RACK1 interacts with E1A and rescues E1A-induced yeast growth inhibition and mammalian cell apoptosis.

The adenoviral E1A proteins are able to promote proliferation and transformation, inhibit differentiation, induce apoptosis, and suppress tumor growth. The extreme N terminus and conserved region one of E1A, which are indispensable for transcriptional regulation and for binding to p300/CBP, TBP, and pCAF, play essential roles in these abilities. These observations strongly suggest an intrinsic link between E1A-mediated transcriptional regulation and other effects. In this report, we show that E1A inhibits the normal growth of Saccharomyces cerevisiae HF7c, and this inhibition also depends on the domains required for transcriptional regulation. We demonstrate that E1A associates with histone acetyltransferase activity and represses the transactivation activity of transcription factor in S. cerevisiae, suggesting that E1A may suppress the expression of genes required for normal growth. Based on yeast growth rescue, we present a genetic screening strategy that identified RACK1 as an E1A antagonizing factor. Expression of human RACK1 efficiently relieves E1A-mediated growth inhibition in HF7c and protects human tumor cells from E1A-induced apoptosis. Finally, we show that RACK1 decreases E1A-associated histone acetyltransferase activity in yeast and mammalian cells, and physically interacts with E1A. Our data demonstrate that RACK1 is a repressor of E1A, possibly by antagonizing the effects of E1A on host gene transcription.

The adenovirus E1A proteins are functionally important for viral infection and replication. The most important role they play is preparation of a favorable environment for viral replication by reprogramming the host cell processes. E1A proteins have been implicated in the promotion of cell proliferation, transformation, and inhibition of differentiation of certain cell types (1)(2)(3)(4). Because E1A transforms cultured rodent cells when co-transfected with another oncogene, such as Ras or E1B, it has been considered an oncogene (1,3,(5)(6)(7). E1A also has been implicated in induction of apoptosis through either p53-dependent or p53-independent pathways (8 -11), and in suppression of tumor growth (12)(13)(14)(15)(16).
Recently, we showed that the extreme N terminus (ENT) 1 (aa  and the conserved region one (CR1) (aa 40 -80) of E1A possess transactivation activity in both yeast and mammalian cells (32). In addition, we observed that these regions are able to regulate host cell gene expression, either positively or negatively, in a pRb/E2F-independent manner. The mechanisms underlying the transcriptional regulation may involve the direct interaction with p300/CBP, two nuclear histone acetyltransferases (33,34), and with components of the basal transcriptional machinery, such as TBP (32,35). Furthermore, we reported that the extreme N terminus of E1A interacts with unidentified cellular proteins (36). More recently, it has been shown that these regions directly interact with pCAF, another histone acetyltransferase (26). Interestingly, these regions also are required for E1A-induced apoptosis (37,38) and for tumor suppression (16) in mammalian cells, suggesting that E1A induces apoptosis and tumor suppression by disrupting the regulation of host gene expression.
Yeast is used widely as a model system to study cell growth regulation, DNA replication, and transcriptional regulation, since most of the essential regulatory mechanisms are well conserved between yeast and mammals (39 -41). Genetic complementation approaches in yeast have been used to identify a variety of mammalian genes involved in growth regulation. In addition, identification of the basal transcriptional complexes reveals astonishing similarities between yeast and * This work was supported in part by Associazione Italiana per La Ricerca Sul Cancro and Ministero della Sanità grants (to M. G. P.), by a Federazione Italiana per La Ricerca Sul Cancro grant (to A. D. L.) and by National Institutes of Health Grants RO1 CA 60999 -01A1, P01-CA 56304-08, and PO1 NS 36466 and the Sbarro Institute for Molecular Medicine (to A. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  mammals, providing strong evidence that transcriptional regulation is a highly conserved process between evolutionarily distant species. Particularly, the yeast system has been successfully used for functional and genetic analysis of E1A effects (42,43).
We investigated here the E1A-mediated growth inhibition in yeast cells and the roles of the ENT and CR1 regions. We found that the domains required for yeast growth inhibition also are required for transcriptional regulation, apoptosis, and tumor suppression in mammalian cells. We also found that E1A associated with a HAT activity in yeast, and that expression of E1A repressed transactivation activity of yeast transcription factors. Based on the rescue of yeast growth, we designed a genetic screening strategy and found that human RACK1 successfully antagonizes E1A-mediated growth inhibition in yeast. Finally, we observed that co-transfection of RACK1 protects human tumor cells from undergoing E1A-induced apoptosis and that RACK1 physically interacts with E1A in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Oligonucleotides, Plasmids, and DNA Recombination-Oligonucleotides used as primers for PCR were synthesized by the Nucleic Acid Facility at Kimmel Cancer Institute, Thomas Jefferson University. The pGBT9, pGBKT7, pGAD424 vectors, and pCL1 were purchased from CLONTECH Laboratories. The pEG202 vector and pSH18 -34 (URA3, LexAop-LacZ reporter plasmid) were described previously (32). Plasmids pcDNA3 and pcDNA3T7Tag were obtained from Invitrogen. DNA sequences of PCR-generated fragments were confirmed by sequencing.
The wild-type 243R E1A fusion construct and most of the E1A mutant constructs in pGBT9 were as previously described (32,44). To express 243R E1A alone in yeast, a KpnI-PstI fragment of 243R E1A was generated by PCR. This fragment was then used to replace the KpnI-PstI fragment encoding the GAL4 transactivation domain in the pGAD424 vector, thus generating the construct pG-E1A. The LexA-E1A fusion constructs, pEG202-E1A and pG-LexA-E1A with a Trp selection marker were described previously. To express LexA alone in HF7c, the E1A part was removed by restriction digestion with EcoRI and PstI followed by blunting and re-ligation that resulted in pG-LexA. To express 243R E1A in EGY48 the smaller SnaBI-PstI fragment from pG-E1A was used to replace the smaller SnaBI-PstI fragment in pGBT9, and to generate pE1A. Plasmid pGBT9-Rb2/p130 was constructed by PCR. Plasmids pGBT9-Rb and pGBT9-p107 were described previously (32). Human RACK-1 cDNA was obtained by reverse transcriptase-PCR using total RNA from the 293 cell line as a template. Primers were 5Ј-GATCCCCGGGCATGACTGAGCAGATGACC-3Ј and 5Ј GATCCTG-CAGCTAGCGTGTGCCAATGGTC-3Ј. The amplified fragment was then cloned into a pGAD424 vector fused to GAL4 activation domain.
For mammalian cell transfection assays, all the cDNAs encoding E1A, E1A mutants, DBD, DBD-E1A, DBD-E1A mutants, GST, GST-E1A, and GST-E1A mutants were driven by the cytomegalovirus promoter. The plasmid pON260 that constitutively expresses full-length GAL4 as a control of transfection efficiency was described previously (32). were purchased from CLONTECH Laboratories and the EGY48 strain (MATa, ura3, his3, trp1, LexAop-LEU2) was described previously (32). All the yeast strains were maintained at 30°C on YPD plates. Yeast transformation was performed with the lithium acetate (LiOAc) method described previously (32). Transformants were selected and maintained on synthetic medium lacking selective amino acid(s).
Yeast Growth Rate Analyses-The yeast transformants were grown in 10 ml of SD without tryptophan or other auxotrophic amino acids depending on the selective marker of the plasmids, at 30°C with shaking (270 rpm) to stationary. The liquid culture was diluted to 0.2 (around 0.5-0.6 ϫ 10 6 cells/ml) A 600 in 10 ml of fresh SD without selective amino acids or 10 ml of YPD when transformants harboring plasmids with different selective markers were tested together to prevent variations caused by using different SD medium. The freshly inoculated culture was put back to incubate at 30°C with shaking (270 rpm). One ml of sample was taken from each culture to measure A 600 at various time points post-inoculation. The cell growth rate was measured as the ratio of the A 600 at specific time points to the initial A 600 (Growth Index). To assay the viability of the yeast transformants, 100 l of the original freshly inoculated culture were used to make sequential dilutions (10 Ϫ2 , 10 Ϫ3 , and 10 Ϫ4 ) and were plated out on 150-mm plates with SD-dropout agar and cultured at 30°C for 4 days to monitor colony formation.
Yeast ␤-Galactosidase Assay-Transformants were grown overnight in 2 ml of proper synthetic medium at 30°C with shaking. After adding 8 ml of fresh SD, yeast cells were allowed to grow for a further 4 h. After measuring the A 600 , 1.5 ml of culture was collected and washed once with Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 10 mM MgSO 4 , pH 7.0). After being re-suspended in 300 ml of Z buffer, 100 ml was transferred into a fresh sterile tube. ␤-Galactosidase activity was measured as described before (32) and corrected for cell density (A 600 ) and for the reaction time. The relative activity of the activators was compared with pGBT9 vector alone and expressed as relative transactivation activity.
Cell Lines, Tissue Culture, and Transfections-SAOS-2 (ATCC, human osteosarcoma) and HeLa (ATCC, human cervix carcinoma) cells were maintained in Dulbecco's modified Eagle's medium supplemented with L-glutamine and 10% fetal bovine serum. For transient transfection experiments, cells were seeded 24 h before the transfection to obtain a 50 -70% confluence. Cells were transfected by either Transfectam (Promega) or by standard calcium phosphate protocols. Transfection efficiency was monitored by measuring ␤-galactosidase activity derived from the co-transfected pON260 plasmid (32). To use Transfectam, the instructions for serum-free transfection procedures provided by the manufacturer were followed. Three hours after transfection, fresh medium with serum was added, and the cells then were cultured for 48 h before harvest. When using calcium phosphate transfection, 18 h after transfection, the medium was changed and cells were cultured for an additional 24 h. Transfection efficiency obtained from this method was around 20 -25%. For stable transfection, the calcium phosphate method was used to transfect 10 g of linearized plasmids. Forty-eight hours after transfections, the cells were trypsinized and selected in G418 medium for 4 weeks. Cell labeling with [ 35 S]methionine and GST pull-down assays were performed as previously described (36).
Cell Lysate Preparation, Immunoprecipitation, and Western Blotting-Yeast protoplasts were prepared by standard methods (45). Briefly, overnight grown yeast culture was washed three times with ice-cold phosphate-buffered saline. After being incubated for 10 min at 30°C with stabilization buffer A (1 M sorbitol, 10 mM MgCl 2 , 2 mM dithiothreitol, 50 mM potassium phosphate, pH 7.8, 100 g/ml phenylmethylsulfonyl fluoride, 10 g/ml leupeptin), the yeast was incubated at 30°C for 5 min with stabilization buffer B (1 M sorbitol, 10 mM MgCl 2 , 2 mM dithiothreitol, 25 mM potassium phosphate, pH 7.8, 25 mM sodium succinate, pH 5.5, 100 g/ml phenylmethylsulfonyl fluoride, 10 g/ml leupeptin). Protoplasts then were generated by enzymatic digestion with 25 units/ml lyticase (Sigma) for 15 min at room temperature in stabilization buffer B. The protoplasts were washed with phosphatebuffered saline three times and lysed in IPH buffer as previously described (26,33). Cultured human cells were directly lysed in IPH buffer as described. For immunoprecipitation, lysates containing 250 g of protein were first incubated with a nonspecific mouse serum and cleared by mixing with killed whole Staphylococcus aureus cells (Life Technologies, Inc.). The cleared supernatant was incubated with anti-E1A monoclonal antibody and immunoprecipitation was performed as previously described to obtain E1A complexes (35). The anti-E1A monoclonal antibodies M73 and M37 were described elsewhere (46,47). E1A complexes then were separated by SDS-PAGE followed by Western blotting analysis, or used for HAT assays. The procedure for Western blotting was described previously (32,35).
Histone Acetyltransferase Assays-HAT assays were performed with the HAT-check kit (Pierce). Briefly, the immunoprecipitated complexes were incubated with a synthetic substrate corresponding to the first 23 amino acids of histone H4 and coupled to biotin at 30°C for 1 h with under control of the T7 polymerase in the presence of [ 35 S]methionine. Aliquots of the reaction mixture were added to glutathione-Sepharose beads coupled with 4 g of GST chimerized with E1A. Incubation was carried out in NENT buffer (20 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml leupeptin) for 60 min at 4°C with gentle rocking. Beads were washed three times in NENT buffer, and then electrophoresis was performed on SDS-PAGE. Gel was dried and then exposed at Ϫ70°C using Kodak Bio-Max MS film with DuPont Cronex intensifying screens.
TUNEL Assays-After transfection, cells were fixed in paraformaldehyde, washed in distilled water, and exposed briefly to 3% H 2 O 2 to inactivate endogenous peroxidase. The TUNEL reaction was performed using the peroxidase-based Apoptag kit (Oncor). The TUNEL positive cells were revealed by means of diaminobenzidine and H 2 O 2 , according to the supplier's instructions. Finally, stained cells were slightly counterstained with hematoxylin. TUNEL positive cells were considered as apoptotic.

RESULTS
E1A Inhibits the Growth of the HF7c Yeast Strain-We previously reported that 243R E1A possesses transactivation activity when fused to a DNA-binding domain in yeast and in mammalian cells (32). In that study, we consistently observed that expression of wild-type E1A in yeast strain HF7c led to a slow-growing phenotype. To further understand the E1A-mediated inhibition of yeast growth, we investigated this phenomenon in a carefully controlled study. As shown in Fig. 1A, expression of E1A led to growth inhibition of HF7c. As controls, expression of the RB family gene products (pRb, p107, or pRb2/ p130), a family of growth inhibitors in mammalian cells (23) showed no effect on yeast growth. The fusion of E1A with the DNA-binding domain (DB) of GAL4 was not essential for this effect, because expression of E1A alone, or when fused with either the transactivation domain (TAD) of GAL4 or LexA, resulted in similar inhibition (Fig. 1B). Similar studies were performed in the EYG48 and PJ69-2A strains. Significant growth inhibition was observed in the EYG48 strain, but expression of E1A inhibited the growth of PJ69-2A strain only slightly (data not shown), suggesting that E1A-mediated yeast growth inhibition is a strain-specific event.
Mapping the E1A Domains Involved in Yeast Growth Inhibition-243R E1A is a multifunctional oncoprotein consisting of several well defined functional domains (48). The definition of a solid link between the specific domains and the inhibitory activity is essential for understanding the underlying mechanism and the biological significance in the virus-host interaction. We used an extensive set of E1A mutants to express mutant E1A proteins in yeast cells and compared the effects of these mutants on yeast growth with that of wild-type E1A. The mutant E1A constructs were characterized previously (44) and the mutant protein expression in yeast was confirmed by Western blot analysis (Fig. 2). Two monoclonal antibodies against different epitopes of E1A were used to detect E1A proteins encoded by these constructs (46). As shown in Fig. 2, A and B, wild-type E1A was recognized by both the M73 and M37 antibodies. Mutants were detected by M73, M37, or both, depending on the existence of intact epitopes. The scheme in Fig. 2C represents the structure of the E1A mutants employed, with the position of the epitopes recognized by M73 and M37 indicated.
The data in Fig. 3 show that constitutive expression of 243R E1A in the HF7c yeast strain resulted in a markedly slower growth rate, as did the expression of E1A mutants containing aa 1-228 or 1-120. However, deletion of the N-terminal 70 or 100 aa completely abolished E1A-mediated growth inhibitory activity. While deletion of the major part of CR1 (⌬38 -67) resulted in a slightly lower growth rate, deletion of ENT, while keeping CR1 intact, also led to a loss of inhibition. We concluded that the growth inhibitory domain is contained within the N-terminal aa 1-70 region and must involve aa 38 -67. The slow growth was not caused by a decreased viability of yeast transformants, because the numbers of colonies formed on plates did not differ significantly (data not shown).
E1A-induced Growth Inhibition Was Associated with Transcriptional Inhibition and HAT Interaction-In mammalian cells, E1A proteins function as dual regulators of gene expression. They stimulate the transcription of certain genes, but repress others (1,3,49). The ENT and CR1 regions play roles in both activities (26,32,50). In yeast, while the domains involved in growth inhibition possess transactivating activity when recruited to a promoter (32,50), E1A-mediated growth inhibition is not caused by the stimulation of GAL4-responsive genes as shown above. Therefore, we tested the hypothesis that overexpression of E1A might actually disrupt transcriptional regulation in yeast cells by sequestering cellular factors. Taking advantage of the reporter system engineered in the HF7c yeast strain, we co-transformed pCL1, a plasmid expressing full-length GAL4, with either E1A or E1A mutants. As shown in Fig. 4A, expression of full-length GAL4 strongly activated the reporter gene, measured as ␤-galactosidase activity. E1A efficiently inhibited full-length GAL4 transactivation activity. However, E1A mutants lacking aa 1-70 failed to repress GAL4 transactivation activity, suggesting that the inhibition of yeast growth might be caused by the blockage of transcription of certain host genes that are required for normal growth.
The molecular mechanism underlying E1A-mediated tran-

RACK1 Antagonizes E1A-induced Apoptosis and Growth Inhibition
scriptional regulation in yeast is not clear. In mammalian cells, the N terminus of E1A is involved in interaction with p300, CBP, and pCAF, three nuclear histone acetyltransferases (26,33,34). We looked to see if the N terminus of E1A could interact with a similar HAT activity in yeast. As shown in Fig. 4B, wild-type 243R E1A complex was immunoprecipitated from yeast cells and showed significant HAT activity. Deletion of the N terminus abolished this activity, while deletion of aa 190 -226 had no effect. Western blot analysis revealed that wild-type E1A was expressed at a lower level than in the deletion mutants (Fig. 4C). The similarity of E1A effects on yeast and mammalian cells, with respect to HAT interactions, supports the premise that E1A has an important function that modifies the regulation of host gene expression.
RACK1 Rescues 243R E1A-induced Growth Inhibition in Yeast Cells-Based on the fact that E1A inhibited HF7c growth and resulted in a slow-growing phenotype, we adopted a rescue screening strategy using the yeast system. The hypothesis was that overexpression of a mammalian gene product in yeast might be capable of functionally preventing E1A-mediated growth inhibition. Identification of such genes could greatly facilitate understanding the effects of E1A on host cells. A  4. E1A inhibits the activity of yeast transcription factor and associates with a HAT activity in yeast cells. A, pCL1 was co-transformed with either wild-type E1A or E1A mutants and selected on SD-Trp-Leu agar plates. Colonies were grown in SD-Trp-Leu media in triplicate at 30°C overnight with vigorous shaking. The transcription activity of full-length GAL4 (expressed from pCL1) was measured as ␤-galactosidase activity. E1A and E1A deletion mutants were coexpressed as indicated. Average values and S.D. from three independent experiments are presented. B, yeast transformants were cultured in SD dropout medium and total yeast cell lysates were prepared. Cell lysates, containing 250 g of protein were immunoprecipitated with M73. Half of the immunoprecipitates were used for HAT assays and the other half was used for Western blotting analysis with M73 (C).

FIG. 2. E1A mutants and their expression in yeast.
Western blotting assays with M73 (A) and M37 (B). Total yeast cell lysates were prepared as described under "Experimental Procedures." Twenty g of protein from total lysates were separated on duplicate 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes. The membranes were then detected with either M73 or M37 and developed with ECL. C, schematic structure of the E1A mutants. Epitopes recognized by the M73 and M37 monoclonal antibodies are indicated. yeast expression cDNA library (in plasmids with selective marker Leu) derived from HeLa cells was co-transformed with pGBT9-E1A (Trp) into the HF7c yeast strain and transformants were selected on SD-agar (-Trp, -Leu, -His). Fast-growing colonies appeared after 3 days. We kept the incubation for a maximum of 15 days and picked the 80 largest colonies for analysis. After the recovery of the plasmids in HB101 Escherichia coli and subsequent sequencing analysis, we found that four of the rescued plasmids carried overlapping fragments from a single cDNA. GenBank TM search with the GCG Wisconsin package revealed that they were derived from a known cDNA encoding human RACK1 (GenBank TM access number M24194). Using a pair of primers (see "Experimental Procedures"), we isolated the full-length coding region of RACK1 by reverse transcriptase-PCR.
To confirm the effects of RACK1 on E1A-mediated growth inhibition in yeast, we next co-transformed pGBT9-E1A with RACK1 expression construct into the HF7c yeast strain. As shown in Fig. 5, expression of 243R E1A in HF7c reduced yeast cell growth rate remarkably. This effect was overcome by coexpression of RACK1, which significantly rescued the 243R E1A inhibitory effect on cell growth.
RACK1 Effects on E1A-associated HAT Activity in Yeast and Mammalian Cells-We next wanted to determine if the rescuing effect of RACK1 on E1A-induced growth inhibition in yeast could be correlated to the E1A-HAT interaction. We found that coexpression of RACK1 in yeast cells reduced the amount of E1A-associated HAT activity in yeast (Fig. 6A). Because E1A interacts with pCAF in human cells, we also tested the effect of overexpression of RACK1 on the E1A-pCAF interaction in HeLa cells. Not surprisingly, we observed that overexpression of RACK1 in HeLa cells resulted in an obvious decrease of E1A-associated HAT activity, as compared with cells co-transfected with backbone vector alone (Fig. 6B).
RACK1 Prevents Human Tumor Cells from Undergoing E1Amediated Apoptosis-It has been reported that expression of E1A in tumor cell lines results in apoptosis and growth suppression. In our laboratories, we also observed that expression of E1A led to apoptosis in HeLa cells. Furthermore, this apoptotic effect of 243R E1A requires an intact ENT and CR1 region (Fig. 7A). The expression level of the wild-type 243R E1A was much lower than that of the non-apoptotic mutants in transient transfection assays (Fig. 7B). Similar results were obtained in SAOS-2 cells (not shown). To further demonstrate the role of ENT and CR1 in apoptosis induction, plasmids of wild-type 243R E1A, ENT deletion, or CR1 deletion mutants were transfected into SAOS-2 cells with the calcium phosphate method. After 4 weeks of selection with Geneticin (G418), stable transfected clones were obtained for ENT and CR1 deletion mutants that expressed the mutant E1A protein. However, although repeated several times, only a few colonies formed in wild-type E1A transfected dishes. Western blot assays revealed that none of these expressed wild-type E1A (Table I).
To investigate if RACK1 could protect human tumor cell lines from E1A-mediated apoptosis, we first optimized the amount of 243R E1A expression plasmid and time of recovery after transfection (see "Experimental Procedures") in order to obtain a reasonable, but not excessive, apoptotic effect on SAOS-2 cells. Specifically, 15-20% of apoptotic cells were obtained after 24 h of recovery from transfection with 6 g of 243R E1A expression plasmids. Apoptosis was evaluated by TUNEL, directly performed on the 60-mm culture dishes. Under these conditions, co-transfection of RACK1 drastically decreased the number of apoptotic cells induced by 243R E1A, as indicated in the histogram of Fig. 7C. Co-transfection of RACK1 reduced apoptosis by more than 50%. Overexpression of RACK1 alone produced a number of apoptotic cells that were statistically not distinguishable from the control (data not shown).
RACK1 Interacts with E1A in Vitro and in Vivo-To understand the mechanism by which RACK1 antagonizes the E1A in S. cerevisiae and in mammalian cells, we explored the possibil-  6. Effects of RACK1 on E1A-associated HAT activity. A, total cell lysates from yeast cells that were co-transformed with the indicated plasmids were immunoprecipitated with M73 and assayed for HAT activity. B, lysates from HeLa cells co-transfected with a total of 6 g of indicated plasmids (1:1) were immunoprecipitated with M73 and assayed for E1A-associated HAT activity.
ity of a physical interaction between RACK1 and E1A first by assaying this binding in a cell-free system. A chimeric GST-E1A protein was tested for its ability to bind to 35 S-labeled in vitro translated RACK1. 35 S-Labeled in vitro translated RACK1 interacted specifically with the GST-E1A fusion protein (Fig. 8A). Next, to determine whether RACK1 was able to associate with E1A in vivo, 293 cells, that constitutively express E1A proteins, were transfected with a plasmid expressing RACK1 tagged with a T7Tag. Transfected cells were immunoprecipitated with the M73 anti-E1A monoclonal antibody or with normal mouse serum as a control. Co-precipitated proteins were analyzed by Western blotting using an anti-T7Tag monoclonal antibody and, as shown in Fig. 8B, E1A co-purified with RACK1.

DISCUSSION
The adenoviral E1A protein was initially considered to be an oncogenic protein. Subsequent findings indicated that E1A could function as an anti-oncogene to suppress transformation, metastasis, and tumorigenicity by interfering with the function of genes responsible for the growth or development of cancer cells (51)(52)(53). E1A has also been reported to increase cellular susceptibility to apoptosis in cultured cells, particularly under serum starvation or high cell density conditions (8), as well as in vivo (16). E1A expressing derivatives of primary mouse embryo fibroblasts, as well as several tumor cell lines, underwent rapid apoptosis following treatment with ionizing radiation or several chemotherapy compounds (54 -56). While the transformation effect requires ENT, CR1, and CR2 of E1A, the pro-apoptotic and growth inhibitory effects are attributed only to ENT and CR1.
In the yeast model, E1A down-regulated the transactivation activity of GAL4. This inhibition correlated with the presence of the 70 N-terminal amino acid residues of E1A. These data are consistent with the role of CR1 and ENT in 243R-E1A   8. Physical interaction between RACK1 and E1A. A, GST pull-down assay shows a physical interaction between RACK1 and E1A. B, semi-confluent 293 cells were transfected with 5 g of pcDNA3T7Tag-RACK1 using the calcium phosphate method. After 48 h, lysates were prepared, immunoprecipitated with M73 or a normal mouse serum, resolved on a 10% SDS-PAGE, transferred to polyvinylidene difluoride, probed with the anti-T7Tag antibody, and detected with an ECL kit. IVT, in vitro translated. modulated regulation of gene transcription in mammalian cells, and strongly suggest that the inhibition of cell growth might be caused, at least partially, by interference with the expression of host genes required for normal growth. In addition, the ability of E1A to form a complex with HAT activity in yeast also correlated with the presence of the 70 N-terminal amino acid residues of E1A. This underscores another similarity between the behavior of E1A in yeast and mammalian cells, leading us to conclude that inhibition of yeast transcription factors could be due, at least in part, to the physical interaction of E1A with a significant amount of HAT activity. Currently, several complexes with HAT activity have been identified in yeast, but the nature of this E1A-associated HAT activity is presently unclear.
We adopted a functional rescuing screening strategy by employing the E1A-induced slow-growing phenotype to identify genes whose overexpression in yeast was able to antagonize the E1A-induced growth inhibition, thus rescuing the slow-growing phenotype. Two classes of genes were expected from this screening. The first class includes genes encoding proteins that directly interact with E1A, thus inactivating the function of ENT and CR1. The second class represents genes involved in essential pathways required for normal growth, but disrupted by E1A, either through direct interaction with the proteins or through transcriptional repression. From this screening, we actually isolated several overlapping cDNA fragments derived from human RACK1, a receptor for activated protein kinase C (PKC) and a homologue of the ␤ subunit of G proteins (57). In addition, coexpression of RACK1 efficiently disrupted the E1A-HAT interaction and rescued E1A-induced growth inhibition in yeast cells, confirming that functional targeting of the HAT activity in yeast could be the mechanism underlying E1Ainduced growth inhibition.
The molecular mechanisms through which E1A sensitizes mammalian cells to apoptosis in response to various stimuli are only partially understood (8 -11, 37, 38, 58). Because our data suggested that the effects of E1A on yeast and mammalian cells might involve a similar mechanism, we investigated the effects of RACK1 on E1A-induced apoptosis in mammalian cells. Not surprisingly, overexpression of RACK1 disrupted the interaction between E1A and HAT activity, and protected human cells from E1A-induced apoptosis.
The interplay between E1A and RACK1 is only partially elucidated. Since our data indicate that RACK1 directly interacts with E1A and disrupts E1A-HAT interaction, the most obvious model is that RACK1 inactivates E1A by competitive binding. In addition to activated PKC, RACK1 has been reported to interact with Src kinase (59), the cyclic AMP-specific phosphodiesterase (PDE4) (60), and the ␤-subunit of integrins (61). While the function of RACK1 is not fully understood, it appears to serve as a scaffold or anchor protein. There is evidence that RACK1 determines the localization and functional activity of ␤IIPKC (62). Therefore, it is possible that the RACK1 effect on E1A can be simply ascribed to its ability to sequester E1A to the cytosolic compartment.
Recently, ␤␥ subunits of G proteins were shown to serve as a scaffold for G protein-coupled receptor-mediated Ras activation (63). In considering the ability of E1A to transform cells with activated Ras, it is conceivable that overexpression of RACK1 may actually facilitate the activation of the Ras pathway. It should be noted that a RACK1 homologue, Cpc2, recently has been identified from S. cerevisiae (64) and from Schizosaccharomyces pombe where it interacts with Ran1 and antagonizes Ran1/Pat1 function (65). A Cpc2 null mutation causes a G 2 phase delay of the cell cycle. All defects associated with loss of Cpc2, including the G 2 phase delay of the cell cycle, are re-versed in cells expressing mammalian RACK1 (65), suggesting that RACK1 has a cell cycle regulatory function. We cannot exclude another intriguing model in which E1A could functionally block a pathway that is required for normal growth, but expression of RACK1 restores the normal signaling pathway or activates an alternate salvage pathway. In this case, in considering the interaction between RACK1 and E1A, RACK1 could be a part of that pathway and might be functionally targeted by E1A.
We observed that E1A-mediated growth inhibition in yeast is strain-specific, but we do not know the exact reason or mechanism. However, variations in genetic backgrounds and metabolic signaling pathways may account for this specificity. Previously, Wada and colleagues (66) reported a lethal effect of E1A on haploid yeast, which is strain-specific and depends on intact CR3 and on the 67 C-terminal amino acids of E1A. The mechanism for this lethal effect is presently not clear, however, it seems not to be linked to the inhibitory effect reported here, because the domain requirements are completely different. They also observed that E1A led to a G 1 arrest in diploid yeast, which requires the transforming region of E1A, but is not strain-specific (66). Because of the similarity with respect to the domain requirements, similar mechanisms may underlie both the G 1 arrest in diploid yeast and the strain-specific growth inhibition reported here. In fact, mating of E1A-expression PJ69-2A with Y187 to generate diploid S. cerevisiae resulted neither in G 1 arrest nor in growth inhibition (data not shown), suggesting that the E1A-induced G 1 arrest in diploid yeast originally reported by Wada et al. (66) also is a strainspecific event. Analysis of the strain specificity may facilitate the understanding of E1A functions. By screening for E1A resistant mutants, Miller et al. (42) found that both CDC35 (adenylate cyclase) and CDC25p (Ras GTP-GDP exchange factor) were involved in the E1A functions in yeast.
The protein expression of wild-type E1A is much lower than that of the deletion mutants in yeast. This could be partially caused by a negative selection. Because of the growth inhibitory effect, cells expressing higher levels of E1A grow much slower than cells with none or lower levels of E1A, which could be derived by a gradual loss in plasmid copy numbers, or by a spontaneous genetic alteration. Another possibility is that E1A also inhibits the expression of E1A itself per se, thus forming a negative feedback. In our transient transfection experiments, lower levels of E1A expression have also been observed in HeLa cells. Again, this could be a result of the apoptotic effect of E1A that selects cells with lower or no E1A expression in culture. In addition, the negative feedback mechanism proposed above contributed, at least partially, to this phenomenon. We have observed that wild-type E1A inhibits the promoter activity of the cytomegalovirus promoter, which is the promoter used to express E1A and E1A mutants in transfection assays. 2 Several previous studies suggested that E1A regions for binding to p300/CBP and pRb were required to promote apoptosis and chemosensitivity in mammalian cells (37,38,58), implying that inactivation of pRb and p300/CBP is involved. In addition, E1A-expressing cells accumulated p53, and increased p53 also sensitized the cells to drug-induced apoptosis. The essential regions for E1A to promote apoptosis overlapped with those previously shown to be essential for oncogenic transformation and p53 accumulation. It would be reasonable to suggest that p300/CBP and pRb, which bind to ENT/CR1 and CR1/CR2 of E1A, respectively, are the critical targets for the different activities of E1A. We confirmed that ENT and CR1 are essential for E1A to promote apoptosis in HeLa and SAOS-2 cells. However, because E1A is able to induce apoptosis in SAOS-2 cells (lacking functional pRb and p53), it must be a p53/pRb-independent event. Nevertheless, our data do not rule out the inactivation of pRb as a requirement for E1A-induced apoptosis.