|Home | About | Journals | Submit | Contact Us | Français|
Early region 3 genes of human adenoviruses contribute to the virus life cycle by altering the trafficking of cellular proteins involved in adaptive immunity and inflammatory responses. The ability of early region 3 genes to target specific molecules suggests that they could be used to curtail pathological processes associated with these molecules and treat human disease. However, this approach requires genetic dissection of the multiple functions attributed to early region 3 genes. The purpose of this study was to determine the role of targeting on the ability of the early region 3-encoded protein RIDα to downregulate the EGF receptor. A fusion protein between the RIDα cytoplasmic tail and glutathione S-transferase was used to isolate clathrin-associated adaptor 1 and adaptor 2 protein complexes from mammalian cells. Deletion and site-directed mutagenesis studies showed that residues 71-AYLRH of RIDα are necessary for in vitro binding to both adaptor complexes and that Tyr72 has an important role in these interactions. In addition, RIDα containing a Y72A point mutation accumulates in the trans-Golgi network and fails to downregulate the EGF receptor when it is introduced into mammalian cells as a transgene. Altogether, our data suggest a model where RIDα is trafficked directly from the trans-Golgi network to an endosomal compartment, where it intercepts EGF receptors undergoing constitutive recycling to the plasma membrane and redirects them to lysosomes.
Adenoviruses (Ads) are nonenveloped DNA viruses that replicate and assemble in the host cell nucleus (13). Ads are responsible for approximately 5% of acute upper respiratory tract (3) and 15% of lower respiratory tract (1) infections in infants and children. Similar to other DNA viruses, Ads are also capable of establishing persistent infections, because they have evolved numerous strategies to evade host antiviral surveillance mechanisms. Some of these same mechanisms also facilitate survival of the virus during an acute infection. Thus, a thorough understanding of viral genes that control host immune and inflammatory responses is fundamental to our ability to prevent and treat virus-induced disease at multiple levels. These mechanisms have also influenced strategies for designing Ad-based vectors for gene therapy (14).
The early region 3 (E3) transcription genes of human Ads encode several proteins that exploit the intracellular trafficking machinery to modify host immune or inflammatory responses (11, 29). Thus, E3 proteins have served as novel probes of membrane transport mechanisms in addition to providing insights to Ad pathophysiology. The most abundant E3 protein, E3/19K, suppresses the host adaptive immunity response by retaining class I MHC molecules in the endoplasmic reticulum. Other E3 proteins affect the functions of molecules involved in proliferation and apoptosis, intracellular cell signaling events linked to NF-κB, and secretion of proinflammatory chemokines. Two E3 proteins, RIDα (also called E3-10.4 and E3-13.7) and RIDβ (also called E3-14.5), that act to remove several receptors from the surface of host cells, including death receptors FAS and TRAIL, tumor necrosis factor receptor 1 (TNFR1), and EGF receptor (EGFR), have been identified. These two E3 proteins have independent functions and are also capable of forming a molecular complex that has been named the receptor internalization and degradation (RID) complex. However, mutations that affect the ability of the RID complex to downregulate one receptor molecule have no effect on expression with a second class of receptors, suggesting that regulation of different targets by the RID complex involves multiple mechanisms (9).
RIDα is a type II membrane protein expressed by all human Ad serotypes (27). Its basic structural organization consists of a short N-terminal cytosolic domain (~4 residues), two transmembrane domains connected by an exocytic loop domain (~17 residues), and a C-terminal cytosolic tail domain (~28 residues). We have shown previously that RIDα is sufficient to downregulate the EGFR (25) and that it localizes to the endocytic pathway, where it diverts EGFRs in a constitutive recycling loop to lysosomes (10). However, other laboratories have reported that RIDα functions only in the context of the RID complex that localizes to the plasma membrane, where it regulates EGFR internalization (43). Although the precise role of EGFR downregulation during an Ad infection is not completely understood, the EGFR has been linked to TNF-α-dependent NF-κB activity (22) and to the release of proinflammatory chemokines (38) in other biological settings.
The RIDα cytosolic tail has several consensus sorting signals that likely engage the host cell machinery necessary for its localization and function. The goal of this study was to identify these sorting signals and to disable them by site-directed mutagenesis. The study focused on potential interactions with clathrin adaptor protein (AP) complexes involved in membrane protein sorting at the level of the trans-Golgi network (TGN) (AP class 1 [AP-1]) and the plasma membrane (AP-2) (42). Our results indicate that RIDα residues 71-AYLRH bind to both classes of APs. In addition, we show that a Y72A point mutation traps RIDα in the TGN but not the plasma membrane, suggesting that the Ad protein encounters AP-1 first in the biosynthetic pathway. Our data also support a model where AP-2 serves as a quality control function to capture RIDα molecules that leak to the plasma membrane and redirect them back to endosomes, similar to many other membrane cargoes with overlapping AP binding sites (2). These studies were carried out with RIDα transgenes, confirming the original Ad mapping studies that demonstrated the ability of this protein to modulate EGFR trafficking independent of other E3 proteins (7).
Human hepatocellular carcinoma-derived N-PLC-PRF/5 cells (6) and fetal lung-derived WI-38 cells (20) were maintained in minimal essential medium (MEM). Human epithelial-derived A431 carcinoma cells (12) and human embryonic kidney 293 (HEK-293) cells (16) were maintained in Dulbecco's modified MEM. Chinese hamster ovary (CHO) cells (40) were maintained in MEM alpha modification. Human lung carcinoma-derived A549 cells (12) were maintained in Ham's F12 medium. All media were supplemented with 10% fetal bovine serum and 2 mM glutamine. Cells were grown at 37°C with 5% CO2 and were subcultured with 0.25% trypsin-0.1% EDTA in phosphate-buffered saline (PBS).
The following antibodies were used: α-adaptin, γ-adaptin, FLAG M2, and FLAG-BioM2 monoclonal antibodies from Sigma (St. Louis, MO); transferrin receptor (TfR) monoclonal antibody from Zymed (San Francisco, CA); the human-specific EGFR EGF-R1 monoclonal antibody (46); RIDα rabbit polyclonal antibody produced with a synthetic peptide corresponding to residues 76 to 91 (27); furin convertase rabbit polyclonal antibody from ABR (Golden, CO); glutathione S-transferase (GST) goat polyclonal antibody from Amersham-Pharmacia (Piscataway, NJ); and Golgi gp125 antibody obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biological Sciences, The University of Iowa, Iowa City, IA. Secondary antibodies were purchased from Amersham-Pharmacia (horseradish peroxidase [HRP]) or Jackson ImmunoResearch (fluorescent and HRP; West Grove, PA).
Ad stocks were grown in HEK-293 cells, and titers were determined by plaque assay, using standard techniques. The Ad2-Ad5 recombinant Ad mutant that overproduces the RIDα protein has been described elsewhere (7). Cells were acutely infected with approximately 200 PFU per cell, according to established protocols (26).
CHO cells were transiently transfected using TransIt-CHO (Mirus, Madison, WI) as directed by the manufacturer, and HEK-293 cells were transfected using FuGENE 6 (Roche, Basel, Switzerland) following the manufacturer's instructions.
Wild-type and modified RIDα carboxyl-terminal sequences were subcloned into pGEX-3X (Amersham-Pharmacia) as follows. RIDα residues Asp64 to Leu91 (see Fig. Fig.1A)1A) were subcloned in the M13mp18 single-stranded DNA phage vector (New England BioLabs, Ipswich, MA). Y72A and Y79A point mutations were introduced using the M13mp18 RIDα template, a Sculptor in vitro mutagenesis system (Amersham-Pharmacia), and the following mutagenic primers to introduce tyrosine-to-alanine substitutions (in boldface type): 5′-TGTGCGCATTGCGGCCCTCAGGCACCAT-3′ for Y72A, and 5′-GCACCATCCGCAAGCCAGAGACAGGACT-3′ for Y79A. Wild-type and mutated sequences were amplified by PCR using a forward primer (5′-CAUCAUCAUCAUGATATCGACTGGGTTTGTGTGCGCATTGC-3′) designed to anneal to the codon for Asp64 (in boldface type) and a reverse primer (5′-CUACUACUACUAGATATCTTTCGGAACCACCATCAAAACAGGATTTTCG-3′) to anneal to M13mp18 vector sequences 3′ to the polylinker region. Both primers incorporate EcoRV restriction sites (underlined) and also have noncomplementary dU repeats on the 5′ ends (in italics). The dU repeats in the PCR products were depurinated with uracyl DNA glycosidase to expose single-stranded 3′ ends and enable base-pairing to complementary ends in the linear shuttle vector pAMP (Invitrogen, Gaithersburg, MD). EcoRV fragments were subcloned in frame at the SmaI site in pGEX-3X, which reconstituted the RIDα Ile63 codon. The GST fusion plasmid encoding RIDα carboxyl-terminal sequences truncated to residues Tyr70 or Tyr79 were prepared using PCR and full-length RIDα cloned in the pcDNA1/Amp eukaryotic expression vector (Clontech, Mountain View, CA) as a template. Sequences were amplified with a forward primer 5′-CTAACTAGAGAACCCACTGC-3′ designed to anneal to vector sequences 5′ to the RIDα coding region, and reverse mutagenic primer 5′-AGTGGGATCCTAGTATTGCGGGTGGTGCCTGA-3′ to incorporate an A71Stop codon or 5′-CCTAGGATCCTAAATGCGCACACAAACCCA-3′ to incorporate an R80Stop codon, respectively (both stop codons are in boldface type). The reverse mutagenic primers also incorporated BamHI sites (underlined) compatible with the pcDNAI/Amp polylinker. The cytoplasmic tail sequences were amplified using forward (5′-GTTTCCCGGGATTGACTGGGTTTGTG-3′) and reverse (5′-GAGTCCCGGGTTGAGAGTCAGCAGTAGC-3′) primers complementary to Ile63 and sequences in pcDNAI/Amp 3′ to the RIDα coding region, respectively. Both primers incorporate SmaI sites (underlined) to facilitate in-frame cloning at the SmaI site in the pGEX-3X polylinker region. The GST fusion plasmid encoding RIDα carboxyl-terminal sequences truncated to residue His75 was created using an overlap PCR extension technique, the RIDα/pcDNA1/Amp template, and the following four primers. Primers 1 and 4 are forward and reverse flanking primers that anneal to unique restriction enzymes in the polylinker region at the 5′ and 3′ ends of the RIDα coding region that were already described, and primers 2 and 3 are forward (5′-ACCTCAGGCATAACCGCAATACA-3′) and reverse (5′-CTGTATTGCGGTTAGTGCCTGAG-3′) mutagenic primers that replace His76 with a stop codon (in boldface type). The first round of PCRs were carried out with primers 1 and 3 or primers 2 and 4, creating two overlapping PCR products. Both fragments were gel purified and used as templates for a second PCR with the flanking primers to generate a 627-bp product with the H76Stop mutation. This fragment was inserted into pcDNA1/Amp using restriction enzymes incorporated in the ends of the product compatible with restriction sites in the polylinker region. Carboxyl-terminal sequences encoding the H76Stop were amplified and subcloned in pGEX-3X exactly as described for the A71Stop and R80Stop constructs.
The cDNAs encoding full-length wild-type RIDα and RIDα with a single Y72A or Y79A amino acid substitution for expression in eukaryotic cells were constructed as follows. Wild-type RIDα cloned in pcDNAI/Amp was used as template for a PCR using forward (5′-ATCGTAAAGATCTTGATTCCTCGAGTTCTTATATTATTG-3′) and reverse (5′-CTAAGATCTCCTTAAAGAATTCTGAGAAGATCAGCTATAGTCCTG-3′) primers to amplify the RIDα open reading frame and incorporate flanking BglII restriction sites (underlined). PCR products were digested with BglII and ligated to an amino-terminal FLAG epitope in the polylinker region of the pExchange2 plasmid (Stratagene, La Jolla, CA) digested with the same restriction enzyme. The Y72A and Y79A point mutations were incorporated into pExchange2/RIDα using a QuikChange XL site-directed mutagenesis kit (Stratagene), and PCR primers 5′-GTGTGCGCATTGCGGCCCTCAGGCACCATC-3′ (forward) and 5′-GATGGTGCCTGAGGGCCGCAATGCGCACAC-3′ (reverse) (Y72A substitution in boldface type), or 5′-AGGCACCATCCGCAAGCCAGAGACAGGACTATAG-3′ (forward) and 5′-CTATAGTCCTGTCTCTGGCTTGCGGATGGTGCCT-3′ (reverse) (Y79A substitution in boldface type), according to the manufacturer's instructions.
PCR primers were designed using the DNASTAR software package (DNASTAR, Inc., Madison, WI). PCR amplifications were carried out using a RoboCycler 40 temperature cycler (Stratagene, La Jolla, CA). All PCR products and religated recombinant products were sequenced by automated DNA sequencing (Cleveland Clinic Foundation Genomics Core, Cleveland, OH).
GST fusion proteins were purified from BL21 cells that were freshly transformed with pGEX-3X plasmids. Bacteria were cultured at 37°C to an optical density at 600 nm of approximately 0.6, induced with 0.1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 16 h at room temperature, and collected by low-speed centrifugation. Cells were subjected to one freeze-thaw cycle, resuspended in a solution of 50 mM Tris (pH 7.7), 0.1 M NaCl, 0.2 mM EDTA, and protease inhibitors (0.2 mM phenylmethylsulfonyl fluoride and 1 μM leupeptin), and then digested with lysozyme (0.1 mg/ml) for 1 h at room temperature. MgSO4 was added to a final concentration of 3 mM, and bacterial lysates were digested for an additional 1 h at room temperature with 0.02 mg (each) of DNase and RNase. Lysates were adjusted to pH 7.4 and incubated with 1.5% l-lauryl sarcosine to solubilize inclusion bodies for 15 min on ice, followed by centrifugation at 12,000 × g for 10 min at 4°C. Supernatants were adjusted to 3% Triton X-100 and incubated with glutathione-Sepharose beads (Amersham-Pharmacia) for 20 min at 4°C with rotation. Beads with attached fusion proteins were washed three times with a solution of 50 mM Tris (pH 7.4), 10 mM MgCl2, and 1% Triton X-100 supplemented with 0.4 M NaCl (high-salt wash) and twice with the same solution supplemented with 0.15 M NaCl (low-salt wash). Beads with attached fusion proteins were incubated with crude subcellular fractions enriched for clathrin adaptors, by using the method described in the next paragraph, for 20 min at 4°C, and the beads were washed three times with the high-salt solution and then twice with the low-salt solution. Fusion protein complexes were solubilized with Laemmli buffer and resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) for immunoblotting using standard methods.
Crude subcellular fractionation was carried out essentially as described in reference 41. Briefly, cells were rinsed twice and then scraped in PBS supplemented with 2 mM EDTA, 5 mM EGTA, and protease inhibitors. Cells were resuspended in 0.1 M morpholineethansulfonic acid (MES; pH 6.5), 0.2 M EDTA, 0.5 mM MgCl2, 0.02% NaN3, 10 mg/ml bovine serum albumin, and protease inhibitors and then lysed with 1% NP-40 for 5 min at room temperature. Postnuclear supernatants were centrifuged at 60,000 × g for 30 min at 4°C in an Optima TL Ultracentrifuge (Beckman Instruments, Inc., Palo Alto, CA) to collect the supernatant corresponding to crude cytosol. The resulting membrane pellet was resuspended in the NP-40 lysis buffer and incubated with 0.5 M Na2CO3 for 5 min on ice to release peripheral membrane proteins. The mixture was then centrifuged at 50,000 × g for 20 min at 4°C, and the peripheral membrane protein-enriched supernatant was incubated with 0.5 M KH2PO4 for 1 h on ice. Integral membrane proteins were solubilized by incubating membrane pellets with radioimmunoprecipitation assay (RIPA) detergent (1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS) in 50 mM Tris, pH 8.0, supplemented with 150 mM NaCl, 2 mM EDTA, 5 mM EGTA, and protease inhibitors, for 30 min on ice.
Cell homogenates were also fractionated on Percoll (Amersham-Pharmacia) gradients essentially as described in reference 18. Briefly, cells were rinsed twice with PBS supplemented with 2 mM EDTA and 5 mM EGTA and then scraped in ice-cold homogenization buffer (HB) consisting of 10 mM HEPES (pH 7.5), 0.25 M sucrose, 1 mM EDTA, and protease inhibitors. Cells were collected by centrifugation, resuspended in HB, and homogenized with 22 strokes of a Dounce homogenizer. The homogenate was diluted with an equal volume of fresh HB and centrifuged at 400 × g for 10 min at 4°C to pellet unbroken cells and nuclei. Postnuclear supernatants were adjusted to a final concentration of 27% Percoll in 0.25 M sucrose using a 90% Percoll stock solution and then layered over a 1-ml sucrose cushion consisting of 10× HB. Gradients were centrifuged for 90 min at 25,000 × g in an SS34 fixed-angle rotor (Sorvall Instruments, Newtown, CT) without braking. A total of nine 1.2-ml fractions were collected manually, starting from the top of the gradient. Fractions were analyzed by comparing equal aliquots of total cellular protein determined by Bradford assay (Bio-Rad Laboratories, Hercules, CA) or by immunoprecipitation after membranes were solubilized with the RIPA lysis buffer and centrifuged at 100,000 × g for 30 min at 4°C in a TL 100.3 fixed-angle rotor (Beckman Instruments) to precipitate Percoll. Alternatively, fractions were assayed for β-hexosaminidase activity by incubating 20 μg of protein in a solution of 0.1 M MES (pH 6.5), 1 mM p-nitrophenyl-β-d-glucosaminide, and 0.2% Triton X-100 for 90 min at 37°C. The reaction was quenched with 0.5 M glycine, pH 10, and absorbance was read at 405 nm using a model 3550 automatic microplate reader (Bio-Rad). Gradients were also collected from cells that had been incubated with 125I-EGF (50 ng/ml, 150 μCi/μg; PerkinElmer-NEN) or 125I-transferrin (200 ng/ml, 0.5 μCi/μg; PerkinElmer-NEN) to label plasma membrane receptors. Fractions from these gradients were analyzed by gamma counting (Cobra Auto-Gamma counter; Packard Instruments, Meriden, CT).
Immunoprecipitations were carried out using antibodies adsorbed to protein A-Sepharose CL-4B beads (Sigma), or streptavidin agarose (Pierce, Rockford, IL). Immune complexes were centrifuged at 14,000 × g for 15 min at 4°C, washed five times with lysis buffer, solubilized with Laemmli buffer, and resolved by SDS-PAGE using standard methods. Immunodepleted supernatants were centrifuged a second time to ensure removal of immune complexes before they were added to GST fusion protein pull-down assays. Gels with radioactive proteins were treated with En3Hance (PerkinElmer-NEN, Wilmington, DE) for fluorography. Immunoblotting was carried out using proteins transferred to nitrocellulose membranes by standard methods. Nitrocellulose filters were incubated with primary antibodies and the appropriate HRP-conjugated secondary antibodies (Amersham Life Sciences, Inc., Arlington, IL; Jackson ImmunoResearch Laboratories, Inc.) for detection by enhanced chemiluminescence (Amersham Life Sciences). For quantification, immunoblots were incubated with 125I-labeled goat anti-mouse secondary antibody (1 × 106 cpm/ml, 8.07 μCi/μg; PerkinElmer-NEN) for 1 h at room temperature. Blots were air-dried after extensive washing, and radiolabeled proteins were quantified by phosphor storage autoradiography. Digitized images were analyzed using the ImageQuant software package (Molecular Dynamics, Sunnyvale, CA), which averages five measurements of light emission for each pixel location, to give a pixel value that is proportional to the amount of stored radiation.
HEK-293 cells or CHO cells transiently transfected with RIDα expression plasmids were prepared for staining essentially as described in reference 8. Briefly, cells were permeabilized with 0.5% β-escin in a solution of 80 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), pH 6.8, supplemented with 5 mM EGTA and 1 mM MgCl2 for 5 min and then fixed with 3% paraformaldehyde-PBS for 15 min. Cells were stained with primary antibody or secondary antibodies and 4′,6′-diamidino-2-phenylindole (DAPI) for 45 min at room temperature. Antibodies were diluted in a solution containing 0.5% β- escin and 3% radioimmunoassay-grade bovine serum albumin and were blocked with a solution containing 1% normal serum from the host animal used to generate the secondary antibody. Cells were examined with a Leica TCS SP2 with AOBS confocal microscope (Leica, Mannheim, Germany) using the 405-nm wavelength line of a UV laser and the 488/568-nm lines of an argon-krypton laser. Image resolution using a Leica 100×, 1.4-numerical-aperture oil immersion lens and Leica LCS software was 512 by 512 pixels. Some cells were pretreated with nocodazole (100 μM; Sigma) or vehicle (dimethyl sulfoxide) for 30 min prior to staining. Phase contrast images were collected, and the outline of the cells was drawn using Metamorph (Molecular Devices, Sunnyvale, CA) and overlaid onto the respective confocal image.
Cells were pulse-labeled for 1 h with l-[35S]cysteine (50 mCi/ml, 1,075 Ci/mmol; PerkinElmer-NEN) in cysteine-free MEM supplemented with 10% dialyzed fetal bovine serum and 0.2% bovine serum albumin and then incubated in chase medium (complete Dulbecco's MEM supplemented with 500 μM nonradioactive cysteine) for 3 h to allow proteins to achieve steady-state localization. Radiolabeled cells were incubated twice (25 min/incubation) with an ice-cold solution of 80 mM l-cysteine, 75 mM NaCl, 1 mM MgCl2, 1 mM CaCl2, 0.5 N NaOH, and 1% fetal bovine serum to reduce external disulfide bonds of surface proteins (36). Cells were rinsed twice with PBS containing iodoacetamide (1 mg/ml) and then incubated with the RIPA lysis buffer supplemented with iodoacetamide (1 mg/ml) and protease inhibitors for analysis by immunoprecipitation. Immunoprecipitates were solubilized in Laemmli buffer supplemented with 1 mg/ml iodoacetamide and separated by SDS-PAGE under reducing (0.1 M dithiothreitol [DTT]) or nonreducing (no DTT) conditions.
CHO cells, which lack endogenous EGFRs, were cotransfected with plasmids encoding human EGFR (23) and FLAG-tagged wild-type or Y72A or Y79A RIDα plasmids or empty vector. Cells were radiolabeled with [35S]cysteine for 30 min starting at 48 h posttransfection as described previously. Cells were harvested either immediately or following a 5-h chase in medium supplemented with a 10-fold excess of nonradioactive cysteine. Cells were lysed with RIPA buffer and analyzed by immunoprecipitation, SDS-PAGE, and fluorography.
We have shown previously that RIDα localizes to the endocytic pathway, where it redirects constitutively recycling EGFRs to lysosomes in Ad-infected cells (10, 26). The RIDα cytoplasmic tail contains a number of putative sorting motifs that may regulate these sorting properties (2), including a dileucine motif at residues 87-LL and two possible tyrosine-based sorting motifs containing Tyr72 and Tyr79, respectively (Fig. (Fig.1A).1A). To determine whether the RIDα cytoplasmic tail interacts with clathrin AP complexes involved in membrane protein sorting, we utilized a fusion protein consisting of the 30-amino-acid cytoplasmic domain of RIDα fused to GST (characterized in Fig. Fig.1B)1B) to pull down complexes from mammalian cell extracts. The mammalian cell fractionation scheme involved purification of a peripheral membrane protein fraction to enrich for functional AP complexes according to the method described in reference 41. Figure Figure1C1C validates this purification scheme, showing that α-adaptin and γ-adaptin (subunits of AP-2 and AP-1, respectively) are both enriched in the peripheral membrane fraction, but not in integral membrane fractions that include both of the known molecular weight forms (e.g., 13.7-kDa and 11.3-kDa) of the RIDα protein. A small amount of each of these adaptin molecules was also present in crude cytosol, as expected for AP complexes that cycle on and off membranes. Thus, the crude cytosol and peripheral membrane protein fractions were mixed with GST or the C-terminal fusion protein, precipitated with glutathione agarose, washed, and subjected to SDS-PAGE and immunoblot analysis with monoclonal antibodies to α-adaptin and γ-adaptin. As shown in Fig. Fig.1D,1D, α-adaptin and γ-adaptin from a human hepatocellular cell line were both coprecipitated with the C-terminal fusion protein in contrast to GST alone. Similar results were obtained using peripheral membrane protein fractions from a number of other cell lines, including epithelial carcinoma-derived A431 cells and normal human diploid WI-38 fibroblasts (Fig. (Fig.1E)1E) and human lung carcinoma-derived A549 cells and HEK cells (data not shown).
Our next goal was to identify the sequence(s) within the cytoplasmic tail of RIDα that interacts with each of the AP complexes identified in the first set of experiments. As already noted, RIDα contains putative dileucine and tyrosine motifs that could be involved in adaptor complex formation. Thus, we constructed a series of fusion proteins with cytoplasmic tail truncations that are shown in Fig. 2A and B to assess a role for each of these putative signals. The fusion protein with an R80Stop substitution has the carboxyl-terminal dileucine motif deleted, the H76Stop has the distal tyrosine residue Tyr79 deleted, and the A71Stop has all putative sorting motifs deleted. In vitro pull-down assays with these fusion proteins revealed that RIDα residues 71-AYLRH are responsible primarily for binding both types of AP complexes (Fig. 2D and E). These data also show that the carboxyl-terminal dileucine motif does not bind APs directly, contradicting published reports from another laboratory (21). Consistent with the results shown here, we have also observed that the conserved 679-LLRIL sequence localized to the EGFR juxtamembrane region that regulates ligand-induced postendocytic sorting to lysosomes (32, 33) does not bind APs in vitro (data not shown).
The AP binding site identified in the in vitro mapping studies has one tyrosine at residue Tyr72. To assess its role in AP binding, fusion proteins were constructed with Y72A as well as Y79A point mutations incorporated into the full-length cytoplasmic tail (Fig. (Fig.2C).2C). These fusion proteins were incubated with peripheral membrane protein fractions, and adaptin binding was quantified by immunoblotting with 125I-labeled secondary antibodies and phosphor storage autoradiography after the bound proteins were resolved by SDS-PAGE and transferred to nitrocellulose filters. Results from one set of experiments are summarized in Table Table1,1, and similar results were obtained in three separate determinations. We found that the Y72A substitution was associated with reduced binding to both AP-1 and AP-2, although its effect on AP-2 binding was consistently greater than that on AP-1 binding. Interestingly, the Y79A substitution adjacent to the primary AP binding site identified in Fig. Fig.22 was associated with increased binding to AP-2 but reduced binding to AP-1, suggesting that residues outside of the binding motif may be important for AP interactions.
Results presented to this point suggest that AP-1 and AP-2 bind overlapping but distinct regions in the RIDα cytoplasmic tail. To further test this hypothesis, studies were carried out to determine whether there is competition between AP-1 and AP-2 for binding to the C-terminal fusion protein. These studies were possible because the γ-adaptin antibody immunoprecipitates intact AP-1 complexes comprised of all four of the protein subunits (Fig. (Fig.3A).3A). Thus, peripheral membrane protein fractions were immunodepleted for AP-1 prior to incubation with GST or the C-terminal fusion protein. AP-2 binding was essentially identical when the C-terminal fusion protein was incubated with cell fractions that were precleared with an isotype-matched nonspecific immunoglobulin G versus the γ-adaptin specific antibody; in comparison, AP-1 binding was substantially reduced in the immunodepleted samples in the same experiment (Fig. (Fig.3B).3B). Although the reverse experiment was not possible since immunoprecipitations with the α-adaptin antibody must be carried out under denaturing conditions, these results strongly suggest that AP-1 and AP-2 binding to the RIDα cytoplasmic tail is noncompetitive.
Figure Figure3D3D provides further evidence that AP-1 and AP-2 bind to RIDα by distinct mechanisms. This experiment was carried out by incubating cell fractions with increasing amounts of the C-terminal fusion protein, followed by quantitative immunoblotting for α-adaptin and γ-adaptin binding exactly as described for Table Table1.1. Data are represented as the percent of total adaptin protein detected in combined crude cytosol and peripheral membrane fractions that were also quantified by immunoblotting with 125I-labeled secondary antibodies and phosphor storage autoradiography. This experiment demonstrates that α-adaptin binding is saturated rapidly compared to γ-adaptin. To evaluate the relative importance of ionic interactions in the binding of AP complexes to the cytosolic tail of RIDα, the in vitro binding assay was carried out in the presence of increasing concentrations of salt or by changing the pH of the binding solution. Ionic interactions are highly sensitive to changes in salt or pH, while hydrophobic interactions are not. Binding of α-adaptin and γ-adaptin to RIDα cytosolic tail sequences was not appreciably affected by changes in salt concentration (Fig. (Fig.3E)3E) or in pH (data not shown), suggesting that these protein-protein interactions are enabled primarily by hydrophobic amino acids with nonpolar side-chains. Thus, Ala71 and/or Leu73 in the 71-AYLRH motif may be important for AP binding; in addition, hydrophobic residues in the surrounding tail sequences may not be necessary but nevertheless contribute to the overall strength of AP binding.
The Y72A and Y79A substitutions were engineered into the full-length viral protein to study their expression and localization in mammalian cells. Wild-type and mutant proteins were transiently expressed in HEK-293 cells, and subcellular localization was determined by collecting confocal images of cells stained for the viral protein and then costained for furin or the TfR, which are well-characterized markers of the biosynthetic and endocytic compartments, respectively. The wild-type protein was largely excluded from the biosynthetic compartment defined by furin that is enriched in the TGN (Fig. (Fig.4A)4A) and was found to colocalize with the TfR (Fig. (Fig.4B),4B), a marker for endocytic vesicles as previously shown (10). The Y72A mutant, however, colocalized almost exclusively with the TGN marker furin (Fig. (Fig.4A)4A) and did not colocalize with the TfR (Fig. (Fig.4B),4B), consistent with the idea that the Y72A mutation blocks exit from the biosynthetic compartment. Similar to the wild type, the Y79A mutant did not colocalize with furin (Fig. (Fig.4A),4A), but it partially colocalized with the TfR (Fig. (Fig.4B).4B). This result indicates that while the Y79A mutation does affect AP-1 binding in vitro (Table (Table1),1), the protein is able to efficiently exit the biosynthetic compartment.
The subcellular distribution of wild-type and mutant proteins was confirmed by cell fractionation of membrane compartments on Percoll gradients. Fractions were initially characterized by quantifying radioactivity in fractions from cells that were incubated with 125I-EGF or 125I-transferrin to label plasma membrane receptors, by assaying fractions for β-hexosaminidase activity enriched in lysosomes (Fig. (Fig.5A),5A), and by immunoblotting with antibodies to well-characterized markers of biosynthetic (gp125) and endocytic (TfR) compartments (Fig. (Fig.5B).5B). Based on these analyses, we concluded that biosynthetic compartments are concentrated in fractions 2 through 5, the plasma membrane is enriched in fractions 1 and 2, and TfR-positive endosomes are concentrated in fractions 2 and 3. Both molecular weight forms of the wild-type protein exhibited a subcellular distribution that is most similar to TfR in light fractions 2 and 3, and distribution of the mutant protein was most similar to the Golgi glycoprotein gp125 (Fig. (Fig.5C).5C). Thus, these data are consistent with the images obtained by confocal microscopy showing that the wild-type protein localizes to an endocytic compartment, compared to the mutant protein, which localizes to the biosynthetic pathway.
In contrast to results shown here and published elsewhere (10, 24, 26, 28), other laboratories have reported that RIDα is localized to the plasma membrane (21, 43). Thus, the observation that the wild-type protein is enriched in fraction 2 could also be interpreted as plasma membrane localization, based on the distribution of radioactivity in gradients obtained from cells labeled with 125I-labeled ligands (Fig. (Fig.5A).5A). In addition, disabling a critical AP-2 binding site could result in plasma membrane accumulation of the Y72A mutant protein. Additional experiments were therefore carried out to exclude the possibility that RIDα is targeted to the plasma membrane. RIDα exists in two molecular weight forms: one corresponding to the full-length molecule, which has two membrane-spanning α-helices and cytosolic amino and carboxy termini, and a lower-molecular-weight species lacking the amino-terminal α-helix, which is cleaved by signal peptidase in the endoplasmic reticulum (Fig. (Fig.1A)1A) (27). Sequences connecting the α-helices should be exposed at the cell exterior in either form of the molecule located at the plasma membrane. However, this region cannot be labeled by conventional surface biotinylation, since it lacks residues with free amino groups. We therefore took advantage of the fact that both RIDα molecular weight species form disulfide-linked dimers at Cys31 (highlighted in Fig. Fig.1A)1A) (27). Hence, disulfide bonds on proteins located at the cell surface should be reduced if cells are exposed to an external reducing agent, such as cysteine, while intracellular proteins are protected (4, 36). Transfected cells were subjected to pulse-chase metabolic labeling and then incubated with cysteine-containing medium and lysed in the presence of iodoacetamide to prevent de novo disulfide bond formation. When RIDα immune complexes were resolved under nonreducing conditions (i.e., in the absence of DTT), radiolabeled proteins migrated as high-molecular-weight dimers, regardless of whether or not cells were exposed to extracellular cysteine (Fig. (Fig.5D).5D). Introduction of the Y72A mutation had no effect on the resistance of high-molecular-weight dimers to surface reduction. This result was in contrast to that for disulfide-linked TfRs, which were partially reduced by external cysteine (Fig. (Fig.5E),5E), consistent with approximately 20% to 45% of this molecule exhibiting plasma membrane localization, with the remainder in endocytic compartments (34, 47). These data confirm that the majority of wild-type and mutant proteins are localized to intracellular membrane compartments at steady state.
Endosomal compartments are dispersed throughout the cytoplasm following a brief exposure to the microtubule-depolymerizing agent nocodazole (30). We have shown previously that wild-type RIDα displays the same behavior, consistent with the viral protein's localization to endocytic compartments (A. H. Shah, N. L. Cianciola and C. Carlin, submitted for publication). We therefore reasoned that a similar treatment should have little or no effect on localization of the RIDα protein with the Y72A substitution if the mutant protein accumulates in a biosynthetic compartment, as is indicated by the results shown in Fig. Fig.44 and and5.5. Results in Fig. Fig.66 indicate that nocodazole has a very modest effect on the distribution of the Y72A mutant protein compared to vesicles with wild-type RIDα that were redistributed from the perinuclear region to the peripheral cytosol. Altogether, these data suggest that the mutant protein is retained in the biosynthetic pathway and mostly excluded from a nocodazole-sensitive endocytic pathway.
Although many functions have now been associated with this Ad protein, RIDα was first identified because of its ability to specifically reduce EGFR plasma membrane expression (7). Thus, studies were carried out to determine the effect of the Y72A and Y79A mutations in RIDα on EGFR metabolic stability in CHO cells expressing both proteins. Cells were cotransfected with viral protein and human EGFR plasmids and harvested for total cellular protein or pulse-labeled with [35S]cysteine for 30 min to measure protein stability. When viral protein immune complexes were examined by immunoblotting, we observed that wild-type and mutant viral proteins were expressed at similar levels at 48 h posttransfection (Fig. (Fig.7A).7A). The EGFR acquires seven to nine N-linked high-mannose oligosaccharides cotranslationally that are processed to complex carbohydrates during Golgi body maturation, resulting in retarded migration on SDS-polyacrylamide gels (5). A molecular weight species corresponding to the high-mannose oligosaccharide EGFR precursor was detected in cells cotransfected with an empty vector control or plasmids encoding wild-type or mutant viral protein in cells that were harvested immediately after a 30-min pulse-label (Fig. (Fig.7B),7B), showing that the EGFR is synthesized to similar extents in all four populations of transfected cells. The molecular weight species corresponding to the mature EGFR protein with complex carbohydrates was detected in cells transfected with the empty vector control or with plasmids encoding the Y72A or Y79A mutant proteins after a 5-h chase with nonradioactive amino acid precursors. This was in contrast to cells expressing wild-type RIDα, where mature EGFR protein was not detected, consistent with the known ability of RIDα to target EGFR to lysosomes for degradation in the absence of other Ad proteins (25). Thus, the Y72A mutation appears to block the ability of the viral protein to facilitate EGFR degradation. RIDα with the Y79A mutation also does not downregulate the receptor, even though these mutants appear to localize to endosomes (Fig. (Fig.4B).4B). The loss-of-function phenotype associated with the Y79A mutant can likely be attributed to the fact that this region mediates protein-protein interactions with ORP1L (Shah et al., submitted), a Rab7 effector that links late endocytic vesicles to minus end-directed microtubule motor complexes (31).
The E3 region is not required for viral replication; nevertheless, it plays a critical role in Ad pathogenesis (15). The importance of this region is underscored by the fact that the first generation of Ad gene therapy vectors which contained large E3 deletions were ultimately deemed unsafe (14). E3 genes encode integral membrane proteins that regulate a variety of host cell functions involved in innate immunity and inflammatory responses. The ability of these proteins to modify host cell function is due in part to cytosolic tail sequences that interact with sorting machinery and target membrane proteins to specific intracellular compartments. The E3 protein RIDα was originally identified because of its ability to downregulate the EGFR (7) and other related receptor tyrosine kinases (35). In this study, we have demonstrated that RIDα residues 71-AYLRH comprise a binding site for AP complexes and that Tyr72 is required for RIDα localization to endosomes and its ability to downregulate the EGFR. These findings support previous studies concluding that RIDα acts by targeting EGFRs undergoing constitutive recycling to the plasma membrane (10, 26, 28). The fact that 71-AYLRH is precisely conserved in all Ad serotypes that have been sequenced except for Ad12 suggests its fundamental importance in a majority of Ad-induced diseases (7).
Although the mutation of Tyr72 to an alanine residue leads to a clear reduction in AP binding in vitro (Table (Table1),1), 71-AYLRH does not conform to classical tyrosine-dependent YXXØ motifs that have a preference for hydrophobic residues with bulky side chains at the Ø position (39). Instead, it has a histidine residue that is mildly basic and hydrophilic. However, many other factors contribute to sorting signal recognition, including the position of the signal relative to the membrane and to the carboxyl terminus and the presence of amino acid residues in areas adjacent to the signal. We have reported previously that 71-AYLRH exists in an amphipathic helix that is stabilized by interactions with a membrane-mimetic phospholipid micelle surface based on data obtained using nuclear magnetic resonance spectroscopic methods (45). This close degree of membrane association could have an important role in regulating the availability of the signal for interaction with APs. Thus, 71-AYLRH may be masked when the cytoplasmic tail is intimately associated with membrane, while cellular events that result in its translocation into the cytosol could make it available for binding APs. Some examples that could bring about such a change include modulation by another membrane protein, changes in the endosomal pH or ionic environment as RIDα traverses different cellular compartments, posttranslational modification, or a dramatic shift in the tilt or membrane placement of the adjacent transmembrane helix. For example, we have recently demonstrated that RIDα function is highly dependent on reversible palmitoylation at a residue in the cytosolic tail (N. L. Cianciola and C. Carlin, unpublished data), suggesting that the association of the RIDα cytosolic tail with membranes is tightly regulated in cells (17). The possibility that 71-AYLRH availability is regulated by a transmembrane mechanism is particularly intriguing since the RIDα loop domain connecting its two membrane spanning domains resides in compartmental lumens. Thus, the transmembrane domain could act as a conduit to fine-tune 71-AYLRH recognition at specific subcellular organelles.
Even though 71-AYLRH recognizes two different classes of APs, we have demonstrated that AP-1 and AP-2 do not compete for binding to RIDα in vitro (Fig. (Fig.3C)3C) and that mutation of adjacent residue Tyr79 leads to increased binding of AP-2 but diminished binding to AP-1 (Table (Table1).1). Thus, it seems likely that AP-1 and AP-2 recognize distinct but overlapping sets of tyrosine-based sorting signals in RIDα. The Y72A mutation traps RIDα in the TGN (Fig. (Fig.4B)4B) but not the plasma membrane (Fig. (Fig.5D)5D) suggesting the Ad protein encounters AP-1 first in the biosynthetic pathway. AP-2 is known to interact with a majority of tyrosine-based signals identified for other molecules, in agreement with studies showing that most naturally occurring signals mediate internalization (37). The broad specificity of AP-2 recognition implies that it serves a quality control function to retarget membrane proteins to their correct intracellular location that escape to the plasma membrane (2). This would suggest that the plasma membrane may not be an obligatory membrane transport destination for RIDα and that a majority of RIDα is delivered directly to endosomes (see summary model in Fig. Fig.8).8). Accordingly, even though RIDα can be detected on the plasma membrane, this localization correlates with high levels of protein expression and constitutes a relatively minor fraction of the total protein in the cell (10). This is entirely consistent with the quality control role for AP-2 that has been proposed for membrane proteins that leak to the plasma membrane.
In addition to the Tyr72-based sorting motif, RIDα also has a potential dileucine-based motif located at residues 87-LL (Fig. (Fig.1A).1A). Although another laboratory has published a study saying that these residues constitute an AP-2 binding site (21), those results were not substantiated in the present study. We did observe that AP binding was remarkably insensitive to pH or salt (Fig. (Fig.3D),3D), supporting a role for hydrophobic interactions either within the signal itself or in adjacent regions. It is possible that the dialanine substitution at 87-LL analyzed in reference 21 lowers the overall strength of AP binding. Interestingly, 87-LL is part of a larger motif that is precisely conserved in the EGFR (26), suggesting its involvement in cargo selection and/or targeting EGFRs to lysosomes. This conjecture is supported by evidence that this sequence in the EGFR is necessary for ligand-induced trafficking to lysosomes (32, 33) and also RIDα-mediated diversion of recycling EGFRs to lysosomes (44). Thus, although 87-LL may not be directly involved in AP recognition, it undoubtedly has an important role in RIDα function at least as it relates to EGFR downregulation. The 87-LL motif shared with the EGFR is found in group C Ad2 and Ad5; however, this region is not precisely conserved in other serotypes (7). Thus, different Ads may vary in their ability to specifically target the EGFR.
RIDα has been associated with other activities besides EGFR downregulation. For example, the RID complex (comprised of RIDα and RIDβ) downregulates death receptors, including TNFR1 and FAS (9). However, mutagenesis studies support a model where the RID complex acts on TNFR1 at the plasma membrane, in contrast to FAS, where the functional interaction occurs intracellularly. These seemingly paradoxical results are best understood by considering the many steps involved in receptor downregulation. These include cargo selection, sorting to specific endocytic compartments involved in transport to lysosomes, and coupling to microtubules necessary for transporting MVB intermediates to the perinuclear region (19). We have already discussed the idea that the molecular basis for RIDα-mediated EGFR cargo selection likely involves the dileucine motif that is conserved in EGFR and RIDα encoded by Ad2 and Ad5. In addition, we have recently discovered that RIDα interacts with Rab7 effectors, including RILP and ORP1L, which are necessary for microtubule-dependent transport (Shah et al., submitted). Thus, it is likely that RIDα regulates EGFR downregulation at multiple levels. Other cargoes could require accessory molecules to deliver specific receptors to endosomes once the maturation process is under way. Thus, RIDβ may promote TNFR1 uptake to endosomes, whereupon they are then sorted to lysosomes by a mechanism involving RIDα-dependent endosomal maturation. This conjecture is consistent with the observation that RIDβ binds AP-2 and that AP-2 is required for RID-mediated downregulation of TNFR1 but not FAS (9). The ability to “mix and match” different aspects of RID function may have evolved to allow Ads to fine-tune RID activity in different cell types or during acute versus persistent infections.
These studies were done with the support of the Pediatric Imaging Center of University Hospitals of Cleveland, with the help of Sarah Richer Dawson. We also thank Kim Preston for technical assistance and Song J. Kil for critically reviewing the manuscript.
This work was supported by Public Health Service grant RO1 GM64243. N.L.C. was supported in part by NIH HL-007653, and A.H.S. was supported in part by a Cell and Molecular Biology Training Grant awarded through the National Institute of General Medical Sciences.
Published ahead of print on 18 July 2007.