Adenovirus 5 E1A is responsible for increased expression of insulin receptor substrate 4 in established adenovirus 5-transformed cell lines and interacts with IRS components activating the PI3 kinase/Akt signalling pathway.

Using mass spectrometric analysis insulin receptor substrate 4 (IRS-4) has been identified as a novel adenovirus 5 early region 1A (Ad5E1A)-binding protein. IRS-4 interacts with both the transcriptional activation domain (conserved region 3) and the N-terminal region of Ad5E1A13S. Prolonged expression of Ad5E1A13S is required for the observed dramatic increase in the levels of IRS-4 mRNA and protein in Ad5E1-transformed human cell lines. Once expressed, as well as binding to E1A and the insulin receptor, IRS-4 remains tyrosine phosphorylated and constitutively associates with the regulatory p85 subunit of phosphoinositide 3 kinase, resulting in the phosphorylation of Akt (causing activation) and GSK-3beta (causing inhibition). Reducing IRS-4 expression using small interfering RNA (siRNA) in established Ad5E1A-expressing cell lines decreases the activation of Akt and cellular proliferation. During Ad5 infection, IRS-4 is not expressed. However, Ad5E1A associates with IRS-1, increasing Akt and GSK-3beta phosphorylation and tyrosine phosphorylation of IRS-1 itself. We conclude that the association and altered regulation of IRS proteins by Ad5E1A contribute to the adenovirus-transformed phenotype and modulates viral infection in an Akt-dependent manner.

T-cell responses to human papillomavirus type 16 among women with different grades of cervical neoplasia.

Infection with high-risk genital human papillomavirus (HPV) types is a major risk factor for the development of cervical intraepithelial neoplasia (CIN) and invasive cervical carcinoma. The design of effective immunotherapies requires a greater understanding of how HPV-specific T-cell responses are involved in disease clearance and/or progression. Here, we have investigated T-cell responses to five HPV16 proteins (E6, E7, E4, L1 and L2) in women with CIN or cervical carcinoma directly ex vivo. T-cell responses were observed in the majority (78%) of samples. The frequency of CD4+ responders was far lower among those with progressive disease, indicating that the CD4+ T-cell response might be important in HPV clearance. CD8+ reactivity to E6 peptides was dominant across all disease grades, inferring that E6-specific CD8+ T cells are not vitally involved in disease clearance. T-cell responses were demonstrated in the majority (80%) of cervical cancer patients, but are obviously ineffective. Our study reveals significant differences in HPV16 immunity during progressive CIN. We conclude that the HPV-specific CD4+ T-cell response should be an important consideration in immunotherapy design, which should aim to target preinvasive disease.

Structural determinants outside the PXDLS sequence affect the interaction of adenovirus E1A, C-terminal interacting protein and Drosophila repressors with C-terminal binding protein.

C-Terminal binding protein (CtBP) interacts with a highly conserved amino acid motif (PXDLS) at the C terminus of adenovirus early region 1A (AdE1A) protein. This amino acid sequence has recently been demonstrated in the mammalian protein C-terminal interacting protein (CtIP) and a number of Drosophila repressors including Snail, Knirps and Hairy. In the study described here we have examined the structures of synthetic peptides identical to the CtBP binding sites on these proteins using NMR spectroscopy. It has been shown that peptides identical to the CtBP binding site in CtIP and at the N terminus of Snail form a series of beta-turns similar to those seen in AdE1A. The PXDLS motif towards the C terminus of Snail forms an alpha-helix. However, the motifs in Knirps and Hairy did not adopt well-defined structures in TFE/water mixtures as shown by the absence of medium range NOEs and a high proportion of signal overlap. The affinities of peptides for Drosophila and mammalian CtBP were compared using enzyme-linked immunosorbent assay. CtIP, Snail (N-terminal peptide) and Knirps peptides all bind to mammalian CtBP with high affinity (K(i) of 1.04, 1.34 and 0.52 microM, respectively). However, different effects were observed with dCtBP, most notably the affinity for the Snail (N-terminal peptide) and Knirps peptides were markedly reduced (K(i) of 332 and 56 microM, respectively) whilst the Hairy peptide bound much more strongly (K(i) for dCtBP of 6.22 compared to 133 microM for hCtBP). In addition we have shown that peptides containing identical PXDLS motifs but with different N and C terminal sequences have appreciably different affinities for mammalian CtBP and different structures in solution. We conclude that the factors governing the interactions of CtBPs with partner proteins are more complex than simple possession of the PXDLS motif. In particular the overall secondary structures and amino acid side chains in the binding sites of partner proteins are of importance as well as possible global structural effects in both members of the complex. These data are considered evidence for a multiplicity of CtBPs and partner proteins in the cell.

Structural determinants in adenovirus 12 E1A involved in the interaction with C-terminal binding protein 1.

The interaction between the C-terminal binding protein 1 (CtBP-1) and purified Ad12 E1A protein has been examined through the use of a combination of biophysical techniques. A fragment equivalent to the 77 C-terminal amino acids of Ad12 E1A (Ad12 77-a.a. E1A) was generated by limited proteolysis of Ad12 266-a.a. E1A at Phe(187) and/or Tyr(189) using chymotrypsin. The impact of deletion of the 189 N-terminal amino acids from E1A on the equilibrium dissociation constant K(d) for binding to CtBP was assessed using ELISA in vitro binding assays and intrinsic fluorescence spectroscopy. Values of K(d) of 4.0 and 38 nM were determined for full-length and truncated forms of E1A, respectively. Circular dichroism spectroscopic studies revealed that the conformation adopted by these polypeptides is dependent on the surrounding environment, which is predominately randomly folded when free in solution, but adopting a more ordered alpha-helical secondary structure in the presence of trifluoroethanol. Using nuclear magnetic resonance (NMR) spectroscopy to examine the interaction between Ad E1A and CtBP it was observed that the chemical shift positions of individual backbone amide nitrogen atoms were well resolved in (15)N-(1)H-HSQC NMR spectra performed on samples of isotopically (15)N-labeled Ad12 77-a.a. E1A. In the presence of CtBP, signals of backbone amide nitrogen atoms displayed increased linewidth consistent with an increase in molecular mass upon binding CtBP. In addition, some signals that have been attributed to Val(254/256) and Leu(259), and reside within the binding site for CtBP on E1A, are shifted in the (15)N- and/or (1)H-dimensions, defining specific contacts between E1A and CtBP. These data suggest that structural determinants in the C-terminal PXDLS binding motif in the rest of exon 2 and in exon 1 all contribute to optimizing the conformation of the binding site on Ad12 E1A for CtBP. However, no interaction was observed between CtBP and truncated Ad12 E1A, which no longer contained the C-terminal binding motif.

Regulation of the 26S proteasome by adenovirus E1A.

We have identified the N-terminus of adenovirus early region 1A (AdE1A) as a region that can regulate the 26S proteasome. Specifically, in vitro and in vivo co-precipitation studies have revealed that the 19S regulatory components of the proteasome, Sug1 (S8) and S4, bind through amino acids (aa) 4-25 of Ad5 E1A. In vivo expression of wild-type (wt) AdE1A, in contrast to the N-terminal AdE1A mutant that does not bind the proteasome, reduces ATPase activity associated with anti-S4 immunoprecipitates relative to mock-infected cells. This reduction in ATPase activity correlates positively with the ability of wt AdE1A, but not the N-terminal deletion mutant, to significantly reduce the ability of HPV16 E6 to target p53 for ubiquitin-mediated proteasomal degradation. AdE1A/proteasomal complexes are present in both the cytoplasm and the nucleus, suggesting that AdE1A interferes with both nuclear and cytoplasmic proteasomal degradation. We have also demonstrated that wt AdE1A and the N-terminal AdE1A deletion mutant are substrates for proteasomal-mediated degradation. AdE1A degradation is not, however, mediated through ubiquitylation, but is regulated through phosphorylation of residues within a C-terminal PEST region (aa 224-238).

Neutralisation of adenovirus infectivity by ascitic fluid from ovarian cancer patients.

Animal models and phase I clinical trials have shown that repeat virus delivery and subsequent transgene expression is limited by the generation of humoral and cellular immune responses directed towards the therapeutic vector. The presence of a pre-existing immune response may even prevent initial delivery. In order to determine the presence of pre-existing anti-adenovirus humoral immunity we analysed ascitic fluid, collected from the peritoneal cavity of patients with advanced ovarian cancer. Twelve ascitic fluid and four matched serum samples were examined. The titre and isotype of anti-adenovirus antibodies was determined by ELISA, and Western blotting identified the molecular basis of the immune response, which was primarily directed towards fibre and penton base. Neutralisation of virus infectivity was assessed in vitro by measurement of green fluorescent protein reporter gene expression. We found that the ascitic fluid samples contain antibodies that recognise both adenovirus types 2 and 5, were predominantly IgG and directed towards the viral antigens responsible for cell adhesion, and had virus neutralising activity.

Consequences of disruption of the interaction between p53 and the larger adenovirus early region 1B protein in adenovirus E1 transformed human cells.

The adenovirus early region 1B (Ad E1B) genes have no transforming capability of their own but markedly increase the transformation frequency of Ad E1A following co-transfection into mammalian cells. The larger E1B proteins of both Ad2/5 and Ad12 bind to p53 and inhibit its ability to transcriptionally activate other genes. We have previously demonstrated that synthetic peptides identical to the binding sites for p53 on both the Ad2 and Ad12 E1B proteins will disrupt the interaction in vivo and in vitro. In the work presented here we have examined the effects of complex dissociation on Ad E1-transformed human cells. It has been shown, using confocal microscopy, that when the peptide identical to the p53 binding site was added to Ad5 E1-transformed cells it initally located in the cytoplasmic dense bodies where it caused disruption of the p53/E1B complex. Peptide and p53 then translocated to the nucleus. In Ad12 E1-transformed cells the peptide localized in the nucleus directly and there caused a reorganization of p53 staining from a highly organized, 'flecked' distribution to one in which nuclear staining was homogeneous and diffuse. Peptides added to either Ad5 E1 or Ad12 E1 transformed cells resulted in the release of transcriptionally active p53. Interestingly, the level of p53 then fell presumably as a result of proteasomal action - this was probably a reflection of the short half-life of 'free' (i.e. dissociated) p53 compared to that of the bound protein. Free p53 did not cause apoptosis in target cells probably due to the presence of the smaller (19K) E1B proteins. However, addition of peptide leads to a significant reduction in cell growth rate. We have further demonstrated that a significant proportion of those cells which had taken up peptide had ceased DNA synthesis, probably due to a p53-induced cell cycle arrest. The role of the larger EIB protein during transformation is considered in view of these data.

The replicative capacities of large E1B-null group A and group C adenoviruses are independent of host cell p53 status.

Recent reports suggest that an early region 1B (E1B) 55, 000-molecular-weight polypeptide (55K)-null adenovirus type 5 (Ad5) mutant (dl1520) can replicate to the same extent as wild-type (wt) Ad5 in cells either deficient or mutated in p53, implicating p53 in limiting viral replication in vivo. In contrast, we show here that the replicative capacity of Ad5 dl1520 is wholly independent of host cell p53 status, as is the replicative capacity of comparable Ad12 E1B 54K-null adenoviruses (Ad12 dl620 and Ad12 hr703). Furthermore, we show that there is no requirement for complex formation between p53 and Ad5 E1B 55K or Ad12 E1B 54K for a productive infection, such that wt Ad5 and wt Ad12 will both replicate in cells which are null for p53. In addition, we find that these Ad5 and Ad12 mutant viruses induce S phase irrespective of the p53 status of the cell and that, therefore, S-phase induction does not correlate with the replicative capacity of the virus. Interestingly, the replicative capacities of the large E1B-null adenoviruses correlated positively with the ability to express E1B 19K and were related to the ability to repress premature adenovirus-induced apoptosis. Infection of primary human cells indicated that Ad5 dl1520, wt Ad5, and wt Ad12 replicated better in cycling normal human skin fibroblasts (HSFs) than in quiescent HSFs. Thus, the cell cycle status of the host cell, upon infection, also influences viral yield.

Adenovirus early region 1A protein binds to mammalian SUG1-a regulatory component of the proteasome.

Adenovirus early region 1A (Ad E1A) is a multifunctional protein which is essential for adenovirus-mediated transformation and oncogenesis. Whilst E1A is generally considered to exert its influence on recipient cells through regulation of transcription it also increases the level of cellular p53 by increasing the protein half-life. With this in view, we have investigated the relationship of Ad E1A to the proteasome, which is normally responsible for degradation of p53. Here we have shown that both Ad5 and Ad12 E1A 12S and 13S proteins can be co-immunoprecipitated with proteasomes and that the larger Ad12 E1A protein binds strongly to at least three components of the 26S but not 20S proteasome. One of these interacting species has been identified as mammalian SUGI, a proteasome regulatory component which also plays a role in the cell as a mediator of transcription. In vitro assays have demonstrated a direct interaction between Ad12 E1A 13S protein and mouse SUGI. Following infection of human cells with Ad5 wt and Ad5 mutants with lesions in the E1A gene it has been shown that human SUG1 can be co-immunoprecipitated with full-length E1A and with E1A carrying a deletion in conserved region 1 which is the region considered to be responsible for increased expression of p53. We have concluded therefore that Ad EIA binds strongly to SUGI but that this interaction is not responsible for inhibition of proteasome activity. This is consistent with the observation that purified Ad12 E1A inhibits the activity of the purified 20S but not 26S proteasomes. We have also demonstrated that SUGI can be co-immunoprecipitated with SV40 T and therefore we suggest that this may represent a common interaction of transforming proteins of DNA tumour viruses.

Definition of a major p53 binding site on Ad2E1B58K protein and a possible nuclear localization signal on the Ad12E1B54K protein.

Previous studies have established that adenovirus 2/5 early region 1B (Ad E1B) 58K protein binds p53 strongly and co-localizes with it to cytoplasmic dense bodies whilst the homologous Ad12E1B54K protein binds only weakly and co-localizes primarily to the nucleus in Ad12E1 transformed cells. We have used these properties of the E1B proteins from different viral serotypes to map the p53 binding site on the Ad2/5 protein. A set of chimaeric genes was constructed containing different proportions of the Ad12 and Ad2E1B DNA. These, together with Ad12E1A and E1B19K DNA, were transfected into baby rat kidney cells and transformed lines isolated. From an examination of the properties of these Ad12/Ad2E1B fusion proteins in co-immunoprecipitation and subcellular localization experiments it has been concluded that the p53 binding site on Ad2E1B58K protein lies between amino acids 216 and 235 and that the homologous region on Ad12E1B54K protein also binds p53. In addition, a unique nuclear localization signal is located on Ad12E1B54K between residues 228 and 239. We suggest that primary structure differences in these regions of the Ad2 and Ad12E1B proteins are responsible for the different subcellular localizations in AdE1 transformants.

The structure of the site on adenovirus early region 1A responsible for binding to TATA-binding protein determined by NMR spectroscopy.

Previous detailed mutational analysis has shown that the binding site on adenovirus (Ad) early region 1A (E1A) for TATA-binding protein (TBP) is located toward the N terminus of conserved region 3 (CR3). Here we demonstrate that synthetic peptides of between 15 and 22 amino acids, identical to amino acid sequences of CR3 present in the larger Ad5 E1A (13 S product) and in both the Ad12 E1A (13 and 12 S products) proteins that lie N-terminal to the zinc finger motif, can disrupt binding of E1A to TBP. These findings suggest that the peptides are biologically active in terms of interacting with TBP and must therefore comprise some, if not all, of the TBP binding site on E1A. The interaction between Ad12 E1A and TBP was confirmed by direct co-precipitation experiments. In 1H NMR studies of CR3 peptides, regular patterns of NOEs were observed from which their conformational preferences in aqueous solution were determined. Both Ad5 and Ad12 peptides were shown to contain regions of helical backbone structure in 50% trifluoroethanol. In each case, the type and intensities of NOE cross-peaks observed correlated best to alpha-helical turns. These helices are more extensive in larger peptides and extend from Glu141 to Val147 and from Arg144 to Pro152 in the full-length Ad5 and Ad12 13S E1A proteins, respectively. The structure of a 19-residue Ad5 CR3 peptide carrying the V147L mutation in the full-length protein that abolishes TBP binding was examined. No significant differences between the substituted and wild type peptides were observed, suggesting that this substitution in the intact protein may cause disruption of global rather than local structures.

Structural aspects of the interaction between heterogeneic human papillomavirus type 1 E4-specific T cell receptors and the same peptide/HLA-DQ8 complex.

TCR usage has been studied in a panel of Th cell clones specific for the same peptide epitope (P N S Q D R G R P R R S D), derived from the human papillomavirus type 1 (HPV1) E4 protein, and restricted through HLA-DQ8. After identifying the V, D, and J genes used by the TCRs and sequencing across the V(D)J junctions, five different alpha-chain sequences and five different beta-chain sequences, comprising six independent clones, were identified. A structural model of our E4 peptide/HLA-DQ8 complex predicted that the guanidinyl side chain on the arginine residue at position 6 of the peptide could exist in different orientations. An intramolecular interaction between this arginine and the glutamine residue at position four appeared to control this orientation. Interacting HPV1 E4-specific TCRs would therefore have to recognize the complex in different conformations, and molecular modeling of the TCRs suggested that this could be achieved by changing the dimensions of the central pocket formed where the CDR3 loops of the TCR alpha- and beta-chains converge. It is known that interactions between bound peptide and amino acid residues lining the peptide-binding cleft of HLA molecules are important for determining the conformation and orientation of the peptide/MHC complex. The suggestion here that intramolecular interactions between amino acids of close proximity on the bound peptide are also important adds a further level of complexity to the mechanism by which TCRs interact with Ag.

Structural determinants present in the C-terminal binding protein binding site of adenovirus early region 1A proteins.

The C-terminal binding protein (CtBP) has previously been shown to bind to a highly conserved six-amino acid motif very close to the C terminus of adenovirus early region 1A (Ad E1A) proteins. We have developed an enzyme-linked immunosorbent assay that has facilitated the screening of synthetic peptides identical or similar to the binding site on Ad E1A for their ability to bind CtBP and thus inhibit its interaction with Ad12 E1A. It has been shown that amino acids both C-terminal and N-terminal to the original proposed binding site contribute to the interaction of peptides with CtBP. Single amino acid substitutions across the binding site appreciably alter the Kd of the peptide for CtBP, indicative of a marked reduction in the affinity of the peptide for CtBP. The solution structures of synthetic peptides equivalent to the C termini of both Ad5 and Ad12 E1A and two substituted forms of these have been determined by proton NMR spectroscopy. Both the Ad12 and Ad5 peptides dissolved in trifluoroethanol/water mixtures were found to adopt regular secondary structural conformations seen as a series of beta-turns. An Ad12 peptide bearing a substitution that resulted in only very weak binding to CtBP (Ad12 L258G) was found to be random coil in solution. However, a second mutant (Ad12 V256K), which bound to CtBP rather more strongly (although not as well as the wild type), adopted a conformation similar to that of the wild type. We conclude that secondary structure (beta-turns) and an appropriate series of amino acid side chains are necessary for recognition by CtBP.

Human cells arrest in S phase in response to adenovirus 12 E1A.

It has previously been shown that following viral infection, Ad5 E1A induces cell cycle progression of quiescent rodent cells, leading to DNA synthesis and mitosis. Here we have examined the effect of Ad12 E1A on the cell cycle characteristics of human cells. Human tumor (A549, KB, and HeLa) cells were infected with Ad12 d/620, a mutant virus which has a lesion in the E1B gene and essentially expresses only E1A. These infected cells progressed from being largely in G1 into S phase, where they arrested. Even up to 96 h postinfection (p.i.) the cells remained blocked in S phase. DNA synthesis did, however, proceed in Ad12 d/620-infected cells, giving rise to multiple copies of cellular DNA. Similar results were obtained when primary human skin fibroblasts were infected, although the polyploidy was less marked. The expression of cyclins A, B1, and E in the tumor cells increased appreciably in response to E1A. In contrast, there was a dramatic reduction in the levels of cyclin D1 and D3. Increases in cyclin D1 expression could be detected at very late times p.i. In those cell lines expressing low levels of cdc2 and cdk2 an appreciable increase in expression was seen soon after Ad12 E1A could be detected. The elevated levels of cyclins A, B1, and E were associated with increased protein kinase activity directed against histone H1. An increase in cyclin D1-associated kinase activity against Rb1 was also observed at late times. This deregulation of the cell cycle was not solely dependent on E1A inactivation of Rb, since similar effects were seen in Ad12 d/620-infected retinoblastoma (Y-79) cells, implicating p107 and p130 in E1A-mediated changes in cell cycle progression. We propose that the E1A-induced levels of cyclins A, B1, and E by Ad12 E1A in human cells may lead to an uncoupling of S phase from cell cycle progression.

Regeneration of the binding properties of adenovirus 12 early region 1A proteins after preparation under denaturing conditions.

Adenovirus 12 early region 1A (Ad12 E1A) was expressed in Escherichia coli. Protein was purified in good yield in the presence of 8 M urea and then renatured by dialysis against dilute NH4HCO3 buffer. The affinity of this protein for pRb, C-terminal binding protein (CtBP), TATA binding protein (TBP), and SUG1 was similar to, or greater than, that of Ad12 E1A prepared by immunoaffinity chromatography under nondenaturing conditions. While the binding of the 266- and 235-amino-acid (aa) E1A components to TBP showed similar characteristics the larger E1A protein had a higher affinity for CtBP, pRb, and SUG1. Using nuclear magnetic resonance (NMR) spectroscopy it was shown that structural perturbations occurred in the 266-aa protein in the presence of Zn2+ consistent with binding--no such changes were seen for the 235-aa protein. Limited proteolysis of the 266- and 235-aa E1A proteins gave rise to comparable polypeptide products, suggesting overall similarities in structure. However, the different affinities of the 266- and 235-aa proteins for the partner proteins and the differences seen in the NMR spectra from the two proteins suggested structural differences.

Identification of conserved hydrophobic C-terminal residues of the human papillomavirus type 1 E1E4 protein necessary for E4 oligomerisation in vivo.

Previous studies have shown that human papillomavirus (HPV) E4 proteins undergo oligomerisation, although the precise sequences involved have not been identified. Using the yeast two-hybrid system we have identified HPV 1 E4 sequences that are critical to multimerisation. Fusion proteins were created by linking wild-type and mutant E4 proteins to a LexA DNA-binding domain or a B42 transactivation domain. HPV 1 E4:E4 interactions were examined by expression of these fusion proteins in Saccharomyces cerevisiae. This assay showed that (1) amino acid residues 95 to 115 at the carboxy-terminus were critical for oligomerisation and (2) hydrophobic residues (isoleucine 107, phenylalanine 114) in this domain are major determinants in the formation of oligomers. Interestingly, the carboxy-terminal domain shares homology with other E4 proteins of cutaneous HPV types and, furthermore, positions 107 and 114 are conserved residues. Substitution of the conserved aspartate amino acids (residues 110 and 112) did not abrogate E4 oligomerisation. Chemical cross-linking of wart and recombinant (baculovirus-expressed) HPV 1 E4 protein indicated that in solution this viral protein forms complexes consistent in size with either trimers or tetramers. These complexes were resistant to urea denaturation and are not dependent on the formation of disulphide linkages. A mutant protein containing a deletion of residues 110 to 115 was unable to form oligomers following cross-linking supporting a role for this region in mediating E4:E4 interactions. We conclude that oligomerisation of the HPV 1 E4 protein is likely to be mediated by carboxy-terminal residues and that conserved hydrophobic residues of this domain play a major role in E4 oligomerisation.

Adenovirus type 12 early region 1B 54K protein significantly extends the life span of normal mammalian cells in culture.

The life span of normal human cells in culture is extended by two to four total life spans following retrovirus-mediated transfer of the adenovirus type 12 E1B 54,000-molecular-weight protein (54K protein). This extension of the in vitro growth potential was accomplished without any of the obvious changes in morphology or growth properties that are usually associated with viral transformation. These 54K+ cells escape the normal senescence checkpoint (M1) and show a very extended secondary growth phase. The 54K+ human cells eventually enter crisis (M2), which does not appear to be due to either telomere attrition or the activation of the senescence-associated proteins p21SdilCipIWaf1 and p16INK4A. Even in the absence of telomerase activity, high-molecular-weight heterogeneous telomeres are produced and maintained in both 54K+ adult dermal fibroblasts and embryo kidney cells, indicating that the 54K protein may interfere with the normal metabolism of telomeric structures during cell division. These findings are discussed with reference to the known ability of the 54K protein to influence p53 function.

Differentiation is inhibited and a senescence pathway is activated when simian virus 40 tsA 58-transformed human retinoblasts are grown at the restrictive temperature.

Neonatal human retina cells transformed by the SV40 tumor antigens were shown to leave the cell cycle and differentiate following treatment with agents that raise intracellular levels of cyclic AMP. This was true for both precrisis and immortal cell lines. However, with time, some of the differentiated retinoblasts withdrew neurites and returned to the cell cycle. Attempts to inhibit this process by developing cell lines transformed using SV tsA 58 with a temperature-sensitive phenotype for growth did not enhance but inhibited retinoblast-differentiating capacity. Growth restriction at the nonpermissive temperature was found to activate a senescence pathway. We propose that at the nonpermissive temperature, stable SV40 T-ag-p53 complexes fragment releasing p53, which transactivates p21waf1/cip1/sdi1 with the subsequent accumulation of p21 culminating in growth inhibition and senescence.

Mutational analysis of the human papillomavirus type 16 E1--E4 protein shows that the C terminus is dispensable for keratin cytoskeleton association but is involved in inducing disruption of the keratin filaments.

The function of the human papillomavirus (HPV) E4 proteins is unknown. In cultured epithelial cells the proteins associate with the keratin intermediate filaments (IFs) and, for some E4 types, e.g., HPV type 16 (HPV-16), induce collapse of the keratin networks. An N-terminal leucine-rich motif (LLXLL) is a conserved feature of many E4 proteins. In a previous study we showed that deletion of this region from the HPV-1 and -16 E4 proteins abrogated the localization of the mutant proteins to the keratin cytoskeleton in a simian virus 40-transformed human keratinocyte cell line (S. Roberts, I. Ashmole, L. J. Gibson, S. M. Rookes, G. J. Barton, and P. H. Gallimore, J. Virol. 68:6432-6445, 1994). The E4 proteins of HPV-1 and -16 have little sequence homology except at the N terminus. Therefore, to establish the role of sequences other than those at the N terminus, we have performed a mutational analysis of the HPV-16 E4 protein. The results of the analysis were as follows: (i) similar to findings for the HPV-1 protein, no mutation of HPV-16 E4 sequences (other than the N-terminal leucine motif) results in a mutant protein which fails to colocalize to the keratin IFs; (ii) the C-terminal domain (residues 61 to 92) is not essential for association with the cytoskeleton; and (iii) deletion of C-terminal sequences (residues 84 to 92; LTVIVTLHP) corresponding to part of a domain conserved between mucosal E4 proteins affects the ability of the mutant protein to induce cytoskeletal collapse, despite colocalization with the keratin IFs. Further analysis of this region showed that conserved hydrophobic residues valines 86 and 88 are important. In addition, we show that the HPV-16 E4 protein is detergent insoluble and exists as several disulfide-linked, high-molecular-weight complexes which could represent homo-oligomers. The C-terminal sequences (residues 84 to 92), in particular valines 86 and 88, are important in the formation of these insoluble complexes. The results of this study support our postulate that the E4 proteins include functional domains at the N terminus and the C terminus, with the intervening sequences possibly acting as a flexible hinge.