A new crucial protein interaction element that targets the adenovirus E4-ORF1 oncoprotein to membrane vesicles

Abstract

Human adenovirus type 9 exclusively elicits mammary tumors in experimental animals, and the primary oncogenic determinant of this virus is the E4-ORF1 oncogene, as opposed to the well-known E1A and E1B oncogenes. The tumorigenic potential of E4-ORF1, as well as its ability to oncogenically stimulate phosphatidylinositol 3-kinase (PI3K), depends on a carboxyl-terminal PDZ domain-binding motif (PBM) that mediates interactions with several different membrane-associated cellular PDZ proteins, including MUPP1, PATJ, MAGI-1, ZO-2, and Dlg1. Nevertheless, because certain E4-ORF1 mutations that alter neither the sequence nor the function of the PBM abolish E4-ORF1-induced PI3K activation and cellular transformation, we reasoned that E4-ORF1 must possess an additional crucial protein element. In the present study, we identified seven E4-ORF1 amino acid residues that define this new element, designated domain 2, and showed that it mediates binding to a 70-kDa cellular phosphoprotein. We also discovered that domain 2 or the PBM independently promotes E4-ORF1 localization to cytoplasmic membrane vesicles and that this activity of domain 2 depends on E4-ORF1 trimerization. Consistent with the latter observation, molecular-modeling analyses predicted that E4-ORF1 trimerization brings together six out of seven domain 2 residues at each of the three subunit interfaces. These findings importantly demonstrate that PI3K activation and cellular transformation induced by E4-ORF1 require two separate protein interaction elements, domain 2 and the PBM, each of which targets E4-ORF1 to vesicle membranes in cells.

FIG. 1.

Mutations in domain 2, TRI, and PBM elements of the Ad9 E4-ORF1 protein. A multiple-sequence alignment of Ad subgroup A to D E4-ORF1 protein sequences reveals 33 conserved amino acid residues (boldface). The filled circles are placed above 25 such Ad9 E4-ORF1 residues mutated in the current study. Also depicted are the locations of residues defining domain 2, TRI (VKI, residues 119 to 121), and core PBM (ATLV, residues 122 to 125) elements of the Ad9 E4-ORF1 polypeptide. The names of all Ad9 E4-ORF1 mutants are also shown, with boldface names indicating the seven mutants that define domain 2. For PBM and TRI/PBM mutants, underlined residues or dashes represent substituted or deleted residues, respectively.

FIG. 2.

Protein expression and cellular transformation phenotypes of 25 new Ad9 E4-ORF1 mutants. (A) Transient expression levels of E4-ORF1 mutant proteins. CREF fibroblasts on 6-cm dishes were transfected with 3.5 μg of empty pJ4Ω plasmid (vector) or pJ4Ω plasmid coding for either wt E4-ORF1 (wt) or the indicated mutant E4-ORF1. An equivalent amount of each cell extract (100 μg protein) was immunoblotted with E4-ORF1 antibodies. Two separate controlled experiments are presented in the top and bottom rows, with double vertical lines denoting the locations of deleted irrelevant lanes. Similar results were obtained in four independent experiments. (B) Cellular transforming activities of E4-ORF1 mutants. CREF fibroblasts were transfected with empty pJ4Ω plasmid (vector) or pJ4Ω plasmid coding for either wt E4-ORF1 (wt) or the indicated mutant E4-ORF1, as for panel A. Transformed foci were counted at 3 weeks posttransfection. Focus-forming activity is reported as the percentage of foci generated by each E4-ORF1 mutant relative to wt E4-ORF1, which was set at 100%. Data compiled from three independent experiments include the means plus standard deviations. The names of mutants with or without protein expression defects, as determined by data presented in panel A, are colored gray or black, respectively. The arrows point to candidate domain 2 mutants. (C) Photographs of selected focus formation assays quantified in panel B. Only expression-competent Ad9 E4-ORF1 mutants are shown. (D) V41A is the crucial mutation in mutant IA. Focus formation assays in CREF fibroblasts were carried out, quantified, and presented as described for panel B. Data compiled from two independent experiments show the mean plus the range. The immunoblot assay comparing E4-ORF1 protein levels (top) was conducted as described for panel A. (E) Photographs of focus formation assays quantified in panel D.

FIG. 3.

All five candidate domain 2 mutants fail to activate PKB. PKB activation by stably expressed candidate domain 2 mutants. An equivalent amount of extract (60 μg protein) from serum-starved CREF fibroblasts stably transfected with empty pBABE plasmid (vector) or pBABE plasmid coding for either wt E4-ORF1 (wt) or the indicated mutant E4-ORF1 were immunoblotted with antibodies to phospho-PKB (T308), PKB, or E4-ORF1.

FIG. 4.

Domain 2 is a new protein interaction element in Ad9 E4-ORF1. (A) Domain 2 mutants bind to cellular PDZ proteins but not to cellular pp70 in GST pull-down assays. GST pull-down assays conducted with extracts (750 μg protein) of [³²P]orthophosphate-labeled CREF fibroblasts were performed as described in Materials and Methods. The recovered proteins were resolved by SDS-PAGE and visualized by autoradiography. An asterisk marks the location of the cellular pp70 band. The control Coomassie-stained gel (bottom) verified the use of equivalent amounts of each GST-E4-ORF1 fusion protein. (B) Cellular pp70 is a phosphoprotein that binds to PBM mutant IIIA but not domain 2 mutant F91S. GST pull-down assays were conducted as described for panel A. Recovered proteins were either mock treated (−) or treated with proteinase K (+) for 30 min at 37°C and then analyzed as described for panel A. The control Coomassie-stained gel (bottom) verified the use of an equivalent amount of each GST-E4-ORF1 fusion protein. Putative domain 2-interacting 150-kDa and 190-kDa proteins are marked with arrowheads. (C) Cellular pp70 coimmunoprecipitates with E4-ORF1. Human TE85 cells or the same cells stably transduced with empty pBABE plasmid (vector) or pBABE plasmid encoding either wt E4-ORF1 (wt) or mutant F91S were metabolically labeled with [³²P]orthophosphate. Extracts (800 μg protein) of TE85 cells were subjected to GST pull-down assays with the indicated fusion proteins, whereas extracts (3 mg protein) of the indicated stably transduced TE85 lines were immunoprecipitated with E4-ORF1 antibodies. Proteins recovered by pull-down (left) and immunoprecipitation (IP) (right) were resolved on the same SDS-PAGE gel, and portions of the gel containing either cellular PDZ proteins (top) or pp70 (middle) were visualized by autoradiography. The data are assembled into four separate properly aligned boxes to permit the presentation of optimal exposures. Other portions of the gel were either Coomassie stained to verify the use of equivalent amounts of each GST fusion protein (lower left) or immunoblotted (IB) with E4-ORF1 antibodies to verify comparable immunoprecipitations of E4-ORF1 protein (lower right).

FIG. 5.

Domain 2 mutants form functional homotrimers and monomers. (A) Domain 2 mutants properly trimerize. 293T cells on 6-cm dishes were transfected with 3 μg of GW1 plasmid coding for either wt E4-ORF1 (wt) or the indicated mutant E4-ORF1 using TransIT LT1, and cell extracts were subjected to analytical size exclusion chromatography in the presence of 0.1% deoxycholate as described in Materials and Methods. One-tenth of each eluted fraction was immunoblotted with E4-ORF1 antibodies. Only fractions containing E4-ORF1 are shown. The peak elution fraction of each size marker is also indicated. (B) Endogenous Dlg1 coimmunoprecipitates with domain 2 mutants stably expressed in cells. Extracts (250 μg protein) from CREF fibroblasts transduced by empty pBABE retrovirus (vector) or pBABE retrovirus coding for either wt E4-ORF1 (wt) or the indicated mutant E4-ORF1 were immunoprecipitated with E4-ORF1 antibodies. Recovered proteins or total cell extracts (60 μg protein) were immunoblotted with antibodies to E4-ORF1 or Dlg1. (C) Domain 2 mutants aberrantly sequester ZO-2 within detergent-insoluble complexes. The CREF fibroblast lines shown in panel B were fractionated into detergent-soluble supernatant (S) and detergent-insoluble pellet (I) fractions as described in Materials and Methods. The soluble fraction (60 μg protein) and an equivalent amount of the insoluble fraction were subjected to SDS-PAGE and immunoblotted with antibodies to E4-ORF1 or ZO-2.

FIG. 6.

Domain 2 or the PBM independently targets Ad9 E4-ORF1 to membrane vesicles. (A) E4-ORF1 colocalizes with clathrin-coated vesicles in cells. CREF fibroblasts stably transfected with pBABE plasmid coding for wt E4-ORF1 were stained with E4-ORF1 (green) and clathrin (red) antibodies and then visualized by epifluorescence microscopy, as described in Materials and Methods. The arrow points to a representative vesicle where E4-ORF1 and clathrin colocalize. The area of interest is magnified in the inset. In the merged image, DAPI-stained nuclei are blue. E4-ORF1 membrane staining is not evident under these experimental conditions. (B) Domain 2 mutant F91S and the PBM mutant IIIA, but not domain 2-PBM double mutant F91S/IIIA, retain vesicular staining patterns. CREF fibroblasts stably transfected with pBABE plasmid coding for either wt E4-ORF1 or the indicated mutant E4-ORF1 were stained with E4-ORF1 antibody and visualized by epifluorescence microscopy. DAPI-stained nuclei are pseudocolored red. (C) E4-ORF1 trimers, but not monomers, possess domain 2 activity. CREF fibroblasts transduced with empty pBABE plasmid (vector) or pBABE plasmid coding for the indicated E4-ORF1 mutant were stained with E4-ORF1 antibody and visualized by epifluorescence microscopy. DAPI-stained nuclei are pseudocolored red.

FIG. 7.

Predicted clustering of domain 2 residues at subunit interfaces of the E4-ORF1 trimer. Surface representations of the E4-ORF1 trimer are shown from top (1), side (2), and bottom (3) viewpoints, with different monomer subunits colored white (subunit A), light gray (subunit B), or dark gray (subunit C) and their corresponding domain 2 residues (indicated by residue number) colored blue, green, or red, respectively. In the top and side viewpoints, a cyan-colored line circumscribes the six domain 2 residues that group together at one subunit interface. TRI and PBM elements are colored purple and orange, respectively, with subscript A, B, and C indicating their subunit origins.