The retinoblastoma (RB) tumor suppressor protein and the TATA-box-binding protein TimD form contacts with a number of viral transactivator proteins. One of these, the adenovirus EMA protein, can bind to both proteins.Here we present evidence that the cellular transcription factor PU.l can bind to both RB and TFIID. Like ElA, PU.1 binds to the conserved C-terminal domain of TFMM and to the RB "pocket" domain. The PU.1 sequences required to bind either protein lie within a 75-amino acid region which functions as an independent activation domain in vivo. The ability of PU.1 to contact directly both RB and TFiD through the same 75-residue domain prompted us to look for sequence silarity between these two proteins. We find that the previously defined domain A of the RB pocket shows sequence similarity to the conserved C terminus of TFID, whereas domain B shows sequence similarity to a second general transcription factor, TFHB. The potential for RB to influence transcription by using TFYD-and TFIB-related functions is discussed. Direct interaction between cellular regulatory and general transcription factors has not yet been established. However, four viral transactivating proteins, the adenovirus ElA (10, 11), the herpes virus VP16 (12), the Epstein-Barr virus Zta (13), and the cytomegalovirus IE2 (14), have been shown to contact directly the TATA box binding protein TFIID and, in the case of VP16, the general factor TFIIB (15).Here we show that PU.1, a lymphoid-specific transcription factor with an Ets-like DNA-binding domain (16), can bind directly to both TFIID and RB. Both of these interactions (PU.1-RB and PU.1-TFIID) are mediated by a 75-amino acid activation domain of PU.1. These results led to sequence comparisons, which show that domains A and B of the RB pocket have sequence similarity to the two general transcription factors TFIID and TFIIB, respectively. The ability ofRB to modulate transcription, using the TFIID and TFIIB similarity regions, is discussed. MATERIALS AND METHODSProteins Produced by in Vitro Translation. Phagemid vectors for PU.1 (pBSPU.1), ElA (pTM1EIA), Spl (pBS-Spl), IE1 (pGEMIEl), and gelsolin were gifts from R. Maki (La

Fos and Jun form a tight heterodimeric complex that activates transcription by AP1 sites. We have recognized that two adjacent regions of the Jun A1 activation domain are conserved in the Fos protein, and we refer to these two homologous regions as homology box 1 (HOB1) and homology box 2 (HOB2). Using GAIA chimeras, we show that the HOB1/HOB2 region of Fos and Jun is an independent activation domain in which HOB1 and HOB2 act cooperatively to activate transcription. This cooperativity is retained after the replacement of

It has been suggested that the dominant effect of mutant p53 on tumor progression may reflect the mutant protein binding to wild-type p53, with inactivation of suppressor function. To date, evidence for wild-type/ mutant p53 complexes involves p53 from different species. To investigate wild-type/mutant p53 complexes in relation to natural tumor progression, we sought to identify intraspecific complexes, using murine p53. The mutant phenotype p53-2460 was used because this phenotype is immunologically distinct from wild-type p53-246+ and thus permits immunological analysis for wild-type/mutant p53 complexes. The p53 proteins were derived from genetically defined p53 cDNAs expressed in vitro and also from phenotypic variants of p53 expressed in vivo. We found that the mutant p53 phenotype was able to form a complex with the wild type when the two p53 variants were cotranslated. When mixed in their native states (after translation), the wild-type and mutant p53 proteins did not exhibit any binding affinity for each other in vitro. Under identical conditions, complexes of wild-type human and murine p53 proteins were formed. For murine p53, both the wild-type and mutant p53 proteins formed high-molecular-weight complexes when translated in vitro. This oligomerization appeared to involve the carboxyl terminus, since truncated p53 (amino acids 1 to 343) did not form complexes. We suggest that the ability of the mutant p53 phenotype to complex with wild type during cotranslation may contribute to the transforming function of activated mutants of p53 in vivo.

The retinoblastoma (RB) tumour suppressor protein is capable of repressing the activity of promoters containing DNA binding sites for the transcription factor E2F. Recently a protein which binds RB and possesses the DNA binding characteristics of E2F has been cloned. Here we show that the E2F activation domain is the target for RB-induced repression. RB can silence the 57 residue E2F activation domain but cannot effectively repress an E2F mutant which has reduced RB binding capacity. Extensive mutagenesis of E2F shows residues involved in RB binding are required for transcription activation. Mutations which affect both functions most dramatically lie within the minimal RB binding region. A further subset of sensitive residues lies within a new repeat motif E/DF XX L X P which flanks the minimum RB binding site. These data show that RB can mask E2F residues involved in the activation process, possibly by mimicking a component of the transcriptional machinery. Consistent with this model, we find that the TATA box binding protein TBP can bind to the E2F activation domain in vitro in a manner indistinguishable from that of RB.

The cell cycle-regulating transcription factors E2F1/DP1 activate genes whose products are required for S phase progression. During most of the G1 phase, E2F1/DP1 activity is repressed by the retinoblastoma gene product RB, which directly contacts the E2F1 activation domain and silences it. The E2F1 activation domain has sequence similarity to the N-terminal activation domain of E1A(12S), which contains binding sites for CBP as well as RB. Here, we present evidence that the CBP protein directly contacts E2F1/DP1 and stimulates its activation capacity. We show that CBP interacts with the activation domain of E2F1 both in vitro and in vivo. Deletion of four residues from the E2F1 activation domain reduces CBP binding as well as transcriptional activation, but still allows the binding of RB and MDM2. This deletion removes residues which are conserved in the N-terminal activation domain of E1A and which are required for the binding of CBP to E1A. When the E1A N-terminus is used as a competitor in squelshing experiments it abolishes CBP-induced activation of E2F1/DP1, whereas an E1A mutant lacking CBP binding ability fails to do so. These results indicate that CBP can act as a coactivator for E2F1 and suggest that CBP recognises a similar motif within the E1A and E2F1 activation domains. The convergence of the RB and CBP pathways on the regulation of E2F1 activity may explain the cooperativity displayed by these proteins in mediating the biological functions of E1A. We propose a model in which E1A activates E2F not only by removing the RB repression but also by providing the CBP co-activator.