A tumor host range selection procedure identifies p150(sal2) as a target of polyoma virus large T antigen.

Cancer cells may undergo loss or alterations in functions that certain viruses normally target to promote virus replication. Virus mutants that have lost the targeting function(s) should be able to grow in such cancer cells but not in normal cells. A "tumor host range" (t-hr) selection procedure has been devised and applied to polyoma virus based on this rationale. Studies of one t-hr mutant have led to the identification of the mSal2 gene product (p150(sal2)) as a binding partner of the large T antigen. mSal2 encodes a multizinc finger protein and putative transcription factor homologous to the Drosophila homeotic gene Spalt. The t-hr mutant encodes an altered large T protein that fails to interact with p150(sal2) and is defective in replication and tumor induction in newborn mice.

Polyoma virus: old findings and new challenges.

The accidental discovery of the mouse polyoma virus nearly 50 years ago opened up an experimental system unique in opportunities for investigating virus-host interactions leading to the development of tumors. Extensive studies of the virus in tissue culture have provided a detailed understanding of its genetics and molecular biology. Knowledge of the virus as a transforming agent in culture can now be tested in the animal where multiple cell types are targets for tumorigenic conversion and where a variety of host factors, both immunological and nonimmunological, come into play. Studies in the animal using well-characterized wild-type and mutant virus strains have led to some unexpected findings. Some of these run counter to certain widely held beliefs in cancer biology. This minireview focuses on these surprising findings and the challenges they raise.

Binding of p300/CBP co-activators by polyoma large T antigen.

Small DNA tumor viruses such as simian virus 40 (SV40) and polyomavirus (Py) take advantage of host cell proteins to transcribe and replicate their DNA. Interactions between the viral T antigens and host proteins result in cell transformation and tumor induction. Large T antigen of SV40 interacts with p53, pRb/p107/p130 family members, and the cyclic AMP-responsive element-binding protein (CREB)-binding protein (CBP)/p300. Py large T antigen is known to interact only with pRb and p300 among these proteins. Here we report that Py large T binds to CBP in vivo and in vitro. In co-transfection assays, Py large T inhibits the co-activation functions of CBP/p300 in CREB-mediated transactivation but not in NF-kappa B-mediated transactivation. p53 appears not to be involved in the functions of CREB-mediated transactivation and is not essential for large T:CBP interaction. Mutations introduced into a region of Py large T with homology to adenovirus E1A and SV40 large T prevent binding to the co-activators. These mutant large T antigens fail to inhibit CREB-mediated transactivation. The CBP/p300-binding Py mutants are able to transform established rat embryo fibroblasts but are restricted in their ability to induce tumors in the newborn mouse, indicating that interaction of large T with the co-activators may be essential for virus replication and spread in the intact host.

Early steps of polyomavirus entry into cells.

The mechanism by which murine polyomavirus penetrates cells and arrives at the nucleus, the site of viral replication, is not well understood. Simian virus 40 and JC virus, two closely related members of the polyomavirus subfamily, use caveola- and clathrin-mediated uptake pathways for entry, respectively. The data presented here indicate that compounds that block endocytosis of both caveola- and clathrin-derived vesicles have no effect on polyomavirus infectivity. Polyomavirus does not appear to colocalize with either clathrin light chain or caveolin-1 by immunofluorescence microscopy. Additionally, expression of a dominant-negative form of dynamin I has no effect on polyomavirus uptake and infectivity. Therefore, polyomavirus uptake occurs through a class of uncoated vesicles in a clathrin-, caveolin-1-, and dynamin I-independent manner.

Accelerated development of polyoma tumors and embryonic lethality: different effects of p53 loss on related mouse backgrounds.

Molecular evidence linking polyoma virus to p53 inactivation is thus far lacking, setting this highly oncogenic virus apart from other DNA tumor viruses. As a biological test for interaction, we studied the effects of p53 loss on development of virus-induced tumors. The absence of p53 led to more rapid tumor development on two different mouse backgrounds, indicating synergism between p53 loss and oncogenic pathways controlled directly by the virus. No effects of p53 on tumor type or frequency were noted. Polyoma tumor-derived cells in culture retained p53, and most of these showed induction of p21CIP1/WAF1 in response to DNA damage. These results indicate that p53 functions are not directly and fully impaired by the virus in the intact host. On one mouse background, it was discovered that loss of p53 resulted in complete embryonic lethality prior to 11 days of gestation. This lethality could be rescued by inclusion of gene(s) from a 129/SvJ background.

Wild-derived inbred mice have a novel basis of susceptibility to polyomavirus-induced tumors.

Polyomavirus induces a broad array of tumors when introduced into newborn mice of certain standard inbred strains, notably those bearing the H-2(k) haplotype. Susceptibility in these mice is conferred by an endogenous mouse mammary tumor virus superantigen (Mtv-7 sag) that acts to delete T cells required for polyomavirus-induced tumor immunosurveillance. In the present study we show that mice of two wild-derived inbred strains, PERA/Ei (PE) and CZECH II/Ei (CZ), are highly susceptible to polyomavirus but carry no detectable Mtv sag-related sequences and show no evidence of Vbeta deletion. C57BR/cdJ (BR) mice, which are H-2(k) but lack the endogenous Mtv-7, are highly resistant based on an effective anti-polyomavirus tumor immune response. When crossed with BR, both PE and CZ mice transmit their susceptibility in a dominant fashion, indicating a mechanism(s) that overrides the immune response of BR. Susceptibility in PE and CZ mice is not based on interference with antigen processing or presentation since cytotoxic T cells from BR can efficiently kill F(1)-derived tumor cells in vitro. The expected precursors of polyomavirus-specific cytotoxic T cells are present in both the wild inbred animals and their F(1) progeny. These findings indicate a novel basis of susceptibility that operates independently of endogenous superantigen and prevents the development of tumor immunity.

Discrimination between sialic acid-containing receptors and pseudoreceptors regulates polyomavirus spread in the mouse.

Variations in the polyomavirus major capsid protein VP1 underlie important biological differences between highly pathogenic large-plaque and relatively nonpathogenic small-plaque strains. These polymorphisms constitute major determinants of virus spread in mice and also dictate previously recognized strain differences in sialyloligosaccharide binding. X-ray crystallographic studies have shown that these determinants affect binding to the sialic acids. Here we report results of further experiments designed to test the importance of specific contacts between VP1 and the carbohydrate moieties of the receptor. With minor exceptions, substitutions at positions predicted from crystallography to be important in binding the terminal alpha-2,3-linked sialic acid or the penultimate sugar (galactose) destroyed the ability of the virus to replicate in cell culture. Substitutions that prevented binding to a branched disialyloligosaccharide were found to result in viruses that were both viable in culture and tumorigenic in the mouse. Conversely, substitutions that allowed recognition and binding of the branched carbohydrate chain inhibited spread in the mouse, though the viruses remained viable in culture. Mice of five different inbred strains, all highly susceptible to large-plaque virus, showed resistance to the spread of polyomavirus strains bearing the VP1 type which binds the branched-chain receptor. We suggest that glycoproteins bearing the appropriate O-linked branched sialyloligosaccharide chains are effective pseudoreceptors in the host and that they block the spread of potentially tumorigenic or virulent virus strains.

Radiation-resistant and radiation-sensitive forms of host resistance to polyomavirus.

Newborn mice of several inbred strains develop few or no tumors following inoculation with highly tumorigenic strains of polyomavirus. Here we show that such resistant strains can be divided into two groups based on the responses of adult mice to radiation followed by virus inoculation. Most strains show a radiation-sensitive form of resistance (Rr-s) and develop tumors following radiation and virus challenge. This type of resistance has previously been recognized as immunological, based on T-cell responses against virus-encoded neoantigen(s) expressed in tumor cells. Other strains exhibit a radiation-resistant form of resistance (Rr-r) and fail to develop tumors when treated in the same manner. Three additional properties of Rr-r mice distinguish them from Rr-s mice: (i) survival of newborns following inoculation with a highly virulent and usually lethal strain of virus, (ii) resistance to virus spread in newborns inoculated with either tumorigenic or virulent virus strains, and (iii) dominant or semidominant transmission of resistance in crosses with a highly susceptible strain. The Rr-r phenotype reflects a constitutive nonimmunological type of resistance that is targeted to the virus and blocks its dissemination.

Evidence of a role for phosphatidylinositol 3-kinase activation in the blocking of apoptosis by polyomavirus middle T antigen.

A polyomavirus mutant (315YF) blocked in binding phosphatidylinositol 3-kinase (PI 3-kinase) has previously been shown to be partially deficient in transformation and to induce fewer tumors and with a significant delay compared to wild-type virus. The role of polyomavirus middle T antigen-activated PI 3-kinase in apoptosis was investigated as a possible cause of this behavior. When grown in medium containing 1D-3-deoxy-3-fluoro-myo-inositol to block formation of 3'-phosphorylated phosphatidylinositols, F111 rat fibroblasts transformed by wild-type polyomavirus (PyF), but not normal F111 cells, showed a marked loss of viability with evidence of apoptosis. Similarly, treatment with wortmannin, an inhibitor of PI 3-kinase, stimulated apoptosis in PyF cells but not in normal cells. Activation of Akt, a serine/threonine kinase whose activity has been correlated with regulation of apoptosis, was roughly twofold higher in F111 cells transformed by either wild-type virus or mutant 250YS blocked in binding Shc compared to cells transformed by mutant 315YF. In the same cells, levels of apoptosis were inversely correlated with Akt activity. Apoptosis induced by serum withdrawal in Rat-1 cells expressing a temperature-sensitive p53 was shown to be at least partially p53 independent. Expression of either wild-type or 250YS middle T antigen inhibited apoptosis in serum-starved Rat-1 cells at both permissive and restrictive temperatures for p53. Mutant 315YF middle T antigen was partially defective for inhibition of apoptosis in these cells. The results indicate that unlike other DNA tumor viruses which block apoptosis by inactivation of p53, polyomavirus achieves protection from apoptotic death through a middle T antigen-PI 3-kinase-Akt pathway that is at least partially p53 independent.

Serine 257 phosphorylation regulates association of polyomavirus middle T antigen with 14-3-3 proteins.

Polyomavirus middle T antigen (MT) is phosphorylated on serine residues. Partial proteolytic mapping and Edman degradation identified serine 257 as a major site of phosphorylation. This was confirmed by site-directed mutagenesis. Isoelectric focusing of immunoprecipitated MT from transfected 293T cells showed that phosphorylation on wild-type MT occurred at near molar stoichiometry at S257. MT was previously shown to be associated with 14-3-3 proteins, which have been connected to cell cycle regulation and signaling. The association of 14-3-3 proteins with MT depended on the serine 257 phosphorylation site. This has been demonstrated by comparing wild-type and S257A mutant MTs expressed with transfected 293T cells or with Sf9 cells infected with recombinant baculoviruses. The 257 site is not critical for transformation of fibroblasts in vitro, since S257A and S257C mutant MTs retained the ability to form foci or colonies in agar. The tumor profile of a virus expressing S257C MT showed a striking deficiency in the induction of salivary gland tumors. The basis for this defect is uncertain. However, differences in activity for the wild type and mutant MT lacking the 14-3-3 binding site have been observed in transient reporter assays.

A novel 39-kilodalton membrane protein binds GTP in polyomavirus-transformed cells.

To investigate the possible involvement of GTP-binding proteins in transformation by the DNA tumor virus polyomavirus, the GTP-binding activities of Ras-like proteins and G protein alpha subunit proteins were examined in polyomavirus-transformed cells. No differences in the degrees or patterns of expression of Ras-like proteins were observed. However, a 39-kDa protein specifically bound GTP in membranes from polyomavirus-transformed cells. This protein was not seen in nontransformed or lytically infected cells or in phenotypically normal revertants of polyomavirus-transformed cells. It reappeared, however, in spontaneous retransformants derived from the revertants. The 39-kDa protein was not found stably associated with polyomavirus T antigens, nor was it phosphorylated on tyrosine. The 39-kDa protein was not recognized by an antiserum specific for members of the Gi alpha subfamily of G proteins or by antisera against all other known GTP-binding proteins of similar molecular mass. These results suggest that this novel 39-kDa GTP-binding membrane protein is observed as part of a long-term response that accompanies stable transformation by the virus.

Studies of partially transforming polyomavirus mutants establish a role for phosphatidylinositol 3-kinase in activation of pp70 S6 kinase.

Infection of mouse fibroblasts by wild-type polyomavirus results in increased phosphorylation of ribosomal protein S6 (D.A. Talmage, J. Blenis, and T.L. Benjamin, Mol. Cell. Biol. 8:2309-2315, 1988). Here we identify pp70 S6 kinase (pp70S6K) as a target for signal transduction events leading from polyomavirus middle tumor antigen (mT). Two partially transforming virus mutants altered in different mT signalling pathways have been studied to elucidate the pathway leading to S6 phosphorylation. An upstream role for mT-phosphatidylinositol 3-kinase (PI3K) complexes in pp70S6K activation is implicated by the failure of 315YF, a mutant unable to promote PI3K binding, to elicit a response. This conclusion is supported by studies using wortmannin, a known inhibitor of PI3K. In contrast, stable interaction of mT with Shc, a protein thought to be involved upstream of Ras, is dispensable for pp70S6K activation. 250YS, a mutant mT which retains a binding site for PI3K but lacks one for Shc, stimulates pp70S6K to wild-type levels. Mutants 315YF and 250YS induce partial transformation of rats fibroblasts with distinct phenotypes, as judged from morphological and growth criteria. Neither mutant induces growth in soft agar, indicating that an increase in S6 phosphorylation, while necessary for cell cycle progression in normal mitogenesis, is not sufficient for anchorage-independent cell growth. In the polyomavirus systems, the latter requires integration of signals from mT involving both Shc and PI3K.

Genetic and structural analysis of a virulence determinant in polyomavirus VP1.

The LID strain of polyomavirus differs from other laboratory strains in causing a rapidly lethal infection of newborn C3H/Bi mice. This virulent behavior of LID was attenuated by dilution, yet at sublethal doses LID was able to induce tumors at a high frequency, like its parent virus PTA. By constructing and assaying LID-PTA recombinant viruses and by DNA sequencing, the determinant of virulence in LID was mapped to the major viral capsid protein, VP1. The VP1s of LID and PTA differed at two positions: at 185, LID has phenylalanine and PTA has tyrosine, and at 296, LID has alanine and PTA has valine. Results obtained with viruses constructed by site-directed mutagenesis showed that alanine at position 296 is sufficient to confer a fully virulent phenotype regardless of which amino acid is at position 185. However, with valine at position 296, an effect of phenylalanine at position 185 is apparent, as this virus possesses an intermediate level of virulence. A crystal structure of polyomavirus complexed with 3'-sialyl lactose previously indicated van der Waals contacts between the side chain of valine 296 and the sialic acid ring (T. Stehle, Y. Yan, T. L. Benjamin, and S. C. Harrison, Nature [London] 369:160-163, 1994). When this interaction was modeled with alanine, these contacts were greatly reduced. Direct confirmation that the substitutions in VP1 affected receptor binding was obtained by studying virus hemagglutination behavior. The ensemble of results are discussed in terms of the idea that a lower affinity of the virus for its receptor can result in more rapid spread and increased pathogenicity.

Phosphatidylinositol 3-kinase binding to polyoma virus middle tumor antigen mediates elevation of glucose transport by increasing translocation of the GLUT1 transporter.

Elevation in the rate of glucose transport in polyoma virus-infected mouse fibroblasts was dependent upon phosphatidylinositol 3-kinase (PI 3-kinase; EC 2.7.1.137) binding to complexes of middle tumor antigen (middle T) and pp60c-src. Wild-type polyoma virus infection led to a 3-fold increase in the rate of 2-deoxyglucose (2DG) uptake, whereas a weakly transforming polyoma virus mutant that encodes a middle T capable of activating pp60c-src but unable to promote binding of PI 3-kinase induced little or no change in the rate of 2DG transport. Another transformation-defective mutant encoding a middle T that retains functional binding of both pp60c-src and PI 3-kinase but is incapable of binding Shc (a protein involved in activation of Ras) induced 2DG transport to wild-type levels. Wortmannin (< or = 100 nM), a known inhibitor of PI 3-kinase, blocked elevation of glucose transport in wild-type virus-infected cells. In contrast to serum stimulation, which led to increased levels of glucose transporter 1 (GLUT1) RNA and protein, wild-type virus infection induced no significant change in levels of either GLUT1 RNA or protein. Nevertheless, virus-infected cells did show increases in GLUT1 protein in plasma membranes. These results point to a posttranslational mechanism in the elevation of glucose transport by polyoma virus middle T involving activation of PI 3-kinase and translocation of GLUT1.

Susceptibility to tumors induced by polyoma virus is conferred by an endogenous mouse mammary tumor virus superantigen.

A dominant gene carried in certain inbred mouse strains confers susceptibility to tumors induced by polyoma virus. This gene, designated Pyvs, was defined in crosses between the highly susceptible C3H/BiDa strain and the highly resistant but H-2k-identical C57BR/cdJ strain. The resistance of C57BR/cdJ mice is overcome by irradiation, indicating an immunological basis. In F1 x C57BR/cdJ backcross mice, tumor susceptibility cosegregates with Mtv-7, a mouse mammary tumor provirus carried by the C3H/BiDa strain. This suggests that Pyvs might encode the Mtv-7 superantigen (SAG) and abrogate polyoma tumor immunosurveillance through elimination of T cells bearing specific V beta domains. DNA typing of 110 backcross mice showed no evidence of recombination between Pyvs and Mtv-7. Strongly biased usage of V beta 6 by polyoma virus-specific CD8+ cytotoxic T lymphocytes in C57BR/cdJ mice implicates T cells bearing this Mtv-7 SAG-reactive V beta domain as critical anti-polyoma tumor effector cells in vivo. These results indicate identity between Pyvs and Mtv-7 sag, and demonstrate a novel mechanism of inherited susceptibility to virus-induced tumors based on effects of an endogenous superantigen on the host's T cell repertoire.

Host range and cell cycle activation properties of polyomavirus large T-antigen mutants defective in pRB binding.

We have examined the growth properties of polyomavirus large T-antigen mutants that are unable to bind pRB, the product of the retinoblastoma tumor suppressor gene. These mutants grow poorly on primary mouse cells yet grow well on NIH 3T3 and other established mouse cell lines. Preinfection of primary baby mouse kidney (BMK) epithelial cells with wild-type simian virus 40 renders these cells permissive to growth of pRB-binding polyomavirus mutants. Conversely, NIH 3T3 cells transfected by and expressing wild-type human pRB become nonpermissive. Primary fibroblasts from mouse embryos that carry a homozygous knockout of the RB gene are permissive, while those from normal littermates are nonpermissive. The host range of polyomavirus pRB-binding mutants is thus determined by expression or lack of expression of functional pRB by the host. These results demonstrate the importance of pRB binding by large T antigen for productive viral infection in primary cells. Failure of pRB-binding mutants to grow well in BMK cells correlates with their failure to induce progression from G0 or G1 through the S phase of the cell cycle. Time course studies show delayed synthesis and lower levels of accumulation of large T antigen, viral DNA, and VP1 in mutant compared with wild-type virus-infected BMK cells. These results support a model in which productive infection by polyomavirus in normal mouse cells is tightly coupled to the induction and progression of the cell cycle.

Structure of murine polyomavirus complexed with an oligosaccharide receptor fragment.

The polyomaviruses are non-enveloped, icosahedrally symmetrical particles with circular double-stranded DNA genomes. The outer shell of the virion contains 360 copies of viral protein VP1 (M(r) approximately 42K) arranged in pentamers. We report here the structure at 3.65 A resolution of murine polyomavirus ('polyoma') complexed with an oligosaccharide receptor fragment. This structure has been determined using the previously described model of simian virus 40 (SV40). Although very similar in structure to SV40, polyoma has interesting biological differences. Cell-surface N-acetyl neuraminic acid (sialic acid) is required for polyoma infectivity, but not for SV40. Polyoma attaches to the surface of susceptible cells by stereospecific recognition of oligosaccharides terminating in (alpha 2,3)-linked sialic acid. Studies of pathogenicity show that the specificity of viral binding to such oligosaccharides is an important determinant of the virus' ability to establish a disseminated infection and to induce tumours in the natural host. The complex described here show how polyoma recognizes the receptor fragment and how strains with different receptor specificities can distinguish between alternative ligands. The results also suggest an explanation for the large disparity in pathogenicity exhibited by strains differing in only one amino-acid residue of VP1.

Synthesis and evaluation of 3-modified 1D-myo-inositols as inhibitors and substrates of phosphatidylinositol synthase and inhibitors of myo-inositol uptake by cells.

A number of 3-substituted 1D-myo-inositols were synthesized and evaluated as substrates for phosphatidylinositol synthase and uptake by intact cells. 1D-3-Amino-, -3-chloro-, and -3-(acetylthio)-3-deoxy-myo-inositols were all synthesized by nucleophilic displacement of the 6-O-(trifluoromethyl)sulfonyl group of 1L-1,2:3,4-di-O-cyclohexylidene-5-O-methyl-6-O-[(trifluoromethyl)-sulfon yl] - chiro-inositol (which was prepared from L-quebrachitol), respectively, by reaction with LiN3, followed by reduction of the azido function, and with LiCl and KSAc to give the O-protected compounds. O-Demethylation using BBr3 and concomitant acetal hydrolysis furnished the free-hydroxy 3-amino- and 3-chloro-3-deoxy-1D-myo-inositols. The 3-mercapto analogue was obtained by removal of the acetal groups of the acetylthio analogue, followed by acetylation and purification of the peracetate, and subsequent O-demethylation and deacetylation. The 3-deoxy derivative was synthesized from the 6-O-(imidazol-1-ylthiocarbonyl) compound via Barton-McCombie deoxygenation. The 3-azido derivative was directly synthesized from 1L-1-O-tosyl-chiro-inositol via displacement with azide. The 3-keto analogue was prepared by Pt-catalyzed air oxidation of 1L-chiro-inositol. The compounds were all evaluated as substrates for phosphatidylinositol (PtdIns) synthase from mouse brain. The 3-NH2, 3-F, 3-deoxy, and 3-keto analogues all showed activity as substrates, as measured by liberation of cytidine monophosphate. These compounds also showed inhibition of the reaction of myo-[3H]inositol with PtdIns synthase. These results taken together indicate that these compounds are likely to be incorporated into phospholipids. As a further indication that these compounds might be useful as probes for the PtdIns pathway, it was demonstrated that the 3-NH2, 3-F, and 3-deoxy compounds are taken up by intact fibroblast cells as evidenced by their competing with myo-[3H]inositol uptake.

Pyvs: a dominantly acting gene in C3H/BiDa mice conferring susceptibility to tumor induction by polyoma virus.

Results in a previous report established the presence of dominant genetic factor(s) in C3H/BiDa mice which contributed to the unusually high susceptibility of this strain to polyoma virus-induced tumors (Freund et al. (1992) Virology 191, 724-731). Here we show that C57BR/cdJ mice, while identical to C3H/BiDa at the H-2 locus, are almost completely lacking in susceptibility. Analysis of crosses between these two H-2k strains has shown that susceptibility is due to a single dominant autosomal gene, designated Pyvs. On an H-2k background, Pyvs acts in a highly penetrant manner to confer susceptibility to induction of a broad range of tumors by the virus. Analysis of F1 mice between C3H/BiDa and the highly resistant C57BL/6 (H-2b) strain shows that Pyvs can partially overcome the immunologically mediated resistance associated with an H-2b haplotype. Pyvs does not encode cell receptors for the virus, nor does it affect levels of virus replication or anti-viral humoral responses in the host. Studies with early passage embryo fibroblasts in culture show that Pyvs does not affect intracellular events essential for either productive infection or cell transformation by the virus. Pyvs thus determines a generalized susceptibility of the host to polyoma-induced tumors but apparently does not act at the level of target cells for tumor induction.

Changes in frequency, morphology, and behavior of tumors induced in mice by a polyoma virus mutant with a specifically altered oncogene.

Alterations in the tumor-inducing ability of a polyoma virus mutant encoding a partially defective middle T oncogene have been investigated. The mutant middle T associates with and activates the tyrosine protein kinase pp60c-src normally but does not promote binding of a second enzyme, phosphatidyl-inositol 3-kinase. Compared with the wild type virus, this mutant shows an altered and reduced ability to induce tumors after inoculation into newborn mice, as judged by the following criteria: lower frequency of tumors, reduced morbidity and increased survival times of host mice, changes in the spectrum of tumor types, and altered morphologic properties of tumors at several target organ sites. These results indicate an important role of changes in 3-phosphoinositide metabolism in induction of a variety of tumors in this experimental system.