INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The E4 transcription unit of adenovirus type 5 (Ad5) is predicted to encode seven different polypeptides, the products of E4orf1, E4orf2, E4orf3, E4orf4, E4orf3/4, E4orf6, and E4orf6/7, all but one of which (E4orf3/4) have been demonstrated to exist in infected cells. Mutant viruses with a deletion of the entire E4 region display profound defects in viral DNA replication, viral late mRNA accumulation, viral late protein synthesis, and the shutoff of host cell protein production. These effects have been attributed to overlapping activities of the viral E4orf6 (E4-34kDa) and E4orf3 (E4-11kDa) proteins, whereas all other E4 gene products are dispensable for virus replication in culture (reviewed by Leppard [26]).

The lytic functions of E4orf6 and E4orf3 involve interactions with the 55-kDa protein expressed from the early region 1B (E1B) transcription unit (E1B-55kDa). In the course of infection, both E4 gene products can individually bind to E1B-55kDa and change its localization from cytoplasmic to predominantly nuclear (10, 23, 27, 44, 49). While the significance of the E1B-E4orf3 interaction to Ad infection is uncertain, the E1B-E4orf6 complex is known to promote the nuclear export of viral mRNA and to simultaneously inhibit the transport of most cellular mRNA species (reviewed in reference 17). The complex may serve as a transporter for viral mRNA as it has been shown to shuttle continuously between the nucleus and the cytoplasm (6). A leucine-rich nuclear export signal (NES) that is required for transport of the viral protein complex out of the nucleus has been identified within E4orf6. However, in the absence of E1B-55kDa, a putative dominant nuclear retention signal (NRS) may prevent export of E4orf6 (6). The NRS seems to be part of a recently identified, arginine-faced amphipathic alpha helix within the carboxy-terminal region of the E4 protein. It has been proposed that this structure mediates an association with E1B-55kDa, which blocks nuclear retention of E4orf6, thereby promoting the export of the E1B-E4 protein complex into the cytoplasm (6). The amphipathic alpha helix has also been shown to be crucial for targeting E1B-55kDa to the nucleus and for E4orf6 function during productive viral infection (43).

Besides its essential role in viral lytic replication, there is accumulating evidence that E4 genes also contribute to the transforming and oncogenic potential of adenoviruses, which has been traditionally ascribed to the E1 region. It is well established that the E1A and E1B oncoproteins cooperatively mediate transformation of primary human and rodent cells by interacting with a number of key cellular growth regulators (reviewed in reference 51). The major E1A gene products, referred to as 12S (or 243R) and 13S (or 289R) proteins, associate with members of the pRb and p300 families of cellular proteins, thereby inducing cell cycle progression. However, E1A also causes programmed cell death (apoptosis), in part through metabolic stabilization and activation of the tumor suppressor protein p53. The major E1B proteins (E1B-19kDa and E1B-55kDa) individually block this apoptotic response, which at least partly explains their ability to cooperate with E1A in the transformation process (for review, see reference 63). The transforming properties of Ad5 E1B-55kDa involve direct binding to p53 (22, 50), repression of p53-activated transcription (66, 67) and maybe sequestration of the tumor suppressor protein in cytoplasmic bodies (1, 69, 70). Although expression of the E1A and E1B oncoproteins is sufficient to oncogenically transform cells, earlier work by Shiroki et al. already demonstrated that the in vitro growth properties and tumorigenicity of rat 3Y1 cells transformed by E1 gene products of highly oncogenic Ad12 are substantially enhanced by coexpression of Ad12 E4 (52). Further, it was shown that the E4 genes of nononcogenic Ad2 potentiate Ad2 E1-induced focus formation in CREF cells (42) and that the presence of E4 is not only a prerequisite but even the major determinant for the tumorigenic capacity of Ad9 (19, 58). The E4 function critical for tumor induction by Ad9 was genetically mapped to E4orf1, which by itself is able to transform CREF cells in vitro (20, 62). Recent evidence suggests that the Ad5 E4orf6/7 protein, which is known to interact with the S-phase-specific transcription factor E2F (16), may also positively or negatively influence oncogenesis by induction of transformation or p53-dependent apoptosis, respectively (65), while E4orf4 exclusively antagonizes cellular transformation through a unique p53-independent mechanism (25, 32, 53, 54).

Recently, we and others have shown that the Ad5 E4orf6 and E4orf3 gene products exhibit E1B-like transforming properties in that they can individually cooperate with Ad5 E1A to initiate focus formation in primary baby rat kidney (BRK) cells. Both E4 proteins also synergize with E1A plus E1B to increase the number and size of transformed foci (36, 38, 40). Moreover, E4orf6 confers profound morphological and growth-related changes on rat cells that additionally express the Ad5 E1 region, as opposed to E4orf3, which only modestly affects the transformed cellular phenotype (39, 40). Consequently, E4orf6 expression leads to increased tumorigenicity of these cells in nude mice and to dramatically enhanced malignancy of the corresponding tumors (36, 39). The oncogenic potential of E4orf6 most probably involves the physical interaction with p53 (7). Several activities of E4orf6 have been described that may be direct consequences of the interaction with the tumor suppressor protein. These include blockage of p53-mediated transcriptional activation and repression (7, 38), interference with suppression of focus formation by p53 (38), inhibition of p53-induced apoptosis, and subcellular redistribution of the tumor suppressor (36). Moreover, E4orf6 interferes with the E1A-E1B-55kDa-induced accumulation of p53 in both lytically infected human and adenovirus-transformed human and rat cells (12, 13, 36, 38, 39, 45, 47, 55). This effect is only apparent when E1B-55kDa is coexpressed with E4orf6 and may require complex formation between the two viral proteins (47, 55). E4orf6 and E1B-55kDa most likely target p53 for active degradation through the proteasome, thereby decreasing the half-life and consequently the steady-state levels of the tumor suppressor protein (36, 45, 55). We recently reported that the ability of E4orf6 to target p53 for degradation resulted in dramatically decreased p53 levels in BRK cells transformed by E1A, E1B, and E4orf6 (ABS cells) compared to cells that only express E1A and E1B (AB cells) (38). The p53 levels in different ABS cell lines inversely correlated with tumor growth rates in nude mice, indicating a relationship between p53 degradation by E1B-55kDa-E4orf6 and oncogenesis (39).

To confirm correlations between E4orf6-dependent p53 destabilization and oncogenic transformation and to investigate if transforming and lytic functions of the viral protein are related, we tested a set of E4orf6 mutant proteins for focus formation in cooperation with E1 genes. Subsequently, cell lines that stably express the E4orf6 variants were derived from these foci and tested for morphological alterations, steady-state levels, and subcellular distribution of p53, as well as tumor development in nude mice.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. Plasmid pAd5XhoI-C contains the left end of the Ad5 genome and expresses the E1A and E1B proteins (29). All other plasmids used in this study express wild-type or mutant E4orf6 under the control of the cytomegalovirus (CMV) immediate-early promoter. Plasmid pCMV-E4orf6 and all pCMV constructs expressing E4orf6 deletion mutants have been previously described (48). The amino-terminal deletion mutants dl1-55, dl1-108, and dl1-203 are fused to the influenza virus hemagglutinin (HA) epitope for immunological detection. For the construction of pCMV-E4orf6flu, which expresses an HA ("flu")-tagged full-length E4orf6 protein (Wt-HA), the E4orf6flu sequence was PCR amplified from template pE4orf6-flu (48) by using primers orf6fw (forward primer) and flufix (reverse primer) (48), which introduce BamHI and EcoRI sites, respectively. The PCR product was subsequently cloned into the BamHI and EcoRI sites of pcDNA3 (Invitrogen). Constructs expressing single- or double-amino-acid substitution mutants of E4orf6 were derived from pCMV-E4orf6flu (L90A/I92A, C124A/C126A, L245P, and R248E) or pCMV-E4orf6dl1-203 (dl1-203/L245P) by site-directed mutagenesis using the QuikChange Kit from Stratagene (oligonucleotide sequences available upon request).

Focus transformation assays and cell lines. Primary cells were obtained from kidneys of 6- to 7-day-old Sprague-Dawley rats, as previously described (38, 40). They were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS). For focus assays, subconfluent cells on 90-mm² dishes were transfected by the calcium-phosphate procedure (11) with the indicated plasmid amounts as previously described (38). Total DNA was adjusted to 20 µg with empty vector and salmon sperm carrier DNA (Boehringer Mannheim). Three weeks after transfection, plates were stained with crystal violet and dense foci of morphologically transformed cells were counted. Alternatively, foci were pooled and expanded into permanent cell lines. The transformed BRK cell lines AB7, ABS1, and ABS11 (which is identical to ABSN) have been described previously (38-40). The generation and identities of all other cell lines are described in the Results section. All BRK cell lines except AB7 were grown in DMEM with 10% FCS and 500 µg of G418 (Calbiochem) per ml. For AB7 cells, the same medium without G418 was used.

Immunoprecipitation and Western blotting. For analysis of protein expression by immunoprecipitation and Western blotting, cells were lysed in radioimmunoprecipitation buffer (50 mM Tris-chloride [pH 8.0], 150 mM NaCl, 0.1% sodium dodecyl sulfate [SDS], 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM phenylmethylsulfonyl fluoride, 0.3 µM aprotinin, 1 µM leupeptin, 1 µM pepstatin). The protein concentrations were normalized by using the Bio-Rad protein assay, and equal amounts of total protein were subjected to immunoprecipitation and/or Western blotting exactly as previously described (40). The following primary antibodies were used in this study. RSA3 and M45 are monoclonal antibodies directed against the amino terminus of the E4orf6 protein (33, 41). Monoclonal antibodies 12CA5 and 3F10 recognize the HA epitope and were obtained from Boehringer Mannheim. PAb421 reacts with p53 proteins from a broad range of mammalian species, including rats (14). Rat- and mouse-specific horseradish peroxidase-coupled secondary antibodies were obtained from Amersham, and chemiluminescence detection was performed with the NOWA kit from Energene. Luminograms were scanned, and figures were prepared by using Adobe Photoshop 5.0 and Microsoft Powerpoint 97 software.

Indirect immunofluorescence and phase-contrast microscopy. Cells were grown on glass slides to subconfluent densities. For immunofluorescence analyses, they were fixed with methanol for 10 min at 20°C and incubated with 5% FCS in phosphate-buffered saline (PBS) for 1 h at room temperature. After that, samples were reacted with undiluted hybridoma supernatants (M45, PAb421) or a 1:100 dilution of purified antibody (3F10), followed by appropriate fluorescein isothiocyanate (FITC)-conjugated secondary antibodies (Jackson). For studies on cell morphology, cells were washed once with PBS and examined without fixation. All samples were analyzed and photographed with a camera-mounted Olympus AX70 microscope. Slides were scanned, and figures were prepared by using Adobe Photoshop 5.0 software.

Tumor formation in nude mice. Cells with similar culture passage numbers were harvested with a cell scraper, washed twice with PBS, and resuspended in serum-free DMEM at a concentration of 10⁷ cells per ml. NMRI(nu/nu) mice were injected subcutaneously with 10⁶ cells, and tumor growth was recorded as previously described (39). For ethical reasons, mice were killed after the tumors reached a maximum size of approximately 100 mm² or 6 weeks after injection.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Design of E4orf6 mutant constructs. A series of previously described CMV-based mammalian expression constructs with carboxy- and amino-terminal truncation mutants derived from the full-length E4orf6 sequence was used for this study (48). Additionally, point mutations were introduced into the E4orf6 cDNA to allow expression of variant proteins with one or two nonconservative amino acid substitutions (Fig. 1). These substitutions were designed to affect three recently identified functional and/or structural motifs within the viral polypeptide, an NES, a conserved cysteine-rich (CCR) motif, and an arginine-faced amphipathic alpha helix that may contain an NRS.

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Schematic representation of E4orf6 mutant proteins used in this study. The 294-residue wild-type E4orf6 protein (Wt) is represented by the black bar at the top of the figure. Structural and functional motifs within the viral polypeptide (NES, CCR motif, and amphipathic alpha helix) are indicated. The deletion mutants (dl) are named according to the numbers of their first and last deleted residues and have been described in more detail previously (48). The single- or double-amino-acid substitution mutants are named by referring to the single-letter code of the original amino acid followed by a number indicating the position of the respective residue within the E4orf6 polypeptide and a letter representing the new amino acid. To the right, the names of the transformed rat cell lines expressing the respective wild-type or variant E4orf6 proteins are indicated (see also Table 1). The HA epitope at the carboxy terminus of some of the E4orf6 variant proteins is indicated. Wt, wild type.

View larger version (37K): [in this window] [in a new window]   FIG. 2.   A highly conserved, cysteine-rich motif within E4orf6 and related proteins. Primary sequence alignment of CCR motifs from polypeptides of different human and animal adenoviruses and one protein of cellular origin (AO7). Sequences were aligned by using the Clustal X program (gap penalty, 10; gap length penalty, 10) (59). Residues that fit the consensus exactly are boxed. Asterisks indicate the two cysteine residues in the HCHC tetrapeptide that were changed to alanines in the Ad5 E4orf6 sequence to create the double-point mutant C124A/C126A. HAV, human adenovirus; PAV, porcine adenovirus; BAV, bovine adenovirus; CAV, canine adenovirus; MAV, mouse adenovirus; OAV, ovine adenovirus; EDS, egg drop syndrome virus; AAV, avian adenovirus; CELO, chicken embryo lethal orphan virus. AO7 is a cellular protein that interacts with E2 enzymes of the ubiquitin-dependent proteolytic system (30).

Focus formation in cooperation with Ad5 E1 proteins requires the CCR motif and the amphipathic alpha helix of Ad5 E4orf6. The Ad5 E4orf6 protein cooperates with Ad5 E1A and E1A plus E1B genes to mediate focal transformation of primary BRK cells (36, 38). We have previously shown that a carboxy-terminal fragment of E4orf6 comprising amino acids 109 to 294 (dl1-108) is sufficient for this activity, while deletion of 143 amino acids from the carboxy terminus (dl152-294) abolishes the focus-promoting function of the viral protein (38).

View larger version (57K): [in this window] [in a new window]   FIG. 3.   Cooperative focus formation by Ad5 E1 genes and wild-type or mutant Ad5 E4orf6. (A) Primary BRK cells were transfected with 2.5 µg of pAd5XhoI-C and 1 µg of empty vector or plasmids encoding the indicated E4orf6 proteins. Colonies of morphologically transformed cells were scored 3 weeks after transfection. Focus-forming activity is represented as a percentage of the number of foci obtained by pAd5XhoI-C plus empty vector. The mean and standard deviation of at least three independent experiments are presented. (B) Representative crystal violet-stained plates showing foci from transfections with either just salmon sperm DNA (Carrier) or pAd5XhoI-C (E1) with empty vector or combinations of pAd5XhoI-C with plasmids encoding the indicated E4orf6-derived proteins.

Expression and subcellular localization of E4orf6 mutant proteins in cell lines derived from transformed foci. Pools of foci from individual plates transfected with pAd5XhoI-C and plasmids expressing E4orf6 variants were pooled, selected for G418 resistance, and expanded into permanent cell lines (ABS cells) (Table 1). Immunoprecipitation analysis showed that all established cell lines expressed the respective E4orf6 proteins, although to different levels (Fig. 4). Additionally, genomic DNA was obtained from ABS cell lines transfected with E4orf6 point mutant cDNAs, and E4 sequences were PCR amplified and subjected to DNA sequencing to confirm the presence of the corresponding nucleotide exchanges (data not shown).

View larger version (61K): [in this window] [in a new window]   FIG. 4.   Expression of wild-type and mutant E4orf6 proteins in transformed BRK cell lines. Equal amounts of total cell extracts were subjected to immunoprecipitation, and precipitates were subsequently resolved on SDS-15% polyacrylamide gels, transferred to nitrocellulose by Western blotting, and probed with the respective antibodies followed by enhanced chemiluminescence detection. Numbers correspond to the positions of marker bands. IgG, immunoglobulin G light chains. (A) Detection of wild-type E4orf6 and carboxy-terminal deletion mutants by immunoprecipitation and Western blotting using antibody RSA3. (B) Detection of HA-tagged amino-terminal deletion mutants by immunoprecipitation and Western blotting using antibodies 12CA5 and 3F10, respectively. (C) Detection of HA-tagged mutant dl1-203/L245P by immunoprecipitation and Western blotting using antibody 12CA5. (D) Detection of HA-tagged E4orf6 wild type and point mutants by immunoprecipitation and Western blotting using antibody 12CA5.

View larger version (140K): [in this window] [in a new window]   FIG. 5.   Subcellular localization of wild-type and mutant E4orf6 proteins in early-passage BRK cell lines. Indirect immunofluorescence microscopy using monoclonal antibody M45 (a, c, e, g, i) or 3F10 (k, m, o) was followed by anti-mouse or anti-rat FITC conjugates, respectively. Epifluorescence images and the corresponding phase-contrast micrographs are shown for the following cell lines expressing either only the Ad5 E1A and E1B proteins (AB7) or E1A or E1B plus the wild-type or variant E4orf6 protein given in parentheses (see Fig. 1 and Table 1 for details): AB7 (a, b), ABS1 (Wt) (c, d), ABS37 (dl224-294) (e to h), ABS26 (dl271-294) (i, j), ABS12 (dl1-55) (k, l), ABS19 (dl1-203) (m, n), and ABS52 (dl1-203/L245P) (o, p). The subcellular localization of all E4orf6 derivatives which are not shown was either undetectable (dl152-294, dl1-108) or undistinguishable from the wild-type staining pattern (Wt-HA and all point mutants). The arrowhead in panel g points to a perinuclear body. Magnification, ×500.

Morphological hypertransformation by the carboxy-terminal domain of E4orf6. We have previously shown that expression of Ad5 E4orf6 confers a number of striking, morphological changes on BRK cells that additionally express Ad5 E1 proteins (ABS cells) (39). When examined by phase-contrast microscopy, ABS cells appear much smaller, rounder, and less differentiated than cells expressing E1A and E1B only (AB cells). Moreover, they tend to form extremely dense, bizarrely shaped colonies rather than growing to confluency. These alterations in cell morphology become more profound during extended culture passages, resulting in cells that completely lose adherence and grow in suspension (39). Since it is known that combinatorial effects of E1B-19kDa and E1B-55kDa are sufficient for morphological transformation of rat cells in cooperation with E1A proteins (64), we refer to the additional, E4orf6-induced changes in cell size and shape as morphological "hypertransformation." Figure 6 shows the typical appearance of ABS cells expressing wild-type or HA-tagged E4orf6 (Fig. 6c and d) in comparison to primary BRK cells (Fig. 6a) and AB cells (Fig. 6b). Intriguingly, the carboxy-terminal 91 amino acids of E4orf6 (dl1-203) were sufficient to induce a hypertransformed cellular phenotype which was virtually identical to that of wild-type ABS cells (Fig. 6e). While the R248E mutant also displayed wild-type characteristics with respect to cell morphology (Fig. 6h), cells expressing full-length E4orf6 or dl1-203 containing the L245P mutation (Fig. 6g and f) closely resembled AB cells. These observations indicate that morphological hypertransformation depends on an intact alpha-helical structure within the carboxy-terminal domain of E4orf6.

View larger version (97K): [in this window] [in a new window]   FIG. 6.   Morphology of selected ABS cell lines in comparison to AB and primary BRK cells. Phase-contrast microscopy showing primary BRK cells (a) and the following BRK-derived cell lines expressing either only the Ad5 E1A and E1B proteins (AB7) or E1A or E1B plus the wild-type or variant E4orf6 protein given in parentheses (see Fig. 1 and Table 1 for details): AB7 (b), ABS1 (Wt) (c), ABS36 (Wt-HA) (d), ABS19 (dl1-203) (e), ABS52 (dl1-203/L245P) (f), ABS51 (L245P) (g), and ABS32 (R248E) (h). Magnification, ×100.

Multiple regions within E4orf6 contribute to the destabilization of p53. Several independent reports established that Ad5 E4orf6 efficiently antagonizes the accumulation of p53 that is mediated by the combined action of E1A proteins and E1B-55kDa (5, 31, 70). This effect is evident in lytically infected human as well as adenovirus-transformed human and rat cells and most likely occurs through metabolic destabilization of p53 at the protein level (12, 13, 36, 38, 39, 45, 47, 55). It appears that simultaneous expression of both E4orf6 and E1B-55kDa is necessary to target p53 for degradation (45, 55). To investigate which regions in the E4orf6 polypeptide are necessary for destabilization of the tumor suppressor protein and to examine a possible correlation between transforming functions of E4orf6 and p53 stability, we tested our transformed rat cell lines for p53 steady-state levels (Fig. 7). In accordance with previous observations, AB cells accumulated p53 to high levels due to the expression of E1A and E1B (38-40). In contrast, cells containing wild-type or HA-tagged E4orf6 in addition to the E1 proteins (ABS1 or ABS36 cells, respectively) exhibited dramatically reduced p53 levels. Similarly, ABS32 cells expressing E4orf6 mutant R248E accumulated only small amounts of the tumor suppressor protein. However, in all other cell lines, p53 levels were significantly elevated, suggesting that all E4orf6 variants except the R248E mutant do not efficiently interfere with p53 stability. As our cell lines are of polyclonal origin, it is unlikely that these results represent accidental variations between individual cell populations. However, for some of the most critical mutants (dl152-294, dl1-203, L90A/I92A, C124A/C126A, L245P, and R248E), we tested at least two separate cell pools for p53 steady-state levels and got very similar results. Although we cannot exclude the possibility that some of the mutants fail to show detectable effects on p53 levels due to weak expression, we do not think this is the case, as mutant R248E still efficiently interferes with p53 accumulation despite relatively low expression levels (Fig. 4 and 7).

View larger version (64K): [in this window] [in a new window]   FIG. 7.   Steady-state levels of p53 in transformed BRK cell lines. Equal amounts of whole-cell lysates were subjected to electrophoresis on SDS-10% polyacrylamide gels, transferred to nitrocellulose by Western blotting, and probed with monoclonal antibody PAb421 followed by a horseradish peroxidase-coupled anti-mouse secondary antibody and enhanced chemiluminescence detection. In addition to the p53-specific signal, a cross-reacting band is shown as a control for quantitative loading. For assignment of cell lines to the respective E4orf6 wild-type or variant proteins, refer to Fig. 1 and Table 1.

Expression of E4orf6 disrupts p53-containing perinuclear bodies and relocalizes p53 to the nucleus. To confirm and extend our Western blotting results on p53 steady-state levels, we additionally performed indirect immunofluorescence studies using monoclonal antibody PAb421 directed against the tumor suppressor protein. In accordance with previous observations, p53 accumulated in large, spherical perinuclear bodies in AB cells (Fig. 8Ab) (1, 39, 69, 70). In ABS cells expressing wild-type or HA-tagged E4orf6 (ABS1 or ABS36 cells, respectively) or mutant R248E (ABS32 cells), these aggregates were much smaller, reflecting the reduced p53 levels in these cells (Fig. 8Ac, d, and n). Consistent with our Western blotting results, all other cell lines accumulated p53 in large perinuclear bodies (Fig. 8Ae-m, o). However, the number and shape of these structures differed substantially between cell lines. Most of the cells expressing mutants dl152-294 (ABS14 [Fig. 8Ae]), dl224-294 (ABS37 [Fig. 8Af]), dl271-294 (ABS26 [Fig. 8Ag]), and C124A/C126A (ABS22 [Fig. 8Al]) contained single perinuclear bodies that morphologically resembled the ones in AB cells, in that they had a round shape. In contrast, in the majority of cells expressing dl1-55 (ABS12 [Fig. 8Ah]), dl1-108 (ABS11 [Fig. 8Ai]), dl1-203 (ABS19 [Fig. 8Aj]), L90A/I92A (ABS29 [Fig. 8Ak]), L245P (ABS51 [Fig. 8Am]), and dl1-203/L245P (ABS52 [Fig. 8Ao]), p53 localized to multiple perinuclear bodies that were reorganized to filamentous, worm-like structures. Moreover, in the presence of an intact amphipathic alpha helix, many of these reorganized aggregates localized to the nucleus and the carboxy-terminal 91-amino-acid fragment of the E4orf6 protein (dl1-203) were sufficient to mediate this relocalization (Fig. 8B).

View larger version (40K): [in this window] [in a new window]   FIG. 8.   Subcellular localization of p53 in transformed BRK cell lines. Antibody pAb421 followed by an anti-mouse FITC conjugate was used for all panels. (A) Indirect immunofluorescence microscopy showing primary BRK cells (a) and the following cell lines expressing either only the Ad5 E1A and E1B proteins (AB7) or E1A or E1B plus the wild-type or variant E4orf6 protein given in parentheses (see Fig. 1 and Table 1 for details): AB7 (b), ABS1 (Wt) (c), ABS36 (Wt-HA) (d), ABS14 (dl152-294) (e), ABS37 (dl224-294) (f), ABS26 (dl271-294) (g), ABS12 (dl1-55) (h), ABS11 (dl1-108) (i), ABS19 (dl1-203) (j), ABS29 (L90A/I92A) (k), ABS22 (C124A/C126A) (l), ABS51 (L245P) (m), ABS32 (R248E) (n), and ABS52 (dl1-203/L245P) (o). Magnification, ×300. (B) Combined phase-contrast and epifluorescence microscopy showing AB7 (a), ABS19 (b), and ABS52 (c) cells. Magnification, ×500.

Destabilization of p53 contributes to E4orf6-induced acceleration of tumor growth. We and others previously reported that BRK cells expressing Ad5 E1 and E4orf6 proteins (ABS cells) exhibit increased tumorigenicity compared to AB cells when subcutaneously injected into athymic mice (36, 39). More significantly, ABS-derived tumors displayed dramatically accelerated malignant tumor growth in these mice when compared with tumors induced by AB cell injections (39). Interestingly, p53 levels in different BRK cell lines were inversely correlated with E4orf6 expression and the respective tumor growth rates, indicating that p53 destabilization significantly contributes to the increased oncogenicity of ABS cells (39). To further investigate this issue, we tested our transformed rat cell lines expressing E4orf6 derivatives for tumor development in nude mice (Table 2 and Fig. 9). All ABS cell lines were tumorigenic, although to different extents, as opposed to AB7 cells, which did not form tumors in nude mice (Table 2). As expected, tumors derived from cells expressing wild-type or HA-tagged full-length E4orf6 (ABS1 and ABS36 cells, respectively) developed extremely rapidly, resulting in tumor sizes of over 200 or 100 mm² 12 or 19 days after injection, respectively (Fig. 9A and B). Among the mutants, only the R248E variant (ABS32 cells) showed wild-type characteristics with respect to promotion of tumor growth (Fig. 9B). Besides, ABS19 cells expressing the carboxy-terminal 91-amino-acid fragment of E4orf6 (dl1-203) displayed significantly enhanced tumor development compared to all other deletion mutants, although clearly less pronounced than cells expressing wild-type E4orf6 (ABS1 cells) (Fig. 9A).

View larger version (21K): [in this window] [in a new window]   FIG. 9.   Growth kinetics of tumors derived from different AB and ABS cell lines. Mean tumor areas of 5 to 8 NMRI(nu/nu) mice per cell line (Table 2) monitored during a 35-day period after injection are shown. Tumor areas were calculated as the product of two perpendicular diameters, one measured across the greatest width of the tumor. For assignment of cell lines to the respective E4orf6 wild-type or variant proteins, refer to Fig. 1 and Table 1. (A) Growth of tumors formed by ABS cells expressing E4orf6 deletion mutants compared to AB and wild-type ABS-derived tumors. (B) Tumor growth by E4orf6 point mutant-derived ABS cells compared to the respective control cell lines.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The present study represents the first comprehensive mutational analysis of the Ad5 E4orf6 protein. Our results show that the transforming properties of Ad5 E4orf6 involve at least two distinct, genetically separable activities of the viral oncoprotein. Our data indicate that the ability of E4orf6 to antagonize p53 stability in conjunction with E1B-55kDa and its tumor growth-promoting potential are related. Both activities map to multiple regions within E4orf6, including the extreme amino terminus, the NES, a CCR motif, an arginine-faced amphipathic alpha-helix, and the extreme carboxy terminus (Fig. 10). In contrast, promotion of focus formation in cooperation with Ad5 E1 proteins and induction of morphological hypertransformation by E4orf6 are completely independent of p53 stability. These transforming functions require exclusively the carboxy-terminal part of E4orf6, which contains the CCR motif and the amphipathic alpha helix, the latter being most critical in this respect (Fig. 10).

View larger version (21K): [in this window] [in a new window]   FIG. 10.   Summary of functional domains within the Ad5 E4orf6 protein. The 294-residue wild-type E4orf6 protein, including the NES, a CCR motif, and an amphipathic alpha helix, is represented by the wide bar at the top. Regions required for the indicated functions are shown as narrow bars, according to their positions along the E4orf6 polypeptide. Hatched bars correspond to activities that have been previously mapped (6, 13, 38, 43, 48), while functions represented by black and white bars were allocated in the present study. Black bars indicate protein segments which are essential, whereas white bars represent regions which have an auxiliary role for the respective function. At the bottom, the proposed E4orf6 oncodomain, spanning amino acids 204 to 294, is shown. AAV, adeno-associated virus.

Interestingly, a segment comprising only 91 carboxy-terminal amino acids of E4orf6, including the amphiphathic helix, is independently active for almost all of the functions that we have investigated in the present study. It induces morphological hypertransformation, reorganizes the perinuclear body, and relocalizes p53 to the nucleus. Moreover, it can promote focus formation and tumor growth, although less efficiently than the wild-type protein. These observations led us to refer to the region spanning amino acids 204 to 294 as the E4orf6 oncodomain (Fig. 10). This segment may represent a true protein domain, because it contains a functional protein structure (the amphipathic alpha helix) and it is stably expressed and properly targeted to the nucleus by an NLS intrinsic to the alpha helix. In this context, it is noteworthy that our results on the subcellular localization of E4orf6 are in complete agreement with previous observations by Orlando et al. (43). Our data indicate that E4orf6 contains at least one additional NLS outside the helix structure. A basic, arginine-rich sequence between amino acids 13 and 29 of the viral protein (RPTRSRLSRRTPYSRDR) is a likely candidate for such a nuclear targeting sequence.

In spite of its transforming potential, coexpression of the E4orf6 oncodomain with E1B-55kDa is apparently not sufficient to target p53 for degradation. Instead, multiple regions within the E4orf6 polypeptide are required for this effect (Fig. 10). It has been proposed that complex formation between E1B-55kDa and E4orf6 may be a prerequisite for p53 destabilization (47). We have previously mapped the physical interaction with E1B-55kDa to the extreme amino terminus of E4orf6 (amino acids 1 to 55) (48). This may explain why deletion of amino acids 1 to 55 prevents p53 destabilization (Fig. 7, ABS12). Besides, a functional interaction between the E4 protein and E1B-55kDa has been shown to be mediated through the amphipathic alpha helix of E4orf6. This interaction is essential for E4orf6 to target the E1B protein to the nucleus and perhaps for E1B-55kDa to relieve nuclear retention of the E4 protein (6, 43). Despite these evidences for a functional E1B-E4 interaction via the alpha helix, no physical binding involving this structure has been demonstrated yet. Nevertheless, our analyses with mutant L245P show that the helix integrity is essential for E4orf6-dependent p53 degradation as well as for all transforming activities of the viral oncoprotein, including cooperative focus formation, promotion of tumor growth, and morphological hypertransformation (Fig. 10). Thus, the arginine-faced amphipathic alpha helix, in addition to being critical for E4orf6 function during adenovirus lytic growth (43), is crucial for the oncogenic potential of the viral protein, suggesting that lytic and transforming functions of E4orf6 may be linked. In addition, the ability of the E4 protein to augment transduction of recombinant adeno-associated virus has also very recently been mapped to the same region (amino acids R243 and L245), emphasizing its importance for different aspects of E4orf6 function (Fig. 10) (13).

As opposed to the E4orf6 variant carrying the L245P substitution, point mutation of the putative NRS within the amphipathic alpha helix (R248E) had no obvious effect on E4orf6-dependent p53 degradation, and this mutant exhibited wild-type characteristics in all of our assays. These observations indicate that the R248E mutation does not adversely affect the helix structure. In contrast, mutation of the NES (L90A/I92A) interferes with p53 degradation and the tumor growth-promoting activity of E4orf6 (Fig. 10). This finding suggests that nuclear export of E4orf6 may be necessary for its effect on p53 stability. Similarly, degradation of the tumor suppressor via the hdm2-dependent feedback loop requires shuttling of hdm2, and p53 destabilization by the human papillomavirus (HPV) E6 protein depends, at least in part, on the nuclear export of the viral polypeptide (8, 46, 57). However, preliminary experiments employing the nuclear export inhibitor leptomycin B indicate that shuttling of E4orf6 is not required for p53 destabilization and results from heterokaryon assays suggest that wild-type E4orf6 is not at all exported from the nuclei of ABS cells (S. Rubenwolf, unpublished data). Thus, the NES mutation may interfere with unidentified E4orf6 functions which are unrelated to nuclear export.

The HCHC motif of E4orf6 appears to be also involved in p53 destabilization and some aspects of transformation. This tetrapeptide is part of a cysteine-rich sequence which has been previously noted to be conserved in proteins of many, if not all, human and animal adenoviruses (Fig. 2) (61). Even the avian adenovirus CELO, which has been reported to lack sequences homologous to E4 region genes of human adenoviruses except E4orf1 (4), encodes a predicted polypeptide (ORF14) that contains a CCR-like motif and may thus represent an E4orf6-related protein (Fig. 2). Very recently, the RING finger, a cysteine-rich, zinc-chelating protein motif, has been demonstrated to mediate binding to ubiquitin-conjugating enzymes (E2s) and ubiquitination of several otherwise unrelated proteins (21, 30, 37, 68). The region surrounding the HCHC tetrapeptide of E4orf6, although no bona fide RING finger, has similarities to RING motifs. Moreover, the E2-interacting human RING finger protein AO7 shares the HCHC tetrapeptide with E4orf6 (Fig. 2). Strikingly, mutation of the two cysteine residues within the HCHC sequence (C158 and C160) leads to the loss of E2 binding and ubiquitination of AO7 (30). Thus, it is tempting to speculate that E4orf6 and related proteins from different adenoviruses interact with E2 enzymes through their CCR motifs, leading to self-ubiquitination and/or ubiquitination and degradation of other proteins, such as p53. In sum, our data provide evidence that multiple protein interactions are required for E1B- and E4-mediated p53 degradation, at least in rat cells.

Despite the fact that p53 destabilization appears to be a major determinant of the oncogenic potential of E4orf6, our data clearly show that other activities contribute to the transforming properties of the viral protein. This conclusion has been anticipated by studies on the immortalizing and tumorigenic functions of the HPV-16 E6 gene product, the paradigmatic p53-destabilizing viral protein (e.g., as described in references 18 and 28). Tumor growth promotion by E4orf6 appears to have a component which is independent of p53 stability. More importantly, the ability of E4orf6 to promote focus formation in cooperation with Ad5 E1 genes and its effects on cell morphology are completely independent of p53 stability. These activities of E4orf6 may therefore involve other interactions with p53. It is possible that the relief of p53-mediated transcriptional repression contributes to the promotion of focus formation, since we mapped both activities of E4orf6 to the carboxy-terminal region between amino acids 109 and 294 (Fig. 10) (38). Moreover, some aspects of E4orf6-induced transformation may involve the relocalization of p53 from the cytoplasm to the nucleus, which has been observed in both human 293 cells (36) as well as in rat ABS cells (Fig. 8B). Our data show that the amphipathic alpha helix of E4orf6 confers nuclear localization on p53. The fact that nuclear targeting of E1B-55kDa also requires the alpha-helical motif indicates that both the cellular tumor suppressor protein and the viral E1B protein are simultaneously relocalized to the nucleus by E4orf6. This effect may be due to a change in the steady-state distribution of the affected proteins rather than a true relocalization, as both E1B-55kDa and p53 are now known to shuttle between the nucleus and the cytoplasm (24, 34). It seems contradictory that the nuclear localization of p53 may be related to oncogenesis, since the nucleus is supposed to be the compartment where p53 performs its tumor suppressor functions. However, it has been shown that E1-transformed 3Y1 rat cells containing low steady-state levels of E1B-55kDa form tumors in nude mice more rapidly than cells with high levels of the viral oncoprotein (60). Concomitantly, 3Y1 cells with low E1B levels accumulate p53 in the nucleus rather than in cytoplasmic bodies, just as we have observed for cells expressing the carboxy-terminal domain of E4orf6 (ABS19 cells) (60).

Alternatively or additionally, the oncogenic potential of E4orf6 may involve functions which are completely p53 independent. In this context, it is noteworthy that the E4orf6 protein has been demonstrated to bind to several cellular proteins other than p53 (2, 15). While the identity of most of these proteins is still unknown, one has been uncovered as the p53-related protein p73 (15). In contrast to other p53-binding viral oncoproteins like simian virus 40 large T antigen and Ad5 E1B-55kDa, E4orf6 has been demonstrated to physically associate with p73 and to block transcriptional activation and cell killing mediated by the p53 homologue (15, 57). However, E4orf6 cannot target p73 for degradation in conjunction with E1B-55kDa (47, 57), and it is not clear if p73 has true tumor suppressor functions like those of p53. Very recently, it has been reported that E4orf6 and E4orf3 are physically associated with the catalytic subunit of the DNA-dependent protein kinase (DNA-PK) and that E4orf6 inhibits double-stranded break repair (3). This finding is highly intriguing, because it suggests an unanticipated role for E4orf6 (and E4orf3) in DNA repair and the DNA damage response. Moreover, both E4 proteins have also recently been implicated in cell cycle regulation during viral infection (9, 13), and a putative cyclin-binding RXL motif has been identified within the amphiphathic alpha-helical region of E4orf6 (13). It remains to be determined if interactions with p73, DNA-PK, cyclins, or other cellular proteins involved in DNA repair, cell cycle control, or protein degradation contribute to the transforming and oncogenic functions of E4orf6.