mUBC9, a Novel Adenovirus E1A-interacting Protein That Complements a Yeast Cell Cycle Defect*

Adenovirus E1A encodes two nuclear phosphopro- teins that can transform primary rodent fibroblasts in culture. Transformation by E1A is mediated at least in part through binding to several cellular proteins, including the three members of the retinoblastoma family of growth inhibitory proteins. We report here the cloning of a novel murine cDNA whose encoded protein in- teracts with both adenovirus type 5 and type 12 E1A proteins. The novel E1A-interacting protein shares sig- nificant sequence homology with ubiquitin-conjugating enzymes, a family of related proteins that is involved in the proteasome-mediated proteolysis of short-lived proteins. Highest homology was seen with a Saccharomyces cerevisiae protein named UBC9. Importantly, the murine E1A-interacting protein complements a cell cycle defect of a S. cerevisiae mutant which harbors a temper-ature-sensitive mutation in UBC9 . We therefore named this novel E1A-interacting protein mUBC9. We mapped the region of E1A that is required for mUBC9 binding and found that the transformation-relevant conserved region 2 of E1A is required for interaction. Early region 1A (E1A) of human adenovirus type 5 (Ad5) encodes two differentially spliced mRNAs that specify proteins of 243 and 289 amino acids, respectively (243R and

Early region 1A (E1A) of human adenovirus type 5 (Ad5) encodes two differentially spliced mRNAs that specify proteins of 243 and 289 amino acids, respectively (243R and 289R E1A). In infected cells, the E1A proteins are the first viral proteins that are expressed and serve two important functions. First, E1A proteins contribute to the regulation of both viral and cellular gene expression during the viral infection cycle (1,2). Furthermore, E1A proteins stimulate quiescent cells to enter a cell division cycle following adenovirus infection, allowing the virus to use host-encoded enzymes for replication of its DNA during S phase. In vitro, E1A proteins can immortalize primary rodent cells and cooperate with an activated ras oncogene in oncogenic transformation (3,4).
The activities of E1A proteins are mediated primarily through three functional domains that are highly conserved between the E1A proteins of different adenovirus serotypes (5). The 243R and the 289R E1A proteins share conserved regions 1 and 2 (CR1 and CR2), 1 while CR3 is uniquely present in the 289R E1A protein. CR1 and CR2 are largely responsible for the cell cycle stimulatory activity and transforming activity of E1A proteins (5,6). CR3 encodes a strong transcriptional activation domain (7)(8)(9)(10). CR1 and CR2 are required for the suppression of enhancer activity (11)(12)(13), for the induction of DNA synthesis (5,14,15), and for the induction of apoptosis (16). Both domains are thought to contribute to deregulation of the cell cycle by binding to a set of cellular proteins (17,18). Subsequent functional inactivation of these proteins with growth inhibitory activity results in the transformed phenotype of the cells. Over the past decade, several of these CR1-and CR2-interacting cellular E1A-binding proteins have been identified. They include the retinoblastoma gene product pRb (19), the pRb-like proteins p107 and p130 (19 -22), and the transcriptional coactivator proteins p300 and CBP (23)(24)(25)(26). A cellular 48-kDa phosphoprotein named CtBP has also been identified that negatively modulates E1A transformation through binding to a carboxyl-terminal epitope of E1A proteins (27,28). More recently, the p27 inhibitor of cyclin-dependent kinases has been found to interact with E1A, although it has not been reported where the p27 binding epitope is located (29).
We have used a yeast two-hybrid screen to clone additional cellular proteins involved in cell cycle control and that bind to CR2 of adenovirus E1A proteins. We report here the structure and characterization of mUBC9, a novel E1A-interacting protein.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screen-Yeast strain Y190, containing the "bait" plasmid pPC97-E1A, encoding the Gal4 DNA binding domain (DBD) fused to an adenovirus 5 E1A 12 S cDNA fragment encoding amino acids 76 -243 of E1A, was transformed with a day 14.5 CD1 mouse embryo library using the lithium acetate method. 6 ϫ 10 5 transformants were selected for growth on plates lacking histidine and supplemented with 25 mM 3-aminotriazole. His ϩ colonies were subsequently analyzed for ␤-galactosidase activity as described previously (38). cDNAs library plasmids derived from double positive yeast colonies were tested for bait specificity by retransformation with different Gal4 DBD fusion plasmids: pPC97-p107, pPC97-bmi, and pPC97 without an insert.
Plasmids-pPC97-E1A was generated by cloning an adenovirus 5 E1A 12 S cDNA fragment encoding amino acids 76 -243 of E1A in-frame with the Gal4 DNA binding domain (amino acids 1-147) of pPC97. Deletion mutants of E1A were constructed by subcloning fragments of * This work was supported by grants from the Dutch Cancer Society and the Netherlands Organization for Scientific Research. 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 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) X99739.
Northern Blot Analysis-For mUBC9 expression analysis, total cytoplasmic RNA was prepared from a panel of mouse tissues. 20 g of total cellular RNA was electrophoresed through a 1% formaldehydeagarose gel, transferred to nitrocellulose, and probed with a 32 P-labeled mUBC9 cDNA.
Cell Culture, Transfections, and Labeling-Human osteosarcoma (U2-OS) cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transfections were performed overnight using the calcium phosphate method. Total amounts of transfected DNA was 20 g/10-cm dish. For pulse-chase experiments, cells were starved for 1 h in methionine/cysteine-free medium and subsequently incubated with 100 Ci of [ 35 S]methionine/cysteine per 10-cm dish for 1 h. After this, cells were washed twice in PBS, and further maintained in normal Dulbecco's modified Eagle's medium, 10% fetal calf serum. At the indicated time points, cells were lysed by incubation in RIPA for 20 min on ice. Equal amounts of radioactive lysates were incubated on ice with 5 l of nonimmune serum for preclearing. Subsequently, lysates were incubated with protein A-Sepharose beads (Pharmacia Biotech Inc.), precomplexed with 50 l of M73 (anti-E1A) hybridoma supernatant, for 2 h at 4°C while rocking. Then, beads were washed four times in RIPA, heated in SDS-containing sample buffer, and loaded on a 12% SDS-polyacrylamide gel. For detection of endogenous UBC9 and the size of mUBC9 encoded by the novel cDNA, U2-OS cells were transfected with the indicated plasmids overnight using the calcium phosphate method. Cells were fed the following morning, lysed 36 h post-transfection by sonication in SDS-containing sample buffer, and subjected to gel electrophoresis and immunoblotting using the mUBC9 rabbit polyclonal antiserum.
Antibodies-To generate antibodies against mUBC9, the production of GST-mUBC9 protein was induced in E. coli and purified using glutathione-Sepharose beads (Pharmacia). For immunization and three rounds of subsequent boosts, a rabbit was injected with beads carrying at least 100 g of purified GST-mUBC9 fusion protein to obtain polyclonal serum.
Immunoblotting-10-cm dishes of transfected U2-OS cells were lysed by sonication in 1 ml of SDS-containing sample buffer. 100 l of each sample was separated on a 15% SDS-polyacrylamide gel, and transfer was performed overnight to nitrocellulose. The filter was blocked in TPBS, 5% milk (PBS supplemented with 0.1% Tween 20 and 5% nonfat milk; Protifar, Nutricia) for 2 h at room temperature, incubated with anti-mUBC9 polyclonal rabbit serum in a dilution of 1:10,000 in TPBS, 2% milk for 2 h at room temperature, and incubated with secondary antibody (GARPO) in TPBS, 2% milk. The filter was washed three times in TPBS, and visualization was performed by enhanced chemiluminescence (Amersham).
GST Pull-down Assay-mUBC9, HA-mUBC9, and yeast UBC9 proteins were made by in vitro transcription and translation of the cDNAs using rabbit reticulocyte lysates (Promega) and [ 35 S]methionine. Equal amounts of proteins were incubated with 1.0 g of either purified GST-13S-E1A protein or GST alone, for 4 h in 0.5 ml of ELB (18) on ice. 50 l of glutathione-Sepharose beads (Pharmacia, 30% slurry) was added and rocked for 1 h at 4°C. Beads were washed four times in ELB, heated in SDS sample buffer, and loaded on a 15% SDS-polyacrylamide gel.
Screening E1A Mutants, Defective in mUBC9 Binding-PCR was performed on pPC97-E1A with a 5Ј primer, identical to a sequence present in the Gal4 DBD and a 3Ј primer of which the sequence is located downstream of the multiple cloning site. PCR was performed under normal conditions using recombinant Taq polymerase (Life Technologies, Inc.) as the mutating agent, following the instructions of the manufacturer. In the first screen the entire PCR pool, containing the randomly point mutated cDNAs, was directly transformed with competent Y190 yeasts in combination with a gapped pPC97. The Y190 strain contained already the pPC86-E1A-BP5 plasmid, carrying the TRP1 gene, which enables this strain to grow on plates without tryptophan. After homologous recombination and thus recirculation of pPC97, yeast clones were selected for growth on plates lacking leucine and tryptophan. 257 colonies were subsequently analyzed for ␤-galactosidase activity as described (38). Six clones did not turn blue, suggesting loss of interaction with mUBC9. White colonies were lysed; the lysates were separated on a 12% SDS-polyacrylamide gel, and protein size was checked by immunoblotting, using an antibody directed against the carboxyl terminus of E1A (M73). Plasmid DNA was extracted from the yeast clone that still produced full-length but mutated E1A, and the insert was sequenced completely using the Pharmacia T7 sequencing kit.

Cloning of a Novel Adenovirus E1A-interacting Protein-To
identify cDNAs encoding polypeptides that interact with adenovirus E1A, we first tested which domains of the 243R adenovirus 5 E1A protein activate transcription in yeast. The largest fragment of the 243R E1A protein that failed to activate transcription in yeast specified amino acids 76 -243 (data not shown). Since this peptide harbors the transformation-relevant CR2 of E1A, a yeast two-hybrid screen was performed with this fragment of E1A as a bait. Yeast strain Y190, which contains two Gal4-inducible reporter genes, HIS3 and LacZ, was cotransformed with the E1A bait plasmid and a day 14.5 CD1 mouse embryo cDNA library. Of 6 ϫ 10 5 yeast transformants, 66 colonies appeared on plates lacking histidine, 19 of which stained positive when tested for expression of ␤-galactosidase. To test whether E1A was required for interaction with the products of the 19 identified cDNAs, all these clones were retransformed in yeast strain Y190 together with plasmids encoding other Gal4 DBD fusions. All hybrid proteins were found to interact only with Gal4-E1A. Subsequent analyses indicated that 12 of the 19 clones contained an approximately 1-kilobase insert and were derived from the same gene.
We determined the complete nucleotide sequence of the insert of E1A-BP5, one of the 12 identical E1A-interacting clones. Analysis of the cDNA sequence (Fig. 1A) shows a single open reading frame of 158 amino acids. Comparison of the deduced amino acid sequence with GenBank revealed that the protein encoded by the newly identified cDNA had significant homology to a class of related proteins, named ubiquitin-conjugating enzymes (42). These proteins play a role in the proteasomemediated targeted proteolysis of short-lived cellular proteins. In particular, the E1A-interacting protein showed high homology to Saccharomyces cerevisiae UBC9 and a protein from Schizosaccharomyces pombe, hus5 (identities: 56 and 66%, respectively; Refs. 41 and 43). Fig. 1B shows the predicted amino acid sequence of E1A-BP5 and its comparison to the amino acid sequence of S. cerevisiae UBC9. Although E1A-BP5 is more closely related to hus5 than to UBC9, we named this novel E1A-interacting protein mUBC9 (see below).
To address whether mUBC9 also interacts with E1A in another context than in yeast, we performed an in vitro binding assay with a bacterially expressed GST-E1A fusion protein and different radioactively labeled UBC9 proteins generated by sequential in vitro transcription and translation. Fig. 2A (lanes mUBC9 Interacts with Adenovirus E1A [1][2][3] shows 10% of the input of the in vitro translated proteins. After incubation, full-length and HA-tagged mUBC9 (lanes 1 and 2) interacted with GST-E1A (lanes 7 and 8), but not with GST alone (lanes 4 and 5). Moreover, the highly homologous UBC9 protein (lane 3) from S. cerevisiae also interacted specifically with GST-E1A (lane 9), but not with GST alone (lane 6).
To detect endogenous mUBC9 protein and to ask whether the protein encoded by the mUBC9 cDNA is full-length, we generated a rabbit polyclonal mUBC9 antiserum. We then used this serum to detect UBC9 in whole cell extracts from mock transfected cells and in extracts from cells transfected with an mUBC9 expression vector. Fig. 2B shows that the mUBC9 antiserum detects a protein of approximately 20 kDa in mocktransfected cells (lane 1), the abundance of which is increased substantially in the mUBC9-transfected cells (lane 2). Furthermore, an HA-tagged version of mUBC9 is also recognized by the rabbit serum, which migrates somewhat slower due to the presence of the HA tag (lane 3). A preimmune serum from the same rabbit did not detect any protein of the indicated sizes (data not shown). We conclude that endogenous UBC9 is of similar molecular weight as the cDNA-encoded mUBC9, indicating that the cDNA specifies the full-length mUBC9 protein.
mUBC9 Complements a S. cerevisiae ubc9 ts Mutation-To test whether mUBC9 can functionally replace S. cerevisiae UBC9, we used a yeast strain (ubc9 -1) with a temperaturesensitive mutation in UBC9 (41). Wild type yeast and ubc9 -1 mutants (YWO1 and YWO102, respectively) were transformed with either vector plasmids alone or vector expressing murine UBC9 and were streaked out on galactose-containing agar plates and incubated at either 23°C or 37°C for 3 days. Fig. 3 shows that the temperature-sensitive growth of the yeast ubc9 -1 mutant is suppressed by expression of the murine UBC9 homologue. Surprisingly, mUBC9 did not complement a mutation in the S. pombe hus5 gene 2 This was unexpected since mUBC9 is more homologous to hus5 than to UBC9. These data indicate that mUBC9 can functionally replace the S. cerevisiae UBC9, but not S. pombe hus5. We therefore named this novel E1A-interacting protein mUBC9.
mUBC9 Is Broadly Expressed-To analyze the pattern of expression of mUBC9 mRNA, the murine UBC9 cDNA was 2 A. M. Carr, personal communication. mUBC9 Interacts with Adenovirus E1A used to probe a Northern blot containing total cytoplasmic RNA from several rat tissues. The mUBC9 probe detected two transcripts of 1.0 and 3.2 kilobases in all tissues tested (Fig. 4). We have recently cloned several additional mUBC9 cDNAs from a mouse embryo cDNA library and found that the larger transcript only differs in the 3Ј-noncoding region from the 1.0kilobase mRNA (data not shown). Immunostaining of cells transiently transfected with a plasmid encoding HA-tagged mUBC9 revealed that transfected HA-mUBC9 was present in the cytosol as well as in the nucleus (data not shown).
Mapping of the UBC9 Binding Site on E1A-To determine which domain of 243R E1A is responsible for binding to mUBC9, we generated a set of expression vectors in which different parts of the 243R E1A protein are fused to the Gal4 DBD. These proteins were then coexpressed in yeast strain Y190 with a vector that directs the synthesis of mUBC9 fused to the Gal4 transactivation domain. The results of these experiments are summarized in Table I. Interaction (as evidenced by the blue staining of the yeast colonies) was seen with 243R E1A residues 76 -140, but not 120 -140 or 76 -120. These data indicate that CR2 (residues 121-127) is likely to be involved in mUBC9 binding, but is not sufficient for interaction. In support of a role for CR2 in the binding of UBC9, we found that residues 80 -154 of adenovirus type 12 E1A also mediate binding to mUBC9, indicating that mUBC9 binding is a conserved feature of different adenovirus serotypes. However, a double point mutant in CR2 of Ad5 E1A that abolishes binding to pRb (928/961) still binds mUBC9, indicating that the pRb binding site and the mUBC9 binding site are not identical (Table I).
To further study the amino acid residues of E1A that are involved in mUBC9 interaction, we performed a yeast PCRbased mutagenesis screen. This screen is schematically represented in Fig. 5. This procedure takes advantage of the high frequency of recombination between homologous DNA fragments in yeast (44). In short, a part of the DNA encoding Gal4-E1A (amino acids 76 -243) fusion protein was amplified by PCR under conditions in which random mutations are induced at low frequency. This mutagenized PCR product was then introduced in yeast, together with a gapped expression vector that lacks an E1A insert. After homologous recombination of sequences up-and downstream of the E1A insert, a functional Gal4-E1A expression vector is created in vivo. Using this approach, we created a "library" of mutated Gal4-E1A expression vectors that were tested for interaction with mUBC9 using the standard two-hybrid assay. In the initial screen 257 yeast colonies were assayed, of which six failed to interact with mUBC9. From these six, five yeast colonies turned out to express a truncated Gal4-E1A fusion protein, resulting from the introduction of an early stop codon by the mutagenesis procedure. One white colony expressing full-length Gal4-E1A protein, as judged by immunoblot analysis of whole yeast protein extracts (data not shown), was then subjected to DNA sequence analysis. This mutant E1A protein turned out to have a single amino acid substitution, changing leucine 122 of E1A to an isoleucine. We named this mutant E1A-L122I. Since leucine 122 is part of the highly conserved pRb binding motif of E1A (the LXCXE motif), these data indicate that mUBC9 interacts directly with CR2 of E1A, raising the possibility that mUBC9 binding may contribute to transformation by E1A.
E1A Protein Stability Is Not Affected by mUBC9 -Since E1A itself is a short-lived protein (45,46), one possibility would be that E1A itself is a target for UBC9-mediated proteolysis as was suggested by Ciechanover et al. (47). To investigate this, we used the L122I mutant, which fails to bind to mUBC9. To   FIG. 3. Murine E1A-BP5 complements yeast ubc9 -1. Yeast wild type (UBC9; right side of the plates) and ubc9 -1 mutants (left side) transformed with either vector plasmids (top) or vector expressing murine E1A-BP5 (pGAL1-E1A-BP5; bottom) were streaked out on galactose-containing agar plates and incubated at either 23°C or 37°C for 3 days. The temperature-sensitive growth of the yeast ubc9 -1 mutant is suppressed by expression of the UBC9 homologue E1A-BP5 from mice.

TABLE I
Mapping of the mUBC9 binding epitope of adenovirus E1A in a yeast two-hybrid assay A set of yeast expression vectors encoding Gal4-DBD-12S E1A deletion mutants was generated (left column). These vectors were introduced in the yeast strain Y190 with and without pPC86-mUBC9 (center), and ␤-galactosidase activity was determined. Only two E1A fusion proteins, comprising residues 1-76 and the full-length 12 S E1A (1-243), transactivated in yeast (blue, right column). The largest E1A part that did not transactivate (white) contained amino acids 76 -243 of 12 S E1A. This deletion mutant of E1A was used in the initial two-hybrid screen. All non-transactivating fusions were checked for the interaction with a fusion of mUBC9 and the transactivation domain of Gal4 (Gal4-TA). The smallest part of E1A that still interacts with mUBC9 contains residues 76 -140. The 928/961 mutant contains residues 76 -140 of Ad5 E1A, except for two mutations: C124G and E135K, together resulting in an E1A mutant that is unable to interact with the family of pocket proteins (pRb, p107, and p130). Residues 80 -154 of Ad12 E1A are the equivalent amino acids of 76 -140 in Ad5 E1A, both including CR2.  Fig. 6 shows that both transiently expressed proteins and especially the faster migrating species have similar half-lives of approximately 2 h, which agrees well with data from previous studies (45,46). We conclude that mUBC9 interaction does not affect the half-life of E1A protein. DISCUSSION We have used a yeast two-hybrid screen to identify a novel murine cDNA whose encoded protein interacts with adenovirus E1A-transforming proteins. The newly identified protein is a member of a family of related proteins, named ubiquitin-conjugating enzymes (UBCs), that play a role in proteasome-mediated proteolysis of short-lived proteins (42,48). It shares significant homology to both S. cerevisiae UBC9 (41) and S. pombe hus5 (43). However, in a functional complementation assay, the murine UBC only complements the S. cerevisiae ubc9 defect, but not the S. pombe hus5 mutation. For this reason, we have named the novel E1A-interacting protein mUBC9. Recently, the human homologue of UBC9 was reported by two groups. Strikingly, the amino acid sequence of human UBC9 is 100% identical to the mouse sequence reported here (49,50).
There are several possible explanations for the observed E1A-mUBC9 interaction, which are schematically represented in Fig. 7. First, it is possible that E1A itself is a target for UBC9-mediated ubiquitin-dependent degradation (Fig. 7A). This is unlikely, because a mutant E1A protein that fails to bind mUBC9 has the same half-life as wild type E1A protein (Fig. 6). It does, however, not exclude the possible ubiquitination of E1A through a number of essential residues in the entire amino terminus of E1A, which are involved in the rapid breakdown of E1A (51,52). Nevertheless, we did not find evidence supporting ubiquitination of E1A using a method as was described for c-Jun (53). Second, by binding UBC9, E1A could interfere with the targeted proteolysis of certain cellular proteins, thereby stabilizing proteins with growth-stimulatory activity (Fig. 7B). Alternatively, E1A could simultaneously bind FIG. 7. Three possible models for the E1A-UBC9 interaction. A, UBC9 (indicated as a black dot) targets E1A for breakdown through the ubiquitin-proteasome pathway, by conjugating ubiquitin molecules (Ub) to E1A. B, E1A binds and subsequently inactivates UBC9, thereby preventing the breakdown of protein X, which can no longer be ubiquitinated by UBC9. C, E1A binds a certain protein X, acting as a targeting factor, and because of the E1A-UBC9 interaction, X becomes a target for breakdown by ubiquitination through UBC9.
FIG. 5. PCR-based screen for point mutants of E1A that lost the ability to interact with mUBC9 in yeast. PCR was performed on pPC97-E1A using primers upstream and downstream of the E1A insert. To generate a pool of mutated products, it was sufficient in these studies to use recombinant Taq polymerase (Life Technologies, Inc.) under normal PCR conditions. Products were subsequently transformed together with a gapped pPC97 plasmid as indicated. Homologous recombination takes place in a transformed Y190 yeast strain, which already contains the pPC86-mUBC9 (pPC86-E1A-BP5) plasmid, and occurs on both sites of the point-mutated E1A cDNA. A circularized pPC97-E1A plasmid allows growth on plates without leucine. Yeast colonies are subsequently screened for ␤-galactosidase activity, where a blue colony means interaction between E1A and mUBC9, while a white colony means loss of interaction. A subsequent analysis of the white colonies is then performed to see whether the mutated Gal4-E1A fusion is full-length. If so, plasmid DNA is extracted and the insert is sequenced to locate the generated point mutation(s).  1-4) or pCMV-L122I-E1A (lanes 5-8), were labeled with [ 35 S]methionine/cysteine for 1 h and subsequently chased in Dulbecco's modified Eagle's medium for the indicated times. E1A protein was immunoprecipitated by M73, an E1A-specific monoclonal antibody, collected with protein A-Sepharose beads, and separated on a 12% SDS-polyacrylamide gel. The different E1A species derived from the 12 S cDNAs are indicated by a bracket on the right. mUBC9 Interacts with Adenovirus E1A both UBC9 and a growth-inhibitory cellular protein, thereby targeting the cellular E1A-interacting protein for ubiquitinmediated proteolysis (Fig. 7C). In this scenario, E1A would act as a targeting factor. A possible candidate in such a model is the cyclin-dependent kinase inhibitor p27, a protein degraded through ubiquitination (54) and one which was recently shown to interact with E1A (29).
That viral infections can lead to a significant deregulation of cell cycle control was suggested in studies from the laboratory of P. Howley, in which it was shown that the tumor suppressor protein p53 is targeted for degradation by the ubiquitin pathway, through binding of the papilloma virus protein E6 to a cellular polypeptide called E6-AP (55,56). In this case, E6-AP serves as a ubiquitin ligase targeting p53 for rapid breakdown. That binding of adenovirus E1A to UBC9 somehow contributes to deregulation of the cell cycle is suggested only indirectly by our finding that E1A uses conserved region 2 for mUBC9 interaction, a region known to be crucial for the transforming activity of E1A. Our finding that mUBC9 complements a yeast cell cycle defect is also consistent with a role for mUBC9 in cell cycle control.
Recently, a number of cell cycle regulatory proteins have been shown to be targets of ubiquitin-mediated targeted proteolysis. These proteins include the yeast cyclins CLN2, CLN3, CLB2, and CLB5 (41,57,58), the yeast p40 sic1 Cdk inhibitor (59), A and B type cyclins (60), the immediate early gene products c-Jun and c-Fos (53,61), and the two proteins mentioned earlier, the mammalian p27 Cdk inhibitor and the tumor suppressor p53 (54,55). We have recently found that the E2F transcription factors are also degraded by ubiquitin-mediated proteolysis. Significantly, E1A coexpression causes a significant increase in the half-life of E2F. 3 We are currently investigating the role of mUBC9 in E1A-mediated transformation and cell cycle progression of mammalian cells.