Quantitative Protein Expression in the Human Retinal Pigment Epithelium: Comparison Between Apical and Basolateral Plasma Membranes With Emphasis on Transporters

Abstract

Purpose: Retinal pigment epithelium (RPE) limits the xenobiotic entry from the systemic blood stream to the eye. RPE surface transporters can be important in ocular drug distribution, but it has been unclear whether they are expressed on the apical, basal, or both cellular surfaces. In this paper, we provide quantitative comparison of apical and basolateral RPE surface proteomes.

Methods: We separated the apical and basolateral membranes of differentiated human fetal RPE (hfRPE) cells by combining apical membrane peeling and sucrose density gradient centrifugation. The membrane fractions were analyzed with quantitative targeted absolute proteomics (QTAP) and sequential window acquisition of all theoretical fragment ion spectra mass spectrometry (SWATH-MS) to reveal the membrane protein localization on the RPE cell surfaces. We quantitated 15 transporters in unfractionated RPE cells and scaled their expression to tissue level.

Results: Several proteins involved in visual cycle, cell adhesion, and ion and nutrient transport were expressed on the hfRPE plasma membranes. Most drug transporters showed similar abundance on both RPE surfaces, whereas large neutral amino acids transporter 1 (LAT1), p-glycoprotein (P-gp), and monocarboxylate transporter 1 (MCT1) showed modest apical enrichment. Many solute carriers (SLC) that are potential prodrug targets were present on both cellular surfaces, whereas putative sodium-coupled neutral amino acid transporter 7 (SNAT7) and riboflavin transporter (RFT3) were enriched on the basolateral and sodium- and chloride-dependent neutral and basic amino acid transporter (ATB⁰⁺) on the apical membrane.

Conclusions: Comprehensive quantitative information of the RPE surface proteomes was reported for the first time. The scientific community can use the data to further increase understanding of the RPE functions. In addition, we provide insights for transporter protein localization in the human RPE and the significance for ocular pharmacokinetics.

Retinal pigment epithelium (RPE) is a polarized cell monolayer located between the photoreceptors of the retina and choroid. The apical microvilli of the RPE face the photoreceptors whereas the basal surface faces Bruch's membrane on the choroidal side. The RPE is polarized; thus, many of its functions have specific direction either from the apical-to-basolateral (from the retina into the choroidal blood flow) or basolateral-to-apical (from choroid into the retina) direction. These include important vision supporting functions of some transporter and channel proteins: the RPE provides nutrients for the photoreceptors from the choroidal blood flow and removes metabolic waste products and excess water from the subretinal space.^1,2 

The RPE is also an important tissue in ocular pharmacokinetics because it serves as the outer blood–retinal barrier. Tight junctions between the RPE cells limit the nonspecific diffusion of molecules, thereby protecting the eye from xenobiotics present in the systemic blood flow. As the RPE cells are damaged in ocular disorders, such as age-related macular degeneration (AMD) and diabetic retinopathy,³ the RPE itself is also an important drug target. It has been proposed that efflux proteins on the RPE surface would serve as a functional component for the outer blood–retinal barrier by preventing the ocular entry of their substrates.⁴ However, it is unclear which efflux transporters contribute and what is the extent of their impact. Many studies show conflicting results on the expression of the drug transporters, presumably due to the use of different cell models and antibody-based methods.⁵ In most cases, the localization of the transporters (apical or basal surface) is unclear, and their transport directionality has not been investigated. In addition, the functional assays of efflux transport have been performed mostly in the RPE cell lines^4,6 that may differ from the situation in vivo. Knowledge on transporter localization on the RPE surface is important for understanding the ocular pharmacokinetics, because the transporters may contribute to the inward (from the choroid across the RPE into the eye) or outward (from the subretinal space across the RPE into the choroid) transport depending on their localization. 

Quantitative targeted absolute proteomics (QTAP) enables the quantitative assessment of protein expression. Thus, the expression of proteins can be compared between tissues and cell models quantitatively, as in our previous study that compared the transporter expression in the plasma membranes of the ARPE19 cell line and human fetal RPE (hfRPE).⁷ However, our previous study did not provide information on transporter localization (apical or basolateral plasma membrane), and the localization has remained mostly unclear in other earlier studies. In this paper, we quantified the expression of the previously studied 36 proteins in the apical and basolateral plasma membranes of primary RPE cells. These proteins include important drug transporting proteins (e.g., multidrug resistance-associated proteins [MRPs], p-glycoprotein [P-gp]) and other transporters that are important in RPE functions (e.g., GLUT1, MCTs). Because 15 of the detected proteins were quantitated also in nonfractionated cell samples, we scaled the expression of those 15 transporters to the RPE tissue level. In addition, we show the relative expression of >1300 proteins detected with SWATH-MS (sequential window acquisition of all theoretical fragment ion spectra mass spectrometry) technology in the apical and basolateral plasma membranes. SWATH-MS has recently been developed as a novel data-independent acquisition method and enables quantitative, sensitive, and reproducible proteomic analysis.^8–11 

Materials and Methods

Cell Culture

Commercially available human fetal retinal pigment epithelial cells (hfRPE cells) were purchased from ScienCell (Carlsbad, CA, USA) (HRPEpiC, 6540) and expanded and maintained as described previously (in EpiCM medium 4101, at +37°C in 5% CO₂ atmosphere; ScienCell).⁷ At passage 3, the cells were seeded at high density (285,000 cells/cm²) and retained in culture until fluid-filled domes appeared, indicating proper apical and basolateral polarity.¹² This took 11 to 13 days after high-density seeding. 

Separation of Apical and Basolateral Plasma Membrane Fractions

Apical and basolateral membrane fractions were separated from polarized hfRPE cell monolayers with a peeling method described by Fong-ngern et al.¹³ The method was modified for this study, and the schematic presentation is displayed in Figure 1. The membrane separations were performed on two separate assay days on which two individual membrane separations were performed. This resulted in a total of four individually separated apical and basolateral plasma membrane fractions and two whole cell lysates. The membrane fractions (n = 4) were analyzed with both QTAP and SWATH, whereas the whole cell lysates (n = 2) were analyzed with QTAP. 

Figure 1

 

Schematic presentation of the apical and basolateral plasma membrane fraction separation. The RPE cells formed a monolayer in the culture. Prewetted nitrocellulose membrane was applied on top of the cells (A). The membrane was lifted resulting in the peeling of the apical microvilli (B). Both apical and basolateral fractions were purified with differential centrifugation (C), resulting in apical plasma membrane fraction and crude basolateral plasma membrane fractions (D). The crude basolateral fraction was further purified with sucrose density gradient centrifugation (E), resulting in the purified basolateral plasma membrane fraction (F).

The cells were rinsed twice with membrane-preserving buffer (1 mM MgCl₂ and 0.1 mM CaCl₂ in PBS; Gibco BRL, Grand Island NY, USA). A nitrocellulose membrane (GE Healthcare, Chicago, IL, USA) was prewetted with sterile water and placed on top of the cell monolayer (Fig. 1A). Suction was used to remove excess water, and the cell plates with nitrocellulose membranes were placed in the incubator (+37°C, 5% CO₂) for 5 minutes. After the incubation, the apical membrane was peeled by lifting the nitrocellulose membrane from the cell monolayer (Fig. 1B). The peeling efficiency was confirmed visually with light microscope. The apical membrane was scraped from the nitrocellulose membrane and collected with sterile water. The remaining cellular fraction containing the basolateral membranes was rinsed twice with PBS (Gibco BRL) and then scraped from the cell plates. The membrane fractions were isolated with differential centrifugation (Fig. 1C). The whole cells were removed with 1000g centrifugation from both membrane fractions, and the supernatants were collected for further purification. The membrane fractions were purified by removing the light mitochondrial fraction with three consecutive 15,000g centrifugations for 10 minutes at +4°C, followed by membrane pelleting at 100,000g for 40 minutes at +4°C (Sorvall WX Ultra Centrifuge, T1250 Rotor; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The resulting membrane pellets contained purified apical plasma membrane fraction or crude basolateral membrane fraction (Fig. 1D). The crude basolateral membrane fraction was further purified with sucrose density gradient centrifugation as described earlier⁷ (Figs. 1E, 1F), resulting in basolateral plasma membrane fraction. 

The Bradford and Lowry method (Bio-Rad Protein reagent, DC protein assay reagent, respectively; Bio-Rad, Hercules, CA, USA) was used to measure the protein concentrations. 

Immunofluorescence Analysis

Cells were seeded on Ibidi μ-slides at 200,000 cells/cm² (80826; Ibidi GmbH, Martinsried, Germany) and fixed with methanol after 2 weeks in culture. The immunofluorescence staining was conducted as described previously,⁷ with the primary antibodies described in Table 1 and specific Alexa Fluor (Thermo Fisher Scientific, Inc., Bleiswijk, The Netherlands) secondary antibodies. The cells were visualized with a confocal microscope (Zeiss LSM 800; Carl Zeiss Microimaging GmbH, Jena, Germany). 

Table 1

 

Antibodies Used in the Immunocytochemical Analysis

 

Protein Quantification by QTAP

The protein digestion was performed as described previously.¹⁴ A description of the procedures is detailed in the Supplementary Methods. Proteins were quantitated with QTAP as described previously.^7,14 A description of the procedures is detailed in the Supplementary Methods. 

Comprehensive Quantitative Protein Expression Profiling by SWATH-MS

Procedures for comprehensive quantitative protein expression profiling by SWATH-MS are described in the Supplementary Methods. 

Results

hfRPE Cells Display Similar Transporter Protein Expression Levels After 2- and 4-Week Cultures

We compared the abundances of transporter proteins of the cells that were cultured for 11 to 13 days (raw data values presented in Supplementary Table S3 and mean values in Table 2) to the cells cultured for 28 to 31 days (reported values⁷) to ensure that our cell model had appropriate transporter protein expression profile after 2-week culture. The comparison was made between the whole cell lysate samples for the proteins whose expression levels were quantitated in both studies by QTAP (Fig. 2A). The transporter expression was stable between the different culture times (Fig. 2A), as the protein expression differences of each transporter were statistically insignificant (unpaired t-test with Welch's correction; GraphPad Prism 5 Software, San Diego, CA, USA). Two-week culture time was chosen because it resulted in proper differentiation of the cells without large fluid-filled domes that would have caused problems in membrane separation. 

Table 2

 

Transporter Protein Expression Levels in hfRPE Cells

 

Figure 2

 

(A) Transporter protein expression is similar in the RPE cells after 2 and 4 weeks in culture. Transporter protein expression in hfRPE cells cultured for 2 weeks (11–13 days, two biological replicates) and 4 weeks (28–31 days, three biological replicates) were compared after high-density seeding at passage 3. The protein abundances were determined with QTAP (cells cultured for 2 weeks; Table 2) and the abundances of the cells cultured for 4 weeks were reported earlier.⁷ The differences of each transporter protein abundance between 2- and 4-week culture times were statistically insignificant (unpaired t-test with Welch's correction; GraphPad Prism 5 Software). (B) Most transporter proteins displayed nonpolarized expression in the RPE. Protein expression levels (data extracted from Table 2) between the apical and basolateral membrane fractions of hfRPE cells were compared. Each value is described as mean ± SEM (n = 3-4). The solid line represents the equal expression levels, and the broken lines represent 1.5-fold differences. More than 1.5-fold differences are described with black circles, whereas smaller differences are represented with gray circles. The expression level of GLUT1 was detected by QTAP, but it is not shown in this figure. (C, D) Protein expression ratios (apical/basolateral) detected with SWATH-MS. Dotted lines indicate 1.5-fold and solid lines indicate 3-fold expression differences, respectively. The apical/basolateral expression ratio of all the detected 1201 proteins were classified into transporters, receptors, other transmembrane proteins, and nontransmembrane proteins (C). (D) Selected important RPE proteins were classified according to their function (expression ratios of each protein presented in Table 3).

Expression of 15 Transporters Was Scaled Onto RPE Tissue Level

We used the transporter expression levels in the whole cell lysates determined with QTAP (Table 2, detailed data in Supplementary Table S3) to scale the protein expression onto human RPE tissue level (Table 2). The calculations are based on our hfRPE cell culture areas and their corresponding total protein content (see Supplementary Table S4 for details) and previously reported human RPE surface area¹⁵ (Table 2). 

Transporters Displayed Nonpolarized Expression in the RPE

The isolated apical and basolateral plasma membrane fractions were analyzed separately by QTAP to determine whether their transporter abundances are similar. Among 36 analyzed proteins, 4 ABC transporters, 12 solute carriers (SLC) transporters, and Na⁺/K⁺ ATPase were quantified in both apical and basolateral membrane fractions (Table 2). The comparison revealed that MRP1, MRP4, MRP5, GLUT1, 4F2hc, TAUT, CAT1, MCT4, MCT3, RFC1, OAT3, PCFT, MATE1, and Na⁺/K⁺ ATPase had similar abundance within 1.5-fold difference in both membrane fractions (Fig. 2B; Table 2). However, monocarboxylate transporter 1 (MCT1), large neutral amino acids transporter 1 (LAT1), and MDR1 (P-gp) were more than 1.5-fold enriched in the apical membrane compared with the basolateral membrane (Fig. 2B; Table 2). The expression levels of 11 ABC transporters, 7 SLC transporters, and membrane protein villin-1 remained under the limit of quantification (ULQ; Supplementary Material; Fig. 2B). 

Marker Protein Localization in Intact Cells Confirms the Success of the Membrane Separation

We analyzed marker protein expression in intact cells cultured for 2 weeks after high-density seeding (Fig. 3) to verify the success of plasma membrane separation. As in the QTAP analyses, MCT1, LAT1, and P-gp displayed modest apical enrichment, whereas MRP1 expression was detected mainly on the lateral cell surfaces, and MCT4 was present on all cellular surfaces. CD81, which was basally enriched in SWATH analysis (Table 3), localized mainly on the basal cell membrane. RPE-specific basolateral marker protein bestrophin 1 (BEST1) was enriched to the basal side, indicating proper differentiation (Fig. 3). 

Figure 3

 

Localization of marker proteins in hfRPE cells confirmed membrane separation success and proper cell differentiation. Immunofluorescence analysis showed apical enrichment of LAT1 (top two image panels), P-gp (third row), and MCT1 (bottom panel). MRP1 was found mostly on the lateral cell surfaces (third image panel). CD81 (first row) was enriched in the basal plasma membrane, whereas MCT4 (bottom panel) was detected on all cell surfaces (not included in the overview image on the right). Scale bars denote 20 μm.

Table 3

 

Localization of Selected Proteins Involved in Important RPE Functions or Involved in Macular Degeneration Detected With SWATH-MS

 

SWATH-MS Revealed the Expression and Localization of Important RPE Proteins and Transporters Relevant for Ocular Drug Delivery

To make a comprehensive protein expression atlas and determine the protein localization, the apical and basolateral membrane fractions of hfRPE cells were analyzed with SWATH-MS. We identified 1201 proteins in both apical and basolateral membranes of hfRPE cells and determined their relative expression differences. The validation of SWATH-MS results was conducted by comparing the transporter expression ratios obtained by SWATH-MS with those determined by QTAP (for details, see Supplementary Material). 

All the proteins detected with SWATH-MS and their expression ratios (apical/basolateral) are presented in Supplementary Tables S4 to S7. In total, 29 SLC and 3 ABC transporters and 55 receptors were identified. We detected additional 251 transmembrane and 863 nontransmembrane proteins in the hfRPE apical and basolateral membranes and calculated their relative expression differences (Supplementary Tables S6, S7). Graphical presentation of the apical and basolateral expression ratios of detected transporters, receptors, other transmembrane proteins, and nontransmembrane proteins is illustrated in Figure 2C. We selected important RPE proteins and classified them according to their role into six different categories (perception of light stimulus, extracellular matrix and adhesion, pigmentation, phagocytosis, transporters and ion channels, and metal ion homeostasis; Fig. 2D; Table 3). The classification was based on the categorization presented earlier by Hongisto et al.¹⁷ The relative expression of apical and basolateral membrane transporters that have clinical drugs as their substrates or are potential prodrug targets are presented in Table 4. 

Table 4

 

Transporters With Drug Substrates or Potential Prodrug Targets Detected With SWATH-MS

 

Discussion

In this study, we present quantitative differences in the hfRPE apical and basolateral plasma membrane proteomes determined with SWATH-MS. In total, 1201 proteins were identified in both membrane fractions, providing important database for understanding the cellular functions of the RPE. Furthermore, our SWATH data can be further used to map different functions on the cellular membranes via pathway analysis. We also used QTAP to determine transporter protein abundances separately on both cellular surfaces and showed that many important drug transporting membrane proteins are present on both sides of the hfRPE cell monolayer. The finding is important for ocular pharmacokinetics as the RPE forms the outer blood–retinal barrier and drug permeation across the RPE is an important factor in the ocular drug distribution. 

Membrane Separation and Polarization of the Cell Model

We performed the hfRPE membrane separation once the cell cultures had differentiated properly: fractionation was conducted when the first signs of dome formation were evident in the cultures at 11 to 13 days after high-density seeding. This assured proper polarity of the cells and successful isolation. In line with earlier literature on hfRPE cell differentiation in 11 to 14 days,¹² we detected many proteins that are important for RPE function (Table 3), indicating proper cellular maturity. The cells displayed apical enrichment of MCT1 and basal enrichment BEST1 (Fig. 3), both considered to be RPE polarization markers.³³ In addition, the total transporter protein content in the cells did not differ significantly between 2- and 4-week culture times (Fig. 2A). Taken together, the culture time, dome formation, and the marker protein expression confirm that the cells had properly differentiated before the membrane separation. 

Separation of the membrane fractions appears successful. Expression of the marker proteins (LAT1, P-gp, MCT1, MCT4, MRP1) was consistent in immunolabeling and QTAP experiments. LAT1, P-gp, and MCT1 expression were modestly enriched onto the apical membrane (QTAP; Fig. 2B; Table 2; immunofluorescence; Fig. 3). MCT4 was observed on all cellular surfaces (Fig. 3), and it was equally abundant in both membrane fractions (Table 2; Fig. 2B). MRP1 had equal abundance in both membrane fractions (Table 2; Fig. 2B) and was observed mainly on the lateral surface via immunocytochemistry (Fig. 3). This is in line with earlier literature as hfRPE cells cultured on transwells for 4 weeks showed lateral localization of MRP1.³⁴ Because laterally located MRP1 had similar expression in both fractions (Figs. 2, 3), the amount of lateral membrane in both apical and basolateral membrane fractions is similar. CD81, basally enriched in SWATH-MS data (Table 3), showed basal enrichment in intact cells (Fig. 3), confirming the success of the membrane separation. Furthermore, we detected similar Na⁺K⁺ATPase enrichment ratios (2.4- to 3.3-fold; calculated from the values in Table 2) as with previously established plasma membrane isolation method.⁷ As Na⁺K⁺ATPase is also present in melanosomes, and therefore plasma membrane/whole cell lysate enrichment remains relatively low; we also compared the enrichment between the basolateral plasma membrane and crude basolateral membrane. This ratio (2.9-fold enrichment, calculated from Supplementary Table S3) was similar as with purified plasma membrane compared with crude membrane fraction enrichment (2.1-fold, from reference 7). All in all, our fractionation method produces plasma membrane fractions that have similar level of membrane enrichment as the previously published plasma membrane isolation method used widely in the proteomics field.^7,35 

Na⁺K⁺ATPase is often enriched to the apical plasma membrane in hfRPE cultures, and this enrichment is sometimes suggested to indicate proper RPE maturity and polarization. However, we found equal levels of proteins expressed in both plasma membrane fractions with QTAP and SWATH-MS (Fig. 2B; Table 3). Our result is similar with earlier findings showing Na⁺K⁺ATPase expression on both cellular surfaces in cultured human adult RPE cells with immunocytochemistry.³³ In native human RPE, Na⁺K⁺ATPase is actually expressed on both cell surfaces.³⁶ Na⁺K⁺ATPase helps to maintain ideal ion concentration in the subretinal space, and it is vital for the functions of the RPE and photoreceptors. Another important protein for K⁺ concentration maintenance, Kir7.1, was apically enriched (Table 3), as indicated in the earlier literature.³⁶ 

Most of the detected proteins were expressed in a nonpolarized manner in hfRPE cells (Fig. 2C). This result is surprising due the well-described polarity of the human RPE in the literature,^1,2 but it can partly be explained by the underestimation of apical/basolateral expression ratios caused by ion suppression in SWATH-MS and the presence of the lateral membrane in both membrane fractions. Because our cell model consists of fetal cells, it cannot be assumed that they fully resemble the adult RPE in vivo, even though polarity of the marker proteins was evident according to the immunofluorescence imaging (Fig. 3). Furthermore, this is the first study evaluating the RPE polarization with quantitated protein expression, making a direct comparison to earlier RPE literature difficult. Due to the poor tissue availability of adult human RPE tissue, the differentiated hfRPE cells are the available material to study the proteome in the human RPE. 

Detection Limit and Ion Suppression Effect in SWATH-MS Analysis

Due the sensitivity difference between QTAP and SWATH-MS, only 7 transporters were detected with the SWATH-MS among the 16 transporters quantified by QTAP (Supplementary Table S1). The transporter expression levels in apical and basolateral membrane fractions detected with SWATH-MS were 1.02 to 1141 fmol/μg protein, whereas the expression levels of nondetectable transporters were 0.197 to 3.59 fmol/μg protein. This indicates that the detection limit of SWATH-MS is about 1 to 3 fmol/μg protein. 

SWATH-MS analyses revealed the comprehensive protein expression patterns in the apical and basolateral membrane fractions of hfRPE. The quantitative value of SWATH-MS was validated by comparing the result with QTAP results (Supplementary Table S1). Our result showed that differences of apical/basolateral ratios of SWATH-MS and QTAP were less than 50%. However, we cannot exclude the possibility that SWATH-MS results were affected by ion suppression effects that may occur in a sample-dependent manner. Most apical/basolateral expression ratios from SWATH-MS analyses were lower than those obtained by QTAP (Supplementary Table S1), suggesting that ion suppression effects take place in apical membrane. Thus, the expression ratios (apical/basolateral) indicated in Tables 3 and 4 and in Supplementary Tables S5 to S8 are possibly underestimated. 

RPE Surface Proteins and Ocular Pharmacokinetics

ABC Efflux Transporters

In the current study, MRP1 showed the highest abundance on the RPE cell surface among ABC transporters (Table 2). In vivo studies in rodents^37–39 and a recent positron emission tomography (PET)-based study in humans⁴⁰ indicated that MRPs had similar impact on the drug uptake to the brain and retina in rats,³⁹ whereas P-gp had a more modest role in the transport across the blood–retinal barrier. Overall, MRPs are suggested to have stronger impact on the ocular pharmacokinetics than P-gp, and our expression data support this conclusion as P-gp expression in the RPE was lower than the expression of MRP1 and MRP5 (Table 2). 

Most of the quantitated transporters displayed nonpolarized expression in the RPE, and the three apically enriched proteins (MCT1, LAT1, and P-gp) were detected and quantitated in both membrane fractions (Table 2; Fig. 2B). This finding seems different from many other tissues, where the polarized drug transporter localization has been reported.⁴¹ Importantly, the efflux proteins are not always expressed in a polarized manner. For instance, MRP5 did not show enrichment into luminal or abluminal membranes in porcine brain capillaries,⁴² which is similar to our finding regarding the MRP5 in the hfRPE cells (Table 2). Furthermore, P-gp was detected on both sides of human brain capillary endothelial cells with approximately 1.4-fold enrichment onto abluminal membrane.⁴³ As emphasized earlier, the previous literature regarding most RPE transporters has not been able to conclude their consistent localization either on the basal or apical membrane.⁵ However, because P-gp was detected on both sides of human RPE tissue,⁴⁴ the P-gp expression herein at both hfRPE membrane fractions is consistent with real human tissue and similar to the enrichment in human blood–brain barrier.⁴³ Kennedy et al.⁴⁴ suggested that RPE's basal P-gp function would include elimination of metabolites formed in the subretinal space and restrict the xenobiotic entry from the choroidal blood stream. Because the protein was localized on both cell surfaces in native noninduced tissue, apically located P-gp was suggested to participate in the transport functions of RPE by delivering bioactive lipids, steroids, and retinoids into the subretinal space. P-gp is unlikely to function in an opposite direction as an influx pump removing substrates from the subretinal space into the RPE. In contrast, immunofluorescent images indicated mainly apical expression of MRP1, MRP4, and MRP5 in cultured stem cell–derived RPE cells (hESC-RPE),⁴⁵ whereas we found these proteins to be expressed at equal levels in the hfRPE plasma membrane fractions (Table 2). On the other hand, fluorescein transport assays with porcine RPE suggested basal rather than apical enrichment of MRP proteins (i.e., clear directional apical-to-basolateral permeation),⁴⁶ but transport of fluorescein may be affected also by OCTs and not only MRPs. Because many substrates overlap between efflux and influx proteins, firm conclusions on the function of a specific transporter are difficult to reach. The physiologic role of the highly abundant MRP1 in the RPE involves maintenance of cellular thiol homeostasis and participation in efflux of cellular glutathione.³⁴ 

Our findings suggest that drug permeation across the RPE may not display strong directionality, because only modest differences were found between their expression on RPE apical and basolateral membranes (Fig. 2B; Table 2). As the permeation of transporter protein substrates across primary RPE cultures has not been reported,⁶ functionality data are needed. The nonpolarized expression pattern of transporters is, however, consistent with QSPR models describing the drug kinetics in the eye after intravitreal injection⁴⁷ or systemic administration.⁴⁸ The models, based on simple physical–chemical descriptors for passive diffusion, were able to describe and predict drug permeation between the vitreous and blood circulation for large compound sets without significant outliers. This suggests that strong transport directionality was absent in those data, and permeation was dominated by passive diffusion. However, as the RPE itself is an important drug target, we recently evaluated the potential influence of efflux proteins for the drug exposure in the RPE.⁵ Our simulations showed that efflux proteins can significantly lower the RPE drug exposure after intravitreal or systemic administration regardless their localization. Our current finding describing the efflux protein expression on both cell surfaces indicates that drug exposure can be reduced by these proteins, even further highlighting the importance of avoiding efflux protein substrates in drug discovery toward targets in the RPE. 

SLC Transporters

Our data show that many SLC family proteins (e.g., MATE1, TAUT, SMVT, LAT1, MCT1, MCT3, MCT4) are also present on both RPE surfaces (Tables 2, 3). The finding is interesting from the drug delivery point of view, because influx proteins may enable drug delivery to the RPE or neural retina. Lactate, which is a byproduct of photoreceptor energy metabolism, is removed from the subretinal space into the systemic circulation via monocarboxylate transporters (MCT1 and MCT3), but these transporters are also important in pharmacokinetics, because MCTs can transport drugs, such as valproate (Table 4). Analogously, LAT1 and TAUT have been under investigation as potential pathways for CNS drug targeting.^26,49 Also, SMVT and sodium- and chloride-dependent neutral and basic amino acid transporter (ATB⁰⁺) have been proposed as avenues for drug targeting; both displayed modest enrichment onto the apical membrane of hfRPE cells. Furthermore, basolaterally enriched glutamate transporter putative sodium-coupled neutral amino acid transporter 7 (SNAT7) (Table 4) has been suggested to enhance blood–brain barrier permeation of glutamate conjugates in rodents.⁵⁰ However, the transporter-mediated drug targeting to the retina has not been proven, and further investigations are needed. 

Summary

The protein enrichment values we present in this paper reveal the localization of the detected proteins among RPE surfaces. The summary of our main findings regarding the drug transporter localization in hfRPE cells is illustrated in Figure 4. As stated previously, most of the drug transporting proteins localize on both cellular surfaces. Together with the previous literature,^5,47,48 our findings suggest that the transporters on the RPE surface have higher impact on RPE drug exposure than on the permeation across the outer blood–retinal barrier. The transporter quantification and localization data of this report can be further used in building pharmacokinetic simulation models for the posterior eye segment. 

Figure 4

 

Localization of transporter proteins with drug substrates in the hfRPE cells. In the case of enrichment, the protein is illustrated with an oval shape instead of a circle. White background indicates the transporter with clinical drugs as substrates, whereas yellow background shows the proteins (SMVT, ATB⁰⁺, TAUT, SNAT7, RFT3) that are potential targets for prodrugs. Orange color indicates proteins that can be targets for both clinical drugs and prodrugs (LAT1, RFC1).

Conclusions

In this paper, we provide information of 1201 proteins present on hfRPE plasma membrane fractions (Supplementary Tables S5–S8). The identification and the quantitative expression ratios of these proteins are valuable in ocular research as they can increase the understanding of the outer blood–retinal barrier functions. Our study reveals relatively nonpolarized localization of transporters in human primary RPE cells. This finding highlights the need of avoiding efflux protein substrates in retinal drug discovery, especially when the RPE is the target tissue. We provide the abundances of 15 transporter proteins scaled to RPE tissue level, which may be further used in pharmacokinetic model building. Because hfRPE cell cultures are vital for RPE research, we predict that the membrane proteomes presented here might also be used as a tool to explain possible differences in the observed in vitro and in vivo RPE functions. 

Acknowledgments

The authors thank laboratory technician Lea Pirskanen for valuable help in the hfRPE cell culture and fractionation and Anna Poutiainen, MSc, for important contributions in the membrane separation method development. 

Supported by funding from the Academy of Finland (311122) and Doctoral Programme in Drug Research (University of Eastern Finland). AU acknowledges support from Government of Russian Federation Mega-Grant 14.W03.031.0025 “Biohybrid technologies for modern biomedicine.” This study was also supported in part by a Grant-in-Aid from the Japanese Society for the Promotion of Science for Bilateral Open Partnership Joint Research Program (between Finland and Japan). 

Disclosure: L. Hellinen, None; K. Sato, None; M. Reinisalo, None; H. Kidron, None; K. Rilla, None; M. Tachikawa, None; Y. Uchida, None; T. Terasaki, None; A. Urtti, None