Molecular Vision 2016; 22:213-223
Received 02 September 2015 | Accepted 01 March 2016 | Published 03 March 2016
1Doheny Eye Institute, Los Angeles, CA; 2Department of Ophthalmology, Keck School of Medicine of the University of Southern California, Los Angeles, CA; 3Department of Pathology, Keck School of Medicine of the University of Southern California, Los Angeles, CA; 4Department of Molecular Microbiology and Immunology, Keck School of Medicine of the University of Southern California, Los Angeles, CA; 5USC Eye Institute, Keck School of Medicine of the University of Southern California, Los Angeles, CA
Correspondence to: Henry K. W. Fong, Department of Ophthalmology, Mudd Memorial Research Building MMR 322, 1333 San Pablo Street, Los Angeles, CA 90089; Phone: (323) 442-5559; FAX: (323) 442-6564; email: firstname.lastname@example.org
Purpose: Rare mutations in the human RGR gene lead to autosomal recessive retinitis pigmentosa or dominantly inherited peripapillary choroidal atrophy. Here, we analyze a common exon-skipping isoform of the human retinal G protein-coupled receptor opsin (RGR-d) to determine differences in subcellular targeting between RGR-d and normal RGR and possible association with abnormal traits in the human eye.
Methods: The terminal complement complex (C5b-9), vitronectin, CD46, syntaxin-4, and RGR-d were analyzed in human eye tissue from young and old donors or in cultured fetal RPE cells by means of immunofluorescent labeling and high-resolution confocal microscopy or immunohistochemical staining.
Results: We observed that RGR-d is targeted to the basolateral plasma membrane of the RPE. RGR-d, but not normal RGR, is expressed in cultured human fetal RPE cells in which the protein also trafficks to the plasma membrane. In young donors, the amount of RGR-d protein in the basolateral plasma membrane was much higher than that in the RPE cells of older subjects. In older donor eyes, the level of immunoreactive RGR-d within RPE cells was often low or undetectable, and immunostaining of RGR-d was consistently strongest in extracellular deposits in Bruch’s membrane. Double immunofluorescent labeling in the basal deposits revealed significant aggregate and small punctate co-localization of RGR-d with C5b-9 and vitronectin.
Conclusions: RGR-d may escape endoplasmic reticulum-associated degradation and in contrast to full-length RGR, traffick to the basolateral plasma membrane, particularly in younger subjects. RGR-d in the plasma membrane indicates that the protein is properly folded, as misfolded membrane proteins cannot otherwise sort to the plasma membrane. The close association of extracellular RGR-d with both vitronectin and C5b-9 suggests a potential role of RGR-d-containing deposits in complement activation.
In the human eye, the RPE and Müller cells express an exon-skipping mRNA, the equivalent of which has not been found in other species . This common human mRNA variant encodes a presumably nonfunctional or dysfunctional splice isoform of the RPE retinal G protein-coupled receptor (RGR OMIM 600342) opsin. RGR is a seven transmembrane-domain (TMD) visual pigment homolog that is located in the profuse smooth endoplasmic reticulum (ER) membrane of RPE cells [2-4]. The intracellular RGR opsin binds the endogenous ligand all-trans-retinal, which is photoisomerized stereospecifically to 11-cis-retinal by blue or near-ultraviolet (UV) light [5,6]. Analyses of Rgr−/− mutant mice indicate that RGR positively influences the rate of 11-cis-retinal synthesis both in light and in darkness after irradiation [7-10]. In the human RGR gene, at least two different mutations are associated with severe retinal degeneration . One of these mutations (c.196A>C, p.S66R) is a rare cause of autosomal recessive retinitis pigmentosa, and another (c.824dupG, p.I276Nfs*77) leads to progressive peripapillary choroidal atrophy that is dominantly inherited .
The exon-skipping isoform of human RGR, referred to as RGR-d, results from an in-frame deletion of exon 6 and the complete loss of TMD6 from RGR . The copy number of extraneous RGR-d mRNA is as high as 17% of the quantity of normal RGR mRNA in human RPE . Immunological assays and mass spectrometric analysis independently confirm the existence of the RGR-d protein in human donor retina and RPE . Unlike normal RGR, RGR-d does not localize in the ER and therefore it lacks a working ER retention signal. Instead, the protein trafficks to the basal region of RPE cells, and some quantity of RGR-d or peptide fragment thereof is released from the epithelium. Deposits of extracellular RGR-d accumulate at intercapillary regions in Bruch’s membrane and in all types of drusen in older donors, including patients with age-related macular degeneration (AMD). Additionally, the distribution pattern of RGR-d within the RPE-Bruch’s membrane-choriocapillaris complex is dissimilar between young and old subjects .
To better understand trafficking and processing of RGR-d in RPE cells, we investigated the target localization of intracellular RGR-d, using high-resolution confocal microscopy. We also analyzed extracellular RGR-d to determine the potential association with other components in human Bruch’s membrane. These results provide evidence that the protein-sorting path of RGR-d differs significantly from that of normal RGR and that extracellular RGR-d eventually becomes closely associated with elements of local inflammation.
Commercially available monoclonal antibodies were directed against a neoepitope of the terminal complement complex C5b-9 (M0777/aE11; DAKO, Carpinteria, CA), human vitronectin (MAB1945; Chemicon, Temecula, CA), human CD46 (#555948; BD Biosciences, San Jose, CA), and syntaxin-4 (SC-101301; Santa Cruz Biotechnology, Santa Cruz, CA). The HRGR-DE7 antibody, which is directed against the identical carboxyl termini of human RGR and RGR-d, and an RGR-d-specific antibody DE21 were produced as described previously. Briefly, rabbit antisera were generated by Cocalico Biologicals, Inc. (Reamstown, PA) by immunization with synthetic peptides conjugated to keyhole limpet hemocyanin. The polyclonal antibodies were purified from antisera by affinity-binding to immobilized peptide attached to Affi-Gel 10 resin (Bio-Rad, Hercules, CA). DE21 was shown to bind recombinant RGR-d protein specifically without binding to full-length RGR. The DE21 antibody is directed against the peptide sequence (GKSGHLQVPALIAK) that corresponds to the sequence of human RGR-d at the splice junction of exons 5 and 7.
All experiments and procedures were conducted in compliance with applicable regulatory guidelines at the University of Southern California and the principles of human research subject protection in the Declaration of Helsinki. The Institutional Review Board of the University of Southern California approved the use of fetal eyes for culture of human RPE cells. Postmortem eyes were obtained from the Doheny Eye and Tissue Transplant Bank (Los Angeles, CA) or the National Disease Research Interchange (NDRI, Philadelphia, PA) and processed within 36 h of the time of death. Tissues for frozen sections were dissected from the central retina and processed with or without fixation. When fixation was performed, tissues were fixed with 4% paraformaldehyde in PBS (D-5652; Sigma-Aldrich, St. Louis, MO) for 4–6 h at 4 °C and then infiltrated overnight with 30% sucrose in PBS. The RPE-choroid complex was dissected from the sclera, embedded in optimum cutting temperature compound (Miles, Elkhart, IN), and frozen. The frozen tissues were sectioned with a cryostat at –20 °C to a thickness of 5–8 μm and mounted on Superfrost/Plus slides (Fisher Scientific, Pittsburgh, PA). Tissue blocks and slides were stored at −80 °C.
All RPE-choroid sections were treated with cold acetone for 5 min. The sections were then incubated with blocking buffer that consisted of 0.2% dodecylmaltoside, 3% (W/V) bovine serum albumin, and 5% (V/V) normal goat serum in PBS. After blocking, the sections were incubated with the primary antibody for 2 h at 22 °C and washed with 0.1% (V/V) Tween-20 in PBS. Primary antibodies were diluted with 0.2% dodecylmaltoside in PBS before immunostaining. Control slides were treated in the same manner, except that the primary antibodies were omitted from the binding buffer.
Immunohistochemical staining was detected by incubation of tissue-bound primary antibody with Impress solution (Vector Laboratories, Burlingame, CA) containing peroxidase-conjugated secondary antibody. The sections were visualized with the Vector VIP substrate (Vector Laboratories). The sections were dehydrated sequentially with increasing concentrations of ethanol, cleared with xylene, and mounted with VectaMount Permanent Mounting Medium (Vector Laboratories). Images were photographed using a Nikon Optiphot microscope with a Nikon D50 camera or the Aperio Scanscope Model CS (Leica Biosystems, Buffalo Grove, IN).
Immunofluorescence labeling was performed by incubating tissue sections with a primary antibody and then with the fluorochrome-conjugated secondary antibody FITC-conjugated anti-rabbit immunoglobulin G (IgG) (FI-1000; Vector Laboratories) or Texas Red- (TI-2000; Vector Laboratories) or Cy3-conjugated (715-165-150; Jackson Immunoresearch Laboratories, West Grove, PA) anti mouse IgG. For double-labeling, the sections were washed with 0.1% Tween in PBS after the first primary antibody, and immunostaining was repeated with a different set of primary and secondary antibodies, as indicated in the figure legends. Negative controls were performed in parallel by omitting the primary antibodies. The sections were mounted using VECTASHIELD Mounting Medium with DAPI (Vector Laboratories) for fluorescence applications. Images were analyzed using a PerkinElmer 6-line Spinning Disk Laser Confocal Microscope (PerkinElmer, Waltham, MA).
Primary RPE cells were isolated from 16- to 18-week-old fetal eyes (Advanced Bioscience Resources, Inc., Alameda, CA) and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) with L-glutamine, penicillin/streptomycin, and 10% fetal bovine serum (FBS), as described previously . RPE cells from single donors were grown to confluence in a fibronectin-coated cell culture flask or on Transwell filters (12 mm internal diameter, 0.4 μm pore size; Corning Costar, Tewksbury, MA). Polarized RPE monolayer cultures were used when transepithelial resistance was 300 Ohms·cm2 or greater, as measured with the EVOM epithelial voltohmmeter (World Precision Instruments, Sarasota, FL).
Proteins were electrophoresed in a 12% polyacrylamide–0.1% sodium dodecyl sulfate (SDS) gel and then transferred to Immun-Blot PVDF membrane (Bio-Rad). The blots were incubated with affinity-purified primary antibody at ambient temperature and then with a secondary antibody that was conjugated to horseradish peroxidase. Immunoreactive antigens were detected by chemiluminescence using the horseradish peroxidase-based ECL (Amersham, Arlington Heights, IL) or SuperSignal West Femto (Pierce Biotechnology, Rockford, IL) substrate systems. Chemiluminescence was detected by exposure to BioMax XAR film.
Whole cell extracts from cultured cells, RGR-d baculovirus-transduced cells, and untreated control Sf9 cells were prepared by lysing the cells in gel-loading buffer containing 62.5 Tris-HCl, pH 6.8, 2.5% β-mercaptoethanol, and 2% SDS. The recombinant human RGR-d protein was produced using the Bac-to-Bac Baculovirus Expression System (Invitrogen, Carlsbad, CA), as described previously . Spodoptera frugiperda (Sf9) cells were cultured in serum-free medium (Sf-900 II SFM) at 27 °C in a nonhumidified ambient air incubator. The cells were transfected with a baculovirus RGR-d expression bacmid to produce stocks of recombinant RGR-d baculovirus.
To produce FLAG-RGR and FLAG-RGR-d fusion proteins, we used PCR to amplify RGR and RGR-d cDNAs from previously constructed pcDNA3-hRGR and pcDNA3-hRGR-d expression vectors, respectively . The EcoRI restriction endonuclease sequences were added to both ends of the cloned fragments during PCR amplification. The sense strand primer was designed to exclude the start codons of the RGR and RGR-d pcDNA3 templates since the start codon of the pFLAG-CMV-4 vector (Sigma-Aldrich, St. Louis, MO) was to be employed. The antisense strand primer was designed to incorporate the stop codon replicated from the RGR and RGR-d pcDNA3 templates. The amplified cDNAs were then cloned into the EcoR1 cloning site of the pFLAG-CMV-4 expression vector. The sequence integrity of the new constructs, pFLAG-hRGR and pFLAG-hRGR-d, was confirmed by DNA sequencing upon comparison to published data .
ARPE-19 cells (passage 10) were kindly provided by Dr. Leonard Hjelmeland (University of California Davis) . The cells were cultured in DMEM/Ham’s F12 (1:1) medium supplemented with L-glutamine, penicillin/streptomycin, and 10% FBS. To produce FLAG-RGR and FLAG-RGR-d as protein markers, ARPE-19 cells were transfected with the pFLAG-hRGR and pFLAG-hRGR-d DNA expression vectors using Lipofectamine 2000 Transfection Reagent, according to the manufacturer’s guidelines (Invitrogen). Transfected cells were cultured initially in the presence of antibiotic G418 at 1200 μg/ml. Stable transformants were maintained in culture medium supplemented with 300 μg/ml G418.
The RGR-d isoform has a basal subcellular distribution in human RPE that differs from that of normal RGR, which tends to be concentrated in the apical portion of the cell . In RPE-choroid sections from a 47-year-old male donor, immunohistochemical labeling of RGR-d with Vector VIP substrate coincided well with the basolateral surface of individual RPE cells (Figure 1A). High-resolution analysis of immunofluorescence by spinning disk confocal microscopy clearly identified labeling of the RPE basolateral plasma membrane (Figure 1B). RGR-d was observed in the RPE basolateral plasma membrane from different donors (Figure 1 and Figure 2); however, RGR-d signals within RPE cells, typically from older donors, were often weak or undetectable, whereas immunostaining of RGR-d was strong in Bruch’s membrane.
Immunostaining of RGR-d in extracellular regions within Bruch’s membrane and the intercapillary pillars indicates that the RGR-d isoform is released from the basal side of the RPE. We compared the distribution of RGR-d with another membrane protein that localizes to the basolateral plasma membrane, CD46 (membrane cofactor protein, MCP) [18-21]. We confirmed the presence of CD46 in the basolateral plasma membrane of RPE from a 64-year-old female donor (Figure 2B). Only a small amount of CD46 co-localized with RGR-d within the plasma membrane. Unlike RGR-d, immunostaining of CD46 was not predominantly localized in Bruch’s membrane. Furthermore, CD46 was not found in the RGR-d-containing extracellular deposits.
We analyzed the expression of RGR-d in cultured human fetal RPE cells. The cells were cultured under conditions that promote epithelial cell polarization. Whole cell extracts were probed with the RGR-specific HRGR-DE7 antibody, which is directed against the identical carboxyl terminus of human RGR and RGR-d. The immunoblot assays indicated that the RGR gene was expressed in the human fetal RPE cells (Figure 3A,B). We detected a highly specific protein band in each of three different donors. The single immunoreactive protein band corresponded in size most closely to the RGR-d variant, rather than the full-length RGR protein.
Since polarized human fetal RPE cells appear to express predominantly the RGR-d variant, we analyzed RGR-d in the cells by immunofluorescence, using RGR-d specific antibody DE21. The results confirmed the expression of RGR-d in the cultured cells (Figure 4A). RGR-d resided mainly in the plasma membrane, along with syntaxin-4 (Figure 4B), a known plasma membrane marker in RPE cells . The subcellular distribution of RGR-d and syntaxin-4 was juxtaposed, although the two proteins did not co-localize within the plasma membrane (Figure 4C). Only faint immunofluorescence of RGR-d was present in intracellular regions of these cultured cells, indicating negligible RGR-d in the ER.
The immunostaining pattern for RGR-d in Bruch’s membrane differs between young and old donors . To compare the distribution and relative amount of RGR-d in the basolateral plasma membrane of young and old RPE cells, we probed for RGR-d by immunolabeling of unfixed tissues under parallel conditions. Diffuse immunohistochemical staining results indicated that the amount of RGR-d in the RPE is significantly higher in younger tissue than in tissue of older donors (Figure 5A,B). Furthermore, the immunofluorescent labeling of RGR-d in young RPE is primarily in the basolateral plasma membrane and is present in the majority of RPE cells. In older donors, it is generally difficult to detect RGR-d in RPE cells. In contrast, extracellular immunolabeling of RGR-d is relatively intense in Bruch’s membrane, intercapillary regions, and drusen of older donors.
To determine whether RGR-d associates with pathological features other than drusen, we compared the distribution of RGR-d with that of vitronectin and the terminal complement complex C5b-9 by means of double-label immunofluorescence. The results showed a close relationship between extracellular RGR-d and both C5b-9 (Figure 6A and Figure 7) and vitronectin (Figure 8A,B) in Bruch’s membrane and small basal deposits. The spatial distribution of C5b-9 as well as vitronectin closely matched the presence of RGR-d over an extended area with a significant amount of co-localization.
RGR-d is a major isoform of RGR opsin in humans. In contrast to normal RGR, the RGR-d isoform is concentrated at the basal pole of the RPE, as determined by immunohistochemical analysis. Previous immunohistochemical staining was often diffuse and did not allow precise structural localization [13-15]. Using immunofluorescent labeling and high-resolution confocal microscopy, we have localized RGR-d to the RPE basolateral plasma membrane in addition to extracellular sites in Bruch’s membrane. The localization of RGR-d in both intracellular and extracellular domains differed significantly in relation to donor age.
The targeting of RGR-d to the plasma membrane was corroborated by the analysis of human fetal RPE cells. When cultured on Transwell filters, these cells are able to form a highly differentiated polarized monolayer with high transepithelial resistance and morphological characteristics of RPE . The human RGR gene, in addition to RPE65, is expressed in these cultured fetal RPE cells. Surprisingly, only the RGR-d isoform was found, and full-length RGR protein was not detected in the fetal RPE cells from any of several different donors. RGR-d trafficked mainly to the plasma membrane of the cultured cells, with a localization pattern similar to that of the transmembrane SNARE protein, syntaxin-4. Syntaxin-4 localizes to the basolateral plasma membrane of RPE and other epithelial cells [22,23].
The presence of RGR-d in the plasma membrane of donor RPE may indicate important protein folding behavior of the human RGR variant. The loss of exon 6 and the corresponding 38 amino acids, which encompass the entire sixth transmembrane domain of RGR, leads to the loss or masking of an ER retention signal that is functional in the full-length RGR. Despite this considerable change in structure, the altered protein is able to assume a folding conformation that allows it to escape the ER and traffick to the plasma membrane. One may expect, if any newly synthesized RGR-d possessed the properties of a misfolded membrane protein, it would be targeted for protein degradation by the ER-associated-degradation (ERAD) pathway [24-26] and would fail to locate in the plasma membrane. Interestingly, the distribution pattern of RGR-d appears to be age related. As demonstrated by the intense immunofluorescent staining of young RPE, RGR-d localizes predominantly in the basolateral plasma membrane over a range of contiguous cells. In contrast, immunostaining of RGR-d in the RPE plasma membrane is often weak or undetectable in older donors. It may be that RGR-d has a low threshold for unfolding during ER stress and other adverse conditions that are believed to disturb protein homeostasis upon aging [27-29].
One may hypothesize that RGR-d is released from human RPE as a result of membrane protein degradation via variations of the ERAD pathway or turnover at the plasma membrane. At the plasma membrane, RGR-d would perforce undergo a basal rate of protein turnover by known pathways of protein degradation [30,31]. The plasma membrane proteins that are targeted for degradation by ubiquitination are internalized by endocytosis and delivered to early endosomes. The ubiquitinated membrane proteins become packaged into intraluminal vesicles and concentrated within multivesicular bodies (MVB). MVBs then fuse with lysosomes, in which the membrane proteins and lipids are degraded. To some extent, the MVBs will fuse with the plasma membrane and release their cargo of intraluminal vesicles into the extracellular space as exosomes. Exosomes have been detected in Bruch’s membrane, as described previously .
Although the mechanism by which RGR-d is released from the RPE is unknown, it appears that protein deposition into Bruch’s membrane is a selective process. Both CD46 and RGR-d were present in the same cells, but extracellular deposits adjacent to the RPE cells contained RGR-d and not CD46 (Figure 2C). Generally, deposits of CD46 and other RPE basolateral membrane proteins, such as monocarboxylate transporter 3  and bestrophin , are not highly abundant in human Bruch’s membrane. Thus, it seems unlikely that large amounts of these basolateral membrane proteins are shed indiscriminately from the RPE.
A possible pathological significance of RGR-d-containing deposits in Bruch’s membrane and drusen is indicated by the close association of these deposits with vitronectin and complement proteins C5b-9. Vitronectin is found in ocular drusen , senile plaques in Alzheimer disease , and other abnormal tissue [36-39]. Analyses of diseased renal tissue [40-42], human atherosclerotic arterial tissue , aged dermal elastic fibers , and diabetic choriocapillaris  suggest that vitronectin is significantly co-localized with the C5b-9 complex in these damaged tissues. Binding of vitronectin to C5b-9 during formation of the terminal complement complex inhibits membrane attachment, C9 polymerization, and cytolytic activity [46,47]. It is likely that vitronectin is bound to C5b-9 in Bruch’s membrane and drusen, although it is also possible that vitronectin forms crosslinked protein aggregates [48,49] or is sequestered into extracelluar aggregates of a mutant protein . In each case, the deposition of vitronectin correlates to regions of damaged or diseased tissue.
RGR-d is not a component or known regulator of the complement cascade, and therefore it was not expected for this epitope to co-localize with the membrane attack complex C5b-9. Thus far, few other proteins that are not part of the immune system have been reported to have widespread immunofluorescent co-localization with C5b-9 in Bruch’s membrane. The close association of RGR-d with both vitronectin and C5b-9 might indicate a role in complement activation. Genetic analyses of complement genes have revealed that complement dysregulation is a major contributing factor in the development or progression of AMD [20,51-54]. Locally acting factors that may trigger complement activation or promote chronic inflammation in Bruch’s membrane include carboxyethylpyrrole adducts , oxidized phospholipids and other oxidation-specific epitopes [56,57], zinc , bisretinoids from the RPE , and altered proteins in the extracellular matrix [60,61]. It is not yet known whether any of these pro-inflammatory factors are co-transported or acquired by RGR-d-containing deposits; however, it is significant that extracellular RGR-d conforms well to the natural histological distribution of the terminal complement complex. As a foreign-like component in human Bruch’s membrane, erstwhile intracellular RGR-d-containing particles from the RPE may become oxidized and present neoantigens that initiate triggers of the complement surveillance system. The otherwise futile cycle of synthesis, degradation, and release of RGR-d may then work to drive long-term formation of RGR-d/complement-associated deposits and early pathogenic traits in the human eye.
This work was supported in part by National Eye Institute grants P30EY03040 and R01EY01545 (DRH), R01EY08364 (HKWF), Research to Prevent Blindness, and the Gordon and Evelyn Leslie Macular Degeneration Fund (HKWF). Some of the results of this work were presented in a poster session at the 2015 Annual Meeting of The Association for Research in Vision and Ophthalmology in Denver, CO.