Proteomic Characterization of Isolated Retinal Pigment Epithelium Microvilli*S

Polarized epithelial cells are characterized by displaying compartmentalized functions associated with differential distribution of transporters, structural proteins, and signaling molecules on their apical and basolateral surfaces. Their apical surfaces frequently elaborate microvilli, which vary in structure according to the specific type and function of each epithelium. The molecular basis of this heterogeneity is poorly understood. However, differences in function will undoubtedly be reflected in the specific molecular composition of the apical surface in each epithelial subtype. We have exploited a method for isolating microvilli from the mouse eye using wheat germ agglutinin (WGA)-agarose beads to begin to understand the specific molecular composition of apical microvilli of the retinal pigment epithelium (RPE) and expand our knowledge of the potential function of this interface. Initially, apical RPE plasma membranes bound to WGA beads were processed for morphological analysis using known apical and basolateral surface markers. The protein composition of the apical microvilli was then established using proteomic analysis. Over 200 proteins were identified, including a number of proteins previously known to be localized to RPE microvilli, as well as others not known to be present at this surface. Localization of novel proteins identified with proteomics was confirmed by immunohistochemistry in both mouse and rat eye tissue. The data generated provides new information on the protein composition of the RPE apical microvilli. The isolation technique used should be amenable for isolating microvilli in other epithelia as well, allowing new insights into additional functions of this important epithelial compartment.

Polarized epithelial cells are characterized by displaying compartmentalized functions associated with differential distribution of transporters, structural proteins, and signaling molecules on their apical and basolateral surfaces. Their apical surfaces frequently elaborate microvilli, which vary in structure according to the specific type and function of each epithelium. The molecular basis of this heterogeneity is poorly understood. However, differences in function will undoubtedly be reflected in the specific molecular composition of the apical surface in each epithelial subtype. We have exploited a method for isolating microvilli from the mouse eye using wheat germ agglutinin (WGA)-agarose beads to begin to understand the specific molecular composition of apical microvilli of the retinal pigment epithelium (RPE) and expand our knowledge of the potential function of this interface. Initially, apical RPE plasma membranes bound to WGA beads were processed for morphological analysis using known apical and basolateral surface markers. The protein composition of the apical microvilli was then established using proteomic analysis. Over 200 proteins were identified, including a number of proteins previously known to be localized to RPE microvilli, as well as others not known to be present at this surface. Localization of novel proteins identified with proteomics was confirmed by immunohistochemistry in both mouse and rat eye tissue. The data generated provides new information on the protein composition of the RPE apical microvilli. The isolation technique used should be amenable for isolating microvilli in other epithelia as well, allowing new insights into additional functions of this important epithelial compartment.

Molecular & Cellular Proteomics 3:1119 -1127, 2004.
Epithelial cells are characterized by the asymmetric distribution of proteins and lipids in their plasma membrane: a basic feature referred to as polarity. The functional polarity of epithelial cells is dependent on the asymmetric distribution of specific enzymes, signaling molecules, and transport proteins between their apical and basolateral surface membranes (1). The apical surface of cuboidal and columnar epithelial cells commonly faces a luminal cavity and is characterized by the presence of numerous surface membrane elaborations referred to as microvilli. Microvilli greatly increase the apical surface area and, consequently, the number of transport and signaling proteins it contains, thereby enhancing the epithelial functional capacity. Absorptive epithelia such as kidney and intestine have their apical surface decorated with highly organized apical microvilli of uniform length and width. More dynamic and less organized structures are present in the epithelial cells of the placenta and the retinal pigment epithelium (RPE), 1 which perform endocytosis and phagocytosis, respectively. The basis of this morphological heterogeneity is poorly understood and is likely to be related to the specialized function and specific molecular composition of the microvilli in each epithelial cell type. A first step toward the understanding of the relationship between the structure and the functions of specialized epithelial surfaces is through the identification of their constituent molecules (2).
The RPE is a low-cuboidal epithelium containing very long sheet-like apical microvilli that project into a complex extracellular matrix, referred to as the interphotoreceptor matrix. At this surface, the microvilli interact with the tips of cylindrical photoreceptor outer segments extending from the outer retinal surface. The RPE basal surface is highly infolded and interacts with the underlying Bruch's membrane (3), an acellular layer separating the RPE from the choriocapillaris. The polarized organization of RPE cells is essential for the vectorial transport of the different molecules between the choriocapillaris and the neural retina and vice versa.
The RPE performs highly specialized, unique functions essential for homeostasis of the neural retina. These include phagocytosis of photoreceptors' shed outer segments, directional transport of nutrients into and removal of waste products from photoreceptor cells, and visual pigment transport and regeneration. The apical microvilli of the RPE play a key role in mediating these activities (3)(4)(5). Another unique characteristic of the RPE is the "reversed polarity" of proteins such as the Na,K-ATPase pump, EMMPRIN, and the adhesion molecule N-CAM at the apical surface, rather than at the basolateral surface where these proteins are found in other epithelia (6 -9). However, the RPE shares many of the com-mon characteristics of other transporting epithelia such as the presence of antioxidative enzymes, amino peptidase (7), and the glucose transporter (10,11) in their apical microvilli.
A more complete definition of the protein composition of the RPE apical microvilli should provide additional insight into other biochemical processes that occur at this interface that are important for the support and maintenance of vision. To this end, we isolated RPE microvilli using wheat germ agglutinin (WGA)-agarose beads (12) and defined the protein composition using of MS. Several new proteins identified in this compartment were confirmed with immunocytochemistry.

EXPERIMENTAL PROCEDURES
Isolation and Protein Identification of Intact RPE Apical Microvilli-RPE microvilli were isolated using the protocol initially described by Cooper (12) with the addition of the mechanical removal of the retina after enzymatic digestion. This procedure was recently described in detail (5) and is only briefly summarized here. C57Bl6 mice were sacrificed by CO 2 asphyxiation, and the eyes were enucleated. The anterior segments were removed and the eyecups with the exposed neural retina were incubated in 320 U/ml bovine testes hyaluronidase (Sigma, St. Louis, MO) in Hank's buffered solution for 1h at 37°C. The neural retina was peeled off from the RPE. Eyecups were extensively washed with TBS plus 1 mM CaCl 2 for 1 h at 4°C followed by incubation with WGA-agarose beads (Sigma) in TBS for 2-3 h at 4°C. WGA beads were gently scrapped from the eyecups, collected into eppendorf tubes, washed extensively with TBS, and processed for biochemical, morphological, or immunohistochemical analysis. For proteomics analyses, the beads were dissolved in 2ϫ Laemmli buffer, boiled, and resolved by SDS-PAGE on 4 -15% gradient gels (Bio-Rad, Hercules, CA). The gel lanes were cut from top to bottom into ϳ2-mm slices. Gel slices were washed, reduced, alkylated, digested with trypsin, extracted, and resultant peptides subjected to LC MS/MS analysis using a QTOF2 mass spectrometer equipped with a CapLC System (Waters Corp., Milford, MA). Protein identifications from MS/MS data utilized the Swiss-Prot, and the NCBI sequence databases and the search engines Protein Lynx TM Global server and Mascot (Matrix Science, London, United Kingdom) (5,13,14).
Morphological Characterization of the Isolated Microvilli-For quality control of the isolation methods, samples of the WGA-agarose beads with attached microvilli and bead-treated eyecups were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer, sequentially dehydrated in ethanol and embedded in Epon as previously reported (15). Thin sections were prepared and electron micrographs were taken on a Tecnai 20 200-kV digital electron microscope (Philips, Hillsboro, OR) using a Gatan image filter and digital camera at 3,600 diameters and are printed at identical magnifications. Additional WGA beads with attached intact microvilli were fixed in 4% paraformaldehyde made in PBS, permeabilized in 0.2% Triton X-100 made in PBS for 10 min at room temperature, and reacted with antibodies to specific apical and basolateral domain markers. Samples were observed under an epifluorescence microscope, and images were collected with a cooled CCD camera (Hamamatsu C5810). Image panels were composed using AdobePhotoshop 5.5 (Adobe, San Jose, CA).
Immunocytochemistry of Identified Apical Microvilli Proteins-To confirm the localization of some of the proteins identified by LC MS/MS analysis, immunohistochemical assays were performed using both paraffin and cryosections of rats and mice eyes. Eyes were enucleated and fixed by immersion in 4% paraformaldehyde made in PBS for 3 h at 4°C, subsequently the anterior segments were removed. For paraffin processing, fixed eyecups were dehydrated and embedded in paraffin. Immunostaining was carried out on 12-m sections. After deparaffinization and rehydration to PBS, sections were subjected to heat-mediated antigen retrieval by pressure cooking in 10 mM citric acid buffer, pH 6.0. For cryosectioning, eyecups were fixed as described above, quenched with 50 mM NH 4 Cl made in PBS for 1 h at 4°C, infused successively with 15 and 30% sucrose made in the same buffer and with Tissue-Tek "4583" (Miles Inc., Elkhart, IN). For labeling, sections were blocked in PBS supplemented with 0.3 mM CaCl 2 , 1 mM MgCl 2 , and 1% BSA (PBS/CM/BSA) for 30 min, and incubated with the monoclonal anti-neuroglycan C antibody in PBS/CM/BSA overnight at 4°C. The sections were washed in PBS/CM/BSA and incubated with secondary antibodies coupled to Alexa 488 and Alexa 594 (Molecular Probes, Eugene, OR) for 1 h at room temperature. Cell nuclei were labeled with 1 M TO-PRO-3 (Molecular Probes) in PBS for 15 min. A series of 1-m xy (en face) sections were collected using a laser scanning confocal microscope (Leica TCS-SP, Exton, PA). Each individual xy image of the retinas stained represents a three-dimensional projection of the entire cryosection (sum of all images in the stack). Polyclonal antibody to Lumican was a generous gift from Chia-Yang Liu (Bascom Palmer Eye Institute, Miami, FL) and used at 1:500. Monoclonal antibody to rat neuroglycan C (BD Biosciences Pharmingen, San Diego, CA) was used at 2.5 g/ml. Phalloidin-FITC (Sigma) was used at 0.5 g/ml. Fig. 1. The procedure relies on the interaction of N-acetylglucosamine and sialic acid-containing glycoconjugates present in abundance on the free surface of epithelial microvilli (12,16) with the WGA lectin conjugated to the agarose beads. Mass interactions of the surface glycoconjugates with the immobilized lectin on the bead allow for the detachment of the RPE microvilli upon physical removal of the WGA beads. Analyzing the residual eyecups by transmission electron microscopy, we observed that the RPE layer remains attached to the Bruch's membrane following the loss of its apical membranes (Fig. 2). The RPE cells that remained following the loss of their apical microvilli displayed normal ultrastructure without any significant damage to the cell body (Fig. 2, A and B). Some cells, however, retained their apical microvilli, demonstrating that not all the microvilli were extracted with this procedure (Fig. 2B). Subsequent analysis of the WGA beads showed extensive surface areas covered with microvilli ( Fig. 2, C and D), indicating that this procedure allows for selective removal of apical microvilli from RPE layer.

RPE Microvilli Isolation-A schematic overview of the microvilli isolation method is presented in
The microvillar fraction while still attached to the WGA beads was also tested for RPE markers using antibodies directed to apical and basolateral proteins. The fixed beads were probed with specific antibodies and observed under an epifluorescence microscope. Fig. 3 shows that beads with microvilli are labeled with antibodies directed to the apical protein Na,K-ATPase (Fig. 3A) and with proteins present both in the apical and basolateral surfaces like ezrin (Fig. 3B) and glucose transporter type 1 (Glut-1) (Fig. 3C) but not with the basolateral marker, laminin (Fig. 3D). A punctate pattern characteristic of microvillar staining was observed. These data suggests that the WGA bead preparations provide a highly purified RPE microvilli fraction (Figs. 2 and 3), lacking significant cytoplasmic and basal protein contaminants.
When the isolated microvilli were subjected to fractionation on a gradient SDS-PAGE, they showed a distinctly different banding pattern compared with total RPE lysates (Fig. 4). The RPE microvilli fraction showed several prominent protein bands (ca. 280, 250, 210, 150, 88, and 49 kDa) that were not as prominent in the total RPE lysate (compare Fig. 4, lanes 5  and 6). In contrast, several major bands in the molecular mass range 33-85 kDa that were prominent in the total RPE lysate were missing in the microvilli fraction. None of these enriched proteins in the microvillar fraction were recovered under control conditions, including WGA bead alone (Fig. 4, lane 4) or protein A-agarose before (Fig. 4, lane 2) and after (Fig. 4, lane 3) exposure to eyecups.
Identification of RPE Microvillar Protein-MS of peptides present in SDS-PAGE following in-gel trypsin digestion resulted in identification of 283 proteins (Supplemental Table I). A summary of selected proteins identified is shown in Table I.  The identified protein profile of the RPE microvilli can be divided into different functional categories: retinoid-metabolizing, cytoskeletal, enzymes, extracellular matrix components, membrane proteins and transporters, unknown, and others (Fig. 5). Several proteins identified in the RPE microvilli fraction were previously identified in the microvilli of other epithelial cells. A number of identified RPE microvilli proteins has been shown independently to localize to this compartment, including basigin, Glut-1, vitronectin receptor ␣ subunit (integrin ␣v), Na,K-ATPase, ezrin, and ERM-binding phosphoprotein 50 (EBP50). In contrast, proteins resident to the basolateral domain like laminin, ZO-1, and SAP-97 were not found in these samples, substantiating the validity that this isolation procedure allows preferential enrichment of RPE apical microvilli. These results therefore provide an unbiased account of proteins present in the apical microvilli.
Many extracellular molecules were identified in this preparation, which were not characterized as being present in the RPE microvilli-interphotoreceptor matrix. Localization of lumican and neuroglycan C in adult mouse and rat eye sections was investigated to gain further understanding of the distribution of these proteins. Lumican was detected by immuno-fluorescence of cryosections of mouse eyecups at the RPE apical and basal surfaces and around the photoreceptor outer segments (Fig. 6A). Lumican localization greatly overlapped with the phalloidin-FITC staining of actin filaments (Fig. 6B), as demonstrated by the yellow color in merged images (Fig. 6C). Similarly, rat paraffin sections probed with neuroglycan C antibody revealed localization in association with both the RPE apical and basal surfaces and the photoreceptor outer segment layers (Fig. 6, D-F). Our data suggests that some of the new extracellular components identified with this method are indeed localized to the interphotoreceptor matrix, which is isolated along with the apical microvilli. DISCUSSION This report describes a simple and efficient method that produces a highly enriched RPE microvilli fraction. This conclusion is supported by several observations. First, we provide morphological evidence for specific removal of microvilli from cells that remain otherwise intact and for association of microvilli with the isolated beads. Second, we show the presence of known apical markers in the microvilli-enriched fraction, whereas there was absence of basolateral. Third, MS analysis allowed the identification of novel proteins confirmed to be present on the apical surface through subsequent morphological analysis of the tissue.
Epithelia involved in the net transport of ions such as sodium, chloride, and calcium are highly susceptible to the reactive oxygen species generated by the ion transport process itself. As a consequence, epithelial cells evolved mechanisms to protect themselves from these reactive molecules (27,28). The presence of several antioxidant enzymes observed in our intact microvilli fraction is consistent with this hypothesis.
The method described here may find application in unraveling both basic mechanisms and pathological conditions in all epithelial microvilli. Examples of pathologies involving epithelial microvilli include microvillus inclusion disease and chronic placental insufficiency. Microvillus inclusion disease is a congenital enteropathy of intractable diarrhea. Both the molecular defect and the pathophysiology of this disease are unclear, but its main feature includes villus atrophy, loss of microvilli, and the presence of intracytoplasmic inclusions, vacuoles containing brush-border-like microvilli called microvilli inclusions (29). More recently, it has been shown that decreased expression of apical membrane transporters such as NHE-2, NHE-3, and sodium glucose transporter is observed in microvillus inclusion disease samples (30). Chronic placental insufficiency is characterized by cytotrophoblast microvilli pathology (31).
Another application for this method can be the identification of age-related changes in epithelia microvilli. Aging studies have shown a decrease in both the number and the length of epithelial microvilli and a declined function of plasma membrane enzymes and receptors (32)(33)(34)(35). Specifically, the decrease in the activity of some of the enzymes Actin, cytoplasmic 1 (␤-actin) detected in RPE microvilli like Na,K-ATPase, lactate-dehydrogenase, GST, phosphoglycerate kinase, adenylate kinase (36), and catalase has been established in various epithelia (23,37,38). The RPE microvillar structure has not been extensively studied. However, some of the information available indicates that RPE microvilli possess an internal core bundle of densely packed actin filaments (39,40). Further studies detected myosin VIIa at the base of apical processes (41,42) but failed to detect villin, fimbrin, and myosin I (43,44). The entire length of the RPE microvilli has been shown to contain ezrin and EBP50 (15,43,(45)(46)(47). Moesin expression was previously detected in rat primary RPE cultures but not in RPE in vivo (15). Our proteomic study identified various types of actin and tubulin, ␤-spectrin, ezrin, moesin, EBP50, and profilin in the RPE microvilli-enriched fraction.
Our results are in agreement with previous studies that detected Na,K-ATPase (6, 48), Glut-1 (49,50), monocarboxylate transporter (11), carbonic anhydrase (51), basigin (52,53), EBP50 (46,47), the vitronectin receptor (54), and the chloride intracellular channel 6 (55) in the RPE apical microvilli. However, proteins such as peropsin, Kir7.1, and organic anion transport protein 2, known to localize to RPE microvilli, were not identified in our study. This may be due to the fact that some of these proteins elute on one-dimensional SDS-PAGE with other proteins like enolase, tubulin, actin, and keratins, and their presence becomes masked after proteolysis by more abundant peptides, as previously reported (5). A previ- ous study identified 278 proteins in normal adult human RPE lysates through cell fractionation and excision of two-dimensional SDS-PAGE gel spots (14). Forty-five of these proteins were also observed in the present analysis of mouse RPE apical microvilli. However, the previous study relied on the manual selection of the protein spots to be processed for MS.
Recently, an interaction was described between cellular retinaldehyde-binding protein and EBP50 in RPE microsomes (47). Our proteomic analyses was notably enriched in several retinoid-processing proteins such as cellular retinaldehydebinding protein, 11-cis-retinol dehydrogenase, cellular retinolbinding protein 1, interphotoreceptor retinoid-binding protein, EBP50, and ezrin. These results were used to support the hypothesis of existence of a visual cycle protein complex in the RPE apical microvilli (5).
Lumican and fibromodulin, members of the small leucinerich proteoglycan (SLRP) gene family (56), were detected for the first time in the RPE microvilli. Both proteins display overlapping expression patterns. Lumican is the major keratan sulfate proteoglycan present in the cornea. Cornea transparency is dependent on the highly ordered lattice-like collagen fibril architecture, which is dependent on lumican (57). Previous studies failed to detect lumican mRNA in retinas of newly born and adult mice (58). However, recent data generated in lumican-fibromodulin double-null mice reported multiple areas of retinal detachment (59). These retinal detachments were credited to the increased ocular axial length and biomechanical weakness due to the collagen fibril-matrix defects of the lumican-fibromodulin double-null mice. The presence of both lumican and fibromodulin in the RPE apical microvilli may also indicate a role for these molecules in stabilizing the attachment of the retina to the RPE.
Neuroglycan C is a transmembrane chondroitin sulfate proteoglycan that is predominantly expressed in the central nervous system (60). RPE and the retinal neuronal cells originate from the same optic vesicle in embryonic stage. A study reported the presence of neuroglycan C in the nerve fiber layer and inner plexiform layer in the postnatal developing rat retina (61). Besides, the presence of neuroglycan C was suggested both in the RPE apical microvilli and basal infoldings but not in the photoreceptor outer segments. Our study detected neuroglycan C in the RPE apical and basal surface and in the outer segments of photoreceptor cells. This apparent discrepancy in the detection of neuroglycan C in the photoreceptor outer segments might be related to differences in tissue processing and antibody antigenicity. The presence of this proteoglycan in both the RPE and photoreceptor outer segments suggests that it may be a component of the interphotoreceptor matrix present between both cell types.
In summary, our morphological and biochemical data shows a rapid, reliable, and reproducible method for generating enriched RPE microvilli without significant cytoplasmic or basal membrane contamination. This method together with an unbiased proteomic approach has the potential of differ-entially identifying disease-linked proteins affecting RPE and other epithelia.