exclude pinosomes (red) and partially internalized or highly aggregated ROS (white). Mean values were calculated from ten captured low-power fields of view. Finally, phagocytic activity per cell was determined by dividing total phagocytosis per field of view by the number of DAPI stained nuclei. Cell density of cells grown on porous polyester or PLC membrane after 6 weeks was not significantly different. 

Suppression Subtractive Hybridization and cDNA Library Construction

mRNA was isolated from ARPE-19 cells grown for 6 weeks on either PLC or on polyester inserts using a QuickPrep Micro mRNA Purification Kit (Amersham Biosciences, Herts, UK) according to the manufacturer’s instructions. cDNA synthesis and suppression subtractive hybridization (SSH) ¹⁴ were carried out using a PCR-Select cDNA Subtraction Kit (Clontech, Palo Alto, CA). Briefly, ds cDNA was prepared from ca. 2 μg of poly(A)⁺ RNA isolated from ARPE-19 cells grown on either PLC membranes or polyester filters, and digested with RsaI. Subsequently, lens capsule-derived cDNA was divided into two portions, and each ligated with a different DNA adaptor. Each of the resulting preparations was then hybridized with an excess of cDNA derived from cells grown on polyester (forward subtraction). The two hybridization samples were combined and allowed to re-hybridize without prior denaturation. Subsequently, two PCR amplification steps were performed using adaptor specific primers which resulted in exponential amplification of cDNAs with different adaptor ends (i.e., differentially expressed sequences). Next, the amplified products were directly subcloned into the A/T cloning vector pT-Adv (Clontech). The resulting subtracted cDNA library was transformed into bacteria and propagated on selective IPTG/X-gal LB plates. 

Differential Screening of Subtracted cDNA Library Arrays

Two thousand random cDNA inserts from the subtracted library were PCR-amplified using pT-Adv-specific primers (Clontech). Four replica arrays were generated by dot-blotting amplified inserts onto Hybond XL membranes (Amersham Biosciences). 

Unsubtracted cDNA was prepared by reverse-transcribing ca. 500 ng of RNA isolated from ARPE-19 cells grown on PLC membranes or polyester filters. Forward- and reverse-subtracted (i.e., polyester cDNA hybridized to an excess of PLC cDNA) cDNAs were generated from the diluted products of the primary PCR described above. All four cDNA preparations were labeled with [α-³²P]-dCTP using standard techniques. ¹⁵ A replica of each dot blot array was then hybridized to either of the two unsubtracted, the forward- or the reverse-subtracted radio-labeled cDNA probes. After hybridization and washing, membrane arrays were exposed to x-ray film overnight at −70°C with an intensifying screen. 

DNA Sequencing

Clones found to hybridize specifically with the unsubtracted, PLC-derived and the forward-subtracted probe were subjected to DNA sequencing using an ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer Corp., Foster City, CA) and an ABI 310 sequencer. DNA sequences were analyzed using the University of Wisconsin GCG software package (Accelrys, Cambridge, UK). Database searches were performed using the Basic Alignment Search Tool (BLAST, Version 2.0) algorithm. ¹⁶  

Northern and Virtual Northern Blot Analysis

Approximately 2 μg of poly(A)⁺ RNA isolated from ARPE-19 cells grown on either PLC membranes or polyester filters was resolved on a formaldehyde-containing 1.2% agarose gels and subsequently transferred onto Hybond N⁺ (Amersham Biosciences). 

For virtual Northern blot analysis, cDNA was prepared using a SMART cDNA synthesis kit according to the manufacturer’s instructions (Clontech). Briefly, ca. 1 μg of total RNA isolated from ARPE-19 cells grown on either PLC membranes or polyester filters was subjected to first strand and second strand synthesis using SMART oligonucleotides. ¹⁷ The double-stranded cDNA was amplified by 15, 18, 21, or 24 PCR cycles. Amplification reactions were analyzed on 1% agarose gels and 18 cycles considered to be within the linear range of amplification for cDNAs up to 3 kb. Subsequently, one-tenth (corresponding to 0.1 μg total RNA input) of 18 cycle PCR reactions were resolved on agarose gels and Southern blotted. 

All Northern and virtual Northern blot analysis were then hybridized to various [α-³²P]-dCTP-labeled DNA probes and autoradiographed using standard techniques. ¹⁵  

Results

Improved Morphology and Enhanced Tight Junction Distribution in ARPE-19 Cultured on PLC Membranes

ARPE-19 cells were grown on either uncoated Transwell polyester filters or PLC membranes for 6 weeks. On both supports ARPE-19 displayed a cobblestone morphology typically found for epithelial cells in culture. However, when grown on PLC membranes, their morphology appeared more uniform and hexagonal (data not shown). Subsequent ultrastructural analysis using electron microscopy revealed close attachment to both underlying substrates. Only limited evidence of apical microvilli was found in ARPE-19 grown on polyester filters (data not shown). However, the apical membrane of ARPE-19 cells grown on PLC membranes was highly organized with clear ultrastructural evidence of a dense microvillar network (Figs. 1A and 1B) . Moreover, these ultrastructural analyses also showed that ARPE-19 cells grown on uncoated polyester filters formed multilayers (Fig. 1C) whereas cells grown on PLC membranes appeared to form perfect monolayers (Fig. 1D) . 

Tight junction (TJ) protein distribution was determined by immuncytochemistry. In ARPE-19 cells grown on either membrane, distribution of the TJ-associated protein ZO-1 was, as expected, mostly restricted to cortical areas delimiting the cell edges (Figs. 2A and 2B) . Whereas occasional discontinuities of ZO-1 staining were observed in cells grown on polyester, the staining was perfectly uninterrupted and also markedly stronger in cells grown on PLC. Significantly stronger differences were found for the distribution of another TJ component, namely the transmembrane protein occludin (Figs. 2C and 2D) . In ARPE-19 cells grown on polyester filters, occludin distribution was poor (Fig. 2C) . The majority of staining was associated with cell nuclei and only sporadic and punctate staining was observed at the points of cell–cell contact. In contrast, occludin was no longer nuclear but lined the cellular junctions in a highly continuous fashion in ARPE-19 cells grown on PLC membranes (Fig. 2D) . Such enhanced distribution of these two TJ proteins suggested that growth on PLC membranes promoted the formation of functional cell-cell junctions. 

Enhanced Polarity and Barrier Function of ARPE-19 Grown on PLC Membranes

The presence of functional intercellular junctions in epithelial cells is a prerequisite for the polarization of their membrane domains and the formation of an effective cellular barrier with regard to transepithelial permeability or electrical resistance. 

To assess ARPE-19 polarization, the localization of the Na/K-ATPase and the monocarboxylate transporter 3 (MCT3) was determined by indirect immunocytochemistry followed by confocal microscopy (Fig. 3) . Consistent with an absence of functional intercellular junctions both proteins were found in apical and basolateral areas of ARPE-19 cells grown on polyester inserts (Fig. 3A) . However, growth on PLC membrane led to a strikingly different distribution: Na/K-ATPase was found almost exclusively in apical sections of the cells, whereas MCT3 appeared to be restricted to basal areas (Fig. 3B) . 

ARPE-19 barrier function was examined by measuring TER. In ARPE-19 cells grown on uncoated Transwell polyester filter, the TER did not rise above 50 Ω.cm² over a 6-week culture period (Fig. 4) . However, the TER of ARPE-19 cells grown on PLC, increased steadily over time and typically reached a plateau with values in excess of 200 Ω.cm² after 4 weeks of culture (Fig. 4) . Taken together, the observations of enhanced distribution of TJ protein as well as increases in membrane polarization and TER indicated that growth on PLC membranes promoted the formation of functional intercellular junctions in ARPE-19 cells. 

Induction of Phagocytosis in ARPE-19 Grown on PLC Membranes

Phagocytic activity of ARPE-19 was measured by fluorescently-labeled ROS. Little evidence of internalization of ROS was found in ARPE-19 cells grown on polyester filters (Fig. 5A , left panel). In marked contrast high amounts of internalized ROS were detected in ARPE-19 grown on PLC membranes. Significantly, the internalized fluorescence was found in ultrastructures resembling phagosomes (Fig. 5A , right panel). When quantified by fluorescence density analysis, ROS phagocytosis was found to be increased by almost two orders of magnitude in ARPE-19 cells grown on PLC (Fig. 5B) . 

Differential Gene Expression in ARPE-19 Grown on PLC Membranes

mRNA from ARPE-19 grown on polyester inserts and PLC membranes were compared by suppression subtractive hybridization PCR. From this, a cDNA library enriched in differentially expressed genes was established as described in the Materials and Methods section. cDNA inserts from 2000 independent library clones were screened with cDNA probes derived from ARPE-19 cells cultured either on polyester filters (Probe A) or PLC membranes (Probe B). Figure 6A shows a representative replica library array hybridized to these two cDNA probes. Only clones which exhibited an at least threefold change in hybridization to the two different probes were considered positive, that is, differentially expressed (example arrowed). From the 2000 clones screened, 35 showed such signal changes. Notably, 34 of these clones were upregulated in PLC-grown ARPE-19 cells, whereas only a single clone (clone 11) was downregulated. All clones considered differentially expressed were sequenced and identified bioinformatically using BLAST ¹⁶ searches of GenBank (www.ucbi.nlm.nih.gov/Genebank) (Table 1) . Twenty-nine distinct transcripts were identified, of which one (CGI-144) and four (annexin 4, Par 1, EBP50, and GRIM-19) were represented three and two times, respectively. 

To determine that the gene products identified by suppression subtractive hybridization displayed altered transcript levels in ARPE-19 after growth on PLC membrane, Northern blot analysis was carried out. Transcripts of the predicted size for annexin 4, intracellular chloride channel 1, and EBP50 were found in ARPE-19 grown on both supports. Significantly, however, all three transcripts were found strongly upregulated in ARPE-19 cells grown on PLC membrane (Fig. 6B) . Due to the low availability of RNA from ARPE-19 cells grown on PLC, all further transcript analysis was conducted by virtual Northern blot, which allows comparison of transcript levels almost as accurately as Northern blot analysis but relies on smaller amounts of RNA. ¹⁴ Using this method, many of the 29 identified transcripts with predicted sizes of up to 3 kb were successfully analyzed. Representative examples of such blots are shown in Figure 6C . As summarized in Table 1 , the transcript levels of all clones analyzed so far were significantly increased in ARPE-19 cells grown on PLC. 

Discussion

Many investigations have highlighted the important role of the RPE in supporting neuroretinal function. To fulfill this specific requirement the RPE cells must adopt a tight epithelial phenotype that is able to regulate the passage of solutes and fluid to and from the retina. In this regard the RPE layer and underlying Bruch’s membrane function as the posterior blood–retinal barrier. In a similar manner to other epithelial cells which act as cellular barriers (e.g., choroid plexus epithelia), these cells require the expression and correct spatial organization of a variety of proteins which form complex intercellular junctions. In vivo, RPE cells are supported by Bruch’s membrane which appears capable of inducing these features within RPE cells. However, in vitro these crucial characteristics are frequently lost. To address this problem, a number of previous studies have used other acellular biological membranes, such as both bovine and porcine lens capsule membrane, ¹¹ amniotic membrane, ¹² as well as complex extracellular matrix proteins ^{11 18} to assess whether these supports can mimic Bruch’s membrane in inducing a tight epithelial phenotype in an in vitro environment. In the present study, the lens capsule membrane was chosen to act as basement membrane for ARPE-19 in vitro. The rationale for using lens capsule is that, on the one hand, it constitutes a readily available physiological basement membrane, and on the other hand, it resembles Bruch’s membrane in its high portion of collagen IV. ^{19 20} Consistent with findings of another study, ¹¹ the immortal RPE cell line, ARPE-19, adopted a more authentic morphologic phenotype when cultured on PLC membrane. This observation is also apparent when RPE cells are cultured on other acellular membrane substrates such as amniotic membrane ¹² and Descemet’s membrane. ¹⁰ It is interesting to note that under certain conditions, synthetic membranes are also capable of inducing an improved RPE morphology and the correct spatial appearance of intercellular junctions. ²¹ Although other studies have suggested the presence of functional intercellular junctions based on the correct distribution of junctional proteins, ²² in the present study, the culture of ARPE-19 cells on PLC membrane was not only associated with a dramatic reorganization of junctional proteins such as occludin and ZO-1, but it also led to membrane polarization and a significant increase in TER. Membrane polarization was visualized by immunocytochemical localization of Na/K-ATPase and MCT3. Na/K-ATPase is a bona fide marker of apical membranes in RPE ²³ and was found exclusively in the apical membrane domains of ARPE-19 grown on PLC. Conversely, MCT3 localizes to basal membranes of RPE in vivo. ²⁴ Interestingly, MCT3 is virtually absent in cultured RPE with modest expression only occurring in long-term cultures of at least 30 days. In our (6-week) ARPE-19 cultures, MCT3 was also detected and that with a predominant basal localization. To our knowledge, the present work is the first to report polarized localization of MCT3 in a RPE culture model. Epithelial polarization and increased TER are not only indicative of the presence of intercellular junctional proteins but more specifically of the presence of functional tight junctions. These findings concur with others recently reported for RPE cultures derived from the embryonic chick. ²⁵  

Microvilli are one of the apical hallmark structures in polarized epithelia. In agreement with all other observations, RPE grown on PLC displayed a well-developed microvillar network. Induction of microvilli has also been observed after culture of RPE on acellular membranes ^{10 11} and indicates that these cells polarize under such conditions. Microvilli normally interdigitate with the outer segments of photoreceptors and mediate the continuous phagocytosis of shed outer segments. Indeed, rodents in which RPE cells are unable to undertake this function effectively, develop retinal dystrophic disease. ²⁶ ARPE-19 cells cultured on polyester filters internalized ROS with barely discernable activity. When grown on PLC membrane they showed an at least 100-fold increase in their ability to phagocytose ROS, which is consistent with the appearance of apical microvilli. Ezrin is a morphogenic determinant of microvilli in RPE cells. ²⁷ Its major interacting partner in polarized epithelia, EBP50, was found significantly upregulated in ARPE-19 grown on PLC membranes (see also below), suggesting that the formation of microvilli in cultures grown on PLC membrane is due to changes in gene expression. 

In ARPE-19 grown on PLC membranes, 28 transcripts were identified to be selectively upregulated whereas a single one was found to be downregulated. It was notable that 13 positive library clones represented only 6 genes. CGI-144 protein, a well-conserved protein of unknown function was represented in three independent clones out of 2000 screened, while annexin 4, cell death regulatory protein GRIM19, the thrombin receptor Par1, and EBP50 were each represented twice. These observations suggested to us that these mRNAs were strongly induced after growth on PLC membrane and provided evidence that the suppressive subtractive PCR procedure yielded a significantly enriched cDNA library. This was further substantiated by additional Northern or virtual Northern blot analyses. 

A number of upregulated sequences identified in ARPE-19 cells grown on PLC membrane are characteristic of differentiated epithelia and implicated in both the maintenance of epithelial phenotype and polarity. Notable were the upregulation of cytokeratin 18, epimorphin, EBP50, FLJ20811, and TGFBI. Cytokeratin 18 is an intermediate filament protein and, together with cytokeratin 8, a specific marker of single layer epithelia. Epimorphin (syntaxin 2) is critically involved in epithelial morphogenesis. ²⁸ EBP50 was identified by two independent clones and is an ezrin/radixin/moesin(ERM)–binding, PDZ domain-containing protein. It is specifically found at the apical face of polarized epithelia and is thought to be critical for various apical activities, such as microvillar organization, receptor localization, and endocytosis. ²⁹ The armadillo-repeat containing protein FLJ20811 (or BAA91387) bears strong homology to ALEX (Armadillo proteins lost in epithelial cancers on chromosome X) proteins and will presumably turn out to belong to this family of proteins. Expression of all bona fide members of this family (ALEX 1 to 3) are lost in a vast number of carcinomas but not in tumors derived from nonepithelial tissues. ³⁰ Together with our observation of its upregulation in more differentiated ARPE-19, it is tempting to speculate that ALEX proteins may be involved in maintaining nonproliferative epithelial differentiation. Finally, TGFBI, the TGFβ-induced gene 3, is a secreted protein that modulates integrin-based matrix adhesion ³¹ and may therefore be important in mediating the effects of the PLC matrix within ARPE-19 cells. 

It is also interesting to speculate on the roles of other genes identified from the library array in RPE physiology. As discussed above, a feature of differentiated RPE is a high metabolic activity, which particularly manifests itself in the phagocytosis of outer segments. Among the 29 differentially expressed transcripts, aminoacyl-peptide hydrolase and the proteasome subunits PSMB7 and Skp1 are implicated in hydrolysis and proteolysis of proteins, while aflatoxin aldehyde reductase and malate dehydrogenase are metabolic enzymes. Their upregulation may reflect the increased metabolic rate of PLC-grown ARPE-19. 

Three differentially-expressed transcripts identified in our gene array analysis suggest that decisive switches in RPE cell signaling are made during growth on PLC. The phosphoinositide-3-kinase-related protein kinase mTOR (mammalian target of rapamycin) is a much-studied key signaling component, inducing switches in the translation of specific sets of mRNAs in response to nutrient availability. It crucially controls selective protein synthesis and turnover during cytokine-dependent lymphocyte proliferation, as well as many other mitogen- or nutrient-induced cellular responses. ³² GRIM-19, represented by two independent array clones, has been identified as a modulator of the cell death response. It appears to act by specifically suppressing the activity of the STAT-3 transcription factor. ³³ Moreover, IGFBP-3, the only transcript found downregulated in PLC-grown ARPE19, is an extracellular IGF scavenger and its expression is associated with a strong pro-apoptotic activity. ³⁴ The simultaneous overexpression of GRIM-19 and downregulation of IGFBP-3 suggests that growth on PLC may also favor a gene expression profile that protects RPE cells from programmed cell death. 

The identification of annexins 2 and 4 is consistent with many studies in which upregulation of annexins accompanies terminal differentiation in various cell types. ³⁵ Annexin 4 is a regulator of Ca-activated chloride channels, and may therefore be involved in the homeostatic function of RPE cells. ³⁶ Annexin 2 is implicated in various forms of inward vesicle trafficking ³⁷ and may form part of the phagocytic machinery in RPE cells. Elevated expression of annexin 2 in differentiated RPE cells is consistent with immunocytochemical analysis of normal human retinal sections, ³⁸ but contrasts with a recent study in which annexin 2 was more highly expressed in de-differentiated RPE cells. ³⁹  

It will be intriguing to analyze further the functions of RPE cell proteins whose mRNA is induced after culture on PLC membrane, such as CGI-144 which is the product of a gene highly conserved between Neurospora, Arabidopsis, Caenorhabditis elegans, and humans, but whose function is currently unknown. Although all these genes cannot yet be functionally linked to the generation or maintenance of a polarized epithelial phenotype, this merely reflects our limited knowledge of the precise mechanism of epithelial cell polarization and survival. 

The ability of PLC membranes to induce an enhanced RPE phenotype underlines the importance of the extracellular matrix in maintaining the functionality of the RPE. It has long been postulated that RPE dysfunction in diseases such as age-related maculopathy may be due to aging of Bruch’s membrane. Indeed, mice lacking the basement membrane component collagen XVIII have abnormal RPE and age-dependent loss of vision. ⁹ In addition, in vitro studies of RPE attachment to severely damaged Bruch’s membrane in both human and rabbits have concluded that a suitable extracellular matrix support is essential for RPE survivability and function. ^{40 41} It is therefore likely that successful implementation of RPE transplantation strategies will be dependent on providing a matrix support which is efficacious in inducing a polarized RPE phenotype. 

²  These authors contributed equally to the work presented here and should therefore be regarded as equivalent senior authors.

 Supported by Moorfields Eye Hospital Trust and Fight for Sight. JS was the recipient of a June Sutor Research Fellowship, Moorfields Eye Hospital Trust.

 Submitted for publication August 29, 2003; revised February 7, 2004; accepted March 19, 2004.

 Disclosure: P. Turowski, None; P. Adamson, None; J. Sathia, None; J.J. Zhang, None; S.E. Moss, None; G.W. Aylward, None; M.J. Hayes, None; N. Kanuga, None; J. Greenwood, 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.

 

Figure 1.

 

Improved ultrastructural morphology of ARPE-19 cells after 6-week culture on PLC membrane. (A) Scanning electron micrograph of ARPE-19 cells grown on PLC membrane showing apical microvilli. (B) Transmission electron micrograph of ARPE-19 cells grown on PLC membrane exhibiting numerous apical microvilli. (C) Transmission electron micrograph of a cross-section of ARPE-19 cells grown on Transwell polyester filter showing propensity to form multilayers. (D) ARPE-19 cells on PLC membrane growing as a monolayer. Bars: (A) 8 μm; (B) 0.5 μm; (C, D) 2 μm.

Figure 2.

 

Enhanced junctional distribution of ZO-1 and occludin in ARPE-19 grown on PLC membrane. Immunocytochemical localization of ZO-1 (A, B) and occludin (C, D) in ARPE-19 cells grown for 6 weeks on Transwell polyester filters (A, C) or PLC membranes (B, D). Micrographs represent full image projections of confocal sections spanning the whole thickness (ca. 8 μm) of the monolayer. Scale bar: 20 μm.

Figure 3.

 

Polarized localization of Na/K-ATPase and MCT-3 in ARPE-19 grown on PLC membranes. ARPE-19 cells were grown for 6 weeks on Transwell polyester filters (A) or PLC membranes (B). Cells were fixed and co-stained for Na/K-ATPase, ZO-1, and DNA (left panels), or for MCT3 and DNA (right panels). Preparations were analyzed under a confocal microscope. Series of overlapping sections spanning the whole thickness of the cells (8–8.5 μm) were recorded. The upper and lower delimitation of the cells were then determined by virtue of the nuclear DNA staining. Finally, sections within the top and bottom 2 μm of the cells were projected into single images representing the apical and basal areas, respectively. Scale bars: 20 μm.

Figure 4.

 

ARPE-19 cells show increased transepithelial electrical resistance when cultured on PLC membranes. ARPE-19 cells were grown on either Transwell polyester filters (•) or PLC membranes (○) for up to 6 weeks. Transepithelial electrical resistance (TER) was measured at weekly intervals. TER values are expressed as mean ± SEM of a minimum of three independent experiments (each consisting of at least ten tissue culture cups).

Figure 5.

 

ARPE-19 cells show a quantitative increase in rod outer segment phagocytosis after culture on PLC membranes. ARPE-19 cells were grown to confluence on either Transwell polyester filters or PLC membranes and maintained at confluence for a further 6 weeks. Cells were cultured in the presence of Alexa Fluor 555-labeled rod outer segments (ROS) for 3 hours, fixed, and the ROS internalization monitored using laser scanning confocal microscopy. Sections through the middle of the cells were recorded and density maps of the fluorescence established by integrative density analysis. (A) Shown are pseudocolored density maps of representative images. Phagocytosed ROS appear in green, very small and faint, possibly pinocytic vesicles in red, and large, very bright objects representing aggregated ROS in white. (B) Quantitative assessment of ROS phagocytosis (green areas in panel A) from ten distinct fields of view.

Figure 6.

 

Identification of genes differentially expressed in ARPE-19 cells grown on PLC membranes. ARPE-19 cells were grown to confluence on either polyester filters (P) or PLC membranes (L) and maintained at confluence for further 6 weeks. (A) Poly(A)-RNA was isolated, reverse transcribed, subjected to subtractive PCR amplification, and subcloned to generate a cDNA library enriched for differentially expressed transcripts. Equal amounts of DNA inserts of 2000 independent clones were replica-spotted to generate four identical filter sets, which were subsequently subjected to hybridization. The arrays were hybridized to probes consisting of cDNAs specific to either ARPE-19 grown on polyester filters (probe A) or on PLC membrane (probe B). Hybridization signals were normalized using β-actin, a house-keeping gene, as an internal control. Shown is one representative set of hybridized replica filters with the arrows designating a clone with increased expression in ARPE-19 cells grown on lens capsule membrane. (B) Approximately 2 μg poly(A)⁺-RNA from ARPE-19 cells was subjected to Northern blot analysis and probed to reveal the expression of GAPDH (as a loading control), annexin 4 (representing array clones 10 and 35), CLIC 1 (array clone 7), and EBP50 (array clones 28 and 36). The sizes of the detected transcripts (indicated underneath each blot) corresponded to that reported in GenBank. (C) Approximately 1 μg of total RNA from ARPE-19 cells was reverse transcribed and PCR-amplified in the linear range using SMART technology. PCR products were resolved on agarose gels and subsequently subjected to Southern hybridization to reveal the expression of GAPDH (as internal control), FLJ12584 (representing array clone 1), FLJ20811 (array clone 3), CGI-144 (array clones 4, 27, 41), and TGFBI (clone 24). The sizes of the detected cDNAs corresponded to that reported in GenBank and are indicated underneath each blot.

Table 1.

 

Molecular Identity of Genes Differentially Expressed in ARPE-19 Cells after Culture on PLC Membrane

Clone ID	Sequence ID	Function	GenBank Reference/Accession	Overexpression in North./Virtual North.
1	FLJ12584	Hypothetical protein containing armadillo repeat	NM_025139	Yes
2	mTOR, FRAP, RAFT	Protein kinase involved in regulating translation	NM_004958	Nd
3	FLJ20811	Belongs to ALEX family of proteins	BAA91387	Yes
			NM_019007
4, 27, 41	CGI-144	Unknown; highly conserved from Neurospora, C. elegans, Arabidopsis to human	NM_014077	Yes
5	Epimorphin, Syntaxin 2	Epithelial morphogenesis	NM_001980	Nd
6	C3orf1	Unknown; transmembrane (ER?) protein with a putative Surf-4 domain	NM_016589	Nd
7	CLIC 1	Nuclear chloride intracellular channel 1	NM_001288	Yes
8	KIAA1404	Unknown; RNA and/or DNA-binding domain; helicase domain	NM_021035	Nd
9	α-Spectrin	Actin crosslinking and molecular scaffold protein	NM_003127	Nd
10, 35	Annexin 4	Ca2+-and phospholipid binding protein	NM_001153	Yes
11	IGFBP-3	Insulin-like growth factor binding protein; secreted, sequesters IGF in the plasma or the ECM; proapoptotic activity	NM_000598 down-regulated	Yes
13	PSMB7	β-7 proteasome subunit; degradation of ubiquinated proteins	NM_002799	Nd
20	Neurotrypsin	Secreted protease involved in neuronal plasticity, but also found in endothelial and epithelial tissues	NM_003619	Nd
21	Aflatoxin aldehyde reductase	Metabolic enzyme – reduces aldehydes	NM_003689	Nd
22, 23	Thrombin receptor; Par 1	G-protein-coupled receptor; role during blood coagulation; also important during angiogenesis and melanoma; expression regulated by AP-2	NM_001992	Nd
24	TGFBI, BIGH-3	TGFβ-induced gene 3; secreted protein modulating integrin-mediated cell adhesion; involved in autosomal dominant corneal dystrophies	NM_000358	Yes
25	PRC1	Regulator of cytokinesis, bundles microtubules and maintains the mitotic spindle midzone	NM_003981	Nd
26	PTD004	Hypothetical protein; strong homology to heterotrimeric GTPases	NM_013341	Nd
28, 36	EBP50	Na+/H+ exchanger; PDZ containing ERM-binding protein; enriched and important within apical face of polarized epithelia	NM_004252	Yes
29, 34	GRIM-19	Cell-death regulatory protein; enhances cell death in response to IFNβ and retinoic acid; inhibitor of STAT3-dependent transcription	NM_015965	Yes
30	Aminoacyl-peptide hydrolase	Hydrolyses amino-terminally acylated polypeptides	NM_001640	Nd
32	PP1201	Unknown function; belongs to UBF0005 family of proteins; displays limited homology to annexins	NM_022152	Nd
33	Malate dehydrogenase 2	Mitochondrial metabolic enzyme	NM_005918	Nd
37	NHP2L1; NHPX	Component of the small nucleolar RNP involved in processing ribosomal RNA	NM_005008	Nd
38	BASP1	Brain abundant, membrane attached signal protein 1; myristoylated protein with several PEST motifs	NM_006317	Nd
39	Skp1; TCEB1L; Ocp2	Identified as (i) cell cycle regulated components of the proteasome; (ii) transcription elongation factor B; (iii) specific marker of the organ of Corti	NM_003197	Nd
40	Cytokeratin 18	Intermediate filament protein; together with CK 8 specific marker of single layer epithelia	NM_000224	Yes
42	Annexin 2	Ca2+-and phospholipid binding protein	NM_004039	Yes
43	LOC253263	Hypothetical protein	XM_173102	Nd

 Following subtractive PCR and array analysis using RNA isolated from ARPE-19 cells grown on Transwell polyester filter or PLC membrane (see Fig. 6 for details), gene sequences which showed altered expression (with differences of greater than threefold) were sequenced and identified using BLAST. ¹⁶ From 2,000 clones screened, 34 showed greater than threefold overexpression in cells derived from lens capsule membrane. Only one, namely IGFBP-3 (clone 11), was found to be significantly downregulated (shown in bold). Note that array clones identified as annexin 4, the thrombin receptor Par1, cell death regulatory protein GRIM-19, and the ezrin-binding protein EBP-50 were represented twice among the 35 differentially expressed library clones, whereas a gene product of unknown function, designated CGI-144, was represented three times. The last column states whether overexpression has been confirmed by Northern or virtual Northern blot analysis (for mRNAs of up to 3 kb).