Search

The human papillomavirus (HPV) capsid is primarily composed of a structural protein denoted L1, which forms both pentameric capsomeres and capsids composed of 72 capsomeres. The L1 protein alone is capable of self-assembly in vivo into capsidlike structures referred to as viruslike particles (VLPs). We have determined conditions for the quantitative disassembly of purified HPV-11 L1 VLPs to the level of capsomeres, demonstrating that disulfide bonds alone are essential to maintaining long-term HPV-11 L1 VLP structure at physiological ionic strength. The ionic strength of the disassembly reaction was also important, as increased NaCl concentrations inhibited disassembly. Conversely, chelation of cations had no effect on disassembly. Quantitative reassembly to a homogeneous population of 55-nm, 150S VLPs was reliably achieved by the re-formation of disulfide linkages following removal of reducing agent at near-neutral pH and moderate NaCl concentration. HPV-11 L1 VLPs could also be dissociated by treatment with carbonate buffer at pH 9.6, but VLPs could not be regenerated following carbonate treatment. When probed with conformationally sensitive and/or neutralizing monoclonal antibodies, both capsomeres generated by disulfide reduction of purified VLPs and reassembled VLPs formed from capsomeres upon removal of reducing agents exhibited epitopes found on the surface of authentic HPV-11 virions. Antisera raised against either purified VLP starting material or reassembled VLPs similarly neutralized infectious HPV-11 virions. The ability to disassemble and reassemble VLPs in vitro and in bulk allows basic features of capsid assembly to be studied and also opens the possibility of packaging selected exogenous compounds within the reassembled VLPs.

Human adenovirus type 9 (Ad9) is unique among oncogenic adenoviruses in that it elicits exclusively mammary tumors in rats and requires the viral E4 region open reading frame 1 (9ORF1) gene for tumorigenicity. The 9ORF1 oncogenic determinant codes for a 14-kDa transforming protein, and three separate regions of this polypeptide, including one at the extreme C terminus, are necessary for transforming activity. In this study, we investigated whether the 9ORF1 transforming protein interacts with cellular factors. Following incubation with cell extracts, a glutathione S-transferase (GST)-9ORF1 fusion protein associated with several cellular phosphoproteins (p220, p180, p160, p155), whereas GST fusion proteins of transformation-defective 9ORF1 C-terminal mutants did not. Similar interactions requiring the 9ORF1 C terminus were revealed with protein-blotting assays, in which a GST-9ORF1 protein probe reacted specifically with cellular polypeptides having gel mobilities resembling those of the 9ORF1-associated cellular phosphoproteins, as well as with additional cellular polypeptides designated p140/p130. In addition, GST fusion proteins containing 9ORF1 C-terminal fragments associated with some of the 9ORF1-associated cellular polypeptides, as did GST fusion proteins of full-length wild-type Ad5 and Ad12 E4 ORF1 transforming proteins. Significantly, the results of coimmunoprecipitation analyses suggested that the same cellular polypeptides also associate with wild-type but not C-terminal-mutant 9ORF1 proteins in vivo. Together, these findings suggest that the 9ORF1 C terminus, which is essential for transformation, participates in specific and direct binding of the 9ORF1 oncoprotein to multiple cellular polypeptides. We propose that interactions with these cellular factors may be responsible, at least in part, for the transforming activity of the 9ORF1 viral oncoprotein.

Papillomavirus early gene expression is regulated by the virus gene-encoded E2 proteins. The best-characterized E2 protein, encoded by bovine papillomavirus type 1 (BPV-1), has been shown to interact with basal transcription factor IIB (TFIIB) and the TATA binding protein basal transcription factor (N. M. Rank and P. F. Lambert, J. Virol. 69:6323-6334, 1995). We demonstrate that the potent E2 transcriptional activator protein encoded by a gene of human PV type 16 also interacts with TFIIB in vitro. Moreover, a direct comparison of domains within human TFIIB (hTFIIB) required for VP16 and BPV-1 E2 indicates that these acidic activators interact with hTFIIB in a qualitatively similar manner. Our mapping experiments identify hTFIIB interaction domains within the amino-terminal activation domain of BPV-1 E2. Finally, we demonstrate in vitro interaction between Saccharomyces cerevisiae TFIIB and BPV-1 E2, an observation that is consistent with the importance of the E2-TFIIB interaction for BPV-1 E2 transactivation in both systems.

The L1 major capsid protein of human papillomavirus type 11 (HPV-11) was expressed in Escherichia coli, and the soluble recombinant protein was purified to near homogeneity. The recombinant L1 protein bound DNA as determined by the Southwestern assay method, and recombinant mutant L1 proteins localized the DNA-binding domain to the carboxy-terminal 11 amino acids of L1. Trypsin digestion of the full-length L1 protein yielded a discrete 42-kDa product (trpL1), determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, resulting from cleavage at R415, 86 amino acids from the L1 carboxy terminus. Sucrose gradient sedimentation analysis demonstrated that trpL1 sedimented at 11S, while L1 proteins with amino-terminal deletions of 29 and 61 residues sedimented at 4S. Electron microscopy showed that the full-length L1 protein appeared as pentameric capsomeres which self-assembled into capsid-like particles. The trpL1 protein also had a pentameric morphology but was unable to assemble further. In an enzyme-linked immunosorbent assay, the trpL1 and L1 capsids reacted indistinguishably from virus-like particles purified after expression of HPV-11 L1 in insect cells. The carboxy terminus of L1 therefore constitutes the interpentamer linker arm responsible for HPV-11 capsid formation, much like the carboxy-terminal domain of the polyomavirus VP1 protein. The trypsin susceptibility of HPV-11 L1 capsids suggests a possible mechanism for virion disassembly.

To assess the potential for cross-protection among genital human papillomavirus (HPV) types in virus-like particle (VLP)-based vaccinations, inhibition of HPV VLP-mediated hemagglutination by rabbit antisera raised against HPV type 6b (HPV-6b), HPV-11, HPV-16, HPV-18, HPV-31, HPV-33, and HPV-45 was analyzed. Only highly homologous types (HPV-6b and HPV-11, and HPV-18 and HPV-45) exhibited detectable serological cross-reaction for the class of antibodies that inhibit virion-to-cell surface binding. However, analysis of neutralizing monoclonal antibodies to several animal and human papillomaviruses indicated that over half of these antibodies do not prevent cell surface binding, but these latter antibodies do not appear to be more cross-reactive in enzyme-linked immunosorbent assays than those that mediate inhibition of hemagglutination. The data strongly suggest that while there may be limited cross-protection between highly (>85% L1 amino acid identity) homologous types, protection by HPV VLP-based vaccines will be predominantly type specific.

The bovine papillomavirus E5 protein is a 44-amino-acid membrane-associated protein that forms a stable complex with the endogenous platelet-derived growth factor (PDGF) beta receptor in rodent and bovine fibroblasts, resulting in sustained receptor activation and cell transformation. We report here that high-level expression of the E5 protein caused a reduction in the level of the mature form of the PDGF beta receptor in acutely and stably transformed mouse C127 cells. To explore in more detail the interaction of the E5 protein and the PDGF beta receptor, we tested the abilities of various E5 point mutants to bind the PDGF receptor, to induce PDGF receptor down-regulation and tyrosine phosphorylation, and to transform cells. A transformation-competent mutant, like the wild-type E5 protein, bound the receptor and induced receptor tyrosine phosphorylation and down-regulation. Transformation-defective E5 proteins either failed to interact with the endogenous PDGF beta receptor in mouse fibroblasts or underwent an aberrant interaction with the receptor. Mutation of glutamine at position 17, aspartic acid at position 33, or both carboxyl-terminal cysteine residues required for E5 homodimerization interfered with stable complex formation with the PDGF receptor, tyrosine phosphorylation and down-regulation of the receptor, and cell transformation. Point mutations at several other carboxyl-terminal positions generated transformation-defective E5 proteins that formed a complex with the PDGF receptor and induced receptor tyrosine phosphorylation but did not induce PDGF receptor down-regulation. Either PDGF receptor activation is not sufficient for transformation of C127 cells or the receptors that are tyrosine phosphorylated in response to these mutant E5 proteins are not fully activated and therefore are not able to deliver a mitogenic signal.

The constitutively active promoter P7535 of the epidermodysplasia verruciformis-associated human papillomavirus type 8 (HPV8) is transactivated by the viral E2 protein. The distribution of potential E2-binding sites (ACCN6GGT) in the viral transcription control region is highly conserved among epidermodysplasia verruciformis-associated human papillomaviruses and differs completely from that of other papillomaviruses. To investigate the role of E2-binding sites P0 to P4 in P7535 regulation, we analyzed their binding affinities in gel retardation experiments using a full-length HPV8 E2 protein expressed from a recombinant baculovirus. Binding site P1 within a transcriptional silencer showed the highest affinity, followed by P0 within the L1 gene and P3 downstream of P7535. P2, 33 nucleotides upstream of the mRNA cap site, and P4 were very weak binders. There is some indication that the number of A/T pairs in the nonconserved core of the recognition sequence is critical for the binding of HPV8 E2. Transient transfection experiments were carried out with an HPV8 E2 expression vector and reporter plasmids containing mutated E2-binding sites in the context of the HPV8 regulatory region. The knockout of the strongest binding site P1 sufficed to clearly diminish transactivation. P0, P3, and P4 mutations had little effect on their own, whereas double mutations P01 and P34 strongly reduced E2 inducibility. Both mutations in P2 severely affected constitutive promoter activity but had opposite effects on transactivation. They revealed an inverse correlation between E2-binding strength and the extent of E2 transactivation. This finding suggests that P2 mediates a negative control of P7535 by E2, counteracting E2 transactivation exerted via the four distal E2 target sequences.

The E6 promoters of all genital human papillomaviruses have a characteristic alignment of transcription factor binding sites. Activation of the basic transcription complex at the TATA box depends upon a sequence-aberrant Sp1 site. Repression of E6 promoters is achieved by two binding sites for the viral E2 protein positioned between the Sp1 site and the TATA box. We have purified the human papillomavirus type 16 E2 protein after expression in Escherichia coli and studied its binding and repression properties with oligonucleotides representing the homologous promoter sequences. A Kd value of 3 x 10(-10) M indicated binding properties expected for a native protein. We found low cooperativity in the binding of two E2 dimers to flanking sites, both when these sites were separated by 3 nucleotides, as in the natural promoter, and when they were further apart. E2 protein, bound close to the distal Sp1 site, displaced the Sp1 factor even when the aberrant sequence was replaced by a typical Sp1 core recognition site. The high affinity of E2 protein for its binding site even led to Sp1 displacement at concentrations of E2 protein nearly 2 orders of magnitude lower than those of Sp1. Functional analyses of mutated E6 promoter sequences showed repression by this distal E2 binding site in the complete absence of binding to the proximal E2 binding site. From our findings and observations published by others, we conclude that each of the E2 binding sites in the E6 promoter of genital human papillomaviruses plays a separate role by displacing the transcription factors Sp1 and TFIID.

The bovine papillomavirus E5 transforming protein appears to activate both the epidermal growth factor receptor (EGF-R) and the platelet-derived growth factor receptor (PDGF-R) by a ligand-independent mechanism. To further investigate the ability of E5 to activate receptors of different classes and to determine whether this stimulation occurs through the extracellular domain required for ligand activation, we constructed chimeric genes encoding PDGF-R and EGF-R by interchanging the extracellular, membrane, and cytoplasmic coding domains. Chimeras were transfected into NIH 3T3 and CHO(LR73) cells. All chimeras expressed stable protein which, upon addition of the appropriate ligand, could be activated as assayed by tyrosine autophosphorylation and biological transformation. Cotransfection of E5 with the wild-type and chimeric receptors resulted in the ligand-independent activation of receptors, provided that a receptor contained either the transmembrane domain of the PDGF-R or the cytoplasmic domain of the EGF-R. Chimeric receptors that contained both of these domains exhibited the highest level of E5-induced biochemical and biological stimulation. These results imply that E5 activates the PDGF-R and EGR-R by two distinct mechanisms, neither of which specifically involves the extracellular domain of the receptor. Consistent with the biochemical and biological activation data, coimmunoprecipitation studies demonstrated that E5 formed a complex with any chimera that contained a PDGF-R transmembrane domain or an EGF-R cytoplasmic domain, with those chimeras containing both domains demonstrating the greatest efficiency of complex formation. These results suggest that although different domains of the PDGF-R and EGF-R are required for E5 activation, both receptors are activated directly by formation of an E5-containing complex.

Human papillomavirus type 18 (HPV-18) E7 proteins bind zinc through Cys-X-X-Cys repeats located at the C terminus of the protein. In order to examine the role of these cysteine motifs in E7 function, we expressed the HPV-18 E7 protein in bacteria and found that purified E7 forms a dimer through interactions with zinc. Mutants with single mutations within the Cys-X-X-Cys motifs bound a reduced level of zinc in a zinc blot assay, while a double mutant lost all zinc-binding activity. When expressed in vivo, none of the mutants cooperated with an activated ras oncogene to transform primary rat embryo fibroblasts, but all mutants retained nearly wild-type Rb-binding activity. The results indicate that the cysteine motifs play an important role in transformation by HPV-18 E7 but do not contribute to Rb binding.

Vaccinia virus vectors were used to express the major (L1) and minor (L2) capsid proteins of human papillomavirus type 1 (HPV-1) with the vaccinia virus early (p7.5K) or late (pSynth, p11K) promoters. All constructs expressed the appropriate-sized HPV proteins, and both L1 and L2, singly or in combination, localized to the nucleus. Capsids were purified by cesium chloride density gradient centrifugation from nuclei of cells infected with a vaccinia virus-L1 (vac-L1) recombinant or a vac-L1-L2 recombinant but not from vac-L2-infected cells. Electron microscopy showed that the particles were 55 nm in diameter and had icosahedral symmetry. Immunogold-labeled antibodies confirmed the presence of the L1 and L2 proteins in the HPV-1 capsids. Capsids containing L1 alone were fewer and more variable in size and shape than capsids containing the L1 and L2 proteins. The L1-plus-L2 capsids were indistinguishable in appearance from HPV-1 virions obtained from plantar warts. The ability to produce HPV capsids in vitro will be useful in many studies of HPV pathogenicity.

Studies of adenovirus E1A oncoprotein mutants suggest that the association of E1A with the retinoblastoma protein (pRB) is necessary for E1A-mediated transformation. Mutational analysis of E1A indicates that two regions of pRB are required for E1A to form stable complexes with the retinoblastoma protein. In addition to pRB binding, these regions are necessary for E1A association with several other cellular proteins, including p130, p107, cyclin A, and p33cdk2. Here we show that short synthetic peptides containing the pRB-binding sequences of E1A are sufficient for interaction with p107, cyclin A, and p130. The E7 protein of human papillomavirus type 16 contains an element that binds to pRB and appears to be functionally homologous to the E1A sequences. Peptides containing this region of the E7 protein were able to interact with p107, cyclin A, and p130 in addition to pRB. These findings suggest that the common mechanism of transformation used by these viral oncogenes involves their association with a set of polypeptides.

To determine the sequences of the oncogenes src (encoded by Rous sarcoma virus [RSV]) and ros (encoded by UR2) that are responsible for causing different transformation phenotypes and to correlate those sequences with differences in substrate recognition, we constructed recombinants of the two transforming protein tyrosine kinases (PTKs) and studied their biological and biochemical properties. A recombinant with a 5' end from src and a 3' end from ros, called SRC x ROS, transformed chicken embryo fibroblasts (CEF) to a spindle shape morphology, mimicking that of UR2. Neither of the two reverse constructs, ROS x SRC I and ROS x SRC II, could transform CEF. However, a transforming variant of ROS x SRC II appeared during passages of the transfected cells and was called ROS x SRC (R). ROS x SRC (R) contains a 16-amino-acid deletion that includes the 3' half of the transmembrane domain of ros. Unlike RSV, ROS x SRC (R) also transformed CEF to an elongated shape similar to that of UR2. We conclude that distinct phenotypic changes of RSV- and UR2-infected cells do not depend solely on the kinase domains of their oncogenes. We next examined cellular proteins phosphorylated by the tyrosine kinases of UR2, RSV, and their recombinants as well as a number of other avian sarcoma viruses including Fujinami sarcoma virus Y73, and some ros-derived variants. Our results indicate that the UR2-encoded receptorlike PTK P68gag-ros and its derivatives have a very restricted substrate specificity in comparison with the nonreceptor PTKs encoded by the rest of the avian sarcoma viruses. Data from ros and src recombinants indicate that sequences both inside and outside the catalytic domains of ros and src exert a significant effect on the substrate specificity of the two recombinant proteins. Phosphorylation of most of the proteins in the 100- to 200-kDa range correlated with the presence of the 5' src domain, including the SH2 region, but not with the kinase domain in the recombinants. This corroborates the conclusion given above that the kinase domain of src or ros per se is not sufficient to dictate the transforming morphology of these two oncogenes. High-level tyrosyl phosphorylation of most of the prominent substrates of src is not sufficient to cause a round-shape transformation morphology.

We have previously demonstrated that a purified bacterially synthesized E1A 289-amino-acid protein is capable of stimulating transcription from the promoters of genes transcribed by RNA polymerase II in vitro (R. Spangler, M. Bruner, B. Dalie, and M. L. Harter, Science 237:1044-1046, 1987). In this study, we show that this protein is also capable of transactivating in vitro the adenovirus virus-associated (VA1) RNA gene transcribed by RNA polymerase III. Pertinent to the transcription of this gene is the rate-limiting component, TFIIIC, which appears to be of two distinct forms in uninfected HeLa cells. The addition of an oligonucleotide containing a TFIIIC binding site to HeLa whole-cell extracts inhibits VA1 transcription by sequestering TFIIIC. However, the addition of purified E1A to extracts previously challenged with the TFIIIC oligonucleotide restores the level of VA1 transcription. When included in the same reaction, an E1A-specific monoclonal antibody reverses the restoration. Incubation of purified E1A with either HeLa cell nuclear or whole-cell extracts alters the DNA-binding properties of TFIIIC as detected by gel shift assays. This alteration does not occur if E1A-specific antibody and E1A protein are added simultaneously to the extract. In contrast, the addition of this antibody to extracts at a later time does not reverse the alteration observed in the TFIIIC binding activities. Never at any time did we note the formation of novel TFIIIC-promoter complexes after the addition of E1A to nuclear extracts. These results clearly establish that E1A mediates its effect on VA1 transcription through TFIIIC in a very rapid yet indirect manner.(ABSTRACT TRUNCATED AT 250 WORDS)

Human papillomavirus type 16 E7 is considered to be a major viral oncoprotein playing an important role(s) in cervical cancers. E7 protein was shown to bind to the protein product of the retinoblastoma gene (RB), while simian virus 40 large T and adenovirus E1A were also shown to possess binding activity to RB protein. The RB protein is a cell cycle regulator that is highly phosphorylated specifically in S, G2, and M, whereas it is underphosphorylated in G0 and G1. Recently, large T was demonstrated to bind preferentially to the underphosphorylated RB protein, which is considered to be an active form restricting cell proliferation. However, it is not known whether E7 can bind to phosphorylated RB protein. We successfully purified large quantities of unfused human papillomavirus type 16 E7 protein expressed in Escherichia coli by using a T7 promoter-T7 RNA polymerase system. The purified E7 protein was demonstrated to bind preferentially to the underphosphorylated RB protein.

Previous work by our laboratory and others has shown that mouse cells normally resistant to tumor necrosis factor can be made sensitive to the cytokine by the expression of adenovirus E1A. The E1A gene can be introduced by either infection or transfection, and either of the two major E1A proteins, 289R or 243R, can induce this sensitivity. The E1A proteins are multifunctional and modular, with specific domains associated with specific functions. Here, we report that the CD1 domain of E1A is required to induce susceptibility to tumor necrosis factor cytolysis in adenovirus-infected mouse C3HA fibroblasts. Amino acids C terminal to residue 60 and N terminal to residue 36 are not necessary for this function. This conclusion is based on 51Cr-release assays for cytolysis in cells infected with adenovirus mutants with deletions in various portions of E1A. These E1A mutants are all in an H5dl309 background and therefore they lack the tumor necrosis factor protection function provided by the 14.7-kilodalton (14.7K) protein encoded by region E3. Western blot (immunoblot) analysis indicated that most of the mutant E1A proteins were stable in infected C3HA cells, although with certain large deletions the E1A proteins were unstable. The region between residues 36 and 60 is included within but does not precisely correlate with domains in E1A that have been implicated in nuclear localization, enhancer repression, cellular immortalization, cell transformation in cooperation with ras, induction of cellular DNA synthesis and proliferation, induction of DNA degradation, and binding to the 300K protein and the 105K retinoblastoma protein.

The E7 protein is one of the principle transforming proteins encoded by human papillomavirus type 16 (HPV16), a virus strongly associated with the development of cervical carcinoma. In the present study we show that cotransfection of wild-type human or murine p53 sequences with E7 and ras markedly reduces transformation in baby rat kidney cells, although no effect of p53 is seen on the ability of E7 to transform an established mouse line to anchorage independence. In contrast, expression of mutant p53 strongly potentiates the transforming function of E7 and confers marked growth factor independence to cells cotransformed by E7 and ras. These data suggest that E7 and p53 function in separate yet complementary biochemical pathways.

The E3/19K protein of human adenovirus type 2 is a resident of the endoplasmic reticulum (ER). Immediately after synthesis, it associates with major histocompatibility complex class I antigens and prevents their intracellular transport and cell surface expression. We have generated several C-terminal deletion mutants of the E3/19K protein that are preterminated at various positions on both sides of the membrane-spanning segment of the protein. One of these mutants is terminated at the luminal side of the membrane (M310), and two are terminated in the hydrophobic segment (M374 and M392), whereas mutant M621 is terminated on the cytoplasmic side of the ER membrane. The M310, M374, and M392 mutants are soluble proteins. They do not associate with HLA antigens in transfected 293 cells, and they are, to some extent, secreted into the medium. The M621 mutant protein is integrated in the ER membrane, associates immediately after its synthesis with HLA antigens, and exits from the ER. By using either an in vitro translation system supplemented with microsomes or overexpression in insect cells, we showed that M374 and E3/19K are able to associate with HLA antigens. These results indicate that the conformation of the luminal part of the E3/19K protein is not grossly altered by the mutations. Rapid transport of the M374 mutant out of the ER and partial degradation of this protein may prevent the interaction with HLA class I antigens in transfected 293 cells.

Previous experiments have demonstrated that the DNA-binding activity of the E2F transcription factor is increased upon adenovirus infection and that both the E1A and E4 genes are required for activation. In this study, we demonstrated that this enhanced binding of E2F to the E2 promoter is the result of two events. (i) There is stimulation of the DNA-binding activity of the E2F factor; this stimulation is E1A dependent but independent of E4. (ii) There is also induction of a stabilized interaction between E2F molecules bound to adjacent promoter sites; induction of stable E2F binding requires E4 gene function. This two-step activation process was also demonstrated in vitro. A heat-stable fraction from extracts of adenovirus-infected cells, which contains the 19-kilodalton E4 protein, was capable of stimulating stable E2F binding in an ATP-independent manner and appeared to involve direct interaction of the E4 protein with E2F. An extract from virus-infected cells devoid of the E4 19-kilodalton protein stimulated E2F DNA binding without forming the stable complex. This reaction required ATP. We conclude that activation of E2F during adenovirus infection is a two-step process involving a change in both the DNA-binding activity of the factor and the capacity to stabilize the interaction through protein-protein contacts.

EIa-dependent transactivation of the adenovirus EIIa early (EIIaE) promoter is correlated with the activation of the cellular transcription factor E2F. In this study we identified a cellular protein, C alpha, that is distinct from E2F and that binds two sites in the EIIaE promoter, one of which overlaps with the proximal E2F binding site of the EIIaE promoter. The possible involvement of C alpha in the EIa responsiveness of this promoter is discussed.

Phosphorylation at serine 89 was shown to be the major cause of the shift in gel migration of the 289R and 243R early region 1A (E1A) proteins of human adenovirus type 5. However, conversion of Ser-89 to alanine by site-directed mutagenesis did not abolish E1A transactivating or transforming activities, suggesting that phosphorylation at this site is not necessary for these E1A functions.

Affinity-purified medium T antigen encoded by NG59, a nontransforming mutant of polyomavirus, is specifically associated with a protein of 72,000 daltons (72K protein). Medium T antigens of wild-type polyomavirus and the transformation-competent mutant dl8 are not associated with the 72K protein. Instead, they form a complex with another protein of 61,000 daltons. Several lines of evidence suggest that the medium T antigen-associated 72K protein is equivalent to the abundant and constitutive 73K heat shock protein. First, on two-dimensional polyacrylamide gels the 72K protein migrated with the same pI (5.6) as did the 73K heat shock protein. Second, the 72K protein was immunoprecipitable with antibodies against heat shock proteins. Third, when digested with V8 protease, the 72K protein gave rise to the same pattern of fragments as did the 73K heat shock protein.

The adenovirus Ad5(pymT) has been used to express middle T antigen at very high levels in 293 cells. The middle T antigen produced was localized to membranes and was modified in the same way as that expressed in polyoma virus-infected mouse cells. It was phosphorylated in vivo on serine residues and in vitro on tyrosine residues. The in vivo phosphorylations occurred between residues 223 and 275. The middle T antigen encoded by A d5(pymT) was phosphorylated in vitro in a complex with human pp60c-src. Interestingly, the extreme overexpression of middle T antigen did not cause a parallel increase in the amount of complex; most of the pp60c-src remained unassociated. Immunoaffinity purification resulted in approximately 100 micrograms of middle T antigen from a 100-mm tissue culture dish. Several cell proteins copurified with the Ad5(pymT)-derived middle T antigen. Two of these, the 74- and 63-kilodalton species, are of particular interest because they were also purified from mouse tumors expressing middle T antigen.